Mineralogy and Petrology (2008) 92: 81–97 DOI 10.1007/s00710-007-0193-5 Printed in The Netherlands
Mobility of palladium chloride complexes in mafic rocks: insights from a flow-through experiment at 25 C using air-saturated, acidic, and Cl-rich solutions S. A. Wood1 and C. Normand2 1
Department of Geological Sciences, University of Idaho, Moscow, Idaho, USA Departement des Sciences de la Terre et de l’Atmosphere, Universite du Quebec a Montreal, Montreal (Quebec), Canada 2
Received January 2, 2006; revised version accepted May 8, 2007 Published online July 13, 2007; # Springer-Verlag 2007 Editorial handling: T. Alapieti Summary A flow-through chromatographic experiment was carried out to determine the ability of an oxidized, acidic, chloride-bearing solution to transport Pd through a column of crushed Columbia River basalt at 25 C. An initial plug of solution containing 13.8 mg Pd=kg H2O was loaded onto the column of crushed basalt and eluted with an airsaturated, 1 m NaCl solution of pH 3. Even after 3606 h of elution, corresponding to a fluid-rock mass ratio of 1338, less than 10% of the Pd originally loaded on the column was recovered. On the other hand, in a similar experiment involving relatively unreactive quartz sand, the Pd was almost completely recovered after the passage of only 4 pore masses of eluent (or a fluid-rock mass ratio of approximately one), a behavior similar to that of Rb. The results suggest that unrealistically large amounts of oxidizing and acidic fluids are required to react with a given mass of basalt in order to overcome the acid- and redox-buffering capacity of the rock and to mobilize palladium as a chloride complex. On the other hand, Pd-chloride complexes can be easily transported through a rock with minimal acid- and redox-buffering capacity, such as clean quartz sandstone. The implications of this study for PGE enrichments in sediment-hosted stratiform copper (Kupferschiefer type) deposits are explored briefly.
Introduction Thermodynamic calculations based on both theoretical predictions and experimental solubility studies indicate that significant amounts of Pd can be transported as
82
S. A. Wood and C. Normand
chloride complexes at temperatures less than 350 C only under extremely oxidizing and acidic conditions, irrespective of salinity (Mountain and Wood, 1988; Gammons et al., 1992; Wood et al., 1992; Wood, 2002). It is predicted that such solutions can not maintain their Pd-bearing capability while interacting with mafic or ultramafic rocks, until the acid- and redox-buffering capacities of these rocks have been exceeded. In other words, it is unlikely that significant Pd could be transported through mafic or ultramafic rocks, as chloride complexes at temperatures less than 350 C, until minerals containing Fe(II) have been converted to Fe(III)oxyhydroxides and basic cations such as Mg2þ and Ca2þ have been removed. On the other hand, there have been numerous reports in the literature that hydrothermal fluids were involved in the transport and concentration of palladium and other platinum-group elements in basalts, gabbros or other mafic rocks. Examples of PGE mineralization in mafic rocks with a possible hydrothermal component include those at: the Salt Chuck Intrusion, Alaska (Watkinson and Melling, 1992); the Kruuse Fjord gabbro complex, Greenland (Arnason et al., 1997); the Baula Complex, India (Auge et al., 2002); the McBratney prospect, Manitoba (Olivo and Theyer, 2004); the Duluth Complex, Minnesota (Mogessie et al., 1991); Lac des Iles, Ontario (Hinchey and Hattori, 2005); the Coldwell Complex, Ontario (Watkinson and Ohnenstetter, 1992; Watkinson and Jones, 1996); Rathbun Lake, Ontario (Rowell and Edgar, 1986); Reeser’s summit diabase, Pennsylvania (Belkin, 1989); Lac Sheen, Quebec (Cook and Wood, 1994); the Lukkulaisvaara layered intrusion, Russia (Barkov et al., 1995); the New Rambler mine, Wyoming (McCallum et al., 1976); and the Wellgreen intrusion, Yukon Territory, Canada (Marcantonio et al., 1994), among others. In general, the temperatures of the hydrothermal activity in the above-mentioned deposits are poorly constrained, and it is likely that hydrothermal activity occurred over a comparatively wide range of temperature in many cases. Nevertheless, in all the studies in which estimates of the temperature of the hydrothermal PGE-mobilizing events were provided, relatively low-temperatures (all less than 500 C and most less than or equal to 300–350 C) have been suggested (Belkin, 1989; Mogessie et al., 1991; Watkinson and Melling, 1992; Barkov et al., 1995; Auge et al., 2002; Olivo and Theyer, 2004; Hinchey and Hattori, 2005). Seafloor hydrothermal vent fluids at mid-ocean ridges, which acquire their chemical characteristics by interaction of seawater and basalt at elevated temperatures (approximately 350–400 C; e.g., Seyfried et al., 1988; Allen and Seyfried, 2003; Foustoukos and Seyfried, 2005), appear to be capable of transporting small but significant quantities of PGE (Crocket, 1990; Vaganov et al., 1995; Pasava et al., 2004). Moreover, PGE mineralization in large ultramafic-mafic layered intrusions such as the Bushveld (Cawthorn et al., 2002) and the Stillwater Complexes (Zientek et al., 2002) has been proposed by some workers to be at least partially the result of hydrothermal processes (e.g., Kinloch, 1982; Schiffries, 1982; Volborth and Housely, 1984; Ballhaus and Stumpfl, 1985, 1986; Boudreau and McCallum, 1986, 1992; Boudreau et al., 1986; Volborth et al., 1986; Schiffries and Skinner, 1987; Boudreau, 1988; Boudreau and Kruger, 1990; Willmore et al., 2000; Hanley et al., 2005a). Many of these authors envision hydrothermal PGE transport at or near magmatic temperatures (1200 C), but hydrothermal systems in these large intrusions also likely persist over a wide temperature range.
Mobility of palladium chloride complexes in mafic rocks
83
It is commonly assumed that the mobility of Pd in basaltic=gabbroic rocks is a result of chloride complexation, based largely on the ubiquity of chloride as a ligand in hydrothermal fluids in general (Seward and Barnes, 1997; Wood and Samson, 1998), the association of saline fluid inclusions and chloride-bearing minerals with some PGE mineralization (e.g., Schiffries, 1982; Ballhaus and Stumpfl, 1986; Boudreau, 1988, 1993, 1995; Boudreau and Kruger, 1990; Springer, 1989; Boudreau et al., 1986, 1993, 1995; Nyman et al., 1990; Mogessie et al., 1991; Dahlberg and Saini-Eidukat, 1991; Farrow and Watkinson, 1992; Li and Naldrett, 1993a, b; Barkov et al., 1995; Watkinson and Jones, 1996; Willmore et al., 2000; Hanley et al., 2004, 2005a, b), and the relatively high stability constants for Pd-chloride complexes at room temperature (e.g., Elding, 1972). However, as mentioned above, theoretical calculations and experimental measurements (Mountain and Wood, 1988; Gammons et al., 1992; Wood et al., 1992; Wood, 2002) suggest that, whereas Pd transport as bisulfide complexes is possible in mafic rocks, transport as chloride complexes is very unlikely at temperatures less than approximately 350 C under rock-buffered conditions. High chloride content of the fluid is a necessary but not sufficient criterion for the hydrothermal mass transfer of significant amounts of Pd as chloride complexes under these conditions. Significant PGE transport as chloride complexes under mafic rock-buffered conditions probably requires temperatures considerably in excess of 350 C. Here we report the results of a flow-through experiment in which the mobility of palladium through crushed basalt in an acidic, Cl-rich fluid was studied at room temperature. Although hydrothermal Pd deposits form at temperatures higher than 25 C, initial experimentation at room temperature offers several advantages including precise determination of reaction parameters such as pH, visual monitoring of the progress of reaction fronts, and use of a metal-free experimental system, thus avoiding PGE metal loss by alloying or reduction on the walls of vessels, tubing, valves, etc. These low-temperature experiments provide an important starting point for subsequent experiments at elevated temperatures, and suggest that the above-mentioned theoretical predictions are valid. Experimental approach The behavior of Pd was studied during interaction of an air-saturated, acidic, 1 m Cl solution with basalt (Columbia Plateau) by allowing gravity-fed fluid to pass through a rock-filled glass column at 25 C. The basalt in the column first was saturated with a HCl–NaCl solution of pH ¼ 3. Subsequently, a 2-g aliquot of solution with the same HCl and NaCl concentrations, but also containing 13.8 mg Pd per kg of H2O was introduced. After the top of the slug of the Pd-spiked solution reached the top of the rock column, the glass column was filled to the top with the Pd-free HCl–NaCl solution and connected with a reservoir containing the same solution. In essence, the experiment consisted of a gravity-fed chromatographic elution. The HCl–NaCl solution in the reservoir remained in contact with air and was allowed to flow continuously for 3515 h. After that time, the solution in the reservoir was changed to a NaCl–HCl solution with the same ionic strength but a pH of 2 and the fluid was allowed to flow through the column for a further 91 h. Therefore, the total duration of the experiment was 3606 h.
84
S. A. Wood and C. Normand
The Columbia River Basalt sample employed was taken from a quarry a few miles from Moscow, Idaho in the direction of Pullman, Washington. Care was taken to collect homogeneous specimens free of alteration. A further selection was done in the lab by trimming off any part containing veinlets or signs of hydrothermal or weathering imprints. Sample chips were then crushed first in a jaw crusher, and second in a disk grinder to less than 2 mm. The crushed sample was then sieved and the 74–100 mm size fraction retained. The basalt grains were cleaned first three times in a deionized water-filled beaker suspended in an ultrasonic bath, and then 10 times in alcohol until the supernatant was clear. The basalt was analyzed by X-ray diffraction and scanning electron microscopy (SEM) which showed that the predominant phases are plagioclase, clinopyroxene, and glass. No sulfide was detected. The clear glass tube employed to construct the basalt column was 31.2 cm long and had an internal diameter of 0.6 cm. The column was loaded with 6.83 g of crushed basalt (grain size 74–100 mm). At the bottom of the tube a plug of silica wool was employed to retain the crushed basalt in the column. The bottom end of the glass tube was fitted with a short piece of Tygon tubing, then a 4-cm long, opaque, black, plastic reducing connector, followed by another short piece of Tygon tubing, and finally two 0.02-mm pore-size, 25-mm diameter, disposable Anapore+ membrane filters in series. The effluent from the last filter was collected in a borosilicate glass volumetric flask. The basalt occupied 18 cm of the column corresponding to a volume of 5.04 cm3 for a volume of rock of 2.28 cm3. The calculated porosity of the portion of the column filled by rock was 45.2%. This corresponded to a volume of 2.76 cm3 of fluid required to saturate the porosity of the rock column. Fluid flowed through the rock column at a variable rate between 0.9 and 4.7 g=h for a total duration of 3606 h. The total mass of fluid circulated during this time was 9135 g for a total fluid-rock mass ratio of 1338. In addition to the basalt experiment, a control experiment was conducted using pure quartz sand cleaned five times in aqua regia and then rinsed with deionized water. In this experiment the glass column was 30.5 cm long and had an inner diameter of 0.6 cm. The column was loaded with 9 g of quartz sand, which resulted in a calculated porosity of 37% and a fluid saturation volume of 2 cm3. The eluent fluid was the same as in the basalt experiment, but the initial plug of Pd-bearing solution was 3.02 g of a solution of 12.5 mg kg1 Pd and 1.0 mg kg1 Rb (added as a conservative tracer). The flow rate varied between 0.1 and 20 g hr1, and the experiment was carried out for 617 h. A total of 2373 g of fluid passed through the column during the experiment, yielding a cumulative fluid-rock mass ratio of 264. In order to closely follow the behavior of the Pd injected at the start of the basalt experiment, particularly during the critical initial stages, twelve samples weighing between 1 and 28 g were collected during the first day, sixteen samples the second and third days (1.6 g each), and nine samples during the fourth day. After that the number of samples was gradually decreased to two per day until 963 h, after which time only one sample per day was collected for the remainder of the experiment. All of the fluid that emerged from the column was collected. A portion of each sample of fluid was used to measure pH and another diluted ten times with water and acidified to 2% HCl and 4% HNO3 for preservation until
Mobility of palladium chloride complexes in mafic rocks
85
analysis. A similar sampling schedule was employed for the quartz-sand experiment, with 27 samples taken on the first day, 6 samples the second day, 3 samples the third day, 4 samples the fourth day, 3 samples the fifth day, and between 1 and 2 samples per day for the remainder of the experiment. All fluid samples were analyzed by ICP-AES (inductively coupled plasmaatomic emission spectroscopy) using a Perkin Elmer Optima 3000XL at the University of Idaho. Analyses were performed for Si, Al, Fe, Mn, Mg, Ca, K, and Pd using standard solutions prepared from commercial standards. Each standard was made up to 0.1 m NaCl and acidified to 2% HCl and 4% HNO3 to match the matrix of the diluted samples. Many of the solution samples were below the detection limit of the ICP-AES for Pd (50 mg kg1 accounting for the dilution factor), and so they were reanalyzed by ICP-MS (inductively coupled plasma-mass spectrometry) using a ThermoFinnigan Element 2 instrument at nearby Washington State University. The detection limit for this technique, taking in account of the dilution factor, was 0.2 mg kg1 . The pH of each fluid sample was determined using an Orion combination glass electrode and a Radiometer Copenhagen meter. The electrode and meter were calibrated using standard solutions prepared at the same ionic strength as the experimental solutions to minimize liquid junction and activity coefficient errors. Results Because pH is generally a major control on the mobility of platinum-group elements (Mountain and Wood, 1988; Gammons et al., 1992; Wood, 2002), it is useful to discuss first the behavior of the measured pH of the effluent samples before turning to the concentrations of metals. The behavior of pH in the quartz-sand experiment is shown in Fig. 1. The pH was not measured in the first several samples, because we expected that the Pd might elute quickly and we did not want to take the chance that a significant portion of Pd would be lost to aliquots taken for pH measurement. However, the initial measured pH of the eluent prior to passing through the column was 3.2, and the first sample in which pH was measured
Fig. 1. Plot of measured pH as a function of time for the quartz-sand column experiments
86
S. A. Wood and C. Normand
Fig. 2. Plot of measured pH as a function of time for the basalt column experiments. The arrow indicates the point at which the pH of the eluent was lowered from 3 to 2
(1.78 h) had a pH of 3.3. The pH remained constant until after 2.5 h, at which point it rose slowly to 3.8 at 6.9 h. Although the experiment continued for a total of 617 h, the total amount of Pd initially placed on the column was recovered within the initial 6.9-h period, so we did not analyze any of the subsequent samples taken. That the pH remained comparatively close to the initial pH of the eluent is consistent with the unreactive nature of pure quartz sand. It is not clear why the pH increased somewhat during the experiment. In the basalt experiment, the pH of the eluent fluid dropped sharply from a value slightly greater than 7 to a value of 6.3 within the first 39 h of reaction, and then smoothly to a value of 5.8 after about 200 h of reaction (Fig. 2). The pH then decreased at a slower rate to a value of 5.2 after 2534 h of reaction. After this, the pH dropped suddenly to a new plateau value of 4.8. When the lower-pH eluent was introduced at 3515 h, the pH dropped again but did not fall below 4.3 by the termination of the experiment. These measured pH values indicate that the basalt had considerable buffering capacity that was not completely consumed even after a fluid-rock mass ratio of more than 1300 or a factor of ten increase in the proton activity. The various plateaus of pH attained during the experiment may represent portions of the experiment where buffering was dominated by different phases or phase assemblages. Minerals with higher buffering capacity or faster reaction rates were consumed first, leaving less-reactive primary phases or lessreactive, newly formed secondary phases to buffer pH at subsequent steps in the reaction path. With the exception of Si and Al, the concentrations of the rock-forming elements were very high at the start of the basalt experiment (reflecting dissolution of adhering fine-grained material, sharp edges and sites of high strain) and dropped sharply within the first 24 h of reaction. Following this initial stage, Fe concentrations increased sharply between 30 and 40 h of reaction from a value of 3108 m to a value of 2105 m which was approximately maintained until 95 h. Iron
Mobility of palladium chloride complexes in mafic rocks
87
concentrations increased again abruptly to a value of 3105 m after 115 h. From 115 to 2534 h, the Fe concentrations varied within the restricted range of 3 to 4105 m. After this, the Fe concentrations dropped sharply, and this sharp drop was accompanied by a sharp increase in Al concentrations. In the basalt column experiment, an oxidation front characterized by the formation of ochre-yellow secondary iron oxide minerals progressed with time through the basalt column at a rate of 50 mm=h for the first 817 h of reaction. This rate increased to a maximum of 70 mm=h after 1613 h of reaction, after which time it decreased to a value of 64 mm=h by the time the front had passed through the entire column (3110 h). The passage of the oxidation front through the bottom of the basalt column accompanied the relatively sharp decrease in pH from a value of 5.2 to 4.8, the sharp decrease in the iron concentrations, and concomitant increase in Al concentrations. Figure 3 shows that the initial plug of Pd was completely eluted after only 4 pore masses of fluid had passed through the quartz sand column (i.e., all the initial Pd was recovered after the pore fluid had been replaced 4 times). The behavior of Pd is similar to Rb, which we included as a non-reactive tracer. This finding implies that Pd and Rb were only slightly retarded during passage through the quartz sand, possibly as a result of relatively weak sorption onto the quartz surfaces. The behavior of Pd on the basalt column was quite different (Fig. 4). We did not detect Pd at a level greater than 0.2 mg kg1 in any effluent samples until after 797 h at which point 2859 g of fluid (998 pore masses of fluid) had passed through the column. The cumulative fraction of recovered Pd gradually increased from 0.004 at 998 pore masses (at which point Pd concentration was 7 mg kg1 ) to 0.085 at 1061 pore masses (at which point Pd concentration was at a maximum
Fig. 3. Plot of the cumulative fraction of Pd or Rb recovered from the quartz-sand column experiment as a function of the number of pore masses of fluid eluted. A pore mass is the mass of fluid in grams completely filling the pore space of the column. The cumulative fraction of Pd recovered slightly exceeds 1.0 (1.05), probably as a result of the accumulation of small errors in the Pd analysis
88
S. A. Wood and C. Normand
Fig. 4. Plot of the cumulative fraction of Pd recovered from the basalt column experiment as a function of the number of pore masses of fluid eluted. Only the portion of the experiment in which detectable Pd was found in the effluent sample is shown in this figure. Note that the fraction of Pd recovered never exceeded 0.1
at 18 mg kg1 ), and thereafter the Pd concentration dropped below detection (i.e., 0.2 mg kg1 ). Palladium was not detected in the subsequent effluent samples even after the pH of the influent was decreased from 3 to 2, or after the oxidation front had passed. Thus, the calculated percentage of Pd recovered never exceeded 10%. We can conclude from this finding that either most of the Pd was never eluted and remained on the column in some form or the Pd was eluted fairly continuously at concentrations below the detection limit of ICP-MS. If we assume that Pd was eluted continuously at a concentration just at our detection limit, then over the course of the entire experiment, only 6.5% of the initial Pd would have eluted. Combining this with the 8.9% that eluted between 998 and 1061 fluid pore masses when Pd was detectable, we calculate that a maximum of 15.4% of the total initial Pd was eluted, and therefore at least 85% must have remained on the column. Discussion Before discussing the results, it is necessary to determine the probable predominant dissolved species for Pd under the conditions of the experiments. Calculations show that the speciation of palladium in aqueous NaCl solutions containing 1 m total Cl (based on experimentally determined equilibrium constants provided in van Middlesworth and Wood, 1999) is dominated by the species PdCl4 2 at pH below 6.7, the species PdCl3OH2 at pH 6.7 to 7.9, and the species Pd(OH)2 at pH above 7.9 (and < 12). Under reducing conditions (log f O2 <22.47) in a sulfur-free system, metallic palladium is the stable solid phase, and its solubility decreases precipitously with decreasing f O2 and increasing pH (Fig. 5). Under oxidizing conditions, Pd(OH)2(s) is the stable phase; its solubility is independent of f O2 but strongly dependent on pH.
Mobility of palladium chloride complexes in mafic rocks
89
Fig. 5. log f O2 – pH diagram at 25 C and 1 bar showing the stability fields of amorphous Pd(OH)2 and Pd metal (separated by solid line), the fields of predominance of the aqueous palladium complexes (separated by long dashed lines), and the solubility contours for total palladium in mg kg1 water (dotted lines). The diagram was constructed assuming total free Cl ¼ 1.0 m. Filled circles represent calculated conditions for selected samples assuming saturation with goethite. Path A corresponds to samples collected before the oxidation front passed through the entire rock column. Path B corresponds to samples collected after the oxidation front passed through the entire column. Arrows indicate the direction of reaction progress
In the quartz-sand experiment, the pH was always less than 4, so PdCl4 2 was almost certainly the predominant species. The results of this experiment showed that PdCl4 2 can be transported readily through quartz sand with minimal retardation, which implies that Pd could be transported in significant quantities (>10 mg kg1 ) as a chloride complex through relatively pure quartz sandstone. In this case, Pd is transported as easily as a large alkali element, Rb, which is essentially conservative. In order to interpret the basalt column experiments, it is useful to consider the oxidation state of the fluid within the column. Based on the color of the altered basalt, we suggest that iron oxyhydroxides formed an important part of the secondary mineral assemblage. Calculations using the Geochemist’s Workbench show that iron and manganese oxides-hydroxides are greatly supersaturated if f O2 is fixed to a value corresponding to that in air (0.21 bar). However, the relatively high concentrations of dissolved Fe at the pH values of the experiment suggest that the conditions were relatively reduced, probably due to the reaction of Fe(II) contained in the silicates with the fluid along the lines of FeðIIÞ þ Hþ ¼ FeðIIIÞ þ 0:5H2 ðgÞ to form the secondary iron oxyhydroxides. Furthermore, the sudden drop in the iron concentration of the fluid after the oxidation front passed through the entire rock column suggests that the capacity of the rock to maintain low f O2 was greatly diminished thereafter.
90
S. A. Wood and C. Normand
As a first-order approximation, f O2 was calculated by assuming that the elevated iron concentrations reflect formation of goethite at saturation. The results (Fig. 5) suggest that f O2 was maintained to a value where the equilibrium solubility of palladium in the fluid was extremely low (1 mg kg1 ) for the duration of the experiment, consistent with the lack of detectable Pd in the effluent throughout most of the experiment. Recall that the pH was 5.7 at the point where the maximum dissolved Pd concentration (18 mg kg1 ) was attained. From Fig. 5 it is evident that, at this pH along the calculated reaction path for the experiment, the equilibrium solubility of Pd metal as PdCl4 2 is more than ten orders of magnitude less than the measured Pd concentration. This finding suggests that either there are additional species contributing to Pd transport that have not been taken into account in the modeling or, more likely, the experiments did not reach full oxidation-reduction equilibrium at the low temperature of the experiment. If the latter is the case, then the amount of Pd transported over time scales longer than those in the experiments could potentially be orders of magnitude less than that measured. More importantly, because a modest increase (100–300 C) in temperature above 25 C would significantly increase the rates of equilibration, but would not result in a sufficient increase in rock-buffered Pd solubility (see below), temperatures moderately higher than that employed in the experiments could actually decrease the mobility of Pd. As mentioned in the Results section, the results of the basalt column experiment could be explained by retention of the bulk of the Pd on the column or by a continuous release of Pd at quantities below the method detection limit. We favor the former explanation based on our thermodynamic calculations which suggest that the equilibrium solubility of Pd under the conditions present in the column was always likely to be low, and our calculation of the small percentage of total Pd that would have been released if the concentration was continuously at the detection limit or below throughout the experiment. Reduction and acid-neutralization of the fluid by the constituents of the basalt were probably the main factors that caused retention, most likely via precipitation of a solid Pd phase in the basalt column, although a role for adsorptive retardation cannot be excluded. The experiments therefore indicate that Pd cannot be transported as a chloride complex through rocks such as basalts until the buffering capacity of primary and secondary minerals has been destroyed by interaction with the fluid at very high fluid-rock ratios. In other words, the presence of minerals containing Fe(II) and minerals containing alkali and alkaline earth elements, such as plagioclase, pyroxene, olivine, K-feldspar, etc., maintain the oxygen fugacity too low and the pH too high, respectively, for significant Pd transport as chloride complexes. Even if the alternate hypothesis is true, i.e., that very small quantities of Pd were removed continuously from the column, then it still would be difficult to invoke hydrothermal transport of Pd (in the form of chloride complexes) through basalt as an ore-transporting mechanism. An ore-transporting mechanism should lead to the concentration of a metal, not its dispersion. It is likely that our conclusions regarding basalt apply to ultramafic rocks as well, because such rocks contain significant amounts of olivine, which is even more reactive towards fluid alteration than pyroxene and plagioclase, and therefore will buffer the pH to comparatively high values where Pd solubility as chloride complexes is negligible. For example, Barnes et al. (1967, 1972) and Barnes and
Mobility of palladium chloride complexes in mafic rocks
91
O’Neill (1969) have reported that springs emerging from ultramafic rocks undergoing active serpentinization have pH values up to 12. Experiments by Janecky and Seyfried (1986) demonstrate that fluids with chemistries related to seawater can acquire pH values up to 2 units more alkaline than neutral upon interaction with harzburgite or lherzolite at elevated temperatures. On the other hand, fluids moving through more felsic rocks, particularly those that may have been subjected to prior alteration or weathering, may be capable of overcoming the smaller redoxand pH-buffering capacities of these rocks at more reasonable fluid-rock ratios. However, little Pd transport as a chloride complex would occur until minerals such as K-feldspar and albite are nearly completely destroyed, leaving a relatively unreactive residue of quartz, micas, and clays. We plan to test these predictions in future column experiments using ultramafic and felsic rocks. The question that remains is: How relevant are these low-temperature experiments to Pd transport in higher-temperature, hydrothermal solutions? From experimental measurements and theoretical calculations (in particular, see Gammons et al., 1992) it is apparent that, although the solubility of Pd as a chloride complex increases with increasing temperature, solubilities remain less than 1 mg kg1 up to 350 C, as long as the system is buffered with respect to pH and redox conditions at values typical of a fluid in equilibrium with a mafic or ultramafic rock. Assuming that a rock initially contains an insignificant concentration of Pd and that the depositional process is 100% efficient, then a fluid-rock ratio of more than 1000 would be required to produce a concentration of 1 mg kg1 Pd in that rock at any temperature at or below 350 C. Thus, conclusions reached from our room-temperature experiments are probably valid to at least 350 C, and as noted above, the faster reaction kinetics at moderately elevated temperatures might lead to an actual reduction in the mobility of Pd compared to that observed experimentally at 25 C. A significant number of the studies cited in the introduction (but by no means all) have postulated significant hydrothermal Pd redistribution at temperatures less than or equal to 400 C, so our results are likely applicable to many instances of hydrothermal mass transfer of Pd in mafic=ultramafic rocks. One type of PGE-bearing mineralization to which our experimental results are most directly relevant are sediment-hosted stratiform copper (SHSC) deposits (e.g., those in the Kupferschiefer and the Zambia-Zaire Copperbelt). These deposits have been reported to contain small but significant enrichments in PGE (Mertie, 1969; Wedepohl, 1971; Kucha, 1981, 1982, 1985; Unrug, 1985; Kucha and Przybylowicz, 1999; Oszczepalski, 1999; Coveney, 2000; Piestrzynski and Wodzicki, 2000; Stribrny et al., 2000). Several authors (Rose, 1976; Eugster, 1985; Haynes, 1986; Sverjensky, 1987, 1989; Vaughan et al., 1989; Brown, 1993) have suggested a model for the formation of SHSC deposits in which late diagenetic or epigenetic fluids at temperatures of less than 120 C first interact with red beds and evaporites, becoming oxidized (and possibly acidic), saline and Cu-laden, and then with black shales where the fluids are reduced and Cu is precipitated. Experimental and theoretical investigations mentioned previously on the solubility of Pt and Pd as chloride complexes suggest that mg kg1 quantities of these metals could be transported in the type of fluid envisaged for Cu transport during the formation of SHSC deposits. Encounter of such fluids with a black shale or other reduced lithology would cause Pt and Pd to precipitate along with the Cu. Metal-rich iron-oxide
92
S. A. Wood and C. Normand
coatings on quartz and other mineral grains, and basalt-derived clastic components of redbed sediments, have been proposed as sources of Cu in SHSC (Haynes, 1986; Brown, 1993), and it is possible that these same sources provide the PGE. Assuming mafic rock=mineral fragments in redbed sediments to be the source of the PGE in SHSC deposits, our results suggest that Pd (and probably also Pt) could not be mobilized through the redbeds until the buffering capacities of any reactive minerals, e.g., those in the mafic fragments, and any others such as feldspar, calcite etc., are overcome. In the redbed sediments, the buffering capacity of the minerals could be consumed during weathering and transport to the site of sediment deposition. For example, the oxygen buffering capacity could be reduced by oxidation of Fe(II) in ferromagnesian silicates during transport, resulting in the formation of hematite and other Fe-oxyhydroxides that give the redbeds their name. Any remaining mineral buffer capacity would be consumed when later diagenetic or epigenetic, oxidized hydrothermal fluids passed through the redbeds. If the amount of mafic material in the sediments is relatively small compared to quartz, and the mafic material is ‘‘preconditioned’’ by weathering during sediment transport, then a realistic mass of hydrothermal fluid would be sufficient to overcome any remaining buffer capacity and permit the transport of Pt and Pd through the redbeds. An alternative model, and one that is favored by Brown (2005, 2006), is that the sediments were not highly oxidized at the time of their deposition, but rather underwent diagenetic ‘‘reddening’’ due to influx of oxidizing meteoric water. The timing of this preconditioning is critical. According to Brown (2005, 2006) Cu is optimally dissolved and transported at the moderately oxidizing conditions that would occur before diagenetic reddening was complete, whereas we would argue that strongly oxidizing conditions (perhaps only possible after significant volumes of meteoric water have completely oxidized the redbeds and exhausted their oxygen buffering capacity) are required to dissolve and transport PGE. These considerations have important implications for the relative timing of PGE and Cu mineralization in sediment-hosted stratiform type deposits. Copper could be released to the ore-forming fluid well before the PGE. Thus, where present, the PGE might be expected to be paragenetically late. On the other hand, if diagenetic reddening ceased prior to the complete exhaustion of oxygen buffer capacity, then the PGE might never be released to the ore-forming fluid and these elements would be absent in the final deposit. The results of this study should not be taken to mean that significant transport of Pd through mafic or ultramafic rocks can never take place. The conclusions only apply to Pd transport as chloride complexes at temperatures less than 350 C. Transport of Pd as chloride complexes through such rocks at significantly higher temperatures (greater than 400 C) is quite likely owing to significant increases in the rock-buffered solubility of Pd with increasing temperature above 350 C. Also, as pointed out by Wood (2002) and references therein, if reduced sulfur is present, some mobility of Pd as bisulfide complexes is expected in mafic and ultramafic rocks even at relatively low temperatures, and it is likely that PGE mobility observed in the examples given in the introduction (except for the Bushveld-= Stillwater-type examples) is accomplished via complexation with bisulfide. Future experiments with sulfide-rich basalts are planned to test whether significant lowtemperature transport of Pd through basalt is possible via bisulfide complexes.
Mobility of palladium chloride complexes in mafic rocks
93
Conclusions The results of this study lead us to the following conclusions: 1) Acidic, oxidized, chloride-bearing fluids maintain their ability to transport Pd while passing through relatively unreactive rocks with low pH- and redox-buffering capacities, such as pure quartz sandstone. Transport of Pd through such rocks is only mildly retarded, and only to approximately the same extent as the conservative alkali element Rb. 2) Passage through a rock with relatively high pH- and redox-buffering capacity, such as a basalt, causes an initially oxidized and acidic fluid to be sufficiently neutralized and reduced that it loses its capacity to transport Pd (in sulfide-poor systems at temperatures less than 350 C). At low temperatures, fluid-rock mass ratios in excess of 1400 are required to overcome the innate buffering capacity of basalt, such that the oxidizing and acidic conditions required to transport Pd as a chloride complex are realized. At this point, the rock would be so leached of alkalis, alkaline earths, and possibly aluminum, and its iron would be so completely oxidized, that it would be unrecognizable as a basalt. 3) The results of this study thus support conclusions based on experimental solubility studies and thermodynamic calculations that, at temperatures less than 350 C, significant transport of Pd as chloride complexes is not possible at pH and oxygen fugacity values fixed by the mineralogy of mafic and ultramafic rocks. However, these results do not preclude transport of Pd as bisulfide complexes where sulfide is present, or transport as chloride complexes at temperatures significantly above 350 C. Acknowledgments This work was funded by a grant from the National Science Foundation (EAR-0229495). We thank Tom Williams for assistance with X-ray diffraction and scanning electron microscopic analyses of the basalt and quartz-sand samples. Comments on an earlier version of the manuscript by Jim Mungall and Alan Boudreau helped improve the presentation.
References Allen DE, Seyfried WE Jr (2003) Compositional controls on vent fluids from ultramafichosted hydrothermal systems at mid-ocean ridges: an experimental study at 400 C, 500 bars. Geochim Cosmochim Acta 67: 1531–1542 Arnason JG, Bird DK, Bernstein S, Kelemen PB (1997) Gold and platinum-group element mineralization in the Kruuse Fjord Gabbro Complex, East Greenland. Econ Geol 92: 490–501 Auge T, Salpeteur I, Bailly L, Mukherjee MM, Patra RN (2002) Magmatic and hydrothermal platinum-group minerals and base-metal sulfides in the Baula Complex, India. Can Mineral 40: 277–309 Ballhaus CG, Stumpfl EF (1985) Occurrence and petrological significance of graphite in the Upper Critical Zone, western Bushveld Complex, South Africa. Earth Planet Sci Lett 74: 58–68 Ballhaus CG, Stumpfl EF (1986) Sulfide and platinum mineralization in the Merensky Reef: evidence from hydrous silicates and fluid inclusions. Contrib Mineral Petrol 94: 193–204
94
S. A. Wood and C. Normand
Barkov AYu, Savchenko YeE, Zhangurov AA (1995) Fluid migration and its role in the formation of platinum-group minerals: evidence from the Imandrovsky and Lukkulaisvaara layered intrusions, Russia. Mineral Petrol 54: 249–260 Barnes I, O’Neill JR (1969) The relationship between fluids in some fresh Alpine-type ultramafics and possible modern serpentinization, Western United States. Geol Soc Amer Bull 80: 1947–1960 Barnes I, Lamarche VC Jr, Himmelberg G (1967) Geochemical evidence of present-day serpentinization. Science 156: 830–832 Barnes I, Rapp JB, O’Neill JR, Sheppard RA, Gude AJ III (1972) Metamorphic assemblages and the direction of flow of metamorphic fluids in four instances of serpentinization. Contrib Mineral Petrol 35: 263–276 Belkin HE (1989) Fluid inclusion and mineralogic studies of a palladium-platinum anomaly zone in the Reeser’s Summit Diabase, Pennsylvania: evidence for hydrothermal transport. US Geol Surv Circ 1035: 4–6 Boudreau AE (1988) Investigations of the Stillwater Complex. IV. The role of volatiles in the petrogenesis of the J-M Reef, Minneapolis adit section. Can Mineral 26: 193–208 Boudreau AE (1993) Chlorine as an exploration guide for the platinum-group elements in layered intrusions. J Geochem Explor 48: 21–37 Boudreau AE (1995) Fluid evolution in layered intrusions: evidence from the chemistry of halogen-bearing minerals. In: Thompson JFH (ed) Magmas, Fluids and Ore Deposits. Min. Assoc. Can. Short Course Series 23, pp 25–45 Boudreau AE, Kruger FJ (1990) Variation in the composition of apatite through the Merensky Cyclic Unit in the western Bushveld Complex. Econ Geol 85: 737–745 Boudreau AE, MCCallum IS (1986) Investigations of the Stillwater Complex: III. The picket Pin Pt=Pd deposit. Econ Geol 81: 1953–1975 Boudreau AE, MCCallum IS (1992) Concentration of platinum-group elements by magmatic fluids in layered intrusions. Econ Geol 87: 1830–1848 Boudreau AE, Mathez EA, McCallum IS (1986) Halogen geochemistry of the Stillwater and Bushveld Complexes: evidence for transport of the platinum-group elements by Cl-rich fluids. J Petrol 27: 967–986 Boudreau AE, Love C, Hoatson DM (1993) Variation in the composition of apatite in the Munni Munni Complex and associated intrusions of the West Pilbara Block, Western Australia. Geochim Cosmochim Acta 57: 4467–4477 Boudreau AE, Love C, Prendergast MD (1995) Halogen geochemistry of the Great Dyke, Zimbabwe. Contrib Mineral Petrol 122: 289–300 Brown AC (1993) Sediment-hosted stratiform copper deposits. In: Sheahan PA, Cherry ME (eds) Ore deposit models, Vol II. Geosci. Can Reprint Ser 6: 99–115 Brown AC (2005) Refinements for footwall red-bed diagenesis in the sediment-hosted stratiform copper deposits model. Econ Geol 100: 765–771 Brown AC (2006) Close linkage of copper (and uranium) transport to diagenetic reddening of ‘‘upstream’’ basin sediments for sediment-hosted stratiform copper (and roll-type uranium) mineralization. J Geochem Explor 89: 23–26 Cawthorn RG, Merkle RKW, Viljoen MJ (2002) Platinum-group element deposits in the Bushveld Complex, South Africa. In: Cabri LJ (ed) The geology, geochemistry, mineralogy and mineral beneficiation of platinum-group elements. Can Inst Min Metall Spec 54: 389–429 Cook NJ, Wood SA (1994) Platinum-group minerals in the Lac Sheen Cu–Ni–PGE prospect, Quebec. Can Mineral 32: 703–712 Coveney RM Jr (2000) Metalliferous shales and the role of organic matter with examples from China, Poland and the United States. In: Giordano TH, Kettler RM, Wood SA (eds) Ore genesis and exploration: the roles of organic matter. Rev Econ Geol 9: 251–280
Mobility of palladium chloride complexes in mafic rocks
95
Crocket JH (1990) Noble metals in seafloor hydrothermal mineralization from the Juan de Fuca and mid-Atlantic ridges: a fractionation of gold from platinum metals in hydrothermal fluids. Can Mineral 28: 639–648 Dahlberg EH, Saini-Eidukat B (1991) A chlorine-bearing phase in drill core of serpentinized troctolitic rocks of the Duluth Complex, Minnesota. Can Mineral 29: 239–244 Elding LI (1972) Palladium(II) halide complexes. I. Stabilities and spectra of palladium(II) chloro and bromo aqua complexes. Inorg Chim Acta 6: 647–651 Eugster HP (1985) Oil shales, evaporites and ore deposits. Geochim Cosmochim Acta 49: 619–635 Farrow CEG, Watkinson DH (1992) Alteration and role of fluids in Ni, Cu and platinumgroup element deposition, Sudbury Igneous Complex contact, Onaping-Levack area, Ontario. Mineral Petrol 46: 67–83 Foustoukos DI, Seyfried WE Jr (2005) Redox and pH constraints in the subseafloor root zone of the TAG hydrothermal system, 26 N Mid-Atlantic Ridge. Earth Planet Sci Lett 235: 497–510 Gammons CH, Bloom MS, Yu Y (1992) Experimental investigation of the hydrothermal geochemistry of platinum and palladium: I. Solubility of platinum and palladium sulfide minerals in NaCl=H2SO4 solutions at 300 C. Geochim Cosmochim Acta 56: 3881–3894 Hanley JJ, Mungall JE, Bray CJ, Gorton MP (2004) The origin of bulk and water-soluble Cl and Br enrichments in ore-hosting Sudbury breccia in the Fraser copper zone, Strathcona Embayment, Sudbury, Ontario, Canada. Can Mineral 42: 1777–1798 Hanley JJ, Mungall JE, Spooner ETC, Pettke T (2005a) Fluid and melt inclusion evidence for platinum-group element transport by high salinity fluids and halide melts below the J–M reef, Stillwater Complex, Montana, U.S.A. In: T€orm€anen TO, Alapieti TT (eds) Platinumgroup elements from genesis to beneficiation and environmental impact. Extended Abstracts of the 10th International Platinum Symposium, August 8–11, 2005, Oulu, Finland, 94–97 Hanley JJ, Mungall JE, Pettke T, Spooner ETC, Bray CJ (2005b) Ore metal redistribution by hydrocarbon-brine and hydrocarbon-halide melt phases, North Range footwall of the Sudbury Igneous Complex, Ontario, Canada. Mineral Dep 40: 237–256 Haynes DW (1986) Stratiform copper deposits hosted by low-energy sediments: II. Nature of source rocks and composition of metal-transporting water. Econ Geol 81: 266–280 Hinchey JG, Hattori KH (2005) Magmatic mineralization and hydrothermal enrichment of the High Grade Zone at the Lac des Iles palladium mine, northern Ontario, Canada. Mineral Dep 40: 13–23 Janecky DR, Seyfried WE Jr (1986) Hydrothermal serpentinization of peridotite within the oceanic crust: experimental investigations of mineralogy and major element chemistry. Geochim Cosmochim Acta 50: 1357–1378 Kinloch ED (1982) Regional trends in the platinum-group mineralogy of the Critical Zone of the Bushveld Complex, South Africa. Econ Geol 77: 1328–1347 Kucha H (1981) Precious metal alloys and organic matter in the Zechstein copper deposits, Poland. Tschermaks Mineral Petrogr Mitteil 28: 1–16 Kucha H (1982) Platinum-group metals in the Zechstein copper deposits, Poland. Econ Geol 77: 1578–1591 Kucha H (1985) Platinum-group metals in the Zechstein copper deposits, Poland – a reply. Econ Geol 80: 519–520 Kucha H, Przybylowicz W (1999) Noble metals in organic matter and clay-organic matrices, Kupferschiefer, Poland. Econ Geol 94: 1137–1162 Li C, Naldrett AJ (1993a) High chlorine alteration minerals and calcium-rich brines in fluid inclusions from the Strathcona Deep Copper Zone, Sudbury, Ontario. Econ Geol 88: 1780–1796
96
S. A. Wood and C. Normand
Li C, Naldrett AJ (1993b) Platinum-group minerals from the Deep Copper Zone of the Strathcona deposit, Sudbury, Ontario. Can Mineral 31: 31–44 Marcantonio F, Reisberg L, Zindler A, Wyman D, Hulbert L (1994) An isotopic study of the Ni–Cu–PGE-rich Wellgreen intrusion of the Wrangellia Terrane: evidence for hydrothermal mobilization of rhenium and osmium. Geochim Cosmochim Acta 58: 1007–1017 McCallum ME, Loucks RR, Carlson RR, Cooley EF, Doerge TA (1976) Platinum metals associated with hydrothermal copper ores of the New Rambler Mine, Medecine Bow Mountains, Wyoming. Econ Geol 71: 1429–1450 Mertie JB Jr (1969) Economic geology of the platinum metals. United States Geological Survey Professional Paper, vol 630, p 120 Mogessie A, Stumpfl EF, Weiblen PW (1991) The role of fluids in the formation of platinumgroup minerals, Duluth Complex, Minnesota: mineralogic, textural and chemical evidence. Econ Geol 86: 1506–1518 Mountain BW, Wood SA (1988) Chemical controls on the solubility, transport, and deposition of platinum and palladium in hydrothermal solutions: a thermodynamic approach. Econ Geol 83: 492–510 Nyman MW, Sheets RW, Bodnar RJ (1990) Fluid-inclusion evidence for the physical and chemical conditions associated with intermediate-temperature PGE mineralization at the New Rambler deposit, southeastern Wyoming. Can Mineral 28: 629–638 Olivo GR, Theyer P (2004) Platinum-group minerals from the McBratney PGE-Au prospect in the Flin Flon greenstone belt, Manitoba, Canada. Can Mineral 42: 667–681 Oszczepalski S (1999) Origin of the Kupferschiefer polymetallic mineralization in Poland. Mineral Dep 34: 599–613 Pasava J, Vymazalova´ A, Petersen S, Herzig P (2004) PGE distribution in massive sulfides from the PACMANUS hydrothermal field, eastern Manus basin, Papua New Guinea: implications for PGE enrichment in some ancient volcanogenic massive sulfide deposits. Mineral Dep 39: 784–792 Piestrzynski A, Wodzicki A (2000) Origin of the gold deposit in the Polkowice-West mine, Lubin-Sieroszowice mining district, Poland. Mineral Dep 35: 37–47 Rose AW (1976) The effect of cuprous chloride complexes in the origin of red-bed copper and related deposits. Econ Geol 71: 1036–1048 Rowell WF, Edgar AD (1986) Platinum-group element mineralization in a hydrothermal Cu-Ni sulfide occurrence, Rathbun Lake, northeastern Ontario. Econ Geol 81: 1272–1277 Schiffries CM (1982) The petrogenesis of a platiniferous dunite pipe in the Bushveld complex: infiltration metasomatism by a chloride solution. Econ Geol 77: 1439–1453 Schiffries CM, Skinner BJ (1987) The Bushveld hydrothermal system: field and petrologic evidence. Amer J Sci 287: 566–595 Seward TM, Barnes HL (1997) Metal transport by hydrothermal ore fluids. In: Barnes HL (ed) Geochemistry of hydrothermal ore deposits, 3rd edn. John Wiley & Sons, New York Seyfried WE Jr, Berndt ME, Seewald JS (1988) Hydrothermal alteration processes at midocean ridges: Constraints from diabase alteration experiments, hot-spring fluids and composition of the oceanic crust. Can Mineral 26: 787–804 Springer G (1989) Chlorine-bearing and other uncommon minerals in the Strathcona Deep Copper zone, Sudbury district, Ontario. Can Mineral 27: 311–313 Stribrny B, Wellmer F-W, Burgath K-P, Oberth€ur T, Tarkian M, Pfeiffer T (2000) Unconventional PGE occurrences and PGE mineralization in the Great Dyke: metallogenic and economic aspects. Mineral Dep 35: 260–281 Sverjensky DA (1987) The role of migrating oil field brines in the formation of sedimenthosted Cu-rich deposits. Econ Geol 82: 1130–1141 Sverjensky DA (1989) Chemical evolution of basinal brines that formed sediment-hosted Cu–Pb–Zn deposits. In: Boyle RW, Brown AC, Jefferson CW, Jowett EC, Kirkham RV
Mobility of palladium chloride complexes in mafic rocks
97
(eds) Sediment-hosted stratiform copper deposits. Geol Assoc Canada Spec Paper 36, pp 127–134 Unrug R (1985) Platinum-group metals in the Zechstein copper deposits, Poland – a discussion. Econ Geol 80: 518–519 Vaganov PA, Krasnov SG, Kulikov VD, Pshenichnova TG, Stepanova TV (1995) Gold and platinoid fractionation in oceanic hydrothermal sulfide ores. Geochem Int 32: 118–125 van Middlesworth JM, Wood SA (1999) The stability of palladium(II) hydroxide and hydroxyl-chloride complexes: An experimental solubility study at 25–85 C and 1 bar. Geochim Cosmochim Acta 63: 1751–1765 Vaughan DJ, Sweeney M, Friedrich G, Diedel R, Haranczyk C (1989) The Kupferschiefer: an overview with an appraisal of the different types of mineralization. Econ Geol 84: 1003–1027 Volborth A, Housely RM (1984) A preliminary description of complex graphite, sulphide, arsenide, and platinum group element mineralization in a pegmatoid pyroxenite of the Stillwater Complex, Montana, U.S.A. Tschermaks Mineral Petrogr Mitt 33: 213–230 Volborth A, Tarkian M, Stumpfl EF, Housely RM (1986) A survey of the Pd-Pt mineralization along the 35-km strike of the J–M Reef, Stillwater Complex, Montana. Can Mineral 24: 329–346 Watkinson DH, Jones PC (1996) Platinum-group minerals in fluid inclusions from the Marathon deposit, Coldwell Complex, Canada. Mineral Petrol 57: 91–96 Watkinson DH, Melling DR (1992) Hydrothermal origin of platinum-group mineralization in low-temperature copper sulfide-rich assemblages, Salt Chuck intrusion, Alaska. Econ Geol 87: 175–184 Watkinson DH, Ohnenstetter D (1992) Hydrothermal origin of platinum-group mineralization in the Two Duck Lake intrusion, Coldwell Complex, northwestern Ontario. Can Mineral 30: 121–136 Wedepohl KH (1971) ‘Kupferschiefer’ as a prototype of syngenetic sedimentary ore deposits. Soc Min Geol Japan Special Issue 3: 268–273 Willmore CC, Boudreau AE, Kruger FJ (2000) The halogen geochemistry of the Bushveld Complex, Republic of South Africa; implications for chalcophile element distribution in the lower and critical zones. J Petrol 41: 1517–1539 Wood SA (2002) The aqueous geochemistry of the platinum-group elements with applications to ore deposits. In: Cabri LJ (ed) The geology, geochemistry, mineralogy and mineral beneficiation of platinum-group elements. Can Inst Min Metall Spec 54: 211–249 Wood SA, Samson IM (1998) Solubility of ore minerals and complexation of ore metals in hydrothermal solutions. In: Richards JP, Larson PB (eds) Techniques in hydrothermal ore deposits geology. Rev Econ Geol 10: 33–80 Wood SA, Mountain BW, Pan P (1992) The aqueous geochemistry of platinum, palladium and gold: Recent experimental constraints and a re-evaluation of theoretical predictions. Can Mineral 30: 955–982 Zientek ML, Cooper RW, Corson SR, Geraghty EP (2002) Platinum-group element mineralization in the Stillwater Complex, Montana. In: Cabri LJ (ed) The Geology, Geochemistry, mineralogy and mineral beneficiation of platinum-group elements. Can Inst Min Metall Spec 54: 459–481 Authors’ addresses: Scott A. Wood (e-mail:
[email protected]), Department of Geological Sciences, University of Idaho, Box 443022, Moscow, Idaho 83844-3022, USA; Charles Normand (e-mail:
[email protected]), Departement des Sciences de la Terre et de l’Atmosphere, Universite du Quebec a Montreal, Case postale 8888, Succursale Centre-ville, Montreal (Quebec), H3C 3P8 Canada