Katahdin deposit (Maine), 8.3 + 3.1 in the Moxie pluton. (Maine), 10.8 I 2.3 in the Dumont deposit (Quebec), and. 1 1.4 t 4.8 in the Platinum Reef of the Stillwater ...
Canadian Mineralogist Yol.22, pp. 55-66(1984)
THE DISTRIBUTION OF NICKELAND IRONBETWEENOLIVINEAND MAGMATICSULFIDES IN SOME NATURAL ASSEMBLAGES JOHN F.H. THOMPSON{.ANDSTEPHEN J. BARNES DeparTment of Geology,UniversityoJ Toronto, Toronto, OntarioMSSIAI J. MURRAY DUKE GeologicalSurveyof Canada,601Booth Street,Ottawa,OntarioKIA 088 ABSTRACT The distribution of nickel and iron betweenolivine and bulk sulfide has been determinedin four nickel-copperPGE (platinum-group-element) sulfidedeposits.The mean distribution-coetticient 1f$iff{$!il is 5.0 + 1.7 in the Katahdin deposit(Maine), 8.3 + 3.1 in the Moxie pluton (Maine), 10.8 I 2.3 in the Dumont deposit(Quebec),and 11.4 t 4.8 in the Platinum Reef of the Stillwatercomplex (Montana). The weight of the evidence from this and previousstudiesof natural assemblages, as well asvarious dfuectand indirect experimental determinations, indicates that the distribution coefficient to be expectedfrom the equilibration of olivine and molten sulfide under normal magmaticconditionswill be in the range of 5 to 20, and most likely in the low part of the range. Keywords: nickel, iron, distribution coefficient, natural assemblages, olivine, sulfide melt. SoMNaarns On a d6termin6 la distribution du nickel et du fer entre l'olivine et l'ensembledessulfures dansquatre gites de sulfures de nickel, cuivre et 6lementsdu groupe du platine. Le coefficientde distribution moyen tKffi{$fil estde s.O t 1.7 dans le glte Katahdin (Maine), 8.3 t 3.l dans le pluton Moxie (Maine), 10.8 t 2.3 dans le glte Dumont (Qu6bec)et ll.4 t 4.8 dans le banc platinifbre du complexede Stillwater(Montana).Dansl'ensemble,cesddterminations, les rdsultats ant6rieurs obtenus sur assemblagesnatuels, et diversedeterminationsexp6rimentalesdirectes et indirectesindiquentqu'on peut pr€voir, pour l'6quilibre entre I'olivine et le bain sulfurd dans des conditions magmatiquesnormales, une valeur de Ki1 situde entre 5 et 20, probablementvers la limite inf6rieure. (Traduit par la R6daction)
terpretedas having formed by the separationand accumulation of an immiscible sulfide liquid from silicate parent-magma(Vogt 1918, Scholtz 1937, Hawley 1962,Naldrett 1979,l98l). The composition ofthe sulfideliquid that separatesfrom a given magmais controlledby the bulk compositionof the system,the relative proportions of sulfide liquid, silicatemagmaand crystallinephasessuchasolivine, and the exchangeequilibria governingthe distribution of chalcophile elementsamong the various phases(Maclean & Shimazaki 1976,Duke & Naldrett 1978,Campbell& Naldrett l979,Duke 1979). havebeen Someof the relevantexchange-equilibria studied experimentally,but there are a number of ambiguities in the resulting data (Maclean & Shimazaki 1976, Rajamani & Naldrett 1978, Medvedev& Almukhamedov 1978,Fleet et al. L977, 1981,Boctor 1981,1982).In particular,a controversy hasarisenoverthe distribution of nickel and iron by the betweenolivine and molten sulfide,expressed distribution coefficient
0d*""$(XPJ!"to.roJ
r(;Nieitit= (x?t$ GRIlio,o)
(r)
in which X/ is the mole fraction of componenti in phase,4. Valuesof the distribution coefficient determined directly and indirectly in experimentsand range from less deducedfrom natural assemblages than 5 to more than 60. Opinions regardingthe "correct" valueof the distribution coefficienthavebeen used as a basisto question or support the validity of the magmaticseglegationhypothesisfor the origin of Ni-Cu-PGE sulfide ores (e.g., Fleet 1979, Naldrett 1979).Our purposein this paper is to preMots-clds:nickel, fer, coefficient de distribution, assemsent data on the distribution coefficient in several blagesnaturels, olivine, bain sulfur6, natural assemblagesin which we consider the geological evidence for a magmatic origin for the sulfidesto be conclusive.Our data are derivedfrom INTRoDUCTIoN sulfide depositsassociatedwith four different intrusivebodiesthat representa rangein composition Sulfide deposits that contain nickel, copper, cobalt of the parent magma and in petrological enand platinum-group elements occur in association vironments: the sulfide-bearingMoxie and Katahdin with mafic and ultramafic rocks, and have been ingabbrosin Maine, the platinum reef of the Stillwater *p..r*t uaa..ss: SeltrustMining Corporation Pty. Ltd., complex,Montana, and the dunite-hostedDumont 50 St, George'sTerrace,Perth, WesternAustralia 6000. deposit in northwesternQuebec.
55
56
THE CANADIAN MINERALOGIST
GEoLocy oF THEMarxs, SrrLLwersn ANp DunaoNr OcCURRENCES The Moxie and Kstshdin occurrcnces The Moxie pluton is a large mafic intrusive body in central Maine, emplaced during the Acadian orogeny @evonian). The Katahdin gabbro is a small maJicpluton that occursto the southeastof the Moxie pluton and is probably geneticallyrelated to it. Both inlrusive bodieswere formed by the injection of variably fractionated magmas into irregular magma chambers and feeder zones (Thompson 1982).The southernhalf of the Moxie pluton hosts three small accumulationsof sulfide: Black Narrows (BL), Burnt Nubble (BN) and Big Indian Pond (BIP). The sulfide assemblages in theseoscurrences are depletedin Ni and Cu relative to typical accumulationsof magrnaticsulfide in mafic rocks. The Katahdin gabbroicpluton hostsan enormousbody of pyrrhotite (KI), whoseoverlyinglimonitic gossan was onqeexploitedfor iron ore. Thesesulfidesalso havea low tenor of Ni and Cu. Despitethe unusual composition of the sulfide, there is sufficient evidenceto indicate the magmatic origin of theseaccumulations,and their compositionmay be explained by magmatic processes(Thompson & Naldrett, in press). The BL accumulation contains disseminated sulfide interstitial to cumulus olivine, whereasthe BN and BIP sulfides are hosted by olivineplagioclasecumulates.Olivine in the BL accumulation is Fo334a,as comparedwith For3-rein the BN and BIP occurrences.A typical texture illustrating sulfide interstitial to olivine at BIP is shownin Figure I . The KI body is largely massive( > 759o) pynhotite containingeuhedralolivine @os6),plagioclaseand rare chromite. The concentration mechanismof this massof sulfide remainsambiguous,but its associa-
tion with magmatic silicate minerals and the igneous textlues support its magmatic origin (Thompson 1982).Minor secondaryalteration is presentat all the Maine sulfide occurrencesbut is only significant at the BL locality, where serpentinization and chloritization of olivine-plagioclasecumulatesare relatively common. Alteration of this kind may change the composition of primary sulfides @ckstrand 1975,Groves& Keays1979)but is consideredto be quantitativelyinsignificantin this case (Thompson & Naldrett, in press). The Stillwater platinum-palladium horizon The Stillwater complex in Montana contains stratiform sulfide accumulationsat a number of stratigraphiclevels,the most economicallysignificant beingthe platinum-rich zoneor reef (Conn 1979, Bow e/ al. 1982).The reef occurswithin the lower BandedZone of the complex(McCallum et al. 1980), 400 m above the first appearanceof cumulus plagioclase.It consistsof a I to 2 m thick, laterally continuouslayer of plagioclaseand olivine cumulates containing l-290 sulfides that are extremely platinum-rich. The sulfides occur as compositeblebs of pyrrhotite - pentlandite - chalcopyrite - PGE minerals, interstitial to cumulusminerals and enclosed in intercumuluspyroxene (Bow el al. 1982).A typical texture is illustrated in Figure 2. The stratigraphiccontinuity of the reef and the textural relationship of sulfidesto cumulus and intercumulus minerals leave no doubt as to a magmatic origin. The Dumont deposit
The Dumont intrusive complexis a layeredsill of Archean ageoccurring in the Abitibi greenstonebelt in the SuperiorProvince of the CanadianShield.The sill is believedto haveformed by the differentiation of a parent magmaof peridotitic komatiite composition and comprisesa lower Ultramafic Zone and an upper Mafic Zone. The Ultramafic Zone has been subdividedinto three units, which are, from the base upward, the Lower Peridotite Subzone,the Dunite Subzoneand the Upper Peridotite Subzone(Duke sulfidesoc1980).An extensivezoneof disseminated curs within the centralpart of the Dunite Subzone; the strict conformity of the sulfide-enrichedhorizons with the primary modal and cryptic layering is strong evidencefor the magmatic origin of the mineralization (Duke 1982).The externalmorphology of the disseminated sulfide grains is also indicative of a magmatic origin, as illustrated in Figure 3 (seealso Eckstrand 1975).The ultramafic rocks are variably serpentinized, but the primary textures are usually preserved.The mineralogyand pseudomorphously Ftc. l. Sulfide net-texture,Big Indian Pond occurrence, Moxie pluton. Magmatic sulfide is interstitial to cumulus texture of serpentineare indicative of type-3 serpentinization (Wicks & Whittaker L977),i.e., that ocolivine (dark grey). Reflectedlight.
NICKEL AND IRON IN OLIVINE AND MACMATIC SULFIDES
57
Ftc. 2. Olivine-plagioclasecumulate,PGE reef, Stillwater complex.A. PGE-rich sulfide blebs (S) occur within intercumuluspyroxene(B) and interstitial to cumulusolivine (ol) and plagioclase@). Incident light. B. Magmaticsulfide occurring interstitially to cumulus olivine. Transmitted light.
curred in a regimeof falling or constant temperature without significant shearing or nucleation of antigorite. The effectsof regional metamorphismare negligible. Prior to alteration, the mineralizeddunite typically comprisedmore than 9590cumulus olivine, lessthan 3Vocumulussulfide and lessthan2t/o ir^tercumulus chromite and clinopyroxene.The fact that the bulk compositionof unmineralizeddunite very closely matches that of the contained relict olivine indicates that the olivine has retained its primary composition. However, the modal and bulk chemicalcompositionsof the sulfide f'raction have changedduring serpentinization(Eckstrand 1975, Duke, in press).The grainsof magmaticsulfide are polymineralic, and consist of pentlandite -r magnelite + awaruite t heazlewoodite t native copper. The assemblagepentlandite + magnetite + awaruite is the most common. The distribution coefficients given below are therefore not equilibrium values,inasmuchas they are calculatedusing compositions of primary olivine and of the secondary sulfides. There is, however,good evidengeto support the contention that serpentinization has led to an increasein the Ni,/Fe ratio of the sulfide fraction, such that the calculateddistribution-coefficients are somewhatgreater than the primary values. ANeryrrcar
1975)reduction procedure. The method provided satisfactory accuracyand precision, better than I mole 9o forsterite overall. The Ni content of the olivine was determinedby the wavelength-dispersion techniquewith a graphicalmethod of reductionapplied from a seriesof appropriatestandards(Thompson 1982).The errors introduced by this reduction techniqueare insignificant at the concentrationlevels of interest.The extremelylow Ni content of olivine from tJreMaine occurrencespresentedproblems,and the low instnlmental precisionis the major source of drrors in the determinationof Kp. Olivine in the Stillwater rocks containsmuch higher levelsof Ni; the variation in individual samplesis thought to be due to inhomogeneities.Olivine in contact with
TrcnNrquns
Olivine compositions The distribution coefficients for the Maine and Stillwater samples were determined by the same methods. The major-element concentrations in olivine were determinedby energy-dispersiontechniqueson the University of Toronto ARL microprobe using a modified version of the PESTRIPS (Statham
Ftc. 3. Olivine-sulfide cumulate, Dunite subzone,Dumont intrusive complex. The primary magmatic morphology of the sulfide grain is preserved but the minsp4l assemblage, pentlandite + magnetite, results from secondary alteration. The olivine is partly altered to serpentine + brucite + magnetite. Reflected ligbt.
58
THE CANADIAN MINERALOGIST
sulfide was analyzed wherever possible in order to obtain the best possibleapproach to equilibrium. Olivine compositionsfrom the four Maine sulfide localitiesand the Stillwater PGE reef are shown in Table 1.
The composition of the Dumont olivine was determined using a MAC electron microprobe at the GeologicalSurvey of Canada. The concentrations of SiOr, FeO, MnO, MgO and CaO weremeasured with the instrument operatedin the energy-dispersien
TABLE I. COMPOSITIONOF OLIVINE FROM MOXIE, KATAHDIN AND STILLV/ATER MOXIE PLUTON SAMPLE Analyses SiOz FeO MnO Mgo Nio
ToTAL Fo
BL2 6 40.29 15.58 0.15 43,92 0.090 100.03 83.4
SAMPLE Analyses88644 Sioz Feo Mno M8o Nio ToTAL Fo
(0. 14) (0.82) (0.06) (0.52) (0.007) (0.9)
79Mt12
38.37 22.69 0.27 38.46 0.044 99.83 75.2
(0.20) (0.48) (0.03) (0.17) (0.018) (0.4)
BLIOB-I 2? 40.77 14.40 0.t7 44.81 0.074
t00.22 84.7
BLIOB-2
79BN-7
(0.02) (0.55) (0.01) (0.16) (0.001)
40.19 16.37 0.20 43.33 0.080
(0. l4) (0.08) (0.04) (0.01) (0.004)
38.75 19.05 0.22 40,97 0.020
(0. l5) (0.4r) (0.04) (0.V3) (0.014)
(0.4)
82.5
(0.1)
79.3
(0.5)
79BB 38.56 24.32 0.32 37.09 0.022 100.31 7t.r
79BNI 6
798n3 (0.41) (0.26t (0.03) (0.06) (0.009) (0.2)
38.48 23.97 0.34 37.74 0.032 100.56 73.7
(0.19) (0.38) (0.02) (0.43) (0.008) (0.5)
79En7-1 38.47 22.22 0.29 39.09 0.023 100.09 75,8
(0.30) (0.28) (0.06) (0.01) (0.006) (0.3)
38.13 24,7r 0.33 36.96 0.022 I 0 0 .l 5 72.8
( 0 . 3 1) (0.33) (0.03) (0.71) (0.01t) (0.7)
798127-2 37.89 24.96 0.28 36.24 0.043 99.41 77.r
(0.25') (0.01) (0.04) (0.28) (0.013) (0.1)
KATAHDIN SAMPLE Analyses SiOz Feo MnO Mto Nio TOTi ToTAL Fo
78KT50 39.65 19.04 0.32 41.51 0.009 100.55 79.9
(0.r0) (0.18) (0.02) (0.04) (0.004) (0.1)
78K'r62 62
/8KT60 39.10 2r.22 0.25 )9,49 0.020 100.08 76.8
E2
(0.41) (0.29') (0.03) (0,47) (0.008)
19.79 18.56 0.28 41.48 0.016
@.37, @.27) (0.04) (0.46) (0.009)
39.08 23.18 039 37,85 0.014
(0.21) rc35) (0.06) (0.18) (0.004)
(0.4)
80.1
(0.3)
74.5
Q.2)
STILLVATER SAMPLE Analyses24854 Sioz Feo Mso Nio ToTAL Fo
38.54 20.23 41.83 0382 100.98 78.3
SAMPLE Analyses4463 sioz Feo Mso Nio ToTAL Fo
346 (0.51) (0.48) (0.36) (0.018)
IzE-MI
39.45 2r,09 39.59 0.339 100.47 7 7. 0
(0.46) (0.32) (0.20) (0.01r)
4W-C2A 38.59 25.92 35.50 0.498 100.51 7O,8
(0.52) (0.53) rc.42) (0.014)
4W-C3 39.57 19.04 40.90 0.560 100.03 79.4
I2\('-BI 38.2v 24.03 )7.08 0.42r 99,76 73,4
(0.43) rc.)6) (0.18) (0.057)
(0.48) @.21, (0.41) (0.018)
8E-K3 '8.E3 24,78 t6.2r 0.438 100,26 72.4
All concentrations in weight p€rcenti Fo = mole percent forsterite.
(0.24') (0.40) (0.43) (0.019)
4W-D2B 38.68 22.53 39.45 0.473 l0l.l3 76.1
(0.44) (0.15) (0.25') (0.014)
8E.K3A 38.46 25.09 )6.53 0,422 100.50 72.2
(0.21) (0.15) (0.19) (0.013)
LZE-L39.54 r9.r7 41.00 0.395 100.10 79.2
(0.r1) (0.19) rc36) (0.011)
NICKEL AND IRON IN OLIVINE AND MAGMATIC SULFIDES
59
TABLE2. COMPOSITION OF SULFIDEFROMMOXIE.KATAHDINAND STILL\YATER BL BN BIP KI STILLVATER Ni 1 . 8 7 ( 0 . 4 1 ) 0 . 4 4 (0.04) 0 . 8 4 (0.08) 0 . 2 1 (0.004) 9.3 (0.4r) (0.003) 6.9 (0,46) 0 . 7 2 ( 0 . 2 2 ' o . 3 2 ( 0 . 0 7 ) 0 . 6 5 ( 0 .1 1 ) 0 . l 0 ( 0 . 0l ) Co 0 . 2 2 ( 0 . 0 4 ) 0 .t 3 ( o .o l ) 0 . 1 8 (0.004) NiS/Fes
0 . 0 3 0 3( 0 . o t z t ) 0 . 0 0 6 9( 0 . 0 0 1 2 )0 , 0 t 3 3 @ . 0 0 2 6 )0 . 0 0 3 3( 0 . 0 0 0 r ) 0 . r 8 0 ( 0 . 0 1 )
mode, and the concentrationof NiO was determined using wavelength-dispersionanalysis. A large number of analysesof olivine from sulfide-free dunites from Dumont have beencarried out; the excellent agreement between these results and the whole-rock compositionsconfirms the accuracyof the method. The low standard deviations associated with the averageanalysesreported in Table 3 (see below) reflect the extremehomogeneityof the olivine at Dumont.
danceof sulfidesin individual samplesis generally lessthan 570, making the calculation of a sulfide norm unreliable. However, a strong positive correlation is observedbetweenwhole-rock Ni, Cu, PGE and S (Barnes,unpubl. data) for over 40 sulfidebearingsamplesfrom the PGE reef. Ni, Cu and Co concantrationsfor eachsamplemay, therefore, be determinedby normalizing to 3590 S, the concentration of S assumedfor the pure sulfide fraction. Iron is then found by differencefor the sulfidefraction. The calculationsinvolve a correction fbr silicate Sulfide compositions Ni obtained from the Ni content of olivine and the modal proportion of olivine in each sample. T'he The direct analysisof the sulfide fraction in a uncertaintyin this calculationintroducesa maximum sulfide-bearing rock present$ major problems. error of 30Voin the Ni-in-sulfide value for eachsamDirectJeachingtechniques-(Dlavis 1972,Lynch 1971, ple and is the largestcontribution to the variancein Czamanske& Ingamells 1970, Karapdtyan 1968) the Stillwater [NiS]/[FeS] ratio. The mean and were investigatedfor individual samplesfrom the standard deviation for Ni, Cu and Co, and the Maine deposits,but wereunsuccessfglowing to the [NiS]/[FeS] ratio of the sulfide fraction for the minor dissolution of silicates. The approach Maine and Stillwater occurrences,are given in Table ultimately adopted was to calculatethe composition of the sulfide fraction from whole-rock analytical The compositionsof grains of magmaticsulfide data, or from the proportions of sulfides of a in the Dumont samplesweredeterminedby combinpredetermined composition. Ni, Cu, Co and S ing the modal proportions and chemicalcomposianalyses were carried out by X-Ray Assay tions of the constituentminerals in the polymineralic LaboratoriesLtd. on sulfide-bearingsamplesfrom grains. The modal proportions of pentlandite, BL, BN, BIP and Stillwater usingX-ray-fluorescence magnetite,awaruite and native copper were detertechniques.Assay data provided by the Superior mined using a Quantimet image analyzer. The Mining Company were used for the KI deposit. In mineral compositions were determined by wavethe caseof the Maine deposits,the abundanceof length-dispersionelectron-microprobeanalysis.A sulfides for each sample was determined by a large number of analysesof the sulfide and alloy "sulfide norm" calculation,whereNi, Cu, Co and mineralsfrom the mineralizedpart of the intrusive S are apportioned to normative chalcopyrite, body havebeencarried out; they show that the range pentlanditeand pyrrhotite compositions.The resul- of compositionsis very limited. Therefore, average tant Ni, Cu, Co, Fe and S concentrationsare nor- mineral compositionshavebeenusedin calculating malizedto 10090sulfide (Table2). This methodhas the bulk-sulfidecompositionsin Table 3. The uncerbeenusedpreviouslyin the stgdy of magmaticsulfide tainties quotedreflect only the variancein the average deposits(Hoffman et al. L978). The effect of silicate compositions.An undeterminedbut probably much Ni on this calculation is minor, owing to the very larger uncertaintyarisesfrom possibleerrors in the low concentrationsof Ni in olivine and the low pro- modal analyses. portion of olivine in the majority of samples.A maximum error of 590 of the amount presentmay be Results introduced from the silicates in the whole-rock analysis(Thompson 1982).An additional seriesof The sulfide compositionsvary considerablyamong Moxie sampleswas analyzed after physical separa- the occurrences discussed here. The scale of tion of the sulfide component.The Ni, Cu and Co equilibration betweenolivine and sulfide is difficult concentrationsof the sulfide sepaxatesfell within the to definefor a particular sulfide accumulation,parrangedeterminedby the abovemethod for eachoc- ticularly at low concentrationsof sulfide. Results currence,further justifying the approach. from the Maine and Stilwater examples are, In the caseof the Stillwater samples,the abun- therefore,presentedin terms of mean samplecom-
60
THE CANADIAN MINERALOGIST
positions of olivine relative to mean compositions equilibration betweenan immiscible sulfide liquid of sulfide for eachaccumulation.This approachis and a silicatemagma with olivine on the liquidus. reasonablein terms of the magmatic model of Data are plotted from eachsamplefor eachdeposit in Figures4 through 6. For the Maine and Stillwater deposits,the comlosition of the sulfidesis taken to be constant for each deposit, and the horizontal error-bar representsplus or minus twice the standard 6.0 error of the meanfor samplevaluesof [NiS]/[FeSl. Vertical error-barsrepresent95.590confidencelimits on [NiO]/[FeO] calculatedusingtwo standarddevia50 tions on NiO and FeO determinations for the mean compositionsof olivine in eachsample.The olivine .variance in the Maine examplesis clearly the major 40 sourceof error in the determination of the Kp, par.E at low [NiS]/[FeS] ratios where K, lines ticularly n converge.The valuesofK2 calculatedfrom the ac; tual data-points,however, all fall between3.7 and oxo .!, 14.8. Determinationsof the sulfide compositionin inz dividual samplesfor the Dumont body are illustrated 4 in Figure 6. In the context of the Dumont values, we emphasizethat the sulfide compositionshavebeen somewhatmodified by serpentinization.It is likely that the primary sulfide fraction contained more sulfur and lessorygen than the presentcompositions. Moreover, Duke (in press)has argued,on the basis calculations,that the relativemetal of mass-balance o oot 0.@ 0.6 o.o4 content of the sulfideshasincreasedin the order iron, NiS/FeS in sutfide nickel and cobalt. We infer that the Ni/Fe, and thus Ftc. 4. NiO,/FeOratio of olivine plotted againstNiS/FeS the value of ffit-$$, has increasedduring serpenratio of sulfide for the Moxie pluton (BL, BN, BIP) tinization. This is supported in a qualitative sense and Katahdin @I) occurrences.Lines of constantdisthe distribution coefficientsin Table by comparing tribution-coefficient are indicated for values of Kiz of 3 with the specificgravitiesof the samples.Serpen5. 10 and 20. Ko=5
o
!o
o.oz
o
u 9_ z
NiS,zFeSin sulfide Frc. 5. NiO/FeO ratio of olivine plotted againstNiS,/FeSratio of sulfide for the PGE reef of the Stillwater complex. Lines of constant distribution-coefficient are indicatedfor valuesof Kp of 1,5, 10, 20 and 40.
6l
NICKEL AND IRON IN OLIVINE AND MAGMATIC SULFIDES
tinization leadsto a decreasein rock density;in the caseof the Dumont samples,there is a direct linear relationshipbetweenspecificgravity and the modal percentageof unalteredprimary minerals.The least alteredsamples,with specificgravitiesgreaterthan 2.7, have,I$N/_9.5" of l0 or less,whereasthe samples with lower specificgravitieshave distribution coefficients of 11 or more. Nine of the ten distributioncoefficients fall in the range from 7 to 15, with a mean value of 10.8 t 2.3. One sample has an anomalouslyhigh coefficient(23.5),but it is sienificant that this sample contains only 0.1 modal Vo sulfide, much lessthan the other samples,and thus its modal analysisis least likely to be representative.
D$cussIoN Theoretical considerationsand experimentalstudies The distribution coefficient may in principle be calculatedfrom the free-energychangeof the reaction FeS + Nisio.sOz= NiS + FeSie.rO2
e)
Thus
rf$ zRlii".o" Ke ^s,r-pr; INiS IFeSig.5O2
rcNi"#y:
where the equilibrium constant K, is given by -AGz K"- R=T
(4)
_-
a\d 7l is the activity coefficient of component i in phase.4, AG, is the free-energychange,R is the gas constant, and T is absolute temperature. The equilibrium constant is plotted as a function of temperaturein Figure 7 using the pure solidsas the standardstatesof all four cornponents.It would be more appropriate to usethe pure liquids as the standard statesfor NiS and FeS,sincewe are interested in the distribution of elementsbetweensolid olivine and molten sulfide. However, the limited data available are impreciseand do not yield internally consistentresults. Clark & Naldrett (1972)studiedthe distribution of nickel and iron between fayalitic olivine and monosulfidesolid solution at 900oCand found that
o
I
2
o
o.2
(3)
0.4 0.6 NiS/FeS in sulfide
0.8
t.o
Ftc. 6. NiO,/FeO ratio of olivine plotted aginst NiS/FeS ratio of sulfide for the Dumont intrusive complex. Lines of constant distribution-coefficient are indicated for valuesof ffp of 5, l0 and 20.
62
THE CANADIAN MINERALOGIST TABLE3. COMPOSITIONs OFOLIVINEANDSULFIDEIN DUMONTDEPOSIT SAMPLE
EI6-II8O
Et6-1380
Et6-1425
Et6-14E0
El6-1r00
Ollvlne Compositions Analyses sio2 Feo Mno M8O cao Nio TOTAL Fo
5 40.86 (0.16) 7.44 (0.11\ 0.09 (0.03) 50.34 (0.23) 0.04 (0.01) 0.391 (0.00E) 99.16 92.3
54 4r.36 (0.31) 7.r0 (0.11) o.08 (0.03) 50.24 (0.3t) 0.07 (0.04) 0.176(0.00r) 99.43
4 0 . 3 2 ( 0 .l l ) 8.13 (0.06) o.13 (0.07) 50.77 (0,06) 0.14 (0.01) 0.1J6 (0.009) 99.65
4t.54 (0.24) 7.18 (0.0r) o.o9 (0.02) 50.53 (0.33) 0.08 (0.02) o.2oo (o.oo9) 99.62
4t.7t (0.32) 6.63 (0.05) 0.09 (0.03) 5t.36 (0.2E) 0.04 (0.02) 0.329 (0.006) 100.16
91E .
92.6
93,2
1.80 56.19 (0.50) r.2.t6 (0.29)
2.00 4J,09 (0.78) t9.42 (0.46')
L70 38.96 (0.88) 21.99 (0.52)
0.67 (0.49) t4.70 (0.27) 15.4E
t,04 (0,76) 22.7t Q.42) 7.74 0.43t (0.012)
l.18 (0.86) 25,72 Q.48) 4.15 0 , 5 6 4( 0 . 0 1 8 )
1t.4 2.64
il.4 2.63
Sultlde Compositions % slf. Fe Ni cu co 5 Nis/Fes Kp Sp. cr. SAMPLE
0,24 44.18 28.23 0.09 0.94 19.37
0.10 (0.67) (0.411 (0.02) (0.6r) (0.16)
0.639 (0.011)
43.E8 (0.83) 24.49 (0.49, o.o, (o.ol) l. 13 (0.82) 24.99 (0.45) 6.08 0.158 (0.015)
2.55
23,8 2,55
EI6.I''O
0.222 Q.OO6\ ll.6 2,62
Et6-t575
El6-1600
4.oo (o.oo)
El6-1760
8.oo (o.oo)
E16-1910
Ollvine compositions 4
Analyses
4
5
SiOz Feo Mno Mgo Cao Nio TOTAL
40.58 (0.1r) 6.82 (0.04) 0.10 (0.04) 50.72 (0.54) 0.09 (0.02) 0.318 (0.008) 98,63
fo
>2.v
4t.90 (0.32, 6.72 (0,14) 0.10 (0.03) 51.04 (0.3r) 0.t2 (0.02) 0.527 (0.0t5) 100.41
40.40 (0.)7) 7.51 (0.11) 0. l0 (0.04) ,0. I I (0.28) 0,07 (0.03) 0.452 (0.008) 98.64
40.06 (0.45) 9.80 (0.19) 0.14 (0.03) 49.57 (0,56) 0. l0 (0.0r) 0.442 (0.008) 1 0 0 i. I
(0.10) ( 0 .l 2 ) (0.03) (0.t2)
0.330 (0.007) 100,70 89,2
90.0
93. I
40.83 10.14 0 .l 4 48.79
Sultide Comp6ltions % sulf. Fe Ni Cu C, s o
1.80 48.69 2t.38 0.04 0.90 19.04 9.9'
(0.6r) (0.391 (0.01) (0,64) (0.35)
NiS/FeS
0 . 4 3 9( 0 . 0 1 0 )
KD Sp.cr.
9,4 2,73
39.99 35.)7 0. l5 0.95 18.74 4. E0
(0.66) {0.42) (0.03) (0.62) (0.34)
0.884 (0.018) rt.3
0.83 49.4r 20.81 0.04 0.87 rE.t7 10.'
1,20
0.66 (0.63) (0.37) (0.01) (0.621 (0.34\
54.16 24,25 0. l6 0.34 ,.34 15.76
(0,27) @.24) (0,03) (0.18) (0.10)
0,421 (0.009)
0.448 (0.005)
7,0 2,77
2.44
14.8'
(0.rr)
olao (o-.:s)
t7.37 (0.32) 13.27 0.276 (0.007) 8.8
xRlbio.ror,YPS$o, ,S$i{?iYis 33.2 + 3.4. The fact that their distribu(6) tqN{9*i= tion coefficient is independentof the relative proxRfiio: x8JSto.ro, poftions of nickel and iron in the system, and somewhatgreaterthan the calculatedequilibriumconstant (Fig. 7), implies that the activity-coefficient being rypically about 3. Since.Klil('9o[iis simply the quotient or r