IntroductionTHE ROLE of external sulfur in the genesis of magmatic Ni-Cu-(PGE) sulfide deposits has been widely debated. Traditionally,ore genesis models emphasize crustal incorporation of sulfuras the key process that triggers sulfide liquid immiscibility ina magmatic system (Lesher and Campbell, 1993; Ripley et al.,1999; Lesher and Burnham, 2001; Lesher et al., 2001). Specif-ically, previous authors envisaged bulk or partial assimilationof sulfide-bearing pelitic rocks (e.g., Ripley and Al-Jassar,1987), coupled assimilation of evaporite (adding sulfate to themagma) and coal (contributing a reductant to convert sulfateto sulfide; Naldrett, 2004), and upward migration of gaseoussulfur compounds from heated sediments into the magma(e.g., Ripley, 1981). Another possibility is cannibalization, ei-ther through physical remobilization or dissolution, of sul-fides deposited in and near a conduit by a previous, possiblyunrelated magma surge, such as at Kabanga, Tanzania (Maierand Barnes, 2010) and at the Mount Keith, Western Australiakomatiite-hosted deposits (Bekker et al., 2009). However, the
spectrum of mineralization styles that are represented in nature may reflect a diversity of ore-forming processes and,potentially, a diversity of S sources (Penniston-Dorland et al.,2008; Bekker et al., 2009; Seat et al., 2009).
The majority of magmatic Ni-Cu-(PGE) sulfide depositsare hosted within sulfide-bearing country rocks (e.g., Noril’sk(Lightfoot and Keays, 2005) and Pechenga (Barnes et al.,2001; Fiorentini et al., 2008)), which are inferred to havebeen assimilated or devolatilized by magmas upon emplace-ment (e.g., Li and Naldrett, 1999; Ripley et al., 2002), therebyextracting sulfur from crustal lithologic units and triggeringsulfide saturation in the magma. However, several exceptionsare documented. As an example, on the basis of δ34S values,Seat et al. (2009) argued that the Nebo-Babel Ni-Cu-(PGE)sulfide deposit in Western Australia reached sulfide satura-tion without incorporation of any significant crustal sulfur.
Metallogenic models have had a critical impact on targetingand exploration. Consequently, the emphasis on external S ad-dition as a major trigger for sulfide saturation in orthomagmaticsystems has directed exploration toward areas with sulfide- orsulfate-bearing country rocks. However, even though crustal
Multiple Sulfur and Iron Isotope Composition of Magmatic Ni-Cu-(PGE) Sulfide Mineralization from Eastern Botswana
MARCO L. FIORENTINI,1,† ANDREY BEKKER,2 OLIVIER ROUXEL,3 BOSWELL A. WING,4 WOLFGANG MAIER,5AND DOUGLAS RUMBLE6
1 Centre for Exploration Targeting, School of Earth and Environment, ARC Centre of Excellence for Core to Crust Fluid Systems, The University of Western Australia, Crawley, WA 6009, Australia
2 Department of Geological Sciences, University of Manitoba, Winnipeg, Manitoba, Canada R3T 2N23 European Institute for Marine Studies, Technopôle Brest-Iroise, France
4 Department of Earth and Planetary Sciences, McGill University, 3450 University Street, Montreal, Quebec, Canada H3A 2A75 Department of Geology, University of Oulu, Linnanmaa, 90014 Oulu, Finland
6 Geophysical Laboratory, Carnegie Institution, Washington
AbstractWe report here high-precision multiple sulfur and iron isotope compositions for a series of mineralized sam-
ples from Ni-Cu-(PGE) sulfide deposits in the Archean Tati greenstone belt and the Phikwe Complex of east-ern Botswana. Mineralized samples from the Phoenix and Selkirk Ni-Cu-(PGE) deposits in the Tati greenstonebelt display slightly positive δ34S isotope values, ranging from 0.2 to 0.8‰ V-CDT. ∆33S values of sulfides atPhoenix and Selkirk are −0.01 to −0.08‰ V-CDT, suggesting either a dominantly mantle sulfur source or effective eradication of a crustal ∆33S anomaly through equilibration with large amounts of silicate melt. In theSelebi-Phikwe belt, a granite-gneiss terrane with abundant amphibolite lenses of either volcanic and/or intru-sive nature, mineralized lower grade samples from the Phikwe, Phokoje, and Dikoloti Ni-Cu-(PGE) depositshave more variable δ34S values ranging from −3.1 to +0.3‰ and display significant mass independent anom-alies (∆33S values ranging from −0.89 to −0.27‰), suggesting that barren sulfides associated with distal or low-temperature sea-floor hydrothermal activity contributed sulfur to these deposits. Iron isotopes of sulfides fromthese deposits show a relatively small range of negative ∆56Fe values (−0.29 to −0.04‰), consistent with high-temperature fractionations in magmatic systems, with the exception of one sample from the Dikoloti Ni-Cu-(PGE) deposit of the Selebi-Phikwe greenstone belt, which shows a more negative δ56Fe value of −0.61‰,consistent with assimilation of sedimentary or hydrothermal sulfides rather than fractionations in high-tem-perature magmatic systems. Data from this study highlight the complexity and variability that characterize ore-forming processes in magmatic systems. We suggest that the presence of sulfur-bearing lithologic units in hostrocks of mafic and ultramafic intrusions may not be essential toward the assessment of the prospectivity of aprovince to host orthomagmatic nickel sulfides. Geologic settings without any or little sulfur in the stratigra-phy, which have been traditionally neglected in terms of their prospectivity, should thus be revisited and pos-sibly reassessed considering the potential importance of external source of sulfur to generate ore deposits.
Submitted: November 28, 2010Accepted: June 29, 2011
sulfur assimilation is crucial for some ore-forming processes,the potential of terranes, where sulfide- and sulfate-bearingcountry rocks are not known near mafic-ultramafic rocksshould not be neglected. In fact, recent studies have shownthat the location of sulfur assimilation and deposition may notnecessarily be adjacent (Ding et al., 2011), because basaltic andkomatiitic magmas are capable of transporting entrained sul-fide blebs for great distances (Lesher and Groves, 1986; Bre-mond d’Ars et al., 2001; Bockrath et al., 2004; Barnes, 2007;Barnes et al., 2008). Massive sulfides may also be transportedfor significant distances, as discussed in the earliest papers onKambalda (e.g., Lesher, 1989) in numerous papers on Sudburyoffset ores (e.g., Lightfoot and Farrow, 2002) and in severalpapers on Voisey’s Bay (e.g., Evans-Lamswood et al., 2000).
Metallogenic models have traditionally emphasized addi-tion of external sulfur as a key ore-forming mechanism butwithout specifying the sulfur reservoir that has been most ac-cessible to mafic-ultramafic magmas. Conversely, Bekker etal. (2009) argued that in the case of komatiites, giant depositsform when ultramafic magma interacts with hydrothermalmassive sulfide lenses hosted within felsic volcanic rocks, asopposed to interacting with sulfide-bearing shales (cf.Menard et al., 1996). This interpretation, which is largelybased on the application of multiple sulfur isotope data (e.g.,Farquhar and Wing, 2003), provides new insight into the sul-fur source for some Ni-Cu-(PGE) deposits.
In this study, we use multiple sulfur isotope data to investi-gate the sulfur source for magmatic Ni-Cu-(PGE) deposits inthe Tati and Selebi-Phikwe belts of eastern Botswana. Thedeposits in these belts are hosted by magmatic rocks that havelithophile trace element similarities to arc basalts. However,they exhibit a wide range of platinum group element (PGE)contents (Maier et al., 2008). The source of sulfur, which ispoorly constrained for these deposits (Maier et al., 2008),would help to elucidate how these geochemical characteris-tics developed. Our multiple sulfur data suggest that diversemechanisms have led these Botswana magmatic systems tosulfide saturation. Our results thereby provide new insight onthe prospectivity of magmatic provinces where the host rocksare sulfide poor. The findings have thus the potential tobroaden the scope of exploration targets for new Ni-Cu-(PGE) mineralized provinces.
Sulfur and Iron Isotope Systematics of Magmatic SystemsMeasurements of δ34S values have been applied to discrim-
inate the source of sulfur (i.e., mantle vs. crust) in magmaticsystems (e.g., Green and Naldrett, 1981; Naldrett, 1981;Lesher and Groves, 1986;). However, in the Archean, mantle-derived and crustal sulfides are isotopically similar and bothpossess δ34S values in 0 ± 5‰ VCDT range. Significant ex-ceptions have been documented, such as the sulfides hostedin the banded iron formations at Langmuir, in which Greenand Naldrett (1981) reported δ34S concentrations of ca.−8‰, and the komatiite-hosted nickel-sulfides at Alexo, inwhich Naldrett (1966) reported δ34S values up to ca. +6‰.
Sulfur isotope compositions of magmatic sulfides are alsoaffected by mass-dependent fractionation of S isotopes,which is largely controlled by oxygen fugacity and oxidationreactions in magmatic and hydrothermal systems. Variablefractionations are possible, with 32S enrichments in the melt
at low oxidation state due to H2S loss, and 32S depletions inthe melt at high oxidation state due to SO2 loss (de Hoog etal., 2001). However, model calculations suggest that evenwhen the isotopic consequences of these processes are am-plified through Rayleigh fractionation, they produce differ-ences of less than 4‰ from initial magmatic δ34S values (deHoog et al., 2001). In addition, δ34S values of magmatic sul-fides may be modified during regional metamorphism andhydrothermal alteration, although the effect of these postde-positional processes on sulfur isotope composition of sulfidesseems to be relatively small (cf. Oliver et al., 1992; Crowe,1994; Alirezaei and Cameron, 2001; Wagner and Boyce,2006). As a result, equilibrium fractionations correspondingto temperatures of the latter processes can be establishedamong adjacent sulfides (e.g., Crowe, 1994). However, Sec-combe et al. (1981) found little metamorphic modification ofδ34S values.
Conversely, mass independent fractionation of S isotopes(i.e., nonzero ∆33S values) in Archean sulfates and sulfidesprovides a new tracer for geochemical cycling of S on theearly Earth (Farquhar et al., 2000, 2001). The production ofnonzero ∆33S values has been linked directly to photochemi-cal processes in an atmosphere low in oxygen (<10−5 presentatmospheric levels: Farquhar et al., 2000, 2001; Bekker et al.,2004). Since it is produced by an atmospheric source, subse-quent high-temperature, metamorphic, and hydrothermalprocesses can only dilute the original ∆33S signal, but cannotenhance it. Therefore ∆33S values provide a unique signatureof an S source. Once this signature is passed on to a given sul-fur reservoir, it is preserved there unless there is addition ofsulfur with a different composition. Sulfates and sulfides inArchean sediments typically carry large mass independentfractionation of sulfur isotopes, whereas Archean mantle-de-rived sulfides do not have this signature (Farquhar and Wing,2003). The combination of ∆33S with δ34S values provides arobust method to fingerprint sulfur reservoirs in a wide rangeof geologic settings.
Iron-bearing minerals from and bulk-rock analyses of sedi-mentary rocks deposited during the Archean and Proterozoic,such as organic matter-rich, sulfidic shales and iron forma-tions, show a large range of iron isotope values (Rouxel et al.,2005; Dauphas and Rouxel, 2006), which is not observed inclastic sedimentary rocks or igneous rocks (Beard et al.,2003). The large range of iron isotope fractionations of up to5‰, observed in iron formations and black shales, has beenexplained by extensive redox cycling in the Precambrianoceans. Magmatic and mantle-derived rocks, including basaltsand peridotites, show a much smaller range of iron isotopevalues, generally between 0.03 to 0.09‰ relative to interna-tional standard IRMM-14 (Beard et al., 2003; Dauphas et al.,2009). High-temperature magmatic processes such as mantlemelting and silicate and/or sulfide crystallization generallyproduce small, less than 0.4‰, fractionations in iron isotopecomposition (e.g., Schuessler et al., 2007; Dauphas et al., 2009).Rock alteration processes may also produce significant Fe iso-tope variability (Rouxel et al., 2003), whereas metamorphicprocesses do not completely obliterate primary iron isotopecomposition and heterogeneity (e.g., Frost et al., 2007).
Previous studies that applied mass independent fractiona-tion of sulfur isotopes and iron isotopes to decipher the
106 FIORENTINI ET AL.
0361-0128/98/000/000-00 $6.00 106
source of sulfur in magmatic systems focused on Ni-Cu-(PGE) mineralization associated with komatiites (Bekker etal., 2009) and the PGE mineralized Platreef in the BushveldComplex (Penniston-Dorland et al., 2008). In this study, weinvestigate the multiple sulfur and iron isotope signature oforogenic Ni-Cu-(PGE) sulfide-bearing mafic intrusions ineastern Botswana. Maier et al. (2008) proposed an oceanic arcsetting for these intrusions on the basis of their lithophiletrace element patterns.
Geologic BackgroundWe investigated the multiple sulfur isotope signature of min-
eralized samples from selected intrusive Ni-Cu-(PGE) sulfidedeposits in the Tati and Selebi-Phikwe belts of eastern Bot -swana (Fig. 1). Several deposits in these belts are of consider-able economic interest, including the Phoenix, Selebi-Phikwe,
Tekwane, and Selkirk orebodies. The deposits are of variablesize, ranging from 31 million metric tons (Mt) of ore (1.36%Ni, 1.12% Cu) at Phikwe to 0.6 Mt of ore (1.2% Ni, 0.6% Cu)at Tekwane (cf. Maier et al., 2007). Some of the deposits, no-tably Phoenix, contain significant concentrations of platinum-group elements (ca. 5−10 ppm Pt + Pd in sulfides).
The eastern Botswana Ni-Cu-(PGE) deposits may be sub-divided into two groups. The first group of the deposits,hosted by the Phoenix, Selkirk, and Tekwane intrusions oc-curs within and at the periphery of the Tati greenstone belt(Fig. 2). The second group of deposits, comprising Phikwe,Dikoloti, Lentswe, and Phokoje form part of the Selebi-Phikwe mafic-ultramafic belt of intrusions that occur withingneisses of the Limpopo metamorphic belt (Fig. 3), some 200km to the south of the Tati greenstone belt (Gordon, 1973;Baldock et al., 1976).
S & FE ISOTOPE COMPOSITION, MAGMATIC Ni-Cu-(PGE) SULFIDE MINERALIZATION, E. BOTSWANA 107
FIG. 1. Schematic map of the Limpopo belt and adjacent cratons, showing the location of the Tati and Selebi-Phikwegreenstone belts. Modified from Maier et al. (2008).
Selkirk
Sikukwe
Tekwane
Phoenix
2
km
4
Francistown Zimbabwe
Mafic-ultramafic intrusions
Lady Mary FormationPenhalonga FormationSelkirk Formation
FIG. 2. Geologic map of the central portion of the Tati greenstone belt, showing the Phoenix, Selkirk, and Tekwane de-posits. Modified from Maier et al. (2008) and Johnson (1986).
The Tati greenstone belt forms part of the Francistown ArcComplex (McCourt et al., 2004), which is located along thesouthwestern margin of the Zimbabwe craton (Fig. 1). Basedon lithostratigraphic similarities, Carney et al. (1994) corre-lated the volcanosedimentary rocks of the Francistown ArcComplex with the ca. 2.7 Ga Upper Bulawayan greenstones inZimbabwe. Correlation with the adjacent ca. 2.7 Ga Vumbaand Matsitama greenstone belts further supports this age forsedimentary and volcanic units of the Tati greenstone belt(Majaule et al., 1997; Majaule and Davis, 1998; Bagai et al.,2002; Døssing et al., 2009). The Tati greenstone belt com-prises lower greenschist to lower amphibolite facies volcanicand sedimentary rocks intruded by granitoids of unknown age(Fig. 2). The volcanosedimentary succession has been subdi-vided into three formations (Key, 1976). At the base is the1,400 to 1,650 m thick Lady Mary Formation, which com-prises altered komatiite and komatiitic basalts, and lesseramounts of quartzitic schist, calcitic marble, and iron forma-tion. The overlying Penhalonga Formation is of variablethickness (~1 to >10 km) and comprises basaltic, andesitic,and rhyolitic volcanics and volcaniclastic rocks, as well asphyllites, pyrite-bearing black shales, calcitic marbles, andjaspilites. The Penhalonga Formation is overlain by the up to
several kilometers-thick Selkirk Formation, which comprisesdacitic and rhyolitic volcaniclastic rocks and minor amountsof mafic volcanic rocks, quartzites, and quartz-sericite schists.The Selkirk Formation hosts the Phoenix, Selkirk, and Tek-wane meta-gabbronoritic intrusions and the Sikukwe meta-peridotite intrusion (Fig. 2).
The Selebi-Phikwe area forms part of the Limpopo belt, anArchean to early Paleoproterozoic granulite-facies metamor-phic belt situated between the Kaapvaal and Zimbabwe cra-tons (Fig. 1a; cf. Barton et al., 2006). The Limpopo belt com-prises several terranes with different ages ranging from ca. 3.2to 2.6 Ga and tectono-metamorphic histories. These terraneshave been accreted onto the Zimbabwe and Kaapvaal cratonsover a period of ca. 700 m.y. and are separated by wide, steeplydipping shear zones (Barton et al., 2006). The major terranescomprise the Central Zone, Southern Marginal Zone, andNorthern Marginal Zone. The Phikwe Complex is locatedwithin the Central Zone and largely consists of Archean(Zeh et al., 2009) hornblende-bearing and quartzofeldspathictonalitic and trondhjemitic gneisses, which host the mafic-ul-tramafic intrusions of the Selebi-Phikwe belt. Most of thehost rocks for the Selebi-Phikwe belt contain <200 ppm S,but sedimentary rocks with higher S contents reaching sev-eral 1,000 ppm occur locally (Brown, 1988).
Few age determinations are available to constrain relation-ships in the Selebi-Phikwe area. Granite-gneisses have beendated at 2.6 to 2.65 Ga (McCourt et al., 2004). According toWright (1977) and Brown (1988), the gneisses have intrusiverelationships with the supracrustal rocks, implying that the lat-ter are older than ca. 2.6 Ga. The mafic-ultramafic intrusionsare older than ca. 2.1 Ga, the age of the most recent tectono-metamorphic event in the Limpopo belt that affected themafic-ultramafic intrusions of the Selebi-Phikwe belt (Holzeret al., 1999). However, their absolute and relative ages remainunclear, as contacts between units are mainly tectonic.
Description of the Deposits
Phoenix
The Phoenix deposit is located on the northern periphery ofthe Tati greenstone belt (Figs. 1, 2). The intrusion that hoststhe deposit comprises medium- to coarse-grained, weakly de-formed gabbronorites (Fig. 4). The country rocks are tonaliticparagneisses, which may locally contain accessory chalcopyrite(Key, 1976). Magmatic sulfide mineralization generally occursin disseminated, massive, and veinlike form throughout thegabbronorites (Fig. 4a-c). Disseminated sulfides form submil-limeter- to centimeter-sized intergranular masses of pyrrhotite,chalcopyrite, and pentlandite (Fig. 4a). Massive sulfides forma series of relatively thin (up to 1.5 m) subvertical sheets andveins, which may represent fissure and fracture fillings. Mas-sive and/or vein-type sulfides show considerable composi-tional variation on a centimeter to meter scale. Most samplesare dominated by pyrrhotite containing flamelike lamellaeand granular aggregates of pentlandite, but chalcopyrite-rich(10−90% modal) massive and/or vein-type sulfides also occur.
Selkirk
The medium-grained layered metagabbronorite-anorthositeintrusion that hosts the Selkirk deposit is located in the center
108 FIORENTINI ET AL.
0361-0128/98/000/000-00 $6.00 108
Phikwe
Selebi North
Selebi
Dikoloti
Phokoje
Lentswe4 km
Undifferentiatedpara- and orthogneissAnorthositic gneissGranite gneissUltramafic rocks & ores
FIG. 3. Schematic geologic map of the Selebi-Phikwe area, with locationsof the deposits discussed in the text shown. Modified from Maier et al. (2008)and Marsh (1978).
of the Tati greenstone belt (Fig. 2). The intrusion is hosted bypredominantly dacitic volcanic rocks of the Selkirk Formationand contains a ca. 20 m thick and up to 250 m long lens ofmassive sulfide, which is mantled by a zone of disseminatedsulfides (ca 20 vol %) of uncertain width. Pyrrhotite consti-tutes up to 90 vol % of the massive ores. Pentlandite occursas flamelike lamellae and granular aggregates in pyrrhotite.Chalcopyrite predominantly occurs in the disseminated sul-fides (Fig. 4d). Magnetite locally constitutes up to 15% of the
opaque fraction, occurring as subhedral grains, which may bedistinctly rounded. In some cases, pyrite may constitute ca.5% of the sulfides, forming late-stage veins and subhedralcrystals.
Selebi-Phikwe belt
The Phikwe, Dikoloti, and Phokoje intrusions form part ofthe Selebi-Phikwe intrusive belt (Fig. 3). In all these deposits,the sulfide ores are mainly associated with boudinaged lenses
S & FE ISOTOPE COMPOSITION, MAGMATIC Ni-Cu-(PGE) SULFIDE MINERALIZATION, E. BOTSWANA 109
0361-0128/98/000/000-00 $6.00 109
0.5 cm
A
0.5 cm
C
0.5 cm
B
0.5 cm
D
0.5 mm
0.5 cm
E
0.5 mmF
FIG. 4. Sulfide ores from the Tati and Selebi-Phikwe greenstone belts. A. Medium- to coarse-grained gabbronorite withdisseminated sulfide mineralization. Sample PS 148-215.76, Phoenix. B. Massive pyrrhotite with granular pentlandite andsmall rounded silicate inclusions. Sample P10, Phoenix. C. Chalcopyrite and pyrrhotite veins in gneiss. Sample P5, Phoenix.D. Disseminated chalcopyrite-pyrrhotite ore, collected 4 m below massive sulfide body, Sample S1, Selkirk. E. Massivepyrrhotite ore containing rounded inclusions of silicate rocks and magnetite. Note concentration of chalcopyrite around largeamphibolite inclusion on the right. Sample SP13-1, Phikwe. F. Semimassive pyrrhotite ore with rounded silicate rock inclu-sions. Sample DK 17-2, Dikoloti.
and layers of fine- to medium-grained amphibolite interlay-ered with various types of gneisses, including gray horn-blende-rich paragneiss and pink granite gneiss (Gordon,1973; Key, 1976; Wakefield, 1976; Gallon, 1986; Brown,1988). The inferred parental magmas to the intrusions weretholeiitic basalts (with ca. 8 wt % MgO), which crystallizedvariable proportions of olivine (Fo85), pyroxene (En85), andplagioclase (An83).
Disseminated and massive sulfides occur throughout theamphibolites, locally forming up to 38 m thick massive andsemimassive concentrations, particularly near the contactsbetween the amphibolites and the host gneisses (Fig. 4g). Thesulfides are intensely tectonized, as evidenced by their con-centration in crosscutting veins and in fold closures. Pyrrhotite,pentlandite, and chalcopyrite are the main sulfide mineralsand occur in variable relative proportions within individualsamples of massive and disseminated ore.
Analytical MethodsBetween 2 and 4 mg Ag2S or 1 and 2 mg of sulfide mineral
separate was reacted with elemental fluorine at 25 to 30 torrwith the assistance of a 25W CO2 infrared laser in a vacuumchamber at the Geophysical Laboratory, Carnegie Institutionof Washington. The SF6 so produced was purified by gaschromatography before being introduced into a ThermoElectron MAT 253 mass spectrometer for multiple sulfur iso-tope measurements in a dual-inlet mode (Ono et al., 2006).δ33S and δ34S values are conventional δ notations with respectto VCDT (Vienna-Ca on Diablo Troilite) defined as δxS =1,000[(xS/32S)sample/(xS/32S)VCDT − 1], where x is 33 and 34, re-spectively. ∆33S value (∆33S = δ33S* − 0.515 δ34S*), a measureof mass-independent fractionation of S isotopes, is based onδ33S* and δ34S* values defined as δxS* = 1,000 ln[(dxS/1,000)+ 1], where x is 33 and 34, respectively. The uncertainty (1σ)for δ34S and ∆33S values based on multiple S isotope analysesof CDT material and internal laboratory reference materialsis better than 0.4 and 0.03‰, respectively. A subset of sam-ples was analyzed in the Stable Isotope Laboratory of the De-partment of Earth and Planetary Sciences at the McGill Uni-versity. Sulfide-bearing samples were microdrilled, and the Sfrom the resulting powder was extracted to Ag2S through aCr(II) reduction procedure after Canfield et al. (1986). Theresulting Ag2S was fluorinated at 225°C in a Ni bomb under~20X stoichiometric excess F2 for >9 h. The SF6 so producedwas purified cryogenically and chromatographically and ana-lyzed on a Thermo Electron MAT 253 mass spectrometer formultiple sulfur isotope measurements in a dual-inlet mode.Repeat analyses through the entire analytical procedure re-turn uncertainties (1σ) on δ34S and ∆33S values better than 0.1and 0.01‰, respectively.
Sulfide-rich rocks were crushed in agate mortar and the re-sulting chips were cleaned through several rinses with deion-ized water and ultrasonification. Handpicked mineral sepa-rates were dissolved in HNO3-HCl acid mixture on hot plate.Fe was purified on Bio-Rad AG1X8 anion resin, and iron iso-tope ratios were determined with a Thermo-Electron Neptunemulticollector inductively coupled plasma mass spectrometer(MC-ICP-MS) at Pole Spectrometrie Ocean, IFREMER, fol-lowing previously published methods (Rouxel et al., 2005,2008). The MC-ICP-MS was operated in a medium-resolution
mode, and we used Ni as an internal standard for mass biascorrection. Fe isotope values are reported relative to the stan-dard IRMM-14, using the conventional delta notations (Table1). Internal precision of the data reported in Table 1 is deter-mined through duplicated analysis of the reference standard.Several georeference materials, including banded iron forma-tion (IF-G) and Hawaiian Basalt (BHVO-1) standards, werealso measured. We obtained a δ56Fe value of 0.67‰ for IF-Gstandard, which is similar to previously reported values(Dauphas and Rouxel, 2006). Based on duplicate chemicalpurifications and isotope analyses, the long-term external re-producibility is 0.08‰ for δ56Fe and 0.11‰ for δ57Fe (2σ).
ResultsSulfur isotope data for mineralized sulfide-bearing samples
from eastern Botswana deposits are listed in Table 1 and areshown in Figure 5. Vein-hosted and massive sulfide samplesfrom the Phoenix and Selkirk Ni-Cu-(PGE) deposits in theTati greenstone belt display δ34S values that are only slightly34S enriched (0.2−0.8‰ VCDT) with respect to Canon Dia-blo meteorite (MacNamara and Thode, 1950; Hulston andThode, 1965), which is assumed to reflect the average Earthand mantle composition (e.g., Sakai et al., 1982, 1984; Far-quhar et al., 2002). The ∆33S values of Phoenix and Selkirksulfides are also similar to those of Ca on Diablo meteorite(Hulston and Thode, 1965), with a small range from −0.08 to−0.01‰. δ56Fe data show a small range of negative iron iso-tope values from −0.29 to −0.09‰.
In the Selebi-Phikwe greenstone belt, vein-hosted andbasal massive sulfide samples from the Phikwe, Phokoje, andDikoloti Ni-Cu-(PGE) deposits display S isotope values witha small but significantly larger range of δ34S values (−3.1 to+0.3‰) than that displayed by sulfides from the Tati green-stone belt. In addition, ∆33S values show a significant mass-in-dependent anomaly with respect to the mantle reservoir rang-ing from −0.89 to −0.27‰. Iron isotope data also show alarger range of negative iron isotope values from −0.61 to−0.04‰.
DiscussionAnalysis of multiple sulfur isotopes permits testing of
crustal assimilation models and identification of crustal sulfurreservoirs that contributed to ore genesis. Results from thisstudy on ore samples from various deposits in the greenstonebelts of eastern Botswana indicate that magmas can reach sul-fide saturation through multiple processes. Accordingly, theprocess through which sulfide saturation was attained mayhave affected the tenor and mineralization style recorded inthe various systems. According to Maier et al. (2008), the Ni-Cu-(PGE) deposits in the Tati and Selebi-Phikwe belts showimportant compositional differences as far as the former arerelatively PGE rich, whereas the latter are relatively PGEpoor. In addition, there is also significant difference in theirsize. The deposits of the Tati greenstone belt are relativelysmall (~4.5 Mt of ore at 2.05% Ni and 0.85% Cu) and have ahigh Ni tenor, whereas the deposits of the Selebi-Phikwegreenstone belt are much larger (up to 31 Mt of ore grade) buthave lower tenors. The question arises whether these compo-sitional and size differences could be due, at least in part, todifferent processes involved in reaching sulfur saturation.
110 FIORENTINI ET AL.
0361-0128/98/000/000-00 $6.00 110
S & FE ISOTOPE COMPOSITION, MAGMATIC Ni-Cu-(PGE) SULFIDE MINERALIZATION, E. BOTSWANA 111
0361-0128/98/000/000-00 $6.00 111
Table 1. Sulfur and Iron Isotope Data for Mineralized Samples from the Tati and Selebi-Phikwe Greenstone Belts, Eastern Botswana
δ34S δ33S ∆33Sln
Sample no. Host rock Country rock Mineralization type (‰ V-CDT) (‰ V-CDT) (‰ V-CDT) δ56Fe 1s δ57Fe 1s
Notes: δ33S and δ34S values are conventional δ notations with respect to VCDT defined as δxS = 1,000[(xS/32S)sample/(xS/32S)VCDT – 1], where x is 33 and 34,respectively; δ33S* and δ34S* values are defined as δxS* = 1,000ln([δxS/1,000] + 1), where x is 33 and 34, respectively, and used in ∆33S calculations (∆33S =δ33S* – 0.515δ34S*; cf. Miller, 2002; Bekker et al., 2004); nd = not determined
FIG. 5. δ34S vs. ∆33S data for the studied samples with the schematically shown fields for various Neoarchean sulfur reser-voirs. δ34S values for the mantle sulfur is from Sakai et al. (1982, 1984), Chaussidon et al. (1987), Eldridge et al. (1991), Rud-nick et al. (1993), Farquhar et al. (2002), and Seal (2006) and for Neoarchean massive sulfide lenses from Hannington et al.(1999). ∆33S values for the Archean mantle are inferred from Farquhar et al. (2002), Penniston-Dorland et al. (2008), andBekker et al. (2009) and the Neoarchean massive sulfide lenses are from Farquhar and Wing (2005), Jamieson et al. (2005),and Bekker et al. (2009). Field of δ34S and ∆33S values for Archean komatiite-hosted Ni-Cu-(PGE) deposits is from Bekkeret al. (2009), whereas the abridged field of δ34S and ∆33S values for sedimentary sulfides is from Farquhar et al. (2005) andBekker et al. (2009).
Deposits in the Tati greenstone beltSulfur and iron isotope data from this study apparently do not
show any evidence of crustal signatures. The simplest interpre-tation is that the magmas from which the deposits in the Tatigreenstone belt crystallized did not assimilate any significantcrustal material during generation, transport, or emplacement.For example, the small range in iron isotope values of ~0.2‰is consistent with high-temperature magmatic fractionationprocesses (e.g., Schuessler et al., 2007). Alternatively these mag-mas may have assimilated crustal material, which either was notcharacterized by a significant mass-independent sulfur isotopecomposition or was thoroughly homogenized to erase evidenceof mass-independent fractionation in sulfur isotopes. An addi-tional working hypothesis is that the magmatic sulfides of theTati greenstone belt do not show any crustal signature if a highsilicate to sulfide mass ratio (R factor; Campbell and Naldrett,1979) that characterized these mineralized systems. Follow-ing the argument of Lesher and Burnham (2001), equilibra-tion at a relatively high R factor—as deduced from the metaltenors and mineralization styles of the Phoenix and Selkirkdeposits—may have diluted some of the original crustal S iso-tope signal. We consider this possibility in a separate section.
The range of mass-independent fractionation in sulfur iso-topes is dramatically smaller in sediments deposited between2.7 and 3.2 Ga with respect to a much larger range in sedi-ments deposited between 2.5 and 2.7 Ga (Farquhar et al.,2000, 2007; Ohmoto et al., 2006; Domagal-Goldman et al.,2008). Poor geochronologic constraints allow for sediments ofthe Tati greenstone belt to be in the older age group. How-ever, even if 2.7 to 3.2 Ga sediments are inferred as the mainsulfur source for ultramafic intrusions in the Tati greenstonebelt, the range of δ33S values in these sediments are still anorder of magnitude larger than we measured in our samplesfrom the Tati greenstone belt.
Furthermore, black shales of the Manjeri Formation fromthe Belingwe greenstone belt, Zimbabwe, which are broadlylithostratigraphically and geochronologically correlative withsediments from the Tati greenstone belt, show a large rangein ∆33S values with negative ∆33S anomalies, largely restrictedto early diagenetic pyrite nodules, and highly positive ∆33Svalues typical for disseminated sulfides (Bekker et al., 2008).If shales from the Tati greenstone belt, which are the only sul-fur-rich rocks in the stratigraphy (Key, 1976), exhibit similarisotopic characteristics, the lack of a ∆33S anomaly at thePhoenix and Selkirk deposits suggests that it would be neces-sary to incorporate a mixture of the shale sulfur with an aver-age δ33S value near zero. Total homogenization of this mixtureto mantlelike ∆33S (and δ34S) values should be regarded asfortuitous. We note that if diagenetic pyrite nodules havehighly negative ∆33S values and if disseminated sulfides in thesame samples have highly positive ∆33S values, local assimila-tion of both might generate a sulfide mixture with an appro-priate near-zero ∆33S value. This process would circumventthe physical difficulties required by large-scale incorporationand homogenization of crustal material in a mafic magma.
Deposits in the Selebi-Phikwe greenstone beltSulfur isotope data for mineralized samples of the Selebi-
Phikwe greenstone belt are notably different from those ofthe Tati greenstone belt. Despite a small range of δ34S values
from −3.1 to +0.3‰, which could be explained by mass- dependent processes affecting mantle-derived S, sulfidesfrom the Selebi-Phikwe greenstone belt display a significantnegative ∆33S anomaly, which reflects input from a crustalreservoir. Ripley and Li (2003) outlined that degassing associ-ated with low-pressure emplacement and extrusion, as well aschanges in the redox state of the magma affecting the pro-portion of dissolved sulfide and sulfate, are important factorsfor sulfur isotope variations in mafic magmas and could ac-count up to several per mil δ34S variations.
The negative ∆33S anomaly in the Selebi-Phikwe greenstonedeposits suggests that sulfur may have been assimilated frommassive sulfide lenses, either barren or mineralized, associatedwith felsic rocks or black shales with early diagenetic pyritenodules. Iron isotope data show a larger range of negativeδ56Fe values down to −0.6‰, beyond the expected range ofhigh-temperature magmatic fractionations (+0.2 to −0.3‰;Bekker et al., 2009). These values are consistent with iron as-similation from either hydrothermal sulfide deposits or pyritenodules in organic matter-rich shales, which generally displaynegative δ56Fe values (Rouxel et al., 2005, 2008). A detailedpetrologic study is required to determine if the metamorphicrocks in the footwall of the deposits in the Selebi-Phikwegreenstone belt (e.g., hornblende-bearing gneisses bearingup to 9,400 ppm S and quartzites with up to 47,000 ppm S;Brown, 1988) are consistent with this type of protolith.
Sulfur and iron isotope data from this study support the con-clusion of Maier et al. (2008), who inferred on the basis of traceelement geochemistry that magmas from which the Ni-Cu-(PGE) sulfide deposits at Phikwe, Phokoje, and Dikoloti likelyincorporated a significant component of external sulfur. In thenext section, we consider the implications of low R factors thatapparently characterized the mineralizing systems of the Se-lebi-Phikwe belt for the preservation of crustal isotopic signa-ture recorded in the ore sulfides (Lesher and Burnham, 2001).
Sensitivity analysis of the R factor hypothesis
Geochemical and isotopic signatures of crustal assimilation,either directly through melting or indirectly through de-volatilization, can be damped and potentially erased if silicatemelt-sulfide melt mass ratios (R factors) are high enough (seeLesher et al. 2001; Lesher and Burnham, 2001). This situa-tion would give the false impression that the S in a Ni-Cu-(PGE) sulfide deposit was magmatic. Instead it may simplyindicate equilibration of the sulfide xenomelt with enough sil-icate melt to hide evidence of its crustal origin, even thoughthe deposit contains a mixture of crustal S from externalsources with mantle derived S from the magma. Multiple Sisotopes provide a unique approach for differentiating thevarious magmatic and crustal sources in a mineralizing envi-ronment and may offer a fresh constraint on R factors in Ni-Cu-(PGE) systems. While the limited size of the multiple Sisotope dataset in this pilot study does not permit a quantita-tive assessment of the role that R factors played in setting thespecific S isotope characteristics of Ni-Cu-(PGE) mineraliza-tion in the Tati and Selebi-Phikwe belts, the chemically con-servative nature of δ33S values does allow for broad compar-isons to be made between the two systems.
We examined the implications of R factors for multiple Sisotopes by applying generalized mass-balance expressions for
112 FIORENTINI ET AL.
0361-0128/98/000/000-00 $6.00 112
batch equilibration between a sulfide xenomelt and a sulfur-bearing silicate melt (Lesher and Burnham, 2001). We usethe following equation to calculate the effects of R factors onthe final D33S value of the sulfide melt:
∆33S final sulfide 1—————— = ———————, (1)CS
silicate∆33S initial sulfide 1 + ——— × RCS
sulfide
where ∆33S final sulfide refers to the isotopic composition of thefinal sulfide melt, ∆33S initial sulfide refers to the isotopic compo-sition of the initial sulfide melt, CS
silicate is the concentration ofsulfur in the silicate melt, CS
sulfide is the concentration of sulfurin the sulfide melt, and R is the mass ratio of silicate melt tosulfide melt. Derivation of this equation assumes whole-sys-tem mass balance and full sulfur isotope equilibration be-tween the sulfide and silicate melts. As ∆33S anomalies canonly be diluted, not created, during geologic processes, theratio of ∆33S values in the final sulfide melt to those in the ini-tial sulfide melt can be no greater than one under the as-sumption that primary ultramafic melts are characterized by∆33S values near zero (cf. Bekker et al., 2009). The ex-ploratory calculations in Figure 6 reveal that the resolvingpower of multiple S isotopes extends over a range of R factorsthat is characteristic of magmatic Ni-Cu-(PGE) ores (Lesherand Burnham, 2001).
We do not have estimates of the initial ∆33S value in the sul-fide melt or the final ∆33S value in the silicate melt for eitherthe Tati or Selebi-Phikwe systems so we cannot make a quan-titative assessment of sulfur isotope-based R factors for either
environment. However, we can use some of the measuredmetal contents of putative sulfide melts (Table 1) to constrainpotential variability in the R factors that would be associatedwith both systems. As the sulfide melt-silicate melt partitioncoefficient for Pd (~30,000; Lesher and Burnham, 2001) ismuch larger than the typical R factor for a magmatic deposit,the ratio of the Pd content in the final sulfide melt, [Pd]final sul-
fide, to the Pd content in the silicate melt, [Pd]silicate, can be ap-proximated as follows:
[Pd]final sulfide [Pd]initial sulfide————— ≈ R + ——————,(2)[Pd]silicate [Pd]silicate
where [Pd]final sulfide refers to the Pd content of any initial sul-fide xenomelt (cf. eq. (5) in Lesher and Burnham, 2001).Given the likely Pd contents of representative parental mag-mas and sulfide xenomelts (~6 and ~55 ppb, respectively;table 3 in Lesher and Burnham, 2001), the final Pd ratios forthe Tati deposits can be interpreted essentially as R factors.These are on the order of 1,000. Making the conservative as-sumption that no Pd was contributed by the initial sulfidexenomelt to the Selebi-Phikwe deposits, we estimate that themaximum R factor there was between 10 and 100. These es-timates imply that the Selebi-Phikwe deposits would essen-tially preserve the ∆33S anomaly of the crustal S they incor-porated, while the Tati deposits should retain only 20 to 50%on any original crustal ∆33S anomaly. We recognize that thePd/Pt ratios at Tati are above the range typically expected fora magmatic ore deposit, possibly reflecting hydrothermal up-grading of Pd. The R factors estimated from Pd are thereforelikely to be maxima. For example, a similar exercise with Ptcontents returns R factors that are an order of magnitudelower. If the silicate melt associated with the Tati depositswere already depleted in PGEs, however, much larger R fac-tors could be implied.
As is the case for δ34S values (Lesher and Burnham, 2001),the large disparity between S contents in sulfide and silicatemelts means that the ∆33S anomaly will be one of the lastcrustal signals to be erased through silicate-sulfide melt equi-libration. An R factor explanation for the near-zero ∆33S val-ues in the values of Phoenix and Selkirk sulfides, therefore,are consistent with the conclusion of Maier et al. (2008), whopostulated mainly on the basis of trace element geochemistryand field observations, that the Tati magmas did not containevidence for significant contamination by crustal material.Similarly, our observations of uniformly negative ∆33S valuesin the Tati deposits, rather than a more random distributionaround a mean ∆33S value of zero, are consistent with the Rfactor hypothesis. While the calculations described here arepreliminary since we do not have a reliable estimate for theinitial metal content of the silicate melt, they suggest that ifthe R factor hypothesis is valid, the crustal S incorporatedinto the Tati deposits should be characterized by relativelysmall original ∆33S anomalies (−0.05 to −0.40‰).
As pointed out by Lesher et al. (2001), the compositions ofthe ores and magmatic host rocks in the open-system envi-ronment that characterize most Ni-Cu-(PGE) mineralizingsystems may be decoupled, as a result of interactions betweenof early-formed sulfide liquids with subsequent pulses of pris-tine mantle-derived magma. A more rigorous approach would
S & FE ISOTOPE COMPOSITION, MAGMATIC Ni-Cu-(PGE) SULFIDE MINERALIZATION, E. BOTSWANA 113
0361-0128/98/000/000-00 $6.00 113
0.0
0.2
0.4
0.6
0.8
1.0
∆33
Sfin
al s
ulfid
e m
elt
∆33
Sin
itial
sul
fide
mel
t
1 2 3 4 5
15001000
500
100
log10 R factor
FIG. 6. Graphic solution of the δ33S consequences of complete batch equi-libration between a silicate and sulfide melt. Contours are labeled for repre-sentative constant values of S concentration in parts per million in the silicatemelt. Complete equilibration implies that δ33S values in the final silicate andsulfide melt are equal. We took the sulfide melt to have an S concentrationof 38 wt % (Lesher and Burnham, 2001). Given the magnitude of the aver-age δ33S value of Archean sediments (~1‰) and a typical measurement un-certainty of ~0.01‰, the resolving power of S multiple isotopes spans a rel-atively wide range of R factors.
be to apply N-factor upgrading models (Naldrett, 2004)rather than the single-stage bulk equilibration model de-scribed by the R factor, but the distinction is subtle and doesnot substantially affect the results; in either case, isotopic sys-tems are decoupled from one another to different degrees. Inorder to fully constrain mass-balance effects during contami-nation and sulfide liquid equilibration, a wider range of mon-itors of crustal materials (e.g., Pb-Re-Os isotopes, S/Se ratios)should be applied to ore compositions in these deposits.However, our initial results suggest that for the range of Rfactors applicable to most magmatic Ni-Cu deposits, ∆33S val-ues are potential indicators of presence or absence of a crustalS component.
Concluding RemarksResults from this study suggest that sulfur in deposits situ-
ated within and along the Tati greenstone belt in Botswanamay have been largely derived from the mantle. This impliesthat incorporation of crustal S may not have been an impor-tant mechanism in triggering sulfide saturation in these mag-mas, but that other processes including fractional crystalliza-tion, assimilation of silicates, and changes in pressure andtemperature may have been critical for the mineralizationprocess to form these relatively small, PGE-rich deposits.Similar conclusions have also been reached in studies of otherlarge magmatic sulfide deposits based on δ34S values alone(e.g., the Proterozoic Nebo-Babel deposit in the MusgraveComplex of Western Australia; Seat et al., 2009). The alterna-tive, hypothesis of large-scale batch equilibration betweensulfide and silicate melts is testable by ∆33S analysis of crustalS in the Tati greenstone belt. Relatively small anomaliesshould be present if the ore sulfides preserve coupled metalcontents and S isotope signatures.
Conversely, magmas that formed the deposits of the Selebi-Phikwe greenstone belt underwent significant crustal conta-mination during crystallization and assimilated significantamounts of crustal sulfur during this process. An analog set-ting may be represented by the Eagle Ni-Cu-(PGE) depositin the United States, where recent multiple S isotope studiesindicate a complex history of sulfur assimilation and interac-tion between mafic magma and both Archean and Protero-zoic country rocks (Ding et al., 2011). Although the specificsource of S in the Selebi-Phikwe greenstone belt remains un-determined, this study suggests that ∆33S measurements,even in cases of S-poor terranes hosting mineralization, mayunveil the significant contributions of crustal sulfur that werenecessary to saturate the magma. In this light, it is interestingto note that most of the granitic and gneissic rocks in theLimpopo belt are extremely S poor (< ~200 ppm sulfur; cf.Brown, 1988), and a crustal assimilation model would tradi-tionally not be applied to the deposits of the Selebi-Phikwegreenstone belt. It may be that the gneisses lost most of theirsulfur during high-grade metamorphism after mineralizationformed, thereby creating the appearance that these lithologicunits could not contribute significant amounts of sulfur todrive the mafic-ultramafic magma to sulfide saturation. Thereis little outcropping in the belt, and so the apparent paucity ofsulfide may be misleading.
In magmatic systems, near-zero δ34S or ∆33S values permita mantle source but do not prove it if the full complexity of
homogenization processes is considered. However, nonzeroδ34S or ∆33S values prove a nonmantle source. In conclusion,data from this study highlight the variation that characterizesore-forming processes in magmatic systems. The presence ofsulfur-bearing lithologic units in the footwall hosting maficand ultramafic intrusions should not be considered as essen-tial toward the assessment of the prospectivity of a province.Geologic provinces without any or with little sulfur in thefootwall stratigraphy, which have been traditionally neglectedin terms of their prospectivity, should be revisited and possi-bly reassessed.
AcknowledgmentsWe thank Tati Nickel for facilitating sampling of the
Phoenix and Selkirk deposits. WDM acknowledges fundingby the Centre for Research on Magmatic Ore Deposits at theUniversity of Pretoria. MF acknowledges support fromAMIRA and the Australian Research Council. AB’s participa-tion was supported by NSF grant EAR-937 05-45484, NAIaward NNA04CC09A, and NSERC 938 Discovery grant, andthe TGI-3 Program operated by the Geological Survey ofCanada. BAW was supported by NSERC, FQRNT, and theCRC programs. We are grateful to Steve Beresford for in-sightful comments and thorough criticism. We also thankC.M. Lesher and S.J. Barnes for their very insightful com-ments and revisions.
REFERENCESAlirezaei, S., and Cameron, E.M., 2001, Variations of sulfur isotopes in meta-
morphic rocks from Bamble Sector, southern Norway: A laser probe study:Chemical Geology, v. 181, p. 23−45.
Bagai, Z., Armstrong, R.A., and Kampunzu, A.B., 2002, U-Pb single zircongeochronology of granitoids in the Vumba granite-greenstone terrane (NEBotswana): Implications for the evolution of the Archean Zimbabwe craton:Precambrian Research, v. 118, p. 149−168.
Baldock, J.W., Hepworth, J.V., and Marengwa, H., 1976, Gold, base metals,and diamonds in Botswana: ECONOMIC GEOLOGY, v. 71, p. 139−156.
Barnes, S.-J., Melezhik, V.A., and Sokolov, S.V., 2001, The composition andmode of formation of the Pechenga nickel deposits, Kola Peninsula, north-western Russia: Canadian Mineralogist, v. 39, p. 447−471.
Barnes, S.J., 2007, Cotectic precipitation of olivine and sulfide liquid fromkomatiite magma, and the origin of komatiite-hosted disseminated nickelsulfide mineralization at Mount Keith and Yakabindie, Western Australia:ECONOMIC GEOLOGY, v. 102, p. 299−304.
Barnes, S.J., Fiorentini, M.L., Austin, P., Gessner, K., Hough, R., andSquelch, A., 2008, Three-dimensional morphology of magmatic sulfidessheds light on ore formation and sulfide melt migration: Geology, v. 36, p.655−658.
Barton, J.M., Klemd, R., and Zeh, A., 2006, The Limpopo belt: A result ofArchean to Proterozoic, Turkic-type orogenesis?: Geological Society ofAmerica Special Paper 405, p. 315−332.
Beard, B.L., Johnson, C.M., Skulan, J.L., Nealson, K.H., Cox, L., and Sun,H., 2003, Application of Fe isotopes to tracing the geochemical and bio-logical cycling of Fe: Chemical Geology, v. 195, p. 87–117.
Bekker, A., Holland, H.D., Wang, P.-L., Rumble, D., Stein, H.J., Hannah,J.L., Coetzee, L.L., and Beukes, N.J., 2004, Dating the rise of atmosphericoxygen: Nature, v. 427, p. 117−120.
Bekker, A., Holmden, C., Beukes, N.J., Kenig, F., Eglington, B., and Patter-son, W.P., 2008, Fractionation between inorganic and organic carbon dur-ing the Lomagundi (2.22–2.1 Ga) carbon isotope excursion, EPSL, v. 271,p. 278–291.
Bekker, A., Barley, M.E., Fiorentini, M., Rouxel, O.J., Rumble, D., andBeresford, S.W., 2009, Atmospheric sulfur in Archean komatiite-hostednickel deposits: Science, v. 326, p. 1086−1089.
Bockrath, C., Ballhaus, C., and Holzheid, A., 2004, Fractionation of the platinum-group elements during mantle melting: Science, v. 305, p.1951−1953.
114 FIORENTINI ET AL.
0361-0128/98/000/000-00 $6.00 114
Bremond d’Ars, J.De, Arndt, N.T., and Hallot, E., 2001, Analog experimen-tal insights into the formation of magmatic sulfide deposits: Earth and Plan-etary Science Letters, v. 186, p. 371–381.
Brown, P.J., 1988, Petrogenesis of Ni-Cu ore bodies, their host rocks andcountry rocks at Selebi-Phikwe, eastern Bostswana: Unpublished Ph.D.thesis, UK, University of Southampton, 333 p.
Campbell, I.H., and Naldrett, A.J., 1979, The influence of silicate:sulfide ra-tios on the geochemistry of magmatic sulfides: ECONOMIC GEOLOGY, v. 74,p. 1503−1506.
Canfield, D.E., Raiswell, R., Westrich, J.T., Reaves, C.M., and Berner, R.A.,1986, The use of chromium reduction in the analysis of reduced inorganicsulfur in sediments and shales: Chemical Geology, v. 54, p. 149−155.
Carney, J.N., Aldiss, D.T., and Lock, N.P., 1994, The geology of Botswana:Geological Survey of Botswana Bulletin 37, 113 p.
Chaussidon, M., Albarède, F., and Sheppard, S.M.F., 1987, Sulfur isotopeheterogeneity in the mantle from ion microprobe measurements of sul-phide inclusions in diamonds: Nature, v. 330, p. 242−244.
Crowe, D.E., 1994, Preservation of original hydrothermal δ34S values ingreenschist to upper amphibolite volcanogenic massive sulfide deposits:Geology, v. 22, p. 873−876.
de Hoog, J.C.M., Taylor, B.E., and van Bergen, M.J., 2001, Sulfur isotope sys-tematics of basaltic lavas from Indonesia: Implications for the sulfur cycle insubduction zones: Earth and Planetary Science Letters, v. 189, p. 237−252.
Dauphas, N., and Rouxel, O., 2006, Mass spectrometry and natural variationsof iron isotopes: Mass Spectrometry Reviews, v. 25, p. 515−550.
Dauphas, N., Craddock, P.R., Asimow, P.D., Bennett, V.C., Nutman, A.P.,and Ohnenstetter, D., 2009, Iron isotopes may reveal the redox conditionsof mantle melting from Archean to Present: Earth and Planetary ScienceLetters, v. 288, p. 255–267.
Ding, X., Ripley, E.M., Shirey, S.B., Li, C., and Moore, C.H., 2011, Os, Nd,O and S isotope constraints on the importance of country rock contamina-tion in the generation of the conduit-related Eagle Cu-Ni-(PGE) deposit inthe Midcontinent Rift System, Upper Michigan: Geochimica et Cos-mochimica Acta (in press).
Domagal-Goldman, S.D., Kasting, J.F., Johnston, D.T., and Farquhar, J.,2008, Organic haze, glaciations and multiple sulfur isotopes in the Mid-Archean Era: Earth and Planetary Science Letters, v. 269, p. 29−40.
Døssing, L.N., Frei, R., Stendal, H., and Mapeo, R.B.M., 2009, Characteri-zation of enriched lithospheric mantle components in ~2.7 Ga banded ironformations: An example from the Tati Greenstone Belt, northeasternBotswana: Precambrian Research, v. 172, p. 334−356.
Eldridge, C.S., Compston, W., Williams, I.S., Harris, J.W., and Bristow, J.W.,1991, Isotope evidence for the involvement of recycled sediments in dia-mond formation: Nature, v. 353, p. 649−653.
Evans-Lamswood, D.M., Butt, D.P., Jackson, R.S., Lee, D.V., Muggridge,M.G., Wheeler, R.I., and Wilton, D.H.C., 2000, Physical controls associ-ated with the distribution of sulfides in the Voisey’s Bay Ni-Cu-Co depositLabrador: ECONOMIC GEOLOGY, v. 95, p. 749–770.
Farquhar, J., and Wing, B.A., 2003, Multiple sulfur isotopes and the evolu-tion of the atmosphere: Earth and Planetary Science Letters, v. 213, p.1−13.
——2005, The terrestrial record of stable sulphur isotopes: A review of theimplications for evolution of Earth’s sulphur cycle, in McDonald, I., A.J.Boyce, A.J., Butler, I.B., Herrington, R.J., and Polya, D.A., eds., MineralDeposits and Earth Evolution: London: Geological Society, Special Publi-cation 248, p. 167–177.
Farquhar, J., Bao, H.M., and Thiemens, M., 2000, Atmospheric influence ofEarth’s earliest sulfur cycle: Science, v. 289, p. 756−758.
Farquhar, J., Savarino, J., Airieau, S., and Thiemens, M., 2001, Observationof wavelength-sensitive mass-independent sulfur isotope e¡ects during SO2
photolysis: Implications for the early atmosphere: Journal of GeophysicalResearch-Solid Earth, v. 106, p. 32829−32839.
Farquhar, J., Wing, B.A., McKeegan, K.D., Harris, J.W., Cartigny, P., andThiemens, M.H., 2002, Mass-independent sulfur of inclusions in diamondand sulfur recycling on early Earth: Science, v. 298, p. 2369−2372.
Farquhar, J., Wing, B.A., Johnston, D.T., Ono, S., Canfield, D.E., andHabicht, K., 2005, Multiple isotope effects associated with biogeochemicalnetworks [abs.]: Geochimica et Cosmochimica Acta, v. 69, 10S, SpecialSupplement, p. A215.
Farquhar, J., Johnston, D.T., and Wing, B.A., 2007, Implications of conser-vation of mass effects on mass-dependent isotope fractionations: Influenceof network structure on sulfur isotope phase space of dissimilatory sulfatereduction: Geochimica et Cosmochimica Acta, v. 71, p. 5862−5875.
Fiorentini, M.L., Beresford, S.W., Deloule, E., Hanski, E., Stone, W.E., andPearson, N.J., 2008, The role of mantle-derived volatiles in the petrogene-sis of Palaeoproterozoic ferropicrites in the Pechenga Greenstone Belt,northwestern Russia: Insights from in-situ microbeam and nanobeamanalysis of hydromagmatic amphibole: Earth and Planetary Science Let-ters, v. 268, p. 2−14.
Frost, C.D., von Blanckenburg, F., Schoenberg, R., Frost, B.R., and Swapp,S.M., 2007, Preservation of Fe isotope heterogeneities during diagenesisand metamorphism of banded iron formation: Contributions to Mineralogyand Petrology, v. 153, p. 211−235.
Gallon, M.L., 1986, Structural re-interpretation of the Selebi-Phikwe nickel-copper sulphide deposits, eastern Botswana, in Anhaeusser, C.R., andMaske, S., eds., Mineral deposits of southern Africa: Geological Society ofSouth Africa, p. 1663−1669.
Gordon, P., 1973, The Selebi-Pikwe nickel-copper deposits, Botswana: Geo-logical Society of South Africa Special Publication 3, p. 167−187.
Green, A.H., and Naldrett, A.J., 1981, The Langmuir volcanic peridotite-as-sociated nickel deposits: Canadian equivalents of the Western Australianoccurrences: ECONOMIC GEOLOGY, v. 76, p. 1503−1523.
Hannington, M.D., Poulsen, K.H., Thompson, J.F.H., and Sillitoe, R.H.,1999, Volcanogenic gold in the massive sulfide environment: Reviews inEconomic Geology, v. 8, p. 319–350.
Holzer, L., Barton, J.M., Paya, B.K., and Kramers, J.D.1999, Tectonothermalhistory of the western part of the Limpopo belt: Tectonic models and newperspectives: Journal of African Earth Sciences, v. 28, p. 383−402.
Hulston, J.R., and Thode, H.G., 1965, Variations in the S33, S34, and S36 con-tents of meteorites and their relation to chemical and nuclear effects: Jour-nal of Geophysical Research, v. 70, p. 3475–3484.
Jamieson, J.W., Wing, B.A., Hannington, M.D., and Farquhar, J., 2005, Iso-topic disequilibrium in sulfide mineral pairs from multiple sulfur isotopes[abs.]: Geochimica et Cosmochimica Acta, v. 69, 10S, Special Supplement,p. A450.
Johnson, R.S., 1986, The Phoenix and Selkirk nickel-copper sulphide ore de-posits, Tati Greenstone Belt, eastern Botswana, in Anhaeusser, C.R., andMaske, S., eds., Mineral deposits of southern Africa: Geological Society ofSouth Africa, p. 243−248.
Key, R.M., 1976, The geology of the area around Francistown and Phikwe,northeast and central districts, Botswana: Geological Survey of BotswanaDistr Mem, v. 3, p. 15−25.
Lesher, C.M., 1989, Komatiite-associated nickel sulfide deposits: Reviews inEconomic Geology, v. 4, p. 45−101.
Lesher, C.M., and Burnham, O.M., 2001, Multicomponent elemental andisotopic mixing in Ni-Cu-(PGE) ores at Kambalda, Western Australia:Canadian Mineralogist, v. 39, p. 421−446.
Lesher, C.M., and Campbell, I.H., 1993, Geochemical and fluid dynamicmodeling of compositional variations in Archean komatiite-hosted nickelsulfide ores in Western Australia: ECONOMIC GEOLOGY, v. 88, p. 804−816.
Lesher, C.M., and Groves, D.I., 1986, Controls on the formation of ko-matiite-associated nickel-copper sulfide deposits, in Friedrich, G.H., ed.:Geology and metallogenesis of copper deposits: Proceedings of theTwenty-Seventh International Geological Congress, Berlin, Springer Ver-lag, p. 43–62.
Lesher, C.M., Burnham, O.M., Keays, R.R., Barnes, S.J., and Hulbert, L.,2001, Geochemical discrimination of barren and mineralized komatiites as-sociated with magmatic Ni-Cu-(PGE) sulphide deposits: Canadian Miner-alogist, v. 39, p. 673−696.
Li, C., and Naldrett, A.J., 1999, Geology and petrology of the Voisey’s Bay in-trusion: Reaction of olivine with sulfide and silicate liquids: Lithos, v. 47, p.1–31.
Lightfoot, P.C., and Farrow, C.E.G., 2002, Geology, geochemistry, and min-eralogy of the Worthington offset dike: A genetic model for offset dike min-eralization in the Sudbury Igneous Complex: ECONOMIC GEOLOGY, v. 97, p.1419−1446.
Lightfoot, P.C., and Keays, R.R., 2005, Siderophile and chalcophile metalvariations in flood basalts from the Siberian Trap, Norilsk region: Implica-tions for the origin of the Ni-Cu-PGE sulfide ores: ECONOMIC GEOLOGY, v.100, p. 439−462.
MacNamara, J., and Thode, H.G., 1950, Comparison of the isotopic consti-tution of terrestrial and meteoritic sulfur: Physical Review, v. 78, p.307−308.
Maier, W.D., and Barnes, S.-J., 2010, The Kabanga Ni sulfide deposits, Tan-zania. II: Chalcophile and siderophile element geochemistry: MineraliumDeposita, v. 45, p. 443−460.
S & FE ISOTOPE COMPOSITION, MAGMATIC Ni-Cu-(PGE) SULFIDE MINERALIZATION, E. BOTSWANA 115
0361-0128/98/000/000-00 $6.00 115
Maier, W.D., Barnes, S.J., Chinyepi, G., Barton, J.M., Eglington, B., and Set-shedi, I., 2008, The composition of magmatic Ni-Cu-(PGE) sulfide de-posits in the Tati and Selebi-Phikwe belts of eastern Botswana: MineraliumDeposita, v. 43, p. 37−60.
Majaule, T., and Davis, D.W., 1998, U-Pb zircon dating and geochemistry ofgranitoids in the Mosetse area, NE Botswana, and tectonic implications[ab.]: Geological Survey of Botswana, 50th Anniversary International Con-ference, Abstract Volume, p. 46−48.
Majaule, T., Hall, P., and Hughes, D., 1997, Geochemistry of mafic and ul-tramafic igneous rocks of the Matsitama supracrustal belt, northeasternBotswana—provenance implications: South African Journal of Geology, v.100, p. 169–179.
Marsh, S.C.K., 1978, Nickel-copper occurrences at Dikoloti and Lentswe,eastern Botswana: Geological Society of South Africa Special Publication 4,p. 131−148.
McCourt, S., Kampunzu, A.B., Bagai, Z., and Armstrong, R.A., 2004, Thecrustal architecture of Archean terranes in northeastern Botswana: SouthAfrican Journal of Geology, v. 107, p. 147−158.
Menard, T., Lesher, C.M., Stowell, H.H., Price, D.P., Pickell, J.R., Onstott,T.C., and Hulbert, L., 1996, Geology, genesis, and metamorphic history ofthe Namew Lake Ni-Cu deposit, Manitoba: ECONOMIC GEOLOGY, v. 91, p.1394−1413.
Naldrett, A.J., 1966, The role of sulphurization in the genesis of iron-nickelsulphide deposits of the Porcupine district, Ontario: Canadian Institute ofMining and Metallurgical, Transactions, v. 69, p. 147−155.
——1981, Nickel sulphide deposits: Classification, composition and genesis:ECONOMIC GEOLOGY 75TH ANNIVERSARY VOLUME, p. 628−687.
——2004, Magmatic sulfide deposits: Geology, geochemistry and explo-ration: Heidelberg, Springer, 727 p.
Ohmoto, H., Watabane, Y., Ikemi, H., Poulson, R.P., and Taylor, B.E., 2006,Sulphur isotope evidence for an oxic Archean atmosphere: Nature, v. 442,p. 873−874.
Oliver, N.H.S., Hoering, T.C., Johnson, T.W., Rumble, D., III, and Shank,W.C., III, 1992, Sulfur isotopic disequilibrium and fluid-rock interactionduring metamorphism of sulfidic black shales from the Waterville-Augustaarea, Maine, USA: Geochimica et Cosmochimica Acta, v. 56, p. 4257−4265.
Ono, S., Wing, B., Johnston, D., Farquhar, J., and Rumble, D., 2006, Mass-dependent fractionation of quadruple stable isotope system as a new tracerof sulfur biogeochemical cycles: Geochimica et Cosmochimica Acta, v. 70,p. 2238−2252.
Penniston-Dorland, S.C., Wing, B.A., Nex, P.A.M., Kinnaird, J.A., Farquhar,J., Brown, M., and Sharman, E.R., 2008, Multiple sulfur isotopes reveal amagmatic origin for the Platreef platinum group element deposit, BushveldComplex, South Africa: Geology, v. 36, p. 979−982.
Ripley, E.M., 1981, Sulfur isotopic studies of the Dunka Road Cu-Ni deposit,Duluth Complex, Minnesota: ECONOMIC GEOLOGY, v. 76, p. 619−620.
Ripley, E.M., and Al-Jassar, T.J., 1987, Sulfur and oxygen isotope studies ofmelt-country rock interaction, Babbitt Cu-Ni deposit, Duluth Complex,Minnesota: ECONOMIC GEOLOGY, v. 82, p. 87−107.
Ripley, E.M., and Li, C., 2003, Sulfur isotope exchange and metal enrich-ment in the formation of magmatic Cu-Ni-(PGE) deposits: ECONOMIC GE-OLOGY, v. 98, p. 635−641.
Ripley, E.M., Li, C., and Shin, D., 2002, Paragneiss assimilation in the gene-sis of magmatic Ni-Cu-Co sulfide mineralization at Voisey’s Bay, Labrador:δ34S, δ13C and Se/S evidence: ECONOMIC GEOLOGY, v. 97, p. 1307–1318.
Ripley, E.M., Park, Y.-R., Li, C., and Naldrett, A.J., 1999, Sulfur and oxygenisotopic evidence of country rock contamination in the Voisey’s Bay Ni-Cu-Co deposit, Labrador, Canada: Lithos, v. 47, p. 53−68.
Rouxel, O., Bekker, A., and Edwards, K., 2005, Iron isotope constraints onthe Archean and Paleoproterozoic ocean redox state: Science, v. 307, p.1088−1091.
Rouxel, O., Dobbek, N., Ludden, J.and Fouquet, Y., 2003, Iron isotope frac-tionation during oceanic crust alteration (Site ODP 801): Chemical Geol-ogy, v. 202, p. 155−182.
Rouxel, O., Shanks, W.C., Bach, W., and Edwards, K.J., 2008, Integrated Fe-and S-isotope study of seafloor hydrothermal vents at East Pacific rise9°−10° N: Chemical Geology, v. 252, p. 214−227.
Rudnick, R.L., Eldridge, C.S., and Bulanova, G.P., 1993, Diamond growthhistory from in situ measurement of Pb and S isotopic compositions of sul-fide inclusions: Geology, v. 21, p. 13−16.
Sakai, H., Casadevall, T.J., and Moore, J.G., 1982, Chemistry and isotope ra-tios of sulfur in basalts and volcanic gases at Kilauea Volcano, Hawaii:Geochimica et Cosmochimica Acta, v. 46, p. 729−738.
Sakai, H., Des Marais, D.J., Ueda, A., and Moore, J.G., 1984, Concentrationsand isotope ratios of carbon, nitrogen and sulfur in ocean-floor basalts:Geochimica et Cosmochimica Acta, v. 48, p. 2433−2441.
Schuessler, J.A., Schoenberg, R., Behrens, H., and von Blanckenburg, F.,2007, The experimental calibration of the iron isotope fractionation factorbetween pyrrhotite and peralkaline rhyolitic melt: Geochimica et Cos-mochimica Acta, v. 71, p. 417−433.
Seal, R.R., 2006, Sulfur isotope geochemistry of sulfide minerals: Reviews inMineralogy and Geochemistry, v. 61, p. 633−677.
Seat, Z., Beresford, S.W., Grguric, B.A., Gee, M.A.M., and Grassineau, N.V.,2009, Reevaluation of the role of external sulfur addition in the genesis ofNi-Cu-(PGE) deposits: Evidence from the Nebo-Babel Ni-Cu-(PGE) de-posit, west Musgrave, Western Australia: ECONOMIC GEOLOGY, v. 104, p.521−538.
Seccombe, P.K., Groves, D.I., Marston, R.J., and Barrett, F.M., 1981, Sulfideparagenesis and sulfur mobility in Fe-Ni-Cu sulfide ores at Lunnon andJuan Main shoots, Kambalda; Textural and sulfur isotopic evidence: ECO-NOMIC GEOLOGY, v. 76, p. 1675−1685.
Wagner, T., and Boyce, A.J., 2006, Pyrite metamorphism in the DevonianHunsrück Slate of Germany: Insight from laser microprobe sulfur isotopeanalysis and thermodynamic modeling: American Journal of Science, v.306, p. 525−552.
Wakefield, J., 1976, The structural and metamorphic evolution of the PhikweNi-Cu sulfide deposit, Selebi-Phikwe, eastern Botswana: ECONOMIC GEOL-OGY, v. 71, p. 988−1005.
Wright, L., 1977, A structural cross section across the north margin of theLimpopo belt: Unpublished Ph.D. thesis, UK, University of Leeds.
Zeh, A., Gerdes, A., and Barton, J.M., 2009, Archean accretion and crustalevolution of the Kalahari Craton: the zircon age and Hf isotope record ofgranitic rocks from Barberton/Swaziland to the Francistown arc: Journal ofPetrology, v. 50, p. 933−966.
