The cover photo shows a water-rich melt inclusion in pegmatite quartz from Ehrenfriedersdorf/Germany, re-homogenized at 600°C and 1 kbar using the cold-seal pressure vessel technique. The vessel was pressurized with CO2. The melt inclusion glass contains 25wt% H2O determined with Raman spectroscopy. The results are given in the paper by Thomas, Kamenetsky and Davidson - example b of Table 2. The yellow color corresponds to 25 wt.% water and the dark green color at the basis corresponds to about 0 wt.% (quartz). The blue ridge at the right side comes from a small secondary fluid inclusion trail perpendicular to the quartz chip. The water distribution image was taken with the confocal high resolution Raman device LabRam HR800 using the 488 nm excitation Laser line. The image is composed of 750 single points, each 2µm. Behind each point is a complete Raman spectrum taken in the frequency range from 2800 to 4000 cm–1. For the quantification of the water we used the integral intensity of the broad asymmetric H2O-OH Raman band between 3100 and 3750 cm–1.
Distribution
Copies of Volume 36 and preceding volumes are obtainable from: The Business Manager Mineralogical Association of Canada P.O. Box 78087 Meriline Postal Outlet 1460 Merivale Road Ottawa, Ontario Canada K2E 1B1 e-mail:
[email protected] www.mineralogicalassociation.ca
Series Coordinator: R. Raeside Department of Geology Acadia University Wolfville, Nova Scotia Canada, B4P 2R6 Fax: +1 902-585-1816 e-mail:
[email protected] ISBN 0-921294-36-0 Copyright 2006 – Mineralogical Association of Canada Printed in Canada
CHAPTER 1: MELT INCLUSIONS IN PLUTONIC ROCKS: PETROGRAPHY AND MICROTHERMOMETRY R. J. Bodnar Fluids Research Laboratory, Virginia Tech, Blacksburg VA 24061, USA E-mail:
[email protected] and James J. Student Dept. of Geology, Central Michigan University, Mt. Pleasant, MI 48859, USA E-mail:
[email protected] Erzgebirge, Germany, were smaller than 20 µm, with a mean diameter of about 10 µm. R. Thomas (personal communication, 2006) also noted a correlation between inclusion size and volatile content, with more volatile-rich inclusions reaching 200 µm in diameter. Thomas also suggested that MI in Precambrian rocks are generally smaller than those in younger rocks. MI in plutonic rocks are commonly completely crystalline and contain a distorted bubble that is not always recognizable, and are much more likely to have reequilibrated following entrapment, owing to slow cooling and contact with subsolidus hydrothermal fluids. This chapter summarizes progress in the development of techniques to study MI from plutonic/intrusive rocks and focuses on MI from more silicic (granitic) environments (Table 1-1). Veksler (2006) discusses crystallized MI from gabbroic rocks, and Veksler & Lentz (2006) consider MI in carbonatites and related rocks. The application of MI in understanding pegmatite formation is discussed by Thomas et al. (2006). In a more general sense, this chapter focuses on MI that have experienced significant devitrification or crystallization following entrapment, or subsolidus aqueous alteration. This discussion thus applies to MI from the plutonic environment, as well as inclusions from the volcanic or extrusive environment that have undergone slow cooling and/or hydrothermal alteration.
INTRODUCTION Melt inclusions (MI) are small droplets of melt now containing some combination of crystals, glass and vapor that are trapped in crystals formed in magmas. Over the past few decades, the study of MI has matured to become an accepted technique to investigate melt evolution in volcanic systems (Clocchiatti 1975, Roedder 1979, Lowenstern 1995, 2003, Sobolev 1996, Frezzotti 2001, Hauri et al. 2001, Anderson 2003, Schiano 2003). MI in volcanic rocks are commonly large (>50 µm), glassy and contained in fresh and transparent minerals, and are generally easy to identify. MI in volcanic rocks generally provide consistent and reasonable results concerning the chemistry of melts at depth and provide the best tool available for assessing the volatiles in magmas (Anderson 1973, 1974, Lowenstern 1994, 1995, 2003). The best MI are often found in the most rapidly cooled parts of pyroclastic deposits and glassy rinds of lavas, which are not so easy to find (A.T. Anderson, Jr., personal communication, 2006). Compared to studies of MI in extrusive rocks, there are many fewer studies of MI from plutonic rocks, and there is still some uncertainty concerning the interpretation of data obtained from these samples. Roedder (1984) noted, “The lack of evident silicate melt inclusions in many igneous intrusive rocks, particularly those formed at greater depths, is puzzling.” Roedder further stated, “One of the major problems in the study of melt inclusions in intrusive rocks is the difficulty in recognizing the inclusions.” MI in intrusive rocks are usually small (5–20 µm) compared to those in extrusive rocks. Thomas et al. (1996) noted that about 80% of the MI in granitic rocks from the
IN THE BEGINNING …… The first detailed discussion of MI within a rigorous geological and petrological context was by Sorby (1858). In his classic 1858 paper on fluid and melt inclusions, Sorby introduced the section on
Mineralogical Association of Canada Short Course 36, Montreal, Quebec, p. 1-25.
1
2003 1984 2004 1997 2003 2001 1979 2001 2005 2000 1992 2000 1990 1986 2003 2004 1999 1995 1997 2004 1999 2004 2000 2002 2002 2004 1975 1988 1994a 1994b 1994c 1997 2000 2002 2003 2005 1998 2001 2004 1994
2
qtz Many qtz qtz, fspar, ap, tpz and zn qtz qtz qtz qtz qtz qtz, tpz qtz qtz, fspar qtz qtz cpy qtz qtz, fspar qtz qtz qtz qtz, tpz qtz qtz qtz, pyr qtz qtz qtz, ap qtz, tpz, ap, fspar, gar, fl qtz, K-fspar qtz, fspar, ap, tpz qtz zn qtz qtz qtz qtz qtz qtz
Xl, FI Gl, Xl Xl Xl Gl Xl Xl Xl, Gl, FI Gl, Xl Dev, crypto-Xl Gl + Xl + V Xl partly Xl, FI Gl, FI Xl Xl Mix, FI Xl Mix Xl Xl Xl Xl Xl Xl Xl, Mix, FI Gl, Xl Mix Xl, V Xl, FI Xl Xl Xl Xl Xl Xl Xl Gl Xl Xl, Mix
No Yes Yes Yes No Yes Yes Yes Some Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes No Yes Yes
Hmzd? (4) NA NR RQ CSVP (100) Furn NR NR NR HS HS (?) Furn HS Furn HS HS HS Furn, HS HS CSVP(?) (300) Furn CSVP (?) HS IHPV (500) HS, IHPV (500) CSPV(?) Furn, HS, CSPV (50-150) Furn, HS, CSPV (50-150) Furn, HS Furn, HS Furn Furn Furn Furn RQ CSVP (100) CSVP RQ CSVP (100) RQ CSVP (100), Furn HS NR Furn Furn
Heating Method (5) LA-ICPMS EMPA EMPA, Raman EMPA EMPA, LA-ICPMS EMPA None EMPA, SEM, Raman, LA-ICPMS EMPA, LA-ICPMS, SEM, PIXE, Raman EMPA, PIXE, SIMS EMPA EMPA EMPA EMPA PIXE SEM, PIXE SEM, EMPA, LA-ICPMS, Raman EMPA, SIMS EMPA EMPA, SEM, Raman, SIMS None EMPA EMPA, SEM EMPA, SIMS EMPA EMPA EMPA, Laser Probe-MA None None None None EPMA, SIMS EMPA, Raman EMPA, SIMS EMPA, SIMS, Raman EMPA, SIMS, Raman, SXRF SEM EMPA, SIMS EMPA, SIMS EPMA
Analyses Conducted (6)
(1). PCD = porphyry copper deposit; Peg = pegmatite; Rhy = rhyolite; Gran = granite; Monz = monzonite; QM = quartz monzonite; Dac = dacite; Alb = albite; zinn = zinnwalldite; (2). qtz=quartz; fspar = feldspar; ap = apatite; tpz = topaz; cpy = clinopyroxene; zn = zircon; pyr = pyroxene; and = andalusite; gar = garnet; fl = fluorite; (3). Xl = crystalline; Gl = glass; Dev = devitrified glass; Mix = glass + crystals; V = vapor; FI = fluid inclusions; (4). Were the inclusions heated to homogenization?; (5). Furn = 1-atmosphere furnace; HS = microscope heating stage; CSPV = cold seal pressure vessel (pressure in MPa); IHPV = internally heated pressure vessel (pressure in MPa);RQ = Rapid-quench; PCA = Piston cylinder apparatus; NR = not reported; NA = not applicable; (6). EMPA = electron microprobe; SIMS = secondary ion mass spectrometry; SEM = scanning electron microscope; Raman = Raman spectroscopy; LA-ICPMS = laser ablation inductively coupled plasma mass spectrometry; PIXE = particle induced X-ray emmission; SXRF = synchrotron X-ray fluorescence; MA = mass analyzer
Barren Gran Various LiF-rich Gran P-rich rare-metal Peg & Gran Anatectic metapelite Charnockitic orthogneiss Rapakivi Gran PCD Cu-Mo PCD PCD rhy-dac intrusive bodies Gran Topaz Gran peralkaline Gran Plutonic ejecta Quartz veins Dac porphyry Gran Syenitic dike in Cu-Au PCD Peralkaline Gran Granophyric blocks in rhy ejecta Tourmaline rich gem Peg Granodiorite Peralkaline Gran LiF-enriched Gran Peralkaline Gran PCD QM PCD QM Qtz Porphyry, Gran; Rhy Gran, porphyry, ongonite, kersanite Gran Variscan Gran Precambrian Gran Gran Tin Peg Tonalite Peraluminous Gran-Peg Tin Gran Gran, qtz monzodiorite Gran Xenoliths Alb-zinn porphyritic micro-Gran Deep crustal granitoid intrusions
MI Types (3)
Audétat & Pettke Bakumenko et al. Badanina et al. Breiter et al. Cesare et al. Chupin et al. Chupin et al. Davidson & Kamenetsky Davidson et al. Dietrich et al. Frezzotti Haapala & Thomas Hansteen & Lustenhouwer Harris Harris et al. Kamenetsky & Naumov Kamenetsky et al. Kovalenko et al. Lowenstern et al. Peretyazhko et al. Prokofiev et al. Reyf Reyf et al. Schmitt et al. Student Student & Bodnar Takenouchi & Imai Thomas Thomas Thomas Thomas Thomas & Klemm Thomas et al. Thomas et al. Thomas et al. Thomas et al. Vapnik Webster & Rebbert Webster et al. Yang & Bodnar
Host Minerals (2)
Reference
Rock Type/Environment (1)
OF STUDIES OF MELT INCLUSIONS IN SILICIC PLUTONIC ROCKS
Date
TABLE 1-1: SUMMARY
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
MELT INCLUSIONS IN PLUTONIC ROCKS: PETROGRAPHY AND MICROTHERMOMETRY
melt inclusions with the statement “The formation of crystals from a state of igneous fusion is in every respect analogous to what takes place when crystals are formed in water”. Later he stated “There is thus a most perfect analogy between glass- and stonecavities and fluid-cavities in every respect except the nature of the included substances”. The implication of this statement is that assumptions applied to the interpretation of FI should also apply to MI. Sorby referred to MI containing glass (± a vapor bubble) as glass-cavities; those in which the melt has crystallized are referred to as stonecavities. Sorby equated the stone-cavities to
aqueous FI that are cooled to low temperature and contain ice plus various salts and hydrates that were originally in solution. Sorby examined both glassy and crystallized MI and arrived at interpretations that are still valid today. Cameras were unavailable to Sorby to photograph features observed with the microscope, and he drew illustrative sketches of inclusions that he observed in thin sections and grain mounts. Several of Sorby’s drawings of crystallized MI from plutonic environments are shown in Figure 1-1. Much of Sorby’s understanding of MI is based on studies of
FIG. 1-1. Drawings of crystallized melt inclusions from Sorby (1858). The number above each drawing refers to the figure number in the original publication. Sorby did not include a scale as in modern publications but, rather, indicated the number of times the inclusions are magnified in linear dimensions, indicated by the number in parentheses following each description. (A) Crystallized melt inclusion in a crystal of iron silicate from a copper slag (X1600); (B) Crystallized melt inclusion in pyroxene from a blast furnace slag (X400), (C) Crystallized melt inclusion in feldspar in a xenolith from Vesuvius, Italy (X500); (D) Crystallized melt inclusion in quartz from Ponza, Italy (X400); (E) Crystallized melt inclusion in trachyte from Ponza, Italy (X800); (F) Altered melt inclusion from a porphyry from Arthur’s Seat, near Edinburgh, UK (X400); (G) Crystallized melt inclusion in quartz from an elvan near Penrhyn, Cornwall, UK (X250); (H) Crystallized melt inclusion in quartz from the granite at St. Austell, UK (X1000); (I) Altered melt inclusion from a porphyry from Arthur’s Seat, near Edinburgh, UK (X200); (J) Crystallized melt inclusion in quartz from an elvan near Penrhyn, Cornwall, UK (X800); (K) Crystallized melt inclusion in quartz from a coarse-grained granite near Cape Cornwall, UK (X800); (L) Crystallized melt inclusion in quartz from the granite at St. Austell, UK, with radiating fractures (X600).
3
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
inclusions in slags from iron furnaces in England (Fig. 1-1A, B). Sorby also described many natural crystallized MI, including “stone-cavities” from the island of Ponza (Italy) (Fig. 1-1D, E). Fedele et al. (2003) studied similar samples from Ponza and heated the MI to produce a homogeneous glass that was then analyzed by electron microprobe. The results were used to develop a model for igneous petrogenesis in the Ponza trachyte. Sorby was among the first to recognize that coexisting FI and MI indicated entrapment in a volatile-saturated magma. Today melt-volatile immiscibility is recognized as an important process in many magmatic–hydrothermal systems (Roedder 1992), ranging from shallow granites (Frezzotti 1992) to porphyry copper deposits (Bodnar 1995, Kamenetsky et al. 1999, Davidson & Kamenetsky 2001) to orthomagmatic Au–Te deposits such as Cripple Creek, CO (Webster 2004). Referring to inclusions trapped in a volatile-saturated melt, Sorby concluded “Some crystals might be deposited from solution in the highly heated water, and catch up small portions of the fused stone [to form melt inclusions], whilst others might be formed by the crystallization of the melted stone, and catch up small portions of liquid water [to form aqueous inclusions]”. He went on to state “… we may, I think, conclude that the crystals would contain glass- [glass melt inclusions] or stone-cavities [crystallized melt inclusions], and perhaps gas- and vapour-cavities [fluid inclusions], ....” Today the occurrence of coexisting MI and FI that were trapped from the same magma is considered to be unequivocal evidence for melt–volatile immiscibility in the magma, as was documented at Ascension Island by Roedder & Coombs (1967) based on coexisting MI and halite-rich FI. In fact, one could conclude that an assumption of volatile saturation in the absence of coexisting MI and FI is highly questionable.
recognizing completely crystallized MI in plutonic rocks (Roedder 1984, Yang & Bodnar 1994), and MI that have been altered by later hydrothermal processes (Frezzotti 1992, Varela 1994, Student & Bodnar 2004). Altered, crystallized MI are especially common in the porphyry copper deposits (Student & Bodnar 2004), where the rock has undergone subsolidus alteration associated with hydrothermal activity during formation of the ore deposit. In this environment, planes of secondary aqueous FI commonly intersect the MI, and the compositions of the MI reflect alteration by the hydrothermal fluid (Frezzotti 1992, Student & Bodnar 2004). Crystallized MI usually appear as opaque, poorly defined patches at low magnification (Fig. 1-2). At higher magnification, the inclusions appear to be dark masses of crystals lacking sharp borders between the MI and the host (Fig. 1-3). Under crossed polars the inclusions often show a few bright spots representing birefringent daughter minerals, or an overall sparkly appearance in finergrained inclusions. In many cases it may be difficult to distinguish between crystallized MI and trapped solids in the host (Rapien et al. 2003). The largest MI in plutonic rocks are generally less than several 10s of micrometres in longest dimension, whereas MI up to several hundred micrometres are not uncommon in extrusive rocks.
PETROGRAPHIC ANALYSIS OF MELT INCLUSIONS IN PLUTONIC ROCKS Sorby noted that MI in plutonic rocks are often difficult to identify because “In the quartz of very coarse grained granites the stone cavities [crystallized melt inclusions] are generally obscure and of irregular shape….” and “Those in the feldspar are often so much obscured by the partial decomposition of that mineral, that it is difficult to distinguish them from small decomposed patches [of feldspar]…” (see Fig. 1-1F-L). More recent workers have also discussed the difficulty in
FIG. 1-2. Quartz phenocryst from Stage IV granodiorite from the Tyrone, New Mexico, porphyry copper deposit (Student & Bodnar 2004). The crystal is approximately 2 mm across. The phenocryst contains abundant crystallized MI, including hourglass inclusions (Anderson 1991), large, randomly distributed MI (larger opaque areas in crystal) and numerous, small MI along a growth zone near the outer edge of the crystal.
4
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
FIG. 1-3. Examples of melt inclusions in quartz phenocrysts from Stage IV granodiorite from the Tyrone, New Mexico, porphyry copper deposit (Student & Bodnar 2004). All of the inclusions occur in the same phenocryst, which is approximately 2 mm across (see Figure 1-2). The scale bar, shown in (Q) represents 30 µm and applies to all images. MI that occur along growth zones (G) are generally smaller than those that are randomly distributed throughout the crystal. Some MI (K, P, Q) appear to have nucleated on solid inclusions in the phenocryst; other melt inclusions (M–O) have anomalously large vapor bubbles, suggesting either entrapment of vapor along with melt or later reequilibration. Some MI (D, E) have fractures that extend into the surrounding quartz.
Many workers have discussed processes that lead to the formation of crystallized MI (Roedder 1979, 1984, Lowenstern 1995, Frezzotti 2001). In general, for a given inclusion size, the slower the cooling rate, the more likely that the melt in the inclusions will crystallize. Similarly, for a given cooling rate, larger inclusions are more likely to crystallize compared to smaller inclusions. In addition to cooling rate and size, the composition of the melt may affect the crystallization behavior of MI (Roedder 1984). Student & Bodnar (1999) noted that synthetic MI trapped under H2O-saturated conditions in the quartz-saturated haplogranite system were often partially to completely crystallized (Fig. 1-4), even though the samples were quenched from formation conditions (720°C and 200 MPa) to room temperature in a few
minutes. These workers suggested that the high H2O contents of the melts, and the fact that the melts were saturated in H2O at the time of trapping, might have promoted crystallization during cooling. Inclusion Selection The most important aspect of any MI or FI study is determining which inclusions to study and whether those inclusions record the physical and chemical conditions at the time of trapping. As such, the timing of inclusion trapping relative to formation of the host phase, and the position of the host within the overall paragenesis, must be constrained. There is surprisingly little discussion in the literature concerning the temporal classification of MI. Sobolev & Kostyuk (1975) described the use of MI in studies of magmatic crystallization,
5
R. J. BODNAR & J.J. STUDENT
and summarized the different temporal occurrences of MI. These workers distinguished between zonal and azonal primary MI. Zonal inclusions define a growth zone and there is little debate concerning their primary origin. [Note, however, that a primary origin does not guarantee that the inclusions record the physical and chemical conditions of crystal growth, as the inclusions may have reequilibrated following entrapment, as described in more detail below.] Zonal inclusions are common in some minerals, such as nepheline, plagioclase (Halter et al. 2004b) and pyroxene (Yang & Scott 2002, Rapien et al. 2003) (Fig. 1-5), but are less common in other minerals such as olivine (Anderson 1974) and zircon (Thomas et al. 2002). Isolated or randomly distributed inclusions that cannot be associated with a specific growth feature are referred to as azonal inclusions (Fig. 1-6). Such isolated inclusions are generally considered to be primary if no evidence of fracturing or mineral dissolution (that could allow secondary inclusions to form) is observed. Sobolev & Kostyuk (1975) emphasized that a negative crystal shape is not in itself sufficient evidence for a primary origin of MI, as has also been noted for FI (Roedder 1984, Bodnar 2003a). Some minerals, such as quartz (Anderson et al. 2000, Halter et al. 2004a) commonly contain both zonal and azonal MI (Fig. 1-2). Several workers (Yang & Scott 2002, Student & Bodnar 2004) have noted that within a given phenocryst zonal inclusions tend to be considerably smaller than azonal MI. Within the FI community, much effort has been devoted to the issue of inclusion selection. A procedure has been developed that allows one to be confident that the FI selected are related to the process being studied, and that the inclusions have not been affected by later events. The first step is to identify a group of FI that were all trapped at the same “time” and, by inference, at the same temperature and pressure and from fluids of the same composition. This group of inclusions, representing the most finely discriminated trapping event that can be identified based on petrography, is referred to as a fluid inclusion assemblage or FIA (Goldstein & Reynolds 1994). The amount of time represented by an FIA will vary, depending on the geologic environment. Thus, the amount of time represented by FI trapped along a growth surface of a halite crystal forming as a result of diurnal temperature variations in a sabkha environment is less than 24 hours. Similarly, the amount of time required to heal a fracture in a high temperature
FIG. 1-4. Natural (top) and synthetic (bottom) crystallized MI. The natural MI is from a quartz phenocryst in Stage IV granodiorite from the Tyrone, New Mexico, porphyry copper deposit (Student & Bodnar 2004). The synthetic MI is in quartz and was trapped at 720°C and 200 MPa under H2O-saturated conditions (Student & Bodnar 1999).
FIG. 1-5. Photomicrograph of primary MI outlining growth zones in a pyroxene phenocryst from the White Island, New Zealand, volcano (from Rapien et al. 2003).
6
MELT INCLUSIONS IN PLUTONIC ROCKS: PETROGRAPHY AND MICROTHERMOMETRY
the host crystal. However, MI may be preferentially trapped during episodes of anomalous and rapid crystal growth associated with changing temperature and/or pressure and/or volatile content in the magma (Roedder 1984), effectively decreasing the amount of time represented by the inclusion assemblage. It is important to emphasize that the fluid inclusion assemblage definition implies nothing about the temporal relationship between the inclusions and the host mineral. An FIA may be composed of either primary or secondary or pseudosecondary inclusions, The FIA concept does not constrain when the inclusions were trapped relative to growth of the host mineral (other than that they must have been trapped during or after the host formed). Thus, secondary (or pseudosecondary) FI along a well-defined fracture represent an FIA that was formed after precipitation of the bulk of the mineral in which those inclusions occur. While the FIA approach to selecting inclusions to study is used routinely in studies of FI, an analogous method has not been developed for MI. However, because similar petrographic techniques are (and should) be used to identify groups of MI that were trapped at the same time, we propose the term melt inclusion assemblage (MIA) to describe a group of MI trapped at (essentially) the same time and, by analogy, at the same temperature and pressure and from a melt of the same composition. Thus, MI along an individual growth zone in a phenocryst, such as those shown in Figures 1-2 and 1-5, represent MIAs containing primary MI. Yang & Bodnar (1994) studied MI in quartz from deep crustal granodiorite plutons in the Gyeongsang Basin, South Korea, and identified various MIAs based on petrographic analysis of the samples. One type of MIA occurred near the interface between plagioclase and quartz crystals that extended into vugs, while other MIAs were observed along growth surfaces near the outer portions of quartz crystals (Fig. 1-7). In both cases, petrographic evidence for simultaneous trapping of the MI in the MIA was conclusive (although there is some uncertainty as to whether the MIA at the plagioclase–quartz interface represents primary or secondary MI). Compared to FI, secondary MI are much less common. Thus, at a minimum the MI in a given phenocryst can be considered to be an MIA composed of inclusions that were all trapped during the (unknown) amount of time required for the phenocryst to grow.
FIG. 1-6. (Top) Quartz phenocryst from the Bishop Tuff (upper left) containing numerous randomly distributed (azonal) glassy MI. (Bottom) Transmitted light photographs of melt inclusions (MI) in zircons from the Quottoon Igneous Complex, British Columbia, Canada (from Thomas et al. 2002).
hydrothermal environment to produce a plane of secondary FI might be on the order of days to weeks (Brantley 1992, Sterner & Bodnar 1984, Bodnar & Sterner 1987). Dowty (1980) and Bacon (1989) compiled crystal growth rates from melts for various minerals. In general, growth rates fall in the range from about 10–2 to 10–11 cm/s. These growth rates would require 0.1 sec to 1011 sec (3.17 x 103 years) to trap a 10 µm MI. Tomiya & Takahashi (2005) estimated average growth rates for plagioclase and pyroxene in the magma beneath Usu volcano (Japan) of 0.1 to 0.7 µm/a. Thus, an FIA composed of 10 µm MI along a growth zone in a plagioclase or pyroxene phenocryst from this system would represent a minimum of about 15–100 years. This amount of time assumes a constant growth rate for 7
R. J. BODNAR & J.J. STUDENT
evaluating the quality of data obtained from MI is to identify MI that represent a MIA, as described below (Fig. 1-8). If MI do not occur in growth zones or along fractures, it is sometimes possible to group MI that show similar petrographic characteristics and, by inference, trapped a melt of the same composition. Anderson et al. (2000) noted that quartz phenocrysts from the Bishop Tuff contain clear, faceted MI in the interior, and round, brown MI close to the edge of the crystal. These workers used this zonal arrangement to infer the sequence of trapping – thus the clear MI would be assigned to one MIA, and the brown MI to a later MIA, both associated with growth of the crystal. Recently, many workers have successfully used cathodoluminescence (CL) analysis to identify FI that belong to the same FIA (Landtwing et al. 2005). Peppard et al. (2001) used CL to reveal growth zones in quartz from the Bishop Tuff, and used this information to identify MI with a common origin. CL is a powerful technique that can be used to reveal growth textures in phases that show no such features during normal petrography, and its use in MI studies is expected to increase. MICROTHERMOMETRY In FI studies, once a FIA has been identified, the next step is to determine if the inclusions record the physical and chemical conditions of trapping. To address this, the FIA must be tested to determine if the inclusions adhere to “Roedder’s Rules”, which are based on criteria first proposed by Sorby (1858). According to Roedder (1984), FI record the original trapping conditions if the following requirements are met: 1. the inclusions trap a single, homogeneous phase, 2. nothing has been added to or removed from the inclusion following trapping, 3. the inclusion represents an isochoric (constant volume) system. Note that the requirement of constant volume applies to the volume of the original cavity in which the fluid was trapped. During cooling of a MI to room temperature some (considerable?) amount of material may precipitate on the inclusion walls, resulting in a volume that is apparently smaller than the original MI volume at the trapping conditions. If this material is incorporated back into the melt when the MI is heated to homogenization (see Fig. 1-9), the condition of constant volume is satisfied. The condition would not be satisfied if the host mineral surrounding the MI deformed
FIG. 1-7. Schematic representation of the distribution of mostly crystallized MI in quartz from deep crustal granitoid intrusions from the Gyeongsang Basin, South Korea. Abundant small (<8 µm) MI occur in the quartz near the boundary between euhedral quartz and alkali feldspar (A). Larger (>12 µm) MI show random or isolated distribution (B). The outer growth zone of the quartz is decorated by numerous MI (C). Some MI in the outermost portion of the crystal contain devitrified glass (D). The scale corresponds to the crystal only – MI have been drawn larger for illustrative purposes. (from Yang & Bodnar 1994).
It is relatively easy to assign MI along growth surfaces or other interfaces to the same MIA. However, many minerals or phenocrysts contain only one or a few melt inclusions, and these are commonly distributed irregularly within the grain (Fig. 1-6). Where two or more randomly distributed MI occur within a given crystal, petrographic observations usually do not provide sufficient information to determine if the inclusions represent an MIA, i.e., were trapped at the same time and at the same temperature and pressure and from a melt of the same composition. This distinction is important because the first step in 8
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
Identify a Melt Inclusion Assemblage (MIA) based on petrography
Stop
No
Consistent Phase Relations?
No Why do MI have variable phase relations and/or microthermometry?
Yes
Homogenize MI: Consistent Microthermometry?
No Mixed Trapping Reequilibration in nature
Yes
Reequilibration in the laboratory
Analyze MI
MI do not represent an MIA
Interpret Results
FIG. 1-8. Flow chart showing the steps that should be followed to ensure that data obtained from MI represent conditions in the magma at the time of trapping. The first step is to identify MI that were trapped at the same time, and at the same temperature and pressure and from a melt of the same composition, and thus represent a melt inclusion assemblage (MIA). Then, petrographic, microthermometric and chemical data are evaluated to test for reequilibration of the MI after trapping.
FIG. 1-9. Series of photomicrographs showing the behavior of two crystallized MI in quartz from the Red Mountain, Arizona, USA, porphyry copper deposit during heating. Evidence of melting is first observed at about 675°C. At 790°C the inclusions contain a vapor bubble and a small feldspar crystal (not visible). At 810°C both the vapor bubble and feldspar disappear (from Student & Bodnar 2004).
9
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
1400°C range. These workers heated the sample at a rate of 30–40°C/minute for temperatures above 900°C and held the sample at the maximum temperature for 7 minutes before quenching. They noted that slower heating rates or longer time at the maximum temperature resulted in an increased Th and oxidation of the sample. Lowenstern (1994) heated MI in quartz phenocrysts from Pantelleria, Italy, in a one atmosphere heating stage at a rate of 50°C up to 600°C. The sample was held at 600°C for 10–20 minutes, then heated to 700°C and maintained at this temperature for 10 minutes. The sample was then heated in 50°C increments with 10 minute equilibration periods after each step up to 800°C. At higher temperatures 25°C heating steps were used. Lowenstern (1994) noted that these heating rates were likely too rapid, resulting in observed Th that were 25–75°C higher than the actual temperatures. Reyf (1997) heated small (3–5 µm) MI in quartz from a granite in a microscope heating stage, using heating steps of 30–50°C, and holding the sample at temperature for 1.5–2 hours after each step. Although this technique involved long heating experiments, the results were consistent with previous work that suggested host granites were hypersolvus and crystallized outside of the region of immiscibility in the Ab–An–Or system at 50–100 MPa. Massare et al. (2002) studied MI in olivine phenocrysts and found that when heated at one atmosphere the Th increased systematically with time as a result of deformation of the host crystal and loss of H2O from the MI. Danyushevsky et al. (2002) conducted a detailed study of the effect of heating schedule on the measured Th. They found that Th decreases with decreasing heating rate and approaches a constant value. Using this kinetic technique, the “correct” heating rate for each host phase and melt composition can be determined. Using faster heating rates will result in measured Th that are too high, while using heating rates that are slower than that at which the Th levels off increases the likelihood that the inclusion composition will change as a result of diffusion of components out of (or into?) the inclusion (Qin et al. 1992, Danyushevsky et al. 2000, Gaetani & Watson 2000). One of the most detailed studies of the effect of heating rate, inclusion size and volatile content on the homogenization behavior of MI is that of Thomas (1994a). Thomas used this information to develop a method to determine a minimum homogenization temperature for any given MI, which he interpreted to represent the trapping temperature of the MI.
plastically during heating to produce a volume that was larger than the original MI volume (Bodnar 2003b). How does one test the inclusions in a FIA to determine if “Roedder’s Rules” are satisfied? Studies of FI have shown rather conclusively that when FI re-equilibrate following entrapment the microthermometric and chemical properties of the FI show a wide range compared to inclusions that have not changed (Bodnar 2003b). Thus, if all of the inclusions within an FIA have the same number of phases and in the same volume proportions when observed at room temperature, and if the temperatures of phase changes are the same in all of the inclusions in the assemblage (indicating that the inclusions all have the same composition), one can be reasonably certain that the inclusions record the original formation conditions. While one can usually determine if all of the FI in an FIA contain the same phases and are in the same volume proportions, this is usually not possible with crystallized MI because the individual phases within the MI cannot be discerned (see Fig. 1-3). As such, crystallized MI must be homogenized and analyzed to determine if all of the inclusions in the MIA have the same composition and phase behavior during heating. Techniques that are commonly used to analyze MI include electron microprobe, SIMS (Layne 2006), FTIR, Raman spectroscopy and laser ablation ICP–MS (Pettke 2006, Halter et al. 2006). All of these techniques (except LA–ICP–MS; Halter et al. 2002) require a homogeneous glass, thus necessitating that the MI be heated to homogenization and quenched. Several different techniques have been used to homogenize MI. These techniques can broadly be divided into those in which the MI are heated at one atmosphere confining pressure and those in which the MI are heated under an elevated confining pressure. These techniques may be further divided based on whether the MI are heated continuously in one step or are heated incrementally. The most commonly used technique to homogenize MI is to heat the inclusions at one atmosphere in a microscope-mounted heating stage (Clocchiatti 1975, Magakyan et al. 1993, Lowenstern 1994, Reyf 1997, Fedele et al. 2003). The heating is usually conducted in an inert or reduced gas (He±H2) atmosphere to prevent oxidation of the sample. Magakyan et al. (1993) heated MI in clinopyroxene from boninite in a one atmosphere stage and report an error of ±20°C for homogenization temperatures (Th) in the 900– 10
MELT INCLUSIONS IN PLUTONIC ROCKS: PETROGRAPHY AND MICROTHERMOMETRY
Rather than heating MI in a microscopemounted heating stage, some workers heated inclusions at one atmosphere in a tube furnace (Yang & Bodnar 1994, Webster et al. 1997, Raia et al. 2000, Thomas & Webster 2000, Stockstill et al. 2005). While this technique is similar to heating at one atmosphere in a microscope-mounted stage, it does not allow MI to be monitored continuously during heating to determine temperatures of phase changes and to look for “anomalous behavior” during heating. Some workers heated the sample in one step to the final temperature (Thomas & Webster 2000). Other workers (Yang & Bodnar 1994, Student & Bodnar 2004) heated the samples incrementally and observed the sample under the microscope between heating steps to monitor the melting behavior (Fig. 1-9). Heating MI at one atmosphere pressure, either in a microscope-mounted stage or in a tube furnace, works well for MI with low concentrations of volatiles. However, if the MI contain significant amounts of H2O (or CO2?) the inclusions will commonly decrepitate before Th is reached, owing to the high internal pressures generated during heating. As a result, many workers heat crystallized volatile-rich MI only until the solid phases have dissolved, producing an inclusion that contains melt plus vapor bubble (Fig. 1-10), or heat the MI in a pressure vessel (Skirius et al. 1990, Webster & Duffield 1991, Anderson et al. 2000, Audétat et al. 2000, Thomas et al. 2002, Student & Bodnar 2004) or in a pressurized microscope stage (Massare & Clocchiatti 1987). This technique minimizes decrepitation of volatile-rich MI, although
Anderson et al. (2000) reported that some CO2 was lost from MI as a result of heating at 800–900°C under ≈200 MPa of Ar pressure for 20 hours. Student & Bodnar (1999) investigated the effect of heating technique on the observed Th, using synthetic silicate MI trapped at known P–T conditions. These workers found that, regardless of the technique used, Th increased with increasing inclusion size (Fig. 1-11), suggesting that homogenization of MI is a diffusion-controlled process, as previously argued by Thomas (1994a) and Thomas et al. (1996). MI heated in one step at one atmosphere in a tube furnace produced Th that most closely matched known temperatures. Inclusions heated continuously in a one atmosphere stage with a heating rate of 1°C/minute were about 10–15°C higher than the correct temperature, and those heated continuously at 3°C/minute were about 25°C too high (Fig. 1-12). An additional complication associated with crystallized MI from magmatic–hydrothermal ore deposits is that the host phase is often altered and/or crosscut by numerous planes of aqueous FI. Many of these planes intersect MI, and promote the decrepitation of MI during heating. Student & Bodnar (2004) tested various methods to homogenize MI in quartz phenocrysts containing abundant planes of secondary FI, including heating in a microscope heating stage, heating in a one atmosphere vertical tube furnace, and heating under pressure in a cold-seal pressure vessel. The method that proved most satisfactory to homogenize crystallized MI in phenocrysts with abundant planes of aqueous FI involved heating
FIG. 1-10. Melting sequence of a crystallized MI in plagioclase in trachyte from Ponza, Italy. At room temperature the inclusion consists of a mass of intergrown, fine-grained crystals ± glass. By 1092°C most of the silicate minerals have melted leaving only a mass of fine-grained opaque minerals, which melt between 1092 and 1197°C. After quenching, the inclusion contains a homogeneous glass and a vapor bubble (from Fedele et al. 2003).
11
R. J. BODNAR & J.J. STUDENT
870 3°C/
860
min
Homogenization Temperature (°C)
1° C / m
in
850 Frequency 5
840
4
3
2
1
3 °C /mi n Thermal Cycling
830
Th= 819-839 °C n=16
820
810
Tu b e F u r n a c e
Tra p p i n g Te m p e r a t u r e 800 ± 5 °C
800
T h = 8 0 0 ± 1 0 °C n=18
790 1 °C /mi n T h = 8 0 6 - 8 5 3 ° C n = 1 6 3 °C /mi n T h = 8 1 9 - 8 6 3 ° C n = 1 6 780
0
10
20
30
50
40
60
70
80
90 100 110
Inclusion Area (mm2) FIG. 1-11. Effect of inclusion size and heating technique on the measured homogenization temperature (Th) of synthetic silicate MI. The Th ranges for heating in a tube furnace (at one atmosphere) and the thermal cycling experiments are shown by the shaded boxes, along with the experimental conditions. These experiments did not consider inclusion size. A histogram of frequency versus Th for the thermal cycling experiment is inset along the right side of the diagram. Homogenization temperature as a function of inclusion area (as viewed through the microscope) for continuous heating experiments (1°C/min and 3°C/min) are indicated with unfilled and filled circles, respectively. Black vertical tie lines link data for the same inclusion. The solid line drawn through the data points for the continuous heating experiments shows the general relationship between inclusion size and Th, with Th approaching the trapping temperature as inclusion size approaches zero. The Th for the tube furnace experiment coincides most closely with the known trapping temperature (from Student & Bodnar 1999).
12
MELT INCLUSIONS IN PLUTONIC ROCKS: PETROGRAPHY AND MICROTHERMOMETRY
400
(a)
350
Pressure (MPa)
Haplogranite minimum curve
Entrapment P-T estimate
300 250
Trapping Conditions 800 °C and 200 MPa
200
e
or
uid
150
h oc
is
fl Cl
Fluid inclusion Th
100
%
.9 11
wt
Na
Melt inclusion Th range
50
200
300
400
500
600
700
800
900
1000
Temperature °C
Pressure (MPa)
230
(b)
3 °C/min Thermal Cycling 3 °C/min Continuous Heating
220
1 °C/min Continuous Heating 210
Tube Furnace
11.9 wt% NaCl fluid isochore
200
Trapping Conditions 800 ± 5 °C, 200 ± 5 MPa
190 750
800
850
Temperature °C FIG. 1-12. Estimated P–T formation conditions for coexisting melt and aqueous synthetic inclusions, calculated using vapor/melt Th for the four different heating experiments. The top diagram shows the P–T estimate for the ≈3°C/min thermal cycling heating experiment, showing the aqueous inclusion liquid/vapor curve and isochore to illustrate the "intersecting isochore" technique (Roedder & Bodnar 1980) used to determine formation pressure. The bottom diagram shows the P–T estimates for four different heating experiments shown in Figure 1-11 (from Student & Bodnar 1999).
13
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
samples under an elevated confining pressure. Phenocrysts were placed in a 5 mm long, 5 mm OD platinum capsule and loaded into a pressure vessel. A small hole was punched into the capsule bottom and top to permit the argon pressure medium to freely enter the capsule at run conditions to prevent collapse of the capsule and possible crushing of the sample. The vessel was sealed and pressurized with argon to 30–50 MPa, then lowered into a preheated furnace. Once the pressure reached 100 MPa during heating, argon was continuously bled off such that the sample was heated isobarically to the final run temperature. The technique for homogenizing MI in a pressure vessel involves heating in one step to some elevated temperature and holding the sample at this temperature for a sufficient amount of time to assure homogenization of a significant proportion of the inclusions. In general, the smaller inclusions in the sample homogenize first (at lower temperature and/or shorter run durations). Thomas et al. (1996) observed similar behavior. To determine the minimum run time and temperature required to homogenize a significant portion of the inclusions while at the same time minimizing the number of MI that decrepitate, a few phenocrysts from one sample were heated incrementally and examined after each heating step to determine the homogenization progress (Fig. 1-9). In most samples, a large portion of the MI homogenized over a range of a few tens of degrees – at this point the heating experiment was stopped. If heating were continued in an attempt to homogenize all inclusions in the phenocryst, those with lower Th would have decrepitated. Once the minimum Th was determined in this way, several phenocrysts from the same sample were heated to that temperature in one step and held for 24 hours. Heating under pressure significantly reduced decrepitation of MI. However, even under these conditions the larger inclusions (greater than about 30–50 µm) still decrepitated. Other workers (Sterner & Bodnar 1989, Skirius et al. 1990, Schmidt et al. 1998) have previously shown that heating FI or MI under confining pressure eliminates (or minimizes) decrepitation. Additionally, Massare & Clocchiatti (1987) reported that, when MI are heated in a pressurized microscope heating stage, Th of rhyolitic MI in sanidine decrease by 70°C/100 MPa in the temperature range 560– 850°C. As noted above, MI containing H2O-rich compositions often decrepitate, even when heated
under an elevated pressure. Decrepitation results because the internal pressure in the MI exceeds the strength of the host mineral (Bodnar 2003b). Thomas (1994b) described qualitatively the P–T path followed by H2O-rich MI during heating to the solidus and during melting, and Student & Bodnar (1996) quantified the effect of H2O on the P–T path followed by MI during cooling (or heating). The partial molar volume of H2O in hydrous melts is less than the molar volume of H2O in the vapor phase, and this difference becomes greater at lower pressures where the molar volume of H2O is large. As a result, as H2O exsolves from the melt phase in an MI of constant volume (i.e., assumes the volume of the host phase does not change with changing temperature and pressure), the pressure in the inclusion increases. For example, if an MI is trapped on the H2O-saturated solidus at 50 MPa and 782°C, H2O will begin to exsolve from the melt as the sample cools and the MI begins to crystallize feldspar and quartz (Fig. 1-13A). [This assumes that phase equilibrium is maintained during cooling. While synthetic MI do appear to maintain equilibrium during heating and cooling, it is unknown whether this applies to natural MI.] The pressure in the MI will continue to increase from 50 MPa as crystallization proceeds and the P–T path follows the H2O-saturated solidus, reaching a maximum pressure of about 150 MPa when the MI is completely crystallized at 708°C (Fig. 1-14A, point II). With further cooling the MI will follow the isochore corresponding to the H2O density in the crystallized inclusion (path II – I, Fig. 1-14A). The high pressures generated as the MI cools along the H2O-saturated solidus may cause the quartz host surrounding the MI to fracture with loss of H2O. The relative pressure increase during cooling decreases with increasing trapping pressure owing to the smaller difference between the partial molar volume of H2O in the melt and the molar volume of the free H2O phase. Thus, the pressure in a MI trapped on the H2O-saturated solidus at 200 MPa and 682°C increases to only about 230 MPa as the sample cools along the H2O-saturated solidus (Fig. 1-14B). Water-rich MI may not decrepitate during cooling in nature, owing to the elevated confining pressure. However, when heated in the laboratory to homogenize the MI in preparation for later analysis, the MI will follow the reverse of the P–T path followed during cooling in nature. Thus, the MI trapped on the H2O-saturated solidus at 50 MPa and 782°C will generate an internal pressure of about 14
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
FIG. 1-13. Calculated phase behavior of crystallized MI that trapped an H2O-saturated melt in the haplogranite system at 50 (A) and 200 (B) MPa. The calculated pressure in the MI during cooling from trapping conditions (and heating to homogenization) assumes a partial molar volume for H2O in the melt of 22 cm3/mole. The heating sequences A & B correspond to the P–T paths A & B, respectively, shown in Fig. 1-14, and the heating increments labeled I – V correspond to the P–T points I – V on Fig. 1-14. See text for additional details. (from Student & Bodnar 1996).
FIG. 1-14. Calculated P–T path during heating of crystallized MI that trapped an H2O-saturated melt in the haplogranite system at 50 (A) and 200 (B) MPa. The calculated pressure in the MI during cooling from trapping conditions (and heating to homogenization) assumes a partial molar volume for H2O in the melt of 22 cm3/mole. The P–T paths shown on A & B correspond to the heating sequences A & B, respectively, shown in Fig. 1-13, and the P–T points labeled I – V correspond to the heating increments I – V on Fig. 1-13. See text for additional details. (from Student & Bodnar 1996).
15
R. J. BODNAR & J.J. STUDENT
inclusions within the FIA, and is confirmed by consistent microthermometric results from all FI in the assemblage. For crystallized MI, it is generally not possible to determine the phase relations of the MI owing to the poor optics of MI and host mineral. In this case, the inclusions within the MIA should be tested for consistency in microthermometric and compositional data. As an example, Figure 1-15 shows a quartz phenocryst from the Red Mountain, Arizona, porphyry copper deposit with a zone of primary MI trapped along a growth (or possibly resorbtion) surface. A portion of this phenocryst was heated under 100 MPa confining pressure as described above, and the phase relations of 86 MI were monitored. It was previously determined (Student & Bodnar 2004) that the MI were trapped on the H2O-saturated solidus at about 810°C. Thus, when heated, the last daughter mineral (in this case a feldspar crystal) and the vapor bubble should disappear at the same temperature (i.e., similar to the MI in Fig 1-13A), which also represents the trapping temperature. Thirty-seven of the 86 inclusions from the MIA showed this behavior (Fig. 1-15), indicating that these 37 inclusions trapped only the H2O-saturated melt phase and did not reequilibrate after trapping. This conclusion is based on the fact that any change in composition and/or volume of the MI would have produced variability in the mode and temperature of homogenization (Bodnar 2003b). Thirty of the 86 inclusions contained a feldspar crystal of varying
230 MPa before melting begins (i.e., before the path intersects the H2O-saturated solidus) (point II, Fig. 1-14A). Once melting begins, the path follows the H2O-saturated solidus and the pressure in the inclusion decreases with continued heating as H2O dissolves into the melt phase. SUMMARY OF PETROGRAPHY AND MICROTHERMOMETRY The first step in any MI study is to identify a melt inclusion assemblage, or MIA, that represents a group of inclusions that were all trapped at the same time. By extension, this requirement implies that all of the MI in the assemblage trapped a melt of the same composition and at the same temperature and pressure. It is important to emphasize that a MIA is identified based initially on petrographic analysis of the sample, and not on MI compositions. Subsequent analysis of the MI in the assemblage can help to confirm that the inclusions being studied do indeed represent a MIA, and to distinguish between those MI that trapped a single, homogeneous melt and maintained the original composition during later cooling in nature and heating in the lab, and those that have not. After an MIA has been identified, the next step is to determine if the MI record the original physical and chemical conditions of trapping. For FI, this is most easily accomplished based on observation of consistent phase relations in all the
FIG. 1-15. Results of heating experiments on 86 crystallized MI from a single growth (or resorbtion) zone in a quartz phenocryst from the Red Mountain, Arizona, porphyry copper deposit. See text for details.
16
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
sizes after heating to 810°C. These inclusions are interpreted to have trapped H2O-saturated melt plus a feldspar crystal. Twelve of the MI contained both feldspar and a vapor bubble after heating to 810°C, indicating that these inclusions either trapped feldspar and vapor along with the melt, or reequilibrated, following entrapment. The remaining 7 MI had various combinations of melt, crystals and vapor bubble present after heating to 810°C, again suggesting mixed trapping and/or reequilibration. Thus, analyses of any of these 49 MI that did not show simultaneous dissolution of feldspar and vapor bubble at 810°C would provide a melt composition that is not representative of the melt that was present at the time of trapping. Random selection of MI in this MIA would provide erroneous (and misleading) information if the behavior of the MI during heating was not monitored before analyses were conducted. Observations of MI behavior during laboratory heating thus provide a test of assumptions regarding the timing of inclusion trapping and possible reequilibration. The protocol that one should follow in the selection of MI for study is outlined in Figure 1-8. To summarize, one should first select two or more MI that define a melt inclusion assemblage (MIA), based on petrographic examination of the sample. MI in the same growth zone or along a healed fracture are examples of MI occurrences that could be used to define an MIA. If MI cannot be related to a MIA, one should reconsider whether a MI study should be undertaken, as it is unlikely that one will be able to argue convincingly that the results obtained represent the melt that was present at the time of MI formation. After an MIA is identified, the MI should be examined to determine if they contain consistent phase relations (Fig. 1-8). If not, the MIA should be abandoned because the MI have either trapped mixtures of phases or have reequilibrated after trapping, or both. In either case, it is unlikely that the MI will provide useful information concerning the physical and chemical conditions in the magma. If the phase relations cannot be determined at room temperature because of poor optics or because the phases are too finegrained, the MI should be heated to homogenization. Assuming that the inclusions trapped a single, homogeneous melt phase, all MI should show similar temperatures of phase changes during heating. If the MI show variability in the order in which various phases disappear (i.e., solids and vapor bubble) and/or in the temperatures of
phase changes, the MIA should be abandoned because, as with the example of inconsistent phase relations described earlier, the MI do not represent the original melt that was present at the time of trapping. Those MI in the MIA that show similar phase behavior and temperatures of phase changes, such as the 37 MI in the Red Mountain sample described above, are most likely to have trapped the melt that was present and have not reequilibrated following trapping. It is these inclusions, and only these inclusions, that should be selected for further analysis and interpretation. By applying these simple tests and selecting only inclusions that satisfy the criteria outlined above, one can obtain data that accurately reflect conditions in the magma at the time that the MI were trapped. MELT INCLUSIONS IN PORPHYRY-TYPE DEPOSITS A plutonic environment in which MI have contributed significantly in recent years is in understanding ore-forming processes in porphyrytype deposits (see Student & Bodnar 2004 for review). Porphyry copper deposits are associated with epizonal siliceous intrusions emplaced at convergent margins. While FI studies have improved our understanding of the nature and role of magmatic–hydrothermal fluids in porphyry systems (Roedder 1971, Nash 1976, Bodnar 1995, Beane & Bodnar 1995, Roedder & Bodnar 1997, Davidson & Kamenetsky 2001), it has been only recently that workers have begun to study the melt MI in these systems. Student & Bodnar (2004) summarized a protocol for studying crystallized MI in samples that have undergone extensive subsolidus hydrothermal alteration, such as those in porphyry copper deposits. The technique produces glassy MI with a homogeneous composition that are amenable to analysis by a variety of techniques, including synchrotron XRF, PIXE, SIMS, electron microprobe, Raman spectroscopy, FTIR spectroscopy and laser ablation ICP–MS. Importantly, the ore metal content of MI representing different stages in the magmatic history can be determined, and relative differences between pre-, syn- and post-mineralization samples (Fig. 1-16) can be compared with models for metal partitioning between melt and coexisting magmatic–hydrothermal fluids (Candela & Holland 1986, Candela 1989, 1997, Cline & Bodnar 1991, Lynton et al. 1993: Candela & Piccoli 1995). 17
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
FIG. 1-16. Comparison of Zn and Cu concentrations in MI from the porphyry copper deposits at Red Mountain, Arizona, and Tyrone, New Mexico, with concentrations in MI from the White Island volcano, New Zealand. Metal concentrations in pre-mineralization quartz latite at Red Mountain and at White Island are higher than those in syn- and post-mineralization MI from Red Mountain and Tyrone. (from Student & Bodnar 2004). quartz
quartz
A
RM QL MI
B
MI RM D2 Ty QM MI ph c
ylli
ylli
ph c RMD2
potassic
albite
orthoclase
albite
orthoclase
FIG. 1-17. Compositions of MI from the porphyry copper deposits at Red Mountain, Arizona (A), and Tyrone, New Mexico (B). Hydrothermal fluids associated with phyllic alteration in porphyry copper deposits lie along the quartz–potassium feldspar (orthoclase) join near the quartz apex, whereas fluids associated with potassic alteration lie near the orthoclase apex. MI compositions at Red Mountain produce a trend (arrows, Fig. 1-17A) that projects towards the composition of fluids associated with phyllic alteration, whereas MI at Tyrone show trends (arrows, Fig. 1-17B) that project to both phyllic and potassic alteration fluids. (from Student & Bodnar 2004).
18
MELT INCLUSIONS IN PLUTONIC ROCKS: PETROGRAPHY AND MICROTHERMOMETRY
As noted above, MI in porphyry copper and related deposits are often intersected by one or more planes of FI. It is thought that most of these fractures occur at subsolidus conditions associated with hydrothermal alteration and may affect the compositions of MI. This interpretation is supported by the observation that compositions of MI fall along trends that project to compositions of fluids associated with potassic and/or phyllic alteration in porphyry copper deposits (Fig. 1-17). An interesting application of MI is to compare the geochemistry of magmatic systems that host (or have the potential to host) economic mineralization and those that are barren or subeconomic. As an example, Rapien et al. (2003) studied MI from the White Island, New Zealand, volcano and compared the results to bulk rock compositions in productive and barren porphyry intrusions. These workers concluded that the White Island magma has the potential to generate economic porphyry copper-type mineralization, but that the magmatic system has not evolved to the productive stage, i.e., the magmatic–hydrothermal system is too “young”. Similarly, Audétat & Pettke (2003) studied MI and coexisting FI from two barren plutons in New Mexico, USA. They concluded that the absence of mineralization was related to the low salinity of the exsolving magmatic fluids, resulting in less efficient extraction of metals from the melt. Grancea et al. (2001) studied MI in mineralized and barren intrusions in Romania and found that MI in mineralized systems were enriched in S and had a lower Al/(K+Na+2Ca) compared to barren systems. Kamenetsky et al. (1999) studied mixed silicate glass and crystalline silicate–sulfate– carbonate–sulfide–oxide inclusions from the Dinkidi Cu–Au porphyry deposit, Philippines. The inclusions are enriched in ore metals, and they interpreted the inclusions to have originally formed early in the ore-forming process as a result of immiscibility. Harris et al. (2003) observed coexisting silicate MI and high-salinity and vaporrich FI in magmatic-hydrothermal quartz veins from the Bajo de la Alumbrera porphyry copper deposit, Argentina. The close association of these three different inclusion types was interpreted to represent melt–aqueous fluid immiscibility, and compositions of the coexisting inclusions were used to calculate bulk partition coefficients for ore metals between the melt and magmatic aqueous phase. Schmitt et al. (2002) studied MI in peralkaline
granites from the Amis Complex in Namibia. The MI are enriched in Nb and REE and proved that the peralkaline composition and rare metal enrichments are primary magmatic features and not the result of later hydrothermal activity. SUMMARY This chapter attempts to summarize a methodology to recognize, select and study MI in plutonic rocks. In general, MI from this environment are small, mostly to completely crystallized, and difficult to recognize during normal petrographic observations. Crystallized MI must be homogenized to confirm that all MI in a melt inclusion assemblage show similar modes and temperatures of phase changes and, thus, likely trapped samples of the original homogeneous melt and did not reequilibrate following entrapment. The technique that proves most reliable to homogenize crystallized, volatile-rich MI is to heat the samples in one step in a pressure vessel under an elevated confining pressure. The most important take-home message from this chapter concerns the selection of MI to study. Numerous studies of natural and synthetic FI and MI show clearly that the best evidence that inclusions have trapped a single homogeneous phase and have not reequilibrated following entrapment is if all of the MI or FI in an assemblage show the same phase behavior and temperatures of phase changes. Trapping mixtures of phases, leakage, or change in the volume of the inclusion, all result in a wide range in phase relations and temperatures of phase changes. Thus, if a group of MI that were all trapped at the same time (MIA) show similar phase behavior and temperatures of phase changes (including homogenization), one can have a high level of confidence that the MI provide information on the physical and chemical environment at the time of trapping. ACKNOWLEDGEMENTS During the past decade many students, colleagues and visitors to the Fluids Research Laboratory at Virginia Tech have contributed to our understanding of melt inclusions in plutonic environments. We wish to acknowledge contributions from Andreas Audétat, Claudia Cannatelli, Benedetto DeVivo, Luca Fedele, John Mavrogenes, Maria Rapien, Nobu Shimizu, Karen Stockstill, Csaba Szabo, Jay Thomas, Rainer 19
R. J. BODNAR & J.J. STUDENT
equilibrium crystallization of magmas, according to data obtained by studies of melt inclusions. Soviet Geology and Geophysics (Geologiya i Geofizika) 25, no. 8, 73–81.
Thomas and Kyounghee Yang. Jim Reynolds led us to appreciate the importance of petrography in the selection of fluid and melt inclusions. The authors also thank Fred Anderson, Ilya Veksler and Jim Webster for comments and suggestions on an earlier version of this manuscript. Funding for recent work on FI and MI was provided by NSF Grants EAR0001168, EAR-0125918, and EAR-0337094 to RJB.
BADANINA, E.V., VEKSLER, I.V., THOMAS, R., SYRITSO, L.F. & TRUMBULL, R.B. (2004): Magmatic evolution of Li–F, rare-metal granites: A case study of melt inclusions in the Khagnilay complex, Eastern Transbaikalia (Russia). Chem. Geol. 210, 113–134.
REFERENCES
BEANE, R.E. & BODNAR, R.J. (1995): Hydrothermal fluids and hydrothermal alteration in porphyry copper deposits. In Pierce, F.W., and Bohm, J.G., eds., Porphyry Copper Deposits of the American Cordillera. Arizona Geological Society Digest 20, Tucson, Arizona, pp. 83–93.
ANDERSON, A.T., JR. (1973): The before-eruption water content of some high-alumina magmas. Bull. Volcanology 37, 530–552. ANDERSON, A.T., JR. (1974): Evidence for a picritic, volatile-rich magma beneath Mt. Shasta, California. J. Petrol. 15, 243–267.
BODNAR, R.J. (1995): Fluid inclusion evidence for a magmatic source for metals in porphyry copper deposits: In Magmas, Fluids and Ore Deposits (J.F.H. Thompson, ed.) Mineral. Assoc. Can. Short Course 23, 139–152.
ANDERSON, A.T., JR. (1991): Hourglass inclusions: Theory and application to the Bishop rhyolitic Tuff. Am. Min. 76, 530–547. ANDERSON, A.T., JR. (2003): Melt (glass ±crystals) inclusions. In Fluid Inclusions: Analysis and Interpretation (I. Samson, A. Anderson & D. Marshall, eds.) Mineral. Assoc. Can. Short Course 32, 353–364.
BODNAR, R.J. (2003a): Introduction to fluid inclusions. In Fluid Inclusions: Analysis and Interpretation (I. Samson, A. Anderson & D. Marshall, eds.) Mineral. Assoc. Can. Short Course 32, 1–8.
ANDERSON, A.T., JR., DAVIS, A.M. & LU, F. (2000): Evolution of Bishop Tuff rhyolitic magma based on melt and magnetite inclusions and zoned phenocrysts. J. Petrol. 41, 449–473.
BODNAR, R.J. (2003b): Reequilibration of fluid inclusions. In Fluid Inclusions: Analysis and Interpretation (I. Samson, A. Anderson & D. Marshall, eds.) Mineral. Assoc. Can. Short Course 32, 213–230.
AUDÉTAT, A., GÜNTER, D. & HEINRICH, C.A. (2000): Magmatic–hydrothermal evolution in a fractionating granite: A microchemical study of the Sn–W–F-mineralized Mole Granite (Australia). Geochim. Cosmochim. Acta 64, 3373–3393.
BODNAR, R.J. & STERNER, S.M. (1987): Synthetic fluid inclusions. In Hydrothermal Experimental Techniques, G.C. Ulmer and H.L. Barnes, eds., Wiley–Interscience, New York. p.423–457. BRANTLEY, S.L. (1992): The effect of fluid chemistry on quartz microcrack lifetimes. Chem. Geol. 113, 145–156.
AUDÉTAT, A. & PETTKE, T. (2003): The magmatichydrothermal evolution of two barren granites: a melt and fluid inclusion study of the Rio de Medio and Cañada Pinatebe plutons in northern New Mexico (USA). Geochim. Cosmochim. Acta 67, 97–121.
BREITER, K., FRYDA, J., SELTMANN, R. & THOMAS, R. (1997): Mineralogical evidence for two magmatic stages in the evolution of an extremely fractionated P-rich rare-metal granite: the Podlesi stock, Krusne hory, Czech Republic. J. Petrol. 38 1723–1739.
BACON, C.R. (1989): Crystallization of accessory phases in magmas by local saturation adjacent to phenocrysts. Geochim. Cosmochim. Acta 53, 1055–1066.
CANDELA, P.A. (1989): Magmatic ore-forming fluids: Thermodynamic and mass transfer calculations of metal concentrations, In Ore Deposition Associated with Magmas (J.A.
BAKUMENKO, I.T., BAZAROVA, T.YU., PANINA, L.I., SIMONOV, V.A. & KOSUKHIN, O.I. (1984): Distinctive features of the equilibrium and non-
20
MELT INCLUSIONS IN PLUTONIC ROCKS: PETROGRAPHY AND MICROTHERMOMETRY
Whitney & A.J. Naldrett eds.) Reviews in Economic Geology 4, Economic Geology, El Paso, TX (223–233).
DANYUSHEVSKY, L.V., MCNEILL, A.W. & SOBOLEV, A.V. (2002): Experimental and petrological studies of melt inclusions in phenocrysts from mantle-derived magmas: an overview of techniques, advantages and complications. Chem. Geol. 183, 5–24.
CANDELA, P.A. (1997): A review of shallow, orerelated granites: Textures, volatiles and ore metals. J. Petrol. 38, 1619–1633.
DAVIDSON, P. & KAMENETSKY, V.S. (2001): Immiscibility and continuous felsic melt–fluid evolution within the Rio Blanco porphyry system, Chile: Evidence from fluid inclusions in magmatic quartz. Econ. Geol. 96, 1921–1929.
CANDELA, P.A. & HOLLAND, H.D. (1986): A mass transfer model for copper and molybdenum in magmatic hydrothermal systems: Origin of porphyry-type ore deposits. Econ. Geol. 81, 1–19. CANDELA, P.A. & PICCOLI, P.M. (1995): Model oremetal partitioning from melts into vapor and vapor–brine mixtures. In Thompson, J.F.H. (ed.), Magmas, Fluids and Ore Deposits, Mineralogical Association of Canada, Short Course 23, 101– 127.
DAVIDSON, P., KAMENETSKY, V., COOKE, D.R., FRIKKEN, P., HOLLINGS, P., RYAN,C., VAN ACHTERBERGH, E., MERNAGH, T., SKARMETA, J., SERRANO, L. & VARGAS, R. (2005): Magmatic precursors of hydrothermal fluids at the Rio Blanco Cu–Mo deposit, Chile: links to silicate magmas and metal transport. Econ. Geol. 100, 963–978.
CESARE, B., MARCHESI, C., HERMANN, J. & GOMEZPUGNAIRE, M.T. (2003): Primary melt inclusions in andalusite from anatectic graphitic metapelites; implications for the position of the Al2SiO5 triple point. Geology 31, no. 7, 573–576.
DIETRICH, A., LEHMANN, B. & WALLIANOS, A. (2000): Bulk rock and melt inclusion geochemistry of Bolivian tin porphyry systems. Econ. Geol. 95, 313–326.
CHUPIN, V.P., BAKUMENKO, I.T. & SOBOLEV, V.S. (1979): Crystallized melt inclusions in quartz or rapakivi granite. Transactions (Doklady) of the U.S.S.R. Academy of Sciences: Earth Science 248, no. 1–6, 109–111
DOWTY, E. (1980): Crystal growth and nucleation theory and the numerical simulation of igneous crystallization. In Physics of Magmatic Processes (R.B. Hargraves, ed.), Princeton University Press, Princeton, 419–485.
CHUPIN, V.P., SOROKINA, G.A. & BARTON JR., J.M. (2001): Late Archean melt inclusions in zircon from charnockitic orthogneiss of Bulai Pluton (Limpopo Belt, South Africa) (abs.). Proceedings of ECROFI XVI, Porto 2001, 93–94.
FEDELE, L., BODNAR, R.J., DEVIVO, B. & TRACY, R.J. (2003): Melt inclusion geochemistry and computer modeling of trachyte petrogenesis at Ponza, Italy. Chem. Geol. 194, 81–104.
CLINE, J.S. & BODNAR, R.J. (1991): Can economic porphyry copper mineralization be generated by a typical calc-alkaline melt? J. Geophys. Res. 96, 8113–8126.
FREZZOTTI, M.L. (1992): Magmatic immiscibility and fluid phase evolution in the Mount Genis granite (southeastern Sardinia, Italy). Geochim. Cosmochim. Acta 56, 21–33.
CLOCCHIATTI, R. (1975): Les inclusions vitreuses des cristaux de quartz. Étude optique, thermooptique et chimique. Applications geologiques. [Glass (melt) inclusions in quartz. Optical, microthermometric and chemical study. Geologic applications] Memoires de la Société Géologique de France, LIV, 122, 1–96 [in French].
FREZZOTTI, M.L. (2001) Silicate melt inclusions in magmatic rocks: applications to petrology. Lithos 55, 273–299. GAETANI, G.A. & WATSON, E.B. (2000): Open system behavior of olivine-hosted melt inclusions. Earth & Planet. Sci. Letters 183 27–41.
DANYUSHEVSKY, L.V., DELLA-PASQUA, F.N., & SOKOLOV, S. (2000): Re-equilibration of melt inclusions trapped by magnesian olivine phenocrysts from subduction-related magmas: petrological implications. Contr. Min. Pet. 138, 68–83.
GOLDSTEIN, R.H. & REYNOLDS, T.J. (1994): Systematics of Fluid Inclusions in Diagenetic Minerals. Society for Sedimentary Geology Short Course 31, Tulsa, Oklahoma, 199 pp.
21
R. J. BODNAR & J.J. STUDENT
KAMENETSKY, V.S. & NAUMOV, V.B. (2004): Immiscibility between silicate magmas and aqueous fluids; a melt inclusion pursuit into the magmatic–hydrothermal transition in the Omsukchan Granite; NE Russia. Chem. Geol. 210, 73–90.
GRANCEA, L., CUNEY, M. & LEROY, J.L. (2001): Mineralized versus barren intrusions: a melt inclusion study in Romania’s gold quadrilateral. C. R. Acad. Sci. Paris, Sciences de la Terre et des planets 333, 705–710. HAAPALA, I. & THOMAS, R. (2000): Melt inclusions in quartz and topaz of the topaz granite from Eurajoki, Finland. J. Czech. Geol. Soc. 45, 149– 154.
KAMENETSKY, V.S., WOLFE, R.C., EGGINS, S.M., MERNAGH, T.P. & BASTRAKOV, E. (1999): Volatile exsolution at the Dinkidi Cu–Au porphyry deposit, Philippines: A melt–inclusion record of the initial ore-forming process. Geology 27, 691–694.
HALTER, W.E. & HEINRICH, C.A. (2005): Magmatic processes and volatile phase generation in porphyry-type environments: A laser ablation– ICP–MS study of silicate and sulfide melt inclusions. In Melt Inclusions in Plutonic Rocks (J.D. Webster, ed.) Min. Assoc. Can. Short Course 36, 151-164.
KOVALENKO, V.I., TSARYEVA, G.M., GOREGLYAD, A.V., YARMOLYUK, V.V., TROITSKY, V.A., HERVIG, R.L. & FARMER, G.L. (1995): The peralkaline granite-related Khaldzan–Buregtey rare-metal (Zr, Nb, Ree) deposit, Western Mongolia. Econ. Geol. 95, 313–326.
HALTER, W.E., HEINRICH C.A. & PETTKE T. (2004b): Laser ablation ICP–MS analysis of silicate and sulfide melt inclusions in an andesitic complex II: evidence for magma mixing and magma chamber evolution. Contr. Min. Pet. 147, 397–412.
LANDTWING, M.R., PETTKE, T., HALTER, W.E., HEINRICH, C.A., REDMOND, P.A., EINAUDI, M.T. & KUNZE, K. (2005): Copper deposition during quartz dissolution by cooling magmatic– hydrothermal fluids: The Bingham porphyry. Earth & Planetary Sci. Letters 235, Issues 1–2, 229–243
HALTER, W.E., PETTKE T. & HEINRICH C.A. (2004a): Laser ablation ICP–MS analysis of silicate and sulfide melt inclusions in an andesitic complex I: analytical approach and data evaluation. Contr. Min. Pet. 147, 385–396.
LAYNE, G.D. (2006): Application of secondary ion mass spectrometry to the determination of traditional and non-traditional light stable isotopes in melt inclusions. In Melt Inclusions in Plutonic Rocks (J.D. Webster, ed.) Min. Assoc. Can. Short Course 36, 27-49.
HALTER W.E., PETTKE T., HEINRICH C.A. & ROTHEN-RUTISHAUSER B. (2002): Major to trace element analysis of melt inclusions by laser ablation–ICP–MS; methods of quantification. Chem. Geol. 183, 63–86.
LOWENSTERN, J.B. (1994): Dissolved volatile contents in an ore-forming magma. Geology 22, 893–896.
HANSTEEN, T.H. & LUSTENHOUWER, W.J. (1990): Silicate melt inclusions from a mildly peralkaline granite in the Oslo paleorift, Norway. Min. Mag. 54, 195–205.
LOWENSTERN, J.B. (1995): Applications of silicate– melt inclusions to the study of magmatic volatiles. In Magmas, Fluids and Ore Deposits (J.F.H. Thompson, ed.), Mineral. Assoc. Can. Short Course 23, 71–98.
HARRIS, C. (1986): A quantitative study of magmatic inclusions in the plutonic ejecta of Ascension Island. J. Petrol. 27, 251–276.
LOWENSTERN, J.B. (2003): Melt inclusions come of age: Volatiles, volcanoes, and Sorby’s legacy. In Developments in Volcanology 5, Melt Inclusions in Volcanic Systems: Methods, Applications and Problems, (B. DeVivo & R.J. Bodnar, eds.), Elsevier, Amsterdam, 1–21.
HARRIS, A.C., KAMENETSKY, V.S., WHITE, N.C., VAN ACHTERBERGH, E. & RYAN, C.G. (2003): Melt inclusions in veins: Linking magmas and porphyry Cu deposits. Science 302, 2109–2111. HAURI, E.H., KENT, A.J.R. & ARNDT, N. (2001): Melt inclusions at the miilenium: toward a deeper understanding of magmatic processes. Chem. Geol. 183, 1–3.
LOWENSTERN, J.B., CLYNNE, M.A. & BULLEN, T.D. (1997): Comagmatic A-type granophyre and rhyolite from the Alid volcanic center, Eritrea, northeast Africa. J. Petrol. 38, 1707–1731. 22
MELT INCLUSIONS IN PLUTONIC ROCKS: PETROGRAPHY AND MICROTHERMOMETRY
LYNTON, S.J., CANDELA, P.A. & PICCOLI, P.M. (1993): An experimental study of the partitioning of copper between pyrrhotite and a high silica rhyolite melt. Econ. Geol. 88, 901–915.
inclusions. Am. Min. 77, 565–576. RAIA, F., WEBSTER, J.D. & DE VIVO, B. (2000): Pre-eruptive volatile contents of Vesuvius magmas: constraints on eruptive history and behavior. I. The medieval and modern interplinian activities. Eur. J. Mineral. 12, 179–193.
MAGAKYAN, R., SOBOLEV, A.V., ZAKARIADZE, G.S. & KONONKOVA, N.N. (1993): Petrology of evolved boninite magmas: evidence from the Mesozoic lesser Caucasus island arc. Petrology 1, 378–394.
RAPIEN, M.H., BODNAR, R.J., SIMMONS. S., SZABÓ. CS., WOOD, C.P. & SUTTON, S.R. (2003): Melt inclusion study of the embryonic porphyry copper system at White Island, New Zealand. Society of Economic Geologists Special Publication 10, 41– 59.
MASSARE, D. & CLOCCHAITTI, R. (1987): Influence de la pression de confinement sur les temperatures d’homogénéisation des inclusions fluids et vitreuses: premières données obtenues à l’aide d’une platine chauffante dams le domaine 0.0–400 MPa; 2008–50°C. C. R. Acad. Sci. Paris 305 1207–1210.
REYF, F.G. (1997): Direct evolution of W-rich brines from crystallizing melt within the Mariktikan granite pluton, west Transbaikalia. Mineral. Dep. 32 475–490.
MASSARE, D., METRICH, N. & CLOCCHAITTI, R. (2002): High-temperature experiments on silicate melt inclusions in olivine at 1 atm: inference on temperatures of homnogenization and H2O concentrations. Chem. Geol. 183, 87–98.
REYF, F.G. (2004): Immiscible phases of magmatic fluid and their relation to Be and Mo mineralization at the Yermakovka F–Be deposit, Transbaikalia, Russia. Chem. Geol. 210, 49–71. REYF, F.G., SELTMANN, R. & ZARZISKY, G.P. (2000): The role of magmatic processes in the formation of banded, Li,F-enriched granites from the Orlovka tantalum deposit, Transbaikalia, Russia: Microthermometric evidence. Can. Mineral. 38, 915–936.
NASH, T.J. (1976): Fluid-inclusion petrology – data from porphyry copper deposits and applications to exploration: U. S. Geological Survey Prof. Paper 907–D, 16 pp. PEPPARD, B.T., STEELE, I.M., DAVIS, A.M., WALLACE, P.J. & ANDERSON, A.T., JR. (2001): Zoned quartz phenocrysts from the rhyolitic Bishop Tuff. Am. Min. 86, 1034–1052.
ROEDDER, E. (1971): Fluid inclusion studies on the porphyry-type ore deposits at Bingham, Utah, Butte, Montana, and Climax, Colorado: Econ. Geol., 66, 98–120.
PERETYAZHKO, I.S., ZAGORSKY, V.Y., SMIRNOV, S.Z. & MIKHAILOV, M.Y. (2004): Conditions of pocket formation in the Oktyabrskaya tourmalinerich gem pegmatite (the Malkhan field, Central Transbaikalia, Russia). Chem. Geol. 210, 91–111.
ROEDDER, E. (1979): Origin and significance of magmatic inclusions: Bull. de Miner. 102, 487– 510. ROEDDER, E. (1984): Fluid Inclusions. Mineralogical Society of America, Reviews in Mineralogy, 12, 646 pp.
PETTKE, T. (2006): In situ laser ablation-ICP-MS chemical analysis of melt inclusions and prospects for constraining subduction zone magmatism. In Melt Inclusions in Plutonic Rocks (J.D. Webster, ed.) Min. Assoc. Can. Short Course 36, 51-80.
ROEDDER, E. (1992): Fluid inclusion evidence for immiscibility in magmatic systems. Geochim. Cosmochim. Acta 56, 5–20.
PROKOFIEV, V., KAMENETSKY, V.S., KOVALENKER, V., BODON, S.B. & JELEN, S. (1999): Evolution of magmatic fluids at the Banska Stiavnica precious and base metal deposit, Slovakia – Evidence from melt and fluid inclusions. Econ. Geol. 97, 949– 955.
ROEDDER, E. & BODNAR, R.J. (1980): Geologic pressure determinations from fluid inclusion studies. Ann. Rev. Earth Planet. Sci. 8, 263–301. ROEDDER, E. & BODNAR, R.J. (1997): Fluid Inclusion Studies of Hydrothermal Ore Deposits. in Geochemistry of Hydrothermal Ore Deposits, 3rd ed., H. L. Barnes, ed., Wiley & Sons, Inc, New York (657–698).
QIN, Z., LU, F. & ANDERSON, A.T., JR (1992): Diffusive reequilibration of melt and fluid
23
R. J. BODNAR & J.J. STUDENT
ROEDDER, E. & COOMBS, D.S. (1967): Immiscibility in granitic melts, indicated by fluid inclusions in ejected granitic blocks from Ascension Island. J. Petrol. G. B. 8, 417–451.
Virginia Polytechnic Institute and State University, Blacksburg, VA. 119 p. http://scholar. lib.vt.edu/theses/available/etd-08192002-150340/ STUDENT, J.J. & BODNAR, R.J. (1996): Melt inclusion microthermometry: Petrologic constraints from the H2O-saturated haplogranite system. Petrology, 4, 291–306.
SCHIANO, P. (2003): Primitive mantle magmas recorded as silicate melt inclusions in igneous minerals. Earth Sci. Reviews 63, 121–144.
STUDENT, J.J. & BODNAR, R.J. (1999): Synthetic Fluid Inclusions XIV: Microthermometric and compositional analysis of coexisting silicate melt and aqueous fluid inclusions trapped in the haplogranite–H2O–NaCl–KCl system at 800°C and 2000 bars. J. Petrol. 40, 1509–1525.
SCHMIDT, C., CHOU, I-M., BODNAR, R.J. & BASSETT, W.A. (1998): Microthermometric analysis of synthetic fluid inclusions in the hydrothermal diamond-anvil cell. Am. Min. 83, 995–1007. SCHMITT, A.K., TRUMBULL, R.B., DULSKI, P & EMMERMANN, R. (2002): Zr–Nb–REE mineralization in peralkaline granites from the Amis Complex, Brandberg (Namibia): Evidence for magmatic pre-enrichment from melt inclusions. Econ. Geol. 97, 399–413.
STUDENT, J.J. & BODNAR, R.J. (2004): Silicate melt inclusions in porphyry copper deposits: Identification and homogenization behavior. Can. Min. 42 1583–1599. TAKENOUCHI, S. & IMAI, H. (1975): Glass and fluid inclusions in acidic igneous rocks from some mining areas in Japan. Econ. Geol. 70, 750–769.
SKIRIUS, C.M., PETERSON, J.W & ANDERSON, A.T. JR. (1990): Homogenizing rhyolitic glass inclusions from the Bishop Tuff. Am. Min. 75, 1381–1398.
THOMAS, J.B., BODNAR, R.J., SHIMIZU, N. & SINHA, A.K. (2002): Determination of zircon/melt trace element partition coefficients from SIMS analysis of melt inclusions in zircon. Geochim. Cosmochim. Acta, 66, 2887–2902.
SOBOLEV, A.V. (1996): Melt inclusions in minerals as a source of principal petrologic information. Petrology 4, 228–239.
THOMAS, R. (1988) Results of investigations on melt inclusions in various magmatic rocks from the northern border of the Bohemian Massif. Proceedings of the 1st international Conference on the Bohemian Massif, Prague, Czechoslovakia, Sept. 26–Oct. 3, 1988, 298–306.
SOBOLEV, V.S. & KOSTYUK, V.P. (1975): Magmatic crystallization based on a study of melt inclusions. Fluid Inclusion Research 9, 182–253 (translated from original publication in Russian). SORBY, H.C. (1858): On the microscopical structure of crystals, indicating origin of minerals and rocks. Quart. J. Geol. Soc. London 14, 453–500.
THOMAS, R. (1994a): Estimation of viscosity and water content of silicate melts from melt inclusion data. Eur. J. Mineral. 6, 511–535.
STERNER, S.M. & BODNAR, R.J. (1984): Synthetic fluid inclusions in natural quartz. I. Compositional types synthesized and applications to experimental geochemistry. Geochim. Cosmochim. Acta 48, 2659–2668.
THOMAS, R. (1994b): Fluid evolution in relation to the emplacement of the Variscan granites in the Erzgebirge region: A review of the melt and fluid inclusion evidence. In Metallogeny of collisional orogens focussed on the Erzgebirge and comparable metallogenetic settings (Seltmann, R., Kampf, H. & Moller, P., eds.) Czech Geol. Survey, pp. 70–81.
STERNER, S.M. & BODNAR, R.J. (1989): Synthetic fluid inclusions. VII. Reequilibration of fluid inclusions in quartz during laboratory-simulated metamorphic uplift. J. Met. Geol. 7, 243–260. STOCKSTILL, K.R., MCSWEEN, H.Y., JR. & BODNAR, R.J. (2005): Melt inclusions in augite of the Nakhla Martian meteorite: Evidence for basaltic parental melt. Meteor. Planet. Sci. 40, 377–395.
THOMAS, R. (1994c): Thermometric study on granites from Bornholm Island, Denmark. Z.Geol. Wiss. 22, 139–145. THOMAS, R., FOERSTER, H.-J. & HEINRICH, W. (2003): The behavior of boron in a peraluminous granite–pegmatite system and associated hydro-
STUDENT, J.J. (2002): Silicate melt inclusions in igneous petrogenesis. Unpub. Ph.D. diss., 24
MELT INCLUSIONS IN PLUTONIC ROCKS: PETROGRAPHY AND MICROTHERMOMETRY
thermal solutions: A melt and fluid inclusion study. Contrib. Mineral. Petrol. 144, 457–472.
VEKSLER, I.V. & LENTZ, D. (2006): Parental magmas of plutonic carbonatites, carbonatesilicate immiscibility and decarbonation reactions: evidence from melt and fluid inclusions. In Melt Inclusions in Plutonic Rocks (J.D. Webster, ed.) Min. Assoc. Can. Short Course 36, 123-149.
THOMAS, R., FOERSTER, H.-J., RICKERS, K. & WEBSTER, J.D. (2005): Formation of extremely F-rich hydrous melt fractions and hydrothermal fluids during differentiation of highly evolved tingranite magmas: a complex melt/fluid inclusion study. Mineral. Dep. 148, 582–601.
WEBSTER, J.D. (2004): The exsolution of magmatic hypersaline chloride liquids. Chem. Geol. 210, 33–48.
THOMAS, R. & KLEMM W. (1997): Microthermometric study of silicate melt inclusions in Variscan granites from SE Germany: Volatile contents and entrapment conditions. J. Petrol. 38, 1753–1765.
WEBSTER, J.D. & DUFFIELD, W.A. (1991): Volatiles and lithophile elements in Taylor Creek rhyolite: constraints from glass inclusion analysis. Am. Min. 76, 1628–1645.
THOMAS, R., RHEDE, D. & TRUMBULL, R.B. (1996): Microthermometry of volatile-rich silicate melt inclusions in granitic rocks. Z. Geol. Wiss. 24, 505–526.
WEBSTER, J.D. & REBBERT, C.R. (2001): The geochemical signature of fluid-saturated magma determined from silicate melt inclusions in Ascension Island granite xenoliths. Geochim. Cosmochim. Acta 65, 123–136.
THOMAS, R. & WEBSTER, J.D. (2000): Strong tin enrichment in a pegmatite-forming melt. Mineral. Dep. 35, 570–582.
WEBSTER, J.D., THOMAS, R., FOERSTER, H.-J., SELTMANN, R. & TAPPEN, C. (2004): Geochemical evolution of halogen-enriched, granite magmas and mineralizing fluids of the Zinnwald tin–tungsten mining district, Erzgebirge, Germany. Mineral. Dep. 39, 452– 472.
THOMAS, R., WEBSTER, J.D. & DAVIDSON, P. (2006): Understanding pegmatite formation: the melt and fluid inclusion approach. In Melt Inclusions in Plutonic Rocks (J.D. Webster, ed.) Min. Assoc. Can. Short Course 36, 189-210. THOMAS, R., WEBSTER, J.D. & HEINRICH, W. (2000): Melt inclusions in pegmatite quartz: Complete miscibility between silicate melts and hydrous fluids at low pressure. Contrib. Mineral. Petrol. 139, 394–401.
WEBSTER, J.D., THOMAS, R., RHEDE, D., FÖRSTER, H.-J. & SELTMANN, R. (1997): Melt inclusions in quartz from an evolved peraluminous pegmatite: Geochemical evidence for strong tin enrichment in fluorine-rich and phosphorus-rich residual liquids. Geochim. Cosmochim. Acta 61, 2589– 2604.
TOMIYA, A & TAKAHASHI, E. (2005): Evolution of the magma chamber beneath the Usu volcano since 1663: a natural laboratory for observing changing phenocryst compositions and textures. J. Petrol. 46, 2395–2426.
YANG, K. & BODNAR, R.J. (1994): Magmatic– hydrothermal evolution in the “bottoms” of porphyry copper systems: Evidence from silicate melt and aqueous fluid inclusions in granitoid intrusions in the Gyeongsang Basin, South Korea. Int. Geol. Rev. 36, 608–628.
VAPNIK, YE. (1998): Melt inclusions in granitoids of the Timma igneous complex, southern Israel. Min. Mag. 62, 29–40.
YANG, K. & SCOTT, S.D. (2002): Magmatic degassing of volatiles and ore metals into a hydrothermal system on the modern sea floor of the eastern Manus back-arc basin, western Pacific. Econ. Geol. 97, 1079–1100.
VARELA, M.E. (1994): Silicate–melt and fluid inclusions in rhyolitic dykes, Los Manantiales mining district, Argentina. Eur. J. Mineral. 6, 837–854. VEKSLER, I.V. (2006): Crystallized melt inclusions in gabbroic rocks. In Melt Inclusions in Plutonic Rocks (J.D. Webster, ed.) Min. Assoc. Can. Short Course 36, 99-122.
25
SIMS IN THE DETERMINATION OF LIGHT STABLE ISOTOPES IN SILICATE MELT INCLUSIONS
CHAPTER 2: APPLICATION OF SECONDARY ION MASS SPECTROMETRY TO THE DETERMINATION OF TRADITIONAL AND NON-TRADITIONAL LIGHT STABLE ISOTOPES IN SILICATE MELT INCLUSIONS Graham D. Layne Dept. of Earth Sciences Memorial University of Newfoundland INCO Innovation Centre, Room 1047 St. John’s, NL, Canada A1B 3X5 E-mail:
[email protected] accelerated into a solid sample at potentials of a few kV. The impact of these primary ions gradually erodes a shallow crater in the sample. All instruments routinely used for geochemical research use finely focused microbeams of primary ions and are thus generally referred to as “ion microprobes”. A portion of the material sputtered from the sample is ionized, and these secondary ions are the analyte species that are introduced into the mass spectrometer of the instrument, hence the name secondary ion mass spectrometry. When practitioners of SIMS refer to the “source” of the instrument, they are generally referring to the source of primary ions, for example, the duoplasmatron (DP) device commonly used to generate O– (or O2– or O2+). However, in concept, SIMS instruments are solid source mass spectrometers, with the sputtering crater comprising the “source” of ionized sample for mass spectrometry – analogous to the filament source in thermal ionization mass spectrometry (TIMS), or the plasma source in inductively coupled plasma mass spectrometry (ICP–MS). The mass spectra of secondary ions produced by sputtering are often complex. In addition to the monatomic singly charged ions most commonly used as analytes for light stable isotope determinations (e.g., 11B+/10B+ for δ11B, 37Cl–/35Cl– for δ37Cl), there are numerous possible isobaric interferences at each integral mass. For example, 10 BH+ interferes with 11B+ at 11 Da, and 34SH– interferes with 35Cl– at 35 Da. The appropriate level of mass resolving power (∆M/M; MRP) may be used to separate these interfering species with the magnetic prism of the instrument according to small differences in their mass. Alternatively, energy filtering (Shimizu et al. 1978) may sometimes be used to suppress interferences. This latter technique exploits differences in the initial energy spectrum of analyte and interfering species as they leave the
INTRODUCTION In the context of analyzing silicate melt inclusions, or similar materials, secondary ion mass spectrometry (SIMS) has the potential to achieve sub-per mil reproducibilities for a wide variety of light stable isotope ratio determinations. These can generally be accomplished with a lateral spatial resolution of better than 10µm where necessary, and with sputtered pit depths of less than a few µm. For glassy materials, this represents a total sample consumption of less than 10 ng for a single analysis. Sample preparation is normally fairly simple. Sample mounts must be compatible with the ultra-high vacuum of the sample chamber (ideally, better than 10–8 torr), and present a flat polished surface of the objects of interest. For analysis of insulating samples (such as silicate melt inclusions) mounts must generally be coated with a thin conductive layer of gold or carbon (300–500 Å). Most SIMS instruments used for light stable isotope determinations limit sample size to a maximum 25.4 mm diameter. Overall sample requirements are, therefore, quite similar to those for electron probe microanalysis (EPMA) of major or trace elements. SIMS shares the advantage with other forms of microanalysis such as EPMA, LA–ICP– MS and PIXE of allowing in situ analysis of low destructivity, preserving the textural context of the objects analyzed within the rock. This is, of course, invaluable when studying melt inclusions, where the extremely small size of an individual object severely limits study of their elemental or isotopic composition by other means. Abbreviations used repeatedly in this chapter are defined in Table 2-1. PRINCIPLES OF SIMS SIMS relies on the physical phenomenon of “sputtering”. A primary beam of ions is
Mineralogical Association of Canada Short Course 36, Montreal, Quebec, p. 27–49.
27
GRAHAM D. LAYNE
TABLE 2-1. ABBREVIATIONS CHAPTER TEXT. CCD cps Da
DP ECL EEF em EPMA eV fc IMF kV LA–ICP– MS MRP nA NEG pA PIXE SIMS TIMS
USED WITHIN
mass spectrometric techniques, the isotope ratio measured by the mass spectrometer is finitely different from the actual ratio in the sample. This effect increases with descending mass of the analyte ions and is of substantial magnitude. For example the measured ratio of 11B+/10B+ in many materials in the IMS 3f instrument is 4.8 % lighter than the actual sample ratio (Chaussidon et al. 1997). IMF is a consequence of multiple effects in the mass spectrometer, most of which are consequent to the relatively small differences in initial energy spectra of secondary ions of the various isotopes of the same element. Since these differences vary with the bulk chemical composition of the sample matrix, careful attention must be paid to characterizing appropriate reference materials for calibrating IMF. However, although IMF effects are of large magnitude in relationship to the (sub-) per mil differences to be measured in isotope ratios of scientific interest, IMF may be effectively held constant at these levels, allowing very reproducible results. It is commonly valuable to determine the concentration of the element itself, in tandem with its light stable isotope ratio, for example S with δ34S. This is often as simple as measuring a single additional “reference” peak related to the inclusion matrix, in this example, 30Si–. The concentration of S is then calculated simply by comparing 32S–/30Si– * wt.% SiO2glass in the sample to that in a reference glass of similar bulk composition. The term for Si concentration normalizes the denominator of the measured ratio. Determinations (e.g., δ34S, δ37Cl) that utilize Cs+ primary beams for the production of negative secondary ions from insulators like melt inclusion glasses also require additional charge compensation of the sample during analysis. Since this combination of primary and secondary ions produces an electron deficit at the sputtering site, an electron flood gun such as the Cameca NEG (see Instrumentation, below) may be effectively used for this purpose. One useful strategy is to use the minimum primary beam density necessary to produce sufficient signals for the analyte ions of interest, minimizing the sample charging effect, and reducing the dependence of IMF on charge compensation. This can be useful even in analysis of positive secondary ions (e.g., δ11B), where charge compensation depends exclusively on the conductive sample coating. Where primary currents of less than 10 nA are sufficient, O2– primary beams can sometimes be used to advantage for sputtering
charge-coupled cevice, as used as a sensor in digital cameras counts per second equivalent to the unified atomic mass unit (u), defined as 1/12 the mass of one atom of 12C. Preferred by many journals to the older term amu (atomic mass unit). duoplasmatron emitter-coupled logic extreme energy filtering electron multiplier detector electron probe microanalysis electron volt Faraday cup detector instrumental mass fractionation kilovolt laser ablation – inductively coupled plasma – mass spectrometry mass resolving power, 10% peak height definition used throughout text nanoampere normal incidence electron gun picoampere proton induced X-Ray excitation secondary ion mass spectrometry thermal ionization mass spectrometry
sample surface. In SIMS instruments that possess an energy-dispersed crossover in their mass spectrometer, the manipulation of an energy slit position, or of the initial accelerating potential of the secondary ions, allows preferential sampling of the monatomic species. However, many of the important interfering species in light stable isotope determinations are, in fact, hydride species like 10 BH+ and 34SH–, which do not have sufficiently different energy spectra from the monatomic analyte species to be suppressed effectively by energy filtering. As a consequence, most techniques for light stable isotope determination discussed herein involve the use of appropriate MRP to separate interfering species. A central concern in the quantitative determination of isotope ratios by SIMS is instrumental mass fractionation (IMF). As in many 28
SIMS IN THE DETERMINATION OF LIGHT STABLE ISOTOPES IN SILICATE MELT INCLUSIONS
positive secondary ions, even though they generally focus to somewhat larger diameters. This is because the singly charged oxygen dimer produces approximately double the sputtering rate for a given charging current. Further, although poorly documented in the published literature, it appears that collecting a broad band of secondary ion energies, through the use of wide energy slits (>50eV) in the mass spectrometer, may also be beneficial in minimizing the effect of sample charging on IMF drift in many applications. As light stable isotope measurements with SIMS become possible with precisions and reproducibilities approaching a few tenths of a per mil, as is now the case with δ18O using multicollection instruments, additional subtle physical effects of sputtering ionization may need to be incorporated in data reduction. For example, the phenomenon of quasi-simultaneous arrivals (QSA) of secondary ions at the conversion dynode of pulse counting ems – an effect related to the ejection of more than one analyte secondary ion in response to a single primary ion – may require a correction procedure at the sub-per mil level (Slodzian et al. 2001, Slodzian et al. 2004). The preparation of melt inclusions as polished mounts creates a thin (10–100 nm) layer of altered or damaged material on the surface of the sample. Exposure to air, cleaning solvents and the process of conductive coating, all add to this contamination burden. The presence of surface contaminants, if not addressed, can introduce strong biases into melt inclusion analyses. However, SIMS is intrinsically capable of eliminating surface contamination by pre-sputtering the site of analysis before data accumulation – essentially depth profiling through these layers. Most instruments also allow the insertion of a field aperture in the mass spectrometer (see Instrumentation, below), or other means to control the effective field of view for collecting secondary ions from the sample. This can be used to eliminate the collection of secondary ions from areas outside the melt inclusion boundaries or beyond the pre-sputtering preparation. It is, however, usually important to collect secondary ions from the entire diameter of the sputtered crater itself during analysis, avoiding irreproducibility in IMF caused by local spatial variations in the relative initial energies of the analyte ions. For this reason the total field of view is generally maintained somewhat larger than the area of the sputtered crater.
A second source of biased results can originate from contaminants that occur dispersed in three dimensions; mineral inclusions, microlites, fluid inclusions and contaminant-laden cracks or fissures. Fortunately, these too may often be detected, and their effect eliminated, by monitoring and discarding short-term excursions in the depth resolved data stream for analyte peak signals. INSTRUMENTATION Cameca IMS 3f/4f/5f/6f/7f Many important advances in the use of SIMS for light stable isotope determinations have been made using instruments from the Cameca IMS 3f/4f/5f series. A useful summary of the detailed features of these and other SIMS instruments is found in Ireland (1995). The Cameca f-series was first manufactured as the IMS 3f, beginning in 1978. A simplified synoptic of the original IMS 3f design is shown in Figure 2-1. Primary ions, usually O– or O2–, are produced in a duoplasmatron ion source (DP), and extracted into the primary column through potentials of up to 12.5 kV. A pair of Einzel-type electrostatic lenses, lens 1 (L1) and lens 3 (L3), is used to focus the primary ions to a small diameter spot on the sample surface. Simple fourplate deflectors are used to center the primary ion beam through these lenses. The L3 deflector plates can also be enabled to deliver a square-rastered primary beam to the sample, or to deflect the primary beam into a Faraday cup for measurement of the primary current. An eight-plate stigmator (L3 stigmator) is also associated with L3, to enable further shaping of the beam. L1 and L3 are used in tandem to deliver a de-magnified image of the exit aperture of the DP (source aperture) to the sample surface (critical illumination). An aperture placed before lens 3 (L3 aperture) is used as a beam limiter to control aberration of the focused spots. Subsequent f-series instruments added a Cs+ ion source, twinned with the DP on the primary column by means of a primary beam magnetic filter (PBMF). The PBMF selects explicitly between 16O– or 16O2–, or 133Cs+, depending on the desired mode of operation. They were also equipped with an additional electrostatic lens (L2) immediately after the PBMF, to improve the current density of the primary beam spot. This third lens, in tandem with an additional aperture (PBMF aperture), allows alternate strategies for primary beam focusing. These include using L2 to deliver maximum beam intensity to illuminate the PBMF aperture. An 29
GRAHAM D. LAYNE
Electrostatic Analyzer Energy Slit Duoplasmatr on
Deflectors Magnet
Field Aperture Contrast Aperture
Spectrometer Lens
Entrance Slit
Deflector 1 Exit Slit
Transfer Lenses L3 Stigmator Secondary Projector Beam L3 Lenses Deflectors
Lens 1 Deflector 2 L3 Aperture
Projection Deflector Stigmator Faraday Electron Cup Multiplier
Transfer Deflector Sample Lens 3 Airlock
Electrostatic Analyzer Immersion Lens
Detection Deflector
Channel Plate/ Phosphor Screen
FIG. 2-1. A simplified synoptic showing the major optical elements of the Cameca IMS 3f ion microprobe.
image of the PBMF Aperture is then focused on the sample by L3 in a manner analogous to that utilized on SHRIMP-type instruments (Kohler illumination, Ireland 1995). Secondary ions sputtered from the sample are extracted through a potential of ±4.5kV, and focused through the immersion lens into the transfer optics of the instruments. The transfer optics comprise three electrostatic lenses, used to control the maximum field of view for secondary ion collection from the sample, and the effective magnification for direct ion imaging. The transfer optics focus the secondary ions on the entrance slit of the mass spectrometer. The contrast aperture may be used to limit the cross section of the ion beam entering the mass spectrometer for high spatial resolution ion imaging, but is usually maintained at its maximum diameter for light stable isotope determinations. Cameca f-series instruments utilize a double focusing (Nier-type) mass spectrometer, which will accommodate either energy filtering or modest mass resolution approaches to the elimination of isobaric interferences. Energy filtering approaches, as discussed above, are
enabled by the energy-dispersed crossover at the energy slit located between the electrostatic analyzer (ESA) and the magnet of the mass spectrometer (Fig. 2-1). Alternatively, mass resolution can be increased from a base value of approximately MRP 300 by narrowing the entrance slit and exit slit. Secondary ions leaving the exit slit are delivered through the first projection lens, a smaller ESA, and the detection deflector to either a Faraday cup (fc) or electron multiplier (em) detector for quantitative ion detection. The isotopic ratios of different analyte masses are counted sequentially by cyclical switching of the magnetic field. The f-series design also allows for ion microscopy, either by direct ion imaging, or by scanning ion imaging using rastered primary beams registered to the pulse-counting em detector. For direct ion imaging, both projection lenses are used in tandem (and the detection ESA de-energized) to project a secondary ion image of the sample plane onto either a channel plate/phosphor screen/CCD camera combination, or a resistive anode encoder (RAE) detector. This imaging capability can be particularly useful for the very exact location and 30
SIMS IN THE DETERMINATION OF LIGHT STABLE ISOTOPES IN SILICATE MELT INCLUSIONS
centering of the analytical spot within small melt inclusions. For example, most melt inclusions within olivine hosts are significantly lower in Mg, and consequently register as a recognizable dark area in CCD imaging of 24Mg+. Cameca f-series instruments may also be equipped with the self-balancing normal incidence electron gun (NEG) designed by Slodzian for charge compensation when analyzing insulators (Migeon et al. 1990). This device uses a Wehnelt electron gun, and additional optics, to supply a cloud of electrons with an opposite balanced potential to that used to extract secondary ions from the sample. Any charging of the sample during analysis effectively changes this potential balance, drawing down a small compensating current of electrons to the sample surface. Direct CCD imaging for optimization of a bright spot signal (e.g., 16O–) from the sample is valuable for the routine alignment of compensating electron flux from the NEG for successive analytical spots. The IMS 6f and 7f are more recent models of the Cameca f-series. In addition to substantially increased automation and computerization of the instrument controls and data acquisition, they include several updates to the ion optics that benefit light stable isotope determination. Most significantly, they have an improved NEG design, with an Einzel lens for more uniform and stable focusing of the charge compensating electrons over a 150 µm diameter on the sample. The extraction potential for secondary ions has been increased from ±4.5 kV to ±10 kV, with incremental benefits for ion collection efficiency and em response. A laminated mass spectrometer magnet slightly reduces the waiting time for peak switching for analyses not involving the fc detector. The most recent IMS 7f instrument has an additional lens for the duoplasmatron source that substantially increases O– (or O2–) primary beam densities, advantageous for the analysis of positive secondary ions (e.g., Li+, B+). The IMS 7f also includes electrostatic deflection for rapid switching between fc and em detection for high dynamic range measurements. The automated controls for the ion optics allow easy alignment and programming of higher current coincident primary beams that can be used for fast and efficient pre-sputtering for the removal of surface contamination. Versions of the DP lensing system, improved NEG, and electrostatic detector switching are all available for retrofitting to the older f-series instruments.
In laboratories attempting sub-per mil reproducibility for light stable isotope analyses, it has been common to retrofit more sophisticated detector preamplifier and counting systems to all but the very newest instruments. Many different, customized approaches have been taken to these improvements but the objectives are in common: i) for the fc detector, to reduce and stabilize background noise (current) while maintaining a reasonably fast response rise time for peak switching analyses, ii) for the em detector, to maintain a stable pulse height distribution from the detector and stable overall dead time for the pulse counting system. Cameca IMS 1270/1280 The Cameca IMS 1270, first manufactured in 1993, is a large format high resolution, high transmission ion microprobe originally conceived and designed for U–Pb geochronology. It will accommodate a 5 device multicollector array with sufficient nominal mass dispersion to collect simultaneously 206Pb+, 207Pb+, 208Pb+, 232Th+ and 238 + U (Fig. 2-2). The primary ion column, sample chamber, immersion and transfer optics, and NEG are virtually identical to the IMS 6f. It is the much larger and more powerful mass spectrometer of the IMS 1270 that distinguishes these architectures. The mass spectrometer performance necessary for U–Pb geochronology has also translated into a highly versatile instrument with many features valuable for light stable isotope analysis. The multicollection array of the IMS 1270 is valuable in many light stable isotope determinations. Each of the five detector chariots can be used to deploy either an fc or miniaturized em detector. The ability to collect simultaneously two or more analyte peaks has obvious and readily calculable benefits for precision/time since the duty cycle for data collection on every peak can be virtually 100% (excluding counting on mass-free positions to establish background signals). It must be borne in mind, however, that considerable time must often be devoted to detector gain matching and calibration, sometimes severely diluting the precision/time efficiency gained over the monocollection peak switching approach. The fundamental strength of multicollection for many natural materials is in eliminating the cycle to cycle bias induced by primary ion current and sample-induced fluctuations in signal during peak switching analyses. This is of enormous value in increasing analytical precision 31
GRAHAM D. LAYNE
FIG. 2-2. Detailed synoptic of the Cameca IMS 1270/1280 multicollection array (adapted from Cameca, SA product catalog literature).
in samples where the analyte element is heterogeneously distributed within a glassy matrix. However, it is not always viable to use SIMS on very phase-heterogeneous inclusions. For some isotope systems, IMF is too highly dependent on bulk composition of the individual phases, introducing substantial errors. The IMS 1270 has a maximum mass dispersion (∆M/MMEAN) across the multicollection focal plane of 15.6%. This will nominally support simultaneous detection for any of the ratio measurements discussed herein. In some cases, combined analyses of more than one ratio (e.g., δ7Li and δ11B) might be accommodated using a single magnetic peak switch, and appropriate detector positions. However, several significant issues require attention for quantitative analysis with multicollection: i) periodic matching of detector gains is essential to maintain accuracy and reproducibility. Electron multipliers, in particular, may age quite rapidly, especially when incident ion currents exceed 105–106 cps. These effects are substantial, even on an intraday basis, when attempting to maintain sub per mil reproducibility while using miniaturized ems.
ii) the use of multicollection necessitates disengaging the first projection lens in the IMS 1270, disabling the capability for direct ion imaging of the sample. This reduces the capability for accurate spot centering, and eliminates the ability to objectively monitor NEG alignment on a spot to spot basis. The former capability will be partially restored, though in a less rapid and well resolved form, in future instruments, which will have scanning ion imaging capabilities tied to one or more multicollector ems. In addition to multicollection capability, the IMS 1270 offers fundamental advantages for some analyses in terms of superior transmissions at high MRP. The IMS 1270 maintains almost 100% transmission through the mass spectrometer for MRP as high as 3000 (10% definition), while the IMS 6f maintains <20% transmission at this MRP (Fig. 2-3). Even at the relatively high MRP of 5250 used for δ37Cl determinations, the IMS 1270 retains transmission >50%. This superior transmission offers no substantial advantage over the f-series instruments for major element analytes such as O (for δ18O) in silicates, where MRP of 2300 is sufficient to resolve the significant isobaric
32
GRAHAM D. LAYNE
Relative Transmission
10.00
1.00
IMS 6f
0.10
IMS 1270/1280 NanoSIMS 50
0.01 100
1000
10000
100000
Mass Resolution
FIG. 2-3. Relative transmission versus mass resolution curves for the Cameca IMS 1270 and IMS 6f instruments (adapted from Cameca, SA product catalog literature).
interferences, and signals are substantial even on the minor isotope (18O). However, for trace-element analytes such as B, it is a tremendous advantage for maintaining precise analyses even for the B concentrations of 10 ppm, or less, which are common in igneous melt inclusions. One other convenient feature of the IMS 1270 optics is a continuously variable square field aperture. This allows perfect matching of the desired field of view for secondary ion collection from the sample for a given spot size or raster presputtered area. The newest, IMS 1280 instruments were first delivered in 2005. They feature the same ion optics as the IMS 1270, supported with modernized “e7”-type electronics. Most of their “improved features” are a consequence of more sophisticated software for the automated control and centering of ion optics, and are gradually being made available for existing IMS 1270 installations through software upgrades. Foremost among these are automated precentering of secondary ions in the field aperture using deflectors in the transfer optics. Optionally, the entrance slit of the mass spectrometer may be mechanically re-centered for each analysis as well. The object is to maintain a uniform flight path (and energy cross section) of secondary ions through the transfer optics to the mass spectrometer from spot to spot on the sample. Field aperture re-centering, in particular, appears useful in refining reproducibility of light stable isotope analyses at the sub-per mil level. Other newly automated features include monitoring of pulse height distributions for em
detectors, conceived to enable correction of small changes in em gain during successive spot analyses. The simple permanent magnets traditionally used to correct flight path differences for the lightest ions (due to the Earth’s vertical magnetic field in high latitude installations) have been supplanted in some versions of the IMS 1280 by a series of Helmholtz coils mounted along the flight tubes of the mass spectrometer. These are supposed to provide tunable compensatory fields to address concerns about low order effects of stray magnetic fields on IMF at the sub-per mil level. SPECIFIC TECHNIQUES AND APPLICATION EXAMPLES δ18O: Stable isotope ratio analysis of a major element Quantitative determination of δ18O by SIMS has been in use for some time (e.g., McKeegan 1987, Hervig et al. 1992, Riciputi & Paterson 1994, Valley et al. 1998). The major challenge in achieving routine sub-per mil reproducibility has been to stabilize IMF, particularly for insulators (like most melt inclusions) and to calibrate for it accurately across a wide range of major element compositions. Also, the large dynamic range of this ratio (16O/18O ≈ 487) requires careful attention to detector characteristics and stability. Eiler et al. (1997), closely following Riciputi & Paterson (1994), summarized an approach that is typical of one that has been generally successful in obtaining reproducibilities better than ±1‰ (1σ) for δ18O in many laboratories 33
GRAHAM D. LAYNE
equipped with Cameca f-series instruments. They used an IMS4f, with a primary ion beam of 8 nA of Cs+, accelerated through a nominal potential of 10 kV and focused to a 20–30 µm spot on the sample. Extreme energy filtering (EEF, Schauer & Williams 1990, Hervig et al. 1992) was used to eliminate 17 OH– interference effectively from the 18O– peak, and to reduce sensitivity to variations in sample charge compensation with the NEG. The use of EEF is based on the observation that IMF becomes insensitive to variations in charge compensation, even of several tens of volts, for secondary ions with initial energies in excess of 350 eV as they leave the sample surface. In this case, EEF was performed using a sample offset potential of 350V (±25eV energy slit). Despite the high degree of signal loss to the energy filter, the substantial primary beam current used allows typical count rates on silicate matrices to be maintained at 106 cps on 16O– (≈2*104 cps on 18O–) or more, with detection of both peaks by a single regular format multiple-dynode em. Each spot analysis involved 120 cycles of counting by magnetic peak switching on 16O– (1s) and 18O– (5s) (or 200 cycles of 16O– (1s) and 18O– (10s)) with a 50 ms settling time between peak switches. Typical analyses therefore take between 726 and 2210 s (12–37 minutes), not including any pre-sputtering. Pre-sputtering times of a few minutes remain invaluable in removing surface contamination, and allowing steady state charging of the sample surface to be attained. Eiler et al. (1997) reported that actual inrun precisions (standard error of the mean for the 120 cycles) of ±0.92 ‰ (±0.60‰ for 200 cycle analyses) closely conform to the theoretical precisions based on Poisson counting statistics, which were typically ±0.90‰ (±0.50‰ for 200 cycle analyses). This confirms that instrumental instabilities and drifts during individual analyses were an order of magnitude lower than the counting precision limit. Riciputi & Paterson (1994), using a similarly equipped IMS 4f, had earlier demonstrated that the above method using 200 cycles of data collection would yield external reproducibilities routinely better than ±1‰ (1σ), typically ±0.7‰ (1σ), for quartz and carbonate materials. They also showed that the reproducibility for IMF (18O–/16O–) on their Brazilian quartz standard (–7.26 %) was consistent to ±2.1‰ (1σ) over a 15-day period. With periodic intraday calibration for IMF, this procedure can therefore yield overall reproducibilities that approach ±0.5 ‰ (1σ) for chemically
simple matrices. These, and other studies of δ18O and other light stable isotope ratios, have benefited from improved em preamplifier and counting systems, such as those manufactured by Pulse Counting Technologies, coupled with improved electron multipliers such as the ETP 133H. These ECLbased systems are indeed faster, maintaining overall system dead time as low as 9 ns. This attribute is beneficial in allowing quantification of count rates in excess of 106 cps, as is the improved linearity of the low resistance ETP 133H. However, the core benefit of such systems for high dynamic range measurements, such as δ18O, actually stems from improved stability of the pulse height distribution for individual ion strikes on the em, as processed by the em preamplifier and its contribution to improved reproducibility of the 18O/16O measurement. These systems can effectively limit the intraday effects of aging of the large format em to a few tenths of a per mil, even at count rates of 2*106 on 16O–. Gurenko et al. (2001) presented one of the first published studies of δ18O in phenocryst-bound melt inclusions, using the IMS 1270 at CRPG– Nancy. They used 1–10 nA primary Cs+ beams focused to 5–30 µm spot diameters. The NEG was employed for charge compensation, in an analogous manner to that used in earlier studies with f-series instruments. No energy filtering was used. The energy slit of the instrument was instead set to an extremely wide bandpass (120 eV; with energy slit center offset –25eV from the 0eV axis), and MRP 5000 was used to eliminate isobaric interferences on O analyte species. Monocollection mode measurements of δ18O used a combination of electron multiplier (18O–) and Keithley 642-supported Faraday cup (16O–) with gain intercalibration of these paired detectors monitored using 18O–/16O– ratios measured on standard glasses. Gain calibration was observed to drift by 1‰ per day, or more, an effect attributed to gradual aging of the electron multiplier. These measurements yielded precision and reproducibility similar those produced using f-series instruments with EEF, typically ±0.4‰ (1σ) and ±0.7‰ (1σ), respectively. Data for 34S–/32S– were collected during the same analysis as for 18O–/16O– (see discussion below, in δ34S, δ37Cl section). Additional δ18O determinations were also accomplished using paired Faraday cup multicollection, with ion intensities on 18O– maintained at 3–5 million cps. This enabled internal precision of better than ±0.5‰ (1σ) for only 3 34
SIMS IN THE DETERMINATION OF LIGHT STABLE ISOTOPES IN SILICATE MELT INCLUSIONS
of charge compensation by NEG for each spot, overall reproducibility for δ18O in flat, well-coated samples can be reduced to better than ±0.5‰ (1σ). A wide energy slit (>50eV) seems to provide a benefit in improving the reproducibility of IMF. This recent experience with IMS 1270/1280 instruments operated in monocollection mode suggests that the tradition of using EEF for δ18O on f-series instruments might bear careful reexamination. As discussed above, the primary column and transfer optics of the IMS 1270/1280 are virtually identical to those of the IMS 6f/7f. The major architectural changes from the IMS 4f instruments originally used in the studies of Riciputi & Paterson (1994) and Eiler et al. (1997) are i) the increase in secondary ion extraction potential from 4.5 kV to 10 kV, and ii) the improvement of the NEG to include an Einzel lens for electron focusing. It appears possible, therefore, that any f-series instrument equipped with the newer style of NEG, particularly 6f/7f instruments, should be capable of highly reproducible δ18O analysis using MRP to eliminate 17OH– interference. This would allow substantially lower primary beam currents, which would act in concert with the improved NEG charge compensation to increase the reproducibility of IMF, along with higher lateral spatial resolution (10–15 µm versus 20–30 µm). With multicollector IMS1270/1280 instruments, several detector pairings are possible for δ18O. However, the most commonly used are either fc for 16O– and em for 18O– , or fc for both peaks, utilizing an increased primary beam current. This latter arrangement allows precision of better than ±0.2‰ (1σ) to be developed for spot analyses as short as 5 minutes, while maintaining lateral spatial resolution of 10–15 µm. At this level of precision, overall reproducibility then becomes dominated by variations in IMF caused by sample behavior, small drifts or variations in the gain matching between the detector pairs, and matrix effects on IMF for chemically variable materials (such as natural glasses in melt inclusions). The first two effects are intrinsically combined within measurements of IMF on a standard material, which must be performed periodically to maintain intraday reproducibility. Given that several means exist to make SIMS determinations of δ18O at the sub-per mil level of precision and reproducibility, the lingering issue for the analysis of melt inclusions is that of accurate calibration for IMF for these chemically
minutes of data accumulation. Cross-calibration of Faraday cup gains showed intraday drifts limited to less than 0.1‰. Typical precision and reproducibility for δ18O were improved in this manner to ±0.2‰ and ±0.3‰ (1σ), respectively. In all cases, matrix-dependent changes in IMF of 18O–/16O– were calibrated using a set of eight natural glasses ranging in composition from 45.3 to 72.3 wt.% SiO2. Negative correlations of IMF with SiO2 and (Na+K)/Al atomic ratios, as well as positive correlations with FeO and CaO were also observed in this study. No statistical correlation of IMF with either Al2O3 or H2O (0.16 to 1.55 wt.%) was noted. In practice, a simple linear regression of observed IMF versus SiO2 for these materials was used to correct for IMF in the melt inclusion measurements. Overall uncertainty in δ18O, including that derived from the IMF correction scheme, was estimated at ±1.7 ‰ for monocollection and ±0.7 ‰ (1σ) for multicollection measurements. Approaches to δ18O determination with IMS 1270/1280 instruments have generally used mass resolution (MRP >2275) to resolve 17OH– from 18O–. This is sensible, since these instruments still retain virtually 100% transmission at this MRP. The consequently higher transmission achieved over EEF techniques allows the option of substantially reducing primary beam current, current density, and spot size., for example, 200 pA of Cs+ rastered over 15µm2 versus the 8 nA in 25 µm spots commonly used in the f-series analyses described above. Pre-sputtering of surface contamination becomes critical at these lower currents, given the drastically reduced rate of material removal. A 200pA primary Cs+ beam will produce the equivalent of >5*107 cps on 16O– (≈105 cps on 18 O) using the above conditions. Using a combination of fc (16O– (1s)) and em (18O– (4s)) detectors in monocollection mode with magnetic peak switching (and electrostatic detector switching) the IMS 1270 can develop a typical inrun precision of better than ±0.5‰ (1σ) for a fast 30 cycle analysis (3.5 minutes, including (1s) wait times for magnetic switching, and 1s background measurement cycle for fc noise); with external reproducibilities (σ/√n) of better than ±2‰ (1σ) for 10 spots and ±1.5‰ (1σ) per mil for 20 spots, using fully automated analysis of points within 15 mm of the center of a homogeneous sample such as SiO2. With longer counting times (12–15 minutes), and some manual attention to the centering and stability 35
GRAHAM D. LAYNE
variable materials. Hervig et al. (1992), Riciputi & Paterson (1994), Eiler et al. (1997) and Gurenko et al. (2001) have presented various semi-empirical models for the variation of IMF of 18O/16O in a variety of natural and synthetic minerals and glasses. None of these models, however, currently provide a basis to extrapolate IMF calibration adequately at a sub-per mil level across a wide range of melt inclusion chemistries. Limited data for natural glass compositions in Eiler et al. (1997) illustrate that IMF variations between basalt and rhyolite can exceed 4‰, and may vary by 1‰ even over restricted ranges of basalt compositions. If working within limited compositional ranges (e.g., high SiO2 rhyolite), it may suffice to calibrate against a single wellcharacterized standard of similar bulk composition. However, for studies involving systems with more variable melt inclusion chemistry, it will be necessary to develop and utilize a series of IMF standards that cover a wider compositional range.
plus supporting trace element evidence (determined separately by SIMS), they concluded that the inclusions represent 18O-enriched phonolite melts produced by low degree partial melting of a highly depleted peridotite that was metasomatized by slabderived fluids. On the basis of this interpretation, these inclusions provide the first direct demonstration of the existence of mantle fluids or melts with strongly elevated δ18O values. δ34S, δ37Cl: Stable isotope ratio analysis of minor elements with excellent sputtering ionization yields Unlike O, S and Cl exist only at minor to trace element levels in most glassy melt inclusions. However δ34S and δ37Cl are both readily amenable to SIMS determination, due to the excellent yields of S– and Cl– secondary ions under sputtering by Cs+ primary beams. Natural glasses routinely contain significant S, producing the isobaric interference of 34 SH– on 35Cl– in the secondary ion spectrum. Attempts to utilize SIMS for δ37Cl were initially limited, using f-series instruments, because hydride interferences like SH– are difficult to eliminate effectively using energy filtering. However, Layne et al. (2004) have recently published a detailed technique for determining δ37Cl in natural glasses using high MRP with the IMS 1270. Figure 2-4 displays the mass spectrum at 35 Da using an IMS 1270 at MRP 5250 (10% definition) for a sample of synthetic basalt glass with an extreme enrichment in Cl (2.79 wt.% Cl).
Applications: Using procedures similar to those of Riciputi & Paterson (1994) and Eiler et al. (1997) for the IMS 4f, Eiler et al. (1998) reported δ18O data for two vein-bounded silicate glass inclusions of phonolite composition, hosted by a sample of mantle olivine from the island arc setting of Simberi Island, New Guinea. These two inclusions showed δ18O of +11.3 ±1.3‰ (1σ, n=4) and +8.8 – 9.2‰ (n=2), respectively – values much heavier than the δ18O of +5.7±0.5‰ inferred for most mantlederived magmas. Based on these measurements,
FIG. 2-4. Secondary ion spectrum (em detection) at nominal mass 35 Da for a synthetic basalt glass (“Basalt glass B” of Godon et al. 2004, 2.79 wt.% Cl) to illustrate isobaric interferences at MRP 5250 (10% definition) (after Layne et al. 2004).
36
GRAHAM D. LAYNE
systems where individual samples span a broad range of melt compositions, a series of standards is required. Godon et al. (2004) have demonstrated that IMF is a highly correlated function of SiO2, Al2O3, CaO and FeO, permitting correction of the measured δ37Cl value to ±0.5 ‰ over this wide compositional range, using a simple linear fit based on a set of synthetic standard materials (Fig. 2-5). It is also notable that the IMF of 37Cl–/35Cl– for many SiO2 rich glasses is positive, as also noted for 7 + 6 + Li / Li (see δ7Li, δ11B, below). Analyses accumulated in 12 minutes routinely yield internal precision better than ±0.45‰ (1σ). Individual analyses of homogeneous materials have internal precision values (0.30– 0.45‰, 1σ) that approach the theoretical limits calculated from Poisson counting statistics (0.25– 0.30‰, 1σ). Overall reproducibility of individual spots is better than ±0.7‰ (1σ), even for sessions spanning several days, and can be maintained for 12 minute analyses of samples with as little as 250 ppm Cl. The multicollector of the IMS 1270/ IMS1280 will permit simultaneous collection of 35 – Cl and 37Cl– signals, with obvious advantages for reducing total analysis time and/or improving precision. However, any expected increase in overall counting time efficiency will be partially offset by the necessity of periodic intercalibration of the multiple em detectors. Further, the acquisition
The closest isobaric interference apparent is indeed that of 34SH– on 35Cl–. The average S/Cl ratios documented for many oceanic basalt examples are in fact 103–104 times larger than those for the Clenriched basalt of Figure 2-4, and so would produce a 34SH– peak of 10–100% the height of the 35Cl– peak. Total Cl contents of these same basalts are also commonly <500 ppm. MRP 5250 provides the best balance of Cl– transmission and mass resolution, effectively reducing SH– interferences to <<0.1 ‰ level, even for high S/Cl samples. Simultaneous δ37Cl and Cl concentration analyses are performed by bombarding the sample with a primary ion beam of 150–300 pA of Cs+ focused into a 10–20 µm diameter spot. The NEG is used for charge compensation. Any exotic Cl contaminating the immediate surface of the sample is removed by 1–2 minutes of pre-sputtering with a rastered beam before data accumulation begins. Any surface Cl remaining on the periphery of this pre-sputtered area is excluded by restricting the field of view of the mass spectrometer to the immediate area of sputtering during the subsequent spot analysis. This is accomplished using the continuously variable square field aperture of the IMS 1270, in concert with the second (center) transfer lens, to produce an effective field of view of the sample surface of 25 x 25 µm. The extremely high ionization efficiency of Cl under these conditions, coupled with the high transmission of the IMS 1270, provides very high count rates, typically >650 cps/ppm Cl/nACs+ in high SiO2 glasses. Peaks are counted by cyclical magnetic peak switching with ion counting using a regular format em (ETP 133H). The counting times and peak sequence used are: background position (29.67 Da, 1.0 s), 30Si– (2.0 s), 35Cl– (2.0 s) and 37Cl– (4.0 s). Waiting times of 0.5 s are inserted before each peak counting position, and 1.0 s before background, to allow for magnet settling. A typical analysis consists of accumulating 60 of these peak cycles, which takes less than 12 minutes. Simultaneous determination of Cl concentration is easily accomplished by comparing 35Cl–/30Si– (normalized for SiO2 content) for unknown and standard glasses. IMF of 37Cl–/35Cl– is highly matrixdependent for natural glasses ranging from rhyolite to basalt (Godon et al. 2004). IMF for high-SiO2 rhyolite is on the order of +6 to +10‰ depending on instrumental conditions. That for basalt is as much as 8‰ lower. Consequently, in studies of
FIG. 2-5. Plot of the instrumental mass fractionation (IMF, in ‰) as a function of major element composition (0.1319 SiO2 – 0.3829 Al2O3 – 0.6174 CaO + 0.2888 FeO) of seven silicate glass standards ranging from basalt to rhyolite. Two additional synthetic glass standard candidates (Cl-Phono-I, ClPhono-II, Godon et al. 2004) proved inhomogeneous for δ37Cl and are not shown in this figure (after Godon et al. 2004).
37
GRAHAM D. LAYNE
of 30Si– will necessitate at least one magnetic peak switch. Nonetheless, large format ion microprobes with multicollection arrays have the potential to improve the internal precision further and, consequently, the reproducibility of SIMS determinations of δ37Cl. With these enhanced detection systems overall reproducibilities should improve to well below ±0.5 ‰ (1σ) for natural glasses and melt inclusions. One of the only published descriptions of δ34S determination in melt inclusions is that contained within Gurenko et al. (2001), using an IMS 1270 in monocollection mode. Data for 34S– /32S– were collected in the same analysis as those for 18O–/16O– (see details of instrument settings above, in the δ18O section). Primary beam current was adjusted to maintain intensity on the minor isotopes (18O–, 34S–) at 4*105 – 7*105 cps. A typical analysis involved 50 cycles of data acquisition (2s 16 – O ; 6s 18O–; 2s 32S–; 6s 34S–), for a total accumulation time approaching 15 minutes. Precision and reproducibility for δ34S were typically ±0.8‰ and ±1.5‰ (1σ), respectively. IMF for 34S–/32S– appears far less sensitive to matrix composition in silicate glasses than for 18 – 16 – O / O . Consequently, the magnitude of IMF in this study was calibrated using a single natural MORB glass of known δ34S. Although the overall uncertainty cited for δ34S in this study (typically ±1.5‰ (1σ)) was probably rendered higher than may be optimal by being determined in monocollection mode in tandem with δ18O, they nonetheless achieved useful determinations of δ34S in melt inclusions containing 450–2500 ppm S. They also demonstrated the potential utility of SIMS determination of δ34S in tandem with ΣS and S speciation as determined by EPMA. Hauri et al. (2002) made brief mention of SIMS for δ34S determination in basaltic glasses. They followed a procedure essentially similar to that which they used with the IMS 6f for δ13C (see δD, δ13C, below), including an MRP of 3200– sufficient to resolve 16O2– from 32S– and 33SH– from 34 – S . The IMF for 34S–/32S–, as determined on a single low H2O (<1 wt.%) basalt glass reference material (ALV 519-4-1) was –10.0 ±3.0‰. For basaltic glasses containing >700 ppm S, a count rate >106 cps was easily obtained on 32S–, yielding in-run precisions for 30 min analyses of ±0.3‰ (1σ). However, overall reproducibility was far inferior, typically no better than ±1–2‰ (1σ). This is attributable to their procedure of collecting secondary ions only from the center of a 40 µm
diameter sputter crater, producing the propensity for variable IMF from spot to spot (see δD, δ13C, below). Reproducibility for δ34S in glasses with the IMS 6f would almost certainly be greatly improved by collecting secondary ions from the entire sputter crater, just as for the δ37Cl procedures of Layne et al. (2004) summarized above. In fact, it is completely practical to obtain δ34S and δ37Cl in the same analysis with the IMS 1270/1280, simply using additional magnetic peak switching to 32S– and 34S–, providing the potential for simultaneous high quality determinations of these two complementary light stable isotope ratios in melt inclusions. Applications Gurenko et al. (2001) examined the S and O isotopic composition of clinopyroxenehosted melt inclusions from Miocene basaltic hyaloclastite recovered southwest of Gran Canaria (ODP Leg 157) in an effort to evaluate the origin of the unusually S-rich basaltic magmas described from this intraplate volcanic environment by Gurenko & Schmincke (2000). The S and O isotopic compositions of melt inclusions in this basalt suite are both strongly variable (–1.0 to +8.5‰ for δ34S, +5.0 to +7.8‰ for δ18O). δ34S values of the inclusions correlate positively with total concentrations of Cl, and with the proportion of S dissolved in the melt as sulfate (S6+/Stot). The authors argued that processes involving direct contamination by seawater, hydrothermal vent fluids, or assimilation of seawater-derived sulfate do not properly explain the observed stable isotope variations. Instead, this variability is best explained by varying degrees of crustal contamination having occurred in the underlying magma ponding system. They concluded that contamination of the magma by 10% or less of hydrothermally altered basaltic crust (containing up to 5 wt.% anhydrite), plus a contribution from marine sediments, is the most plausible explanation for the observed variations in δ34S and δ18O. Further, they argued that this style of crustal assimilation is likely to have developed beneath other oceanic intraplate volcanoes. δ7Li, δ11B: Stable isotope ratio analysis of traceelement analytes The primary challenge of δ7Li and δ11B determinations in silicate melt inclusions is that one is working with trace element levels of the Li and B analytes (routinely less than 100 ppm, and often less 38
SIMS IN THE DETERMINATION OF LIGHT STABLE ISOTOPES IN SILICATE MELT INCLUSIONS
106
than 10 ppm). A key requirement, therefore, is to maintain high enough signals on the analyte masses to achieve sufficient precision. Most published δ11B determinations by SIMS have utilized the general approach published by Chaussidon et al. (1997) for their study of meteorites and mantle rocks. A primary beam of O– accelerated through 10–12.5 kV is used to produce B+ secondary ions. Charge compensation is effectively maintained by a 300–500A gold coating. No energy filtering is used (the energy slit is opened to >120eV), and a resolution of 1800 MRP is applied to separate the interference of 10BH+ on 11B+ (Fig. 2-6). This level of MRP is also sufficient to eliminate any other possible interferences, such as 9 BeH+ on 10B+. Ion counting is by an em detector, using cyclical magnetic peak switching between 10 + B and 11B+. However, the Chaussidon et al. (1997) study was adapted in many ways to the ultra-low concentrations in their samples (100 ppb–1 ppm), as approached using an IMS 3f instrument. This included using primary beam currents as high as 100 nA to produce sufficient B+ signal, counting times of up to 90 minutes, and completely separate measurement of B concentration via 11B+/30Si+ using a classical energy filtering technique (±10 eV energy slit, –60 V sample offset potential). A useful revelation of the Chaussidon et al. (1997) study was the relative independence of B+ ionization efficiency from matrix effects over a
MRP 1800 11B+
105
I (cps)
104 10BH+
103 102 101 100 10-1
10.95
10.975
LC 19 (4.5 wt%)
log (Bppm /SiO2wt% )
1
JV 1 (1920 ppm)
RM 6 (252 ppm)
0 DK 89 (14 ppm)
-1
UTR 2 (17.7 ppm)
BCR 1 (3.1 ppm) NBS 614 (1.3 ppm)
-2
NBS 617 (0.2 ppm)
-3
-5
-4
-3
-2 -1 log (11B+/ 30Si+)
11.05
wide range of natural and synthetic glass compositions including natural basalt and pantellerite, and synthetic compositions that approximated high-silica rhyolite. In practice this allows a single universal working line for determinations of B concentration (Fig. 2-7). They also demonstrated that the IMF for 11B+/10B+ is extremely consistent over a very wide range of matrix compositions (silicate glasses, marine salts and even boric acid),
LQ 24 (1.0 wt%) GB4 (970 ppm)
11.025
FIG. 2-6. Secondary ion spectrum at nominal mass 11 Da for the GB4 glass reference material showing separation of the 10BH+ on 11B+ peaks at MRP 1800 (10% definition) (after Chaussidon et al. 1997). Originally published in Geostandards Newsletter: The Journal of Geostandards and Geoanalysis (1997, Vol. 21 No. 1, pp. 7-17), reproduced by permission of the copyright holder - Association Scientifique pour la Geologie et ses Applications (ASGA).
3 2
11
Da
0
39
1
FIG. 2-7. Calibration line for B concentration measurements – B/SiO2 versus 11B+/30Si+ (data from Chaussidon & Libourel 1993). The linear fit obtained for standards of diverse composition shows that the background for B is low (<0.05 µg.g–1) and that matrix effects on ionization yield of B+ are negligible. As presented here on logarithmic axes, the estimated precision of these analyses is smaller than the plotted symbols. (after Chaussidon et al. 1997). Originally published in Geostandards Newsletter: The Journal of Geostandards and Geoanalysis (1997, Vol. 21 No. 1, pp. 7-17), reproduced by permission of the copyright holder - Association Scientifique pour la Geologie et ses Applications (ASGA).
GRAHAM D. LAYNE
4.05 solutions evaporated on Si wafer 4.00
synthetic standard glasses Seawater
11B+/ 10B+
SIMS
3.95 3.90
NBS 951
3.85 3.80
11B/10 B SIMS
3.75
11B/10B accepted
3.70 3.90
3.95
4.00
= 0.9538+0.0015
4.05 4.10 4.15 11B+/ 10B+ accepted
with a value of –4.84 ±0.16% for their analytical conditions (Fig. 2-8). Very long term reproducibility of δ11B for the main reference material used to calibrate for IMF (GB4 synthetic glass) is cited as ±1.3‰ (1σ, n=173 over five years). Reproducibility for week long sessions was generally better than ±1.0‰ (1σ). Schmitt et al. (2002) used both Cameca IMS 6f (GFZ-Potsdam) and modified IMS 3f (ASUTempe) instruments to determine δ11B and B in silicic glasses and melt inclusions. Their conditions for δ11B analysis were similar to those used by Chaussidon et al. (1997), although the energy slit was narrowed to 50eV. A primary O– beam of 20 nA (25 µm diameter) was used, with 10 minutes of pre-sputtering for each spot, as a precaution against the inclusion of surface B contamination in the analysis. Precision of individual 15–50 minute δ11B analyses of glasses containing 40–70 ppm B was typically better than ±1‰ (1σ). IMF was corrected by reference to NIST 610 reference glass (δ11B, –1.2 ±0.7‰, by TIMS). Replicate analyses of NIST 610, during sessions as long as 36 hours, had a typical external reproducibility of ±1.7‰ (1σ). Schmitt et al. (2002) therefore estimated the overall reproducibility of their individual sample measurements as better than ±2‰ (1σ). As in Chaussidon et al. (1997), boron concentrations (as 11B+/30Si+) were measured in separate runs, using 1 nA primary O– beams, and energy filtering (–75eV offset). Concentrations were deduced by comparison with NIST 610 (356.4
4.20
4.25
FIG. 2-8. Calibration of IMF for 11 + 10 + B / B during one session of analyses. Standards of extremely variable composition and structure (glasses, salts) are co-linear, demonstrating that matrix effects on IMF are negligible at the per mil level (after Chaussidon et al. 1997). Originally published in Geostandards Newsletter: The Journal of Geostandards and Geoanalysis (1997, Vol. 21 No. 1, pp. 7-17), reproduced by permission of the copyright holder Association Scientifique pour la Geologie et ses Applications (ASGA).
±7.3ppm B, Pearce et al. 1997). Straub & Layne (2002) determined δ11B using an IMS 1270. A primary beam of 8 nA of O– was accelerated through a nominal potential of 12.5 kV. The sample was pre-sputtered for 5 min with a 15 µm2 raster applied, producing a shallow 25 µm2 pit. Analyses were then performed with the primary beam focused to a 12 µm diameter spot centered within this pre-sputtered area. The field aperture size was adjusted to restrict collection of secondary ions to an area of 15–20 µm2, just slightly larger than this spot diameter. Secondary ions were extracted into the mass spectrometer through a nominal potential of 10 kV. Signals were collected cyclically, by magnetic switching, for the following analyte peaks: 9.33Da (background, 1s), 10B+ (6s), 11B+ (4s) and 30Si+ (2s). Waiting times of 0.5–1.0 s were allowed before counting each peak to permit magnet settling. Measurements were made for between 80 and 160 cycles (25–50 min), depending on the concentration of B in the sample. At these primary beam currents, signals were typically 103 cps/ppm B for 11B+. B peak intensities, and cyclical stability of measured 11B+/10B+, were monitored during each analysis for any indication of either surface contamination or inclusions within the glass. IMF was corrected for by comparison with intraday measurements of the synthetic glass standard GB4 of Chaussidon et al. (1997). However, this material is only homogeneous to approximately ±1.5 ‰ (1σ) for δ11B (Straub & 40
SIMS IN THE DETERMINATION OF LIGHT STABLE ISOTOPES IN SILICATE MELT INCLUSIONS
Layne 2002, Chaussidon & Jambon 1994). In the study by Straub & Layne (2002), overall reproducibility for δ11B was demonstrated as better than ± 0.3 to 0.6 ‰ on individual shards of a dacite glass sample (IZB 90, SiO2, 65 wt.%, B, 48–51 ppm), with an overall reproducibility for a one week session of ±0.7‰ (1σ). This reproducibility was routinely possible even for glass with 10 ppm B. In subsequent work at Woods Hole Oceanographic Institute (WHOI), GB-4 has been replaced by the glass B6 (Gonfiantini et al. 2003) a natural obsidian from Lipari Island. It is enriched in B (~210 ppm ) and, based on replicate IMS 1270 analyses at WHOI (summarized in Gonfiantini et al. 2003), appears homogeneous to better than ±0.2‰ (1σ) for δ11B. Kasemann et al. (2001), Schmitt et al. (2002) and LeRoux et al. (2004) presented overlapping values for δ11B in NIST 610, as determined by TIMS and ICP–MS. Therefore NIST 610 also offers the potential of a more homogeneous and widely available standard (versus GB-4) for future SIMS studies. Straub & Layne (2002) and Schatz et al. (2004) demonstrated that 30Si++ (14.987 Da) was an appropriate alternative to 30Si+ as a reference peak for B concentration determination, with the advantage of reducing the range of magnetic field required for peak switching determinations. This should encourage simultaneous determination of B and δ11B in future studies including, potentially, those by multicollection instruments. Boron is highly mobile in the laboratory environment as aerosols and in water and cleaning solvents and exotic B is routinely present in sample surfaces. Effective elimination of this exotic B is essential for accurate determination of δ11B. Chaussidon et al. (1997) detailed extended precautions, including pre-cleaning of samples in ultra pure water. Studies of materials with more complex surface characteristics, such as clays (e.g., Williams et al. 2001), have incorporated precleaning of samples with a mannitol solution before analysis. However, for the study by Straub & Layne (2002) no special precautions were taken in sample preparation apart from a final vigorous buffing of the sample surface with a dry lint-free clean room cloth before Au coating to remove solvent residues from pre-cleaning with isopropanol. Subsequent to Au coating, the main procedure for effectively eliminating the contribution of exotic B (to less than the equivalent of 1 ppb B) was pre-sputtering an area slightly larger than the field of view (as determined by the field aperture size) with a
rastered beam before spot analysis of glass. For the purpose of melt inclusion analysis, therefore, careful use of pre-sputtering will often be sufficient to eliminate exotic B from determinations of δ11B and B. There are also several recent publications on the determination of δ7Li in natural glasses using SIMS. Chaussidon & Robert (1998) performed a cosmochemical study of δ11B and δ7Li from the Semarkona LL3 chondrite meteorite. They used an IMS 3f, and determined δ7Li and δ11B in tandem, using similar conditions to those of Chaussidon et al. (1997) for δ11B. IMF for δ7Li was calibrated using a synthetic glass to which the NIST LSVEC Li2CO3 standard had been added. Measurements made on a natural MORB glass, and an assemblage of olivine, pyroxene and glass in a natural peridotite, did not show significant matrix effects on IMF for this wide range of compositions. This effect was confirmed by Decitre et al. (2002), who performed δ7Li analyses using the same IMS 3f for their study of serpentinization of oceanic peridotites. In the Decitre et al. (2002) study, a 10–20 nA primary beam of O–, accelerated through 10 kV, was delivered as a 25 µm diameter spot on the sample. MRP of 1100 was used to separate 6LiH+ from 7Li+. As for Chaussidon et al. (1997) the energy slit was kept at maximum width (>120 eV). Magnetically switched peaks were collected for 120 cycles of 6Li+ (6 s) and 7Li+ (3 s), with the magnetic field position for the peaks scanned and re-centered every 10 cycles. A wide range of standard compositions plotted on a single IMF calibration line (Fig. 2-9). Consequently, it was possible to use a single sample of fused natural basalt (BHVO, 5 ppm Li) to calibrate IMF for glass and mafic mineral samples. Chaussidon et al. (1997) and subsequent studies, including Decitre et al. (2002), have noted a seemingly unusual feature of the IMF for 7Li+/6Li+ – it is positive (+4% to +6%), meaning that the heavier ion (7Li+), is enriched in the measured ratio. The reproducibility of δ7Li for replicate BHVO analyses was better than ±1‰ (1σ) during single analytical sessions. Decitre et al. (2002) determined Li concentrations separately using energy filtering, counting 7Li+ and 30Si+ at an MRP of 500, with an energy slit of ±10 eV and a sample offset potential of 80 V. Olivine, pyroxenes, biotite, amphibole, and basalt plot on a single working line for Li concentration (Fig. 2-10), demonstrating the absence of large matrix effects on Li ion yield for 41
GRAHAM D. LAYNE
these bulk compositions. The serpentine standard UB-N scatters significantly from this line, due to an inherently inhomogeneous content of Li. Gurenko & Schmincke (2002) used an IMS 1270 to emulate the same conditions established by Chaussidon & Robert (1998) in their earlier studies of δ11B and δ7Li using the IMS 3f. Their precision and reproducibility for these two quantities in replicate analyses of the GB-4 standard glass were consequently similar to this earlier study. Since the natural glasses examined in this newer study contained less than 1 ppm of total B, the precision of individual analyses of δ11B was limited to ±2.2 to ±3.8‰ (1σ). Kobayashi et al. (2004) performed both δ11B and δ7Li determinations using an IMS 1270 in multicollection mode, for their study of olivinehosted melt inclusions from Hawaiian lavas. Primary ion beams were 15 nA O– for Li, and 20 nA O– for B, producing final pit diameters of ~20, and ~30 µm, respectively. MRP was ~2000 for both determinations, near the intrinsic minimum MRP of the IMS 1270. Secondary ions were extracted through the standard 10 kV potential. A 50 eV energy window was utilized, with no sample offset
FIG. 2-9. Accepted value of δ7Li by TIMS (Bristol University) versus δ7Li measured by SIMS (IMS 3f). High correlation of both mafic mineral and basalt glass standards (BHVO) illustrates a negligible difference in matrix effect on IMF for these compositions (after Decitre et al. 2002). Copyright 2002 American Geophysical Union, reproduced in modified form by permission of American Geophysical Union.
FIG. 2-10. Ion yield calibration lines for mineral and basalt standards (four different analytical sessions): Li/Si (as normalized from 7Li+/30Si+) measured by IMS 3f SIMS versus an accepted Li/Si measured conventionally on bulk powders. Standards include biotite, clinopyroxene MC and olivine MC (Massif Central, France), clinopyroxene BZ29 and olivine BZ29 (Zabargad lherzolite), amphibole (Kipawa) and basaltic glass (Nazca). Serpentine UB-N (Vosges) appears intrinsically inhomogeneous for Li/Si. This line illustrates the independence of Li+ ion yield under sputtering from matrix effects for these minerals and glasses (after Decitre et al. 2002). Copyright 2002 American Geophysical Union, reproduced in modified form by permission of American Geophysical Union.
42
GRAHAM D. LAYNE
potential. All peaks were counted with miniaturized ems positioned on the multicollector array. For both δ7Li and δ11B determinations a single analysis consisted of 550 s counting after 5 min of presputtering. Typical sensitivities were ~2000 cps/ ppm/nA for 7Li+ and ~50 cps/ppm/nA for 11B+. Concentrations of Li and B in the glass inclusions were subsequently determined independently, using a Cameca IMS 5f (after Nakano & Nakamura 2001). The reference materials for IMF were a series of five synthetic glasses, all fused from a homogeneous powder of a basaltic andesite from Izu-Oshima, Japan, but spiked with increasing concentrations of Li and B. IMF and differences in em gain were calibrated using linear regressions of these five glass standards as measured both before and after analysis of unknown samples. Typical precisions for individual determinations were ±0.6‰ (1σ) for δ7Li (for ~25000 cps on 7Li+, equivalent to 0.83 ppm Li) and ±0.5‰ (1σ) for δ11B (for ~23000 cps on 11B+; equivalent to 23 ppm B). Reproducibility, based on repeat measurements of the set of reference glasses was better than ±1.0‰ (1σ) for both δ7Li and δ11B. The uncertainty related to the accuracy of the IMF calibrations was estimated to be ±0.4–0.6‰ (1σ) for δ7Li, and ±0.3–0.8‰ (1σ) for δ11B. Kasemann et al. (2005) assessed the δ7Li of available basalt glass reference materials using several techniques, including SIMS. They determined δ7Li with an IMS 4f, using conditions very similar to those of Decitre et al. (2002), with the exception that the energy slit was narrowed to 50 eV. They concluded that USGS glasses GSD-1G (δ7Li 31.1±0.4‰, 1σ) and BCR-2G (δ7Li 4.1±0.5‰, 1σ) are useful standards for the calibration of IMF in δ7Li determinations of basaltic glass. However, they also noted that SIMS analyses of the NIST 610-612-614 series of microanalytical glass standards (~72 wt.% SiO2) imply a substantially different magnitude of IMF in high silica glasses. For SIMS determinations of δ7Li (IMS 4f) in USGS reference glass GSD-1G (37 ppm Li), they achieved ±0.4‰ (1σ) precisions for individual spots (~960 s analyses), external reproducibilities (σ/√n) of ±0.6‰ (1σ, n =10) for single day sessions, and ±0.2‰ (1σ) for four days. In subsequent trials of GSD-1G, using an IMS 1270 in monocollection mode, they achieved precisions of ±0.2‰ (1σ) for individual spots (~800 s analyses), external reproducibilities (σ/√n) of ±0.5‰ (1σ) for single day
sessions, and of ±0.4‰ (1σ) over two days. Applications. Gurenko & Chaussidon (1997) successfully used the analytical approach of Chaussidon et al. (1997) to establish a uniform δ11B for the Icelandic mantle (–11.3 ±1.9‰), based on their analysis of olivine-hosted melt inclusions of primary mantle melt. Gurenko & Schmincke (2002) studied the late Pliocene (~2 Ma) orthopyroxene-bearing tholeiite of the Iblean Plateau, which is believed to represent a large volume, high degree (16–30%) melting event of the mantle beneath Sicily. They determined δ7Li and δ11B in olivine- and orthopyroxene-hosted glass inclusions and their host pillow-rim glasses. Major element variations, H2O (0.2–0.5 wt.%), Cl (100–350 ppm), as determined by EPMA, and Li (4.6–5.8 ppm), Be (0.5–0.8 ppm) and B (0.6–1.1 ppm), as determined by SIMS, were deemed to reflect heterogeneity of the Iblean magma source. The δ7Li (–3.4 to +1.2‰) and lightest δ11B (–17.1‰ to –12.9‰) values observed in the glass inclusions (n=8) and their host glasses (n=3) are lighter than the average MORB, but very similar to OIB magmas. This is attributed to the presence of isotopically light mantle domains beneath the Iblean Plateau, most probably resulting from previous subduction of crust. Further, the wider variations of δ11B (as heavy as –3.1‰) over restricted ranges in δ7Li in a small subset (n=3) of the measured samples are explained by the addition of less than 2 wt.% of altered basaltic rocks to the crystallizing magma in shallow crustal reservoirs. Schmitt et al. (2002) determined δ11B and B in quartz phenocryst-hosted primary melt inclusions (n=31) from the calc-alkaline ignimbrites and lavas of the Neogene-Pleistocene central Andean Altiplano-Puna Volcanic Complex. Strongly devitrified inclusions were rehomogenized in a hydrothermal bomb apparatus for 20 hours at 800°C and 100 MPa with a CO2 pressurizing medium before SIMS analysis. The average δ11B of both melt inclusion and matrix glasses from individual ignimbrite units are relatively uniform with an overall average δ11B of 3.8 ±2.8‰ (1σ). This range overlaps that of local basement rocks (δ11B, –5 to –11‰), implying a dominantly crustal source for these magmas. Straub & Layne (2002) determined δ11B and B concentrations in matrix glasses (n=24) and glassy plagioclase-hosted melt inclusions (n=12) from Neogene fallout tephra of the Izu arc volcanic 43
GRAHAM D. LAYNE
δD, δ13C: Stable isotope ratio analysis of volatile trace element analytes with strong exotic contribution issues The primary challenge in both δD and δ13C determinations by SIMS is controlling the contribution of exotic or contaminant H and C to the secondary ion signals. In the case of δD, the extremely high dynamic range of the measured ratio (H/D >6000) provides an additional degree of difficulty. Enhanced sample chamber pumping with cryopumps or N2 cold fingers can help to reduce the partial pressure of H and C-based gas species near the sample. However, most exotic contributions to H and C signal appear to originate from the sample surface itself, with localized migration to the site of sputtering during analysis. Special mounting procedures, such as replacing epoxide mounting media with indium metal, may have some benefit in reducing surface contamination (Hauri et al. 2002). Pre-preparation of samples in vacuum ovens, and/or high vacuum chambers for periods of up to days before analysis, has a marked impact on contaminant signals. However, careful attention to pre-sputtering of the sample, and the exact conditions of secondary ion collection, are crucial in performing accurate analyses. The first detailed published application of SIMS to the determination of δD in terrestrial materials was that of Deloule et al. (1991) who explored the δD analysis of hydrous minerals. They utilized a primary beam of 2–5 nA of O– rastered over an approximately 30 µm2 area. The dynamic transfer optical system (DTOS) of the IMS 3f was used to maintain MRP, even with this larger rastered beam spot. Mass resolution was set at MRP 1300 to ensure separation of H2+ interference from D+ (Fig. 2-11). The energy slit was maintained wide open (>120 eV) and no sample offset was applied. Ion detection was accomplished with a regular format em in pulse counting mode. Peak switching analyses were run for 90 to 120 minutes, to achieve precision of approximately ±5‰ (1σ). To mitigate against contaminant H, both samples and the IMS 3f sample chamber were baked at 120ºC. A liquid nitrogen cold finger was also active within the sample chamber during analysis. The ratio of H2+/H+ appears diagnostic of moisture contamination of the sample, and thus analyses were only carried out on samples that had reached a condition where the measured H2+/H+ was <8 * 10–4.
front (Izu VF, ODP Site 782A). These samples display high B abundances (10–60 ppm) and heavy δ11B (+4.5‰ to +12.0‰), substantially enlarging the range previously reported for Izu VF rocks (δ11B, +7.0‰ to +7.3‰). Further, these samples display clear negative correlations of δ11B with LILE/Nb ratios. These correlations cannot be properly explained by the mixing of two separate slab fluids; i.e., fluids originating from the subducting sediment and the subducting basaltic crust, respectively. Instead, the authors’ preferred model involves a mixture of a low δ11B (~+1‰) ‘composite’ slab fluid (a mixture of metasedimentary rock and metabasalt-derived fluids) with a metasomatized mantle wedge containing elevated B (~1–2 ppm) and heavy δ11B (~+14‰). The mantle wedge was likely originally metasomatized by 11B-rich fluids beneath the outer forearc, and subsequently down dragged to arc-front depths by the descending slab. Incorporation of Pb–B isotope systematics into this model yields an estimate that >50% of the B in the Izu VF rocks was derived from an underlying mantle wedge. Straub & Layne (2002) thus concluded that down-dragged serpentinized mantle wedge is an important reservoir in the eventual recycling of B through arc front volcanism and it has an important effect on the long term recycling of crustal constituents in the subduction factory. Kobayashi et al. (2004) determined δ7Li 11 and δ B (and Pb isotopes) in olivine-hosted melt inclusions (n=28) from lava samples of the Hawaiian volcanoes Kilauea Iki, Mauna Loa, and Koolau. Host olivine grains were heated in Pt crucibles (1 atm, QFM buffer) for 30 minutes at 1190ºC to rehomogenize all melt inclusions to glass prior to analysis. Overall, these melt inclusions showed substantial variations in both δ7Li (–10.2 to +8.4‰) and δ11B (–10.5 to +5.2‰), markedly exceeding those previously measured in whole-rock samples from these same volcanoes. The authors interpreted the lowest observed values for δ7Li (minimum –10.2‰ in Koolau) and δ11B (minimum –10.5‰ in Mauna Loa) as indicative of an isotopically light Li and B source. They considered this source most likely involved the recycling of crustal materials that had experienced near-surface alteration and then dehydration during subduction, and that it played a potentially important role in creating the geochemical and isotopic heterogeneity observed in Hawaiian lavas. 44
SIMS IN THE DETERMINATION OF LIGHT STABLE ISOTOPES IN SILICATE MELT INCLUSIONS
analytical approach to δD and δ13C may be the minimization of H– desorbed directly from the sample surface due to the impact of “stray” or continuum flux electrons consequent to small imperfections in the potential matching of the electron flood. Once good vacuum was established in the sample chamber (<5 * 10–9 torr), Hauri et al. (2002) routinely measured an H2–/D– ratio of < 1.5 * 10–3 in volcanic glasses containing 0.1 to 5.6 wt.% H2O. δD analyses were thus performed, at the 2–3‰ level of accuracy, without excluding H2– interference on D– by either MRP or energy filtering. Instead, δD was determined at the lowest mass resolution of the IMS 6f (MRP 300), to preserve the greatest possible transmission of D–. Hauri et al. (2002) adopted an additional strategy for minimization of contaminant H signal, using a combination of image field magnification (100 µm maximum field of view) and limiting field aperture selected so that only secondary ions from the central 20 µm diameter of the 40 µm diameter sputtered crater are collected by the mass spectrometer. This produces an effective barrier to the incursion of exotic H from the perimeter of the sputtered crater, but produces a potentially substantial variation in measured ratios from spot to spot, due to the lateral variation expected in D–/H– across the diameter of the sputtered crater (Shimizu & Hart 1982). An alternative approach is one used in earlier studies of H2O concentrations in basaltic glasses (Sisson & Layne 1993) where the sample is pre-sputtered with a rastered beam of slightly greater area than the field of view, as limited by the field aperture. Analyses are then performed with a focused spot beam. This also produces an effective barrier to surface H migration, but allows collection of virtually all H– sputtered from the sample. Hauri et al. (2002) used ion detection by means of a regular format em (model ETP133H). Primary beam current was adjusted from sample to sample to maintain a count rate on H– of >106 cps. This required primary beam currents as high as 10– 15 nA for 0.1 wt.% H2O. IMF drift for D/H was usually limited to 5‰ per day. For samples with >0.5 wt.% H2O, 2–3‰ (1σ) precision was generally achieved for a 20–40 min analysis, and reproduceibility for replicate analyses of the same sample approached that governed by the precision of the individual spots. IMF calibration relied on a series of basalt, andesite and rhyolite standards with known δD. Pending a more detailed presentation of matrix effects on IMF, Hauri et al. (2002)
103
MRP 1250
H2+ 2
I(cps)
10
101
D+
100
10-1 2.014
Da
2.016
FIG. 2-11. Spectrum at nominal mass of 2 Da showing H2+ and D peaks resolved at MRP 1250. Primary beam was 5nA of O– focused to a 15µm spot. (after Deloule et al. 1991).
Deloule et al. (1991) used their results to undertake a detailed examination of the variation of IMF with major element chemistry in amphiboles and micas. The most detailed published procedures for δD and δ13C by SIMS in natural glasses are those of Hauri et al. (2002) and Hauri (2002) using an IMS 6f. In contrast to most previous studies of δD and H2O concentrations in glassy samples, Hauri et al. (2002) used negative secondary ions. This approach is inherently well suited to δ13C determination, since C ionizes quite efficiently as C– under Cs+ primary ion bombardment. The use of H– as the analyte for δD is a strategy that appears beneficial in reducing exotic contamination. The H+ analyte used in previous studies (e.g., Deloule et al. 1991) seems inherently prone to contamination effects, since 1H+, a simple proton, is expected to be omnipresent in sample surfaces. Cs+ primary ion beams (10 kV) were collimated using Kohler illumination (see Instrumentation, above) of the final primary beam aperture, to produce spots of 40 µm diameter by slightly defocusing the beam on the sample. Charge compensation with the NEG was employed, using procedures essentially similar to those commonly employed for δ18O determinations with f-series instruments. The total electron current generated by the NEG was reduced to the minimum value effective for charge compensation. Hauri et al. (2002) presented this as a precaution against H– desorption from or within the transfer/electron gun optics. However, the larger benefit for many types of
45
GRAHAM D. LAYNE
demonstrated that IMF for D/H is –20 to – 26% and is correlated with the H2O and Fe contents of their reference glasses. IMF for rhyolite was often 2–4% greater than for basalt and andesite. Overall uncertainty for δD analyses was cited as ±5‰ (1σ, Hauri 2002). Hauri et al. (2002) also performed δ13C analyses with the IMS 6f using similar conditions to those for δD, except that mass resolution was increased to MRP 3200 to eliminate the interference of 12CH– on 13C–. IMF calibration was based on andesite and basalt standards. Assorted basalt glass standards gave the same IMF to ±1‰, on the order of –3.3 to –3.9% for D/H. IMF was lower in andesite glasses (–2.8 to –3.2%). The precision of individual (60 min) δ13C analyses was ±2–3‰ (1σ), limited largely by counting statistics. Overall uncertainty for δ13C analyses was stated as ±5‰ (1σ, Hauri 2002).
H2O values back along open system degassing curves using estimates of the primary H2O content of each inclusion. The author concluded that there are significant contrasts in the δD of the primary magmas for each of the Hawaiian volcanoes in this study, ranging from approximately –48‰ for Kilauea to –120‰ for Koolau. This implies a δD heterogeneity in the underlying Hawaiian mantle plume that is spatially extensive (> tens of km) and/or relatively young (<1 Ga), allowing it to persist despite the rapid rate of H diffusion expected for this regime. A small number of CO2-bearing melt inclusions from Koolau (>200 ppm CO2) were also determined for δ13C, displaying light values (–12 to –29‰). The author interpreted the broad correlation of δ13C with δD in these data as consistent with open-system degassing of CO2-rich magmas followed by mixing with less degassed magmas.
Applications: Using the techniques described in Hauri et al. (2002), Hauri (2002) used an IMS 6f instrument to measured δD as well as H2O, CO2, F, S and Cl in olivine-hosted melt inclusions from five lava samples of the Hawaiian volcanoes Loihi, Kilauea, Mauna Loa and Koolau. All melt inclusions were first briefly reheated to 1275ºC in a 1 atm furnace (QFM buffered) to rehomogenize quench crystals in the glass. Heating time was limited to 10 minutes to minimize diffusive loss of hydrogen to the host olivine. After SIMS analysis, several chemical criteria were applied to filter out inclusions containing melts that might have suffered modification of their primitive δD compositions through natural processes of H diffusion, or assimilation, which preceded or accompanied lava emplacement. These criteria included discarding data for: i) inclusions suspected of suffering significant diffusive loss of H to the olivine host (all samples with positive δD values), ii) inclusions that may have experienced assimilation of seawater components as evidenced by apparent Cl contamination (all samples with Cl/K ratios in excess of those considered “normal” for each specific volcano), iii) inclusions that may have experienced shallow degassing (all samples with CO2 <500 ppm). Of the remaining inclusions, S-saturated glasses from Loihi and Kilauea were treated as fully preserving primary pre-eruptive δD compositions. Original δD compositions for low-S glasses were estimated by extrapolating the measured δD and
CONCLUSIONS SIMS offers unique capabilities for the determination of light stable isotope ratios in melt inclusions, and other very small objects. For many systems (δ7Li, δ11B, δ18O, δ34S, δ37Cl) analyses are now readily possible at a sub-per mil level of overall reproducibility. For others (δD, δ13C) the errors are somewhat larger, but still provide unique information invaluable to the study of magmatic processes. Ongoing improvements in the available instrument technology for SIMS, and the evergrowing experience of geochemists using this technique in their research, will undoubtedly lead to further improvements in the overall accuracy and capabilities of this analytical technique in the years to come. As detailed in chapters of this volume, LA–ICP–MS is playing an increasingly important and valuable role in the in situ chemical analysis of melt inclusions. However, potential applications of LA–ICP–MS to the determination of δD, δ13C and δ18O are complicated by the expectation of strong contamination and background effects for these elements in the sub-atmospheric environment of the ablation cell. Determinations of δ34S and δ37Cl at trace concentrations in glass may be rendered suboptimal due to the need to utilize positive analyte ions, as produced in the plasma. Conversely, S– and Cl– are produced extremely efficiently in SIMS under Cs+ sputtering, providing excellent sensitivity for stable isotope ratio analysis. LeRoux et al. (2004) have demonstrated considerable success in the accurate determination of δ11B in 46
SIMS IN THE DETERMINATION OF LIGHT STABLE ISOTOPES IN SILICATE MELT INCLUSIONS
basalt glass with LA–ICP–MS. However, the sample volume consumed for a precise δ11B analysis by LA–ICP–MS is 105 times higher than for SIMS, rendering analysis of individual melt inclusions impossible. It is telling that most of the published applications examples cited in this chapter involve phenocryst-hosted melt inclusions from volcanic rocks. To date, SIMS has had very limited application to the determination of light stable isotopes in melt inclusions from plutonic rocks. It is my hope and intention in writing this chapter that it will stimulate and encourage additional applications of SIMS to the determination of light stable isotopes in this important field of igneous geochemistry.
CHAUSSIDON M. & LIBOUREL G. (1993): Boron partitioning in the upper mantle: an experimental and ion probe study. Geochim. Cosmochim. Acta. 57, 5053–5062.
ACKNOWLEDGEMENTS I owe an enormous debt to the proprietors of other SIMS facilities who have let me visit and lurk around their instruments (sometimes even when they weren’t home), and have magnanimously let me tap their vast experiential and technical knowledge over the years. In particular, amongst these many colleagues are Rick Hervig (ASU), Kevin McKeegan, Marty Grove and Axel Schmitt (UCLA), Martin Whitehouse (NORDSIM, Stockholm), Marc Chaussidon, Etienne Deloule and Denis Mangin (CNRS–CRPG, Nancy), John Craven and Richard Hinton (Edinburgh). The National Science Foundation has provided support for my involvement in multi-user facilities since 1993, first at the University of New Mexico Advanced Materials Laboratory, and more recently through NSF–EAR Instrumentation and Facilities Program grants to the Northeast National Ion Microprobe Facility at Woods Hole Oceanographic Institution. My engineer at WHOI, Peter Landry, has been unfailingly patient and expert in keeping the NENIMF instruments in perfect order over the past eight years. Lucinda Gathercole applied her considerable expertise to the improvement of the figures contained within this paper. The original manuscript was greatly improved in response to thoughtful and constructive comments by Nicole Metrich and an anonymous reviewer.
DECITRE, S., DELOULE, E., REISBERG, L., JAMES, R., AGRINIER, P. & MEVEL, C. (2002): Behavior of Li and its isotopes during serpentinization of oceanic peridotites. Geochem. Geophys. Geosyst. 3:10.1029/2001GC000178.
CHAUSSIDON, M., ROBERT, F., MANGIN, D., HANON, P. & ROSE, E.F. (1997): Analytical procedures for the measurement of boron isotope compositions by ion microprobe in meteorites and mantle rocks. Geostandards Newsletter. 21, 7–17. CHAUSSIDON, M. & ROBERT, F. (1998): 7Li/6Li and 11 10 B/ B variations in chondrules from the Semarkona unequilibrated chondrite. Earth Planet. Sci. Lett. 164, 577–589.
DELOULE, E., FRANCE-LANORD, C. & ALBAREDE, F. (1991): D/H analysis of minerals by ion probe. In: Taylor, H.P., O’Neil, J., Kaplan, I.R. (Eds.), Stable Isotope Geochemistry: A Tribute to Samuel Epstein. Geochemical Society Special Publication. 3, 53– 62. EILER, J.M., GRAHAM, C. & VALLEY, J.W. (1997): SIMS analysis of oxygen isotopes: Matrix effects in complex minerals and glasses. Chem. Geol. 138, 221–244. EILER, J.M., MCINNES, B., VALLEY, J.W., GRAHAM, C.M. & STOLPER, E.M. (1998): Oxygen isotope evidence for slab-derived fluids in the sub-arc mantle. Nature 393, 777–781. GODON, A., WEBSTER, J.D., LAYNE, G.D., JENDRZEJEWSKI, N. & PINEAU, F. (2004): Secondary Ion Mass Spectrometry for the determination of δ37Cl. Part II. Intercalibration of SIMS and IRMS for aluminosilicate glasses. Chem. Geol. 207, 291–303. GONFIANTINI, R., TONARINI, S., GRÖNING, M., ADORNI-BRACCESI, A., AL-AMMAR, A.S., ASTNER, M., BÄCHLER, S., BARNES, R.M., BASSETT, R.L., COCHERIE, A., DEYHLE, A., DINI, A., FERRARA, G., GAILLARDET, J., GRIMM, J., GUERROT, C., KRÄHENBÜHL, U., LAYNE, G.D., LEMARCHAND, D., MEIXNER, A., NORTHINGTON, D.J., PENNISI, M., REITZNEROVÁ, E., RODUSHKIN, I., SUGIURA, N., SURBERG, R., TONN, S., WIEDENBECK, M., WUNDERLI, S., XIAO, Y. &
REFERENCES CHAUSSIDON M. & JAMBON A. (1994): Boron content and isotopic composition of oceanic basalts: geochemical and cosmochemical ations. Earth Planet. Sci. Lett. 121, 277–291.
47
GRAHAM D. LAYNE
ZACK, T. (2003): Intercomparison of boron isotope and concentration measurements, Part II: Evaluation of results. Geostandards Newsletter: J. Geostandards and Geoanalysis. 27, 41–57.
Chemistry 77, 5251–5257. KASEMANN, S., MEIXNER, A., ROCHOLL, A., VENNEMANN, T., ROSNER, M., SCHMITT, A. K. & WIEDENBECK, M. (2001): Boron and oxygen isotope composition of certified reference materials NIST SRM 610/612 and reference materials JB-2 and JR-2. Geostandards Newsletter, Journal of Geostandards and Geoanalysis. 25, 405–416.
GURENKO, A.A. & CHAUSSIDON, M. (1997): Boron concentrations and isotopic compositions in the Icelandic mantle: evidence from glass inclusions in olivine. Chem. Geol. 135, 21–35. GURENKO, A.A., CHAUSSIDON, M. & SCHMINCKE, H-U. (2001): Magma ascent and contamination beneath one intraplate volcano: Evidence from S and O isotopes in glass inclusions and their host clinopyroxenes from Miocene basaltic hyaloclastites southwest of Gran Canaria (Canary Islands). Geochim. Cosmochim. Acta. 65, 4359–4374.
KOBAYASHI, K., TANAKA, R., MORIGUTI, T., SHIMIZU, K. & NAKAMURA, E. (2004): Lithium, boron, and lead isotope systematics of glass inclusions in olivines from Hawaiian lavas: evidence for recycled components in the Hawaiian plume. Chem. Geol. 212, 143–161.
GURENKO A.A. & SCHMINCKE H.-U. (2000): S concentrations and its speciation in Miocene basaltic magmas north and south of Gran Canaria (Canary Islands): Constraints from glass inclusions in olivine and clinopyroxene. Geochim. Cosmochim. Acta. 64, 2321–2337.
LAYNE, G.D., GODON, A., WEBSTER, J.D. & BACH, W. (2004): Secondary Ion Mass Spectrometry for the determination of δ37Cl. Part I. Ion microprobe analysis of glasses and fluids. Chem. Geol. 207, 277–289. LEROUX, P.J., SHIREY, S.B., BENTON, L., HAURI, E.H. & MOCK, T.D. (2004): In situ, multiplemultiplier, laser ablation ICP–MS measurement of boron isotopic composition (δ11B) at the nanogram level. Chem. Geol. 203, 123–138.
GURENKO, A.A. & SCHMINCKE, H-U. (2002): Orthopyroxene-bearing tholeiites of the Iblean Plateau (Sicily): Constraints on magma origin and evolution from glass inclusions in olivine and orthopyroxene. Chem. Geol. 183, 305–331.
MCKEEGAN, K.D. (1987): Oxygen isotopes in refractory stratospheric dust particles: proof of extraterrestrial origin. Science 237, 1468–1471.
HAURI, E.H. (2002): SIMS investigations of volatiles in volcanic glasses, 2. Isotopes and abundances in Hawaiian melt inclusions. Chem. Geol. 183, 115–141.
MIGEON, H.N., SCHUHMACHER, M. & SLODZIAN, G. (1990): Analysis of insulating specimens with the Cameca IMS 4f. Surface Interface Analysis 16, 9–13.
HAURI, E.H., WANG, J., DIXON, J.E., KING, P.L., MANDEVILLE, C. & NEWMAN, S. (2002): SIMS investigations of volatiles in volcanic glasses, 1. Calibration, sensitivity and comparisons with FTIR. Chem. Geol. 183, 99–114
NAKANO, T. & NAKAMURA, E. (2001): Boron isotope geochemistry of metasedimentary rocks and tourmalines in a subduction zone metamorphic suite. Phys. Earth Planet. Inter. 127, 233–252.
HERVIG, R.P., WILLIAMS, P., THOMAS, R.M., SCHAUER, S.N. & STEELE, I.M. (1992): Microanalysis of oxygen in insulators by secondary ion mass spectrometry. Int. J. Mass Spectrom. Ion Processes 120, 45–63.
PEARCE, N.J.G., PERKINS, W.T., WESTGATE, J.A., GORTON, M.P., JACKSON, S.E., NEAL, C.R. & CHENERY, S.P. (1997): Compilation of new and published major and trace element data for NIST SRM 610 and NIST SRM 612 glass reference materials. Geostandards Newsetter 21, 115–144.
IRELAND, T.R. (1995): Ion microprobe mass spectrometry: Techniques and applications in cosmochemistry, geochemistry and geochronology. Advances in Analytical Geochemistry. 2, 1–118.
RICIPUTI, L.R. & PATERSON, B.A. (1994): High Spatial Resolution Measurement of O isotope ratios in silicates and carbonates by ion microprobe. Am. Mineral. 79, 1227–1230
KASEMANN, S.A., JEFFCOATE, A.B. & ELLIOT, T. (2005): Lithium isotope composition of basalt glass reference material. Analytical
48
SIMS IN THE DETERMINATION OF LIGHT STABLE ISOTOPES IN SILICATE MELT INCLUSIONS
SCHATZ, O.J., DOLEJŠ, D., STIX, J., WILLIAMSJONES, A.E. & LAYNE, G.D. (2004): Partitioning of boron among melt, brine and vapor in the system haplogranite–H2O–NaCl at 800oC and 100 MPa. Chem. Geol. 210, 135–147.
SLODZIAN, G., CHAINTREAU, M., DENNEBOUY, R. & ROUSSE, G. (2001): Precise in situ measurements of isotopic abundances with pulse counting of sputtered ions. Eur. Phys. J. Appl. Phys. 14, 199. SLODZIAN, G., HILLION, F. & STADERMANN, F.J. (2004): QSA influences on isotopic ratio measurements. Applied Surface Science. 231– 232, 874–877.
SCHAUER, S.N. & WILLIAMS, P. (1990): Elimination of cluster interferences in secondary ion mass spectrometry using extreme energy filtering. Int. J. Mass Spectrom. Ion Processes 103, 21–29.
STRAUB, S.M. & LAYNE, G.D. (2002): The systematics of boron isotopes in Izu arc front volcanic rocks. Earth Planet. Sci. Lett. 198, 25– 39.
SCHMITT, A.K., KASEMANN, S., MEIXNER, A. & RHEDE, D. (2002): Boron in central Andean ignimbrites: Implications for crustal boron cycles in an active continental margin. Chem. Geol. 183, 333–347.
VALLEY, J.W., GRAHAM, C.M., HARTE, B., EILER, J.M. & KINNY, P.D. (1998): Ion microprobe analysis of oxygen, carbon and hydrogen isotopes. Reviews in Economic Geology 7, 73–98.
SHIMIZU, N. & HART, S.R. (1982): Applications of the ion microprobe to geochemistry and cosmochemistry. Ann. Rev. Earth Planet. Sci. 10, 483–526.
WILLIAMS, L.B., HERVIG, R.L., HOLLOWAY, J.R. & HUTCHEON, I. (2001): Boron isotope geochemistry during diagenesis. Part I. Experimental determination off fractionation during illitization of smectite. Geochim. Cosmochim. Acta 65, 1769–1782.
SHIMIZU, N., SEMET, M.P. & ALLEGRE, C.J. (1978): Geochemical applications of quantitative ion microprobe analysis. Geochim. Cosmochim. Acta. 42, 1321–1334. SISSON, T.W. & LAYNE, G.D. (1993): H2O in Basalt and Basaltic Andesite Glass Inclusions from Four Subduction–Related Volcanoes. Earth Planet. Sci. Lett. 117, 619–635.
49
GRAHAM D. LAYNE
50
CHAPTER 3: IN SITU LASER-ABLATION-ICP-MS CHEMICAL ANALYSIS OF MELT INCLUSIONS AND PROSPECTS FOR CONSTRAINING SUBDUCTION ZONE MAGMATISM Thomas Pettke University of Bern, Institute of Geological Sciences, Baltzerstrasse 1+3, CH–3012 Bern, Switzerland
[email protected] (EPMA) and secondary ion mass spectrometry (SIMS) have been utilized to analyze a fraction of the total MI exposed at the sample surface. Pioneering work using laser ablation–inductively coupled plasma–mass spectrometry (LA–ICP–MS) followed the same approach, drilling pure MI material out of the sample (Taylor et al. 1997). Most LA–ICP–MS MI applications since have provided an alternative to SIMS analysis, notably for middle to heavy trace elements (e.g., Taylor et al. 1997, Kamenetsky et al. 1999, de Hoog et al. 2001, Kamenetsky et al. 2002). Obviously, polyphase MI have to be homogenized in the lab to produce a homogeneous single phase (glass) followed by exposure to the sample surface for analysis. Having obtained a glass, the analyst can test for homogeneity by measuring multiple spots on exposed large MI. However, the mere presence of a homogeneous glass in the MI after homogenization does not guarantee that this glass is chemically representative of the originally trapped melt. An incorrect amount of host mineral may have been remelted into the MI, or the melting conditions of the MI were modified in response to volatile loss, e.g., H2 or H2O. As a consequence, a large variety of crystallized MI exists from various geotectonic settings, notably from volatile-rich plutonic rocks, that cannot be homogenized to a glass by heating them up to independently estimated entrapment temperatures. If heating to considerably higher temperatures is required to remelt a crystallized MI, this suggests volatile loss, resulting in a glass composition that will not be identical to that of the melt at the time of entrapment. In light of the above concerns, some researchers have tried to identify and investigate samples with minimal extent of post-entrapment modifications in subduction zone settings, for example, from tephra, small diameter lapilli or tuff/ash. Shortcomings of possible sampling bias towards naturally glassy MI or MI that could be homogenized in the lab at entrapment
INTRODUCTION Melt inclusions (MI) may provide direct samples of parental liquids during growth of their host crystals (e.g., Anderson 1974, Clocchiatti 1975, Watson 1976, Roedder 1979, Sobolev 1996, Lowenstern 1995, Frezzotti 2001, Schiano 2003). They may form in virtually all types of phenocrysts; hence MI compositions monitor the chemical changes of the residual liquid during magma evolution. Thanks to petrographic and petrologic control of the host mineral crystallization and MI entrapment sequence, such data provide powerful constraints on source magma characteristics and on (deep) processes such as fractional crystallization, assimilation, magma mixing and volatile saturation. Such detailed resolution of chemical signatures using MI holds potential for resolving the chemical evolution of magmatic systems in great detail. This, in turn, provides an excellent tool for elucidating igneous processes at various scales and stages in different geotectonic settings. Melt inclusions are commonly only up to a few tens of micrometres in diameter; hence, an in situ microbeam technique is required to analyze their chemical composition (for reviews see e.g., Roedder 1984, Ihinger et al. 1994, Lowenstern 1995). Melt inclusions may undergo several transformations after entrapment such as crystallization, fluid exsolution or diffusive equilibration with host minerals and external magma, collectively called post-entrapment modifications (e.g., Danyushevsky et al. 2000, 2002, Tait 1992, Watson 1976, Sobolev & Chaussidon 1996, Gaetani & Watson 2000, Qin et al. 1992, Kress & Ghiorso 2004). Therefore, it is not trivial to backtrack to the correct chemical composition of the original liquid. Limited understanding of these complications is probably the principal reason for the limited acceptance of quantitative MI data in igneous petrology. Traditionally, electron probe microanalysis
Mineralogical Association of Canada Short Course 36, Montreal, Quebec, p. 51-80.
51
T. PETTKE
temperatures is thus a serious concern notable for investigations of hydrous plutonic systems. A unique feature of LA–ICP–MS is the possibility to analyze heterogeneous MI as entire inclusions and to reconstitute their bulk chemical composition. Consequently, all types of MI can now be analyzed for major to trace element compositions, considerably expanding the MI applications in igneous petrology. This paper focuses on the technique of bulk chemical MI analysis by LA–ICP–MS as an alternative approach in MI research in an attempt to illustrate its potential for investigations related to hydrous plutonic systems where crystallized MI are ubiquitous. This development builds on pioneering work devoted to the chemical analysis of entire fluid inclusions by LA–ICP–MS (e.g., Ramsey et al. 1992, Günther et al. 1998). The analysis of entire MI drilled out of the host crystal thereby represents the most general case of bulk inclusion analysis. This contribution presents first a discussion of fundamental parameters in LA–ICP– MS. A strategy for the optimization of LA–ICP–MS analytical conditions for geological materials is then presented, with special attention to the analysis of entire MI in minerals. Analysis and data quantification strategies of LA–ICP–MS bulk MI analysis are then outlined and compared primarily to those of SIMS and EPMA, followed by a discussion of the strengths and limitations of these techniques. A detailed discussion of SIMS applications is provided by Layne (2006); an excellent book on principles of EPMA published by Goldstein et al. (1992). This chapter then concludes with an introduction to the small, nascent literature bearing on bulk MI analysis from plutonic rocks in subduction zones and on experimental work constraining the chemical composition of aqueous fluids and hydrous melts and supercritical fluids coexisting with eclogite. These examples shall elucidate the prospects of the novel approach of LA–ICP–MS analysis of heterogeneous samples such as crystallized MI to constraining subduction zone magmatism.
These issues are relevant for MI research in general, yet their significance may vary considerably with analytical techniques employed. For a comprehensive treatment of principles of MI research the reader is referred to the contribution of Bodnar & Student (2006). Careful petrographic characterization of MI types in phenocrysts and xenocrysts in a given rock sample is a prerequisite for later interpretation of analytical data. The aim of this characterization is to establish the entrapment sequence of MI present in the sample, providing a window on the chemical evolution of the magmatic system. Fig. 3-1 shows a sketch illustrating some petrographic key observations and illustrates the basic concept of MI assemblages. When phenocrysts grow, they commonly trap the liquid from which they crystallize as inclusions. Several coevally entrapped inclusions are termed an inclusion assemblage (Fig. 3-1). We distinguish homogeneous and heterogeneous assemblages. Homogeneous assemblages (Fig. 3-1) contain inclusions of originally identical chemical composition (excluding possible boundary layer effects). Each inclusion of such an assemblage thus represents an isolated sample of a chemically uniform melt at the time of entrapment. Heterogeneous assemblages are coevally entrapped inclusions of variable chemical composition, for example coexisting melts (such as silicate and sulfide melts) or coexisting silicate melt and aqueous fluids (Fig. 3-2C). Such assemblages contain inclusions of both immiscible phases, possibly together with inclusions representing mixtures between the two end members. An assemblage of primary MI represents inclusions coevally trapped while the host crystal grew (Fig. 3-1). Hence, such MI assemblages commonly line growth zones in phenocrysts (Figs. 3-1, 2A). Assemblages of secondary MI represent inclusions that formed after the host crystals grew (Fig. 3-1). Hence, the chemical compositions of the host mineral and the secondary inclusions are most likely not related to each other. Typically, assemblages of secondary MI fill former cracks oblique to growth zoning in host minerals, and such cracks commonly cut through grain boundaries. Secondary MI assemblages may be common in xenocrystic minerals or grain aggregates (Fig. 3-1). Consequently, the correct identification of secondary inclusions is crucial for MI studies. It is important to appreciate that inclusions in proper assemblages may appear petrographically variable
Background Before discussing the various analytical approaches in more detail, I consider it necessary to briefly review some inclusion terminology, MI entrapment conditions and the various possibilities of reversible or irreversible post-entrapment modifications (as far as is currently recognized).
52
IN SITU LASER-ABLATION-ICP-MS ANALYSIS AND CONSTRAINING SUBDUCTION ZONE MAGMATISM
FIG. 3-1: Sketch of a rock thick section showing different petrographic types of melt inclusions (MI). One xenocrystic aggregate consisting of 4 crystals (number 1) and two phenocrysts (crystal 2 and 3) are set in a fine-grained groundmass. Each crystal shows former growth zones outlined in dashed, grey lines. Cracks in crystals are shown by thin dashed, black lines. Crystal 2 hosts primary MI assemblages (labeled 2p1, 2p2, 2p3, 2p4). Individual MI are either arranged along planes (as can be recognized by varying focal depth during thick section microscopic observation) parallel to the crystal surfaces or cluster in a part of the crystal, here shown in the core. One assemblage in aggregate 1, labeled 1p, is also primary, while the other two MI assemblages, labeled 1s1 and 1s2, are secondary. The latter cut across growth zoning of individual crystals, across grain boundaries within the xenocrystic aggregate, and they sometimes reach the crystal surfaces (assemblage 1s2). MI in phenocryst 3 are characterized by a geometrically random occurrence; hence, they cannot be grouped into assemblages. MI labeled 1a in xenocryst aggregate 1, and 2a in phenocryst 2, are cut by a crack. These MI should be avoided because of the possibility of alteration through fluids that circulated through the crack (indicated by the dark appearance of the respective MI). Petrographic interpretation of this sample is that the oldest MI are represented in assemblage 1p, recording a magmatic event prior to formation of this rock. Assemblages 2p1, 2p2, 2p3 and 2p4 were successively trapped while phenocryst 2 grew; hence, they record the melt evolution while this rock crystallized. The MI in phenocryst 3 are randomly distributed throughout the crystal. They can only be arranged into an entrapment sequence when growth zoning of the host crystal is independently visible. This is a rare case. Commonly, such MI cannot be reliably used to establish the chemical evolution of the melt while the crystal grew and, therefore, such MI should be avoided whenever possible. Note in the example of phenocryst 3 that the MI closest to the rim of the phenocryst is actually the oldest one in this phenocryst. Secondary inclusion assemblages 1s1 and 1s2 are not useful either because they may have trapped a melt prior to inclusion of the xenocrystic aggregate 1 in this magma, or they may contain melt from this magma trapped shortly before its solidification.
(Figs. 3-1, 2A) due to different extents of postentrapment modifications explained in detail below. Melt inclusion populations commonly encompass all inclusions trapped, for example, in one phenocryst type. In Figure 3-1, the MI hosted by phenocrysts 2 and 3 could be unified into a MI population. A MI population, therefore, commonly contains several inclusion assemblages successively entrapped while the host phenocryst grew (obvious for phenocryst 2 but not for phenocryst 3) and the
residual melt evolved. Hence, MI of a population are expected to be chemically variable and may thus not provide precise average chemical compositions of temporally well-resolved crystallization stages. Melt inclusions are useful probes of the melt present in magmatic evolution only when they represent equilibrated melt droplets. Two principal processes of MI entrapment exist, illustrated based on MI formation in olivine, (a) diffusion-controlled and (b) equilibrium entrapment (see Faure &
53
T. PETTKE
crystals. Equilibration of this boundary layer with the rest of the melt through diffusion is not possible because diffusion is too slow. Therefore, the boundary layer will be relatively depleted with elements compatible in the host, and enriched in the incompatible ones; diffusive supply of components limits the growth rate of the crystal. Resulting MI are chemically heterogeneous and may display chemical trends as a function of MI size which may be revealed when working with MI assemblages (the “boundary layer effect” commonly referred to in the literature). During equilibrium entrapment, interface attachment processes control MI formation (i.e., host mineral growth rate controlled) and the host forms polyhedral crystals. Resulting MI are chemically representative of the residual melt, irrespective of their size. Careful petrography is thus required to study the morphology of host phenocrysts in order to avoid MI assemblages that may suffer from boundary layer effects. Another point of concern is the possibility of accidentally trapped minerals in MI, having once served as a nucleus for MI formation. Such MI, too, can be identified when working with MI assemblages. Finally, MI in primitive (i.e., Fo-rich) olivine may not have sampled “geologically significant melt bodies” but rather the product of localized dissolution–reaction–mixing processes in mush zones of the magmatic plumbing system (e.g., Danyushevsky et al. 2004). The above statements emphasize the prerequisite of careful sample selection and petrographic inspection of MI for conducting a “meaningful MI study” including addressing issues of post-entrapment modifications (e.g., Roedder 1979, 1984). Two types of post-entrapment modifications need clear distinction, those which can be reversed in the lab by bringing the inclusion back to entrapment conditions, and those which irreversibly modify inclusion compositions. Every MI evolves after entrapment and may eventually nucleate a bubble inside the glass (Fig. 3-1). This bubble either forms in response to fluid exsolution from the trapped melt (for inclusions that trapped a melt near or at fluid saturation), or it may represent a shrinkage bubble resulting from post-entrapment crystallization of the host mineral onto the inclusion walls (for inclusions that trapped a strongly fluid-undersaturated melt). Such bubbles form before the MI reaches the temperature of the glass transition; hence, they provide indications of significant high-temperature fluid exsolution or sidewall crystallization. The
FIG. 3-2: Photomicrographs of various inclusion types. Image (A) shows a variably crystallized melt inclusion (MI) assemblage in plagioclase from an andesite, lining a former growth zone. Note the variable appearance of inclusions all with negative crystal shapes, from glassy with bubble (bottom left) to completely devitrified (black, without visible bubble). (B) Glassy MI in plagioclase from a dredged MORB sample showing variably sized shrinkage bubbles. (C) Heterogeneous assemblage of co-existing silicate melt (MI cluster outlined by dashed white line) and vaporlike fluid inclusions (FI identified by white arrows) in quartz (from Audétat and Pettke, 2003). Note the presence of inclusions of various mixing proportions between the two end member phases.
Schiano, 2005, for a detailed discussion and excellent MI images). Diffusion-controlled MI formation represents disequilibrium entrapment; the material enclosed in the MI corresponds to the boundary layer forming around dendritic or skeletal 54
IN SITU LASER-ABLATION-ICP-MS ANALYSIS AND CONSTRAINING SUBDUCTION ZONE MAGMATISM
Metrich & Clocchiatti 1996, Sisson & Layne 1993, Cottrell et al. 2002). Model predictions by Gaetani & Watson (2000) suggested that diffusive equilibration may be a problem when strong chemical gradients are present between the MI and its surroundings. Some MI may even decrepitate, notably in host minerals other than quartz, or MI may suffer from decrepitation of fluid inclusions in the same mineral grain when reheated in the lab at atmospheric pressure conditions (e.g., Roedder 1984, Tait 1992, Webster et al. 1997, Student & Bodnar 2004). Introduction of Cu and Ag into the MI from the crucible in which quartz phenocrysts were heated for 24h at 850°C has been demonstrated (Kamenetsky & Danyushevsky 2005) and illustrate that even the environment of lab treatment may adulterate the bulk MI composition. It is important to appreciate that while the concentrations of some elements or species in the inclusions may have been modified irreversibly in nature or in the lab, others (e.g., the large ion lithophile elements and the rare earth elements in MI hosted by plagioclase, olivine or orthopyroxene) may have remained largely unaffected (see e.g., Schiano, 2003). There is a clear need to improve our understanding of the extent of diffusive equilibration in natural samples. Volume diffusion data only provide a conservative estimate of the problem, because imperfections providing fast diffusion tracks are likely to control the rate of diffusive exchange. The artistry in MI research is to determine which element concentrations of any given MI assemblage have remained unadulterated and thus can be used for petrogenetic modeling.
extent of this crystallization obviously varies with the extent of chemical difference between the MI and host mineral composition (the more similar they are, the larger the mass fraction of post-entrapment host crystallization may be) and with the postentrapment history of the MI. Intuitively, these post-entrapment effects are minimized for quenched extrusive rocks as illustrated by the large amount of research dedicated to MI studies on volcanic ejecta from various geotectonic settings. Such MI can commonly be reverted to entrapment composition by reheating to entrapment temperature in the lab (e.g., Anderson 1974). Yet, much microbeam analytical work has been done on glassy untreated MI, thereby measuring a composition that may be more evolved than that present at the time of entrapment (e.g., Danyushevsky et al. 2002, Pettke et al. 2004). If such precise microbeam measurements are directly used for petrogenetic modelling, results may be seriously misleading. Such data require correction by modelling the reverse of host mineral crystallization onto the inclusion walls (e.g., Danyushevsky et al. 2000, de Hoog et al. 2001¸ Kress & Ghiorso 2004, Pettke et al. 2004) to obtain the correct chemical composition at entrapment. For the case of fluid bubble exsolution prior to solidification of the MI, volatile elements and some metals (e.g., Cu, Zn, Pb) will be strongly enriched in the bubble, and the analysis of the exposed glass of unheated MI will seriously underestimate the contents of such elements in the melt at the time of entrapment. Irreversible post-entrapment modifications are readily indicated if homogeneously trapped MI do not homogenize to a glass at entrapment conditions (note that designing a MI reheating experiment requires optimization of the heating rate: as fast as possible to minimize H2 diffusion out of MI but slow enough to satisfy kinetic limitations of melting; Danyushevsky et al. 2002). These modifications include all cases where mass transfer occurred through the MI–host mineral interface. The extent of such modifications ranges considerably, from diffusive loss of volatiles (most importantly H+ or H2O), to diffusive re-equilibration with the host mineral (e.g., Fe and Mg in mafic minerals) or even the external magma at elevated temperatures (either during natural slow cooling and/or during reheating in the lab), to chemical exchange with extraneous components introduced along cracks in the host (e.g., Gaetani & Watson 2000, Danyushevsky et al. 2000, 2002, Qin et al. 1992, Tait 1992, Hauri et al. 2002, Luhr 2001,
MICROBEAM INSTRUMENTATION The following section will summarize key application characteristics of LA–ICP–MS and compare them with other techniques (e.g., EPMA, SIMS, Fourier Transform Infrared Spectroscopy (FTIR) and RAMAN scattering) that are applicable to the analysis of MI. This discussion is preceded by a more fundamental consideration of key parameters and an optimization strategy for the accurate analysis of geological materials by LA– ICP–MS. The comprehensive description of these analytical techniques is beyond the scope of this paper, and the interested reader is referred to the literature cited below. Rather, a conceptual path is followed. We have geoscientific questions to be answered through the analysis of a series of MI. We try to appraise the type of data that can be obtained by using the various microbeam techniques, and we
55
T. PETTKE
try to appreciate the significance of these data sets. It will be illustrated that LA–ICP–MS is the only method to date for the bulk chemical analysis of MI that cannot be homogenized reliably in the lab. Such MI may be dominant in some hydrous magma types; hence, LA–ICP–MS MI analysis has an enormous potential to constrain better subduction zone magmatism and plutonic activity in particular. All analytical instruments share the common principle that a beam smaller than ca. 200 µm is focused on the sample and emits radiation or sample material to be measured by a detection device. An important consequence of this approach is that the response of any analytical setup is calibrated against standard materials of known composition and state (crystalline, amorphous). We distinguish two principal analytical setups. In setup one, the signal (ions or radiation) emitted from the analytical spot on the sample is directly measured (EPMA, SIMS, Raman, FTIR, or Laser-Induced Breakdown Spectroscopy (LIBS)). LIBS will not be considered further here because this technique is extremely matrix–dependent. In setup two, sample material is liberated and transported as an aerosol to a site where the signal (ions or radiation) is produced independently of the sample liberation process (LA–ICP–MS and LA–ICP–Optical Emission Spectroscopy (OES)). LA–ICP–OES will not be considered further in this contribution because limits of detection (LOD) are not competitive with LA–ICP–MS for the vast majority of elements (e.g., Pettke et al. 2000). LA–ICP–MS has a key advantage, namely that it allows independent optimization of two fundamentally different processes, sample ablation and production of ions in the ICP for analysis. This can considerably reduce matrix effects on analyte signals, provided that other key LA–ICP–MS parameters are properly taken into account. Consequently, the need for strictly matrix-matched calibration, a prerequisite for SIMS analysis and recommended for EPMA, may be strongly relaxed for LA–ICP–MS analysis.
on laser beam properties and ICP–MS settings I consider particularly relevant for the analysis of geological samples in general and MI in particular. For a complete list of parameters the reader is referred to Günther & Hattendorf (2005). Note that most results in the literature were obtained by using a specific LA–ICP–MS instrumental setup, and the various setups used possess considerably different specifications. It is therefore delicate to generalize conclusions that were obtained by one given setup. This is also the reason why there are not many numbers provided in the following section. Rather, the reader should get confronted with some basic principles about LA–ICP–MS parameters and their mutual dependence. Laser energy and its spatial energy profile (i.e., energy distribution across the ablation pit) are of paramount importance. First, enough energy density on the sample surface is required to effect ablation. This energy density is also called irradiance (energy per area, W/cm2) per pulse and relates to the laser fluence. The laser fluence is the total amount of energy per area (J/cm2) per pulse reaching the target, which is proportional to the output energy of the laser system. For fixed laser output energy, the irradiance will increase with decreasing laser pulse duration. Each material has a characteristic energy density below which ablation does not occur and the material may crack or splinter. For quartz, this ablation energy threshold corresponds to a narrow window in energy density that is variable between ~10 and 20 J/cm2 (for 193 nm excimer laser light of ~20 ns pulse duration) for different quartz samples. For example, when the ablation energy threshold is 15 J/cm2 for a given quartz sample, then no ablation occurs below ~13 J/cm2 while perfect ablation is observed above ~17 J/cm2. The ablation energy threshold is generally lower for mafic minerals. For a laser fluence above the ablation threshold and silicate or oxide matrices, typical ablation rates for 193 nm excimer laser light are 100–200 nm per pulse. Ablation rates for a given analytical setup increase linearly with increasing laser fluence for ns-pulsed UV laser light (e.g., Horn et al. 2001). It follows from the above observations that identical ablation conditions throughout the entire pit, between successive laser shots and for variable pit sizes facilitate controlled ablation. For nonhomogenized energy laser beams, some parts of the ablation pit (e.g., rim domains) may not reach controlled ablation conditions because of
LA–ICP–MS analytical setup All commercially available LA–ICP–MS setups consist of a pulsed monochromatic laser source, laser beam modulation optics, a modified petrographic microscope with TV screen observation and an ICP–MS instrument (Fig. 3-3). Various LA–ICP–MS systems have significantly different but interrelated key parameter specifications. The following paragraphs elaborate
56
IN SITU LASER-ABLATION-ICP-MS ANALYSIS AND CONSTRAINING SUBDUCTION ZONE MAGMATISM
FIG. 3-3: Sketch of a laser – ablation inductively – coupled – plasma mass – spectrometry (LA–ICP–MS) setup, consisting of a laser source, beam modulation optics, a microscope equipped with camera and TV screen (showing a 40µm pit in quartz), and a commercial quadrupole ICP–MS without dynamic reaction or collision cell. The sample located in the ablation chamber is hit by a laser beam schematically shown in grey. Detailed description of various LA–ICPMS components is provided in the text.
insufficient energy density (irradiance below the ablation threshold value), and the sample may crack or break out. Such “non-ideal” and variable ablation conditions produce more large particles, and their presence in the aerosol is considered to be a primary cause for elemental fractionation (outlined below). Laser beams with a homogeneous energy density allow for the analysis of standards and unknowns with identical laser-ablation conditions and with variable pit sizes to optimize the analytical resolution on the samples. Moreover, because uniform sample layers are ablated, the resulting pit has a flat-bottomed pan shape, and depth profiling with a sub-micrometre resolution becomes feasible.
somewhat longer pulses of ca. 20 ns. Laser pulse duration appears to be a parameter of subordinate importance for the analysis of silicates and oxides relative to metals, however, when the irradiance is clearly above the ablation threshold values for all matrices to be analyzed. Therefore, nanosecond laser sources are perfectly suitable for most geochemical applications (e.g., Günther & Hattendorf 2005). Laser wavelength is a key parameter simply because laser beam coupling with any given matrix strongly depends on laser wavelength, i.e., monochromatic light absorbance and reflectivity vary with wavelength and target matrix. Historically, almost all available wavelengths (e.g., 694 nm, Gray et al. 1985; 1064 nm, Jackson et al. 1992; 266 nm, Jenner et al. 1993) have been used for LA–ICP–MS, but research soon demonstrated that shorter wavelengths (notably 213 nm quintupled Nd–YAG and 193 nm ArF excimer lasers) are more suitable for geochemical applications when using ns-pulsed lasers (e.g., Jeffries et al. 1996, Günther et al. 1997, Guillong et al. 2003a). This is because shorter wavelengths generally couple better with silicates and other
Laser pulse duration is essential, because the irradiance increases with decreasing pulse duration for a given laser fluence. Short pulse durations allow for less heat dissipation into the matrix and, consequently, a larger fraction of sample material is vaporized (modeled by Bogaerts & Chen 2005). Present day commercial laser-ablation systems are predominantly equipped with lasers producing nanosecond pulses. Solid source lasers, e.g., the Nd–YAG lasers, are characterized by a pulse duration of ca. 5 ns, while excimer lasers produce 57
T. PETTKE
transparent materials; hence, the volume affected by a given laser pulse is reduced and thus the energy density in the affected material increased. Although this reduces the ablation rate, the aerosol produced can be processed better by the ICP, resulting in accuracy improvements and lower LOD (e.g., Guillong et al. 2003a). Because of the ease of operation of commercially available ns-pulsed UV lasers down to 193 nm wavelength together with their demonstrated performance on silicate and oxide matrices, most LA–ICP–MS geochemical labs currently acquire systems with either 213 or 193 nm laser sources.
very recently, it has become accepted that processes in the ICP fundamentally affect overall element fractionation, and these processes directly relate to the particle size and particle size distribution of the aerosol, which in turn influences aerosol transport phenomena. Additionally, the mass of aerosol per unit time reaching the plasma may be an important parameter, notably for solid state rf generators as commonly used in commercial instruments. All these and possibly other yet unidentified parameters interact to produce elemental fractionation — so, what can we do to fundamentally minimize this problem for geochemical applications?
Elemental fractionation refers to the changes of element responses (i.e., element sensitivity ratios) with changing LA–ICP–MS analytical conditions (e.g., Longerich et al. 1996b). The first parameterization of fractionation was proposed by Fryer et al. (1995) who demonstrated for a long transient (i.e., time-resolved) signal from one analytical spot that integrating the first and the second half of the signal and normalizing the resulting intensities with that of Ca does not provide the same values. It implies that sensitivity ratios may evolve with progressive laser drilling at a single spot. Consequently, results for samples may vary as a function of where the signal integration interval is set (i.e., across the entire recorded signal or only across a part of this signal). It has been shown repeatedly in the literature since then, that this type of elemental fractionation can be controlled for various LA–ICP–MS systems when ablation pits are not drilled too deeply. As a rule of thumb, ablations with pit depth to diameter ratios (pit aspect ratios) of less than about two should not suffer from such fractionation with ablation depth (e.g., Borisov et al. 2000), again keeping in mind that the extent of this phenomenon strongly depends on LA–ICP–MS setup and operating parameters. The latest progress in fundamental research on elemental fractionation has identified different locations where fractionation occurs, has constrained their relative importance, and has provided possible processes that cause them. Elemental fractionation relates to processes occurring (i) at the ablation site (aerosol particle size distribution), (ii) during aerosol transport (aerosol sorting effects related to particle size, accepting that the chemical composition of aerosol particles may vary with particle size) and, importantly, (iii) during ionization in the ICP–MS (as reviewed by Günther & Hattendorf 2005). Only
Aerosol particle size: ideally, every aerosol particle arriving in the ICP is completely vaporized, atomized, and all the atoms are ionized. Reality has shown, however, that this is not necessarily the case. Each aerosol cloud produced by laser ablation is characterized by a particle size distribution, i.e., the range and abundances in particle size fractions. These two parameters have been shown to be highly variable for different laser-ablation conditions and setups. Evidence that large particles may not be quantitatively ionized comes from different experimental results. Removal of large particles from the aerosol does not result in a proportional signal reduction, suggesting that large particles are not quantitatively ionized in the ICP–MS (Guillong & Günther 2002, Guillong et al. 2003b). Highspeed digital photography demonstrates that large particles indeed may at least partially survive transit through the ICP without being completely broken down and ionized (Aeschliman et al. 2003). Particle size fractions of the total aerosol generated with a 266 nm laser system have been shown to be chemically variable (Kuhn & Günther 2004). Incomplete vaporization of large particles preferentially liberates volatile compounds relative to refractory compounds, according to their vaporization indices. Results from copper isotope measurements by laser ablation – multiple collector ICP–MS have suggested that isotope fractionation is smallest for smallest possible particle sizes (Jackson & Günther 2003). In summary, to minimize elemental fractionation in the ICP it is best to optimize the ablation process so that the smallest possible particle size with a narrow size distribution is produced. The largest mass of silicate particles produced by an energy-homogenized 193 nm LA– ICP–MS system from silicate glass has a particle size below ca. 200 nm (Guillong et al. 2003a), and
58
IN SITU LASER-ABLATION-ICP-MS ANALYSIS AND CONSTRAINING SUBDUCTION ZONE MAGMATISM
these particles appear to be ionized completely in commercially available ICP–MS instruments run at robust plasma conditions (see below). Controlled ablation of any matrix only occurs well above the ablation energy threshold. It is therefore essential that there is enough energy at any site in the ablation pit in order to ablate the sample properly. This may not be ensured for laser systems that do not have a homogenized energy distribution at the ablation spot but simply focus their Gaussian energy distribution emitted from the laser source onto the sample surface. Ablation near the threshold energy tends to be less well controlled, showing splintering and producing poorly defined rims of ablation craters. It is intuitive that such catastrophic ablation conditions form much larger, and many more, large particles. Hence, avoiding such uncontrolled ablation conditions near the laser-ablation threshold irradiance will considerably reduce elemental fractionation.
i.e., the laser energy can almost fully be deposited into the target material before the plasma plume above the ablation spot starts to form (it forms only after a few ns of irradiance; e.g., Bogaerts & Chen 2005). As a consequence, plasma shielding, the absorption of a considerable amount of laser pulse energy by the evolving plasma above the spot, is virtually eliminated, and the sample irradiance is much higher. This may lead to an ablation process controlled by photo-physical bond breaking instead of melting, boiling and vaporization that is probably the dominant process in ns-pulsed laser ablation (e.g., Russo et al. 2004). Such a change in predominant laser ablation processes may significantly reduce elemental fractionation and matrix dependence (e.g., Poitrasson et al. 2003). Hence, the potential of picosecond and femtosecond-pulsed laser technology for LA–ICP– MS applications in the geosciences is really promising.
The ablation chamber gas environment strongly affects the resulting aerosol characteristics, too. Improved sensitivities for a nanosecond-pulsed LA– ICP–MS setup by using He as the ablation chamber gas instead of Ar (Eggins et al. 1998) probably relate to reduced deposition of particles around the ablation pit. The aerosol plume above the ablation pit may expand more freely in He when compared to Ar, thus reducing the probability of material condensation to large particles that may get deposited in the aerosol transport system. Therefore, the observed improvement in sensitivity when using He in the ablation chamber and keeping all other parameters constant probably relates to the transport of a larger mass of smaller particles to the ICP–MS.
Optimization strategy of the ICP–MS for laserablation applications The sample aerosol arriving at the plasma site of an ICP–MS instrument is vaporized, atomized and ionized; and the ions are extracted by differential underpressure through the interface, focused, filtered according to their mass-to-charge ratio and finally detected. As outlined above, it is best when the aerosol arriving at the ICP is ionized completely. Experiments have identified that this may only be approached for “robust plasma conditions”, conditions at which fractionation effects resulting from incomplete ionization are minimized. Traditionally, ICP–MS operating conditions have been optimized for low oxideproduction rates (plasma temperature-sensitive species, monitored by Th/ThO intensity ratios to be <0.5%) and maximum signal-to-noise ratio for the ions of interest. Günther & Hattendorf (2005) have summarized in detail that other optimization criteria are required to ensure robust plasma conditions. For example, the intensity ratio of U/Th, two elements with nearly equal first ionization energies, mass, and abundance of major isotopes, should correspond to the concentration ratio in the reference material. The silicate glass standard SRM610 from NIST contains 461 µg/g U and 457 µg/g Th, respectively; hence, robust plasma conditions should result in a 238U/232Th intensity ratio of one. Optimizing an ICP–MS instrument to Th/ThO <0.5% and maximum signal/noise ratio of analytes may well result in U/Th ratios much higher
Future LA–ICP–MS developments: In summary, a nanosecond-pulsed laser-ablation system providing homogeneous energy distribution across the entire pit and sufficient output energy to ablate all matrices of interest in a controlled manner appears to be currently the best choice from commercially available systems for the analysis of complex geological samples, with 193 nm as the most appropriate wavelength for silicate and oxide matrices (Günther & Hattendorf 2005). The future will show how well such samples are treated by currently much more expensive and more difficultto-control picosecond or femtosecond laser sources available with various laser wavelengths. The pulse duration of these laser sources can be shorter than the reaction time of the target to the applied energy,
59
T. PETTKE
than one (Fig. 2 in Günther & Hattendorf 2005). Analyzing samples with such an ICP–MS setting bears considerable danger that aerosol particles are not completely converted into ions. If so, the ionization rate in the ICP will be variable at least for some elements (refractory or non-volatile ones) as a function of sample matrix, and one of the most important advantages of LA–ICP–MS is lost, namely the matrix-independent calibration. The possibility of matrix-independent calibration was demonstrated for silicates by Jackson et al. (1992), aqueous fluids from inclusions (Günther et al. 1998, Heinrich et al. 2003), oxides (e.g., Heinrich et al. 2003), carbonates (e.g., Eggins et al. 2003) and even for Fe, Ni, Co and Cu in some sulfides (chalcopyrite, pyrrhotite and millerite; Halter et al. 2004a). Therefore, unless we employ matrixmatched calibration, the careful analyst has to ensure robust plasma conditions for the variable matrices as commonly analyzed in geochemistry.
For sulfide and carbonate melts, crystallites will vary in composition accordingly. As a consequence, minor and trace elements may become strongly enriched in any tiny phase in the crystallized inclusion. Signals produced from such small phases (e.g., fluid bubbles or accessory minerals in silicate MI or noble metal nuggets in sulfide MI) may be highly transient, notably for very small volume ablation chambers. Pettke et al. (2000) have illustrated and discussed causes and effects of nonrepresentative recording of such fast signals in detail. Of quintessential importance is that the transient intensity structure of the signal is defined correctly by all the isotopes of interest in sequential recording mode. This requires short (10 ms or less) dwell times (duration of analysis per isotope in one sweep, where a sweep consists of one sequential analysis of all isotopes of interest) for routines analyzing many isotopes (e.g., >20) as commonly employed in bulk MI LA–ICP–MS analysis. Consequently, the analyst has to know the transient signal intensity curve and duration of a single pulse ablation for the LA setup used (i.e., the signal dispersion of the aerosol transport system) and the number of isotopes to be measured. This allows one to establish compromise conditions between fast scanning protocols, to ensure proper recording of
Representative recording of fast, transient signals: The analysis of entire, crystallized silicate MI by LA–ICP–MS aims at correctly recording cation signals from a mixture of small phases (Fig. 3-4), including silicate, oxide and phosphate minerals, and possibly a volatile phase present in the bubble.
FIG. 3-4: Typical transient LA–ICP–MS signal of a crystallized melt inclusion in plagioclase analyzed in bulk. Note the nonparallel evolution of the inclusion signal interval, e.g., for Rb and Mg, reflecting the ablation of phases enriched in these elements. Elements enriched in the host (e.g., Ca and Sr) display a signal depression in the inclusion interval. Even such elements can be quantified in bulk MI, albeit at lower precision as explained in text.
60
IN SITU LASER-ABLATION-ICP-MS ANALYSIS AND CONSTRAINING SUBDUCTION ZONE MAGMATISM
respectively; (40Ca12C)+ on 52Cr+ in carbonates; or (56Fe32S)+ on 88Sr+ in iron sulfides. For some of these, dynamic reaction cell or collision cell ICP– MS technologies coupled with laser ablation may reduce the problem significantly (e.g., Hattendorf & Günther, 2000). Doubly charged ions may also interfere, e.g., 138Ba++ on 69Ga+ (recall that the mass filter of an ICP–MS resolves ions according to their mass/charge ratio). The mass resolution of quadrupole instruments is commonly insufficient to resolve all these types of interferences. Magnetic sector field instruments have tunable resolution and are thus capable of resolving some of these interferences; yet, increasing the mass resolution inevitably lowers the sensitivity, and consequently LOD are compromised. And only recently, sector field instruments have become fast enough for transient signal recording from inclusions (e.g., Latkoczy et al. 2002). They are still considerably slower than quadrupole instruments, however, and thus not the instrument of choice for recording of isotopes across the entire mass range (Li to U) of fast transient signals produced from MI ablation.
short transient signals, and slow scanning protocols, since long dwell times lower the LOD when keeping all other parameters constant. Another possibility of extending signals from small crystallites is to use large volume aerosol transport systems. This extends a given signal across a longer time span, thus resulting in lower signal to noise ratios that translate into higher LOD. Therefore, this is not considered to be a viable alternative for bulk MI LA–ICP–MS analysis. The concern of representative sampling of such short transient signals (Pettke et al. 2000) from tiny phases within crystallized MI may be relaxed by at least partially homogenizing the inclusions of interest. For volatile-rich inclusions, this procedure always bears the danger of decrepitation during atmospheric heating (e.g., Student & Bodnar 2004). Moreover, partial homogenization of MI is no guarantee that the content of tiny daughter crystals has become homogeneously distributed throughout the entire MI. Probably the most reliable test for representative sampling of elements exclusively contained in tiny daughter crystals is the external (i.e., inclusion-to-inclusion) reproducibility achieved for element concentrations in a MI assemblage. For signals not limited by counting statistics uncertainty, the uncertainty of average assemblage element concentrations should be uniform.
Limits of detection are a key parameter in LA–ICP– MS applications, notably for the analysis of inclusions where the sample mass available for analysis is limited. The ablated mass of sample per unit time exerts the dominant control on the resulting LOD from the laser ablation side. For spot analyses of a homogeneous phase using a 193 nm excimer laser, the LOD very roughly decreases by a factor of 4 when doubling the pit diameter. Additionally, aerosol dispersion (dominantly in the ablation chamber volume, less importantly in the transport tubing) prior to entering the ICP leads to reduced signal/background ratios for short transient signals (a given signal is smeared over a longer time), resulting in lower element sensitivities and thus higher LOD. Therefore, small volume aerosol transport systems (notably the ablation chamber) will increase the signal/background ratios for a given mass of sample as available for inclusion analysis and thus minimize the LOD. However, for minimized signal duration (resulting in highest signal/background ratios) the requirement of representative signal recording introduced above has to be properly accounted for. The laser energy density used on the sample is of subordinate importance here, provided it is clearly above the ablation threshold. The recommendations by Longerich et al. (1996a) for the calculation of LOD reveal that the quality of recording the gas
Mass interferences may also plague LA–ICP–MS analysis. Among them, plasma gas based interferences are properly accounted for by background subtraction. Isobaric interferences (e.g., 58 Fe+ on 58Ni+, being the most abundant Ni isotope) can largely be avoided by proper mass selection (for Ni in Ca-rich matrices this is 62Ni because of (44Ca16O)+ on 60Ni+) and accepting higher LOD. Alternatively, isobaric overlap can be corrected mathematically, after analysis, based on known isotopic abundances of interfering elements. Problematic interferences are polyatomic ions that form by combination of elements abundant in the plasma gas with elements abundant in the analyzed matrix. Such “dangerous” polyatomic ions include all the element oxides (simply because oxygen is the most abundant element in silicates) or, more specifically, (27Al16O)+ and (28Si16O)+ on 43Ca+ and 44 Ca+ (many major silicates); metal argides, (M40Ar)+, e.g., transition metal argides on Rb, Sr, Zr, Nb, Mo, Ru, Rh or Pd or (23Na40Ar)+ on 63Cu+; or simply pure matrix-sourced interferences, e.g., (16O16O)+ or (16O18O)+ on 32S+ and 34S+,
61
T. PETTKE
background (i.e., the scatter of the background measurement) most prominently influences the resulting LOD from the ICP–MS side. LOD values are only useful numbers for known LA–ICP–MS parameters and LOD filtering criteria. LOD using the 3 sigma criterion (Longerich et al. 1996a) can be as good as single µg/g values for elements from the middle to heavy mass range when analyzing a homogeneous phase with a pit of 80 µm and 10 Hz laser repetition rate in single spot or raster mode. For bulk MI analysis drilled out of the host mineral, resulting LOD are between a few thousand µg/g and several tens of ng/g, depending primarily on the isotope analyzed, MI size, ablation quality and element compatibility in the host (compare Fig. 12 in Heinrich et al. 2003, for LOD estimates in 25 µm inclusions). When analyzing bulk MI, best LOD values are achieved for elements incompatible in the host mineral, with fast drilling and with 30–50 µm spherical MI. Larger MI do not allow for better LOD values when analyzed in bulk because the reconstruction of the pure MI from the mixed signal is the limiting factor.
measurement of exposed glassy MI by spot analysis poses no additional problem. ANALYSIS AND DATA QUANTIFICATION STRATEGIES OF MI BY EPMA, SIMS, RAMAN AND LA–ICP–MS The most important difference between the analysis of MI by EPMA, SIMS, Raman, FTIR and LA–ICP–MS is that with all the former techniques only a tiny fraction of the total MI mass can be measured, while LA–ICP–MS also has the capability of analyzing an entire inclusion. Consequently, for all analytical approaches not analyzing the entire inclusion the analyzed mass of sample must be representative of the bulk MI composition. It requires that reversible postentrapment modifications are indeed accurately corrected in the lab prior to analysis. For MI that cannot be homogenized in the lab at entrapment P, T, for example, due to volatile loss as commonly observed in hydrous volcano-plutonic rocks, LA– ICP–MS currently remains as the only microbeam technique for chemical analysis. The same is true for sulfide MI that hardly ever quench to a homogeneous glass (Halter et al. 2002b, 2004a, 2005, Halter & Heinrich 2006). The data-reduction scheme for LA–ICP–MS bulk MI analyses outlined below will demonstrate that multiphase (i.e., partially crystallized) MI can be accurately (Pettke et al. 2004, Halter et al. 2004a) quantified, including the fraction of post-entrapment (sidewall) crystallization.
In summary, accurate LA–ICP–MS measurements of geological materials including multiphase inclusions in minerals should follow the philosophy of keeping all parameters as uniform as possible in order to reduce the potential for complications. This approach requires: • enough laser energy density on the sample to ablate all matrices of interest in a controlled manner, • homogenized energy distribution across the ablation pit, to minimize energy-densitydependent changes in aerosol production (notably aerosol particle size) and to control the ablation process, • laser ablation in He to maximize analytical sensitivity, • robust plasma conditions, to minimize ICP–MS– induced element fractionation, and to provide matrix-independent external calibration, • low and constant gas backgrounds to achieve low LOD, • representative recording of short transient signals, not to (partially) miss trace elements enriched in tiny crystallites or in exsolved fluid bubbles, and • keeping interferences in mind for proper analyte isotope selection, notably matrix-based polyatomic interferences. Obviously, if the analysis of entire MI drilled out of the host mineral is accurate, the
Signal quantification strategies All microbeam techniques are relative analytical methods, i.e., the instrument response needs to be calibrated against standards of known element abundances. This external standardization is best done with matrix matching the sample material; for EPMA it is important and for SIMS it is a prerequisite. Matrix-dependence of analyte signals may also be corrected for by establishing matrix-dependent calibration curves, e.g., as done for the analysis of volatile species by vibrational spectroscopy techniques. FTIR (e.g., Stolper 1982, Newman et al. 1986) analysis achieves LOD for H and C bearing species similar to those obtained by SIMS, at similar spatial resolution (e.g., Hauri et al. 2002), and the analysis of water in silicate glasses of a wide compositional and structural range has become possible by confocal Raman microprobe (Zajacz et al. 2005). The aim of standardization for all these techniques is to determine the sensitivity of 62
IN SITU LASER-ABLATION-ICP-MS ANALYSIS AND CONSTRAINING SUBDUCTION ZONE MAGMATISM
the elements on an external standard (i.e., the amount of signal per unit time and concentration, e.g., counts per second per µg/g) and to analyze the sample at identical instrument conditions. Provided that signal intensity variations are generated exclusively by variable analyte concentrations, element concentrations in the sample are uniquely defined by the use of external standardization alone. For the case where analyte signal intensities change in response to variables other than analyte concentrations, external standardization alone fails to quantify a measurement. Such additional variables may include (a) chemical and structural differences in matrices between external standard and sample, (b) differences in analytical conditions, e.g., variable beam size, between external standard and sample or (c) variability in sample introduction, e.g., nonuniform flushing of the ablation chamber in LA– ICP–MS. If the sum of this additional variability results in a uniform signal intensity change, an intensity shift at constant element sensitivity ratios, independent knowledge of one element concentration in the sample is sufficient to quantify such signals. This element is named the internal standard and is used to determine the relative sensitivity factor, unique for every analytical spot, between external standard and sample. Combined external and internal standardization is vital for the quantification of LA–ICP–MS data (e.g., Longerich et al. 1996a). The condition of variable signal intensities at constant analyte sensitivity ratios just described may not be fulfilled in some cases. For example, the analyte sensitivity ratios can vary with changing matrix chemistry or changing analytical conditions, e.g., beam size or duration of drilling in LA–ICP– MS, or both. For an analytical setup where element sensitivity ratios vary between measurements on the external standard and the sample, the only way of getting accurate data is strictly matched analytical conditions to eliminate such variability. Most commonly, the matrix between external standard and sample is matched. This matrix-matched calibration is essential for EPMA and vital for SIMS. For both techniques, other instrumental parameters are kept strictly uniform between standard and sample analysis. These considerations illustrate the potential problems we encounter for MI analysis. The simplest case is glassy silicate MI exposed on the sample surface. Matrix-matched standardization at uniform beam size is easily ensured for any of the
microbeam techniques. The analysis of multiphase (crystallized) MI is more complex. Different matrices are all analyzed at once, requiring matrixindependent analytical conditions for LA–ICP–MS (i.e., robust plasma conditions). And the entire MI needs to be analyzed in order to reconstitute its chemical composition at the time of trapping. In order to optimize the MI-to-host-mineral mixing proportion during LA–ICP–MS MI analysis, the laser beam size needs to be adjusted for every inclusion, thus requiring beam size-independent external calibration. The analyst is therefore confronted with a large extent of “mismatch” between analytical conditions used for the measurement of external standards and MI samples. This required flexibility in sample-related parameters holds much potential for inaccurate analysis. Therefore, an approach for the accurate analysis of entire MI drilled out of the host mineral using LA–ICP–MS is outlined now. LA–ICP–MS bulk MI analysis The bulk analysis of MI (be they crystallized or glassy) by LA–ICP–MS is schematically shown in Fig. 3-5. Data recording by the ICP–MS is started to acquire the gas background signal. After ~50 s, the laser is switched on and drills first through the host mineral, then hits the inclusion, and the laser is switched off only when the entire MI is consumed. Data recording is stopped. The detailed mathematical formulation for quantification of such signals including a rigorous uncertainty assessment is provided in Halter et al. (2002a); hence, only the conceptual approach behind the quantification strategy is described here. The analytical signal consists of a pure host mineral interval and an inclusion interval representing an a priori unknown mixture between host and inclusion contribution. Quantification of the MI composition thus requires two steps. The two signal intervals, host mineral and inclusion plus host mixture, must be quantified first. Bracketing external standardization (commonly done using SRM61X glasses from NIST) determines the instrument responses for all analytes and corrects linearly for instrument drift. These element responses are then converted to concentration data for the two signal intervals simply by normalizing the element abundances to a fixed element–oxide total (e.g., Leach & Hieftje, 2000), i.e., the sum of all element oxides measured (commonly 100 wt.% minus the amount of non-analyzed volatiles such as H2O). This quantification step is straightforward,
63
IN SITU LASER-ABLATION-ICP-MS ANALYSIS AND CONSTRAINING SUBDUCTION ZONE MAGMATISM
FIG. 3-5: Schematic representation of the LA–ICP–MS analysis of an entire melt inclusion (MI) modified from Halter et al. (2002a), with the resulting signal for a MI in plagioclase shown to the right. The approximate mass consumed or damaged for analysis by EPMA or SIMS is shown in black for comparison.
following the procedure detailed in Longerich et al. (1996a). The second step is then to unmix the inclusion signal from the host signal, i.e., to determine the ratio between the mass of the MI and the total mass of the ablated inclusion interval (Fig. 3-6). This ratio, which is unique for every MI analysis, can be calculated provided that a second constraint is available: (i) The concentration of one element in the melt at the time of inclusion entrapment is known. Ideally, this element is abundant in the melt and scarce in the host. (ii) The concentration ratio of an element pair present in the melt at the time of inclusion entrapment is known. Again, such a ratio in the melt should be very different from that in the host. (iii) The mass ratio between inclusion and host for the signal interval is known independently. This second constraint together with the pure host mineral composition then allows the calculation of the pure MI composition by subtracting the appropriate amount of host mineral
from the mixed signal. Obviously, the accuracy of inclusion composition primarily depends on the quality of this second constraint. Various approaches exist for determining this second constraint for LA–ICP–MS MI quantification. • MI are homogenized in the lab and then measured for their major element composition by EPMA. These results can then be used as an internal standard for the bulk MI analysis and the spot analysis of exposed glassy inclusions. • A series of bulk-rock compositions from a magmatic complex define the range in magma composition with progressive magma evolution. Provided that one element concentration remains essentially invariable throughout, it may be used as an internal standard. In evolved volcanoplutonic centers at convergent margins, bulk Al2O3 can commonly be used (Halter et al. 2002a, Audétat & Pettke, 2003). • This last approach can be refined by comparing bulk rock element correlation plots with the superposed mixing line for the inclusion–host mixed signal interval (Fig. 3-6). By incrementally
64
IN SITU LASER-ABLATION-ICP-MS ANALYSIS AND CONSTRAINING SUBDUCTION ZONE MAGMATISM
FIG. 3-6: Schematic illustration of LA–ICP– MS bulk melt inclusion (MI) signal quantification based on the determination of the mass ratio between inclusion and the total mass ablated in the inclusion signal interval (compare Fig. 3-5). The constraint to calculate the mass ratio between inclusion and total mass ablated in the inclusion interval is defined by the intersection of the trend in FeO/Al2O3 concentration ratio of the MI obtained via incrementally subtracting host mineral component from the mixed signal interval with the trend in whole rock data. See text for explanations.
varying the ratio of inclusion mass divided by total mass ablated for the inclusion signal interval, we obtain a line defined by apparent melt compositions as a function of MI–host mixing proportions. The intersection of this line with the correlation line of the bulk rock data then defines the correct mass ratio for this inclusion signal interval (Halter et al. 2002a), assuming that MI and bulk rock have similar element concentration ratios. • Any internal standard derived from bulk rock element concentrations does not take into account the elements fixed in phenocrysts at the time of MI trapping. Notably for MI trapped at high rock crystallinity, the best way of estimating an element concentration in the original MI is to forward model the chemical evolution of residual melt in the rock relative to the crystallinity estimated for the time of MI entrapment. • Use published element distribution coefficients between host mineral and melt at entrapment conditions, and calculate the element concentration in the melt based on that measured for the host mineral in the same analysis. • Inverse modeling of rock crystallization (starting from matrix-glass data) and of host mineral
crystallization from the MI onto inclusion walls should result in an intersection of trends that defines the element concentrations at the time of MI entrapment (Pettke et al. 2004). • Determine volumetric proportions between MI and the host mineral in the mixed signal interval (Fig. 3-5) optically and convert this to the mass ratio between the two. This last option has turned out to be not precise enough (Halter et al. 2002a); hence, it is not considered further. The above considerations shall serve as a conceptual approach on how to derive an internal standard for the quantification of LA–ICP–MS bulk MI data. What follows is a discussion about advantages, complications and limitations of this bulk MI approach by LA–ICP–MS. This LA–ICP–MS data reduction scheme automatically corrects for the fraction of host mineral crystallized onto the inclusion walls after entrapment. Therefore, it is not necessary to know the amount of post-entrapment crystallization onto the inclusion walls nor is it required to re-melt this rim of crystallized host into the melt prior to bulk MI analysis by LA–ICP–MS. The quality of this internal standard is obviously key to the accuracy of the MI
65
T. PETTKE
composition. First of all, relevant data may not be available, e.g., element distribution coefficients at P and T for the magma of interest, or data bases for modeling the reverse of host mineral crystallization out of the trapped MI may not be adequate for some host minerals (e.g., Kress & Ghiorso, 2004). Pettke et al. (2004) have mathematically explored the effects of inappropriate internal standardization for LA–ICP–MS MI analyses. An apparently simple sample of glassy MI from a dredged MORB sample (Fig. 3-2B) was used for this purpose. The various scenarios presented in Pettke et al. (2004) illustrate well that MI concentrations may be overestimated or underestimated, and it is intuitive that inaccurate host mineral subtraction affects elements compatible in the host less than those incompatible in the host. The variability of results exceeded 20% (e.g., 7.68 < MgO < 9.42) even for the petrographically pristine MI in a fresh, dredged MORB sample (Fig. 3-2B) containing less than ca. 15% phenocrysts. Impressive for this case is the degree of post-entrapment crystallization onto the inclusion walls – it amounted to 11 wt.% on average. It is fair to state that for the complex case of MI from strongly evolving systems, the danger for inappropriate internal standardization needs to be taken seriously. The most robust check on the possibility of inappropriate standardization is to apply a variety of internal standards for MI data reduction, to explore their effects on the final result.
spots on individual inclusions). Heterogeneities within petrographic MI assemblages can be explored by highly precise spot analyses of several inclusions from one assemblage in order to constrain MI entrapment processes (e.g., equilibrium vs. disequilibrium entrapment; Faure & Schiano 2005). Accuracy at useful analytical precision is required to constrain the chemical composition of source melts. Such data obtained on MI of known entrapment history may then be used to trace the chemical evolution of the residual melt in magmatic systems (e.g., Halter et al. 2004b) or to investigate element distribution between coexisting phases such as fluid and melt (e.g., Audétat & Pettke 2003). As outlined above and illustrated in Fig. 3-1, the petrographically most reliable definition of MI entrapment sequence is the geometric arrangement of MI assemblages in rock-forming crystals. Average MI assemblage compositions are thus perfectly suited for studying igneous processes. Consequently, the time frame and scale of investigation defines what type of data is ideally used to attack the problems. The analytical uncertainties associated with data sets obtained by the various microbeam techniques are addressed now. First, it should be appreciated that uncertainties for trace element concentrations of individual spot analyses near the machine-specific LOD are dominated by the uncertainty in counting statistics for all instruments unless instrument backgrounds are large. These uncertainties are easily 50% or more and provide a minimum uncertainty for each element analysis in individual inclusions. Error propagation for LA– ICP–MS analysis of individual MI includes analytical uncertainties on the host mineral and the mixed signal interval (including counting statistics and plasma flicker), and the uncertainty on the extrapolation of MI composition from signal deconvolution of the mixed signal interval (Halter et al. 2002a). To my knowledge, a similar propagation of analytical uncertainties is not available for individual spot analyses by EPMA and SIMS. Consequently, the uncertainties on LA–ICP– MS analyses of individual MI allow one to identify “bad” analyses, for example those suffering from dominant host contribution to the mixed signal, and discard them (Halter et al. 2002a). For the case of analyzing entire MI by LA– ICP–MS, additional uncertainty on the MI composition stems from subtraction of the host mineral contribution to the mixed signal. Therefore,
STATISTICAL RELEVANCE OF DATA SETS GENERATED FROM LA–ICP–MS, SIMS AND EPMA ANALYSIS The type of geochemical problem to be solved defines the required precision of the analytical data set. Two basic types of data sets need clear distinction. Data for individual inclusions with associated analytical uncertainty may either represent a single spot analysis (internal precision; where available) or averaged multiple spot analyses obtained on a single inclusion (external precision). Alternatively, averages may be calculated from the analysis of several individual inclusions belonging to a MI assemblage. The resulting inclusion-toinclusion reproducibility should not exceed analytical scatter as indicated by mean square weighted deviates (MSWD) values for MI assemblages (Pettke et al. 2004, and further below). The best possible analytical precision at high spatial resolution for individual spot measurements of MI is required to check for homogeneity of reheated inclusions (using multiple
66
IN SITU LASER-ABLATION-ICP-MS ANALYSIS AND CONSTRAINING SUBDUCTION ZONE MAGMATISM
LA–ICP–MS data of unexposed MI are less precise than spot analyses of exposed MI by SIMS, EPMA and LA–ICP–MS itself. This additional uncertainty is larger for chemically similar MI-host mineral pairs (e.g., andesitic MI in amphibole; Halter et al. 2002a) or, generally, for trace elements enriched in the host, as nicely illustrated for Sr in plagioclasehosted MI (Halter et al. 2002a) or Zr in MI hosted by two chemically distinct clinopyroxene crystals from one rock sample (Pettke et al. 2004). Typical uncertainties for individual element concentrations in a bulk LA–ICP–MS MI analysis are between a few percent for elements incompatible in the host and may be as high as a few tens of percent for those dominantly present in the host. Comprehensive data sets including uncertainties of individual bulk MI measurements by LA–ICP–MS are provided elsewhere (e.g., Halter et al. 2002a, Pettke et al. 2004, Halter et al. 2004b). Obviously, the analyst has to strive for maximal inclusion to host mass ratios for the signal interval to minimize uncertainties on MI compositions. Hence, the ablation pit has to be large enough to ablate the entire MI plus crystallized rim of host after entrapment, but as small as possible to minimize the host contribution to the signal interval (Figs. 3-5,6). Therefore, analytical uncertainties on individual bulk MI measurements by LA–ICP–MS need to be quantified, and the detailed formalism to do so is provided in Halter et al. (2002a). Having quantified uncertainties on individual MI measurements then allows us to calculate uncertainty-weighted, average element concentrations for MI assemblages, following the concept that the analyzed MI represent individual samples of a chemically uniform melt at the time of their entrapment (i.e., homogeneous assemblages). This is advantageous because precise analyses exert a larger influence on the resulting average element concentrations than do imprecise ones. This is especially important for element concentrations near their LOD. Moreover, MSWD values can be obtained for uncertainty-weighted averages (commonly employed in isochron dating; e.g., Ludwig et al. 1994), serving as a test whether the variability in averaged data sets can be explained by analytical uncertainty alone. If not, initial heterogeneity within the MI assemblage is indicated (illustrated in Pettke et al. 2004, and further below). As there is no generally accepted way of determining element concentration uncertainties on a single spot analysis by SIMS or EPMA, simple multi-spot averages with standard deviation
uncertainties are used to assess analytical precision (i.e., the external spot-to-spot reproducibility is used instead of the single spot uncertainty-weighted average as available in LA–ICP–MS bulk MI analysis). This can be done by either averaging multiple spot analyses obtained from a single MI, or by averaging spot analyses from a series of MI of one assemblage. This procedure weights each analytical point equally. A comparison between average MI element concentrations obtained by different analytical methods is shown in Figure 3-7 (data from table 3 in Pettke et al. 2004). It can be seen that average concentrations obtained on exposed MI by spot analysis (SIMS and EPMA) largely agree with those obtained on crystallized MI analyzed in bulk by LA–ICP–MS, and that one standard deviation uncertainties are similar for the different analytical techniques employed (Fig. 3-7A). The calculation of uncertainty-weighted average element concentration for the same LA– ICP–MS data set changes the numbers only slightly but results in considerably smaller uncertainties. Comparison of these data with average SIMS and EPMA data (Fig. 3-7B) shows less agreement than between the averaged data shown in Fig. 3-7A. Elements that disagree generally show high MSWD values, indicating sample variability in addition to analytical uncertainty. This suggests that MI from different assemblages with evolving element concentrations were averaged in the current data set. Because different MI were analyzed by the various techniques, the respective average element concentrations of elements present at variable abundances will not overlap. Consequently, uncertainties on uncertainty-weighted average element concentrations will in this case underestimate the variability of the data set. Uncertainty-weighted average compositions of MI assemblages determined by LA–ICP– MS are therefore considered to be a very robust determination of the melt chemistry present at various stages during magma evolution. It infers that LA–ICP–MS analysis of entire crystallized MI from assemblages constrains the assemblage bulk composition with analytical precisions (appreciating the statistical differences inherent in the various data sets) similar to those obtained for EPMA and SIMS averages from exposed glassy MI. STRENGTHS OF THE VARIOUS MICROBEAM TECHNIQUES The following section about key competences of each analytical setup shall illustrate
67
IN SITU LASER-ABLATION-ICP-MS ANALYSIS AND CONSTRAINING SUBDUCTION ZONE MAGMATISM
FIG. 3-7: Comparison between average element concentrations determined by secondary ion mass spectrometry (SIMS) or electron probe microanalysis (EPMA) and LA–ICP–MS bulk melt inclusion (MI) data. The MI were hosted in clinopyroxene of a sample from Mt. Somma Vesuvius. Data for each individual inclusion are reported in Pettke et al. (2004). Data are expressed in wt.% element oxides or in µg/g for trace elements, and plotted linearly, scaled with the preceding factor as labeled. Figure A compares simple averages obtained for both data sets. Error bars are 1 standard deviation and show that uncertainties in average element concentrations of these MI are comparable between the analytical methods. Elements in gray do not agree within 1SD uncertainty between the methods (the reasons for disagreement are discussed in Pettke et al. 2004). Figure B shows the comparison between uncertainty-weighted average MI composition obtained by bulk MI LA–ICP–MS and average SIMS or EPMA data. Again, elements that disagree between the methods are shown in grey. Most of these elements show high MSWD values (reported in table 3 of Pettke et al. 2004), indicating that variability in individual MI element concentrations exceeds pure analytical variability. This would imply that MI from different assemblages are averaged in the current data set.
instrument because matrix-based O2+ interferences cannot be resolved – high-resolution sector field instruments would be required for this (e.g., Evans et al. 2001, Lahaye et al. 2004). Because of poor ionization efficiency and elevated background signals, C and Cl may not give useful LOD values in MI. Provided that inclusions can be homogenized reliably in the lab, data for the above elements can be obtained by SIMS and even EPMA with much better detection power. Volatiles in MI are extremely important because their abundance in melts strongly affects magma dynamics and equilibrium crystallization,
their complementary character. Therefore, the geochemical questions to be attacked will dictate which instrument(s) are ideally employed for analysis. It is well known that LA–ICP–MS cannot analyze all elements of the periodic system – the notoriously difficult if not impossible elements are those in the upper right corner of the periodic table. The impossible elements are those with a first positive ionization potential higher than that of Ar (e.g., F). Volatiles such as H, N and O are currently not possible either. Sulfur in silicates, carbonates or phosphates, cannot be analyzed on a quadrupole 68
IN SITU LASER-ABLATION-ICP-MS ANALYSIS AND CONSTRAINING SUBDUCTION ZONE MAGMATISM
and MI are likely to preserve the least modified volatile content of the original melt (e.g., Metrich & Clocchiatti 1989, Grove et al. 2002, Wallace 2005). Water is among the most important volatile species, explaining why much effort has been put in its accurate analysis in MI. Applications of FTIR to the analysis of volatile species in MI has been essential in advancing our understanding of volatile contents in melts (e.g., Lowenstern & Mahood 1991, Wallace et al. 1999). The main disadvantage of this technique is the need for exposing MI glasses on both sides for measurement by doubly polishing while maintaining sufficient thickness to perform reliable absorbance measurements. SIMS may be used as an alternative analytical technique to FTIR for the analysis of some volatiles, and Layne (2006) addresses this topic. Notably thanks to the introduction of the use of a Cs+ ion source, allowing for the analysis of negative ions, such volatile elements can now be detected down to µg/g concentrations (e.g., Hauri et al. 2002). For the analysis of water in silicate glasses at concentrations of ca. 2 wt.% or more, the use of Raman spectroscopy has been explored (e.g., Thomas, 2000). Recently, a novel quantification approach for confocal Raman water measurements has been developed, allowing fast measurements of inclusions of variable composition entirely enclosed in the host mineral (Zajacz et al. 2005), at analytical precisions inferior to those obtained by SIMS, however. The fact that MI need not be exposed to the surface for confocal Raman analysis minimizes problems of contamination or exchange between MI glass and ambient air; hence, this technique has great potential for constraining melt water contents particularly when contained in MI. Moreover, as this approach is truly non-destructive, the MI are preserved for analysis by other microbeam techniques. Analytical resolution is inferior with LA– ICP–MS when compared to the other microbeam techniques. Subtle, small-scale trends in element concentrations are best resolved by EPMA or SIMS, notably for small (<20 µm) inclusions. Such data may be required for example to test for boundary layer problems or diffusional equilibration between the MI and host mineral, i.e., problems where we need precise data at a resolution higher than that achieved with bulk MI LA–ICP–MS (a recent example is provided in Faure & Schiano 2005). A small beam size also allows for multiple spot analyses on one MI to check for homogeneity across the inclusion (serving as a test for the
effectivity of homogenization in the lab). SIMS analysis achieves LOD comparable to those obtained with LA–ICP–MS, but consumes much smaller sample amounts, and so SIMS may be the method of choice for small MI. And small MI may be important in constraining mantle melt characteristics (compare Danyushevsky et al. 2004). Especially with the advent of nano-SIMS, we may soon have a technique with an analytical resolution hitherto unavailable to the earth sciences (e.g., Stadermann et al. 2005, Hellebrand et al. 2005). High resolution chemical imaging can be done with BSE imaging as available in electron probe instruments (e.g., Faure & Schiano 2005), an excellent tool to identify and visualize small scale heterogeneities such as diffusional zoning patterns. SIMS also offers the capability to generate reasonable resolution chemical maps. Such element distribution maps may then be used for designing the next analytical steps. For EPMA and SIMS analysis, the MI must be exposed at the sample surface. For LA– ICP–MS and confocal Raman analysis, the inclusions may be completely enclosed in the host mineral; hence, many more MI are actually available for measurement in one analytical session (i.e., at one set of instrument optimization). The analysis of all MI at set experimental conditions is essential for precisely constraining average element concentrations for MI assemblages. LA–ICP–MS is a destructive method, consuming the entire inclusion for bulk MI analysis and thus rendering a revisit impossible. EPMA and SIMS consume or damage much less sample (Fig. 3-5). When diverse analytical techniques are combined, LA–ICP–MS will be done in spot mode on exposed MI. Also in this mode, LA–ICP–MS consumes most sample material. On the other hand, LA–ICP–MS is the only method allowing for analysis of major and trace elements within one analysis (thanks to the dynamic range of ICP–MS detectors of up to 9 orders of magnitude). Consequently, the geologic problem to be solved again determines which instrumental setup is most favorable. Matrix-matched external standardization at similar concentration levels, which is mandatory for SIMS and recommended for EPMA, is not required for LA–ICP–MS systems operated under the robust plasma conditions detailed above. Therefore, widely used and well-characterized reference materials can be employed for data quantification, improving the comparability of data generated in different labs and
69
T. PETTKE
thus, ultimately, the reliability of results. The most efficient tool for determining general chemical characteristics of a series of MI assemblages is LA–ICP–MS, due to the large number of bulk MI analyses obtainable within one analytical session (close to 100 MI plus adjacent host at 20–50 major to trace elements per day). Large data sets from MI assemblages also allow identification and thus exclusion of nonrepresentative MI from further data calculations. Such MI commonly include those having accidentally trapped a solid phase or inclusions of poor analytical quality, e.g., those stemming from non-representative recording of highly transient signals produced from tiny daughter crystals. Robust uncertainty-weighted average compositions of assemblages can thus be obtained from the large LA–ICP–MS data sets. Moreover, each host mineral is measured adjacent to the MI within one analytical shot during identical LA–ICP–MS analytical conditions. This provides a check for host mineral homogeneity (recall that host mineral zoning can be used to identify proper primary MI assemblages), and it allows for the direct determination of element distribution coefficients for major and trace elements alike (recalling limitations imposed by post-entrapment diffusive equilibration). The above comparison between advantages of the various microbeam techniques for the analysis of MI again emphasizes that the analytical method of choice is dictated by the geochemical problem to be solved. For cases where homogenization of MI to entrapment composition is not possible in the lab, LA–ICP–MS provides the only technique for bulk chemical analysis. In any case, LA–ICP–MS MI analysis is cost and time efficient; hence, it could be performed on a series of assemblages in order to constrain the overall evolution of the igneous system and to identify critical assemblages on which to obtain specific analyses by other techniques.
results imply that the signal deconvolution of the mixed inclusion interval is correct and document that crystallized bulk MI can be analyzed accurately by LA–ICP–MS. Analytical precision for average compositions of MI assemblages is comparable between EPMA, SIMS and LA–ICP–MS, and analytical uncertainties can be as good as a few percent at the 1 standard deviation level. Analyses for individual inclusions are less precise for LA– ICP–MS except for large exposed MI analyzed in single spot mode. Here, SIMS and LA–ICP–MS data are of comparable analytical precision. Carefully acquired data by these microbeam techniques are analytically accurate at these precisions. It is now essential to assess how “geologically correct” such microbeam data are, and to identify potential differences in “correctness” among the various microbeam techniques. A large non-quantifiable source of uncertainty in microbeam data other than those obtained by bulk MI LA–ICP–MS is the amount of host mineral to be remelted into the inclusion during homogenization in the lab prior to analysis. LA– ICP–MS bulk MI analysis does not require this homogenization step because the correct internal standard (itself the limiting parameter for bulk MI LA–ICP–MS analysis; see below) does account for post-entrapment crystallization of host mineral on to inclusion walls. LA–ICP–MS analysis of unexposed MI should therefore be considered as a check on the correctness of host remelting for the analysis of MI by EPMA and SIMS, by using an element concentration or concentration ratio predicted from independent constraints as the internal standard for LA–ICP–MS data reduction. The accuracy of LA–ICP–MS data most strongly depends on the quality of the internal standard used for data reduction, irrespective of whether the analysis is done as a spot on exposed MI (e.g., Taylor et al. 1997, de Hoog et al. 2001, Kamenetsky et al. 2002, Danyushevsky et al. 2000) or by drilling the entire inclusion out of the host mineral (e.g., Audétat et al. 2000, Halter et al. 2002a,b, Audétat & Pettke 2003). For cases where MI can be reliably homogenized in the lab, the internal standard element concentration may have been determined by EPMA or SIMS; hence any inaccuracy of the values obtained by EPMA or SIMS (e.g., due to inappropriate host mineral remelting) translates directly to the LA–ICP–MS data quantification. Inverse modeling may also be used to predict initial melt compositions from
POTENTIAL INACCURACIES FOR IN SITU MICROBEAM DATA The analytical accuracy of MI data obtained from the various techniques has been repeatedly shown elsewhere. Pettke et al. (2004) demonstrated that MI data on crystallized and reheated MI acquired with EPMA, SIMS and LA– ICP–MS overlap within their uncertainties (see also Fig. 3-7), and Halter et al. (2004a) showed that bulk MI analyzed from coexisting clinopyroxene and plagioclase have identical compositions. These
70
IN SITU LASER-ABLATION-ICP-MS ANALYSIS AND CONSTRAINING SUBDUCTION ZONE MAGMATISM
analytically accurate microbeam analyses; hence, limitations imposed by the modeling affect the accuracy of all microbeam data equally. The sum of the above uncertainties is probably larger than the mere analytical uncertainty irrespective of which analytical technique was used. The overall uncertainty on average element concentrations of MI assemblages as determined by bulk MI LA– ICP–MS analysis is conservatively estimated to be about 10% for elements enriched in the melt. The possibility to analyze heterogeneous MI that cannot be homogenized in the lab as required for all the other microbeam techniques renders the LA–ICP–MS bulk MI approach so promising. The fact that diffusive loss of H2O and H+ are likely to be the dominant cause for the impossibility of homogenizing some MI in the lab at entrapment temperatures renders it possible that many trace element concentrations in such MI have remained essentially unmodified by diffusive processes. This circumstance alludes to the potential of the new technique of LA–ICP–MS bulk MI analysis. Data may now be obtained from MI hitherto inaccessible by microbeam techniques, allowing the study of additional types of MI that are commonly abundant in hydrous magmatic systems ubiquitous in subduction zone settings. It is foreseen that such new data will provide novel constraints on sources and processes in the genesis of hydrous magmas in diverse geotectonic settings.
coexisting melt and fluid phases at the magmatichydrothermal interface using the bulk inclusion approach (e.g., Simon et al. 2005, Hanley et al. 2005a) have demonstrated significant mobility of Au and Pt in variably salty fluids. The Farallón Negro volcano-plutonic complex in Argentina is the only published example where the chemical evolution of the shallow level magma chamber has been reconstructed by using LA–ICP–MS bulk MI data. The comprehensive data set demonstrates, inter alia, that the magma formed by mixing of basaltic magma into resident dacitic magma, and that the mixed magma lost its Cu almost quantitatively to an exsolving aqueous fluid phase, which formed the world class Bajo de la Alumbrera porphyry Cu±Au deposit (Halter et al. 2002a,b: 2004b, 2005). So far, there have been only sparse first efforts towards constraining deeper processes including melt–melt element partitioning in deeper magma chambers (Kamenetsky 2006; Halter & Heinrich 2006). Igneous rocks above subduction zones provide compelling chemical evidence that slab components ascend into the hot zone of the mantle wedge and induce partial melting (e.g., Ulmer 2001). During buoyancy-driven ascent, these magmas react with various sources, including the lower arc crust, and magma mixing has been identified as an important process in the formation of andesitic magmas. These magmas may reside in shallow interconnected reservoirs, evolve and feed volcano-plutonic centers, some of which host the largest magmatic-hydrothermal Cu–Mo–Au ore deposits (e.g., Bingham and Pinatubo: Hattori & Keith 2001; the central Andes: Kay & Mpodozis 2001). The thorough chemical characterization of the resultant magmatic rocks was used in various ways to extrapolate towards the processes that formed them and to constrain the nature of source components from the arc crust, the mantle wedge and the subducted slab (e.g., Perfit et al. 1980, Tatsumi 1989, Morris et al. 1990, McGulloch & Gamble 1991, Hawkesworth et al. 1993, Plank & Langmuir 1993, and many subsequent publications). Most of the data are from lavas that underwent shallow level fractionation, crystallization, magma mixing and degassing. Original magma characteristics are blurred by these shallow level processes, however. It is thus not surprising that the various source components and their relative importance for the generation of arc magmas have been hotly debated for more than two decades, and
PROSPECTS FOR CONSTRAINING SUBDUCTION ZONE MAGMATISM The final part of this chapter tries to illustrate the potential applications of bulk MI analysis by LA–ICP–MS to constrain deep magma processes in subduction zones. To date, it is not possible to review a comprehensive set of results, simply because applications using the LA–ICP–MS bulk inclusion technique in these environments are only emerging. The unique value of LA–ICP–MS bulk inclusion analysis for investigating the chemical composition of shallow, fluid-saturated melts has been demonstrated. Table 3-1 provides a collection of contributions published so far, demonstrating that the research emphasis was on element distribution between co-existing phases in shallow volcano-plutonic complexes and shallow granites (silicate melts and aqueous fluids, including brine and vapor, and halide melts), and to constrain the mobility and enrichment processes of ore-forming metals. First attempts to constrain experimentally the metal distribution between
71
T. PETTKE
TABLE 3-1: TECHNIQUES DEVELOPMENT AND APPLICATIONS OF THE BULK MI ANALYSIS BY LA-ICP-MS Development
Application
Comments
Reference
First quantitative analysis of entire crystallized silicate MI First analysis of entire silicate MI in LA–ICP–MS DRC mode
Fluid–melt element distribution at magmatichydrothermal stage Improvement in LOD for Ca and Fe using Dynamic Reaction Cell (DRC) technology Chemical quantification of entire MI drilled out of host minerals of any chemical composition Estimation of Cu/Au ratios in source magma of porphyrytype ore deposits Fluid – melt element distribution in highly fractionated miarolitic cavities Data comparison between bulk MI (crystallized or reheated) by LA–ICP–MS, and spot analysis by EPMA and SIMS MI data from co-existing plagioclase and pyroxene identical
Quantification using EPMA data of homogenized MI Constant Al concentration from EPMA used for quantification Individual MI compositional data with associated analytical uncertainties Calibration of sulfide MI using silicate glass SRM610 from NIST Mg/Al concentration ratio used to quantify MI
Audétat et al. 2000
Evaluation of potential inaccuracies in bulk MI LA–ICP–MS data sets
Pettke et al. 2004
Accurate LA–ICP–MS sulfide Fe, Co, Ni, Cu data by standardization on SRM610 from NIST Magma mixing, not evident from bulk-rock data, demonstrated using MI analyses
Halter et al. 2004a
Combined with EPMA and Raman data to constrain S speciation and fO2 Identification of sinks for Cu and Au in andesitic magma Halide melts distinguished from halite inclusions by trace element inventories
Audétat et al. 2004
General case of mathematical signal quantification strategy for entire MI First quantitative analysis of entire sulfide MI Data for highly fractionated MI from barren intrusions
Demonstration of accuracy of bulk LA–ICP–MS MI analysis
Demonstration of accuracy of bulk LA–ICP–MS MI analysis First comprehensive study of the life-time evolution of a supra-subduction zone magmatic complex Analysis of heterogeneous melt and mineral inclusions
Cu–Au distribution between co-existing sulfide and silicate MI Chemical characterization of halide MI
Quantify chemical characteristics of magma of the ore-forming Farallón Negro Volcano-Plutonic complex Constrain magma characteristics prior to porphyry-type ore formation Constrain source magma processes relevant for porphyry-type ore formation Constraining the genesis of PGE mineralization in quartz veins of the Sudbury Igneous complex
that geochemical arguments appeared to be sometimes at odds with geophysical constraints. Only recently, the value of MI to explore characteristics of deep subduction zone magmas has become broadly appreciated (e.g., Lowenstern
Günther et al. 2001
Halter et al. 2002a
Halter et al. 2002b Audétat & Pettke, 2003
Halter et al. 2004b
Halter et al. 2005 Hanley et al. 2005b
1995, Sobolev 1996, Kamenetsky et al. 1997, de Hoog et al. 2001, Grove et al. 2002, Schiano et al. 2004a, Wallace 2005), notably because it could be demonstrated that MI preserve more closely the volatile budget of melts present early in the 72
IN SITU LASER-ABLATION-ICP-MS ANALYSIS AND CONSTRAINING SUBDUCTION ZONE MAGMATISM
evolution of eruptive rocks. In other words, the MI represent a more pristine sample than the bulk rock provides. Accordingly, MI trapped in early crystallized minerals better approach pristine samples of mantle-derived melts before shallow level processes disturb their characteristics. This increasing acceptance of MI data currently stimulates research efforts including urgently needed investigations on how MI form in mantle minerals (e.g., Faure & Schiano 2005) and on the fundamental issues of post-entrapment modifications of bulk MI compositions. Such data are prerequisite for the proper application of MI to the study of subduction zone processes in particular, because here, the mobile phase(s) is (are) a driving force in this dynamic environment. Water contents as high as 10 wt.% in MI from high-Mg andesites have been reported (e.g., Anderson 1974, Sisson & Layne 1993, Grove et al. 2002, Straub & Layne 2003). Such high water contents are not known from island arc rocks, revealing that these magmas degas significantly before solidification or eruption. A strong link between water and elements typically enriched in subduction zone magmas such as large ion lithophile elements (LILE) and light rare earth elements (LREE; e.g., Sobolev & Danyushevsky 1994, Stolper & Newman 1994, Grove et al. 2002, Cervantes & Wallace 2003) demonstrate that these elements are enriched in the mobile phase ascending from the slab. This hypothesis has been corroborated by experiment (e.g., Brenan et al. 1995, Keppler 1996, Johnson & Plank 1999), most recently for basalt equilibrated with water at pressures of 4–6 GPa, corresponding to slab depth encountered beneath arcs (Kessel et al. 2005a,b). This mobile component, irrespective of whether it is an aqueous fluid or a hydrous melt, or a supercritical liquid beyond the second critical end point at 5–6 GPa, produced a multiphase quench product trapped in a diamond trap. This diamond trap in the experimental capsule resembles “a very large MI” that can be quantitatively analyzed at ambient pressure conditions only by cryogenic LA–ICP–MS (Kessel et al. 2004). The various mobile components are relatively enriched in LILE and LREE, and element concentrations generally increase with increasing pressure. Supercritical liquids are essentially identical in trace element signatures with hydrous melts existing at lower pressure (Kessel et al. 2005a), implying that “established” trace element fingerprints for slab melting are no longer unique in their interpretation. Such “slab melt
signatures” may thus indicate that the slab component corresponds to a supercritical liquid of potentially highly variable water contents, notably in mature and fast subduction zone settings. This also signifies relaxing of the apparent temperature discrepancy between geochemical arguments and geophysical models for the temperature regimes in slabs beneath island arcs (Kessel et al. 2005a). MI provide the most promising approach to explore further the least modified slab component interpreted to be inherently water-rich and present in “undegassed” magma, allowing a test of how generally applicable these experimental data obtained on a Cl- and S-free system (Kessel et al. 2005a,b) are in nature. Notably the effect of chlorinity may be profound in shifting some of these slab element signatures, and there is evidence for highly variable chlorinity in slab components (e.g., Kent et al. 2002; summarized in Wallace 2005). MI having trapped such undegassed magma will also provide the closest approach to pristine concentrations of mobile trace elements in general, e.g., Li, B, Cu, Pb or halogens to name only a few. The mere occurrence of some of these elements in volcanic exhalatives (e.g., Nho et al. 1996) suggests that their primary abundance in degassed magma may be severely reduced. Consequently, MI trapped early in magma genesis at subduction zones may represent, most closely, the fertile component that renders subduction zone magmas so distinct; hence, it is these MI that constrain the slab input to the magma most reliably. But it is exactly these MI that are most prone to diffusive loss of volatiles during prolonged residence at depth, notably water, rendering their proper homogenization impossible – if recognized as inclusions in primitive samples, e.g., in forsterite-rich olivine, LA–ICP–MS is currently the only tool to constrain their chemical composition including most of the mobile elements. Careful geological interpretation of such data will significantly advance our understanding of the slab contribution to magma formation at convergent margins. Another window on chemical characteristics of deep subduction zone magmas is provided by MI from xenoliths or exposed root zones of volcano-plutonic systems including cumulate rocks. Here, MI homogenization is generally difficult because of the extended residence time of such samples at high temperature and pressure conditions, enhancing diffusive equilibration. The investigation of MI in such samples (e.g., Cortini et al. 1985, Debari & Sleep 1991, Schiano et al. 1995,
73
T. PETTKE
De Vivo et al. 1995, Frezzotti 2001, Spandler et al. 2003, Schiano et al. 2004b) has already provided fundamental constraints on igneous processes in subduction zones. LA–ICP–MS analyses of heterogeneous inclusions will provide novel chemical data on these deep systems, and the comparison with MI data obtained on Fo-rich olivine from mafic lavas will be most interesting. Accepting the difficulty of deriving a reliable internal standard for MI quantification, LA– ICP–MS data will nevertheless provide unprecedented information on such (originally) volatile-rich MI, notably for MI trapped in olivine or orthopyroxene. Here, the host mineral is chemically “simple”, because most trace elements of interest are present at very low concentrations only. Therefore, removing inappropriate amounts of host mineral from the mixed inclusion signal interval will change the abundances of such trace elements in the MI by the same factor, i.e., the trace element ratios and, thus, normalized trace element patterns (e.g., REE spectra, spider diagrams) of the MI will remain the same. Such high precision trace element signatures do not require internal standardization at all, they are uniquely defined by the use of external standardization only in LA–ICP– MS analysis. Many supra-subduction igneous processes are constrained by using element abundance ratios alone, e.g., Sr/Y, U/Th, Ba/La or B/Be, or isotopic ratios such as 10Be/11Be (e.g., Morris et al. 1990, McCulloch & Gamble 1991, Plank & Langmuir 1993, Hawkesworth et al. 1993, Johnson & Plank 1999), and bulk MI LA–ICP–MS will significantly increase this data set on hydrous igneous systems. Evidently, LA–ICP–MS bulk MI data from inclusions that cannot be homogenized reliably in the lab hold great potential for further constraining the signature of the (originally) most volatile-rich components in subduction zone magma systems, thus potentially shrinking the “black box” between what is liberated from the slab as known from high pressure experiments and what can be sampled in primitive arc magma end members.
reviews by Terry Plank and Leonid Danyushevsky, and the great editorial help by Jim Webster, substantially improving the present contribution. Constructive suggestions on an early version of this manuscript by O. Müntener and János Kodolányi are appreciated. I gratefully acknowledge funding from the Swiss National Science Foundation. REFERENCES AESCHLIMAN, D.B., BAJIC, S.J., BALDWIN, D.P. & HOUK, R.S. (2003): High-speed digital photographic study of an inductively coupled plasma during laser ablation: comparison of dried solution aerosols from a microconcentric nebulizer and solid particles from laser ablation. J. Anal. Atom. Spectrom. 18(9), 1008–1014. ANDERSON, A.T. (1974): Evidence for a picritic, volatile-rich magma beneath Mt Shasta, California. J. Petrol. 15(2), 243–267. AUDÉTAT, A. & PETTKE, T. (2003): The magmatichydrothermal evolution of two barren granites: A melt and fluid inclusion study of the Rito del Medio and Canada Pinabete plutons in northern New Mexico (USA). Geochmim. Cosmochim. Acta 67(1), 97–121. AUDÉTAT, A., PETTKE, T. & DOLEJS, D. (2004): Magmatic anhydrite and calcite in the oreforming quartz- monzodiorite magma at Santa Rita, New Mexico (USA): genetic constraints on porphyry-Cu mineralization. Lithos 72(3–4), 147–161. AUDÉTAT, A., GÜNTHER, D. & HEINRICH, C.A. (2000): Magmatic-hydrothermal evolution in a fractionating granite: A microchemical study of the Sn–W–F-mineralized Mole Granite (Australia). Geochmim. Cosmochim. Acta 64(19), 3373–3393. BODNAR, R.J. & STUDENT, J.J. (2006): Melt inclusions in plutonic rocks: petrography and microthermometry. In Melt Inclusions in Plutonic Rocks (J.D. Webster, ed.) Min. Assoc. Can. Short Course 36, 1-25.
ACKNOWLEDGEMENTS My experience in the analysis of MI by LA–ICP–MS has greatly benefited from the various research projects in the group of Chris Heinrich at ETH Zurich, notably from the close collaboration with Werner Halter, and from collaborative work with Andreas Audétat, Adam Simon and Jacob Hanley. I am very grateful for the thoughtful
BOGAERTS, A. & CHEN, Z.Y. (2005): Effect of laser parameters on laser ablation and laser-induced plasma formation: A numerical modelling investigation. Spectrochim. Acta Part B–At. Spectrosc. 60(9–10), 1280–1307. BORISOV, O.V., MAO, X.L. & RUSSO, R.E. (2000): Effects of crater development on fractionation 74
IN SITU LASER-ABLATION-ICP-MS ANALYSIS AND CONSTRAINING SUBDUCTION ZONE MAGMATISM
and signal intensity during laser ablation inductively coupled plasma mass spectrometry. Spectrochim. Acta Part B–At. Spectrosc. 55(11), 1693–1704.
Arc, Alaska – implications for primary magmas and the nature of arc crust. Geol. Soc. Amer. Bull. 103(1), 37–47. DE HOOG, J.C.M., MASON, P.R.D. & VAN BERGEN,
BRENAN, J.M., SHAW, H.F., RYERSON, F.J. & PHINNEY, D.L. (1995): Mineral-aqueous fluid partitioning of trace elements at 900 degrees C and 2.0 GPa: Constraints on the trace element chemistry of mantle and deep crustal fluids. Geochim. Cosmochim. Acta 59(16), 3331–3350.
M.J. (2001): Sulfur and chalcophile elements in subduction zones: Constraints from a laser ablation ICP–MS study of melt inclusions from Galunggung Volcano, Indonesia. Geochmim. Cosmochim. Acta 65(18), 3147–3164. DE VIVO, B., TOROK, K., AYUSO, R.A., LIMA, A. & LIRER, L. (1995): Fluid inclusion evidence for magmatic silicate/saline/CO2 immiscibility and geochemistry of alkaline xenoliths from Ventotene Island, Italy. Geochmim. Cosmochim. Acta 59(14), 2941–2953.
CERVANTES, P. & WALLACE, P.J. (2003): Role of H2O in subduction-zone magmatism: New insights from melt inclusions in high-Mg basalts from central Mexico. Geology 31(3), 235–238. CLOCCHIATTI, R. (1977): Melt inclusions in olivine, plagioclase and pyroxene phenocrystals as samples of magmatic liquid during host mineral crystallization – application to a low K basalt of emerged mid-oceanic ridge (Asal, Tfai). Comptes Rendus Hebdomadaires Des Seances De L' Academie Des Sciences Serie D 284(22), 2203.
EGGINS, S., DE DECKKER, P. & MARSHALL, J. (2003): Mg/Ca variation in planktonic foraminafera tests: implications for reconstructing palaeoseawater temperature and habitat migration. Earth Planet. Sci. Lett. 212(3–4), 291–306. EGGINS, S., KINSLEY, L.P.J. & SHELLEY, J.M.G. (1998): Deposition and element fractionation processes during atmospheric pressure laser sampling for analysis by ICP–MS. Appl. Surf. Sci. 129, 278–286.
CORTINI, M., LIMA, A. & DE VIVO, B. (1985): Trapping temperatures of melt inclusions from ejected Vesuvian mafic xenoliths. J. Volcanol. Geotherm. Res. 26(1–2), 167–172. COTTRELL, E., SPIEGELMAN, M. & LANGMUIR, C.H. (2002): Consequences of diffusive reequilibration for the interpretation of melt inclusions. Geochem. Geophys. Geosys. 3, art. no. 1026.
EVANS P., WOLFF-BRICHE, C. & FAIRMAN, B. (2001): High accuracy analysis of low level sulfur in diesel fuel by isotope dilution high resolution ICP–MS, using silicon for mass bias correction of natural isotope ratios. J. Anal. Atom. Spectrom. 16(9), 964–969.
DANYUSHEVSKY, L.V., DELLA-PASQUA, F.N. & SOKOLOV, S. (2000): Re-equilibration of melt inclusions trapped by magnesian olivine phenocrysts from subduction-related magmas: petrological implications. Contrib. Min. Petrol. 138(1), 68–83.
FAURE F. & SCHIANO P. (2005): Experimental investigation of equilibration conditions during forsterite growth and melt inclusion formation. Earth Planet. Sci. Lett. 236(3–4), 882–898.
DANYUSHEVSKY, L.V., LESLIE, R.A.J., CRAWFORD, A.J. & DURANCE, P. (2004): Melt inclusions in primitive olivine phenocrysts: The role of localized reaction processes in the origin of anomalous compositions. J. Petrol. 45(12), 2531–2553.
FREZZOTTI M.L. (2001): Silicate-melt inclusions in magmatic rocks: applications to petrology. Lithos 55(1–4), 273–299. FRYER, B.J., JACKSON, S.E. & LONGERICH, H.P. (1995): Design, operation and role of the laserablation microprobe coupled with an inductively coupled plasma mass spectrometer (LAM–ICP– MS) in the earth sciences. Can. Mineral. 33, 303– 312.
DANYUSHEVSKY, L.V., MCNEILL, A.W. & SOBOLEV, A.V. (2002): Experimental and petrological studies of melt inclusions in phenocrysts from mantle-derived magmas: an overview of techniques, advantages and complications. Chem. Geol. 183(1–4), 5–24.
GAETANI, G.A. & WATSON, E.B. (2000): Open system behavior of olivine-hosted melt inclusions. Earth Planet. Sci. Lett. 183(1–2), 27– 41.
DEBARI, S.M. & SLEEP, N.H. (1991): High-Mg, low-Al bulk composition of the Talkeetna Island
75
T. PETTKE
GOLDSTEIN, J.L., NEWBURY, D.E., ECHLIN, P., JOY, D.C., ROMIG, A.D., LYMAN, C.E., FIORI, C. & LIFSHIN, E. (1992): Scanning electron microscopy and X-ray microanalysis. 2nd edition. Plenum Press, New York, 820p.
GÜNTHER, D., FRISCHKNECHT, R., HEINRICH, C.A. & KAHLERT, H.J. (1997): Capabilities of an Argon Fluoride 193 nm excimer laser for laser ablation inductively coupled plasma mass spectrometry microanalysis of geological materials. J. Anal. Atom. Spectrom. 12(9), 939– 944.
GRAY, A.L. (1985): Solid sample introduction by laser ablation for inductively coupled plasma source–mass spectrometry. Analyst 110(5), 551– 556.
HALTER, W.E. & HEINRICH, C.A. (2005): Magmatic processes and volatile phase generation in porphyry-type environments: A laser ablation– ICP–MS study of silicate and sulfide melt inclusions. In Melt Inclusions in Plutonic Rocks (J.D. Webster, ed.) Min. Assoc. Can. Short Course 36, 151-164.
GROVE, T.L., PARMAN, S.W., BOWRING, S.A., PRICE, R.C. & BAKER, M.B. (2002): The role of an H2O-rich fluid component in the generation of primitive basaltic andesites and andesites from the Mt. Shasta region, N California. Contrib. Min. Petrol. 142(4), 375–396.
HALTER W., HEINRICH C.A. & PETTKE T. (2005): Magma evolution and the formation of porphyry Cu–Au ore fluids: evidence from silicate and sulfide melt inclusions. Mineral. Dep. 39(8), 845–863.
GUILLONG, M. & GÜNTHER, D. (2002): Effect of particle size distribution on ICP-induced elemental fractionation in laser ablation– inductively coupled plasma–mass spectrometry. J. Anal. Atom. Spectrom. 17(8), 831–837.
HALTER, W., HEINRICH, C.A. & PETTKE, T. (2004b): Laser–ablation ICP–MS analysis of silicate and sulfide melt inclusions in an andesitic complex II: evidence for magma mixing and magma chamber evolution. Contrib. Min. Petrol. 147(4), 397–412.
GUILLONG, M., HORN, I. & GÜNTHER, D. (2003a): A comparison of 266 nm, 213 nm and 193 nm produced from a single solid state Nd : YAG laser for laser ablation ICP–MS. J. Anal. Atom. Spectrom. 18(10), 1224–1230.
HALTER, W., PETTKE ,T. & HEINRICH, C.A. (2002b): The origin of Cu/Au ratios in porphyrytype ore deposits. Science 296(5574), 1844–1846.
GUILLONG M., KUHN H.R. & GÜNTHER D. (2003b): Application of a particle separation device to reduce inductively coupled plasma-enhanced elemental fractionation in laser ablation– inductively coupled plasma–mass spectrometry. Spectrochim. Acta Part B–At. Spectrosc. 58(2–3), 211–220.
HALTER, W., PETTKE, T. & HEINRICH, C.A. (2004a): Laser–ablation ICP–MS analysis of silicate and sulfide melt inclusions in an andesitic complex I: analytical approach and data evaluation. Contrib. Min. Petrol. 147(4), 385– 396.
GÜNTHER, D., AUDÉTAT, A., FRISCHKNECHT, R. & HEINRICH C.A. (1998): Quantitative analysis of major, minor and trace elements in fluid inclusions using laser ablation inductively coupled plasma mass spectrometry. J. Anal. Atom. Spectrom. 13(4), 263-270.
HALTER, W., PETTKE, T. & HEINRICH, C.A. & ROTHEN-RUTISHAUSER, B. (2002a): Major to trace element analysis of melt inclusions by laserablation ICP–MS: methods of quantification. Chem. Geol. 183(1–4), 63–86.
GÜNTHER D. & HATTENDORF, B. (2005): Solid sample analysis using laser ablation inductively coupled plasma mass spectrometry. Trac-Trends in Anal. Chem. 24(3), 255–265.
HANLEY, J.J., PETTKE, T., MUNGALL, J.E. & SPOONER, E.T.C. (2005a): The solubility of platinum and gold in NaCl brines at 1.5 kbar, 600 to 800 degrees C: A laser ablation ICP–MS pilot study of synthetic fluid inclusions. Geochmim. Cosmochim. Acta 69(10), 2593–2611.
GÜNTHER, D., HATTENDORF, B. & AUDÉTAT, A. (2001): Multi-element analysis of melt and fluid inclusions with improved detection capabilities for Ca and Fe using laser ablation with a dynamic reaction cell ICP–MS. J. Anal. Atom. Spectrom. 16(9), 1085–1090.
HANLEY, J.J., MUNGALL, J.E., PETTKE, T., SPOONER, E.T.C. & BRAY, C.J. (2005b): Ore metal redistribution by hydrocarbon–brine and hydrocarbon–halide melt phases, North Range 76
IN SITU LASER-ABLATION-ICP-MS ANALYSIS AND CONSTRAINING SUBDUCTION ZONE MAGMATISM
and sources of laser induced isotopic fractionation in laser ablation–multicollectorinductively coupled plasma– mass spectrometry. J. Anal. Atom. Spectrom. 18(3), 205–212.
footwall of the Sudbury Igneous Complex, Ontario, Canada. Mineral. Dep. 40(3), 237–256. HATTENDORF, B. & GÜNTHER, D. (2000): Characteristics and capabilities of an ICP–MS with a dynamic reaction cell for dry aerosols and laser ablation. J. Anal. Atom. Spectrom. 15(9), 1125–1131.
JACKSON, S.E., LONGERICH, H.P., DUNNING, G.R. & FRYER, B.J. (1992): The application of laserablation microprobe–inductively coupled plasma– mass-spectrometry (LAM–ICP–MS) to in situ trace-element determinations in minerals. Can. Mineral. 30, 1049–1064.
HATTORI, K.H. & KEITH, J.D. (2001): Contribution of mafic melt to porphyry copper mineralization: evidence from Mount Pinatubo, Philippines, and Bingham Canyon, Utah, USA. Mineral. Dep. 36(8), 799–806.
JEFFRIES, T.E., PEARCE, N.J.G., PERKINS, W.T. & RAITH, A. (1996): Chemical fractionation during infrared and ultraviolet laser ablation inductively coupled plasma mass spectrometry – Implications for mineral microanalysis. Analyt. Commun. 33(1), 35–39.
HAURI, E., WANG, J.H., DIXON, J.E., KING, P.L., MANDEVILLE, C. & NEWMAN, S. (2002): SIMS analysis of volatiles in silicate glasses 1. Calibration, matrix effects and comparisons with FTIR. Chem. Geol. 183(1–4), 99–114.
JENNER, G.A., FOLEY, S.F., JACKSON, S.E., GREEN, T.H., FRYER, B.J. & LONGERICH, H.P. (1993): Determination of partition coefficients for trace elements in high pressure temperature experimental run products by laser ablation microprobe–inductively coupled plasma–mass spectrometry (LAM–ICP–MS). Geochmim. Cosmochim. Acta 57(23–24), 5099–5103.
HAWKESWORTH, C.J., GALLAGHER, K., HERGT, J.M. & MCDERMOTT, F. (1993): Trace-element fractionation processes in the generation of island-arc basalts. Philosophical Transactions of the Royal Society of London Series A – Mathematical Physical and Engineering Sciences 342(1663), 179–191.
JOHNSON, M.-C. & PLANK, T. (1999): Dehydration and melting experiments constrain the fate of subducted sediment. Geochem. Geophys. Geosys.1, doi:10.1029/1999GC000014 1999.
HEINRICH, C.A., PETTKE, T., HALTER, W.E., AIGNER-TORRES, M., AUDÉTAT, A., GÜNTHER, D., HATTENDORF, B., BLEINER, D., GUILLONG, M. & HORN, I. (2003): Quantitative multi-element analysis of minerals, fluid and melt inclusions by laser-ablation inductively-coupled-plasma massspectrometry. Geochmim. Cosmochim. Acta 67(18), 3473–3497.
KAMENETSKY, V.S. (2006): Melt inclusion record of magmatic immiscibility in crustal and mantle magmas. In Melt Inclusions in Plutonic Rocks (J.D. Webster, ed.) Min. Assoc. Can. Short Course 36, 81-98.
HELLEBRAND, E., SNOW, J.E., MOSTEFAOUI, S. & HOPPE P. (2005): Trace element distribution between orthopyroxene and clinopyroxene in peridotites from the Gakkel Ridge: a SIMS and NanoSIMS study. Contrib. Min. Pet. 150(5), 486–504.
KAMENETSKY, V.S., CRAWFORD, A.J., EGGINS, S. & MUHE, R. (1997): Phenocryst and melt inclusion chemistry of near-axis seamounts, Valu Fa Ridge, Lau Basin: insight into mantle wedge melting and the addition of subduction components. Earth Planet. Sci. Lett. 151(3–4), 205–223.
HORN, I., GUILLONG, M. & GÜNTHER, D. (2001): Wavelength dependant ablation rates for metals and silicate glasses using homogenized laser beam profiles – implications for LA–ICP–MS. Appl. Surf. Sci. 182(1–2), 91–102.
KAMENETSKY, V.S. & DANYUSHEVSKY, L.V. (2005): Metals in quartz-hosted melt inclusions: Natural facts and experimental artifacts. Am. Mineral. 90(10), 1674–1678.
IHINGER, P.D., HERVIG, R.L. & MCMILLAN, P.F. (1994): Analytical methods for volatiles in magmas. In: Carrol, M.R. and Holloway, J.R. (eds), Rev. Mineral. 30, 67–129
KAMENETSKY, V.S., SOBOLEV, A.V., EGGINS, S.M., CRAWFORD, A.J. & ARCULUS, R.J. (2002): Olivine-enriched melt inclusions in chromites from low-Ca boninites, Cape Vogel, Papua New Guinea: evidence for ultramafic primary magma,
JACKSON, S.E. & GÜNTHER, D. (2003): The nature
77
T. PETTKE
refractory mantle source and enriched components. Chem. Geol. 183(1–4), 287–303.
LAHAYE, Y., BENDALL, C., BREY, G., FIEBIG, J. & WEYER S. (2004): In situ sulfur isotope ratio measurements by laser ablation high resolution MC–ICP–MS. Geochim. Cosmochim. Acta 68(11), A69–A69.
KAMENETSKY, V.S., WOLFE, R.C., EGGINS, S.M., MERNAGH, T.P. & BASTRAKOV, E. (1999): Volatile exsolution at the Dinkidi Cu–Au porphyry deposit, Philippines: A melt–inclusion record of the initial ore-forming process. Geology 27(8), 691–694.
LATKOCZY, C. & GÜNTHER, D. (2002): Enhanced sensitivity in inductively coupled plasma sector field mass spectrometry for direct solid analysis using laser ablation (LA–ICP–SFMS). J. Anal. Atom. Spectrom. 17(10), 1264–1270.
KAY, S.M & MPODOZIS, C. (2001): Central Andean ore deposits linked to evolving shallow subduction systems and thickened crust. GSA Today 11(3), 4–9.
LAYNE, G.D. (2006): Application of secondary ion mass spectrometry to the determination of traditional and non-traditional light stable isotopes in melt inclusions. In Melt Inclusions in Plutonic Rocks (J.D. Webster, ed.) Min. Assoc. Can. Short Course 36, 27-49.
KENT, A.J.R., PEATE, D.W., NEWMAN, S., STOLPER, E.M. & PEARCE J.A. (2002): Chlorine in submarine glasses from the Lau Basin: seawater contamination and constraints on the composition of slab-derived fluids. Earth Planet. Sci. Lett. 202(2), 361–377.
LEACH, A.M. & HIEFTJE, G.M. (2000): Methods for shot-to-shot normalization in laser ablation with an inductively coupled plasma time-of-flight mass spectrometer. J. Anal. Atom. Spectrom. 15(9), 1121–1124.
KEPPLER, H. (1996): Constraints from partitioning experiments on the composition of subductionzone fluids. Nature 380(6571), 237–240.
LONGERICH, H.P., GÜNTHER, D. & JACKSON, S.E. (1996b): Elemental fractionation in laser ablation inductively coupled plasma mass spectrometry. Fresen. J. Anal. Chem. 355(5–6), 538–542.
KESSEL, R., SCHMIDT, M.W., ULMER, P. & PETTKE, T. (2005a): Trace element signature of subduction-zone fluids, melts and supercritical liquids at 120–180 km depth. Nature 437(7059), 724–727.
LONGERICH H.P., JACKSON, S.E. & GÜNTHER, D. (1996a): Laser ablation inductively coupled plasma mass spectrometric transient signal data acquisition and analyte concentration calculation. J. Anal. Atom. Spectrom. 11(9), 899–904.
KESSEL, R., ULMER, P., PETTKE, T., SCHMIDT, M.W. & THOMPSON, A.B. (2005b): The water– basalt system at 4 to 6 GPa: Phase relations and second critical endpoint in a K-free eclogite at 700 to 1400 degrees C. Earth Planet. Sci. Lett. 237(3–4), 873–892.
LOWENSTERN, J.B. (1995): Applications of silicate–melt inclusions to the study of magmatic volatiles. In MAC, Short Course Series 23 (ed. F.H. Thompson), pp. 71–99.
KESSEL, R., ULMER, P., PETTKE, T., SCHMIDT, M.W. & THOMPSON, A.B. (2004): A novel approach to determine high-pressure hightemperature fluid and melt compositions using diamond-trap experiments. Am. Mineral. 89(7), 1078–1086.
LOWENSTERN, J.B. & MAHOOD, G.A. (1991): New data on magmatic H2O contents of pantellerites, with implications for petrogenesis and eruptive dynamics at Pantelleria. Bull. Volcanol. 54(1), 78–83.
KRESS, V.C. & GHIORSO, M.S. (2004): Thermodynamic modeling of post-entrapment crystallization in igneous phases. J. Volcanol. Geotherm. Res. 137(4), 247–260.
LUDWIG, K.R. (1994): ISOPLOT: A plotting and regression program for radiogenic isotope data, for IBM–PC compatible computers, version 2.71. USGS Open file Report 91–445.
KUHN, H.R. & GÜNTHER, D. (2004): Laser ablation–ICP–MS: particle size dependent elemental composition studies on filter-collected and online measured aerosols from glass. J. Anal. Atom. Spectrom. 19(9) 1158–1164.
LUHR, J.F. (2001): Glass inclusions and melt volatile contents at Paricutin Volcano, Mexico. Contrib. Min. Petrol. 142(3), 261–283. MCCULLOCH, M.T. & GAMBLE, J.A. (1991): Geochemical and geodynamical constraints on 78
IN SITU LASER-ABLATION-ICP-MS ANALYSIS AND CONSTRAINING SUBDUCTION ZONE MAGMATISM
QIN, Z.W., LU, F.Q. & ANDERSON, A.T. (1992): Diffusive reequilibration of melt and fluid inclusions. Am. Mineral. 77(5–6), 565–576.
subduction zone magmatism. Earth Planet. Sci. Lett. 102(3–4), 358–374. METRICH, N. & CLOCCHIATTI, R. (1989): Melt inclusion investigation of the volatile behavior in historic alkali basalt magmas of Etna. Bull. Volcanol. 51(3), 185-198.
RAMSEY, M.H., COLES, B. J., WILKINSON, J.J., & RANKIN, A.H. (1992): Single fluid inclusion analysis by laser ablation inductively coupled plasma atomic emission spectrometry – quantification and validation. J. Anal. Atom. Spectrom. 7(4), 587–593.
METRICH, N. & CLOCCHIATTI, R. (1996): Sulfur abundance and its speciation in oxidized alkaline melts. Geochmim. Cosmochim. Acta 60(21), 4151–4160.
ROEDDER, E. (1979): Origin and significance of magmatic inclusions. Bull Minéral. 102(5–6), 487–510.
MORRIS, J.D., LEEMAN, W.P. & TERA, F. (1990): The subducted component in island arc lavas – constraints from Be isotopes and B–Be systematics. Nature 344(6261), 31–36.
ROEDDER, E. (1984): Fluid Inclusions. Rev. Mineral. 12. Mineralogical Society of America, pp 645.
NEWMAN, S., STOLPER, E.M. & EPSTEIN, S. (1986): Measurement of water in rhyolitic glasses – calibration of an infrared spectroscopic technique. Am. Mineral. 71(11–12), 1527–1541.
RUSSO, R.E., MAO, X.L., LIU, C. & GONZALEZ, J. (2004): Laser assisted plasma spectrochemistry: laser ablation. J. Anal. Atom. Spectrom. 19(9), 1084–1089.
NHO, E.Y., LECLOAREC, M.F., ARDOUIN, B. & TJETJEP, W.S. (1996): Source strength assessment of volcanic trace elements emitted from the Indonesian arc. J. Volcanol. Geotherm. Res. 74(1–2), 121–129.
SCHIANO, P. (2003): Primitive mantle magmas recorded as silicate melt inclusions in igneous minerals. Earth Sci. Rev. 63(1–2), 121–144. SCHIANO P., CLOCCHIATTI R., BOIVIN P. & MEDARD E. (2004b): The nature of melt inclusions inside minerals in an ultramafic cumulate from Adak volcanic center, aleutian arc: implications for the origin of high-Al basalts. Chem. Geol. 203(1–2), 169–179.
PERFIT, M.R., GUST, D.A., BENCE, A.E., ARCULUS, R.J. & TAYLOR, S.R. (1980): Chemical characteristics of island arc basalts – implications for mantle sources. Chem. Geol. 30(3), 227–256. PETTKE, T., HEINRICH, C.A., CIOCAN, A.C. & GÜNTHER, D. (2000): Quadrupole mass spectrometry and optical emission spectroscopy: detection capabilities and representative sampling of short transient signals from laser ablation. J. Anal. Atom. Spectrom. 15(9), 1149–1155.
SCHIANO, P., CLOCCHIATTI, R., OTTOLINI, L. & SBRANA, A. (2004a): The relationship between potassic, calc-alkaline and Na-alkaline magmatism in South Italy volcanoes: A melt inclusion approach. Earth Planet. Sci. Lett. 220(1–2), 121–137.
PETTKE, T., HALTER, W.E., WEBSTER, J.D., AIGNER-TORRES, M. & HEINRICH, C.A. (2004): Accurate quantification of melt inclusion chem.istry by LA–ICP–MS: a comparison with EMP and SIMS and advantages and possible limitations of these methods. Lithos 78(4), 333–361.
SCHIANO, P., CLOCCHIATTI, R., SHIMIZU, N., MAURY, R.C., JOCHUM, K.P. & HOFMANN, A.W. (1995): Hydrous, silica-rich melts in the sub-arc mantle and their relationship with erupted arc lavas. Nature 377(6550), 595–600.
PLANK, T. & LANGMUIR, C.H. (1993): Tracing trace elements from sediment input to volcanic output at subduction zones. Nature 362(6422), 739–743.
SIMON, A.C., FRANK, M.R., PETTKE ,T., CANDELA, P.A., PICCOLI, P.M. & HEINRICH, C.A. (2005): Gold partitioning in melt–vapor–brine systems. Geochmim. Cosmochim. Acta 69(13), 3321–3335.
POITRASSON, F., MAO, X.L., MAO, S.S., FREYDIER, R. & RUSSO, R.E. (2003): Comparison of ultraviolet femtosecond and nanosecond laser ablation inductively coupled plasma mass spectrometry analysis in glass, monazite, and zircon. Anal. Chem. 75(22), 6184–6190.
SISSON, T.W. & LAYNE, G.D. (1993): H2O in basalt and basaltic andesite glass inclusions from 4 subduction-related volcanos. Earth Planet. Sci. Lett. 117(3–4), 619–635.
79
T. PETTKE
SOBOLEV, A.V. (1996): Melt inclusions in minerals as a source of principle petrological information. Petrology 4(3), 209–220.
TATSUMI, Y. (1989): Migration of fluid phases and genesis of basalt magmas in subduction zones. J: Geophys. Res. – Solid Earth 94(B4), 4697–4707.
SOBOLEV, A.V. & CHAUSSIDON, M. (1996): H2O concentrations in primary melts from suprasubduction zones and mid-ocean ridges: Implications for H2O storage and recycling in the mantle. Earth Planet. Sci. Lett. 137(1–4), 45–55.
TAYLOR, R.P., JACKSON, S.E., LONGERICH, H.P. & WEBSTER, J.D. (1997): In situ trace element analysis of individual silicate melt inclusions by laser ablation microprobe inductively coupled plasma–mass spectrometry (LAM–ICP–MS). Geochmim. Cosmochim. Acta 61(13), 2559–2567.
SOBOLEV, A.V. & DANYUSHEVSKY, L.V. (1994): Petrology and geochemistry of boninites from the north termination of the Tonga Trench – constraints on the generation conditions of primary high-Ca boninite magmas. J. Petrol. 35(5), 1183–1211.
THOMAS, R. (2000): Determination of water contents of granite melt inclusions by confocal laser Raman microprobe spectroscopy. Am. Mineral. 85(5–6), 868–872. ULMER, P. (2001): Partial melting in the mantle wedge – the role of H2O in the genesis of mantlederived 'arc-related' magmas. Physics of the Earth and Planetary Interiors 127(1–4), 215–232.
SPANDLER, C.J., ARCULUS, R.J., EGGINS, S.M., MAVROGENES, J.A., PRICE, R.C. & REAY, A.J. (2003): Petrogenesis of the Greenhills Complex, Southland, New Zealand: magmatic differentiation and cumulate formation at the roots of a Permian island-arc volcano. Contrib. Min. Petrol. 144(6), 703–721.
WALLACE, P.J. (2005): Volatiles in subduction zone magmas: concentrations and fluxes based on melt inclusion and volcanic gas data. J. Volcanol. Geotherm. Res. 140(1–3), 217–240.
STADERMANN, F.J., CROAT, T.K., BERNATOWICZ, T.J., AMARI, S., MESSENGER, S., WALKER, R.M. & ZINNER, E. (2005): Supernova graphite in the NanoSIMS: Carbon, oxygen and titanium isotopic compositions of a spherule and its TiC sub-components. Geochmim. Cosmochim. Acta 69(1), 177–188.
WALLACE, P.J., ANDERSON, A.T. & DAVIS, A.M. (1999): Gradients in H2O, CO2, and exsolved gas in a large volume silicic magma system: Interpreting the record preserved in melt inclusions from the Bishop Tuff. J: Geophys. Res. – Solid Earth 104(B9), 20097–20122.
STOLPER, E. (1982): Water in silicate glasses – an infrared spectroscopic study. Contrib. Min. Petrol. 81(1) 1–17.
WATSON, E.B. (1976): Glass inclusions as samples of early magmatic liquid – determinative method and application to a South-Atlantic basalt. J. Volcanol. Geotherm. Res. 1(1), 73–84.
STOLPER, E. & NEWMAN, S. (1994): The role of water in the petrogenesis of Mariana Trough magmas. Earth Planet. Sci. Lett. 121(3–4), 293– 325.
WEBSTER, J.D., THOMAS, R., RHEDE, D., FORSTER, H.J. & SELTMANN, R. (1997): Melt inclusions in quartz from an evolved peraluminous pegmatite: Geochemical evidence for strong tin enrichment in fluorine-rich and phosphorus-rich residual liquids. Geochmim. Cosmochim. Acta 61(13), 2589–2604.
STRAUB, S.M. & LAYNE, G.D. (2003): The systematics of chlorine, fluorine, and water in Izu arc front volcanic rocks: Implications for volatile recycling in subduction zones. Geochmim. Cosmochim. Acta 67(21), 4179–4203.
ZAJACZ, Z., HALTER, W.E., MALFAIT, W.J., BACHMANN, O., BODNAR, R.J., HIRSCHMANN, M.M., MANDEVILLE, C.W., MORIZET, Y., MÜNTENER, O., ULMER, P. & WEBSTER, J.D. (2005): A composition-independent quantitative determination of the water content in silicate glasses and silicate melt inclusions by confocal Raman spectroscopy. Contrib. Min. Petrol. 150(6), 631–642.
STUDENT, J.J. & BODNAR, R.J. (2004): Silicate melt inclusions in porphyry copper deposits: Identification and homogenization behavior. Can. Mineral. 42, 1583–1599. TAIT, S. (1992): Selective preservation of melt inclusions in igneous phenocrysts. Am. Mineral. 77(1–2), 146–155.
80
CHAPTER 4: MELT INCLUSION RECORD OF MAGMATIC IMMISCIBILITY IN CRUSTAL AND MANTLE MAGMAS Vadim S. Kamenetsky School of Earth Sciences and ARC Centre of Excellence in Ore Deposits, University of Tasmania, Private Bag 79, Hobart, Tasmania 7001, Australia E-mail:
[email protected] process. If separation of immiscible phases was efficient, the residual magma should be significantly depleted in incompatible volatiles and metals relative to the parental magma, and reconstructing the original metal and volatile content is extremely difficult. One rapidly developing approach to this problem is the use of melt and fluid inclusions trapped and preserved in magmatic minerals (e.g., Roedder 1992, De Vivo & Frezzotti 1994, Bodnar 1995, Lowenstern 1995, Student & Bodnar 1999, Frezzotti 2001, Kamenetsky et al. 2003, Lowenstern 2003 and references therein). Such inclusions provide the closest approximation to samples of continuously evolving (and thus ephemeral) melts and magmatic fluids. Many studies of magmatic inclusions have made possible the recognition of several types of magmatic immiscibility (e.g., between silicate melts, sulfide melts, aqueous and carbonic liquids and vapors, hydrosaline liquids and various combinations of these). For brevity, in this work only those examples from plutonic systems of which the author has first-hand experience will be presented and discussed in detail.
INTRODUCTION Immiscibility (unmixing of melts and fluids) should be almost inevitable at some point in the evolution of most mantle and crustal magmas during cooling and crystallization. Formation of two immiscible phases “results in a major geochemical fractionation – all chemical species present, the elements (and their isotopes), and their various compounds, become distributed between these two phases…the compositional divergence between the two phases is…extreme” (Roedder 2003). The variations in compositions of melts undergoing immiscibility, and physical parameters of their evolution in the plutonic environments, mean that each intrusion should be expected to show differences in the processes of exsolution and compositions of exsolving phases. One of the immiscible phases is universally volatile-rich, and this has important consequences for further magma evolution and geological processes related to it. More specifically, volatile-rich phases generally have significant density and viscosity contrasts with parental silicate magmas, and thus rapid separation of the newly exsolved phases is expected. Further, the exsolution of volatile-rich phases exerts major controls on the chemistry of a magmatic system, particularly on metal partitioning between immiscible melts and fluids, so the volatile phase is highly efficient at sequestering the metals (e.g., Candela 1989, Candela & Piccoli 1995, Williams et al. 1995, Heinrich et al. 1999, Webster 2004). Magmatic immiscibility and the related formation of volatile-rich melts and fluids are prerequisites for the origin of mineralized hydrothermal solutions that may transport metals to a suitable depositional site. Immiscible separation, however, is not restricted to magmas that form mineralized rocks. The fugitive nature of magmatic immiscibility involves problems in unraveling physical and chemical characteristics of this fundamental
TYPES OF MAGMATIC IMMISCIBILITY Immiscible silicate melts Among the reports on silicate-silicate melt immiscibility (e.g., Roedder & Weiblen 1970, De 1974, Dixon & Rutherford 1979, McBirney 1980, Philpotts 1982, Jakobsen et al. 2005) only a few are related to mineralized magmatic systems (e.g., pegmatites of the Ehrenfriedersdorf Sn–W deposit, Germany, Thomas et al. 2000, the La Copa felsic intrusions within the Rio Blanco Cu–Mo deposit, Chile, Davidson & Kamenetsky 2001, Davidson et al. 2005). Quartz-hosted silicate melt inclusions from the La Copa complex, coexisting within the same grains and growth planes, belong to two contrasting
Mineralogical Association of Canada Short Course 36, Montreal, Quebec, p. 81-98. 81
V.S. KAMENETSKY
mapping, is elevated metal (Fe, Mn, Cu, Zn, Pb) and alkali element abundances in a volatile-rich silicate melt (Fig. 4-2). Davidson & Kamenetsky (2001) emphasized that dark silicate melt inclusions, enriched in volatiles and metals, are important to the understanding of late-stage magmatic immiscibility and suggested revisiting the results of other studies on similar inclusions. For example, the melt inclusions from the Lower Bandelier Tuff (Dunbar & Hervig 1992) can be “strongly devitrified” and “darker in color…and some contain clots of crystals”, and also contain up to 8 wt.% H2O, in contrast to clear glass inclusions (3–5 wt.% H2O). It is possible that crystallization of trapped melts was promoted by high H2O abundances. The interpretation of coexisting volatile-poor, and volatile-rich, silicate melts (melt inclusions) as immiscible liquids at La Copa (Davidson & Kamenetsky 2001) was partly based on the phase diagram derived from homogenization experiments with melt inclusions from the Ehrenfriedersdorf pegmatite (Thomas et al. 2000). According to their ground-breaking study, the original H2Oundersaturated felsic magma separates at a certain temperature into two immiscible liquids, a volatilepoor silicate melt and a volatile-rich silicate melt (or a silicate-bearing fluid). The latter inevitably evolves via one or more consecutive unmixing events into aqueous metal-, salt-rich (“hydrothermal”) fluids with a potential for economic mineralization (see below).
types: clear rhyolitic glass with one or several shrinkage bubbles, and dark fine-grained crystalline aggregates of feldspar, mica and quartz (Fig. 4-1, Davidson & Kamenetsky 2001, Davidson et al. 2005). Dark inclusions may also contain daughter crystals of sphalerite, chalcopyrite, Fe oxides and hydroxides, halite, carbonates and unidentified phases. They are also characterized by a significant and variable amount of fluid components, identified by laser Raman spectroscopy as vapor and liquid H2O. A common feature of dark inclusions is an associated halo of aqueous vapor-dominated bubbles (< 5 µm) forming a discontinuous rim, which probably formed as a result of postcrystallization decrepitation of inclusions. Decrepitation of seemingly intact large dark inclusions at T~600°C is responsible for their failure to melt and homogenize during heating experiments. Some smaller inclusions show melting at 750–800°C, although complete homogenization (bubble disappearance) was not achieved even at higher temperatures. Clear evidence of silicate-silicate melt immiscibility in the La Copa samples is observed in inclusions that comprise both clear glass and rounded, commonly spherical, crystalline and amorphous dark masses of dominantly silicate phases with interstitial aqueous fluid (Fig. 4-2). The relative proportions of trapped immiscible melts in such composite inclusions are highly variable, and their compositions are distinctly different (Fig. 4-2). The main difference, as indicated by PIXE
FIG. 4-1. A growth plane in a quartz phenocryst from La Copa complex, Chile (Davidson & Kamenetsky 2006), containing coexisting crystallized silicate melt inclusion (1), glass inclusions (2) and two-phase aqueous fluid inclusions (3). The twophase aqueous inclusions show Brownian motion of the vapor bubbles, and freeze at low temperature.
82
MELT INCLUSION RECORD OF MAGMATIC IMMISCIBILITY IN CRUSTAL AND MANTLE MAGMAS
FIG. 4-2. Optical images and proton-induced X-ray emission (PIXE) element maps of quartz-hosted composite inclusions from La Copa complex, Chile (Davidson et al. 2005). The inclusions are composed of clear silicate glass and a globule of crystallized volatile-rich melt (dark). Outlines on element maps mark boundaries of inclusions and their volatile-rich silicate globules. Scale bars are 15 µm.
trapped in the same growth plane, is provided in Fig. 4-1 (after Davidson & Kamenetsky 2006). Notably, the dark silicate melt inclusion in Fig. 4-1 belongs to the volatile-rich type described above. Another example of explicit immiscibility between the silicate melt and magmatic aqueous fluid prior to or simultaneous with quartz crystallization is recorded by Davidson & Kamenetsky (2006) as composite inclusions containing very large numbers of one- or two-phase aqueous bubbles in a silicate glass (microemulsion, Fig. 4-3).
Immiscible silicate melt and aqueous saline fluids Quartz phenocrysts in felsic intrusive rocks commonly contain numerous aqueous fluid inclusions, but in most cases they are clearly later than the silicate melt inclusions (if present) in the same grains. Trails of secondary aqueous inclusions commonly obliterate any evidence of primary aqueous inclusions, at least presenting a considerable challenge to their confident identification. An unambiguous example of silicate melt and aqueous two-phase fluid inclusions, co-
83
V.S. KAMENETSKY
FIG. 4-3. Optical images and proton-induced X-ray emission (PIXE) element maps of quartz-hosted composite inclusions from La Copa complex, Chile (Davidson et al. 2005). The inclusions are composed of clear silicate glass and numerous bubbles of aqueous fluid (vapor- and liquid-rich). Outlines on element maps mark boundaries of inclusions and aqueous bubbles (on B). Scale bars are 25 µm.
On the other hand, in cases where crystallization occurs deeper than the magma saturation in H2O, primary aqueous fluid inclusions are not trapped. Thus, the search for such inclusions in deep-seated plutons can be a futile task. A homogeneous H2O-undersaturated melt trapped at higher pressure should crystallize and evolve in a closed system (such as within a melt inclusion) towards saturation and immiscibility (Naumov 1979, Roedder 1984, Thomas & Webster 2000). If this is the case, a separate aqueous phase is associated with the silicate content of melt inclusions (Davidson & Kamenetsky 2001, 2006, Davidson et al. 2005). Quartz-hosted melt and fluid inclusions from the La Copa magmatic complex, Chile and Taupo Volcanic Zone, New Zealand (thought to be trapped at 2–3 kb pressure according to their high H2O abundances of up to 7–8 wt.%; Dunbar et al. 1989, Dunbar & Kyle 1993, Kamenetsky & Danyushevsky 2005) demonstrate
post-entrapment exsolution of the aqueous fluid. Partly crystallized silicate melt inclusions from both localities are occasionally surrounded by swarms of aqueous vapor-rich and two-phase inclusions (Fig. 4-4). These fluids either heal fractures radiating from silicate melt inclusions, or decorate what appear to be oriented structural defects (dislocations, channels) in the quartz (Fig. 4-4). The spatial association of a “parental” silicate melt inclusion with systematically aligned aqueous inclusions, in conjunction with their elongated, commonly tubular shapes (Fig. 4-4), suggests the host quartz may have experienced hydraulic fracturing, caused by fluid overpressure within the silicate melt. The build-up of fluid pressure was a result of the melt crystallization, saturation in volatiles, immiscible separation of a fluid phase, and also possibly the compaction of host phenocrysts at the α–β quartz transition.
84
MELT INCLUSION RECORD OF MAGMATIC IMMISCIBILITY IN CRUSTAL AND MANTLE MAGMAS
FIG. 4-4. Systematic distribution of aqueous fluid inclusions around crystallized silicate melt inclusions (shown by arrows) in quartz from the Taupo Volcanic Zone (New Zealand). Scale bars are 50 µm.
Glass inclusions from the La Copa and Taupo Volcanic Zone contain aqueous vapor, aqueous liquid, or both (Fig. 4-4, 5), sometimes with cubic chloride crystals (Fig. 4-5C, D). Bubbles of an aqueous liquid typically have spherical shape, although in the case of the La Copa samples some bubbles appear to be plastically deformed in a contracting glass (Fig. 4-5A, B). Microthermometric experiments on one- and twophase aqueous bubbles from the La Copa samples (Davidson & Kamenetsky 2006) confirmed that these bubbles contain an aqueous liquid, since they freeze into a mass of ice crystals at temperatures below –40°C. The salinities of aqueous fluids, estimated using the data of Bodnar & Vityk (1994), range from 13–17 wt.% NaCl eq. in liquid-phase bubbles, to ≥40 wt.% NaCl eq. in bubbles containing salt precipitates. Dense aqueous fluid bubbles in melt inclusions have been observed in plagioclase and quartz from andesites and rhyolites in Slovakia, the western Carpathians and central Tien Shan (Naumov et al. 1992, 1994, 1996). This implies that they may not be unusual, and should be looked for in other instances. Aqueous one-phase liquid bubbles in silicate glass are likely to be a metastability phenomenon, reflecting the surface tension effects in very small containers. At certain bubble volumes, for a given composition and density, fluid “will remain stretched indefinitely (i.e., without nucleation of a bubble), as the stable configuration” (Roedder 1984). In other words, the fluids may increase their average inter-molecular distances, rather than nucleate a bubble. It should be noted that the nucleation of vapor phase inside a homogeneous liquid bubble occurs sometimes during heating and freezing experiments and during PIXE or Raman measurements (Fig. 4-5D). But the
fact that, in some cases, the bubbles spontaneously return to a single liquid state means that in this specific situation the “stretched” configuration is actually more stable. Samples of most pristine magmatic fluids, formed in a closed system of their parental melt (melt inclusion), have demonstrable quantities of chlorine and metals. The compositions of glasshosted aqueous bubbles, apart from observed chloride crystals (Fig. 4-5C, D) and moderate chlorine content, are characterized by elevated concentrations (1–3 orders of magnitude with respect to silicate glasses) of Cu, Zn and Fe (Fig. 4, Davidson & Kamenetsky 2006) and Cu and Ag (Kamenetsky & Danyushevsky 2005), but at present this enrichment is not expressed in absolute terms. Our preliminary data indicate that metals are indeed sequestered from the silicate melt into in situ exsolved aqueous bubbles, and that compositional diversity of immiscible fluids can occur at very fine scales. Immiscible silicate melt and hydrosaline liquids/salt melts Exsolution of a moderately saline aqueous fluid from a silicate melt, and its later separation into vapor and brine fluids, has been a cornerstone of the orthomagmatic model (e.g., Burnham 1979). Magmas are also capable of direct exsolution of highly concentrated chloride liquids (brine, salt melt, or hydrosaline fluid; see definitions in Halter & Webster 2004) together with low density aqueous vapors (e.g., Cline & Bodnar 1994, Shinohara 1994, Bodnar 1995). It has been demonstrated that the initial melt composition puts strong constraints on “the type of volatile phase that exsolves” and “the relative timing of volatile phase exsolution” (Webster 2004). Given the existing evidence, a 85
V.S. KAMENETSKY
FIG. 4-5. Aqueous globules in quartz-hosted silicate melt inclusions (Davidson & Kamenetsky 2006) from La Copa complex, Chile (A, B) and Taupo Volcanic Zone, New Zealand (C, D). Single-phase aqueous liquid bubbles have elliptical, “deformed” (A, B) or spherical shapes (C, D). Salt crystals are commonly present inside aqueous liquid (C, D). Note that the vapor phase in Fig. 5D (upper close-up) only appeared after laser Raman analysis, prior to this it was single phase aqueous liquid. gl – silicate glass, al – aqueous liquid, v – vapor. Scale bars are 20 µm.
continuum from high-density salt melts to lowdensity aqueous vapors, coexisting with each other and with their “parental” silicate magma, occurs in nature (e.g., Roedder & Coombs 1967, Reyf & Bazheyev 1977, Reyf 1984, 1997, Solovova et al. 1991, Frezzotti 1992, Roedder 1992, Solovova et al. 1992, De Vivo & Frezzotti 1994, Lowenstern 1994, De Vivo et al. 1995, Kamenetsky et al. 1999). Quartz-hosted inclusions from the La Copa magmatic complex, Chile (Davidson et al. 2005, Davidson & Kamenetsky 2006), mineralized Omsukchan granite, NE Russia (Kamenetsky et al. 2004b), dacitic porphyries from the Bajo de la Alumbrera deposit, Argentina (Harris et al. 2003, 2004), granite clasts in the Gawler Craton rhyolites, South Australia, and Panguna Cu–Au porphyry deposits, Bougainville Island, Papua New Guinea, are used below to confirm common silicate melt – salt melt immiscibility in plutonic felsic systems. Primary magmatic inclusions in quartz phenocrysts are aligned along growth planes and
syn-crystallization fractures, and belong to three main types: syngenetic crystals (e.g., feldspar, mica, apatite, zircon), glassy or crystallized silicate melt, and non-silicate volatile-rich phases. Crystallized silicate melt inclusions are commonly surrounded by radiating fractures and halos of tiny vapor-rich aqueous bubbles (Fig. 4-4), suggesting the melt originally had a high volatile content, which was released during post-entrapment crystallization and decrepitation of these inclusions. Volatile-rich inclusions are much more common than silicate melt inclusions and characterized by round to negative crystal shapes and relatively small size (typically <15 µm). They contain variable proportions of aqueous vapor, liquid and nonsilicate crystals, ranging from essentially vapor to dominantly crystalline end-members (Fig. 4-6). Composite inclusions, consisting of trapped combinations of cotectic crystals, silicate melt and aqueous saline fluids (Figs. 4-5A, B, 7), are also common. 86
MELT INCLUSION RECORD OF MAGMATIC IMMISCIBILITY IN CRUSTAL AND MANTLE MAGMAS
FIG. 4-6. Magmatic inclusions of crystallized hydrosaline liquid from the Gawler Craton, South Australia granites (A, B), Omsukchan granite, NE Russia (C, D), Mt. Vesuvius endoskarn (E) and Panguna porphyry, PNG. Note a large-sized vapor bubble (dark) in all inclusions. Host minerals are quartz (A-D, F) and sanidine (E). Scale bars are 20 µm.
Co-trapping of silicate melt with hydrosaline liquid is most evident in the inclusions that escaped post-trapping crystallization of their silicate melt component (i.e., glass; Fig. 4-7B-D). However, in most cases fine-grained crystalline silicate masses inside composite inclusions obscure evidence of non-silicate phases that are present. Heating such inclusions and subsequent quenching (Kamenetsky et al. 2003) are used to make heterogeneously trapped non-silicate phases available for observation and analysis. After heating, melt inclusions consist of clear silicate glass and spherical globules of salt melt (Fig. 4-7E-H). Globules are crowded with microcrystals of chlorides (± anhydrite ± magnetite ± chalcopyrite) and always contain a significant quantity (up to 50 vol.%) of a spherical to deformed
vapor phase (Fig. 4-7B-H). The number (from 1–2 to 100’s) and sizes (<1–15 µm) of globules vary significantly even in neighboring inclusions. Commonly in larger composite inclusions these crystal- and vapor-bearing globules form an emulsion (globules are suspended in the glass “matrix”, Fig. 4-7E-H). This presents a clear case of magmatic immiscibility in which at least two liquids were stable simultaneously during quartz growth. The size of the trapped continuous silicate melt phase is random, depending on the size of the growth irregularities on the surface of the host crystal (Roedder 1984), whereas the size of the trapped globules is suggested to be representative of the actual size of dispersed salt melt phase at the time of immiscibility, and before coalescence (Reyf 1984). 87
V.S. KAMENETSKY
FIG. 4-7. Magmatic inclusions composed of silicate melt (crystallized and glassy - gl) and heterogeneously trapped hydrosaline liquid from the Omsukchan granite, NE Russia (A, E, F), Gawler Craton, South Australia granites (B), Mt Vesuvius endoskarn (C, D), La Copa complex, Chile (G) and Bajo de la Alumbrera porphyry, Argentina. Note cubic crystal, possibly halite (ha?), associated with aqueous fluid in the lower right corner of the inclusion in (A). Numerous spherical globules of dominantly chloride composition suspended in the silicate glass matrix (“microemulsion” in E-H) were observed only after heating (800-850oC) and partial homogenization of the silicate content of these inclusions. Host minerals are quartz (A, B, E-H) and sanidine (C, D). Scale bars are 20 µm.
88
MELT INCLUSION RECORD OF MAGMATIC IMMISCIBILITY IN CRUSTAL AND MANTLE MAGMAS
Morphologically similar hydrosaline globules hosted in quartz (Fig. 4-6) and co-trapped with the silicate melt (Fig. 4-7) are confirmed to be compositionally alike by microthermometric experiments and microbeam analyses (Harris et al. 2003, 2004, Kamenetsky et al. 2004b). The globules are all metal- (1000’s–10000’s ppm) and chloride-rich (Fig. 4-8), although the element abundances and ratios vary significantly, even for co-trapped globules (Kamenetsky et al. 2004b). Compositional variability among hydrosaline globules suggests strong fractionation of most elements between immiscible liquids and the disequilibrium character of exsolution. The latter means that if immiscibility is a continuous process in highly evolved magmatic systems, the components of a dispersed hydrosaline phase must have varying composition, because of the variability of diffusion rates for different elements. In the residual granitic system, where crystallization and immiscibility drive chemical fractionation to the extreme, chemical disequilibrium may occur on very small spatial and temporal scales, and is maintained by slow element diffusion and delayed mixing in a relatively cold, viscous and strongly crystalline environment. Identification of immiscible silicate melts and hydrosaline liquids is the first important step in linking typical aluminosilicate magmas and typical hydrothermal chloride solutions. For example, zoned quartz phenocrysts from the mineralized Panguna porphyries, Papua-New Guinea, show systematic distribution of magmatic inclusions from the dominantly silicate melt in the cores to domin-
antly aqueous saline fluid inclusions in the rims (Fig. 4-9). Another example from the Bajo de la Alumbrera porphyry records silicate and salt melt inclusions occurring together within quartz phenocrysts, as well as mineralized (chalcopyrite, bornite) quartz veins (Fig. 4-7H; Harris et al. 2003, 2004). The hydrosaline liquid inclusions in the vein quartz were interpreted to be the most primitive and copper-rich ore fluids identified to date, exsolved from the crystallizing melt at ~100 MPa. These inclusions appear to represent a “broth” of the silicate melt and hydrosaline metal-enriched fluids, characteristic of the magmatic to hydrothermal transition that is a prerequisite to the formation of economic deposits. Immiscible silicate melt and carbonic (CO2, carbonate) fluids Natural and experimental carbonate-silicate systems show almost ubiquitous evidence of multiphase liquid/fluid immiscibility involving different silicate and carbonate melts; as well as CO2 fluid and chloride, sulfate and phosphate phases (see reviews in Roedder 1994, Frezzotti 2001, Lowenstern 2001, Kamenetsky et al. 2002a, Veksler 2004, Panina 2005). Compelling evidence for this type of immiscibility comes from the studies of melt and fluid inclusions in diamonds, peridotite xenoliths, alkaline intrusive complexes, and carbonatites. In contrast to widespread occurrence of the silicate-CO2-carbonate immiscibility in mantle-derived rocks and magmas, this type of unmixing in the silicic crustal magmas is by far less pronounced (see reviews in De Vivo & Frezzotti
FIG. 4-8. Optical image and proton-induced X-ray emission (PIXE) element maps of a heated composite inclusion in quartz from the Omsukchan granite, NE Russia (Kamenetsky et al. 2004). The inclusion is composed of clear silicate glass and a heterogeneously trapped globule of hydrosaline liquid. Outlines on element maps mark boundaries of the inclusion and the globule. Inclusion size is ~25 µm.
89
V.S. KAMENETSKY
FIG. 4-9. Zoned quartz phenocryst in the Panguna porphyry, PNG and its magmatic inclusions present in different growth zones. The phenocryst core is dominated by crystallized silicate melt inclusions (A). The magmatic-hydrothermal transition is recorded in increasing numbers of hydrosaline (B) and brine fluid (C) inclusions towards the phenocrysts rims. Inclusion sizes are ~20-25 µm.
hosting the Timbarra Au deposit (Mustard et al. 2003, 2006) showed that metal enrichment occurred as a result of crystal fractionation (enrichment factors: 40 for Au, 5–20 for Mo, W, Bi, As and Sb, and 1–3 for Zn, Pb and Cu). Individual inclusions of the crystallized silicate melt in the most fractionated Timbarra rocks contain daughter ore minerals (gold, senarmontite (Sb2O3), scheelite (CaWO4), wulfenite (PbMoO4), molybdenite (MoS2), sphalerite (ZnS) and stibnite (Sb2S3)). Importantly, magmatic inclusions do not record immiscibility between the silicate melt and Cl-bearing fluids; instead the magmatic fluids are CO2-rich (Fig. 4-10A) with up to 12.6% N2 and 1.2% CH4. Such chemically “inert” fluids were largely responsible for preferential partitioning of metals into the silicate melt, and thus crystallization-related metal enrichment in the residual magmas (Mustard et al. 2006).
1994, Lowenstern 2001). The data available to the author show coexistence of silicate melt and dense CO2 fluid (Fig. 4-10) among magmatic inclusions in quartz from granitic clasts in the Gawler Craton rhyolites, South Australia and the Timbarra granite, New England Fold Belt, East Australia (Mustard et al. 2003, 2006), and in the silicate phenocrysts from rocks forming marginal parts of the Mt. Vesuvius magma chamber (Fulignati et al. 2001). The latter example offers an opportunity to study fluids involved in magma-wall rock reactions. The CO2-rich composition of granitederived magmatic fluids is of particular importance in genetic modeling of intrusion-related deposits of Au (e.g., Timbarra) and lithophile elements (e.g., Fe–Cu–Au–U–REE mineralization in the eastern Gawler Craton). Melt inclusion studies performed on a composite 250 Ma I-type granite pluton 90
MELT INCLUSION RECORD OF MAGMATIC IMMISCIBILITY IN CRUSTAL AND MANTLE MAGMAS
FIG. 4-10. Magmatic inclusions composed of silicate melt (crystallized and glassy) and high-density CO2 fluid (shown by arrows) from Timbarra granite, New England Fold Belt, East Australia (A), Mt. Vesuvius endoskarn (B), and Gawler Craton, South Australia granites (C, D). Host minerals are quartz (A, C, D) and nepheline (B). Scale bars are 10 µm.
granite, NE Russia (Kamenetsky et al. 2002b, 2004b) and Udachnaya East pipe kimberlite, Siberia (Kamenetsky et al. 2004a, 2006) demonstrate the role of heating experiments with melt inclusions in deciphering immiscibility processes in deep-seated carbonate-chloride liquids. Multiphase melt inclusions in nepheline from the endoskarn xenoliths of the A.D. 472 Mt. Vesuvius eruption, representing the magma chamber – carbonate wall-rock interface, are composed of Na–Ca carbonates, Na–K chlorides and minor Ca– and Na–K sulfates and Fe–Cu–Zn–Pb sulfides (Fulignati et al. 2001). On heating at 1 atm. they show complete melting of daughter crystals at 800–830°C, followed by vapor-bubble homogenization between 860 and 885 °C, although some solids (possibly heterogeneously trapped silicates) may persist up to higher temperatures. Cooling of homogenized inclusions always results in immiscibility between two clear liquids (Fig. 4-11) at ~670–700°C. One new melt forms either a single spherical globule around a vapor bubble (Fig. 4-11A) or several globules that later coalesce
Immiscible non-silicate melts Cooling, crystallization and unmixing of silicate magmas inevitably results in the exhaustion of the aluminosilicate components, and separation of essentially non-silicate volatile-rich phases from the aluminosilicate solids at, and below, the silicate magma solidus. Very little is known about these residual phases, especially in the plutonic environment, because they are highly fugitive and reactive in nature, and prone to chemical modifications via boiling, mixing, crystallization and interaction with wall-rocks. Unmixing of these residual non-silicate phases into compositionally more simple components is expected, based on experimental and melt inclusion studies (e.g., Panina 2005, and references therein), and a variety of rocks (different carbonatites, skarns and pegmatites) and styles of alteration and mineralization can be genetically linked to ephemeral, still hypothetical, syn- and postmagmatic fluids and melts. Examples below from the endoskarn samples of Mt. Vesuvius (Fulignati et al. 2001, 2005); miarolitic quartz in the Omsukchan
91
V.S. KAMENETSKY
FIG. 4-11. Photomicrographs illustrating cooling behavior of carbonate-chloride melt inclusions in nepheline from the Mt. Vesuvius endoskarn (Fulignati et al. 2001). Note formation and increase in size of a single (A) or multiple (B) globules of immiscible chloride melt. The arrow points to a sulfide globule formed at the carbonate-chloride meniscus. Inclusion sizes are ~30-35 µm.
(Fig. 4-11B). Repeated heating and cooling of the same inclusion show that the immiscibility is reversible at a constant temperature (~670-700°C). Below 550°C spontaneous crystallization within the two unmixed liquids obscures original phase boundaries and deforms the vapor bubble. These inclusions were interpreted to reflect compositions formed as a result of interactions between hightemperature, magmatic-derived hydrosaline liquids and carbonate country rocks. The liquids they represent are important in transferring elements into skarn environments, and the observed carbonatechloride melt immiscibility, if it occurs in nature, can account for specific metasomatic reactions (carbonate-related and chloride-related) and mineral assemblages in the ore deposits. Miarolitic quartz from the tin-mineralized Omsukchan granite intrusion, NE Russia, contains several populations of spectacular inclusions of crystallized hydrosaline liquids (Kamenetsky et al. 2002b, 2004b). The daughter phase assemblage in these inclusions is, without doubt, the most complex that an investigator may come across. Cubic, pseudocubic, hexagonal, pyramidal, and prismatic crystals are common, as well as minerals with high relief and birefringence. However, there is no order
or pattern to the alignment of the crystals and no clear relationships between the shape, color (except for hematite plates), birefringence, and relief of crystals. Even at room temperature, the contents of these inclusions are metastable and prone to spontaneous recrystallization followed by changes in the number, shape and volume ratios of phases. In addition to common halite, sylvite and hematite, at least 10 different solids have been distinguished by Raman spectroscopy. Most of them exhibited very intense OH-stretching bands between 3400 and 3450 cm–1, however, the overall Raman spectra do not correspond to those of the common minerals. No carbonate, sulfate, phosphate, nitrate, or borate bands were observed in these spectra, but antarcticite (CaCl2·6H2O) and Sn chloride (possibly abhurite Sn3O(OH)2Cl2) have been tentatively identified. The PIXE analysis of these inclusions showed significant Br, Fe, Mn, Zn and Pb enrichment, associated with individual daughter crystals (Kamenetsky et al. 2002b), and the LA– ICP–MS analysis confirmed that Cl is the major anion. Kamenetsky et al. (2002b) suggested that the inclusions are largely composed of complex hydrous chlorides.
92
MELT INCLUSION RECORD OF MAGMATIC IMMISCIBILITY IN CRUSTAL AND MANTLE MAGMAS
The inclusions in miarolitic quartz homogenize (except for a few newly formed tiny crystals) into liquid at 550–650oC. On cooling they remain homogeneous for ~100–150oC below homogenization temperature, and then undergo spontaneous unmixing into at least two liquids and nucleation of several vapor bubbles (Fig. 4-12). At the onset of immiscibility one liquid forms wormlike segregations that float freely in the matrix of another liquid, change their shape and size continuously, until they finally coalesce at ~320– 350°C (Fig. 4-12). With further cooling the boundary between two liquids (colloids?) fades out, and spontaneous changes in the phase composition (possibly coagulation of gels) in both phases occurs within a few degrees at 120–140oC. Multiphase melt inclusions in the groundmass olivine from the Udachnaya East kimberlite consist of three principal components – calcite, Na–K–Ca carbonate and Na–K chloride minerals (Golovin et al. 2003, Kamenetsky et al. 2004a, 2006). Other daughter phases, such as olivine, phlogopite-tetraferriphlogopite, Fe–Ti–Cr oxides, aphthitalite (K3Na(SO4)2) and djerfisherite (K6(Fe,Ni,Cu)24S26Cl), are present in subordinate abundances. Homogenization of these inclusions occurs by vapor bubble disappearance at 660– 760°C. Cooling to 610–580°C results in a spontaneous process in which inclusions acquire a ‘foggy” appearance but only momentarily. This process can best be described as the formation of an emulsion, i.e, microglobules of one liquid in another (i.e., melt immiscibility). Microglobules coalesce immediately into larger, elongate, sausage-like globules (Fig. 4-13B). The neighboring globules (“boudins”) are subparallel, and are grouped into
regularly aligned formations with a common angle of ~75–80o (Fig. 4-13B). A resemblance to the skeletal or spinifex texture is evident for several seconds, after which the original “pinch-and-swell structure” pulls apart giving rise to individual blebs of melt. The latter coalesce and become more spherical with time or further cooling (Fig. 4-13C). The immiscible phases are recognized as carbonateand chloride-dominated, respectively, on the basis that these minerals are dominantly present in the unheated melt inclusions. Remarkable textures, observed in kimberlitic melt inclusions at the exact moment of melt unmixing (Fig. 4-13B), are governed by the carbonate crystallographic properties. Similar textures, but on a much larger scale, are recognized in the round chloride-carbonate segregations (nodules, 5–30 cm across) in the same kimberlite rocks. The nodules are composed of regularly interspersed layers (sheets) of carbonates and chlorides (Fig. 4-13D). The groups of aligned carbonate sheets make up larger (2–2.5 cm) formations that appear as isosceles triangles in cross-sections or as rhombohedra in three dimensions. The surfaces of individual carbonate sheets are locally flat, but more typically bumpy, or boudin-like. The thickness of boudins rarely exceeds 1–1.5 mm, and symmetrical zoning is visible at the cross sections of inflated parts. Sugary aggregates of chlorides appear to fill in the intrasheet spaces and cracks in the carbonates. The chlorides are dominated by halite, whereas round and amoeboid blebs of sylvite in halite often show textures reminiscent of liquid immiscibility rather than unmixing of solid solution (Kamenetsky et al. 2006).
FIG. 4-12. Photomicrographs showing cooling behavior of a hydrosaline fluid inclusion in miarolitic quartz from the Omsukchan granite (NE Russia). This and other co-trapped inclusions homogenize at ~550oC and remain homogeneous on cooling until immiscibility occurs at ~425oC. See text for details. Inclusion size is 50 µm.
93
V.S. KAMENETSKY
FIG. 4-13. Photomicrographs demonstrating carbonate-chloride immiscibility in an olivine-hosted melt inclusion (A-C) and residual melt pocket (“nodule”) from the Udachnaya East kimberlite, Siberia (Kamenetsky et al. 2006). Note textural similarity between olivine-hosted melt inclusion at the moment of immiscibility (B) and chloride-carbonate nodule (D). See text for details. Inclusion size is 50 µm.
The presence of similar textures in olivinehosted melt inclusions at ~600oC and chloridecarbonate nodules in the kimberlite groundmass (Fig. 4-13) is the first record of unambiguous chloride-carbonate melt immiscibility in mantlederived magmas. The preservation of magmatic immiscibility among residual non-silicate components of the original kimberlite magma correlates well with the enrichment of the Udachnaya East kimberlite in alkali elements and chlorine (2.3–3.2 wt.% Cl, 2.6–3.7 wt.% Na, and 1.6–2.0 wt.% K). Such enrichment is inherited from the kimberlite parental/primary magma, and it can be responsible for kimberlite’s low liquidus temperatures, low viscosities, and rapid ascent, as well as potential catalytic effects on the growth of diamonds. Under mantle conditions, the chlorideand carbonate-rich kimberlite would be highly volatile, mobile, reactive, and capable of pervasively percolating and wetting ambient peridotite. Thus, such kimberlite magmas, kimberlite-derived chloride-carbonate melts, and their immiscible products, may play a previously unrecognized role as potent metasomatic agents in both mantle and crust.
CONCLUSIONS Immiscible magmatic liquids and vapors trapped as melt and fluid inclusions in phenocrysts are the closest approximation of naturally exsolved volatile-rich phases. Such inclusions can be used as a natural experimental laboratory to model volatile phase exsolution from cooling and crystallizing magmas (a proxy for some large-scale magma chamber processes, such as sequestering of metals and degassing), and thus may make possible quantification of partitioning of metals and volatiles. Since magmatic immiscibility and consequent separation of a volatile-rich phase from a cooling silicate magma is a keystone of the orthomagmatic models of ore formation, more insights into late magmatic–early hydrothermal processes and phases are required. If we are to apply melt/fluid inclusion studies to unravel the details of immiscibility event(s), then the choice of samples becomes critical. Apart from technical problems (e.g., presence of phenocrysts, availability and preservation of inclusions) we face a dilemma, which rocks are most suitable for such a study? At first glance, the choice of rocks spatially and 94
MELT INCLUSION RECORD OF MAGMATIC IMMISCIBILITY IN CRUSTAL AND MANTLE MAGMAS
temporally related to mineralization (e.g., host intrusions) seems most attractive, because such rocks are interpreted as originating from magmas that supplied the volatiles and ore-forming elements. However, if this viewpoint is correct, such rocks (and likely their melt inclusions) must be depleted in volatiles and metals relative to the precursor (parental) magma. Moreover, in such rocks the magmatic signature is commonly overprinted by later hydrothermal events. On the other hand, rocks cogenetic with those having a demonstrated direct relationship to mineralization, but without a proven link of their own, may be a case in which the evolution from silicate magmas to mineralizing fluids has not gone to completion. By studying examples in which immiscibility does not go all the way to ore formation, it should be possible to use the melt/fluid inclusion approach to record consecutive “snapshots” of the immiscibility processes and production of immiscible phases. The first occurrence of immiscibility in magmas appears to be most important in the magmatic-hydrothermal transition, and thus our studies of magmatic immiscibility should be primarily directed towards recognition of coexisting silicate melt and essentially non-silicate melts or fluids.
Jim Webster for inviting this contribution and editorial handling. My studies were made possible by financial support (research grants and fellowships) from the Australian Research Council over a period of years.
ACKNOWLEDGEMENTS This work was inspired by Edwin Roedder, who first recognized the significance of approaching immiscibility processes and immiscible phases by studying naturally trapped inclusions. Special thanks go to Maya Kamenetsky for her enormous help with the melt inclusion studies – almost every sample described here was skillfully prepared by her. Maya also provided me with her unpublished data on kimberlites. I am grateful to Paul Davidson for his great contribution to this work, for many fruitful discussions, and for help in editing this manuscript. My immiscibility studies have benefited from valuable contributions from many people, in particular, Sharon Allen, Adam Bath, Dave Braxton, Tony Crawford, Leonid Danyushevsky, Benedetto De Vivo, Steve Eggins, Paolo Fulignati, Anthony Harris, Dave Lentz, Jake Lowenstern, Roland Maas, Paola Marianelli, Terry Mernagh, Nicole Metrich, Roger Mustard, Vladimir Naumov, Oded Navon, Chris Ryan, Viktor Sharygin, Sergey Smirnov, Alex Sobolev, Irina Solovova, Rainer Thomas, Ilya Veksler, Jim Webster and Greg Yaxley. The manuscript was thoroughly reviewed and edited by Harvey Belkin, Charles Mandeville and Jim Webster. I also thank
CANDELA, P.A. & PICCOLI, P.M. (1995). Model ore-metal partitioning from melts into vapor and vapor/brine mixtures. In Magmas, Fluids, and Ore Deposits. (J.F.H. Thompson, ed.). Mineral. Assoc. Canada (101–127).
REFERENCES BODNAR, R.J. (1995). Fluid-inclusion evidence for a magmatic source of metals in porphyry copper deposits. In Magmas, Fluids, and Ore Deposits. (J.F.H. Thompson, ed.). Mineral. Assoc. Canada (139–152). BODNAR, R.J. & VITYK, M.O. (1994): Interpretation of microthermometric data for H2O-NaCl fluid inclusions. In Fluid inclusions in minerals: methods and applications. (B. De Vivo & M.L. Frezzotti, eds.). Virginia Tech. (117–130). BURNHAM, C.W. (1979). Magmas and hydrothermal fluids. In Geochemistry of Hydrothermal Ore Deposits. (H.L. Barnes, ed.). J.Wiley & Sons, New York (71–136). CANDELA, P.A. (1989). Felsic magmas, volatiles and metallogenesis. In Ore Deposition Associated with Magmas (J.A. Whitney & A.J. Naldrett, eds). (223–233).
CLINE, J.S. & BODNAR, R.J. (1994): Direct evolution of brine from a crystallizing silicic melt at the Questa, New Mexico, molybdenum deposit. Econ. Geol. 89, 1780–1802. DAVIDSON, P. & KAMENETSKY, V. (2001): Immiscibility and continuous melt-fluid evolution within the Rio Blanco porphyry system, Chile: Evidence from inclusions in magmatic quartz. Econ. Geol. 96, 1921–1929. DAVIDSON, P. & KAMENETSKY, V. (2006): Primary aqueous fluids in rhyolitic magmas: melt inclusion evidence for pre- and post-trapping exsolution. Chem. Geol. (in review). DAVIDSON, P., KAMENETSKY, V., COOKE, D.R., FRIKKEN, P., HOLLINGS, P., RYAN, C., VAN ACHTERBERGH, E., MERNAGH, T., SKARMETA, J., SERRANO, L. & VARGAS, R. (2005): Magmatic precursors of hydrothermal fluids at
95
V.S. KAMENETSKY
the Rio Blanco Cu–Mo deposit, Chile: links to silicate magmas and metal transport. Econ. Geol. 100, 963–978.
evidence of second immiscibility within magmatic fluids (79AD eruption of Mt. Vesuvius). Per. Mineral. LXXIV, 43–54.
DE, A. (1974): Silicate liquid immiscibility in the Deccan Traps and its petrogenetic significance. Geol. Soc. Am. Bull. 85, 471–474.
FULIGNATI, P., KAMENETSKY, V.S., MARIANELLI, P., SBRANA, A. & MERNAGH, T.P. (2001): Melt inclusion record of immiscibility between silicate, hydrosaline and carbonate melts: Applications to skarn genesis at Mount Vesuvius. Geology 29, 1043–1046.
DE VIVO, B. & FREZZOTTI, M.L. (1994): Evidence for magmatic immiscibility in Italian subvolcanic systems. In Fluid inclusions in minerals: methods and applications. (B. De Vivo & M.L. Frezzotti, eds.). Virginia Tech. (345– 362).
GOLOVIN, A.V., SHARYGIN, V.V., POKHILENKO, L.N., MAL'KOVETS, V.G., KOLESOV, B.A. & SOBOLEV, N.V. (2003): Secondary melt inclusions in olivine from unaltered kimberlites of the Udachnaya East pipe, Yakutia. Dokl. Earth Sci. 388, 93–96.
DE VIVO, B., TOROK, K., AYUSO, R.A., LIMA, A. & LIRER, L. (1995): Fluid inclusion evidence for magmatic silicate/saline/CO2 immiscibility and geochemistry of alkaline xenoliths from Ventotene Island, Italy. Geochim. Cosmochim. Acta 59, 2941–2953.
HALTER, W.E. & WEBSTER, J.D. (2004): The magmatic to hydrothermal transition and its bearing on ore-forming systems. Chem. Geol. 210, 1–6.
DIXON, S. & RUTHERFORD, M.J. (1979): Plagiogranites as late-stage immiscible liquids in ophiolite and mid-ocean ridge suites: an experimental study. Earth Planet. Sci. Lett. 45, 45–60.
HARRIS, A.C., KAMENETSKY, V.S., WHITE, N.C. & STEELE, D.A. (2004): Volatile phase separation in silicic magmas at Bajo de la Alumbrera porphyry Cu–Au deposit, NW Argentina. Res. Geol. 54, 341–356.
DUNBAR, N.W. & HERVIG, R.L. (1992): Volatile and trace element composition of melt inclusions from the Lower Bandelier Tuff: Implications for magma chamber processes and eruptive style. J. Geophys. Res. 97, 15151– 15170.
HARRIS, A.C., KAMENETSKY, V.S., WHITE, N.C., VAN ACHTERBERGH, E. & RYAN, C.G. (2003): Melt inclusions in veins: Linking magmas and porphyry Cu deposits. Science 302, 2109–2111. HEINRICH, C.A., GÜNTHER, D., AUDÉTAT, A., ULRICH, T. & FRISCHKNECHT, R. (1999): Metal fractionation between magmatic brine and vapor, determined by microanalysis of fluid inclusions. Geology 27, 755–758.
DUNBAR, N.W. & KYLE, P. (1993): Lack of volatile gradient in the Taupo plinian-ignimbrite transition: evidence from melt inclusion analysis. Am. Mineral. 78, 612–618.
JAKOBSEN, J.K., VEKSLER, I.V., TEGNER, C. & BROOKS, C.K. (2005): Immiscible iron- and silica-rich melts in basalt petrogenesis documented in the Skaergaard intrusion. Geology 33, 885–888.
DUNBAR, N.W., HERVIG, R.L. & KYLE, P.R. (1989): Determination of pre-eruptive H2O, F and Cl contents of silicic magmas using melt inclusions: examples from Taupo volcanic center, New Zealand. Bull. Volcanol. 51, 177– 184.
KAMENETSKY, M.B., SOBOLEV, A.V., KAMENETSKY, V.S., MAAS, R., L.V., THOMAS, R., DANYUSHEVSKY, POKHILENKO, N.P. & SOBOLEV, N.V. (2004a): Kimberlite melts rich in alkali chlorides and carbonates: A potent metasomatic agent in the mantle. Geology 32, 845–848.
FREZZOTTI, M.L. (1992): Magmatic immiscibility and fluid phase evolution in the Mount Genis granite (southeastern Sardinia, Italy). Geochim. Cosmochim. Acta 56, 21–33. FREZZOTTI, M.L. (2001): Silicate-melt inclusions in magmatic rocks: applications to petrology. Lithos 55, 273–299.
KAMENETSKY, V.S. & DANYUSHEVSKY, L.V. (2005): Metals in quartz-hosted melt inclusions: Natural facts and experimental artefacts. Am. Mineral. 90, 1674–1678.
FULIGNATI, P., KAMENETSKY, V.S., MARIANELLI, P. & SBRANA, A. (2005): Fluid inclusion 96
MELT INCLUSION RECORD OF MAGMATIC IMMISCIBILITY IN CRUSTAL AND MANTLE MAGMAS
(J.F.H. Thompson, ed.). Mineral. Assoc. Canada (71–99).
KAMENETSKY, V.S., DAVIDSON, P., MERNAGH, T.P., CRAWFORD, A.J., GEMMELL, J.B., PORTNYAGIN, M.V. & SHINJO, R. (2002a): Fluid bubbles in melt inclusions and pillow-rim glasses: high-temperature precursors to hydrothermal fluids? Chem. Geol. 183, 349–364.
LOWENSTERN, J.B. (2001): Carbon dioxide in magmas and implication for hydrothermal systems. Mineral. Depos. 36, 490–502. LOWENSTERN, J.B. (2003). Melt inclusions come of age: Volatiles, Volcanoes, and Sorby’s Legacy. In Developments in Volcanology, 5. Melt inclusions in volcanic systems: methods, applications and problems. (B. De Vivo & R.J. Bodnar, eds.). Elsevier, Amsterdam (1–21).
KAMENETSKY, V.S., DE VIVO, B., NAUMOV, V.B., KAMENETSKY, M.B., MERNAGH, T.P., VAN ACHTERBERGH, E., RYAN, C.G. & DAVIDSON, P. (2003). Magmatic inclusions in the search for natural silicate-salt melt immiscibility: methodology and examples. In Developments in Volcanology, 5. Melt inclusions in volcanic systems: methods, applications and problems. (B. De Vivo & R.J. Bodnar, eds.). Elsevier, Amsterdam (65–82).
MCBIRNEY, A.R. (1980): Mixing and unmixing of magma. J. Volcanol. Geotherm. Res. 7, 357– 371. MUSTARD, R., KAMENETSKY, V.S. & MERNAGH, T.P. (2003): Carbon dioxide bearing melt inclusions within a gold-mineralized felsic granite. In Mineral Exploration and Sustainable Development, Proceedings of the 7th Biennial SGA Meeting, vol. 1. (D.G. Eliopoulos, et al., eds.). Millpress, Rotterdam (351–354).
KAMENETSKY, V.S., KAMENETSKY, M.B., SHARYGIN, V.V., FAURE, K. & GOLOVIN, A.V. (2006): Chloride and carbonate immiscible liquids at the closure of the kimberlite magma evolution (Udachnaya East kimberlite, Siberia). Chem. Geol. (in review). KAMENETSKY, V.S., NAUMOV, V.B., DAVIDSON, P., VAN ACHTERBERGH, E. & RYAN, C.G. (2004b): Immiscibility between silicate magmas and aqueous fluids: a melt inclusion pursuit into the magmatic-hydrothermal transition in the Omsukchan Granite (NE Russia). Chem. Geol. 210, 73–90.
MUSTARD, R., ULRICH, T., KAMENETSKY, V.S. & MERNAGH, T. (2006): Gold and metal enrichment in natural granitic melts during fractional crystallization. Geology 34, 85–88. NAUMOV, V. (1979): Determination of concentrateion and pressure of volatiles in magmas from inclusions in minerals. Geochem. Inter. 16, 33–40.
KAMENETSKY, V.S., VAN ACHTERBERGH, E., RYAN, C.G., NAUMOV, V.B., MERNAGH, T.P. & DAVIDSON, P. (2002b): Extreme chemical heterogeneity of granite-derived hydrothermal fluids: An example from inclusions in a single crystal of miarolitic quartz. Geology 30, 459– 462.
NAUMOV, V., KOVALENKER, V.A., RUSINOV, V.L. & KONONKOVA, N.N. (1994): Inclusions of high-density magmatic water in phenocrysts from acid volcanics of the Western Carpathians and Central Tien Shan. Petrology 2, 480–494. NAUMOV, V., KOVALENKER, V., RUSINOV, V., SOLOVOVA, I. & KAMEN, M. (1992): High density fluid inclusions of magmatic water in phenocrysts from rhyolite of the Stiavnica Stratovolcano (central Slovakia). Geol. Sbornik– Geol. Carpath. 43, 85–89.
KAMENETSKY, V.S., WOLFE, R.C., EGGINS, S.M., MERNAGH, T.P. & BASTRAKOV, E. (1999): Volatile exsolution at the Dinkidi Cu-Au porphyry deposit, Philippines: A melt-inclusion record of the initial ore-forming process. Geology 27, 691–694.
NAUMOV, V., TOLSTYKH, M.L., KOVALENKER, V.A. & KONONKOVA, N.N. (1996): Fluid overpressure in andesite melts from Central Slovakia: Evidence from inclusions in minerals. Petrology 4, 265–276.
LOWENSTERN, J.B. (1994): Chlorine, fluid immiscibility, and degassing in peralkaline magmas from Pantelleria, Italy. Am. Mineral. 79, 353–369.
PANINA, L.I. (2005): Multiphase carbonate-salt immiscibility in carbonatite melts: data on melt inclusions from the Krestovskiy massif minerals (Polar Siberia). Contrib. Mineral. Petrol. 150,
LOWENSTERN, J.B. (1995). Applications of silicatemelt inclusions to the study of magmatic volatiles. In Magmas, Fluids, and Ore Deposits.
97
V.S. KAMENETSKY
19–36. SOLOVOVA, I.P., GIRNIS, A.V., NAUMOV, V.B., KOVALENKO, V.I. & GUZHOVA, A.V. (1991): Mechanism of degassing of silicic magmas: formation of two fluid phases during pantellerite crystallisation, Pantelleria. Trans. (Doklady) USSR Acad. Sci. 320, 982–985.
SOLOVOVA, I.P., GIRNIS, A.V., GUZHOVA, A.V. & NAUMOV, V.B. (1992): Magmatic salt inclusions in minerals from eastern Pamirs alkaline basalts. Geokhimiya 1, 68-77. SOLOVOVA, I.P., GIRNIS, A.V., NAUMOV, V.B., KOVALENKO, V.I. & GUZHOVA, A.V. (1991): Mechanism of degassing of silicic magmas: formation of two fluid phases during pantellerite crystallisation, Pantelleria. Trans. (Doklady) USSR Acad. Sci. 320, 982-985.
PHILPOTTS, A.R. (1982): Compositions of immiscible liquids in volcanic rocks. Contrib. Mineral. Petrol. 80, 201-218. REYF, F.G. (1984): Microemulsion state of fluid saturated granitic magmas: characteristics and petrological implications. Trans. (Doklady) USSR Acad. Sci. 276, 1197-1201.
STUDENT, J.J. & BODNAR, R.J. (1999): Synthetic fluid inclusions XIV: Coexisting silicate melt and aqueous fluid inclusions in the haplogranite–H2O–NaCl–KCl system. J. Petrol. 40, 1509–1525.
REYF, F.G. (1997): Direct evolution of W-rich brines from crystallizing melt within the Mariktkan granite pluton, west Transbaikalia. Mineral. Depos. 32, 475-490.
THOMAS, R. & WEBSTER, J.D. (2000): Strong tin enrichment in a pegmatite-forming melt. Mineral. Depos. 35, 570–582.
REYF, F.G. & BAZHEYEV, Y.D. (1977): Magmatic chloride solutions and tungsten mineralizations. Geochem. Inter. 14, 45-51.
THOMAS, R., WEBSTER, J.D. & HEINRICH, W. (2000): Melt inclusions in pegmatite quartz: Complete miscibility between silicate melts and hydrous fluids at low pressure. Contrib. Mineral. Petrol. 139, 394–401.
ROEDDER, E. (1984): Fluid inclusions, Reviews in Mineralogy, vol. 12: Book Crafters Inc. ROEDDER, E. (1992): Fluid inclusion evidence for immiscibility in magmatic differentiation. Geochim. Cosmochim. Acta 56, 5-20.
VEKSLER, I.V. (2004): Liquid immiscibility and its role at the magmatic-hydrothermal transition: a summary of experimental studies. Chem. Geol. 210, 7–31.
ROEDDER, E. (1994): Fluid inclusion evidence of mantle fluids. In Fluid inclusions in minerals: methods and applications. (B. De Vivo & M.L. Frezzotti, eds.). Virginia Tech. (283-295).
WEBSTER, J.D. (2004): The exsolution of magmatic hydrosaline chloride liquids. Chem. Geol. 210, 33–48.
ROEDDER, E. (2003). Significance of melt inclusions. In Developments in Volcanology, 5. Melt inclusions in volcanic systems: methods, applications and problems. (B. De Vivo & R.J. Bodnar, eds.). Elsevier, Amsterdam (xv-xvi).
WILLIAMS, T.J., CANDELA, P.A. & PICCOLI, P.M. (1995): The partitioning of copper between silicate melts and two-phase aqueous fluids; an experimental investigation at 1 kbar, 800oC and 0.5 kbar, 850oC. Contrib. Mineral. Petrol. 121, 388–399.
ROEDDER, E. & COOMBS, D.S. (1967): Immiscibility in granitic melts, indicated by fluid inclusions in ejected granitic blocks of Ascension Island. J. Petrol. 8, 417-451. ROEDDER, E. & WEIBLEN, P.A. (1970): Silicate immiscibility in lunar magmas, evidenced by melt inclusions in lunar rocks. Science 167, 641644. SHINOHARA, H. (1994): Exsolution of immiscible vapor and liquid phases from a crystallizing silicate melt: Implications for chlorine and metal transport. Geochim. Cosmochim. Acta 58, 52155221. 98
CHAPTER 5: CRYSTALLIZED MELT INCLUSIONS IN GABBROIC ROCKS Ilya V. Veksler GeoForschungsZentrum Potsdam, Department 4.1, Telegrafenberg, Potsdam 14473, Germany E-mail:
[email protected] nature, there are also psychological biases, the burden of conventional truths, and local scientific traditions. Very often people see only what they are prepared to see. The discovery of silicate liquid immiscibility, at first in melt inclusions from lunar samples brought by Apollo missions, and afterwards in ordinary terrestrial basalts (Roedder 1984a, b), is a vivid historic example of roundabout ways that scientific progress may take. Keen observers like H.C. Sorby or E. Roedder are admired so much, and have such a durable scientific impact (Lowenstern 2003; see also Bodnar & Student 2006) probably because of their ability to look at rock specimens with open minds. Tiny inclusions have often been ignored simply because they are small and look unimportant. But let us look at them closer and try to analyze them with proper analytical tools before we issue our assessment. It is impossible to know a priori or after a brief look at a thin section what the inclusions actually are: trivial crystal inclusions, or “fossilized” melt droplets carrying indispensable information about the parental magma. I wish I could present more published data on gabbroic melt inclusions here, but unfortunately I managed to find only a few dozens of papers, and some of them contain dubious results or arrive at dubious conclusions. Because of that, this contribution largely deals with theoretical considerations; it also reflects on the potential scientific output of melt inclusion research aimed at better understanding of gabbroic rocks. Hopefully, the discussion of theoretical principles presented here, and a few recent examples of successful applications of melt inclusion techniques to gabbros will inspire younger scientists, and encourage new studies in the future. The chapter primarily deals with silicate melt inclusions. It should be noted, however, that other types of magmatic fluids, such as the H2O– CO2 gas, strongly concentrated magmatic brines, immiscible sulfide melts, and maybe also Fe–Ti–P oxide nelsonitic liquids do form inclusions in
INTRODUCTION The greatest scientific value of melt inclusion research lies in its ability to constrain the compositions of parental magmatic liquids, and reveal the trends of liquid evolution in specific suites of natural igneous rocks. The cumulative number of melt inclusion papers has been growing exponentially during the last three decades (Bodnar & De Vivo 2003), and many have dealt with the most voluminous basaltic liquids, which play an exceptionally important role in magmatic differentiation of the Earth and other terrestrial planets. Although it is fair to conclude that melt inclusions have “come of age” in volcanic systems (Lowenstern 2003), the research on melt inclusions in mafic plutonic rocks is still in its infancy. The number of papers on melt inclusions in gabbros is at least 10 times smaller than the number of published studies of melt inclusions in the basaltic volcanic equivalents. Furthermore, many researchers express serious doubts about the very existence of gabbroic melt inclusions. The principles of identification and interpretation of crystallized melt inclusions in plutonic products of basaltic magma remain unclear. Much has been written already about the difficulty of recognizing fully crystallized melt inclusions in plutonic rocks (e.g., Roedder 1984a, Bodnar & Student 2006). The difficulty is probably greater in gabbroic rocks than in diorite, syenite, or granite, because of the broader temperature interval, in which basaltic magma crystallizes. Chemical compositions of basaltic liquids change more extensively in the course of magma crystallization than the composition of semi-eutectic syenitic or granitic melts, and prolonged magma evolution is likely to result in a diverse sequence of daughter mineral assemblages in crystallized melt inclusions. The identification and interpretation of mafic melt inclusions are further complicated by host–liquid reactions and other post-entrapment phenomena, which are examined in this chapter in detail. Apart from the difficulties imposed by
Mineralogical Association of Canada Short Course 36, Montreal, Quebec, p. 99-122
99
ILYA V. VEKSLER
gabbroic rocks (Ballhaus & Stumpfl 1986, Larsen et al. 1992, Wörner et al. 1993, Knudsen & Lidwin 1996, Ripley et al. 1998, Clark & Kontak 2004, Hanley et al. 2004). Non-silicate inclusions are only briefly discussed here not because they are unimportant, but simply to avoid overlaps with other chapters, which deal with those fluids and processes responsible for their formation in more detail. While writing the chapter, I have assumed that readers do already have a general idea of what silicate melt inclusions are, and understand the distinction between primary and secondary inclusions explained, for example, in the classical review by Roedder (1984a). Useful introductory information about basaltic melt inclusions can be also found in recent insightful reviews by Frezzotti (2001), Lowenstern (2003), and Danyushevsky et al. (2002a, b, 2004).
equivalent, basalt, have been studied for a long time. We know a good deal about geology, geochemistry and petrography of gabbroic intrusions, and we know a lot about liquidus phase equilibria in basaltic melts from observations in natural lavas and numerous experimental studies. Do we know enough? And what new, if anything, can we learn from melt inclusions? The answer to the first question is no. Igneous petrologists continue to argue about the most fundamental aspects of basaltic magma evolution. As to the second question, melt inclusions may be of great help for resolving long-standing controversies, and in my view, gabbroic inclusions have been underused and underestimated. This section lists some examples of current and potential applications of melt inclusion research to unresolved problems of gabbro petrogenesis.
BACKGROUND ON GABBRO PETROGENESIS AND THE USE OF MELT INCLUSIONS Gabbroic rocks are plutonic products of basaltic magma crystallized at low pressures under conditions of plagioclase stability. Deep-sea drilling, sea-bottom dredging, and studies of ophiolitic complexes show that the lowermost layer of oceanic crust is largely composed of layered gabbro (e.g., Kelemen et al. 1997). On the continents, gabbroic rocks form large isolated intrusions, or crystallize at sub-volcanic levels in magma chambers beneath basaltic volcanoes, and within continental trap formations. Gabbroic intrusions vary in size from gigantic layered complexes, like the Bushveld Complex and the Great Dyke in southern Africa, to small subvolcanic dykes and sills. The size of the body directly affects its cooling and crystallization rates, and thus the conditions for melt inclusion formation may change greatly from one plutonic body to another. Gabbroic rocks often associate in large plutons with co-genetic segregations of ultramafic crystal cumulates and silicic, residual magma products. In some intrusions, like the Skaergaard intrusion in East Greenland (Wager & Brown 1968), the whole range of plutonic rocks from early, high-temperature mafic cumulate layers to late pegmatite and granophyre appears to form from a single batch of basaltic magma. That is why, for sake of consistency, this chapter sometimes touches upon rocks, which are strictly speaking outside the gabbroic family, but form together with gabbro from a single parental melt. Gabbroic rocks and their volcanic
The general trend of melt evolution The greatest controversy of basalt petrogenesis concerns the general trend of differentiation and liquid evolution. The start of the dispute dates back to the 1920s, when the opposing views were clearly formulated in classical works by Bowen (1928) and Fenner (1929). Since that time, the followers of Fenner have argued that fractional crystallization drives a tholeiitic liquid composition towards strong enrichment in iron oxides, while the silica content of the liquid remains almost constant, or slightly decreases. Only at the very end of crystallization does massive precipitation of Fe–Ti oxides turn liquid evolution towards Fe depletion, rapid silica enrichment, and the terminal rhyolitic eutectic. The proponents of the alternative Bowen trend claim that the turn towards silica enrichment and Fe depletion starts much earlier, at higher temperatures, and much lower maximal (FeO+Fe2O3) concentrations. Clearly, the onset of Fe–Ti oxide crystallization is the key. It was realized long ago, and repeatedly shown in numerous experiments that oxidized conditions (high fO2 and Fe3+/Fe2+ values) favor early crystallization of magnetite and the Bowen trend, while the reduced conditions may result in much higher total Fe oxide concentration in the liquid, and a general evolution similar to the Fenner trend (see, for example, Osborn 1979). Since the early study by Wager & Deer (1939), the Skaergaard intrusion in East Greenland has been considered for decades as the prime example of the Fenner (strong Fe enrichment) trend due to extensive fractional crystallization of 100
CRYSTALLIZED MELT INCLUSIONS IN GABBROIC ROCKS
Mg–Fe silicates and plagioclase from a tholeiitic parental magma. The Skaergaard intrusion is much smaller than Precambrian ultramafic gabbroid complexes like the Bushveld, South Africa, or Stillwater, USA, and is thought to represent a straightforward case of fractional crystallization, not complicated much by multiple injections of parental magma, crustal rock assimilation, and other processes. Average bulk compositions of the Skaergaard layered gabbros, and the evolution of Fe–Mg solid solutions show a progressive Fe enrichment from the bottom Lower Zone to the top of the Upper Zone, where the Layered Series and the Upper Border Series converge (see Wager & Brown 1968, and McBirney 1989 for explanation of stratigraphic sub-division, and a more detailed account of Skaergaard stratigraphy). Surprisingly, the appearance of ilmenite and magnetite in the Skaergaard Layered Series seems not to put an end to the Fe enrichment. Thus, it appears that simple schemes based on the timing of Fe–Ti-oxide crystallization are not easily applicable to natural layered gabbroic intrusions. Many attempts have been made to reconstruct liquid lines of descent in Skaergaard by: (1) mass-balance calculations (Wager & Brown 1968, Hunter & Sparks 1987a, 1987b, Nielsen 2004); (2) experimental simulation of crystallization or partial melting of Skaergaard rocks (McBirney & Naslund 1990, Toplis & Carroll 1995); (3) tracing specific elements using mineral–melt partition coefficients (Jang et al. 2001, Tegner 1997); (4) geochemical thermometry of cumulus mineral assemblages (Ariskin 1999, 2002); and (5) comparisons with compositions of coeval and spatially associated dykes (Brooks & Nielsen 1978). The disparate results of the reconstructions are illustrated in Fig. 5-1. In general, the disagreement among researchers regarding the starting composition of the Skaergaard parental magma is not very dramatic (see also Nielsen 2004), and the models show a reasonable agreement at the early stages of magma evolution, up to about 50–60% crystallization. But from this point on, the proposed liquid evolution paths start to fan off in different directions (Fig. 5-1). The fO2 evaluations, which are key for magnetite saturation, show strange discrepancies too. It is generally agreed that the redox conditions in the Skaergaard liquid became progressively more reducing from the middle to the
Toplis & Carroll 1995 Wager & Brown 1968 Hunter & Sparks 1987 McBirney & Naslund 1990
FeO, wt. %
30
20
10
0 40
50
60
70
80
SiO2, wt. % FIG. 5-1. Lines of liquid descent in the Skaergaard intrusion presented in terms of iron oxide (FeO) versus silica (SiO2) variations. Average compositions of immiscible Fe-rich and Si-rich melt inclusions in apatite (Fig. 5-3) reported by Jakobsen et al. (2005) are shown by stars. See text for the discussion of different models presented on the plot.
top of the Layered Series. However, thermodynamic calculations by Ariskin (2002) showed an extraordinary decrease of the fO2 values by three orders of magnitude within a temperature interval of 60oC from the formation of the Lower Zone to the bottom of the Upper Zone. Such a decrease cannot be explained by evolution in a closed system, and, if an open system is assumed, no reasonable mineral buffer can maintain such a reduction. It appears that neither mass-balance calculations, nor thermodynamic modeling based on experimental studies have been able to resolve the long-time Fenner-Bowen controversy at Skaergaard, and for basaltic magma in general. Genuine samples of parental liquid would be of great help in this situation, and they can be found in melt inclusions. In Skaergaard, crystallized melt inclusions have been found in plagioclase throughout the Middle Zone, as described in the pilot study by Hanghøj et al. (1995), and apatite-hosted melt inclusions have been identified in the Upper Zone and studied by Jakobsen et al. (2005). Both studies support strong Fe enrichment in the Layered Series, and Jakobsen et al. (2005) confirmed silicate liquid immiscibility in the Upper Zone (see below). The results of melt inclusion studies have been very encouraging so far, but more work is needed in the future to better constrain melt evolution in the Skaergaard magma, and fully understand the nature of the Feenrichment trend. 101
ILYA V. VEKSLER
may exist many genetic types of layering. It is not possible and hardly necessary to give here a detailed account of the existing models. To put it shortly, there are two main opposing groups of ideas, and a number of intermediate points of view (e.g., dynamic and non-dynamic types of layering in terms used by McBirney & Nicolas 1997, and Boudreau & McBirney 1997). Early models of rhythmic layering tried to explain the origin of igneous modal stratification in terms of crystal gravitational settling and sorting in a large reservoir of convecting magma (Wager & Brown 1968). By analogy, the layering in gabbros has been thought to form like the stratification in river sands. In later models, which have become increasingly popular since the 1980s, the origin of rhythmic layering has been attributed to an interplay of thermal and chemical gradients, and rhythmic changes in crystal nucleation and growth rates due to under-cooling and/or over-saturation at an advancing crystallization front. In this case, the presumable role of crystal settling is minor, and banded chemical sediments can be used as a distant analogy (McBirney & Noyes 1979). In the context of this chapter, it is important to note that the first group of models assumes crystal settling from a homogeneous melt of uniform chemical composition, while the second group of models implies significant chemical variations in the melt at the crystallization front (e.g., Boudreau & McBirney 1997, McBirney & Nicolas 1997). Melt inclusions may help to reveal the periodic fluctuations in melt composition within rhythmic stratigraphic units, and thus test the validity of the models. Unfortunately, I am not aware of any published systematic study of gabbroic melt inclusions at the scale of individual rhythmic units aimed at deciphering the origin of modal layering.
Crystal settling, crystal sorting, and rhythmic layering Mass-balance calculations in gabbroic intrusions are greatly complicated by extensive differentiation and numerous types of layering: phase, cryptic, rhythmic, etc. (Wager & Brown 1968). Not all the variations are easily explained in terms of fractional crystallization, and the classical cumulate model proposed by Wager & Brown (1968). Rhythmic layering manifested by pronounced periodic variations in modal proportions of the main rock-forming minerals is probably the most mysterious feature of gabbroic intrusions. This spectacular stratification can be easily seen in natural outcrops (Fig. 5-2), and it has long attracted much attention of petrologists. The scale of the layering in gabbroic-ultramafic intrusions varies from centimetres to dozens and hundreds of metres, and there exists a great number of different types and styles of layering (e.g., McBirney & Noyes 1979, McBirney & Nicolas 1997, Boudreau & McBirney 1997). Modal proportions of minerals may range from those of anorthosite with only minor poikilitic pyroxene to mafic assemblages of oxides and olivine but little if any feldspar. Some rhythmic units are evenly graded from a mafic base to a felsic top; others are sharply bimodal. Individual layers usually extend laterally over large distances, and some layers less than a metre thick can be traced throughout the intrusion. Numerous models have been proposed over the last decades, but the origin of rhythmic layering remains enigmatic. In fact, the great diversity of types and styles of layering is likely to result from not one but several processes, and there
Sub-liquidus and sub-solidus re-equilibration Igneous rocks in ultramafic-gabbroic complexes may be so much reworked by late magmatic and/or sub-solidus recrystallization and chemical re-equilibration that they start resembling metamorphic rocks. Because of very slow cooling rates, and long and complicated emplacement history, gigantic plutonic complexes such as the Bushveld or Stillwater are in general more affected by late-magmatic and post-magmatic re-equilibration than smaller intrusions like Skaergaard or sub-volcanic dykes and sills. Bowen (1928) was probably the first who emphasized the reactive nature of magmatic crystallization, and pointed out
FIG. 5-2. Natural outcrops of rhythmic layering; the Layered Series of the Skaergaard intrusion, East Greenland. Photograph is a courtesy of Christian Tegner.
102
CRYSTALLIZED MELT INCLUSIONS IN GABBROIC ROCKS
various types of crystal–liquid reactions taking place during crystallization of natural common melt compositions, and basaltic liquids in particular. Liquidus crystals constantly re-equilibrate in a cooling magma, and relative movement of crystals and liquid due to gravitational crystal settling, thermal convection, compaction of a cumulus crystal mush, and other processes in a large magma chamber may result in non-trivial effects, which are hard to understand without detailed geochemical studies, close examination of rock textures, and mathematical modeling. For instance, Sonnenthal & McBirney (1989) presented vivid examples of extensive textural and compositional reworking of foundered blocks of gabbro, which had detached from the Upper Border Series at the roof of the Skaergaard magma chamber and settled into the mush of crystals on the floor. The blocks tend to be more plagioclase-rich than the original rocks at the upper border of the intrusion, and seem to have lost mafic components to the surrounding gabbro. Plagioclase grains in the blocks have patchy resorption and replacement textures with large compositional contrasts between different zones. The patches of replacement show fine oscillatory zoning with much more refractory, Ca-rich compositions than the plagioclase in the zones from which the blocks came or in the gabbros in which they now reside. The importance of complex heat and material transport processes at the upper crystallization front, and the original proportion of phenocrysts during magma emplacement have been emphasized by Marsh (1996). According to his Null Hypothesis, phenocryst-free magma does not differentiate much, and the extensive complex layering exhibited by the majority of large gabbroic intrusions, as well as some smaller sills, indicates a significant load of suspended phenocrysts carried by the magmas during emplacement. A large proportion of early liquidus crystals in the Skaergaard magma during the emplacement (up to 30 volume %) has been proposed by Ariskin (2002) on geochemical and thermodynamic grounds. Melt inclusions may help to reveal the population of early phenocrysts, and decipher local compositional variations of liquid in response to cooling, compositional convection, and other processes at crystallization fronts. However, in those cases when early liquidus minerals are completely recrystallized and compositionally reworked, melt inclusions are not likely to survive either. Thus, it is very important to
examine textural and mineralogical evidence in order to determine, to what extent high-temperature primary magmatic features (including primary melt inclusions) have been preserved in a particular specimen of gabbroic rock. In recent years, the research on igneous rock textures has developed from purely descriptive models (e.g., the classical cumulus model introduced by Wager & Brown 1968) to better formalized, numerical models based on the measurements of dihedral angles between crystals (Holness et al. 2005) or statistical 3D image analysis of mineral grain distribution (Jerram et al. 2003). The most important implication of the modern research on rock textures for melt inclusion studies is the general conclusion that although igneous rocks in large mafic intrusions indeed may show local examples of complete recrystallization, and totally equilibrated “metamorphic” textures with dihedral angles approaching 120o, primary magmatic features are often preserved even in very large intrusions. For example, 3D frameworks of touching plagioclase crystals have been proven to form in basaltic flows as early as at only 25% crystallization, and similar frameworks were revealed in plutonic rocks from large anorthositic and gabbroic complexes (Philpotts et al. 1999). Such observations imply that primary magmatic crystals and melt inclusions may not be totally destroyed by sub-solidus processes in slowly cooling plutons. The descriptions and images of crystallized plagioclase-hosted melt inclusions from anorthositic layers of the Stillwater Complex presented by Loferski & Arculus (1993) give an impression that the inclusions were more recrystallized and better texturally equilibrated than similar inclusions in Skaergaard (Hanghøj et al. 1995, see also Fig. 5-3a), probably because of slower cooling in Stillwater. However, daughter mineral assemblages are essentially the same, and primary magmatic zoning around the inclusions has been preserved in plagioclase crystals from both plutons. Magma mixing and assimilation of country rocks Not all gabbroic intrusions can be viewed as closed systems, and the Skaergaard intrusion is rather an exception than the rule. Many other intrusions demonstrate unequivocal evidence of repeated magma replenishments, mixing of contrasting magma compositions, and major assimilation of country rocks (Wager & Brown 1968). For example, multiple replenishments of hot, picritic magma have been proposed in the Rum 103
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
FIG. 5-3. Back-scattered electron SEM images of melt inclusions in gabbroic rocks. a. Crystallized melt inclusion in plagioclase, Middle Zone of the Skaergaard intrusion. Scale bar is 100 µm. b. Crystallized melt inclusion in olivine, Upper Zone of the Skaergaard intrusion. Scale bar is 20 µm. c. Large devitrified melt inclusion (mi) in apatite, Upper Zone of the Skaergaard inclusion. Scale bar is 200 µm. d. Glassy melt inclusion of two immiscible silicate liquids in native iron, Khungtukun sill in the north of the Siberian trap province. Scale bar is 500 µm. Abbreviations for phases: ap – apatite; cpx – clinopyroxene; fe – native iron; ilm – ilmenite; lf – iron-rich silicate glass; ls – silica-rich rhyolitic glass; mt – magnetite; ol – olivine; pl – plagioclase; qz – quartz.
layered intrusion, Scotland (Upton et al. 2002, and references therein), and it was hypothesized that the Stillwater Complex had formed by mixing of two parental magmas: high alumina (“anorthositic”) basaltic liquid and a boninite-like magma (Irvine et al. 1983). Complex mixing of up to four different magma types has been proposed in the Bushveld Complex (Eales 2002, and references therein). Melt inclusions can be of great help for revealing magma mixing events, and may also record contamination of magma by country rocks. For example, chromitehosted melt inclusions from the main chromitite layer of the Stillwater Complex were interpreted as melt droplets produced by partial melting and assimilation of country rocks by hot primitive magma at the roof of the chamber (Spandler et al. 2005), and similar inclusions in the Bushveld Complex were thought to result from mixing of the magma at the bottom of the crystallizing magma chamber with Na–K rich fluids expelled from the underlying crystal pile (Li et al. 2005). Gabbroic
melt inclusions in common rock-forming silicates may prove to be very useful in the future as indicators of magma mixing. It should be noted, however, that in some instances it might be difficult to distinguish between magma mixing events and liquid immiscibility. The latter process is discussed below in more detail. The role of silicate liquid immiscibility Observations of volcanic rocks (Philpotts 1982, Roedder 1984b, Roedder & Weiblen 1970), and rock-melting experiments (Dixon & Rutherford 1979, McBirney & Nakamura 1974, Philpotts 1979, Roedder & Weiblen 1970) have demonstrated that some ferrobasaltic liquids undergo phase separation and split into two coexisting melts, one of which is rhyolitic and the second is strongly enriched in Fe oxides, CaO, MgO, TiO2 and P2O5, and depleted in SiO2, Al2O3 and alkalis. Despite the significant body of well-established evidence, very few studies have regarded the phenomenon of silicate liquid 104
CRYSTALLIZED MELT INCLUSIONS IN GABBROIC ROCKS
immiscibility as an important petrogenetic factor in terrestrial basalts and gabbroic intrusions (e.g., De 1974, Philpotts 1982, Wiebe 1979, Ryabov 1989, Loferski & Arculus 1993). The process was occasionally brought up as an indication of the extreme Fe enrichment in late-stage Skaergaard liquid (McBirney & Nakamura 1974, McBirney & Naslund 1990). The fact that the majority of researchers of gabbroic intrusions have little interest in silicate liquid immiscibility is somewhat surprising, because the process offers in fact an elegant solution to the major Fenner-Bowen controversy, which was discussed in a previous section. The dispute about the Fe-rich or silica-rich evolved liquids becomes irrelevant, because both liquids are formed at some point of magma evolution by liquid immiscibility. Recent study of melt inclusions in apatite and olivine by Jakobsen et al. (2005) clearly demonstrated the presence of two immiscible silicate liquids in the Upper Zone of Skaergaard (Fig. 5-4), and the fact that compositionally immiscible liquids found in the inclusions represent two end-members of the diverse liquid evolution models proposed for the intrusion (Fig. 51). The presence of melanogranophyre lenses and patches in the Upper Zone is additional, outcropscale evidence of immiscibility (e.g., McBirney 1989). The observations of melt inclusions leave no doubt that silicate liquid immiscibility did take place in Skaergaard, and it was also reproduced in rock-melting experiments (McBirney & Nakamura 1974). The crucial question now is: when did the process start? Experimental studies in tholeiitic compositions (Dixon & Rutherford 1979, McBirney & Nakamura 1974, Philpotts 1979) usually place the onset of liquid immiscibility at 1010–1030oC and 90–95% crystallization. Preliminary results on plagioclase-hosted melt inclusions imply that the unmixing in Skaergaard may have started at higher temperature (around 1100oC) and 60–65% crystallization (J.K. Jakobsen, personal communication). If it is proven that immiscibility in tholeiitic magma chambers can start so early, the process is likely to play a much greater petrogenetic role than was previously thought, and may potentially give clues to many unresolved problems including those outlined in the previous sections. When stable liquid immiscibility takes place at sub-liquidus temperatures, both conjugate liquids are required to be in thermodynamic equilibrium with the same mineral assemblage. This is one of the reasons why even large-scale immiscibility may leave no obvious traces after
FIG. 5-4. Photomicrographs of primary melt inclusions in cumulus apatite, of the Upper Zone of the Skaergaard intrusion. The coexisting Fe-rich (dark) and silica-rich (light) fine-grained crystallized melt inclusions trapped in the neighboring apatite crystals were interpreted as products of silicate liquid immiscibility (Jakobsen et al. 2005). Abbreviations for phases: ap – apatite; ilm – ilmenite; mt – magnetite; ol – olivine; pl – plagioclase. Melt inclusions are indicated by arrows. Scale bar is 0.2 mm.
complete crystallization in fully crystalline plutonic rocks (Bowen 1928). For correct interpretation of the melt inclusion record, it is important to keep in mind that some crystal hosts may preferentially trap only one of the conjugate liquids. The mechanism of preferential trapping was explained in detail by Roedder (1984b). It arises from different growth rates of crystals in coexisting immiscible liquids. In the case of ferrobasalt–rhyolite immiscibility, Fe–Ti oxides and Fe–Mg silicates, such as olivine and pyroxenes, are compositionally close to the Fe-rich immiscible liquid, and apparently grow much faster in that liquid than in the rhyolitic conjugate melt. Small droplets of the rhyolitic liquid would represent obstacles for the propagating crystals, and would tend to form inclusions in the Fe-rich solids. Accordingly, only silicic immiscible liquid formed melt inclusions in the fayalitic olivine from the Upper Zone of Skaergaard (Jakobsen et al. 2005). In contrast, plagioclase borrows components from both liquids, as CaO is concentrated in the Fe-rich melt, while Na2O, Al2O3 and SiO2 mostly reside in the silicic melt. Thus, plagioclase is likely to be less selective in terms of melt entrapment, and examples from volcanic rocks support the universal presence of both immiscible liquids in variable proportions in the plagioclase-hosted melt inclusions (Fig. 5-5). Finally, apatite is a minor component in both 105
ILYA V. VEKSLER
liquids, but P2O5 concentrations are higher in the Fe-rich liquid (Philpotts 1979, 1982). Abundant Ferich inclusions in cumulus apatite from Skaergaaard (Fig. 5-4) also probably result from preferential wetting of apatite surfaces by the Fe-rich liquid. Silicate liquid immiscibility in ferrobasaltic compositions is still poorly constrained by experiments, and the broadly used computation codes like MELTS (Ghiorso & Carmichael 1985) or COMAGMAT (Ariskin 1999) do not model immiscibility. Melt inclusions provided the key evidence for liquid immiscibility in volcanic rocks (Roedder & Weiblen 1970, Roedder 1984b), and they prove to be equally useful for the same task in plutonic settings. Apart from Skaergaard, melt inclusion evidence of silicate liquid immiscibility has been reported in doleritic sills from the Siberian trap province (Ryabov 1989; see also Fig. 5-3d), and proposed in anorthositic rocks from the Stillwater (Loferski & Arculus 1993) and Nain Complexes (Wiebe 1979). The inclusions from anorthositic rocks are enclosed in Ca-rich plagioclase, and appear to represent immiscible liquids formed at much higher temperature than those in the Upper Zone of Skaergaard. This type of silicate liquid immiscibility has not been reproduced in experiments yet, and remains hypothetical.
FIG. 5-5. Back-scattered electron SEM image of plagioclase-hosted devitrified inclusions of the Fe-rich (LFe) and silica-rich (LSi) immiscible liquids in olivine basalt, Mauritius Island, sample BM1932, 103(5) from the rock collection of the Natural History Museum, London. The inclusions are surrounded by rims of sodic plagioclase due to host crystallization on inclusion walls. Scale bar is 0.1 mm.
content than in the main pool of magma. Interesting occurrences of rounded droplets of sulfide in apatite, olivine and pyroxene were reported by Ripley et al. (1998) in the Duluth Complex, Minnesota. In addition to sulfide-silicate immiscibility, the authors proposed also that apatiteand oxide-rich rocks of the Boulder area formed by exsolution of Fe–Ti–P-rich nelsonitic liquid from magma. Microscopic to micrometre-sized droplets of the Fe–Ti–P oxide melt were documented in interstitial glass from the Antauta subvolcanic center, Peru (Clark & Kontak 2004). Electron microprobe study of the droplets revealed strong variations in SiO2, Al2O3, CaO, FeO, TiO2 and P2O5. Positive co-variation of the CaO and P2O5 components closely followed the apatite-control line, implying that some droplets may represent mixtures of tiny apatite and oxide crystals with the immiscible Fe-rich type of silicate liquid discussed in the previous section. Exsolution of Fe- and Tirich nelsonitic liquids has been documented in melt inclusions hosted by clinopyroxene, kaersutite, plagioclase and zircon in gabbroic and syenitic xenoliths carried by alkali basalts of the Western Carpathians, Slovakia (Hurai et al. 1998), and in quartz-hosted melt inclusions from silicic volcanic rocks (Naumov et al. 1993). Associations of melt inclusions with primary or pseudo-secondary fluid inclusions in gabbroic rocks document magma degassing and exsolution of orthomagmatic hydrothermal fluids. Much attention has been given to highly concentrated, NaCl-rich
Magma degassing and exsolution of non-silicate liquids Silicate melt inclusions in basaltic phenocrysts commonly contain tiny sulfide droplets, which are interpreted as exsolution products of immiscible sulfide liquid (Roedder 1984a, Chapter 16). The occurrence of sulfide blebs in the inclusions can be used to determine the timing of sulfide-silicate liquid immiscibility in the magma. In plutonic rocks, timing can be important for the formation of Fe–Ni–Cu sulfide ores and economic concentrations of gold and platinum group elements. Danyushevsky et al. (2002a) have pointed out, however, that the formation of sulfide blebs in melt inclusions trapped by Mg–Fe minerals may not correctly indicate sulfide saturation in the main pool of magma, and that melt inclusion evidence should be treated with caution. Silicate melts enclosed by olivine and other Mg–Fe silicates tend to loose FeO by post-entrapment diffusion and re-equilibration with the host. In addition, because S solubility in silicate liquids strongly depends on the FeO content (Wallace & Carmichael 1992) the exsolution of sulfide droplets in the inclusions may take place at significantly higher temperature and/or lower S 106
CRYSTALLIZED MELT INCLUSIONS IN GABBROIC ROCKS
magmatic brines, because of their capacity to dissolve significant amounts of noble metals and redistribute Au and platinum group elements in magmatic sulfide deposits. Inclusions of chloride magmatic brines were documented at the Merensky Reef of the Bushveld Complex (Ballhaus & Stumpfl 1986), and in the Sudbury Breccia (Hanley et al. 2004). Recent experimental studies using synthetic fluid inclusions confirmed high solubilities of Pt and Au in concentrated chloride hydrothermal solutions at magmatic temperatures (Hanley et al. 2005). Fluid inclusions in gabbros also document exsolution of low-density C–O–H fluids from silicate melts and hydrothermal solutions (e.g., Ballhaus & Stumpfl 1986, Larsen et al. 1992, Wörner et al. 1993, Knudsen & Lidwin 1996). Further information on non-silicate magmatic fluids and vapors can be found in other chapters of this volume.
pyroxene, which are the major constituents of basaltic melts. After crystallization, the original volume of a melt inclusion would significantly shrink because of the host crystallization on inclusion walls, and the final shape of the crystallized inclusion will be determined by interfaces of the host and daughter crystals (Figs. 53 and 5-4). In the absence of severe host-melt postentrapment reactions (see below), most common daughter minerals in the inclusions are the same as the main minerals in the bulk rock. Thus, in a gabbroic rock, plagioclase-hosted inclusions are dominated by pyroxenes, Fe–Ti oxides, and in some cases contain abundant olivine, whereas pyroxenehosted inclusions normally contain plagioclase, oxides, and in some cases mica. Olivine-hosted melt inclusions are composed of feldspars, pyroxenes, oxide minerals, and may contain quartz in evolved compositions (Jakobsen et al. 2005, see also Fig. 53b). Accessory minerals, such as apatite, or sulfides minerals are normally present in daughter mineral assemblages in small amounts, and, as discussed below, serve as important indicators of the primary origin of the inclusions. Because the volatile content in common basaltic magma is low, cavities filled by fluid or vapor in crystallized basaltic melt inclusions are very small (if present at all), and normally not recognizable. Visible vapor/fluid cavities may appear however in evolved, volatilerich inclusion compositions.
CHARACTERISTICS OF MELT INCLUSIONS IN GABBROIC ROCKS Typical examples of primary melt inclusions from gabbroic rocks are presented in Figs. 5-3 and 5-4. Normally the inclusions are completely crystallized to multi-solid fine-grained assemblages of daughter minerals, and glasses in the inclusions can be found only in samples from small subvolcanic intrusions or at chilled margins of large plutons. The size of a typical primary inclusion is relatively large, usually a few dozens of micrometres, up to a hundred micrometres, in rare cases even bigger. Shapes vary broadly, from oval, spherical to irregular, but euhedral negative crystal shapes, which are often observed in basalts (Roedder 1984a, b, Lowenstern 2003, Danyushevsky et al. 2002a, b, 2004), are generally uncommon in gabbro. It should be noted, however, that regular oval or negative crystal inclusion shapes are common in oxide minerals and accessories like apatite (Figs. 5-3 and 5-4). The final shape of crystallized melt inclusions depends to a large extent on the content of host mineral components in trapped melt and the amount of host mineral precipitating on inclusion walls. Because of the excess crystal nucleation energy, components of the hosts normally crystallize on inclusion walls and rarely form separate daughter crystals. As the content of dissolved apatite component in basaltic melts is low, the layer of apatite precipitation on inclusion walls is thin, and melt inclusions are likely to retain the original regular shapes. The situation is quite different in plagioclase or
CHARACTERISTICS OF HOST MINERALS The following brief overview of host minerals available in gabbro is given here for two main purposes. First, before anything else, one has to know where to look for melt inclusions, and which minerals actually contain them. Secondly, host minerals are by no means inert containers, and trapped fluids more often than not are significantly modified by post-entrapment chemical reactions with the hosts (see below). Investigators familiar with experimental petrology know that even noble metal containers are not perfectly inert. Studies of melt inclusions are in many ways like experimental petrologic research, but in case of inclusions the container problems are further aggravated by much smaller sample-to-container mass ratios. In general, gabbro mineralogy does not offer many good hosts or inclusion containers. Olivine and quartz, broadly used in melt inclusion studies elsewhere, are often not available, being present only in some transitional types of gabbros. 107
ILYA V. VEKSLER
These and other potential host minerals are evaluated below in terms of their suitability for melt inclusion studies.
Their dark color makes pyroxenes less transparent, and a protective atmosphere (in common practice, He or Ar inert gases used in some types of modern heating stages) should be used during homogenization runs to avoid further darkening due to oxidation. Nevertheless, pyroxene-hosted melt inclusions should not be ignored, especially where they coexist with plagioclase-hosted inclusions in the same rock.
Plagioclase Plagioclase is universally present in all gabbroic rocks; it constitutes the bulk of anorthosite, and is the defining mineral of the whole gabbro rock family. It is not a bad host for melt inclusions. The great abundance of tabular glassy inclusions in plagioclase phenocrysts from volcanic rocks implies that rapidly growing plagioclase crystals readily trap primary melt inclusions (Fig. 55). Large euhedral plagioclase in plutonic rocks commonly hosts proportionately large melt inclusions (Fig. 5-3a). Plagioclase-hosted melt inclusions from volcanic rocks have been extensively studied, for example, by the research group in Corvallis, Oregon (e. g., Souse-Page et al. 2002, and references therein). Plagioclase solid solution crystallizes over a broad temperature interval, and thus the melt inclusion record in plagioclase normally covers a significant part of parental magma evolution. Inclusions are easily found and viewed in plagioclase because the host is colorless and transparent. The composition of the host is rather simple; some important groups of trace elements are poorly compatible with the plagioclase structure and thus have a good chance of remaining unfractionated and well preserved in the inclusions. However, many other major and minor melt components, such as, Sr, REE, and even Fe to some extent do enter the plagioclase structure in large amounts, can effectively diffuse out of a melt inclusion, and thus a potential post-entrapment modification of inclusions can be significant. What exactly happens with melt inclusions in plagioclase during slow cooling is thoroughly discussed in one of the sections below. Finally, the mechanical strength of plagioclase containers is diminished by cleavage planes, and larger inclusions may leak during heating in homogenization runs.
Olivine Lack of cleavage, enhanced mechanical strength, and its relatively simple chemical composition make olivine a much better inclusion container than pyroxenes. Nevertheless, rapid Fe– Mg olivine-melt re-equilibration and possible Fe loss from the melt into the host by diffusion may greatly complicate estimations of the entrapment temperature and the original melt composition (Gaetani & Watson 2002, Dayushevsky et al. 2002b, 2004). This is particularly true for slowly cooled plutonic rocks. Olivine is abundant in troctolite, olivine gabbro and ultramafic cumulates of layered intrusions, but is typically absent from more evolved gabbroic rocks. Forsteritic olivine is indispensable for sampling of primitive, hightemperature melts, and studies of early stages of melt evolution. Most trace elements and volatile components are not compatible with olivine and stay well preserved in melt inclusions. Spinel and Fe–Ti oxides Chromian spinel is a typical mineral of primitive, high-temperature basaltic magmas, but magnetite and ilmenite solid solutions normally crystallize from more evolved melts at temperatures below 1150oC. Chromian spinel crystals are commonly enclosed, together with some melt, in early forsteritic olivine or plagioclase, and the spinel crystal inclusions, in their turn, may also contain small melt inclusions (Kamenetsky 1996, Danyushevsky et al. 2002a). Chromite-rich layers in the Bushveld and Stillwater intrusions appear to have formed at lower temperatures from more evolved or hybrid melt compositions, and are believed to have been triggered by magma mixing or contamination events (Li et al. 2005, Spandler et al. 2005, and references therein). One would expect chromite to be a stable, inert, and very refractory host, and the studies of chromite-hosted melt inclusions in volcanic rocks (Kamenetsky 1996) and ophiolitic gabbro (Schiano et al. 1997) support this view. Because chromite is poorly soluble in silicate
Pyroxenes High-Ca pyroxene is a typical mineral of gabbro proper, while pigeonite is characteristic of norite and two-pyroxene gabbro. Both minerals commonly contain melt inclusions, but as hosts they have all the imperfections of plagioclase plus some more. Augite can incorporate almost all the components of basaltic melt, and consequently severe post-entrapment melt re-equilibration and modification should be expected in the inclusions. 108
CRYSTALLIZED MELT INCLUSIONS IN GABBROIC ROCKS
characterization of the enclosed mineral phases (see below).
melts, there is generally little danger in over-heating inclusions during homogenization runs, and the homogenized glasses usually show reasonable basaltic compositions. The examples from chromitite layers in large layered complexes tell a different story, however. The homogenized inclusions show strangely variable compositions, far from any reasonable parental melt (Table 5-1), and may indicate a significant host–melt postentrapment interaction. I will come back to the Bushveld and Stillwater chromite-hosted inclusions later in the chapter, when the host-melt reactions are discussed in more detail. The opaqueness of Fe–Ti oxides, and typically dark-brown color of Cr-spinel combined with its very high refractive index greatly complicate the search and optical observation of melt inclusions. Accidental slices of silicate melt inclusions in opaque oxides are sometimes found in normal petrographic thin sections, or may be also seen in reflected light on polished surfaces. Obviously, inclusions exposed on the surface are not suitable for homogenization experiments, and may actually represent silicate embayments or crystal intergrowths rather than true crystallized melt inclusions. Nevertheless, the true nature of the inclusions can be established on the basis of detailed petrography and electron microprobe
Silica polymorphs Tridymite is a structural form of SiO2 that is stable above 867oC (Deer et al. 1966), and it normally transforms to quartz at lower temperatures. With further temperature decrease, the quartz α–β phase transition takes place at 573oC (atmospheric pressure), and it is associated with a density increase of about 5 vol.%. This shrinkage may cause partial decrepitation of volatile-rich fluid-bearing inclusions. In all other respects, silica polymorphs are very good, almost ideal containers for melt and fluid inclusions, and it is no wonder that quartz is the prime host mineral for inclusion studies in granitic rocks. Evolved liquids derived from basaltic magma may precipitate tridymite, and minor quartz may be occasionally found in some evolved, low-temperature gabbroic assemblages. For example, quartz-hosted fluid inclusions helped to identify very reduced, methane-rich lowtemperature fluids in the Skaergaard intrusion (Larsen et al. 1992), and highly concentrated magmatic brines in Bushveld, South Africa, and Sudbury, Canada, magmas (Ballhaus & Stumpfl 1986, Hanley et al. 2004).
TABLE 5-1. EXAMPLES OF MELT COMPOSITIONS (WT.%) FROM CRYSTALLIZED MELT INCLUSIONS IN GABBROIC ROCKS. Intrusion Host Method Reference SiO2 TiO2 Al2O3 FeO MnO MgO CaO Na2O K2O P2O5 Total
Stillwater, USA Plagioclase Modal calculations
Bushveld, S. Africa Chromite
a
Loferski & Arculus, 1993 43.61 4.82 13.45 9.78 0.54 7.92 18.43 0.43 0.02 0.8 99.80
Chromite b
homogenization
modal calculations c
Spandler et al., 2005 46.07 0.98 9.54 19.80 0.33 16.66 0.76 5.18 0.17 n.a. 99.49
a
Modal calculations are based on microprobe analyses and average modal proportions of daughter minerals. b Complete homogenization in a heating stage. c GSA Data Repository item 2005173.
109
Li et al. 2005 47.65 2.50 9.51 4.41 0.05 26.16 1.16 2.41 1.48 n.a. 95.33
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
when they grow in skeletal regimes. This is the main mechanism of melt inclusion formation in volcanic rocks (Roedder 1984a, b, and references therein). In large plutonic magma chambers, episodes of rapid crystal growth may take place locally (at chilled margins) or during degassing or magma mixing events, but in general cooling and crystallization in intrusions are very slow, perhaps taking 1 million years for crystallization of a body 2000 km3 in volume (Morse 1980). Skeletal crystal growth at such conditions is not likely, and there is a good chance that crystals and liquid will closely approach equilibrium with respect to diffusion. Crystal defects and inclusions are much more likely to form in the plutonic environment by heterogeneous trapping of two or more phases. The processes are illustrated in Fig. 5-6. In such a
Apatite and other accessory minerals Liquidus apatite frequently forms euhedral crystals with large, spherical or elongated melt and fluid inclusions (Fig. 5-3c). It is a good and very interesting host mineral, but unfortunately not readily available, as ordinary basaltic magmas reach apatite saturation very late, at the final stages of crystallization. Apatite host crystals have been very helpful in melt inclusion studies of the Skaergaard intrusion (Jakobsen et al. 2005). Some important petrogenic trace elements, such as REE and Sr, and almost all volatile components are compatible with apatite, and thus their distribution in inclusions can be strongly affected by the host. Minor and accessory minerals such as kaersutite or zircon may also contain interesting and informative melt inclusions (e.g., Hurai et al. 1998). Similar to apatite, zircon is a stable, inert and refractory container poorly soluble in basaltic melts, but its use in melt inclusion studies is limited by small crystal size and late crystallization from basaltic liquids.
a
Native iron This rare, exotic host is worth mentioning because it sometimes crystallizes from terrestrial basaltic magmas, and presents a unique record of high-temperature silicate liquid immiscibility. Grains, nuggets and massive blocks of native iron alloy with iron carbide (cohenite) and abundant silicate crystal and glassy inclusions have been found and examined in a few gabbroic sills within the Siberian trap province (Ryabov 1986). Very similar iron segregations with silicate inclusions have been described in basalts of the Disco Island in west Greenland (Bird et al. 1981). In the gabbroic sills, native iron occurs in the roof endocontact zones, and like on Disco, it is believed to form by Fe reduction due to reaction of basaltic magma with host carbon-rich sedimentary rocks. Immiscibility textures in glassy inclusions are truly remarkable (Fig. 5-3d), and the melt droplets represent the closest possible natural analog of experimental runs in metal containers.
b
FORMATION AND POST-ENTRAPMENT MODIFICATION OF MELT INCLUSIONS Crystal defects and inclusion entrapment Mechanisms of melt inclusion entrapment have been thoroughly discussed by Roedder (1984a, b). Theoretical reasoning and observations of real rocks suggest that inclusions and other defects are likely to form in rapidly growing crystals, especially
FIG. 5-6. Heterogeneous trapping of melt inclusions. a. Crystals (rectangles) and suspended globules of a second liquid or a fluid (shaded circles) are trapped together with the main parental melt by a growing crystal. b. The same crystal with crystallized melt inclusions after complete crystallization of the rock. Dotted line signifies growth zones.
110
CRYSTALLIZED MELT INCLUSIONS IN GABBROIC ROCKS
scenario, melt droplets are trapped together with obstacles to crystal growth, and the latter may be other minerals, or gas bubbles, or small droplets of a second liquid suspended in the main parental melt. The droplets of a second liquid may be due to either magma mixing (dynamic, disequilibrium case) or stable liquid immiscibility (equilibrium case). In the first case, the formation of heterogeneous melt inclusions depends upon relationships between the crystal growth rate and the rate of assimilation (dissolution) of the second newly injected liquid. In the second case, emulsions of two conjugate melts may exist indefinitely, if the thermodynamic conditions do not change. Two consequences arise from heterogeneous trapping of melt in plutonic rocks. First, the modal proportions of daughter minerals in fully crystallized inclusions may vary (Fig. 5-6b), and this greatly complicates the recognition and correct interpretation of the inclusions. Secondly, on a more optimistic note, the very appearance of abundant crystallized melt inclusions in plutonic rocks may serve as an important indicator of abrupt changes in magma evolution, and massive populations of melt inclusions may keep a record of notoriously elusive processes, such as degassing, magma mixing or liquid immiscibility. Because of the slow cooling rates and relatively low viscosity of basaltic melts, boundary layer effects (chemical gradients around growing crystals) are not likely to pose a great problem in gabbroic rocks. These potential border effects have been evaluated many times by a number of different methods, and the current consensus seems to be that the effects are insignificant for mafic melt inclusions larger than a few micrometres (e.g., Kuzmin & Sobolev 2003).
one has to make sure that the suspicious groups of small crystals are indeed crystallized melt inclusions, and not accidental aggregates of solid inclusions or crystal intergrowths. Poikilitic crystals and crystal inclusions are common in gabbroic rocks, and distinguishing them from crystallized melt inclusions may be difficult at times. The criterion of uniform proportions of daughter phases proposed by Roedder (1984a) may not always work because of heterogeneous trapping (see above). However, if a stable association of three, four or more small crystals (much smaller than the average grain size of the rock) forms intergrowths, and is repeatedly found inside a certain host, the intergrowths are likely to represent crystallized melt inclusions. They are very likely to be crystallized melt inclusions when rimmed by compositional zoning of the host. They are certainly melt inclusions, if at least one of the typical daughter minerals was not a liquidus phase at the time of inclusion entrapment and is not present among cumulus phases. For example, characteristic tiny needles of apatite (Fig. 5-3a) were crucial for identification of the plagioclase-hosted melt inclusions from the Middle Zone of Skaergaard (Hanghøj et al. 1995), because the Skaergaard magma was known to be undersaturated in apatite at the stage of the Middle Zone formation. In a similar way, plagioclase-hosted melt inclusions in Stillwater anorthosite contain distinct Mn-rich daughter ilmenite absent from the rocks (Loferski & Arculus 1993). Li et al. (2005) identified crystallized melt inclusions in Bushveld chromite based on an unusual Na-rich daughter phlogopite, not present in the bulk-rock mineral assemblage. In fact, Narich phlogopite seems to be a typical daughter phase in many other chromite-hosted inclusions from podiform chromite and chromitite layers worldwide (Schiano et al. 1997, Spandler et al. 2005). Thus, once again, detailed and thoughtful petrographic studies and microprobe analyses of daughter mineral assemblages are absolutely crucial for correct genetic interpretation of crystallized melt inclusions.
Crystallization of daughter minerals The presence of glass, as mentioned above, is very unlikely in slowly cooled inclusions of lowviscosity basaltic melts, and glasses are normally found only in sub-volcanic environments. Occasionally, even in large gabbroic plutons, some melt inclusions may quench to glasses, e.g., at chilled margins (Fig. 5-3d), but in general they completely crystallize to aggregates of daughter minerals. It is always a good idea to start with detailed petrographic and microprobe characterization of daughter minerals before venturing into destructive manipulations, such as heating inclusions back to homogenous liquids. First of all,
Post-entrapment reactions between host minerals and melt inclusions In contrast to rapidly chilled volcanic rocks, large slowly cooling magma chambers provide practically limitless time for crystallization of melt inclusions, as well as heat and mass transfer between inclusions and the hosts. As already mentioned above, magmatic crystallization involves 111
ILYA V. VEKSLER
numerous crystal–melt reactions, in accordance with the Bowen reaction principle (Bowen 1928). The net outcome of a chemical reaction (the composition of the products) strongly depends in general on the mass proportions of the reactants. Separation of a small melt droplet from the main reservoir of magma, and its entrapment inside a mineral host means a dramatic change in the liquid/solid mass proportion. Because of the continuous chemical re-equilibration between liquidus minerals and gradually cooling silicate liquid, the sequence of events in cooling and crystallizing melt inclusions may significantly differ from the crystallization processes in the main batch of magma. Those potential differences and complications should be fully understood before melt inclusions are used as snapshots of the parental melt evolution. When a crystallized melt inclusion is re-heated in a furnace and quenched to glass, the glassy inclusion may look very much like a natural quenched analog from volcanic rock. However, because of the history of melt–host interaction, the temperature of total melting of daughter phases may notably deviate from the entrapment temperature, and the compositions of the natural and experimental glasses may also be different.
Plagioclase and the continuous reaction series. For a start, let us examine the evolution of melt inclusions trapped in plagioclase using the diopside (Di) – anorthite (An) – albite (Ab) model system. A detailed description of liquidus phase relationships in the system can be found in many textbooks, e.g., by Ernst (1976) or Morse (1980). An analysis is illustrated in Fig. 5-7. Consider bulk melt composition X, which starts to crystallize liquidus plagioclase at 1400oC. Plagioclase is a solid solution of the Ab and An end-members, and the initial composition of liquidus plagioclase at 1400oC is P1. In a cooling system, crystallization may proceed either in equilibrium or fractionation mode, or follow some intermediate trend between the two. In any case, the plagioclase solid solution becomes progressively enriched in less refractory Ab component, and the liquid composition moves first along a curved path towards the plagioclase– diopside cotectic, reaches the cotectic at a point where diopside crystals join in, and then moves along the cotectic until crystallization of the liquid is complete. In case of equilibrium crystallization, the bulk composition of the system is fixed, and in order to meet the mass balance constraints, the starting bulk composition should remain inside the
Di
Di
a
b E2
1300
diopside Fr
diopside
Eq
Eq
S X
E1
Ab
00
MI
S
1200
13
E2
1300
13
plagioclase 14
00
1
P3
P2
X
00 14
0 50
00
15
00
X2
P1
An
P2 P1
An
FIG. 5-7. Crystallization in the Diopside (Di) – Anorthite (An) – Albite (Ab) model system at atmospheric pressure. a. Fractional and equilibrium crystallization of liquid X. b. Equilibrium crystallization of the liquid X in a large pool of magma, and in a small melt inclusion inside plagioclase P1. See text for discussion.
112
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
triangle formed by tie-lines between equilibrium phases: the liquid, diopside and plagioclase. As soon as the initial bulk system composition moves out of the triangle, the liquid is exhausted. In our example of bulk composition X, the plagioclase– diopside cotectic is reached during equilibrium crystallization at the point Eq (Fig. 7a), when the plagioclase solid solution has composition P2, and the final liquid composition is at point S, when the plagioclase reaches composition P3. Mass balance restrictions do not apply however in the case of fractional crystallization because crystal phases are instantly removed, and the bulk composition of the system is not constant. In result, the bulk composition of the system and the liquid become progressively enriched in low-melting components excluded from the crystals (that is, Ab), and in our example illustrated in Fig. 5-6a the enrichment results in the shift of the line of liquid descent to the left from the equilibrium curve, towards the Ab corner. The saturation in diopside is reached at slightly lower temperature at point Fr, and in the ideal fractionation case, crystallization should proceed completely along the cotectic until a small liquid fraction reaches the eutectic point E1 on the Ab–Di border join. The final composition of the plagioclase solid solution corresponds in this case to pure Ab. Crystallization in natural gabbroic intrusions is thought to be close to ideal fractionation. So, if the point X represents the composition of a parental liquid in a large magma chamber, the liquid will evolve along the X–Fr–E1 path; it will form anorthositic cumulates followed by a layer of gabbro, and the composition of cumulus plagioclase will change from P1 to almost pure albite. Imagine now that the first liquidus plagioclase P1 traps at 1400oC a melt inclusion of liquid X. The plagioclase host with the inclusion is then buried in crystal cumulus and thus becomes separated from the main magma reservoir. Accordingly, the melt inclusion and the host crystal can be considered as a new isolated system. Contrary to the main magma chamber, crystallization in the inclusion is not likely to proceed in fractionation mode. Slow cooling provides plenty of time for chemical diffusion and approach to perfect equilibration between the host and the small melt droplet. Evolution of the newly formed inclusion system is illustrated in Fig. 5-7b, where it is compared with the earlier outlined equilibrium liquid evolution path in the main magma reservoir. The entrapment changes not only
the crystal–liquid equilibration mode, but also the bulk composition of the system and the crystal/liquid mass proportion. The inclusion system is dominated by the host, and the new bulk is represented by the point X2 located on the X–P1 tieline very close to P1. Because of the much greater crystal/liquid mass ratio, the equilibrium liquid evolution path in the inclusion will not follow the equilibrium path X–Eq–S of the main magma reservoir. Instead, the liquid in the inclusion will evolve along the segment X–MI of the curve X2–X– MI (Fig. 5-7b) and be shifted to the right, towards higher An contents. The plagioclase–diopside cotectic will be reached in the inclusion at somewhat higher temperature (point MI instead of Eq or Fr) where diopside daughter crystal(s) will nucleate in the inclusion, while the composition of the host will change only slightly, from the original point P1 to P2 (Fig. 5-7b). The small amount of remaining liquid will be exhausted at the point MI or very close to it, and crystallization will stop without evolving much further along the cotectic. This is required by mass balance considerations, as the bulk composition of the system X2 has already reached the tie-line Di–P2. Let us compare two imaginary heating experiments: (1) melting of a crystal mixture with bulk composition X, and (2) re-melting of the crystallized melt inclusion X in plagioclase P2. The experiment involving the crystal mixture (or, in other words, rock melting experiment) will represent a reversal of equilibrium crystallization in the gabbroic intrusion with a solidus at the temperature of the point S, and liquid composition evolving along the path S–Eq–X with the final dissolution of plagioclase at point X and 1400oC (Fig. 5-6b). In contrast, the inclusion will not melt up to the temperature of point MI. Diopside daughter crystals will be dissolved isothermally at this point. If the MI melt is quenched into glass and analyzed by a microprobe, it will be found to be in thermodynamic equilibrium with the plagioclase host P2, so there will be no reason for heating the inclusion further up to 1400oC and the liquid composition X. This example demonstrates that we should not expect perfect agreement between rock melting and inclusion homogenization runs, and that crystallized melt inclusions should be used with caution for evaluation of entrapment temperatures and reconstruction of the original trapped melt composition. In general, one should expect daughter crystal assemblages of plagioclase-hosted inclusions to be somewhat more refractory than the bulk rocks 113
ILYA V. VEKSLER
in which they form. The Di–Ab–An ternary does not, of course, reflect the full complexity of natural basaltic melts with 10 or 12 components. Nevertheless, the general tendency still holds, and natural plagioclase-hosted crystallized melt inclusions from gabbroic rocks are likely to melt at somewhat higher temperatures than those expected from rockmelting studies. Such components as Na2O, K2O and FeO, which normally accumulate in residual liquids and strongly reduce solidus temperatures, will dissolve in plagioclase solid solution in significant amounts, while water content in trapped liquids may decrease by hydrogen diffusion through the host phase. The irreversible losses of these components by diffusion are expected to take place during slow cooling and crystallization of host plagioclase crystals in the magma chamber, and the net result should be similar to that in the model system as illustrated in Fig. 5-7b. The plagioclase solid solution belongs, according to Bowen (1928), to a continuous reaction series, and other host minerals, such as Mg–Fe silicates, may affect the composition of crystallized melt inclusions in a similar way. For example, significant FeO losses were implied for some olivine-hosted melt inclusions in volcanic rocks (Danyushevsky et al. 2002b). The diffusion effects are likely to be more pronounced in the plutonic rocks than in their volcanic analogues because of much greater crystallization/equilibration time, which may bring diffusion to completion.
from the spinel host. In the spinel-hosted crystallized melt inclusions from the Merensky Reef, Bushveld Complex (Li et al. 2005) the average modal proportion is 41% Na-phlogopite, accompanied by 32% orthopyroxene, 14% Kphlogopite and 13% hornblende. Hence, not only the Na-rich phlogopite, but all of the other major daughter minerals, including orthopyroxene, may form by peritectic reactions of volatile-rich felsic melt with the spinel host. The bulk inclusion composition estimated from the mineral modal proportions is extremely MgO-rich (more than 26 wt.% MgO, Table 1), and obviously this cannot be a feature of the original trapped melt composition, but rather the result of peritectic reactions. Phlogopite melts incongruently with the formation of olivine and melt (Yoder & Kushiro 1969), so re-melting of the spinel-hosted crystallized melt inclusions is expected to produce abundant olivine. This was confirmed in homogenization runs by Spandler et al. (2005) with inclusions from the Stillwater Complex. Complete dissolution of olivine has been observed by the authors only at very high temperatures close to 1450oC. Melts with high initial silica content may react in a similar way not only with spinel, but also with olivine and pyroxene hosts. In the Skaergaard intrusion, the formation of abundant daughter orthopyroxene was observed in olivine-hosted inclusions of immiscible silica-rich melts from the Upper Zone (Fig. 5-3b; see also Jakoben et al. 2005), and mica appears to form by reaction of melts trapped in augite. Large amounts of MgO, Cr2O3 and other refractory components not present in trapped melts can be incorporated in daughter crystal assemblages by peritectic reactions with their hosts.
Chromian spinel and discontinuous reaction series. While cooling, trapped melts and their hosts may move out of equilibrium, and some daughter minerals may form by peritectic reactions, which are characteristic of discontinuous reaction series (Bowen 1928). The formation of abundant Naphlogopite daughter crystals which is typical for crystallized inclusions of felsic melts in chromian spinel (Li et al. 2005, Schiano et al. 1997, Spandler et al. 2005) seems to be an example of such a reaction, as expressed by the following equation:
MELT INCLUSIONS AND MELT RECONSTRUCTIONS Primary glassy melt inclusions in phenocrysts from volcanic rocks are broadly used as snapshots of melt composition at the moment of inclusion entrapment, and similar reconstructions of melt composition are also attempted using melt inclusions in plutonic rocks. Fully crystallized daughter mineral assemblages are however not so straightforward and applicable as naturally quenched volcanic glasses. As discussed in the previous section, post-entrapment processes are likely to significantly affect melt inclusions in plutonic rocks. What can be done, if anything, to compensate for these effects, and how to correctly
3MgO (spinel) + NaAlSi3O8 (melt) + H2O (melt) = NaMg3AlSi3O10(OH)2 The equation shows that the formation of daughter phlogopite during crystallization of a melt inclusion requires large amounts of the MgO component on a molar basis, apparently coming 114
CRYSTALLIZED MELT INCLUSIONS IN GABBROIC ROCKS
interpret glass compositions produced by re-melting of crystallized plutonic melt inclusions? I address these issues in this section, but the diversity of actual situations, which could be encountered in nature, is so great that only some general guidelines and principles can be given here. My main advice is to always check observations of melt inclusions against general thermodynamic criteria, and to use the data from relevant quench experiments with basaltic compositions as much as possible.
experimental points spread over a broad field at high temperatures, but it also appears that the lowtemperature evolution of gabbroic liquids and solid solutions analogous to the Upper Zone in Skaergaard is poorly covered by experiments. Liquid compositions from the same experiments are plotted in Fig. 5-9 upon a (Ab+Or)–An–(Di+Hy) projection. Only liquids saturated in both plagioclase and pyroxene (+/– other phases) have been selected. For the projection, liquid compositions were first recalculated into the conventional CIPW norms (by weight), and all the pyroxene norms were combined in one corner, while the plagioclase norms were distributed between the other two. The projection is analogous to the Ab–An–Di model ternary (Fig. 5-7), and the anorthite–diposide cotectic from the ternary is shown in Fig. 5-8 for comparison. Notably, the multicomponent cotectic liquid compositions are shifted towards the pyroxene corner from the position of the cotectic in the ternary. The apparent expansion of the plagioclase field probably results from lower melting temperatures of Fe-bearing pyroxenes. The plot also reveals a large overall scatter of the experimental points implying once again that plagioclase and pyroxenes can cocrystallize from a broad range of liquids with
Gabbroic liquids interpreted via rock-melting experiments Parental melts of gabbros are by definition those which crystallize liquidus mineral assemblages of plagioclase with one or two pyroxenes characteristic of natural gabbroic rocks. Chemical compositions of such liquids are constrained by numerous quench experiments with synthetic and natural bulk-rock compositions. They can be modeled reasonably well by experimentbased computer simulation models, and research on melt inclusions should take full advantage of the available experimental data. As mentioned above, plagioclase and pyroxenes, key minerals of natural gabbros, are phases characterized by complex multicomponent solid solutions. Apart from the physical intensive parameters, such as pressure P and temperature T, the composition of plagioclase strongly depends on the chemical potentials of CaO, Na2O, Al2O3, and SiO2, while the composition of low and high-Ca pyroxenes depends primarily on the MgO, FeO, CaO and SiO2 components. There is a general tendency for plagioclase to become more Na-rich (at the expense of Ca), while pyroxene becomes more Fe-rich (at the expense of Mg) with falling temperature. However, the Ca/Na and Mg/Fe substitutions in coexisting phases are to a large extent independent, and equilibrium compositions of coexisting plagioclase and pyroxene can vary broadly. This is illustrated in Fig. 5-8 where nearly 200 analyses of coexisting plagioclase and pyroxenes from experiments (Juster & Grove 1989, Sano et al. 2001, Snyder et al. 1992, Thy & Lofgren 1994, Thy et al. 1998, 1999, Toplis et al. 1994, Toplis & Carroll 1995, Yang et al. 1996) are plotted in terms of their Na/Ca and Fe/Mg values and compared with the natural trend in the Layered Series of the Skaergaard intrusion. Most of the experiments compiled here were carried out at atmospheric pressure, but a recent study at 1 GPa (Villiger et al. 2004) is also included. Overall,
(Na+K)/(Ca+Na+K) in plagioclase
0.8
0.6
UZ MZ
0.4
LZ
1 atm runs 1 GPa 0.2
Skaergaard
0.0
0.2
0.4
0.6
0.8
1.0
Fe/(Mg+Fe) in clinopyroxene FIG. 5-8. Coexisting compositions of plagioclase and pyroxene solid solutions in experimental products and in the Layered Series of the Skaergaard intrusion. Compositional intervals of the Lower, Middle and the Upper Zones of Skaergaard are indicated as LZ, MZ, and UZ. The experiments at 1 GPa are by Villiger et al. 2004. The arrow shows the evolution trend in the Layered Serties of the Skaergaard intrusion from the Lower Zone upwards.
115
ILYA V. VEKSLER
Di+Hy
gabbroic rocks resulted in no more than a dozen relevant references. This modest number is in striking contrast with the large and ever-growing number of papers dealing with glassy or partly crystallized melt inclusions from basaltic rocks. Two approaches are routinely used for evaluations of bulk-melt compositions of crystallized melt inclusions (see also Bodnar & Student 2006). In the first method, melt compositions are calculated from compositions and statistically representative modal proportions of daughter minerals (e.g., Loferski & Arculus 1993, Hanghøj et al. 1995). In the second method, which is broadly used also in volcanic rocks, inclusions and their hosts are heated in furnaces or special heating stages under the microscope, quenched to glasses, and the glasses are analyzed by microprobe (e.g., Kamenetsky 1996, Schiano et al. 1997, Spandler et al. 2005). Both methods have their advantages and drawbacks, but the first one is generally less reliable, because it does not account for a possible heterogeneous trapping, and does not provide clear criteria for the amount of the host that should be incorporated into the calculated melt. In the second method, the inclusion liquid is allowed to dissolve as much host as it can, but the question remains, at which temperature the heating should be stopped. Complications arising from post-entrapment reactions have been already addressed in previous sections. Surprisingly, I could not find a single published analysis of a partially homogenized melt inclusion from gabbroic rocks. Multiply saturated liquids are greatly praised among experimental petrologists, but seem to be neglected by melt inclusion specialists. Partially molten inclusions, which retain some of their daughter crystal phases, could provide valuable information about crystallization sequences and liquid evolution, and obviously deserve more attention in the future. For example, glasses in re-heated plagioclase-hosted inclusions with remnants of daughter pyroxene would directly provide temperature and compositional constraints for a liquid on the plagioclase–pyroxene cotectic. Even if only one cotectic point is determined, a full line of liquid descent could be reconstructed with the help of experiments or computer modeling. On the other hand, it is not clear why one would want to completely homogenize those melt inclusions, which almost surely were compromised by post-entrapment peritectic reactions, or resulted from heterogeneous trapping. Complete
pyroxene
plagioclase
Ab+Or
An
FIG. 5-9. Gabbroic experimental liquids saturated with plagioclase and clinopyroxene (diamonds) at 1 atm and pressures up to 1 GPa. Experiments on the Skaergaard bulk compositions (Toplis & Carroll 1995) are highlighted by open circles and the solid trendline. The diopside–plagioclase cotectic in the Di – An – Ab system (Fig. 6) is shown by the dashed line.
variable (Fe/Mg) and ((Na+K)/Ca) values. A single fixed starting composition produces, however, a tighter liquid evolution trend, as demonstrated for example by the Skaergaard experimental liquids of Toplis & Carroll (1995). The plot in Fig. 5-9 shows that the compositional field of gabbroic liquids is broad, and it will probably broaden more when all other available experiments are plotted. More importantly, however, the field is very well defined. The experiments put rigorous constraints upon the compositional span of possible gabbroic liquids and crystal-melt equilibria. A proposed melt inclusion composition from a gabbroic cumulate should normally plot somewhere within the experimental field, and the liquid composition should be in reasonable agreement with the compositions of cumulus crystals. If this is not the case, and the deviation from equilibrium experimental liquids is strong, the inclusion should be considered as suspicious and treated with caution. Some explanations should be offered to account for the discrepancy. Unfortunately, this is not always a common practice when working with crystallized melt inclusions in gabbroic rocks. Gabbroic liquids from crystallized melt inclusions My search for published melt compositions based on studies of crystallized inclusion in 116
CRYSTALLIZED MELT INCLUSIONS IN GABBROIC ROCKS
homogenization of such inclusions would require significant over-heating, and the resulting glass compositions are likely to be strongly contaminated by the host. Let us examine three examples (Table 5-1). The first column presents a calculated average of plagioclase-hosted melt inclusions from anorthositic layers of the Stillwater complex (Loferski & Arculus 1993). The inclusions are found in central parts of cumulus plagioclase crystals with An75–An77; they are composed of high-Ca pyroxene, ilmenite, and apatite and occasionally contain small grains of baddeleyite (ZrO2). The inclusions are surrounded by rims of more Ca-rich plagioclase with up to An90–An93. The average bulk composition in Table 5-1 is based on visual estimates of the modal proportions and quantitative microprobe analyses of daughter minerals, and it includes 34 vol.% of plagioclase An90, which was added to account for host crystallization at the inclusion rims. The unusual liquid composition was interpreted as a melt formed by silicate liquid immiscibility from a parental anorthositic magma. In my view, the calculation method is dubious in two important aspects. First, olivine- and plagioclase-hosted melt inclusions from volcanic rocks (Philpotts 1982, Roedder & Weiblen 1970; see also Fig. 5-5) demonstrate that crystals growing in emulsions of two immiscible silicate liquids usually trap both liquids in variable proportions, and almost never trap a pure endmember composition. If liquid immiscibility had been taking place during the formation of anorthositic cumulate layers, the inclusions were likely to form by heterogeneous trapping, and the average composition in Table 5-1 is likely to represent some effective mixture of the conjugate liquids. Secondly, the addition of large amounts of plagioclase An90 seems poorly justified. In complex, heterogeneous inclusions it is very hard to correctly compensate for the host crystallization on inclusion walls. The observed Ca-rich rims around the inclusions may have been formed by Ca diffusion into the host from trapped liquid(s), while alumina and silica, which could have resided in the host, were re-distributed to accommodate for the additional Ca. In view of the potential for heterogeneous trapping and apparently extensive postentrapment re-equilibration with the host, correct and meaningful modal mass-balance calculations are hardly possible in this particular case. Heating experiments could work much better for checking the immiscibility hypothesis and evaluating the conjugate liquid compositions. For example, such
experiments have been successfully carried out with crystallized plagioclase-hosted melt inclusions in volcanic rocks (Krasov & Clocchiatti 1979). The remaining two columns in Table 5-1 represent estimated melt compositions of chromitehosted melt inclusions from the Stillwater and Bushveld. Different methods were used, but MgO and/or FeO contents seem to be over-estimated in both cases. The reasons have been already discussed in one of the previous sections, and they mostly lie in extensive post-entrapment melt–host peritectic reactions. It should be noted that the Bushveld inclusion composition was also estimated to contain 1.8 wt.% Cr2O3, mostly residing in mica daughter minerals (Li et al. 2005). Liquidus phase equilibria of the proposed “melts” make little sense, even if the melts are thought to be hybrid. Modeling of equilibrium crystallization of the bulk composition from the second column of Table 5-1 using the COMAGMAT software (Ariskin 1999) gives olivine (Fo84) crystallization at 1500oC joined by ilmenite at about 1490oC, and no other crystal phases down to at least 1400oC and 50% crystallization. A similar modeling of the Bushveld composition by Li et al. (2005) from the third column in Table 5-1 predicts liquidus olivine Fo94 at 1620oC, and no other phases down to at least 1340oC. It is not clear how well such exotic compositions can be handled by the COMAGMAT or MELTS computation codes, but obviously the compositions and crystallization sequences have little to do with realistic processes in real magma chambers. CONCLUSIONS Evidence from melt inclusions can be of great help for resolving long-standing enigmas and controversies of basalt and gabbro petrogenesis. Recent examples of successful melt inclusion studies demonstrate the effectiveness of the method for revealing general chemical trends of melt evolution, magma mixing events, unusual local liquid compositions, and various types of liquid immiscibility in gabbroic magma chambers. Whatever the objectives, great care and clear understanding of potential pitfalls are needed for correct interpretation of the melt inclusion record in slowly cooling plutonic rocks. For deeper insights and correct interpretations, inclusions should be considered in a broader context of experimental, geochemical and textural evidence from bulk-rock specimens. Melt inclusion research in gabbro is 117
ILYA V. VEKSLER
lagging far behind contemporary studies of melt inclusions in basalt and other volcanic rocks. This is disappointing, because very few, if any, mafic igneous rocks available to us come directly from the mantle; they come from magma chambers instead. The debate on key issues of basalt and gabbro petrogenesis is as heated as ever, implying that some essential parts are missing from our conventional picture of igneous processes in magma chambers. We have to understand magma chambers better before we try to explain subtle geochemical variations in volcanic rocks by peculiarities of mantle sources. Otherwise, we may risk obscuring poorly understood phenomena by processes that we understand even less. Natural glasses and quenched aphyric lavas are not likely to cover the full diversity of basaltic liquids, because not all melt compositions are expected to erupt, and some of the liquids might be sampled only in melt inclusions formed in plutonic environments at depth. Crystallized melt inclusions appear to be among those features of gabbroic rocks, which have not received enough attention yet, and could bring important discoveries in the future.
2. Try to heat up representative multi-solid inclusions, preferably in a heating stage under the microscope. Carefully document the disappearance of daughter phases. The homogenization temperatures may be somewhat higher than expected for that particular cumulus composition. Analyze quenched glasses generated at different temperatures, in partly and totally homogenized inclusions. This can give valuable information about the crystallization sequence and liquid line of descent in the main magma reservoir. 3. Compare liquid compositions in the inclusions with relevant liquids from rock-melting experiments. One can also try to run liquids with MELTS or COMAGMAT. Make sure that the liquid compositions and their liquidus phase equilibria are reasonable. 4. Some inclusions should be analyzed for trace elements and volatile components. Very few studies of this kind have been carried out in gabbroic rocks so far. A systematic study may give important insights into the origin of rhythmic layering and the fate of volatiles in the course of magma evolution. Degassing, assimilation, magma mixing, and liquid immiscibility events may be also revealed by melt inclusions.
A SHORT MANUAL OF MELT INCLUSION STUDIES IN GABBROIC ROCKS Finally, I present here a summary of tips and hints (adapted to specific conditions in gabbro environment) that may be useful for novices in melt inclusion research.
ACKNOWLEDGEMENTS I am grateful to my former colleagues at the Vernadsky Institute in Moscow, Russia, Alexander Sobolev, Leonid Danyushevsky, Andrey Gourenko, Boris Romanchev, Dima Kamenetsky, and Igor Nikogosyan for sharing with me their insights and views on numerous aspect of melt inclusion research. Discussions with C. Kent Brooks, Troels Nielsen, Christian Tegner and Jakob Jakobsen have been of great help during the preparation of the chapter. The original version of the manuscript was revised in response to critical remarks by Ed Mathez, Alexander McBirney, Jim Webster, and an anonymous reviewer. My work on basaltic and gabbroic melt inclusions at GFZ Potsdam, and the Natural History Museum, London was supported by DFG priority programme “Formation, transport and differentiation of silicate melts”, by CERCAMS, and a grant from the European SYNTHESYS program.
1. Look for groups of small crystals enclosed in cumulus hosts. Document the petrography, check for chemical zoning and rims around the inclusions; and study crystal inclusions by microprobe. To qualify as crystallized melt inclusions, the crystal groups should meet certain criteria. Modal proportions of individual crystal phases in the groups should be reasonably stable, but mineral compositions may deviate from those in the cumulus. Unusual, exotic compositions of enclosed solids and/or occurrences of phases, which were not stable in the cumulus assemblages (e.g., apatite, mica or baddeleyite), present strong arguments implying that the crystal groups are indeed melt inclusions. Compositional zoning of the host around the inclusions is also a good indication that one is dealing with crystallized melt inclusions.
118
CRYSTALLIZED MELT INCLUSIONS IN GABBROIC ROCKS
techniques, advantages and complications: Chem. Geol. 183, 5–24.
REFERENCES ARISKIN, A.A. (1999): Phase equilibria modelling in igneous petrology: use of COMAGMAT model for simulating fractionation of ferrobasaltic magams and the genesis of high-alumina basalts. J. Volcanol. Geotherm. Res. 90, 115–162.
DANYUSHEVSKY, L.V., SOKOLOV, S. & FALLOON, T.J. (2002b): Melt inclusions in olivine phenocrysts: using diffusive re-equilibration to determine the cooling history of a crystal, with implications for the origin of olivine-phyric volcanic rocks. J. Petrol. 43, 1651–1671.
ARISKIN, A.A (2002): Geochemical thermometry of layered series rocks of the Skaergaard intrusion. Petrology 10, 495–518.
DANYUSHEVSKY, L.V., LESLIE, R.A., CRAWFORD, A.J. & DURANCE, P. (2004): Melt inclusions in primitive olivine phenocrysts: the role of localized reaction processes in the origin of anomalous compositions. J. Petrol. 45, 2531– 2553.
BALLHAUS, C.G. & STUMPFL, E.F. (1986): Sulfide and platinum mineralization in the Merensky Reef: evidence from hydrous silicates and fluid inclusions. Contrib. Mineral. Petrol. 94, 193– 204.
DE, A. (1974): Silicate liquid immiscibility in the Deccan Traps and its petrogenetic significance. Geol. Soc. Am. Bull. 85, 471–474.
BIRD, J.M., GOODRICH, C.A. & WEATHERS, M.S. (1981): Petrogenesis of Uifaq iron, Disko Island, Greenland. J. Geophys. Res. 86, 11787– 11805.
DEER, W.A., HOWIE, R.A. & ZUSSMAN, J. (1966): An introduction to the rock-forming minerals. Longmans, London, 528 p.
BODNAR, R. & DE VIVO, B. (2003): Introduction. In Melt Inclusions in Volcanic Systems: Methods, Applications and Problems (B. De Vivo & R.J. Bodnar, eds.), Elsevier, ix–xi.
DIXON, S. & RUTHERFORD, M.J. (1979): Plagiogranites as late-stage immiscible liquids in ophiolite and mid-ocean ridge suites: An experimental study. Earth Planet. Sci. Lett. 45, 45–60.
BODNAR, R.J. & STUDENT, J.J. (2006): Melt inclusions in plutonic rocks: petrography and microthermometry. In Melt Inclusions in Plutonic Rocks (J.D. Webster, ed.) Min. Assoc. Can. Short Course 36, 1-25.
EALES, H.V. (2002): Caveats in defining the magmas parental to the mafic rocks of the Bushveld Complex, and the manner of their emplacement: review and commentary. Mineral. Mag. 66, 815–832.
BOUDREAU, A.E. & MCBIRNEY, A.R. (1997) The Skaergaard Layered Series. Part III. Nondynamic layering. J. Petrol. 38: 1003–1020.
ERNST, W.G. (1976): Petrologic phase equilibria. W. H. Freeman and Company, San Francisco, 333 p.
BOWEN, N. L. (1928): The Evolution of the Igneous Rocks. Princeton University Press, Princeton, NJ, 332 p.
FENNER, C.N. (1929): The crystallization of basalts. Am. J. Sci. 18, 225–253.
BROOKS, C.K. & NIELSEN, T.F.D. (1978): Early stages in the differentiation of the Skaergaard magma as revealed by a closely related suite of dike rocks. Lithos 11, 1–14.
FREZZOTTI, M.-L. (2001): Silicate melt inclusions in magmatic rocks: applications to petrology. Lithos 55, 273–299.
CLARK, A. & KONTAK, D.J. (2004): Fe–Ti–P oxide melts generated through magma mixing in the Antuata subvolcanic center, Peru: implications for the origin of nelsonite and iron oxidedominated hydrothermal deposits. Econ.Geol. 99, 377-395.
GAETANI, G.A. & WATSON, E.B. (2002): Modelling the major-element evolution of olivine-hosted melt inclusions. Chem. Geol. 183, 25–41. GHIORSO, M.S. & CARMICHAEL, I.S.E. (1985): Chemical mass transfer in magmatic processes. II. Applications in equilibrium crystallization, fractionation and assimilation. Contrib. Mineral. Petrol. 90, 121–141.
DANYUSHEVSKY, L. V., MCNEIL, A.W. & SOBOLEV, A.V. (2002a): Experimental and petrological studies of melt inclusions in phenocrysts from mantle-derived magmas: an overview of 119
ILYA V. VEKSLER
JANG, Y.D., NASLUND, H.R. & MCBIRNEY, A.R. (2001): The differentiation trend of the Skaergaard intrusion and the timing of magnetite crystallization: Iron enrichment revisited. Earth Planet. Sci. Lett. 189, 189–196.
HANGHØJ, K., ROSING, M. & BROOKS, C.K. (1995): Evolution of the Skærgaard magma: evidence from crystallized melt inclusions. Contrib. Mineral. Petrol. 120, 265–269. HANLEY, J.J., MUNGALL, J.E., BRAY, C.J. & GORTON, M.P. (2004): The origin of bulk and water-soluble Cl and Br enrichments in Sudbury Breccia in the Fraser Copper Zone, Strahcona embayment, Sudbury, Ontario, Canada. Can. Mineral. 42, 1777–1798.
JERRAM, D.A., CHEADLE, M.J. & PHILPOTTS, A.R. (2003): Quantifying the building blocks of igneous rocks: are clustered crystal frameworks the foundation? J. Petrol. 44, 2033–2051. JUSTER, T.C. & GROVE, T.L. (1989): Experimental constraints on the generation of the Fe–Ti basalts, andesites,and rhyodacites at the Galapagos Spreading Center, 85 W and 95 W. J. Geophys. Res. 94-B, 9251–9274.
HANLEY, J.J., PETTKE, T., MUNGALL, J.E. & SPOONER, E.T.C. (2005): The solubility of platinum and gold in the NaCl brines at 1.5 kbar, 600 to 800 oC: A laser ablation ICP–MS pilot study of synthetic fluid inclusions. Geochim. Cosmochim. Acta 69, 2593–2611.
KAMENETSKY, V. (1996): Methodology for the study of melt inclusions in Cr-spinel, and implications for parental melts of MORB and from FAMOUS area. Earth Planet. Sci. Lett. 142, 479–486.
HOLNESS, M.B., CHEADLE, M.J. & MCKENZIE, D. (2005): On the use of change in dihedral angle to decode late-stage textural evolution in cumulates. J. Petrol. 46, 1565–1583.
KELEMEN, P.B., KOGA, K. & SHIMIZU, N. (1997): Geochemistry of gabbro sills in the crust-mantle transition zone of the Oman ophiolite: implications for the origin of the oceanic lowed crust. Earth Planet. Sci. Lett. 146, 475–488.
HUNTER, R.H. & SPARKS, R.S.J. (1987a): The differentiation of the Skaergaard Intrusion. Contrib. Mineral. Petrol. 95, 451–461. HUNTER, R.H. & SPARKS, R.S.J. (1987b): The differentiation of the Skaergaard Intrusion. A discussion of R.H. Hunter and R.S.J. Sparks (Contrib. Mineral. Petrol. 95: 451–461). Contrib. Mineral. Petrol. 104, 240–244. Only one reference is made to Hunter & Sparks (page 3) – which one?
KNUDSEN, T.-L. & LIDWIN, A. (1996): Magmatic CO2, brine and nitrogen inclusions in Sveconorwegian enderbitic dehydration veins and gabbro from the Bamble sector, Southern Norway. Eur. J. Mineral. 8, 1041–1063. KRASOV, N.F. & CLOCCHIATTI, R. (1979): Immiscibility in silicate melts and its possible petrogenetic importance, as shown by study of melt inclusions. Trans. (Doklady) U.S.S.R. Acad. Sci. 248, 92–95.
HURAI, V., SIMON, K., WIECHERT, U., HOEFS, J., KONEČNÝ, P., HURAIOVÁ, M, PIRONON, J. & LIPKA, J. (1998): Immiscible separation of metalliferous Fe/Ti-oxide melts from fractionating alkali basalt: P–T–fO2 conditions and two-liquid elemental partitioning. Contrib. Mineral. Petrol. 133, 12–29.
KUZMIN, D.V. & SOBOLEV, A.V. (2003): Boundary layer effect on the composition of melt inclusions in olivine. Geophys. Res. Abstr. 5, 05665.
IRVINE, T.N., KEITH, D.W. & TODD, S.G. (1983): The J–M platinum–palladium Reef of the Stillwater Complex, Montana: II. Origin by double-diffusive convective magma mixing and implications for the Bushveld Complex. Econ. Geol. 78, 1278–1334.
LARSEN, R.B., BROOKS, C.K. & BIRD, D.K. (1992): Methane-bearing, aqueous, saline solutions in the Skaergaard intrusion, east Greenland. Contrib. Mineral. Petrol. 112, 428–437. LI, C., RIPLEY, E.M., SARKAR, A., SHIN, D. & MAIER, W. (2005): Origin of phlogopiteorthopyroxene inclusions in chromites from the Merensky Reef of the Bushveld Complex, South Africa. Contrib. Mineral. Petrol. 150, 119–130.
JAKOBSEN, J.K., VEKSLER, I.V., TEGNER, C. & BROOKS, C.K. (2005): Immiscible iron- and silica-rich melts in basalt petrogenesis documented in the Skaergaard intrusion. Geology 33, 885–888.
120
CRYSTALLIZED MELT INCLUSIONS IN GABBROIC ROCKS
LOFERSKI. P.J. & ARCULUS, R.J. (1993): Multiphase inclusions in plagioclase from anorthosites in the Stillwater Complex, Montana: implications for the origin of the anorthosites. Contrib. Mineral. Petrol. 114, 63–78.
PHILPOTTS, A.R. (1979): Silicate liquid immiscibility in tholeiitic basalts. J. Petrol. 20, 99–118. PHILPOTTS, A.R. (1982): Composition of immiscible liquids in volcanic rocks. Contrib. Mineral. Petrol. 80, 201–219.
LOWENSTERN, J.B. (2003): Melt inclusions come of age: volatiles, volcanoes, and Sorby’s legacy. In Melt Inclusions in Volcanic Systems: Methods, Applications and Problems (B. De Vivo & R.J. Bodnar, eds.), Elsevier, 1–22.
PHILPOTTS, A.R. BRUSTMAN, C.M., SHI, J., CARLSON, W.D. & DENISON, C. (1999): Plagioclase-chain network in slowly cooled basaltic magma. Am. Mineral. 84, 1819–1829. RIPLEY E.M., SEVERSON, M.J. & HAUCK, S.A. (1998): Evidence of sulfide and Fe–Ti–P-rich liquid immiscibility in the Duluth Complex, Minnesota. Econ. Geol. 93, 1052–1062.
MARSH, B.D. (1996): Solidification fronts and magmatic evolution. Mineral. Mag. 60, 5–40. MCBIRNEY, A.R. (1989): The Skaergaard layered series: I. Structure and average compositions. J. Petrol. 30, 363–399.
ROEDDER, E. (1984a): Fluid inclusions. Reviews in Mineralogy, 12, Mineralogical Society of America, Washington, 646 p.
MCBIRNEY, A.R. & NAKAMURA, Y. (1974): Immiscibility in late-stage magmas of the Skaergaard intrusion. Carnegie Inst. Wash. Yearbook, 73, 348–352.
ROEDDER, E. (1984b): Occurrence and significance of magmatic inclusions and silicate liquid immiscibility. Acta Geol. Polonica 34, 139–185.
MCBIRNEY, A.R. & NASLUND, H.R. (1990): The differentiation of the Skaergaard intrusion. A discussion of R.H. Hunter and R.S.J. Sparks (Contrib. Mineral. Petrol. 95:451–461): Contrib. Mineral. Petrol. 104, 235–240.
ROEDDER, E. & WEIBLEN, P.W. (1970): Lunar petrology of silicate melt inclusions, Apollo 11 rocks. Proc. Apollo 11 Lunar Sci. Conf., Geochim. Cosmochim. Acta Suppl. 1, 507–528. RYABOV, V.V. (1989): Liquation in natural glasses: The example of traps. Nauka, Novosibirsk, 221 (in Russian).
MCBIRNEY, A.R. & NICOLAS, A. (1997): The Skaergaard Layered Series: Part II. Dynamic layering. J. Petrol. 38, 569–580.
SANO, T., FUJII, T., DESHMUKH, S.S., FUKUOKA, T. & ARAMAKI, S. (2001): Differentiation processes of Deccan Trap Basalts: Contribution from geochemistry and experimental petrology. J. Petrol. 42, 2175–2195.
MCBIRNEY, A.R. & NOYES, R.M. (1979): Crystallization and layering of the Skaergaard intrusion. J. Petrol. 20, 487–564. MORSE, S.A. (1980): Basalts and phase diagrams. An introduction to the quantitative use of phase diagrams in igneous petrology. Springer-Verlag, 493 p.
SCHIANO, P., CLOCCIATTI, R., LORAND, J.-P., MASSARE, D., DELOULE, E. & CHAUSSIDON, M. (1997): Primitive basaltic melts included in podiform chromites from the Oman Ophiolite. Earth Planet. Sci. Lett. 146, 489–497.
NAUMOV, V.B., SOLOVOVA, I.P., KOVALENKER, V.A. & RUSINOV, V.L. (1993): Immiscibility in acidic magmas: Evidence from melt inclusions in quartz phenocrysts of ignimbrites. Eur. J. Mineral. 5, 937–941.
SNYDER, D., CARMICHAEL, I.S.E. & WIEBE, R.A. (1992): Experimental study of liquid evolution in an Fe-rich, layered mafic intrusion: Constraints of Fe–Ti oxide precipitation on the T–fO2 and T–P paths of tholeiitic magmas. Contrib. Mineral. Petrol. 113, 73–86.
NIELSEN, T.F.D. (2004): The shape and volume of the Skaergaard intrusion, Greenland: Implications for mass balance and bulk composition. J. Petrol. 45, 507–530.
SONNENTHAL, E.L. & MCBIRNEY, A.R. (1998): The Skaergaard Layered Series. Pert. IV. Reactiontransport simulations of foundered blocks. J. Petrol. 39, 633–661.
OSBORN, E.F. (1979): The reaction principle. In The Evolution of Igneous Rocks. Fiftieth Anniversary Perspectives (H.S. Yoder, Jr., ed.), Princeton University Press, Princeton, 133–169. 121
ILYA V. VEKSLER
WAGER, L.R. & BROWN, G. M. (1968): Layered igneous rocks. Oliver and Boyd, Edinburgh and London, 588 p.
SOURS-PAGE, R., NIELSEN, R. & BATIZA, R. (2002): Melt inclusions as indicators of parental magma diversity on the northern East Pacific Rise. Chem. Geol. 183, 237–261.
WAGER, L.R. & DEER, W.A. (1939): Geological investigation in East Greenland, Part III, The petrology of the Skaergaard intrusion, Kangerdluqssuaq, East Greenland. Meddelser om Grønland 105, 352 p.
SPANDLER, C., MAVROGENES, J. & ARCULUS, R. (2005): Origin of chromitites in layered intrusions: Evidence from chromite-hosted melt inclusions from the Stillwater Complex. Geology 33, 893–896.
WALLACE, P. & CARMICHAEL, I.S.E. (1992): Sulfur in basaltic magmas. Geochim. Cosmochim. Acta 56, 1863–1874.
TEGNER, C. (1997): Iron in plagioclase as a monitor of the differentiation of the Skaergaard intrusion. Contrib. Mineral. Petrol. 128, 45–51.
WIEBE, R.A. (1979): Fractionation and liquid immiscibility in an anorthositic pluton of the Nain Complex, Labrador. J. Petrol. 20, 239–269.
THY, P. & LOFGREN, G.E. (1994): Experimental constraints on the low-pressure evolution of transitional and mildly alkalic basalts: the effect of Fe–Ti oxide minerals and the origin of basaltic andesites. Contrib. Mineral. Petrol. 116, 340– 351.
WÖRNER, G., FRICKE, A. & BURKE, E.A.J. (1993): Fluid-inclusion studies on lower crustal gabbroic xenoliths from the Mt. Melbourne Volcanic field (Antarctica): Evidence for the post-crytsallization uplift history during Cenozoic Ross Sea Rifting. Eur. J. Mineral. 5, 775–785.
THY, P., LESHER, C.E. & FRAM, M.S. (1998): Low pressure experimental constraints on the evolution of basaltic lavas from site 917, southeast Greenland continental margin. Proc. Ocean Drilling Progr. Sci. Results 152, 359–372.
YANG, H.-J., KINZLER, R.J. & GROVE, T.L. (1996): Experiments and models of anhydrous, basaltic olivine–plagioclase–augite saturated melts from 0.001 to 10 kbar. Contrib. Mineral. Petrol. 124, 1–18.
THY, P., LESHER, C.E. & MAYFIELD, J.D. (1999): Low-pressure melting studies of basalt and basaltic andesite from the southeast Greenland continental margin. Proc. Ocean Drilling Progr. Sci. Results 163, 95–112.
YODER, H.S. & KUSHIRO, I. (1969): Melting of a hydrous phase: phlogopite. Am. J. Sci. 267-A, 558–565.
TOPLIS, M.J. & CARROLL, M.R. (1995): An experimental study of the influence of oxygen fugacity on Fe–Ti oxide stability, phase relations, and mineral-melt equilibria in ferro-basaltic systems. J. Petrol. 36, 137–1171. TOPLIS, M.J., LIBOUREL, G. & CARROLL, M.R. (1994): The role of phosphorus in crystallization processes of basalt: an experimental study. Geochim. Cosmochim. Acta 58, 797–810. UPTON, B.G.J., SKOVGAARD, A.C., MCCLURG, J., KIRSTEIN, L., CHEADLE, M., EMELEUS, C.H., WADSWORTH, W.J. & FALLICK, A.E. (2002): Picritic magmas and the Rum ultramafic complex, Scotland. Geol. Mag. 139, 437–452. VILLIGER, S., ULMER, P., MÜNTENER, O. & THOMPSON, A.B. (2004): The liquid line of descent of anhydrous, mantle-derived, tholeiitic liquids by fractional and equilibrium crystallization – an experimental study at 1.0 GPa. J. Petrol. 45, 2369–2388. 122
CHAPTER 6: PARENTAL MAGMAS OF PLUTONIC CARBONATITES, CARBONATE– SILICATE IMMISCIBILITY AND DECARBONATION REACTIONS: EVIDENCE FROM MELT AND FLUID INCLUSIONS Ilya V. Veksler GeoForschungsZentrum Potsdam, Department 4.1, Telegrafenberg, Potsdam 14473, Germany E-mail:
[email protected] and David Lentz Department of Geology, University of New Brunswick, Box 4400, 2 Bailey Drive, Fredericton, New Brunswick E3B 5A3 Canada E-mail:
[email protected] alkali carbonatitic melt inclusions, and (3) alkalirich melt (or fluid) inclusions. These types of inclusions are commonly found together in the same rock, or even the same host mineral, and are believed to be genetically related either via fractional crystallization or liquid immiscibility. It is often very hard, or almost impossible, to distinguish between carbonatitic melt and hydrothermal fluid inclusions. The term “melt” is used here as a synonym of liquid, and applied to sub-critical conditions, where there is a clear physical distinction between liquid and vapor phases. The word “fluid” is used, firstly, as a more general term that includes liquids and vapors at sub-critical P-T conditions, but also defines a single mobile phase at supercritical conditions, where the liquid-vapor phase transition ceases to exist. In common with granitic pegmatites, plutonic carbonatites in complex silicate-carbonatitic intrusions normally crystallize very late, at temperatures, which are thought to correspond to the magmatic-tohydrothermal transition. Keppler (2003) demonstrated that the solubility of water in a synthetic CaMg-Na carbonatitic melt reached 10-11 wt.% at 100 MPa, and up to 40 wt.% of dissolved H2O was implied at the same pressure in molten Na2CO3 by in situ DTA study in the Na2CO3–H2O system (Koster van Groos 1990). Alkali carbonates in their turn are highly soluble in hydrothermal fluids at the same conditions (Ravich 1974, Koster van Groos 1990). Thus, populations of water-rich carbonatitic melt inclusions may look very much like comagmatic solute-rich hydrothermal fluid inclusions. Heterogeneous trapping and postentrapment necking-down further complicate the
INTRODUCTION The influence of melt and fluid inclusion research on petrogenetic models of carbonatite genesis has been profound, and probably much greater than for any other type of igneous rock. The exceptional role of melt and fluid inclusion data in the case of carbonatites has been due to the fact that genetic and compositional relationships between plutonic and rare extrusive carbonatite occurrences are not clear, and the extrusive rocks are not likely to provide reliable and comprehensive constraints on the composition of parental carbonatitic magma. Unlike silicate melts, carbonatitic liquids do not quench to glasses, and carbonatitic rocks, both intrusive and extrusive, may not retain the volatile and fluxing components, believed to play an important role in the origin and evolution of the parental magmas. The only carbonatitic lava, which has ever been observed to erupt, is the unique natrocarbonatite at the nephelinitic-phonolitic volcano Oldoinyo Lengai, Tanzania (Dawson 1962). Natrocarbonatite is compositionally very different from other known extrusive and intrusive carbonatites, and its origin remains, as discussed below, contentious (Bailey 1993, Bell & Keller 1995, Nielsen & Veksler 2002). Because comparisons with extrusive analogs are lacking or unreliable, melt and fluid inclusions have been widely used as a source of crucial information regarding the composition of carbonatitic melts and fluids. This chapter primarily deals with carbonatitic melts, but in fact three major types of inclusions are considered here: (1) crystallized inclusions of carbonated silicate melts, (2) low-
Mineralogical Association of Canada Short Course 36, Montreal, Quebec, p. 123-149. 123
ILYA V. VEKSLER & DAVID LENTZ
interpretation of inclusions, which show variable proportions of daughter crystals, liquid and vapor at room temperature (Roedder 1984). The distinction between carbonatitic melt and solute-rich hydrothermal fluid inclusions in many cases becomes purely semantic. In effect, they all represent liquids containing variable amounts of water and large proportions of highly soluble salts, such as alkali carbonates, sulfates and chlorides. Exsolution of carbonatitic melts and hydrothermal fluids is a very important factor in carbonatite petrogenesis, and key evidence from hydrous fluid inclusions will be also considered in this chapter. Carbonatitic glasses do not form in nature, and only rare, exotic compositions have been quenched to glasses in laboratory (e.g., Genge et al. 1995). Because of this, and the frequent decrepitation of carbonatitic melt inclusions during heating runs at atmospheric pressure, heating/ quenching methods are of limited use for the study of carbonatitic inclusions, and are often replaced by detailed petrographic and microprobe studies of daughter minerals. The common presence of unstable, water-soluble daughter minerals in the inclusions makes sample preparation and microprobe analyses also very challenging. Correct interpretation of carbonatitic inclusions is further complicated by post-entrapment modification, and in some cases, by extensive reactions of trapped fluids with the host. These and other specific challenges posed by carbonatitic melt inclusions are examined in this chapter in detail (see also Veksler et al. 1998a). The genetic relationships between carbonatites and associated silicate igneous rocks are very complex, and no existing model is likely to account for the whole diversity of carbonatite occurrences. Geologic relationships between silicate and carbonatite rock types in ultramafic-alkalinecarbonatitic intrusions are commonly chaotic, implying rapid, explosion-like emplacement of lowviscosity, volatile-rich magma. It is usually assumed that at least four or five mechanisms are needed to explain the main genetic types of extrusive and intrusive carbonatite (e.g., Gittins 1989, Kjarsgaard & Hamilton 1989, Wyllie 1989, Bailey 1993). As demonstrated below, the evidence from melt and fluid inclusions has been very helpful for deciphering the origin of silicate-carbonate igneous rock associations, and has provided important insights into carbonatite petrogenesis. Carbonatite enthusiasts like to stress the great scientific and economic importance of
carbonatites and associated silicate rocks, which is strikingly disproportionate to the small total volume of known carbonatitic intrusions and volcanic eruptions. It is certainly true that carbonatites have been extensively used as probes of the deep mantle (e.g., Harmer & Gittins 1998, Wyllie 1989, Bell & Blenkinsop 1989, Bailey 1993), and the world’s largest deposits of strategic rare metals, such as Nb and rare earth elements (REE), are associated with carbonatites (Mariano 1989, Mitchell 2005a). The interest in carbonatites has been great, but many important aspects of carbonatite petrogenesis and rare metal ore formation are still far from being fully understood. Melt and fluid inclusion techniques, in conjunction with experimental phase equilibria studies, and detailed geochemical characterization of natural carbonatitic occurrences, will no doubt continue to play a key role in carbonatite research. THE DIVERSITY OF CARBONATITES AND ASSOCIATED SILICATE ROCKS Carbonatite is defined in the IUGS system of classification as an igneous rock composed of more than 50 modal % primary (magmatic) carbonate and containing less than 20 wt.% SiO2 (Le Maitre 2002). It is sub-divided into varieties according to their modal or chemical compositions (Woolley & Kempe 1989). Thus, calcite carbonatites of the modal classification correspond to calciocarbonatites of the chemical classification (more than 80 wt.% of CaO on the volatile-free basis). Dolomite carbonatite is synonymous with magnesiocarbonatite. Ferrocarbonatite (MgO/(FeO+ Fe2O3+MnO) < 0.5 by weight) is dominated by ankerite. Natrocarbonatite is composed mostly of Na–Ca–K carbonates nyerereite [Na2Ca(CO3)2] and gregoryite [(Na, Ca, K)2CO3]. The mineralogical and chemical criteria are straightforward, but proving igneous origin has never been easy for carbonatites. Although a magmatic origin of carbonate-rich rocks in alkaline plutonic complexes was first postulated more than a century ago (Högbohm 1895, at the Alnö intrusion in Sweden), it was not widely accepted until the 1960s following experimental evidence that low-temperature carbonatitic melts can form at low, crustal pressures in the system CaO–CO2–H2O (Wyllie & Tuttle 1959, 1960), and direct observation of natrocarbonatite eruptions from the Oldoinyo Lengai volcano, Tanzania (Dawson 1962). Without reliable criteria of igneous origin, purely chemical or modal classifications may be, as pointed out by 124
PARENTAL MAGMAS OF PLUTONIC CARBONATITES
Mitchell (2005a, b), misleading, because even common calcite hydrothermal veins associated with a variety of parental magmas, could conceivably be termed as “carbonatite”. True carbonatites are usually defined by the presence of characteristic silicate minerals (e.g., phlogopite, acmite-diopside, forsteritic olivine, nepheline, and monticellite) or accessories (apatite, perovskite-type minerals, pyrochlore, REE fluor-carbonates, etc.), or by association with characteristic silicate rocks. A detailed account of the geology of carbonatite occurrences worldwide can be found in the comprehensive volumes by Woolley (1987, 2001) and Kogarko et al. (1995). Woolley (2003) summarized the geological data, and outlined six distinct silicate rock series typically found in association with carbonatites: 1. [Peridotite, pyroxenite] melilitite-nephelinitephonolite-trachyte (melilitolite). 2. [Peridotite, pyroxenite] nephelinite-phonolitetrachyte (ijolite). 3. [Peridotite, pyroxenite] basanite-trachyte (alkali gabbro). 4. Phonolite-trachyte (nepheline, sodalite and cancrinite syenites). 5. Trachite (syenite). 6. Kimberlite. Ultramafic rocks in square brackets appear to be cumulates, which do not represent melt compositions; and rock types in parentheses are the principal members of the intrusive equivalent series. Importantly, a significant number of intrusive and extrusive carbonatites do not show spatial or temporal associations with any silicate rock other than fenite (Woolley 2003). Extrusive dolomiticankeritic carbonatites in Zambia are a notable example of such occurrences (Bailey 1990). The origin and relationships with silicate rocks are also not clear for REE–F–carbonate-rich rocks such as Bayan Obo (China), Goldie (Colorado, USA), Rock Canyon Creek (British Columbia, Canada), and some others (Drew et al. 1990, Mitchell 2005a, Rankin 2005). Woolley (2003) calculated the relative abundance of the igneous rock types associated with carbonatites (Fig. 6-1). The original data have been compiled for 377 carbonatite occurrences out of the total 453 known to him at that time. (Since 2003, the total exceeds 500, Woolley 2005, personal communication). The occurrences were initially divided into two groups, depending on whether or not extrusive carbonatite is present. Extrusive carbonatites are less abundant than the plutonic
melilite-bearing rocks ijolite/nephelinite neph.syenite/phonolite syenite/trachyte basalt meimechite pyroxenite other none
extrusive carbonatites n = 40
0
10
20
30
40
50
occurrences, % peridotite pyroxenite gabbro ijolite/nephelinite neph.syenite/phonolite syenite/trachyte melilite-bearing rocks leucite-bearing rocks
intrusive carbonatites n = 337
lamprophyre kimberlite other none 0
10
20
30
40
50
occurrences, %
FIG. 6-1. Relative abundance of silicate rocks in association with extrusive and intrusive carbonatites. The data from Woolley (2003).
equivalents, and the total number of known extrusive occurrences was 40. These data showed some systematic and significant differences in the nature and proportions of the silicate rocks associated with extrusive carbonatite (Fig. 6-1). Localities of extrusive carbonatites with no associated silicate rocks are numerous, almost reaching 50%, although this could simply reflect lack of outcrop in comparatively young occurrences (Woolley 2003). The most significant observation is that nearly half of all extrusive carbonatites were erupted from centers containing melilite-bearing rocks. This is a major difference from the intrusive occurrences. An explanation for this observation offered by Woolley (2003) was that the melilitebearing rocks were more readily preserved in rapidly emplaced diatremes and dike complexes, while in larger and deeper slowly cooling complexes melilite was likely to be lost via hydrothermal alteration and replacement by later intrusions. It seems probable, therefore, that the greater abundance of melilite-bearing rocks in the 125
ILYA V. VEKSLER & DAVID LENTZ
(Dawson 1962, Bell & Keller 1995). The discovery of natrocarbonatite melt implied that liquidus temperatures of carbonatitic liquids can be decreased not only by water vapor pressure, but also by alkalis even at atmospheric pressure. These observations also corroborated earlier ideas about the presence and the important role of alkalis in parental liquids of calcitic carbonatites (von Eckermann 1948). Subsequently, silicate-carbonate liquid immiscibility was reported in the synthetic NaAlSi3O8–Na2CO3 system (Koster van Groos & Wyllie 1968). This pioneering work laid the foundation for numerous studies and models linking the origin of carbonatites to liquid unmixing. The diversity of contemporary models of carbonatite petrogenesis falls into four major groups. Because phase equilibria studies have been crucial for the formulation and development of the models, key experimental evidence is briefly outlined below for each group of hypotheses. This overview of experimental data should also help to understand the evidence from melt and fluid inclusions, as the inclusions are usually interpreted in the context of phase equilibria constraints. Isotopic, geochemical and other evidence, although important, are only briefly mentioned, or not considered at all, because a more detailed review exceeds the scope of this chapter.
extrusive carbonatite association reflects more accurately the close genetic relationship of these two rock types. Other factors controlling the melilite–carbonatite–ijolite association are discussed at the end of the chapter in more detail. OVERVIEW OF PETROGENETIC HYPOTHESES Before the 1960s, the existence of natural carbonatitic magma and igneous carbonatite was rejected primarily on physicochemical grounds. Calciocarbonatites, which were encountered in the field but commonly not recognized as igneous rocks, range in their mode to monomineralic calcite rocks. On heating, pure calcite either decomposes to CaO and CO2 or, when stabilized at a particular CO2 pressure, melts at temperatures in excess of 1300oC (Wyllie & Tuttle 1960). The high temperature of calcite crystallization was in apparent conflict with geologic observations implying late-stage, low-temperature emplacement of calciocarbonatite. Thus, limestone assimilation hypothesis was the first popular model, which attempted to explain the origin of the commonly observed association of silica-deficient igneous rocks with calcitic or dolomitic bodies. Originally, assimilation by basaltic magma was considered (Daly 1910), but later the same ideas were applied to granitic melts (Shand 1922). In the Daly-Shand hypothesis, many rocks now known as carbonatites were viewed as metasomatized limestone, marble or sedimentary dolomite. Although chemically plausible, the idea of assimilation was not supported by detailed field observations, did not agree with heat balance considerations, and the growing body of isotopic evidence, which implied a mantle origin for the vast majority of alkaline igneous rocks and carbonatites (Wyllie 1974, Bell & Blenkinsop 1989, Deines 1989). Watkinson & Wyllie (1969) also criticized the hypothesis from the experimental and phase equilibria point of view. As noted above, the crucial development in carbonatite research was due to progress in highpressure experimental studies (Wyllie & Tuttle 1959, 1960, Koster van Groos & Wyllie 1968), and the discovery of natrocarbonatite lava in Africa (Dawson 1962). Two types of low-temperature carbonatitic liquids were postulated simultaneously at that time. Calcite has been proven to melt at low, geologically reasonable temperatures in the presence of water (Wyllie & Tuttle 1959, 1960), and natrocarbonatite lava was found to have eruption temperatures as low as 540-590oC
Primary origin by partial melting of carbonated mantle source According to these models, carbonatitic liquids form by partial melting of carbonated mantle peridotite and, in order to survive decarbonation reactions, they must be brought rapidly to the surface by violent, explosive eruptions. Experimental data confirm that carbonatitic melts may be generated indeed by partial melting of carbonated peridotite at depths greater than ~70 km (Wyllie & Huang 1976, Eggler 1978, Wallace & Green 1988, Thibault et al. 1992, Dalton & Wood 1993, Dalton & Presnall 1998, Gudfinnsson & Presnall 2005). The near-solidus liquid compositions from 2 GPa to at least 7 GPa are dominated by CaCO3 and MgCO3 so that the melts are essentially dolomitic, and precise compositions, e.g., the concentrations of alkalis and other minor components, are defined by peridotite mineralogy and the bulk composition of the mantle source. Melt compositions in the CaO–MgO–Al2O3–SiO2–CO2 (CMAS–CO2) modal system show systematic variation with increasing temperature (and degree of melting) ranging from carbonatitic to kimberlitic 126
PARENTAL MAGMAS OF PLUTONIC CARBONATITES
melt compositions, that is, from about 5 to 40 wt.% SiO2 (Dalton & Presnall 1998, Gudfinnsson & Presnall 2005) During ascent, carbonate-rich magmas retaining equilibrium with mantle lherzolite will react, crystallize and release CO2 vapor, increasing the proportion of clinopyroxene to orthopyroxene in the host mantle rock. Calciocarbonatite liquids can be generated at lower pressures (depth between 40 and 70 km), but the shallowest experimental liquids were shown to contain a maximum of 73 wt.% CaCO3 with 18 wt.% silicate components (Wyllie & Lee 1998). A primary mantle origin of natrocarbonatite has also been examined experimentally, but appeared unlikely (Sweeney et al. 1995). In natural occurrences, a primary origin has been proposed for extrusive dolomitic carbonatites in Africa (e.g., Bailey 1990), and dolomitic carbonatite-kimberlite or carbonatite-aillikite association reported in the Safartoq region in West Greenland (Larsen & Rex 1992, Dalton & Presnall 1998, Mitchell et al. 1999, Tappe et al. 2004). Harmer & Gittins (1998) advocated the primary origin for more common carbonatite occurrences based on isotopic evidence from a number of plutonic complexes in Africa where calcitic and dolomitic carbonatites are found in spatial association with alkaline silicate rocks. As discussed below, melt inclusion evidence supports primary carbonatitic liquids in kimberlite and some other rocks of deep mantle origin.
Origin by silicate-carbonate liquid immiscibility This is another plausible mechanism for an origin from carbonated silicate parental magma, and the favored petrogenetic model for Oldoinyo Lengai natrocarbonatite. Silicate-carbonate liquid immiscibility at crustal and mantle pressures has been studied in numerous experiments (Kjarsgaard & Hamilton 1989, Kjarsgaard 1998, Brooker 1998, Lee & Wyllie 1998, Wyllie & Lee 1998, and references therein). Broad miscibility gaps were found in alkali-rich compositions, but it was concluded that they do not propagate to alkali-free systems (Lee et al. 1994, Lee & Wyllie 1998). Silicate–carbonate immiscibility is also unlikely in natural magmatic liquids at mantle pressures (Wyllie & Lee 1998). Melt inclusions (e.g., Rankin & Le Bas 1974, Romanchev & Sokolov 1980, Panina 2005) offer spectacular examples of carbonate-silicate liquid immiscibility (see below).
Derivative origin from a carbonated silicate magma by fractional crystallization Some types of silica-undersaturated mafic extrusive rocks (e.g., olivine melilitite or potassic kamafugite) are believed to represent primary, mantle-derived liquids, and their generation at mantle depth requires the significant involvement of CO2 and carbonate components (Brey & Green 1977, Foley et al. 1987). Such silica-deficient, carbonated liquids are favored candidates for parental magmas of ultramafic-alkaline-carbonatitic intrusive complexes (see the previous section). Low-alkali carbonatitic liquids may be generated as residua by extensive fractional crystallization of silicates, and typical monominerallic calcite or dolomite rocks may crystallize from the residual carbonatitic liquids after, or during the removal of alkalis by fenitizing hydrothermal fluids (von Eckermann 1948). The possibility of a gradual transition from silicate-precipitating melts to carbonate-precipitating residual liquids has been
Alternative models for natrocarbonatite and anatectic crustal carbonatites Bailey (1993) and Nielsen & Veksler (2002) put forward a number of phase equilibria and geochemical arguments against the origin of natrocarbonatite lava by direct liquid immiscibility from peralkaline nephelinitic or phonolitic magma. They proposed a two-stage process, in which sodarich salt precipitates are first deposited in the volcanic plumbing system, and within the edifice of the nephelinitic-phonolitic volcano by orthomagmatic hydrothermal fluids, and the dried fluid residue is then re-melted and re-mobilized by heat coming from a newer pulse of ascending silicate magma. The model links the origin of the unique natrocarbonatite to fenitizing fluids, which are thought to exsolve from more common, low-alkali, calcite-precipitating carbonatitic liquids, and helps to explain the liquid evolution across the calcitenyerereite cotectic, and the nyerereite thermal barrier of the CaCO3–Na2CO3–K2CO3 system
examined experimentally at 200 MPa in the alkalifree CaO–MgO–SiO2–CO2–H2O synthetic system (Otto & Wyllie 1993), and with NaAlSiO4–CaCO3– H2O mixtures containing 25 wt.% H2O (Watkinson & Wyllie 1971). The experimental systems are, however, lacking important components of natural carbonated magmas, and may not be directly applicable to natural magmatic liquids. The evidence from natural melt inclusions provides, as described below, crucial insights in the process of fractional crystallization of carbonated silica-deficient melts.
127
ILYA V. VEKSLER & DAVID LENTZ
(Cooper et al. 1975). The fluid inclusion evidence for fenitizing fluids is further discussed below in more detail. Although the magmatic origin of the majority of known carbonatite occurrences is strongly supported by isotopic evidence (Bell & Blenkinsop 1989), some calciocarbonatites may form in the crust by fluid-induced anatectic melting (volatile fluxing to syntectic digestion)of crustal carbonate rocks. Such a mechanism has been proposed, for example, for the Grenville Province “carbonatites” (Lentz 1998, 1999).
carbonatite petrogenesis, where prime examples of natrocarbonatite lava (Dawson 1962) and extrusive dolomitic carbonatite (Bailey 1990) were first found. Accordingly, nephelinitic-phonolitic volcanoes and ijolitic-carbonatitic subvolcanic complexes in East Africa were among the first locations where carbonatitic melt and fluid inclusions were studied. Early papers dealt mostly with easily recognizable inclusions in apatite. Apatite is a common mineral in carbonatites and associated silicate rocks, and commonly contains numerous liquid-vapor and liquid-solid-vapor inclusions ranging from low-density fluids to very concentrated salt melts and brines (Rankin 1975, 1977, Romanchev & Sokolov 1980, Le Bas & Aspden 1981, Roedder 1984). Nahcolite (NaHCO3) is the predominant daughter mineral of carbonatitic fluid inclusions (Rankin 1977, 2005), and studies of these inclusions have been important for the recognition of alkali-rich fluids associated with calciocarbonatites. Highlights of early research on African intrusions include the demonstration of carbonate-silicate liquid immiscibility in apatitehosted inclusions from ijolite pegmatites of the Usaki complex of West Kenya (Rankin & Le Bas 1974), and silicate-carbonate inclusions in pyroxene, wollastonite, nepheline, and sanidine from nephelinite and phonolite of Kwaraha and Oldoinyo Lengai (Romanchev & Sokolov 1980). The discovery of liquid immiscibility in natural inclusions agreed well with earlier experimental demonstrations of silicate-carbonate unmixing in peralkaline synthetic systems (Koster van Groos & Wyllie 1968), and provided key evidence in favor of an important role of immiscibility for carbonatite petrogenesis. Le Bas & Aspden (1981) noted the close similarity between the compositions of apatite-hosted, alkali-carbonate inclusions from ijolite and nephelinite (Tables 6-1 and 6-4), and natrocarbonatite lava erupting at the Oldoinyo Lengai volcano, Tanzania.
EXAMPLES OF MELT INCLUSION STUDIES IN CARBONATITES Comprehensive studies of carbonatitic melt and fluid inclusions started soon after, or in some cases before, the recognition of igneous carbonatite. The presence of minute, fluid-filled cavities in carbonatite minerals was noted in many of the early studies of carbonatites (e.g., von Eckermann 1948, Johnson 1961, Yevzikova & Moskalyuk 1964, Melcher 1966, Giraut 1966). Roedder (1973) described CO2-rich aqueous inclusions in apatite from the Amba Dongar carbonatite of India, and summarized the early reports on carbonatitic fluid inclusions in a review volume (Roedder 1984). The geographic distribution of the studies has been uneven. The largest aggregate number of melt and fluid inclusion studies has been carried out on carbonatite occurrences in Africa and the former Soviet Union, but important results were obtained also for intrusions in North America, Greenland and elsewhere. Carbonatitic fluid inclusions were recently reviewed by Rankin (2005). The brief overview presented here is directed primarily towards presumably comagmatic inclusions of carbonated silicate and carbonatitc melts with emphasis on the petrography and bulk composition of trapped liquids. Case studies are sorted by geographic location, and presented more or less in chronological order. Key examples of modal and chemical compositions of the inclusion types are listed in Tables 6-1 to 6-4, and some characteristic optical and back-scattered electron images are presented in Figures 6-2, 6-3 and 6-4.
Magnet Cove, Arkansas, USA The Magnet Cove Complex, central Arkansas, USA, consists of a series of ring dikes intruding Paleozoic sedimentary rocks. The core of the complex contains a coarse-grained calciocarbonatite believed to be the youngest in the sequence of associated igneous rocks, which includes trachyte, phonolite, syenite, ijolite, and jacupirangite (Woolley 1987). The focus of the paper by Nesbitt & Kelly (1977) was the identification, chemical analysis, and interpretation of multi-
Subvolcanic complexes in East Africa Numerous intrusive and extrusive, carbonatite and alkaline silicate rock complexes occurring in the East African rift system (Le Bas 1977, Woolley 2002) have been used as the most important testing ground for the hypotheses of 128
129
0
a
n. d. 30-45 3-5 n. d. n. d. n. d.
0 22-63 0-10 0-19 0 0
Na-Ca-K carbonates
Silicates
Fe-Ti oxides
Phosphates
Alkali sulfates
Fluorite
n. d. Aldous 1980b
24-43 Nesbitt & Kelly 1977
L+V bubble
Reference
Solovova et al. 1998
n. d.
0c
d
~5
0
0
2
12 0
0
4
0
traces
7
79
2
11
0
1
4
57
8
n. d.
ap
d
8
Krestovskiy, Polar Siberia, Russia
Kovdor, Russia
0
<1
2-3 0
0
5-7
45-70
1
4-6
13-25
7-8
0
20-35
9
~1
4
20-30
16
2
> 50
0
3-4
2
4-5
8-13
23-30
0-60
~1
0
~1
0-8
40-90
Panina 2005
not specified, but approx. 10 vol.% on photographs
0
1-9
1-4 0
0
5-9
3-6
~1
2-3
15-24
53-64
0
11-13
1-6
~1
1-8
35-52
10-52
mon, mon, mon, mon, mon, ol, cpx, mel, ap, mel, mel, ol mel mon pvt pvt cpx 850-980 810-970 860-890 850-980 670-840 780-810
n.d. Veksler, unpublished
n.d.
0
0
0 3
0
10
2
2
g
f
20
60
6
n. d.
cpx
pyroxenite
Gardiner, East Greenland
0
0
12
1
19
49
e
16
n. d.
ol
urtite olivine-melilite and olivine-monticellite rocks; pyroxenites phoscorite
n. d.
ap
ijolte
Small intrusions in Kenya
Abbreviations for minerals: ap, apatite; cpx, clinopyroxene; mel, melilite; mon, monticellite; ol, olivine; pvt, perovskite. a. a large proportion of unidentified phases containing P. S. Fe b. compile from Rankin 2005 c. portlandite was reported as solid inclusions in apatite, and as a daughter phase in fluid inclusions d. Ca, Ba, and Sr carbonates e. including eitelite f. bradleyite g. northupite
n. d.
0
n. d.
0 5
25
a
0
15
a
Portlandite
Chlorides
55
40-60
20-57
Ca-Mg-Fe carbonates
850-870
>750
>800
Thom, oC
ap
ol
mon
carbonatite
Host mineral
carbonatite phoscorite
Rock type
Palabora, RSA
Magnet Cove, USA
Pluton
TABLE 6-1. MODAL COMPOSITION OF CRYSTALLIZED CARBONATITIC MELT AND FLUID INCLUSIONS (VOL.%). MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
ILYA V. VEKSLER & DAVID LENTZ
TABLE 6-2. MANTLE-DERIVED CARBONATED SILICATE LIQUIDS REPRESENTED BY MELT INCLUSIONS AND KIMBERLITIC GROUNDMASS (WT.%). Pluton
Kugda, Russia
Krestovskiy, Guli, Russia Russia meliliteolivineRock olivinite melilitolite monticellite melilite rock rocks Host perovskite perovskite melilite perovskite olivine Quench T, °C 1100 1050 1050 1170 1221 35.66 SiO2 34.78 36.64 34.50 35.06 Ti°O2 5.53 3.85 4.99 6.40 5.43 Al2O3 4.07 8.44 8.76 7.70 11.23 FeO 7.9 8.61 9.07 9.40 8.17 MgO 10.2 4.75 5.02 6.10 4.66 CaO 22.46 17.05 18.99 15.10 21.0 Na2O 4.24 4.93 4.87 4.20 8.48 K2O 3.18 2.24 1.56 4.90 3.14 P2O5 0.53 2.96 2.36 2.40 0.78 SrO n.d. 0.65 0.31 0.40 n.d. BaO n.d. 0.36 0.66 0.40 n.d. SO3 n.d. n.d. n.d. 0.60 n.d. Cl n.d. n.d. n.d. 0.20 n.d. CO2 n.d. n.d. n.d. n.d. n.d. Total 92.89 90.50 90.65 92.30 98.2a Ref.
Gardiner, East Greenland
Nielsen et al. 1997
Panina 2005
Rass & Plechov 2000
Zeughaus dike, Germany
Lac de Gras, Canada lherzolitic olivine melilitite xenoliths in kimberlite olivine ap, cpx, ol Cr-diopside 982 855 unheated 34.52 21.57 30.0 4.22 4.36 0.26 13.0 13.27 1.3 10.35 10.83 6.0 4.9 0.8 22.0 17.83 37.68 16.0 5.61 0.16 0.05 3.15 0.04 0.2 2.72 0.07 0.03 n.d. n.d. 0.03 n.d. n.d. 0.04 0.8 1.10 0.9 0.53 0.05 0.01 2.0b 9.8b 13 c c 97.96 90.18 89.82 Seifert & Thomas 1995
van Achterbergh et al. 2004
UdachnayaEast, Russia kimberlite groundmass unheated 26.10 1.30 2.00 7.20 28.90 12.40 4.30 2.10 0.40 0.11 0.13 0.70 2.80 10.3 98.74 Kamenetsky et al. 2004
a – homogenized glassy inclusions contain 1.4 wt.% Cr2O3; b – volatile content estimations based on low totals of electron microprobe analyses; c – excluding volatiles. Abbreviations same as in Table 6-1.
TABLE 6-3. COMPOSITIONS OF CARBONATITIC MELT INCLUSIONS AND CHLORIDE-CARBONATE GLOBULES IN THE KIMBERLITIC GROUNDMASS. ABBREVIATIONS FOR MINERALS ARE THE SAME AS IN TABLE 6-1. Pluton
Magnet Palabora, Cove,USA RSA Rock carbonphos- carbonatite corite atite Host mon ol ap 15.7 11.6-17.4 7.28 SiO2 TiO2 n. d. n. d. 0.07 Al2O3 n. d. n. d. 1.11 FeO 4.38 8.2-13 0.74 MnO n. d. 0.1 0.09 MgO 0.98 13.9-19.8 3.19 CaO 49.67 17.5-23.3 48.77 Na2O n. d. n. d. 0.06 K2O n. d. 2.9-4.3 0.77 P2O5 1.13 n. d. 11.46 SrO n. d. n. d. 0.36 BaO n. d. n. d. 0.1 SO3 n. d. 0.02-0.7 n. d. F n. d. 2.3-3.4 3.41 Cl n. d. n. d. n. d. CO2b 16.71 20.9-27.9 22.59 Total 88.57 100.0 Ref.
Nesbitt & Kelly 1977
Lac de Gras, Kovdor, Russia Gardiner, E. Krestovskiy, Russia Udachnaya, Canada Greenland Russia kimberlite peridotite phoscorite melilitolite melilitolite, pyroxenite kimberlite Cr-diopside 20.4 16.4 0.24 0.22 0.85 0.67 4.44 2.3 0.14 0.14 22.6 11.3 26.4 37.2 0.03 0.12 0.72 0.48 0.1 0.13 0.86 1.18 1.35 1.8 n. d. n. d. 0.02 0.01 n. d. n. d. 21.6 27.9 99.75 99.85
ol 10.65 0.03 0.34 3.50 0.06 6.86 36.25 5.74 1.42 2.0 n. d. n. d. n. d. n. d. n. d. 33.56 100.43
ol 7.87 0 0 3.57 0.51 16.09 18.5 16.2 1.22 4.18 n. d. n. d. n. d. n. d. 0.4 30.38 99.10
mel 0.26 0.11 0.04 0.44 0.01 0.84 44.59 8.73 1.30 2.01 0.34 0.01 n. d. n. d. n. d. 39.83 98.49
Aldous Solovova van Achterbergh Nielsen & Veksler Nielsen et al. 1980 et al. 1998 et al. 2004 2002, and unpublished 1997
mel groundmass pvta mon 6.97 1.41 1.88 1.32 2.09 0.0 0.0 0.05 0.96 0.0 0.13 0.09 1.80 0.56 8.1 0.37 n. d. n. d. n. d. n. d. 2.30 1.68 0.45 2.47 19.01 32.84 35.66 31.9 5.68 13.09 5.42 18.87 10.59 7.8 10.2 2.41 2.98 5.35 2.73 0.23 0.56 3.5 0.9 0.25 0.90 0.35 0 0.05 7.89 5.32 0.63 1.45 n. d. n. d. n. d. n. d. 1.08 0.06 4.59 4.32 n. d. n. d. n. d. 35.27 62.82 71.96 70.69 99.05 Panina 2005
Kamenetsky et al. 2006
a – average composition of immiscible globules in silicate-carbonate inclusions; b – CO2 estimations are based on modal proportions of carbonate daughter minerals. Abbreviations same as in Table 6-1.
130
PARENTAL MAGMAS OF PLUTONIC CARBONATITES
TABLE 6-4. THE EXAMPLES OF ALKALI-RICH CARBONATITIC FLUID INCLUSIONS AND IMMISCIBLE SALT MELTS ASSOCIATED WITH CARBONATITES; COMPOSITIONS IN WT.%. Pluton
Kovdor, Russia
Gardiner, East Greenland
Rock
carbonatite
melilitolite
Kenya ijolite
urtite
Krestovskiy, Russia
Mushugai-Khuduk, Mongolia
melilitolite, pyroxenite
celestite-fluorite rock
Host Thom., C SiO2 TiO2 Al2O3 FeO MnO MgO CaO Na2O K2O P2O5 SrO BaO SO3 F Cl CO2a H2O
apatite melilite apatite apatite perovskite fluorite 620-870 970 n. d. n. d. 1030 670 600 0.0 0.73 2.0 3.7 0.26 0.0 0.0 0.0 0.04 0.0 0.0 0.96 0.0 0.0 n. d. 0.10 1.0 1.8 0 0.0 0.0 n. d. 0.95 0.9 0.0 0.06 0.0 0.0 n. d. 0.03 n. d. n. d. n. d. 0.0 0.0 6 - 7.2 0.68 0.0 0.0 0.08 0.0 0.0 20 - 30 19.19 19.1 25.0 2.95 35.0 26.0 12.5 - 14 28.83 26.3 20.9 10.78 2.7 11.1 2.3 - 5.2 0.48 6.1 5.9 19.99 3.9 3.7 4.5 - 9.5 0.84 n. d. n. d. 0.42 0.0 0.0 n. d. 1.46 0.5 0.0 0.19 9.9 5.7 n. d. 12.05 4.4 3.9 0.11 1.2 1.9 n. d. n. d. 4.4 1.8 20.07 33.1 29.0 n. d. n. d. n. d. n. d. n. d. 5.5 2.1 n. d. n. d. 6.9 2.6 0.16 0.1 10.9 n. d. 39 29.2 34.9 n. d. 2.1 4.5 n. d. n. d. n. d. n. d. n. d. 6.0 4.7 Veksler et al. Nielsen et al. Le Bas & Aspden Ref. Panina 2005 Andreeva et al. 1998 1998a 1997 1981 a , CO2 estimations are based on modal proportions of carbonate daughter minerals.
a
mt
Kalkfeld, Namibia quartzite at contact with carbonatite quartz n. d. n. d. 1-1.4 n. d. 3-4.1 0.6 1 - 1.7 13 - 16 18 - 21 7 - 10.8 n. d. up to 3.2 up to 0.9 n. d. n. d. n. d. ~20 ~20 Bühn & Rankin 1999
b
mel cpx
sht mt
cpx
cpx
50 mm
40 mm ol
c
d
ol pvt pvt
ol
phl ol 50 mm
50 mm
FIG. 6-2. Photomicrographs of crystallized melt inclusions from carbonatites and associated silicate rocks. a. Crystallized melt inclusion in forsteritic olivine, Kovdor peridotite. b. Melilite-hosted inclusion from Kovdor olivine-melilite rock. c. Large silicate-carbonate inclusion in forsteritic olivine from Kodor phoscorite surrounded by swarms of smaller inclusions. d. Small crystallized melt inclusion in perovskite (arrow), phoscorite, Vuoriyarvi carbonatite intrusion, Kola peninsula, Russia. See text for more details. Abbreviations for phases: cpx – clinopyroxene, mel – melilite, mt – magnetite, ol – olivine, pvt – perovskite, sht – shortite.
131
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
solid and fluid inclusions in monticellite, apatite, and calcite with the employment of scanning electron microscopy (SEM) and energy-dispersive X-ray analysis (EDX). Inclusions in apatite, similar to those studied at the volcanic and sub-volcanic localities in East Africa (see above), were extremely abundant, but were difficult to characterize because of the extreme variation in phase proportions (solids, liquid and vapor). The variations were ascribed to necking-down and/or heterogeneous entrapment of immiscible fluids. In contrast, primary and secondary inclusions in early-formed monticellite exhibited a greater uniformity in their modal composition with over 50 vol.% solids (Table 6-1), water and a vapor bubble (presumably CO2-rich). Attempts to homogenize the inclusions were unsuccessful, even after heating them to 800oC for a week. Bulk compositions of the inclusion phase assemblages were calculated on the basis of the EDX analyses of solid phases and statistically representative modal proportions of the daughter minerals (Table 6-3). Calcite was the predominant daughter phase among those identified, and the inclusions were interpreted as trapped droplets of carbonatite magma with an estimated density of 2.2 to 2.3 g.cm–3. The enrichment of silica and Fe oxides in the melt relative to the calciocarbonatite, was attributed to crystal settling and relative enrichment of calcite at shallower levels. The calcite from Magnet Cove was found to contain abundant secondary and pseudo-secondary fluid inclusions, but none appeared to be unequivocally primary. Palabora, Republic of South Africa The 2012–2047 Ma Palabora complex of Limpopo Province comprises a large oval-shaped body of pyroxenite, syenite and carbonatite intruded into Archean gneiss (Eriksson 1989, Woolley 2001). The central carbonatite zone of the complex is famous for extensive economic Cu-sulfide mineralization. Three different carbonatites, emplaced in the following order, have been recognized (Aldous 1980, see also Rankin 2005): 1) phoscorite (olivine-apatite-magnetite rock) 2) banded apatite–magnetite carbonatite 3) cross-cutting, transgressive, apatite-magnetite carbonatite. The late-stage, transgressive carbonatite is the dominant host for the primary Cu-sulfide mineralization, where it occurs in a series of discontinuous veinlets.
FIG. 6-3. Back-scattered electron images of carbonatesilicate crystallized melt inclusions. a, b – olivinehosted inclusions from Kovdor phoscorite. c – inclusion in clinopyroxene from the Gardiner complex, east Greenland. Abbreviations for phases: brad – bradleyite, Ca-Ba – Ca-Ba carbonates, c-hum – clinohumite, dol – dolomite, eit – eitelite, mag – magnesite (?), Na-sul – Na sulfate, she – scheelite; other abbreviations as in Fig. 6-2.
132
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
FIG. 6-4. Back-scattered electron image and X-ray elemental dot maps of a silicate-carbonate inclusion in olivine from peridotite, the Kovdor complex; nye, nyererite, other abbreviations same as in Fig. 6-2.
because of oxidation in air, precluding further observations. On the basis of electron microprobe analyses of the daughter minerals, and an estimate of their relative proportions, Aldous (1980, see also Rankin 2005) determined the bulk composition of melt inclusions (Table 6-3), and suggested that the olivine-hosted inclusions probably represented the earliest-formed carbonatite melt at Palabora. Importantly, melt trapped in the olivine-hosted inclusions with up to 0.3 wt.% Cu was already exceptionally enriched in this metal.
Aldous (1980) studied multi-solid, carbonate melt inclusions in olivine from the early-forming phoscorites. These inclusions typically contained magnesian calcite, tetraferriphlogopite (phlogopite characteristic of evolved carbonatite showing an extensive Fe3+–Al3+ substitution at the tetrahedral sites), low-Ti magnetite, and sulfide, but no obvious vapor bubble or associated aqueous fluid (Tables 6-1 and 6-3). In homogenization runs, some melting of the daughter phases was observed around 750oC, but above that temperature olivine darkened 133
ILYA V. VEKSLER & DAVID LENTZ
Solovova et al. (1998) studied solid and highly concentrated fluid (salt melt) inclusions in apatite from phoscorite, and pyroxene from pyroxenite. Apatite was found to contain numerous solid inclusions up to 20–50 µm in size. Calcite is the predominant mineral of the solid inclusions, although clinopyroxene, amphibole, phlogopite, zeolites and some accessory phases are also abundant. A few thin hexagonal plates of portlandite [Ca(OH)2] were found in apatite. Portlandite was clearly identified by its refractive indexes in immersion oil, and analyzed by electron microprobe. The portlandite inclusions typically associate with a vapor bubble, and the importance of portlandite is discussed below. The multi-solid, salt melt inclusions in pyroxene and apatite are less abundant and usually very small (2–3 µm). Some salt melt inclusions in pyroxene are 20 µm across and large enough for freezing and homogenization experiments. The proportion of solids in the pyroxenehosted inclusions is 80–95 vol.%. The first signs of melting were observed at 700–730oC, but inclusions decrepitated above these temperatures. Modal proportion of daughter phases and bulk-chemical composition of the trapped liquid were calculated for a large (50 µm) inclusion in apatite (Tables 6-1 and 6-3). The composition is dominated by calcite, and phlogopite is the only identified alkali-bearing phase. In common with the melt composition calculated by Aldous (1980), the liquid is Ca-rich and very low in alkalis.
to the nahcolite- and nyerereite-bearing inclusions in apatite from sub-volcanic complexes in east Africa (Rankin 1975, 1977, Roedder 1984, Le Bas & Aspden 1981, see also Table 6-1), alkali sulfates or chlorides, which would suggest a hydrothermal origin, are absent from the Guli inclusions. Thus, the inclusions were interpreted as trapped solids, or melt inclusions. Regardless, the study demonstrated that water-soluble, Na–K–Ca carbonate can be preserved as micro-inclusions in a typical calciocarbonatite. The Guli complex of Triassic age (240 Ma) covers an area of 1500–1600 km2, and it is the largest intrusion within the Maimecha–Kotui province of ultramafic-alkaline-carbonatitic rocks (Kogarko et al. 1991, 1995). The Guli pluton includes a variety of rock types, which were intruded in the following order: dunite, pyroxenite, melilitic rocks, melteigite and jacupirangite, melanephelinite, syenite, and carbonatite. The carbonatite bodies outcrop in the northern (3 km2) and southern (5 km2) parts of the complex. Melilitic rocks were found to form small intrusive bodies in dunite. Rass & Plechov (2000) studied crystallized melt inclusions in early, forsteritic (Fo86-92) olivine from one small (120 m across) body of olivinemelilite plutonic rock (kugdite). The principal daughter phases of the inclusions are monticellite, Ca-rich pyroxene, nepheline, Cr-spinel; some inclusions also contain small amounts of apatite and calcite. The authors were able to homogenize and quench the inclusions to glass, although many inclusions quenched at 1221–1320oC contained quench (or residual?) monticellite. A typical composition of the quenched glass is presented in Table 6-2. The composition is close to the calculated bulk of unheated carbonate-bearing multi-solid inclusions in perovskite from early olivine cumulates of Kugda, another multiphase intrusion of the Maimecha-Kotui province. The inclusions in olivine from the Guli kugdite have been interpreted as samples of high-temperature, mantle-derived liquids, which probably had experienced a significant loss of CO2 by shallowlevel magma degassing, and presumably were parental to the olivine-melilite rock, as well as some other, more evolved rock types of the Guli complex.
Guli, Maimecha-Kotui Province, north-eastern Russia Kogarko et al. (1991) found Na-rich and SrBa-rich carbonate inclusions in accessory perovskite (CaTiO3) and calzirtite (Ca2Zr5Ti2O16) crystals from calciocarbonatite of the Guli complex of ultramafic-alkaline rocks and carbonatites at the northern border of the Siberian platform. The Narich phase was compositionally close to nyerereite Na2Ca(CO3)2, which is a typical phenocryst mineral of the Oldoinyo Lengai natrocarbonatite lava (see above). The inclusions are very abundant, irregular, and small (normally less than 30 µm). The discovery of the Na-rich inclusions has important implications for the composition of carbonatitic liquids parental to calciocarbonatite, because it demonstrates that although the Guli carbonatitic magma had obviously crystallized large masses of calcite, some derivatives evolved to very alkali-rich (and also Sr- and Ba-rich) compositions. In contrast
The Zeughaus dike, south-eastern Saxony, Germany Seifert & Thomas (1995) described crystallized melt inclusions form the Tertiary olivine melilitite (polzenite) Zeughaus dike in 134
PARENTAL MAGMAS OF PLUTONIC CARBONATITES
Saxony. They found inclusions in olivine, pyroxene, apatite, phlogopite, and magnetite. The inclusions in olivine quenched to glass, and two types of average compositions (A and C, according to Seifert & Thomas 1995) are presented in Table 6-2. The more refractory olivine-hosted melt composition A has been interpreted as parental mantle-derived melilititic liquids, the other two compositions were proposed to result from silicate-carbonate liquid immiscibility.
of melilite- and perovskite-bearing melteigite and ijolite, and the younger comprises melilitolite with associated small dikes and veins of calciocarbonatite and peralkaline nephelinite. Detailed petrography of the multi-solid inclusions in Kovdor and Gardiner revealed numerous assemblages of daughter minerals, which include all the principal mineral species observed in the host rocks, but also unstable, water-soluble Na– Ca–K and Na–Mg carbonates, carbonate-chlorides, and carbonate-phosphates (Table 6-1). The widespread occurrence of shortite, Na2Ca2(CO3)3, in the inclusion assemblages was taken as a strong indicator of the primary origin and closed-system behavior of the inclusions, because shortite is a subsolidus phase not stable above 400oC (Cooper et al. 1975) and thus cannot be trapped as a solid in association with refractory silicate daughter phases. One of the puzzling features of the inclusion petrography, which was encountered early in the study, was the observation that, commonly, daughter mineral assemblages were very different in host minerals from the same rock. For example, olivine-hosted inclusions from Kovdor peridotite (olivine–clinopyroxene cumulate rock) are composed of clinopyroxene, phlogopite, magnetite, abundant Na–Ca carbonates, minor calcite, and apatite (Fig. 6-4), whereas the pyroxene-hosted inclusions from the same sample did not contain daughter olivine, and are composed of a richteritic amphibole, phlogopite, monticellite, magnetite, calcite, and only very minor Na–Ca carbonates (Veksler et al. 1998a). The difference in mineral assemblages was attributed to the following reaction:
Gardiner, East Greenland and Kovdor, Kola Peninsula, Russia Nielsen et al. (1997), Veksler et al. (1998a), and Sokolov (2002) have carried out a systematic study of crystallized melt inclusions from all the principal rock types of the Kovdor and Gardiner plutonic complexes. The study built upon the discovery of Na-Ca carbonates in the Guli complex (see above), and reported many new alkali-rich phases in the inclusions from calciocarbonatite and associated silicate rocks. The Kovdor complex is a concentrically zoned subvolcanic pluton emplaced at about 370 Ma into Archean granites and gneisses of the Baltic shield (see Kogarko et al. 1995 and Krasnova et al. 2004 for an overview of Kovdor geology). The earliest rocks, olivinite, occupy the central part of the intrusion, and are surrounded by irregular zones of peridotite, pyroxenite, melilite-bearing rocks, and rocks of the ijolitejacupirangite series. The youngest suite of rocks in Kovdor includes phoscorite and carbonatite, which form a stock in the southwestern part of the complex. Carbonatite evolves from calciocarbonatite to dolomite carbonatite. The carbonatitic-phoscoritic stock has been mined for magnetite, apatite and baddeleyite, and mapped and studied in great detail (Krasnova et al. 2004). The Gardiner complex in east Greenland formed at about 50 Ma during the early Tertiary break-up in the North Atlantic. The complex is about 5 km in diameter and dominated by early ultramafic cumulates (Nielsen 1981). They are interpreted as in situ cumulates formed on the walls of an open, shallow-level (50–100 MPa) subvolcanic magma chamber of a highly alkaline, carbonatite-bearing volcano (Nielsen et al. 1997). The ultramafic cumulates are intruded by a radial and ring dike system of increasingly alkaline composition (Nielsen 1980). Two major ring dikes, more than 100 m in width, are emplaced in the center of the complex. The older dike is composed
4 CaMgSi2O6 + Mg2SiO4 + Na2Ca(CO3)2 + H2O + CO2 = Na2CaMg5Si8O22(OH)2 + CaMgSiO4 + 3CaCO3 (1) With the excess of diopside component in pyroxene-hosted inclusions, nyerereite and olivine may be totally consumed by the reaction, and richterite becomes the main sink for Na. Another host-inclusion reaction has been considered to be responsible for the lack of daughter mineral assemblages with coexisting monticellite and diopsidic pyroxene: CaMgSi2O6 + Mg2SiO4 + 2 CaCO3 = 3 CaMgSiO4 + 2 CO2
(2)
Three types of crystallized melt inclusions were distinguished in Gardiner melilitolite (Nielsen 135
ILYA V. VEKSLER & DAVID LENTZ
1999), and Sr-Ba carbonates and sulfates. Apart from the host olivine, tetraferriphlogopite is the only silicate phase stable in the evolved inclusion assemblages. Notably, the characteristic shortite and northupite were found as major constituents of carbonatitic inclusions and groundmass in kimberlite of the Udachnaya pipe (Kamenetsky et al. 2004, 2006, and Kamenetsky 2006).
et al. 1997). These are similar to those observed in Kovdor melilite-bearing rocks. Large primary silicate-carbonate inclusions (30–100 µm, see Fig. 6-2b) were interpreted as crystallized droplets of carbonated melilititic melt. Commonly, a significant interstitial volume of the inclusions is occupied by gas bubbles. Two additional types of the inclusions are not readily distinguished morphologically, but have different bulk compositions, and are believed to represent coexisting immiscible alkali carbonatite and natrocarbonatite melts (Tables 6-3 and 6-4). The larger melilite-hosted inclusions of the first type from Kovdor olivine melilitite are composed of diopside, phlogopite, apatite and magnetite, contain gas vesicles, but no carbonate daughter phases were found in them (Veksler et al. 1998a). The absence of carbonate daughter minerals was explained by the decarbonation reaction involving daughter pyroxene and the melilite host:
Sulfate-bearing and halogen-rich carbonatites from Mongolia and Transbaikalia, Russia Andreeva et al. (1998) reported unusual salt melt inclusions in fluorite crystals from calcitic carbonatite of the Mushugai-Khuduk alkalinecarbonatitic volcano-plutonic complex in southern Mongolia. The complex is composed of potassic volcanic and subvolcanic rocks (nepheline melaleucitite, melanephelinite, phonolite, trachyte, trachydacite, nepheline syenite, quartz-bearing syenite, and shonkinite). The carbonatite contains variable amounts of fluorite, chalcedony, celestine, barite, bastnaesite, dolomite or ankerite, and show gradual transition to a series of vein rocks, in which the modal proportion of calcite goes down, and the compositions are dominated by fluorite, celestine, magnetite, apatite, phlogopite, barite, and other minerals. The inclusions were studied mostly in the celestine–fluorite vein rock. Andreeva et al. (1998) called the rock carbonatite; however, it may represent a post-magmatic product of hydrothermal fluids, an example of “carbothermal” vein rocks (see Mitchell 2005b for discussion). The inclusions were sub-divided into 5 compositionally different types, and interpreted as portions of sulfate, phosphate and halogen-rich salt melts and/or very concentrated hydrothermal fluids. Some representative compositions of the inclusions are presented in Table 6-4. Doroshkevich et al. (2003) studied primary and secondary salt melt and concentrated fluid inclusions in bastnaesite, pyroxene and titanite from sulfate-rich carbonatite occurrences of western Transbaikalia, Russia. Late Mesozoic carbonatites in that area are related to potassic mafic alkaline igneous rocks, and commonly contain large amounts of sulfates (barite and celestine). Bulk rock SO3 content in some carbonatites exceeds 10 wt.% (e. g., up to 13.6 wt.% at the Khalyuta occurrence). Based on the inclusion evidence, the authors proposed that sulfate-rich inclusions implied immiscible separation of sulfate-carbonate liquid from a silicate melt.
CaMgSi2O6 + CaCO3 = Ca2MgSi2O7 + 2CO2 (3) Perovskite from Gardiner melilitolite contains abundant crystallized primary melt inclusions (Tables 6-2 and 6-3). Nielsen et al. (1997) and Veksler et al. (1998a) studied dozens of the inclusions, but daughter melilite was never found. The daughter minerals include wollastonite, pectolite and hydrogrossular, which are the typical products of low-temperature, fluid-present alteration of melilite. In addition, the inclusions contain clinopyroxene, nepheline, cancrinite and phlogopite. Calcite and Na–Ca carbonate (mostly shortite) are present in minor amounts. The most striking feature of the inclusions in both complexes is the strong variation in the proportion and the compositions of daughter carbonate minerals (Figs. 6-2 to 6-4). Two types of Na–Ca–K carbonate, compositionally close to nyerereite and shortite, are present in olivine-hosted inclusions from early dunite olivinite, and peridotite. The most evolved olivine-bearing rocks in Kovdor are phoscorite and forsterite-calcite carbonatite. Daughter calcite is nearly absent from olivine-hosted inclusions in phoscorite and carbonatite, although host rocks are dominated by calcite. Nyerereite and shortite in these inclusions are joined by dolomite and complex Na–Mg carbonates and mixed salts, such as eitelite Na2Mg(CO3)2, bradleyite Na3Mg(CO3)(PO4), and northupite Na3Mg(CO3)2Cl (Figs. 6-2c, 6-3a and 6-3b). Minor characteristic daughter minerals include K–Fe sulfides, such as djerfisherite K6(Fe,Cu,Ni)25S26Cl (see also Henderson et al. 136
PARENTAL MAGMAS OF PLUTONIC CARBONATITES
sulfate- and chloride-rich inclusions were interpreted as products of carbonate–sulfate, or carbonate–chloride liquid immiscibility, but alternatively may represent “carbothermal” fluids (see previous section).
The Krestovskiy Complex, Polar Siberia, Russia The Krestovskiy complex is a multiphase silicate-carbonatite intrusion of the MaimechaKotui province of ultramafic–alkaline–carbonatitic igneous rocks in the northern part of the Siberian platform, 50 km southwest off the large Guli pluton (see above). The massif is oval in shape, and represents a layered body of ultramafic rocks consisting of thick (tens to hundred metres) layers of olivinite, wehrlite, and pyroxenite. Small (0.12 and 0.54 km2) bodies of melilitolite and monticellite-bearing rocks crop out at the western and eastern margins of the intrusion. Host rocks of the Krestovskiy plutonic complex are basalt and melanephelinite. The intrusion and the hosting effusive series are cut by dikes of dolerite, nepheline and melilite lamprophyres, and alkali picrite. Panina (2005) carried out a detailed study of melt and fluid inclusions in olivine, pyroxene, monticellite, melilite, perovskite, and garnet from all principal rock types of the complex. The most interesting results were obtained for inclusions in different growth zones of large euhedral perovskite crystals from the olivine–monticellite–melilite rocks. Dark-colored perovskite I in the core of the crystals contain rare, oval primary melt inclusions; the inclusions in the light-colored perovskite II at the rim are more abundant, and commonly have elongated or irregular shapes. The size of the inclusions ranges from 20 to 30 x 150 µm. The inclusions are composed of multiphase crystal aggregates comprising pyroxene, kalsilite, phlogopite, combeite, rankinite, apatite, djerfisherite, titanite, pectolite, and other accessory minerals. On heating, the inclusions exhibit liquid immiscibility between silicate and carbonate–sulfate melt in the temperature interval from 1130 to 1250oC. Some inclusions homogenize at 1230–1250oC, others do not, because they decrepitate on further heating. The inclusions that remained heterogeneous up to very high temperatures were interpreted as products of heterogeneous trapping of two immiscible liquids. The compositions of the immiscible silicate glass and sulfate–carbonate globules from the quenched perovskite-hosted inclusions are listed in Tables 6-2 and 6-3. Melilite, monticellite, pyroxene, and other silicate minerals host a broad variety of other melt and fluid inclusion types (Tables 6-3 and 6-4), and some compositions are much enriched in sulfates, phosphates, or chlorides. Six inclusion types have been recognized by Panina (2005) in total. The
Carbonatitic liquids in kimberlites: Lac de Gras, Canada, and the Udachnaya East pipe, Russia Van Achterbergh et al. (2002) reported quenched carbonate–silicate inclusions in clinopyroxene megacrysts (crystal size 1–20 cm) presumably excavated from 200 km depth beneath the Slave craton in northern Canada by kimberlite eruptions at Lac de Gras. Globular carbonate– silicate inclusions occur along distinct planes cutting across the clinopyroxene crystals, and show a bimodal size distribution (0.8–1.3 mm and 3.2–5.8 mm in diameter). Orthopyroxene exsolution lamellae in the host clinopyroxene extend to the edges of the inclusions, indicating that the entrapment of the globules preceded exsolution and therefore probably occurred at a temperature above 1240oC. Numerous planes of small fluid inclusions (1–20 µm), presumably filled with CO2, radiate from the carbonate inclusions into the clinopyroxene host. The inclusions consist of Tirich phlogopite, elongated forsteritic olivine (Fo89), Cr-spinel, and minor perovskite and sulfides, set in a matrix of calcite. Porous poikilitic clinopyroxenes line the walls of some inclusions, and enclose small grains of calcite and phlogopite. The inclusions were sub-divided into high-silica and low-silica groups, and the average compositions of the groups listed in Table 6-3 were calculated from the compositions of individual minerals, weighted by modal abundance. The carbonate-rich globular inclusions were interpreted as primary mantle carbonatite melts trapped and quenched at shallow depths during or after the kimberlite eruption. The authors proposed a temporal link between the entrapment of carbonatite and the Pliocene eruption of the kimberlite. Low bulk concentrations of incompatible Na and REE in the carbonatitic melt were thought to be inconsistent with an origin by low-degree partial melting of carbonated peridotite with low carbonate content (van Achterbergh et al. 2004). The low alkali and REE abundances were explained by melting of a source with a large modal proportion of Ca-Mg carbonate minerals, and the involvement of recycled crustal material has been inferred based on the oxygen, carbon, and strontium isotope data (van Achterbergh et al. 2002). 137
ILYA V. VEKSLER & DAVID LENTZ
The composition in Table 6-2 represents another inclusion type found in Cr-diopside from lherzolite xenoliths brought to the surface by the Lac de Gras kimberlite bodies. They are concentric ultramafic silicate–carbonate globules, varying from 0.1 to 2 mm in diameter (van Achterbergh et al. 2004). The carbonate component, which crystallized as calcite, separates an outer layer of silicate glass (silicate-a) from a silicate-glass core (silicate-b). Small grains of pyrite and pyrrhotite occur in the silicate glass of some inclusions. The calcite is low in Sr and Ba (less than 1 wt.%), and enriched in rare earth elements. There is almost complete separation of chemical elements between the concentric layers, with very little CaO in the silicate glass (0.3-0.4 wt.%), and very little MgO and SiO2 in the carbonate. The composition in Table 6-2 is the calculated bulk average composition of the concentric silicate–carbonate globules, which are believed to represent trapped droplets of silicate-carbonate melt generated by partial melting of carbonated peridotite at high pressure. A very different composition has been found in olivine-hosted inclusions and the groundmass of group I kimberlite from the Udachnaya East diamondiferous pipe in Siberia, Russia (Kamenetsky et al. 2004, 2006, see also Kamenetsky 2006). In contrast to the melts at Lac de Gras, the inclusions and groundmass of the Udachnaya pipe (Tables 6-2 and 6-3) are extremely enriched in water-soluble alkali chlorides, alkali carbonates, and sulfates, and show high concentrations of REE and other incompatible trace elements. The composition in Table 6-2 is the bulk of the kimberlitic groundmass (Kamenetsky et al. 2004), while the composition in Table 6-3 represents a chloride–carbonate globule of presumably immiscible origin (Kamenetsky et al. 2006). Although it was argued that alkali carbonates, sulfates and chlorides may result from contamination by crustal evaporites (Golovin et al. 2003), the detailed petrography, and oxygen, carbon, and strontium isotopes (Kamenetsky et al. 2004, 2006) strongly support the mantle origin of water-soluble salt components. The mantle origin of the exotic melt composition at Udachnaya is further corroborated by findings of Na–Ca carbonates and K-rich brines as abundant and characteristic inclusions in diamonds (Izraeli et al. 2001, 2004).
IMPLICATIONS OF MELT INCLUSION STUDIES FOR CARBONATITE PETROGENESIS Compositions of primary mantle-derived liquids Melt inclusions in early liquidus minerals from ultramafic–alkaline–carbonatitic plutonic complexes worldwide document silica-deficient (34–37 wt.% SiO2), carbonated silicate melts with high TiO2 contents (~4–6 wt.%) and CaO/Al2O3 weight ratios up to about 5 (Table 6-2). The melts are also characterized by rather high alkali contents, and they are commonly peralkaline (molar (Na2O+K2O)/Al2O3 > 1). Broad variations in MgO/(FeO+MgO) of the compositions listed in Table 6-2 probably result from fractionation of early Mg–Fe silicate minerals, such as olivine and diopsidic clinopyroxene. More evolved inclusion compositions (mostly from melilite-bearing rocks) are close to silicate–carbonate rocks such as bergalite or okaite (Barker 2001). The highest MgO contents and Mg# are documented in the inclusions from kimberlite. Notably, the kimberlitic compositions having about the same Mg# show very different alkali contents, and the alkali-rich composition at the Udachnaya pipe (Kamenetsky et al. 2004) is also several times to orders of magnitude higher in Ti, P, Ba, Sr, and Cl than the composition at Lac de Gras (van Achterbergh et al. 2004). The compositions in Table 6-2 are in general agreement with the diversity of liquids produced in partial melting experiments on a carbonated mantle peridotite at pressures above 2 GPa (Wyllie & Huang 1976, Eggler 1978, Wallace & Green 1988, Thibault et al. 1992, Dalton & Wood 1993, Dalton & Presnall 1998, Wyllie & Lee 1998). However, compositions with Cl contents at a weight percent level have never been considered in experimental work to date. Recent discoveries of KCl- and NaCl-rich brines in diamond- and olivine-hosted melt and fluid inclusions from different kimberlite localities worldwide (Izraeli et al. 2001, 2004, Kamenetsky et al. 2004, 2006) imply that chloriderich liquids may be important metasomatic agents in the mantle, and deserve more attention in the future. The composition of perovskite-hosted melt inclusions from early olivinite cumulate of the Kugda intrusion, the Maimecha-Kotui province, Russia (Nielsen et al. 1997), appears to be the leastevolved of the silicate–carbonate liquid compositions from ultramafic–alkaline–carbonatitic complexes listed in Table 6-2. It also seems to retain volatile components, and has not degassed as much as similar inclusions in olivine from the Guli 138
PARENTAL MAGMAS OF PLUTONIC CARBONATITES
complex (Rass & Plechov 2000). It appears to be a good candidate for a “primitive” (not much affected by shallow-level crystallization and magma degassing) mantle-derived parental liquid of Kugda and the whole clan of similar ultramafic–alkaline– carbonatitic complexes.
(Fig. 6-5). The join was later revised and amended by Schairer et al. (1962), but the earlier, incomplete version in Fig. 6-5 better demonstrates the essence of phase equilibria expressed by equation 4. It should be noted that in the synthetic system, and in the majority of natural ijolitic and melilitolitic rocks, the albite component on the right side of the equation does not form a separate feldspar phase, but resides in melt and/or nepheline solid solution. Natural rocks and crystallized melt inclusions imply that carbonates and CO2 are important constituents of ijolitic and melilititic magmas. Thus, the reaction between the diopside + nepheline association and melilite transforms into a series of decarbonation reactions. The first reaction, which involves diopside and akermanite components of clinopyroxene and melilite solid solutions, is presented above by equation 3. Another decarbonation reaction involves nepheline and Namelilite components:
Mineral reactions and relationships between ijolite, melilitite and carbonate At any scale, from relative abundances of principal volcanic and plutonic rock types (Woolley 2003, and Fig. 6-1 of this chapter) to variations in modal proportions of daughter minerals in microscopic crystallized melt inclusions (e.g., Veksler et al. 1998a, and the equations 1 to 3 above), mineral reactions vividly manifest themselves in ultramafic–alkaline–carbonatitic complexes. Bowen (1928), who introduced the fundamental reaction principle, was also probably the first to point out that ijolite (nepheline + clinopyroxene) and melilitolite rock associations are related to each other by the following reaction: 3CaMgSi2O6 + 2NaAlSiO4 = diopside nepheline Ca2MgSi2O7 + NaCaAlSi2O7 + Mg2SiO4 + melilite solid solution olivine NaAlSi3O8 albite
NaAlSiO4 + CaCO3 + SiO2 = NaCaAlSi2O7 + CO2
In effect, reactions 3, 4 and 5 show that the same Ca-rich and silica-deficient melt can crystallize either as a calcite-bearing ijolite or as a melilitolite. The decarbonation reactions, which produce CO2, obviously strongly depend on pressure, and consequently the formation of melilite-bearing rocks appears to be favored by magma degassing. This may be one of the reasons why melilitic rocks,
(4)
Bowen (1922) encountered the reaction in natural rocks and on the synthetic diopside–nepheline join o
T, C 1500 Diopside+Liquid
1400 1300
Olivine+Diopside +Liquid 1200 1100
Carnegieite +Liquid
Olivine+Liquid
Nepheline+Liquid Melilite+ Olivine +Liquid
Melilite+Diopside+Liquid
CaMgSi2O6 Diopside
20
(5)
40
Nepheline+Melilite +Olivine+Liquid
60
Weight %
80
NaAlSiO4 Nepheline
FIG. 6-5. The diopside – nepheline phase diagram, after Bowen (1922).
139
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
as pointed out by Woolley (2003), are more common in volcanic rock associations than in silicate–carbonate plutonic complexes (Fig. 6-1). Some reactions (like reaction 5) also involve silica, and Barker (2001) recently used such reactions for calculations of silica activities in natural carbonatitic liquids. Veksler et al. (1998a) described numerous examples of decarbonation and other mineral reactions in crystallized melt inclusions from Kovdor and Gardiner (see above). They proposed two scenarios of shallow level magma evolution depending on the stability of liquidus melilite. Melts crystallizing early liquidus melilite, but retaining a significant amount of CO2, were expected to evolve towards the silicate–carbonate miscibility gap, and produce immiscible silicate and carbonate melts. Three groups of melt inclusions in melilite and perovskite from Gardiner melilitolite (Nielsen et al. 1997, Veksler et al. 1998a) were considered as examples of such types of melt evolution. It was proposed that when melilite is not stable, the crystallization paths may not intersect the twoliquid field, but reach the calcite saturation boundary via fractionation of the nepheline plus clinopyroxene (± phlogopite) liquidus assemblage. Further evolution along the calcite-saturated cotectic is expected to produce a series of silicate– carbonate liquids with decreasing SiO2 and increasing alkali contents. A sequence of inclusion compositions from Kovdor (Table 6-3) may represent such a liquid evolution trend along the silicate–carbonate multiply-saturated cotectic toward late-stage alkali-rich compositions sampled by melt inclusions in the Kovdor phoscorite.
halogen-rich, or sulfate-rich compositions reported for some evolved crustal carbonatites (e.g., Krestovskiy, Mushgai-Khuduk, or above-mentioned occurrences in Transbaikalia) may also form in the deep mantle. Many compositions observed in the inclusions are not properly covered by experiments in synthetic systems. The available melt inclusion record is still too sketchy and incomplete, and thus universal models explaining the apparent diversity of natural carbonatitic liquids would probably be premature. Water-soluble daughter phases, including Na–Ca carbonates are hard to detect, and could have been overlooked in early studies of carbonatitic inclusions. Host–melt reactions (Veksler et al. 1998a, see also equations 1-4 above) may also destabilize Na–Ca carbonates in inclusions. However, the evidence for low-alkali, calciocarbonatitic liquid at Palabora (Tables 6-1 and 6-3) has been recently supported by the occurrences of portlandite in apatite-hosted inclusions (Solovova et al. 1998, see above). The occurrence of portlandite links low-alkali melt inclusions at Palabora with the well-characterized CaO–CO2– H2O and CaO–SiO2–CO2–H2O synthetic systems (Wyllie 1989, Wyllie & Tuttle 1959, 1960), where low-temperature calciocarbonatitic liquids are stabilized by the presence of dissolved water, and portlandite appears as an important component. The occurrence of daughter fluorite in inclusions at Palabora (Solovova et al. 1998) also suggests a potential role of F as a fluxing component (Jago & Gittins 1990). The role of silicate-carbonate liquid immiscibility Direct observations of silicate–carbonate immiscibility in natural melt inclusions (Rankin & Le Bas 1974, Romanchev & Sokolov 1980, Panina 2005) support the views that this process can be responsible for the origin of some natural carbonatites (Kjarsgaard & Hamilton 1989, Bailey 1993, Kjarsgaard 1998). The experimental constraints on silicate–carbonate unmixing at pressures up to 0.5 GPa are especially robust (e. g. Kjarsgaard & Hamilton 1989, Brooker 1998, Kjarsgaard 1998, Veksler et al. 1998b and references therein). For instance, the partitioning of key alkalis and alkaline earth elements between the immiscible silicate and carbonate minerals is well constrained experimentally for a broad range of bulk compositions and P–T conditions. Nernst partition coefficients (D = CLCi/CLSi, where CLCi and CLSi are weight concentrations of element i in
Parental liquids of plutonic carbonatites According to experimental phase equilibria constraints (e.g., Lee & Wyllie 1998), lowtemperature carbonatitic liquids at shallow, crustal depth can theoretically range from water-bearing calciocarbonatite compositions to dry natrocarbonatite fluxed by high contents of alkali carbonate components. Melt and fluid inclusion compositions presented in Tables 6-3 and 6-4 appear to support the notion that chemical types of natural carbonatitic liquids are diverse. The inclusions from Magnet Cove, USA, and Palabora, South Africa, seem to represent the liquids at the low-alkali end of the spectrum. The inclusions at Kovdor, Gardiner, and Krestovskiy are in general more alkali-rich. The example of the Udachnaya pipe (Tables 6-2 and 6-3) implies that exotic 140
PARENTAL MAGMAS OF PLUTONIC CARBONATITES
carbonate liquid LC and silicate liquid LS) are the simplest and most straightforward measure of liquid–liquid element partitioning. When plotted against ionic potentials Z/r (where Z is the nominal charge of the metal cation, and r is the ionic radius), the D values of the alkali and alkaline earth elements form a convex curve with the maximum at Sr (Fig. 6-6a). Two examples presented in Fig. 6-6a and Table 6-5 are a peralkaline nephelinitenatrocarbonatite liquid pair modeling the Oldoinyo Lengai compositions (Veksler et al., unpublished; later, and refined experiments of the same type as described by Veksler et al. 1998b), and a low-alkali liquid pair in equilibrium with melilite, melanite, and nepheline (Kjarsgaard 1998). The examples show that although melt composition, pressure, and temperature do affect element partitioning, the effects are not very great, and amount to a shift of the convex curve to the right or to the left from the maximum (see also Brooker 1998 for the effects of pressure up to 2.5 GPa). In a broader range of experimental conditions, the absolute D values may increase up to 10 for Sr (e.g., Kjarsgaard 1998, sample BK316), but the general convex trend does not change. In fact, D values of the mono- and divalent cations plot along the convex upward trend against Z/r not only for carbonate-silicate, but any other salt–silicate immiscible pair of liquids (Veksler 2004). Earlier inclusion studies (Rankin & Le Bas 1974, Romanchev & Sokolov 1980) did not report the compositions of the conjugate liquids. With a few exceptions (Le Bas & Aspden 1981, Nielsen et al. 1997, Panina 2005), which are discussed below, electron microprobe data (Tables 6-3 and 6-4) reveal relatively Ca-rich immiscible carbonatite compositions, closer to the experimental liquids in equilibrium with a melilitic melt (Kjarsgaard 1998) than peralkaline compositions at Oldoinyo Lengai. Average compositions of two presumably immiscible groups of inclusions were reported by Nielsen et al. (1997), and Panina (2005) analyzed quench silicate glass and carbonate–sulfate globules in individual perovskite-hosted inclusions (Tables 6-2 and 6-3). The calculated D values for these examples are presented in Fig. 6-6b. Sr and Ba were rather low for reliable electron microprobe determination in the inclusions studied by Nielsen et al. (1997), but the remaining D values in the Gardiner samples are in general agreement with the experimental convex curve. Although the inclusions in Krestovskiy appear to represent a straightforward case of liquid immiscibility, the D values calculated
D carbonate/silicate
10
Ba
Sr
a
Ca
Na Mg
K
Mn
1
Rb Cs
OL41
BK254
0.1 0
10
Z/r
20
30
D carbonate/silicate
10
K Na
b
Ba Sr Ca
1
Mg
0.1
Gardiner Krestovskiy
Mn
0.01 0
10
Z/r
20
30
FIG. 6-6. Carbonate-silicate liquid-liquid Nernst partition coefficients in synthetic systems (a) and melt inclusions (b). See Table 6-5 and text for more detail.
from the available electron microprobe data (Panina, 2005, see also Tables 6-2, 6-3 and 6-5) do not fit well with the convex D–Z/r curve. In particular, the D value for K appears to be too high, and DSr is too low. The reasons for the discrepancy are not clear, and surely more microprobe analyses of immiscible inclusions would be very welcome. Immiscible silicate and carbonate liquids are traditionally plotted on the so-called Hamilton projection (e.g., Kjarsgaard & Hamilton 1989 and references therein), and such a plot for the inclusion compositions in Tables 6-2, 6-3 and 6-4 is presented in Figure 6-7. The extent of the experimentally determined miscibility gap in the high-Ca nephelinite–carbonatite system at 0.2 GPa and a few tie-lines (Kjarsgaard 1998) are shown for 141
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
TABLE 6-5. LIQUID-LIQUID CARBONATE/SILICATE NERNST PARTITION COEFFICIENTS D’S (SEE TEXT FOR THE DEFINITION) IN EXPERIMENTS WITH SYNTHETIC MIXTURES, AND NATURAL IMMISCIBLE MELT INCLUSIONS. Ionic properties Sample T, C P, GPa H2Oa, wt.% Ion Cs+ Rb+ K+ Na+ Li+ Ba2+ Sr2+ Ca2+ Mg2+ Mn2+ Fe3+ Al3+ Ti4+ Si4+ Nb5+ P5+ S6+ FClReference
CN 8 6 6 6 5 6 6 6 6 6 6 4 5 4 6 4 4 -
r, pm 188 166 152 116 82 149 132 114 86 81 69 53 66 40 78 17 14 133 181
Z/r x 1000 5.3 6.0 6.6 8.6 12.2 13.4 15.2 17.5 23.3 24.7 43.5 56.6 60.6 100.0 64.1 294.1 428.6 -7.5 -5.5
Shannon 1976
Experiments with synthetic mixtures
Natural inclusions Gardiner, Krestovskiy, OL41 OL2 BK254 BK316 melilite perovskite 850 850 950 900 1050 1170 0.1 0.1 0.2 0.5 n. d. n. d. dry 11 dry dry n. d. n. d. Carbonate/silicate partition coefficients (D) 0.69 0.65 n. d. n. d. n. d. n. d. 0.87 0.64 n. d. n. d. n. d. n. d. 1.29 0.96 0.32 0.28 0.83 2.16 2.02 1.01 0.93 1.25 1.79 1.35 2.51 1.35 n. d. n. d. n. d. n. d. 4.03 2.00 2.47 7.79 0.02 2.25 4.24 1.58 4.10 10.08 1.10 1.40 2.77 2.34 3.72 18.38 2.35 1.26 0.49 0.62 1.17 4.93 0.17 0.38 0.30 0.35 0.67 2.87 0.04 n. d. 0.04 0.34 0.42 0.47 0.05 0.19 0.006 0.19 0.013 0.014 0.005 0.13 0.09 n. d. 0.066 0.221 0.02 0.33 0.03 0.33 0.079 0.012 0.007 0.20 0.21 0.42 n. d. n. d. n. d. n. d. n. d. 6.04 4.93 n. d. 0.85 1.24 n. d. n. d. n. d. n. d. n. d. 13.2 6.11 4.15 3.83 7.93 n. d. n. d. 5.94 2.05 2.24 5.27 n. d. 5.38 Nielsen Veksler et al. Panina 2005 Kjarsgaard 1998 et al. 1997 unpublished
a
– bulk water content in starting mixtures. CN – coordination number to oxygen anions; r – ionic radius; Z – nominal ionic charge; Z/r – ionic potential; n. d. – not determined. Ionic radii are according to Shannon (1976).
comparison. The projection in Figure 6-7 shows that a number of melt inclusion compositions plot inside the experimental miscibility gap. This is probably not surprising, and one should not expect perfect agreement with the experimental data in this case, because the inclusions represent a broad diversity of liquid compositions and P–T conditions. All of those factors are known to affect the extent and exact position of the silicate–carbonatite miscibility gap. The projection in Figure 6-8 attempts to use a few key reference inclusion compositions for tracing the main topological elements of the Hamilton projection, that is, the liquid miscibility gap, and the carbonate saturation curve. The miscibility gap is constrained by the immiscible inclusions from Krestovskiy (Panina 2005), and the silicate–calcite cotectic is traced by the inclusion compositions from Magnet Cove, Palabora, and Kovdor (Nesbitt & Kelly 1977, Aldous 1980, Veksler et al. 1998a, see Tables 6-2, 6-3 and 6-4).
The miscibility gap constrained by the natural inclusion compositions appears to be somewhat narrower than the maximal extent reported in experimental systems (Fig. 6-7), but more analyses of the inclusions are needed for a more accurate comparison. Melt-fluid immiscibility Melt and fluid inclusion studies have provided crucial information about the nature and composition of aqueous fluids exsolved from carbonatites and coexisting silicate rocks. The best fluid data come from inclusions in apatite (Rankin 1975, 1977, 2005, Roedder 1984, and references therein), but also from rare examples of quartzhosted inclusions from fenitization zones around carbonatite bodies (Bühn & Rankin 1999). Fluid compositions are in general very alkali-rich; some inclusions may contain weight percent concentrations of halogens and SO3, up to 3 wt.% 142
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
Na2O + K2O
nephelinitic-melilititic liquids kimberlitic liquids low-alkali carbonatitic melts
Kr
phoscoritic melt high-alkali melts and fluids Kf Ic Gr Uc
900
Kr
950 Gu Gr
Kr
Ko
Gr
960 Ko
Kr
Pa
Ud
Ku
SiO2 + TiO2 + Al2O3
Ko
Pa
LG MC
LG
CaO + MgO + FeO
FIG. 6-7. Compositions of melt and fluid inclusions from Tables 2, 3 and 4 plotted onto the Hamilton projection. The silicatecarbonatite miscibility gap for low-alkali compositions at 0.2 GPa (Kjarsgaard 1998) is shown by the heavy solid curve; experimental tie-lines are shown by dashed lines with numbers indicating the experimental temperatures. Abbreviations for the locations: Gr – Gardiner, Gu – Guli, Ic and Uc – inclusions from ijolite and urtite in Kenya (Le Bas & Aspden 1981), Kr – Krestovskiy, Ko – Kovdor, Kf – Kalkfeld, MC – Magnet Cove, LG – Lac de Gras, Pa – Palabora, Ud – Udachanaya pipe.
Na2O + K2O Kr
Kf Ic
Two liquids
?
Gr cc-nye
Uc Kr
70
Ko
Ca
10
lc
Kr
Pa MC
SiO2 + TiO2 + Al2O3
e
S i l i c a t e s
it
Ko
CaO + MgO + FeO
FIG. 6-8. Main topological elements of the Hamilton projections inferred from melt and fluid inclusions. See text for discussion; abbreviations same as in Fig. 6-7.
143
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
ΣREE (Bühn & Rankin 1999), and up to a few weight percent of Sr and Ba (Tables 6-1 and 6-4). Some compositions show no detectable silica (e.g., apatite-hosted inclusions in Kovdor, Veksler et al. 1998a); others have SiO2 concentrations at a level typical for immiscible carbonatite liquids (Le Bas & Aspden 1981). In the Hamilton projection (Figs. 6-7 and 6-8), the inclusion compositions form a group of points stretching from the calcite–nyerereite cotectic of the Na2CO3–K2CO3–CaCO3 system (Cooper et al. 1975) to the alkaline corner. The topology of the alkali-rich portion of the projection is poorly constrained. It is not easy to distinguish between two-fluid and three-fluid equilibria (Nielsen et al. 1997, Nielsen & Veksler 2002, Veksler 2004). It is not quite clear how fluid compositions cross the calcite-nyerereite cotectic, although some explanations involving the effect of fluorine addition have been offered (Jago & Gittins 1990; see discussions by Nielsen & Veksler 2002). The positions of critical curves in relevant multicomponent silicate–carbonate–H2O systems have not been defined. Some very preliminary experimental constraints on the topology are provided by the studies of the Na2CO3–H2O system (Ravich 1974, Koster van Groos 1990), in which the upper critical end-point has been located at ~480oC and 160 MPa (Ravich 1974). The partitioning of Ca, Mg, and Na between hydrothermal fluids and carbonatitic melts was experimentally studied by Veksler & Keppler (2000), and, as one would expect, the study confirmed a higher solubility of Na in the fluids relative to Ca and Mg. In view of the unclear general topology, the alkali-rich compositions of apatite-hosted inclusions from ijolite and urtite in Kenya (Le Bas & Aspden 1981, see also Table 6-4 and Figs. 6-7 and 6-8) can be interpreted either as immiscible natrocarbonatite melt, or as highly concentrated hydrothermal brines. As discussed by Nielsen & Veksler (2002, and references therein), a similar dilemma stands for the natrocarbonatite lava at Oldoinyo Lengai. Some geochemical features of natrocarbonatite agree with the long-proposed origin by dry silicatecarbonate liquid immiscibility, however, other significant details do not comply with the one-stage exsolution model, and hint towards the involvement of hydrothermal fluids. Clearly, more experimental and melt inclusion studies are needed to clarify the issues.
Other types of liquid unmixing Direct observations on melt and fluid inclusions during heating experiments reveal liquid immiscibility between halogen-rich brines and molten carbonates (Fulignati et al. 2001, Webster & De Vivo 2002, Kamenetsky et al. 2004, 2006, and Kamenetsky 2006). Chloride–carbonate liquid unmixing has not been reproduced experimentally in synthetic systems, and Mitchell (1997) presented evidence for this type of liquid immiscibility in the groundmass of the Oldoinyo Lengai natrocarbonatite. The chloride–carbonate unmixing, and other types of immiscibility in salt melts and brines are discussed in detail by Kamenetsky (2006). This phenomenon requires better experimental characterization in the future, because it may play an important petrogenetic role not only in crustal carbonatites and skarns, but also in kimberlite from the deep mantle. CONCLUSIONS Melt and fluid inclusion studies have been indispensable for the development of petrogenetic models for the origin of carbonatites and the carbonatite–silicate igneous rock association. Melt inclusions in early cumulus silicates constrain the compositions of silica-undersaturated, carbonated melts, which may represent parental magmas of ultramafic–alkaline–carbonatitic plutonic complexes. Silica content at about 34–37 wt.%, high TiO2 concentrations (4–6 wt.%), CaO/Al2O3 up to 5 by weight, and total alkalis at about 7–8 wt.% are the main characteristic features of early liquids. Melt and fluid inclusions serve as direct examples of silicate–carbonate liquid immiscibility, alkalirich fenitizing fluids, and complex mineral reactions controlling the evolution of natural carbonatitic magma. Melt and fluid compositions in the inclusions appear to evolve to alkali-rich carbonatitic liquids with high contents of Cl, F, SO3, Sr, Ba, and REE, approaching the composition of the Oldoinyo Lengai natrocarbonatite lava. ACKNOWLEDGEMENTS We thank Jim Webster for his outstanding organizational and editorial efforts in preparing this short course volume. Reviews by Roger Mitchell and Dima Kamenetsky helped to improve the earlier version of the chapter. I.V.V. acknowledges the support from the research fellowship program at the American Museum of Natural History (New York, N.Y.), the CERCAMS grant (Natural History Museum, London), and the European SYNTHESYS 144
PARENTAL MAGMAS OF PLUTONIC CARBONATITES
COOPER, A.F., GITTINS, J. & TUTTLE, O.F. (1975): The system Na2CO3–K2CO3–CaCO3 at 1 kilobar and its significance in carbonatite petrogenesis. Am. J. Sci. 275, 534–560.
program. The research by DL was supported by a NSERC Discovery Grant. REFERENCES ALDOUS, R. (1980): Ore genesis in copper bearing carbonatites: a chemical, mineralogical and fluid inclusion study. Unpublished Ph.D. thesis, Imperial College, Univ. London.
DALTON, J.D. & PRESNALL, D.C. (1998): The continuum of primary carbonatitic-kimberlitic melt compositions in equilibrium with lherzolite: data from the system CaO–MgO–Al2O3–SiO2– CO2 at 6 GPa. J. Petrol. 39, 1953–1964.
ANDREEVA, I.A., NAUMOV, V.B., KOVALENKO, V.I. & KONONKOVA, N.N. (1998): Fluoride–sulfate and chloride–sulfate salt melts in the carbonatite-bearing complex Mushugai-Khuduk, Southern Mongolia. Petrology 6, 284–292.
DALTON, J.D. & WOOD, B.J. (1993): The composition of primary carbonate melts and their evolution through wall rock reaction in the mantle. Earth Planet. Sci. Lett. 119, 511–525.
BAILEY, D.K. (1990): Mantle carbonatite eruptions: crystal context and implications. Lithos 26, 37–42.
DALY, R. A. (1910): Origin of alkaline rocks. Geol. Soc. Am. Bull. 21, 87–118.
BAILEY, D.K. (1993): Carbonate magmas. J. Geol. Soc. (Lond.) 150, 637–651.
DAWSON, J.B. (1962): Sodium carbonate lavas from Oldoinyo Lengai, Tanganyika. Nature 195, 1075– 1076.
BARKER, D.S. (2001): Calculated silica activities in carbonatite liquid. Contrib. Mineral. Petrol. 141, 704–709.
DEINES, P. (1989): Stable isotope variations in carbonatites. In Carbonatites: Genesis and Evolution. (K. Bell, ed.), London, Unwin Hyman, p. 301–359.
BELL, K. & BLENKINSOP, J. (1989): Neodymium and strontium isotope geochemistry of carbonatites. In Carbonatites: Genesis and Evolution (Bell, K., ed.). London, Unwin Hyman, 278–300.
DOROSHKEVICH, A.G., KOBYLKINA, O.V. & RIPP, G.S. (2003): Role of sulfates in the formation of carbonatites in the Western Transbaikal region. Doklady Earth Sci. 388, 131–134.
BELL, K. & KELLER, J. (1995): Carbonatite Volcanism. Oldoinyo Lengai and Petrogenesis of Natrocarbonatites. Springer, Berlin, Heidelberg, New York.
DREW, L.J., QUINGRUN, M. & WEIJUN, S. (1990): The Bayan Obo iron–rare-earth–niobium deposits, Inner Mongolia, China. Lithos 26, 43–65.
BOWEN, N.L. (1922): Genetic features of alnöitic rocks at Isle Cadieux, Quebec. Am. J. Sci. 203(3), 1–13.
EGGLER, D.H. (1978): The effect of CO2 upon partial melting of peridotite in the system Na2O–CaO– Al2O3–MgO–SiO2–CO2 to 35 kb, with an analysis of melting in a peridotite–H2O–CO2 system. Am. J. Sci. 278, 305–343.
BOWEN, N. L. (1928): The Evolution of the Igneous Rocks. Princeton University Press, Princeton, NJ.
ERIKSSON, S.C. (1989): Palabora: a saga of magmatism, metasomatism and miscibility. In Carbonatites: Genesis and Evolution (K. Bell, ed.). Unwin Hyman, London, 221–254.
BREY, G. & GREEN, D.H. (1977): Systematic study of liquidus phase relations in olivine melilitite– H2O+CO2 at high pressures and petrogenesis of an olivine melilitite magma. Contrib. Mineral. Petrol. 61, 141–162.
FOLEY, S.F., VENTURELLI, D.H., GREEN. D.H. & TOSCANI, L. (1987): The ultrapotassic rocks: characteristics, classification, and constraints for petrogenetic models. Earth Sci. Rev. 24, 81–134.
BROOKER, R.A. (1998): The effect of CO2 saturation on immiscibility between silicate and carbonate liquids: an experimental study. J. Petrol. 39, 1905–1915.
FULIGNATI, P., KAMENETSKY, V.S., MARIANELLI, P., SBRANA, A. & MERNAGH, T.P. (2001): Melt inclusion record of immiscibility between silicate, hydrosaline, and carbonate melts: applications to skarn genesis at Mount Vesuvius. Geology 29, 1043–1046.
BÜHN, B. & RANKIN, A.H. (1999): Composition of natural, volatile-rich Na–Ca–REE–Sr carbonatitic fluids trapped in fluid inclusions. Geochim. Cosmochim. Acta 63, 3781–3797. 145
ILYA V. VEKSLER & DAVID LENTZ
JOHNSON, R.L. (1961): The geology of the Dorowa and Shawa carbonatite complexes, southern Rhodesia. Geol. Soc. S. Africa Trans. 64, 101– 145.
GENGE M.J., JONES A.P & PRICE G.D. (1995): An infrared and Raman study of carbonate glasses: Implications for the structure of carbonatite magmas. Geochim. Cosmochim. Acta 59, 927– 937.
M.B., SOBOLEV, A.V., KAMENETSKY, KAMENETSKY, V.S., MAAS, R., DANYUSHEVSKY, L.V., THOMAS, R., POKHILENKO, N.P., & SOBOLEV, N.V. (2004): Kimberlite melts rich in alkali chlorides and carbonates: a potent metasomatic agent in the mantle. Geology 32, 845–848.
GIRAUT, J. (1966): Genèse et géochemie de l’apatite et de la calcite dans les roches liées ou complex carbonatique et hyperalcalin d’Oka (Canada). Bull. Soc. Franç. Min. Crist. 89, 496– 513. GITTINS, J. (1989): The origin and evolution of carbonatite magmas. In Carbonatites: Genesis and Evolution (K. Bell, ed.). Unwin Hyman, London, 580–600.
KAMENETSKY, V.S. (2006): Melt inclusion record of magmatic immiscibility in crustal and mantle magmas. In Melt Inclusions in Plutonic Rocks (J.D. Webster, ed.) Min. Assoc. Can. Short Course 36, 81–98.
GOLOVIN, A.V. SHARYGIN, V.V. POKHILENKO, L.N., MAL’KOVETS, V.G., KOLESOV, B.A. & SOBOLEV, N.V. (2003): Secondary melt inclusions in olivine from unaltered kimberlites of the Udachanya East pipe, Yakytia. Doklady Earth Sci. 388, 93–96.
KAMENETSKY, V.S., KAMENETSKY, M.B., SHARYGIN, V.V., FAURE, K. & GOLOVIN, A.V. (2006): Chloride and carbonate immiscible liquids at the closure of the kimberlite magma evolution (Udachnaya East kimberlite, Siberia). Chem. Geol. (in review)
GUDFINNSSON, G.H. & PRESNALL, D.C. (2005): Continuous gradations among primary carbonatitic, kimberlitic, melilititic, basaltic, picritic and komatiitic melts in equilibrium with garnet lherzolite at 3–8 GPa. J. Petrol. 46, 1645– 1659.
KEPPLER, H. (2003): Water solubility in carbonatite melts. Am. Mineral. 88, 1822–1824. KJARSGAARD, B.A. (1998): Phase relations of a carbonated high-CaO nephelinite at 0.2 and 0.5 GPa. J. Petrol. 39, 2061–2075.
HARMER, R.E. & GITTINS, J. (1998): The case for primary, mantle-derived carbonatite magma. J. Petrology 39, 1895–1903.
KJARSGAARD, B.A. & HAMILTON, B.A. (1989): The genesis of carbonatites by immiscibility. In Carbonatites: Genesis and Evolution (K. Bell, ed.). Unwin Hyman, London, 388–404.
HENDERSON, C.M.B., KOGARKO, L.N. & PLANT, D.A. (1999): Extremely closed system fractionation of volatile-rich, ultrabasic peralkaline melt inclusions and the occurrence of djerfisherite in the Kugda alkaline complex, Siberia. Mineral. Mag. 63, 433–438.
KOGARKO, L.N., KONONOVA, V.A., ORLOVA, M.P. & WOOLLEY, A.R. (1995): Alkaline Rocks and Carbonatites of the World. Part 2: Former USSR. Chapman & Hall, London.
HÖGBOHM, A.G. (1895): Über das Nephelinsyenotgebeit auf der Insel Alnö. Fören. Stockholm Förh. 17, 100–160.
KOGARKO, L.N., PLANT, D.A., HENDERSON, C.M.B. & KJARSGAARD, B.A. (1991): Na-rich carbonate inclusions in perovskite and calzirtite from the Guli intrusive Ca-carbonatite, Polar Siberia. Contrib. Mineral. Petrol. 109, 124–129.
IZRAELI, E.S., HARRIS, J.W. & NAVON, O. (2001): Brine inclusions in diamonds: a new upper mantle fluid. Earth Planet. Sci. Lett. 187, 323–332.
KOSTER VAN GROOS, A.F. (1990): High-pressure DTA study of the upper three-phase region in the system Na2CO3–H2O. Am. Mineral. 75, 667–675.
IZRAELI, E.S., HARRIS, J.W. & NAVON, O. (2004): Fluid and mineral inclusions in cloudy diamonds from Koffiefontein, South Africa. Geochim. Cosmochim. Acta 68, 2561–2575.
KOSTER VAN GROOS, A.F. & WYLLIE, P.J. (1968): Liquid immiscibility in the join NaAlSi3O8– Na2CO3–H2O and its bearing on the genesis of carbonatites. Am. J. Sci. 266, 932–967.
JAGO, B. & GITTINS, J. (1990): The role of fluorine in carbonatite magma evolution. Nature 349, 56–58. 146
PARENTAL MAGMAS OF PLUTONIC CARBONATITES
KRASNOVA, N.I., BALAGANSKAYA, E.G. & GARCIA, D. (2004): Kovdor – classic phoscorites and carbonatites. In Phoscorites and Carbonatites from Mantle to Mine: the Key example of the Kola Alkaline Province (F. Wall & A.N. Zaitsev, eds.), Mineralogical Society Series 10, 99–132.
MITCHELL, R.H. (1997): Carbonate-carbonate immiscibility, neighborite and potassium iron sulphide in Oldoinyo Lengai natrocarbonatite. Mineral. Mag. 61, 779–789. MITCHELL, R.H. (2005a): Mineralogical and experimental constraints of the origins of niobium mineralization in carbonatites. In Rare Element Geochemistry and Mineral Deposits (R.L. Linnen, and I.M. Samson, eds.). Geological Association of Canada Short Course 17, 201–215.
LARSEN L.M. & REX, D.C. (1992): A review of the 2500 Ma span of alkaline-ultramafic, potassic and carbonatitic magmatism in west Greenland. Lithos 28, 367–402. LE BAS, M.J. (1977): Carbonatite–Nepheline Volcanism. Wiley, New York.
MITCHELL, R.H. (2005b): Carbonatites and carbonatites and carbonatites. Can. Mineral. 43, 2049–2068.
LE BAS, M.J. & ASPDEN, J.A. (1981): The comparability of carbonatitic fluid inclusions in ijolites with natrocarbonatite lava. Bull. Volcanol. 44, 429–438.
MITCHELL, R.H., SCOTT SMITH, B.H. & LARSEN, L.M. (1999):Mineralogy of ultramafic dikes from the Sarfartoq, Sisimiut and Maniitsoq areas, West Greenland. In Proc. Seventh Int. Kimberlite Conf. 2. Red Roof Designs, Cape Town, South Africa, 574–583.
LE MAITRE, R.W., compiler (2002): Igneous Rocks: a Classification and Glossary of Terms. Cambridge University Press, Cambridge, UK.
NESBITT, B.E. & KELLY, W.C. (1977): Magmatic and hydrothermal inclusions in carbonatite of the Magnet Cove Complex, Arkansas. Contrib Mineral. Petrol. 63, 271–294.
LEE, W.-J. & WYLLIE, P.J. (1998): Processes of crustal carbonatite formation by liquid immiscibility and differentiation, elucidated by model systems. J. Petrol. 39, 2005–2013.
NIELSEN, T.F.D. (1980): The petrology of melilitolite, melteigite, carbonatite and syenites ring dyke system in the Gardiner complex, East Greenland. Lithos 13, 181–197.
LEE, W.-J., WYLLIE, P.J. & ROSSMAN, G.R. (1994): CO2-rich glass, rounded calcite crystals, and no liquid immiscibility in the system CaO–SiO2– CO2. Am. Mineral. 79, 1135–1144.
NIELSEN, T.F.D. (1981): The ultramafic cumulate series, Gardiner complex, East Greenland. Contrib. Mineral. Petrol. 76, 60–72.
LENTZ, D. R. (1998): Late-tectonic U–Th–Mo–REE skarn and carbonatitic vein-dyke systems in the southwestern Grenville Province: A pegmatiterelated pneumatolytic model linked to marble melting. In Mineralized Intrusion-related Skarn Systems (Lentz, D.R., ed.) Min. Assoc. Can. Short Course 26, 519-657.
NIELSEN, T.F.D., SOLOVOVA, I.P. & VEKSLER, I.V. (1997): Parental melts of melilitolite and origin of alkaline carbonatite: evidence from crystallized melt inclusions, Gardiner complex. Contrib. Mineral Petrol. 126, 331–344.
LENTZ, D.R. (1999): Carbonatite genesis: a reexamination of the role of intrusion-related pneumatolytic skarn processes in limestone melting. Geology 27, 335–338.
NIELSEN, T.F.D. & VEKSLER, I.V. (2002): Is natrocarbonatite a cognate fluid condensate? Contrib. Mineral. Petrology 142, 425–435.
MARIANO, A.N. (1989): Nature of economic mineralization in carbonatites and related rocks. In Carbonatites: Genesis and Evolution (K. Bell, ed.). Unwin Hyman, London, 149–176.
OTTO, J.W. & WYLLIE, P.J. (1993): Relationships between silicate melts and carbonate-precipitating melts in CaO–MgO–SiO2 –CO2 –H2O at 2 kbar. Mineral. Petrol. 48, 343–365.
MELCHER, G.C. (1966): The carbonatite of Jacupiranga, São Paulo, Brazil. In Carbonatites. (Tuttle, O.F. & Gittins, J. eds.). Interscience, New York, 169–181.
PANINA, L.I. (2005): Multiphase carbonate–salt immiscibility in carbonate melts: data on melt inclusions from the Krestovskiy massif minerals (Polar Siberia). Contrib. Mineral. Petrology 150, 19–36. 147
ILYA V. VEKSLER & DAVID LENTZ
RANKIN, A.H. (1975): Fluid inclusion studies in apatite from carbonatites of the Wasaki area of western Kenya. Lithos 8, 123–136.
SHAND, S.J. (1922): The alkaline rocks of the Transport Line, Pretoria District. Geol. Soc. S. Africa Trans. 25, 85.
RANKIN, A.H. (1977): Fluid-inclusion evidence for the formation conditions of apatite from the Tororo carbonatite complex of eastern Uganda. Mineral. Mag. 41, 155–164.
SHANNON, R.D. (1976): Revised effective ionic radii and systematic studies of interatomic distances in halides and chalcogenides. Acta Crystallog. A32, 751–767.
RANKIN, A.H. (2005): Carbonatite-associated rare metal deposits: Composition and Evolution of OreForming Fluids – The Fluid Inclusion Evidence. In Rare-Element Geochemistry and Mineral Deposits (R.L. Linnen & I.M. Samson, eds.). Geological Association of Canada Short Course 17, 299–314.
SOKOLOV, S. (2002): Melt inclusions as indicators of the magmatic origin of carbonatite rare metal and rare earth minerals. Chem. Geol. 183, 373– 378. SWEENEY, R.J., FALLOON, T.J. & GREEN, D.H. (1995): Experimental constraints on the possible mantle origin of natrocarbonatite. In Carbonatite Volcanism. Oldoinyo Lengai and Petrogenesis of Natrocarbonatites (K. Bell & J. Keller, eds.) Springer, Berlin, Heidelberg, New York, 191–207.
RANKIN, A.H. & LE BAS, M.J. (1974): Liquid immiscibility between silicate and carbonate melts in naturally occuring ijolite magma. Nature 250, 206–209.
TAPPE, S., JENNER, G.A., FOLEY, S.F., HEAMAN, L., BESSERER, D., KJARSGAARD, B.A. & RYAN, B. (2004): Torngat ultramafic lamprophyres and their relation to North Atlantic alkaline province. Lithos 76, 491–518.
RASS, I.T. & PLECHOV, P.Y. (2000): Melt inclusions in olivines from the olivine-melilitite rock of the Guli Massif, northwestern Siberian platform. Doklady Earth Sci, 275A, 1399–1402. RAVICH, M.I. (1974): Water–Salt Systems at Elevated Temperatures and Pressures. Nauka, Moscow, (in Russian).
THIBAULT, Y., EDGAR, A.D. & LLOYD, F.E. (1992): Experimental investigation of melts from a carbonated phlogopite lherzolite: implications for metasomatism in the continental lithospheric mantle. Am. Mineral. 77, 784–794.
ROEDDER, E. (1973): Fluid inclusions from the fluorite deposits associated with carbonatite at Amba Dongar, India, and Okorusu, South west Africa: metasomatic feldspar rocks (potash fenites) associated with the fluorite deposits and carbonatites of Amba Dongar, Gujarat, India: discussion and contributions. Trans. Inst. Mining & Metallurgy B-82, B35-B39.
VAN ACHTERBERGH, E., GRIFFIN, W.L., RYAN, C.G., O’REILLY, S.Y., PEARSON, N.J., KIVI, K. & DOYLE, B.J. (2002): Subduction signature for quenched carbonatites from the deep lithosphere. Geology 30, 743–746.
ROEDDER, E. (1984): Fluid Inclusions. In (Ribbe, P.H., ed.), Reviews in Mineralogy 12, Washington, D.C., 644 p.
VAN ACHTERBERGH, E., GRIFFIN, W.L., RYAN, C.G., O’REILLY, S.Y., PEARSON, N.J., KIVI, K. & DOYLE, B.J. (2004): Melt inclusions from the deep Slave lithosphere: implications for the origin and evolution of mantle-derived carbonatite and kimberlite. Lithos 76, 461–474.
ROMANCHEV, B.P. & SOKOLOV, S.V. (1980): Liquation in the production and geochemistry of rocks in carbonatite complexes. Geochem. Intern. 16, 125–135.
VEKSLER, I.V. (2004): Liquid immiscibility and its role at the magmatic–hydrothermal transition: a summary of experimental studies. Chem. Geol. 210, 7–31.
SCHAIRER, J.F., YAGI, K. & YODER, H.S., JR. (1962): The system nepheline – diopside. Carnegie Inst. Wash. Yearbook 61, 96–98.
VEKSLER, I.V. & KEPPLER, H. (2000): Partitioning of Mg, Ca, and Na between carbonatite melt and hydrous fluid at 0.1–0.2 GPa. Contrib. Mineral. Petrology 138, 27–34.
SEIFERT, W. & THOMAS, R. (1995): Silicate– carbonate liquid immiscibility: a melt inclusion study of olivine melilitite and wehrlite xenoliths in tephrite from the Elbe Zone, Germany. Chem. Erde 55, 263–279. 148
PARENTAL MAGMAS OF PLUTONIC CARBONATITES
WYLLIE, P.J. (1974): Limestone assimilation. In The Alkaline Rocks (H. Sørensen, ed.). John Wiley & Sons, New York, 459–474.
VEKSLER, I.V., NIELSEN, T.F. & SOKOLOV, S.V. (1998a) Mineralogy of crystallized melt inclusions from Gardiner and Kovdor ultramafic alkaline complexes: implications for carbonatite genesis. J. Petrol. 39, 2015–2031.
WYLLIE, P.J. (1989): Origin of carbonatites: Evidence from phase equilibrium studies. In Carbonatites: Genesis and Evolution (K. Bell, ed.). Unwin Hyman, Boston, 500–545.
VEKSLER, I.V., PETIBON, C., JENNER, G.A., DORFMAN, A.M. & DINGWELL, D.B. (1998b): Trace element partitioning in immiscible silicate– carbonate liquid systems: an initial experimental study using a centrifuge autoclave. J. Petrol. 39, 2095–2104.
WYLLIE, P.J. & HUANG, W.-L. (1976): Carbonation and melting reactions in the system CaO–MgO– SiO2–CO2 at mantle pressures with geophysical and petrological applications. Contrib. Mineral. Petrol. 54, 79–107.
VON ECKERMANN, H. (1948): The genesis of the Alnö alkaline rocks. Int. Geol. Congr., 18th Session, Great Britain 3, 94–101.
WYLLIE, P.J. & LEE, W-J. (1998): Model system controls on conditions for formation of magnesiocarbonatite and calciocarbonatite magmas from the mantle. J. Petrol. 39, 1885– 1993.
WALLACE, M.E. & GREEN, D.H. (1988): Experimental determination of primary carbonatite magma composition. Nature 335, 343–346.
WYLLIE, P.J. & TUTTLE, O.F. (1959): Melting calcite in the presence of water. Am. Mineral. 44, 453–459.
WATKINSON, D.H. & WYLLIE, P.J. (1969): Phase equilibrium studies bearing on the limestone assimilation hypothesis. Geol. Soc. Am. Bull. 80, 1565–1576.
WYLLIE, P.J., TUTTLE, O.F. (1960): The system CaO–CO2–H O and the origin of carbonatites. J. Petrol. 1, 1–46.
WATKINSON, D.H. & WYLLIE, P.J. (1971): Experimental study of the join NaAlSO4–CaCO3– H2O and the genesis of alkalic rock–carbonatite complexes. J. Petrol. 12, 357–378.
YEVZIKOVA, N.Z. & MOSKALYUK, A.A. (1964): Gas–liquid inclusions in carbonatite carbonates. Acad. Sci. U.S.S.R. Doklady Earth Sci. 159, 108– 111.
WEBSTER, J.D. & DE VIVO, B. (2002): Experimental and modeled solubilities of chlorine in aluminosilicate melts, consequences for magma evolution, and implications for exsolution of hydrous chlorine melt at Mt. Somma–Vesuvius. Am. Mineral. 87, 1046–1061. WOOLLEY, A.R. (1987): Alkaline Rocks and Carbonatites of the World. Part 1: North and South America. British Museum Natural History, London. WOOLLEY, A.R. (2001): Alkaline Rocks and Carbonatites of the World. Part 3: Africa. The Geological Society, London. WOOLLEY, A.R. (2003): Igneous silicate rocks associated with carbonatites: their diversity, relative abundances, and implications for carbonatite genesis. Per. Mineral. 72, 9–17. WOOLLEY, A.R. & KEMPE, P.R.C. (1989): Carbonatites: nomenclature, average chemical composition and element distribution. In Carbonatites: Genesis and Evolution (Bell, K., ed.), London, Unwin Hyman, p. 1–14. 149
ILYA V. VEKSLER & DAVID LENTZ
150
CHAPTER 7: MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRYTYPE ENVIRONMENTS: A LASER ABLATION–ICP–MS STUDY OF SILICATE AND SULFIDE MELT INCLUSIONS
Werner E. Halter and Christoph A. Heinrich Isotope Geochemistry and Mineral Resources Department of Earth Sciences, ETH Zürich 8092 Zürich, Switzerland E-mail:
[email protected] of the magma chamber, and to find out which processes lead to the formation of silica-rich magmas and volatile saturation. Our results are published in a series of six papers that report the discovery of sulfide melt inclusions (Halter et al. 2002a) and successively established the micro-analytical technique (Halter et al. 2004c, Halter et al. 2002b), the evolution of the magmatic complex based on field mapping, wholerock chemical and isotopic analysis and 40Ar/39Ar geochronology by stepwise laser heating (Halter et al. 2004a), the origin of andesitic magmas through mixing of basalt and rhyodacite (Halter et al. 2004b), and the origin of the ore fluid (Halter et al. 2005). The combination of geological constraints on the magmatic evolution of the complex with the mass balance and fluid–chemical requirements imposed by the formation of the 600 Mt ore deposit permits a detailed reconstruction of the timing of volatile saturation, the consequences of sulfide melt saturation, the source for the hydrothermal fluid and the chronology of the ore-forming events. Here, we use this geological and geochemical framework to determine the source magmas for fluids, ore metals, and sulfur, and to trace the behavior of Cu and Au during magma evolution. Our results contrast with current views of the processes responsible for orefluid formation, which generally assume that porphyry-mineralizing magmas have to be oxidized (see Rowins 2000) to prevent sulfide saturation and permit metal enrichment. The resulting model has the potential of practical application to regional mineral exploration.
INTRODUCTION Large porphyry-type hydrothermal systems are generally associated with silica-rich subvolcanic intrusions that form in several cycles of intrusion and alteration or mineralization (Gustafson & Hunt 1975, Proffett 2003). Due to this close temporal association, the link between the magmatic and the hydrothermal event is commonly accepted and intrusions hosting the mineralization have been investigated intensively. In particular, several studies have focused on reconstructing the evolution of the magmatic system that generated the porphyritic intrusions (e.g., Dilles & Proffett 1995, Halter et al. 2005, Hattori & Keith 2001, Keith et al. 1997, Kesler 1997, Landtwing et al. 2002, Sillitoe 1973). A powerful approach to do this involves the study of melt inclusions, trapped at depth during crystal growth. However, only few such studies have been conducted, mostly because these melts are rich in volatiles and, thus, difficult to homogenize (Student & Bodnar 2004). Homogenization under pressure is possible in quartz-hosted inclusions (Schmitt et al. 2002), but in minerals depicting a strong cleavage, successful homogenization can be quite difficult. The only analytical technique that does not require previous homogenization of the inclusions is laser ablation–ICP–MS. Analyses are done by ablation of entire inclusions below the sample surface. As part of the host mineral is ablated with the inclusion, quantification of the melt composition requires that the appropriate amount of material from the host mineral is subtracted from the analysis. This is done using the composition of the host phase, obtained within the same analysis, and an internal standard. We used this approach on samples from a large andesitic complex that hosts a world-class porphyry Cu–Au deposit to reconstruct the history of the complex, to identify the evolution
Melt inclusions in porphyry systems Studies that investigated melt inclusions in porphyry systems mainly focused on inclusions in quartz as they are the best candidates for a successful homogenization during microthermo-
Mineralogical Association of Canada Short Course 36, Montreal, Quebec, p. 151-164.
151
WERNER E. HALTER & CHRISTOPH A. HEINRICH
metric experiments. Due to their high water contents, inclusions in other phases generally decrepitate before the formation temperature is reached (Student & Bodnar 2004). Audétat & Pettke (2003) used quartz-hosted melt inclusions to quantify the element enrichment during crystallization of granitic systems, and their results indicate that very high degrees of fractionation and late volatile saturation need to be reached to increase significantly the Cu content in the residual melt. High Cu concentrations in silicate melt inclusions have been measured in extremely fractionated melts (Harris et al. 2003), but there is no evidence that sufficiently large quantities of such melts are associated with porphyry-type mineralization to explain the formation of large deposits. Recently, Student & Bodnar (2004) measured Cu concentrations of a few hundreds of ppm in melt inclusions of intermediate silica content in quartz phenocrysts from the Tyrone and Red Mountain porphyries. Entrapped melts with higher degrees of fractionation have lower Cu concentrations, and the authors suggested that this is consistent with the exsolution of a Cu-rich fluid from the melt. Thus, the highest Cu contents in the melt are not due to a progressive increase during fractional crystallization. Quartz is formed only in silica-rich magmas, and melt inclusions therein might not provide a record that spans the entire evolution of the system and a potential mechanism that enriches Cu in the melt prior to volatile saturation. Therefore, we extended the investigations to melt inclusions trapped in other phases formed earlier in the magma chamber. Below we present the results from the ore-forming Farallón Negro andesitic complex, Argentina, in which silicate and sulfide melt inclusions were analyzed in amphibole, pyroxene, plagioclase and quartz using laser ablation–ICP–MS.
vertical topography of the edifice (Ulrich 1999) and exposed the deep, internal parts of the volcano. The present study focused on the largest and bestexposed part of the complex, which records most of the intrusive rocks and hydrothermal events. Most of the volcanic pile could be sampled in sequence from the interior to the rim of the volcanic complex. 40 Ar/39Ar stepwise laser heating geochronology is consistent with this interpretation and allowed absolute timing of subvolcanic intrusions within this sequence (Halter et al. 2004a). Note that hereafter, volcanic rocks are called "extrusive" and subvolcanic stocks are called "intrusive" to distinguish their mode of emplacement. However, as no "intrusion" is really plutonic and the texture of extrusive rocks and subvolcanic stocks are very similar, we used the same terminology (from volcanic rocks) to describe their compositions. Emplacement of intrusive and extrusive rocks Volcanic rocks are mainly high-potassium calcalkaline basalt, basaltic andesite and andesite, containing 45 to 66 wt.% SiO2 and phenocryst assemblages of amphibole (hornblende) + plagioclase + magnetite ± pyroxene. Andesite is by far the most abundant extrusive rock type. Biotite-bearing volcanic rocks appear only after 8.0 Ma. The main volcanic activity ceased at 7.5 Ma with the crystallization of the equigranular Alto de la Blenda stock, interpreted to be the central volcanic conduit (Llambías 1972). The only younger extrusive rocks are quartz-bearing dacitic ignimbrite extruded in a flank eruption at 7.35 Ma. The coeval Agua Tapada stock is the likely intrusive feeder, as indicated by dykes connecting it to the ignimbrite. Subvolcanic stocks were first emplaced at 9.0 Ma and become abundant after 8.5 Ma. Individual stocks consist of up to eight intrusions, emplaced within a restricted time frame, but comprising a wide range in compositions. Stocks are mostly dacitic but range from andesitic to rhyolitic and contain amphibole + plagioclase + magnetite ± pyroxene or plagioclase + quartz + magnetite ± amphibole ± biotite as the main phenocryst phases. Pyroxene is restricted to stocks prior to 7.5 Ma and represents at most 10 vol. % of phenocrysts. Biotite appears as a major phenocryst phase in stocks after 7.5 Ma. Amphibole is ubiquitous in all fresh intrusive stocks, except in the latest and most silica-rich ones dominated by biotite.
Farallón Negro Volcanic complex Geological setting The Farallón Negro Volcanic Complex (FNVC) is located in northwestern Argentina, ~200 km east of the present Andean volcanic arc and 500 km from the trench. Previous mapping and analysis of this system (Llambías 1970, Sasso 1997, Proffett 2003, Halter et al. 2004a) have shown that the complex represents the remnants of a 20 km wide and 4.5 km high stratovolcano (Sillitoe 1973, Halter et al. 2004a) intruded by numerous subvolcanic stocks (Fig. 7-1). Extensive erosion removed over 3.5 km of the
152
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
Tertiary Igneous Rocks
27° 10'
A
Volcanic Rocks Basalt to dacite
Campo de Arenal
Porphyries Rhyolite Rhyodacite and dacite Andesite Basaltic andesite
Cerro Durazno
Tertiary Sediments Continental and marine
27° 20'
Basement Farallón Negro lita
s
Paleozoic Granite Metagreywacke Migmatite and gneiss
pil Ca 10°
27° 30'
Overburden Known fault Inferred fault Reverse fault
Agua Rica
Neogene foreland thrust belt
Subduction angle 5-10°
modern magmatic arc
66° 40'
66° 30'
20
Loma Morada
23
25
Inf ere d
fa
Ca ld eFaurtl
ul
t
Agua Tapada
na Pu
Farallon Negro Volcanic Complex volcanoes
Porphyries Rhyolite Macho Muerto rhyodacite Agua Tapada dacite Alto de la Blenda monzonite Chilca andesite Durazno basaltic andesite Tampa Tampa andesite Basaltic andesite dyke Andesitic dyke
Inf er re d
ca
l de
ra
6 985 000
hile Tre nch
500 km
Volcanic rocks Gravel Quartz dacite Basement Biotite andesite Veins Hornblende andesite Bedding Andesite (undiff.) Pyroxenehornblende andesite 2 km
B
Peru-C
27° 40'66° 50'
5 km
Tucuman Transform Zone
Sierras nas Pampea
0
Subduction angle 5-10°
no
30°
NAZCA PLATE
Subduction angle ~30°
la Altip
20°
Macho Muerto Tampa Tampa
30 21
Alto de la Blenda
Leo
6 980 000
Bajo el Durazno
nes
Las Pampitas
30 fa
ul
t
Chilca
Bajo de la Alumbrera Casitas
27 25
3 440 000
3 435 000
3 430 000
6 975 000
FIG. 7-1. A. Geological setting of the Farallón Negro Volcanic Complex. The volcanic sequence covered some 700 km2 and was variably uplifted through reverse faults during late compression (modified from Martinez et al. 1995 and Sasso 1997). B. Detailed geological map of the northwestern part of the complex. Erosion of some 3.5 km of the complex exposed the various intrusions and volcanic rocks erupted over most of the magmatic activity. Modified from Llambías (1970).
153
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
approximately 15 km3 of mafic magma (Halter et al. 2005).
Some 1.2 m.y. after initiation of extrusive activity, a continuous compositional evolution of subvolcanic intrusions with time is taken as evidence for the gradual build-up and growth of a subvolcanic magma chamber. Heat is provided by continued advection of magma through the center of the volcano (Halter et al. 2004b). Cessation of the main volcanic activity at 7.5 Ma was probably caused by the end of magma supply to the chamber, concurrently ending the heat supply and forcing the system to crystallize progressively. Following this event, only dacitic and more silica-rich intrusive rocks were emplaced. A change in the tectonic regime from locally extensive (Sasso & Clark 1998) to dominantly transpressive probably contributed to the termination of extrusive activity, and constrained the magma chamber to evolve internally towards bulk fluid saturation (Halter et al. 2005). The NE–SW alignment of intrusions emplaced after 7.5 Ma corresponds to a fracture zone that controlled the ascent and spatial emplacement of volatile-saturated magmas.
Analytical approach for melt inclusions Melt inclusions in andesitic systems have been analyzed with various techniques, most of which require preliminary homogenization and exposure to the sample surface (see Pettke 2006). In this contribution, we focus on analyses of heterogeneous inclusions by laser ablation– inductively coupled plasma–mass spectrometry (LA–ICP–MS). Melt compositions and their uncertainties were quantified following the procedure described in Halter et al. (2002b). This procedure is reviewed below. Since multiphase inclusions are not homogenized to a glass, critical data evaluation has to be applied to identify inclusions that are not representative of a melt because of heterogeneous entrapment, because of back reactions between the host mineral and the melt or because they behaved as open systems after entrapment. This is done by analyzing several inclusions in a given assemblage of simultaneously trapped inclusions (e.g., on a same growth zone, Fig. 7-2). Analyses are considered representative of the melt only if they yield consistent results within an inclusion assemblage. Individual inclusions were not analyzed and inclusions which differed significantly from other inclusions of the same assemblage were not retained.
The Alumbrera porphyry The Bajo de la Alumbrera porphyry Cu–Au deposit formed between 7.1 and 6.8 Ma, toward the end of the magmatic activity. Most of the stocks formed prior to the Alumbrera deposit show some hydrothermal alteration and Cu and Au enrichment to various extents, but only Alumbrera shows economic metal concentrations. It formed as a series of eight intrusions (Proffett 2003) showing a general evolution toward more mafic rocks, from early dacite to late andesite (Ulrich & Heinrich 2001, Proffett 2003). This evolution of intrusion composition within the Alumbrera stock is taken as evidence for a chemically structured magma chamber, with initial extraction of silica-rich melts from the top and progressively more mafic melt from deeper parts of the chamber (Halter et al. 2004b). Emplacement of intrusions was probably caused by the buoyancy of silica- and volatile-rich magma in the roof zone of the magma chamber, together with structural focusing. Following a barren intrusion that predated the porphyry stock, Cu–Au mineralization is associated with each magma pulse up to the fourth intrusion, with a general trend toward less intense alteration and mineralization with each pulse. Post-mineralization andesitic dykes that cut the stock are unaltered and barren, but they carry sulfide melt inclusions. The total amount of Cu deposited in the porphyry stock requires a minimum source melt volume of
Analytical procedure Only entire, unexposed melt inclusions 5 to 30 µm beneath the sample surface were considered for analysis. Analysis of only parts of heterogeneous inclusions would yield a nonrepresentative composition of the melt. Inclusions were ablated using an optically homogenized UV beam from a 193 nm ArF Eximer laser (Gunther et al. 1997). Details of run conditions are reported in Halter et al. (2002b). The size of the laser spot can be adjusted between 8 and 80 µm using pinhole apertures. For each inclusion, the analytical pit size was selected to be slightly larger than the inclusion diameter. The ablated material was transported by He carrier gas to an Elan 6100 quadrupole mass spectrometer, which sequentially recorded signals for all the elements of interest. Ablation was monitored through an optical microscope and the recorded signal was displayed in real time on a computer monitor to assess the control of the ablation procedure (Fig. 7-3). Each transient signal is composed of 20 to 30 s of instrument background
154
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
FIG. 7-2. A to C Partially crystallized silicate melt inclusions along growth zones in a plagioclase crystal. Due to partial or complete crystallization of inclusions, heterogeneous entrapment cannot be recognized petrographically if inclusions are not rehomogenized. Thus analyses were considered representative of a melt only if several inclusions in the same assemblage (e.g., along a same growth zone) yielded the same compositions. Sulfide melt inclusion in amphibole in transmitted (D) and combined transmitted and reflected light (E).
measurement, followed by the analytical signal from the ablation of the host mineral and the inclusion. The analytical signal was composed of a steady response from the ablation of the host mineral before and after the inclusion, and an intermediate transient signal from the combined ablation of host and inclusion in varying proportions. Details of the analytical set-up are given in Gunther et al. (1997, 1998) and Heinrich et al. (2003), and further details are presented in Pettke (2006). Sulfide melts were analyzed using the same approach as described for silicate melt inclusions (Halter et al. 2004c). As for silicate inclusions, the signal consists of a background measurement, monitored prior to ablation, a signal from the host before and after the ablation of the inclusion and an evolving signal reflecting the mixture of host plus inclusion.
Quantification of melt inclusion compositions Quantification of the chemical composition of silicate melt inclusions is obtained through a threestep procedure. First, analytical signals of the host and the host plus inclusion mixture are converted into element ratios using element sensitivity factors determined through external standards analyzed before and after the unknown (SRM–610 from NIST, hereafter NIST 610). Second, element concentrations in the host and the mixed signal are calculated by normalizing them to a total concentration of 100 wt.% major elements on an anhydrous basis (or less if a certain water concentration is expected). Finally, the relative contributions from the host and the inclusion to each element concentration in the mixed host + inclusion signal needs to be assessed, to determine element concentrations in the originally trapped melt. This is done for each inclusion using an internal standard (Halter et al. 2002b). An internal
155
Concentration
WERNER E. HALTER & CHRISTOPH A. HEINRICH
C iHOST C iMIX
Extr
apo
latio
n
CiINCL
0
0.2 X 0.4
0.6
0.8
1
Mass inclusion
(mass inclusion + host)
Host 7
Laser on
Host
Host + inclusion
Laser off 27 23
Log (counts/second)
6
Al Na
39
K Si 88 Sr 29
5 Background 57
4 86 232
3
Th
Fe
Rb
2 0 20
40
60 Time (seconds)
80
100
FIG. 7-3. Typical laser ablation–ICP–MS signal for various elements during ablation of inclusions in a plagioclase host and schematic representation of the procedure to quantify melt inclusion compositions. CiHOST, CiMIX and CiINCL are the concentrations of an element, i, in the host, the host + inclusion mixture and the inclusion, respectively. The signal consists of an initial signal from the host mineral, followed by an evolving signal of host + inclusion mixture and then again from the host only, after the inclusion is completely ablated. Quantification is done by using the relative proportion, X, of [inclusion/(inclusion + host)] in the mixed signal determined using an internal standard (e.g., an element for which the concentration CiINCL in the inclusion is know). Element concentrations in the inclusion, CiINCL, are calculated by removing the host contribution from the mixture, i.e., by extrapolating from CiHOST and CiMIX to an inclusion/(inclusion + host) ratio of 1. standard is a known element concentration in the melt, by which the mass ratio between the host and the inclusion can be uniquely determined (Halter et al. 2002b). In this study, we took advantage of the essentially constant concentration of Al2O3 in the bulk rock and used the well-defined correlation between Al2O3 and FeO in bulk rocks to constrain the Al2O3 content of the melt inclusions. Once the inclusion/host ratio, X, is known, the concentration of all the elements in the melt can be quantified through a simple mathematical procedure, graphically presented in figure 7-3. This approach also corrects for any quantity of host mineral crystallized from the melt onto the inclusion wall.
Thus, remelting of the inclusion wall is not necessary with this approach. The composition of the various host phases obviously varies significantly during magma evolution, but since we always use the host analyzed just above or below the inclusion, this has no effect on the quantification of the inclusion composition. There is no evidence for significant compositional changes due to small scale zonation, which could generate differences in the host composition at the scale of a few tens of micrometres (i.e., the vertical distance between the analysis of the inclusion and the host). Important to note is that uncertainties of individual element analysis can be calculated from 156
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
most elements (Fig. 7-4), except for small but diagnostic deviations observed for some elements at particular stages of the evolution. Inclusions depict a dominantly bimodal distribution with either basaltic or rhyodacitic compositions, in strong contrast to the predominantly andesitic bulk rock compositions. Moreover, various phenocrysts contain melt inclusions with contrasting compositions. The mafic component is preserved as melt inclusions with 42 to 45 wt.% SiO2 in amphibole. Plagioclase dominantly crystallized from the felsic magma prior to mixing, as rhyodacitic melt inclusions are present almost exclusively in this phase. This shows that phenocryst minerals observed in a given rock crystallized from different melts, and are in fact xenocrysts in the bulk rock. These observations and linear compositional variations of intrusive and extrusive rocks are best explained by a process of mixing varying proportions of a mafic and a felsic (rhyodacitic) magma that already contained phenocrysts. Both magmas may have been at similar temperatures and viscosities, thanks to the high H2O content of the mafic magma, thus favoring complete hybridization of the melts. Intrusions and extrusions have characteristic differences, as evidenced by a systematic distinction in the compositions of intermediate melt inclusions and in their occurrence in the various phases. In volcanic rocks, intermediate melts generated upon mixing are mainly recorded in plagioclase and pyroxene. Amphibole is partially resorbed, indicating a shift from an amphibole (±plagioclase) stable assemblage to the plagioclase + pyroxene stability field. This is explained by decompression during the ascent of the magma towards the surface. Sulfide melt inclusions are generally absent from any phenocrysts in the extrusive andesite, suggesting that these magmas never exsolved a sulfide melt on a broad scale. Rapid degassing upon magma mixing likely suppressed the formation of a sulfide melt (Keith et al. 1997) and induced eruption selectively at times when volatile exsolution occurred or slightly thereafter. In intrusive rocks, melts of intermediate compositions are trapped preferentially by amphibole. Pyroxene with melt inclusions of intermediate composition is only present in the Alto de la Blenda stock, which is the inferred volcanic conduit. The coexistence of amphibole and plagioclase and the absence of pyroxene indicate high water contents and suggest that the source
the LA–ICP–MS signal. These uncertainties increase with decreasing signal intensity and duration and, more significantly, with increasing concentration of the element in the host phase. Typical values are between a few weight percent for incompatible elements and a few tenths of a weight percent for elements abundant in the host mineral. Details of the uncertainty calculation are given in Halter et al. (2002b), and uncertainties from the data used in this study are in Halter et al. (2004b). Sulfide melt inclusions were quantified using the same approach and the same external standard as for silicate melt inclusions, i.e., a silicate glass NIST 610. This is possible because ablation with the Eximer 193 nm laser and the homogenized beam profile induces no matrixdependent fractionation (Halter et al. 2004c). Sulfur was not quantified and element concentrations were obtained by assuming that the inclusion was stoichiometric (Fe,Cu)S. Copper concentrations were sufficiently small (max. 3 wt.%) that the valence and stoichiometric states of Cu are irrelevant within the analytical uncertainty (5–10 wt.%). The mass ratio between the inclusion and the host was calculated by assuming that the inclusion contained no silica, i.e., element contributions from the host mineral (mostly Fe from amphibole in this case) were subtracted in proportion of the SiO2 content of the host mineral. To increase the number of determinations and the counting time on Au, the number of elements analyzed was reduced to four (Si, Fe, Cu, Au). This significantly decreased the uncertainty and the limit of detection for Au. Moreover, it granted a more representative sampling (Pettke et al. 2000) of the Au signal by increasing the number of measurements over the short time interval over which tiny nuggets are ablated. Silicate and sulfide melt inclusion analyses Results for approximately 200 individual melt inclusions in 19 samples are presented graphically for selected major and trace elements in Figure 7-4 in comparison with bulk-rock compositions. The full data set is provided in Halter et al. (2004b). Melt inclusions are identified with respect to the bulk rock composition and the host mineral phase in which they occur. Silicate melt inclusions have SiO2 contents between 42 and 75wt. %, covering a similar compositional range as that of the bulk rocks. Changes in the composition of melt inclusions with increasing SiO2 content and other inter-element correlation trends follow those of bulk rocks for
157
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
4 3.5
Volcanic rocks
Intrusive rocks
4 3.5
TiO2 (wt.%)
3
3
2.5
2.5
2
2
1.5
1.5
1
1
0.5
0.5
0 10 9 8 7 6 5 4 3 2 1 0 18
0 10 9 8 7 6 5 4 3 2 1 0 18
K2O (wt.%)
16
16
CaO (wt.%)
14
14
12
12
10
10
8
8
6
6
4
4
2
2
0 2000 1800 1600 1400 1200 1000 800 600 400 200 0 10000
0 2000 1800 1600 1400 1200 1000 800 600 400 200 0 10000
Ba (ppm)
Cu (ppm) 1000
1000
100
100
10
10
1 40
45
50
55
60
65
70
75
80
1 40
45
50
Host Mineral amphibole
60
65
70
80
Bulk rock
Host rock plagioclase
75
SiO2 (wt.%)
SiO2 (wt.%) pyroxene
55
quartz
basaltic andesite
andesite
dacite
rhyolite
FIG. 7-4. Variation in major- and trace-element concentrations in melt inclusions and bulk rocks as a function of the SiO2 content. Symbols identify host minerals; grey shades represent the host rock. Linear changes, a bimodal distribution of mostly mafic and silica-rich inclusions as well as contrasting melt inclusions in various phases of the same host rock suggest that most rock types result from mixing between a very mafic and a rhyodacitic magma. Most element variations in melt inclusions follow the trend determined by bulk rocks. Only Cu shows deviation in melt inclusions from bulk rocks with a 2- to 5-fold enrichment in volcanic rocks and a strong depletion in intrusive rocks. Cu-poor inclusions of intermediate compositions in intrusive rocks are associated with sulfide melt inclusions.
158
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
magmas for intrusions were not actively degassed during mixing. This is consistent with the common association of sulfide melt inclusions with silicate melt inclusions of intermediate composition in amphibole phenocrysts of subvolcanic intrusions. Degassing would destabilize sulfides through the loss of sulfur to the volatile phase and suppress sulfide saturation.
could be measured in the largest inclusions, where limits of detection are as low as 0.1 ppm (Fig. 7-5), and the concentrations range from 0.1 to 4.6 ppm. Early high-temperature fluid inclusions, characterized texturally and associated with the oreforming event in the Alumbrera deposit, are hightemperature brines (salinity ~50% equivalent NaCl, entrapped in pre-ore vein quartz at T >700°C, Ulrich et al. 2001) containing Cu and Au in the same concentration ratio as the bulk ore (fig. 5). The same Cu/Au ratio is also recorded by the sulfide melt inclusions in barren porphyry intrusions of the Farallón Negro Volcanic Complex (Halter et al. 2002a) and, in particular, in the late, unaltered andesitic intrusions of the Alumbrera stock. The combined textural and metal ratio data, therefore, indicate that the magmatic sulfides and the ore fluids share a common magmatic source.
Copper and gold contents of silicate and sulfide melt inclusions In volcanic rocks, the abundance of Cu in silicate melt inclusions displays a very large variation, sharply contrasting with the simple mixing trend of all other elements that closely mimic the composition of bulk rocks (Halter et al. 2004b, Fig. 7-4). In further contrast with other elements, Cu is almost systematically more abundant in silicate melt inclusions than in bulk rocks with the same silica content. In particular, Cu contents of the most primitive melts are approximately 100 to 200 ppm, corresponding to 2–5 times the bulk Cu concentration in mafic rocks. In silica-rich melts, Cu concentrations of 20 to 50 ppm in inclusions represent a similar enrichment factor over bulk rocks. In intrusive rocks, melt inclusions of intermediate composition (50 to 65 wt.% SiO2) are mostly trapped in amphibole and have very low Cu contents of a few ppm. Most Cu concentrations are indeed below the limit of detection of 2 to 3 ppm. Such inclusions are always associated with small (<10 µm), irregularly shaped sulfide melt inclusions. Only inclusions of intermediate composition in sulfide-free pyroxene from the Alto de la Blenda stock have Cu contents of several tens of ppm, comparable to those of volcanic rocks and consistent with the interpretation that this stock was the main conduit during volcanic activity. Silicarich melt inclusions in intrusive rocks have Cu concentrations between 20 and 50 ppm, which is similar to inclusions in volcanic rocks. These inclusions are not associated with sulfide melt inclusions. Sulfide melt inclusions are almost exclusively present in amphiboles in unaltered intrusive rocks that also host silicate melt inclusions with andesitic compositions. They are notably absent from minerals other than amphibole and from volcanic rocks. Primary sulfides could never be identified in the matrix of either volcanic or intrusive rocks. Analyses of sulfide melt inclusions, indicate that they are mainly iron sulfides (pyrrhotite) with Cu contents of up to several wt.%. Gold
Discussion: the formation of a coppermineralizing magmatic–hydrothermal fluid Although ore occurs in dacitic rocks at Alumbrera, dacitic magmas were apparently not saturated with sulfides, because silicate melt inclusions of dacitic and more silica-rich compositions are not associated with sulfide melt inclusions. Moreover, the Cu content of silica-rich melt inclusions is on the order of a few tens of ppm, which is much below the concentration measured in the most mafic inclusions. Thus, Cu is not enriched in silica-rich melts, and melt inclusions indicate that mixing is more significant than fractionation in this system. This is consistent with the data presented in Student & Bodnar (2004), and shows that Cuenrichment during fractionation is not a necessary mechanism to generate porphyry-type ore deposits. Melt inclusions in dacitic intrusions, which host the mineralization at Alumbrera, also indicate that the Cu content of these melts is too low to allow generation of the ore fluid from these melts. Even though the actual Cu content of the exsolving fluid is determined by the distribution coefficient, i.e., mostly by the Cl-content of the fluid (Audétat & Pettke 2003, Candela & Holland 1984, Candela & Holland 1986, Candela & Piccoli 1995) it is unlikely that fluids generated by dacitic melts at Alumbrera carry large amounts of Cu. We suggest that the poor grade of Cu and Au mineralization in early intrusions of the Farallón Negro Volcanic Complex results from the fact that volatile exsolution was at that time restricted to small regions within the upper parts of the chamber, where dacitic melts predominate (Fig. 7-6).
159
WERNER E. HALTER & CHRISTOPH A. HEINRICH
10.0
Halter et al. (2002a) suggested that these sulfides are a likely source for the ore-forming components. This implies that the ore fluid that generated mineralization in the dacitic intrusions was indeed released from deeper parts of the magma chamber, where andesitic melts are present (Halter et al. 2005). Thus, it is likely that after the supply of fresh magma had ceased and extrusive activity had been stopped at 7.5 Ma (possibly aided by a change in stress regime surrounding the magma chamber), the bulk of the magma chamber cooled and progressively crystallized as a whole. After ~0.4 Ma of cooling, emplacement of the early intrusions at Alumbrera induced a pressure drop and generated massive exsolution of a relatively dense, chloriderich volatile phase from the bulk of the residual melt. This volatile loss took place in a major part of the chamber that included sulfide-bearing andesitic melts. Hydrofracturing and stockwork veining in the porphyry neck provided a pathway through which the fluid could escape (Ulrich et al. 2001, Fig. 7-6). Subsequent intrusions were each associated with their own alteration and mineralization events, as recorded by a repeatedly opened vein network (Ulrich & Heinrich 2001, Proffett 2003). Each mineralization event thus extracted a new pulse of fluid from the magma chamber. Precipitation of Cu sulfides requires that the fluids cooled to T <420°C (Hezarkhani & Williams-Jones 1998, Ulrich et al. 2001, Landtwing et al. 2005) implying that successive intrusions cooled prior to emplacement of the next intrusion, which is easily possible in the time span of ~0.3 m.y. over which the Alumbrera stock was formed. The only intrusion at Alumbrera carrying amphiboles with sulfide melt inclusions is the unaltered andesite dike. This dike, which was the last magmatic rock to be emplaced, is a likely sample of the source magma for the mineralizing fluid. The absence of magmatic sulfides in the matrix of all rock types is further indication that sulfides were destabilized in the residual melt prior to solidification of the intrusive rocks. Melt inclusions in volcanic rocks indicate that the average andesitic melt contained approximately 100 ppm Cu. This is 4 to 5 times more than the Cu content measured in the bulk rock. This suggests that most of the Cu (~75 ppm) has been lost during magma degassing upon eruption. Based on the estimated volume of extruded magma of ~750 km3 (Halter et al. 2004a) and assuming that the magmas prior to eruption contained about 70% melt, this amounts to some 150 Mt of Cu that
Cu (wt.%)
9.0
Host intrusions of sulfide melt inclusions
8.0 7.0 6.0
Alumbrera
5.0
Chilca
4.0
Durazno
3.0
Fluid inclusions
2.0
Bulk ore
1.0 0.0 5.00
Au (ppm)
4.50 4.00 3.50 3.00 2.50 2.00 1.50 1.00 0.50 0.00 100000
Au/Cu
10000 1000 100 10 1 0.1 0.01 0
20
40
60
80
100
120
Analyses
FIG. 7-5. Copper and gold concentrations and gold/copper ratios in sulfide melt inclusions. Also shown are gold/copper ratios in ore forming fluid (data from fluid inclusions in pre-ore quartz) and bulk ore at Alumbrera. Sulfide melts concentrate Cu and Au, causing the strong depletion observed in intermediate silicate melt inclusions. The similar Cu/Au ratio in sulfide melt inclusions and the ore-forming fluid was used as evidence for a common magmatic source (Halter et al. 2002a).
In the source region of intrusive rocks, the only Cu-rich phase is the sulfide melt. This melt is apparently generated upon magma mixing as it only occurs in the presence of melt inclusions of intermediate composition. Based on similar Cu/Au ratios in these melts and in the mineralizing fluid,
160
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
150 Million tonnes of Cu
Caldera fault
6000
a
7.5
3000
Basalt
Alumbrera ~7.0 - 6.8 Ma
Elevation (m)
Topography
7.3
Ma
M .0 Ma 8 Ma 5 . 8 Ma 9.0
Dacite
8.0 Ma 8.7 Ma 9.0 Ma
7.5 Ma
7.3 Ma
Andesite
Sulfide melt
8.3 Ma
0
Basement
B
A
-3000 Mixing Silica-rich primary melt Mafic melt
Alumbrera Mineralization
Pre-Alumbrera >7.5 Ma
Alumbrera Post-ore intrusion
Fluid path Melt path Amphibole with sulfide melt inclusions
Brittle Ductile
B1
A
B2
FIG. 7-6. Schematic cross section of the Farallón Negro Volcanic Complex and illustration of the formation of the ore fluid. The complex was formed through volcanic activity between 9.7 and 7.5 Ma and subvolcanic intrusions emplaced between 8.5 and 6.3 Ma. After an initial period of formation, the upper part of the magma chamber became more silicic. During active volcanism, the magma chamber was continuously filled (and heated). Intrusions emplaced over the same time period extracted volatiles only from the upper (silica-rich) part of the chamber. After volcanism had ceased and the system cooled, emplacement of the Alumbrera stock induced volatile exsolution in andesitic magmas from deeper parts of the chamber, where sulfide melts were present. These volatiles scavenged the metal content of the sulfide melt and transported them to the depositional site in the earlier formed dacitic intrusion. Andesitic dykes emplaced after ore formation carry relicts of the sulfide melt as inclusions in amphibole that likely represent the source magma for the ore fluid.
were lost to the atmosphere and hydrosphere during the entire lifetime of the volcanic complex, equating to ~50 times the economic Cu reserves of the Alumbrera deposit (Proffett 2003). Thus, Cu dispersion rather than concentration seems to be the prevailing process even in this ore-forming magmatic system. As suggested by Cline & Bodnar (1991), special conditions within the magma chamber and at the depositional site are required to
generate an economic ore deposit and not primary Cu-rich magmas. CONCLUSIONS The investigation of melt inclusions is an essential tool when trying to understand the evolution of magmatic systems in general, and oreforming processes in particular. As homogenization of these melt inclusions is often difficult (e.g.,
161
WERNER E. HALTER & CHRISTOPH A. HEINRICH
water-rich systems) or even impossible (sulfide melt inclusions) the application of laser ablation–ICP– MS to analyze heterogeneous inclusions is ideally suited to address some of these issues. Moreover, the large amount of data that can be generated with this technique (up to 50 inclusion analyses in one day) allows for a much more comprehensive study than would be possible with other analytical techniques. Complementary techniques are necessary for the analysis of some elements such as S, Cl or water, which are not measurable with the current set-up. Thus, homogenization of melt inclusions is still required to generate data on some key elements in the ore-forming process. Laser ablation–ICP–MS allows construction of the appropriate framework in which to conduct such specialized investigations. This study further demonstrates the capability of LA–ICP–MS for geochemical investigations of magmatic–hydrothermal oreforming processes. The suggested process for the formation of the ore fluid has important implications, not only for ore genesis but possibly also for practical exploration. It implies that the dacitic intrusions hosting the mineralization do not represent the source magma from which most of the ore fluid and its metal load were exsolved. The actual fluid source was an andesitic or perhaps a basaltic magma, which was present in the deeper parts of the magma chamber and may not always have been emplaced at the erosion level of the deposit. At Alumbrera, andesitic magmas intruded after mineralization carry sulfide melt inclusions in amphibole. These late intrusions are the best candidates for yielding information on the conditions of ore fluid formation. Cu/Au ratios measured in the sulfide melt inclusions of these rocks provide the best indication of the Cu/Au ratio in the mineralizing fluid and the ore body. Such post-mineralization, sulfide-bearing mafic magmas are not only present at Alumbrera, but were also described in the giant Bingham porphyry deposit, where Keith et al. (1997) interpreted them to represent a potential metal source for the mineralization. At Bingham, the ore is also hosted by a much more felsic quartz monzonite porphyry than the mafic melts, which carry magmatic sulfides (Redmond et al. 2001). Systematic laser ablation– ICP–MS analysis of sulfide melt inclusions in such post-ore intrusions could be applied to predict potentially Au-rich porphyry systems during the early stages of exploration in similar magmatichydrothermal systems elsewhere.
ACKNOWLEDGEMENTS The authors would like to thank Jean Cline, Alan Anderson and Jim Webster for their comments and help to improve the quality of the manuscript. This study has benefited from extensive input by Thomas Pettke. It was partly financed by ETH research funds and by the Swiss National Science Foundation (grant 20–59544–99 to CAH). REFERENCES AUDÉTAT, A. & PETTKE, T. (2003). The magmatichydrothermal evolution of two barren granites: A melt and fluid inclusion study of the Rito del Medio and Canada Pinabete plutons in northern New Mexico (USA). Geochim. Cosmochim. Acta 67(1): 97–121. CANDELA, P.A. & HOLLAND, H.D. (1984): The partitioning of copper and molybdenum between silicate melts and aqueous fluids. Geochim. Cosmochim. Acta 48, 373–380. CANDELA, P.A. & HOLLAND, H.D. (1986): A masstransfer model for copper and molybdenum in magmatic hydrothermal systems – the origin of porphyry-type ore deposits. Econ. Geol. 81, 1– 19. CANDELA, P.A. & PICCOLI, P.M. (1995): Model ore–metal partitioning from melts into vapor and vapor/brine mixtures. In Magmas, fluids and ore deposits (J.F.H. Thompson, ed.). Mineralogical Association of Canada Short Course 23, 101– 127. CLINE, J.S. & BODNAR, R.J. (1991): Can economic porphyry copper mineralization be generated by a typical calc-alkaline melt. J. Geophys. Res., Solid Earth & Planets 96(B5), 8113–8126. DILLES, J.H. & PROFFETT, J.M. (1995): Metallogenesis of the Yerington Batholith, Nevada. In Porphyry copper deposits of the American Cordillera. (F.W. Pierce & J.G. Bolm, eds.). Arizona Geological Society, Tucson, AZ, U.S., 306–315. GUNTHER, D., AUDÉTAT, A., FRISCHKNECHT, R. & HEINRICH, C.A. (1998): Quantitative analysis of major, minor and trace elements in fluid inclusions using laser ablation inductively coupled plasma mass spectrometry. J. Anal. At. Spectrom. 13, 263–270.
162
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
GUNTHER, D., FRISCHKNECHT, R., HEINRICH, C.A. & KAHLERT, H.J. (1997): Capabilities of an argon fluoride 193 nm excimer laser for laser ablation inductively coupled plasma mass spectrometry microanalysis of geological materials. J. Anal. At. Spectrom. 12, 939–944.
HATTORI, K.H. & KEITH, J.D. (2001): Contribution of mafic melt to porphyry copper mineralization: evidence from Mount Pinatubo, Philippines, and Bingham Canyon, Utah, USA. Miner. Depos. 36, 799–806. HEINRICH, C.A., PETTKE, T., HALTER, W.E., AIGNER-TORRES, M., AUDÉTAT, A., GUNTHER, D., HATTENDORF, B., BLEINER, D., GUILLONG, M. & HORN, I. (2003): Quantitative multielement analysis of minerals, fluid and melt inclusions by laser ablation–inductively coupled plasma–mass spectrometry. Geochim. Cosmochim. Acta 67, 3473–3497.
GUSTAFSON, L.B. & HUNT, J.P. (1975): The porphyry copper deposit at El Salvador, Chile. Econ. Geol. 70, 857–912. HALTER, W.E., BAIN, N., BECKER, K., HEINRICH, C.A., LANDTWING, M., VONQUADT, A., CLARK, A.H., SASSO, A.M., BISSIG, T. & TOSDAL, R.M. (2004a): From andesitic volcanism to the formation of a porphyry Cu–Au mineralizing magma chamber: the Farallon Negro Volcanic Complex, northwestern Argentina. J. Volcanol. Geotherm. Res. 136, 1–30.
HEZARKHANI, A. & WILLIAMS-JONES, A.E. (1998): Controls of alteration and mineralization in the Sungun porphyry copper deposit, Iran: Evidence from fluid inclusions and stable isotopes. Econ. Geol. 93, 651–670.
HALTER, W.E., HEINRICH, C.A. & PETTKE, T. (2004b): Laser ablation–ICP–MS analysis of silicate and sulfide melt inclusions in an andesitic complex II: evidence for magma mixing and magma chamber evolution. Contrib. Mineral. Petrol. 147, 397–412.
KEITH, J.D., WHITNEY, J.A., HATTORI, K., BALLANTYNE, G.H., CHRISTIANSEN, E.H., BARR, D.L., CANNAN, T.M. & HOOK, C.J. (1997): The role of magmatic sulfides and mafic alkaline magmas in the Bingham and Tintic mining districts, Utah. J. Petrol. 38, 1679–1690.
HALTER, W.E., HEINRICH, C.A. & PETTKE, T. (2005): Magma evolution and the formation of porphyry Cu–Au ore fluids: evidence from silicate and sulfide melt inclusions. Miner. Depos. 39, 845–863.
KESLER, S.E. (1997): Metallogenic evolution of convergent margins: Selected ore deposit models. Ore Geol. Rev. 12, 153–171. LANDTWING, M.R., DILLENBECK, E.D., LEAKE, M.H. & HEINRICH, C.A. (2002): Evolution of the breccia-hosted porphyry Cu–Mo–Au deposit at Agua Rica, Argentina: Progressive unroofing of a magmatic hydrothermal system. Econ. Geol. 97, 1273–1292.
HALTER, W.E., PETTKE, T. & HEINRICH, C.A. (2002a): The origin of Cu/Au ratios in porphyrytype ore deposits. Science 296, 1844–1846. HALTER, W.E., PETTKE, T. & HEINRICH, C.A. (2004c): Laser ablation–ICP–MS analysis of silicate and sulfide melt inclusions in an andesitic complex I: Analytical approach and data evaluation. Contrib. Mineral. Petrol. 147, 385– 396.
LANDTWING, M.R., PETTKE, T., HALTER, W.E., HEINRICH, C.A., REDMOND, P.B., EINAUDI, M.T. & KUNZE, K. (2005): Copper deposition during quartz dissolution by cooling magmatic– hydrothermal fluids: The Bingham porphyry. Earth Planet. Sci. Lett. 235, 229–243.
HALTER, W.E., PETTKE, T., HEINRICH, C.A. & ROTHEN-RUTISHAUSER, B. (2002b): Major to trace element analysis of melt inclusions by Laser ablation–ICP–MS: methods of quantification. Chem. Geol. 183, 63–86.
LLAMBÍAS, E.J. (1970): Geologia de los Yacimientos Mineros Agua de Dionisio, Prov. de Catamarca, Rep. Argentina. Revista de la Asociacion Argentina de Mineralogia Petrologia y Sedimentologia. 1, 2–32.
HARRIS, A.C., KAMENETSKY, V.S., WHITE, N.C., VAN ACHTERBERGH, E. & RYAN, C.G. (2003): Melt inclusions in veins: Linking magmas and porphyry Cu deposits. Science 302, 2109–2111.
LLAMBÍAS, E.J. (1972): Estructura del grupo volcanico Farallon Negro, Catamarca, Republica Argentina. Revista de la Asociacion Geologica Argentina 27, 161–169.
163
WERNER E. HALTER & CHRISTOPH A. HEINRICH
MARTINEZ, L., MEILÁN, D. & MAZA, A.E. (1995): Mapa geológico de la provincia de Catamarca, República Argentina. Secretaria de Minería, Dirección Nacional del Secvicio Geológico.
SILLITOE, R.H. (1973): The tops and bottoms of porphyry copper deposits. Econ. Geol. 68, 799– 815. STUDENT, J.J. & BODNAR, R.J. (2004): Silicate melt inclusions in porphyry copper deposits: Identification and homogenization behavior. Can. Mineral. 42, 1583–1599.
PETTKE. T. (2006): In situ laser ablation–ICP–MS chemical analysis of melt inclusions and prospects for constraining subduction zone magmatism. In Melt Inclusions in Plutonic Rocks (J.D. Webster ed.) Mineral. Assoc. Canada Short Course 36, p. 51-80.
ULRICH, T. (1999): Genesis of the Bajo de la Alumbrera porphyry Cu–Au deposit, Argentina: Geological, fluid geochemical, and isotopic implications. Doctoral dissertation, Swiss Federal Institute of Technology, Zurich.
PETTKE, T., HEINRICH, C.A., CIOCAN, A.C. & GUNTHER, D. (2000): Quadrupole mass spectrometry and optical emission spectroscopy: detection capabilities and representative sampling of short transient signals from laser ablation. J. Anal. At. Spectrom. 15, 1149–1155.
ULRICH, T., GUNTHER, D. & HEINRICH, C.A. (2001): Evolution of a porphyry Cu–Au deposit, based on LA–ICP–MS analysis of fluid inclusions: Bajo de la Alumbrera, Argentina (vol. 96, pg 1743, 2001, correctly printed in 2002). Econ. Geol. 97, 1888–1920.
PROFFETT, J.M. (2003): Geology of the Bajo de la Alumbrera porphyry copper–gold deposit, Argentina. Econ. Geol. 98, 1535–1574.
ULRICH, T. & HEINRICH, C.A. (2001): Geology and alteration geochemistry of the porphyry Cu–Au deposit at Bajo de la Alumbrera, Argentina (vol 96, pg 1719, 2001, correctly printed in 2002). Econ. Geol. 97, 1863–1888.
REDMOND, P.B., LANDTWING, M.R. & EINAUDI, M.T. (2001): Cycles of porphyry dike emplacement, veining, alteration and mineralisation in the Bingham porphyry Cu–Au–Mo deposit, Utah. In Mineral deposits at the beginning of the 21st century. (A. Piestrzynski, et al., eds.), Krakow, Poland. Society for Geology Applied to Mineral Deposits (SGA), International, 2001, Proceedings of the 6th Biennial SGA Meeting. 6, 473–476. ROWINS, S.M. (2000):. Reduced porphyry copper– gold deposits: A new variation on an old theme. Geology 28, 491–494. SASSO, A.M. (1997): Geological evolution and metallogenetic relationships of the Farallon Negro volcanic complex, NW Argentina. Ph.D., Queen's University, Kingston, Ontario. SASSO, A.M. & CLARK, A.H. (1998): The Farallón Negro Group, northwest Argentina: Magmatic, hydrothermal and tectonic evolution and implications for Cu–Au metallogeny in the Andean back-arc. Soc. Econ. Geol. Newslett. 34, 8–17. SCHMITT, A.K., TRUMBULL, R.B., DULSKI, P. & EMMERMANN, R. (2002): Zr–Nb–REE mineralization in peralkaline granites from the Amis Complex, Brandberg (Namibia): Evidence for magmatic pre-enrichment from melt inclusions. Econ. Geol. 97, 399–413.
164
CHAPTER 8: SILICATE MELT INCLUSIONS IN FELSIC PLUTONS: A SYNTHESIS AND REVIEW James D. Webster Department of Earth and Planetary Sciences, AMNH, Central Park West at 79th Street, New York, NY 10024-5192, USA E-mail:
[email protected] and Rainer Thomas GeoForschungsZentrum Potsdam, Telegrafenberg B120, D-14473 Potsdam, Germany E-mail:
[email protected] (Student 2002). Last but not least, numerous investigators have obtained accurate thermobarometric constraints on granitoid emplacement from H2O contents and homogenization behavior of melt inclusions (e.g., Naumov 1969, Weisbrod 1981, Roedder 1984, Thomas 1988, Förster et al. 1989, Thomas et al. 1991, Frezzotti 1992, Thomas 1994a, b, Koval & Prokofiyev 1999, many others). This chapter reviews investigations of MI from felsic granitoid rocks and focuses on recent insights, techniques, and applications. Meltinclusion data from granitoid rocks are evaluated against MI data from compositionally similar and evolved, high-silica rhyolite and topaz rhyolite bodies to compare and contrast processes of magma evolution, mineralization, and fluid exsolution in these related environments. The latter process (i.e., immiscible fluid exsolution) has particular relevance and significance to a variety of compositionally evolved magmas. Abundant literature exists on MI in evolved epizonal, alkaline intrusive rocks that are found as xenoliths in materials brought to the surface through eruptive processes, and these samples provide key information on the exsolution of hydrosaline liquids and/or CO2-rich vapors in shallowly emplaced felsic magmas (Belkin et al. 1985, Frezzotti et al. 1991, De Vivo et al. 1993). Discussion of MI in these environments is presented by De Vivo et al. (2006), and other examples of magmatic fluid immiscibility are addressed by Kamenetsky (2006). Although this chapter addresses felsic magmas, it does not address MI in felsic aplite-forming or pegmatite-forming magmas, because processes characterizing these environments are discussed in the chapter by Thomas et al. (2006a). Bodnar & Student (2006)
INTRODUCTION Silicate melt inclusions (MI) are microscopic samples of silicate melt enclosed in phenocrysts of intrusive and extrusive igneous rocks. Many MI are chemically representative of the magmatic melt phase, and hence, they provide unique and important information on magmatic and magmatic-hydrothermal processes. Melt inclusions from felsic magmas are of particular interest and significance because these magmas crystallize to form the cores of many of the Earth’s large mountain belts and because some exhibit explosive eruptive behavior and/or are genetically associated with a variety of metallic ores. Melt inclusion-based investigations of felsic plutons offer distinct benefits over petrologic studies based on granitic mineral and whole-rock compositions. First and foremost, the presence of naturally glassy MI, or of crystallized MI that rehomogenize to glasses of granitic composition, argues strongly against the hypothesis of granite generation via metasomatic granitization. In addition, MI from granites can provide important compositional information on pockets of late-stage, volatile-enriched granitic melt, which can be quite difficult to constrain otherwise. For example, Thomas et al. (2000) and Müller et al. (2003) have successfully determined the chemistry of residual melts by analyzing MI in the outer growth zones of quartz. Moreover, the chemistry of many granitic rocks and their minerals is often modified by latestage hydrothermal and/or deuteric alteration; the minerals of some rocks, in fact, show complete metasomatic destruction. Barring complete host phase replacement, however, MI trapped in quartz may be preserved because of the comparative stability of this mineral during metasomatism
Mineralogical Association of Canada Short Course 36, Montreal, Quebec, p. 165-188.
165
JAMES D. WEBSTER & R. THOMAS
noted, however, the evidence for boundary layers in igneous systems has been more theoretical than observational. A number of investigators have concluded that boundary layer-influenced MI in volcanic rocks are not common, given certain constraints on MI sizes (Lu et al. 1995, Thomas et al. 2002). This issue may pose problems in some igneous situations, but it has not been proven to be generally so. Readers are referred to London (2005) for observations on the compositional consequences of boundary layer growth in hydrothermal experiments. It has been generally observed that many MI in volcanic rocks provide poor constraints on maximum volatile abundances in magma. For instance, pre-eruptive volatile saturation has been established definitively for many intermediate to felsic, calc-alkaline volcanoes including Augustine, Alaska (Johnston 1978), Mt. St. Helens, Washington (Gerlach & McGee 1994), and Mt. Pinatubo, Philippines (Gerlach et al. 1996); for high-silica rhyolitic magmas such as the Bishop Tuff, California (Anderson et al. 1989), and for intermediate to felsic alkaline volcanoes such as Mt. Somma-Vesuvius, Italy (Raia et al. 2000, Marianelli et al. 1995, Signorelli & Capaccioni 1999) based on MI research. Thus, many MI in volcanic rocks represent magma that has degassed to some extent, and similarly, many MI in plutonic rocks may have also formed from and represent partially degassed magma. Melt inclusions in plutonic rocks may pose concerns in addition to those associated with MI of volcanic rocks. For example, the necessity to remelt and quench crystallized MI to a homogenous glass (i.e., revitrify) that represents the initial melt composition at the time of entrapment is challenging. Investigators have traditionally had two methods of MI revitrification at their disposal. Individual, MI-bearing phenocrysts may be observed directly with a microscope-mounted, hightemperature heating stage, in order to monitor the disappearance of each mineral and fluid phase. This approach permits the investigator to use minimum heating durations and temperatures, and thus diminish H2O loss during this process (details given below), because the moment at which the second-to-last phase dissolves into the silicate melt is observed directly. Alternatively, following some process of mineral separation, multiple phenocrysts of a given MI-host mineral may be heated either at one atmosphere or at elevated pressure with a hydrothermal vessel. This approach can generate a
describe criteria useful for evaluating MI. Several issues involving fluid terminology used in this chapter require clarification up front. Regarding magmatic volatile phases (MVP), we use the term ‘fluid’ in a broad or generic sense to refer to any of the following MVP types: vapor (i.e., comparatively low density, H2O- and/or CO2dominated phase); liquid (i.e., comparatively high density, H2O-dominated phase); magmatic hydrosaline liquid (referred to as MHL), salt melt, or brine (higher density phase dominated by alkali and other chloride species); and supercritical fluid. Herein, we treat H2O, CO2, the various S species, and Cl as volatile magma components; and F, P, and B are considered fluxing agents even though they do exhibit volatile behavior in some magmatichydrothermal environments. CHALLENGES IN INTERPRETING MELT INCLUSIONS FROM PLUTONS Melt inclusions, regardless of the igneous environment they represent, are prone to problems that may modify their chemistry and render them unrepresentative of bulk melt, so it is necessary to judge the quality of MI data sets before they are used to interpret magmatic processes. For instance, MI are crystallized, glassy, or can range between both physical states. Other problems include the competency versus incompetency of the host phase as regards cracking and volatile leakage after melt entrapment, post-entrapment MI crystallization and related challenges of MI refusion to homogeneous glass that is compositionally equivalent to original entrapped melt (without inclusion decrepitation), loss or gain of H2O and other constituents via diffusion through the host phenocryst after entrapment and while rehomogenizing MI in the laboratory, accidental co-entrapment of microphenocrysts in MI during host phase crystallization, and modification of MI chemistry during postentrapment fracturing of the host phase and subsequent alteration by influx of a metasomatizing fluid. Evidence, recognition of, and recommended treatments for these issues have been addressed by Roedder (1984), De Vivo & Frezzoti (1994), Lowenstern (1995, 2003), Lu et al. (1992), Nielsen et al. (1998), Frezzotti (2001), Danyushevsky et al. (2002), Cottrell et al. (2002), Anderson (2003), and references cited therein, so interested readers are referred to these sources for specific details. It has also been suggested that some MI may represent the entrapment of boundary layer melt not in equilibrium with bulk melt. As London (2005) 166
SILICATE MELT INCLUSIONS IN FELSIC PLUTONS: A SYNTHESIS AND REVIEW
larger database because hundreds of phenocrysts can be heated simultaneously, but it also requires more time because of its trial-and-error nature of determining optimal heating conditions. Rehomogenization of some crystallized MI requires elevated temperatures and extensive heating durations. These conditions enhance the loss of H2O from trapped melt (Massare et al. 2002), because water diffuses from the melt and through the host phase as H2O and H2 during the heating event (Roedder 1984, Qin & Anderson 1992). It follows that progressive loss of H2O steadily increases solidus and liquidus temperatures of entrapped melt, which may necessitate even higher homogenization temperatures and/or longer heating durations. An additional concern is that the dissociation of H2O and loss of H2 by diffusion vary with the ambient oxygen fugacity imposed by the host grain and the heating apparatus. The consequences of this process were shown by Webster et al. (2004) in their study of highly crystallized, granitic MI from Zinnwald, Germany. After heating MI-bearing quartz phenocrysts at 1025–1100°C for 20 to 30 hours in closed quartzglass tubes at one atmosphere the observed range of H2O in the glass of all but one MI was 0.3–0.7 wt.%. Such H2O contents are extremely low for granitic magmas, and they imply loss of H2O during revitrification. Subsequent work provided confirmation of this (Thomas et al. 2005). Crystallized MI from coeval, compositionally equivalent, Zinnwald granites were revitrified with different procedures that dramatically reduced the loss of water. The quartz host grains were: (1) enclosed in evacuated quartz glass ampules (also containing a small mass of nickel wire and a drop of oil to lower the ambient oxygen fugacity) and heated in a one-atmosphere tube furnace at 800–1040° C for 20 hours, (2) loaded into a partially closed Au capsule (30 mm long and 5 mm diameter) and placed in a horizontal, cold-seal pressure vessel. The vessel was pressurized with CO2 to 100 or 300 MPa, inserted into the preheated furnace and subsequently heated isobarically. These experiments were performed at temperatures between 600 and 850°C and for heating times of 20 to 50 hours. At the conclusion of the experiments, the vessel was removed from the furnace and quenched with cold air. For (3) the quartz grains were enclosed in H2O-bearing gold capsules and heated in cold-seal hydrothermal vessels at 850°C and 100 MPa for 24 hours. The resultant glasses contained from 3 to more than 20 wt.% H2O. This approach follows from other
studies that used elevated pressure to minimize inclusion decrepitation and reduce H2O loss through leakage (Thomas et al. 2000, Schmitt et al. 2002). An alternative means of studying H2O in MI was introduced by Thomas (1994c) who developed a method of estimating the viscosity and water content of hydrous silicate melts. This follows from Naumov (1979) who determined the H2O concentrations of MI by combining estimates of the inclusion volume with thermometric measurements. The recognition of MI that have leaked volatiles and other mobile constituents along cracks, after entrapment, is a particularly vexing concern. Melt inclusions have been found in quartz, feldspars, amphiboles, pyroxenes, apatite, topaz, and iron-titanium oxides in granitoid rocks. These minerals, however, are not equivalent as regards their competence as a host phase. For instance, they exhibit differing cleavage and parting characteristics, and theoretically, as the number of cleavage planes increases in a given mineral, so does the likelihood of MI or FI decrepitation and leakage (Tait 1992). The vast majority of the MI described herein are hosted by quartz, and as most quartz has no cleavage it is more likely to remain competent during sample preparation. It is noteworthy that some quartz may exhibit a weak form of pseudocleavage, but this is atypical. Melt inclusions that have leaked have been distinguished by homogenization behavior as well as textural and chemical evidence (Roedder 1984, Lowenstern 1995, Student 2002). Artificially high temperatures of rehomogenization have been used as evidence of volatile loss (Bazarova & Krasnov 1975, Roedder 1984). Textural evidence of leakage includes anomalously large or multiple shrinkage/vapor bubbles in glass and the presence of MI on cracks in the host phenocryst. Other researchers have culled individual MI, as “leakers”, from a data set because they contained anomalously low abundances of one or more volatiles (Lowenstern 1995). This approach assumes that lower volatile contents do not represent true magmatic values as might occur in highly degassed magma entrapped at shallow crustal pressures. Interestingly, recent work also suggests that some metals actually exhibit significant volatility in melt while reheating crystallized MI (Kamenetsky & Danyushevsky 2005). This observation requires further analysis. Other difficulties arise when establishing optimal heating temperatures and times for crystallized MI. Entrapped melt typically 167
JAMES D. WEBSTER & R. THOMAS
precipitates along the inclusion–host phase contact during post-entrapment MI crystallization, and it is difficult to establish how much of the host phase to remelt in order to return the inclusion to the composition of the initial entrapped melt. Prior studies have modeled the worst-case effects of overor under-heating MI-bearing quartz grains to compute the maximum possible influence on traceelement and volatile abundances in the MI (Webster et al. 1997, 2004). These computations are particularly straightforward for quartz because of its compositional simplicity. Most such studies have determined that the observed range in volatiles, fluxing components, and trace elements far outweighs the potential compositional influence of revitrifying MI at other than optimal conditions. Thus, for some applications of MI research, like determining abundances of volatiles, fluxing components, and trace elements, this issue may not be particularly significant. A final cautionary note involves MI located in phenocrysts characterized by complex zoning and embayment. Müller et al. (2003) discussed the utility of examining MI-bearing quartz phenocrysts with scanning electron microscope cathodoluminescence to search for and identify zoning and embayment patterns prior to analysis. This issue has bearing because relative ages of individual MI are not necessarily a linear function of distance from the phenocryst rim toward the core. For instance, it is practical to know if a particular MI-bearing phenocryst has sampled multiple magma batches, and clear recognition of this, using cathodoluminescence analysis, can assist in such situations.
plutonic rocks are partially to completely crystallized, contain trace to no observable glass, are subject to subsolidus reequilibration and metasomatism, and rarely show distinct shrinkage/vapor bubbles so they are difficult to recognize (Thomas 1979, Weisbrod 1981, Roedder 1984, Student 2002, Sirbescu & Nabelek 2003a, Bodnar & Student 2006). Interestingly, the comparative rarity of MI in plutonic rocks is in stark contrast with fluid inclusion (FI) abundances and types; Roedder (1984) observed that intrusive rocks show the widest variety of FI types of all environments studied. These inclusions range from aqueous, to sulfide-, carbonate-, and halogenenriched compositions; they may be quite abundant; and they provide definitive evidence for fluid– silicate melt immiscibility. For this chapter, we conducted a comprehensive literature search for MI from felsic plutonic rocks in order to determine granitic magma geochemistry and abundance ranges of volatiles and fluxing components in granitic magmas. We located 24 investigations and assembled a chemical database composed of more than 320 analyzed MI. Some data represent barren granitoid rocks in Scandinavia, Canada, Sardinia, South Africa, and at Ascension Island. The great majority of the plutons studied, however, are halogen-enriched, highly evolved granites genetically associated with lithophile mineralization including Be, F, Sn, Ta, W, U, Th, and/or Mo. Curiously, there are few published analyses of MI from porphyritic, Cu- and Momineralized granitoid rocks given that many investigations have reported highly saline, magmatic FI in these systems. The only studies relevant to the latter environments include mineralized porphyritic granitoid bodies in Japan, New Mexico, U.S.A., Chile, and the Philippines, and unmineralized but chemically related plutons of South Korea. Before addressing the volatile and fluxingcomponent abundances of these MI, it is instructive to assess the quality of these data sets. Nearly all of the MI, exclusive of the 15 Ascension Island MI and a few, rare MI in a few other studies, were reheated before analysis, so many of these MI may contain artificially low H2O concentrations. We include the latter data, nevertheless. An additional complication is that only 26% of the MI were analyzed for H2O directly by secondary ion mass spectrometry (SIMS) or confocal micro-Raman spectroscopy. Because of the paucity of directly measured H2O values, we have turned to the next best approach. We computed apparent H2O
BACKGROUND AND GENERAL OBSERVATIONS Melt inclusion research has an extensive history; MI have been recognized as quenched samples of silicate melt since the early observations of Sorby (1858) and Zirkel (1873). Research during the past two decades has seen an exponential rise in the number and quality of scientific investigations based on MI (Bodnar & De Vivo 2003). The large body of analytical data amassed for MI from felsic magmas has been derived primarily from volcanic rather than intrusive materials. This disproportion in available data (i.e., MI representative of melt in volcanic versus plutonic environments) is a consequence of several factors. For instance, phenocrysts from granitoid rocks generally contain fewer MI (Roedder 1984). Moreover, most MI in 168
SILICATE MELT INCLUSIONS IN FELSIC PLUTONS: A SYNTHESIS AND REVIEW
contents with the by-difference technique (i.e., the difference between 100 wt.% and the measured electron microprobe analytical totals). This approach, however, can be problematic for several reasons: all major and minor elements other than H2O in the glass must be measured, and their concentrations must be determined accurately, meaning that alkali mobility in hydrous glasses must be accounted for. Moreover, the maximum precision in apparent H2O content is limited by counting statistics. For example, SiO2 alone can be imprecise by 0.5 to 1.0 wt.% (using typical analytical conditions of 15 keV, 10–20 nA current, and a counting time for SiO2 of 20–30 s). As a consequence, the combined errors on H2O contents determined with the by-difference method may be ≥1 wt.%. The abundance of the dominant magmatic volatile, e.g. H2O, varies widely in MI in plutonic rocks; these MI glasses contain dissolved concentrations of hundreds of ppm to tens of wt.% H2O. Figure 8-1a displays H2O concentrations measured directly and those inferred with the by-difference
technique. Notably, 25% of the MI contain extremely low H2O contents of <0.5 wt.% (negative values determined with the by-difference technique are assumed to reflect abundances near zero). Such low values are anomalous and presumably reflect H2O loss during MI revitrification or possibly earlier, but they are included because some of them could represent accurate volatile contents of shallowly emplaced magma. In summary, the combined data show that most MI in granitoid rocks (i.e., 75%) contain less than 6.75 wt.% H2O, but the potential for artificially low values due to water loss while reheating is significant. Other MI contain higher to much higher H2O concentrations. For instance, prior research on volatile-rich granitic and pegmatitic rocks has identified two types of primary, coexisting, coeval melt inclusions (Thomas et al. 2000). They include type-A MI, which are dominated by aluminosilicate constituents, exhibit bubble volumes of 10–20%, and quench largely to glass after heating. The other variety, type-B MI which are very low density melts at high temperature, contains >80 mole % H2O, 0.8
B.
A. Cl (wt.%) in Melt Inclusions
H2O (wt.%) in Melt Inclusions
40
30
20
10
0.6
0.4
0.2
0
0 16
18
20
22
24
26
28
16
18
20
22
24
26
28
Modified Larson Index
Modified Larson Index
Fig. 8-1. Water (a) and Cl (b) concentrations of silicate melt inclusions (MI) versus modified Larson Index ((0.33)(SiO2)–
(FeO+CaO+MgO+TiO2)) (constituents in wt.%) from granite samples. In (a) small open crosses are revitrified MI with H2O computed with the by-difference method; small gray crosses are revitrified type-A/type-B inclusions analyzed for H2O by secondary ion mass spectrometry or confocal micro-Raman spectroscopy; small black crosses are revitrified MI analyzed for H2O by secondary ion mass spectrometry or confocal micro-Raman spectroscopy; and large filled crosses are unheated (naturally glassy) MI analyzed for H2O by secondary ion mass spectrometry or confocal micro-Raman spectroscopy. In (b) glassy and revitrified MI are undifferentiated. Some MI have probably lost some H2O during reheating; see text for discussion and clarification. Modified Larson Index does not include K2O and is used because some electron microprobe data may involve alkali migration or loss, thus giving incorrect estimates of the extent of melt evolution. Data sources: Takenouchi & Imai (1975), Thomas (1988), Förster et al. (1989), Hansteen & Lustenhouwer (1990), Frezzotti (1992), Thomas (1994a,b,c), Yang & Bodnar (1994), R. Thomas (unpub. data), Breiter et al. (1997), Thomas & Klemm (1997), Kamenetsky et al. (1999), Reyf et al. (2000), Haapala & Thomas (2000), Thomas et al. (2000), Webster & Rebbert (2001), Student (2002), Reyf (2004), Badanina et al. (2004), Chupin et al. (2001), Davidson & Kamenetsky (2001), Thomas et al. (2002), Schmitt et al. (2002), Audétat & Pettke (2003), Thomas et al. (2003), Webster et al. (2004), Thomas et al. (2005), and Davidson et al. (2005).
169
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
an average of 1.83 wt.% CO32– for 30 measurements of extremely H2O-enriched alkaline MI from the Amis Complex, Namibia. We note, however, that numerous primary FI in granites are CO2-enriched (Frost & Touret 1989, Takenouchi & Imai 1975, Thomas et al. 2000, Sirbescu & Nabelek 2003a), so carbon-bearing species in a MVP do play an important role in degassing and other processes in granitic magmas. Only 25 of the granitic MI were analyzed for S (Naumov & Kovalenko 1997), and the maximum S concentration measured is 0.11 wt.%. A modest number of granite-hosted MI have been analyzed for boron. Concentrations exceeding 2 wt.% B2O3 have been determined for MI representative of a few of the lithophile elementenriched granitoid bodies, but most MI from chemically evolved, F-enriched granitoid rocks contain ≤a few thousand ppm B2O3 (Thomas 2000, 2002, Schmitt et al. 2002, Thomas et al. 2003, Webster et al. 2004, Badanina et al. 2004, Thomas et al. 2005). Interestingly, FI containing several wt.% B2O3 are coeval with MI in samples of a pegmatite-related intrusive in Transbaikalia, Russia, (Peretyazhko et al. 2004). These FI represent trapped samples of granite-sourced MVP, and are generally consistent with the elevated B abundances in MI from some of the highly evolved granitic magmas described.
exhibits comparatively large bubble volumes with or without a visible aqueous liquid phase, contains comparatively less aluminosilicates, and homogenizes to an H2O-dominated solution. Directly measured H2O concentrations of the glass portions of type-A inclusions are up to 34 wt.%; whereas, the total reported H2O concentrations for several type-B MI (Thomas et al. 2005) actually approach 55 wt.%. The latter values were determined by summing the H2O measured in glass with the estimated H2O contents of the aqueous liquid and the vapor bubbles coexisting with glass. Details on these paired inclusion types are given by Thomas et al. (2000, 2006a). The other volatiles and fluxing components are generally less abundant than H2O in granitic MI. The range in Cl abundance is roughly an order of magnitude less than the range in F content (Figs. 8-1b and 8-2a). Most MI contain ≤0.3 wt.% Cl and ≤5 wt.% F. Magmatic abundances of P2O5 in felsic plutons cover a wide range (e.g., 0.01–2.8 wt.%), but most values exceeding 0.2 wt.% have been reported for highly evolved, F-enriched samples from the Eibenstock pluton of the German Erzgebirge (Fig. 8-2b). Constraints on C and S abundances in pluton-forming magmas are extremely poor. We located only one analysis for carbon in MI in granitic phenocrysts; Thomas et al. (2006) reported 10
3
P2O5 (wt.%) in Melt Inclusions
F (wt.%) in Melt Inclusions
A. 7.5
5
2.5
0 16
18
20
22
24
26
28
Modified Larson Index
B.
2
1
0 16
18
20
22
24
26
28
Modified Larson Index
Fig. 8-2. Fluorine (a) and P2O5 (b) concentrations of silicate melt inclusions versus modified Larson Index ((0.33)(SiO2)–
(FeO+CaO+MgO+TiO2)) (constituents in wt.%) from granite samples. Modified Larson Index does not include K2O and is used because some electron microprobe data may involve alkali migration or loss, thus giving incorrect estimates of the extent of melt evolution. Glassy and revitrified melt inclusions are undifferentiated. Data sources: Thomas (1988), Hansteen & Lustenhouwer (1990), Webster & Duffield (1991), Frezzotti (1992), Webster et al. (1993), Thomas (1994a,b), Yang & Bodnar (1994), Webster & Duffield (1994), R. Thomas (unpub. data), Breiter et al. (1997), Webster et al. (1996), Reyf et al. (2000), Haapala & Thomas (2000), Thomas et al. (2000), Webster & Rebbert (2001), Reyf (2004), Badanina et al. (2004), Thomas et al. (2002), Schmitt et al. (2002), Audétat & Pettke (2003), Webster et al. (2004, 2006), and Thomas et al. (2005).
170
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
application. To provide the proper context for this comparison, we have categorized the H2O data by their method of analysis, whether or not they represent revitrified MI prior to analysis, and if the MI were sampled from granites showing type-A and type-B inclusion behavior. All H2O concentrations reported here (Fig. 8-3) for granites containing typeA and type-B MI were measured directly and most range from 10–34 wt.%. In contrast, most other MI (i.e., those sampled from granites not exhibiting type-A and type-B inclusion behavior) contain <10 wt.% H2O. Of the latter, most directly measured and unheated MI from granitic rocks contain greater H2O than directly measured, heated MI from granites. Comparison of high-silica MI from rhyolite and directly measured, heated MI from granite not exhibiting type-A and type-B inclusion behavior shows roughly equivalent H2O contents. In contrast, the low H2O contents (i.e., both measured and apparent values) determined for many of the reheated MI from granite appear to be
COMPARISON OF MELT INCLUSIONS FROM FELSIC PLUTONS WITH THOSE OF EQUIVALENT VOLCANIC SYSTEMS Studies of silicate MI in felsic volcanic rocks have provided key constraints on magmatic processes. As summarized by Lowenstern (1995, 2003) and Wallace & Anderson (2000), melt inclusions in felsic volcanic samples have helped to understand the volatile abundances of magmas better, substantiate “early” magmatic volatile exsolution, and constrain compositions of volatile phases, depths of degassing and crystallization, and processes of magma evolution and hydrothermal mineralization. Given the vast insights and knowledge gained from hundreds of studies of MI in volcanic rocks and because of the comparative rarity of MI in granitic plutons, it is useful to compare the granite-hosted MI with their volcanic equivalents. The MI data for the granites summarized herein, however, require final qualification before 40
Granite-hosted, high-silica, & topaz rhyolite-hosted MI
H2O (wt.%) in Melt Inclusions
30
20
10
0 18
20
22
24
26
28
Modified Larson Index Fig. 8-3.
Water concentration of silicate melt inclusions (MI) versus modified Larson Index ((0.33)(SiO2)– (FeO+CaO+MgO+TiO2)) (constituents in wt.%) from granite samples (symbols same as in Figure 1) compared with H2O in MI of topaz rhyolites (open triangles enclosed by faint lines) and high-silica rhyolites (open circles enclosed by bold lines). Some MI have probably lost some H2O during reheating; see text for discussion and clarification. Modified Larson Index does not include K2O and is used because some electron microprobe data may involve alkali migration or loss, thus giving incorrect estimates of the extent of melt evolution. Data sources are those of Figure 1 and Anderson et al. (1989), Hervig & Dunbar (1992), Gerlach & McGee (1994), Lowenstern (1994), Webster et al. (1995), and Gerlach et al. (1996).
171
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
granite-hosted MI contain <0.2 wt. % Cl, whereas only 25% of MI from topaz rhyolites contain similar Cl abundances. For the former, the highest F contents occur in those MI with comparatively low Cl, and the highest Cl contents occur in comparatively F-deficient MI. The F and Cl contents of MI in high-silica rhyolites are less than those of most MI of granites and topaz rhyolites. The S abundances of most granite-hosted MI are similar to those of high-silica rhyolites. Seventy-five % of the granite-hosted MI contain <250 ppm S; Wallace & Anderson (2000) reported that MI of silicic volcanic rocks area generally ≤200 ppm. They attributed the low S values of highsilica rhyolitic magmas to low S solubility (resulting from low iron contents and low temperature ranges of crystallization). An alternate interpretation is that low S values may reflect sequestration of S by an exsolved MVP which is consistent with the conclusions of Hansteen & Lustenhouwer (1990) for Norwegian granites.
unreasonably low for felsic melts when considering other means of constraining H2O in pluton-forming magmas, e.g., experimental petrology and the presence of hydroxyl-bearing minerals (Clemens 1984, Burnham 1997). Nearly 25% of the MI contain <0.5 wt.% and 50% contain <3.2 wt.% H2O. The anomalously H2O-deficient nature of these reheated, granite-hosted MI is further substantiated through comparison with other MI sets. More than 96% of unheated, high-silica MI in volcanic rocks contain >0.5 wt.%, and 75% contain >3.5 wt.% H2O (Fig. 8-3). It is noteworthy that these MI in volcanic rocks data are average values representing more than 200 individual MI from 3 large-volume volcanic systems and 2 smaller volcanoes. Interestingly, the high-silica MI of rhyolite samples show much smaller ranges in H2O and in melt evolution index than MI from granitoid rocks. This suggests that the magmas experienced a smaller range in melt evolution during the time period of entrapment represented by the rhyolite-hosted MI. The H2O concentrations of MI from topaz rhyolites and those of felsic plutons not exhibiting type-A and type-B inclusions are roughly equivalent (Fig. 8-3). It is noteworthy that most of the topaz rhyolitic MI were reheated for extended times at one atmosphere, so significant diffusive H2O loss may have occurred in these samples as well. Interestingly, the range in F of MI from topaz rhyolites is similar to that of the granitehosted MI, but the former generally contain greater Cl contents (Fig. 8-4 a & b). Eighty percent of
GRANITE PETROGENESIS Magma evolution Investigations involving MI provide insights into magma genesis (Pettke 2006), the crystallization history of felsic intrusions, and information on other mechanisms of magma evolution. With MI research, one can also determine constraints on solidus and liquidus temperatures and the conditions of melt entrapment in order to understand better the emplacement and
0.8
0.6 Topaz & high-silica rhyolite MI
B
Granite-hosted MI & granitic whole rocks
Cl (wt.%) in Melt Inclusions
Cl (wt.%) in Melt Inclusions
A 0.6
0.4
0.2
0
0.4
0.2
0 0
2.5
5
7.5
10
F (wt.%) in Melt Inclusions
0
2.5
5
7.5
10
F (wt.%) in Melt Inclusions
Fig. 8-4. Chlorine versus F for (a) granite (melt inclusions as crosses, whole-rock samples as filled squares) and (b) rhyolite (topaz rhyolite as triangles and high-silica rhyolite as circles). References as in Figures 8-1 and 8-2 plus Förster et al. (1995, 1999).
172
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
crystallization of granitic magmas in the crust (Thomas & Klemm 1997). However, it should be kept in mind that phase homogenization experiments can lead to misinterpretations and yield artificially high or low solidus and liquidus temperatures depending on the microthermometric techniques employed (Roedder 1984). Further details on this application are beyond the scope of this chapter, but see Bodnar & Student (2006) for details. Early research on MI in plutonic rocks concentrated on interpretation of progressive changes in bulk-melt chemistry with particular attention paid to the dominant volatiles and fluxing components (Naumov 1969, Takenouchi & Imai 1975, Reyf & Bazheyev 1977, Hansteen & Lustenhouwer 1990). Subsequently, Thomas (2000) made a major advance in quantitative analysis for H2O in glass of very small MI with the introduction and application of confocal microRaman spectroscopy. More recent studies have taken the opportunity afforded by the advent of advanced, high-spatial resolution analytical techniques (Layne 2006, Halter & Heinrich 2006, Kamenetsky 2006, Pettke 2006), to constrain the maximum extent of enrichment of trace and potential ore elements in barren and mineralizing magmas. For example, Thomas et al. (2002) modeled granite melt production and subsequent melt evolution, via fractional crystallization, using SIMS analyses of trace elements in MI hosted by granitic zircon grains. Investigation of the range in abundance of highly incompatible trace elements in MI allows computation of magmatic enrichment factors which quantify the extent of crystallization that must occur to drive the composition of the more primitive melt (i.e., the MI with the lowest trace-element concentration) toward that of the more evolved, residual melt (i.e., the MI exhibiting the highest trace-element abundance). This approach, in effect, determines the magmatic crystallization history sampled by a set of MI, and the results of two recent studies are pertinent. In investigations of MI in quartz of xenoliths of barren granite erupted at Ascension Island (Webster & Rebbert 2001) and of coexisting, unambiguously coeval MI and FI in magmatic and hydrothermal quartz grains of two barren plutons of northern New Mexico (Audétat & Pettke 2003), modeling determined that the extent of melt evolution captured by these MI involved at least 50 and 99 percent crystallization, respectively. Few MI studies have found distinct evidence of magma mixing. Clearly, magma
differentiation in intrusions must be controlled by this process, as was shown by Müller et al. (2003) for plutons of Erzgebirge, Germany. It is just that most MI studies have not yet shown this to be the case. In summary, it has been generally observed from the preceding case studies that magma evolution is dominated by crystal fractionation with or without the exsolution of one or more immiscible volatile-rich phases. However, in distinct contrast to results of investigations involving volcanically erupted MI (Hervig & Dunbar 1992, Lowenstern 1995), most studies of MI in granitic intrusions have not yet determined distinct evidence of other processes of magma evolution such as magma mixing, assimilation, or contamination. Rheological properties of late-stage, volatilecharged magma The viscosity of silicate melts exerts a strong control on magmatic processes and properties including magma movement (i.e., ascent) and crystallization. For most felsic magmas, melt viscosity increases dramatically as temperature decreases. In contrast, however, MI-based studies of lithophile-element mineralized plutons have described volatile-rich, granitic residua that are apparently quite fluid and mobile. Investigators (Thomas 1994a, c, Thomas & Webster 2000, Thomas et al. 2000, Peretyazhko et al. 2004, Thomas et al. 2005) have estimated the viscosities of final melt fractions like these to have been extremely low (e.g., <10 Pa.s). This dramatic suppression of viscosity, even with significant cooling, is a consequence of elevated H2O, F, P2O5, and/or B contents (i.e., those measured and predicted) in granitic magmas, and a significant number of the MI in the data-set of this chapter come from evolved magmas like these. Some of these magmas are strongly peraluminous or peralkaline, and the presence of excess, networkmodifying aluminum or alkali ions in silicate magmas has a demonstrated influence on reducing melt viscosity (Dingwell 1987, Mysen 1988). These observations are particularly surprising given that late-stage crystallization of such volatile- and fluxing component-rich magmas may actually occur at extremely low temperatures (Sirbescu & Nabelek 2003a, b). Moreover, the H2O-rich type-B MI observed in some granite bodies also indicate that these media behave in a particularly fluid manner and thus are capable of highly effective late-stage 173
JAMES D. WEBSTER & R. THOMAS
at least, compositionally equivalent to felsic plutons and for which relevant constraints on volatile solubilities are available (Lowenstern 1994, Shinohara et al. 1995, Rapien et al. 1997, Webster 2004). The potential for MVP saturation in plutons exceeds that for compositionally similar eruptive magmas. This is true because felsic plutons crystallize completely and because crystallization is dominated by the volatile-deficient minerals quartz and feldspar. Total crystallization compels volatile concentrations of intermediate- to late-stage melts to achieve MVP saturation values at any depth in the crust. The last gasp of residual silicate melt in all felsic magmas must be MVP saturated no matter the depth or pressure of emplacement, because volatiles are, in effect, incompatible melt constituents. In contrast, the extent of crystallization in rhyolite-forming magmas is far less. The phenocryst contents of most rhyolitic rocks are <35 volume % (Hildreth 1981), and the modal abundances of late-stage microphenocrysts and microlites are comparatively small. Consequently, crystallization-driven degassing, also known as second boiling (Burnham 1997), is far more likely in plutonic environments, even though the bulk compositions, temperatures, pressures, and volatile abundances characteristic of these eruptive magmas are very similar to those of most intrusively emplaced felsic plutons. Granitic rocks show a variety of evidence for MVP exsolution in felsic plutons. Primary, saline FI characterized by magmatic homogenization temperatures are found in phenocrysts of Cu-, Au-, and Mo-mineralized plutons (Roedder 1984, Bodnar & Cline 1991, Cline & Bodnar 1991, 1994, Cline & Vanko 1995). Similar features have been observed for Sn- and W-bearing granitic plutons of the Erzgebirge, Germany, (Thomas 1994b) and at Beauvoir, France (Fabre et al. 2001). The presence of CO2-enriched FI, some of which exhibit enhanced salinity, in mineralized and nonmineralized granitoid rocks has also led investigators to conclude that the magmas in question exsolved one or more volatile phases during crystallization (Weisbrod 1981, Frost & Touret 1989, Roedder 1992, Sirbescu & Nabelek 2003a). The exsolution of a MVP in plutons is also evident in the observation of coeval primary MI, primary FI, and melt- plus fluid-bearing inclusions in other felsic plutonic rocks (Roedder & Coombs 1967, Reyf 1997, Kamenetsky et al. 1999, Yang & Bodnar 1994, Khetchikov & Pakhomova 1996). These observations are consistent with other
movement and element transport. It follows that the presence of low viscosity, late-stage melt pockets may play a key role in the formation of pegmatitic and aplitic facies in and around felsic plutons (Thomas et al. 2000, Peretyazhko et al. 2004, Thomas et al. 2000, 2006a). MAGMATIC VOLATILE BEHAVIOR Volatile exsolution The study of volatile phases exsolved from granitic plutons is crucial because these phases have the potential to influence the evolution of late-stage granitic melt and because they are genetically associated with a variety of metallic ores. As noted, granitic rocks may show partial to strong chemical modification resulting from hydrothermal alteration associated with these and other volatile-rich phases, so MI trapped in unaltered or less-altered phenocrysts preserve unique information on original magma chemistry (Davidson et al. 2005), and they provide an important means of studying magmatic volatile phases. The primary means of establishing magmatic volatile abundances and gradients in volatile content, the occurrence of pre-eruptive volatile saturation and the depths of MVP release, as well as the compositions of MVP in volcanic systems involves the analysis of MI glass and comparison of these data with experimentally determined volatile solubilities for known pressure and temperature conditions (Wallace & Anderson 2000). It was recently observed “Combined with the ever-increasing data set on volatile solubilities, MI may yield more reliable estimates of the depths to magma chambers than are currently available through mineral geobarometers” (Lowenstern 2003), so this approach is also appropriate for characterizing magmatic processes in plutonic systems. Water and CO2 are the dominant magmatic volatiles, and their solubilities in melt have been well established experimentally and modeled theoretically. This approach has helped to recognize and characterize numerous examples of MVP exsolution and volcanic eruption in high-silica rhyolitic magmas (Holloway & Blank 1994, Newman & Lowenstern 2002, Wallace & Anderson 2000). However, because constraints from MI on CO2 contents of felsic plutons are essentially null and knowledge of plutonic H2O abundances is poor, we cannot apply H2O–CO2 data to granitic magmas at present. In the meantime, we are forced to use other means of study. One method has been to apply MI from volcanic systems that are cogenetic to or are, 174
SILICATE MELT INCLUSIONS IN FELSIC PLUTONS: A SYNTHESIS AND REVIEW
behavior. For example, a functional experimental database exists for H2O and Cl in a variety of alkaline to aluminous, felsic melts at shallow crustal conditions. This is particularly relevant because constraints on CO2 abundances of felsic MI from plutonic rocks are unavailable for interpreting MVP exsolution (Webster 1992a, b, Webster & Rebbert 1998, Webster et al. 1999, Signorelli & Carroll 2000, Webster 2004, Webster et al. 2006). The H2O and Cl contents of many granitic MI are well determined (Fig. 8-1), and MI analyses have been compared with corresponding experimental solubility data to interpret degassing processes in a variety of felsic volcanic (Lowenstern 1994, 1995, Webster & Rebbert 1998, Signorelli & Carroll 2000, Wallace & Anderson 2000) and plutonic magmas (Reyf 1997, Kamenetsky et al. 1999, Reyf et al. 2000, Webster & Rebbert 2001, Schmitt et al. 2002, Audétat & Pettke 2003, Reyf 2004, Badanina et al. 2004). Toward this goal, we have culled the granitoid data set to identify and interpret only those plutonic systems for which H2O was measured directly, for which the MI were either unheated or heated only briefly during revitrification, and are free of mixed type-A and type-B inclusions (Fig. 85). The latter concern is important because Cl and H2O solubility experiments have not involved volatile- and fluxing component-charged systems with compositions analogous to those exhibiting type-A and type-B MI behavior described previously, so it would be imprudent to apply extant H2O and Cl solubility data to magmas exhibiting complex compositions and exsolution behavior like these. Given the lack of MI that were revitrified at conditions optimal for H2O loss and were analyzed for H2O directly, we can only apply published H2O–Cl solubility data to a small number of the granitic MI studied. Plotting H2O versus Cl for granite-hosted MI predicts pressure ranges of magmatic degassing for three of the plutonic systems: Ascension Island, the Cañada Pinabete pluton (New Mexico), and the Norwegian granites (Fig. 8-5). This figure shows volatile solubilities for two pressures and includes data for two endmember granite compositions at each pressure. The issue of composition is significant, because Cl solubility varies dramatically with the dominant cations in aluminosilicate melts (Webster & De Vivo 2002) and with the F and S abundances of MVP-saturated systems in particular (Botcharniko et al. 2004, Webster et al. 2006). These two granitic, end-member compositions were selected
textural signals of MVPs. Miarolitic cavities in granitoid rocks and unidirectional solidification textures in Climax-type, W- and Mo-mineralized porphyries are clear demonstrations of MVP saturation in granitic magmas (Lowenstern & Sinclair 1996). The compositions of MI from plutonic magmas provide key information on the timing of MVP exsolution relative to the degree of crystallization, the rate of magma ascent, and the extent of magma evolution. Several such investigations focused primarily on unmineralized (i.e., barren) plutons. One particularly interesting example, given the vast age of MI entrapment, involved crystallized MI in zircon of a metamorphosed Archean granodiorite pluton. The MI were analyzed after revitrification, and their extremely low Cl contents were construed to indicate exsolution of saline MVP from melt (Chupin et al. 2001). A study by Audétat & Pettke (2003) of MI and FI coexisting in quartz phenocrysts from two unmineralized plutons modeled the timing of MVP exsolution relative to the extent of crystallization and constrained the evolution of melt and associated volatile phases. They observed that the abundances of B, Mn, Rb, Mo, Cs, and W increased in residual melt and in coexisting “fluid” with progressive crystallization of MVP-saturated magma. It was concluded that <30% crystallization was required for MVP saturation and that the poor mineralizing potential of the volatile phases in these barren magmas was largely a consequence of the low salinity of the volatile phases. It is noteworthy that the analysis of MI and FI by laser ablation inductively coupled mass spectrometry (LA–ICP– MS) was crucial to this work. In a recent investigation of MI from barren Ascension Island granites, it was observed that the Li, Rb, Nb, Zr, Cs, Y, Ce, and Th contents exhibit negative correlations with the (Cl/H2O) of residual melt (Webster & Rebbert 2001). Some of these MI were coeval with highly saline, magmatic FI. The geochemical correlations observed in the MI were determined to be identical to theoretical relationships expressing the predicted increase in trace-element abundances associated with decreasing (Cl/H2O) in granitic melts saturated in Cl-enriched volatile phases during crystallization at closed-system conditions. Thus, the trace-element data were consistent with the textural evidence (i.e., coeval MI and FI in the granites) of MVP saturation. Other geochemical parameters characterizing felsic magmas are indicative of volatile phase 175
H2O Concentration (wt.%) of Silicate Melt
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
8
vapor
6
200 MPa
4 vapor
50 MPa
2 hydrosaline liquid
hydrosaline liquid 0 0
0.2
0.4
0.6
0.8
Cl Concentration (wt.%) of Silicate Melt Fig. 8-5. Water versus Cl concentration of experimentally determined and modeled (Webster & Rebbert 1998, Webster & De Vivo 2002) F-bearing, felsic silicate melts saturated in vapor ± hydrosaline liquid at 50 and 200 MPa. Maximum H2O solubility varies with pressure, and maximum Cl solubility varies with pressure and melt composition. Compositions of melt inclusions (Hansteen & Lustenhouwer 1990, Webster & Rebbert 2001, Audétat & Pettke 2003) from Ascension Island, Canada Pinabete, New Mexico, and the Oslo rift, Norway, are compared with volatile solubilities. Some MI have probably lost some H2O during reheating; see text for discussion and clarification. Each of these systems were fluid saturated, based on textural evidence, so the magmas exsolved a fluid phase at pressures between 50–200 MPa. See text for discussion.
because they extend over the predicted range in maximum Cl solubility for the array of melt compositions represented by these specific MI. The comparison indicates that the magmas would have saturated in vapor and/or MHL at pressures of 50– 200 MPa. Fortunately, we have independent information that these magmas were saturated in one or more MVP, (i.e., the MI in each system coexist with coeval FI), so we can also constrain the specific depths of fluid exsolution to be equivalent to this pressure range. It is noteworthy that using this approach one can only constrain the pressure at which volatile exsolution should theoretically occur if a magma ascends to a depth equivalent to that pressure, without independent textural or other chemical evidence of MVP saturation. We have also compared the observed Cl concentrations of some MI with the maximum Cl solubilities expected for the bulk composition of each MI, respectively, to establish first-order pressures estimates of MHL exsolution (Fig. 8-6). This technique applies only to Cl-enriched, felsic magmas that exsolved a MHL with or without
aqueous vapor. Any MI composition that lies on the 1:1 correlation line signifies an aliquot of granitic melt that would exsolve MHL ± vapor at 200 MPa, and any MI located to the right of the line corresponds to melt that would exsolve MHL ± vapor at P >200 MPa. Interestingly, none of the granitic MI plot on or right of the line, so these MI represent aliquots of granitic melt that were not MHL saturated at pressures ≥200 MPa at the time of melt entrapment. Melt inclusions lying to the left of the line can denote aliquots of granitic melt at 200 MPa that would not exsolve MHL at the time of entrapment, vapor-saturated melt at 200 MPa, or melt saturated in MHL ± vapor at P <200 MPa. Some MI lying to the left of the line could also represent magma that had exsolved a MHL ± vapor at 200 MPa prior to melt entrapment and had evolved to lower Cl contents by the time of entrapment, because the (H2O/Cl) of melt increases significantly during fractional crystallization at closed-system conditions (Webster & Rebbert 2001). The gap for predicted Cl concentrations <0.26 wt.% in the figure is a consequence of the 176
2.5
2.0
1.5 H2O generally >6 wt.% H2O generally < 6 wt.%
1.0
e n lin o i t ela corr 1 : 1
0.5
0.0 0
0.25
0.5
0.75
1
Measured Cl Concentration (wt.%) in Melt Inclusions
Modeled Cl solubility (wt.%) in Hydrosaline Liquid-saturated Silicate Melt at 200 MPa
Modeled Cl solubility (wt.%) in Hydrosaline Liquid-saturated Silicate Melt at 200 MPa
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
2.5
2.0
1.5
1.0
e n lin o i t ela corr 1 : 1
0.5
0.0 0
0.25
0.5
0.75
1
Measured Cl Concentration (wt.%) in Melt Inclusions
Fig. 8-6. Chlorine concentrations of (a) melt inclusions from granites (references same as in Figure 1) and (b) melt inclusions from granites, topaz rhyolites, and high-silica rhyolites (symbols and references same as in figure 2) compared with predicted Cl solubility for hydrosaline liquid- ± vapor-saturated silicate melt at 200 MPa (Webster & De Vivo 2002). Melt inclusions differentiated by approximate H2O concentrations (dashed line); some MI have probably lost some H2O during reheating; see text for discussion and clarification. All inclusions lie to left of Cl solubility line (i.e., 1:1 correlation line) so they represent aliquots of granitic melt that were not saturated in hydrosaline liquid at pressure ≥ 200 MPa.
crystallization of MVP-saturated felsic magma while Cl is strongly sequestered by the fluid, because the fluid/melt partition coefficients for these halogens are so dramatically different (Audétat et al. 2000, Webster et al. 2006). Thus, the generally low Cl contents of most granitic MI may indicate they represent partially degassed melt. The alkali abundances of some granitic MI may also reflect the consequences of volatile exsolution. Plotting Cl versus the N/NK ratio (i.e., the molar Na2O/Na2O+K2O) of the MI demonstrates that the greatest dispersion in N/NK occurs with comparatively Cl-poor granitic melts and the smallest range in N/NK correlates with Cl-enriched melts (Fig. 8-7a). We remind the readers that accurate analysis of alkalis in such small areas of hydrous glass can be quite difficult, so some fraction of the broad spread observed in the molar N/NK may be a result of analytical difficulties. Most of the observed variability in N/NK of these MI, however, may well be representative of granitic melt. Furthermore, the Cl concentrations of MI decrease with decreasing abundances of melt CAFEMIC (CaO, TiO2, FeO, and MgO) constituents (Fig. 8-7b), and this is inconsistent with expected behavior unless the melts are MVP saturated, because increasing Cl in residual melt is a consequence of fractional crystallization of fluidundersaturated melts. During the final stages of
fact that no known MHL-saturated granitic melt exhibits Cl solubilities less than this value at 200 MPa (Webster & De Vivo 2002). Moreover, Cl solubilities decrease (they shift downward and toward the correlation line) with decreasing pressure; so at 50 MPa, the Cl solubilities are approximately 90% of those for 200 MPa. Consequently, some fractions of granitic melt represented by these MI would exsolve a MHL if the corresponding magma ascended to depths equivalent to 50 MPa of pressure, and this is consistent with the presence of primary magmatic, highly saline FI in some of the granites. Partially degassed melt in melt inclusions It follows from the preceding discussion that many of these granitic MI represent fractions of melt that lost volatiles to a MVP prior to entrapment. Most granitic MI contain less Cl than their volcanic counterparts, and it is difficult to accept that granitic magmas should contain less Cl than rhyolitic magmas at equivalent conditions of magma evolution. Many topaz rhyolite MI contain similar F abundances as those in granitic MI, and without the presence and influence of a volatile phase in magma there is no reason to expect volcanic and plutonic systems to contain different Cl contents. Modeling has shown that F is increasingly retained in melt during fractional 177
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
0.10
1 Felsic Intrusions
Felsic Intrusions
B.
Moles (CaO+MgO+FeO+TiO2) in Melt Inclusions
Molar (Na2O/Na2O+K2O) of Melt Iinclusions
A. 0.8
0.6
0.4
0.2
0.05
0.00
0 0
0.1
0.2
0.3
0.4
0.5
0
0.6
Cl Concentration (wt.%) of Melt Inclusions
0.2
0.4
0.6
0.8
Cl Concentration (wt.%) in Melt Inclusions
Fig. 8-7. Molar (Na2O/Na2O+K2O) (a) and moles (CaO+MgO+FeO+TiO2) versus Cl concentration (b) of granite-hosted
melt inclusions. Decreasing Cl in melt as a function of decreasing moles (CaO+MgO+FeO+TiO2) reflects fractional crystallization of fluid-saturated melt. With progressive evolution and exsolution, molar (Na2O/Na2O+K2O) of residual melt shows strongly increasing dispersion which may reflect evolution of separate pockets of melt.
solidification, the effective MVP/melt ratio should be quite variable in dispersed pockets of residual, MVP-saturated melt, because at some point in the crystallization history the melt pockets should not necessarily have been in physical contact (i.e., in chemical communication) with each other. The S concentrations of these MI also express volatile exsolution of granitic magma. Most MI analyzed for this volatile contain low abundances; ¾ of the MI, in fact, contain <250 ppm S. This can be interpreted to indicate that granitic plutons simply contain low S abundances (which is consistent with some observations on solubilities in melt, Wallace & Anderson 2000) and/or that S was sequestered by MVP prior to entrapment of the MI (Hansteen & Lustenhouwer 1990). In summary, with all other variables equal, small variations in the effective MVP/melt ratio cause strong changes in the N/NK ratio and Cl concentration of silicate melts (Webster 1992a). Thus, we interpret these relationships to indicate that progressive evolution of MVP-saturated melts reduced the concentrations of Cl, S, and CAFEMIC constituents and increased the dispersion in the N/NK of very late stage, residual melt fractions.
2004). This is apparent in the F-rich granites exhibiting the two unique inclusion types (type-A and type-B MI of Thomas et al. 2005). These granites provide key constraints on late- to endstage processes of magma evolution and MVP exsolution as well as implications for late mineralization processes. Some of these granitehosted MI show distinct trends involving F in melt and bulk-melt composition (Thomas et al. 2005). Increasing F in melt is accompanied by decreasing melt peraluminosity which probably reflects crystal fractionation of topaz and aluminous micas found in the granites. When these magmas achieve approximately 3.5 wt.% F in melt, they undergo melt-fluid immiscibility. Subsequent melt and fluid evolution is characterized by increasing peraluminosity, Cl, and F in melt represented by type-A MI and increasing alkalinity with fairly constant F in melt represented by coexisting type-B MI. Figure 8-8 shows other related geochemical correlations associated with changes in F and H2O in these same inclusions. Recall from the preceding discussion that the highest F enrichments in granitic magmas like these involve extreme crystallization levels (Webster et al. 2004, Thomas et al. 2005). With cooling and progressive crystallization, the F and H2O contents increase in residual melt until, at about 28 wt.% H2O, the magma unmixes to form the two compositionally distinct “fluids” (e.g., aluminosilicate-rich aqueous fluid and H2O-rich aluminosilicate melt) which were trapped as the two types of MI. These relationships can be expressed as a solvus (Fig. 8-9) based on analyses of these
Immiscibility in late-stage fractions of granitic melt The residual melts of some highly evolved granitic magmas exhibit complex volatile solubility behavior, and the solubility relationships involved exert a strong control on MVP exsolution (Veksler & Thomas 2002, Veksler et al. 2002, Veksler 178
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
6
Fig. 8-8.
trend 2 increasingly aluminous
5
Fluorine in Inclusions (wt.%)
4
3 trend 1 decreasingly aluminous to increasingly alkaline
2
1
0 0
10
20
30
40
50
60
Fluorine versus H2O concentrations of melt inclusions from high-silica rhyolites (circles), topaz rhyolites (triangles), and granites (crosses) for which H2O was measured directly. Trends in data include (1) increasing H2O, presumably resulting from fractional crystallization, accompanied by decreasing aluminosity leading to increasing alkalinity. Fluid saturation occurs at about 28 wt.% H2O in F-enriched silicate melts, forming two fluids signified by type-A (follows trend 2) and type-B (follows trend 1 from point of intersection of two trends) inclusions in granites. With continued crystallization type-A melts become increasingly aluminous and type-B melts become increasingly alkaline. Melt inclusions with high F and low H2O contents may reflect H2O loss during revitrification and/or naturally H2O-poor silicate melts. See text for discussion. Data sources: Anderson et al. (1989), R. Thomas (unpub. data), Webster & Duffield (1991), Hervig & Dunbar (1992), Webster et al. (1993), Gerlach & McGee (1994), Lowenstern (1994), Webster & Duffield (1994), Gerlach et al. (1996), Thomas & Klemm (1997), Thomas et al. (2000), Webster & Rebbert (2001), Thomas et al. (2002), Schmitt et al. (2002), Audétat & Pettke (2003), Thomas et al. (2003), Webster et al. (2004), and Thomas et al. (2005).
H2O in Inclusions (wt.% - directly analyzed) 750
Temperature (oC)
Zinnwald granitic rocks
700 type-B FI
type-A MI 650
Fig. 8-9. Solvus for type-A, solute-poor
600
550 0
10
20
30
40
50
60
H 2O (wt.%) Concentration in Inclusions 179
melts and solute-rich type-B melts determined from melt inclusions in granites (Thomas et al. 2000, 2005) as a function of temperature and H2O concentration. See text for discussion.
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
inclusions (Thomas et al. 2005). The flanks of the solvus express the influence of temperature on the H2O contents of the two “fluids”. At the shallow crustal pressures of emplacement and crystallization of these magmas, both phases display complete miscibility at ≈728°C. These inclusions also show that the solubility of H2O in the peraluminous melt fractions, represented by type-A MI, decreases while F in residual melt increases (Fig. 8-8). Simultaneously, as evolution progresses, the solubility of H2O in the fluids represented by the type-B MI increases as the F content of melt remains comparatively constant. It is noteworthy, but beyond the scope of this chapter, that Thomas et al. (2005) discussed clear evidence that continued crystallization and evolution of both phases (e.g., those represented by the A- and B-type inclusions) leads to subsequent exsolution of other, secondary fluids from both phases.
crystallization, based on the observed ranges in H2O, F, Zr, and Nb in MI, that was similar to those of Schmitt et al. (2002). In another study of a highly evolved granitic system, Reyf et al. (2000) computed that up to 75 percent crystallization was necessary to explain the observed range in Ta for revitrified MI of the Li- and F-enriched granites of the Orlovka Ta deposit, Russia. Their study modeled the conditions required to saturate the granites in columbite and tantalite; thus determining the extent of pre-enrichment needed to generate magmatic lithophile mineralization. The orthomagmatic model can also account for subsequent processes of hydrothermal ore deposition associated with and occurring as a result of the magma-hydrothermal transition stage of granitic magma evolution (Burnham & Ohmoto 1980). At this transition, which follows the orthomagmatic stage, ore metals are sequestered from granitic magmas by hydrothermal fluids as the magmas undergo crystallization- and depressurization-driven degassing. The ore-laden fluids, many of which tend to be enriched in Cl and/or S, subsequently transport, deposit, and concentrate the metals as veins and disseminations within granitic plutons and their surrounding carapace of altered host rock. Clearly, many of these studies have involved lithophile element-mineralized intrusions, but MI-based investigations have also focused on base-metal mineralization. Audétat & Pettke (2003), for instance, recently studied MI in a copper-mineralized porphyry and concluded that magmatic pre-enrichment preceding mineralization required late-stage volatile exsolution as well as extreme degrees of fractionation. In another study, felsic MI in clinopyroxene of a syenitic dike associated with monzonitic intrusions of the Dinkidi Cu–Au porphyry deposit, Philippines, are accompanied by a variety of FI including hypersaline inclusions (Kamenetsky et al. 1999). The MI glasses exhibit significant depletions in alkalis, which are presumably the result of modified melt chemistry due to the exsolution of magmatic, alkali-rich brines. These same fluids were implicated in the transport of Cu and Au. In a related study, Student (2002) revitrified and analyzed crystallized MI in quartz phenocrysts of quartz latite dikes and quartz monzonite stocks associated with Cu-porphyry ores. He observed that the MI represented magmas trapped before, during, and subsequent to mineralization processes. The behavior of Cu and Zn was also investigated,
MAGMATIC AND MAGMATICHYDROTHERMAL MINERALIZATION Prior investigations of barren and mineralized felsic plutons that were based on the interpretation of MI ± FI chemistries have addressed the behavior of ore metals and volatiles during magma evolution, fluid exsolution, and mineralization. MI from mineralized granites can constrain the extent of pre-enrichment of ore and potential ore elements in residual melt that is a result of magmatic crystallization and is associated with ore formation. Many studies probing these issues follow the orthomagmatic model for granitoid-related mineralization which holds that Cu, Mo, Au, Sn, and many other metals are gradually enriched in residual melt as crystallization proceeds (Burnham & Ohmoto 1980, Hedenquist & Lowenstern 1994, Burnham 1997, Mustard et al. 2004). We review the results of a few, exemplary case studies here. In one investigation, Schmitt et al. (2002) measured volatiles and lithophile elements in revitrified, peralkaline granitic MI from the Zr-, Nb-, and REE-mineralized AMI Complex, Namibia. They also determined enrichment factors for Zr, Nb, and REEs in the alkaline magmas at a stage of melt evolution resulting from 50–75 percent crystallization. Audétat et al. (2000) analyzed MI in quartz and magmatic topaz and coexisting FI to track the enrichments of chalcophile as well as lithophile ore elements in residual melt. Reyf (2004) studied revitrified MI from alkaline F- and Be-mineralized granites of Yermakovka, Russia, and determined an extent of 180
SILICATE MELT INCLUSIONS IN FELSIC PLUTONS: A SYNTHESIS AND REVIEW
Y, Nb, Rb, and F involved >90 percent crystallization (Webster et al. 2004). The magma also contained one or more Li-, F-, and Cl-enriched volatile phases, but their presence simply slowed the rate of enrichment of Li and F in the residual melt during progressive melt evolution and did not lead to overall depletions of these constituents in residual melt. In a follow-up study involving different but related Zinnwald granites, the observed ranges in Rb, Cs, F, and H2O in MI were studied and determined to require greater than 99 percent crystallization (Thomas & Webster 2000, Thomas et al. 2005).
and it was determined that the Cu contents of MI representing more evolved melt were lower than those of the comparatively primitive MI (i.e., those in pyroxene from older andesitic magma) thus indicating the loss of Cu to mineralizing fluids as magma evolution progressed. Not all such studies agree that extensive magma differentiation is required to generate porphyry copper ores, however (Halter & Heinrich 2006). For one porphyry deposit, Halter et al. (2005) have shown that the porphyry serves only as the host rock for mineralization and alteration and that the fluid and ore metals are actually derived from an associated, but modestly evolved andesitic melt. Thus, the MI of this host porphyry actually yield little information on the genesis and geochemistry of the ore fluid(s). Study of lithophile-element mineralized and MVP-saturated granites has also provided important information on magmatic-hydrothermal processes. Interpretation of degassing in many of these systems is complicated, however, by the presence of elevated F abundances and the strong capacity of F to enhance H2O and Cl solubilities in melt, which tends to retard MVP saturation during progressive melt evolution. In studies cited previously, Reyf (1997), Reyf et al. (2000), Audétat et al. (2000), Schmitt et al. (2002), and Reyf (2004) concluded that Cl-bearing magmatic fluids had a key role in effecting ore element transport. In addition, Badanina et al. (2004) analyzed revitrified, peraluminous MI of Li- and F-rich granites of the Khangilay complex, Russia, to track magma evolution. They concluded that late-stage volatile depletions in magma and separation of Na from K in melt was caused by unmixing of residual melt into two liquid phases (silicate melt and a MHL). In their study of MI and coexisting FI in magmatic quartz and topaz, the modeling of Audétat et al. (2000) indicated that MVP exsolution occurred at roughly 40% crystallization of granitic magma. The results of extended fractional crystallization and MVP exsolution were also determined in investigations of revitrified, quartzhosted MI representative of the Sn- and Wmineralized, Cl-, F-, and B-enriched granitic magmas of the Zinnwald mining district, Germany. These MI are highly enriched in Sn, with concentrations far exceeding those of unaltered granitic whole-rock samples, and the Rb2O and Cs2O concentrations of some MI are in the wt.% range and thus exceedingly high. One study determined that the observed enrichments in REE,
SUMMARY AND RECOMMENDATIONS FOR FUTURE WORK Silicate melt inclusions in felsic plutonic rocks show interesting similarities and important differences in their volatile- and fluxing componentabundance ranges as compared to MI in compositionally equivalent volcanic rocks. The H2O concentrations of unheated or judiciously revitrified, granite-hosted MI are similar to and in some cases much higher than those of MI representative of high-silica or topaz rhyolite magmas. The presence of coexisting H2O-rich MI and solute-rich FI implies complete miscibility between some granitic melts and aqueous MVP(s), and hence the H2O concentrations of residual melt may reach extreme values (e.g. tens of wt.%) during magmatic differentiation. Chlorine contents of MI of granitoid rocks, high-silica rhyolite, and topaz rhyolite are typically <0.6 wt.%, but most granitoidhosted MI contain roughly half the Cl that is present in MI of topaz rhyolite. Dissolved S values are roughly equivalent for MI of granitoid rocks, highsilica rhyolite, and topaz rhyolite; and the F concentrations of lithophile-element enriched and/or mineralized granites and topaz rhyolite are equivalent. Comparison of these data with experimentally determined volatile solubilities determines pressures and timing of MVP exsolution, which are generally equivalent to those of high-silica and topaz-rhyolite forming magmas. There are critical differences in the compositions of MI from granitoid rocks versus those of eruptive silicic magmas, however. Most MI in plutonic rocks show progressive chemical evolution leads to decreasing abundances of CaO, TiO2, FeO, and MgO (as expected) and decreasing Cl in melt, but interestingly the molar (Na2O/Na2O+K2O) ratio shows significantly increasing variability. Coupled with the low S 181
JAMES D. WEBSTER & R. THOMAS
values, we interpret these characteristics to reflect the consequences of extreme compositional differentiation of small, isolated pockets of MVPsaturated residual melt. Strong to extreme melt evolution is also demonstrated for all MI studies involving mineralized plutons. These results are particularly significant given that eruptive magmas cannot achieve similar extremes of evolution resulting from fractional crystallization, or otherwise, they simply would be too viscous to erupt. Thus, MI from felsic plutonic systems provide unique constraints on volatile behavior during late-stage magma evolution. Analysis of the ranges of abundance of volatiles and trace and ore elements in MI of these systems establishes the timing of MVP exsolution, the scale and extent of crystallization, and in some cases the degree of crystallization leading to magmatic and/or magmatic-hydrothermal mineralization. In fact, the results of most studies cited herein are consistent with essential features of the orthomagmatic model for ore genesis. Future research should be directed at determining abundances of C-bearing species in granite-hosted MI, and more data are needed for S as well. The advent of new and emerging analytical technologies offers great promise to constrain volatiles, fluxing components, trace elements and ore elements better in MI and coexisting FI. Some methods permit direct analysis of crystallized MI without the need for prior revitrification. Reflection-based (as opposed to transmittancebased) spectroscopic techniques will permit the analysis of H- and C-bearing species in much smaller MI than currently possible. Moreover, in situ isotopic analysis of MI is becoming increasingly common, and the resulting data will provide crucial and unique constraints on numerous geological and geochemical processes.
REFERENCES ANDERSON, A.T., JR. (2003): Melt (glass ± crystals) inclusions. In Fluid Inclusions: Analysis and Interpretation (I. Samson, A. Anderson, & D. Marshall, eds.). Min. Assoc. Can. Short Course Series 32, 353–364. ANDERSON, A.T., JR., NEWMAN, S., WILLIAMS, S.N., DRUITT, T.H., SKIRIUS, C. & STOLPER, E. (1989): H2O, CO2, Cl and gas in Plinian and ash flow Bishop rhyolite. Geology 17, 221–225. AUDÉTAT, A., GUENTHER, D. & HEINRICH, C.A. (2000): Magmatic-hydrothermal evolution in a fractionating granite: A microchemical study of the Sn–W–F-mineralized Mole Granite (Australia). Geochim. Cosmochim. Acta 64, 3373–3393. AUDÉTAT, A. & PETTKE, T. (2003): The magmatichydrothermal evolution of two barren granites: A melt and fluid inclusion study of the Rito del Medio and Cañada Pinabete plutons in northern New Mexico (U.S.A.). Geochim. Cosmochim. Acta 67, 97–121. BADANINA, E.V., VEKSLER, I.V., THOMAS, R., SYRITSO, L.F. & TRUMBULL, R.B. (2004): Magmatic evolution of Li–F, rare-metal granites: A case study of melt inclusions in the Khagnilay complex, Eastern Transbaikalia (Russia). Chem. Geol. 210, 113–134. BAZAROVA, T. YU. & KRASNOV, A.A. (1975): Temperatures and sequence of crystallization of some leucite-bearing basaltoids. Akad. Nauk SSSR, Dokl. 222 (4), 935–938 (in Russ.). BELKIN, H.E., DE VIVO, B., ROEDDER, E. & CORTINI, M. (1985): Fluid inclusion geobarometry from ejected Mt Somma–Vesuvius nodules. Am. Mineral. 70, 288–303. BODNAR, R.J. & CLINE, J.S. (1991): Fluid inclusion petrology of porphyry copper deposits revisited; Re-interpretation of observed characteristics based on recent experimental and theoretical data. Plinius 5, 24–25.
ACKNOWLEDGEMENTS We acknowledge and appreciate reviews by Werner Halter, Jake Lowenstern, and James Student. This chapter also reflects the results of numerous prior discussions with Bob Bodnar, Benedetto De Vivo, Bobby Fogel, Werner Halter, Jake Lowenstern, Thomas Pettke, Jim Student, Ilya Veksler, and Paul Wallace, but we claim ownership of all errors within it. Research for this chapter was partially supported by NSF award EAR 0308866 to JDW.
BODNAR, R.J. & DE VIVO, B. (2003): Melt Inclusions in Volcanic Systems. Elsevier, Amsterdam, ix–x. BODNAR, R.J. & STUDENT, J.J. (2006): Melt inclusions in plutonic rocks: petrography and microthermometry. In Melt Inclusions in Plutonic Rocks (J.D. Webster, ed.) Min. Assoc. Can. Short Course 36, 1-25. 182
SILICATE MELT INCLUSIONS IN FELSIC PLUTONS: A SYNTHESIS AND REVIEW
BOTCHARNIKOV, R.E., BEHRENS, H., HOLTZ, F., KÖPKE, J. SATO, H. (2004): Sulfur and chlorine solubility in Mt. Unzen rhyodacite melt at 850°C and 200 MPa. Chem. Geol. 213, 207-225.
overview of techniques, advantages complications. Chem. Geol. 183, 5–24.
and
DAVIDSON, P. & KAMENETSKY, V.S. (2001): Immiscibility and continuous felsic melt-fluid evolution within the Rio Blanco porphyry system, Chile: Evidence from inclusions in magmatic quartz. Econ. Geol. 96, 1921–1929.
BREITER, K., FRYDA, J., SELTMANN, R. & THOMAS, R. (1997): Mineralogical evidence for two magmatic stages in the evolution of an extremely fractionated P-rich rare-metal granite: the Podlesi stock, Krusne hory, Czech Republic. J. Petrol. 38, 1723–1739.
DAVIDSON, P., KAMENETSKY, V., COOKE, D.R., FRIKKEN, P., HOLLINGS, P., RYAN, C., VAN ACHTERBERGH, E., MERNAGH, T., SKARMETA, J., SERRANO, L. & VARGAS, R. (2005): Magmatic precursors of hydrothermal fluids at the Rio Blanco Cu–Mo deposit, Chile: links to silicate magmas and metal transport. Econ. Geol. 100, 963–978.
BURNHAM, C.W. (1997): Magmas and hydrothermal fluids. In Geochemistry of Hydrothermal Ore Deposits (H.L. Barnes, ed.). 3rd edition, John Wiley & Sons, New York, p.63–123. BURNHAM, C.W. & OHMOTO, H. (1980): Late-stage processes of felsic magmatism. In Granitic Magmatism and Related Mineralization (S. Ishihara & S. Takenouchi, eds.). Min. Geol. Spec. Issue 8 (1–11).
DE VIVO, B. & FREZZOTTI, M.L. (1994): Evidence for magmatic immiscibility in Italian subvolcanic systems. In Fluid Inclusions in Minerals: Methods and Applications. (B. De Vivo and M.L. Frezzotti, eds.). Virginia Polytechnic and State Univ. (346–362).
CHUPIN, V.P., SOROKINA, G.A. & BARTON JR., J.M. (2001): Late Archean melt inclusions in zircon from charnockitic orthogneiss of Bulai Pluton (Limpopo Belt, South Africa) (abs.). XVI ECROFI, Porto 2001, 93–94.
DE VIVO, B., FREZZOTTI, M.L. & LIMA, A. (1993): Immiscibility in magmatic differentiation and fluid evolution in granitoid xenoliths at Pantelleria: fluid inclusion evidence. Acta Vulcanol. 3, 195–202.
CLEMENS, J.D. (1984): Water contents of silicic to intermediate magmas. Lithos 17, 273–287.
DE VIVO, B., LIMA, A., KAMENETSKY, V.S. & DANYUSHEVSKY, L.V. (2006): Fluid and melt inclusions in the sub-volcanic environments from volcanic systems: Examples from the Neapolitan area and Pontine Islands, Italy. In Melt Inclusions in Plutonic Rocks (J.D. Webster, ed.) Min. Assoc. Can. Short Course 36, 211-237.
CLINE, J.S. & BODNAR, R.J. (1991): Can economic porphyry copper mineralization be generated by a typical calc-alkaline melt? J. Geophys. Res. 96, 8113–8126. CLINE, J.S. & BODNAR, R.J. (1994): Direct evolution of brine from crystallizing silicic melt at Questa, New Mexico, molybdenum deposit. Econ. Geol. 89, 1780–1802.
DINGWELL, D.B. (1987): Melt viscosities in the system NaAlSi3O8–H2O–F2O-1. In Magmatic Processes: Physicochemical Principles (B.O. Mysen ed.). Geochem. Soc. Spec. Pub. 1 (423–432).
CLINE, J.S. & VANKO, D.A. (1995): Magmatically generated saline brines related to molybdenum at Questa, New Mexico, USA. In Magmas, Fluids, and Ore Deposits (J.F.H. Thompson, ed.). Min. Ass. Can. Short Course 23 (153–174).
FABRE, C., BOIRON, M.-C., MARIGNAC, CH. & AISSA, M. (2001): Li–F-rich magmatic fluids exsolved from rare metal granites: The example of the Beauvoir granite (French Massif Central); A microthermometric and LIBS study. (abs.). XVI ECROFI, Porto 2001, 145–147.
COTTRELL, E., SPIEGELMAN, M. & LANGMUIR, C.H. (2002): Consequences of diffusive reequilibration for the interpretation of melt inclusions. Geochem. Geophy. Geosys. 3, 10.1029/ 2001GC000205.
FÖRSTER, H.-J., SELTMANN, R. & TISCHENDORF, G. (1995): High-fluorine, low-phosphorus A-type (post-collision) silicic magmatism in the Erzgebirge. In Second Symposium on Permocarboniferous Igneous Rocks. Potsdam, Terra Nostra 7 (32–35).
DANYUSHEVSKY, L.V., MCNEILL, A.W. & SOBOLEV, A.V. (2002): Experimental and petrological studies of melt inclusions in phenocrysts from mantle-derived magmas: An 183
JAMES D. WEBSTER & R. THOMAS
FÖRSTER, H.-J., THOMAS, R. & TISCHENDORF, G. (1989): Physicochemical conditions as controlling factors on magmatism and metallogenesis. In Silicic Magmatism and Metallogenesis of the Erzgebirge (G. Tischendorf, ed.). ILC Publication 0171 (221– 243).
porphyry-type environments: A laser ablation– ICP–MS study of silicate and sulfide melt inclusions. In Melt Inclusions in Plutonic Rocks (J.D. Webster, ed.) Min. Assoc. Can. Short Course 36, 151-164. HALTER, W., HEINRICH, C.A. & PETTKE, T. (2005): Magma evolution and the formation of porphyry Cu–Au ore fluids: evidence from silicate and sulfide melt inclusions. Miner. Depos., 39(8): 845–863.
FÖRSTER, H.-J., TISCHENDORF, G., TRUMBULL, R.B. & GOTTESMAN, B. (1999): Late collisional granites in the Variscan Erzgebirge, Germany. J. Petrol. 40, 1613–1645.
HANSTEEN, T.H. & LUSTENHOUWER, W.J. (1990): Silicate melt inclusions from a mildly peralkaline granite in the Oslo paleorift, Norway. Min. Mag. 54, 195–205.
FREZZOTTI, M.L. (1992): Magmatic immiscibility and fluid phase evolution in the Mount Genis granite (southeastern sardinia, Italy). Geochim. Cosmoch. Acta 56, 21–33.
HEDENQUIST, J.W. & LOWENSTERN, J.B. (1994): The role of magmas in the formation of hydrothermal ore deposits. Nature 370, 519–527.
FREZZOTTI, M.L. (2001): Silicate-melt inclusions in magmatic rocks: Applications to petrology. Lithos 55, 273–299.
HERVIG, R.L. & DUNBAR, N.W. (1992): Cause of chemical zoning in the Bishop (California) and Bandelier (New Mexico) magma chambers. Earth Planet. Sci. Lett. 111, 97–108.
FREZZOTTI, M.L., DE VIVO, B. & CLOCCHIATTI, R. (1991): Melt–mineral fluid interactions in ultramafic nodules from alkaline lavas of Mount Etna (Sicily, Italy): melt and fluid inclusion evidence. J. Volc. Geotherm. Res. 47, 209–219.
HILDRETH, W. (1981): Gradients in silicic magma chambers: Implications for lithospheric magmatism. J. Geophys. Res. 86, 10153–10192.
FROST, B.R. & TOURET, J.L.R. (1989): Magmatic CO2 and saline melts from the Sybille Monzosyenite, Laramie Anorthosite Complex, Wyoming. Contrib. Mineral. Petrol. 103, 178– 186.
HOLLOWAY, J.R. & BLANK, J.G. (1994): Application of experimental results to C–O–H species in natural melts. In Volatiles in Magmas (M.R. Carroll & J.R. Holloway, eds.). Rev. Mineral. 30 (187–230).
GERLACH, T.M. & MCGEE, K.A. (1994): Total sulfur dioxide emissions and pre-eruption vaporsaturated magma at Mount St. Helens, 1980–88. Geophys. Res. Lett. 21, 2833–2836.
JOHNSTON, D.A. (1978): Volatiles, magma mixing, and them mechanism of eruption of Augustine Volcano, Alaska. Unpub. Diss., Univ Washington, 177p.
GERLACH, T.M., WESTRICH, H.R. & SYMONDS, R. (1996): Preeruption vapor in magma of the climactic Mount Pinatubo eruption: source of the giant stratospheric sulfur dioxide cloud. In Fire and Mud: Eruptions and Lahars of Mount Pinatubo, Philippines (C.G. Newhall, and R.S. Punongbayan, eds.). Philippine Institute of Volcanology and Seismology, Quezon City, and University of Washington Press, Seattle (415– 533).
KAMENETSKY, V.S. (2006): Melt inclusion record of magmatic immiscibility in crustal and mantle magmas. In Melt Inclusions in Plutonic Rocks (J.D. Webster, ed.) Min. Assoc. Can. Short Course 36, 81-98. KAMENETSKY V.S. & DANYUSHEVSKY, L. V. (2005): Metals in quartz-hosted melt inclusions: Natural facts and experimental artifacts. Amer. Mineral. 90, 1674–1678.
HAAPALA, I. & THOMAS, R. (2000): Melt inclusions in quartz and topaz of the topaz granite from Eurajoki, Finland. J. Czech. Geol. Soc. 45, 149– 154.
KAMENETSKY V.S., WOLFE, R.C., EGGINS, S.M., MERNAGH, T.P. & BASTRAKOV, E. (1999): Volatile exsolution at the Dinkidi Cu–Au porphyry deposit, Philippines: A melt-inclusion record of the initial ore-forming process. Geology 27, 691–694.
HALTER, W.E. & HEINRICH, C.A. (2005): Magmatic processes and volatile phase generation in
184
SILICATE MELT INCLUSIONS IN FELSIC PLUTONS: A SYNTHESIS AND REVIEW
KHETCHIKOV, L.N. & PAKHOMOVA, V.A. (1996): Fluids in biotite granite of the Voznessnskiy Complex, Khanka Massif: Evidence from fluid inclusions. Geol. Pac. Ocean 12, 855–862.
Min. Petrol. 120, 159–169. MASSARE, D., MÉTRICH, N. & CLOCCHIATTI, R. (2002): High-temperature experiments on silicate melt inclusions in olivine at 1 atm: Inference on temperatures of homogenization and H2O concentrations. Chem. Geol. 183, 87–98.
KOVAL, P.V. & PROKOFIYEV, V.YU (1999): P–T and H2O concentration of Mesozoic granitoid formation in the Mongol–Okhotsk zone, based on melt inclusion investigations, Mongolia, Siberia. Rev. Bras. Geosc. 29, 67–72.
MÜLLER, A., BREITER, K., SELTMANN, R. & PECSKAY, Z. (2003): Quartz and feldspar zoning in igneous rocks of the Eastern Erzgebirge pluton (Germany, Czech Republic): evidence of multiple magma mixing. Lithos 80, 201–227.
LAYNE, G.D. (2006): Application of secondary ion mass spectrometry to the determination of traditional and non-traditional light stable isotopes in melt inclusions. In Melt Inclusions in Plutonic Rocks (J.D. Webster, ed.) Min. Assoc. Can. Short Course 36, 27-49.
MUSTARD, R., ULRICH, T. & MERNAGH, T. (2004): Magmatic gold: A breakthrough in understanding the origin of gold through melt inclusion research. Econ. Geol. Res. Unit, James Cook University, April issue, 4-6.
LONDON, D. (2005): Granite pegmatites: As assessment of current concepts and directions for the future. Lithos 80, 281–303.
MYSEN, B.O. (1988): Structure and properties of silicate melts. Elsevier, 354pp., 1988.
LOWENSTERN, J.B. (1994): Dissolved volatile concentrations in an ore-forming magma. Geology 22, 893–896.
NAUMOV, V.B. (1969): Thermometric investingations of phenocrysts of quartz in quartz porphyry. Geokhimiya N4, 494–498 (in Russ.).
LOWENSTERN, J.B. (1995): Applications of silicatemelt inclusions to the study of magmatic volatiles. In Magmas, Fluids, and Ore Deposits (J.F.H. Thompson, ed.). Min. Ass. Can. Short Course 23, pp. 71–99.
NAUMOV, V.B. (1979): Determination of concentration and pressure of volatiles in magmas from inclusions in minerals. Geochem. Int. 16, 33–40. NAUMOV, V.B. & KOVALENKO, V.I. (1997): Sulfur concentration in magmatic melts: Evidence from inclusions in minerals. Geochem. Internat. 35, 887–94.
LOWENSTERN, J.B. (2003): Melt inclusions come of age. In Melt Inclusions in Volcanic Systems (B. De Vivo & R.J. Bodnar, eds.). Elsevier, Amsterdam (1–21).
NEWMAN, S. & LOWENSTERN, J.B. (2002): VOLATILECALC: a silicate melt–H2O–CO2 solution model written in visual basic for EXCEL*. Comp. & Geosc. 28, 597–604.
LOWENSTERN, J.B. & SINCLAIR, W.D. (1996): Exsolved magmatic fluid and its role in the formation of comb-layered quartz at the Cretaceous Logtung W–Mo deposit, Yukon Territory, Canada. Trans. Royal Soc. Edin. Earth Sci. 81, 291–303.
NIELSEN, R.L., MICHAEL, P.J. & SOURS-PAGE, R. (1998): Chemical and physical indicators of compromised melt inclusions. Geochim. Cosmochim. Acta 62, 831–838.
LU, F.Q., ANDERSON, A.T. & DAVIS, A.M. (1992): Melt inclusions and crystal–liquid separation in rhyolitic magma of the Bishop Tuff. Contrib. Mineral. Petrol. 110, 113–120.
PERETYAZHKO, I.S., ZAGORSKY, V.Y., SMIRNOV, S.Z. & MIKHAILOV, M.Y. (2004): Conditions of pocket formation in the Oktyabrskaya tourmalinerich gem pegmatite (the Malkhan field, Central Transbaikalia, Russia). Chem. Geol. 210, 91–111.
LU, F.Q., ANDERSON, A.T. & DAVIS, A.M. (1995): Diffusional gradients at the crystal/melt interface and their effect on the compositions of melt inclusions. J. Geol. 103, 591–597.
PETTKE, T. (2006): In situ laser ablation-ICP-MS chemical analysis of melt inclusions and prospects for constraining subduction zone magmatism. In Melt Inclusions in Plutonic Rocks (J.D. Webster, ed.) Min. Assoc. Can. Short Course 36, 51-80.
MARIANELLI, P., MÉTRICH, N., SANTACROCE, R. & SBRANA, A., (1995): Mafic magma batches at Vesuvius: a glass inclusion approach to the modalities of feeding stratovolcanoes. Contrib. 185
JAMES D. WEBSTER & R. THOMAS
QIN, L.R. & ANDERSON, A.T., JR. (1992): Diffusive reequilibration of melt and fluid inclusions. Am. Mineral. 77, 565–576.
mineralization. Geology 23, 1091–1094. SIGNORELLI, S. & CAPACCIONI, B. (1999): Behavior of chlorine prior and during the 79 AD Plinian eruption of Vesuvius (southern Italy) as inferred from the present distribution in glassy mesostases and whole pumices. Lithos 46, 715–730.
RAIA, F., WEBSTER, J.D. & DE VIVO B. (2000): Preeruptive volatile contents of Vesuvius magmas: contraints on eruptive history and behavior. I – the medieval and modern interplinian activities. Eur. J. Mineral. 12, 179–193.
SIGNORELLI, S. & CARROLL, M.R. (2000): Solubility and fluid–melt partitioning of Cl in hydrous phonolitic melts. Geochim. Cosmochim. Acta 64, 2851–2862.
RAPIEN, M., BODNAR, R.J., SZABÓ, CS.,& SIMMONS, W. (1997): Magma evolution at White Island, New Zealand: Evidence from melt inclusions (abs). EOS 78, S330–331.
SIRBESCU, M.-L. & NABELEK, P.I. (2003a): Crystallization conditions and evolution of magmatic fluids in the Harney Peak Granite and associated pegmatites, Black Hills, South Dakota – Evidence from fluid inclusions. Geochim. Cosmochim. Acta 67, 2443–2465.
REYF, F.G. (1997): Direct evolution of W-rich brines from crystallizing melt within the Mariktikan granite pluton, west Transbaikalia. Mineral. Deposita 32, 475–490. REYF, F.G. (2004): Immiscible phases of magmatic fluid and their relation to Be and Mo mineralization at the Yermakovka F–Be deposit, Transbaikalia, Russia. Chem. Geol. 210, 49–71.
SIRBESCU, M.-L. & NABELEK, P.I. (2003b): Crustal melts below 400° C. Geology 31, 685–688. SORBY, H.C. (1858): On the microscopical structure of crystals, indicating the origin of minerals and rocks. Quart. J.Geol. Soc. Lond. 14, 453–500.
REYF, F.G. & BAZHYEV, Y.D. (1977): Magmatogenic chloride solutions and tungsten mineralization. Geoch. Intern. 14, 45–51.
STUDENT, J.J. (2002): Silicate melt inclusions in igneous petrogenesis. Unpub. Ph.D. diss. Virginia Polytechnic Institute and State University, 119 p.
REYF, F.G., SELTMANN, R. & ZARZISKY, G.P. (2000): The role of magmatic processes in the formation of banded, Li, F-enriched granites from the Orlovka tantalum deposit, Transbaikalia, Russia: Microthermometric evidence. Can. Mineral. 38, 915–936.
TAIT, S. (1992): Selective preservation of melt inclusions in igneous phenocrysts. Am. Mineral. 77, 146–155. TAKENOUCHI, S. & IMAI, H. (1975) Glass and fluid inclusions in acidic igneous rocks from some mining areas in Japan. Econ. Geol. 70, 750–769.
ROEDDER, E. (1984): Fluid Inclusions (P.H. Ribbe, ed.). Rev. Mineral. 12, 646 p. ROEDDER, E. (1992): Fluid inclusion evidence for immiscibility in magmatic differentiation. Geochim. Cosmochim. Acta 56, 5–20.
THOMAS, J.B., BODNAR, R.J., SHIMIZU, N. & SINHA, A.K. (2002): Determination of zircon-melt trace element partition coefficients from SIMS analysis of melt inclusions in zircon. Geochim. Cosmochim. Acta 66, 2887–2901.
ROEDDER, E. & COOMBS, D.S. (1967): Immiscibility in granitic melts, indicated by fluid inclusions in ejected granitic blocks from Ascension Island. J. Petrol. 8, 417–451.
THOMAS, R. (1979): Investigations of inclusions for the thermodynamic and physico-chemical characterization of ore-forming solutions and processes in the magmatic and postmagmatic range. Unpub. Ph.D. dissertation, Bergakademie Freiberg, 245 p (in German).
SCHMITT, A.K., TRUMBULL, R.B., DULSKI, P. & EMMERMANN, R. (2002): Zr–Nb–REE mineralization in peralkaline granites from the Amix complex, Brandberg (Namibia): Evidence for magmatic pre-enrichment from melt inclusions. Econ. Geol. 97, 399–414.
THOMAS, R. (1988): Results of investigations of melt inclusions in various magmatic rocks from the northern border of the Bohemian Massif. Proc. First. Inter. Conf. on the Bohemian Massif (Z. Kukal, ed.), Czech Geol. Surv. pp. 298–306.
SHINOHARA, H., KAZAHAYA, K. & LOWENSTERN, J.B. (1995): Volatile transport in a convecting magma column: Implications for porphyry Mo 186
SILICATE MELT INCLUSIONS IN FELSIC PLUTONS: A SYNTHESIS AND REVIEW
THOMAS, R. (1994a): Thermometric study on granites from Bornholm Island, Denmark. Z. Geol. Wiss. 22, 139–145.
THOMAS, R., WEBSTER, J.D. & DAVIDSON, P. (2006): Understanding pegmatite formation: the melt and fluid inclusion approach. In Melt Inclusions in Plutonic Rocks (J.D. Webster, ed.) Min. Assoc. Can. Short Course 36, 189-210.
THOMAS, R. (1994b): Fluid evolution in relation to the emplacement of the Variscan granites in the Erzgebirge region: A review of the melt and fluid inclusion evidence. In Metallogeny of Collisional Orogens (R. Seltmann, H. Kaempf & P. Müller, eds.). Czech Geol. Surv. (70–81).
THOMAS, R., WEBSTER, J.D. & HEINRICH, W. (2000): Melt inclusions in pegmatite quartz: Complete miscibility between silicate melts and hydrous fluids at low pressure. Contrib. Mineral. Petrol. 139, 394–401.
THOMAS, R. (1994c) Estimation of the viscosity and the water content of silicate melts from melt inclusion data. Eur. J. Mineral. 6, 511–535
THOMAS, R., WEBSTER, J.D., RHEDE, D., SEIFERT, W., RICKERS, K., FÖRSTER, H.-J., & HEINRICH, W. (2006): The transition from peraluminous to peralkaline granitic melts in nature: evidence from melt inclusions and accessory minerals. Lithos, in press.
THOMAS, R. (2000): Determination of water contents of granite melt inclusions by confocal laser Raman microprobe spectroscopy. Amer. Mineral. 85, 868–872.
VEKSLER, I.V. (2004): Liquid immiscibility and its role at the magmatic-hydrothermal transititin: a summary of experimental studies. Chem. Geol. 210, 7–31.
THOMAS, R. (2002): Determination of the H3BO3 concentration in fluid and melt inclusions in granite pegmatites by laser Raman microprobe spectroscopy. Amer. Mineral. 87, 56–68.
VEKSLER, I.V. & THOMAS, R. (2002): An experimental study of B-, P- and F-rich synthetic granite pegmatite at 0.1 and 0.2 GPa. Contrib. Mineral. Petrol. 143, 673–683.
THOMAS, R. & KLEMM, W. (1997) Microthermometric study of silicate melt inclusions in Variscan granites from SE Germany: Volatile content and entrapment conditions. J. Petrol. 38, 1753–1765
VEKSLER, I.V., THOMAS, R., & SCHMIDT, C. (2002): Experimental evidence of three coexisting immiscible fluids in synthetic granite pegmatite. Amer. Mineral. 87, 775–779.
THOMAS, R., FÖRSTER, H.-J. & HEINRICH, W. (2003): The behavior of boron in a peraluminous granite-pegmatite system and associated hydrothermal solutions: A melt and fluid inclusion study. Contrib. Mineral. Petrol. 144, 457–472.
WALLACE, P.J. & ANDERSON, A.T., JR. (2000): Volatiles in magma. In Encyclopedia of Volcanoes, (H. Sigurdsson, ed.), Academic Press, San Diego (149–170).
THOMAS, R., FÖRSTER, H.-J., RICKERS, K. & WEBSTER, J.D. (2005): Formation of extremely Frich hydrous melt fractions and hydrothermal fluids during differentiation of highly evolved tingranite magmas: a complex melt/fluid-inclusion study. Mineral. Dep. 148, 582–601.
WEBSTER, J.D. (1992a): Fluid–melt interactions in Cl-rich granitic systems: Effects of melt composition at 2 kbar and 800° C. Geochim. Cosmochim. Acta 56, 659–678. WEBSTER, J.D. (1992b): Fluid–melt interactions involving Cl-rich granites: Experimental study from 2 to 8 kbar. Geochim. Cosmochim. Acta 56, 679–687.
THOMAS, R., FÖRSTER, H-J & TISCHENDORF, G. (1991): PTX–signatures of Hercynian oreproducing granites, Erzgebirge, Germany. In Source, Transport and Deposition of Metals (M. Pagel & J.L. Leroy, eds.). Balkema, Rotterdam (231–234).
WEBSTER, J.D. (2004): The exsolution of magmatic hydrosaline melts. Chem. Geol. 210, 33–48. WEBSTER, J.D. & DE VIVO, B. (2002): Experimental and modeled solubilities of chlorine in aluminosilicate melts, consequences of magma evolution, and implications for exsolution of hydrous chloride melt at Mt. Somma–Vesuvius. Amer. Mineral. 87, 1046–1061.
THOMAS, R. & WEBSTER, J.D. (2000): Strong tin enrichment in a pegmatite-forming melt. Mineral. Dep. 35, 570–582.
187
JAMES D. WEBSTER & R. THOMAS
WEBSTER, J.D. & DUFFIELD, W.A. (1991): Volatiles and lithophile elements in Taylor Creek Rhyolite: Constraints from glass inclusion analysis. Amer. Mineral. 76, 1628–1645.
Vivo, ed.). Developments in Volcanology 5, Elsevier, pp. 221–236. WEBSTER, J.D., TAYLOR, R.P. & BEAN, C. (1993): Pre-eruptive melt composition and constraints on degassing of a water-rich pantellerite magma, Fantale Volcano, Ethiopia. Contrib. Mineral. Petrol. 114, 53–62.
WEBSTER, J.D. & DUFFIELD, W.A. (1994): Extreme halogen abundances in tin-rich magma of the Taylor Creek rhyolite, New Mexico. Econ. Geol. 89, 840–850.
WEBSTER, J.D., THOMAS, R., RHEDE, D., FÖRSTER, H.-J. & SELTMANN, R. (1997): Melt inclusions in quartz from an evolved peraluminous pegmatite: geochemical evidence for strong tin enrichment in fluorine-rich and phosphorus-rich residual liquids. Geochim. Cosmochim. Acta 61, 2589– 2604.
WEBSTER, J.D. & REBBERT, C.R. (1998): Experimental investigation of H2O and Cl solubilities in F-enriched silicate liquids: implications for volatile saturation of topaz rhyolite magmas. Contrib. Mineral. Petrol. 132, 198–207. WEBSTER, J.D. & REBBERT, C.R. (2001): The geochemical signature of fluid-saturated magma determined from silicate melt inclusions in Ascension Island granite xenoliths. Geochim. Cosmochim. Acta 65, 123–136.
WEBSTER, J.D., THOMAS, R., FÖRSTER, H.-J., SELTMANN, R. & TAPPEN, C. (2004): Geochemical evolution of halogen-enriched, granite magmas and mineralizing fluids of the Zinnwald tin–tungsten mining district, Erzgebirge, Germany. Mineral. Dep. 39, 452– 472.
WEBSTER, J.D., BURT, D.M. & AGUILLON, R.A. (1996): Volatile and lithophile trace-element geochemistry of Mexican tin rhyolite magmas deduced from melt inclusions. Geochim. Cosmochim. Acta 60, 3267–3283.
WEISBROD, A. (1981): Fluid inclusions in shallow intrusives. In Fluid Inclusions: Applications to Petrology (L.S. Hollister & M.L. Crawford, eds.). Min. Assoc. Can. Short Course 6, pp. 241–271.
WEBSTER, J.D., CONGDON, R.D. & LYONS, P.C. (1995): Determining pre-eruptive compositions of late Paleozoic magma from kaolinized volcanic ashes: Analysis of glass inclusions in quartz microphenocrysts from tonsteins. Geochim. Cosmochim. Acta 59, 711–720.
YANG, K. & BODNAR, R.J. (1994): Magmatichydrothermal evolution in the “bottoms” of porphyry copper systems: Evidence from silicate melt and aqueous fluid inclusions in granitoid intrusions in the Gyeongsang Basin, South Korea. Inter. Geol. Rev. 36, 608–628.
WEBSTER, J.D., KINZLER, R.J. & MATHEZ, E.A. (1999): Chloride and water solubility in basalt and andesite liquids and implications for magmatic degassing. Geochim. Cosmochim. Acta 63, 729–738.
ZIRKEL, R. (1873): Die mikroscopische Beschaftenheit der Mineralien und Gesteine. Wilhelm Englemann, Leipzig, 502 p.
WEBSTER, J.D., SINTONI, M.F. & DE VIVO, B. (2006): The role of sulfur in promoting magmatic degassing and volcanic eruption at Mt. Somma– Vesuvius. In Volcanism in the Campania Plain: Vesuvius, Campi Flegrei and Ignimbrites (B. De
188
CHAPTER 9: UNDERSTANDING PEGMATITE FORMATION: THE MELT AND FLUID INCLUSION APPROACH Rainer Thomas GeoForschungsZentrum Potsdam, Telegrafenberg B120, D-14473 Potsdam, Germany E-mail:
[email protected] James D. Webster Department of Earth and Planetary Sciences, AMNH, Central Park West at 79th Street, New York, NY 10024-5192, USA and Paul Davidson ARC Centre of Excellence in Ore Deposits, University of Tasmania, Hobart 7001, Australia MI and FI. Processes and evidence of the late hydrothermal evolution stage, characterized by dilute aqueous solutions, are beyond the scope of this paper.
INTRODUCTION Despite extensive studies on pegmatites in the past (Barth 1928, Bjørlykke 1935, Brotzen 1959, Niggli 1920, 1937) and more recently (Černý 1982, Gordienko 1996, London 2005, Peretyazhko et al., 2004 Shmakin & Makagkon 1999, Strunz 1974) their origin and genesis remain controversial, particularly so for complexly zoned pegmatites. Some researchers (e.g., Fersmann 1931, Turner & Verhoogen 1960) consider them to be of igneous origin, while others believe that pegmatites are metamorphically or metasomatically generated (e.g., Barth 1962, Gresens 1967). In this paper we present evidence in favor of the igneous origin of pegmatites and provide insights into some major pegmatite-forming processes. Our evidence is based primarily on fluid and melt inclusions (FI and MI, respectively) and experimental studies, and suggests that the vast majority of granitic pegmatites are of magmatic origin, even though many may be affected by recrystallization and metasomatism. The very existence of silicate melt inclusions in minerals of pegmatites provides proof of their magmatic origin. In this contribution, we also discuss melt and fluid inclusion evidence concerning the nature and evolution of some granitic pegmatites. Our observations bear primarily on F-enriched and/or alkaline pegmatites, but they are also useful for considering processes involving simple pegmatites composed of feldspar, quartz, and little else. We confine ourselves to major processes of pegmatite formation that can be constrained by the study of
BACKGROUND ON PEGMATITE GENESIS According to Fersmann (1931), granite pegmatites can be defined as crystallization products of residual solutions that exsolve during solidification of granitic magma. In contrast, we suggest that many are crystallization products of residual H2O-rich melt fractions produced by melt– melt immiscibility during late-stage fractionation of granitic magma. Pegmatites are characterized by an arranged sequence of mineral associations, a remarkable size range of the individual crystals, simultaneous crystallization of different mineral phases, and commonly by an enrichment of highly volatile constituents and incompatible elements. Fersmann (1931) regarded the pegmatite-forming process as part of a continuous physicochemical transition that accompanies cooling of granitic magma. It starts at magmatic conditions and proceeds to hydrothermal temperature and pressure conditions. This definition can mutatis mutandis be applied to pegmatites of other magma types. During the evolution of felsic magmas the crystallization of anhydrous quartz and alkali feldspars enriches the residual homogeneous melt in incompatible constituents, and the enrichment in water is of particular importance (Roedder 1984a). Water is the most important constituent thus concentrated in the melt by crystal fractionation,
Mineralogical Association of Canada Short Course 36, Montreal, Quebec, p. 189-210. 189
RAINER THOMAS, JAMES D. WEBSTER & PAUL DAVIDSON
Kravchuk (1978) using the system SiO2–Na2O– H2O), may be useful in explaining certain aspects of pegmatite development. Mustart (1972) has shown, using phase relations in the peralkaline portion of the Na2O–Al2O3–SiO2–H2O system, that natural peralkaline melts may dissolve much higher concentrations of H2O than haplogranite melts. For example, the addition of sodium disilicate (Na2Si2O5) to the albite–H2O system dramatically increases H2O solubility at 5 kbar: 15 wt.% H2O at 700°C, 21 wt.% at 600°C, and up to 43 wt.% at 530°C (see also Preston et al. 2003). However, F-rich peralkaline pegmatites can also contain very high concentrations of alkali carbonates and H2O (Amis granite/Namibia, see Thomas et al. 2006b). Strongly peralkaline melt or fluid fractions are very reactive, and therefore will exchange components with their surroundings until equilibrium is attained. Thus, the peralkaline magmatic stage may be transitory and the rocks may show little evidence of the former peralkalinity (Mustart 1972). As a result, a “peralkaline history” is commonly conserved only in the isolated melt, fluid and mineral inclusions (Thomas et al. 2006b). This applies particularly to simple pegmatites composed only of quartz, feldspar, and mica, since volatiles and semivolatiles like F, B, and P are only present in very low concentrations. It follows that determining the formation mechanism of simple pegmatites, which are commonly genetically related to Precambrian granites formed at depths of 7–11 km, is a challenge since H2O and CO2 have comparatively low solution capacities for aluminosilicate components (Burnham 1967) and normally the solubility of CO2 in granitic melts is very low. Therefore, any applicable model of formation requires a medium of transport and crystallization like the alkaline varieties described herein, that does not involve the low concentrations of H2O, CO2, and F which would be expected in typical mid-crustal granitic melts. Another important issue bearing on pegmatite formation is the solubility of aluminosilicate components in coexisting fluids. Based on extensive field and laboratory studies Jahns & Burnham (1969) emphasized the role of H2O as the dominant constituent of a separate supercritical fluid phase. However, they also stressed the importance of an initial phase of closedor restricted-system crystallization in the presence of hydrous silicate melt, followed by exsolution of a coexisting supercritical aqueous fluid from melt, and finally by crystallization of pegmatite-forming
due to its greater abundance in most granitic magmas and its major effects on melting and crystallization of minerals and rocks by lowering liquidus and solidus temperatures. Water also decreases the viscosity of aluminosilicate melts, and together with other volatiles and semi-volatiles (particularly F, B2O3, and Cl), it enhances the transfer of ions through the melt to the growing crystal interfaces, leading to unusual sizes of crystal growth. Černý (1975) wrote “the separation of a supercritical fluid marks the beginning of fundamental changes in texture and distribution of the crystallizing materials, and also the beginning of reactions and exchanges between the earlier solids…”. Given the high partition coefficients (concentration in fluid/concentration in melt) for alkalis and volatiles from silicate melts to coexisting aqueous fluids it follows that hydrothermal fluids at the melt-dominated pegmatite stage are likely to be particularly concentrated in these constituents. Kamenetsky et al. (2002) and Webster (2004), for example, have demonstrated that some granitic magmatichydrothermal fluids are dominated by coexisting vapor and complex multiphase brines. Niggli (1920) also emphasized the outstanding significance of H2O for the formation of pegmatites: “Melt solutions in which water is plentiful as an essential volatile component besides large amounts of non-volatile components are called pegmatitic solutions”. However, because H2O solubility in aluminosilicate melts depends strongly on pressure and less so on temperature and because an important number of granite pegmatites form at shallow crustal conditions (i.e., at depths of about 3–5 km, Kozlowski 1978), H2O enrichment is limited. In the system Qz–Ab–Or–H2O at 2 kbar and 680°C the H2O content at the eutectic point is about 6.4 wt.%, and the viscosity is approximately 105.5 Pa.s (Holtz et al. 2001). At such high viscosities, the formation of typically coarsegrained or giant-textured pegmatites is hard to conceive. London (1999, 2005) attempted to solve the viscosity problem with the assumption of a boundary layer enriched in incompatible fluxing components that is present along phenocryst surfaces. Alternatively, Beus (1983) proposed a model for the giant growth of crystals embedded in hard rock based on metasomatic processes. It has been suggested that some properties of alkaline systems, and the importance of critical phenomena and immiscibility between fluids and alkaline melts (as observed by Valyashko & 190
UNDERSTANDING PEGMATITE FORMATION: THE MELT AND FLUID INCLUSION APPROACH
intriguing peculiarity of this system is the existence of a solvus linking a H2O-rich silicate melt and a silicate-rich fluid at pressures ~1 kbar (see Fig. 9-2). There is increasing evidence for the coexistence of aluminosilicate melt, hydrosaline melt, and lower salinity aqueous fluid (herein characterized as melt–melt immiscibility) in natural complex, multi-component systems over a broad pressure–temperature range. This is supported by the MI and FI data described herein, and has been confirmed experimentally by Veksler et al. (2002) and Veksler & Thomas (2002). The recognition that three coexisting immiscible liquid phases may be stable is, in fact, an important advance over older, classical concepts of pegmatite development. Multiple coexisting phases influence element distribution and, in particular, crystallization behavior. Because the wetting and solvent behaviors of these liquids are entirely different,
minerals from the aqueous fluid alone. However, experiments by Bureau & Keppler (1999) showed that haplogranite melts only exhibit complete miscibility with aqueous supercritical fluid at pressure of ≈16.9 kbar at ≥825°C when H2O is the sole volatile present. Such P–T values greatly exceed those for natural processes in the upper crust where most granites and granitic pegmatites are emplaced and evolve. This implies that the addition of other components may be critical. In a study on the effect of F, B, and excess Na on the critical curve in the albite–H2O system, Sowerby & Keppler (2002) observed complete miscibility between melt and fluid in the final stages of crystallization of a complex pegmatitic system at comparatively low temperatures and pressures (Fig. 9-1). These experimental results are directly applicable to natural pegmatite-forming systems. Studies of MI and FI provide critical evidence for liquid compositions and phase relations in nature. The MI studies of Thomas et al. (2000) demonstrated that complete miscibility is possible even at the considerably lower pressures characteristic of the Ehrenfriedersdorf granite– pegmatite complex. Complete miscibility between melt and fluid(s) was attained at 712°C and 21.5 wt.% H2O. The cause for this peculiar behavior can be seen in the complex interplay of the volatiles H2O, F, and Cl; the semi-volatiles B2O3 and P2O5; and fluxing components such as Li, Rb, and Cs; along with SiO2 and Al2O3 in the melt. The most
Glass Liquid Vapor
Temperature (°C)
750
a)
C.P.
700 650 600
Two liquids 550 500 0
20
40
60
80
100
H2O (mol%)
b)
Type-A MIs
Type-B MIs
FIG. 9-2. Schematic characteristics of three different melt inclusion types (type-A, near critical, and type-B) in quartz from the Ehrenfriedersdorf pegmatite, Germany, at room temperature after rehomogenization at given temperatures and a pressure of 1 kbar (from Thomas et al. 2000, 2003, 2006b). The arrows show the relations of the melt inclusions to the measured solvus of the pseudobinary melt–H2O– system of Ehrenfriedersdorf. C.P. = critical point. See text for discussion.
FIG. 9-1. P–T projection of the critical curves in the albite–H2O system and the synthetic pegmatite–H2O systems according to Sowerby & Keppler (2002) completed by our own data. a) albite–H2O system, b) pegmatite systems. See text for discussion.
191
RAINER THOMAS, JAMES D. WEBSTER & PAUL DAVIDSON
observed and experimental data for H2O solubility in peralkaline melts. We add a critical caveat to this comment on experimental studies. Niggli (1937) observed “The following consideration is particularly important, however. If a mineral aggregate that formed from magma containing volatile constituents is later liquefied again, and if the highly volatile substances had left the magma in the latest stage, then the re-melting process, without a re-supply of the same constituents can in no way be reverse of the solidification process”. Thus, the recognition of the prior existence of fugitive volatile phases can be particularly difficult, but is essential.
immiscible phases are not necessarily trapped in inclusions in the same proportions as present in the parental melts. Each of the different phases may also involve varying rates of crystal growth. Our studies on natural MI and FI and our experimental work on liquid immiscibility suggest that the formation of peralkaline melt fractions and the subsequent exsolution of multiple liquids are crucial to the melt-dominated stage of pegmatite formation (Thomas et al. 2000). Moreover, liquid immiscibility need not occur as a single event; multiple exsolution events are possible as long as the aluminosilicate melts are H2O enriched (Thomas et al. 2006b). It is likely that at the end of the meltdominated stage, this process occurs in small pockets of residual melt separated by large crystals of quartz, feldspar and other minerals (Thomas et al. 2006b). It is noteworthy that primary melt signatures are very often modified or destroyed by re-crystallization in the concluding hydrothermal stage. For example, in large pegmatite bodies (e.g., Tanco, Canada, Mozambique, Africa, and Borborema Pegmatite Province, NE Brazil) the relationship between the different types of FI and MI generated during different magmatic and magmatic–hydrothermal events is often not discernable. This is also true for the large chamber pegmatites of Volyn, based on Lemmlein et al. (1962) study of MI in topaz crystals. In recent years, models based on rapid recrystallization of supercooled silicate melt have become popular (Webber et al. 1999, London 1992, 2005). Proponents of this view (London 1992, 2005) start out with the premise of low H2O concentrations in pegmatite-forming melts, and, in fact, London et al. (1989) have developed this theory extensively. For example, Webber et al. (1999) used an initial H2O content of 3 wt.% in the melt to model the cooling rates and crystallization dynamics of shallow-level pegmatite aplite dikes, San Diego County, California. London (2004) wrote “The high viscosities of hydrous granitic melts at liquidus temperatures (~105 Pa.s, the viscosity of window putty, at 680°C, 2 kbar PH2O) or below (e.g., ~108 Pa.s, the viscosity of cold asphalt, at 400°C, 2 kbar PH2O) were thought to work against long-range diffusion, but also would impede the buoyant separation of an aqueous phase.” However, we contend that such low-H2O conditions are contrary to pegmatite-forming processes in systems we have studied, according to the available MI and FI analyses and the
COMMON PEGMATITES VS. RAREELEMENT PEGMATITES During fractional crystallization of some granitic magmas, melt evolution leads to the formation of silicic, alkali-rich hydrous melts which, when they crystallize, form texturally unique granitic rocks. They are comparatively coarse grained and are called “common pegmatite”, owing to the increased concentration of volatiles and the comparatively simple bulk composition of the magmas (Roedder 1981). If crystallization starts in a closed system, such as occurred in the chamber pegmatites in Volynia (Bakumenko et al. 1979), the first minerals to crystallize are feldspars and quartz and they often form graphic intergrowths. Our MI studies of such systems show that crystallization starts from H2Orich aluminosilicate melts. Crystallization of feldspars and quartz from the melt further enriches the residual melt fraction in incompatible components, and particularly in H2O. However many other species, in particular Li, Be, B2O3, CO2, F, Cl, and S as well as Rb, Cs, W, Nb, Ta, Sn, and others, that initially are present in low concentrations can also be concentrated to very high levels. With cooling and further enrichment in volatiles, the residual melt divides into two or more melt fractions by melt–melt immiscibility (Thomas et al. 2006b). These enriched residual melts are volumetrically minor and are likely to be much more mobile than a crystallizing melt. The latter characteristic facilitates its separation within, or escape from, the larger magma body. Any further melt evolution is associated with a gradual or sudden transition to a fluid-dominated system, in which highly concentrated hydrosaline brines coexist with a vapor phase. The latter phases play a key role in the subsequent concentration of rare 192
UNDERSTANDING PEGMATITE FORMATION: THE MELT AND FLUID INCLUSION APPROACH
elements in ore minerals. It is noteworthy, that in contrast to the relatively simple situation described above for chamber pegmatites or miarolitic cavities, observations and relationships like these can be easily overlooked. This may apply, in particular, to large irregular pegmatite bodies, like those at Tanco or to Stockscheider pegmatites of the Erzgebirge and similar localities.
directly from other sources. Melt inclusions from pegmatitic rocks have been observed and studied in a variety of phenocrysts including quartz, feldspar, topaz, tourmaline, beryl, garnet, diopside, mica, oxide minerals, and other rare minerals. Through improvements in analytical techniques in recent years it is possible to study very small MI. For example it is now possible to determine H2O concentrations of 0.05 to over 35 wt.% in small glassy MI deep within host phenocrysts using confocal micro-Raman spectroscopy (CMRS, see Thomas (2000) and Thomas et al. (2006a)). To investigate compositional differences between granitic bulk rocks and MI more thoroughly, a large number of MI in quartz and topaz from the chemically and mineralogically complex topaz–zinnwaldite–albite granites of Zinnwald (Eastern Erzgebirge, Germany) and pegmatites from the Ehrenfriedersdorf Sn–W deposits (Central Erzgebirge, Germany) were studied (Thomas et al. 2000, 2003, Webster et al. 2004, Rickers et al. 2006). Additionally, other, related pegmatites with less complex mineralogy and composed mainly of quartz and feldspar were also investigated (Thomas et al. 2000). These studies determined very high concentrations of H2O in MIs, as well as relatively large differences between the bulk-rock and MI compositions. Large differences in the abundances of volatiles and semi-volatiles H2O, F, B2O3, and Cl as well in the alkalis between syngenetic MI and FI were observed in single growth zones of quartz crystals.
MELT AND FLUID INCLUSIONS IN PEGMATITE MINERALS General remarks A basic requirement of any model of pegmatite formation is an understanding of the initial composition of the pegmatite-forming melt. As previously noted, initial volatile concentrations are fugitive and reactive components readily equilibrate with their surroundings, so the whole-rock composition of a pegmatite may hold little similarity to its parental melt composition. With some qualification, MI and FI provide the best method of constraining parental melt compositions, so generally, if inclusion work is ignored or skimped, further model development is difficult. Melt inclusions Melt inclusions are small blebs of silicate melt that are trapped within phenocrysts at magmatic temperatures and pressures (Roedder 1979, 1981, Lowenstern 1995, 2003). Generally such inclusions are small (1–200µm) in plutonic rocks, but larger MIs (very rarely >1 mm) are trapped in minerals of extrusive rocks. However, the small inclusions in minerals of intrusive rocks are often only 5 µm in diameter. If such trapped droplets of melt are cooled relatively quickly (as is typical of extrusive environments), the inclusion forms a uniform glass phase, typically with a well-formed shrinkage bubble. However, the slow cooling that is typical of intrusive environments causes the glass to devitrify to a dense, finely crystalline mass, often with deformed vapor bubbles or only a fluid film between the crystal phases (Davidson & Kamenetsky 2001, Davidson et al. 2006, Bodnar & Student 2006). Because MIs are often enclosed in stable mineral containers, they can remain chemically isolated for millions or even billions of years (Roedder, 1981) and can provide data about temperatures, pressures, and compositions of aluminosilicate melts which cannot be obtained
Fluid inclusions Fluid inclusions observed in pegmatites include vapor-rich, two-phase liquid-rich and commonly multiphase crystal-rich brine inclusions (Kamenetsky et al. 2004) containing Na, K, Mg, and Ca chlorides. We note here that the use of nomenclature in this regard can be problematic. Roedder (1984a) has shown that a continuum exists between the two inclusion extremes (melt vs. fluid inclusions). While we recognize this, in this paper we follow convention and use the term “fluid inclusion” for those inclusions that remain fluid at room temperature. However, we note that pegmatite-forming fluids can be preserved as volatile-rich aluminosilicate MI having H2O concentrations >40 wt.%, and as such are almost intermediate between MI and FI.
193
RAINER THOMAS, JAMES D. WEBSTER & PAUL DAVIDSON
others (Rickers et al. 2006, Samson et al. 2003, Veksler et al. 2003).
Sample preparation and microprobe analysis Doubly polished mineral plates, with thicknesses of 200 to 1000 µm depending on the transparency, inclusion dimension, and number of inclusions per unit volume of host, were used for MI and FI study. The first step is careful petrographic microscope study of the samples to determine which types of inclusions are present, and which are important for investigation. For example, it is important to establish which fluid inclusions are primary and which are secondary and to detect compromised inclusions. Petrographic study should also detect all inclusion types and varieties present, as well as any spatial or temporal relations between them. We recommend the use of Köhler illumination (Köhler 1893) to achieve high quality microscopy, because the contrast difference between inclusion glass and quartz hosts is commonly near zero. Abundances of major and minor elements in included glasses are typically determined by electron microprobe analysis (EMPA). However, the analysis of hydrous glasses requires particular care to avoid alkali migration or loss (Morgan & London 2005, Preston et al. 2003). Another problem with this technique is that the backscatter secondary electron (BSE) image contrast of volatile-rich glasses and quartz is often quite small. However, in quartz samples involving MI which are visible in BSE illumination, we have observed: • if the BSE contrast of the inclusion glass is larger than of the quartz host (i.e., the image of the inclusion appears bright against the surrounding quartz), heavy elements (e.g., Rb, Cs) may be present, • if the contrast is similar to quartz, then the light elements H and B may be present in high concentrations, and • if the contrast of the inclusion is great, but lower than that of the quartz host (i.e., the inclusion glass is darker or black) then Li may be present in high concentrations. Other techniques which may be useful in constraining the composition of MI and FI quantitatively, semi-quantitatively or qualitatively include: Synchrotron radiation induced X-ray fluorescence analysis (SXRF), analytical transmission electron microscopy (TEM), X-ray absorption fine structure analysis (XAFS), protoninduced X-ray emission method (PIXE), laser ablation ICPMS, Fourier transform infrared spectroscopy (FTIR), secondary ion mass spectrometry (SIMS) (Webster et al. 1997), and
How well do MIs represent magmatic compositions? Apart from technical issues (see below) the investigator needs to consider the applicability of any data produced, and specifically, how the MI and FI analyses relate to the magmatic phases from which they were trapped. The great differences in wetting ability between H2O-rich aluminosilicate melts and CO2 vapor is important in the formation of MI (Lowenstern 1995) and may introduce the possibility of selective trapping of the more wettable phases. For example, the nearly ubiquitous presence of CO2-rich FI which dominate the inclusion population of some pegmatite quartz crystals, seemingly suggests that quartz crystallized from a CO2-rich fluid although this is quite unlikely (see Watson & Brenan 1987). Since the crystallization of pegmatite minerals is often rapid and because metastable conditions are common, heterogeneous trapping of melts can influence MI and FI in ways that ultimately cause severe difficulties in analyzing them. This is particularly true for entrapment under melt–melt immiscibility conditions. If the inclusion size is similar to, or larger than, the particle size of the emulsion of magmatic phases, co-trapping of multiple phases, i.e., “composite inclusions” (Davidson & Kamenetsky 2001, Davidson et al. 2006), may occur. However, inclusions with sizes that only allow trapping of an emulsion having particles larger than the inclusion are only capable of trapping a pure end member, and inclusions larger than the average particle size could theoretically still trap an approximately pure end member, particularly if the emulsion is inhomogeneous on scales >200 µm. Davidson et al. (2006) demonstrated that fluid bubbles in magmatic emulsions at Río Blanco, Chile, range from 1–2 µm up to at least the size of the largest magmatic inclusions (e.g., >200 µm). Moreover, such emulsions mature by coalescence of bubbles into progressively larger bubbles (due to their great density and viscosity contrast) under gravity and/or compaction. Any MI study, therefore, needs to identify composite inclusions, and fortunately there is a practical test that is currently available. If a composite inclusion traps multiple immiscible phases, heating the inclusion back to the trapping temperature will not homogenize the inclusion, since it was never homogeneous in the first place. 194
UNDERSTANDING PEGMATITE FORMATION: THE MELT AND FLUID INCLUSION APPROACH
ratios of daughter crystals, liquid, and vapor. These inclusions were large (up to 200 µm in diameter) and geometrically well shaped. The errors in volume estimation by normal optical observation are in the order of ±20% or better (Naumov 1979, Diamond 2001). This error was reduced to about ±10% by three-dimensional survey of the inclusions using a modified spindle stage equipped with a micrometer ocular (Anderson & Bodnar 1993). The element concentrations in the MI were calculated from the estimated volumes, compositions, and densities (given in Anthony et al. 1990, 2003) of each mineral phase. However, this method does not account for problems resulting from incorrect estimation of the quantity of SiO2 that crystallizes on inclusion–quartz host contact walls after entrapment. For instance, compositions of type-B inclusions trapped along a solvus crest are shown in Table 9-2, where column (1) represents the calculated bulk composition, and column (2) shows the bulk composition with an added 20 wt.% of SiO2. Column (2) values were computed this way in order to make the results comparable to previous compositions obtained by rehomogenizing the MI and analyzing them by EMPA (column 3) – see Rickers et al. (2006). This comparison shows that it is important to account for post-entrapment host phase crystallization. In a practical sense, the estimation of the amount of quartz crystallized on inclusion–host contact walls will vary between samples, and it is best constrained by observing the increase in inclusion volume during heating experiments as was done in this case.
It is possible that heating to much greater temperatures may homogenize such inclusions, but if an inclusion homogenizes at a geologically reasonable temperature there is a good chance it is not composite. This is consistent with practices of typical homogenization experiments, because they end at a geologically reasonable upper temperature for that sample suite and any inclusions that have not homogenized are abandoned. Some of the latter inclusions are probably compromised by leakage or other problems, but composite inclusions would also tend to be ignored as well. Raman spectroscopy and estimation of bulk compositions of melts and inclusions The identification of mineral phases within FI and MI is useful for estimating the compositions of primary pegmatite-forming melts and investigating their subsequent cooling history. Element speciation in inclusions is also of particular interest, and CMRS is an indispensable tool for this. Table 9-1 lists daughter minerals that were found in MI and FI hosted by quartz of pegmatites of the Ehrenfriedersdorf region of southern Germany. In marked contrast to porphyry copper deposits and related magmatic systems in which halite and sylvite form the principal daughter phases (Roedder 1984a), other daughter mineral phases are usually dominant in pegmatites and in miarolitic quartz from granites (Kamenetsky et al. 2002, 2004). For example, miarolitic quartz of the Eibenstock granite of southern Germany contains, in addition to rare silicate MI, many multiphase FI with halite, sylvite, and Ca- and Mg-chlorides and borates as principal daughter minerals. One can estimate bulk compositions of inclusions by carefully determining the volume ratios of daughter crystals, liquid, and vapor, and by computing the composition as the weighted sum of the individual phase compositions. Non-halide daughter crystal compositions are readily identified by Raman spectroscopy (Table 9-1), and the volumes of crystals, liquid, and vapor are estimated by optical microscopy. Although more precise analytical tools are available (e.g., SXRF, PIXE and laser ablation–ICP–MS) this simple technique provides inexpensive, non-destructive first-pass estimates that are applied before these other techniques. As an example, the compositions of some non-rehomogenized H2O-rich MI (herein referred to as type-B MI, Fig. 9-2, and described in detail below) were estimated by determining volume
Melt inclusion homogenization experiments As noted, the method of determining compositions of MI that best represent the bulk melt composition at the time of entrapment is to reheat and homogenize the MI and quench them to homogeneous glass for analysis. This is important because many natural, unheated MIs are inhomogeneous on scales smaller than the beamsize commonly employed for most micro-beam analytical techniques. However, because of the high H2O concentrations of many pegmatite MI, and their tendency to decrepitate during heating, the homogenization process cannot be conducted safely or successfully at atmospheric pressure on a microscope-mounted heating stage. During heating, the pressure inside the MI increases significantly. We offer a cautionary note, here, regarding the characteristics and consequences of metastable homogenization. If the homogenization procedure 195
RAINER THOMAS, JAMES D. WEBSTER & PAUL DAVIDSON
TABLE 9-1. DAUGHTER MINERAL PHASES IN DIFFERENT INCLUSION TYPES IN PEGMATITE QUARTZ FROM EHRENFRIEDERSDORF. Mineral
Orthoclase KAlSi3O8 Albite NaAlSi3O8 Nepheline NaAlSiO4 Muscovite KAl2(Si3Al)O10(OH,F)2 Diopside CaMgSi2O6 Topaz Al2SiO4(F,OH)2 Boromuscovite KAl2(Si3B)O10(OH,F)2 Berlinite AlPO4 Fluorapatite Ca5(PO4)3F Amblygonite LiAl(PO4)(F,OH) Lacroixite NaAl(PO4)F Herderite CaBe(PO4)F Triplite (Mn,Fe,Mg,Ca)2(PO4)(F,OH) Halite NaCl Sylvite KCl Li2CO3 Calcite CaCO3 Cryolite Na3AlF6 Elpasolite K2NaAlF6 Sassolite H3BO3 Ca-Mg-hexaborates MgB6O10, CaB6O10 with 4 - 7.5 H2O Alkali tetrafluoroborate MBF41) Lithium metaborate LiBO2 Cassiterite SnO2 Sphalerite ZnS Huebnerite Mn(WO4) W-Nb-tantalites (wolframoixiolite) (Fe,Mn,Nb)(Nb,W,Ta)O4 Hematite Fe2O3
Characteristic Raman bands (cm-1) 512.8 475.0, 502.8 399, 415, 920, 967 192.0, 260.1, 408.9, 636.2, 700.4 (triplet) 391, 666, 1012 238.0, 266.0, 285.0 (triplet) 555, 717, 3614 1110.9 962.0 601.0, 644.0, 1011.2 608.6, 622.7, 1000.8 584, 595, 983, 1005 980.2 1090.8 1085.0 552 325.7, 558.5 500, 880 634, 638, 641, 852, 855, 861, 953, 964 772 713, 1459 632.8 348.8 881 395, 526, 868 225, 245, 291
Melt Inclusion Type- TypeA B X X X
X
X X
X
X
X X
X
X X X
X X X
X X X X
X X X X X
X X X
X
High- Medium- LowT T T
X X X
X X
Fluid Inclusion
X
X
X
X
X
X X
X
X
X X
X X X X X
X
Analyzed by electron microprobe, Raman spectroscopy and optical microscopy, after Rickers et al. (2006). 1) M = Na, K, Rb, Cs The main mineral phases (quartz, orthoclase, albite, topaz, lacroixite) in the type-A and type-B inclusions are identical, only the phase proportions are different.
196
UNDERSTANDING PEGMATITE FORMATION: THE MELT AND FLUID INCLUSION APPROACH
TABLE 9-2. CHEMICAL COMPOSITIONS OF TYPE-B MELT INCLUSIONS
SiO2 B2O3 Al2O3 FeO(1) BeO CaO Li2O Na2O K2O Rb2O(1) Cs2O(1) F Cl P2O5 SnO2 H2O Sum Sum(2) ASI
1 (wt%) 16.9 4.7 16.6 0.2 0.3 0.8 0.9 6.0 6.1 0.5 0.2 6.2 5.2 6.9 0.1 32.0 103.6 99.8 0.77
2 (wt%) 30.8 4.0 13.8 0.1 0.2 0.7 0.7 5.0 5.1 0.4 0.1 5.2 4.3 5.7 0.1 27.0 103.2 100.0 0.78 + 0.24(3)
3 (wt%) 24.3 4.1 11.1 3.7 0.4 0.8 0.4 5.9 7.5 0.2 0.1 5.8 6.7 6.7 0.2 26.0 103.9 99.9 0.53
Daughter mineral phases orthoclase, topaz meyerhofferite, H3BO3 orthoclase, topaz, amblygonite, lacroixite, elpasolite hematite herderite meyerhofferite, herderite amblygonite amblygonite, lacroixite, halite, elpasolite, (+ solution) orthoclase, (+ solution) topaz, lacroixite, amblygonite, elpasolite, herderite halite, sylvite, (+ solution) amblygonite, lacroixite, herderite cassiterite
Analyses of type-B melt inclusions, trapped near the solvus crest at 650 ± 50°C, estimated from the daughter crystal and liquid volumes in very large inclusions (up to 200 µm in diameter) and measured by SRXRF analyses. The mean standard deviation of complete determinations is 27% (SiO2 + 5%). Column 1 shows the estimated composition (from mineral compositions), column 2 the same composition recalculated to equilibrium with host quartz (with the addition of about 20 wt.% SiO2, see text). Column 3 is from Thomas et al. (2003) measured by electron microprobe from rehomogenized inclusions (from Rickers et al. 2006). (1) determined by SXRF; (2) sum corrected for F and Cl; (3) Standard deviation of the ASI (aluminum saturation index) results from the mean standard deviation of the element determinations (+ 26.7%)
does not achieve complete equilibrium, a false or metastable homogenization may occur. In this situation, the result is a melt that is perfectly stable at magmatic conditions, but the melt quenches to a glass that is only metastable at room temperature. This is a non-trivial problem, because it complicates subsequent analyses of resultant glasses. For example, during homogenization experiments that were quenched rapidly we have observed the presence of just such a metastable glass (immediately after the quench) that separated into fluid, vapor and completely different glass after only a few hours. During cooling, these processes are apparently accompanied by a large volume change resulting from the increase of the melt and/or glass density and by a change in the speciation of volatiles. Based on experimental work involving rapid quench experiments and Raman spectroscopic studies (Thomas unpublished data) as well as on
theoretical treatments and FTIR spectroscopic studies (Burnham 1979, Stolper 1982) it appears that OH– converts to molecular H2O and CO32– to CO2 during cooling. Conversely, these reactions go in the reverse direction, although slowly, during heating. Therefore, heating on a microscope heating stage can increase the internal pressure to very high values that trigger decrepitation. It is worthy of note that if these processes are not recognized and accounted for it is likely that the analyst would mistakenly interpret the phase assemblage in one of the inclusions to indicate heterogeneous trapping of multiple phases. Cold-seal pressure vessels and hydrothermal rapid-quench homogenization One method of homogenizing MI at elevated pressure involves the use of standard coldseal pressure vessels and small polished mineral chips of 1 mm thickness that were loaded into a
197
RAINER THOMAS, JAMES D. WEBSTER & PAUL DAVIDSON
punctured Au capsule (30 mm long, 5 mm in diameter) and placed horizontally into the vessel. The vessel was pressurized with CO2 to 1 or 3 kbar, and inserted into the preheated furnace and heated isobarically. Depending on the experimental objective, the run temperatures were between 850°– 500°C and run durations of 20 to 50 hours. Experience shows that a temperature range from 750°–500° is adequate for the study of MI in pegmatite minerals. Subsequently, the vessel was quickly removed from the furnace and quenched isobarically with compressed air. After quenching, the samples were removed, re-polished to a thickness between 100 and 500 µm, and then used for microscopic examination and chemical and spectroscopic analyses. A second useful technique involves hydrothermal rapid quench homogenization experiments (Thomas et al. 2000, 2005, Veksler & Thomas 2002). Here, doubly polished plates of quartz were placed into an Au capsule (5 x 30 mm) with a fixed amount of pure H2O. The MI were remelted in their hosts at temperatures between 500°– 850°C at 1–3 kbar, typically with a run time of 24 hours. After each run, the samples were quenched isobarically. This rapid quench technique was used to homogenize H2O-rich, type-B MI to a homogeneous glass for Raman spectroscopic study. With this technique it was possible to homogenize H2Orich melts, containing up to ~36 wt.% H2O, to a homogeneous, although metastable, glass, which could be studied with Raman spectroscopy immediately after quenching. However, it was not possible to obtain stable homogeneous glasses from such H2O-rich melts with this technique for electron microprobe analysis. These methods do not permit direct observation of phase changes during homogenization of MI and FI. The tendency of H2O-rich inclusions to decrepitate means that one can observe the results of the heating process only with use of a hydrothermal pressure cell that incorporates an optically transparent window. For example, the hydrothermal diamond anvil cell (HDAC) is laborious to use and expensive, but at this time it is the only method with which one can observe phase changes during homogenization experiments. Using HDAC experiments, it was previously shown that a silicate melt is completely soluble in H2O-rich solutions, demonstrating that there is a continuous transition between the two extremes (Roedder 1984a, p. 397).
Techniques for determining H2O concentrations of melt inclusion glasses The determination of accurate H2O concentrations of MI glasses is of critical importance, and this can be particularly difficult when working with samples from pegmatites because many of their MI contain very high H2O concentrations (e.g., up to 30 wt.%) and because accurate analysis of high concentrations can be problematic with some methodologies. Most of the more functional analytical methods, such as EMPA, FTIR, or SIMS, have not been used previously to measure such high H2O concentrations in glasses. As an alternative, CMRS was developed (Thomas 2000), improved (Thomas 2002), and further developed (Thomas et al. 2006a) for the determination of H2O concentrations in MI unexposed on polished mineral surfaces (i.e., those completely contained within the host). This method involves direct comparison of results from samples to those of a reference glass, and it is necessary to investigate the sample and standard glass under the same conditions. It is independent of the composition of the glasses to be investigated, so it is applicable to glasses of widely differing compositions. The strength of the method is illustrated in Thomas et al. (2006a), which showed very good correlation (r2 = 0.999) of the measured integral signal intensity versus the H2O content of synthetic albite, granite, and basalt glasses determined by Karl Fischer titration. Figure 9-3 shows two Raman spectra in the high frequency region taken from the same H2O-rich MI in pegmatite quartz from Ehrenfriedersdorf, Germany, immediately after it was heated and quenched to a glass (upper part of figure) and taken several hours later (lower part of figure) after decomposition into another glass plus liquid and vapor phases. The H2O contents of the glasses were 36 wt.% before decomposition and 10 wt.% afterwards. The spectra also show that the H2O–OH stretching band changes from a rather symmetrical one to a more asymmetrical form before and after decomposition, respectively (Thomas et al. 2006a). RECENT KEY OBSERVATIONS FROM REHOMOGENIZED INCLUSIONS The early inclusion work of one of the authors (RT) involved searching for inclusions in graphic orthoclase–quartz intergrowths from the Königshain pegmatites, SE Germany (Woitschach 1881, Fersmann 1928). Initially, the smoky quartz from these pegmatites appeared to contain only FI 198
UNDERSTANDING PEGMATITE FORMATION: THE MELT AND FLUID INCLUSION APPROACH
1400
volatile-rich silicate MI. Using the microvolumetric/microthermometric method of Naumov (1979), which was later improved by Reyf (1990), our preliminary estimates of H2O concentrations in quartz-hosted MI of small miaroles in the Ehrenfriedersdorf samples implied about 20 wt.% in the glasses (unpublished data). And, once volatilerich MI were recognized in the Ehrenfriedersdorf samples, subsequent searches of the graphic orthoclase–quartz samples from Königshain that were studied previously revealed similar, very H2Orich MI (Figs. 9-4 and 9-5, Table 9-3). Our samples of the Ehrenfriedersdorf pegmatites that are genetically related to the tungsten–tin granites there contain two different types of MI: a silicate-rich, H2O-poor (type-A)
3453 cm-1
Intensity (a.u.)
1200 1000
3297 cm-1
800
3560 cm-1
600 400 200 0 2800 3000 3200 3400 3600 3800 4000
Wavenumber (cm-1) 250
3458 cm-1 3558 cm-1
Intensity (a.u.)
200 150
3180 cm-1
100 50 0 2800 3000 3200 3400 3600 3800 4000
Wavenumber (cm-1)
FIG. 9-4. Type-B melt inclusions in graphic pegmatite quartz from Königshain, Germany, after rehomogenization at 650°C and 1 kbar in 20 hours.
FIG. 9-3. Raman spectra at the high frequency region from a water-rich melt inclusion (type B in Fig. 9-2) in pegmatite quartz from Ehrenfriedersdorf, homogenized at 650°C and 1 kbar, immediately after quenching using the rapid quench technique (upper part). Some hours later the primary homogeneous glass broke down to form glass, liquid, and vapor phases spontaneously. According to Chabiron et al. (2004) the main Gaussian components correspond to molecular water, only the weak band at about 3560 cm–1 corresponds to hydroxyl groups. See text for discussion.
with low salinities and low homogenization temperature and no MI, suggesting (quite improbably) the pegmatites formed from large quantities of dilute solutions. Subsequent, careful examination of analogous sample material from pegmatites of the Variscan Ehrenfriedersdorf complex, Germany, showed the presence of
FIG. 9-5. A large type-A melt inclusion in graphic pegmatite quartz from Königshain, exposed by polishing, with a close-up b) showing details of the EMPA spots. The sample with the melt inclusion was re-homogenized at 700°C and 1 kbar in 20 hours. See text for discussion.
199
RAINER THOMAS, JAMES D. WEBSTER & PAUL DAVIDSON
TABLE 9-3. ELECTRON MICROPROBE ANALYSIS OF A LARGE MELT INCLUSION IN GRAPHIC QUARTZ FROM A MIAROLITIC PEGMATITE IN THE KÖNIGSHAIN GRANITE, EASTERN SAXONY. SiO2 TiO2 Al2O3 FeO MnO MgO CaO Na2O K 2O Rb2O Cs2O F Cl P2O5 Sum Sum1) H2O-calculated H2O-Raman2) ASI n
TABLE 9-4. REPRESENTATIVE COMPOSITIONS OF TYPE-A AND TYPE-B MELT INCLUSIONS IN QUARTZ FROM THE MALKHAN PEGMATITE FIELD/CENTRAL TRANSBAIKAL REGION, RUSSIA.
Glass 64.4 + 1.8 0.02 12.9 + 0.3 1.6 + 0.2 0.08 + 0.02 0.01 0.19 + 0.01 3.94 + 0.23 4.12 + 0.20 0.25 + 0.01 0.12 + 0.01 2.28 + 0.05 0.01 0.01 88.33 87.37 12.63 10.1 + 0.25 1.12 + 0.05 6
SiO2 TiO2 B2O3 Al2O3 MgO CaO Li2O (estimated) Na2O K2O Rb2O Cs2O F Cl P2O5 H2O ASI ASI (+Li) n
Type-A MIs 72.0 0.03 2.9 + 0.9 12.6 d.l. 0.11 1.5 2.95 3.05 0.21 2.95 1.61 d.l. 0.04 11.7 + 2.5 1.32 0.86 18
Type-B FIs 66.3 0.01 7.0 + 0.8 10.4 d.l. 0.05 1.3 4.34 4.00 0.14 3.49 2.89 0.01 0.02 25 + 6.0 0.81 0.60 30
Components other than H2O normalized to anhydrous to emphasize the composition of the silicate component of these inclusions, and the H2O concentration (± 1 standard deviation) is also provided. Analyses by EMPA and CMRS after re-homogenization at 700°C and 1kbar, d.l. = below detection limit.
ASI: Aluminum saturation index; n, Number of analyses; 1) Fluorine and chlorine-corrected sum; 2) Water determination after EMPA. The glass was initially very water-rich (18.4 + 0.8 wt.% H2O), but not stable during measurement; reducing to 13.7 + 0.6 wt.% during analysis due to irreversible damage by the electron beam (see Fig. 3b). The water concentration after the EMPA reduced further with time due to structural changes in the exposed glass (see Thomas et al. 2006a).
Central Transbaikal region, Russia. Moreover, we have also found similar MI in graphic quartz in numerous, graphic granite-bearing pegmatite samples including Königshain and Hühnerkobel, Germany, Striegau, Poland, Bornholm Island, Denmark, Tanco, Canada, Borborema Pegmatite Province, NE Brazil, Mozambique Pegmatite Province (e.g., Muiane, Naipa, and Nuaparra), and the Amis Complex, Namibia. This research follows that of Bakumenko et al. (1979) who described similar MI in the famous chamber pegmatites at Volyn, in the Ukraine. These observations broadly support the presence of MI in graphic quartz as a general characteristic of many pegmatites.
inclusion and a silicate-poor, H2O-rich (type-B) inclusion (Thomas et al. 2000). They represent coexisting fluid fractions that define a solvus (Fig. 9-2) in an XH2O–T pseudobinary system (Thomas et al. 2000, 2003). Interestingly, if characterized on a molar basis the H2O-dominated type-B MI could be described as a variety of silicate-rich FI. Both inclusion types are found in the same growth zones of quartz crystals, demonstrating their syngenetic origin. We also observed contrasting chemical and physical properties in the two coexisting fluid fractions. With cooling and chemical evolution one of the fractions (type-A melts) shows increasing peraluminosity and the other (type-B melts) display increasing peralkalinity. Another important example of this (Table 9-4) includes MI in quartz of a tourmaline pegmatite in the Malkhan field of the
DISCUSSION Viscosity Viscosity is a fundamental property that influences the dynamic behavior of silicate melts (e.g., melt segregation, magma mixing, crystal fractionation, fluid exsolution, and the ascent rate of magma). The viscosity of silicate melts is significantly affected by bulk composition, volatile 200
UNDERSTANDING PEGMATITE FORMATION: THE MELT AND FLUID INCLUSION APPROACH
contents, as well as temperature and pressure (Dingwell 1987, Thomas 1994a). The presence of volatiles generally reduces the viscosity of aluminosilicate melts (Dingwell 1987, Mysen 1988). In addition, solidification of volatile-enriched melts begins at comparatively low temperatures since increased activities of highly volatile constituents also reduce crystallization temperatures. Volatile constituents accumulate in residual melts as crystallization progresses (e.g., Roedder 1984b). In fact, volatiles may attain increasingly high abundances and finally determine the character of silicate melts, and this has extremely important consequences. For instance, although viscosity typically increases with reduced temperature, the viscosity of volatile-rich melts does not always increase with reduced temperature, as otherwise occurs in volatile-poor melts, since the presence of volatile components may overcome the effect of temperature. The viscosities of some volatile-charged melts may be extremely low. Experimental observations demonstrate that silicate melts become completely miscible with aqueous fluids at upper mantle conditions (Bureau & Keppler 1999) and at upper crustal conditions (Sowerby & Keppler 2002, Thomas et al. 2000, 2003), thus the viscosities of such liquids should lie somewhere between that of pure H2O (10–4 Pa.s) and those of hydrous silicate melts (102 to 106 Pa.s, Audétat & Keppler 2004). Conservatively, we (Thomas et al. 1996, 2003, 2005) have estimated the viscosity of topaz–albite granitic liquids, like those emplaced at Zinnwald, Germany, to be about 1 Pa.s at 700°C. In this regard, Audétat & Keppler (2004) demonstrated that the viscosity of aqueous fluids increases only slowly with increasing silicate content. For example, the viscosity of a liquid with 20 wt.% silicate at 800°C is equal to that of pure H2O at room temperature (0.001 Pa.s), and that of a fluid with 50 wt.% silicate components (35 mol%) is comparable to that of olive oil at the same conditions. Such extremely low viscosities have a major effect on the behavior of pegmatite-formation and associated mineralization. For example, decreasing viscosity that accompanies cooling can cause the crystal nucleation number to decrease dramatically during cooling which has obvious consequences for pegmatite formation (Niggli 1937). In addition, because of the reciprocal relationship between viscosity and diffusivity (Stokes–Einstein equation), i.e., that the diffusivity
increases as the viscosity drops, it follows that ionic mobility increases with decreasing viscosity. Components, which are necessary for crystal growth, are supplied to growing crystals more rapidly. Moreover, viscosities less than about 102 Pa.s are necessary for a melt to move with respect to its crystalline fraction (McKenzie 1985). Audétat & Keppler (2004) have suggested that the low viscosity of such silicate-rich liquids, together with their favorable wetting properties, allow them to migrate even at low volume fractions. Thus, such liquids may move over long distances along channels or grain boundaries. Because of the very low viscosity and favorable wetting angle, extremely H2O- and F-rich or silicate-rich melts are excellent media for the extraction and concentration of trace elements. These components are concentrated at grain boundaries, because they are not taken up into the crystal lattice of quartz and feldspar – the main minerals of granites and pegmatites. The high mobility and dissolution capacity of such “fluids” causes the percolating fluid medium to collect trace elements that are available as they pass through crystallizing granitic magma. These processes enhance element transport, permitting extreme enrichment of many trace elements, even to ore-forming levels. Peralkaline pegmatite-forming melts: indirect indications of peralkaline melt and fluid fractions According to our observations and research on inclusions, at least three different mechanisms of generating alkaline melt compositions are possible: (i) crystallization and fractionation, for example, of topaz, (ii) liquid immiscibility and (iii) formation of carbonates at high CO2 confining pressures. This is consistent with prior work. According to Brotzen (1959), for example, excess alkalis are not derived from normal granitic magmas by crystallization differentiation, because all mica would be converted into feldspar and alkali amphiboles. Recent observations on simple pegmatites that are composed principally of quartz and feldspar show that many are formed from H2O and CO2-rich, peralkaline aluminosilicate melts and fluids (Thomas et al. 2006b). These liquids differ from late-stage peraluminous melt fractions, which are often enriched in F, B, and P. In Table 9-5 is listed the composition of large MI in pegmatite quartz from Ehrenfriedersdorf. This quartz is characterized by appearance of nepheline crystals beside the MI. 201
RAINER THOMAS, JAMES D. WEBSTER & PAUL DAVIDSON
TABLE 9-5. REPRESENTATIVE COMPOSITION OF LARGE PERALKALINE MELT INCLUSIONS (UP TO 60 µM IN DIAMETER) IN PEGMATITE QUARTZ FROM EHRENFRIEDERSDORF, GERMANY. Glass 67.0 + 2.6 0.02 7.0 + 0.4 0.43 + 0.10 0.08 d.l. 0.19 + 0.02 4.03 + 0.44 4.58 + 0.20 0.38 + 0.05 0.04 + 0.01 d.l. 0.21 + 0.05 d.l. 99.96 16.0 0.58 + 0.05 10
30 µm
H2O 3
ICE
-5°
9 12 -8° 15 -10°
Cl . 2H 2O
27
30 NaCl
ASI, Aluminum saturation index; d.l., detection limit; n, Number of analyses, 1) Water determination after EMPA (Thomas et al. 2003). Analyzed by electron microprobe (rehomogenized at 650°C and 1 kbar).
6 9
NaHCO3 12 15
68.8% H2O 18 10.3% NaCl 20.9% NaHCO3 21
18 21 24
3 -3°
6
Na
SiO2 TiO2 Al2O3 FeO MnO MgO CaO Na2O K2O Rb2O Cs2O F Cl P2O5 Sum H2O-calculated1) ASI n
NaHCO3
60°
24 27 30 NaHCO3
FIG. 9-6. Nahcolite-rich fluid inclusion in pegmatite quartz from Pechtelsgrün, Germany, with a simplified ternary NaHCO3–NaCl–H2O diagram (according to Borisenko 1977). The position of the fluid inclusion shown in the photomicrograph is marked with an arrow in the ternary diagram. See text for discussion.
Given that the high reactivity of peralkaline melts or liquids destroys most evidence of preexisting peralkalinity, as noted previously, it is important to find surviving evidence of such a geochemical evolution trend. Reliable evidence is found in the form of inclusions preserved from pos9-entrapment alteration processes. We have observed, for example, nahcolite (NaHCO3)-rich FI, but these inclusions do not necessarily indicate the presence of nahcolite in residual, pegmatite-forming melts (Fig. 9-6). According to Mustart (1972), sodium disilicate is unstable in the presence of CO2 at room temperature:
immiscible liquids in the NaAlSi3O8–Na2CO3–H2O system, given suitable conditions of pressure, temperature, and bulk composition. These phases included a silicate liquid containing dissolved Na2CO3, CO2, and H2O; a carbonate-dominated liquid consisting of Na2CO3 with dissolved H2O; and a vapor phase composed of CO2, H2O, and a very small proportion of dissolved silicates. Careful Raman spectroscopic studies of FI in quartz from different pegmatites show that nahcolite- and other bicarbonate- (KHCO3) and carbonate–(Li2CO3 and Na2CO3) rich FI are not rare. In some instances, the concentrations of bicarbonates and carbonates can achieve relatively high values, e.g., 20–40 wt.%. Such inclusions are observed in quartz, feldspar, or beryl of many compositionally simple pegmatites and in some more evolved pegmatites (e.g., Zinnwald, Erzgebirge, Pechtelsgrün/Vogtland, Zwiesel in Eastern Bavaria, Germany; Precambrian pegmatites from the Rønne granite, Denmark; Orlovka, Transbaikalia; Tanco pegmatite, Canada; Naipa and Muiane, Mozambique – see Thomas et al. 2006b).
Na2Si2O5 + CO2 = Na2CO3 + 2 SiO2 Na2Si2O5 + 2 CO2 + H2O = 2 NaHCO3 + 2 SiO2 and these reactions demonstrate that, if CO2 is present, we are not likely to find sodium disilicate in FI at room temperature. Instead, the FI should be Na2CO3- and/or nahcolite-rich, and both are a strong indication of the presence of sodium disilicate at magmatic temperatures. These observations are consistent with the experimental results of Koster van Groos & Wyllie (1968) who demonstrated the coexistence of three analogous 202
UNDERSTANDING PEGMATITE FORMATION: THE MELT AND FLUID INCLUSION APPROACH
Such high concentrations of alkali carbonates and bicarbonates cannot be explained by normal hydrothermal activity, since this commonly produces very low carbonate and bicarbonate concentration. We suggest that, originally, the phases in the inclusions were alkali disilicates (or their derivatives) which formed the carbonates and/or bicarbonates while cooling. Silica was deposited on the inclusion wall while sodium remained in the liquid during cooling. We have also found muscovite and euhedral crystals of cristobalite in such inclusions, which may indicate re-crystallization of the inclusion contents. Indications of CO2-rich peralkaline melt fractions Thomas & Schmidt (1986) and Schmidt & Thomas (1990) have observed quartz from Nb–Ta– REE–U pegmatites from the Mozambique province (e.g., Muiane, Naipa and Nuaparra) that also contain spodumene, cookeite, lepidolite, morganite, and tourmaline. The quartz of the quartz core contains extremely CO2-rich MI, and these MI were rehomogenized at 650°C and 3 kbar. After microscopic observation and analysis by CMRS, two different types of MI were found: (i) H2O-rich and CO2-poor type-A and (ii) H2O-poor and CO2rich type-B inclusions (see Fig. 9-7). At room temperature, the type-B inclusions are characterized by four phases with three phase boundaries: CO2– vapor, CO2–liquid, H2O-rich alkali carbonate solution, and alkali carbonate-rich aluminosilicate glass. From Raman studies, the glasses of type-A and type-B inclusions contain 32 and 12.7 wt.% H2O, respectively. But, the H2O contents of the glasses do not characterize the total H2O in the inclusions. For example, the bulk H2O of the typeB inclusions (i.e., H2O in the glass + the H2O in the aqueous solution) is about 33 wt.%. The two different types of inclusions were trapped near the crest of a pseudoternary solvus with a critical temperature close to 650°C at 3 kbar. A conservative estimation of the CO2 concentration is about 10 wt.%. These data, however, are preliminary estimates. A similar inclusion assemblage was found in quartz of the Tanco pegmatite in Canada (Fig. 9-8), and this assemblage is understandable only because the aluminosilicate melt was, in part, strongly peralkaline (see Thomas et al. 2006b). Other pegmatitic systems provide evidence of alkaline carbonic melts. Recently, Thomas et al. (2006b) observed CO2-rich, type-B inclusions (Fig. 9-9) in core quartz of the Hühnerkobel pegmatite
FIG. 9-7. Complex melt inclusions in pegmatite quartz from Naipa, Mozambique, containing carbonate-rich glass (G), carbonate-rich H2O solution (L), liquid CO2 (CO2–L), and CO2 vapor (CO2–V), rehomogenized at 650°C and 3 kbar in 20 hours. See text for discussion.
FIG. 9-8. Melt inclusions in quartz of the Tanco pegmatite, Canada, rehomogenized at 650°C and 1 kbar in 20 hours; a) type-A melt inclusion, b) to d) type-B melt inclusions. See text for discussion.
near Rabenstein (Strunz 1974). After rehomogenization at 700°C and 3 kbar, the inclusions are characterized by four different phases at room temperature: glass (53 vol.%) with 21.9 wt.% H2O, an aqueous liquid phase (22 vol.%), a liquid CO2 phase (25 vol.%), and a CO2–vapor bubble. The bulk H2O content is 36.5 wt.% and the bulk density of such inclusions is 1.34 g.cm-3. From microscopic observations it follows that this inclusion type was formed by liquid immiscibility in a pseudoternary melt–H2O–CO2 system. A similar assemblage of inclusions was described by Sirbescu & Nabelek (2003a and b) from the internally zoned,
203
MAGMATIC PROCESSES AND VOLATILE PHASE GENERATION IN PORPHYRY-TYPE ENVIRONMENTS
FIG. 9-9. Silicate melt + H2O + CO2 containing type-B melt/fluid inclusions in pegmatite quartz from Hühnerkobel near Bodenmais, Bavaria. G – silicate glass, L – H2O rich fluid, CO2–L – liquid CO2, CO2–V – CO2-rich vapor, re-homogenized at 650°C and 3 kbar in 20 hours. See text for discussion. The volume change is produced by melting of SiO2 from the quartz host during rehomogenization. It follows that during cooling in nature significant amounts of SiO2 will deposit on inclusion walls which depletes the silica content of the residual glass. According to Webster (2004) melting of 30 wt.% SiO2 from the walls of the quartz host increases the silica concentration of a MI from 66 to 76 wt.% and dilutes alumina from 17 to 12 wt.%. This means, the exact determination of the real trapping temperature is not a trivial problem.
Li-bearing Tin Mountain pegmatite in the Harney Peak granite–pegmatite system of the Black Hills, South Dakota (USA). From their inclusion work, they concluded the pegmatite was characterized by low crystallization temperatures between 400 and 350°C because of the combined fluxing effects of Li, B, P, H2O, and carbonate anions. Their temperatures might require some correction to higher values, because the influence of SiO2 was not considered. The issue of the role of SiO2 is not trivial, however. Rickers et al. (2006) have demonstrated that the type-B inclusions are undersaturated in SiO2 at room temperature despite the fact that they are trapped in quartz hosts. This may appear to be counterintuitive. However, rehomogenization of type-B inclusions is always accompanied by a volume increase of up to 25%.
Graphic textures in pegmatites The existence of MI in graphic quartz of some pegmatites is crucial for the understanding of their genesis. Furthermore, there are significant differences between the inclusions found in quartz 204
UNDERSTANDING PEGMATITE FORMATION: THE MELT AND FLUID INCLUSION APPROACH
grown in open spaces (i.e., with only FI of lowtemperature characteristics) and those (i.e., MI) in graphic quartz associated with feldspar. The presence of MI in graphic quartz provides unambiguous evidence of the formation of graphic granite under melt-dominated conditions. This fact is not always recognized, however, because graphic texture has also been interpreted as a reaction texture (e.g. Drescher-Kaden 1948). The formation temperatures of graphic textures in pegmatites (~750–700°C) assessed by Fersmann (1928) are probably close to the general crystallization temperatures of granitic pegmatites.
granite pegmatites under melt-dominated conditions, pass through the fluid-dominated pneumatolytic and hydrothermal states (see also Fersmann (1931) table II), and end at room temperature. One critical concept is that pegmatiteforming processes as a whole should not be considered as representing pure equilibrium, because at best some phases may achieve only partial or local equilibrium. “Equilibrium” in pegmatites can be a difficult concept to deal with in detail because it is both time- and scale-dependent. The time dependence means that diffusion is a distinctly finite process in silicate melts (even H2Orich melts). The scale dependence means that if immiscible separation occurs (or the formation of any other kind of strong concentration gradients), the newly formed phases may be in equilibrium within their local region, but not in equilibrium with respect to the whole pegmatite-forming body. Moreover, once these phases form, each undergoes its own evolution, potentially pushing it further from any previous equilibrium position. Thus, the thermodynamics of irreversible processes bear on conceptual modeling of pegmatite-forming processes. We note that single, generalized models fail because of the equifinality problem, i.e. that the same final state or condition of a system may be reached from different initial conditions and in different ways (von Bertalanffy et al. 1977). We consider the accurate determination of the concentrations and speciation of H2O and other volatiles in pegmatite-forming melt as crucial, since they determine key properties such as pressure, viscosity, and diffusivity within such melts. The study of MI and FI provides major constraints in this regard. Traditionally, the study of granite– pegmatite systems by experimental petrology was restricted mainly to systems with low H2O concentrations (i.e., ≤8wt.%) and generally involved use of the haplogranite model system. But, clearly, these conditions are contrary to those that are representative of pegmatite-forming melts as presented herein from MI data. However, there have been important attempts to study pegmatite genesis experimentally using volatile-rich melts with volatiles in addition to H2O. London (2005) provides a summary of current concepts and of his own work. It has often been emphasized that the solutions preserved in FI in pegmatite minerals are very dilute and therefore should not have any great transport capacity for silicate materials. In light of new research (Thomas et al. 2000), however, the
SYNTHESIS It is clear that there is no uniform and uncontested model for the explanation of all phenomena which can be observed in pegmatites worldwide, nor is it our aim to introduce a new model for pegmatite genesis. Rather, we present some new results bearing on a few pegmatitic systems, which should be incorporated into existing and future models. It is clear that the crystallization of anhydrous minerals (principally quartz and feldspar) causes the concentration of H2O in residual melt to increase continuously up to the point at which fluid exsolves and becomes dominant. Subsequently, the crystallization of new minerals under hydrothermal conditions followed by hydrothermal recrystallization of primary minerals grown under meltdominated conditions become dominant. Primary magmatic signatures are often largely destroyed, or at best only traces remain, during late-stage processes, and it is the search for definitive evidence of them that remains decisive for understanding pegmatite genesis. From our inclusion studies we suggest that the following processes are of critical importance in pegmatite formation: 1. formation of extremely volatile-rich aluminosilicate melts 2. multi-stage, melt–melt immiscibility 3. generation of peralkaline, residual melt fractions (sometime preserved as type-B MIs, sometimes as carbonate/bicarbonate-rich FIs), and 4. the transition from melt-dominated to fluiddominated systems and subsequent metasomatic recrystallization. The formation of pegmatites in the shallow crust is characterized by non-equilibrium processes operating in a broad temperature–pressure field. These processes start at about 700°C in the case of 205
RAINER THOMAS, JAMES D. WEBSTER & PAUL DAVIDSON
crystals. These data are inconsistent with models that invoke H2O-poor or very dilute pegmatiteforming media, or purely solid-state metasomatic models. Evidence also suggests that melt–melt immiscibility processes in H2O-saturated felsic melts may be an important factor in pegmatite formation, since they produce various very H2Orich peralkaline melt fractions implicated in the pegmatite-forming processes. These immiscibility processes may also help overcome the problem of low H2O-solubility in haplogranite melts at middle and upper crustal pressures, which have tended to be the basis of much of the experimental work in the field.
transport capacity of “hydrothermal solutions” related to pegmatite-forming processes must be assessed differently than in previous work (e.g,. Jahns & Burnham 1969). The complex interplay of the volatiles H2O, F, and Cl; the semi-volatiles B2O3 and P2O5; and fluxing components such as Be, Li, Rb, and Cs along with SiO2 and Al2O3 in the silicate melt is responsible for elevated H2O solubility in melt, elevated melt solubility in fluid and, perhaps, for apparently unusual immiscibility behaviors. We have also observed MI with up to 40 wt.% H2O, and selective trapping of CO2-rich FI, all of which suggest that dilute aqueous solutions in pegmatite inclusions are not generally representative of pegmatite-forming media. In this regard, it is also significant that H2O solubility can be quite significant in peralkaline melts. The main information that can be derived from MI and FI in minerals of pegmatites is the composition of mineral-forming media and, in particular, the concentration of H2O and other volatiles at magmatic temperatures and pressures. Evidence from these inclusions suggests that pegmatite-forming melts or fluids have very low viscosities, typically very high H2O concentrations, very high diffusivities, and also high element mobilities. The process of melt–melt immiscibility is very important for the formation of contrasting melts and fluids with strongly different chemical and physical characters.
ACKNOWLEDGEMENTS One author (RT) thanks Edwin Roedder for his helpful interest (starting at 1975) in the melt inclusion work on granites and pegmatites up to now. Thanks go to Andrzej Kozlowski who kindly reviewed an early version of the manuscript. The paper has benefited from discussion with Ilya Veksler, Wilhelm Heinrich, and discussion and analytical work of Karen Rickers. We thank Felix Reyf, Dima Kamenetsky and Thomas Pettke for careful and insightful reviews that significantly improved the manuscript. The study was supported by the Deutsche Forschungsgemeinschaft (DFG) through the grant Th 489 to R. Thomas, and was in part conducted within the framework of the DFG Priority Program SPP 1055 “Bildung, Transport und Differenzierung von Silikatschmelzen”.
CONCLUSIONS Despite more than a century of study there is still no universally accepted model for the formation of pegmatites. Herein, we have given both a methodology and evidence which may provide a basis for further advances. Melt and fluid inclusion studies are important in this field since they can give a direct estimate of the composition of pegmatite-forming melts, some constraints on the pressure and temperature conditions during pegmatite formation, and may provide clues to the evolution of silicate magmas into pegmatiteforming melts, and the evolution of these melts during the processes of pegmatite formation. Without consensus on these most basic factors there can be no progress in this field. The melt and fluid inclusion evidence presented herein shows that pegmatite-forming melts are very H2O-rich, low-viscosity, highdiffusivity, alkali-rich aluminosilicate melts, that provide excellent transport media for silica, alkalis, and trace elements, and growth media for large
REFERENCES ANDERSON, A.J. & BODNAR, R.J. (1993): An adaptation of the spindle stage for geometric analysis of fluid inclusions. Am. Mineral. 78, 657–664. ANTHONY, J.W., BIDEAUX, R.A., BLADH, K.W. & NICHOLS, M.C. (1990, 2003): Handbook of Mineralogy. Mineral Data Publishing Tucson, Arizona, Vol. 1 to 5 AUDÉTAT, A. & KEPPLER, H. (2004): Viscosity of fluids in subduction zones. Science 303, 513–516. BAKUMENKO, I.T., KOSUKHIN, O.N., PAVLISHIN, V.I. & CHUPIN, V.P. (1979): On the magmatic stage in the formation of chamber pegmatites from Volyn. Dokl. Akad. Sci. SSR 248, 1194– 1197. BARTH, T. (1928): Zur genesis der pegmatite im Urgebirge. Chem. Erd. Band IV, 95–136.
206
UNDERSTANDING PEGMATITE FORMATION: THE MELT AND FLUID INCLUSION APPROACH
BARTH, T.F.W. (1962): Theoretical Petrology. 2nd edition, John Wiley & Sons, New York, 416 p.
magmatic quartz. Econ. Geol. 96, 1921–1929. DAVIDSON, P., KAMENETSKY, V., COOKE, D. R., FRIKKEN, P., HOLLINGS, P., RYAN, C., VAN ACHTERBERGH, E., MERNAGH, T., SKARMETA, J., SERRANO, L., & VARGAS, R. (2006): Magmatic precursors of hydrothermal fluids at the Río Blanco Cu–Mo deposit, Chile: links to silicate magmas and metal transport. Econ. Geol. 100, 963–978.
BEUS, A. (1983): On the possible mechanism of formation of euhedral crystals in metasomatic processes. Bull. Minéral. 106, 411–415. BJØRLYKKE, H. (1935): The mineral paragenesis and classification of the granite pegmatites of Iveland, Setesdal, Southern Norway. Norsk geologisk tidsskrift B, XIV, 211–231.
DIAMOND L.W. (2001): Review of the systematics of CO2–H2O fluid inclusions. Lithos 55, 69–99.
BODNAR, R.J. & STUDENT, J.J. (2006): Melt inclusions in plutonic rocks: petrography and microthermometry. In Melt Inclusions in Plutonic Rocks (J.D. Webster, ed.) Min. Assoc. Can. Short Course 36, 1-25.
DINGWELL, D.B. (1987): Melt viscosities in the system NaAlSi3O8–H2O–F2O-1. Magmatic Processes: Physicochemical Principles. Geochem. Soc. Spec. Pub. 1, 423–431.
BORISENKO, A.S. (1977): Cryometric technique applied to studies of the saline composition of solution in gaseous fluid inclusions in minerals. Geologija i geofizika, AN SSSR, SO, 8, 16–27 (in Russ.).
DRESCHER-KADEN, F.K. (1948): Die feldspat– quarz-reaktionsgefüge der granite und gneise und ihre genetische Bedeutung. Springer Berlin, 259 p.
BROTZEN, O. (1959): Outline of mineralization in zoned granitic pegmatites. A quantitative and comparative study. Geol. Fören. Stock. Förhand. 81, 98 p.
FERSMANN, A.E. (1928): Die Schriftstruktur der Granitpegmatite und ihre Entstehung. Zeitschrift für Kristallographie, Bd. 69, 77–104. FERSMANN, A.E. (1931): Über die geochemischgenetische Klassifikation der Granitpegmatite. Miner. Petrograph. Mitteilungen 41, 64–83.
BUREAU, H. & KEPPLER, H. (1999): Complete miscibility between silicate melts and hydrous fluids in the upper mantle: experimental evidence and geochemical implications. Earth Planet. Sci. Lett. 165, 187–196.
GORDIENKO, V.V. (1996): Granite Pegmatites. St. Petersburg, 272 p. (Russ.) GRESENS, R.L. (1967): Tectonic-hydrothermal pegmatites. Contrib. Mineral. Petrol. 15, 345– 355.
BURNHAM, C.W. (1967) Hydrothermal fluids at the magmatic stage. In Geochemistry of Hydrothermal Ore Deposits. (H. Barnes, ed). John Wiley. New York, pp 34–76.
HOLTZ, F., JOHANNES, W., TAMIC, N. & BEHRENS, H. (2001): Maximum and minimum water contents of granitic melts generated in the crust: a reevaluation and implications. Lithos 56, 1–14.
BURNHAM, C.W. (1979) The importance of volatile constituents. In The evolution of the ignous rocks: Fiftieth Anniversary Perspectives. (H.S. Yoder Jr., ed). Princeton Univ Press, 439– 482.
JAHNS, R.H. & BURNHAM, C.W. (1969): Experimental studies of pegmatite genesis: A model for the derivation and crystallization of granite pegmatites. Econ. Geol. 64, 843–864.
ČERNÝ, P. (1975): Granitic pegmatites and their minerals: selected examples of recent progress. Fortschr. Miner. 52, 225–250
KAMENETSKY, V.S., NAUMOV, V.B., DAVIDSON, P., VAN ACHTERBERG, H. E., & RYAN P. (2004): Immiscibility between silicate magmas and aqueous fluids: a melt inclusion pursuit into the magmatic-hydrothermal transition in the Omsukchan Granite (NE Russia). Chem. Geol. 210, 73–90.
ČERNÝ, P. (Ed.) (1982): Granitic Pegmatites in Science and Industry. Mineralogical Association of Canada Short. Course Handbook 8, 555 p. DAVIDSON, P. & KAMENETSKY, V.S. (2001): Immiscibility and continuous melt-fluid evolution within the Rio Blanco porphyry system, Chile: evidence from inclusion in
207
RAINER THOMAS, JAMES D. WEBSTER & PAUL DAVIDSON
KAMENETSKY, V.S., VAN ACHTERBERG, H. E., RYAN, C.G., NAUMOV, V.B., MERNAGH, T.P., & DAVIDSON, P. (2002): Extreme chemical heterogeneity of granite-derived hydrothermal fluids: An example from inclusions in a single crystal of miarolitic quartz. Geology 30, 459–462.
In Melt Inclusions in Volcanic Systems – Methods, Applications and Problems (B. De Vivo & R.J. Bodnar, eds.) Elsevier Science B.V., 1–2 MCKENZIE, D. (1985): The extraction of magma from the crust and mantle. Earth Planet. Sci. Lett. 74, 81–91.
KÖHLER A. (1893) Ein neues Beleuchtungsverfahren für mikrophotographische Zwecke. Zeit. Wisse. Mikro. 10, 433.
MORGAN, G.B. & LONDON, D. (2005): Effect of current density on the electron microprobe analysis of alkali aluminosilicate glasses. Am. Mineral. 90, 1131–1138.
KOZLOWSKI, A. (1978): Pneumatolytic and hydrothermal activity in the Karkonosze–Izera block. Acta Geol. Polon. 28, 171–222.
MUSTART, D.A. (1972): Phase relations in the peralkaline portion of the system Na2O–Al2O3– SiO2–H2O. Ph.D. Diss., Stanford University, 187 pp.
KOSTER VAN GROOS, A.F. & WYLLIE, P.J. (1968): Liquid immiscibility in the join NaAlSi3O8– Na2CO3–H2O and its bearing on the genesis of carbonatites. Amer. J. Sci. 266, 932–967.
MYSEN, B. O. (1988): Structure and properties of silicate melts. Elsevier, 354 pp.
LEMMLEIN, G.G., KLIJA, M.O. & OSTROVSKI, I.A. (1962): About the conditions of the mineral formation in pegmatites from primary inclusions in topaz. Dokl. AH SSSR 142, 81–83 (Russ.).
NAUMOV, V.B. (1979): Determination of concentration and pressure of volatiles in magmatic melts based on study of inclusions in minerals. Geochimija 7, 997–1007.
LONDON, D. (1992): The application of experimental petrology to the genesis and crystallization of granitic pegmatites. Can. Mineral. 30, 499–540.
NIGGLI, P. (1920): Die leichtflüchtigen Bestandteile im Magma. B.G. Teubner, Leipzig, 272 p.
LONDON, D. (1999): Melt boundary-layers and the growth of pegmatitic textures. Can. Mineral. 37, 826–827.
NIGGLI, P. (1937): Das Magma und seine produkte. I. Teil: Physikalisch-chemische Grundlagen. Akademische Verlagsgesellschaft, M.B.H. Leipzig, 379 p.
LONDON, D. (2004): Alkali fractionation and feldspar zonation in granitic pegmatites. Granite Pegmatites: Recent Advances in Mineralogy, Petrology, and Understanding I, (Abstr) Geological Society of America Annual Meeting.
PERETYAZHKO, I.S., ZAGORSKY, V.Y., SMIRNOV, S.Z. & MIKHAILOV, M.Y. (2004): Conditions of pocket formation in the Oktyabrskaya tourmalinerich gem pegmatite (the Malkhan field, Central Transbaikalia, Russia). Chem. Geol. 210, 91–111.
LONDON, D. (2005): Granitic pegmatites: an assessment of current concepts and directions for the future. Lithos 80, 281–303.
PRESTON, R.F., STEVENS, G. & MCCARTHY, T.S. (2003): Fluid compositions in equilibrium with silica-undersaturated magmas in the system Na2O–Al2O3–SiO2–H2O: clues to the composition of fenitizing fluids. Contrib. Mineral. Petrol. 144, 559–569.
LONDON, D., MORGAN, G.B., IV, & HERVIG, R.L. (1989): Vapor-undersaturated experiments with macusani glass + H2O at 200 MPa, and the internal differentiation of granitic pegmatites. Contrib. Mineral. Petrol. 102, 1–17.
REYF, F.G. (1990): Ore-forming potential of granites and conditions for their realization. Moskva “Nauka”, 181 p.
LOWENSTERN, J. B. (1995): Applications of silicatemelt inclusions to the study of magmatic volatiles, In Magmas, Fluids, and Ore Deposits (J.F.H. Thompson, ed.) Mineralogical Association of Canada, Short Course 23, p. 71– 99
RICKERS, K., THOMAS, R. & HEINRICH, W. (2006): The behaviour of trace elements during the chemical evolution of the H2O, B, and F-rich granite-pegmatite-hydrothermal system at Ehrenfriedersdorf, Germany: A SXRF study of melt and fluid inclusions. Mineral. Dep. (accepted)
LOWENSTERN, J. B. (2003): Melt inclusions come of age: volatiles, volcanoes, and Sorby’s legacy. 208
UNDERSTANDING PEGMATITE FORMATION: THE MELT AND FLUID INCLUSION APPROACH
fluid inclusion evidence. In Metallogeny of Collisional Orogens (R. Seltmann, H. Kämpf & P. Möller, eds.) Czech Geological Survey, Prague, 70–81.
ROEDDER, E. (1979): Origin and significance of magmatic inclusions. Bull. of Minéral. 102, 487– 510. ROEDDER, E. (1981): Natural occurrence and significance of fluids indicating high pressure and temperature. Phys. Chem. Earth 13, 9–39.
THOMAS, R. (2000): Determination of water contents of granite melt inclusions by confocal laser Raman microprobe spectroscopy. Am. Mineral. 85, 868 – 872.
ROEDDER, E. (1984a): Fluid inclusions. Reviews in Mineralogy Vol. 12, 644 p.
THOMAS, R. (2002): Determination of water contents in melt inclusions by laser Raman spectroscopy. Workshop – Short course on volcanic systems, geochemical and geophysical monitoring. In Melt inclusions: Methods, Applications and Problems (B. De Vivo & R.J. Bodnar, eds.) Naples, Italy, Proceedings, 211– 216
ROEDDER, E. (1984b): Occurrence and significance of magmatic inclusions and silicate liquid immiscibility. Acta Geol. Polon. 34, 139–178. SAMSON, I., ANDERSON, A. & MARSHALL, D. (Eds.) (2003): The analysis and interpretation of fluid inclusions. Mineralogical Association of Canada Short Course 32, 374 p.
THOMAS, R., FÖRSTER, H.-J. & HEINRICH, W. (2003): The behavior of boron in a peraluminous granite-pegmatite system and associated hydrothermal solutions: a melt and fluid inclusion study. Contrib. Mineral. Petrol. 144, 457–472.
SCHMIDT, W. & THOMAS, R. (1990): Scheme of the genesis of Nb–Ta–pegmatites outlined by the study of inclusions. Zeitschrift für Geol. Wiss. 18/5, 443–446. SHMAKIN, B.M. & MAKAGON, V.M. (1999): Granite pegmatites, Vol. 3: Miarolitic Pegmatites. Novosibirsk “Nauka”, 488 p. (Russ.).
THOMAS, R., FÖRSTER, H.-J., RICKERS, K. & WEBSTER, J.D. (2005): Formation of extremely Frich hydrous melt fractions and hydrothermal fluids during differentiation of highly evolved tingranite magmas: a melt/fluid-inclusion study. Contrib. Mineral. Petrol. 148, 582-601.
SIRBESCU, M.-L. C. & NABELEK, P.I. (2003a): Crustal melts below 400°C. Geology 31, 685– 688. SIRBESCU, M.-L. C. & NABELEK, P.I. (2003b): Crystallization conditions and evolution of magmatic fluids in the Harney Peak Granite and associated pegmatites, Black Hills, South Dakota – Evidence from fluid inclusions. Geochim. Cosmochim. Acta 67, 2443–2465.
THOMAS, R., KAMENETSKY, V.S. & DAVIDSON, P. (2006a): Laser Raman spectroscopic measurements of water in unexposed glass inclusions. Am. Mineral. 91, 467–470. THOMAS, R., RHEDE, D. & TRUMBULL, R. B. (1996): Microthermometry of volatile-rich silicate melt inclusions in granitic rocks. Z. Geol. Wiss. 24, 505 – 526.
SOWERBY, J.R. & KEPPLER, H. (2002): The effect of fluorine, boron and excess sodium on the critical curve in the albite–H2O system. Contrib. Mineral. Petrol. 143, 32–37.
THOMAS, R. & SCHMIDT, W. (1986): Erste Ergebnisse thermobarometrischer Untersuchungen an Einschlüssen mineralbildender Medien aus Pegmatiten der VR Mozambique und einige Bemerkungen zu methodischen Problemen. Freiberger Forschungsheft C 402, 31–48.
STOLPER, E. (1982) Water in silicate glasses: an infrared spectroscopic study. Contrib. Mineral. Petrol. 81, 1–17. STRUNZ, H. (1974): Granites and pegmatites in Eastern Bavaria. Fortschr. Mineral. 52, 1–32. THOMAS, R. (1994a): Estimation of the viscosity and the water content of silicate melts from melt inclusion data. Eur. J. Mineral. 6, 511 – 535.
THOMAS, R., WEBSTER, J.D. & HEINRICH, W. (2000): Melt inclusions in pegmatite quartz: complete miscibility between silicate melts and hydrous fluids at low pressure. Contrib. Mineral. Petrol. 139, 394–401.
THOMAS, R. (1994b): Fluid evolution in relation to the emplacement of the Variscan granites in the Erzgebirge Region: A review of the melt and 209
RAINER THOMAS, JAMES D. WEBSTER & PAUL DAVIDSON
THOMAS, R., WEBSTER, J.D., RHEDE, D., SEIFERT, W., RICKERS, K., FÖRSTER, H.-J., HEINRICH, W. & DAVIDSON, P. (2006b): The transition from peraluminous to peralkaline granitic melts: evidence from melt inclusions and accessory minerals. Lithos (in press).
WATSON, E.D. & BRENAN, J.M. (1987): Fluids in the lithosphere: 1.Experimentally determined wetting characteristics of CO2–H2O fluids and their implications for fluid transport, host-rock physical properties, and fluid inclusion formation. Earth Planet. Sci. Lett. 85, 497–515.
TURNER, F.J. & VERHOOGEN, J. (1960): Igneous and Metamorphic Petrology. 2nd edition, McGraw Hill, New York, 694 pp.
WEBBER, K.L., SIMMONS, W.B., FALSTER, A.U. & FOORD, E.E. (1999): Cooling rates and crystallization dynamics of shallow level pegmatite–aplite dikes, San Diego County, California. Am. Mineral. 84, 708–717.
VALYASHKO, V.M. & KRAVCHUK, K.G. (1978): P– T–X parameters of critical phenomena in solutions of the system SiO2–Na2O–H2O. Akad. Nauk SSSR Dokl. 242, No. 5, 1104–1107
WEBSTER, J.D. (2004): The exsolution of magmatic hydrosaline melts. Chem. Geol. 210, 33–48.
VEKSLER, I.V. & THOMAS, R. (2002): An experimental study of B-, P- and F-rich synthetic granite pegmatite at 0.1 and 0.2 GPa. Contrib. Mineral. Petrol. 143, 673–683.
WEBSTER, J.D., THOMAS, R., RHEDE, D., FOERSTER, H.-J., AND SELTMANN, R. (1997): Melt inclusions in quartz from an evolved peraluminous pegmatite: geochemical evidence for strong tin enrichment in fluorine-rich and phosphorus-rich residual liquids. Geochim. Cosmochim Acta 61, 2589–2604.
VEKSLER, I.V., THOMAS, R., & SCHMIDT, C. (2002): Experimental evidence of three coexisting immiscible fluids in synthetic granite pegmatite. Am. Mineral. 87, 775–779.
WEBSTER, J.D., THOMAS, R., FOERSTER, H.-J., SELTMANN, R. & TAPPEN, C. (2004): Geochemical evolution of halogen-enriched, granite magmas and mineralizing fluids of the Zinnwald tin–tungsten mining district, Erzgebirge, Germany. Mineral. Deposita 39, 452–472.
VEKSLER, I.V., THOMAS, R. & WIRTH, R. (2003): Crystallization of AlPO4–SiO2 solid solutions from granitic melts and implications for P-rich melt inclusions in pegmatite quartz. Am. Mineral. 88, 1724–1730. VON BERTALANFFY, L., BEIER W., & LAUE, R. (1977): Biophysik des Fließgleichgewichtes. Akademieverlag Berlin, 157 p.
WOITSCHACH, G. (1881): Das Granitgebirge von Königshain in der Oberlausitz mit besonderer Berücksichtigung der darin vorkommenden Mineralien. Abh. d. Naturf. Gesellsch. Görlitz 17, 141–197.
210
CHAPTER 10. FLUID AND MELT INCLUSIONS IN THE SUB-VOLCANIC ENVIRONMENTS FROM VOLCANIC SYSTEMS: EXAMPLES FROM THE NEAPOLITAN AREA AND PONTINE ISLANDS, ITALY Benedetto De Vivo1, Annamaria Lima1, Vadim S. Kamenetsky2 and Leonid V. Danyushevsky2 1
Dipartimento di Scienze della Terra, University of Naples Federico II, Via Mezzocannone 8, 80134, Napoli, Italy
2
School of Earth Science and Centre for Ore Deposit Research, University of Tasmania, Hobart, Tasmania, Australia. E-mail:
[email protected] and MI data are used to address the problem of frequent ground movements (bradyseism) in the Campi Flegrei, interpreted as representing a modern analog behaving physically like a porphyry system.
INTRODUCTION The study of fluid (FI) and melt inclusions (MI) can be a powerful tool for understanding melt generation, crystallization, mixing histories of magmas and the conditions of magma evolution during their ascent to the surface. FI and MI data from the alkali syenite xenoliths of different subvolcanic igneous systems in the Neapolitan volcanic area (Vesuvius, Campi Flegrei, Ponza and Ventotene) in Italy provide valuable information on the nature of fluid and melt phases trapped during the late evolutionary stages of these alkaline magmatic systems. They also document liquid immiscibility at pre-eruptive magma conditions and furnish evidence that high salinity fluids (brines) exsolve directly from magma in the upper part of chambers at the magmatic/hydrothermal transition and play critical roles in ore metal transport. Magma chamber margins are of particular significance because FI and MI may record the various evolutionary processes during the crystallization of the magmatic system. The complex daughter crystal assemblages seen in the silicate melt + CO2 + H2O and silicate melt, hypersaline or S-rich aqueous inclusions found in xenoliths of some samples record high solute contents in the fluid(s) during entrapment and provide direct evidence of the magmatic source of these metals. The latter inclusions could be of considerable interest for the interpretation of ore genesis, because FI and MI demonstrate a linkage of these systems with low sulfidation epithermal deposits and some porphyry systems. In addition, FI
BACKGROUND FI are small droplets of fluid trapped within mineral samples (Samson et al. 2003). Most inclusions contain two phases, a liquid and a bubble of gas or vapor. They are generally less than 1 µm in size, but usually those in the range of 1 – 10 µm are the most abundant. More rarely they can reach or exceed 1 cm in length. FI occur in many geologic environments and preserve samples of ancient to very recent fluids, enabling the geologist to unravel the history of the host rocks (Roedder 1972). In the literature the term “fluid inclusion” is used for those inclusions that remain in large part fluid at surface temperature, while the term “melt inclusion” is used for those that have become predominantly solid at surface temperature (Roedder 1984). In this usage, in a general way, MI may refer to silicatedominated inclusions and FI may refer to vapordominated or liquid-dominated inclusions. During magmatic differentiation, the compositions of residual liquids change drastically, and may be recorded by inclusions trapped in minerals. Magmatic phenomena can thus be studied directly from observations of inclusions in igneous minerals that crystallized at depth. The observed composition of FI and MI is dependent on the original composition of the trapped melt and by post-magmatic processes, (e.g., hydrothermal
Mineralogical Association of Canada Short Course 36, Montreal, Quebec, p. 211-237.
211
B. DE VIVO, A. LIMA, V.S. KAMENETSKY & L.V. DANYUSHEVSKY
activity). The latter frequently mask the original composition of trapped inclusions, particularly when the host igneous rock is in contact with the water table (Touret & Frezzotti 1993). Most early inclusion studies have thus concentrated primarily on the evolution of aqueous fluids in subvolcanic systems in both ancient (low sulfidation epithermal deposits) and modern geothermal systems. Indeed, such studies recognized that the majority of hydrothermal systems in ancient volcanic rocks or sub-volcanic, plutonic, suites were very similar to geothermal systems active today (Henley & Ellis 1983). FI and MI in nodule-bearing rocks from sub-volcanic environments are important because they trap brines (hydrosaline liquids), CO2-rich fluids and melts. These phases are representative of a transition from a magma-dominated system to a fluid-dominated hydrothermal system along the margins of magma chambers. The record of these fluids and melts is important because of the critical role they play in ore formation and in the control of eruption styles. In the studies of volcanoes, FI and MI represent a vital source of information about the pre-eruptive volatile gases (i.e., H2O, CO2, S, Cl, and others), which are released during eruption leaving little trace of their former presence in bulk tephra and groundmass glass (Anderson et al. 2000, Lowenstern 2003, and references therein) In recent years, there have been numerous studies on FI and MI, conducted in parallel with studies on FI and MI in volcanic systems (De Vivo & Bodnar 2003, and references therein), to characterize the processes of ore formation in magmas (Audétat et al. 2000, Audétat & Pettke 2003, Grancea et al. 2001, Kamenetsky et al. 1999, 2004, Webster 2004). These studies have emphasized and stressed the critical role played by magmatic hydrosaline chloride liquids (Webster 2004), which are equivalent to chloride-rich aqueous liquids, hypersaline brines or H2O-bearing salt melts (Candela & Piccoli 1995, Roedder 1992, Shinohara 1994) that are normally associated with unmineralized magmas. Chlorine-rich fluids are also shown to play critical roles in ore metal transport. Good examples of these types of fluids occur in Ascension (Webster & Rebbert 2001) and Austral Islands (Lassiter et al. 2002). Webster (2004) emphasized that hydrosaline liquids represent the most Cl-enriched volatile phase that occurs in magmas and that the exsolution of this phase has important consequences for processes of hydrothermal mineralization and for
volcanic emission of Cl to the atmosphere. Webster (2004) compared the volatile contents of MI from the unmineralized alkaline magmas of SommaVesuvius (Marianelli et al. 1995, Raia et al. 2000, Webster & De Vivo 2002, Webster et al. 2003) with the computed Cl solubilities of phonolitic (and more primitive) magmas like those associated with the magmatic-hydrothermal gold–tellurium mineralization at Cripple Creek (Colorado) (Kelley et al. 1998, Thompson et al. 1985), and suggested that hydrosaline chloride liquid should have exsolved at Cripple Creek as the magma evolved to phonolitic compositions. This is consistent with the role of Clenriched mineralizing fluids at Cripple Creek gold– tellurium deposits. The importance of the role of hydrosaline fluids and immiscibility processes in plutonic systems was stressed also by Kamenetsky et al. (2004), who noted that exsolution (unmixing) of the volatile element-rich phases from cooling and crystallizing silicate magmas is important for element transport from Earth’s interior into the atmosphere, hydrosphere, and crustal hydrothermal systems and also for the formation of magmatichydrothermal ore deposits (i.e., ore minerals precipitating from hydrothermal fluids that have – at least partially – a magmatic origin). Kamenetsky et al. (2004) gave an excellent example of MI in phenocrystic and miarolitic quartz, which registers the final stages of cooling and volatile exsolution from granitic magmas of Omsukchan (NE Russia). In such granites, primary MI in quartz phenocrysts demonstrate the coexistence of silicate melt and magma-derived, Cl-rich fluids (brine and vapor). Significant progress in studies of MI in minerals from intrusive rocks has been made because of the improvement of microanalytical techniques. For instance, the method of confocal laser Raman spectroscopy (Thomas 2000, Zajacz et al. 2006) is particularly worthy of mention. This method allows direct determination of H2O in very small MI (< 30 µm, in diameter) some of which are, for various reasons, difficult to analyse with other microanalytical methods such as SIMS and FTIR. Other important progress in microanalysis has been made with the use of LA–ICP–MS (Laser Ablation – Inductively Coupled Plasma – Mass Spectrometry) and PIXE (Particle Induced X-ray Emission) (Heinrich et al. 2003, Kamenetsky et al. 2004). The PIXE technique (which allows for non-destructive analysis of fluid inclusions) offers much higher trace-element sensitivities (down to ppm level measurements for certain elements). The method 212
FLUID AND MELT INCLUSIONS IN SUB-VOLCANIC ENVIRONMENTS
of LA–ICP–MS offers the possibility to analyze a wide concentration range from major (tens of wt.%) to trace (ng.g-1) levels in minerals and their microscopic inclusions. LA–ICP–MS offers better detection limits than PIXE), but it is, however, a destructive technique. Melt evolution can be investigated unambiguously when magma immiscibility occurs in trapped inclusions (Roedder 1992, Roedder & Coombs 1967). As discussed below, good examples of this approach are the studies of granitoid (alkali syenite) xenoliths entrained in volcanic units of the Campania alkaline magmatic province in Italy, which is part of the Roman Comagmatic Region as defined by Washington (1906). In the alkali syenite xenoliths, evidence of the transition from magmatic to hydrothermal systems has been found at Campi Flegrei (Fedele et al. 2006), Vesuvius (Fulignati et al. 2001, Gilg et al. 2001), and in the islands of Ponza (Belkin et al. 1996) and Ventotene (De Vivo et al. 1995) (Fig. 10-1). Beyond the Campania Plain, similar evidence has been found at Pantelleria in the Sicily Channel (De Vivo et al. 1992, 1993, Lowenstern 1993, 1994, Lowenstern & Mahood 1991). For the sake of brevity, we report only examples from the Neapolitan area (Campi Flegrei, Vesuvius) and from the Ventotene and Ponza islands. Detailed studies carried out on cognate syenite nodules from the above sub-volcanic and plutonic rocks in Italy document the existence of immiscibility between hydrosaline fluids and silicate melt, providing evidence for the presence of hydrothermal fluids of magmatic origin that are very similar in composition to those reported from low sulfidation epithermal deposits (Andre-Mayer et al. 2005, Killas et al. 2001) and from some
porphyry copper systems (Cline & Bodnar 1994, De Vivo & Lima 2006, Rapien et al. 2003, Roedder 1984, Sasada 2000). The alkali syenite nodules also display convincing evidence of a transition from a magmadominated system to a fluid-dominated hydrothermal system. This transition took place along the margins of a magma chamber where a magma of trachytic composition was sufficiently evolved to exsolve an aqueous fluid carrying complex solutes with high concentrations of incompatible elements. Textures, mineralogy, and the mineral chemistry of the nodules and the nature of FI daughter minerals all point toward this interpretation. The various sulfides and tungstates observed in the nodules suggest that, for a specific time of evolution, there was potential for an ore deposit somewhere in the sub-volcanic systems. METHODS For Campi Flegrei and Vesuvius, FI studies were carried out on doubly polished wafers of about 300 µm thickness using a Linkam THM600 stage. The stage was calibrated at –56.6, –6.6, 0, 93.8, 163.8 and 398°C using both salts of known melting points and synthetic FI. Calibration indicates a precision at the standard reference points of +0.1°C at 0°C; in the heating mode, calibration gives a precision of +6°C at 398°C. High temperature measurements were carried out on a Linkam THS1500 heating stage calibrated using different synthetic chemicals (KI: 681°C, BaCl2: 963°C, Au: 1063°C). The precision at 1063°C is +20°C. Crystals hosting inclusions that contain daughter minerals were hand fractured and carbon coated. The opened inclusions were analyzed by scanning electron microprobe (SEM) equipped with
FIG. 10-1. Location of Neapolitan and Pontine Island volcanic areas, Italy.
213
B. DE VIVO, A. LIMA, V.S. KAMENETSKY & L.V. DANYUSHEVSKY
EDS Ortec System 5000. Semi-quantitative analyses and photomicrographs of FI daughter mineral phases were obtained with a JEOL JSM840 SEM equipped with a Princeton Gamma Tech energy dispersive X-ray detector. The analyses were made at 15 kV accelerating voltage and a probe current of ~20 nA. Count times were >120 s, and data were corrected using the Phi-Rho-Z program as supplied by Princeton Gamma Tech. The relative accuracy of the analyses is estimated to be ~5– 10wt.%. A LABRAM Raman microspectrometer equipped with an Ar+ laser (λ = 514 nm) was used for Raman analyses. Operating conditions were emission condition of 550–600 mW, with accumulation times of 30 seconds for solid phases, 150–300 seconds for gas phases. Inductively coupled plasma–mass spectrometry (ICP–MS) and inductively coupled plasma–atomic emission spectrometry (ICP–AES) were done on solutions using a lithium metaborate/tetraborate fusion procedure before sample dissolution with multiple acids. For Ventotene and Ponza FI studies were carried out on doubly polished wafers of about 300 µm thickness using a modified Chaix-Meca heating freezing stage (Cunningham & Corollo 1980). The stage was calibrated at –56.6, -6.6, 0, 93.8, 163.8, and 398°C using both salts of known melting points and synthetic FI. Calibration indicates a precision at the standard reference points of +0.1°C at 0°C and +1°C at –56.6°C; in the heating mode, calibration gives a precision of +10°C at 398°C. High temperature measurements were carried out on a Leitz 1350 heating stage calibrated by different chemicals (KI: 681°C; NaCl: 802°C; BaCl2: 963°C; Au: 1063°C). The precision at 1063°C was +30°C. Small fragments of the xenoliths were observed with a JEOL-840 scanning electron microscope (SEM) equipped with a Princeton Gamma-Tech energy-dispersive X-ray fluorescence analyzer (SEM–EDS). Standardless quantitative analysis using ZAF corrections yields probable uncertainties of + 15 % considering the variability of surface configuration. Trace elements in FI have been analyzed using laser ablation–ICP–MS at CODES CoE, University of Tasmania. The instrumentation includes a 213 nm NdYAG laser (UP213, Newwave Research) coupled with an Agilent 4500 quadrupole mass spectrometer. Analyzed inclusions were located 20– 50 µm below the polished surface of the sample. The laser beam diameter was set to exceed the diameter of inclusions slightly to ensure ablation of
the entire inclusion volume. Counting time for each element has been set to 20 ms, with the total sweep time of ~0.3 s. CAMPI FLEGREI The Campi Flegrei system is the largest alkaline volcanic complex in the Campanian Province; its last eruption occurred in 1538 CE, creating the small cone of Monte Nuovo. The Campanian province represents the southern extension of the Roman Magmatic Province (Washington 1906). In the Campania Plain, volcanism spans from > 300 ka to present. It started with fissure-sourced ignimbrite events (from > 300 to 19 ka; De Vivo et al. 2001) and is still active in the Neapolitan area (Mt. Vesuvius and Campi Flegrei; De Vivo 2006, and references therein). Within the Campi Flegrei system is the Breccia Museo, a volcanic breccia of complex origin, containing abundant fragments of juvenile lava, country rock, hydrothermally altered rock, and feldspar-dominated cumulate nodules (syenite). The latter are interpreted to represent portions of a magma chamber margin. These alkali syenite nodules illustrate the processes, mineralogy, and chemistry of interactions involving the magma–host rock–hydrothermal system and show the metallogenetic potential of the Campi Flegrei volcanic system (Fedele et al. 2006). The syenite nodules are composed primarily of potassium feldspar (up to ~ 80%) with subordinate plagioclase, scapolite, a S- and Cl-rich member of the cancrinite group (davyne), amphibole, pyroxene, biotite, magnetite, titanite, apatite, and rarely sodalite (Fig. 10-2). The interlocking nature of the feldspar crystals gave rise to small angular cavities or vugs. No glass was identified in these cavities (with the exception of one sample) suggesting that the phases found therein precipitated from a dense, supercritical aqueous fluid. Thus, the nodules are a mix of major phases, crystallized from a silicate liquid, and minor phases precipitated from a latestage aqueous fluid permeating and moving through a volume of earlier-formed crystals. De Vivo et al. (1989) carried out a detailed FI study on hydrothermal minerals from the Campi Flegrei geothermal system. They presented evidence of extensive alteration extending from the basement rocks to the overlying volcanic rocks. They also reported the existence of a shallow, lowsalinity fluid (~4% NaCl equiv.) and deeper hypersaline fluids (> 26% NaCl equiv.) that were 214
FLUID AND MELT INCLUSIONS IN SUB-VOLCANIC ENVIRONMENTS
FIG. 10-2 A. SEM back-scattered image showing biotite mica (m) and amphibole (A) filling a vug formed by earlier crystallized alkali feldspars (K) of Breccia Museo (Campi Flegrei) (from Fedele et al. 2006). Later phases, monazite (p) and titanite (T), enriched in incompatible elements, formed after the mafic minerals (i.e., biotite and amphibole). B. SEM back-scattered image of sample MT20 showing clinopyroxene (px) and plagioclase (pl) forming with alkali feldspar (K).
thermal/hydrothermal systems and porphyry copper systems was studied to develop and enhance an exploration model (Burnham 1979, Beane 1983). The research on active geothermal/hydrothermal systems was directed to understanding the processes, although the ultimate criterion for a "porphyry copper ore deposit" will be the mass of metal deposited. Sasada (2000) discussed the geothermal porphyry copper analogue and concluded that the major alteration difference, at least in the context of Japanese geothermal systems is that porphyry copper systems have a pervasive potassium-rich alteration. The Breccia Museo– Campi Flegrei system is developed in a highpotassium magmatic environment. Further evolution of the hydrothermal system may lead to potassic alteration.
generated either by continuous boiling at depth, near a magmatic body, or by addition of magmatic fluids to hot meteoric/marine fluids. Caprarelli et al. (1997) suggested the existence of two distinct reservoirs and origins for these fluids: (1) sea water infiltrated at relatively shallow depth (< 2000 m) and mixed with steam-heated groundwater; and (2) a deeper (> 2000 m) hypersaline fluid of probable magmatic origin mixed with meteoric water. The fluids of two reservoirs underwent little if no mixing, probably due to the fluid density contrast. The coexistence of MI with high salinity FI provides direct evidence, of fluid exsolution from the magma, indicating also that, at depth, high salinity fluids were not the products of boiling or condensation. The latter processes were, however, certainly active at more surficial levels (De Vivo et al. 1989). All of these data strongly suggest a similarity between the aqueous fluids associated with the Campi Flegrei–Breccia Museo magma chamber(s) and the mineralized brines related to low sulfidation epithermal deposits (Andre-Mayer et al. 2005, Killas et al. 2001) and, at least partially, to porphyry copper systems (Cline & Bodnar 1994, De Vivo & Lima 2006, Rapien et al. 2003, Roedder 1984, Sasada 2000). In these systems there is increasing evidence that metal transport occurs in high salinity brines, hydrosaline melts and/or coexisting vapor phases which exsolved from silicate magmas (Cline & Bodnar 1991, 1994, Kamenetsky et al. 1999, 2003, Kilinc & Burnham 1972, Roedder 1971). The comparison between modern geo-
Fluid and melt inclusions To study FI and MI, their host potassium feldspar crystals were fractured and a detailed SEM–EDS examination of the fracture surfaces was carried out. This process identified a wide variety of daughter crystals (or accidentally trapped crystals). The SEM–EDS analytical studies performed on the daughter crystals (soluble on heating) in FI revealed the presence of the following minerals: chlorides (halite and sylvite), identified in many opened inclusions (Figs. 10-3A, 3B) and Fe- and Mnchlorides, identified as irregular masses or coating the wall of the inclusions; sulfides [pyrite, (in some cases perhaps pyrrhotite), galena, and chalcopyrite, are commonly associated] (Fig. 10-3D) and the rare occurrence of sphalerite and Ag2S (argentite/ 215
FLUID AND MELT INCLUSIONS IN SUB-VOLCANIC ENVIRONMENTS
FIG. 10-3 SEM back-scattered images of opened fluid inclusions in host alkali feldspar from Breccia Museo (Campi Flegrei) (from Fedele et al. 2006), containing: A. chlorides NaCl (halite) and KCl (sylvite). B. CaSO4 (anhydrite) and halite (NaCl). C. pyrite and scheelite. D. pyrite and chalcopyrite.
acanthite); sulfates (anhydrite, CaSO4) (Fig. 10-3B); carbonates [rare calcite with no Mg above the SEM–EDS detection limit (~1000 ppm) and smithsonite, ZnCO3] and a Pb-bearing phase (probably cerussite); tungsten- and molybdenumbearing phases (rare scheelite with a minor powellite component occurring separately or in association with carbonate or sulfides (Fig. 10-3C); and other minerals [mixed Fe- and Mn-oxygenbearing minerals identified as oxides or hydroxides; apatite as single, euhedral crystals or in multiphase inclusions; a small Ca-, Th-, U-, and REE-bearing phosphate (monazite group); U-bearing zircon, baddeleyite, and probable zirconolite; a very Mnrich pyroxene (tentatively identified as johannsenite)]. FI in potassium feldspar were found in healed fractures and are considered to be of secondary origin. Two types of FI can be distinguished: 1) Type one: one phase inclusions [“vapor only” (V)]; 2) Type two: 3-phase inclusions [liquid + vapor + solid (halite)] (Fig. 10-4).
In addition to FI, we also identified several types of partially to totally crystallized MI. Some MI were isolated or occurred in small clusters and were nowhere related to any apparent fracture or cleavage planes, and are thought to be primary. Thermometric experiments determined that the MI start melting at ~950°C and completely homogenize (i.e., total melting and complete bubble disappearance) at ~1150°C. This high temperature range may reflect previous loss of volatiles during cooling. Microthermometry and observations on secondary hypersaline FI in K-feldspars suggest two possible scenarios for fluid trapping, including the circulation of non-boiling, high temperature (up to 525°C) high-salinity fluids (according to the phase equilibrium constraints in the NaCl–H2O system; Bodnar 1994) which possibly were trapped under decreasing P–T conditions. In this case it is possible that high salinity fluids were generated by direct exsolution from the magma during the late stages of crystallization. The other scenario involves low circulation of boiling, hypersaline fluids, trapped at
216
FLUID AND MELT INCLUSIONS IN SUB-VOLCANIC ENVIRONMENTS
FIG. 10-4 Transmitted light photomicrographs of fluid inclusions in alkali feldspar from Breccia Museo (Campi Flegrei) rock sample (from Fedele et al. 2006). Daughter mineral is always halite A, B. Three-phase inclusions, the arrow indicates halite daughter crystals. C. Details of a plane of coexisting three-phase inclusions. D. Three-phase inclusions.
pressures and temperatures up to 300°C. The strong petrographic evidence for the coexistence of one vapor and two hypersaline melt-dominated FI, seems to suggest that at least these particular inclusions trapped boiling fluids and formed in a later hydrothermal stage.
magmatic to hydrothermal aqueous fluid conditions. This transition was essentially isobaric, whereas the major differences occurred in temperature and fluid composition. The volume of melt that any nodule represents would progressively increase further away from the interface of magmatic melt/crystal accumulation as more and more crystals settled or migrated to the chamber margins. Figure 10-5 (A, B, and C) illustrates this development whereby a loosely packed accumulation of early magmatic feldspar and mafic minerals becomes progressively mineralized, first with precipitation from a volatileenriched silicate melt, followed by mineralization from exsolved, dense brines. Figure 10-6 shows a simplified paragenesis of the major mineral phases forming during the evolution of a silicate melt to a hydrothermal (aqueous brine) fluid. All hightemperature crystallization and/or alteration processes operating in the chamber margin would be quickly and effectively quenched during eruption. However, cooling of the erupted products may be associated with further alteration. The transition from higher temperature magmatic processes to lower temperature aqueous processes has been documented by Fedele et al.
Alkali syenite nodules as samples of a magma chamber margin and their bearing on mineralization potential All evidence from chemistry, mineralogy, and petrography suggests that the studied samples are portions of an alkalic magma chamber margin that was broken up, entrained, and erupted with the Breccia Museo deposits. The bulk of each nodule consists of potassium feldspar, precipitated from a trachytic magma and accumulated around the wall or on the roof of a chamber. During pre-eruption crystallization, immiscible aqueous fluids exsolved and become enriched in various incompatible elements. This was especially important in the marginal region of the magma chamber where cooling and magma-wall rock interaction was significant. The mineral paragenesis and their petrographic relationships record a transition from 217
B. DE VIVO, A. LIMA, V.S. KAMENETSKY & L.V. DANYUSHEVSKY
FIG. 10-6 A simplified paragenetic scheme showing the main mineral phases formed during the transition from a silicate magma to a relatively high temperature aqueous brine (from Fedele et al. 2006). Phases such as ferberite, sulfides, and acanthite are late stage but the details of the sequence of deposition is difficult to decipher as these phases are uncommon, hence the vertical scale is to be considered qualitative. We have used the general textural and observed mineralogical and fluid inclusion characteristics to classify the mineral phases.
FIG. 10-5. Schematic cartoon illustrating the mineralogical and textural evolution of the Breccia Museo (Campi Flegrei) nodules, simplified into three main stages (from Fedele et al. 2006). The arrowmarked trends along the base and left side indicate, in general, the evolution of an average representative nodule. Each regime, A, B, and C, corresponds to the same small volume of magma chamber margin which was dislodged, entrained, and erupted with the Breccia Museo products. A. Early magmatic accumulation of alkali feldspar, plagioclase, and mafic minerals (mostly mica, amphibole, and clinopyroxene). This accumulation was volumetrically dominated by Kfeldspar and contained many open spaces. B. The open spaces of the K-feldspar accumulation were filled with more mafic minerals, Na-rich plagioclase, feldspathoids (scapolite and a cancrinite-group mineral), and phases rich in incompatible elements such as titanite, apatite, monazite, and Zr- and Ubearing minerals. C. The last features represent the precipitation of chlorides, sulfates, and base metals, Ag, W, and Mo phases in fluid inclusions or in fractures.
and the hornblende–plagioclase geothermometer is that of Holland & Blundy (1994). Based on textural evidence, we assume that the mineral pairs were in equilibrium. Capitanio & Mottana (1998) pointed out that the recognition of equilibrium pairs is not a trivial matter. In our case, for every mineral pair (especially alkali feldspar and plagioclase) in these nodules, different generations have been observed in a single thin section. For example, in the Breccia Museo nodules (Fedele et al. 2006), early accumulations of euhedral alkali feldspar and plagioclase formed open spaces that were filled later by anhedral alkali feldspar, plagioclase, and cancrinite. Each nodule represents a small volume of a chamber margin, which is different from that of any other nodule. Nevertheless, despite the differences that are expected to occur around the margins of a crystallizing magma, all the nodules display the same evolution – early-formed magmatic–melt phases giving way to late-stage hypersaline brine precipitated phases. In the alkali syenite nodules of the Breccia Museo we found textural, mineralogical, and chemical evidence that suggests they recorded a transition from magmatic to hydrothermal conditions. Microprobe data provide evidence for the partitioning of F, Cl, and S into minerals of the syenite nodules. Complex assemblages of daughter minerals found in multiphase FI in the nodules of
(2006). Figure 10-7 shows the general pressure– temperature regime estimated for the studied nodules using measured temperature data and several assumptions regarding pressure (as indicated in the figure caption). In Figure 10-7, mineral–pair geothermometers and the liquidus curve derived from the program MELTS (Ghiorso & Sack 1995) have been plotted. The two-feldspar geothermometer used is that of Nekvasil & Burnham (1987) from the program SOLVCALC, 218
FLUID AND MELT INCLUSIONS IN SUB-VOLCANIC ENVIRONMENTS
FIG. 10-7 A schematic cartoon in pressure/temperature space that illustrates the evolution of the studied Breccia Museo (Campi Flegrei) nodules (from Fedele et al. 2006). Here we have used various mineral pair geothermometers, fluid inclusion data, and liquidus temperatures to construct a reasonable PT evolution of the nodules. The liquidus curve is derived from MELTS (Ghiorso & Sack 1995) using the average Campanian Ignimbrite (CI) whole rock composition. The general range of temperatures from early alkali feldspar–plagioclase pairs is shown by two curves and the field (small open circles) is shown for texturally late stage alkali feldspar–plagioclase pairs (both using Nekvasil & Burnham 1987). The temperature is extrapolated to higher pressures by the dashed lines. The field for the hornblende–plagioclase geothermometer data (Holland & Blundy 1994) is shown by the “triangle and lines” pattern. The general, estimated PT field for the exsolution of aqueous fluid is shown. The horizontal arrow traces, in general, the PT evolution of the average nodule from the melt environment to the PT conditions just before entrainment and eruption. We have chosen a constant pressure of about 2200 bars, corresponding to a depth of 8 km (assuming an overburden density of 2.7 g/cm3). With regard to the liquidus (CI) and mineral pair geothermometer data, their curves are very steep so the exact pressure regime assumption is not critical. The letters, A, B, and C mark the general PT areas as outlined by figure 10-5.
the Breccia Museo are evidence of the trapping of high solute fluids. The abundance of chlorides, sulfides, and to a lesser extent sulfates and carbonates, suggests that the FI trapped a hypersaline/S-rich fluid (possibly with minor CO2) that may have been exsolving from a crystallizing magma. In the past two decades, many studies of magma chamber margins in alkali-enriched magmatic systems (Belkin et al. 1994, 1996, Federico et al. 1994, Fulignati et al. 1997, Gilg et al. 2001, Renzulli et al. 1995, Tarzia et al. 1999, 2000, Turbeville 1992) have demonstrated that these peripheral areas can show selective enrichment of incompatible elements (i.e., U, Th, Zr, REE) with features similar to those found in the Breccia Museo nodules. Experimental studies have pointed out that many metals tend to partition in favor of a hypersaline chloride-bearing brine exsolved from a silicate melt (Candela 1986, 1989, Candela & Piccoli 1995, Cline & Bodnar 1991, Kamenetsky et al. 2003, Shinohara 1994). Similarly, trace and REE elements can be efficiently extracted from magma by Cl-rich
fluids (Haas et al. 1995, Kravchuk & Keppler 1994). Sulfate-rich fluids are reported to have been present in peralkaline silicic magmatic intrusions and in magmatic-hydrothermal fluids related to porphyry copper deposits (Roedder 1984). The occurrence of sulfides together with chlorides, sulfates, and Fe-Mn-oxides in the nodules and as daughter minerals, suggests that the redox state of the nodule-forming environment shifted progressively towards more oxidized conditions. Concerning the partition of REE into a S-rich fluid, Wood (1990) stated that for temperatures <350°C sulfate–REE complexes can predominate over REE–aqueous species in the absence of other ligands. Furthermore, temperature strongly affects the REE–complex stability constant: increasing temperatures produce an increase in the stability constant values for fluoride, sulfate, and chloride (in that order). From a detailed study of Nd3+ complexation in chloride solutions Migdisov & William-Jones (2002) concluded that with increasing temperature (up to 250°C) Nd-chloride species are dominant in a wide range of chloride 219
B. DE VIVO, A. LIMA, V.S. KAMENETSKY & L.V. DANYUSHEVSKY
concentrations, although these complexes may not be sufficiently stable to account solely for Nd transport reported in natural systems. Therefore chloride and sulfate species, which were present in the peripheral magma chamber system of the Breccia Museo magma, could have played an important role in the selective enrichment of REE in the accessory minerals of the nodules. Although these nodules are not exceptionally mineralized in comparison to typical W-skarn deposit samples, they do contain an interesting assemblage of W (Mo) mineral species (e.g., scheelite–powellite, ferberite). Manning & Henderson (1984) demonstrated that chloride complexes can efficiently extract W from magmatic melts into an aqueous phase, and Webster (1997) showed that Mo partitions strongly into a brine relative to silicate melt. This suggests that an exsolving, Cl-rich aqueous phase most likely extracted W and Mo from the magma and transported these elements to the nodule environment where cooler conditions or some particular reaction caused precipitation. This region also displays seismic evidence of fluid-related activities (De Natale et al. 2001, and references therein). De Vivo & Lima (2006) addressed the problem of frequent ground movements (bradyseism) in the Campi Flegrei volcanic region. Combining FI and MI with other geological and geophysical evidence, the authors explained the bradyseism at Campi Flegrei using a model that emphasizes the role of hydrothermal fluids, as opposed to the classical model of intrusion of new magma at shallow depth (Barberi et al. 1984). The model of De Vivo & Lima (2006) suggested that ground deformation could be generated by conductive heating of the hydrothermal fluids overlying the magma chamber. This model is based on what is well known from porphyry systems (Burnham 1979, Fournier 1999, Henley & McNabb 1978), and suggests that Campi Flegrei might actually represent a modern analogue behaving physically like an ancient porphyry system (Beane & Titley 1981, Beane & Bodnar 1995, Roedder & Bodnar 1997, Sasada 2000) as has been suggested for White Island, New Zealand (Rapien et al. 2003). This comparison does not extend, necessarily, to the mechanism of metal transport and deposition. Porphyry copper systems are, in fact, associated with calc-alkaline magmas and have been shown to exsolve low to moderately saline fluids that subsequently boil or condense to yield high salinity fluids. However, the Campi
Flegrei igneous system is alkaline and the high salinity fluids are interpreted to have exsolved directly from the magma. In addition, in porphyry systems, the mechanism of metal transport in the aqueous phase is implied to be as chloride complexes. This circumstance, although it may hold for copper, likely does not hold for high field strength metals such as tungsten or molybdenum, which are most stable as oxyacids. In the Neapolitan volcanoes, convectively driven fluids are found only in the volcaniclastic sedimentary rocks of the Campi Flegrei caldera and are representative of the brittle, hydrostatic domain. The coexistence of liquid-dominated and vapordominated inclusions in the same FI assemblage is strong evidence of boiling conditions during inclusion trapping, whereas FI with daughter crystals trapped in samples from deeper, hotter levels indicate a high concentration of solute (brines), as confirmed by drilling. The scenario suggested by FI and MI data indicates that the Campi Flegrei system receives an influx of saline water (magmatic + sea water), localized in aquifers at depths of ~2.5–3 km. The fluids are heated by the underlying crystallizing magma and remain under lithostatic pressure for long periods. The pressure in the upper, apical part of the magma chamber increases as water exsolves from the magma and causes uplift of the overlying rocks (positive bradyseism). When the system ruptures, due to the increasing pressure, the regime changes from lithostatic to hydrostatic resulting in boiling, hydraulic fracturing, volcanic tremors and finally pressure release leading to deflation of the ground. Afterward, the system begins to seal again due to the precipitation of newly formed minerals and a new phase of positive bradyseism will occur only after several years, when the system “reloads” under new lithostatic pressure conditions. In this scenario a hydrothermal eruption can still occur, but only if the fluids pass from lithostatic to hydrostatic pressure when the overlying rocks have a thickness <500 m. If this happens, the hydrothermal eruption could even trigger a magmatic eruption, as was probably the case during the Monte Nuovo eruption in 1538 CE. MT. VESUVIUS Samples of FI and MI in skarns and associated cognate and xenolithic cumulate nodules from Mt. Vesuvius have been studied (Fulignati et al. 2001, Gilg et al. 2001). The cumulates comprise a variety of rocks interpreted as solidified crystal 220
FLUID AND MELT INCLUSIONS IN SUB-VOLCANIC ENVIRONMENTS
W-bearing minerals (tungstite – WO3•H2O or monoclinic hydrotungstite – H2WO4•H2O), and apatite. The daughter minerals define two distinct associations: a chloride association and a fluoride association. Fulignati et al. (1998a) proposed that the two associations reflect the chemical variations of the fluid phase during exsolution in response to the different melt/fluid partition coefficients for Cl and F (Signorelli & Carroll 2000, Villemant & Boudon 1999). Based on these partition coefficients, they suggested that a chloride-bearing brine exsolved early, while the fluoride-bearing brine represents fluids released later from the crystallizing magma. The high salinity shown by the multiphase FI, together with the high temperatures revealed by microthermometry (Th ~700–750°C, Fulignati et al. 1997) suggest that the fluids trapped in the inclusions represent magmatic brines. These brines likely originated by exsolution of a fluid phase by crystallizing magma. As at Campi Flegrei (Fedele et al. 2006), tungsten–oxide daughter minerals occur in the chloride-bearing FI, but not in the fluoridebearing FI. This represents further evidence for ore metal transport by means of chloride-bearing brines exsolving from a silicate melt (Candela 1986, 1989, Candela & Piccoli 1995, Cline & Bodnar 1991, Manning & Henderson 1984, Shinohara 1994). Gilg et al. (2001) calculated the pressure of skarn formation from the densities of CO2 inclusions in wollastonite from skarn nodules, to be from 660 to 1368 bars [assuming a formation temperature of ~1000°C and using the Brown & Lamb (1989) equation of state for CO2]. Multiphase aqueous brine inclusions are present in wollastonite, calcite, and vesuvianite in all samples investigated. They invariably contain large isotropic daughter crystals of halite and sylvite (Fig. 10-8). Additionally, these inclusions contain several birefringent daughter minerals, which have been identified by Raman and SEM–EDS as carbonates (calcite and/or Mg-calcite) and sulfates (matteuccite, arcanite), and opaque phases. SEM– EDS analysis of salt residues of opened FI reveals the presence of Ti, Cu, and Zn. Estimated salinities based on halite and sylvite melting temperatures are in the range 43 to 52 wt.% NaCl equivalent. The other daughter crystals always melt before vapor bubble disappearance. The total homogenization temperatures range from 720 to 820°C. Salt-rich MI occur in wollastonite, gehlenite, scapolite and “fassaitic” clinopyroxene of skarn samples. The major phases identified by
mushes that have been studied in detail by Belkin et al. (1985), Belkin & De Vivo (1993), Cioni et al. (1995), Fulignati et al. (1997, 1998 a, b), Hermes & Cornell (1978, 1981) and Joron et al. (1987). The most common members of this group are mafic cumulates including phlogopite-bearing clinopyroxenite, rare biotite-bearing dunite, and wehrlite. Two distinct types of pyroxenite nodules have been recognized on the basis of mineral composition (Cioni et al. 1995, Fulignati et al. 1998b, Hermes & Cornell 1978, 1981). The so-called “metasomatic mafic cumulates” are distinguished from “normal mafic cumulates” by the presence of green spinel, the more aluminous (“fassaitic”) character of diopside, the higher Mg/Fe ratio of phlogopite, and lower Cr and Ni contents in whole rocks. Metasomatic mafic cumulates were initially interpreted as “skarns” by Hermes & Cornell (1978), but a subsequent FI study by Belkin et al. (1985) showed that these rocks crystallized from a silicate melt and not from an aqueous fluid. The silicate melt from which the metasomatic mafic cumulates crystallized was probably contaminated by carbonate material (Belkin et al. 1985, Cioni et al. 1995, Fulignati et al. 1998b). Fluid and melt inclusions Cognate xenoliths of feldspathoid-bearing syenite ejected in the 79 CE eruption (Fulignati et al. 1998a) were investigated through SEM–EDS study of daughter minerals (i.e., minerals soluble on heating) in multiphase hypersaline FI in sanidine crystals. The samples studied represent the magmatic mush crystallized at the magma chamber walls, and the hypersaline FI are indicative of an immiscible fluid phase exsolving from the melt. The melt from which the fluid has separated is probably a hybrid melt of pristine Vesuvius magma affected by wall-rock interaction. FI (aqueous brine + CO2 fluids) and salt-rich MI occur in calcite, wollastonite, “fassaitic” diopside, and vesuvianite of massive unzoned wollastonite- and gehlenitebearing calc-silicate ejecta and zoned vesuvianitebearing skarns (Gilg et al. 2001). These minerals contain various generations of secondary inclusions, so the identification of unambiguously primary inclusions is often very difficult. The SEM–EDS analyses of feldspathoid-bearing syenite (Fulignati et al. 1998a) revealed the presence of chlorides (halite and sylvite), fluorite, calcite, Fe-bearing minerals, pyrite, sulfates [anhydrite or gypsum, glaserite – (K,Na)2SO4], other minerals (sodalite or scapolite or cancrinite group, nahcolite – NaHCO3), 221
FLUID AND MELT INCLUSIONS IN SUB-VOLCANIC ENVIRONMENTS
FIG. 10-8 Photomicrographs of fluid inclusions in Vesuvius skarns (from Gilg et al. 2001). A. A primary type inclusion representing trapping of two immiscible fluids, a silicate melt and a supercritical CO2 fluid in skarn sample SKM3. Inclusion in wollastonite at room temperature with well defined menisci separating CO2 vapor (V), CO2 liquid (L) and silicate glass (SG). B. A primary silicate melt inclusion with shrinkage bubble and a sulfide globule in wollastonite of skarn sample SKM3. C. Multiphase aqueous brine inclusion in wollastonite of skarn sample SKM7 with two isotropic cubic soluble daughter minerals (SDM). D. Multiphase aqueous brine inclusion in wollastonite of skarn sample SKM7 with two isotropic cubes (daughter minerals, DM), an opaque phase (OP) and a birefringent mineral (BM). The vapor bubble contains CO2 and H2S.
small crystals of minor phases, including barite and iron oxide, have been identified. Sylvite and halite crystals melt at temperatures of 200 to 220° and 480 to 500°C, respectively. The total salinity is up to 55 wt.% NaCl equivalent. The temperature of complete homogenization to liquid ranges between 870 and 890°C. The high chloride contents of the FI and
SEM–EDS analyses in opened inclusions (Fig. 109) are halite, sylvite, calcite, Mg-calcite, a K-sulfate (probably arcanite, K2SO4), a Na-sulfate (probably matteuccitte, NaHSO4•H2O), a Ca sulfate anhydrite, a K-Ca sulfate mineral (syngenite?), a K-Na sulfate (aphthitalite?), Na-Ca sulfate (glauberite?), apatite, fluorite, Fe-Cl-bearing biotite, and phlogopite. Very
FIG. 10-9 Representative saline melt inclusion containing chloride, carbonate, sulfate and silicate phases in Vesuvius skarn (from Gilg et al. 2001). A SEM back-scattered image of an inclusion on a fresh fractured surface in wollastonite. Phases identified with a high degree of confidence by SEM energy dispersive X-ray analysis are sylvite (S), barite (B), calcite (C), a hexagonal crystal of biotite mica (M), and anhydrite (An). Other phases more tentatively identified are a K, Na sulfate (X), perhaps aphthitalite, and a Na, Ca sulfate (G), perhaps glauberite.
222
FLUID AND MELT INCLUSIONS IN SUB-VOLCANIC ENVIRONMENTS
magmatic MI are consistent with the prediction of exsolved magmatic hydrosaline liquids from Vesuvius magmas (Webster & De Vivo 2002). During heating and subsequent cooling experiments, most of the larger (60–90 µm diameter) MI show clear evidence of fluid immiscibility, emphasizing the presence of several immiscible fluids (silicate melt–aqueous chloriderich liquid–carbonate/sulfate melt). MI have been found only in wollastonite and clinopyroxene of skarn samples. They homogenize in a narrow temperature interval of 1000–1050°C. During heating experiments a black immiscible fluid (CO2-rich vapor?) separates from the melt at about 600°C. It dissolves in the silicate melt at about the same temperature at which the vapor bubble disappears (total homogenization). The presence of MI coexisting with CO2-bearing vapor-rich FI in a wollastonite skarn sample (Fig. 10-8A) provides clear evidence for fluid immiscibility between a silicate melt and a CO2-rich vapor at the carbonate–magma contact zone. The same fluid association has been described in pyroxenite cumulate nodules from non-Plinian eruptive episodes by Belkin et al. (1985). It is noted, however, that the CO2-rich vapor inclusions are not commonly observed in clinopyroxene phenocrysts of lavas and pyroclastic rocks from Vesuvius that are generally rich in MI (e.g., Cioni et al. 1998, Joron et al. 1987, Roedder 1965, Vaggelli et al. 1992). In some samples, we also found salt-rich MI in close association with silicate melt and CO2 vapor-rich FI. Some saline inclusions have appreciable amounts of CO2 ± H2S in their vapor bubble as indicated by Raman spectroscopy or microthermometry, whereas in most other saline inclusions we could not detect any CO2. This observation may suggest heterogeneous trapping of two immiscible phases (CO2-rich vapor and saltrich melt). The presence of small amounts of silicate daughter minerals in salt-rich MI and the observations of liquid immiscibility during microthermometric measurements of some salt-rich MI may indicate heterogeneous trapping or partial miscibility between a silicate and a salt melt. The homogenization temperatures (Th) of MI (1000– 1050°C) are at least 110°C higher than those of saltrich MI (<890°C). Therefore, salt-rich MI probably exsolved from a silicate melt during rapid cooling. Thus, immiscibility among three fluid phases including a silicate melt, a salt-rich melt and a CO2-
rich vapor seems very probable at the magma– carbonate interface at Mt. Vesuvius. VENTOTENE The compositions of FI in xenoliths of Ventotene, a volcanic island of the Pontine Archipelago in the Tyrrhenian Sea, have been investigated. These xenoliths represent a continuous magmatic sequence from olivine clinopyroxenite (with cumulate texture) to feldspathoid-bearing syenite. The results, described below, suggest that volatiles (F, H2O) and alkalis (Na, K) were partitioned from the crystallizing magma preferentially into the minerals of feldspathoidbearing syenite that resided in the upper part of a magma chamber there (De Vivo et al. 1995). Xenoliths from the ultramafic–mafic group generally contain MI ± vapor bubbles ± droplets of an opaque phase. Rare MI in clinopyroxene record heterogeneous trapping of various phases present in the magmatic system. In these inclusions, the “salt” melt phase, composed of different sulfates, carbonates and probably chlorides, exceeds the silicate glass volumetrically. Two other phases, e.g., a CO2-rich vapor bubble and a Cu-rich Femonosulfide globule, are also present. Inclusions of this type are very rare, but nevertheless they provide information on the nature of immiscibility process during crystallization (Fig. 10-10). Xenoliths from the alkali syenite and from transitional rocks have a much greater variety of FI and MI. They include silicate melt + salt inclusions (with highly variable ratios between the two phases), silicate melt + CO2 vapor-rich FI, aqueous (liquid + vapor) inclusions, silicate melt + “vapor” inclusions (shrinkage bubbles), [CO2 (liquid) + CO2 (vapor) + H2O]-rich FI, silicate melt + salt-rich melt + H2O-rich FI (again with highly variable ratios between the phases), and aqueous FI with one or more isotropic, soluble daughter minerals of salts or other nonsoluble birefringent and/or opaque daughter minerals, and usually with some silicate glass. The Ventotene xenoliths show clear evidence of immiscibility in the system silicate melt–hydrous salt-rich melt–aqueous fluid–CO2 fluid (De Vivo et al. 1995). Silicate melt or hydrosaline melt, and silicate melt or CO2 inclusions indicate heterogeneous trapping of immiscible phases [silicate melt + hydrosaline melt, and silicate melt + CO2 respectively] during crystallization. Such immiscibility may have resulted from either the direct exsolution of a hydrosaline fluid or a CO2dominated vapor from the evolving magma, perhaps 223
FLUID AND MELT INCLUSIONS IN SUB-VOLCANIC ENVIRONMENTS
FIG. 10-10 A single complex inclusion trapped in clinopyroxene from a mafic xenolith (Ventotene island, Italy) (from De Vivo et al. 2005). This inclusion records heterogeneous trapping of various phases present in the magmatic system. The “salt” melt component, composed of different sulfates, carbonates and probably chlorides, exceeds the brownish silicate glass in volume. Two other components, a CO2-rich vapor bubble and a Cu-rich Fe-monosulfide globule, are also present. Inclusions of this type are extremely rare, but nevertheless they provide information on the nature of immiscibility process during crystallization. All of these phases can be found as individual inclusions in phenocrysts. The silicate glass has the composition of a residual melt showing depletion in CaO and MgO after crystallization of clinopyroxene on the inclusion walls and has quite significant abundances of K2O (8.9 wt.%), P2O5 (0.93 wt %), and S (0.45 wt.%). The residual glass is also characterized by low SiO2 (40 wt.%) and a low microprobe total. The latter suggest a significant amount of dissolved H2O (8–10 wt.%). The size of inclusion is 25 µm.
rounded. These changes suggest melting. With further heating, at 400–430oC the number of phases decreases, but the size of the phases increases significantly, whereas the bubbles are almost spherical. At 500oC, melting is very strong, and the remaining crystals become more rounded and move within the liquid (Fig. 10-11B). Some crystals shrink suddenly and disappear, whereas new crystals appear. At 580oC, the crystals rapidly diminish in size in both inclusions, allowing more movement of the bubbles. At 603oC spherical crystals are small, similar in size (~5 µm in diameter), and show Brownian movement (Fig. 1011C). At 610oC, the spherical crystals in both inclusions rapidly shrink and simultaneously disappear (Fig. 10-11D). At 620–625oC, the inclusions acquire strong optical contrast with the host feldspar, and the inclusion shapes become seemingly more round than at lower temperature. At 760–800oC, the
by crystal fractionation processes, or the loss of water from the residual melt, as existing pyroxenes reacted with the melt to crystallize hydrous minerals like biotite and amphibole. Heating/cooling experiments have been conducted on MI in feldspar from the same Ventotene alkali syenite samples (Fig. 10-11). At room temperature, two typical inclusions in Kfeldspar (Fig. 10-11A) exhibit “rectangular’ shapes, a large vapor phase (~20–30 wt.%) and a finegrained texture. Daughter phases (i.e., those soluble on heating) are not easily distinguished from each other because of the small differences in refractive index. Possibly, some crystals in the MI are birefringent. Aqueous liquid is not visible. The first visible changes occur on heating at ~185–190oC, and movements of crystals and fluid phases are recorded as rapid or jerk-like. This might record the beginning of melting or recrystallization. The phase boundaries are better defined at 250–270oC; the phases clearly change shape and size and the fluid phase becomes more
224
FLUID AND MELT INCLUSIONS IN SUB-VOLCANIC ENVIRONMENTS
FIG. 10-11. Melt immiscibility experiments in feldspar from alkali syenite (Ventotene Island, Italy). Photomicrographs show heating (A–I) and cooling (J–L) experiments with essentially non-silicate melt inclusions. Inclusions sizes are ~22 µm. See text for detailed description of phase transformations.
bubbles, decreases in size, and thus makes visible a low-relief translucent phase (melt B) along the original outlines of the inclusions (Fig. 10-11G). At first, melt A shows an ellipsoidal shape, but then the shape gradually changes to spherical (Fig. 10-11H). At the same time, the bubbles decrease in size, and their movement becomes very fast. With shrinking
increased optical contrast between inclusions and feldspar clearly shows that the inclusion corners became rounded (Fig. 10-11E). At 847oC, the bubbles reach 5 µm and start Brownian motion of small amplitude (Fig. 10-11F). At 860–870oC, it becomes clear for both inclusions that a high relief, rounded phase (melt A), containing wobbling vapor
225
B. DE VIVO, A. LIMA, V.S. KAMENETSKY & L.V. DANYUSHEVSKY
of the melt A volume, the volume of melt B increases accordingly, as well as phase contrast with surrounding feldspar. Globules of melt A homogenize by bubble dissolution at 938oC “upper” inclusion) and 946oC (“lower” inclusion) (Fig. 1011I). Close to the homogenization temperature, the globules of melt A are perfect spheres with rapidly moving vapor bubbles. Their size is only 30 vol.% of the entire inclusion. The “coat” of melt B around globules of melt A is voluminously dominant (~70%, respectively). The outer boundaries of the MI define negative crystallographic shapes better than at room temperature, and the relative increase in volume at homogenization temperature is ~15–25 %. Quenching causes a relative increase in size of the globules of melt A and in their deformation with bubble appearance at <900oC (Fig. 10-11J). Cooling further deforms the globules and vapor bubbles, so they become larger and angular (Fig. 10-11K). At the same time melt B diminishes in volume, and its boundary with the host mineral gradually fades. The last of melt B is visible at 670oC. At 405oC a spontaneous change in the phase composition occurs in both inclusions (Fig. 10-11L). Down to 150–200oC, the newly formed phases (gel-like substance) continuously change shape and position, while the bubbles grow and become angular. This experiment points out the following observations: a) Although type-A and type-B melts are clearly immiscible liquids at high temperature (Fig. 1011 F-J), there is uncertainty about their relationships on cooling. Apparently, melt B has an aluminosilicate component, but the activity of this component decreases by crystallization of feldspar on the walls of inclusions, and thus the miscibility gap between original type-A and type-B melts is eventually bridged. In other words, the two melts become increasingly miscible due to melt evolution. b) The low viscosity of the trapped melt, especially type-A, suggests that it is essentially nonsilicate. Chloride is surely present in these inclusions (based on the LA–ICP–MS signal, Fig. 10-12), but the compositions of the nonsilicate components of the dominant melt B and subordinate melt A (at high temperature) remain unknown. Carbonates and/or sulfates cannot be excluded, but Ca is not a major cation (based on LA–ICP–MS) in this case. Major cations
detected by LA–ICP–MS are Na, K, Fe, and Mn; and the most prominent ore metals are Pb and Zn (Fig. 10-12). Incompatible lithophile trace elements (REE, Rb, and U) are also present in amounts of 10s–1000s of ppm. c) Spontaneous change in the phase composition at <415oC (Fig. 10-11L) and continuous phase movements and re-arrangements down to ~150– 200oC (not shown) are unlikely to reflect crystallization, but may represent another liquid immiscibility event or transition from a liquid (melt) to a colloid. Another type of immiscibility in the Ventotene xenoliths (in addition to silicate melt + hydrosaline melt) is the occurrence of silicate melt + CO2 inclusions, and in some cases, these inclusions co-exist with silicate melt + hydrosaline melt. Evidently, a primary assemblage of silicate melt, hydrosaline melt, and CO2 fluids exists in crystallizing syenite and this is consistent with observations of Frost & Touret (1989). This assemblage provides further evidence that the brine and CO2-rich vapor are primary igneous phases. The presence of immiscible NaCl melts in magmatic rocks is explained by the fact that chloride is weakly soluble in silicate melt (Webster & De Vivo 2002) and is partitioned strongly into an aqueous fluid phase (Kilinc & Burnham 1972). At Ventotene the high salinity of the magmatic fluids exsolving from the crystallizing melt was strongly dependent on the pressure of crystallization (Bodnar 1992). In the case of Ventotene xenoliths (De Vivo et al. 1995), crystallization pressures <1300 bar are suggested by study of primary CO2 FI (Fig. 10-13). At such pressures, the salinity of a fluid increases during crystallization and may reach values exceeding 50 wt.% NaCl equivalent (Bodnar 1992). This explanation is in agreement with our results from the Ventotene xenoliths, and confirms the fact that the hydrosaline MI occur preferentially in shallow intrusive rocks (Weisbrod 1981). Subsequent evolution of the fluid phases under subsolidus conditions is represented by secondary aqueous FI. The data from these inclusions indicate a moderate temperature hydrothermal stage (usually between 120 and 290°C), mostly with low salinity water (from 0.2 to 2.4 wt.% NaCl equivalent) circulating in the intrusive system through microfractures. Finally, we want to stress that the immiscibility phenomena recorded by MI, albeit still difficult to explain due to lack of appropriate 226
FLUID AND MELT INCLUSIONS IN SUB-VOLCANIC ENVIRONMENTS
FIG. 10-12. Signal intensities with time for selected metals, major elements and lithophile trace elements recorded on the mass spectrometer during LA–ICP–MS analysis of representative salt melt inclusion trapped in Kfeldspar of the Ventotene felsic xenolith VT5B. Note increase in abundances of Cl, Na, K, and metallic and lithophile elements that corresponds in time to the ablation of the inclusion.
experimental data, are related to the phase relationships in the better known model system NaCl–H2O–CO2, where the presence of NaCl dramatically increases the T–P range of vaporliquid immiscibility (Bowers & Helgeson 1983, Duan et al. 1995, Schmidt & Bodnar 2000). PONZA
thorianite [(U,Th)O2]. The xenolith contains common angular cavities (druse or miarolitic cavities), which occur as the result of the interlocking texture of the feldspars. These cavities are interpreted to be magmatic in origin based on mineralogy and inclusion study, and they represent the probable sites of vapor bubble location during late-stage magmatic fluid exsolution.
FI and MI in a feldspathoid-bearing syenite xenolith entrained in trachyte from Ponza were studied (Belkin et al. 1996). The xenolith, which may be a cognate nodule, provides an opportunity to examine a sample of the zone of magma–wall rock interaction. It is a feldspathoid-bearing syenite composed primarily of K- and Na-rich feldspar with less than 15% of clinopyroxene, amphibole, biotite, manganoan magnetite, titanite, apatite, and nosean. Also identified were baddeleyite (ZrO2), probable zirkelite [(Ca,Ce,Y,Fe)(Ti,Zr,Th)3O8], cheralite [(Ce,La,Th)PO4], pyrrhotite (FeS), and uraninite-
Fluid and melt inclusions. All studied inclusions were hosted by K-feldspar, but rare salt-rich FI and MI were observed in amphibole as well. Trapped silicate melt was observed as glass on the walls of some vapor-phase inclusions. The first observable melting of silicate glass occurred between 700 and 800°C as indicated by trapped bubbles becoming more rounded. One-phase silicate MI are smaller in size. They rarely exceed 40–50 µm in diameter, and they contain transparent colorless to pale brown glass, or devitrified glass. Aqueous FI can be 227
FLUID AND MELT INCLUSIONS IN SUB-VOLCANIC ENVIRONMENTS
FIG. 10-13. P–T plot for fluid inclusions in Ventotene xenoliths (from De Vivo et al. 1995). Isochores for CO2 and H2O–NaCl are from Brown & Lamb (1989). 1, CO2 isochores from sample VT-2A (gabbro); 2, CO2 isochores from VT-4 (transition rock); 3, CO2 isochores from sample VT-5B (alkali syenite); 4, H2O–NaCl isochores from secondary fluid inclusions (samples VT-4; VT-5B). Fields A, B an C represent the P–T conditions of crystallization of gabbro (A), transition rocks (B) and alkali syenite (C), respectively; field D represents the P–T conditions of late stage aqueous secondary fluid inclusion trapping. The large bold dashed arrow refers to probable fluid evolution path.
classified into three types: 1) one-phase vapor inclusions which can reach 200 µm in long dimension. 2) two-phase inclusions, vapor + liquid (vapor-rich inclusions with a small amount of CO2, in most cases, are common). The presence of CO2 was identified by the formation of the CO2-clathrate on cooling. The Th–CO2 hydrates ranged from 5.3 to 8.5°C, corresponding to salinities between 2.9 and 8.5 wt.% NaCl equivalent. The homogenization (Th–V) to the vapor phase occurs at temperatures between 359 and 424°C. 3) Hypersaline/S-rich, three-phase and multiphase inclusions. Isotropic, cubic minerals, which are soluble on heating, were identified as NaCl and KCl. Opaque crystals may be a daughter phase in some inclusions or an accidentally trapped mineral in others. Hypersaline/S-rich inclusions occur individually, in clusters together with vapor + MI or along healed fractures. On heating, the melting of isotropic daughter
minerals occurs between 459°C and 536°C, whereas the total homogenization to the liquid phase occurs at temperatures between 640°C and 755°C. Salinities, calculated from the melting point temperatures (vapor-present), are between 54 and 65 wt.% NaCl equivalent (Zhang & Frantz 1987) assuming that Na is the only cation present in the system. SEM–EDS analyses reveal the presence of a S- and K-bearing phase (arcanite, K2SO4) (Fig. 10-14A), a Na-, Ca-, and S-bearing phase [glauberite, Na2Ca(SO4)2], as blocky prisms (Fig. 10-14B), as single elongated prisms (Fig. 10-14C, 14D) and as masses of prisms, a Ca- and S-bearing phase (anhydrite, CaSO4); and a Na- and S-bearing phase (thernardite). Commonly occurring with sulfates and chlorides were crystals identified as Na-rich K-feldspar (Fig. 10-14D). In many inclusions, small irregular masses of Fe and Mn oxides were observed.
228
FLUID AND MELT INCLUSIONS IN SUB-VOLCANIC ENVIRONMENTS
FIG. 10-14 SEM secondary electron images of opened fluid inclusions from syenite xenolith of Ponza island (from Belkin et al. 1996). Potassium feldspar is the host phase for all images. A. Opened inclusion containing NaCl (H and crystal indicated with arrow), K2SO4 (arcanite) (A). Glauberite was also noted as small crystals to the left of the large NaCl. B. Opened inclusion containing glauberite (G), arcanite (A), and NaCl (H). Also noted was a small KC1 crystal. C. Opened inclusion containing K2SO4 (arcanite) (A). The arrow indicates a small crystal of glauberite. D. Opened inclusion containing Na-rich potassium feldspar (K), Na2SO4 (thenardite), and glauberite (G). The arrow indicates K2SO4 (arcanite); the dark spot on the crystal is electron beam damage from an energy dispersive X-ray fluorescence determination.
Campi Flegrei samples (Fedele et al. 2006). Many studies on shallow intrusive rocks (Roedder 1984) report the occurrence of Cl-rich, high-density aqueous FI with Th and salinities in the same range as found for Ponza sample, i.e., >500°C and >40wt.% NaCl equivalent. These hypersaline/ S-rich FI are interpreted to have formed as immiscible fluids in equilibrium with silicate melt. The presence of high-temperature, hypersaline immiscible fluids has important implications for ore-metal transport. For example, experimental studies of Cu partitioning among vapor, liquid and a silicate melt at 1000 bar and 800°C (Williams et al. 1991), Fe, Pb and Zn at 500 bar and 500°C (Hemley et al. 1992), and Au at hydrothermal conditions (Simon 2003, Simon et al. 2003) showed that these metals partition between the vapor and the liquid in a manner similar to Cl. Model calculations at 0.6 and 1200 bar indicate that a portion of Cl in the system will be partitioned into the liquid phase particularly at lower pressure. This suggests that the majority of chloride-complexed metals exsolving from magma may be transported in the liquid phase
Magmatic/hydrothermal transition. The studied xenolith records the magmatic/hydrothermal transition only in the upper part of the magma chamber because the lower magma chamber cumulates are presumed to be predominantly mafic, e.g., pyroxene-rich. Magma chamber margins are of particular significance because they may record the various evolutionary processes during the crystallization of the magmatic system. The silicate melt + CO2 + H2O and silicate melt/hypersaline/Srich aqueous inclusions provide direct evidence of fluid exsolution and fluid entrapment. For instance, MI are commonly observed coexisting with hypersaline/S-rich aqueous inclusions along healed fractures. Subsequent processes of sub-solidus, hydrothermal fluid evolution are represented by secondary aqueous + CO2 fluid inclusions. The complex daughter crystal assemblages in the Ponza sample record high solute contents in the fluid(s) during entrapment (Roedder 1984, and references therein), and the inclusions show selective enrichments in Th, U, REE, Zr and other incompatible elements similar to that found in 229
B. DE VIVO, A. LIMA, V.S. KAMENETSKY & L.V. DANYUSHEVSKY
under shallow crustal conditions. Furthermore, Clrich fluids are extremely efficient in extracting trace elements and REE out of magmas (Webster et al. 1989, Candela & Piccoli 1995, Haas & Shock 1994, Kravchuk & Keppler 1994, Reed 1995).
in the precipitation of the final melt and the initial stages of aqueous evolution. Belkin (1992) reported baddeleyite forming in the temperature range 500 to 600°C, from high-salinity aqueous fluids exsolving during the last stages of magma crystallization in the ferrogabbro zone of a diabase. In this environment, baddeleyite formed by the replacement of zircon. The hydrothermal mobility of Zr is favored by the same conditions existing along the margins of the Ponza magma chamber in the magmatic/hydrothermal regime. Belkin et al. (1996) reported that in Ponza syenite, some Zr-bearing minerals have textures and petrographic features which are compelling evidence for late stage hydrothermal transport of Zr. The high level, trachytic composition magma chamber at Ponza was sufficiently evolved before eruption for complex immiscibility to develop at least along its margins. Trapping of multiple, complex silicate, salt-rich, S-rich, and/or CO2 fluids occurred as the chamber crystallized. The exsolution and coalescence of the supercritical fluid as “bubbles” formed an abundance of cavities in the magma that subsequently was broken up and carried to the surface as the alkali feldspar xenolith of this study. The mineralogy and mineral chemistry of the cavities are typical of high temperature magmatic environments associated with magmatichydrothermal aqueous fluids. The occurrence of REE-, U-, Th-, Y-, Nb-, and Zr-rich phases is also compatible with residual fluids exsolved from a cooling alkaline magma, whereas the common occurrence of S-bearing daughter phases may represent an assimilated S-bearing component.
Sulfur-rich solutions and REE, U, Th, Y, Nb and Zr mineralization. Although metal transport by chloride-complexes has been demonstrated by modeling and experiments and abundant FI studies have reported the occurrence of high-salinity inclusions in major ore deposits (Roedder 1984, and references therein), the role of S in metal transport associated with sulfide-dominated ores (the bulk of ore deposits) is usually not constrained due to the lack of relevant data. Also, this bears on the studied xenolith because if it formed at relatively oxidized conditions (e.g., supported by the occurrence of magnetite, titanite, sulfates and probable hematite daughter crystals), there is no evidence of the redox state of the magmatic solutions prior to their passage into the magma chamber wall environment. Sulfate-enriched fluids are common in orthomagmatic fluids associated with porphyry copper deposits (Halter & Heinrich 2006, Roedder 1984) and have also been described from peralkaline silicic intrusions (Hansteen & Burke 1990). REE, U, Th, and Nb mineralization was prominent in the xenolith as evidenced by their relatively high levels of abundance in late-stage minerals such as apatite, titanite, and Zr-bearing minerals (see above) as well as rare urarinite– thorianite. The Zr-bearing minerals baddeleyite and especially zirkelite are unusual, although these minerals may be reported more frequently given the advent of new microanalytical techniques that are currently available. Similarly, Gianfagna (1985) reported baddeleyite in an ejected block which may represent a piece of magma chamber wall from Colle Cimino, a scoria cone in the Alban Hill volcanic area, Italy. Baddeleyite is an accessory mineral also found in a wide variety of other environments, e.g., carbonatite (Paarma 1970), gabbro (Keil & Fricker 1974), skarn (Konev 1978), and diabase (Belkin 1992, Siivola 1977). Electron microprobe and SEM–EDS analysis reveals that those minerals, whose texture and petrography suggest late-stage formation, contain significant amounts of REE, U, Th, and other metals. This suggests that late-stage, hightemperature, high-solute aqueous fluids were transporting considerable REE, U, and Th. These incompatible elements reached high concentrations
SUMMARY AND CONCLUSIONS Magmatic phenomena can be studied directly from observations of inclusions in igneous minerals because FI and MI, trapped in igneous minerals crystallized at depth, record the changes in composition during magmatic differentiation. In recent years, inclusion studies have concentrated mostly on the evolution of aqueous fluids in subvolcanic systems. Such studies recognize that the majority of hydrothermal ore deposits (low sulfidation epithermal deposits) in volcanic rocks or sub-volcanic to plutonic suites formed within geothermal systems similar to those active today. The critical role played by magmatic hydrosaline chloride and/or S-rich liquids has become clear. Hydrosaline liquids represent the most Cl-enriched volatile phase that occurs in magmas, and exsolution of this phase has important 230
FLUID AND MELT INCLUSIONS IN SUB-VOLCANIC ENVIRONMENTS
consequences for processes of hydrothermal mineralization. The role of hydrosaline fluids is critical for element transport from the Earth’s interior to other, more surficial environments, such as the atmosphere, the hydrosphere, and the crustal hydrothermal systems where magmatic-hydrothermal ore deposits form. Melt evolution leading to magma immiscibility can be unambiguously investigated with trapped inclusions. Good examples of this approach are seen in the studies of alkaline granitoid (alkali syenite) xenoliths entrained in volcanic units in Italy, where evidence of the transition from magmatic to hydrothermal systems and distinct evidence of mineralization processes has been found at Campi Flegrei, Mt. Vesuvius, and on the islands of Ponza, Ventotene, and Pantelleria. The results of detailed studies carried out on cognate syenite nodules from these volcanic areas, document the existence of immiscibility between hydrosaline fluids and silicate melt and/or S-rich fluids and silicate melt. These studies also prove the presence of hydrothermal fluids of magmatic origin that are very similar to those reported in ancient geothermal systems (low sulfidation epithermal deposits). The FI and MI data can also be used to address the particular problem of frequent ground movements (bradyseism) in the Campi Flegrei volcanic region (De Vivo & Lima 2006). These ground movements have been explained previously by a classical model that involves the intrusion of new magma to shallow depth, or by models which emphasize both magmatic and aquifer effects. De Vivo & Lima (2006) describe a new model that involves only hydrothermal fluids, of magmatic or meteoric/marine origin, with no direct involvement of the magma, other than as a heat source to explain the ground deformation. They explain the bradyseism by a purely hydrothermal model, using processes in porphyry systems (Burnham 1979, Fournier 1999, Henley & McNabb 1978) as an analogue to those to the Campi Flegrei. In this view, Campi Flegrei might very well represent a modern analogue behaving physically like a mineralized porphyry system, as it has been demonstrated for White Island, New Zealand (Rapien et al. 2003). FI and MI data from Campi Flegrei and other volcanoes of the Neapolitan area (Vesuvius, Ponza and Ventotene) also demonstrate a linkage with low sulfidation epithermal deposits and some porphyry systems. FI in these volcanic systems show clear evidence of various stages of silicate melt–hydrosaline melt–aqueous fluid–CO2 immisc-
ibility during magmatic evolution providing evidence of the transition from the magmatic to hydrothermal stage, which is comparable to the plastic, lithostatic domain in porphyry systems. ACKNOWLEDGEMENTS The authors are grateful to M.L. Frezzotti (University of Siena, Italy) and H.E. Belkin (U. S. Geol. Survey, Reston, VA, USA) for critically reviewing a preliminary version of the manuscript, and to A. Audétat (Universität Bayreuth, Germany), A.E. William-Jones (McGill University, Montreal, Canada) and J.D. Webster (AMNH, New York, USA), whose careful reviews and suggestions helped to improve the final version of the manuscript. REFERENCES ANDERSON, A.T. JR., DAVIS, A.M. & LU, F. (2000): Evolution of Bishop Tuff rhyolitic magma based on melt and magnetite inclusions and zoned phenocrysts. J. Petrol. 41, 449–473. ANDRE-MAYER, A.S., BAILLY, L., LEROUGE, C., CHAUVET, A., LEROY, J. & MARCOUX, E. (2005): Constraints on the ore fluids in the Sando Alcalde Au–Ag epithermal deposit, southwestern Peru: Fluid inclusions and stable isotope data. Comptes Rendus Géoscience 337(8), 745–753. AUDÉTAT, A., GÜNTHER, D. & HEINRICH, C.A. (2000): Magmatic-hydrothermal evolution in a fractionating granite: a microchemical study of the Sn–W–F-mineralized Mole Granite (Australia). Geochim. Cosmochim. Acta 64, 3373–3393. AUDÉTAT, A. & PETTKE, T. (2003): The magmatichydrothermal evolution of two barren granites: A melt and fluid inclusion study of the Rito del Medio and Cañada Pinabete plutons in northern New Mexico (USA). Geochim. Cosmochim. Acta 67, 97–121. BARBERI, F., CORRADO, G., INNOCENTI, F. & LUONGO, G. (1984): Phlegraean Fields 1982– 1984: brief chronicle of a volcano emergency in a densely populated area. Bull. Volcanol. 47 (2), 175–185. BEANE, R.E. (1983): The magmatic–meteoric transition. Geothermal Resource Council Special Report 13, 245–253
231
B. DE VIVO, A. LIMA, V.S. KAMENETSKY & L.V. DANYUSHEVSKY
BEANE, R.E. & BODNAR, R.J. (1995): Hydrothermal fluids and hydrothermal alteration in porphyry copper deposits. Arizona Geol. Soc. Digest 20, 83–93.
geologic systems: equation of state for H2O– CO2–NaCl fluids at high pressures and temperatures. Geochim. Cosmochim. Acta 47(7), 1247–1275.
BEANE, R.E. & TITLEY, S.R. (1981): Porphyry copper deposits. Part II. Hydrothermal alteration and mineralization. Econ. Geol. 75th Anniversary Vol., p. 235–263.
BROWN, P.E. & LAMB, W.M. (1989): P–V–T properties of fluids in the system H2O + CO2 + NaCl: new graphical presentations and implications for fluid inclusion studies. Geochim. Cosmochim. Acta 53, 1209–1221.
BELKIN, H.E. (1992): Baddeleyite (ZrO2), a hydrothermal mineral in Mesozoic mafic intrusive rocks, eastern North America: implications for aqueous transport and solubility of zirconium. V.M. Goldschmidt Conference, program and abstracts, 8–10 May 1992, Reston, VA, A–7.
BURNHAM, C.W. (1979): Magmas and hydrothermal fluids, In: Geochemistry of Hydrothermal Ore Deposits, Barnes, H.L., ed., John Wiley & Sons, New York, 71–136. CANDELA, P.A. (1986): Towards a thermodynamic model for the halogens in magmatic systems; An application to melt–vapor–apatite equilibria. Chemical Geology 57, 289–301.
BELKIN, H.E., CAVARRETTA, G. & TECCE, F. (1994): Petrogenesis of leucite-bearing nodules from Colli Albani Volcano ejecta, Latium, Italy. Abstr. Internat. Mineral. Assoc., 16th General Meeting, 4–9 September 1994, Pisa, Italy, p. 35.
CANDELA, P.A. (1989): Calculation of magmatic fluid contributions to porphyry-type ore systems: predicting fluid inclusion chemistries. Geoch. J. 23, 295–305.
BELKIN, H.E. & DE VIVO, B. (1993): Fluid inclusion studies of ejected nodules from Plinian eruptions of Mt. Somma–Vesuvius. J. Volcanol. Geoth. Res. 58, 89–100.
CANDELA, P.A. & PICCOLI, P.M. (1995): Model ore-metal partitioning from melts into vapor and vapor/brine mixtures. In Magmas, fluids and ore deposits. Min. Assoc. Canada Short Course 23, 101–127.
BELKIN, H.E., DE VIVO, B., LIMA, A. & TÖRÖK, K. (1996): Magmatic silicate/saline/sulfur rich/CO2 immiscibility and zirconium and REE enrichment from alkaline magma chamber margins: evidence from Ponza Island and Pontine archipelago, Italy. Eur. J. Mineral. 8, 1401–1420.
CAPITANIO, F. & MOTTANA, A. (1998): Sanidine holocrystalline ejecta from central Sabatini Volcanic District, Latium (Italy), II Intergranular ejecta and minerogenetic deductions. Atti della Accademia Nazionale dei Lincei. Rendiconti Lincei. Scienze Fisiche e Naturali. Serie 9, 9(2), 131–143.
BELKIN, H.E., DE VIVO, B., ROEDDER, E. & CORTINI, M. (1985): Fluid inclusion geobarometry from ejected Mt Somma–Vesuvius nodules. Am. Mineral. 70, 288–303.
CAPRARELLI, G., TSUTSUMI, M. & TURI, B. (1997): Chemical and signatures of the basement rocks from the Campi Flegrei geothermal field (Naples, southern Italy): inferences about the origin and evolution of its hydrothermal fluids. J. Volcanol. Geoth. Res. 76, 63–82.
BODNAR, R.J (1992): Can we recognize magmatic fluid inclusions in fossil systems based on room temperature phase relations and microthermometric behavior? Rept Geol. Survey Japan 279, 26–30. BODNAR, R.J. (1994): Synthetic Fluid Inclusions: XII. The system H2O–NaCl. Experimental determination of the halite liquidus and isochores for a 40 wt.% NaCl solution. Geochim. Cosmochim. Acta 58, 1053–1063.
CIONI, R., CIVETTA, L., MARIANELLI, P., METRICH, N., SANTACROCE, R. & SBRANA, A. (1995): Compositional layering and syn-eruptive mixing of a periodically refilled shallow magma chamber: the AD 79 Plinian eruption at Vesuvius. J. Petrol. 36, 739–776.
BOWERS, T.S. & HELGESON, H.C. (1983): Calculation of the thermodynamic and geochemical consequences of nonideal mixing in the system H2O–CO2–NaCl on phase relations in
CIONI, R., MARIANELLI, P. & SANTACROCE, R. (1998): Thermal and compositional evolution of the shallow magma chambers of Vesuvius:
232
FLUID AND MELT INCLUSIONS IN SUB-VOLCANIC ENVIRONMENTS
evidence from pyroxene phenocrysts and melt inclusions. J. Geophys. Res. 103, 18277–18294.
DE VIVO, B., LIMA, A. & WEBSTER, J.D. (2005) Volatiles in magmatic-volcanic systems. Elements 1, 19-24.
CLINE, J.S. & BODNAR, R.J. (1991): Can economic porphyry copper mineralization be generated by a typical calc-alkaline melt? J. Geophys. Res. 96, 8113–8126.
DE VIVO, B., ROLANDI, G., GANS, P.B., CALVERT, A., BOHRSON, W.A., SPERA, F.J. & BELKIN, H.E. (2001): New constraints on the pyroclastic eruptive history of the Campania volcanic Plain (Italy). Mineral. Petrol. 73, 47–65.
CLINE, J.S. & BODNAR, R.J. (1994): Direct evolution of brine from crystallizing silicic melt at Questa, New Mexico, molybdenum deposit. Econ. Geol. 89, 1780–1802.
DE VIVO, B., TOROK, K., AYUSO, R.A., LIMA, A. & LIRER, L. (1995): Fluid inclusion evidence for magmatic silicate/saline/CO2 immiscibility and geochemistry of alkaline xenoliths from Ventotene Island, Italy. Geochim. Cosmochim. Acta 59, 2941–2953.
CUNNINGHAM, C.G. & COROLLO, C. (1980): Modification of a fluid inclusion heating/freezing stage. Econ. Geol. 75, 335–337. DE NATALE, G., TROISE, C. & PINGUE, F. (2001): A mechanical fluid-dynamical model for ground movements at Campi Flegrei caldera. J. Geodyn. 32, 487–517.
DUAN Z. MOLLER, N. & WEARE, J. H. (1995): Equation of state for the NaCl–H2O–CO2 system: prediction of phase equilibria and volumetric properties., Geochim. Cosmochim. Acta 59(14), 2869–2882.
DE VIVO, B. (2006): Volcanism of the Campania Plain: Vesuvius, Campi Flegrei and Ignimbrites. Developments in Volcanology 9, Elsevier (in press).
FEDELE, L., TARZIA, M., BELKIN, H.E., DE VIVO, B., LIMA, A. & LOWENSTERN, J.B. (2006): Magmatic-hydrothermal fluid interaction and mineralization in alkali-syenite nodules from the Breccia Museo pyroclastic deposit, Naples, Italy. In Volcanism of the Campania Plain: Vesuvius, Campi Flegrei and Ignimbrites (De Vivo B., ed.) Developments in Volcanology, Elsevier (in press).
DE VIVO, B., BELKIN, H.E., BARBIERI, M., CHELINI, W., LATTANZI, P., LIMA, A. & TOLOMEO, L. (1989): The Campi Flegrei (Italy) geothermal system: a fluid inclusion study of the Mofete and san Vito fields. J. Volcanol. Geoth. Res. 36, 303– 326.
FEDERICO, M., PECCERILLO, A., BARBIERI, M. & WU, T.W. (1994): Mineralogical and geochemical study of granular xenoliths from the Alban Hills volcano, Central Italy: bearing on evolutionary processes in potassic magma chambers. Contrib. Mineral. Petrol. 115, 384– 401.
DE VIVO, B. & BODNAR, R.J., EDS. (2003): Melt Inclusions in Volcanic Systems. Methods, Applications and Problems. Developments in Volcanology 5, Elsevier, 258 p. DE VIVO, B., FREZZOTTI, M.L. & LIMA, A. (1993): Immiscibility in magmatic differentiation and fluid evolution in granitoid xenoliths at Pantelleria: fluid inclusion evidence. Acta Vulcanol. 3, 195–202.
FOURNIER, R. O. (1999): Hydrothermal processes related to movement of fluid from plastic into brittle rock in the magmatic-epithermal environment. Econ. Geol. 94(8), 1193–1211.
DE VIVO, B., FREZZOTTI, M.L. & MAHOOD, G. (1992): Fluid inclusions in xenoliths yield evidence for fluid evolution in peralkaline granitic bodies at Pantelleria (Italy). J. Volcanol. Geotherm. Res. 52, 295–301.
FROST, B.R. & TOURET, J.L.R. (1989): Magmatic CO2 and saline melts from the Sybille monzosyenite, Laramie Anorthosite complex, Wyoming. Contrib. Mineral. Petrol. 103, 178– 186.
DE VIVO, B. & LIMA, A. (2006): A hydrothermal model to explain the ground movements (bradyseism) at Campi Flegrei. In Volcanism of the Campania Plain: Vesuvius, Campi Flegrei and Ignimbrites (De Vivo B., ed.) Developments in Volcanology 9, Elsevier (in press).
FULIGNATI, P., GIONCADA, A., MAINERI, C. & SBRANA, A. (1997): A first insight into the magmatic-hydrothermal system related to 79 AD eruption magmatic chamber of Vesuvius (Italy). XIV ECROFI, Nancy, Abstract, 115–116. 233
B. DE VIVO, A. LIMA, V.S. KAMENETSKY & L.V. DANYUSHEVSKY
HALTER, W.E. & HEINRICH, C.A. (2005): Magmatic processes and volatile phase generation in porphyry-type environments: A laser ablation– ICP–MS study of silicate and sulfide melt inclusions. In Melt Inclusions in Plutonic Rocks (J.D. Webster, ed.) Min. Assoc. Can. Short Course 36, 151-164.
FULIGNATI, P., GIONCADA, A. & SBRANA, A. (1998a): The 79 AD Vesuvius magma chamber: a SEM–EDS study of daughter minerals in hypersaline fluid inclusions from cognate syenites. N. Jb. Mineral. Mh. 9, 403–416. FULIGNATI, P., KAMENETSKY, V.S., MARIANELLI, P., SBRANA, A. & MERNAGH, T.P. (2001): Melt inclusion record of immiscibility between silicate, hydrosaline, and carbonate melts: applications to skarn genesis at Mount Vesuvius. Geology 29, 1043–1046.
HANSTEEN, T.H. & BURKE, E.A.J. (1990): Melt– mineral–fluid interaction in peralkaline silicic intrusions in the Oslo Rift, southeast Norway. II: High-temperature fluid inclusions in the Eikeren– Skrim complex. Nor. Geol. Unders. Bull. 417, 15–32.
FULIGNATI, P., MARIANELLI, P. & SBRANA, A. (1998b): New insights on the thermometamorphic-metasomatic magma chamber shell of the 1944 eruption of Vesuvius. Acta Vulcanol. 10, 47–54.
HEINRICH, C.A., PETTKE, T., HALTER, W.E., AIGNER-TORRES M., AUDÉTAT, A., GUNTHER, D., HATTENDORF, B., BLEINER, D., GUILLONG, M. & HORN, I. (2003): Quantitative multi-element analysis of minerals, fluid and melt inclusions by laser ablation–inductively coupled plasma–mass spectrometry. Geochim. Cosmoch. Acta 67(18): 3473–3496.
GHIORSO, M.S. & SACK, R.O. (1995): Chemical mass transfer in magmatic processes IV. A revised and internally consistent thermodynamic model for the interpolation and extrapolation of liquid–solid equilibria in magmatic systems at elevated temperatures and pressures. Contrib. Min. Petr. 119(2–3), 197–212.
HEMLEY, J.J., CYGAN, G.L., FEIN, J.B., ROBINSON, R.B. & D’ANGELO, W.M. (1992): Hydrothermal ore-forming processes in the light of studies in rock-buffered systems: I. Iron–copper–zinc–lead sulfides solubility relations. Econ. Geol. 87, 1– 22.
GIANFAGNA, A. (1985): Occurrence of baddeleyite – ZrO2 – in an ejected block from Colle Cimino, marino (Alban Hills, Italy). Per. Mineral. 54, 129–133.
HENLEY, R.W. & MCNABB, A. (1978): Magmatic vapor plumes and ground water interaction in porphyry copper emplacement. Econ. Geol. 73(1), 1–20.
GILG, H.A., LIMA, A., SOMMA, R., BELKIN, H.E., DE VIVO, B. & AYUSO, R.A. (2001): Isotope geochemistry and fluid inclusion study of skarns from Vesuvius. Mineral. Petrol. 73, 145–176.
HENLEY, R.W. & ELLIS, A.J. (1983): Geothermal systems ancient and modern: a geothermal review. Earth Sci. Rev. 19, 1–50.
GRANCEA, L., CUNEY, M. & LEROY, J.L. (2001): Mineralised versus barren intrusions: a melt inclusion study in Romania’s Gold Quadrilateral. C. R. Acad. Sci. Paris, Sciences de la Terre et des Planètes 333, 705–710.
HERMES, O.D. & CORNELL, W.C. (1978): Pterochemical significance of xenolithic nodules associated with potash-rich lavas of Somma– Vesuvius volcano. NSF Final Technical Report, Univ. of Rhode Island.
HAAS, J.R. & SHOCK, E.L. (1994): Predictions of rare earth element (REE) speciation in a simulated hydrothermal solution as a function of pressure, temperature, and fluid pH. EOS, Trans. AGU vol. 75, n.16, April 19, 1994/Suppl., 361.
HERMES, O.D. & CORNELL, W.C. (1981): Quenched crystal mush and associated magma compositions as indicated by intercumulus glasses from Mt. Vesuvius, Italy. J. Volcanol. Geotherm. Res. 9, 133–149.
HAAS, J.R., SHOCK, E.L. & SASSANI, D.A. (1995): Rare earth elements in hydrothermal systems: estimates of standard partial molal thermodynamic properties of aqueous complexes of the rare earth elements at high pressures and temperatures. Geochim. Cosmochim. Acta 59(21), 4329–4350
HOLLAND T. & BLUNDY J. (1994): Non-ideal interactions in calcic amphiboles and their bearing on amphibole–plagioclase thermometry. Contrib. Mineral. Petrol. 116(4), 433–447.
234
FLUID AND MELT INCLUSIONS IN SUB-VOLCANIC ENVIRONMENTS
B: Applied Earth Science. 110, 221–222. JORON, J.L., METRICH, N., ROSI, M., SANTACROCE, R. & SBRANA, A. (1987): Chemistry and petrography. In: Somma–Vesuvius. (R. Santacroce, ed.) CNR Quad. Ric. Sci. 114, 105–174.
KONEV, A.A. (1978): Les mineraux de titane et de zirconium dans les skarns du massif alcalin de Tagerane (Baikal, Siberie). Bull. Minéral. 101, 387–389.
KAMENETSKY, V.S., DE VIVO, B., NAUMOV, V.B., KAMENETSKY, M.B., MERNAGH, T.P., VAN ACHTERBEGH, E., RYAN, C.G. & DAVIDSON, P. (2003): Magmatic inclusions in the search for natural silicate–salt melt immiscibility: methodology and examples. In: Melt Inclusions in Volcanic Systems. Methods, applications and problems (B. De Vivo and R.J. Bodnar R, Eds). Elsevier, Developments in Volcanology 5, 65–82.
KRAVCHUK, I.F. & KEPPLER, H. (1994): Distribution of chloride between aqueous fluid and felsic melts at 2 kbars and 800°C. Eur. J. Mineral. 6, 913–923. LASSITER, J.C., HAURI, E.H., NIKOGOSIAN, I.K. & BARSCZUS, H.G. (2002): Chlorine–potassium variations in melt inclusions from Raivavae and Rapa, Austral Islands: constraints on chlorine recycling in the mantle and evidence for brineinduced melting of oceanic crust. Earth Planet. Sci. Lett. 202, 525–540.
KAMENETSKY, V.S., NAUMOV, V.B., DAVIDSON, P., VAN ACHTERBERGH, E. & RYAN, C.G. (2004): Immiscibility between silicate magmas and aqueous fluids: a melt inclusion pursuit into the magmatic-hydrothermal transition in the Omsukchan Granite (NE Russia). Chem. Geol. 210, 73–90.
LOWENSTERN, J.B. (1993): Immiscibility between silicate melt, vapor and hydrosaline melt (~70–80 wt.% NaCl) in peralkaline rhyolites from Pantelleria, Italy. EOS 74, 670. LOWENSTERN, J.B. (1994): Chlorine, fluid immiscibility, and degassing in peralkaline magmas from Pantelleria, Italy. Am. Mineral. 79, 353 369.
KAMENETSKY, V.S., WOLFE, R.C., EGGINS, S.M., MERNAGH, T.P. & BASTRAKOV, E. (1999): Volatile exsolution at the Dinkidi Cu–Au porphyry deposit, Philippines: a melt–inclusion record of the initial ore-forming process. Geology 27, 691–694.
LOWENSTERN, J.B. (2003): Melt inclusions come of age: volatiles, volcanoes, and Sorby’s Legacy. In: Melt inclusions in volcanic systems. (R.J. Bodnar & B. De Vivo, eds.) Developments in Volcanology 5, Elsevier, 1–21.
KEIL, K. & FRICKER, P.E. (1974): Baddeleyite (ZrO2) in gabbroic rocks from Axel Heiberg Island, Canadian Arctic Archipelago. Am. Mineral. 59, 249–253.
LOWENSTERN, J.B. & MAHOOD, G.A. (1991): New data on magmatic H2O contents of pantellerites, with implications for petrogenesis and eruptive dynamics at Pantelleria. Bull. Volcanol. 54, 78– 83.
KELLEY, K.D., ROMBERGER, S.B., BEATY, D.W., PONTIUS, J.A., SNEE, L.W., STEIN, H.J. & THOMPSON, T.B. (1998): Geochemical and geochrnological constraints on the genesis of Au– Te deposits at Cripple Creek, Colorado. Econ. Geol. 93, 981–1012.
MANNING, D.A.C. & HENDERSON, P. (1984): The behavior of tungsten in granitic melt vapor system. Contrib. Mineral. Petrol. 86, 286–293.
KILINC, I.A. & BURNHAM, C.W. (1972): Partitioning of chloride between a silicate melt and coexisting aqueous phase from 2 to 8 kilobars. Econ. Geol. 67, 231–235.
MARIANELLI, P., MÉTRICH, N., SANTACROCE, R. & SBRANA, A. (1995): Mafic magma batches at Vesuvius, a glass inclusion approach to the modalities of feeding stratovolcanoes. Contrib. Mineral. Petrol. 120, 159–169.
KILLAS, S.P., NADEN, J., LENG, M.J. & CHELIOTIS, I. (2001): Fluid inclusion studies for identification of steam- and high-grade gold zones in lowsulphidation epithermal gold deposits; the Profitis Ilias Deposit, Milos Island, Greece. In: Mineral Deposits Studies Group annual meeting; abstracts. (Boulter & Clive, compilers), Institution of Mining and Metallurgy, Transactions, Section
MIGDISOV, A.A. & WILLIAMS-JONES, A.E. (2002): A spectrophotometric study of neodymium (III) complexation in chloride systems. Geochim. Cosmochim. Acta 66, 4311–4323.
235
B. DE VIVO, A. LIMA, V.S. KAMENETSKY & L.V. DANYUSHEVSKY
(H.L. Barnes, ed.), Wiley & Sons, 3rd Ed., New York, p. 657–698.
NEKVASIL, H. & BURNHAM, C.W. (1987): The calculated individual effects of pressure and water content on phase equilibria in the granite system. In: Magmatic Processes, Physicochemical Principles; a Volume in Honor of Hatten S. Yoder, Jr., Spec. Publ. of The Geochemical Society 1, 433–445.
ROEDDER, E. & COOMBS D.S. (1967): Immiscibility in granitic melts, indicated by fluid inclusions in ejected granitic blocks of Ascension Island. J. Petrol. 8, 417–451.
PAARMA, H. (1970): A new find of carbonatite in North Finland, the Sokli plug in Savukosli. Lithos 3, 129–133.
SAMSON, I, ANDERSON, A. & MARSHALL, D. (2003): Fluid Inclusions: Analysis and Interpretation. Mineral. Assoc. Canada, Short Course 32, 374 p.
RAIA, F., WEBSTER, J.D. & DE VIVO, B. (2000): Pre-eruptive volatile contents of Vesuvius magmas, constraints on eruptive history and behavior: I. The medieval and modern interplinian activities. Eur. J. Mineral. 12, 179– 193.
SASADA, M. (2000): Igneous-related active geothermal system versus porphyry copper hydrothermal system. Proceedings World Geothermal Congress 2000, Kyushu–Tohoku, Japan, May 28–June 10, p. 1691–1693 SCHMIDT, C. & BODNAR, R.J. (2000): Synthetic fluid inclusions: XVI. PVTX properties in the system H2O–NaCl–CO2 at elevated temperatures, pressures, and salinities. Geochim. Cosmoch. Acta 64(22), 3853–3869.
RAPIEN, M.H., BODNAR, R.J., SIMMONS, S.F., SZABO, C.S., WOOD, C.P. & SUTTON, S.R. (2003): Melt inclusion study of the embryonic porphyry copper system at White Island, New Zealand. Econ. Geol., Spec. Publ. 10, 41–59.
SHINOHARA, H. (1994): Exsolution of immiscible vapor and liquid phases from a crystallizing silicate melt: implications for chlorine and metal transport. Geochim. Cosmochim. Acta 58, 5215– 5221.
REED, M.J. (1995): Distribution of rare earth elements between aqueous fluid and granitic melt. Unpub. PhD diss. 1997. RENZULLI, A., UPTON, B.G.J. & NAPPI, G. (1995): Magma chamber processes preceding the Pitigliano Formation eruption (Latera volcanic complex, central Italy): evidence from cognate plutonic clasts. Acta Vulcanol. 7, 55–74.
SIGNORELLI, S. & CARROLL, M.R. (2000): Solubility and fluid–melt partitioning of Cl in hydrous phonolitic melts. Geochim. Cosmochim. Acta 64(16), 2851–2862.
ROEDDER, E. (1965): Liquid CO2 inclusions in olivine-bearing nodules and phenocrysts from basalts. Am. Mineral. 50, 1746–1782.
SIIVOLA, J. (1977): Baddeleyite – ZrO2 – from Lovasjarvi diabase, south-eastern Finland. Bull. Geol. Soc. Finland 49, 59–64.
ROEDDER, E. (1971): Fluid inclusion studies on the porphyry-type ore deposits at Bingham, Utah, Butte, Montana and Climax, Colorado. Econ. Geol. 66, 98–120.
SIMON, A.C. (2003): An experimental study to elucidate magnetite solubility and metal partitioning in a magnetite–silicate melt–vapor– brine–gold metal assemblage. Ph.D. thesis, Univ. of Maryland, College Park, MD, USA, 190 pp.
ROEDDER, E. (1972.): Fluid inclusions. In: The Encyclopedia of Geochemistry. Van NostrandReinhold, New York, 373–377.
SIMON, A.C., FRANK, M.R., PETTKE, T., CANDELA, P.A., HEINRICH, C.A., PICCOLI, P.M. & GLASCOCK, M.D. (2003): Experiments using prefractured quartz to trap volatile phases; can we believe the data? Geol. Soc. America, Abstracts with Programs 35(6), 356.
ROEDDER, E. (1984): Fluid inclusions. Rev. Mineral., Mineral. Soc. Amer. 12, 643 pp. ROEDDER, E.. (1992): Fluid inclusion evidence for immiscibility in magmatic differentiation. Geochim. Cosmochim. Acta 56, 5–20.
TARZIA, M., DE VIVO, B. & BELKIN, H.E. (2000): Occurrence of uranium, zirconium and REE bearing minerals in alkali syenite nodules from the Breccia Museo pyroclastic deposit, Naples, Italy. EOS 81, no. 48.
ROEDDER, E. & BODNAR, R.J. (1997): Fluid inclusion studies of hydrothermal ore deposits. In: Geochemistry of Hydrothermal Ore Deposits. 236
FLUID AND MELT INCLUSIONS IN SUB-VOLCANIC ENVIRONMENTS
hydrous chloride melt at Mt. Somma–Vesuvius. Am. Mineral. 87, 1046–1061.
TARZIA, M., LIMA, A., DE VIVO, B. & BELKIN, H.E. (1999): Uranium, Zirconium and Rare Earth Element enrichment in Alkali Syenite nodules from the Breccia Museo Deposit, Naples, Italy. Geological Society of America Annual Meeting, Abstracts with Programs Vol. 31, No 7, p. A-69.
WEBSTER, J.D., DE VIVO, B. & TAPPEN, C. (2003): Volatiles, magmatic degassing and eruptions of Mt. Somma–Vesuvius: constraints from silicate melt inclusions, solubility experiments and modeling. In: Melt inclusions in volcanic systems. (R.J. Bodnar & B. De Vivo, eds.) Developments in Volcanology 5, Elsevier, 207– 226.
THOMAS, R. (2000): Determination of water contents of granite melt inclusions by confocal laser Raman microprobe spectroscopy. Amer. Mineral. 85, 868–872.
WEBSTER, J.D., HOLLOWAY, J.R., & HERVIG, R.E. (1989): Partitioning of lithophile trace elements between topaz rhyolite melt and H2O and H2O + CO2 fluids. Econ. Geol. 84, 116–134.
THOMPSON, T.B., TRIPPEL, A.D. & DWELLEY, P.C. (1985): Mineralized veins and breccias of the Cripple Creek District, Colorado. Econ. Geol. 80, 1669–1688.
WEBSTER, J.D. & REBBERT, C.R. (2001): The geochemical signature of fluid-saturated magma determined from silicate melt inclusions in Ascension Island granite xenoliths. Geochim. Cosmochim. Acta 65, 125–136.
TOURET, J.L.R. & FREZZOTTI, M.L. (1993): Magmatic remnants in granites. Bull. Soc. Geol. Franc. II, 229–242. TURBEVILLE, B.N. (1992): Relationship between chamber margin accumulates and pore liquids: evidence from arrested in situ processes in ejecta, Latera Caldera, Italy. Contrib. Mineral. Petrol. 100, 429–441.
WEISBROD, A. (1981): Fluid inclusions in shallow intrusives. In: Fluid Inclusions: Applications to Petrology (L.S. Hollister & M.L. Crawford, eds.) Mineral. Assoc. Canada, Short Course 6, 241– 277.
VAGGELLI, G., BELKIN, H.E., DE VIVO, B. & TRIGILA, R. (1992): Silicate–melt inclusions in recent lavas (AD 1631–1944). I. Petrography and microthermometry. Eur. J. Mineral. 4, 1113– 1124.
WILLIAMS, T.J., CANDELA, P.A. & PICCOLI, P.M. (1991): An experimental determination of copper partitioning between water saturated high-silica rhyolite and a chloride-bearing two-phase aqueous mixture at 800°C and 1 kilobar. Geol. Soc. America, Abstr. Progr. 23, 291.
VILLEMANT, B. & BOUDON, G. (1999): H2O and halogen (F, Cl, Br) behaviour during shallow magma degassing processes. Earth Planet. Sci. Lett. 168(3–4), 271–286.
WOOD, S.A. (1990): The aqueous geochemistry of the rare-earth elements and yttrium 2. Theoretical predictions of speciation in hydrothermal solutions to 350°C at saturation vapor pressure. Chem. Geol. 88, 99–125.
WASHINGTON, H.S. (1906): The Roman Comagmatic Region. Carnegie Inst. Washington, Publ. 57, 199 p.
ZAJACZ, Z., HALTER, W.E., MALFAIT, W.J., BACHMANN, O., BODNAR, R.J., HIRSCHMANN, M.M., MANDEVILLE, C.W., MORIZET, Y., MUN ̈ TENER, O., ULMER, P. & WEBSTER, J.D. (2006) A composition-independent quantitative determination of the water content in silicate glasses and silicate melt inclusions by confocal Raman spectroscopy. Contrib. Mineral. Petrol. (in press).
WEBSTER, J.D. (1997): Exsolution of magmatic volatile phases from Cl-enriched mineralizing granitic magmas and implications for ore metal transport. Geochim. Cosmochim. Acta 61, 1017– 1029. WEBSTER, J.D. (2004): The exsolution of magmatic hydrosaline chloride liquids. Chem. Geol. 210, 33–48.
ZHANG Y. & FRANTZ, J.D. (1987): Determination of the homogenization temperatures and densities of supercritical fluids in the system NaCl–KCl– CaCl2–H2O using synthetic fluid inclusions. Chem. Geol. 64, 335–350.
WEBSTER, J.D. & DE VIVO, B. (2002): Experimental and modeled solubilities of chlorine in aluminosilicate melts, consequences of magma evolution, and implications for exsolution of 237
B. DE VIVO, A. LIMA, V.S. KAMENETSKY & L.V. DANYUSHEVSKY
238
