ARTICLE

A.J. Naldrett

World-class Ni-Cu-PGE deposits: key factors in their genesis

Received: 9 January 1998 /Accepted: 21 September 1998

Abstract Magmatic Ni-Cu sul®de deposits form as theresult of segregation and concentration of droplets ofliquid sul®de from ma®c or ultrama®c magma, and thepartitioning of chalcophile elements into these from thesilicate melt. Sul®de saturation of a magma is not en-ough in itself to produce an ore deposit. The appropriatephysical environment is required so that the sul®de liq-uid mixes with enough magma to become adequatelyenriched in chalcophile metals, and then is concentratedin a restricted locality so that the resulting concentrationis of ore grade. The deposits of the Noril'sk region havedeveloped within ¯at, elongate bodies (15 ´ 2 ´ 0.2 km)that intrude argillites, evaporites and coal measures,adjacent to a major, trans-crustal fault and immediatelybelow the centre of a 3.5 km-thick volcanic basin.Studies of the overlying basalts have shown that lavasforming a 500 m-thick sequence within these have lost75% of their Cu and Ni and more than 90% of theirPGE. Overlying basalts show a gradual recovery in theirchalcophile element concentrations to reach ``normal''values 500 m above the top of the highly depleted zone.The ore-bearing Noril'sk-type intrusions correlate withthose basalts above the depleted zone that contain``normal'' levels of chalcophile elements. The high pro-portion of sul®de (2±10 wt.%) associated with theNoril'sk-type intrusions, the high PGE content of theores, the extensive metamorphic aureole (100±400 maround the bodies), and the heavy sulfur isotopic com-position of the ores (+8±+12 o34S) are explicable if theore-bearing bodies are exit conduits from high level in-trusions, along which magma has ¯owed en route toextrude at surface. The ®rst magma to enter these in-trusions reacted with much evaporitic sulfur, at a low``R'' value and thus gave rise to sul®des with low metal

tenors. Successive ¯ow of magma through the systemprogressively enriched the sul®des in the conduits, losingprogressively less of their chalcophile metals, and thusaccounting for the upward increase in metals in succes-sive lava ¯ows above the highly depleted ¯ows. TheVoisey's Bay deposit lies partly within a 30±100 m-thicksheet of troctolite, interpreted as a feeder for the1.334 Ga Voisey's Bay intrusion, and partly at the baseof this intrusion, where the feeder adjoins it. Studies ofolivine compositions indicate that an early pulse ofmagma through the feeder and into the intrusion was Nidepleted but that subsequent pulses were much less de-pleted. Trace element, Re-Os and S and O isotope data,and mineralogical studies indicate that the magmapulses interacted with country gneiss, probably princi-pally in a deeper level intrusion, extracting SiO2, Na2O,K2O and possibly sulfur form the gneiss, which accountsfor the magma becoming sul®de saturated. The Jinchuandeposit of north central China occurs within a 6 km-long dyke-like body of peridotite. The compositions ofolivine within the dyke, the igneous rocks themselves,and the ore are all inconsistent with derivation of thebody from ultrama®c magma, as originally supposed,and indicate that the structure forms the keel of a muchlarger intrusion of magnesian basalt magma. Flow ofmagma into the intrusion has resulted in olivine andsul®de being retained where the keel was widening outinto the intrusion. The West Australian komatiite-re-lated deposits occur in thermal erosional troughs whichhave developed due to the channelisation of magma ¯owand the resulting thermal erosion of underlying sedi-ments and basalt by the hot komatiite magma. Thesediments are sul®de-rich, and may have contributedsubstantially to the sul®de of the ores. The mineralisa-tion in the Duluth complex occurs in troctolitic intru-sions along the western margin of the complex as a resultof magma interacting with and extracting sulfur fromthe underlying graphite- and sul®de-bearing sediments.No magma ¯ow channels have been identi®ed so far,and the lack of magma ¯ow subsequent to the devel-opment of sul®de immiscibility is regarded as the reason

Mineralium Deposita (1999) 34: 227±240 Ó Springer-Verlag 1999

Editorial handling: DR

A.J. NaldrettDepartment of Geology,University of Toronto, Toronto, ON, M5S 3B1, Canadae-mail: ajn@quartz.geology.utoronto.ca

why these deposits are not of economic grade. Whenmost major Ni-Cu sul®de deposits are compared, theyprove to have a number of features in common; olivine-rich magma, proximity to a major crustal fault, sul®de-bearing country rocks, chalcophile element depletion inrelated intrusive or extrusive rocks, ®eld and/or geo-chemical evidence of interaction between the magmaand the country rocks, and the presence of or proximityto a magma conduit. The features are thought to explainthe three key requirements (sul®de immiscibilty, ade-quate mixing between sul®des and magma, and locali-sation of the sul®des) discussed and have importantimplications with respect to exploration.

Introduction

Magmatic sul®des can be viewed as belonging to one oftwo major groups (Naldrett et al. 1990), those that arericher in sul®de, and that are of interest primarily becauseof their contained Ni and Cu, and those that containmuch less sul®de and for which the Platinum GroupElements (PGE) content is the principal economic inte-rest. This study is concerned with the ®rst group.

The relative importance of a selected group of theworld's largest Ni-Cu sul®de deposits is illustrated inFig. 1. The Noril'sk and Sudbury camps dominate interms of contained Ni, although Noril'sk ore containssigni®cantly more copper than does Sudbury, and thePGE:Ni ratio is more than ®ve times that of Sudbury, sothat the value of the mineralisation is far greater. Duluthrepresents a major Ni resource, but the low grade

(0.2 wt.% Ni, 0.66 wt.% Cu), coupled with environ-mental constraints on operating large, low grade de-posits means that as yet it has not been mined. Jinchuanranks third in the world in terms of contained Ni, withMt. Keith, Thompson, Voisey's Bay, Kambalda andAgnew in 4th to 8th places respectively. The Voisey'sBay ®gure is the current sum of reserves plus resourcesas quoted by INCO Ltd. (July 1998).

Segregation of liquid sul®de, coupled with settling ofthe sul®de to form rich basal accumulations, is not partof the normal cooling and crystallisation of ma®c mag-ma. The world's important deposits of Ni-Cu sul®des (asopposed to those of interest primarily because of theirPGE content) occur almost exclusively at the base oftheir associated igneous bodies, which implies that themagmas involved were saturated in sul®de, and carryingexcess sul®de at the time of their ®nal emplacement. Thehigh PGE content (1±10 ppb Pt, Pd) of most basalticmagma other than MORB implies that these magmasare not sul®de saturated as they leave the mantle, orduring their ascent into the crust (Naldrett and Barnes1986). Something has to happen to speci®c batches ofmagma prior to emplacement to cause sul®de saturation,if a signi®cant magmatic deposit is to form.

Key aspects in the genesis of a magmatic sul®de oredeposit are therefore: (1) that the host magma becomessaturated in sul®de and segregates immiscible sul®de; (2)that these sul®des react with a su�cient amount ofmagma to concentrate chalcophile elements to an eco-nomic level1; and (3) that the sul®des are themselvesconcentrated in a restricted locality where their abun-dance is su�cient to constitute ore. It is my objective toexamine some of the world's major magmatic sul®decamps with a view to determining how these aspectshave been ful®lled. A discussion of Sudbury is omitted,since, as will become clear, Sudbury is unique, and lacksmany of the characteristics common to most other de-posits, probably due to its formation in the superheatedenvironment created by extra-terrestrial impact (Nald-rett 1999).

Noril'sk, Siberia

The deposits of the Noril'sk region comprise both thoseat Noril'sk itself and the more recently discovered (1962)deposits at Talnakh; they occur at the extreme north-western corner of the Siberian platform which is bor-dered by the Khatanga trough to the north and theYenesei trough to the west. Their emplacement accom-panied the ``Siberian Trap'' magmatic event (the extru-sion of approximately 3 ´ 106 km3 of basalt) whichimmediately followed collision of the East European andSiberian plates. This collision gave rise to the uplift of

Fig. 1 Plot of grade in wt.% Ni versus production+reserves inmillions of tonnes for major Ni sul®de deposits of the world, modi®edafter Naldrett (1994). Data for Sudbury are personal communicationsfrom INCO Ltd. and Falconbridge Ltd. (1990); those for Noril'sk arean estimate only; for Duluth are from Listerud and Meineke (1977);for Jinchuan are from Chai and Naldrett (1992a); for Kambalda,Agnew and Mt. Keith are from posters exhibited by Western MiningCorporation at the 13th Australian Geological Convention, Canberra,February (1996); for Thompson are personal communication from,O.R. Eckstrand (1990); for Voisey's Bay are reserves plus indicatedand inferred resources announced by Mike Sopko, Chairman, INCOLtd., in July 1998

1 It is sometimes not appreciated that the grade of a deposit cannever exceed the metal content in the sul®des themselves (i.e. in theoriginal sul®de liquid), and that this metal content is characteristicof a given deposit and does not vary signi®cantly within it

228

the Ural mountains. Recent descriptions of the Noril'skarea that are available in English include Duzhikov et al.(1992) and Lightfoot and Naldrett (1994).

Naldrett et al. (1995) have argued that the depositsoccur within feeder conduits to a speci®c phase of the3.5 km of lavas comprising the Siberian Trap in theNoril'sk region. The sequence of lavas provides a his-torical record as to what was occurring in the underlyingmagma reservoir(s) and conduits feeding them. Figure 2shows variations of La/Sm, Ni, Cu and Pt with depththrough the sequence (La, Sm, Ni and Cu data are fromLightfoot et al. 1990, 1993, 1994 and Hawkesworth et al.1995; Pt data are from Brugmann et al. 1993) whichLightfoot et al. (1994) have divided into three associa-tions. Critical to understanding the development of themineralisation is the abrupt increase in La/Sm ratio atthe base of the IIB1 formation, which coincides withequally abrupt changes in other parameters, including�Sr, �Nd, and wt.% SiO2 and has been interpreted(Lightfoot et al. 1990, Wooden et al. 1993; Lightfootet al. 1994; Fedorenko 1994; Hawkesworth et al. 1995;

Naldrett et al. 1996a) as the consequence of this magmahaving been contaminated by 8±15% of a granitic par-tial melt derived from mid-crustal gneisses. The subse-quent gradual decrease in La/Sm is attributed (Lightfootet al. 1994; Naldrett et al. 1995) to repetitive pulses of adi�erent magma type (typi®ed by the uppermost ¯ows ofthe IIC2 formation) entering the plumbing system forthe lavas and progressively ``diluting'' the compositionalcharacteristics that had resulted from the contamina-tion.

As seen in Fig. 2, the increase in La/Sm coincideswith pronounced decreases in the Ni, Cu and PGEconcentrations of the lavas. Naldrett et al. (1992),Brugmann et al. (1993) and Fedorenko (1994) proposedthat the chalcophile elements missing from the lavas arethose giving rise to the ores. These authors point outthat chalcophile element depletion of this magnitude canonly be explained in terms of equilibration with sul®de.

The initiation of sul®de immiscibility might have beenbrought about by one or both of two mechanisms. Themixing of ma®c magma with a felsic contaminant can solower the ability of the resulting hybrid to dissolve sul-®des that immiscibility can result even if the two mag-mas involved in the mixing are sul®de unsaturated(Irvine 1975; Li and Naldrett 1993). Additionally, theheavy isotopic signature of the ores (o34S is in the range

Fig. 2 Stratigraphic variation in La/Sm, ppm Ni, ppm Cu and ppb Ptacross the 3.5 km vertical succession of ¯ood basalts in the Noril'skregion (data from Hawkesworth et al. 1995; Peter Lightfoot, personalcommunication 1996)

229

+8 to +12, Grinenko 1985) suggests that contamina-tion may have involved the ingestion of crustal sulfurfrom the surrounding anhydrite evaporites.

Signi®cant mineralisation occurs only in the Noril'sk-type intrusions which lie within 7 km of the Noril'sk-Kharayelakh fault and which appear to be slightlyyounger than the Lower Talnakh type. An appreciationof a typical mineralised intrusion can be obtained fromthe Bears Brook open pit in the Noril'sk I intrusion, inwhich it can be seen that the main body (MB) of theintrusion is di�erentiated and cuts at high angle across200 m of strati®cation in the sedimentary and volcaniccountry rocks, a feature which this author can only as-cribe to thermal erosion (Fig. 3). The MB is ¯anked byand connected with peripheral sills which appear to beinjection ``push-apart'' structures.

The Noril'sk-type (mineralised) intrusions are un-usual in a number of ways:1. They contain or are associated with a very high pro-portion of sul®de, ranging between 2 and 10 wt.% oftheir total mass (Naldrett et al. 1992).2. These sul®des contain a very large concentration ofPGE, which, according to Naldrett et al. (1992), musthave come from a mass of magma at least 200 times thatrepresented by the intrusions.3. It has been emphasised repeatedly in the Russian lit-erature on Noril'sk that the mineralised intrusions aresurrounded by an intense metamorphic and metaso-matic aureole (Genkin et al. 1981; Likhachev 1994). Inmany cases this extends farther into the country rocksthan the thickness of the intrusions themselves, in somecases 400 m (Genkin et al. 1981).4. The sulfur isotopic composition of the sul®des is tooheavy for mantle-derived sulfur, ranging from +8 to+12d34S (Godlevsky and Grinenko, 1963; Grinenko,1985).

Naldrett et al. (1995) noted that these aspects areexplicable if the mineralised intrusions have acted asfeeders to the 5000±10 000 km3 of volcanic magma rep-resented by the IIB1-IIC2 formations, and if much of thesul®de has formed at a shallow level in the crust, es-sentially in situ. On the basis of La/Sm, Gd/Yb and Th/Ta ratios coupled with Nd isotopic data they concludedthat the last magma to ¯ow through these bodies wasthat feeding the IIC2 volcanic formation. They propose

that although the bulk of the chalcophile metals wereextracted from magma giving rise to the IIB1 and IIB2lavas, magma continued to ¯ow through the system upuntil the IIC2 lavas.

The ¯ow of a large amount of magma through aplumbing system of which the mineralised intrusions arepart, accounts for the very extensive metamorphismassociated with them. It also accounts for the very highproportion of sul®de, and of PGE associated with thesul®de, in comparison with the amount of magma rep-resented by the mineralised intrusions themselves.Naldrett et al. (1995) noted that the geochemical dataare explicable if the IIB1-IIB2 magma underwent twostages of contamination (Fig. 4), a mid-crustal stagegiving rise to the high La/Sm ratio and the other geo-chemical characteristics indicative of crustal interaction,and a second stage, at essentially the present level of themineralisation, in a chamber developed within the ar-gillites, evaporites and coal measures, to which themineralised intrusions are exit channels. It is likely thatthe magma was eroding the walls of its chamber at thisstage, ingesting Devonian evaporite and C-bearingTungusskaya formation, and reducing the evaporite toproduce sul®de; this accounts for the high d34S values ofthe sul®des. The sul®des were then swept out from thecentral di�erentiation chamber southward in the case ofNoril'sk I and northward in the case of Talnakh, andbecame lodged in the thicker (thermally eroded) parts ofthese magma conduits. Fresh magma then surgedthrough the system, stirring up the trapped sul®deswhich interacted with the magma and, in the process,progressively upgraded the sul®des in chalcophile met-als. The extent of this upgrading depended on theamount of magma ¯owing through the di�erent con-duits of the system. Progressively later pulses of magmaencountered sul®des that had progressively higher con-centrations of Ni, Cu and PGE, with the result that themagmas lost less of their chalcophile elements. This ac-counts for the ``recovery'' in the concentrations of thesemetals observed in lavas from the IIC1 to IIIA units(Fig. 2).

Fig. 3 Vertical section through the east branch of the Noril'sk Iintrusion at the Bear's Brook open pit (from Distler and Kunilov1994)

Fig. 4 A model for the development of the lavas and intrusions atNoril'sk during the eruption of the IIB1 (nd1), IIB2 (nd2), IIB3 (nd3),IIC1 (mr1) and IIC2 (mr2) formations. Modi®ed after and illustratingthe views expressed by Naldrett et al. (1995)

230

Voisey's Bay Ni-Cu-Co deposit

Description of the deposit

The Voisey's Bay deposit (Naldrett et al. 1996b, 1997;Ryan 1997) is associated with the 1.334 Ga (Amelinet al. 1997) Voisey's Bay complex. This is a member ofthe Nain Plutonic Suite (1.35±1.29 Ga) which is an an-orthosite-granite-troctolite-ferrodiorite complex thattransects the east-dipping 1.85 Ga collisional boundarybetween interbanded garnet-sillimanite and quartzo-fe-ldspathic (Tasiuyak) gneisses of the Proterozoic Chur-chill Province to the west and quartzo-feldspathicgneisses of the Nain Province to the east (Ryan et al.1995).

At the present level of understanding (Fig. 5) theVoisey's Bay deposit comprises two, east-plunging in-trusions (Reid Brook and Eastern Deeps intrusions), onesituated about 2 km vertically above the other, linked bya steeply dipping igneous sheet. The lower, Reid Brook,intrusion is situated in Proterozoic (Tasiuyak) gneiss. Todate, the Reid Brook body has only been intersected inthe western part of the mine area, where it is seen tonarrow upward and merge with the linking sheet. At thepoint of merger, the sheet tends to be choked by ``feederbreccia'' comprising numerous partially reacted inclu-sions of gneiss in a matrix of norite or troctolite ac-companied by variable amounts of sul®de.

The thickness of the linking sheet varies from 10±100 m. Mineralisation is restricted to the wider parts,which occur as a series of elongate lenses, within theplane of the sheet, all raking to the east. In the discoveryzone, where the dip of the sheet varies between 70 and40°N, a well developed stratigraphy has been observed,with an upper chill zone giving way downward to 5±20 m of unmineralised olivine gabbro (Feeder OlivineGabbro or FOG), which quickly passes downward intotroctolite with 30±50 percent interstitial sul®de (LeopardTroctolite or LT). The LT is in sharp contact with theBasal Breccia Sequence (BBS, similar to the FeederBreccia), which consists of inclusions of variably reactedgneiss and fresh troctolite, pyroxenite and peridotite in atroctolitic to noritic matrix. Elsewhere, where the atti-tude of the sheet is nearly vertical, this simple stratig-raphy is not present but the FOG is always developedadjacent to one or other of the walls.

The Eastern Deeps intrusion occurs at surface, withits base plunging at 20±25° east southeast (Fig. 6). Sev-eral tens of metres of BBS occur along the base and areoverlain by a troctolite with numerous inclusions, vari-able amounts (commonly 10±20 vol %) of sul®de and avariable texture (Varied Textured Troctolite or VTT).This passes upward into a troctolite with even textureand less sul®de and inclusions (Normal Troctolite)which appears to be intrusive into the uppermost unit ofthe Eastern Deeps observed so far, the Olivine Gabbro.A feeder sheet intersects the base of the Eastern Deepsintrusion along much of its explored length. Naldrett

et al. (1996b) showed that massive sul®de occurs over asigni®cant length of the intrusion close to and within themouth of the feeder.

The richest zone of mineralisation discovered so far isknown as the ``Ovoid''. This is a 300 ´ 600 m basin ofmassive sul®de, up to 110 m deep, that has developedabove and to the south of the sheet linking the two in-trusions. Naldrett et al. (1996b) interpreted it to repre-sent the base of the Eastern Deeps intrusion, nowexposed at surface.

Li and Naldrett (1997) reported that the olivines inthe gabbros and troctolites of the Voisey's Bay complexcomprise two groups on the basis of their Ni content.The olivine gabbro at the top of the Eastern Deeps in-trusion, the FOG and some zones within the troctolite ofthe Reid Brook body contain Ni-depleted olivine, while

Fig. 5 Surface plan of the Voisey's Bay deposit (after Li and Naldrett1998)

231

the NT and VTT of the Eastern Deeps, and LT of thelinking sheet contain olivine with a higher Ni content.Olivines within ultrama®c-ma®c inclusions within theBBS de®ne a trend on the Ni vs Fo diagram (Fig. 7)which is consistent with protracted fractional crystalli-sation of olivine from troctolitic magma, leading to Ni-depleted magma (and thus Ni-depleted olivines). Somegabbros from the Reid Brook intrusion, the FOG andEastern Deeps olivine gabbro lie on an extension of thistrend. They propose a two-stage model of ore genesis.Troctolitic magma rose to form a chamber (Reid Brookintrusion) within the Tasiuyak gneiss, where it frac-tionated, giving rise to ultrama®c cumulates, reactedwith Tasiuyak gneiss, and became sul®de saturated.Application of experimentally derived partition coe�-cients to the Ni-depleted olivines indicate that the sul-®des segregating from the Ni depleted magma wouldhave contained 1.5±2.0 wt.% Ni. This magma thenprogressed up the linking sheet to form the OG of theupper (Eastern Deeps) chamber; the FOG is a relict of

its passage which adhered to the walls of the sheet. Atthe same time, fresh magma was entering the lowerchamber, disrupting the cumulates and mixing with re-sidual magma, BBS and the sul®des within it, enrichingthe sul®des in Ni and Cu and transporting them up toform the mineralisation in the linking sheet, the Ovoid,the massive ore at the base of the Eastern Deeps and thedisseminated ore of the VTT. The sul®des became con-centrated in widened zones within the linking sheet, andat the mouth of the feeder where the rate of magma¯owed decreased as the magma entered the chamber.

Jinchuan, China

Geology

The Jinchuan deposit is located in north central China,within but close to the southern margin of theSino-Korean craton. It lies within one of a series ofma®c-ultrama®c bodies that intrude a marginal, north-west-trending, uplifted belt, the Longshoushan belt.Lower Proterozoic migmatites, gneisses, schists andmarbles form a southeasterly facing monoclinal se-quence, overlain to the southwest by Upper Proterozoicconglomerates, sandstones, limestones and schists (Tang1993). Major northwesterly striking faults bound andoccur within the uplifted block, and the Jinchuan bodyoccupies one of these which cuts Lower Proterozoicstrata. Tang et al. (1992) cite a Sm-Nd age for the in-trusion of 1.501 � 0.031 Ga. Chai and Naldrett (1992b)have concluded that the host intrusion was emplaced ina rift zone that formed at the boundary between theSino-Korean craton and a developing Proterozoicocean; this subsequently closed with the welding offolded Upper Proterozoic strata onto the craton.

The host to the deposit comprises a mass of dunite,lherzolite, olivine websterite and websterite, 6500 m in

Fig. 6 ESE-WNW sectionthrough the ``Eastern Deeps'',modi®ed after Naldrett et al.(1996)

Fig. 7 Plot of ppm Ni in olivine versus Fo content of olivine forultrama®c inclusions, troctolite of the Reid Brook intrusion andFeeder Olivine Gabbro at Voisey's Bay, showing a trend consistentwith protracted crystallisation of olivine from troctolitic magma

232

length by a few m to >500 m in width. It has a V-shapedcross section, tapering downward to pinch out at depthsof 500±1300 m. Northeast-trending faults o�set di�erentportions of the intrusion, which have been developed asseparate mining blocks (Fig. 8).

The internal structure of the intrusion varies along itslength, from block to block. At the northwest end in-ternal contacts are steep (Fig. 8) with dunite predomi-nating at depth, and lherzolite increasing toward the top,and forming thin marginal zones to the dunite at depth.In the central and southeastern parts of the intrusion,dunite is also predominant at depth, but layering be-tween lherzolite, plagioclase lherzolite, websterite andolivine websterite is present, parallel to the margins atthe margins but ¯attening and becoming horizontalclose to the axis of the body (Fig. 8).

The predominant ore types are ``net-textured ore'', inwhich olivine grains are enclosed in a continuous networkof sul®de and lower grade ``disseminated ore'' in whichsul®des occur both interstitial to silicate minerals and asdiscrete blebs. In general, dunite is the principal host tomineralisation, and essentially all dunite is of ore grade.

Origin of the intrusion and its mineralisation

The nature of the magma responsible for the ultrama®cbody hosting the mineralisation (ma®c or ultrama®c)has been a source of controversy. Chai and Naldrett(1992a) have demonstrated using a plot of FeO versusMgO, that the liquid intercumulus to the olivine of theperidotites that constitute the intrusion contained

Fig. 8 (top) Geological map ofthe Jinchuan intrusion;(bottom)longitudinal and transversevertical sections through theJinchuan intrusion showing thedistribution of rock types andthe nature of the Ni-Cu miner-alisation that they contain,from Chai and Naldrett (1992a)

233

12 wt.% MgO and 11.5 wt.% FeO. When coupled withthe other geochemical characteristics, such as the Ni/Curatio of 1.5 and the gabbroic-type PGE pro®les (i.e. high(Pt+Pd)/((Ru+Ir+Os) ratios), this is strong evidencethat the composition of the initial magma at Jinchuanwas not ultrama®c, but that of a magnesian basalt. Chaiand Naldrett (1992a) proposed that the gabbroic cu-mulates which would also have resulted from a di�er-entiating magnesian basalt magma have been removedby erosion, and that the Jinchuan body is the root zoneof a linear body of overall gabbroic composition andtrumpet shaped cross section, similar to but smaller thanthat of the Great Dyke of Zimbabwe, as is illustrated inFig. 9. They interpret the deeper parts of the roots asconduits up which magma entered the intrusion. Densesul®de and olivine were carried along in rapidly ¯owingmagma in the narrow parts of conduits feeding the in-trusion, but collected where these conduits widened outon entering the main body.

Komatiite-related deposits, Norseman-Wiluna belt,Western Australia

Komatiitic ore deposits of the Wiluna-Norseman beltcan be considered, as can all other komatiite-relateddeposits, as falling into one of two groups: Group 1 are

generally small (1 to 5 ´ 106 tonnes), high grade (1.5±3.5% Ni) deposits which occur at the base of mesocu-mulate dunitic ¯ows; group 2 comprise very large (100±500 ´ 106 tonnes), low-grade (0.6% Ni) deposits of®nely disseminated sul®de in channel-like lenses of ad-cumulate dunite. Examples of group 1 include the de-posits of the Kambalda district (Ross and Hopkins1975; Gresham and Loftus-Hills 1981; Gresham 1986)and the Widgiemooltha dome (Fisher 1979; McQueen1981). Examples of group 2 include the Six±Mile andMt. Keith deposits near Yakabindie, Western Australia(Burt and Sheppy 1975; Naldrett and Turner 1977; Hillet al. 1989; 1995).

Hill et al. (1989, 1990) and Hill (1997) have discussedthe internal structure of komatiite ¯ows and have shownthat the komatiitic lavas can be subdivided into anumber of distinctive facies, which relate to their dis-tance from the eruptive source and the rate of eruption(Fig. 10). Flows that are close to their source, andtherefore are hottest, tend to develop as very extensivesheets (in some cases in excess of 35 ´ 150 km by severalhundreds of metres thick) of adcumulate dunite. Whenthis facies develops over a substrate with low meltingtemperature, such as felsic volcanic rocks, the ¯ow maybecome channelled with the principal ¯ow restricted to anumber of channels. These channels have been cut intothe underlying substrate by thermal erosion of the¯owing komatiitic magma (Huppert and Sparks 1985).Farther downstream, where ¯ow is less rapid andprobably lamina, ``lava plains'' composed of sheet ¯ows

Fig. 9 Model for the Jinchuan intrusion after emplacement asenvisaged by Chai and Naldrett (1992b)

234

with orthocumulate rather than adcumulate dunite de-velop. Channelisation under these circumstances cangive rise to lenses of mesocumulus olivine (several hun-dreds of metres wide and up to 100 m thick), whichrepresent the main ¯ow channels separated by lavaplains in which the upper parts of the ¯ows have well-developed spinifex texture.

Group 1 deposits are associated with the more distalportions of komatiite ¯ows, occurring within well-de-veloped erosional troughs that have formed at the baseof lava rivers characterised by olivine mesocumulates(40±44 wt.% MgO). The Group 2 deposits are thoughtto be more proximal in nature, occurring in dunite lenseswhich represent channelised ¯ows of adcumuls dunitelava (45±50 wt.% MgO).

The source of the sulfur in komatiitic lava ¯ows is stillan open question. Values of d34S in the Kambalda oresare generally within �5 of 0 per mil, which is not di-agnostic because the sedimentary sources of sul®de showa similar variation, although Lesher (1989) has demon-strated a close match between d34S in ores and footwallsul®des at a number of localities, including Kambaldaand Mt. Windarra.

Mineralisation of the Duluth complex

The Duluth complex comprises a large composite in-trusion of troctolite-gabbroic anorthosite, that crops outas an arcuate body extending about 240 km northeastfrom Duluth, Minnesota, to the Canadian border, closeto the northwestern shoreline of Lake Superior. LakeSuperior itself overlies a graben, de®ned by major faultsand hinge lines, within which Archean crust has beenthinned to less than 1/4 of its previous thickness, andover 17 km of volcanics and sediments have accumu-

lated. It is situated where the North American ``mid-continent rift'', which is marked by well-de®ned gravityand magnetic anomalies, changes its trend from south-west-northeast to northwest-southeast, and is probablythe triple junction of a failed rift system that developedin response to an underlying hot-spot.

The Duluth Complex is a 1.2 Ga (Faure et al. 1969)composite intrusion which the studies of Grout (1918),Green et al. (1966), Bonnichsen (1972), Phinney (1970,1972), Weiblen and Morey (1975, 1980), Martineau(1989), Severson (1988, 1991) and Severson and Hauck(1990) have shown to consist of two series, an olderseries of anorthositic rocks which is cut towards itswestern margin by a series of later intrusions consistingpredominantly of troctolite. It is closely associatedpetrogenetically with Keewanawan basalts of the LakeSuperior area and was intruded close to the contact ofthese rocks with older Precambrian basement rockswhich are now exposed to the northwest (Fig. 11).These basement rocks consist, in the north, of a com-plex of Archean felsic intrusions and volcanics, pri-marily the Giants Range Granite, which are overlainunconformably to the south by the south-dipping, mid-Proterozoic, Biwabik Iron Formation. The iron for-mation is itself overlain by black argillites, greywackes,siltstones, graphitic slates, and sul®de facies iron-for-mation comprising the Virginia Formation.

The Complex is the host to more than 4 ´ 109 tonnesof mineralisation averaging 0.66 wt.% Cu and 0.2 wt.%Ni (Listerud and Meineke 1977). The mineralisationoccurs along the western margins (stratigraphic bases) ofthe troctolitic intrusions which de®ne the western ex-tremity of the Complex (Fig. 11). In most deposits, themineralisation consists of pyrrhotite, chalcopyrite,pentlandite, and cubanite weakly disseminated introctolite and norite within 300 m of the base of the

Fig. 10 Schematic illustrationof an interpreted regional ko-matiite volcanic complex,formed by cataclysmic sus-tained eruption. Note that ad-cumulate (dunite) sheet ¯owfacies develops close to thevent, and becomes channeledto adcumulate channel ¯owfacies farther from the vent.Downstream, cooling lavagives rise to orthocumulatesheet ¯ow facies within which afacies composed of mesocu-mulate dunite channel ¯owsmay be present. The marginsand farthest extremities of thevolcanic complex consist oflava plains comprising ¯owsintermixed with or capped bythin ``Munro-type'' ¯ows afterHill et al. (1990)

235

complex. The mineralised zones are characterised bynumerous inclusions of country rocks consisting ofhornfelsed Virginia and Biwabik Formations, togetherwith barren gabbro and peridotite. At both the Minna-max and Dunka Road deposits (Fig. 11), norite is morecommon and troctolite less common in the vicinity ofthe sul®de horizons, attesting to reactions between thetroctolite and country rock inclusions. An unusual fea-ture of many of the deposits is the occurrence of as muchas 10% graphite in certain zones. Other unusual featuresof the sul®de environment include the presence of greenhercynite (Mg-Al.) spinel in some of the troctolites, andthe local occurrence of cordierite. As in other deposits,numerous partially resorbed remnants of what appear tobe hornfelsic Virginia slate occur in the troctolite.

Sulfur in the Duluth deposits is heavy, with values ofd34S ranging from 11±16 in the Waterhen intrusion(Mainwaring and Naldrett 1977); and 0.2±15.3 per mil inthe Dunka Road deposit and 2±7 in the Babbit deposit(Ripley 1981, 1986). Pyrrhotite in the underlying Vir-ginia Formation shows a similar range in isotopiccomposition, also supporting a country rock source formuch of the sulfur. The heterogeneous distribution ofthe sulfur isotope values throughout many of the de-posits, coupled with a wide range in the nickel contentsof olivine in the mineralised zones led Ripley (1986) tosuggest that sulfur introduction had occurred essentiallyin situ, and that widespread equilibration between thesul®des and their enclosing silicates had not occurred.Citing the lack of geochemical evidence supportingwidespread bulk assimilation, he proposed that the sul-fur had been introduced in a volatile phase that wasreleased from the footwall rocks as they were meta-morphosed by the intrusion. This is supported by thecommon association of hydrous minerals such as biotiteand amphibole and patches of pegmatite with the min-eralisation.

Conclusions and application to exploration

The preceding discussion indicates that there are manygeological features in common amongst the world'smajor concentrations of Ni-Cu sul®des. Some of themost important of these are summarised in Table 1.They include (1) a magma capable of crystallising sub-stantial amounts of olivine, (2) a prominent crustal su-ture, (3) plentiful sulfur in the adjacent or subjacentcountry rocks, (4) evidence of chalcophile element de-pletion in associated silicate magmas, (5) evidence ofinteraction with country rocks and (6) the occurrence ofmineralisation within or close to a conduit along whichmagma ¯ow has been concentrated.

To discuss each of these features in turn, olivine-richmagmas have been important at Noril'sk, Voisey's Bay,Jinchuan, Kambalda and Duluth. The rocks of theSudbury Igneous Complex stand out in their absence ofolivine, except for a suite of inclusions that are closelyassociated with mineralised Sublayer. The olivine rangesfrom highly forsteritic in the Kambalda komatiites(Fo90±95) to quite forsteritic at Jinchuan (Fo80) andNoril'sk (Fo80) to moderately forsteritic at Voisey's Bay(Fo50±65) and Duluth (Fo50±65). The importance ofthe presence of olivine is thought to be 3-fold; (1) oli-vine-normative magmas are generally hotter than thosewith a higher SiO2 content, (2) the Ni content of olivine-normative magmas tends to be higher than those with nonormative olivine (the Ni content depends, of course, onthe Fo content of the normative olivine, and on whetherthe magma has become chalcophile depleted as discus-sed later), and (3) an olivine normative magma will reactwith SiO2-bearing country rocks more readily than onethat is richer in SiO2 (see later).

Prominent crustal sutures have clearly played animportant role in providing zones of weakness up whichmagma has ascended into supercrustal rocks at Noril'sk(Noril'sk-Kharayelakh fault), Jinchuan (faultingbounding the Sino-Korea platform) and Duluth (mid-continent rift structure). The Voisey's Bay complex liesclose to the southerly extension of the Abloviak shearzone which marks the collisional suture between theNain and Churchill provinces.(Ryan et al. 1995) andalong which the last documented deformation is dated at1.73±1.75 Ga (Van Kranendonk 1996); it is possible thatthis remained a zone of weakness and facilitated em-placement of the 1.35±1.29 Nain Plutonic Suite. Earlystructural elements are much harder to recognise inArchean terrains, but the Eastern Gold®elds are char-acterised by a series of NNW-trending faults which mayre¯ect ancient sutures, possibly those associated withearly rifting. In contrast, the Sudbury deposits and theirassociated astrobleme owe their origin to activationfrom above, rather than below!

Sulfur-bearing country rocks adjacent or subjacent todeposits are present at Noril'sk (Middle Devonian eva-porites containing anhydrite or gypsum), Voisey's Bay(sul®de and graphite-bearing Tasiuyak gneiss), Kam-

Fig. 11 Map of the western end of Lake Superior, showing thelocation of the Duluth complex and its relationship to the countryrocks (after Mainwaring and Naldrett 1977)

236

balda (chemical sediments which lie on the basalt sub-strate to the komatiitic lavas) and Duluth (pyrite andgraphite-bearing Virginia formation). No obviouscrustal source of sulfur is available at Jinchuan or atSudbury. At Noril'sk and Duluth, isotopic evidence ar-gues strongly for the incorporation of crustal sulfur intothe ores. At Voisey's Bay isotopic evidence is suggestivethat sulfur from the Tasiuyak gneiss has played a role inore formation (Ripley et al. 1997) and at Kambalda thesulfur isotopic data are consistent with a crustal sourcebut provide no compelling evidence for it. Values of d34Sin the Sudbury ores are close to zero, and thus provideno support for a crustal source.

At Noril'sk, chalcophile element depletion is welldocumented in basalts representative of magma that justpre-dated that hosting the ores, although this is lessapparent in the basalts correlative with the mineralisedintrusions. The same is true at Voisey's Bay, where theinitial magma to ¯ow through the system is chalcophile-depleted, but that associated most directly with themineralisation is not notably depleted. Duke andNaldrett (1978) pointed out that the komatiites andkomatiitic basalts are Ni-depleted in a regional sense atKambalda, but the rocks most directly related to themineralisation do not show signi®cant depletion, even intheir PGE content (Lesher 1989), which would be moresensitive to chalcophile element depletion than Ni and/or Cu. In each of these cases it appears that fresh magmahas ¯owed through the mineralised magma channelssubsequent to that which shows chalcophile elementdepletion. Chalcophile depletion has not been estab-lished at Duluth, Jinchuan or Sudbury.

The geochemical data discussed demonstrate stronginteraction between certain of the Noril'sk magmas andcrustal rocks. Furthermore the Taxitic Gabbro, whichis a variable-textured rock containing many partiallydigested inclusions of intrusive and country rocks is®eld evidence of such interaction. Lambert et al.'s(1997) Re-Os data, and the intense alteration experi-enced by inclusions in the BBS is compelling evidencethat such interaction has occurred at Voisey's Bay.Geochemical data of Frost and Groves (1989),McNaughton et al. (1988) and Arndt and Jenner (1986)support the interaction of komatiitic magma with crustin the Kambalda district. Field observation, and Rip-ley's (1986) chemical and O isotopic data support atleast limited interaction at Duluth. Crustal interactionhas not been documented at Jinchuan, although thechemical and isotopic measurements which are requiredto prove this geochemically have yet to be made. Themost intense interaction is present at Sudbury, whereNaldrett et al. (1986) suggested that the SIC consistedof ¯ood basalt magma contaminated by 50% of impactmelt, and Faggart et al. (1985) and Grieve (1994),amongst others, have suggested that the whole complexmay be an impact melt. Naldrett et al. (1986) and Liand Naldrett (1993) suggested that 50% intermixing ofsul®de-unsaturated primitive mantle-derived basalt anda felsic impact melt could induce sul®de saturation, andT

able

1ComparisonofNi-Cudeposits

Key

factor

Noril'sk

Voisey'sBay

Jinchuan

Kambalda

Duluth

Sudbury

Olivne-rich

magama

Yes-picrite

Fo

80

Yes-troctolite

Fo

50±65

Yes-harzburgite&

duniteFo

84

Yes-komatiitic

lavas

Fo

90±95

Yes-troctolite

Fo

50±65

No

Prominent

crustalsuture

Yes-N

oril'sk

kharayelakhFault

Yes?-Abloviakshearzone

Yes-continentalmargin

Yes?-NNW

faults

Yes-LakeSuperiorRift

No

Sourceofsulfur

incountryrock

Yes-Evaporites

Yes-TasiuyakGneiss

Notestablished

Yes-underlying

sedim

ents

Yes-V

irginia

Form

ation

No

Chalcophile

depletion

Yes-inNd-M

rvolcanics

Yes-olivinegabbro,

Feeder

troctolite

Notestablished

Yes-onaregionalscale

Notestablished

Notestablished

Evidence

of

interactionwith

countryrocks

Yes-taxitic

gabbro

and

chem

icaldata

Yes-basalbreccia

sequence

and

chem

icaldata

Notestablished

Yes-chem

icaldata

Yes-visualandchem

ical

Yes-chem

icaland

®eldobservation

Magmaconduit

Yes-m

arked

bythickened

zones

ofpicriticand

taxitic

gabbro

Yes-thickened

zones

infeeder

sheetandentryof

sheetto

intrusion

Yes-funnel

shaped

feeder

Yes-lavachannels

No-sul®des

appearto

haveform

edin

situ

Notestablished

237

thus account for the widespread mineralisation atSudbury.

Turning to the importance of some form of magma¯ow channel, it is argued that thermally eroded feederchannels for overlying volcanics have served as traps forsul®des and now constitute the ore zones. Some of theore at Voisey's Bay is in a similar environment, and theremainder occurs at the line of entry of the feeder to theintrusion. The ore at Jinchuan occurs within what isinterpreted to be the funnel-shaped (in cross section)feeder to an intrusion that has largely been removed byerosion. The Kambalda ores occur at the base of chan-nels within which much of the komatiite magma ¯owwas localised. It is only at Duluth and Sudbury thatmagma conduits do not appear to be a major ore con-trol. It is this author's opinion that if the sul®de-bearingmagma at Duluth had continued intruding farther, afterit had reacted with the Virginia Formation and immis-cible sul®des had developed, the related deposits mighthave become much richer in sul®de and would have beenmined long ago. Sudbury is a totally di�erent environ-ment, that of the chaos attending a meteorite impact,and so far one has an excellent picture of where thedeposits occur, but not of why they occur where theyare.

One can now re-visit the introduction to this paper,and discuss possible answers to the three key factorsoutlined there. With respect to the ®rst factor, why didsul®de immiscibilty occur?, in many of the major Ni-Cucamps discussed here deep crustal sutures have facili-tated the ascent of olivine-rich magma into the crust,where this has reacted with crustal rocks, acquiringcrustal sulfur. Extensive assimilation of crust, or mixingwith a crustal melt, has been documented for all oc-currences except Jinchuan, at which it has not beenlooked for. At Sudbury, where the addition of crustalsulfur is unlikely, felsi®cation of a ma®c melt has beensuggested as the cause of sul®de saturation. Turning tothe second factor of the concentration of sul®des from alarge volume of magma, the entrapment of sul®deswithin a magma ¯ow channel stands out as an extremelyimportant mechanism. This feature of many depositsalso provides an answer to the question of the third keyfactor, that of how the sul®des were able to interact withsu�cient magma to concentrate Ni, Cu and PGE toeconomic levels. At Noril'sk it has been shown thatcontinued ¯ow of magma through the conduit, afterinitial deposition of sul®de, and the continued interac-tion of this magma with the sul®des has resulted in theirbeing upgraded. The latest studies at Voisey's Bay andKambalda have indicated that this upgrading processhas also occurred in these deposits, thus providing ananswer to the third key factor. This aspect is extremelyimportant; the battle ®eld of Ni exploration is litteredwith the corpses of projects in which massive sul®deswere located but in which, to the dismay of investors,subsequent analysis proved the sul®des to contain in-adequate tenors of Ni, Cu, Co and PGE.

References

Amelin Y, Li C, Naldrett AJ (1997) Multi-stage evolution of theVoisey's Bay complex, Labrador, Canada revealed by U-Pbsystematics of zircon, baddeleyite and apatite. Volume of Ab-stracts, American Geophysical Union fall meeting, San Fran-cisco, December 1997

Arndt NT, Jenner GA (1986) Crustally contaminated komatiitesfrom Kambalda, Western Australia. Chem Geol 56: 229±256

Bonnichsen W (1972) Southern part of the Duluth complex. In:Sims PK, Morey GB (eds) Geology of Minnesota: a centennialvolume: St. Paul, Minnesota Geological Survey, pp 361±387

Brugmann GE, Naldrett AJ, Asif M, Lightfoot PC, GorbachevNS, Fedorenko VA (1993) Siderophile and chalcophile metalsas tracers of the evolution of the Siberian trap in the Noril'skregion, Russia, Geochim Cosmochim Acta 57: 2001±2018

Burt DRL, Sheppy NR (1975) Mount Keith nickel sul®de deposit,In: Knight CL (ed) Economic geology of Australia and PapuaNew Guinea, I. Metals: Australasian Institution Mining andMetallurgy Mon. 5, pp 159±168

Chai G, Naldrett AJ (1992a) Petrology and geochemistry of theJinchuan ultrama®c intrusion: cumulate of a high-Mg basalticmagma. J Petrol 33: 1±27

Chai G, Naldrett AJ (1992b) PGE mineralisation of the JinchuanNi-Cu sul®de deposit, NW China. Eco Geol 87: 1475±1495

Distler VV, Kunilov VY (1994) Geology and ore deposits of theNoril'sk region, Guidebook to Noril'sk Field Trip, VII IntPlatinum Symp, Moscow-Noril'sk

Duke JM, Naldrett AJ (1978) A numerical model of the fraction-ation of olivine and molten sul®de from komatiite magma.Earth Planet Sci Lett 39: 255±266

Duzhikov OA, Distler VV, Strunin BM, Mkrtychyan AK, Sher-man ML, Sluzhenikin SS Lurje AM (1992) Geology and met-allogeny of sul®de deposits of Noril'sk Region, USSR. (Englishtranslation of 1988 Russian edition) Society of Economic Ge-ologists, Special Publication 1, 242 pp

Faggart BE, Basu AB, Tatsumoto M (1985) Origin of the SudburyComplex by meteorite impact: neodymium isotopic evidence.Science 230: 436±439

Faure G, Chaudhuri S, Fenton M (1969) Ages of the Duluth ga-bbro complex and the Endion sill, Duluth, Minnesota: J Geo-phys Res 74: 720±725

Fedorenko VA (1994) Evolution of magmatism as re¯ected in thevolcanic sequence of the Noril'sk region. In: Lightfoot PC,Naldrett AJ (eds) Proc Sudbury-Noril'sk Symp. Ontario Geo-logical Survey Special Vol. 5: 171±184

Fisher D (1979) The petrology of the Mt. Edwards nickel sul®dedeposit, Widgiemooltha, Western Australia: Unpub PhD The-sis, University of Toronto, 320 p

Frost KM, Groves DI (1989) Ocellar units at Kambalda: evidencefor sediment assimilation by komatiite lavas. In: PrendergastMD, Jones MJ (eds) ``Magmatic sulphides ± The ZimbabweVolume'' Special Publication, Institution of Mining and Met-allurgy, London, pp 207±214

Genkin AD, Distler VV, Gladyshev GD, Filiminova AA,Evstigneeva TL, Kovalenker VA, Laputina IP, Smirnov AV,Grokhovskaya TL (1981) Sul®de copper-nickel ores of theNoril'sk deposits. Moscow, Nauka, 234 pp (in Russian)

Godlevsky MN, Grinenko LN (1963) Some data on the isotopiccomposition of sulfur in the sul®des of the Noril'sk deposit.Geochemistry 1: 335±41

Green JC, Phinney WC, Weiblen PW (1966) Gabbro Lake Quad-rangle, Lake County, Minnesota Geological Survey Miscella-neous Map M2

Gresham JJ (1986) Depositional environments of volcanic perido-tite-associated nickel sulphide deposits with special reference tothe Kambalda Dome. In: Freidrich GH, Genkin AD, NaldrettAJ, Ridge JD, Sillitoe RH, Vokes FM (eds) Geology andmetallogeny of copper deposits Springer-Verlag, HeidelbergBerlin, pp 63±90

238

Gresham JJ, Loftus-Hills GD (1981) The geology of the KambaldaNickel Field, Western Australia. Econ Geol 76: 1373±1416

Grieve RAF (1994) An impact model of the Sudbury Structure, In:Lightfoot PC, Naldrett AJ (eds) Proc Sudbury-Noril'sk SympOntario Geological Survey Special Vol 5: 119±132

Grinenko LN (1985) Sources of sulfur of the nickeliferous andbarren gabbro-dolerite intrusions of the northwest Siberianplatform. Int Geol Rev 27: 695±708

Grout FF (1918) The lopolith; an igneous form exempli®ed by theDuluth Gabbro. Am J Sci (4th Ser) 46: 516±522

Hawkesworth CJ, Lightfoot PC, Fedorenko VA, Blake S, NaldrettAJ, Doherty W, Gorbachev NS (1995) Magma di�erentiationand mineralization in the Siberian continental ¯ood basalts.Lithos 34: 61±88

Hill RET (1997) Komatiite volcanology and associated nickelsulphide deposits. In: Papunen H (ed) Mineral deposits: re-search and exploration, where do they meet? Proc fourth Bi-ennial SGA Meeting, Turku Finland, August 1997, Balkema,Rotterdam, 5±6

Hill RET, Gole MJ, Barnes SJ (1989) Olivine adcumulates in theNorseman-Wiluna greenstone belt, Western Australia: impli-cations for the volcanology of komatiites. In: Prendergast MD,Jones MJ (eds) Magmatic sul®des ± the Zimbabwe volume.Special Publication, Institution Mining and Metallurgy, Lon-don, pp 189±206

Hill RET, Barnes SJ, Gole MJ, Dowling SE (1990) Physical vol-canology of komatiites, Excursion Guide Book 1, GeologicalSociety of Australia (W.A. Division), East Perth, WesternAustralia

Hill RET, Barnes SJ, Gole MJ, Dowling SE (1995) The volcano-logy of komatiites as deduced from ®eld relationships in theNorseman-Wiluna greenstone belt, Western Australia. Lithos34: 159±188

Huppert HE, Sparks RSJ (1985) Komatiites I: eruption and ¯ow.J Petrol 26: 694±725

Irvine TN (1975) Crystallization sequences of the Muskox intru-sion and other layered intrusions ± II. Origin of chromititelayers and similar deposits of other magmatic ores. GeochimCosmochim Acta 39: 991±1020

Lambert DD, Foster JG, Frick LR, Li C, Naldrett AJ, MacKenzieC (1997) Re-Os systematics of the Voisey's Bay Ni-Cu-Comagmatic ore system, Labrador, Canada, Vol Abstr, AmericanGeophysical Union Fall Meeting, San Francisco, December1997

Lesher CM (1989) Komatiite-associated nickel sul®de deposits. In:Whitney JA, Naldrett AJ (eds) Ore deposition associated withmagmas. Reviews in Economic Geology, Society of EconomicGeologists, pp 45±102

Lesher CM, Groves DI (1986) Controls on the formation of ko-matiite-associated nickel-copper sul®de deposits. In: FreidrichGH, Genkin AD, Naldrett AJ, Ridge JD, Sillitoe RH, VokesFM (eds) Geology and metallogeny of copper deposits.Springer-Verlag, Heidelberg Berlin, pp 43±62

Li C, Naldrett AJ (1999) Geology and olivine stratigraphy of theVoisey Bay complex: evidence for multiple intrusion. Lithos (inpress)

Li C, Naldrett AJ (1997) The Voisey's Bay Ni-Cu-Co deposit,Labrador, Canada: variations of olivine composition related tomultiple magma injections and sul®de segregation. In: PapunenH (ed) Mineral deposits: research and exploration where dothey meet? Proc Fourth Biennial SGA Meeting, Turku Finland,August 1997, Balkema, Rotterdam, pp 461±462

Li C, Naldrett AJ (1993) Sul®de capacity of magma: a quantitativemodel and its application to the formation of sul®de ores atSudbury. Econ Geol 88: 1253±1260

Lightfoot PC, Naldrett AJ (1994) Proceedings of the Sudbury ±Noril'sk Symposium, Ontario Geological Survey spec Vol 5. p423

Lightfoot PC, Naldrett AJ, Gorbachev NS, Doherty W, FederenkoVA (1990) Geochemistry of the Siberian trap of the Noril'skarea, USSR, with implications for the relative contributions of

crust and mantle to ¯ood basalt magmatism. Contrib MineralPetrol 104: 631±644

Lightfoot PC, Hawkesworth CJ, Hergt J, Naldrett AJ, GorbachevNS, Fedorenko VA, Doherty W (1993) Remobilisation ofcontinental lithosphere by mantle plumes: major, trace element,and Sr-, Nd-, and Pb-isotope evidence for picritic and tholeiiticlavas of the Noril'sk district, Siberian trap, Russia. ContribMineral Petrol 114: 171±188

Lightfoot PC, Naldrett AJ, Gorbachev NS, Fedorenko VA,Hawkesworth CJ, Doherty W (1994), Chemostratigraphy ofSiberian trap lavas, Noril'sk district, Russia: implications andsource of ¯ood basalt magmas and their associated Ni-Cumineralization. Proc Sudbury-Noril'sk Symp. Ontario Geolog-ical Survey Spec Vol 5: 283±312

Likhachev AP (1994) Ore-bearing intrusions of the Noril'sk region.Proc Sudbury-Noril'sk Symp. Ontario Geological Survey SpecPubl 5: 185±202

Listerud WH, Meineke DG (1977) Mineral resources of a portionof the Duluth Complex and adjacent rocks in St. Louis andLake counties, northeastern Minnesota. Minnesota Depart-ment of Natural Resources, Division of Minerals, MineralsExploration Section, Report 93

Mainwaring PR, Naldrett AJ (1977) Country-rock assimilationand the genesis of Cu-No sul®des in the Waterhen intrusion,Duluth Complex, Minnesota. Econ Geol 72: 1269±1284

Martineau MP (1989) Empirically derived controls on Cu-Nimineralisation: a comparison between fertile and barren ga-bbros in the Duluth complex, Minnesota, USA. In: PrendergastMD, Jones MJ (eds) Magmatic sul®des ± the Zimbabwe vol-ume. Special Publication, Institution Mining and Metallurgy,London, pp 117±138

McNaughton NJ, Frost KM, Groves DI (1988) Ground meltingand ocellar komatiites: a lead isotopic study at Kambalda,Western Australia. Geol Mag 125: 285±295

McQueen KG (1981) Volcanic-associated nickel deposits fromaround the Widgiemooltha dome, Western Australia. EconGeol 76: 1417±1443

Naldrett AJ (1994) The Sudbury-Noril'sk Symposium, an over-view. Ontario Geological Survey Spec Publ 5: 3±10

Naldrett AJ (1999) Summary: the development of ideas on Sudburygeology, 1992±1998. In: Dressler B, Grieve R, and Sharpton VL(eds) Large meteorite impacts and planetary evolution. Geo-logical Society of America, Memoir (in press, accepted August1998)

Naldrett AJ, Turner AR (1977) The geology and petrogenesis of agreenstone belt and related nickel sul®de mineralization atYakabindie, Western Australia. Precamb Res 5: 43±103

Naldrett AJ Barnes, S-J (1986) The behavior of platinum groupelements during fractional crystallization and partial meltingwith special reference to the composition of magmatic sul®deores. Forschritte der Min 64: 113±133

Naldrett AJ, Rao BV, Evensen NM (1986) Contamination atSudbury and its role in ore formation. In: Gallagher MJ, IxerRA, Neary CR, Pritchard HM, (eds) Metallogeny of basic andultrabasic rocks. Spec Publ Inst Mining Metal, London, pp 75±92

Naldrett AJ, Brugmann GE, Wilson AH (1990) Models for theConcentration of PGE in Layered Intrusions. Can Mineral 28:389±408

Naldrett AJ, Lightfoot PC, Fedorenko VA, Gorbachev NS, Do-herty W (1992) Geology and geochemistry of intrusions and¯ood basalts of the Noril'sk region, USSR, with implicationsfor the origin of the Ni-Cu ores. Econ Geol 87: 975±1004

Naldrett AJ, Fedorenko VA, Lightfoot PC, Kunilov VA, Gorb-achev NS, Doherty W, Johan J (1995) Ni-Cu-PGE deposits ofthe Noril'sk region, Siberia: their formation in conduits for¯ood basalt volcanism. Trans Inst Mining Metal 104: B18-B36

Naldrett AJ, Fedorenko VA, Asif M, Lin S, Kunilov VI, StekhinAI, Lightfoot PC Gorbachev NS (1996a) Controls on thecomposition of Ni-Cu sul®de deposits as illustrated by those atNoril'sk, Siberia. Econ Geol 91: 751±773

239

Naldrett AJ, Keats H, Sparkes K, Moore R, MacKenzie C (1997)The Voisey's Bay Ni-Cu-Co deposit, Labrador, Canada: im-plications for exploration in: International Exploration. In-stitution of Mining and Metallorgy

Naldrett AJ, Keats H, Sparkes K, Moore R (1996b) Geology of theVoisey's Bay Ni-Cu-Co deposit, Labrador, Canada. ExplorMining Geol J 5: 169±179

Naldrett AJ, Fedorenko VA, Asif M, Lin S, Kunilov VI, StekhinAI, Lightfoot PC, Gorbachev NS (1996) Controls on thecomposition of Ni-Cu sul®de deposits as illustrated by those atNoril'sk, Siberia. Econ Geol 91, 751±773

Phinney WC (1972) Northwestern part of the Duluth complex. In:Sims PD, Morey GB (eds) Geology of Minnesota: a centennialvolume: St.Paul, Minnesota Geological Survey, pp 335±345

Phinney WC (1970) Chemical relations between Keewanawan lavasand the Duluth complex, Minnesota. Bull Geol Soc Am 81:2487±2496

Ripley EM (1981) Sulfur isotopic abundances of the Dunka RoadCu-Ni deposit, Duluth Complex, Minnesota. Econ Geol 76:619±620

Ripley EM (1986) Applications of stable isotope studies to prob-lems of magmatic sul®de ore genesis with special reference tothe Duluth Complex, Minnesota. In: Friedrich GH, GenkinAD, Naldrett AJ, Ridge JD, Sillitoe RH, Vokes FM (eds)Geology and metallogeny of copper deposits. Society for Ge-ology Applied to Ore Deposits Spec Publ 4. Springer-Verlag,Berlin Heidelberg, pp 25±42

Ripley EM, Park Y-R, Li C, Naldrett AJ (1997) Sulfur and oxygenisotopic studies of the Voisey's Bay Ni-Cu-Co deposit, Labra-dor, Canada, Vol Abstr, American Geophysical Union FallMeeting, San Francisco, December 1997

Ross JR, Hopkins GMF (1975) The nickel sul®de deposits ofKambalda, Western Australia. In: Knight C (ed), Economicgeology of Australia and Papua-New Guinea. Australian In-stitution of Mining and Metallurgy Monograph 5, (1) Metals,pp 100±121

Ryan B (1997) The Voisey's Bay nickel-copper-cobalt deposit,Labrador, Canada: magmatic sulphide in anorogenic troctoliticrocks. Chron Rech Miniere 525: 3±11

Ryan B, Wardle RJ, Gower CF, Nunn GAG (1995) Nickel-coppersulphide mineralisation in Labrador: the Voisey Bay discovery

and its exploration implications, Current Research, 95-1,Geological Survey, Department of Natural Resources, Gov-ernment of Newfoundland and Labrador, 177±204

Severson MJ (1991) Geology, mineralisation, and geostatistics ofthe Minnamax/Babbitt Cu-Ni deposit (Local Boy area), Min-nesota, Part I: Geology Techn Rep NRRI/TR-91/13a NaturalResources Research Institute, University of Minnesota, Duluth,96 pp

Severson MJ (1988) Geology and structure of a portion of thePartridge River intrusion: a progress report. (Technical ReportNRRI/GMIN-TR-88-08) Natural Resources Research Insti-tute, University of Minnesota, Duluth, 78 pp

Severson MJ, Hauck S (1990) Geology, geochemistry and stratig-raphy of a portion of the Partridge River intrusion: a progressreport. Techn Rep NRRI/GMIN-TR-89-11 Natural ResourcesResearch Institute, University of Minnesota, Duluth, 230 pp

Tang Z (1993) Genetic models of the Jinchuan nickel-copper de-posit, In: Kirkham RV, Sinclair WD, Thorpe RI, Duke JM(eds) Mineral deposit modeling, Geological Association ofCanada Spec Pap 40: 389±401

Tang Z et al. (1992) Sm-Nd dating of the Jinchuan ultrama®c rockbody. Chinese Sci Bull 37: 9

Van Kranendonk M (1996) Tectonic evolution of the Paleo-pro-terozoic Torngat orogen: evidence from pressure-temperature-time deformation paths in the North River map area. Tectonics15: 843±869

Weiblen PW, Morey GB (1980) A summary of the stratigraphy,petrology and structure of the Duluth complex. Am J Sci 280-A: 88±133

Weiblen PW, Morey GB (1975) The Duluth complex ± a petro-graphic and tectonic summary: 36th Ann Minn Min SympDepartment of Conferences and Continuing Education, Uni-versity of Minnesota, pp 72±95

Wooden JL, Czamanske GK, Fedorenko VA, Arndt NT, ChauvelC, Bouse RM, King B-SW, Knight RJ, Siems DF (1993) Iso-topic and trace element constraints on mantle and crustalcontributions to characterization of Siberian continental ¯oodbasalts, Noril'sk area, Siberia. Geochim Cosmochim Acta 57:3677±3704

240