American Mineralogist, Volume 76, pages 1363-1372, 1991

Nickel-copper sulfides from the 1959 eruption of Kilauea Volcano, Hawaii:Contrasting compositions and phase relations in eruption pumice and

Kilauea Iki lava lake

Wrr,r,raryr E. SroNnDepartment of Geology, McMaster University, Hamilton, Ontario L8S 4Ml, Canada

Mrcn,lBl E. Fr-rrrDepartment of Geology, University of Western Ontario, London, Ontario N6A 5B7, Canada

AssrRAcr

Iron-nickel-copper sulfides in picritic eruption pumice from the 1959 eruption of Ki-lauea Volcano, Hawaii, and associated Kilauea Iki lava lake have been studied in detail.The sulfides occur as micrometer-sized subspherical blebs in glass inclusions within olivinephenocrysts and in glass within quenched groundmass. The proportion of blebs presentwithin the volume of a thin section and within individual olivine grains is widely variable.Sulfide minerals present in the eruption pumice include monosulfide solid solution (mss),isocubanite, pentlandite, bornite, and an unidentified metal-excess iron-rich sulfide. Bulksulfides in the eruption pumice are enriched in Ni relative to Cu. These minerals are alsopresent in Kilauea Iki lava lake, along with chalcopyrite, but the bulk sulfides are markedlyenriched in Cu relative to Ni.

In the eruption pumice, the high-temperature history of the sulfide liquid precursor tothe blebs involved crystallization entirely to mss and subsequent exsolution of isocubaniteon cooling or crystallization of mss and formation of residual Cu-rich sulfide liquid. InKilauea Iki lava lake, the high-temperature history of the sulfide liquid precursor to theolivine-hosted blebs involved crystallization and subsequent unmixing of isocubanite,whereas that of the sulfide liquid precursor to groundmass glass-hosted blebs involvedexsolution ofbornite solid solution from cubic chalcopyrite + isocubanite and chalcopyrite+ isocubanite from bornite solid solution. The presence ofpentlandite + chalcopyrite andpentlandite + bornite solid solution suggests that the blebs have equilibrated internallybelow approximately 600 "C.

The Ni-rich bulk sulfide composition in the pumice reflects the higher temperature andrate of quenching. The Cu-rich bulk sulfide composition in Kilauea Iki reflects a lowertemperature of quenching and more extensive annealing and crystallization history of thelava. The wide variability in bulk sulfide composition in the samples and lack of equili-bration with coexisting olivine are inconsistent with an origin solely by early magmatic,sulfide liquid immiscibility. Instead, bulk sulfide compositions reflect late magmatic pro-cesses. Blebs in the eruption pumice and Kilauea Iki lava largely represent cooling- andquench-induced segregation of sulfides with metals scavenged from the silicate liquid.

fNrnonucrroN

The origin ofnickel-copper sulfide ore bodies has gen-erally been interpreted as caused by either sulfide liquidimmiscibility in silicate magma during the early mag-matic stage (Naldrett, 1989; Lesher and Groves, 1986) orsulfidation of the host rocks during the late magmaticstage or high-temperature subsolidus stage (Fleet, 1979;Fleet and MacRae, 1983; Stone et al., 1989). Proponentsof sulfide liquid immiscibility cite the associarion of nick-el-copper sulfides with mafic and ultramafic rocks andbulk sulfide compositions in the mss phase field withinthe Cu-Fe-Ni-S system. Alternatively, those favoring sul-fidation note that the bulk compositions of the sulfides

are too variable for equilibration under magmatic con-ditions (Fleet, I 978, I 979; Fleet and MacRae, I 983, I 987,1988) and that in unmetamorphosed to low-grade meta-morphosed rocks, the NiO content of olivine is com-monly too high and the NiS content of the associatedsulfides is too low, relative to results of laboratory par-titioning studies, for equilibration during magmatism(Fleet et al.,1977; Fleet, 1979; Fleet and MacRae, 1983;Fleet and Stone, 1990).

Iron-nickel-copper sulfide blebs have been described inocean floor basalt (Czamanske and Moore, 1977), oceanisland basalt (Skinner and Peck, 1969), olivine mega-crysts from Mount Shasta, California (Stone et al., 1989),hydrothermally altered sediment (Missack et al., 1989),

0003404X/9 l /0708-1 363$02.00 l 363

t364 STONE AND FLEET: SULFIDES FROM KILAUEA VOLCANO

Trer-e 1. Composition of glass in groundmass and inclusions (wt%)

Eruption pumice Kilauea lki lava lake

Groundmass Inclusion Groundmass Inclusion

8o5'

sio,Tio,Alr03CrrO"FeOMnOMgoCaoNaroKrONioCuOI

Total

49.902.5'l

12.320.05

11.240.149 9 6

11.231.670.440.020.010.00

99.50

49.722.76

13.730.00

11 .940.136.828.542.012.840 0 00.030.18

98.99

JC.OC

1.6410.810.07J . I I

0 . 1 65.24

13.540.940.010.000.000.1498.17

49.252.66

12.810.06

10.830.245.64

13.271 . 1 50.810.010.020.15

97.14

61.682 . 1 9

13.630.058.460.092.244.670.961.720.000.000.00

95.69

55.573.63

13.210.009.850.133.817.081.471.570.000.050.00

96.37

52.462.77

13.930.00

10.640.154.47

10.481.330.500.000.020.07

96.93

52.351.70

13.540.03

10.740.154.91

11.241.330.300.010.000.10

96.55

. Sample Kl79-3-163.7.'* Sample Kl81-1-306.7.

xenoliths and megacrysts in kimberlite (Bishop et al.,1978; Tsai et al., 19791, Boctor and Boyd, 1980), alkalibasalt (Lorand and Conquere, 1983), and continental ba-salt (Andersen et al., 1987). These blebs have analogousmineral assemblages and equivalent host rock associa-tions to nickel-copper sulfide ore bodies and may haveformed by similar processes. The petrogenesis of sulfrdeblebs is, therefore, significant because ofthe potential forinsight into the origin and cooling history ofnickel-cop-per sulfide ore bodies.

In this paper, the petrography, composition, and phaserelations of iron-nickel-copper sulfide blebs in samples ofpicritic basalts from the 1959 eruption of Kilauea Vol-cano, Hawaii, including quenched eruption pumice andquenched lava from Kilauea Iki lava lake, are docu-mented, and their origin is discussed. These sulfides aredistinctive because the bulk composition of the blebsranged from extremely Ni rich and Cu poor to Ni poorand extremely Cu rich. The blebs are also generally de-void of evidence for low-temperature oxidation (Helz,1987a), and the thermal history of the host rocks is wellconstrained (Helz and Thornber, 1987). Therefore, studyof these blebs may allow comparison with the results oflaboratory experimentation in the Cu-Fe-S and Cu-Fe-Ni-S systems (Yund and Kullerud, 1966; Kullerud et al.,1969; Craig and Kullerud, 1969) and could yield insightinto the efects of cooling rate on bulk sulfide compositionas well as the origin and cooling history ofnickel-coppersulfide ore bodies. The extent of equilibration of theirbulk sulfide compositions with coexisting olivine is dis-cussed by Fleet and Stone (1990).

Sllvrpr,ns AND ANALYSES

The three thin-section samples studied are of picriticpumice and lava from Kilauea Iki lava lake, Hawaii, ef-fused during the 1959 eruption of Kilauea Volcano (Helz,1987a, 1987b). The pumice sample, Ikt-22, quenched on

eruption at 1216 'C. The lava samples,KlT9-3-163.7 andKISl-l-306.7 (hereafter referred to as KI79 and KI8l),are of drill core recovered from olivine-phyric basalt inKilauea Iki. KI79 and KI8l annealed at temperaturesabove 1053 oC and 1083'C for 20 and 22 yr, respectively,before quenching during drilling (Helz, 1987b; Helz andThornber, 1987).

Analyses of the olivine, glass (Table l), and sulfides(Tables 2, 3) were made with a JEOL JXA-8600 auto-mated superprobe at the University of Western Ontario,using natural silicate minerals, synthetic FeS and NiS,and natural chalcopyrite as standards. Analyses ofolivineand glass were made at 15 kV, 10 nA, and with a beamdiameter of 3 pm. The counting time for Ni in olivinewas 50 s and that for Ni, Cu, and S in glass was 100 s.

Bulk sulfide compositions (Table 2) were calculated onthe basis of modal proportion of phases present, as de-termined by point counting, or were determined by WDSelectron microprobe analysis with a focused or slightlydefocused beam at 20 kV, 20 nA cup current, 20 s or20000 counts, and a beam diameter of 2 to l0 pm. Anal-yses of individual sulfide phases were made under theconditions used for bulk sulfide analysis, except that thecup current was 4 nA and the beam diameter was I pm.The sulfide phase compositions (Table 3) represent eithera spot analysis or the average ofup to five spot analyses.Analysis of the sulfide phases was complicated by the finegrain size. Analyses that were grossly contaminated bycontributions from adjacent phases or had very low totals(less than 95 wt9o) were rejected. Standard deviations forthe averaged composition of the phases are less than 0.9wto/o for the elements present. Repeated analyses of thestandards indicate a precision relative to absolute ele-ment concentrations of within I wt0/0.

PErnocRApHY AND cHEMrsrRY

The sulfide blebs occur in the eruption pumice and inKilauea Iki lava as < I to 40 pm subspherical multiphase

STONE AND FLEET: SULFIDES FROM KILAUEA VOLCANO

TABLE 2. Bulk composition of the sulfide blebs

l 365

CuNiFe

Sample Bleb Host (wt%) Assemblage

tki-22tki-22tki-22tki-22tki-22tki-22tki-22tki-22tki-22tki-22tki-22Kt81Kt81Kt81Kt81Kt81Kt81Kt81Kt81Kr81Kr81Kt81Kt81Kt79Kt79Kt79Kt79Kt79Kt79Kr79Kt79Kt79Kt79Kr79Kt79Kt79Kt79Kt79Kt79Kt79Kt79Kr79Kr79Kr79

1234567I9

1 01 11237

1 01 11 21 61 71 81 9211234

b

78q

1 01 11 21 3't41 51 61 71 81 92021

ololololololololglsglsgtsolololololololololololololgtsglsglsglsglsglsglsgtsgtsglsglsglsglsgtsgtsglsglsglsglsgls

46.1545.6142.2037.7837 3237.9534.9137.7129.0154.7034.0637.8947.9138.7036.8440.6446.9941.83252732.8032.1627.0217.8132.2318.5719.2118.6619.6621.2817.2919.8617.6422.6118.5519.5318.5217.6916.9617.0515.4719.5522.5018.0619.12

14.74'14.5215.898.269.45

14.9414.5915.5831.813.40

27.404.447.734.236 . 1 13.678.807.802.905.493.245.560.255.310.060.100.040 . 1 10 . 1 10.200.080.07o.270.060.020.200.150.120.050.030.250.050.080.04

0.851 . 1 54.18

17.6216.4310.4714.3510.335.006.076.07

22.116.43

22.7021.5020.127.08

14.7537.4524.8329.3234.7752.3527.3350.7449.3949.0848.3245.35s2.8849.7551.9844.0050.2248.0651.8751.5452.8050.9657.9347.5545.O750.3849.03

37.1836.7637.2634.9134.8335.5034.9935.5632.0133.7932.7433.8435.4733.6833.6633.7634.6733.4231.4033.2833.8631.3827.4134.7430.6830.5530.7330.4932.0429.1330.6829.9332.1030.3231.1430.1130.2530.5631.2527.5431.6431.7030.5230.84

m s s + i c b + m tm s s + i c b + m tmss + icbmss + icbm s s + i c b + m t ?m s s + i c b + p nmss + pnmss + icbmss + icbUMIS + issp n + i c b + b nicb + pnmss + icbicb + pnicbmss + icbmss + icbmss + icbi c b + p n ? + b n ?icbicb + cpi c b + b n + p nbnc p + p n + p o ?b n + c p + i c b + m tb n + c p + i c b + m tb n t c p + i c b + m tb n + c p + i c b + m tb n + c p + i c b + m tb n + c p + i c b + m tb n + c p + i c b + m tb n + C p + i c b + m tb n + c p + i c b + m tb n + c p + i c b t m tb n + c p + i c b + m tb n J c p + i c b + m tb n + c p + i c b + m tb n + C p + i c b + m tb n + c p + i c b + m tb n + C p + i c b + m tb n + c p t i c b + m tb n a c p + i c b + m tb n + c p + i c b + m tb n + C p + i c b + m t

Note.'Abbreviations: ol, olivine phenocryst; gls, groundmass glass; mss, monosulfide solid solution; icb, isocubanite; pn, penflandite; UMIS, unidentifiedmetal-excess iron-rich sulfide; bn, bornite; cp, chalcopyrite; mt, magnetite.

TABLe 3. Composition of minerals in sulfide blebs

Fe

Sample Bleb Phase (wt7") (at.o/4

CuNiCuNiFe

Host

tki-22tki-22tki-22tki-22tki-22tki-22tki-22Kt81Kr81Kt81Kt81Kt81Kt81Kr81Kt81Kt81

ololgtsglsglsglsgtsololololololololol

46.0745.7956.8142.7634.2134.3725.3339.5250.7038.9938.1733.1725.6630.5131.6818.49

14.7714.523.781.25

32.154.274.051.839.692.28

10.832.804.721.75

1 7 . 1 80.34

o.770.673.56

20.261.28

27.8541.2722.441.09

23.8715.1028.2935.3831.6616.1349.12

37.8136.9233.9532.8432.8332.5530.6335.4935.4733.9334.2334.0533.4433.4031.5127.62

36.3736.5346.2135.9527.7928.7321.3232.1741.3332.1631.4427.642't.4725.8227.0616.80

1 1.0911.702.930.99

24.853.393.241.457.521.798.492.223.761.41

13.960.29

12

1 01 01 11 11 122I'|

1 81 81 91 91 9

mssmssUMISicbpnicbbn + icbicbmssicbpn + icbicbcpicbpn + icbbn

0.53 52.010.47 51.302.55 48.21

14.97 48.090.91 46.45

20.47 47.4'l30.53 44.9116.05 50.330.78 50.37

17.30 48.7510.93 49.1420.72 49.4226.02 48.7523.54 49.2312.11 46.8739.21 43.70

/votej Abbreviations: mss, monosulfide solid solution; UMIS, unidentified metal-excess iron-rich sulfide; icb, isocubanite; pn, penilandite; bn, bornite;cp, chalcopyrite; ol, olivine phenocryst; gls, groundmass glass.

I 366

Fig. l. Sulfides in eruption pumice. (A) Olivine phenocryst-hosted bleb of mss (gray) and lamellar isocubanite (icb, brieh0and magnetite (mt, darD. Polished section, reflected light, oil;scale bar represents 100 pm. (B) Backscattered-electron imageof a pumice glass-hosted bleb of pentlandite (pn, gray), isocu-banite (icb, dark), and bornite On, brieht).

blebs in glass inclusions within olivine phenocrysts andas inclusions in the olivine phenocrysts, groundmass glass,

and segregation veins (Figs. 1, 2). The blebs are very lim-ited in distribution and are present at much less than 0.01modalo/o of a sample. The sulfide minerals observed in-clude monosulfide solid solution (mss), isocubanite (pre-viously intermediate solid solution or iss; Caye et al.,1988), pentlandite, chalcopyrite, bornite, and an uniden-tified metal-rich iron sulfide.

The sulfide-bearing olivine crystals observed in theeruption pumice comprise a loose aggregate of three phe-

nocrysts, the largest of which is 500 pm across. They areForr-r, in composition and contain 0.26 wto/o NiO. Thesulfide blebs in sample KI81 occur only within olivine,whereas those in sample KI79 occur primarily withingroundmass glass. The sulfide-bearing Kilauea Iki lavaolivines are singular phenocrysts more than 1 mm across.The olivines in KI8l are Fo?4-7s in composition and have

STONE AND FLEET: SULFIDES FROM KILAUEA VOLCANO

0.26 vrto/o NiO. Those in KI79 are Fou and have 0.26wto/o NiO. Analyses of the olivine phenocrysts in the Ha-

waiian samples resembl€ those in the literature (Helz,

1987a, 1987b; Helz and Thornber, 1987) except that theNiO content is higher in the present study. The higher

NiO content is supported by repeated analysis of the Ni

standard (P140 olivine) and selected product olivines fromFleet and MacRae (1988). The discrepancy in NiO con-

tent of olivine is discussed further in Fleet and Stone(1e90).

The sulfide-bearing glass inclusions within olivine are

up to 100 pm in diameter. They are subspherical in out-line, commonly contain a chromite grain, and lack crys-

talline silicate phases. Results from analysis of sulfide bleb-

bearing inclusion glass show that composition varies

widely within a single olivino crystal and between the

eruption pumice and Kilauea Iki lava (Table 1). The in-

clusion glass varies from basaltic to andesitic in range of

composition. It is devoid of Ni and Cu to the detectionlimits of the analysis (0.05 and 0.02 wlo/o, respectively)'Inclusion glass in the eruption pumice contains more S(up to 0.2 wto/o S) than that in Kilauea Iki lava (0.1 wto/o

S). Analyses of pumice glass (basalt) and groundmass glass

(andesite to dacite) from Kilauea Iki (Table l) resemblethose of Helz (1987a,1987b).

Bulk sulfides

Although the sulfide blebs are generally multiphase(Figs. l, 2), few are large enough for analysis of the in-

dividual phases. Therefore, they are best characterized in

terms of their bulk composition. The bulk sulfide com-positions from the eruption pumice and Kilauea Iki lavarange from Ni rich and Cu poor to extremely Ni poor

and Cu rich (Fig. 3 ofFleet and Stone, 1990; Fig. 3; Table2). This extremely wide range in bulk sulfide compositionexceeds that reported for sulfide blebs in other Hawaiianlavas (Desborough et al., 1968; Skinner and Peck, 1969).The sulfides in the eruption pumice are more nickelifer-ous and larger, whereas those in Kilauea Iki lava are more

cupriferous and more abundant.The bulk compositions of sulfide in the eruption pum-

ice plot in the mss field or in the two-phase field betweenisocubanite and mss in the Cu-(Fe + Ni)-S system at 600.C (Fig. 3). The bulk composition of these blebs varieswidely, both for blebs within olivine and those within thepumice elass (Table 2). Among the eight olivine-hostedblebs analyzed, the bulk Ni/Cu ratio is 20.0 and 14.3 in

two of the blebs (e.g., Fig. lA) and it ranges from 1.6 to0.5 in the remaining six. The variability among the Ni-rich compositions may be within analytical error. Muchof that among the more Ni-poor compositions may re-flect sampling bias because the phases in some of theseblebs are more coarsely segregated. However, the extremecompositional disparity is real and significant.

The sulfide blebs in the pumice glass are larger thanthose in the olivine (Fig. 1) and they are metal rich. Amongthe three pumice glass-hosted blebs analyzed, the Ni con-tent of two ftlebs 9 and I l, Table 2) exceeds that of the

Fig. 2. Sulfides in quenched Kilauea Iki lava. (A) Backscat-tered-electron image of an olivine phenocryst-hosted bleb of in-tergrown mss (light gray) and isocubanite (icb, dark gray). (B)Backscattered-electron image of an olivine phenocryst-hosted blebofpentlandite (pn, gray), chalcopyrite (cp, dark), and bornite (bn,brigh|. (C) Backscattered-electron image of an olivine pheno-cryst-hosted bleb of isocubanite (icb, light gray) and lamellarchalcopyrite (cp, brieht).

t367

Fig. 3. Bulk sulfide compositions (at.o/d. Full symbols areeruption pumice; open symbols are Kilauea Iki lava; circles areolivine phenocryst-hosted blebs; squares are rock matrix glass-hosted blebs. Phase fields for 600'C are from Craig and Scott(1974) and Hil l (1983).

most Ni-rich olivine-hosted blebs. The bulk compositionof the third bleb (bleb 10, Table 2) is Fe rich and richerin Cu than Ni.

The bulk compositions of sulfide blebs from KilaueaIki lava (Fig. 3) span the range from the pyrrhotite-mssphase field to the bornite solid solution field in the systemCu-(Fe + Ni)-S (Fig. 3). The bulk composition of rheolivine-hosted blebs varies widely and is distinct fromthat of blebs in the groundmass. Analysis of the olivine-hosted blebs indicates a range in Ni,uCu ratio from <0.01to 1.3 (Table 2). Most of rhe blebs are Cu rich (Fig. 3)and contain significant amounts of Ni. The bulk com-position of the groundmass-hosted blebs is relatively uni-form: extremely Ni poor and Cu rich, with an averageNi./Cu ratio of <0.01. The bleb with the highest bulk Cucontent in the Hawaiian samples (KI8l bleb 21, 52.4 wto/oCu, Table 2) is actually olivine hosted. Although some ofthe variability in bulk composition of the sulfide blebsmay reflect sampling bias, the extremely wide variabilityis considered to be real and significant because the blebsare primarily very fine grained, sulfide phases are intri-cately intergrown (see below), and a large number ofblebswere analyzed (Table 2).

Sulfide phases

The blebs in the eruption pumice are of mss * isocu-banite + pentlandite + magnetite, pyrrhotite * isocu-banite, or pentlandite * isocubanite + bornite. The pre-dominance of nickel sulfides is consistent with thenickeliferous bulk composition of the sulfides in the erup-tion pumice. The blebs in Kilauea Iki lava are of eitherisocubanite + mss + pentlandite + chalcopyrite or born-ite + chalcopyrite + isocubanils + pentlandite + mag-netite. The predominance of copper sulfides in KilaueaIki lava is distinctive from the eruption pumice and con-sistent with the cupriferous composition of the bulk sul-fides.

In the eruption pumice, the mss + isocubanite blebsare the most common and occur in olivine and the pum-ice glass. They range in size from 5 to 40 pm across. The

STONE AND FLEET: SULFIDES FROM KILAUEA VOLCANO

3565

Aguz r lFe + Ni)

bn p*#'43'+

l 368

A

STONE AND FLEET: SULFIDES FROM KILAUEA VOLCANO

Fig.4. Compositions of sulfide phases within the Cu-(Fe +Ni)-S system (at.o/o). (A) Eruption pumice and (B) Kilauea Ikilava. Dots represent averaged analyses ofindividual phases; fullsymbols represent bulk sulfide compositions in the eruptionpumice; open symbols represent bulk sulfide compositions inKilauea Iki. Dashed tie lines represent coexistence with an iden-tified phase of a composition that was not determined quanti-tatively. Phase fields for 600 qC are from Craig and Scott (1974)

and Hill (1983).

mss comprises from 50 to 90 modalo/o of the blebs and isthe matrix phase. Composition of the mss (Table 3) plotsin the mss(l) phase field (Misra and Reet, 1973; Craig,1973; Stone et al., 1989).

The isocubanite coexisting with the mss in these blebsis present in amounts up to 50 modalo/0, as either lamellae(Fig. 1A) or very tiny grains (0.1 pm). In the latter case,the isocubanite is concentrated in the bleb margins. Pent-landite comprises about 5 modalo/o of some of these blebs.It occurs as small specks within mss or along the contactbetween mss and isocubanite. One mss * isocubanitebleb has 3 modalo/o of very fine-grained (< 1 pm) mag-netite as isolated inclusions, inclusion trails, and lamellae(Fig. lA). A second bleb within the same glass inclusionhas a partial rim of Cr-bearing magnetite separating itfrom a chromite grain. Isocubanite, magnetite, and pent-landite present in the mss + isocubanite blebs are toofine grained for quantitative analysis.

The Fe-rich bleb in the eruption pumice consists of anunidentified metal-excess iron-rich sulfide fuMI$ + iso-

cubanite. The bleb is within the pumice glass and is 30by 20 pm in size. It resembles the mss + isocubaniteblebs with lamellar isocubanite except that the isocuban-ite lamellae are up to I pm wide, more isocubanite ispresent (about 15 modalo/o), and magnetite is absent. TheUMIS is metal rich (M/S : 1.07) and contains 2.9 at.o/oNi and 2.6 at.o/o Cu (Table 3). Although similar in com-position to mackinawite (cf. Clarke,1970), the UMIS isisotropic and therefore quite distinct from that mineral.The composition of the coexisting isocubanite approxi-mates that of cubanite (Fig. aA).

One bleb with pentlandite + isocubanite * bornite wasfound in the eruption pumice (Fig. lB). It occurs in thepumice glass and is 40 pm in diameter. The pentlanditeis the matrix phase and it is Fe rich Oleb I l, Table 3).The isocubanite occurs as irregular interstitial areas with-in the pentlandite. It is metal rich and Cu rich relative tonominal CuFe,S. (Table 3; Fig. 4A). The bornite is pres-ent as very fine-grained (< I pm) spots within the isocu-banite and along boundaries between the isocubanite andthe pentlandite. The very fine grain size of the borniteprecluded quantitative analysis.

In Kilauea Iki lava, the blebs consist of either isocu-banite + mss + pentlandite t chalcopyrite or bornite +

chalcopy'rite + pentlandite. The isocubanite-bearing blebsare most common within olivine. One such bleb consistsof 25 modal0/o isocubanite in lamellar intergrowth withmss (Fig. 2A). The isocubanite is close to stoichiometricCuFerS, (KI81 bleb 2,Table 3;Fig.4B). The compositionof the mss corresponds to mss(l) (Misra and Fleet, 1973;stone et al., 1989).

The blebs ofisocubanite + pentlandite + chalcopyritein Kilauea Iki lava are of isocubanite as matrix and 5 to15 modalo/o lamellar pentlandite or chalcopyrite. Thepentlandite in these blebs is too fine grained (< I pm) forquantitative analysis. In one bleb where isocubanite andlamellar chalcopyrite coexist (Fig. 2B), the former is Curich relative to ideal CuFerS, (KI8l bleb l, Table 3). Thecoexisting chalcopyrite is metal rich and Cu rich relativeto ideal CuFeS, (Table 3; Fig. 4B).

The bornite-bearing blebs occur mostly in the ground-mass glass of sample KI79. These blebs are the smallest(<2 pm) and most abundant in the samples studied. Manyconsist of a core with intricately intergrown lamellarbornite + chalcopyrite + isocubanite and a margin ofmore discrete chalcopyrite + isocubanite. Others havebornite in the margin that is more coarsely intergtownwith discrete isocubanite + chalcopyrite in the core. Blebsentirely of bornite are apparently present. The small sizeof the bornite-bearing blebs precluded quantitative anal-ysis of the phases. Magnetite is present in the maryin ofa very few of these blebs.

Three bornite-bearing blebs were observed within asmall olivine phenocryst in KI81. One of the blebs con-tains a small core of intergrown isocubanite and pent-

landite and a margin composed partly of discrete isocu-banite and partly ofintricately intergrown isocubanite andbornite (Fie. 2C). The discrete chalcopyrite is S poor rel-

35

35

ative to ideal CuFeS, (bleb 19, Table 3), whereas thebornite appears to be Cu poor and S rich relative to idealCurFeSo (Fig. aB). The pentlandite is too fine grained forquantitative analysis. The deviations from stoichiometryofthe chalcopyrite and the bornite probably reflect beamoverlap onto adjacent sulfides during analysis.

Prrnsn RELATIoNs

The sulfide blebs of the 1959 eruption pumice and Ki-lauea Iki lava are extremely variable in bulk composition.Solidification of the blebs probably occurred between 1200and 850 "C because this interval includes the liquidustemperatures of the three common sulfide minerals thatcrystallize from Fe-Ni-Cu-S liquid: mss (1 19l-l100 .C;

Naldrett et al., 1967; Kullerud et al., 1969), bornite solidsolution (l129-1059 "C; Craig and Kullerud, 1969; Kul-lerud et al., 1969), and isocubanite (970-850 .C; Craigand Kullerud,1969; Craig and Scott, 1974). Cooling ofsulfide liquid to the range of ll92 to 1000 oC results incrystallization of mss, which removes much of the avail-able Ni and up to 5 wto/o Cu in solid solution (Kullerudet al., 1969). Any sulfide liquid residual of mss crystalli-zation would therefore be more Cu rich. However, thesolubility of Cu in mss increases with decreasing temper-ature and may attain a maximum of 7.5 wto/o at 940 "C(Kullerud et al., 1969). Isocubanite exsolves from mssbelow 940'C.

In the eruption pumice and Kilauea Iki lava, the blebsof lamellar isocubanite within mss or UMIS are inter-preted to represent an unmixing during cooling of a high-temperature, more Cu-rich mss precursor. The originalhigh-temperature mss appears to have contained all ofthe Ni and Cu in these blebs. This apparent congruentcrystallization of complex mss is apparent in sulfide blebsdescribed in the literature (Stone et al., 1989; Boctor andBoyd, 1980).

The present of a Cu-rich margin about a more Ni-richcore in most of the olirrine-hosted blebs suggests that inthe eruption pumice and Kilauea Iki lava, Cu-rich sulfideliquid was residual from mss crystallization. Isocubanitecrystallized from the Cu-rich residual liquid between about970 and 850 "C (Craig and Kullerud, 1969). This crys-tallization history has been inferred for many sulfide blebsin the literature (Lorand, 1989a, 1989b; Stone et al., 1989;Stone and Fleet, 1990).

Although the UMIS * isocubanite bleb encountered isdepleted in Ni relative to Cu and is metal rich, the sulfideliquid precursor appears to have initially crystallized en-tirely to high-temperature UMIS. At high temperature,pyrrhotite can accommodate 5 wto/o Ni and 5 wto/o Cu(Yund and Kullerud, 1966; Kullerud et al., 19691, Craigand Kullerud, 1969). Sulfide liquid enriched in Fe rela-tive to Ni and Cu and containing less than l0 wto/o Ni +Cu would crystallize as hexagonal pyrrhotite rather thanmss. Therefore, the metal-enriched bulk composition ofthe UMIS + isocubanite bleb is a curiosity. The metalenrichment is consistent with the composition of theUMIS and isocubanite. The contents of Fe, Ni, and Cu

r369

in the UMIS resemble those of high-temperature pyrrho-tite (Craig and Kullerud, 1969;Crng,1973), but the UMISappears to be isotropic and S poor. Admittedly, the av-eraged analysis of the UMIS totals only to 98.1 wt0/0. Ifthe S content of this analysis were 2 wto/o higher, then thecomposition would plot on the S-poor mss boundary.Metal-rich analyses of mss are reported from MORBsamples (Kanehira et al., 1973; Czamanske and Moore,1977) and, in experimental samples (Misra and Fleet,1973), but these also remain to be explained.

Although the eruption pumice and Kilauea Iki lavasamples were quenched from the range of 1216 to 1053oC, the sulfide phases in the blebs continued to reequili-brate in the sulfide subsolidus. The exsolution of isocu-banite of cubanite composition from the UMIS is con-sistent with the observation that isocubanite inequilibrium with pyrrhotite has a smaller Cu/Fe ratio thanthat in equilibrium with chalcopyrite (Yund and Kulle-rud, 1966). Theoretically, tie lines formed between iso-cubanite and pyrite at 743'C persist during cooling to590 "C and prevent the coexistence of isocubanite andpyrrhotite (Craig and Kullerud, 1969). In the eruptionpumice, the coexistence of isocubanite and UMIS reflectsthe S-poor bulk composition, which precluded pyrite for-matron.

The occurrence of isocubanite in pentlandite within onebleb (Fig. lB) suggests that the former exsolved from ahigh-temperature mss precursor during cooling rather thancrystallized from residual Cu-rich sulfide liquid. The pau-city of mss and pyrrhotite in this bleb suggests that theprecursor mss decomposed entirely to pentlandite andisocubanite rather than to pentlandite and pyrrhotite. Thisis consistent with the Fe-rich composition of the pent-landite (Misra and Fleet, 1973; Stone et al., 1989) andthe isocubanite and with the S-poor bulk composition ofthe bleb. The exsolution of bornite rather than chalco-pyrite from the isocubanite probably reflects the Cu-richand S-poor composition of the latter (e.g., Yund and Kul-lerud, 1966). The coexistence of pentlandite and bornitesolid solution implies that this bleb equilibrated inter-nally during cooling below 590 "C (Craig and Kullerud,1969; Craig and Scott, 1974).

The predominance of Cu over Ni in the bulk compo-sition of the blebs in Kilauea Iki lava suggests that thecrystallization history of these sulfides reflects more thedecomposition of isocubanite than mss. Given that iso-cubanite at 600 oC may contain as much as 15 wto/o Ni(Craig and Kullerud, 1969), the pentlandite in the Cu-rich blebs could have exsolved from isocubanite. Pent-landite in Ni-rich masses of chalcopyrite within spinelperidotite massifs in France has been attributed to exso-lution from Ni-rich isocubanite (Lorand, 1989c). Indeed,the presence ofpentlandite as a discrete phase intergrownwith chalcopyrite in the core of isocubanite-dominatedblebs (e.g., Fig. 2C) and as lamellae in isocubanite sug-gests that the exsolution ofpentlandite from isocubanitecan predate that ofchalcopyrite.

The single bleb with lamellar chalcopyrite in isocuban-

STONE AND FLEET: SULFIDES FROM KILAUEA VOLCANO

1370

ite (Fig. 2C) is interpreted to represent an unmixing ofhigh-temperature isocubanite into chalcopyrite and moreFe-rich isocubanite. Isocubanite crystallizes from sulfideliquid during cooling between 970 and 850 "C (Craig andKullerud, 1969). It subsequently unmixes into cubic chal-copyrite and isocubanite of about CuFerS, compositionat approximately 590 'C (Yund and Kullerud,1966; Ca-bri, 1973). During unmixing, most of the Ni in the iso-cubanite is partitioned into the chalcopyrite, which inturn is exsolved as pentlandite at about 550'C (Craig andKullerud, 1969).

The bulk sulfide compositions of the extremely Ni-poorand Cu-rich bornite-bearing blebs from Kilauea Iki lavaspan the miscibility gap between bornite solid solutionand chalcopyrite or, in one case, isocubanite at 600'C.At temperatues above 840 qC, this miscibilitygap is muchnarrower. During cooling below 840 oC, bornite exsolvesfrom Cu-rich isocubanite, and chalcopyrite with or with-out isocubanite exsolves from S- and Fe-rich bornite sol-id solution (Yund and Kullerud, 19661, Craig and Kulle-rud, 1969; Sugaki et al., 1975). By analogy, thecrystallization history of the extremely Ni-poor and Cu-rich blebs in Kilauea Iki lava involved exsolution of chal-copyrite + isocubanite from bornite solid solution andexsolution ofbornite solid solution from chalcopyrite +isocubanite during cooling below about 840 C.

The lamellar and inclusion trail magnetite in sulfideblebs within the eruption pumice reflects exsolution fromprecursor high-temperature mss, perhaps during theexsolution of isocubanite. The bleb margin of magnetitein the Kilauea eruption pumice is attributable to late-stage oxidation ofthe sulfide.

Sur-rrnn PETRoLocY

Features of the sulfide blebs from Hawaii, such as theirrounded shape and the fact that many of the bulk sulfidecompositions would plot in the mss phase field of the Ni-Fe-S system, appear to be consistent with an early mag-matic origin. However, the extreme variation in com-position of the sulfide blebs (Fig. 2 of Fleet and Stone,1990; Table 2; Fig. 3) is of fundamental significance tohypotheses of their origin.

Variation in the bulk composition of sulfide blebs andnickel-copper sulfide ore bodies has been attributed var-iously to fractionation of immiscible sulfide liquid (Craigand Kullerud,1969', Mackan and Shimazaki, 1976; Ra-jamani and Naldrett, 1978), filter pressing and mechan-ical dispersion of Cu-rich liquid residual of mss crystal-lization (Lorand, 1989a, 1989b), loss ofFe from sulfideinclusions to chromite hosts during subsolidus cooling(Naldrett and von Gruenewaldt, 1989), and to the effectsof low-temperature alteration such as serpentinization(Abrajano and Pasteris, I 989; Lorand, I 989d). Accordingto the fractionation hypothesis, during olivine crystalli-zation, the immiscible iron-nickel-copper sulfide liquidfractionates to a more Cu-rich composition as crystalli-zation of the silicate liquid proceeds. The bulk compo-sition of olivine-hosted blebs within the volume of a hand

STONE AND FLEET: SULFIDES FROM KILAUEA VOLCANO

specimen should be nearly identical and more Ni richthan those in later crystallized minerals or in the ground-mass.

In the eruption pumice, the variability of the Ni/Curatio of the olivine-hosted blebs and presence of the mostNi-rich blebs in the pumice glass are inconsistent with anorigin by fractionation of immiscible sulfide liquid. Con-versely, in Kilauea Iki lava, the presence of the morenickeliferous blebs within the olivines and the extremelyNi-depleted and Cu-rich blebs in dacite glass within thegroundmass appears to be consistent with nonequilibri-um fractionation. However, here, the restriction of theoccurrence of the sulfide blebs to the olivine in one sam-ple and to the groundmass in the other is unexplained bythis model. Furthermore, the Cu-rich sulfide bleb inclu-sions cannot be in equilibrium with ihe host olivine orbasalt liquid.

Wide variation in the modal percentages of chalcopy-rite and pyrrhotite in the mantle pyroxenite vein-hostedsulfide blebs from the Lherz and Freychindde spinel-peri-dotite massifs in France has been attributed to mechan-ical dispersion of Cu-rich liquid residual to mss crystal-l ization (Lorand, 1989a, 1989b). The mechanicaldispersion is said to have been induced by plastic defor-mation of the host rock in the range of 750 to 950 'C.

Evidence for pervasive plastic deformation in the Ha-waiian samples is lacking. It is difficult to envisage howthe variable bulk sulfide compositions of the olivine-hosted blebs might be explained by this model. Further-more, mechanical dispersion of liquid is unlikely to haveoccurred in the eruption pumice.

Fe loss from sulfide inclusions to chromite hosts is anunlikely explanation because many of the blebs are notassociated with chromite and the enrichment of Cu rel-ative to Fe, Ni, and S is unexplained. Although olivinehas, to varying degrees, lost Mg and gained Fe as a resultof reequilibration during the annealing (Helz, 1987b), thesimilarity in NiO content of these grains suggests that Nihas not been significantly redistributed. The lack of op-portunity for low-temperature alteration in the Hawaiiansamples precludes this explanation as well.

In addition to the wide variability in bulk compositionof the Hawaiian sulfide blebs even within individual ol-ivine crystals, the NiO content of olivine is too high andthe NiS content of the included and associated sulfides istoo low, relative to laboratory partitioning data, for equil-ibration during magmatism (Fleet and Stone, 1990). Therounded shape of the sulfide blebs is not unequivocalevidence for sulfide liquid immiscibility (e.g., Skinner andPeck, 1969). Bulk sulfide compositions that plot in themss phase field could reflect the composition of the hostin which the sulfides occur. Therefore, origin of the Ha-waiian sulfide blebs cannot be attributed solely, if at all,to sulfide liquid immiscibility in the early magmatic stage.Indeed, origin of the most Cu-rich bleb in the 1955 ba-salts from Kilauea Volcano by sulfide liquid immiscibil-ity has been questioned (Desborough et al., 1968).

Alternatively, the sulfide blebs may have been present

in the early magmatic stage of these rocks and were sub-sequently modified by late magmatic processes on cool-ing, particularly during quenching. Fleet and Stone ( I 990)surmised that all the sulfide inclusions from Kilauea Ikilava with less than equilibrium proportions of NiS havegained Fe, Cu, and S during cooling and quenching. TheS capacity of basalt melts decreases during cooling andabruptly on quenching. Excess S could segregate as sul-fides of metals locally abundant in the melt and may pref-erentially nucleate on existing sulfide blebs or be lost tothe gas phase (e.g., Anderson, 1974). Wide variation ofbulk-sulfide compositions would result. The observedtendency of increase in CurS content with decrease inbleb size and lack of equilibrium between olivine hostsand included and associated sulfides is certainly consis-tent with sulfidation during late-stage magmatic process-es.

The bulk compositions of the groundmass sulfides inboth the eruption pumice and Kilauea Iki lava lake arecertainly consistent with an origin by quench- and cool-ing-induced sulfidation, as reflected in their disparity incomposition and their correlation with the compositionof the immediate host material. The melt phase in Kilau-ea Iki lava lake evolved further by continued crystalli-zalion at a lower temperature than the eruption melt,producing extreme Cu enrichment of the sulfide blebs.Ample opportunity was available (about 20 yr) for bothdissolved and segregated sulfides in Kilauea Iki lava laketo be modified, through degassing and sulfidation, beforequenching. Indeed, blebs from Alae appear to have equil-ibrated with fractionated silicate liquid (Maclean andShimazaki, 1976).

The nickeliferous sulfides in the olivine-hosted glassinclusions may represent early magmatic nickel sulfides,possibly formerly in equilibrium with the coexisting ol-ivine, that have been diluted during cooling and quench-ing. The inclusion glass contains abundant S, and the sizeof the sulfide blebs decreases with decrease in area of theinclusion. However, many of the sulfide blebs appear toaccount for 20 or more modalo/o of some glass inclusions.This may reflect the level of section, but the large sulfideblebs appear to be relatively too large to have been sig-nificantly diluted by quench sulfide unless sulfur and met-al were derived from a much larger volume of melt.

Recent study ofthe products ofsulfide-silicate experi-ments indicates late introduction of sulfide blebs and sil-icate melt in olivine rather than entrapment during crys-tal gowth. Inclusions of sulfide blebs and glass arecommonly present within residual unmelted olivine grainsin experiments involving equilibration of immiscible sul-fide liquid and olivine basalt melt (Fig. 5). The sulfideliquid and silicate melt are apparently introduced by pref-erential dissolution along cleavage intersections and de-fects. The sulfide blebs illustrated in Figure 5A are Ferich and Ni poor relative to the principal mass of im-miscible sulfide in the charge. Stone et al. (1990) attrib-uted this disparity in composition to incomplete sulfi-dation of the olivine basalt, but given the low quench

I 3 7 1

Fig. 5. Backscattered electron images of iron nickel sulfideblebs (bright) and glass (pale gay) introduced into residual un-melted olivine grains (gray). Experiments were conducted on theequilibration of R-group elements between immiscible sulfideliquid and olivine basalt melt (Stone et al., 1990). (A) is exper-iment PG31; (B) is experiment PG32.

rate of these experiments, dilution by iron sulfide duringquenching cannot be excluded.

AcxNowr,rocMENTs

We thank R.L. Barnett and D.M. Kingston for their assistanc€ with theelectron microprobe analyses. R.T. Helz kindly provided the samples forthis study and critically reviewed an earlier version of the manuscript.Critical reviews by J.P. Lorand and P.A. Candela were helpful. The re-search was funded by a Natural Sciences and Engineering Research Coun-cil of Canada operating grant to M.E.F.

RnrnnnNcns crrEDAbrajano, T.A., and Pasteris, J.D. (1989) Zambales ophiolite, Philippines

II. Sulfide petrology of the critical zone of the Acoje Massif. Contri-butions to Mineralogy and Petrology, 103,64-77.

Andersen, T., Griffin, W.L., and O'Reilly, S. (1987) Primary sulfide meltinclusions in mantle-derived megacrysts and pyroxenites. Lithos, 20,279-296.

STONE AND FLEET: SULFIDES FROM KILAUEA VOLCANO

1372

Anderson, A.T. (1974) Chlorine, sulfur, and water in magmas and oceans.Geological Society of America Bulletin, 8 5, 1 483 - 1 492.

Bishop, F.C., Smith, J.V., and Dawson, J.B. (1978) Pentlandire-magnetiteintergxowth in De Beer spilel lherzolite. Review of sulfides in nodules.Physics and Chemistry of the Eanh, 9, 323-328.

Boctor, N.2., and Boyd, F.R. (1980) Ilmenite nodules and associated sul-phides in kimberlite from Yakutia, U.S.S.R. Carnegie Institute Wash-ington Year Book, 1979, 302-304.

Cabri, L.J. (1973) New data on phase relations in the Cu-Fe-S system.Economic Geology, 68, 443-454.

Caye, R., Cervelle, B., Cesbron, F., Oudin, E., Picot, P., and Pillard, F.(l 988) Isocubanite, a new definition ofthe cubic polymorph ofcubaniteCuFe'Sr. Mineralogical Magazine, 52, 509-5 14.

Clarke, A.H. (l 970) An unusual copper-iron sulfide, Cuo ,,Feo roS, ., fromthe Ylojarvi copper-tungsten deposit, Finland. Economic Geology, 65,590-59 l .

Craig, J.R. (1973) Pyrite-pentlandite assemblages and other low temprer-ature relations in the Fe-Ni-S system. American Joumal of Science,2734,496-5r0.

Craig, J.R., and Kullerud, G. (1969) Phase relations in the Cu-Fe-Ni-Ssystem and their application to magmatic ore deposits. Economic Ge-ology, Monograph 4, 344-358.

Craig, J.R., and Scott, S.D. (1974) Sulfide phase equilibria In Mineral-ogical Society of America Reviews in Mineralogy, l, CS-I-CS-I10.

Czamanske, G.K., and Moore, J.G. (1977) Composition and phase chem-istry ofsulfide globules in basalt from the Mid-Atlantic Ridge rift valleynear 37'N lat. Geological Society ofAmerica Bulletin, 88, 587-599.

Desborough, G.A., Anderson, A.T., and Wrighr, T.L. (1968) Mineralogyofsulfides from certain Hawaiian basalts. Economic Geology, 63,636-644.

Fleet, M.E. (1978) Origin of disseminated copper-nickel sulfide ore atFrood, Sudbury, Ontario. Economic Geology, 72, 1449-1456.

- (1979) Partitioning ofFe, Co, Ni, and Cu between sulfide liquidand basaltic melts and the composition of Ni-Cu sulfide deposits-adiscussion. Economic Geology, 7 4, 1517 -1 5 19.

Fleet, M.E., and MacRae, N.D. (1983) Partition of Ni between olivineand sulfide and its application to Ni-Cu sulfide deposits. Contributionsto Mineralogy and Petrology, 83, 75-81.

- (1987) Partition ofNi between olivine and sulfide: The effect oftemperature, .fo2 and f",. Contributions to Mineralogy and Petrology,95,336-342.

- (1988) Partition ofNi between olivine and sulfide: Equilibria withsulfide-oxide liquids. Contributions ro Mineralogy and Petrology, 100,462-469.

Fleet, M.E., and Stone, W.E. (1990) Nickeliferous sulfides in xenoliths,olivine megacrysts and basaltic glass. Contributions to Mineralogy andPetrology, 105, 629-636.

Reet, M.E., MacRae, N.D., and Henzberg, C.T. (1977) Partition of nickelbetween olivine and sulfide: A test for immiscible sulfide liquid. Con-tributions to Mineralogy and Petrology, 32, l9l-197.

Helz, R.T. (1987a) Differentiation behavior of Kilauea Iki lava lake, Ki-lauea Yolcano, Hawaii: An overview ofpast and current work. In B.O.Mysen, Ed., Magmatic processes: Physiochemical processes. The Geo-chemical Society, Special Publication I, 241 -258.

- (1987b) Character of olivines in lavas of the 1959 eruption ofKilauea Volcano and its bearing on eruption dynamics. U.S. GeologicalSurvey Professional Paper 1350, 69l -7 22.

Helz, R.T., and Thomber, C.R. (1987) Geothermometry of Kilauea Ikilava lake, Hawaii. Bulletin Volcanologique, 49,651-668.

Hill, R.E.T. (1983) Experimental study of phase relations at 600. in aportion of the Fe-Ni-Cu-S system and its application to natural sul-phide assemblages. In D.L. Buchanan and M.J. Jones, Eds., Sulphidedeposits in mafic and ultramafic rocks, p. 14-21. The Institution ofMining and Metallurgy, London.

Kanehira, K., Yui, S., Sakai, H., and Sasaki, A. (1973) Sulfide globulesand sulfur isotope ratios in abyssal tholeiite from the Mid-AtlanticRidge near 30'N latitude. Geochemical Journal, 7, 89-96.

Kullerud, G., Yund, R.A., and Moh, G.H. (1969) Phase relations in theCu-Fe-S, Cu-Ni-S, and Fe-Ni-S system. Economic Geology, Mono-graph 4,323-343.

STONE AND FLEET: SULFIDES FROM KILAUEA VOLCANO

ksher, C.M., and Groves, D.I. (1986) Controls on formation of komatiiteassociated nickel-copper sulfide deposits. In G. Friedrich, A. Genkin,A.J. Naldrett, J.D. Ridge, R.H. Sillitoe, and F.M. Vokes, Eds., Geologyand metallogeny of copper deposits, p. 43-62. Springer-Verlag, Hei-delberg.

Lorand, J.P. (1989a) Abundance and distribution of Cu-Fe-Ni sulfides,sulfur, copper and platinum-group elements in orogenic-type spinellherzolite massifs ofAridge (northeastern Plrenees, France). Earth andPlanetary Science ktters, 93, 50-64.

- (1989b) Sulfide perrology of spinel and garnet pyroxenite layersfrom manfle-derived spinel lherzolite rnassifs of Aridge, northeasternPyrenees, France. Journal ofPetrology, 30, 987-1015.

- (1989c) The Cu-Fe-Ni sulfide component ofthe amphibole-richveins from the Lherz and Freychindde spinel peridotite massifs (north-eastem Pyrenees, France): A comparison with mantle-derived mega-crysts from alkali basalts. Lithos, 23, 281-298.

- (1989d) Mineralogy and chemistry of Cu-Fe-Ni sulfides in oro-genic-type spinel peridotite bodies from Ariege (northeastern Pyrenees,France). Contributions to Mineralogy and Petrology, 103, 335-345.

Lorand, J.P., and Conquere, F. (1983) Contribution a l'6tude des sulfuresdans les enclaves de lherzolite a spinelle des basaltes alcalins (MassifCentral et I-anquedoc, France). Bulletin Mineralogique, 106, 585-606.

Maclean, W.H., and Shimazaki, H. (1976) The partition of Co, Ni, Cuand Zn between sulfide and silicate liquid. Economic Geology, 7 l, 1049-1057 .

Misra, K.C., and Fleet, M.E. (1973) The chemical compositions of syn-thetic and natural pentlandite assemblages. Economic Geology, 68, 5 I 8-539.

Missack, E., Stoffers, P., and El Goresy, A. (1989) Mineralogy, paragen-esis, and phase relations ofcopper-iron sulfides in the Atlantis II deep,Red Sea. Mineralium Deposita, 24,82-91.

Naldrett, A.J. (1989) Magmatic sulfide deposits, 186 p. Oxford UniversityPress, New York.

Naldrett, A.J., and von Gruenewaldt, G. (1989) Association of platinum-group elements v/ith chromitite in layered intrusions and ophiolitecomplexes. Economic Geology, 84, 180-187.

Naldrett, A.J., Craig, J., and Kullerud, G. (1967) The central portion ofthe Fe-Ni-S system and its bearing on pentlandite exsolution in iron-nickel sulfi de ores. Economic Geology, 62, 826-847 .

Rajarnani, V., and Naldrett, A.J. (1978) Partitioning of Fe, Co, Ni, andCu between sulfide liquid and basaltic melts and the composition ofNi-Cu sulfide deposits. Economic Geology, 73,82-93.

Skinner, B.J., and Peck, D.L. (1969) An immiscible sulfide melt fromHawaii. Economic Geology, Monograph 4, 3 10-322.

Stone, W.E., and Fleet, M.E. (1990) Platinum-iron alloy (PtjFe) in kim-berlite from Fayette County, Pennsylvania. American Mineralogist, 75,88 l -88 5.

Stone, W.E., Fleet, M.E., and MacRae, N.D. (1989) Two-phase nickelif-erous monosulfide solid solution (mss) in megacrysts from Mount Shas-ta, California: A natural laboratory for nickel-copper sulfides. Ameri-oan Mineralogist, 7 4, 981-993.

Stone, W.E., Crocket, J.H., and Fleet, M.E. (1990) Partitioning of palla-dium, iridiurn, platinum, and gold between sulfide liquid and basaltmelt ar 1200'C. Geochimica et Cosmochimicn Acta,54,234l-2344.

Sugaki, A., Shima, H., Kitakaze, A., and Harada, H. (1975) Isothermalphase relations in the system Cu-Fe-S under hydrothermal conditionsat 350 .C and 300 "C. Economic Geology, 70, 806-823.

Tsai, H., Shieh, Y., and Meyer, H.O.A. (1979) Mineralogy and Sr/S,ratios of sulfides associated with kimberlite, xenoliths and diamonds.In F.R. Boyd and H.O.A. Meyer, Eds., Proceedings of the second in-ternational kimberlite conference. The mantle sample: Inclusions inkimberlites and other volcanics, vol. 2, p. 87-103. American Geo-physical Union, Washington, D.C.

Yund, R.A., and Kullerud, G. (1966) Therrnal stability of assernblages inthe Cu-Fe-S system. Journal ofPetrology, 7,454-488.

MlNuscnrr'r RrcErvED Sevrpuser 21, l99OMeruscnrpr ACcEPTED Apnrl 9, 1991