Canadian MineralogistVol. 17, w. 417435 (1979)

Arst.necr

We examine the relationship between the bulkcomposition of the sulfide fraction and the abund-ances of Pd, Pt, Ir and Au in volcanic-type nickelsulfide deposits of Western Australia. We also re-view abundance of these elements in samples re-presenting the liquid and olivine components ofthe komatiitic ultramafic sequences that bost thedeposits. Results from representative samples ofore, recalculated, to l00Vo sulfides, show varyingdegrees of positive correlation between the tenorof precious metals and the Ni content. The cor-relation is good for Ir and Pd and weak to moderatefor Pt, whereas Au values show no regular trends.Our study of ore profiles, sulfide stringers and mine-ral separates, however, has shown us that consider-able effort and care is required to establish averageprecious metal values in volcanic-type deposits.Samples representing the high-Me liquid fractionof the magmas that coexisted with original sulfidemelts have Pd:Ir ratios similar to those of thesulfides; this fact indicates that these liquidsexerted the dominant influence on the relativepartitioning of precious metals between silicatemagmas and sulfide melts. A wide range of sulfide-melt compositions were associated with only anarrow rangp of $ilicate-melt compositions; factorsother than the bulk composition of the silicatemelt controlled the composition of the sulfide melt.

So*trvterns

Nous examinons les gisements nickeliftsres detyF volcanique de lAustralie occidentale b deuxpoints de vue: (1) relation entre composition glo-bale des sulfures et teneurs en Pd, Pt, Ir et Au;(2) concentration de ces 6l€ments dans des €chan-tillons repr6sentatifs de la fraction liquide et delolivine des s6quences encaissantes ultramafiques

tForrnerly with Western Mining Corporation Ltd.,Kambalda. Western Australia.

PRECIOUS METALS IN VOLCANIC.TYPE NICKEL SULFIDE DEPOSITSIN WESTERN AUSTRALIA

I. RELATIONSHIP WITH THE COMPOSITION OF THE ORES ANDTHEIR HOST HOCKS

JAMES R. ROSS1 AND REID R. KEAYS,rReseorch School ol Earth Sciences, Australian National University,

Canberra, AustraliazDepartment of Geologl, University ol Melbourne, Parkville, Victoria 3052, Australia

komatiitiques. I-es analyses d'6chantillons repr€-sentatifs, recalculdes en posant la somme des sulfu-res 6gale e 100, montrent une corr6lation positivevaria6le entre les teneurs en m6taux pr6cieux eten Ni: bonne pour Ir, faible l mod6r€e pour Pt'cette corr6lation n'est pas 6tablie pour Au. Notre6tude de profils de minerai. venues de sulfures etconcentrds nous a montr6 qu'il faut beaucoup d'ef-forts et de soins pour 6tablir les teneurs moyennesen m6taux pr€cieux dans les grsements de typevolcanique. Les 6chantillons de la fraction liquideriche en Mg qui coexistait avec les sulfures liquidesoriginels donnent un rapport Pd:Ir semblable- Icelui des sulfures, ce qui'indique que ce sont desliquides qui ont exerc6 I'influence prEpond6rantesui le partage relatif des m6taux pr6cieux entreliquides silicat6s et sulfur6s. Un large domaine decompositions de sulfures tiquides a coexist6. avecun domaine 6troit pour les liquides silicat6s: ce

"irt-pa. la composition globale du liquide-silicat6

qui gouverne la compo'sition du liquide sulfur6'

(Traduit Par la R6daction)

INTnooucnoN

We have addressed five questions relatingto volcanic-type nickel sulfide deposits at K1m-belda and elslwhere in Western Australia: (1)

how should these orebodies be sampled to de-termine their precious-metal contents? (2) whatis the relationship between the abundance ofprecious metals in different ore zones and theiomposition of the ores? (3) what is the rela-tionship between the precious-metal contents ofthe or& and their host rocks? (4) what is thedistribution of precious metals in individual sul-fide phases? (5) what can precious metals tellus ab-out the influence of metamorphism on thecomposition and distribution of the sulfide ores?

Mbre than 1400 individual precious-metalanalyses have been generated during this projectand results afe presented in two parts. Part I

417

4 1 8 THE CANADIAN MINERALOGIST

(this paper) places emphasis on the relationshipof Pd, Pt, Ir and Au to the composition of oresand host rocks, and on their partitioning be-tween original silicate and sulfide melts. Acompanion paper (Part II: Keays et al., inprep.) examings the distribution of preciousmetals within ore zones and host rocks at Kam-balda in more detail, including their partitioningbetween sulfide minerals and between originalsilicate phases.

Gsor,ocrcer- SsrrrNc

Nearly all of the niskel sulfiCe deposits in

Australia occur in Archean greenstone beltswithin the Eastern Goldfields province of theYilgarn Block (Fig. 1), mostly within the Norse-man-Wiluna belt as defined by Williams (1973).Apart from a few exceptions (e.9., Carr Boyd:Schultz 1975), these deposits are associatedwith the komatiitic ultramafic members of thegreenstone belts (an association discussed morefully by Naldretr & Cabri L976). Two princi-pal types of deposit are recognized: (1) thevolcanic type @inns et al. 1977), typified bythin layen of matrix and massive sulfides at,or near, the base of lenses of altered olivineperidotite up to 50 m thick. These lenses usually

Fro 1. Generalized geologicatr map of the Yilgarn Block of WesternAustralia showing the greenstone bel* and the location of the principalnickel sulfide deposits. Those mentioned in text include Kambalda,Forrestania (F), Nepean (N), Redross (R), Scotia (Sc),Spargoville (S),and Windana (W).

NICKEL DEPOSITS IN WESTERN AUSTRALIA 419

occur at the bottom of much thicker sequencesof komatiitic ultramafic flows with lower MgOcontents. The volcanic-type deposits rarely con-tain more than 5 million tonnes and usually lessthan 2 million tonnes of ore. Most have averageore-reserve grades of 2 ZVo Ni and Ni:Curatios of 1G-15:1; (2) the dunitic type, typifiedby internal low-grade disseminations of sulfideswithin altered dunitic bodies up to 900 m thick(e.g., Mt. Keith: Burt & Sheppy 1975), butdeposits with higher grade, near-contact ac-cumulations of sulfides (e.g., Perseverance:

)a

Martin & Allchurch 1975) form a substantialpart of the nickel metal reserves in this class.Individual deposits contain up to 263 milliontonnes of ore (Mt. Keith) with average gradescommonly about 0.6Vo Ni; Ni:Cu ratios usualyoccupy the range 25-6O:.I.

There is widespread acceptance of a mag-matic origin for essentially all of the sulfidesin these deposits, even though both ore andhost rocks have been metamorphosed to gradesranging from greenschist to upper amphibolitefacies (Binns et al. 1977). Initial studies of

4N

I

TOADS

rt6lc rNTtRAEDrAtttNT.ruStVES

\ .\\\

utTSaaafrc sEouENct.

f@Twalr u$tT

Nlcr[! sulHrD8 otf

fAU!I

/

HffiH"{u .

\@ O S lOOOG.r,.t HUNI SHOOT

v v v v v v v - vV V V V V V V V

:f::;-- .vr v v v v v v: : : : : : : : : l i i j V v ' z v v v v v v . v

"J v v v v v v v v \/

. , V V V V V V V V V

Fto. 2. Geological plan of the Kambalda Dome.

420

volcanic-type deposits (Ewers & Hudson 1972,Hudson 1972, Ross & Hopkins 1975) regardedboth matrix and massive sulfides as the productof magmatic accumulation, but recently Binnset al. (1977) and Barett et al. (1977) haveargued for the generation of massives oresfrom disseminated sulfides in some environ-ments of dynamic-style metamorphism (e.g.,Nepean, Windarra) and implied that similarprocesses occurred in environments of static-style metamorphism such as Kambalda.

The regional setting of the Kambalda districthas been illustrated by Gemuts & Theron(L975), whereas the district geology and oredeposits have been described by Woodall &Travis (1969), Ross (1974), and Ross &Hopkins (1975). Most of the deposits areclustered around the Kambalda Dome, at theoutwardly dipping, correct-facing contact be-tween a footwall of tholeiitic metabasalt andan overlying sequence of komatiitic ultramafic

THE CANADIAN MINERALOGIST

rcn

rocks up to 800 m in thickness (Fig. 2). Thelower portion of this seguence consists of thickunits (usually 1G-50 m in thickness) of meta-olivine peridotite (commonly 34-45Vo M*)which are in turn overlain by multiple thin'units (< 10 m) of metapicrite and metaperi-dotite that average about 3O/o MgO. (Notethat all oxide and element abundance are ex-pressed on a volatile-free basis, in recognitionof the large, variable component of COg andH,O in altered ultramafic rocks.) Both typesof units commonly show chill margins andspinifex textures in the upper sections, and thesequence is regarded as essentially extrusive.Abou| 85/o of the pre-mining reserves of nickelmetal occurs in ribbonlike ore zones in directcontact with the footwall basalt and are re-ferred to as contact ores. Typical contact orezones show a layer of matrix sulfides, up to 2 mthick, overlying a thinner and less continuouslayer of massive sulfides, which rest directly on

rAM 1 . EIAJr VAU'89 t! l8 coilmrB sNrg! s cABm w ne.! guqim AF

orts sF! oRff rtt Pd Il h Srtrl Nlscq Ult&.

m 794,394 2.4O ^ 0.2o 0.06 7.73 2e4 3tA _ 363 3.219 11.73 39.65(0.r7sz) (o.016) (o.oo8) (o.7ro)(1r7) (93.6) (150)

Uct{agor A7,773 2.2A 0.20 0.06 7.75 267 312 - 359 j.39t 11.{S 3A.06(0.302) (0.026) (0.005) (1.242)(r2S) (134, (334'

FrsER 335,016 1.86 O.15 O.O5 5.2O 24O 295 - 340 2.79! L2.O5 40.?6(0.243' (0.O2s' (0.007) (0.737)(18) (83.1) (287)

m 734,432 3.73 0,25 0.08 1O.OZ 347 491 - 3!rg 2.697 L4.77 4A.Zg(0.342) (0.022) (0.o10) (r . t i t7) (2o4) (u4) (3to)

m 1491727 3.55 O.2A O.Oa 10.49 277 45A - 273 2.957 L2.@ 45.59(0.521) (0.M2) (0.01s) (1.528)(r4o) (13! l (193'

DORrCru 6t6.o32 3.33 0.24 0.06 7.23 422 553 _ 312 2.r?O 13.S9 53.15(0.312) (0.0I7) (0.008) (0.662)(r93) (132) (r28'

NEPBAN 150,381 3.66 0.23 O.O5 6.64 315 1gg - 293 1.817 15.69 59.5a(0.499) (0.033) (0.010) (0.94o) (132) (109' (229'

m mao3 3.2s o.z2 0.06 s.ss - 363 59 t l3 2.124 L4,44 so.Lg

(b) uEaN

M SB@I

LUlXtOll

Uc!{Af,d

.tt allm

DMWEA}I

KNM BEAD

L2.204 t .O4 0.31

11.60ta.ooS x.o5 0.311 3 . 9 5 1 . 1 6 0 . 3 414.40 0.9? 0.3013.20rs.od r. t9 0.3314.00 1.30 0.3421.20 1.33 0.35L4.20 .96 .26

39.27

39.3440.69

34.9434.4439.0444.35

3 9 . 0 538.46

33.80

I/t45 1615

L402

L7971339

1169

2217LA24

1635 - 1869

22L3 - 25501894 - 1341

1937 - 11533041 - 16A62490 - 169?

1546 25A 494

(I) .Ni, Cu, co, @al S expressed ia r t . r i pd, I . sal Au in plb. AtL ualyBes by ! l . ! , t .C. ulesBlndlcald othefriae. (2) Stedad d.eviatioa i! pseajrheBes. (3) ConlEgite aahpLe of totaLK@balda nil1 fedl, uleLated in tibe to ihe conlbsite s@lles for j.ndiviaml oie sboote.Preciow Eetal- udyses tuon Keay6 & Dayisoa (19?6). (!) iLt values have ken cdcuLatreal fronS:Ni ratio8, except for the second set of dtata fo! MclGhon & Ken (see f5J. $) Recatcul_atioab6ed on Ni ya1@6 seL*ted o! the b@is of nile exlerience ed routiae slecifj.c_gravitydeasue@lta.oB ore. [he higber s yalues leflect ilcollbration of sul?ide-b€uiog eeili.aenta inore duiag oioi.ng.

NICKEL DEPOSITS IN

footwall basalt. The footwall basalt adjacent tothe contact often contains thin stringers ofchalcopyrite-rich sulfides. The sulfides in theore zones have been recrystallized and themassive layer has experienced'varying degreesof remobilization. Some contact ore zones areoverlain by one or more zones of sulfide mine-ralization within the lower third of the overlyingultramafic sequence. This hanging-wall mine-ralization occurs at, or close to, the base ofultramafic units; although it can show most ofthe features of contact ore it usually consistsof thin, elongate zones of disserninated andblebby sulfides. The collective term shoot em-braces all zones, including hanging wall, withineach of the ore environments at Kambalda.This description of the Kambalda deposits isrepresentative of the other volcanic-type de-posits included in this study.

An important but not well recognized featureof the volcanic-type deposits is the variation incomposition of the sulfide fraction of differentore zones. The composition of the sulfide frac-tion of an individual ore zone is relativelyconstant over its areal extent, but it is usuallydifferent from that of other ore zones. Mostore shoots at Kambalda contain only one zoneof contact ore which is always dislocated byfaulting. Where hanging-wall ore zones are pre-sent they show a different composition of sulfidefraction. Normal pyrrhotite-pentlandite contactore zones at Kambalda have S:Ni values inthe range of. 2,24.3:1, corresponding to nickelvalues of about 9 to ISVo in l00Vo sulfides;most of that range is represented in this study.Lower S:Ni values at Kambalda are associatedwith the millerite-bearing ores at Otter Shoot(Keele & Nickel 1974) and Gibbs Shoot, but asthey have not been adequately sampled, thecompositional range has been extended by theinclusion of material from the nickel-rich pent-landite-pyrrhotite ores at Nepean and Scotia(S:Ni values of about 1.8:l and 1.6:1, respec-tively).

Axarvtrcer, Mergoos

Samples analyzed at the University of Mel-bourne were carefully cleaned and crushed in asoft iron mill known to have low precious-metal contents. Gold, Ir, Pd and Co were de-termined by a neutron-activation-analysis(NAA) procedure modified after that ofCrocket et al. (1968) and Keays et al. (1974).The accuracy of the method is estimated tobe *157o; a comparison of analytical data

WESTERN AUSTRALIA 421

for some rock standards with published datamay be found in Shaw er al. (1976). Cu andNi were measured by AAS and S by a Lecoautomatic S titrator. Whole-rock silicate anal-yses (MgO) were obtained by XRF analysisfollowing the method of Haukka & Thomas(1977 ) .

Precious-metal determinations by WesternMining Corporation (WMC) follow the stand-ard method for spectrographic analysis of ores,minerals and rocks by the fire-assay precon-centration technique (ASTM E4W-71) de-scribed in the 1971 Annual Book of ASTMStandards; good comparative results were ob-tained with six other laboratories. Ni, Cu andCo were determined by AAS and routine pre-cision is no less than * 5Vo for Ni and Cu andslightly higher for Co. Sulfur was measured bythe barium sulfate gravimetric technique witha quoted accuracy of -+ 3%.

Pnncrous MEreLs rN NTcKEL Sulrroe Onrs

Four types of samples were obtained: re-presentative samples of ore production, re-presentative sampling of drill core across orezones, samples across ore profiles at LunnonShoot, and mineral separates from massive ore.

Samples of ore production

Ore is produced from several shoots atKambalda and sampled separately by ore sourceat the weighbridge during transit to the primarycrusher. This proeedure takes a sample of ap-proximately 50 kg from every 5G-100 tonnesof ore. At the close of every 4-week period acomposite sample representing total productionfor that period is prepared for each ore source.During 1974-76, a total of 18 composite samplesfrom each of seven ore shoots were analyzedfor Pt, Pd and Au. Six of these shoots (Lunnon,Juan, Durkin, Ken, Fisher and McMahon) areat Kambalda (Fig. 2) and the seventh is thenearby deposit at Nepean (Fig. 1). The ton-nage-weighted means for the 18 samples fromeach shoot are recorded in Table 1a, togetherwith data for a composite sample of ore prod-uction from all sources at Kambalda collectedin 1973. Table lb shows these values normal-ized to 1007o sulfides on the bases of S:Niratios, sulfide-mineral compositions (Ewers &Hudson 1972, Nickel et al. 1974) and pyritecontent (Woolrich & Giorgetta 1978).

Recalculation of analytical data for the sul-fide component of nickel sulfide ores to 100%

.w f,ii.ii.;.::i..-''*

1 2 l o mwT% t{ErEl n m0% $tulDEs

€B

;pE

5E

=g

;==L

422

Ftc. 3. Plot of Pd versas wt. 7o Ni in 100% sul-fides. The large dots represent carted ore samples(Table l), middle-sized dots represent concen-trates (Table 2), and small dots represent coresamples and hand specimens (Tables 2 and 3),Symbols are: D Durkin Shoot, F Fisher Shoot,Fo Forrestania, H Hitura, J Juan Shoot, K KenShoot, K.H. Kambalda Head, L Lunnon Shoot,L.C. Lunnon contact, M McMahan Shoot, NNepean, P Perseverance, R Redross, S Scotia,Sh Shangani, Su Sudbury, W Windarra. The re-gression line (r - 0.88) is for carted ore sam-ples only. Individual composite samples forLunnon, Juan, Durkin and Nepean fall withinthe large shaded field. The smaller field representsindividual samples of Lunnon contact ore.

wT% ilctct til t0096 sutFItrs

Ftc. 4. Plot of Pt versus wt. 7o Ni in 1007o sul-fides. Symbols as for Fig. 3. Regression line(r = 0.53) is for carted ore samples only. Thelighter line is the regression line derived fromPd:Pt ratioa (see text); large shaded field asfor Fig. 3.

THE CANADIAN MINERALOGIST

sulfides is essential in any comparison betweenore sources because absolute values in indivi-dual samples and ore production samples canbe influenced more by the intensity of sulfidemineralizatiot (e.9., disseminated, matrix ormassive) and host-rosk dilution during miningthan by the composition of the sulfide fraction.Five features of Table 1 should be noted: (1)the ore mined from Juan, Durkin, Ken, Nepeanand McMahon Shoots is essentially from a singleore-zone in each case: hanging wall at Mc-Mahon. contact at the other four. Productionfrom Lunnon and Fisher shoots, however, wasa composite of both contact and hanging-wallore zones in different production periods, andtherefore the mean data in Table 1 for theseshoots should be viewed as a new composite-orezonz; (2) as the ore mined from Ken andMcMahon Shoots is known to have includedsignificant amounts of nickel-poor, pyrite- andpyrrhotite-bearing metasedimentary rock, therecalculation has been based on a value fornickel in 100% sulfides more consistent withextensive mine sampling and specific-gravitymeasurements on ore. The close similarity ofPd and Pt values in Lunnon and McMahonShoots supports this adjustment; (3) calculationof the weighted mean values in Table la assumesa constant specific gravity for ore sampled ineach of the 18 composite samples. Althoughthis assumption is incorrect, the relatively smallstandard deviations for Ni and S indicate afairly constant sulfide content; (4) the meanvalues have been derived from large tonnagesof ore and should be representative; (5) super-gene ore from the violarite-pyrite zone (Nickelet ol. 1974) formed a significant part of thesampled production from McMahon, Ken andDurkin Shoots and this supergene ore may havebeen enriched in Pd and Pt in addition to Ni.

Values for Pd and Ni from Table lb, plottedin Figure 3, show a positive correlation with r- O.88. Figure 3 also shows the envelopecovering practically all of the 18 individualcomposite sample results for Lunnon, Juan,Durkin and Nepean Shoots; considerable scatteris evident even though individual samples re-present up to 9O,@0 tonne$ of ore. The meanvalues for Pt in Table lb have been plottedagainst Ni in Figure 4. They show a weakercorrelation (r = 0.53) and the scatter of indivi-dual samples is far greater. No correlation isevident between Au and Ni, but the large dis-crepancy between the Au value in the Kam-balda head sample, derived from NAA results,and the other mean values in Table 1b hascast doubt on all Au values in Table 1.

NICKEL DEPOSITS IN WESTERN AUSTRALIA 423

Other ore samples

One of us (R.K.) obtained representativedrill core of contact ore zones at Redross,Nepean, and from the dunitic-type deposit atForrestania. Where possible, results for indivi-dual core samples have been weighted by sam-ple length and sulfide content (a measure ofspecific gravity) to give the weighted meanvalues recorded in Table 2. These mean valueshave then been normalized to 1007o sulfideson the basis of S:Ni ratios (Table 2b). Table2 also includes data for the unclassified koma-

I.IEAN VAII'ES

tiitic deposits of Shangani and Hitura, for thetholeiitic deposits at Sudbury and Noril'sk, andfor samples of concentrate from the volcanic-type deposits at Windarra (Roberts 1975) andScotia (Christie 1975). The normalized datafor Pd and Pt in Table 2 have been plotted onFigures 3 and 4, whereas Ir values are shownin Figure 5.

Samples ol contact ore lrom Lunnon Shoot

Primary contact ore at Lunnon Shoot hasbeen sampled from 13 different locations andsummary results are recorded in Table 3a; full

EBBIE 2.

(a) UEe

DEPGI! TYPB PI!F" NI Au StNL NL3Cu NLtco

RsDRoss v 1 6.49t'NEPEAN V 2 11.55TONREST'NIA D 3 3.3IPmsEVERAN@ D 4 2.T9SNEN(T\NI R 5 9.7aIruRA r 6 1.96SI'DBI'E 'T 7 1.5NORrlrS( ! I 1.3(plcrLtLc gabbro,

NORILTSK T A 1.13(tartttc gabbro)

0.54 0.14 15.911 . 4 4 o . 1 3 1 8 . 60 . 1 6 - 5 . 0 9

0 . 2 5 - 1 1 . 6 1 8 01.3 3461 . 6 2 6 0 0

L . 9 2 3 0 0

2 . 4 5 1 1 . 1 8 4 5 . 61 . 6 1 8 . 0 A s . 6L.54 20.23 . 5 4 2 L . 9

- 8 . 1- 7 . 4 4

- 0 . 4 1

9791896

470293

2000IA2

6900

7200

225 8072 2530 34- ro0

t l 123- 230

DEK)gIT

REDROSSNSPEAN

TONRESTINIA

PERSEVERANCE

SMNGANI

BLrruReS1'DRJRYTiORIL I SK(plcri.tj.c gabbro)

NORII,'SK(td1tlc aabbro)

0 . 4 4

38.93 - 1951

37.57 - 347938.90 - L469

1300 200042.6L 661 668

- 1038 1092- 18000 47770

- 13446 42092

L4.7022.4O24.5011.00

7 . 2 0

299 213443 158534 r8a148 L72- 100

33 369

(c) urAS SIJIJ'IDES

DEFOSIT"o. ffi Ni cu Pd Au slNL NL:cu NLrco

IIINDARRA

SCOTIA54L44 14.84 r. .36 0.32 43.89

3206 23.07 r.45 0.36 35.651310 3280 90r 2.96 10.9 46.71461 3110 1020 1.54 15.9 63.1

r V volcal ic i D duit ic i K ko@ti i t ic associat ion; f l rholei i i ic $Bociat ion.,t Values fo! Ni, Cu, Co ad g in rtr; pb, pd, I! @d Au iu ppb.(L) Keays & Davigoo (f9?6), Oaleqno (1975)r reieht'ed ae@ of 5 rdtloh core sa8plesitecalcu-Latio! to 100, sulfides based on the Bee S :lli ratio. (2) Keays (in prep.): aeaof ? s@pl-es of driLl cole represeltiag a driLLed. Letrgih of 3.57a of coatacl, orei leightedby cote leogth dd g cootelt. Descripiioo of deposilr by Sheppy & Rore (L975) aod Barettet 41. (L976). (3) feaye & Davigon (19?6): veighted neo of I 6@pLes of dr iLL coleirecal-culation to 1"00f sutfideB based on tbe ne@ S:lti ratio. (lr) Keays & Dayj.son (1976):nean of 2 heal specieers; recaLcuLatiou to 100% sul.fid.es based on the ne@ S:Ni retio.Description of delFsit by !4a!ti! & Allchuch (19?5). (5) Natdrett & Cabri. (19'f6): datafo! @ ul<nom n@bei of @ssive suLfideBi assueal to be 100tr suLfides, but probabLy coo-taia <9rF on the beiB of extrE ieuce at Kebalda. Analytical nethoal: u!k.o@. (6) Iltkl-iet a]. (L976)r ne@ (ureighted?) of 36 saples rith ihe higheet S content; lecalcul"a-tio! to L00f sulfides ba6ed on the S:Ni latio. AlalyLj.caL nelrbodt fire assay ed emis-aioa spectrcgrepby. (7) Naldlett & Cabri (L976)t preferled tlata fof, overaLl aeetage,ilcLudibg the contrDsilrion of @ssive 6u.Lfi.des. Aralytical nethod: uklom. (8) Sdj.mov(f9561 : neo (weighted.? ) of 36 ud 38 smples, resleclive1y, incl-udilg coblpaiiion f,or@ssive su1fid.e6 (ass@ed to be LOOl sultide6). AnalyticaL neihod: un]rnm, (9) Weigbteddean for saapleB of concentrate rith recsLcu-lation to L006 su.Ifideo bded on earysis ofthe Eilicaie frection. The oodal-oGly bigh S content of ihe Windara sa&pLee reflectsthe incorporation of suLfide-beaing sedinelts iuto ore duiag niliDg. Anafyseg by $.M.C.

424 THE CANADIAN MINERALOGIST

details are given in Keays et al. (in prep.).The results for these three sample categorieshave been normalized to lUOVo sulfides andrecorded in Table 3b. Values for Pd are variableand range from 176-1232 ppb, yet all butthree ore sections show the matrix layer to havehigher values than the massive layer. Iridiumvalues are more uniform; most fall between100-300 ppb. The matrix layer contains less Irthan the massive layer in all but two sections.In the absence of better data, the mean of thecombined values for-the integrated samples andthe core from KA 4-12, given in Table 3b(iv),has been taken. to represent contact ore fromLunnon Shoot; it is plotted in Figures 3 and 5together with an envelope representing thespread of individual samples. The Au valuesin Table 3b vary over the range 11-730 ppband broadly correlate with Cu; with one ex-ception, sulfides in the matrix layer containmore Au than in the massive layer, usuallyby a factor of more than two.

The distribution of Pd, Ir and Au withinthe matrix and massive ore layers is fully dis-cussed by Keays et al. (in prep.), and only datafor the best sampled section (drill hole KA4-12) are plotted. Figure 6 shows the relativeenrichment of the matrix layer in Pd, Au andCu, and of the massive layer in Ir; it alsoshows that the distribution of Pd and Ir withinore layers is not uniform.

Samples of sulfide stringers from Lunnon Shoot

Stringers of primary sulfide within the foot-wall basalt have been sampled in drill holeKA 4-12 (three separate stringers), and beneaththe massive layer in the 505 stope (one stringer).

Sulfide stringers in a quartz vein overlying themassive layer, sampled in the 605 stope, werealso analyzed. When results are normalized tol1OVo sulfides (Keays et al., in prep.), theyshow that four stringers have anomalously highvalues for Pd (1400-17@ ppb) and Au (857-2520 ppb), and four are relatively enrichedin Cu. All five are enriched in Pd, Au andCu relative to the adjacent massive ore. Thelevel of Ir seems normal and is usually com-parable to normalized values in adjacent mas-sive ore. Keays & Crocket (1970) also foundthat footwall stringers below the main orebodiesat Sudbury were enriched in the precious metalsalthough their Ir contents commonly were low.

Sullide mineral separates

Separates of individual sulfide phases (pent-landite, pyrrhotite, pyrite and chalcopyrite)were obtained from samples of primary massiveore from Lunnon, Juan, Long and DurkinShoots (Fig. 2) and analyzed for Pd, Ir andAu. The results demonstrate that pentlanditeis the principal host for Pd, that Ir is almostuniformly distributed between all sulfide phases,and that pyrite and chalcopyrite are the princi-pal hosts for Au (Keays et al., in prep.).

Pnnctous Msrals IN THE Ur,rnervrerrcHosr Rocrs

The ultramafic host rocks have experiencedvarying degrees of metamorphism, serpentiniza-tion and talc-carbonate alteration, but detailedchemical studies at several deposits, e.9., Kam-balda (Ross 1974, Ross & Hopkins 1975),

4laEL

E5ctE

wT% r{tcKEr ril r00% surFtDEs

Frc. :. Plot of Ir versas wt. 7o Ni in IOOVo sulfides. Symbols as forFig. 3. Regression line (r = 0.99) is for L.C., K.H., R and N only.The shaded field represents individual samples of Lunnon contact or€.

NICKEL DEPOSTTS IN WESTERN AUSTRALIA

U1IMMAFIC(with disssminaredsulphidesI

MATRIX ORE

MASSIVE ORE

IT BASATTmnsl

PAtLAIllllMlppb) tFlDlUMlppb)0 400c0u(ppb)

Fro. 6. Plot of Pd, Ir, Au and Ca in tO0Vo sulfides for the contact orezone in drill hole KA4-12 from Lunnon Shoot. Data from Keays &Davison (1976).

425

Nepean (Barrett et al. 1976), Windarra (Watch-man 1971, Santul 1975), Mount Edwards(Hough 1976), and Scotia (Nesbitt 1971, Simon1972) suggest that these processes have usuallyeffected little change apart from the additionof HsO and COz. When these chemical dataare combined with the common preservationof original spinifex textures (representing theliquid-rich fraction) and cumulus textures theyindisate that these rocks formed from high-Mgliquids ()2OVo MgO) containing varying pro-portions of forsteritic olivine ( >Foso). AtLunnon Shoot there is strong evidence that theultramafic sequence formed from varying mix-tures of a liquid containing aboat 20-24Vo Mgpand crystals of about Fooz olivine (Ross 1974).Subsequent studies indicate that these two com-ponents could account for the range of composi-tions and textures encountered around theKambalda Dome, and, in particular, the com-positions of the meta-olivine peridotites directlyoverlying contact ore zones. Available data forthe other volcanic-type deposits in WesternAustralia favor similar components in the origi-nal magmas. This concept of a range of ultra-

mafic compositions being formed by simple two'component systems (liquid f olivine) is sup-ported by the occasional occurrence of relictigneous olivine (e.g., Kambalda, Scotia), andthe experimental studies of Arndt (1976).

If these ultramafic rocks resulted from suchmixtures, measursment of the abundances ofprecious metals in the original liquid and olivineshould determine values in rocks resulting frommixtures of these components. Furthermore, ifthese measurements are combined with datafor the sulfides they should indicate the distri-bution of precious metals betrveen the silicateand sulfide melts. According, we have analyzednumerous samples of spinifex-textured ultra-mafic rock, believed to represent original crystal-free liquid, and three samples of relict olivinefrom the basal unit overlying Victor Shoot(Fig. 2).

As the metapicrites in Table 4 do not containanomalous concentrations of sulfides, theirprecious-metal values should represent originalhighty magnesian liquids (plus or minus a fewpercent olivine) provided these metals have notbeen redistributed by postigneous processes.

G0PPEF(wt%l

426 THE CANADIAN MINERALOGIST

TABI,E 38. VAII',ES IN SAMPI,S OF COIITICT

(1) IMEGRATED SAMPITS

ttif;: LocArroN REF. oRE TYPE l.lrcRt{Es.s Nl cu

tEeErea,I i A u P d i l r

210640 4 Level(403-405)

zIo639zLO636 5 l€vel210635210626 710 stopezLO62521062A 704/8 Stopezto627210630 801 Stopezto629210632 807 StopezL063LzLO634 813 stope2LO63n2t0638 827 stope210637

I t l a t r t x 1 . 8 3 5 . 1 3 * 0 . 9 1

!{a66l.ve 0.46 4.46 1..5Il{atrlx 1.83 4.85 0.38lraslve 0.91 a.74 O.O2t ta t r tx 0 .46 6 .17 0 .65vat r l x 2 .13 5 .90 0 .05ldatl ix 1.00 4.69 O.29llagsl.ve 1.00 8.68 0.17ltatrlx 0.91 5.49 0.11ltassive 0.30 8.23 0.05ltatrlx 1.83 5.47 0.33lrassl.ve 0.30 8.50 0.08llatrlx 1.22 5.31 0.19l tassLve 0 .30 7 .48 l . t5uat r l x 1 .22 5 .16 O.42llassive 0.55 8.0 0.90

15.4 449 55 .3

3tt.8 LO75 221L6.4 106 16 .333,7 279 16322.O 142 t l l20 ,2 9 r .8 67 .9t7.6 345 14834,4 464 2632r.9 r49 12432.3 246 23216.7 437 92,329,O 611 19318.3 416 12033.0 594 20320.5 219 lo734.6 189 213

296 8. 12

4a2 4.422 8 . 7 6 . 5 0

9 , 4 L , ' 1 r92 .4 1 .28

5 0 . 5 2 . 3 393.6 r .7623.2 t ,2036.3 1 .0659.7 4 .7353.7 3 .L6

285 3,47240 2 ,94

54.3 2 .O54 3 . 0 0 . 5 9

(11) CORE SAIIPLES

KA4-12 4 Loel 22

(A4-t 33

t€ t r tx 4 .24 5 .O4 0 .64ltaasLve L.49 8.15 0.50lratrlx 3.57 6. 13 0.30l lasa lve 1 .58 9 .01 0 .16

20.53 3L2 62 .3 91 ,5 5 .6036.51 217 158 56 .2 1 .3 ' l2t.87 339 68 52 4.9836.31 259 181 14 1 .43

(ttt) LUIIP SA!{PLES

4-8 lnc. 703 Stop€, 1

8-U t !c. 801 Stope

1-7 lnc. 605 stope

ldatrLa 6.08 0.42l.laselve 7.92!rat!18 2.O2 5. 15 0.92M a s d i v e 1 . 0 5 1 . 7 A O . 4 5l l a s a t e e 1 . 0 8 . 4 6 0 . 5 0

4t

22.94 68232.50 571L9,42 59432.45 4223 3 . 1 9 3 r 0

99 233 6.89176 '14 .6 3 .2446.3 51,2 L2,83

t 5 0 9 . 2 2 , 8 Ll4 t ! 59 .7 2 ,15

r Value6 fo. Ni, Cu dd S in tl.gt Pd, Ir and Au i! prb. AL1 @alyses by R. Keays

except for s i , cu ed s ia ( i ) (bv g.M.c.) ua ( i i ) ( t rou gters & gudsou 19?2) '(f) gacl e@le is a corposite of cbip Ea.Epl-es ecross the ole layer' (2) t{eighted

averws (by- su.Ifide conient) for 1! suplee of @ttix ore ad 16 6arq)lea of @gsive

ore. -( : ) 'w! iehted

averages (by s content) for.3 saEples of.@t' ix ole @d ? 6ep1e€ of

*uuire-o*; idt-r: *u ;ril1ed aqircent to KA!-l-2. (!) weighted avera.3es (by S

conteDt) for 5 sepLes acrcgs tbe ;attix Eulfide layet dd 2 EaaltLes frod the @asjve

sulfide layer. (5) reietrtea Eve@ges (by s conteBt) for lr smples acrcs6 the @ttix

Iaver ual lr ea@Iee ecrcss the @B6ive Laye!. A dacitic porp\yry foru the f@italf,

to the @saive iaver. (6) weigltted average (by s contebt) of I s@ples acrc6s the

mssive J-ayer; a qwtz vein forc the h@ging laLL.

(Although ore-associated units were not sam-pled, the sample locations range from betweenore-associated units to well above ore-associatedunits; results show no sy$tematic variation inprecious-metal values with location). Some re-distribution of Au has almost cartainly occurred,but Pd and Ir are believed to have been essential-ly stable. Results for the seven sets of metapicritesamples show little variation in MgO, Pd and Irvalues, with the exception of the anomalouslyhigh Ir result from Mount Clifford. Palladiumvalues range from 7.40-11.60 ppb; there is noapparent difference between the first four re-sults from ore-beariFg environments and theremaining three. Iridium values are lower in,thetwo barren environments and have resulted inhigher Pd:Ir ratios, but the lower Ir value inthe Ora Banda sample may relate to its lowerMgO content (Keays et al., in prep.). The singlePt value, from Munro Township, representsonly three samples from the Az zotre, as MaoRae& Crocket (1977) suspect seawater leaching ofPt from the other three flow top samples.

Mean values for three mineral separatescontaining almost pure olivine and negligiblesulfides (Table 4) indicate that original olivineswere enriched in Ir and almost devoid of Pdand Au relative to material representing theoriginal, coexisting high-Mg liquids.

DrscussroN

Precious metals in sulfide ores

Although it is widely accepted that the sulfidesin volcanic-type deposits are of magmatic origin,there is uncertainty about the extent to whichoriginal compositions have been modified dur-ing postigneous processes. Our view is that thecoherence of the Ni:Cu and Ni:Co ratios shownin Figure 7 suggest that the sulfides have exper-ienced very limited modification across the widerange of metamorphic settings and associatedultramafic-rock alteration assemblages (relictolivine, serpentine and talc--carbonate) repre-

NICKEL DEPOSITS IN WESTERN AUSTRALIA

TA3LS 3b. METAI VAI,UB nECALCULATSD rO lOOt SUlqrD.ES 11S SA',PLE 0r'CONTACN ONE FBOU IUI{NOI! SHOOT

(i) INTEGMTED SAMPIJS

tTJ Loc,ArroN REF, oRE rypE TJ"|Hi Nt cu s Pd rr au

427

2|0640 4 l€vel IzLO639 403-405 2

zLO636 5 Levelz|o635

zLO626 710 Stopezto625

2!0628 704/8 stopezLo627

210630 801 StopezLo629

210632 807 StoPezLo63L

210634 813 StoPezto633

210638 827 stopezLO63'1

! {at ! l : 1.83Masalve 0,46CoEblled 2.29!'lat!la 1.83!,taselve 0.91CoEbtred 2.74uatr la 0.46!t6t ! lx 2. I3codbl.ned 2.59l{atrlx 1.00ltadal.ve f.00

Conbhed 2.00

ldatr ls 0.9L

lraaelve 0.30

coDbtned L.zL

Matrt ! 1.83

!,lasslve 0. 30

Co6b1aed 2.L3

ItAtr tx L.22

l,raasLve 0.30

co'blned L'52

l.latltr L'22

l6.alve 0.55

Co'bh€d L.77

L2.66* 2.24 38.00 1108s.u 1.73 39.87 12329.92 2 .O5 38.67 l l53

r r .48 0 .90 38 .83 25r10.07 0 .02 38 ,83 321r0 .76 0 .41 38 .83 24710.89 l .15 38 .83 25r1r .34 0 ,10 38 .83 L761r .25 0 .30 38 .83 190r0 .35 0 .65 38 .83 76r9 .80 0 ,19 38 .83 5249.99 0.34 38.83 6049.73 0 .19 38 ,83 2649.89 0 .06 38 .83 2969.78 0 .15 38 .83 214

12.45 0 .75 38 .00 99411.12 0 .10 38 .00 79912.15 0 .60 38 .00 9501r .26 0 .40 38 .83 8838.80 1 .35 38 ,83 lM

r0 .49 0 .69 38 .83 A279.77 0 .79 38 .83 4158.98 1 .01 38 .83 2129.43 0 .88 38 .83 327

r36255L7938.6

188114196130r43326297307220

2LO

2r9

239250203239218

73055265567 .910.83 9 . 0

1641 8 . 346.O

l 1 lto5r084 1 . r43.64 1 . 9

13670.2

T2L6052A250510348.27 9 , 3

(11) CORE SAIIPLES

KA4-12 4 LeveL

KA4-134

Matl la 4.24ldaaslve L,49Codblled 5.7 3ltatlta 3.57Uasslve 1.58

9.45 1 . 19 38 .41 615 t l2 1719.00 0 .58 40 .59 257 115 70 .99.28 0.96 39.18 1.75 I35 132

10.88 0.53 38.83 602 l2L 92.39 .63 0 .17 38 .83 277 193 15 .0

(rli.) LUIiP sAl.lPLES

4-8 hc. 703 Slope 5

8-l l t t rc. 801 stope 54 - l l U C .l-7 tuc. 605 stope

ltat!t!Ma6siveI'latrtxltasalve!{a6sLve

r0 .29 0 .71 38 .839,46 - 38 .83

1 0 . 2 9 . 1 . 8 4 3 8 . 8 39.31 0 .54 38 .839.90 0 .58 38 .83

n54 157 3946A2 210 89.1

lt88 92,6 LO2505 179 l l .0363 169 69.8

(1v) !{EAN oF co!{BINm DATA FRoll (1) AND (1t)

3,4 codbhed LO'22 0'76 38'75 612 2Og 2LL

r Val-ueB for Ni, Cu ud. g ia rX.%t Pd, Au ed Ir ia ppb. -ALL tecalcul,atioas 8e bsBed

oa the reigbt;d Dee g:Ni ror lunaoo coatact ore oi3.982 (RosB 2. EoPkins^19?9),

mlegs iadicated othetrise. Thig ratio le equival.eot to a s conteot of 3B.83%.

(1) necalcuaation baBeal otr 38.06 6 becaue of the telatively high Ni md cu values'

ipi neceLcutatiou baaed. oD suLfide coateat estj@ted f!@ Fe aalysi8 becaEe of the

relatively 1or Ni valuei the resuLtat S contelt is probably dore reslistic thd tbat

derived flon the reiaht;d eee gtsi for Luaon contact ore' (3) c@biued vaL€s have

been calculated ly reightiag the @trix dd @Esive ote valueB for both tbickleas @dg conteqt. (ll) dr ror-(:),-except the values bave bee! veighted for auLfid'e conteot'

wiqe the a.i"'oi s"e"u -&

iiudEo; (rg?2). (5) Sa.lxpLing @y oot be tldy lepreastative

of the ore sectio6 ud c@bioed values have not beeB ealcuLated'

sented by the deposits plotted in Figure 7. Values for core samples from Redro:: *d

These results encourage ,i, to dir"rrs t[e rela- Nepean and fol concentrates from Scotia fall

tionships between thJ abundance of precious within the envelope and support the wider ap-

metals and composition of sulfide bres in plication of this relationship. The mean value

terms of original igoeo* associations. ?or concentrates from Windarra is anomalously

Figure 3 Jhows good correlation between Pd high, but it represents early mine production

and Ni n IOOVo sulfides in the eight mean with a large component of supergene sulfides,

values for carted ore given in Table 1. As each and may reflect enrichment of Pd by weather'

value represents numerous samples and large ing. Also included_ in Figure 3 are mean values

tonnages'of oreo the correlation is viewed as-a foi core samples from Forrestania and for two

significant guide to the relationship over the hand specimens of matrix ore from Persever-

ringe of at least LOlOlo Ni in 100% sdfides. ance; ioth points for these dunitic-type deposits

The- important qualification is tlat tlese sul- lie very slosg t9 the regression line'-

fides be associated with komatiitic ultramafic The relatively low position 9! th" envelope

rocks for which the composition of the original for Lunnon contact samples (Fig. 3) suggests

liquid fractioD was in thi tuog" 2V25Vo NIgO, that lump and chip sampling of ore zones \ilill

428 THB CANADIAN MINERALOGIST

EEqeclo

-g-g-

F

Ecl

:g-g

-

4 8 1 2 1 0 2 0 2 4ITT% iltGlGL tt{ 100% sutFloEs

Ftc. 7. Plot of Ni:Cu and Ni:Co yersas vrt. VoNi in l0o7o sulfides. Symbols as for Fig. 3. Theregression lines are for carted ore only withr - 0.92 for Ni:Co, and r = 0,77 for Ni:Cu.

give low values for Pd if they do not includethe Pd-enriched footwall stringers. Althoughmany of these stringers are included in oreproduction, most are included during mining ofthe thin, high-grade ore zones such as DurkinShoot. The higher Pd value from Durkin Shootsuggest that the trend in Figure 3 may representthe minimum level of Pd in original sulfidemelts.

The correlation between the Pt and Ni valuesin the representative samples of ore productionis not good (Fig. 4, r = O.53). Individual sam-ples show large variation in Pt relative to Ni,and although other Pt values recorded in Table2 fall within the envelope shown in Fig. 4(with the exception of Scotia), they effect littlei.mprovement on tlte correlation. Although theregression line of Figure 4 has been used inlater calculations to estimate silicate-*ulfidepartitioning of Pt, we should not lose sight of

uYTz iltcK$ rit r00zsuLPHtDEs

Ftc. 8. Plot of Pd:Pt yersas wt. 7o Ni in lMVosulfides (r - 0.66). Symbols as for Fig. 3.

the large measure of uncertainty in the relation-ship between Pt and Ni. Because of this un-certainty, and because Figures 3 and 4 showcorrespondence between the levels of Pd andPt values, we have attempted to assess the re-lationship between Pt and Ni on the basis ofPd:'Pt ratios. These ratios, plotted against Niin Figure 8, yield a correlation coefficient of0.66. The regression equation has been com-bined with that for Pd versus Ni (Fig. 3) toderive an alternative expression for Pt yersas Ni.This derived line of best fit is also shown inFigure 4. The correlation between Pd:Pt andNi suggests a similar distribution of the twoprecious metals in ores and sulfide stringers.

Results for Ir in the volcanic-type deposits(Fig. 5) show the best correlation with Ni(r = 0.99), but it is only based on mean valuesfor four samples: Scotia, Redross, Lunnon con-tact, and the representative Kambalda Headsample. Use of the average value for Lunnoncontact is justified because: (1) Ir is not en-riched in footwall stringers, (2) it has a fairlyuniform distribution in ore profiles when nor-malized to IOOVI sulfides, and (3) the Lunnoncontact sampling is adequately representative.Figure 5 also shows the mean values for coresamples from Forrestania and two samples ofmatrix ore from the Perseverance {eposit (Table2d), and both points for these dunitic-typedeposits lie near the regression line.

The data for Au in these ores do not allow ameaningful discussion of its relationship withNi. Apart from the discrepancy in Table 1 be-tween the result of R.R. Keays (for Kambalda

NICKEL DEPOSITS IN

Head) and those of WMC, referred to previ-ously, the WMC values seem anomalously highin comparison with data given by Naldrett &Cabri (1976). Redistribution of Au is suggestedby its substantial enrichment in sulfide stringersin Lunnon Shoot and also by the large variationsbetween the ore layers at a given sampling site,and within an ore layer at different samplingsites, recorded by the data of Table 3 andFigure 6. Given the chemical mobility of Auand the alteration processes experienced by thesevolcanic-type deposits, e.9., hydrothermal alter-ation and leaching by seawater circulatingthrough a hot, submarine volcanic pile, meta-morphism, serpentinization, and talc-carbonatealteration, il is probably unreasonable to expectthat original igneous relationships can be fullyassessed in the field. Anomalously high concen-trations of Au in interflow sediments withinthe ultramafic sequence at Kambalda have beenrecently measured by Bavinton & Keays (1978);they suggested sea-floor leaching from the vol-canic pile (but not necessarily from the sulfideores) as a possible source.

The results presented in this paper illustratethe difficulties in sampling volcanic-type de-posits to establish the average tenor of preciousmetals, particularly in instances where massiveore layers are present. For example, the Pdvalues in Figure 3 indicate that samples oflarge tonnages of ore production are suitable,whereas results from Lunnon Shoot show thatclose sampling of the matrix and massive layersis very likely to yield lower values for Pd. Theselowervalues probably result from exclusion ofthe Pd-enriched sulfide stringers in the footwall.Figure 6 shows that the Pd content is differentin each ore layer and that it also varies willinore layers at Lunnon Shoot. As we know thatpentlandite is the major host for Pd (Keayset al., in prep.) and that the pentlandite con-tent of the sulfides in the profile shown inFigure 6 is approximately constant, it is evidentthat the Pd content of the pentlandite increasessubstantially up the profile. If our results fromLunnon Shoot are representative, the averagePd content can only be properly assessedthrough sampling of ore production. There arefewer data available to guide comment onsampling for Pt, but the large variation shownby the envelope in Figure 4 suggests that onlysamples representing large tonnages of ore prod-uction should be considered. As the Pd:Ptratio correlates more closely with Ni than doesPt, the distribution of Pt within ore zones andsulfide stringers may resemble that of Pd. Sam-pling for Ir appears to be much simpler for it

WESTERN AUSTRALIA

has a more uniform distribution within thecontact-ore zone at Lunnon Shoot and betweensulfide phases, and it is not enriched in thesulfide stringers. Howeyer, the profile shown inFigure 6 points to differences between andwithin ore layers anil sirggests that compositesamples acrosi'the rinlire ore zone are required.This study' has'not established any systematicpatterns to the distribution of Au in the sulfides;other attempts to establish original relationshipsbetween Au and the composition of the sulfidefraction will probably be seriously inhibited bysampling difficulties. Our results suggest thatmean valnes for Au will probably not exceed500 ppb.

Precious tnetals in the ultramafic host rocks

The results for metapicrites given in Table 4

(a) g!A!!g]fEg

429

LINOn SS001 !0(!er or€)

LUNNON SEM 4(n€r ore)

LOrc Sf,m 6(!@r o!e)

sllmEu I(o€r orc)

loUNl clrmm t0(uqr ote?)

UTJNRO IOTNSEIP 6(dlstet fi@ ore)

ruNA T(d16tet f,r@ ore)

G) Olnlm speG

vtcTof, stm 3(be*1 @it 1! drll1hole xD 6042)

(c) Olf,l8 RoCRSgEA FI,oOR BASALT II-T3

m AnERI ?Lmon 5

9 . ! t t . 1 8 3 . 1 0

r 0 . 7 7 t . 3 6 4 . 0 1

ngo Pt Pd l r & Pd l l r

I

3

4

7

8

9

l 0

r leo val@a in d.ti precioE neta18 iu ppt' [L rasdts haYe becnrecdcdated io volatil+tuee valu$. All saly8es by R.R' Keays

except for those in refelences 6' 9r and 1.0.(1) ke vdu* lor 10 s&ples of the Lurc! aetaficrite (nosB &

ropkins.l9?r) tuom &il1 hof,. KD 12?. (2) ueu vdues fot I smplas

collected st beiglts regina tttu9o-255 E abve the ba6al contecti sec

Kerys 9! d. (in prep.) fo. detailB. (3) ileM duea ror 6 seplo8colf,eted at heights mairy tuon 6-2?0 B &ove thc ka1 conlact; ae€

Kesys d d . ( in p rep . ) fo r de ta i l s . ( I ) seDle f roE eder roud dr i l1

ho le ! /33 tuon imed ia te o re env i ronren ! & ! [he bca l ion 3 dcps i t(tudres 19?5). (5) hs of l0 senles frd lhe upp€r Fdion of a

th in l loe in & i11 ho le f i cD !?B ( (e rys , jn p rep . ) , f re .e la l ionsh ip o f

the unit to binemlization loter in the guccession is wcertain.(6) U€e of 6 s&ples fM the A1 md A2 zones of 2 flor uii6. fre

MgO due lep.esents the ree of col@s 3' ! and 5 in hble ? of hdi

9!.9!. (197?) sd ihe nes l,.O.I. he been uscd to recdcdat€ the

resuLis of hcn& ! crocket (19?7) to vofatil*free Elue8. fre Pt

dues only lepresent the 3 seples ft@ the A2 zone. (l) one sqtc(Hlttc No. Z 11232) of batren eatapicrita disldl lrcn knom ore occa-rencee. (8) l{ee of 3 ddysas ot seFratcs coniaininc >>959 olivine(Foo2-o3) sd neFJ i f l ib ta sd f idcs . (9 ) c rcck , ! & T . ro ta (1977) .( ro l -c iocker & chv i ( r97?) .

indicate that the liquid fraction of these koma-tiitic ultramafic rocks from the Eastern Gold-fields of Western Austlalia contained values forPd, Ir and Au similar to those in almost identical

23.tot

23.55

7 . 5 8

7.44

5 . 9 0

9 , 3 5

1 0 . 3

12.99

24.20 - 7.40 1.25 5.30

2 2 . t 2 - u . 6 0 L . 2 9 0 . 1 7

25.64 - 9.6t 2.87 0.92

2 5 . 6 1 5 . 1 1 0 . 4 1 . 0 1 . 9

! 9 , 3 8 - 1 0 . 7 8 0 . 8 3 ! . 6 1

0.21 5.68 0.25 0,04

< 0 . 7 < 0 . 0 2 2 . 8

7 . 9 2 . 4 l . l 3 . 3

430

rocks from Munro Township in Ontario. Themean values of about 9-10 ppb Pd, 1.0-1.3ppb Ir and 2-5 ppb Au are much higher thanpublished values for sea-floor basalts shown inTable 4, but they are compara'ble with 'results

for peridotite and dunite from the Mount Albertpluton in Quebec. Samples from barren environ-ments seem to contain less Ir, but as the lowervalue in the Ora Banda sample rnay result fromits lower MgO content, the significance of thelower average value for Munro samples mustbe questioned. There is no simple explanafiionfor the anomalous levels of Ir in the 10 samplesfrom Mount Clifford; the. higher values areconsistent and seem to be an original featureof the flow unit.

The high Ir and low Pd values in theolivine separate samples are consistent withresults for MgO-rich, sulfide-free samples fromForrestania reported by Keays & Davison (1976,Fig. 7). Results given in Table 4 indicate Nernstpartition coefficients (D) for the distributionof Ir, Pd and Au between blivine and coexistinghigh-Mg liquid (D = ppb metal in olivine/ppbmetal in silicate melt) of approximately 4.7 f.orIr, 0.02 for Pd, and 0.06 for Au, providing thatequilibrium was attained. trn view of the hightemperatures required for the liquid (>1500'C:Arndt 1976) equilibrium relationships are con-sidered most likely.

Silicate-sulfide partitioning of precious metals

If we assnme the conventional view that sul-fide concentrations have formed by segregationand accumulation of immiscible sulfide dropletsfrom the host silicate magma, then it follows,as indicated by Rajamani & Naldrett (1978),that their composition will be governed by thecomposition of the host magma and the parti-tioning functions for the distribution of metalsbetween the sulfide and silicate liquids. In thecase of Ni and Cu, Naldrett & Cabri (1976)and Rajamani & Naldrett (1978) have showeda trend of decreasing Cu/ (Cu*Ni) with in-creasing host-rock MgO for deposits of tho-leiitic and komatiitic association. However. al-though their observations are true in a generalsense they have simplified the actual relation-ships in volcanic-type deposits. Figure 7 showsthat at Kambalda a range of Ni, Cu and Covalues in sulfides coexisted with high-Mg liquidof essentially uniform composition containingvariable proportions of olivine. Similarly, theresults for Pd, Pt and Ir indicate a range ofpartition coefficients for these metals betweena silicate melt of essentially constant composi-tion and sulfide melts that span the range from

THE OANADIAN MINERALOGIST

<10 to )20Vo Ni. As the systematic variationin the content of precious and transition metalsin the sulfides almost certainly precludes ametamorphic origin, we must consider otherpossibilities. For example, the variation couldresult from postsegregation re-equilibration witholivine during cooling, with different coolingrates giving rise to different nickel contents inthe sulfide. However, the systematic increaseof Pd and Cu with Ni (Figs. 3, 7) arguesagainst olivine as a source of additional Ni,whereas an even more compelling argumentagainst an olivine source is the observation thatthe Pd:Ir ratios in 1007o sulfides with differentNi contents are essentially constant, and similarto the value measured in the Lunnon metapicrite.If we take the regression lines in Figures 3 and5. at l jVo Ni the Pd:Ir ratio is 6.37, and at20/o Ni it is 7.48; these ratios are very similarto the mean Pd:Ir value of 7.68 given in Table4 for l0 samples of Lunnon metapicrite andstrongly suggest that the high-Mg liquid 'was

solely responsible for the composition of thecoexisting sulfides.

A second source of this systematic variationcould be the chemical environment of the'coexisting silicate magma and sulfide melt priorto extrusion. Because Fe and Ni are the mostimportant variables in the sulfides it is logicalto look to factors that affect their partitioning'between the two melts. Rajamani & Naldrett(1978) concluded that their partitioning will belargely influenced by factors that influence theactivity coefficients of NiO and FeO in thesilicate magma. They emphasized variation incomponents such as AlOr, NarO, KeO and CaO,but we seem to be dealing with only minorvariation in the silicate components; we mustlook to other factors, such as l(Os) and l(Sz).If each contact ore shoot and overlying ultra-mafic rocks originated from a separate magmapulse with a different path of ascent in.themantle, then each may have been exposed to' '

different conditions. Small variations in thecontent of HrO could lead to significant varia-tions in l(O,), which in turn are likely to in-fluence the partitioning of Fez* between the '

silicate and sulfide melts. For example, highervalues of l(Oz) should favor the partitioningof Ni'+ into the sulfide melt at the expenseof Fe2+, and lead to lower S:Ni ratios. Similarly,variations in l(Sr) are likely to exert a differen-tial influence on the partitioning of Ni'+ andFe2+. We have also considered the possibilitythat differences in composition of the sulfidefraction resulted from varying departures fromequilibrium, but in view of the high tempera-

NICKEL DEPOSITS IN

tures that must have. prevailed, and the shortequilibration times observed in experimentalstudies (Rajamani & Naldrett 1978, Clark &Naldrett 1972), we consider this an unlikelyalternative. The relative proportion of sulfideto silicate melt may also be a contributing factor.For a given amount of silicate magma theremay be a fixed budget of Ni, Cu, Co, Pt, Pd,Ir and Au entering the sulfide melt. Increasingamounts of S simply add more Fe from thesilicate melt which in turn dilutes the concen-tration of the other metals. Another factorcould be the,timing of the separation of thesulfide melt, from the silicate magma; meltsthat separated early would be enriched in allthese metals relative to Fe whereas later melts,perhaps forming after the removal of smallamounts of such early melts, would be enrichedin Fe.

The small variation in the Pd:Ir ratios of oresulfides noted above probably reflects verysimilar partition coefficients. To show thissimilarity, the Nernst partition coefficient, D(where D = wt.Vo metal in sulfide liquid/wt.%metal in silicate liquid) has been calculated forthe range of l'00/o sulfide compositioris. Thevalues for Pd and Ir in the Lunnon metapicriteClable 4) and the value for Pt in the MunroTownship samples have been used to representthe silicate melt, whereas the sulfide melt hasbeen represented by values taken from the re-

WESTERN AUSTRALIA 431

gressiou lines in Figures 3, 4 and,5. These cal-culations assume equilibrium relationships forsulfide-silicate partitioning of these element$;tlte results are plotted in Figure 9, together withD values for Ni, Cu and Co calculated from thedata of Ross (1974). Space does not permit afull discussion of all data in Figure 9, but it isinteresting to note that the relative position ofD values for Cu and Ni are consistent withthose observed by Rajamani & Naldrett (1978)in experimental charges containing l3,5%o MgO.Although the values for Pd and Pt exceed theestimate of Naldrett & Cabri (1,976) for asingle sulfide melt composition, there is broadagreement and the relative values are the same.

The data of Figure 9 provide a convenientreference for discussion of the relationshipsbetween Cu/(Cu*Ni) and Pt/(Pt+Pd) innickel sulfide deposits of komatiitic and tholeiitisassociation. The results of Naldrett & Cabri(1976) are shown in Figure 10 together withresults from this study. From inspection of theNi:Cu ratios in Figure 7 and the relative posi-tion of the D values for Fd and Pt in Figure 9,it is evident that komatiitic deposits shoulddefine a.steep positive trend at very low Cu/(Cu+Ni) valueso as is the case in Figure 10.Moreover, it is to be expected that there willbe a family of similar trends, displaced tohigher Cu/ (Cu*Ni) values with decreasingMgO in the silicate liquid, and to lower Cu/

o,

q)

€.J

o

.Eq,

o,

!l'E

ah

a!OJ

-'Z:E,fi E'w__y:Hla'ffi 't

t l

trt

COBALT

4 8 1 2 1 6 2 0 2 4wTu NtcKEr til t00 z sutF|DES

FIc. 9. Plot of Nernst partitioning coefficients (D) versus wt. % Ni in1007a sulfides.

432 THB CANADIAN MINERALOGIST

0.2 0.4 0.E 0.8

Cu/Cu' l t l i

Flc. 10. Variation in PtlPt+Pd ratios as a functionof the Cu/CuaNi ratios of nickel sulfide oredeposits. Symbols as in Fig. 3. The stippled areashows the komatiitic field defined by Naldrett &Cabri (1976). The open circle represents theLunnon metapicrite, assuming the Pt value fromthe Munro Township samples (Table 4).

(Cu*Ni) values with increasing MgO. Thekomatiitic field defined by Naldrett & Cabrimay be composite, whereas the trend definedin this study represents a very limited rangeof MgO in the original high-Mg liquid. As thekomatiitic trends in Figure 1O represent a de-crease in Ptl(Pt+Pd) with increasing MgO(decreasing Cul (Cu*Ni) ), the tholeiitic trendof Naldrett & Cabri (1976) shown in Figure 10requires an increase in this ratio with increas-ing MgO. This requirement is better envisagedas a change in the relative position of the trendsfor Pd and Pt in Figure 9 from divergence withincreasing Ni (decreasing Cu/(Cu+Ni)) toconvergense.

Data presented in this paper indicate con-sistent variations in the partitioning of preciousmetals between host silicate liquids and the co-existing sulfide liquids which formed the vol-canic-type deposits, but these results should beapplied with caution. For example, althoughthey may have general application to volcanic-type deposits the results can only be appliedsensu stricto to nickel sulfide deposits associatedwith original ultramafic magmas having a high-Mg liquid fraction containing 2V25Vo MgO.Fortunately, this range includes most, and pos-sibly all, volcanic-type deposits in WesternAustralia.

The anomalous Ir values in samples of meta'picrite from Mount Clifford suggest that somesulfide melts could have coexisted with ano-malous silicate melts and thus would show de-partures from the trends presented here. Inaddition it should be noted that: (1) the'Pdvalues in the carted ore samples of Table 1may represent minimum values, (2) the cor-relation between Ni and Pt in sulfides is notgood, and (3) the correlation between Ni andIr is only based on a small sample population.The anomalously high values for Pd and Pt insamples from Durkin Shoot coincide with highIr values calculated from data for mineral se-parates (Keays et aJ,, in prep.) and indicatethat significant departures from the trends mayoccur even amongst ore shoots associated witha single ultramafic sequence.

An enigmatic feature of the results for meta-picrites (Table 4) is the lack of obvious differ-ence between values in samples from ore-bear-ing and barren areas. If the volume of sulfidemelt was large relative to the volume of silicatemelt we would expect some evidense of scaveng-ing of metals from the latter. However, if thesulfide melt was enriched in precious and transi-tion metals before the silicate and sulfide meltcame into contact it could have enriched thehigh-Mg liquid. Obviously the unknowns pre-clude firm deductions, but the similarity ofsilicate values favors the possibility that thevolume of sulfide melt was small in relation tothat of high-Mg liquid.

The results for olivine (Table 4) enable usto make some general comments on the distribu-tion of precious metals in deposits associatedwith original silicate liquids of higher and lowerMgO content. Liquids containing more thannsay, 25/o MgO can onlY be formed frommelting of residual olivine in the source diapir(Arndt 1977). If the precious-metal content ofthis olivine is similar to that of relict olivineat Kambalda, the resulting silicate liquid willcontain more Ir, less Pd and probably less Authan the Lunnon metapicrite. The lower Ptvalues obtained by MacRae & Crocket (1977)in more olivine-rich sections of units suggestthat Pt will also decrease as MgO increases.These differences should be reflected in lowerPd:Ir values in coexisting sulfide melts, butthey are likely to be small; not surprisingly,values for the Perseverance and Forrestania de-posits (believed to be associated with originalsilicate liquids containing more than 257o MgO)plot close to the regression lines in Figures 3 and5. Conversely, liquids containing less than about20/o MgO should contain less Ir, more Pd, Au,

\\

r-^g

NICKEL DEPOSITS IN

and probably more Pt than the Lunnon meta-picrite, and result in higher Pd:Ir ratios incoexisting sulfide melts.

Surraumv AND CoNcLrJsIoNs

We have studied the distribution of preciousmetals in the sulfides and associated host rocksof volcanic-type nickel sulfide deposits in West-ern Australia. Although the high-Mg liquidfraction of the komatiitic host rocks appearsusually to have been restricted to the range of2O-2570 MgO, we have observed compositionsfor the associated massive and matrix sulfidesthat span the range of. 10-23Vo Ni in 1007asulfides. The level of Pd, Pt and Ir in thesesulfides increases with increasing Ni, whereasAu values are more erratic. Sampling of nu-merous ore sections and footwall sulfide stringersfrom the contact-ore zone at Lunnon Shoot hasshown that Pd is commonly depleted from themassive ore layer and enriched in the footwallstringers together with Cu and Au; we concludethat the only reliable method to accuratelyestablish original Pd values is to sample largetonnages of mine production. A similar con-clusion probably applies to Pt. The distributionof Ir is much more uniform and close samplingof complete ore sections should provide arepresentative sample. Individual ore samplesare inadequate measures of the abundance ofPd, Ir and Au, because their concentration insulfides varies within ore layers and withinindividual sulfide minerals.

Analyses of numerous samples of metapi-crite, believed to represent the high-Mg liquidfraction of the associated ultramafic rocks" showsimilar Pd and Ir values in ore-bearing andbarren environments, with Pd:Ir ratios com-monly in the range of 6-10. Almost pure se-parates of relict olivine from Kambalda areenriched in Ir and impoverished in Pd and Aurelative to the material representing the originalcoexisting liquid. These results have allowed usto calculate Nernst partitioning coefficients forthe coexisting silicate and sulfide melts overthe range of. 10-20% Ni in sulfides. We findvery similar D values for Pd and Ir over thisrange (138-319 for Pd, 167-327 for Ir), andmuch lower values for Pt (73-120). ThePd:Ir ratio of the sulfides ranges from 6.37 atl07o Ni to 7.49 at 20% Ni, and closely approxi-mates that measured in the Lunnon metapicrite(7.68). This similarity indicates that precious-metal values in the sulfides reflect those inthe associated high-Mg liquid, and that their

WESTERN AUSTRALIA

systematic variation together with those ob-served fer Ni, Cu and Cq has resulted from theinfluence of factors other than bulk composi-tion on the partitioning of metals between co-existing silicate and sulfide liquids.

ACKNOWLEDGEMENTS

This study has been made possible by thefinancial support of the Australian ResearchGrants Commission, the Australian Institute ofNuclear Science and Engineering, and WesternMining Corporation Ltd. It has benefited con-siderably from the broader support of WMC;we are also grateful for their permission topublish. We have benefited from the contribu-tion of many colleagues and warmly acknow-ledge their interest and help. Several geologistsat Kambalda helped with sample collection, inparticular Dr. P. Woolrich and Mr. R. Watch-orn, whereas Mr. G.A. Travis provided bothsamples and stimulating comment during thecourse of the project. The analytical work atthe University of Melbourne was assisted byMs. R.M. Davison, and Messrs. P.J. McGold-rick and P. Hannaker. Dr. C.E. Gee carriedout the statistical treatment of data and helpedintroduce much of the valuable WMC data tothe projet. Completion of the study has beenassisted by a Visiting Fellowship (J.R.R.) atthe Research School of Earth Sciences, A.N.U.;that support, and the help of Mr. R. Rudowskiwho separated the olivine samples, is gladlyacknowledged. Mr. O.A. Bavinton offered manyhetpful comments, both during the project andon an earlier version of the manuscript. Thefinal version has been improved by the com-ments of Dr. G. Loftus-Hills, Mr. J.J. Greshamand Dr. P. Woolrich at Kambalda, and thereviews of Professor J.H. Crocket and Dr. D.I.Groves.

RrnBnBNcEs

ANonrws, P.B. (1975): Spargoville nickel de-posits. /z Economic Geology of Australia andPapua New Guinea. 1. Metals (C'L. Knight, ed.),Atstral, Inst. Mining Met. Mon. 5, 89-91.

AnNor, N.T. (1976): Melting relations of ultra-mafic lavas (komatiites) at 1 atm and high pres-xrte. Carnegie Inil. Wash. Yearbook 76, 555's62.

(1977): Ultrabasic magmas and high-de-gree melting of the mantle. Contr. Mineral.Petrology 64,205-221.

433

434 THE CANADIAN MINERALOGIST

-, NALDRETT, A.J. & PYKE, D.R. (1977):Komatiitic and iron-rich tholeiitic lavas ofMunro Township, Northeast Ontario. l. Petrol-ogy L8, 319-369,

BARRETT, F.M., BrNNs, R.A., Gnovrg D.I., M,las-roN, R.J. & McQunEN, K.G. (1977): Structuralhistory and metamorphic modification of Ar-chaean volcanic-type nickel deposits, YilgarnBlock, Western Australia. Econ. Geol. 72, ll95-1223.

---, Gnovrs, D.I. & BrNNs, R.A. (1976): Im-portance of metamorphic processes at the Ne-pean nickel deposit, Western Australia. /nst.Mining Met. Trans. 85,8.252-8273.

BAvrNToN, O.A. & KEAys, R.R. (1978): Preciousmetal values from interflow sedimentary rocksfrom the komatiite sequence at Kambalda, West-ern Australia, Geochim, Cosmocltim. Acta 42,1 151-1 163.

BrNNg R,A., Gnovrs, D.I. & GuNrnonrr, RJ.(1977): Nickel sulphides in Archaean ultramaficrocks of Western Australia. In Correlation ofthe Precambrian 2 (A.V. Sidorenko, ed.), Nauka,Moscow.

Bunr, D.R.L. & Snrruv, N.R. (1975): Mount Keithnickel sulphide deposit. .lz Economic Geologyof Australia and Papua New Guinea. l. Metals(C.L. Knight, ed.), Austral, Inst. Mining Met.Mon. 5, 159-168.

CHRrsrIE, D. (1975): Scotia nickel sulphide de-posit. /r Economic Geology of Australia andPapua New Guinea. l. Metals (C.L. Knight, ed.),Austral. Inst, Mining Met. Mon. 5, l2l-125.

Cl,ln& T. & Neronrrr, A.J. (1972): The distribution of Fe and Ni between synthetic olivineand sulphide at 900"C. Econ. Geo|.67,939-952.

Cnocrsr, J.H. & Crm, L.L. (1972): Abundancesof Pd, Ir, Os and Au in an alpine ultramaficpluton. Int. Geol. Congr. 24th L0, 202-209.

---, Kravs, R.R. & Hsrrn, S. (1968): Deter"mination of some precious metals by neutronactivation analysis. I. Radioanal. Chem. L, 487-507.

& TERUTA, Y. (1977): Palladium, iridium,and gold contents of mafic and ultramafic rocksdrilled from the Mid-Atlantic Ridge, leg 37,Deep Sea Drilling Project. Can. l. Earth Sci. L4,777-784.

DALcARNo, C.R. (1975): Nickel deposits of theWidgiemooltha Dome - Redross, Wannaway,Widgiemooltha, Dordie. /n Economic Geologyof Australia and Papua New Guinea. 1. Metals(C.L. Knight, ed.), Austral. Inst. Mining Met.Mon.5 ,82-86,

EwERs, W.E. & HuDsoN, D.R. (1972): An inter-pretive study of a nickel-iron sulfide ore inter-section, Lunnon Shoot, Kambalda, Western Aus-tralia. Econ. Geol. 67, 1075-1092.

Gruurs, I. & Tnnnory A. (1975): The Archaeanbetween Coolgardie and Norseman - stratigraphy

and mineralization. '/n Economic Geology ofAustralia and Papua New Guinea, 1. Metals(C.L. Knight, ed.), Austral Inst. Mining Met.Mon.5 , 66-74 .

Hirrr, T.A., HiNrvtNnN, E., VuonuernrN, Y. &PArUNEN, H. (1976): Platinum-group mineralsin the Hiturq nickel deposit, Finland. Ecorr.Geol. 7L, 1206-1213.

HAUKKA, M.T. & Tnorans, l.L, (1977): Total X-rayfluorescence analysis of geological samples usinga low-dilution lithium metaborate fusion method.Matrix corrections for major elements. X-raySpectrom. 6, 204-211.

Houcn, M.J. (1976): Archaean Ultantalic Meta-volcanics Close to Nickel-Sulphide Mincraliza-tion, Mt. Edwards, Western Australia. Ph.D.thesis, Australian National Univ., Canberra.

HuosoN, D.R. ( 1972) : Evaluation of genetic modelsfor Australian sulphide nickel deposits. Austral.Inst, Mining Met. Proc. Newcastle Conl., 59-68.

Kreys, R.R. & Cnocrrr, I.H. (1970): A study ofthe precious metals in the Sudbury nickel ir-ruptives ores. Ecott. Geol. 65,438-450.

& DevrsoN, R.M. (1976): Palladium.iridium, and gold in the ores and host rocks ofnickel sulfide deposits in Western Australia.Econ. Geol. 7L, 1214-1228.

_-, GANArATv, R., Leur,, J.C., KniinsNstiHr,,U. & MonceN, J.W. (1974): The simultaneousdetermination of 20 trace elements in terrestrial,lunar and meteoritic material by radiochemicalneutron activation analvsis. Anal. Chim. Acta72, l-29.

Krrr.e, R.A. & Nrcrer., E.H. (1974): The geologyof a primary millerite-bearing sulphide assem-blage and supergene alteration at the Otter Shoot,Kambalda. Western Australia. Econ. Geol. 69.r102-ttt7.

MecRer, W.E. & Cnocrsr, J.H. (1977): Noblemetal content of komatiitic ultramafic volcanicsfrom Munro Township, Ontario. Geol, Assoc.Can, Program Abstr. 2, 34.

MaRrrN, J.E. & Ar,r,cnuncu, P.D. (1975): Per-severance nickel deposit, Agnew. /n EconomicGeology of Australia and Papua New Guinea.l. Metals (C.L. Knight, ed.), Austral. Inst.Mining Met. Mon.5, t49-155.

NALDRETT, A.J. & CesRr, LJ. (1976): Ultramaficand related mafic rocks: their classificationand genesis with special reference to the con-centration of nickel sulfides and platinum-groupelements. Econ. Geol. 71, ll3l-1158.

Nesattr, R.W. (1971): Skeletal crystal forms inthe ultramafic rocks of the Yilgarn Block, West-ern Australia: evidence for an Archaean ultra-mafic liquid. Geol. Soc. Aust. Spec. Pttbl. 3,33 1-350.

NIcKEL, E.H., Ross, J.R. & TnonxsER, M.R.(1974): The supergene alteration of pyrrhotite-

NTCKEL DEPOSITS IN

pentlandite ore at Kambalda, Western Australia.Econ. Geol, 69, 93-107,

RAJ"AMANT, V. & Ner-onrrr, A.I. (1978): Partition-ing of Fe, Co, Ni, and Cu between sulfideliquid and basaltic melts and the composition ofNi{u sulfide deposits. Econ, Geol. 73, 82-93,

Roarnrs, J.B. (1975): Windarra nickel deposits. InEconomic Geology of Australia and Papua NewCuinea 1. Metals (C.L. Knighto ed.), AustralInst. Mining Met. Mon. 5, 129-143.

B"oss, J.R. (1974): Archaean Nickel SulphideMineralization, Lunnon Shoot, Kambalda. Ph.D.thesis, Univ. Calif., Berkeley.

& HornNs, G.M.F. (1975): Kambaldanickel sulphide deposits. le Economic Oeologyof Australia and Papua New Guinea. l. Metals(C.L. Knight, ed.), Austral. Inst. Mining Met.Mon. 5, 100-121.

SeNrur, J. (1975): The Geology, Geochemistry,and Mincralization of the South WindarraNickel Ore Deposit, Western Australia. B.Sc.(Hons.) thesis, Univ. Adelarde, Australia.

Scrulrz, K. (1975): Carr-Boyd Rocks nickel-copper deposits. In Economic Geology of Aus-tralia and Papua New Guinea. (C.L. Knight,ed.), Aust:al. Inst. Met. Mon. 5, 125-128.

Sne.w, D.M., DosrAL, J. & Knevq R.R. (1976):Additional estimates of continental surface Pre-cambrian shield composition in Canada. Geochim.Cosmochim. Actu 44, 73-83.

Snnrrv, N.R. & Rows, J. (1975): Nepean nickeldeposit. In Economic Geology of Australia andPapua New Guinea. l. Metals (C.L. Knight, ed.),Austral. Inst. Mining Met. Mon. 5, 91-98.

WESTERN AUSTRALIA 43s

StuoN, I.K. (1972): A Geochemical and Petrologi-cal Investigation of the Massive Dunite Mem-bers ol the Scotia Ultramafic Bel1. B.Sc. (Hons.)thesis, Univ. Adelaide, Australia.

SunNov, M.F. (1966): The Structure of Noril'skNickel-Bearing Intrusions and the Genetic Typesof their SuUide Ores. All-Union Sci. Res. Inst.Mineral Raw Materials (VIMS), Moscow (inRuss.).

Tnevrs, G.A. (1975): Mount Clifford nickel de-posit. /n Economic Geology of Australia andPapua New Guinea. l. Metals (C.L. Knight, ed.),Austral. Inst. Mining Met, Mon. 5, lM-146.

WetcHrraN, A.L. (1971): A Study of Two Geol-ogical Sections and the Peftology and Geochetrt-istry ol the Ultramafic Rocks - Mount WindarraOrebody, Western Australia. B.Sc. (Hons,) thesis,Univ. Adelaide, Australia.

Wnrreus, I.R. (1973): A proposed subdivisionof the Eastern Goldfields Province of the Ar-chaean Yilgarn Block, Western Australia. Rec.Geol. Surv, W. Aust. 1973/8.

Woooer.r, R. & TRAVrs, G.A. (1969): The Kam-balda nickel deposits, Western Australia. Proc.9tlt Commonw, Mining Met. Congr. Pap, 26.

Woonrctr, P. & GroncErrA, N. (1978): Themineralogy and geochemistry of the iron-nickelsulphides at Kambalda. Western Mining Corp,Tech. Rep. K/2379.

Received August 1978, revised manuscript acceptedDecember 1978.