| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Petrology | Volume 45 | Number 3 | Pages 457-484 | 2004
Journal of Petrology 45(3) © Oxford University Press 2004; all rights reserved.
The Rare Earth Elements and Uranium in Garnets from the Beinn an Dubhaich Aureole, Skye, Scotland, UK: Constraints on Processes in a Dynamic Hydrothermal System
1 SCHOOL OF THE ENVIRONMENT, THE UNIVERSITY OF BRIGHTON, LEWES ROAD, BRIGHTON BN2 4GJ, UK
2 DEPARTMENT OF MINERALOGY, THE NATURAL HISTORY MUSEUM, CROMWELL ROAD, LONDON SW7 5BD, UK
3 DEPARTMENT OF EARTH SCIENCES, THE UNIVERSITY OF CAMBRIDGE, CAMBRIDGE CB2 3EQ, UK
RECEIVED OCTOBER 28, 2002; ACCEPTED JULY 25, 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONGEOLOGICAL SETTINGANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Garnets from skarns in the Beinn an Dubhaich granite aureole, Isle of Skye, Scotland, have a large range of concentrations of uranium (0·2–358 ppm) and the rare earth elements (REE) (23–4724 ppm). Variations in these concentrations correlate with major element zonation within the garnets, and with changes in the shape of REE patterns. Typical patterns in most garnets display light REE (LREE) enrichment, flat heavy REE (HREE) distribution and a negative Eu anomaly. These patterns are interpreted to represent equilibrium trace element exchange between pre-existing pyroxene, hydrothermal fluid and calcic garnets. Iron-rich zones are characterized by positive Eu anomalies and an increase in the abundance of the LREE relative to the HREE. These patterns are interpreted as resulting from changes in REE speciation related to the introduction of externally buffered fluid to the skarn system. Relatively Fe-poor zones show strongly HREE-enriched patterns with negative Eu anomalies and in some instances depletions in Y relative to Ho and Dy, which are interpreted as resulting from surface sorption of the REE during rapid, disequilibrium garnet growth. Strong correlations between U abundance and the REE patterns indicate that the same processes have affected U distribution. Both types of pattern can be modified by the effects of closed-system crystallization on REE abundance in the fluid, and changes in fluid major element chemistry.
KEY WORDS: fractionation; garnet; hydrothermal; rare earth elements; skarn
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTINGANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The behaviour of the rare earth elements (REE) and actinide elements during magmatism, metamorphism and metasomatism is of great interest because of their applications in geochronology and as geochemical tracers (e.g. reviews by Faure, 1986
; Lipin & McKay, 1989). The use of the REE as petrogenetic tracers in magmatic systems is well established, and rests on the basis that the chemistry of the trivalent REE elements is primarily controlled by the regular decrease in ionic radius with increasing atomic number. The application of the REE as tracers in hydrothermal systems, however, is more problematic.
Numerous attempts have been made to use REE patterns as tracers of fluid source, protolith type and water–rock interaction (e.g. Lottermoser, 1992; Fleet et al., 1997); skarn systems have been a particular focus for this type of study (e.g. Giuliani et al., 1987; Auwera & Andre, 1991). However, many studies have shown that hydrothermal processes may fractionate adjacent isovalent elements in the REE series from one another (e.g. Bau, 1991; Sherrell et al., 1999). These studies suggest that, at high temperature and pressure, the trends in the association constant of aqueous complex species across the group are not a regular function of atomic number (Haas et al., 1995). The irregularities are almost certainly a result of the role of electron configuration on bonding in aqueous species (e.g. Kawabe, 1992; Gramaccioli et al., 1999), an influence that has been termed the ‘lanthanide tetrad effect’ (e.g. Masuda et al., 1987; Bau, 1996).
Although these effects may complicate the application of elemental REE patterns as tracers of metal source in hydrothermal systems, they may offer the prospect of the utilization of REE element patterns as even more powerful indicators of hydrothermal processes if they are fully understood. The processes affecting inter-element fractionations in the REE series are also of great interest to geochronology and isotope geochemistry, particularly the potential for fractionation of Sm relative to Nd (Poitrasson et al., 1995, 1998).
In this study we report the variations in REE pattern across single garnet crystals from contact metasomatic skarns from the Beinn an Dubhaich granite aureole, Isle of Skye, Scotland (Fig. 1). These samples were selected on the basis of bulk analyses that had shown the skarns to be unusually enriched in both the REE and uranium, with garnets being the main host for both. Our aim here is to demonstrate the processes responsible for REE fractionation, and the utility of such fractionations in explaining the behaviour of the actinides in this environment.
|
|
GEOLOGICAL SETTING |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTINGANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The Beinn an Dubhaich granite is part of a Tertiary igneous complex in the Isle of Skye, Scotland (Fig. 1). The granite was intruded c. 54 Myr ago into siliceous dolostones (Balnakeil Formation) and dolomitic, chert-bearing, limestones (Croisaphuill Formation) of the Ordovician Durness Group (Moorbath & Bell, 1965
); initial crystallization temperatures were in the range 740–800°C, with late-stage crystallization at around 660–700°C (Ferry, 1985). The associated metamorphic aureole is estimated to have been produced at 280–640 bars and up to 650°C (Hoersch, 1981; Holness et al., 1989). During metamorphism the dolostones developed a prograde sequence of talc– tremolite–olivine–periclase, and the limestones developed a sequence of talc–tremolite–diopside–olivine (Holness, 1992). Extensive fluid advection took place through the carbonates during contact metamorphism, probably involving meteoric waters with a significant component of magmatic water (Holness & Fallick, 1997). Iron-rich skarn zones occur close to and at the contact of the granite and the carbonates, and are sometimes associated with basaltic dykes that are cut by the granite. The skarns are complex, consisting of a wide range of minerals of which several are the products of boron–fluorine metasomatism (Tilley, 1951). The aureole has long been a focus of research on the conditions of fluid flow during contact metamorphism (e.g. Tilley, 1951; Moorbath & Bell, 1965; Hoersch, 1981; Holness et al., 1989, 1992). REE enrichment is associated with several stages of skarn alteration. The samples examined in this study were taken from three sites within 2–5 m of the granite contact (Fig. 1). In two cases the samples consist of an assemblage of garnet + clinopyroxene + magnetite, with the main distinguishing feature being the amount of magnetite present (low in SK-84-20; high in SK-84-22). Sample SK-84-5 contains an assemblage of garnet + white mica + fluorite + calcite + magnetite.
|
ANALYTICAL METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTINGANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Samples were initially examined optically and using the back-scattered electron (BSE) mode of a Hitachi S2500 scanning electron microscope, at an accelerating voltage of 20 kV. Electron microprobe analyses of the main skarn silicates were carried out at the Natural History Museum, London, using a Cameca SX50 electron microprobe at an accelerating voltage of 15 kV and a beam current of 20 nA, using a range of natural and synthetic mineral standards including wollastonite (Ca, Si), jadeite (Na), corundum (Al), rutile (Ti), potassium bromide (K), magnesium oxide (Mg), and iron and manganese metal. The results are summarized in Table 1.
|
Uranium distribution maps were obtained using induced fission-track analysis, the principles of which have been given by Fleischer et al. (1975)
. Onto each polished thin section of the samples studied was affixed a 100 µm thick sheet of Lexan polycarbonate, which acted as an external solid state nuclear track detector. Each sample and detector couple was irradiated to a total thermal neutron dose of c. 1 x 1017 neutrons/cm2. Standard Reference Material standard glasses 612, 614 and 616 were simultaneously irradiated with the samples and detectors. After irradiation, the polycarbonate was detached from the thin sections (and standard glasses) and etched with 6 M NaOH at 70°C for 10 min to reveal the induced fission tracks.
Bulk-rock and mineral trace element data were obtained on mineral separates and bulk-rock powders using instrumental neutron activation analysis (INAA) at the Natural History Museum, London, and employing a solid state intrinsic Ge detector (particularly for the REE, Henderson & Williams, 1981), and a Ge(Li) detector following the technique of Williams & Wall (1991). The results are shown in Table 2.
|
The uranium distribution in samples investigated using fission-track analysis was studied quantitatively by ion microprobe at the University of Cambridge. The instrument used (Long & Gravestock, 1988
) was coupled to a Hratos ‘CONCEPT’ mass spectrometer (30 cm radius magnetic sector) and operated with an unfiltered O- beam of 
20 µm diameter and a current of 5 nA. All measurements were made at high mass resolution (M/
M 9000) with intensities recorded by integration over 11 channels across the limited flat tops (M/30000) of the mass peaks. Uranium was measured as 254UO+ and the REE as M+. Calcium and silicon were used as internal references. All measured concentrations were directly referred to NIST 910 glass without matrix corrections. The results are shown in Table 3.
|
Laser ablation–inductively coupled plasma mass spectrometry (LA–ICP-MS) analyses of garnet and pyroxene were carried out using a VG Plasmaquad 3 system with enhanced sensitivity S-option interface, and a New Wave Research LUV Microprobe II frequency quintupled Nd:YAG laser operating at 213 nm at the Natural History Museum, London (Jeffries, 2001
). Samples were ablated at a pulse energy of 0·3 mJ at 10 Hz. Spot diameter was 30 µm. 43Ca was used as an internal standard, the concentrations of which were measured by electron microprobe analysis. Calibration was performed using the National Institute of Science and Technology (NIST) standard SRM 612. Accuracy was monitored by repeated analysis of the US Geological Survey (USGS) standard SRM BCR-2G, a fused basalt from the Columbia River (Table 4). Limits of quantification were calculated at 10
of the mean background count and all data were filtered at 20. The results are summarized in Table 4. The full dataset is available as an Electronic Appendix (Appendix 1) at http://www.petrology.oupjournals.org.
|
Fluid inclusion microthermometry was carried out using a Linkam TMS91 heating–freezing stage at Imperial College, London, calibrated at -56·6, 10·0, 30·8 and 294°C using synthetic H2O–CO2 fluid inclusion standards. Accuracy is estimated at ±0·2°C in the range -100 to +30°C and ±0·5°C at higher temperatures. All measurements were made during heating runs to avoid problems of metastability. At low temperatures (<30°C) heating rates of 0·5°C/min were used for measurements of first melting temperature (Tfm), final ice melting temperature (Tmice) and hydrohalite melting temperature (Tmhh); rates of 5°C/min were used for measurements of liquid–vapour homogenization temperature (Th L–V) and halite and sylvite dissolution temperatures (Tsol halite, Tsol sylvite). For the measurement of the high Th of some inclusions a Linkam TM1500 stage was used at the University of Kingston. To ensure consistency between the results from the different stages Tsol halite in inclusions in pyroxene were measured using both instruments, and were typically repeatable within ±5°C or better. Salinity was calculated from ice melting temperatures (Tmice) using the equation of Bodnar (1993)
, and from Tsol halite using the equation of Sterner et al. (1988). The results are summarized in Table 5.
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTINGANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Major element chemistry and garnet zonation
The major element chemistry of the main silicate minerals of the studied skarn samples is shown in Fig. 2. Clinopyroxene is typically Mg-rich and dominated by the diopside end-member. All the garnets studied are dominated by the grossular–andradite solid solution with only minor amounts of Mn and Mg present, and show major element zonation, indicated by the variations in grey scale (mean atomic number contrast) in BSE images (Figs 3 –5), in the relative proportions of Fe(III) and Al (Table 1), with the exception of apparently homogeneous garnets from sample SK-84-5 (Fig. 4c). All garnets from sample SK-84-20 display a consistent zonation pattern. This consists of a homogeneous core, which has subsequently been fractured, then contains several stages of fracture fill and overgrowth, culminating in an oscillatory zoned rim (Fig. 3). In virtually all cases the oscillatory zoned rim includes an Fe-rich zone with evidence of dissolution of early garnet on its inner margin (Fig. 3c).
|
|
|
|
Sample SK-84-5 contains three types of garnet distinguishable on the basis of their optical properties. The earliest stage is brown in transmitted light, anhedral, shows weak zonation in BSE images, and has skeletal rims, possibly as a result of reaction with later-stage fluids (Type 1; Fig. 4a). This is followed by euhedral green garnet, which is either chemically homogeneous (Type 2a; Fig. 4c), or displays well-developed oscillatory zonation in BSE images (Type 2b; Fig. 4e).
Garnet from sample SK-84-22 typically has a relatively unzoned core, with an oscillatory zoned rim (Fig. 5a). However, in this case the transition is continuous with no evidence for brecciation or fracturing (Fig. 5b). Iron-poor zones, with evidence for dissolution of the preceding zones, occur within the oscillatory zoned rim (Fig. 5d).
Trace element analysis
Induced fission-track images (Fig. 6) clearly indicate a close correlation of the uranium content of garnet with major element (e.g. Al) zonation. The high density of fission tracks generated precludes direct quantification by point counting and comparison with standards; however, ion microprobe analyses on sample SK-84-20 indicate the range of U contents represented by the fission-track densities (Table 3), and these correspond well to the range of values measured using LA–ICP-MS. Difficulties with polyatomic interferences for the heavy REE (HREE) during ion microprobe analysis limited the number of REE that could be analysed to La, Ce and Pr. The behaviour of U and of the REE was therefore investigated in more detail using quadrupole LA–ICP-MS.
|
element maps (right panels) of skarn garnets from: (a) sample SK-84-20; (b) sample SK-84-5; (c) sample SK-84-22. Circles and line in (c) refer to the sites of spot analyses and the line transect in Significant problems with microanalysis of mineral zonation arise from both the diameter of the analysis pit in relation to the scale of zonation, and the angle of the zonation to the imaged surface. For LA–ICP-MS analyses flat, time-resolved, mass spectra indicated no major changes in concentration during ablation, and therefore that a single homogeneous zone, or group of zones, was analysed at each point. Recent X-ray and transmission electron microscopy (TEM) investigations of chemically zoned garnets have shown that oscillatory zonation may occur down to the tens of nanometres scale (Pollok et al., 2001
). This very fine-scale heterogeneity problem was avoided where possible, by concentrating analyses on homogeneous domains within garnet, or on packages of oscillatory zones within a larger-scale, non-oscillatory zoned structure. In particular, we have focused on zones with evidence of dissolution of earlier garnet on their inner margins (Figs 3–5). These boundaries mark discontinuities during growth, rather than the products of self-organization as a result of enrichment–depletion cycles at the mineral surface during growth from a supersaturated solution (Ortoleva et al., 1987).
Rare earth element ICP-MS analyses from each of the samples studied were normalized to the chondrite values of Wakita et al. (1971). Y was included as a pseudolanthanide, and plotted between Dy and Ho. Each of the garnet samples displays distinct REE patterns, and changes in pattern in separate growth zones (Figs 3–5; Table 4). Overall the patterns are light REE (LREE) enriched with a normalized abundance peak at Pr, and a relatively flat HREE distribution. Eu shows both positive and negative anomalies, and Y is either enriched or depleted relative to Dy and Ho (Figs 3–5). In SK-84-20 the first stage of overgrowth (overgrowth 1) and the outer Fe-rich zone are markedly enriched in Ce and Pr, and have a positive Eu anomaly (Fig. 3b and d). In a nearby grain these differences are less marked, but can still be seen, confirming the correlation in zoning at the scale of the hand specimen (Fig. 3e and f).
In sample SK-84-5 the early brown garnet is relatively low in the REE (Fig. 4a and b), and typically shows a positive Eu anomaly. The patterns in the unzoned garnet show little or no Eu anomaly (Fig. 4d), but in the oscillatory zoned garnet are distinguished by a very marked positive Eu anomaly (Fig. 4f). Despite the major element zonation observed in BSE images, the REE patterns of the oscillatory zoned garnet are relatively constant.
In the single large garnet analysed from sample SK-84-22 the garnet core and the inner margin of the oscillatory zoned rim show relatively simple LREE-enriched patterns (Fig. 5c). The outer oscillatory zoned garnet also has highly LREE-enriched patterns, but with positive Eu anomalies. At the transition to the oscillatory zoned rim (Fig. 5b) an Fe-poor zone (labelled core–rim boundary zone) shows patterns in which La, Ce and Lu are depleted relative to the normalized abundance of most other REE, and both Eu and Y display marked negative anomalies (Fig. 5c). Two Fe-poor zones cut the rim, in one case with good evidence for dissolution of earlier garnet on its inner rim (Fig. 5d). Both these zones have strong middle REE (MREE) to HREE enrichment relative to La and Ce with strong negative Eu anomalies (Fig. 5e).
All the garnets studied contain U at or above the ppm level (Table 4). In samples SK-84-20 and SK-84-22, U enrichments occur in Fe-rich and Fe-poor zones, respectively, and correlate closely with the major element zonation (Fig. 7). In all three samples the U content is positively correlated with the 
REE content of the garnet (Table 6; Fig. 8). In sample SK-84-20 the U content shows strong positive correlations with the LREE and the magnitude of the positive Eu anomaly (Table 6). The high U contents (242–358 ppm) all occur in the Fe-rich zone in the oscillatory zoned rims of the garnets. In sample SK-84-5 the U concentration again correlates strongly with the LREE (La to Pr). Zoned and unzoned green garnets are enriched in U relative to the earlier brown garnet. In SK-84-22 the U correlates with all the REE except La. The negative correlation of U with the Y anomaly, i.e. Ycn/[0·5(Dycn + Hocn)], where cn refers to chondrite-normalized concentration, is a result of the high U values associated with the core–rim boundary zone (Table 6). Thorium occurs at trace levels in all garnets, and shows only very weak correlation with zoning or other variations in trace element chemistry (Tables 4 and 6).
|
|
|
Fluid inclusions
The results of fluid inclusion microthermometry indicate that skarn formation occurred over a wide range of temperature conditions. These are summarized in Table 5. Fluid inclusions in clinopyroxene contain an aqueous liquid plus several daughter crystals including halite, sylvite and an opaque phase, probably magnetite (Fig. 8a). Tsol for sylvite ranges from 137 to 205°C, and for halite ranges from 338 to 513°C. Th ranges from 580 to 840°C, with the mode for the full dataset at 606°C. Much of the range in Th may be due to stretching or leakage of the inclusions at the high temperatures of measurement.
Primary inclusions in garnet occur in bands parallel to the optically visible zonation (Fig. 9b) and contain aqueous liquid plus vapour (Lw + V) at room temperature (Fig. 9c). Secondary inclusions occur in trails cutting across the zonation. In sample SK-84-20 first ice melting temperatures (Tfm) range from -56 to -64°C, indicating a complex range of salts in solution, probably including significant divalent cations. Tmice ranges from -21 to -12°C in primary inclusions and from -14 to -13°C in secondary inclusions, corresponding to salinity ranges of 16–23 wt % NaCl eq. and 17–18 wt % NaCl eq., respectively. Th ranges from 341 to 380°C in primary inclusions and from 281 to 285°C in secondary inclusions. In sample SK-84-5 inclusions in garnet have a similar range in salinity (19–26 wt % NaCl eq.) but homogenize in the range 247–370°C, with a slight trend towards dilution with falling temperature. Primary and secondary inclusions in garnet from sample SK-84-22 have much lower salinity (0·9–7·8 wt % NaCl eq.) and Th (170–270°C).
|
Fluorite deposition forms the final paragenetic stage of skarn alteration in sample SK-84-5, and fluorite contains three distinct fluid inclusion populations defined on the basis of salinity. There are two Lw + V inclusion populations, a low-salinity one ranging from 0·5 to 2·4 wt % NaCl eq. and a higher-salinity one ranging from 12·3 to 12·9 wt % NaCl eq. The third population is halite-bearing, and ranges in salinity from
31 to 37 wt % NaCl eq. Within the halite-bearing inclusions halite always dissolved after L–V homogenization. Homogenization temperatures vary in the range 193–262°C, 285–332°C and 135–271°C in each of the three types, respectively. Experimental data are lacking for situations other than vapour-saturated halite dissolution and equivalent salinities of inclusions estimated by the above method may be slightly in error (Chou, 1987; Bodnar, 1994). |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTINGANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Fluid evolution and the conditions of skarn formation
The fluid inclusion data from clinopyroxene from the skarn localities studied here are in agreement with previous work, which indicated the formation of pyroxene in endo- and exoskarn from high-salinity fluids at
600–700°C (Smith et al., 2002, Fig. 10a and b) for pressures within the granite aureole slightly in excess of 640 bars (Hoersch, 1981). For the pyroxenes studied here the modal Th of 606°C suggests formation pressures of 700 bars, assuming a salinity of 50 wt % NaCl eq. and fluid inclusions trapped on the liquid–vapour curve (Fig. 10b). Errors are introduced into this estimate because of the assumption of a pure NaCl solution. Furthermore, the high temperatures and salinities and the complex mixtures of salts present in the inclusions are beyond the experimentally calibrated range of currently available equations of state and empirical equations for isochore location. On the basis of halite and sylvite dissolution temperatures, bulk salinities are more likely to be in the region of 58–60 wt % salts, with NaCl:KCl ratios of 1·68–1·95, assuming the system NaCl–KCl–H2O. These estimates are also likely to be in error, but insufficient data are available for more complex systems (in this case almost certainly including FeCl2 and CaCl2).
|
Fluid inclusions in garnet are indicative of a lower-temperature and lower-salinity fluid regime during garnet formation relative to pyroxene formation (Fig. 10a). The range in observed Tfm in garnet is indicative of a complex mixture of salts in solution, including divalent cations. Figure 10b illustrates isochores for modal Th for inclusions in garnet calculated using the equation of Zhang & Frantz (1987)
assuming a pure NaCl solution. For the pressure range inferred by Hoersch (1981), this suggests trapping temperatures of 370–410°C at salinities of 20–26 wt % NaCl eq. for primary inclusions in garnet from sample SK-84-20, from 315 to 340°C for primary inclusions in brown garnet from sample SK-84-5, and from 240 to 270°C at salinities of 5–8 wt % NaCl eq. for primary inclusions from sample SK-84-22. Secondary inclusions in samples SK-84-20 and SK-84-5 are inferred to have been trapped from 300 to 325°C and from 260 to 290°C. The inclusions in fluorite measured here probably indicate formation from a fluid with a salinity of 12–13 wt % NaCl eq. at 290–330°C, possibly with periods of boiling responsible for the trapping of high-salinity (31–37 wt % NaCl eq.) fluids. The low-salinity, low-temperature (193–262°C) population in fluorite (Fig. 9a), however, is probably a result of trapping of later-stage fluids along cleavage planes. The decrease in salinity between the skarn samples does not define a continuous trend (Fig. 10a), and suggests intermittent, pulsed fluid flow, exploiting different areas of the granite contact at different times, possibly in response to fracturing.
Phase relations in the system Ca–Fe–Si–O–H2O–CO2 are illustrated in Fig. 11. Equilibrium constants were calculated using the SUPCRT92 computer code (Johnson et al., 1992). The fugacity of CO2 was set at an arbitrary low value, to reflect the low CO2 solubility in saline aqueous fluids at low pressure (Bowers & Helgeson, 1983). For a fluid with the oxygen fugacity (fO2) externally buffered by a granitic assemblage (K-feldspar–annite–magnetite; KFAM), andradite would be stable over the entire temperature range inferred for these garnets.
|
Origins of zonation in garnet
In each of the samples studied here the zonation pattern can be described as an initial homogeneous garnet core, overgrown by oscillatory zoned garnet interrupted by zones with resorbed inner margins and more significant changes in chemistry (Figs 3 –5). An interpretation of isochemical, or near-isochemical growth controlled by diffusion in the local environment during the initial stages of contact metamorphism, followed by metasomatic growth as a result of the infiltration of an external fluid is entirely consistent with the textural evidence from the garnets (Yardley et al., 1991
; Jamtveit & Hervig, 1994). The ingress of fresh batches of granite-derived aqueous fluid led to major chemical overstepping of the andradite-forming equilibria, supersaturation and the formation of oscillatory zoned rims as a result of enrichment–depletion cycles in garnet components in solution during growth (Ortoleva et al., 1987; Jamtveit, 1991; Yardley et al., 1991; Jamtveit & Andersen, 1992; Jamtveit et al., 1993; Holten et al., 1997, 2000). In samples SK-84-20 and SK-84-22, the regular oscillatory zonation is interrupted by broad zones (typically around 150 µm) that display dissolution of earlier zones on their inner margins suggesting significant, externally controlled, changes in the chemical conditions at these stages. In SK-84-20 these zones are Fe-enriched compared with the surrounding garnet (Fig. 3a and c), whereas in SK-84-22 they are Fe-poor (Fig. 5d). In the absence of mechanisms such as boiling, such variations may occur as a result of alternation between internally and externally buffered fluid compositions during pulsed fluid flow (Yardley et al., 1991). The change in the relative iron content of the zones, with evidence of dissolution on their inner margins, probably arises from the change in iron content between the more saline fluids in samples SK-84-20 and SK-84-5, and less saline fluids in sample SK-84-22. The Fe content of skarn-forming fluids is closely related to their total salinity (Kwak et al., 1986).
REE fractionation mechanisms during garnet growth
The behaviour of trace elements during mineral crystallization has been reviewed in detail by a number of workers (e.g. McIntire, 1963). The full process of partition of trace elements between aqueous fluid and a growing mineral involves removal of elements from the fluid (including the breakdown of aqueous complexes), sorption onto a growth surface, incorporation of elements into the bulk mineral via a substitution mechanism and diffusion through the surface layer, and the solution of the substituted cations into the aqueous fluid. Thus the overall distribution coefficient (
) will be a function of fluid chemistry, surface complexation, crystal chemistry and the relative rates of mineral growth and diffusive re-equilibration in the bulk crystal. The distribution coefficient for the REE between the Ca site in andradite and an aqueous fluid can be defined so as to take into account the effects of fluid chemistry by equation (1) (McIntire, 1963):
 | (1) |
; Brugger et al., 2000). Assuming that for andradite garnets the REE substitute onto the eight-fold co-ordinated Ca-site, this can be illustrated using the equations below. For the isovalent substitution of Eu2+ for Ca2+
 | (2) |
The equilibrium constant for this reaction can be written as
 | (3) |
indicates the activity coefficient. At equilibrium the distribution coefficient [equation (1)] for Eu2+ can be related to the equilibrium constant (K2) for reaction (2) by
 | (4) |
However, for altervalent substitutions a charge balancing mechanism must be invoked. Assuming a charge balanced substitution involving trivalent REE and Na+, model reactions such as (5) must be postulated:
 | (5) |
)
 | (6) |
 | (7) |
 | (8) |
). If all activity coefficients in aqueous solution are also assumed to be unity, this equation reduces to
 | (9) |
With these assumptions, for isovalent substitutions (in this case for Eu2+) the distribution coefficients are essentially independent of the major cation chemistry of the solution, whereas for altervalent substitutions they are not. This provides one mechanism for fractionation of the REE during hydrothermal processes. The degree to which this affects the relative partitioning of Eu will also depend on the oxygen fugacity of the fluid (fO2) and hence the relative predominance of Eu2+ and Eu3+.
A distribution coefficient defined in terms of total concentrations in the fluid, as in equation (1), must also take into account the speciation of the trace element in the fluid. This has been demonstrated for crystallization of garnet from a melt by Van Westernen et al. (2000), who showed that the overall thermodynamic control on element partition into garnet is a function of both the lattice relaxation energy (Urel) following the insertion of the element into the mineral lattice, and the overall solution energy (Usol), which incorporates Urel, the dissociation energy of melt species and any energy change involved in their incorporation into the growing crystal. For mineral growth from aqueous fluids, distribution coefficients will, therefore, be controlled not only by the crystal chemistry of the mineral phase, but also by the formation of aqueous complexes, and their relative thermodynamic stability, expressed by the association constant ß. The formation of associated complexes in aqueous solution will therefore also exert a control on mineral–fluid trace element partitioning.
The kinetic steps involved in incorporation of trace elements into the mineral lattice begin with sorption onto the mineral surface. This will be controlled by a distribution coefficient (
) distinct from the distribution coefficient between fluid and the bulk crystal (D). For surface sorption, will approach D only for infinitesimal degrees of supersaturation, otherwise diffusive re-equilibration will be inhibited or prevented by rapid crystal growth rates, leading to the preservation of surface-controlled REE patterns within the crystal (McIntire, 1963; Möller 1998). Sorption processes are essentially controlled by Coulomb's Law and the charge–size ratio of the trace element in question (Bau, 1991). Such rapid growth rates will typically arise during periods of supersaturation, which will also favour the development of oscillatory zonation (Ortoleva et al., 1987). Additional fractionation effects may be introduced into this process depending on the crystallographic nature of available surface sites (e.g. Rakovan & Reeder, 1996).
In addition to these effects the REE pattern of a mineral crystallizing from a hydrothermal fluid will also be affected by changes in temperature, pressure and the total concentrations of the REE in the fluid. Fluid inclusion data indicate a relatively constant fluid salinity during crystallization of each garnet studied here, but do not constrain potential changes in the REE content of the fluid. These controls on mineral REE content are examined in detail below, with respect to the data presented above.
Equilibrium controls on distribution coefficients
No firm inferences can be made about the REE content of the skarn-forming fluid in these samples. Calculations based on the INAA analyses of bulk granite samples and the partition coefficients of Flynn & Burnham (1978) suggest that a granite-derived fluid should have a normalized REE pattern enriched in the LREE relative to the HREE, with a pronounced negative Eu anomaly (Fig. 12a). However, these experimental data were derived for relatively low-salinity fluids (5 wt % NaCl eq.), and fluid inclusion evidence indicates that garnet is forming at significantly lower temperatures than the solidification of the magma, and hence direct melt–aqueous fluid partitioning may no longer be controlling the fluid REE composition. During the extraction of the REE from source rocks during water–rock interaction, processes may be extremely variable, making the prediction of the fluid composition difficult. Leaching of the REE during water–rock interaction may lead to fluids with varying degrees of enrichment or depletion both in the total REE concentration and in individual elements, relative to the source rock (e.g. Lottermoser, 1992; Banks et al., 1994; Campbell et al., 1995). To make an initial assessment of fluid–mineral partitioning we assumed a fluid with an REE composition equivalent to granite, with the content of unanalysed REE interpolated between elements to give a smooth REE pattern. Apparent between representative garnet analyses and this model fluid composition are shown in Fig. 13. These values are not intended to represent actual values of , as the composition of the skarn-forming fluid is unknown, and probably changed during the genesis of the different garnets; nevertheless, they provide a basis for the discussion of mineral–fluid REE partioning within the Beinn an Dubhaich skarn system.
|
|
) for representative garnet analyses using There are no experimental data on the equilibrium partitioning of the REE between bulk grossular–andradite garnet and aqueous solutions, but inferences can be made on the basis of crystal chemistry and the thermodynamics of trace element substitution into a crystal lattice (e.g. Van Westernen et al., 2000
). Comparison between the effective ionic radius of Ca2+ and the trivalent REE in eight-fold co-ordination (Shannon, 1976) shows the closest correspondence in radii between Ca, Pr and Nd (Fig. 12b). Purely on the basis of spatial accommodation into the garnet lattice this would suggest that garnet–aqueous fluid partitioning should produce LREE-enriched chondrite-normalized patterns in grossular–andradite garnet (Nicolescu et al., 1998), typically with the form of an upward-facing parabola with a peak at Pr. This is also consistent with the models of Urel for grossular of Van Westernen et al. (2000), which indicated a minimum in the range La–Pr. The formation and breakdown of REE complexes in solution will also contribute to the thermodynamics of element substitution into hydrothermal garnet. Most models and experimental studies of REE speciation suggest that aqueous complexes of the HREE will be more strongly associated than those of the LREE (e.g. Wood, 1990; Haas et al., 1995). Thus in partitioning of the REE between grandite garnet and aqueous solution it seems likely that the pattern in should show a preference for the entry LREE into garnet, possibly with a preference for Pr entering the Ca-site. Modifications to this pattern of distribution coefficients will occur depending on fluid chemistry and hence REE speciation in solution. This interpretation matches available data from natural grandite garnets (Nicolescu et al., 1998).
An additional effect on the REE patterns of minerals arises from the variation in the oxidation state of Eu in natural systems. The formation of associated aqueous species with F- and Cl- brings the redox boundary between Eu2+ and Eu3+ predominance in aqueous solution into the oxygen fugacity and pH range of natural solutions (Sverjensky, 1984; Bau, 1991; Fig. 14a). The simulations of Urel by Van Westernen et al. (2000) suggest that the relaxation energy for Eu2+ entering the Ca-site in grossular may be lower than for the trivalent REE and so between grandite garnet and aqueous fluid may be higher for Eu2+ than for Eu3+.
|
. Calculated using SUPCRT92 (Johnson et al., 1992Chondrite-normalized REE patterns suggestive of partitioning controlled by crystal chemistry occur in samples SK-84-20 and SK-84-5 and in the garnet core of sample SK-84-22 (Figs 3b, d, f, 4b and 5c). For the simple system assumed for the calculation of apparent
for representative analyses of these areas (Fig. 13), the LREE are partitioned more favourably into garnet than the HREE, Eu is strongly partitioned into garnet, and La is less strongly partitioned than the other REE. Partitioning controlled by bulk crystal chemistry would be expected to occur during slow growth, with the fluid chemistry buffered by the pre-existing local mineral assemblage. This type of environment is inferred for the formation of the garnet cores (Figs 3b and 4a), and in this instance the REE potentially could be derived from the breakdown of pyroxene in the local environment as well as from the skarn-forming fluid (Figs 11 and 12). The slight negative Eu anomaly seen in these examples, despite the inferred preference for Eu2+ on the Ca-site in andradite garnet, is probably inherited from the skarn-forming fluid. Differences in the calculated values of D between the two samples are probably related to differences in temperature and fluid chemistry, which cannot be accounted for with the current dataset and these relatively simple calculations. The changes in apparent across zoned crystals are discussed below.
The inferred comparatively weak partitioning of La into garnet may arise from the greater mismatch in ionic radius between La and Ca than between Ce, Pr and Nd and Ca. Coupled with this, complexes of La3+ with Cl- and F- are predicted to be more strongly associated than those of Nd3+ and Sm3+ by the models of Haas et al. (1995) (Fig. 14b). Aqueous complex formation will inhibit the incorporation of trace elements into minerals. It should be noted, however, that the patterns of variation in shown in Fig. 13 (and used in crystallization models; Fig. 15c and f) are also a function of the assumed fluid composition used.
|
Kinetic controls on distribution coefficients
A significant number of the garnet analyses are enriched in the HREE relative to the patterns expected for partitioning between bulk garnet and hydrothermal fluid, most notably for the Fe-poor zones in garnet from sample SK-84-22 (Fig. 5). These zones also typically display marked negative Eu anomalies, and in at least one case a negative Y anomaly (Fig. 5c). The overall apparent
for a representative analysis of the core–rim boundary zone in sample SK-84-22 is shown in Fig. 13, where preferential uptake of the HREE relative to the LREE, and anomalously low partitioning of Eu are shown. Such patterns are interpreted as the result of control by surface sorption on the uptake of the REE during rapid growth from a supersaturated solution.
Bau (1991) made qualitative predictions regarding the mineral surface sorption behaviour of the REE in hydrothermal systems, suggesting that the degree of sorption would decrease with increasing ionic radius, rising temperature and decreasing ionic charge. These qualitative predictions were confirmed experimentally for the sorption of the trivalent REE onto iron oxy-hydroxides by Bau (1999). There are few data on sorption of the REE onto mineral surfaces at higher temperatures and at conditions relevant to this study. However, assuming similar sorption behaviour onto garnet surfaces, the patterns within the Fe-poor zones from sample SK-84-22 can be explained by sorption control on the incorporation of the REE into garnet during rapid growth at conditions that are far from equilibrium. Modifications to the predicted sorption patterns arise from the redox behavior of Eu, which was probably predominantly in the Eu2+ state, and the effects of REE complexation in the solution, which would be much greater at skarn-forming temperatures than at the ambient temperatures of the experimental studies. Bivalent Eu would be less strongly sorbed than the other REE, resulting in the preservation, and possible enhancement, in garnet of any negative Eu anomaly in the original fluid (Fig. 5c). The increasing stability of aqueous REE complexes with atomic number (Wood, 1990; Haas et al., 1995) accounts for the slight decrease in normalized REE abundance from Ho to Lu in the Fe-poor zones in sample SK-84-22, as the increased stability of the complexes would inhibit surface sorption.
Surface sorption controlled REE patterns will be preserved when the rate of mineral growth exceeds the rate of diffusive re-equilibration in the surface layer. Depending on the relative rates of growth and diffusion, the preserved patterns potentially may also have been modified by partial re-equilibration in the surface zone (see below).
Speciation control on distribution coefficients
Several zones, most notably the Fe-rich zones in sample SK-84-20 (Fig. 3d and f) and the oscillatory zoned garnet from sample SK-84-5 (Fig. 4f), have marked positive Eu anomalies, and the apparent values show an increase in LREE and partitioning into garnet in these zones (Fig. 13), in some instances with a peak in Pr abundance in the iron-rich zone (Fig. 3d). The inner margin of this zone is also marked by dissolution, and it thus seems likely that the change in REE pattern is related to a significant, externally controlled, change in fluid chemistry. The general trend of these changes occurs in both samples (Fig. 13), and so most probably represents the effects of processes during garnet crystallization rather than a change in fluid REE content during mineral growth.
The most plausible cause for an externally controlled shift in the fluid chemistry, without significant temperature change, is the introduction of a fresh batch of externally buffered fluid. Such a fluid, if buffered by the granite mineral assemblage, would have a slightly higher oxygen fugacity (Fig. 11) than a skarn assemblage containing both a hedenbergite component in diopside, and andradite. Given the high salinity of the skarn-forming fluid in samples SK-84-20 and SK-84-5, it would also have an elevated Fe content. Such a fluid would also, for a brief period, have externally buffered Na and Ca contents, leading to a shift in fluid composition; this would result in Eu2+ partitioning more favourably into garnet relative to Sm and Gd, producing the positive Eu anomaly. This is indicated by the sharp increase in the distribution coefficient for Eu estimated assuming a fluid REE composition identical to granite in Fig. 13. Such a shift is illustrated by the initial garnet composition in the fractional crystallization model for garnet growth in Fig. 15a (see below), where the initial crystallization corresponds to open-system conditions with an externally buffered fluid composition. This is apparent in both SK-84-20 and SK-84-5, although the absolute magnitude of the estimated varies between samples.
Externally buffered fluids are also likely to have a low pH relative to a skarn system retaining any relict calcite, or with pH buffered by skarn mineral assemblages. Simple models of changing REE speciation at different pH are shown in Fig. 14c and d, calculated at 300°C and saturated vapour pressure using PHREEQC Version 2 (Parkhurst & Appelo, 1999), and the LLNL database and solution parameters (Wolery, 1992). These are internally consistent with data from SUPCRT92 and Haas et al. (1995). The Cl content was fixed at 3·03 wt % to simulate a fluid similar to that seen in sample SK-84-22, as the salinities in the other samples give ionic strengths beyond the applicable range of the extended Debye–Hückel expression used in the program. The fluid was assumed to contain 50 ppm F. At basic to neutral pH the REE are dominantly present as hydroxide and oxy-complexes, with the high concentrations of oxy-complexes predicted because thermodynamic data for REE-hydroxide complexes with co-ordination numbers >1 are not included in any currently available thermodynamic database. Hydroxide complexes are actually most likely to predominate at high pH (Haas et al., 1995). As the pH drops the REE typically shift towards positively charged complexes with F and then Cl (Fig. 14c–e). For a shift from neutral to slightly acidic pH to an externally controlled pH in equilibrium with a granitic mineral assemblage (two feldspars, white mica and an alumino-silicate phase, pH modelled at 4·5), the LREE (illustrated by La and Nd in Fig. 14c and d) will dominantly be present as LnF2+, LnCl2+ and 
, whereas the HREE will dominantly be present as LnF2+ (illustrated for Lu in Fig. 14e). For fluoro-complexes the LREE are less strongly associated than the HREE, and the chloro-complexes are uniformly less strongly associated than the fluoro-complexes (Fig. 14b). Such a shift in speciation would thus favour the partitioning of the LREE into garnet, and inhibit the partitioning of the HREE both with the mineral surface () and in the bulk material (). Such trends in the change in association constant between zones are apparent in the calculated values of in both samples SK-84-20 and SK-84-5 (Fig. 13).
Simple models of garnet crystallization
The discussion on crystal-chemical and surface sorption control on mineral–fluid partitioning applies primarily to the garnet cores and the zones of highest REE abundance in the sample. The remainder of the data, most notably the oscillatory zoned rims of SK-84-20 and most markedly SK-84-22 (Figs 3d and 5e), show gradual depletion in the overall REE abundance relative to these patterns, accompanied by modification of the chondrite-normalized REE pattern. In any hydrothermal system where fluid flow is controlled by periods of quiescence between episodes of hydrofracturing and the influx of fresh batches of fluid (Sibson et al., 1988; Brugger et al., 2000), there will be periods of growth from a small, essentially isolated batch of fluid. The distribution coefficient as calculated will remain constant only if the fluid composition remains constant. This situation is likely to occur only following the initial introduction of a batch of externally derived fluid into the skarn system. Water–rock interaction (including the crystallization of garnet) will lead to changes in the oxygen fugacity, pH and major element chemistry of the fluid, particularly if the skarn system is periodically isolated and continues to crystallize under closed-system conditions. Closed-system crystallization can cause inter-element fractionations within the REE as well as a gradual depletion in the absolute concentration of some or all REE as a result of the dependence for altervalent substitutions of on the fluid composition. will change continuously throughout a batch of closed-system crystal growth (Brugger et al., 2000).
The closed-system crystallization of garnet from a batch of solution with no re-equilibration of crystals with the solution (a reasonable scenario for rapid hydrothermal growth) can be modelled using equation (4) to simulate for Eu2+, and equation (8) to simulate for the trivalent REE. The Rayleigh fractionation equation [equation (10)] can then be used to calculate the composition of garnet (Cgarnet) for a given fluid composition (Caq), at a particular crystallization fraction (F) [modified from McIntire (1963)]:
 | (10) |
This set of equations was used to model the closed-system crystallization of garnet from a 10 kg batch of hydrothermal fluid containing 20 wt % NaCl and 1 wt % Ca (comparable with SK-84-20), and from a 10 kg batch of hydrothermal fluid containing 2·5 wt % Ca and 5 wt % NaCl (comparable with SK-84-22). The Ca concentration used primarily affects the initial REE concentration in garnet. The concentration in each model was thus fixed to give an initial concentration close to the Fe-rich and Fe-poor zones in samples SK-84-20 and SK-84-22, respectively. The values used for sample SK-84-20 (Model 1; Fig. 15) are proportional to the concentrations of Ca in SW England granite-derived hydrothermal fluids determined by Smith et al. (1996) (4000–6000 ppm in 6–9 wt % NaCl eq. solutions). The values used for sample SK-84-22 (Model 2; Fig. 15) are significantly higher than this, but may be justified by the skarn setting. All trivalent REE were assumed to enter garnet on the Ca-site following equation (6). An initial fluid composition and set of D values were assumed and used to calculate a value for K6. This value was then used to calculate the variation in D with changing fluid composition during the crystallization of garnet, initially assuming an Na content of garnet for each value of F modelled. The values of D were then varied to provide closer simulation of the real garnet, and at the same time the Na concentration in garnet was altered iteratively until it agreed with the total molar concentration of the REE in garnet to within <0·05%. The procedure essentially follows that of Brugger et al. (2000) for modelling the hydrothermal crystallization of scheelite.
The results of these simple models are shown in Fig. 15. In both instances the model calculations can simulate the REE patterns of the studied garnets within reasonable limits. Differing values of the distribution coefficient between the two models are used to reflect the differences in processes inferred from the REE patterns of garnet in samples SK-84-20 and SK-84-22. These have been interpreted above as arising in the case of sample SK-84-20 from near-equilibrium growth of garnet, and hence a close approximation to the bulk crystal controlled distribution coefficient () of the REE. In the case of sample SK-84-22 they are interpreted as resulting from rapid growth during periods of supersaturation of garnet, and hence a close approximation to the surface sorption controlled partition coefficient (). The trends in distribution coefficient in each case roughly follow those predicted above. Distribution controlled by crystal chemistry in the bulk crystal and distribution controlled by surface complexation are end-member processes. The observed distribution coefficient (i.e. that suggested by the models used here) may be intermediate between the two, and controlled by the relative rates of diffusion in the bulk crystal and the crystal growth rate, with equalling D only at infinitesimal degrees of supersaturation [discussed in detail for continuous crystal growth by MacIntire (1963)]. Such intermediate distribution coefficients may account for the relatively small apparent differences between the modelled values of for the LREE and HREE in both these models and the apparent shown in Fig. 13.
The model of closed-system crystallization used to interpret the REE patterns of sample SK-84-22 reproduces the shift in the Eu anomaly from positive to negative relative to the adjacent elements, but does not reproduce the overall magnitude of the depletion in Eu seen with progressive crystallization within the sample. The models assume that only bivalent Eu occurs in the solution, with the change in Eu anomaly arising from the fact the observed 
will not change with fluid composition as it does for the other REE. However, in reality Eu probably occurs in solution in mixed valence, with Eu2+ predominating (Fig. 14a; Sverjensky, 1984). In such a situation the observed will actually be a result of both charge balanced (for Eu3+) and isovalent substitution into garnet. The depletion in the amount of Eu in the fluid caused by the altervalent substitution can account for the observed depletion in Eu at Eu2+/Eu3+ ratios in the fluid of around 1–1·5. The exact value of this ratio is dependent on the assumed for Eu2+ and Eu3+.
In each case only very low degrees of partial crystallization are invoked to account for the observed variation in the garnet REE patterns (a maximum of 9% in Fig. 15). This is a function of the total REE concentration in the fluid used in each model. Using higher values both lowers the values of and increases the fraction of crystallization (F) necessary to reproduce the observed patterns. All the modelling carried out here has been used with the purpose of illustrating the ability of the inferred mechanisms to produce the observed inter-element fractions in the REE. A more fully quantitative model must await more detailed experimental and theoretical work in the areas of surface and bulk distribution coefficients, the effects of speciation on mineral–aqueous solution partitioning, and a fuller knowledge of the REE content of the hydrothermal fluids.
The transport and deposition of U in hydrothermal fluids
Assuming that uranium occupies the Ca-site in andradite, then, on the basis of ionic radius, U(IV) is much more likely to substitute into garnet than U(VI). In eight-fold co-ordination, the radius of U(IV) is closest to those of Er3+, Tm3+ and Yb3+ (Shannon, 1976), but our data (Table 6) do not show an exclusive correlation with these elements. The positive correlation of U with the total REE content of garnet indicates that the uptake of U into garnet from the hydrothermal fluid was not controlled by equilibrium crystal chemistry alone, and the additional processes controlling the behaviour of the lanthanides discussed above also control the uptake of U in skarn-forming hydrothermal solutions. At high temperatures, with fO2 buffered at any level above the Ni–NiO buffer (approximately the KFAM buffer in Fig. 11), U in aqueous solution is most likely to be dominated by hydroxyl, uranyl, carbonate, chloride or fluoride species of U(VI), with carbonate complexes being the most stable (McLennan & Taylor, 1979; Keppler & Wylie, 1990, 1991; Bailey & Ragnarsdottir, 1994; Valsami-Jones & Ragnarsdottir, 1997a, 1997b).
The introduction of externally buffered fluid into the skarn system inferred from textural evidence and the REE patterns would also initially introduce higher concentration of U in the fluid. As noted above, U solubility is strongly fO2 dependent. An initially granite-buffered fluid coming into contact with a skarn assemblage containing andradite, magnetite and hedenbergite (consistent with the assemblage in the Beinn an Dubhaich skarns) would undergo a reduction in fO2 (Fig. 11), reducing U solubility and promoting U incorporation into minerals. Such a shift is entirely consistent with the correlations between U and REE patterns, and the observed garnet textures in samples SK-84-20 and SK-84-5, although U concentrations are lower in SK-84-5, possibly reflecting a smaller shift in fO2, as a result of the lack of pyroxene in this sample.
In sample SK-84-22, U correlates strongly with the total REE content, and with all the REE except La. The strong uptake of all the REE in this garnet is interpreted above as a result of surface sorption control on the incorporation of trace elements during periods with a high degree of supersaturation and hence rapid growth kinetics, leading to disequilibrium partitioning controlled by sorption processes. Such a mechanism could equally lead to increased incorporation of U into the garnet lattice during growth. The proposed mechanisms assume either reduction of U during incorporation into garnet, and/or the incorporation of U(VI) into the garnet structure. Establishing which of these mechanisms predominates requires a more detailed understanding of the location of U within the garnet structure, the incorporation mechanisms at the garnet surface, and the redox state of U in the hydrothermal solution.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTINGANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The REE and uranium in garnet from skarns in the Beinn an Dubhaich granite aureole, Skye, display major variations both within and between crystals. Fluid inclusion data provide no evidence of changes in fluid source, and the variations in trace element distribution in garnet are interpreted to reflect both changes in the chemical environment during crystallization and the kinetics of mineral growth.
Garnet growth occurred under conditions that ranged from near equilibrium to high degrees of supersaturation, and from essentially closed-system crystallization from small batches of fluid, to periods of more open-system behaviour. The changes in REE pattern with zonation are sensitive indicators of these processes. Positive Eu anomalies without significant changes in the REE pattern, or accompanied by marked increases in the normalized abundance of the LREE relative to the HREE, are interpreted as the result of the introduction of new batches of externally buffered fluid into the skarn system as a result of crack seal controlled fluid flow. Dramatic increases in all the REE to give MREE–HREE-enriched patterns are attributed to phases of garnet growth from a supersaturated solution, leading to a surface sorption control on the incorporation of trace elements into garnet. Supersaturation also probably arose as a result of the introduction of fresh batches of hydrothermal fluid into the skarn system. Under these conditions closed-system fractional crystallization after self-sealing of the skarn led to a gradual steepening of the REE pattern, increased time for diffusion in the crystal surface layer, and hence closer approach to equilibrium within the bulk crystal, and the generation of positive Eu anomalies. Uranium correlates with all these effects, indicating that its incorporation in hydrothermal minerals is also dependent on fluid chemistry and the kinetics of crystallization, as well as crystal chemical controls. These processes can be simply modelled to reproduce the changes in REE pattern observed, but are limited in this case by the necessity of assuming a fluid composition. More realistic models must await a more detailed knowledge of REE content and speciation in hydrothermal fluids, to allow changes in both of these factors to be accounted for.
This study provides an insight into some of the mechanisms that may be responsible for Sm–Nd fractionation in hydrothermal systems, and describes some effects of hydrothermal processes and crystallization kinetics on REE fractionation. It provides a clear demonstration of the potential utility of the REE in elucidating hydrothermal processes. The potential for parent–daughter fractionations in radioactive elements from hydrothermal processes suggests that significant variations in Sm–Nd ratio may occur within individual crystals without significant changes in metal source, and this effect subsequently may have major implications for geochronology.
|
ACKNOWLEDGEMENTS |
|---|
We would like to thank John Spratt for his assistance with electron microprobe analyses during the course of this study. This paper has benefited from constructive reviews by J. Brugger, F. Poitrasson, J. Vander Auwera and an anonymous reviewer, and editorial comments by M. Wilson.
|
FOOTNOTES |
|---|
* Corresponding author. Telephone: 01273 642265. Fax: 01273 642285. E-mail: martin.smith{at}brighton.ac.uk
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL SETTINGANALYTICAL METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Auwera, J. V. & Andre, L. (1991). Trace elements (REE) and isotopes (O, C, Sr) to characterise the metasomatic fluid sources: evidence from the skarn deposit (Fe, W, Cu) of Traversella (Ivrea, Italy). Contributions to Mineralogy and Petrology 106, 325–339.[CrossRef][Web of Science]
Bailey, E. H. & Ragnarsdottir, K. V. (1994). Uranium and thorium solubilities in subduction zone fluids. Earth and Planetary Science Letters 124, 119–129.[CrossRef][Web of Science]
Banks, D. A., Yardley, B. W. D., Campbell, A. R. & Jarvis, K. E. (1994). REE composition of an aqueous magmatic fluid—a fluid inclusion study from the Capitan Pluton, New Mexico. Chemical Geology 113, 259–272.[CrossRef][Web of Science]
Bau, M. (1991). Rare earth element mobility during hydrothermal and metamorphic fluid–rock interaction and the significance of the oxidation state of europium. Chemical Geology 93, 219–230.[CrossRef][Web of Science]
Bau, M. (1996). Controls on the fractionation of isovalent trace elements in magmatic and aqueous systems: evidence from Y/Ho, Zr/Hf and lanthanide tetrad effect. Contributions to Mineralogy and Petrology 119, 213–223.
Bau, M. (1999). Scavenging of dissolved yttrium and rare earths by precipitating iron oxyhydroxide: experimental evidence for Ce oxidation, Y–Ho fractionation, and lanthanide tetrad effect. Geochimica et Cosmochimica Acta 63, 67–77.[CrossRef][Web of Science]
Bodnar, R. J. (1993). Revised equation and table for determining the freezing point depression of H2O–NaCl fluid inclusions. Geochimica et Cosmochimica Acta 57, 683–684.[CrossRef][Web of Science]
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. Geochimica et Cosmochimica Acta 58, 1053–1063.[CrossRef][Web of Science]
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 geologic systems—metamorphic equilibria at high-pressures and temperatures. American Mineralogist 68, 1059–1075.[Abstract]
Brugger, J., Lahaye, Y., Costa, S., Lambert, D. & Bateman, R. (2000). Inhomogeneous distribution of REE in scheelite and dynamics of Archaean hydrothermal systems (Mt. Charlotte and Drysdale gold deposits, Western Australia) Contributions to Mineralogy and Petrology 139, 251–264.[CrossRef][Web of Science]
Campbell, A. R., Banks, D. A., Phillips, R. S. & Yardley, B. W. D. (1995). Geochemistry of Th–U–REE mineralising magmatic fluids, Capitan Mountains, New Mexico. Economic Geology 90, 1271–1287.
Chou, I.-M. (1987). Phase relations in the system NaCl–KCl–H2O. III: Solubilities of halite in vapor-saturated liquids above 445°C and redetermination of phase equilibrium properties in the system NaCl–H2O to 1000°C and 1500 bars. Geochimica et Cosmochimica Acta 51, 1965–1975.[CrossRef][Web of Science]
Faure, G. (1986). Principles of Isotope Geology. New York: John Wiley.
Ferry, J. M. (1985). Hydrothermal alteration of Tertiary igneous rocks from the Isle of Skye, northwest Scotland II. Granites. Contributions to Mineralogy and Petrology 91, 283–304.[CrossRef][Web of Science]
Fleet, M. E., Seller, M. H. & Pan, Y. (1997). Rare earth elements, protoliths and alteration at the Hemlo Gold deposit, Ontario, Canada, and comparison with argillic and sericitic alteration in the Highland Valley Porphyry district, British Columbia. Economic Geology 92, 551–568.
Fleischer, R. L, Price, P. B. & Walker, R. M. (1975). Nuclear Tracks in Solids. Principals and Application. Berkeley: University of California Press.
Flynn, R. T. & Burnham, C. W. (1978). An experimental determination of rare earth partition coefficients between a chloride containing vapour phase and silicate melts. Geochimica et Cosmochimica Acta 42, 685–701.[CrossRef][Web of Science]
Giuliani, G., Cheilletz, A. & Mechiche, M. (1987). Behaviour of REE during thermal metamorphism and hydrothermal infiltration associated with skarn and vein-type tungsten ore bodies in central Morocco. Chemical Geology 64, 279–294.[CrossRef][Web of Science]
Gramaccioli, C. M., Diella, V. & Demartin, F. (1999). The role of fluoride complexes in REE geochemistry and the importance of 4f electrons: some examples in minerals. European Journal of Mineralogy 11, 983–992.
Haas, J. R., Shock, E. L. & Sassani, D. C. (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. Geochimica et Cosmochimica Acta 59, 4329–4350.[CrossRef][Web of Science]
Henderson, P. & Williams, C. T. (1981). Application of intrinsic Ge detectors to the instrumental neutron activation analysis for rare earth elements in rocks and minerals. Journal of Radioanalytical Chemistry 67, 445–452.[CrossRef]
Hoersch, A. L. (1981). Progressive metamorphism of the chert-bearing Durness limestone in the Beinn an Dubhaich aureole, Isle of Skye, Scotland: a re-examination. American Mineralogist 66, 491–506.[Abstract]
Holness, M. B. (1992). Palaeohydrology of calcsilicate aureole of the Beinn an Dubhaich granite, Skye, Scotland: a stable isotopic study. Journal of Metamorphic Geology 15, 71–83.
Holness, M. B., Bickle, M. J. & Harte, B. (1989). Textures of forsterite–calcite marbles from the Beinn an Dubhaich aureole, Skye, and implications for the structure of metamorphic porosity. Journal of the Geological Society, London 146, 917–920.
Holness, M. B. & Fallick, A. E. (1997). Palaeohydrology of the calcsilicate aureole of the Beinn an Dubhaich granite, Skye, Scotland: a stable isotopic study. Journal of Metamorphic Geology 15, 71–83.[CrossRef][Web of Science]
Holten, T., Jamtveit, B., Meakin, P., Cortini, M., Blundy, J. & Austrheim, H. (1997). Statistical characterisation and origin of oscillatory zoning in crystals. American Mineralogist 82, 596–606.[Abstract]
Holten, T., Jamtveit, B. & Meakin, P. (2000). Noise and oscillatory zoning of minerals. Geochimica et Cosmochimica Acta 64, 1893–1904.[CrossRef][Web of Science]
Jamtveit, B. (1991). Oscillatory zonation patterns in hydrothermal grossular–andradite garnet: nonlinear dynamics in regions of immiscibility. American Mineralogist 76, 1319–1327.[Abstract]
Jamtveit, B. & Andersen, T. (1992). Morphological instabilities during rapid growth of metamorphic garnets. Physics and Chemistry of Minerals 19, 176–184.[Web of Science]
Jamtveit, B. & Hervig, R. L. (1994). Constraints on transport and kinetics in hydrothermal systems from zoned garnet crystals. Science 263, 505–507.
Jamtveit, B., Wogelius, R. A. & Fraser, D. G. (1993). Zonation patterns of skarn garnets: records of hydrothermal system evolution. Geology 21, 113–116.
Jeffries, T. E. (2001). Elemental analysis by laser ablation ICP-MS. In: Alfassi, Z. B. (ed.) Non-destructive Elemental Analysis. Oxford: Blackwell Science, pp. 115–150.
Johnson, J. W., Oelkers, E. H. & Helgeson, H. C. (1992). SUPCRT92: a software package for calculating the standard molal thermodynamic properties of minerals, gases, aqueous species and reactions from 1 to 5000 bars and 0 to 1000°C. Computers and Geoscience 18, 899–947.[CrossRef]
Kawabe, I. (1992). Lanthanide tetrad effect in the Ln3+ ionic radii and refined spin-pairing theory. Geochemical Journal 25, 31–44.[Web of Science]
Keppler, H. & Wylie, P. J. (1990). Role of fluids in transport and fractionation of uranium and thorium in magmatic processes. Nature 348, 531–533.[CrossRef]
Keppler, H. & Wylie, P. J. (1991). Partitioning of Cu, Sn, Mo, W, U and Th between melt and aqueous fluid in the systems haplogranite–H2O–HCl and haplogranite–H2O–HF. Contributions to Mineralogy and Petrology 109, 139–150.[CrossRef][Web of Science]
Kretz, R. (1983). Symbols for rock-forming minerals. American Mineralogist 68, 277–279.[Abstract]
Kwak, T. A. P., Brown, W. M., Abeysinghe, P. B. & Tan, T. H. (1986). Fe solubilities in very saline hydrothermal fluids: their relation to zoning in some ore deposits. Economic Geology 81, 447–465.
Lipin, B. R. & McKay, G. A. (1989). Geochemistry and mineralogy of rare earth elements. Mineralogical Society of America, Reviews in Mineralogy 21.
Long, J. V. P. & Gravestock, D. C. (1988). An ion microprobe system for geological and other applications. In: Benninghoven, A., Hunter, A. M. & Werner, H. W. (eds) Secondary Ion Mass Spectrometry (SIMS VI). Chichester: John Wiley, pp. 161–164.
Lottermoser, B. G. (1992). Rare earth elements and hydrothermal ore formation processes. Ore Geology Reviews 7, 25–41.[CrossRef][Web of Science]
Masuda, A., Kawakami, O., Dohmoto, Y. & Takenaka, T. (1987). Lanthanide tetrad effects in nature: two mutually opposite types, W and M. Geochemical Journal 21, 119–124.[Web of Science]
McIntire, W. L. (1963). Trace element partition coefficients—a review of theory and applications to geology. Geochimica et Cosmochimica Acta 27, 1209–1264.[CrossRef][Web of Science]
McLennan, S. M. & Taylor, S. R. (1979). Rare earth element mobility associated with uranium mineralisation. Nature 282, 247–250.[CrossRef]
Möller, P. (1998). Rare earth elements and yttrium fractionation caused by fluid migration. In: Novák, M. & Rosenbaum, J. (eds) Challenges to Chemical Geology. Referenced papers from MAEGS-10. Prague: Czech Geological Survey, pp. 9–32.
Moorbath, S. & Bell, J. D. (1965). Strontium isotope abundance studies and rubidium–strontium age determinations on Tertiary igneous rocks from the Isle of Skye, North-West Scotland. Journal of Petrology 6, 37–66.
Nicolescu, S., Cornell, D. H., Södervall, U. & Odelius, H. (1998). Secondary ion mass spectrometry analysis of rare earth elements in grandite garnet and other skarn related silicates. European Journal of Mineralogy 10, 251–259.
Ortoleva, P., Merino, E., Moore, C. & Chadam, J. (1987). Geochemical self-organisation 1: Reaction–transport feedbacks and modelling approach. American Journal of Science 287, 979–1007.
Parkhurst, D. L. & Appelo, C. A. J. (1999). User's guide to PHREEQC (Version 2)—a computer program for speciation, batch-reaction, one-dimensional transport, and inverse geochemical calculations. US Geological Survey Water-Resources Investigations Report 99-4259.
Poitrasson, F., Pin, C. & Duthou, J. L. (1995). Hydrothermal remobilization of rare-earth elements and its effects on Nd isotopes in rhyolite and granite. Earth and Planetary Science Letters 130, 1–11.[CrossRef][Web of Science]
Poitrasson, F., Montel, J. M., Pin, C. & Duthou, J. L. (1998). Importance of late-magmatic and hydrothermal fluids on the Sm–Nd isotope mineral systematics of hypersolvus granites. Chemical Geology 146, 187–203.[CrossRef][Web of Science]
Pollok, K., Jamtveit, B. & Putnis, A. (2001). Analytical transmission electron microscopy of oscillatory zoned grandite garnets. Contributions to Mineralogy and Petrology 141, 358–366.[Web of Science]
Rakovan, J. & Reeder, R. J. (1996). Intracrystalline rare earth element distributions in apatite: surface structural influences on incorporation during growth. Geochimica et Cosmochimica Acta 60, 4435–4445.[CrossRef][Web of Science]
Rollinson, H. R. (1993). Using Geochemical Data: Evaluation, Presentation, Interpretation. London: Prentice–Hall.
Shannon, R. D. (1976). Revised effective ionic radii and systematic studies of interatomic distances in halides and chalcogenides. Acta Crytsallographica, Section A 32, 751–767.[CrossRef]
Sherrell, R. M., Field, M. P. & Ravizza, G. (1999). Uptake and fractionation of rare earth elements on hydrothermal plume particles at 9°45'N, East Pacific Rise. Geochimica et Cosmochimica Acta 63, 1709–1722.[CrossRef][Web of Science]
Sibson, R. H., Robert, F. & Poulsen, K. H. (1988). High-angle reverse faults, fluid-pressure cycling, and mesothermal gold– quartz deposits. Geology 16, 551–555.
Smith, M. P., Banks, D. A., Yardley, B. W. D. & Boyce, A. (1996). Fluid inclusion and stable isotope constraints on the genesis of the Cligga Head Sn–W deposit, S. W. England. European Journal of Mineralogy 8, 961–974.
Smith, M. P., Henderson, P. & Jefferies, T. (2002). Fluid evolution and the formation and alteration of allanite in contact skarn from the Beinn an Dubhaich granite aureole, Skye. European Journal of Mineralogy 14, 471–486.
Sourirajan, S. & Kennedy, G. C. (1962). The system H2O–NaCl at elevated temperatures and pressures. American Journal of Science 260, 115–141.
Sterner, S. M., Hall, D. L. & Bodnar, R. J. (1988). Synthetic fluid inclusions. V. Solubility relations in the system NaCl–KCl–H2O under vapor-saturated conditions. Geochimica et Cosmochimica Acta 52, 989–1006.[CrossRef][Web of Science]
Sverjensky, D. A. (1984). Europium redox equilibria in aqueous solution. Earth and Planetary Science Letters 67, 70–78.[CrossRef][Web of Science]
Tilley, C. E. (1951). The zoned contact-skarns of the Broadford area, Skye: a study of boron–fluorine metasomatism in dolomites. Mineralogical Magazine 29, 621–666.
Valsami-Jones, E. & Ragnarsdottir, K. V. (1997a). Solubility of uranium oxide and calcium uranate in water and Ca(OH)2-bearing solutions. Radiochimica Acta 79, 249–257.[Web of Science]
Valsami-Jones, E. & Ragnarsdottir, K. V. (1997b). Controls on uranium and thorium behaviour in ocean-floor hydrothermal systems: examples from the Pindos ophiolite, Greece. Chemical Geology 135, 263–274.[CrossRef][Web of Science]
Van Westernen, W., Allan, N. L., Blundy, J. D., Purton, J. A. & Wood, B. J. (2000). Atomistic simulation of trace element incorporation into garnets—comparison with experimental garnet–melt partition data. Geochimica et Cosmochimica Acta 64, 1629–1639.[CrossRef][Web of Science]
Wakita, H., Rey, P. & Schmidt, R. A. (1971). Abundances of the 14 rare-earth elements and 12 other trace elements in Apollo 12 samples: five igneous and one breccia rocks and four soils. In: Proceedings of the Second Lunar Science Conference. Cambridge, MA: MIT Press, pp. 1319–1329.
Williams, C. T. & Wall, F. (1991). An INAA scheme for the routine determination of 27 elements in geological and archaeological samples. In: Hughes, M. J., Cowell, M. R. & Hook,D. R. (eds) Neutron Activation and Plasma Emission Spectrometric Analysis in Archaeology: Techniques and Applications. British Museum Occasional Paper 82, 105–119.
Wolery, T. J. (1992). EQ3/6. A Computer Program for Geochemical Aqueous Speciation–Solubility Calculations: Theoretical Manual, User's Guide and Related Documentation (Version 7.0). UCRL-MA-110662-PT-III. Livermore, CA: Lawrence Livermore National Laboratory.
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 water vapour pressure. Chemical Geology 88, 99–125.[CrossRef][Web of Science]
Yardley, B. W. D., Rochelle, C. A., Barnicoat, A. C. & Lloyd, G. E. (1991). Oscillatory zoning in metamorphic minerals: an indicator of infiltration metasomatism. Mineralogical Magazine 55, 357–365.[Web of Science]
Zhang, Y. G. & Frantz, J. D. (1987). Determination of the homogenisation temperature and density of supercritical fluids in the system NaCl–KCl–CaCl2–H2O using synthetic fluid inclusions. Chemical Geology 64, 335–350.[CrossRef][Web of Science]
Zhang, Z. & Saxena, S. K. (1991). Thermodynamic properties of andradite and application to skarn with co-existing andradite and hedenbergite. Contributions to Mineralogy and Petrology 107, 255–263.[CrossRef][Web of Science]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  F. Tornos, C. Galindo, J. L. Crespo, and B. F. Spiro GEOCHEMISTRY AND ORIGIN OF CALCIC TUNGSTEN-BEARING SKARNS, LOS SANTOS, CENTRAL IBERIAN ZONE, SPAIN Can Mineral, February 1, 2008; 46(1): 87 - 109. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Kumral, H. Coban, and S. Caran Th-, U- AND LREE-BEARING GROSSULAR, CHROMIAN FERRIALLANITE-(Ce) AND CHROMIAN CERITE-(Ce) IN SKARN XENOLITHS EJECTED FROM THE GOLCUK MAAR CRATER, ISPARTA, ANATOLIA, TURKEY Can Mineral, October 1, 2007; 45(5): 1115 - 1129. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. G. Spry, A. Heimann, J. D. Messerly, and R. S. Houk Discrimination of Metamorphic and Metasomatic Processes at the Broken Hill Pb-Zn-Ag Deposit, Australia: Rare Earth Element Signatures of Garnet-Rich Rocks Economic Geology, May 1, 2007; 102(3): 471 - 494. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||