Canadian MineralogistVol. 30, pp.289-301 (1992)

ORIGIN OF PLATINUM-GROUP-MINERAL NUGGETS INFERREDFROM AN OSMIUM-ISOTOPE STUDY

KEIKO HATTORIOttawa-Carleton Geoscience Centre and Department of Geology, University of Ottawa, Ottawa, Ontario KIN 6N5

LOUIS J. CABRICanada Centrefor Mineral and Energy Technology, 555 Booth St/eet, Ottawo, Ontario KIA OGI

ABSTRACT

Osmium-isotope ratios of individual PGM (platinum-group minerals) were determined lr sl/a using an ionmicroprobe. All samples except for one came from placers associated with Alpine- or Alaskan-type ultramaficintrusions. Most Alpine+ype samples are Os-Ir-Ru alloys, and most Alaskan-type samples are Pt-Fe alloys(isoferroplatinum or tetraferroplatinum), which contain Os-lr-Ru alloys and laurite/erlichmanite. Osmium-isotopemeasurements for the Pt-Fe alloys were carried out on those Os-bearing inclusions. The Os-isotope values are similarbetween cores and rims of PGM and among different inclusions of different phases within individual nuggets. Noisotopic variations are observed in PGM that show profound chemical zoning. The-lack o^! isotopic heterogeneiry inindividual nuggets precludes a low-temperature origin for rhe nuggets of PGM, 1876t71869, values of all samplesshow a narrow spread, ranging from 0.99 to 1.12, with most of the values between 1.00 and 1.06. These values fallwithin the range of l87Os/l86Os values of the mantle, suggesting essential derivation of platinum-group elements (PGE)from the mantle without a significant contribution of crustal Os. The grains of PGM were formed in intrusions'weathered, eroded, and concentrated in placers by mechanical processes. The Os-isotopic data are consistent with theoccurrence of exsolulion lamellae of PGM, the inclusions of pristine unwealhered olivine, and the lack of "foreign"mineral inclusions, unrelated to ultramafic rocks, within the nuggets. Os-isotopic values for a placer nugget and PGMfrom chromitite in Urals also support the conclusion that the placer nugget was derived from the latter. The irregularshape and nonabraded surfaces of some nuggets may simply reflecr their durable nature and a short distance oftransporl from the eroded ultramafic intrusions.

Keywords; osmium isotopes, placer, platinum nuggets, platinum-group minerals, Alaskan-type, Alpine-type, ultramaficintrusions.

SoMMAIRE

Le rapport des isotopes 187 et 186 de I'osmium dans les min6raux du groupe du platine (MGP) a 6t6 d6termindin sltu au moyen d'une microsonde ionique. Tous les 6chantillons sauf un proviennent de graviers alluvionnairesassoci6s i des massifs ultramafiques alpins ou de type Alaska. La plupart des 6chanrillons de massifs alpins consistentd'alliages de Os-lr-Ru, tandis que la plupart des 6chantillons de massifs de type "Alaska" sont des alliages de Pt-Fe(isoferroplatine ou t6tralerroplatine), qui conriennent aussi des alliages Os-Ir-Ru et laurite ou erlichmanite. Les mesuresisotopiques des 6chantillons d'alliage Pt-Fe ont port6 sur leurs inclusions riches en Os. Le rapport est semblable dela bordure d'un grain de MGP i son coeur, ou parmi les inclusions de diff6rents min6raux d'une seule p6pite. Aucunevariation isotopique a 6t6 d6couverte dans les grains de MGP qui sont zon6s. L'absence de h6t6rog6n6it6 isotopiquedans une p6pite rend impossible une origine des p6pites par m6canisme de faible temp6rature. Les valeurs du rapportl87Os/l86Os ne varient que peu, entre 0.99 et 1.12; la plupart des valeurs se siluent entre 1.00 et 1.06, et correspondenldonc ir des valeurs typiques du manteau. Les 6l6ments du groupe du platine seraient donc d6rivds essentiellement dumanteau, sans contribution importante d'osmium crustal. Les grains de MGP ont 6t6 formes dans des massifs intrusifs,qui ont subi un lessivage et une 6rosion, pour €tre concentr6s dans des alluvions par processus mdcaniques. Lesdonn6es isolopiques concordent avec la pr6sence dans les p6pites de lamelles d'exsolution de MGP, la pr6senced'inclusions d'olivine saine, et l'absence d'inclusions de min6raux x6nocristiques, sans lien avec les roches ultramafiques.Les valeurs du rapport l8r6r7t86gr dans une p€pite et des MGP d'une chromitite provenant de l'Oural confirmentI'hypothbse que la p6pite alluvionnaire a son origine dans la chromitite. La forme xdnomorphe et la surface sans signed'abrasion de certaines p6pites pourraient tout simplement r6sulter de leur tenacit6 e1 de la proximit6 de la sourceultramafique.

(Traduit par la R6daction)

Mots-clds: isotopes d'osmium, min6raux du groupe du platine, pdpites, graviers alluvionnaires, massifs ultramafiques,type alpin, type Alaska.

289

290 THE CANADIAN MINERALOGIST

INTRoDUCTIoN

Alluvial nuggets of platinum-group minerals(PGM) are historically important. Placers in theUSSR and Colombia and, to a lesser extent, inCanada, were the principal source of the PGEthrough to the first quarter of the 20th century,until the discovery of lode deposits in South Africaand Siberia and the increased efficiency of recoveryof the PGE from Ni-Cu sulfide ores in Canada.

The prevailing view on the origin of nuggets ofPGM by the mid 1960s was their derivation fromultramafic rocks by weathering and mechanicalprocesses of transportation. This opinion was based

Frc. I Back-scattered electron image of osmian iridiumnuggel from the Atlin district, British Columbia(sample number 10120-11). The osmian iridium,having average composition of lr7a.60s1.6Pt1.6 (Harris& Cabri 1973), contains a round inclusion ofOs-Ir-S-As (OISA). The PGM showed evidence ofvariable concentrations of Os and Re during theOs-isotope analyses. Grey area partially rimming thenugget is an intergrowth of irarsite and osmian iridium(M-Ia in the diagram). The scale bar represents 500pm. Listed here are 1876t71869, ratios (all t 0.006);the ualues in -parentheses are the correction due to'o'Re for the points shown. Z: 1.081 (14.890), Y: 1.083(8.790), X: 1.081 (16.2s/o), W: 1.083 (9.6V0), V: 1.079(5.9q0), U: 1.078 (8.890), T: 1.083 ( l7.SVo), S: 1.088(18.390), R: 1.082 (9. l t /o), P: 1.081 (17.590).

on the studies of placers in the Urals (Betekhtinl96l), at Tulameen, British Columbia (Rice 1947),and at Joubdo, Ethiopia (Molly 1959). Augustithis(1965) introduced the concept that "nuggets ofnative Pt have been formed in the lateritic soil"capping the Joubdo ultramafic complex. Ottemann& Augustithis (1967) suggested that they wereformed "during the process of early alteration orlateritization". The formation of nuggets wasbelieved to have taken place "by element agglutina-tion under a phase of hydration (low temperature)of the ultrabasic rocks". These ideas were latersupported by Cousins (1973) and expanded byCousins & Kinloch (1976) and Stumpfl (1974) tothe "chemical accretion" hypothesis. More recent-ly, Bowles (1988) proposed crystallization of PGMin placers. Rounded shapes were ascribed toprocesses of chemical accretion, and angular andunabraded textures, to the growth of crystals insediments (e.g., Bowles 1986, Barker & Lamal1989).

Frc. 2. Back-scattered electron image and 1876t71866t

values of iridian osmium nugget from the Atlindistrict, British Columbia (sample number 10120-10).The nugget contains chromite (Chr) and laurite (Lr)and is rimmed by irarsite (Ia). The,s-cale bar represents500 pm. Listed here ar" 187os/l86os values (all t0.006) for the points shown. Z: 1.036, Y:.1.040, X:1.043, W: 1.036, V: 1.036.

291ORIGIN OF PCM NUCGETS

Frc. 3. Values o1 1879171869, for an osmian iridiumnugget (Ir5e.6osru.3Pta.t, Harris & Cabri 1973) rimmedby irarsite (Ia) from .the Atlin district, BritishColumbia (sample number 10120-9). The,.scale barrepresents 300 pm. Listed here are l87os/l86os ratios(all t 0.006) forthepoints shown. Z: 1.064, Y: 1.061,X: 1 .066, W: 1 .067, V : 1 .061, U: 1 .064, T : 1 .065, S :1.065, R: 1.060, Q: 1.068.

Mineralogical and petrographic studies support-ing the earlier view, mechanical derivation of PGMnuggets from intrusions, include publications byMertie (1976), Cabri & Harris (1975), Raicevic &Cabri (1976), Cabri et a/. (1981), Ford (1981),Hagen e/ ol, (1990), Johan el al. (1990), Nixon elal. (1990), and Slansky et al. (1991).

Osmium, one of the six PGE, has seven isotopes(l84Os, 1860S, l87os, l88OS, l8eos, l90OS, le2OS). l87OS

is the decay product of r87Re, which has a half-lifeof 4.56 x l0r0 years (Luck & Alldgre 1983). Ther87os/r86os ratio of the bulk Earth was 0.805 at4.55 Ga, and the present ratio is believed to beapproximately 1.04 (Alldgre & Luck 1980). Thevalues of the mantle have been close to those of

Frc. 4. Back-scattered electron image and 1879r7t866,

values for an iridian osmium nugget of placer originfrom Teshio, Japan (sample number = Te-l). Thegrain showed the highest Re peaks during theOs-isotope analyses, but the Re contents were foundto be below the detection limit of an electronmicroprobe,

292 THE CANADIAN MINERALOGIST

Flc. 5. Back-scattered electron image of isoferroplatinum nugget from Tulameen, containing chromite (Chr), olivine(Ol; Foe3-e5; Nixon et al, 1990), laurite (Lr) and iridian osmium inclusions (sample number C-236). Several thinlamellae of iridian osmium (Ir21.7Os7a.7Ru3.e; Nixon et ol. 1990).ue present in the field shown in Figure 5b, butare not apparent in the photo. The dashed area in Figure 5b (scale bar: I mm) is shown in Figure 5a (scale bar:500 pm). The laurite in the field 5u, l87gr7l869s ratios of 1.087 t 0.008. Other Os-isotope measurements fromthis nugget are: iridian osmium: 1.092 ! 0.008, iridian osmium: 1.083 t 0.006, iridian osmium: 1.088 + 0.006,laurite: 1.077 J 0.007.

Sevpls DescnrprroN

The placer PGM associated with Alpine-typeintrusions used in this study come from Atlin,Ruby, Bullion and Cariboo (British Columbia),Adamsfield (Tasmania), and Teshio and Onnebetsu(Hokkaido, Japan). Most Alpine-type samplesconsist of Os-Ir-Ru alloys. These alloy sampleswere used for the Os-isotope analyses. The nuggetsvary in shape from rounded (Figs. 1, 2) to angular(Figs. 3, 4). They commonly show chemicalheterogeneity and zonation, with variable Os andRu contents (Cabri & Harris 1975). Some nuggetsare mixtures of different PGM (Figs. l, 2), andsome are coated by irarsite (Figs. 2, 3, 4) or by amixture of fine-grained irarsite and Os-Ir alloys(Fie. l).

Samples associated with Alaskan-type ultramaficintrusions are from Tulameen (British Columbia),Choco (Colombia), Omutnaya River and theNizhni Tagil dunite massif in the Urals (Russia),Joubdo (western Ethiopia) and Goodnews Bay(Alaska). Most Alaskan-type samples consist of aPt-Fe alloy, either isoferroplatinum or tetrafer-roplatinum. The shape of the nuggets varies fromwell rounded (Figs. 5, 6) to irregular with branches(Fig. 7). The grains of alloy commonly containinclusions and lamellae of Os-Ir alloys (Figs. 5, 6,7, 8), laurite and erlichmanite (Figs. 5, 6, 8). Theseinclusions were used for the Os-isotope study.Chemical zonation, similar to that observed ingrains of Os-Ir-Ru alloys from the Alpine-typesamples, also is observed in the included phases(Fie. 6).

ORIGIN OF PCM NUGGETS 293

METHODOLOGY

Analyticol procedures

Osmium-isotope ratios were determined onAu-coated PGM grains using a Cameca 3f ionmicroprobe at MIT. Masses 185, 186, 187' 188 and189 were determined in order to correct the ratiosof l87Os/t86Os for the contribution of r87Re,

hydrides and mass fractionation. Some grainsdisplaying high Re peaks required Re correctionsof up to 24c/o, i.e., 240/o of mass 187 wascontributed by r87Re (Fig. a). Some showed aheterogeieous distribution of Re. After the correc-tion for Re, calculated l87os/186os values werefound to be consistent within the grain (Figs. l, 4),confirming that the Re correction does notintroduce additional uncertainty into the measure-ments. Replicate analyses of the same grains, madeover a l0-month period, show a reproducibility (2o)of + 0.6q0. The precision of each result thus isgiven as x. 0.60/o or the in-run error, whichever islarger. Detailed operating condilions andreproducibilities of the analyses are described byHattori et al. (1991).

Because significant amounts of Re were notedin several samples during the Os-isotopic analyses,the Re contents were checked on the grains coatedwith carbon by counting the Relo peak for 60 susing the wavelength-dispersion spectrometer on aJEOL 733 electron-microanalyzer system. Theoperating conditions were 20 kV with a beamcurrent of approximately 25 nA. Several grains ofisoferroplatinum and of Os-Ir-Ru alloys weresubjected to the examination of Re contents,including an iridian osmium grain that displays thehighest Re signals recorded in the ion microprobe(Fig. a). The ReIa peak was calibrated on a pureRe metal standard, but it was not detected on anyof the grains during the 60 s counting time.Therefore, Re contents are below the minimumdetection level for the electron microprobe, 0.06wt,qo.

Frc. 6. Back-scattered electron images of isoferroplatinum(PF) nugget with inclusions of chromite (Chr), iridianosmium (IO) and erlichmanite (ER) from GoodnewsBay (sample number GDN-I). Dashed square inFigure 6a (scale bar: I mm) is the area shown in Figure6b (scale bar: 200 pm). Note zonation from an Os-richcore to a Ru-rich rim in erlichmanite. Values of1876171866r (all t 0.@6) are: Z (erlichmanite): l'M6'Y (iridian osmium): 1.034, X (erlichmanite): 1.030, W(erlichmanite): 1.034.

294 THE CANADIAN MINERALOGIST

FIc. 7. Irregularly shaped Pt-Fe alloy from GoodnewsBay containing iridian osmium platelets (light color)(sample number I4NMA-NM4). The scale bar repre-sents I mm. Values of l87os/l86Os. Z (tabular iridianosmium): 1.104 t 0.007, Y (tabular iridian osmium):l.lll I 0.008, X oamellar iridian osmium): l.ll9 t0.008, W(thin rod of iridian osmium): 1.121 t 0.002.

Re-Os isotope systemutics

The l87os/l86Os ratio of mantle-derived rocks(Walker et al. 1989, l99la) and the chondrite values(AllBgre & Luck 1980) define the range ofl87os/l86os values of the mantle during geologicaltime (Fig. 9). Although there are detectablevariations in 187os/l86os ratio in the mantle at agiven time, the differences are small compared withthe variations in crustal rocks (Figs. 9, l0).

Ultramafic rocks have a low Re,/Os ratio (e.9.,Walker et al. 1988), and the change in r87os/r86osratio over geological time is relatively small.However, the t87os/l86os ratio of other crustalrocks increases rapidly owing to generally low Osand high Re contents in the rocks. For example,the l87os/l86os ratio of tholeiitic rocks, with anaverage Re/Os ratio of 35 (Chou e/ a/. 1983), wouldincrease from 1.0 to 3.4 in 100 Ma. and the ratioof a black shale with a Re/Os ratio of 75 (Ravizza& Turekian 1989) would change from 1.0 to 7.7 in100 Ma. Because rocks of the upper crust generallyhave a high value of l87os/l86os, the ratio ofdissolved Os in surface waters also should be high.

Initial 187Os/t86Os ratio of the PGM

The Re contents in the PGM studied are lessthan 0.06 wt.qo: the Re/Os ralio of the PGM.

containing on the order of 6090 Os, should be lessthan 0.001. Thus the r87os/r86os ratio of anOs-Ir-Ru alloy would increase only by 0.0003 in500 Ma after crystallization owing to the decay of187Re. The contribution of radiogenic l87Os,

therefore, is not significant for samples of Os-bear-ing alloy.

The isoferroplatinum nuggets contain inclusionsof Os-Ir alloy. Assuming that an jsoferroplatinumnugget contains less than 0.06 wt.9o Re and 5 wt.9o

Ftc. 8. Photomicrograph of a portion of isoferroplatinumnugget (PF; Pt74Ir2Fe4Cul) containing iridian os-mium (IO; Os66lr4Ru3Pt2Rh,) and laurite (Lr)

' inclusions, from the Omutnaya River, Urals (samplenumber 53). The nugget is rimmed by thin layer oftulameenite. The scale bar represents 500 pm. Valueso1 1879171866r. z: 1,030 a 0.012, Y: 1.u21 r 0.011.Os-isotope values of different grains of PGM fromthe nugget include: Os-ricb laurite rod: 1.033 t 0.010,other end of the laurite rod: 1-040 t 0.010, tabulariridian osmium inclusion, B: 1.033 t 0,009, tabulariridian osmium inclusion, C: 1.035 t 0.009, tabulariridian osmium inclusion, D: 1.037 + 0.007, iridianosmium inclusion. E: 1.051 t 0,010.

ORIGIN OF PGM NUGGETS

1 . 0 present

Age (Ga)

Frc. 9. Values o1 1876.71869r for the PGM studied compared with the values ofthe contemporaneous mantle. Solid circles: PGM in this study and data fromPGM from AJpine chromitites from Borneo (Hattori et al. 1992). The range ofmantle l87os/l86Os values are based on values of chondrite (Alldgre & Luck1980), mantle xenoliths (Walker al a/. 1989), and komatiite from Gorgona Island(Walker et al. l99la).

295

1 . 4

1 . 2at,o

@@

u,( J 1 n

@

0.8

2.03.04.0

Os in inclusions, the l87os/l86os ratio of theinclusions would increase by less than 0.003 in 500Ma. If the inclusions are 0.5 wt.9o of the totalmass, the increase would be 0.03. The contributionof the radiogenic r87Os must still be small. Theobserved l87os/l86os values are, therefore, safelyassumed to be the initial values at the time of theirformation.

RESULTS

The r87os/r86os ratio of all samples falls in therange from 0.99 to 1.12, with most of the valuesbetween 1.00 and 1.06. Several grains that showchemical heterogeneity were examined in detail.Iridian osmium nuggets from the Atlin district(Figs. I , 2, 3) show variable Os contents, but valuesof the Os-isotopic ratio were found to be similarwithin each grain. There is no evidence ofovergrowth of PGM enriched in crustal l87Os norof a trend of l87Os enrichment toward the rim ofPGM grains. The Os-isotope ratio of the core andrim of the grains that are coated by irarsite issimilar within the limits of precision (Figs. l, 2, 3),indicating that the alteration of the Os-Ir alloy toirarsite was not accompanied by the introductionof radiogenic Os.

Some nuggets contain several PGM, such asPGE alloys of contrasting compositions and

sulfides. The isotopic ratios of all these PGM,independent of mineral species, are similar (Figs.5, 6, 7 , 8). It is remarkable that, in addition, PGMfrom widely separated geographic localities alsohave similar l87os/l86os ratios.

The PGM from the Nizhni Tagil dunite massif,Urals, Russia, have an average l87os/r86os valueof 1.029 t 0.006. The value is identical to theaverage l87os/l86os value of 1.035 a 0.006 for anugget from the Omutnaya River, Sissertj, near .Sverdlosk, Urals, which is located about 180 kmto the south. The nugget was derived from theOmutnaya dunite massif, which belongs to thesame ultramafic belt as the Nizhni Tagil massif(Betekhtin 1961).

The data, together with the description of PGMfrom Joubdo @thiopia), appear in Table l. Therest of the data and description of PGM analyzedare available from the Depository of UnpublishedData, CISTI, National Research Council, Ottawa,Canada KIA 0S2.

DISCUSSION

Os-isotope rotios of PGM nuggets

The observed values of t87os/t86os from placerPGM (0.99 to l'l2) show a narrow spread, andthey are generally lower than the l87os/r86os values

296 THE CANADIAN MINERALOGIST

2.0 4.0 o.u

I

I

III

-

10.0 20.0 30.0l

40.0

1u7os/186os

Frc. 10. Values of 18u6r71869, for the PGM nuggels studied compared with thoseof large mafic intrusions and crustal rocks. The 1876171866r values of the maficintrusions are initial values. The values of sediments and manganese nodulesare present-day measured values. Sources of data: PGM nuggets (this study),present mantle (Walker et al. 1989, l99la, Martin 1991), Bushveld Complex(Hart & Kinloch 1989), Stillwater Complex (Lambert et al. 1989, Martin 1989),Freetown Complex (Hatt ori et a l. I 99 I ), Sudbury Complex (Walker el a/. I 99 I b),manganese nodule (Luck & Turekian 1983, Palmer & Turekian 1986, Esser &Turekian 1988), continental crust (Palmer & Turekian 1986), Paleozoic blackshale (Ravizza & Turekian 1989).

8.0

PGM nuggets

present mantle

Large lntrusionsBushveldStillwaterFreetownSudbury

Mn-nodules

ContinentalCrust

Black Shales

reported for the rocks and grains of PGM fromlarge intrusions (0.88 to 7.55) such as the Bushveld(Hart & Kinloch 1989), Stillwater (e.g., Lambert etal. 1989, Martin 1989), Freetown (Hattori et al.l99l) and Sudbury complexes (Walker et al. l99lb;Fig. l0). The higher values for the large intrusionsare attributed to contamination of the masmas bvcrustal Os.

15. l8:9r7186Os values of the PGM nuggets fallwithin the range of r87Os/l86os values of the mantle(Figs. 9, l0). The low r87os/r86Os values of thenuggets suggest that the PGE in the nuggets werederived from the mantle without a significantcontribution from crustal Os.

Moopr-s FoR THE FoRMATToN or PGM NuGGErs

Formation of PGM by high-temperaturehydrot herma I processes

Oxygen- and hydrogen-isotope studies of manyepizonal intrusions suggest that the emplacementof igneous rocks in shallow levels commonly resultsin large-scale meteoric hydrothermal activity (e.9.,Taylor & Forester 1979). The presence of fluidinclusions in PGE-rich dunite pipes from theBushveld Complex seems to support the involve-ment of a high-temperature (>700oC) fluid forPGE mineralization in large intrusions (e.9.,

ORIGIN OF PGM NUGGETS

TABT,E 1. leo"/r&o" lATIos oF pc![ IN BrrttDUAr, PLAcf,B NUcGEf,s FaoM TsE JouBDoULTMI'AFIC COMPLEK, BIBBIB DISTBIdI, I{XSTERN EISIOPIA

297

Samplo

8M192&247-10

1.022 r .0061.022 a .@6'1.027 r .006

BMlg&247-12

BMl92&247-13

BMlg*247-14

1.027 a .006

'1.023 I .00€1.022 3 .006

'1.034 I .0O8

1.023 a .0071.019 I .0101. t2 . ! .@7

(a) Fouwtog tls rc@clatue of f,auis & Cabtlt (1991),(b) Alalyd€l tlata gtyo h Cablt st al. (1981).

Stumpfl & Rucklidge 1982, Schiffries & Skinner1987). Fluids derived from hydrous magmas wereinvolved in the PGM mineralization in theAlaskan-type intrusions near Fifield, Australia,according to Johan et al. (1989). It is, however,unlikely that the relatively small Alpine- andAlaskan-type ultramafic intrusions generatedmeteoric-hydrothermal activity. Alpine-type in-trusions are believed to be either obducted slices ofupper mantle or cumulates of boninitic magmassolidified in the lower crust (e.9., Serri 1981). Theorigin of Alaskan-type intrusions is controversial,but the dunite bodies are considered to representcumulates formed in the feeder part of arc magmas(Conrad & Kay 1984) or pre-existing Alpine-typeultramafic intrusions (Kelemen & Ghiorso 1986).In either case, there was little opportunity for themto produce hydrothermal activity at an uppercrustal level.

Postulated high-temperature hydrothermal ac-tivity should have altered the primary silicatechemistry, if such fluids were responsible for muchof PGM crystallization. However, pristine silicatemineralogy common in most Alaskan-type in-trusions counters this hypothesis. Furthermore,olivine enclosed in PGM has identical compositionsto the olivine in chromitites. but distinct fromolivine in the rest of intrusions (e.g., Ford 1981,Nixon e/ al. 1990).

In addition, the occurrence of PGM confinedwithin ultramafic bodies and their absence outsideultramafic bodies indicate that the postulatedhydrothermal activity did not take place on a largescale. If the hydrothermal activity involved country

rocks, the mineralization of PGE would have alsooccurred outside the intrusive bodies.

r87os/186os ratios of PGM obtained in this studyalso suggest that the contribution of crustal Os wasinsignificant. Though current data and evidencecannot reject the possible involvement of high-temperature fluids in the Alaskan- and Alpine-typeintrusions, the overall evidence indicates thathydrothermal activity in the case of these intrusionswas not significant in the formation of PGM.

Formqtion of PGM during serpentinization

Platinum and Pd are considered to be moresoluble in hydrothermal fluids compared with therest of PGE (Mountain & Wood 1988). The lackof variation in Pt and Pd contents during theserpentinization of ultramafic rocks (e.9., Ross &Keays 1979, Oshin & Crocket 1982) suggests thatthe less soluble Os would not be mobile duringserpentinization Processes.

Many Alaskan-type intrusions presently displaypristine compositions of the primary minerals,including olivine and clinopyroxene, beneath theweathered zones. Examples include the GoodnewsBay complex (e.g., Southworth & Foley 1986)' theTulameen complex (Findlay 1969)' and severalzoned complexes in the Urals (Betekhtin l96l). Onthe other hand, Alpine-type intrusions have com-monly undergone extensive serpentinization. Ifserpentinization is responsible for the formation ofnugget PGM, one would expect to observe Pt- andPd-bearing nuggets associated with highly serpen-tinized Alpine-type intrusions, because Pt and Pd

fisf€roDlalirum with ir€aulil surta@ with brmch6' 2.5 mm x 1 mm';ontainlio abundarn orisnted irldia Gmlum lmellae,

- '10 pm wide'trd tabuiar iridim osmlum. Rims ot sme iridlm €mium inclwiomare partally replacod by unidentfrod lr-rich phs€)

labular lridhn osmium grain A in lhe nuggottabular lridia cmium grain B in ths nuggeliridian 6mium larnellas

(R-Fs allov wilh very lreqular surtffi, 0.8 mm x 1.3 mm, @ntainingioa snam'd Iridhn cbmium melal. 10/m x "moltm trd rounded lridiil@miuni ind6lon)iridim 6mlum rd lnclusion

(rounded R-Fe alloy, 0.7 mm, @ntajning short rods of mlum,2otm x -6orrm, md l@nd cmium mstalinclBlom)lddlan Grnium rodroundsd lridla Gmium inclBion

(rounded boturoplatnum nqgget @ntaining genkinite,

5}*''%';ff #''k?3lgd.iil'fl H*".F,fl ,n,il,";lf f LTH'j'lridlan osmium rod in th€ c€ntre ol the Pt-F6,iridian osmium rod in tha matgin ot tho nugget Pt-Fs30tm x 70tmirldim Bmlum thin lamellaelridian osmlum rod in the othsr mtrgin ot ihs nugg€tiridim osmium in the @nral part ot ths nugget

298 THE CANADIAN MINERALOGIST

are more soluble than the other PGE. Empiricalobservations, however, suggest that relativelyunserpentinized Alaskan-type intrusions are accom-panied mostly by nuggets of Pt- and Pd-bearingphases, whereas highly serpentinized Alpine-typeultramafic intrusions are associated with Os-Ir-Rualloy nuggets (e.g., Cabri & Naldrett 1984). SomeAlpine-type intrusions are associated with Pt-bear-ing nuggets as well as Os-Ir-Ru phases (Burgath1988), but these cases are not common. Theseobservations appear to suggest that the nugget-sizedPGM were not formed by the hydrothermal activitythat caused serpentinization.

As mentioned earlier, Re,/Os values ofultramafic rocks are low compared with those ofmany other crustal rocks. The values, however,vary within ultramafic rocks and should be verylow in chromitites because of their high Os contents(e.9., Crocket 1981, Table 2). This implies thatsilicates in ultramafic rocks have high Re,/Osvalues. Therefore, radiogenic l87Os should bereleased during the breakdown of silicate minerals,resulting in high r87os/r86os in serpentinizingfluids. If PGM were indeed formed duringserpentinization, high l87os/l86os values should beobserved in some nuggets, but the lack of suchevidence precludes serpentinization as the principalprocess of the formation of PGM.

Tulameenite and some Cu-Pt phases, whichcommonly rim PGM nuggets, are considered tohave been formed during serpentinization(Betekhtin 1961, Nixon et al. 1990, Cabri & Genkinl99l). Owing to the low concentration of Os inthese minerals, Os-isotopic measurements were notpossible during this study. However, these mineralsmay record a l87os/l86os ratio higher than that inthe early-formed PGM.

Formation of PGM at low temperatures duringweothering or sedimentation

This model includes the formation of PGMnuggets by "element agglutination" (Augustithis1965) or "accretion" of fine particles (Cousins1973, Cousins & Kinloch 1976), and their formationby dissolution and precipitation processes (Bowles1986, 1988, Barker & Lamal 1989). The model maybe discounted on two different grounds:mineralogical and textural evidence and geochemi-cal data, including Os-isotope values obtained inthis study.

Mineralogical and textural evidence: Some Pt-Fenuggets contain inclusions of igneous silicateminerals and oxides, exsolution lamellae of otherPGM, and sulfides (Figs. 2, 5, 6,'7 , 8). Olivine andsulfides are not stable during weathering andsedimentary processes. Their presence indicates that

they were protected by the Pt-Fe alloys duringsedimentary processes.

The occurrence of crystallographically orientedlamellae of Os-Ir alloys in Pt-Fe alloys also isdifficult to account for by the postulated formationof PGM at low temperatures. The texture isattributed to exsolution of a once-homogeneousalloy at high temperatures (Cabri et al. 1987,Slansky et al. l99l).

None of the nuggets in our study containinclusions that are foreign to ultramafic rocks. Inaddition, the compositions of the olivine andchromite are typical of the ultramafic source-rocks(e.9., Nixon et al. 1990).If the nuggets were indeedformed by element "agglutination" or accretion offine particles in sediments, it is statisticallyreasonable to assume that the nugget matrix wouldenclose minerals from adjacent lithologies andminerals abundant in placers, such as quartz,zircon, and monazite. The lack of such foreigninclusions also is consistent with the high-tempera-ture formation of the nuggets.

PGM nuggets consist of crystallographically andcompositionally distinct mineral species; they arenot noncrystalline, "amorphous" mixtures ofmetals. The phases observed in the nuggets areisoferroplatinum (cubic Pt3Fe), tetraferroplatinum(tetragonal PtFe), hexagonal osmium, rutheniumand rutheniridosmine. and cubic iridium. Someform euhedral to subhedral crystals. These nuggetsare very diffegent from "minute" "colloform"aggregates of native platinum documented byWagner and Schneiderhcihn (cited in Wagner 1929).The noncrystalline aggregates are believed to havebeen formed during weathering of Pt- andPd-bearing sulfides at low temperatures (Wagner1929).

Geochemicol evidence: Laboratory experimentsand thermodynamic calculations demonstrate thatPGE may be significantly soluble in acidic salinesolutions, even at low temperatures (e.9., Mountain& Wood 1988). Surface waters in ultramaficterranes cannot dissolve a significant amount ofPGE because they usually are alkaline (pH >10)owing to serpentinizing reactions of silicateminerals (e.g., Barnes et al, 1978). Surface run-offmay become acidic upon completion of serpen-tinization. This is not the case, however, for mostAlaskan-type intrusions, because serpentinizationrarely is pervasive.

If "agglutination" were involved in the forma-tion of nuggets of PGM, nuggets would be expectedto show isotopic heterogeneity within individualgrains because they would contain PGM formedfrom PGE derived from different parts of anultramafic intrusion or from different intrusions.

OR]GIN OF PGM NUGGETS 299

Lack of isotopic heterogeneity in individual grainsalso rejects consideration of the "agglutination"hypothesis as the principal process for nuggetformation.

If the nuggets were precipitated in sedimentsfrom dissolved PGE, they would show high andvaried l87os/l86os values. They would have incor-porated 187Os released from decomposing silicateand sulfide minerals of the ultramafic rocks andthe accompanying gabbros and volcanic rocks. Inaddition, dissolved Os in surface waters would notnecessarily have a constant l87os/l86os ratio. If thenuggets grew by overgrowth on pre-existing PGMgrains, the margins should be high in l87Os. Onewould expect a zoned profile in l87Os,/l80s ratio.The lack of a high l87Os content in the rim or inany nuggets precludes the low-temperature growthof nuggets of PGM as the principal process of theirformation.

CoNcLUSIONS

The 187Os/186Os values of PGM within individualnuggets are consistently low and uniform. Somenuggets of PGM show prominent chemical varia-tions, and some consist of different phases, butthey are homogeneous in Os-isotopic composition.Low r87os/l86Os values reject a postulated low-temperature formation of PGM in sedimentaryenvironments. The data indicate that PGE werederived essentially from the mantle without asignificant crustal component. PGM were formedin the intrusions, eroded, transported and mechani-cally concentrated in placers. For eluvial PGM,they remained in the residual soil on ultramaficintrusions during sedimentary processes. The ir-regular surface of some PGM may simply beattributed to their durable nature and the shortdistances of transportation of the grains from thesource.

This conclusion is in agreement with themineralogical and petrological studies carried outby Cabri & Harris (1975) and Ford (1981) fornuggets associated with Alpine-type intrusions, andby Raicevic & Cabri (1976), Cabri et c/. (1981),Hagen et al. (1990), Johar et al. (1990), Nixon elal. (1990) and Slansky et al. (1991) for PGMnuggets associated with Alaskan-type intrusions.

ACKNOWLEDGEMENTS

We thank J.H.G. Laflamme of the CanadaCentre for Mineral and Energy Technology forchecking the detection limit of Re and providingsome of the photos, and S.R. Hart and N. Shimizuof the Woods Hole Oceanographic Instilution fortheir assistance during the Os-isotope analyses ofthe samples at MIT. KH is grateful to Y. Urashima

of Kagoshima University for the samples fromJapan. LJC received samples over a period of manyyears from J.A. Mandarino and R.I. Gait (RoyalOntario Museum), A.J. Criddle (The NaturalHistory Museum, London)' J. Haight (Univ.British Columbia), J.G. Symons (TasmanianDepartment of Mines), A.D. Genkin (Institute forGeology of Ore Deposits, Petrography, Mineralogyand Geochemistry, Moscow), and J. Cochrane ofMallorca, Spain. Careful reading and helpfulcomments by D.C. Harris, B. Ballantyne, R.R.Keays and R.F. Martin are appreciated. Financialassistance of NSERC to KH is acknowledged.

RrrrneNcrs

ALLEcne, C.J. & Lucr, J.-M. (1980): Osmium isotopesas petrogenetic and geological ftacerc. EarthPlanet. Sci. Lett. 4E,148-154.

AucusrrrHrs, S.S. (1965): Mineralogical and geochemi-cal studies of the platiniferrous dunite - birbirite -pyroxenite complex of Yubdo, Birbir, W. Ethiopia.Chem. Erde A, $9-196.

BenrEn, J.C. & Lavel, K. (1989): Offshore extensionof platiniferous bedrock and associated sedimenta-tion of the Goodnews Bay ultramafic complex,Alaska. Marine Mining E, 365-390.

BenNEs, I., O'Nrll, J.R. & TnEscesrs, J.J. (1978):Present-day serpentinization in New Caledonia,Oman and Yugoslavia. Ceochim, Cosmochim.Acta 42, 144-145.

BrrrrurrN, A.G. (1961): Mikroskopische Unter-suchungen an Platinerzen aus dem Ural. NeuesJahrb. Mineral. Abh. 97, l-34.

Bowlrs, J.F.W. (1986): The development of platinum-group minerals in laterites. Econ. Geol. 81,1278-1285.

(1988): Further studies of the development ofplatinum-group minerals in the laterites of theFreetown layered complex, Sierra Leone. /nGeo-Platinum '87 (H.M. Prichard, P.J. Potts,J.F.W. Bowles & S.J. Cribb, eds.). Elsevier, NewYork (273-280).

BuncarH, K.-P. (1988): Platinum-group minerals inophiolitic chromitites and alluvial placer deposits,Meratus-Bobaris area. southeast Kalimantan. 1nGeo-Platinum '87 (H.M. Prichard' P'J. Potts,J.F.W. Bowles & S.J. Cribb, eds.). Elsevier, NewYork (383-403).

Casnr, L.J., CntooLE, A.J., Lerlavrr le, J.H.G.,BEanNa, G.S. & Hennts, D.C. (1981): Mineralogi-cal study of complex Pt-Fe nuggets from Ethiopia.Bull. Mindral. 104, 508-525.

300 THE CANADIAN MINERALOGIST

& GrNnN, A.D. (1991): Re-examination of Pralloys from lode and placer deposits, Urals. Car.Mineral. 29.419-425.

& Hannrs, D.C. (1975): Zoning in Os-Ir alloysand the relation of the geological and tectonicenvironment of the source rocks to the bulk pt:Pt+lr+Os ratio for placers. Can. Mineral. 13,266-274.

& Nalnnrrr, A.J. (1984): The nature of rhedistribution and concentration of platinum-groupelements in various geological environments. Proc.27th Intern. Geol. Congress 10, 17-46.

CHou, Curu-LrN, SHaw, D.M. & Cnocrer, J.H.(1983): Siderophile trace elemenrs in the Earth'soceanic crust and upper mantle. J. Geophys. Res.EE, Suppl., pt. 2, A507-518.

CoNnao, W.K. & Kev, R.W. (1984): Ultramafic andmafic inclusions from Adak Island: crystallizationhistory and implications for the nature of primarymagmas and crustal evolution in the Aleutian arc.J. Petrol.25, 88-125.

CousrNs, C.A. (1973): Notes on the geochemistry ofthe platinum-group elements. Trans. Geol, Soc, S.Afr. 76, 77 -81 .

.- & KrNLocu, E.D. (1976): Some observation ontextures and inclusions in alluvial platinoids. .E'con.Geol. 71, 1377-1398.

CnocKEr, J.H. (1981): Ceochemistry of the platinum-group elements. -fr? Platinum-Group Elements:Mineralogy, Geology, Recovery (L.J. Cabri, ed.).Can. Inst. Min. Metall., Spec. Vol. 23,47-64.

Essen, B.K. & TunrrraN, K.K. (1988): Accretion rateof extraterrestrial particles determined from os-mium isotope systematics of Pacific pelagic clayand manganese nodules. Geochim. Cosmochim,Acta 52 .1383-1388.

FrnoLay, D.C. (1969): Origin of the Tulameenultramafic gabbro complex, southern BritishColumbia. Can. J. Earth $ci.6.399-425.

Fono, R.J. (1981): Platinum-group minerals in Tas-mania, Econ. Ceol. 76, 498-504,

HecEN, D., Wnsrn, T. & Hrev, T. (1990): Platinum-group minerals in Quaternary gold placers in theupper Chindwin area of northern Burma. Mineral,Petrol. 42,265-286.

Hennrs, D.C. & Caanr, L.J. (1973): The nomenclarureof the natural alloys of osmium, iridium andruthenium based on new compositional data ofalloys from world-wide occurrences. Can, Mineral.t2. t04-112.

group-element alloys: review and revision. Car.Mineral. 29,231-237.

Henr, S.R. & KrNr-ocu, E.D. (1989): Osmium isotopesystematics in Witwatersrand and Bushveld oredeposits. Econ. Geol. 84, l65l-1655.

Herronr, K., BuncnrH, K.-P. & Hanr, S.R. (1992):Os-isotope study of platinum-group minerals inchromitites in Alpine-type ultramafic intrusionsand the associated placers in Borneo. Mineral.Mog. 56,157-164.

Cennr, L.J. & Hanr, S.R. (1991): Osmiumisotope ratios of PGM nuggets associated withFreetown layered complex, Sierra Leone, and theirorigin. Contrib. Mineral. Petrol. 109, 70-18.

JoHaN,2., OuNsNsrrrrEn, M., FrscHrn, W. &Alrossr,J. (1990): Platinum-group minerals from theDurance river alluvium, France. Mineral. Petrol.42,287-306.

SLaNsrv, E., Bannox, L.M. & Supprl,D. (1989): Platinum mineralization in the Alaskan-type intrusive complexes near Fifield, New SouthWales, Australia. l. Platinum-group mnerals inclinopyroxenites of the Kelvin Grove Prospect,Owendale intrusion. Mineral. Pet rol. 40. 289-309.

KrlrvrN, P.B. & Guronso, M.S. (1986): Assimilat ionof peridotite in zoned calc-alkaline plutonic com-plexes; evidence from the Big Jim complex,Washington Cascades. Contrib. Minerol. Petrol.94, t2-28.

Lavaenr, D.D., MoncaN, J.W., War-ren, R.J.,Surnrv, S.B., CnnlsoN, R.W., ZrENrEr, M.L. &Kosxr, M.S. (1989): Rhenium-osmium andsamarium-neodymium isotopic systematics of theStillwater complex. Science 244, 1169-1174.

Lucr, J.-M. & Ar-r-Ecnr, C.J. (1983): 1879.-t86gt

systematics in meteorites and cosmochemical con-sequences. Nature 302, 130-132.

& TuneueN, K.K. (1983): Osmium-187,/os-mium-186 in manganese nodules and theCrelaceous-Tertiary boundary. Science 222, 613-6 1 5 .

ManrrN, C.E. (1989): Re-Os isotopic investigation ofthe Stillwater Complex, Montana. Earth Planet.Sci. Lett.93.336-344.

(1991): Osmium isotopic characteristics ofmantle-derived rocks, Geochim. Cosmochim. Act a55, r42t-1434.

Mrnrrr, J.8., Jn. (1976): Platinum deposits in theGoodnews Bay District. U,S, Geol. Surv,, Prof.Pap. 93E.& .- (1991): Nomenclature of platinum-

ORIGIN OF PCM NUOGETS 30r

Mollv, E.W. (1959): Plat inum deposits in Ethiopia.Econ. Geol. 54, 467-477.

MouNrerN, B.W. & Wooo, S.A. (1988): Chemicalcontrols on the solubility, transport and depositionof platinum and palladium in hydrothermalsolutions; a thermodynamic approach. Econ, Geol.83 ,492-510.

NrxoN, G.T., Cannr, L.J. & Larlavue, J.H.G. (1990):Platinum-group-element mineralization in lode andplacer deposits associated with the TulameenAlaskan-type complex, British Columbia. Can.Mineral. 28. 503-535.

OsHrN, I .O. & Cnocrrr, J.H. (1982): Noble metals inThetford Mines ophiolites, Quebec, Canada. I.Distribution of gold, iridium, platinum andpalladium in the ultramafic and gabbroic rocks.Econ. Geol. 77. 1556-1570.

OrreueNN, J. & Aucusrrrurs, S.S (1967):Geochemistry and origin of "platinum-nuggets" inlateritic covers from ultrabasic rocks and birbiritesof W. Ethiopia. Miner. Deposita 1,269-277.

Peluen, M.R., FelrNrn, K.K., Tun*reN, K.K. &CelvEnr, S.E. (1988): Sources of osmium isotopesin manganese nodules. Geochim. Cosmochim.Acta 52. l l97-1202.

& TunerreN, K.K. (1986): 1879171869, inmarine manganese nodules and the constraints onthe crustal geochemistries of rhenium and osmium.Nature 319.216-220.

RarcEvrc, D. & Cennr, L.J. (1976): Mineralogy andconcentration of Au- and Pt-bearing placers fromthe Tulameen River area in British Columbia. Car.Inst. Min. Metall. Bull. 69(770), 111-119.

Rnvrzza, G. & TunxreN, K.K. (1989): Application ofthe 187p. - 1879r system to black shalegeochronometry. Geochim, Cosmochim. Acta 53,32s7-3262.

Rrce, H.M.A. (1947): Geology and mineral depositsof the Princeton map-area, British Columbia.Geol, Surv, Can., Mem, ?/'3.

Ross, J.R. & Krnvs, R.R. (1979): Precious metals involcanic-type nickel sulfide deposits in WesternAustralia. I. Relationship with the composition ofthe ores and their host rocks. Can, Minerol. 17,4t7 -435.

Scurrr-nres, C.M. & SxtrNrn, B.J. (1987): TheBushveld hydrothermal system: field and petrologicevidence. Am. J. Sci. 287. 566-595.

Srnnr, G. (1981): The petrochemistry of ophiolitegabbroic complexes: a key for the classification of

ophiolites into low-Ti and high-Ti types. EarthPlanet. sci. Lett. 52. 203-212.

SlaNsrv, E., Jouar, 2., OuNrNsrerren, M., BannoN,L.M. & SueeEL, D. (1991): Platinum mineralizationin the Alaskan-type intrusive complex near Fifield,N.S.W., Australia. 2. Platinum-group minerals inplacer deposits at Fifield. Mineral. Petrol. 43,1 6 1 - 1 8 0 .

Sourswonru, D.D. & For-Ev, J.Y. (1986): Lodeplatinum-group metals potential of the GoodnewsBay ultramafic complex, Alaska. U.S. Bur. Mines,Open-File Rep. 5l-E6.

Sruuppl, E.F. (1974): The genesis of plat inumdeposits: further thoughts, Minerals. Sci. Engng.6 .120- t4 t .

& Rucrlrocr, J.C. (1982): The platiniferousEcon. Geol.dunite pipes of the Eastern Bushveld,

77. 1419-t43t.

Tevlon, H.P., Jn. & Fonrsrrn, R.W. (1979): Anoxygen and hydrogen isotope study of the Skaer-gaard Intrusion and its country rocks: a descriptionof a 55 m.y. old fossil hydrothermal system' -/.Petrol. 20.355-419.

WacNrn, P.A. (1929): The Platinum Deposits andMines oJ South Africa (2nd imp., 1973). C. Struik(Pty) Ltd., Cape Town, South Africa.

Welren, R.J., CenlsoN, R.W., Sutnri ' , S.B. & Bovo,F.R. (1989): Os, Sr, Nd and Pb isotope systematicsof southern African peridotite xenoliths; implica-tions for the chemical evolution of subcontinentalmantle. Geochim. Cosmochim. Acta 53, 1583-1 595.

EcHrvrnnfa, L.M., Snrnrv, S.B. & Hon,rN,M.F. ( l99la): Re-Os isotopic constraints on theorigin of volcanic rocks, Gorgona Island, Colom-bia: Os isotopic evidence for ancient heterogeneitiesin the mantle. Contrib. Mineral. Petrol. 107,150-t62.

MoRcAN, J.W., Nelonrrr, A.J., Lr Cuusr &Fnssrrr, J.D. (1991b): Re-Os isotope systematicsof Ni-Cu sulfide ores, Sudbury igneous complex,Ontario: evidence for a major crustal component.Earth Planet. Sci. Lett. 105.416-429.

Surnev, S.B. & Srecuan, O. (1988): Compara-tive Re-Os, Sm-Nd and Rb-Sr isotope and traceelement systematics for Archaean komatiite flowsfrom Munro Township, Abitibi belt, Ontario:Earth Planet. Sci. Lett.87, l-12.

Received June 21, 1991, revised manuscript acceptedNovember 20, 1991.