compositional and textural peculiarities of gold-rich
TRANSCRIPT
ROLAND MERKLE, WIEBKE GROTE AND PETER GRÄSER 177
Compositional and textural peculiarities of gold-rich alloysfrom the Merensky Reef
Roland Merkle, Wiebke Grote and Peter GräserBushveld Intelligence Center, Department of Geology, University of Pretoria, Pretoria 0002, South Africa.
e-mail: [email protected]; wiebke.grote@up,ac.xapeter, graser® up.ac, za
© 2008 September Geological Society of South Africa
ABSTRACT
Gold-rich alloys with variable ¡imounrs of Pd from the Merensky Reef, Bushveld Complex, are charatterized by a range of finelydispersed inclusions. The small sizes of these inclasions (median of 1,85 fJLin) make fiilly quantitative analyses virtually impossible.Tentative interpretation of semi-quantitative microprobe analyses demonstrates the presence of atokite. kotulskite, laurite,moncheice, niggliite, paolovlte, rustenburgite, sobolevskite, and spenylite. Some inclusions appear to represent unnamed minerals.
Textures and assemblages imply that these gold grains and their inclusions formed from Huids through occasionalremobilisation and co-precipitation.
IntroductionThe ,Mt.Ten,sky Reef of the Bushveld Complex is not onlya major source of platinum-group elements (PGE),it also contains gold which is recovered as a by-product.Very little information about the concentrations anddistribution of gold in the Merensky Reef is available,but the recommended value for Au in SARM-7, aninternational standard for PGF which was prepared fromMerensky Reef (Steele el al.. 1975), is 310 ± 0.015 ppbfor the 95 % confidence interval, and an average valuefor Au in the Merensky Reef was reported to be 220 ppb(Vermaak, 1976). Regional variation in goldconcentration is linked to variations in SPGE+AU grade(Viljoen, 1994; 1999) and to variable proportions of Au(1.5 to 6.6% of SPGE+AU) in different types of MerenskyReef (Leeb-Du Toit, 1986; Viljocn and Hieber. 1986;Farquhar, 1986; Mossom, 1986; Vitjoen el al., 1986a;1986b).
Even in single boreholes through the Merensky Reef(ignoring dilferences in rock types) with an average Au-content of 460 ppb (Von Gnjenewaldt et al., 1990) and570 ppb (Barnes and Maier, 2002) respectively,individual samples were found to vary between <50and >2000 ppb (Barnes and Maier, 2002; Wilson andChunnett, 2006), Tin contents in the Merensky Reef areassumed to be about 2 ppm (Vermaak, 1976), whilesilver may reach 1.4 ppm in individual samples (Barnesand Maier, 2002) with a large scale average of 0.54 ppm(Vermaak, 1976), Gold shows positive correlations withAg, Pt, and Pd, but the number of reported analyses istoo small to allow a rigorous evaluation of variationsbetween the different rock types that make up theMerensky Reef (i.e. the economically mineable PGE ore,irrespective of the rock type). It is reasonable to assumethat different rock types of the Merensky Reef arepetrologically and genetically different (Cawthorn andBoerst, 2006).
Mineralogical descriptions mention the occurrence ofmetallic gold and electrum (Mihálik et al., 1975;Schwellnus et al.. 1976; Brynard et al., 1976; Vermaak
and Hendriks, 1976; Kingston and El-Dosuky. 1982;Mostert et al., 1982; Franklyn and Merkle, 1999) in theMerensky Reef. For simplicity', all the.se grains will liereferred to as metallic gold in this communication,irrespective of the Au/Ag ratio, Mosi of these grains arevery small. Volume proportions of metallic gold amongstPGE and Au-containing minerals range typically fromtraces to 4,I'M) (Brynard el al., 1976; Vermaak andHendriks, 1976; Kinloch, 1982), although in individualboreholes the proportion of Au-Ag alloys may exceed40% (Kinloch, 1982).
Large metallic gold grains of up to 500 |xm in lengthand up to 100 |i.m in thickness can be found inMerensky Reef concentrate, A microscopic investigationof such grains from Rustenbutg Platinum Mine (Kingstonand El-Dosuky, 1982) revealed that many of themcontain variable amounts of tiny inclusions which wereidentified to be atokite and mertieite II. Such inclusionsare not restricted to Rustenburg Platinum Mine and canbe observed in coarse metallic gold from around theBushveld Complex and have even been observed indetrital gold from the eastern Bushveld drainage system(Oberthüre/ö/,. 2004).
The information presented here is based onobservations of -250 gold grains handpicked fromproduction concentrate, with the main emphasis on thecomposition of the metallic gold, associations and typesof inclusions, and micro-textures. Because suchconcentrates are dependent on the mining operation ata specific time, they are representative for a largevolume of Merensky Reef, They can, however, not betaken to be representative for the Merensky Reef per se.The concentrate was put at our disposal under thecondition that its source may not be revealed.
Analytical ConditionsElectron microprobe analyses were performed with aJEOL 733 electron microprobe and attached ISIS energydispersive system with an accelerating potential of 20kVand a beam current of 2 x lO" A. We decided on this
SOtJTH AFRICAN JOURNAL O F GEOLOGY, 2 0 0 8 , VOLUME 111 PAGE 1 7 7 - 1 8 6
dui: \Ú.2U .VRSsain, 1 11.2-3.177
178 COMPOSITIONAL AND TEXTURAL PECULIARITIES OF GOLD-RICH ALLOYS FROM THE MERENSKY REEF
1. Frequency distribution (if Au content in gold-ricli alloys.
approach because this allows to store the spectnimof any analy.sis and to process it with combinations ofdifferent standards. Because of the very small grain sizesof most inclusions in the metallic gold, it had to beexpected that most analyses will represent mixtures of arange of platinum-group minerals (PGM) with metallicgold of varying Au / Ag ratios. Such complex variationsin the matrix cause a multitude of problems, of which
the non-linear determination in mixtures of Au and Ag(Reid et al., 1988), or Pt and Pd (Merkle and Venyn,1991) are just two. Correction procedures (mainlyabsorption and fluorescence) in electron microprobeanalysis assume that the analyzed material ishomogeneous and all elements are evenly distributedthroughout the excitation volume from which X-rays areproduced (Reed, 1975). Analyses {i.e., excitation areas)that represent two or more distinct pliases inevitablylead to incorrect results.
Attempts to produce metallic standards with complextnatrices failed because heterogeneities on ^tm-scalecould not be overcome. Suitable standards close to theunknown, which are required to achieve very goodquality results in complex matrices (Cabri, 1980), weretherefore not available. We had to resort to a large rangeof metals, alloys, and synthetic PGM in attempts toapproach the matrix in question. Consequently, manyanalyses are too low and in some cases, element ratiosappear to be represented inaccurately. Even if totals ofthe microprolie analyses approach totals of 100%, weconsider this to be coincidence and the analyses areconsidered to be only semi-quantitative. We thereforerefrain from any interpretation of small variations ofcomposition or stoichiometry. but consider the analysesperfectly acceptable for the tentative identification of thephases.
Au
60,
4o /
20/0 ^
10^ 80
at%
Pd* 10100 A
8 0 /
/00.• *
60
at%
V 20
\
40
literaturetlisstudy
40
\60
\ 80
20
Ag
0•
^100
at%
Figure 2. Compositional variation of gold-rich alloy from Merensky Reef concentrate. Literature dala from Kingston &
El-l>(),suky (1982) and Schwellnus et al., (1976),
SOUTH AFRICAN JOURNAL OF GEOLOGY
ROLAND MERKLE, WIEBKE GROTF AND PETFR C.RASER 179
Figure 3- HacLsfailLT ulcciron iiiu^fs of
(k'ivsiiy üv.in gold) define distinct zones.
y irom Mcrcnsky Red'coiicciuratc. Dürk inLlusions ol l'GM (with
ObservationsThe metallic gold hostOur suidy confirms the compí)silions of the gold-richalloys reponed previously (Schwellnus et ai, 1976;Kingston and El-Dosuky, 1982). The gold analyses in ourstudy wert' performed in iireas of the gold grains whichdid nol contain visilile inclusions. It appears thai a smallproportion of these grains are very pure gold with -86to 92 atomic % Au, while the majority containssubstamial amounts of Ag with -74 to 44 atomic % Au(Figure 1). Many of the more Ag-rich alloys contain U]ito - 9 atomic % Fd ("-6 weight % Pd") in solid soiution(Figure 2).
The inclusionsNot all gold grains contain visible inclusions. However,it is not clear whether the lack of inclusions is a functionof the grain exposure. Most grains have areas that arefree of inclusions, with the highest density of inclusionsin specific zones or rims of the gold grains (Figure 3).
Most of the inclusions in the gold-rich alloys areround to lenticular with the largest axis cotnmonlybelow 2 jxm. with a median of 1.85 p-m ft)r43(1 inclusions (Figure 4). It is sometimes possible to
microscopically distinguish more than one type ofinclusion in a particular gold-rich alloy grain, biit thesmall grain sizes make an identification impossible.Therefore, these inclusions can only be distinguishedand identified based on their element combinations.
Fre
quen
cy1J
J-
Ol
rr|1 :
Longest axis [micron]
Figure 4. Sizu distrihiition ct' i.-íl' ititki^inns in jiokl-ricli alloy.
Arilhmetic mean 2, ] 1 micron; median I. HS micron
SOUTH AFRICAN K)tlRNAI. Or GEOLOGY
180 COMPOSITIONAL AND TEXTURAL PECULIARITIES OF GOLD RICH ALLOYS FROM THE MERENSKY REEF
Table 1. Selected analyses in wt.-% of inclusions in gold-rich
aUoy.s to demonstrate the mixed character of analy.se.s of acokite
((Pd,Pt),Sn) and paolovite CCPd.Pt)^n)
Pd
3a7631.6454.0542.18
33.2639.8445.52
63.04
55.75
59,07
60,58
23.64
42.9340.0545.35
Pt
2,48
3,77
1,81
2,25
Sn
1322
10,25
18.93
14.07
12.06
13,98
23.26
35.16
30.81
32.99
34.77
11.59
19.84
20.44
22.54
Au
39-30
46,95
19,69
35,77
40.56
36.52
25.05
3.18
9,17
6.71
2.02
40.76
31.14
32.29
27.90
Ag
9.16
12,00
7.67
8,69
12.51
9.18
7.28
1.23
24.03
8.14
9.18
5,08
Total
100-44
100.84
100.34
100.71
98.39
99.52
101.11
103.86
100,73
100,58
99.62
100,02
102,05
101.96
100,87
Mineral
atokite
atokite
atokite
atnkite
atokite
atokite
paolovilc
paoloviie
paolovite
paolovite
paolovite
paoloviie
paolovite
paolovite
paolovite
Quantitative microprobe analyses are hampered byinterference from the surrounding gold-rich alloys.Although many analyses resulted in totals close to 100weight % (Table 1) and imply element proportions closeto expected stoichiometries, the high proportions of Auand Ag render these analyses only semi-quantitative.
Table 2. Pkitinum-gmiip minerals identified as inclusions in gold-
rich alloys = not observed in this study
iHatlnum-group mineral Formula
atokite
kotulskite
laurite
meitiite II
moncheite
niggliite
paolovite
rustenbutgite
sobolevskite
sperrylite
(Pd.Pt),Sn
Pd(Te,Bi)
CPt,Pd)CTe,Bi)i
(P(.Pd)Sn
PtA.s>
Based on their element combinations and pro-portions, a wide range of inclusions were recognized(Table 2). The most common inclusions contain Pd or Piand Sn. Our observations imply that in the MerenskyReef, Pd-rich inclusions va.stly outnumber the Pl-richinclusions. Niggliite, (Pt,Pd)Sn, is a rare mineral, whilepaolovite ((Pd,Pt)2Sn with variable but low Pt content)and atokite f(Pd.Pt),Sn without detectable Pt) are themo.st common inclusions. Rustenburgite, (Pt,Pd)iSn,inclusions were encountered only rarely, but theircomposition appears to fall inside the range of Pt/Pd
2oy
Z100
Pt at%
60
ei/
40/
/ •
80
Sn at%100 A
80 / \ ^
60 40
atokJteniggliitepaobvitepaobvite with Rruste rburgite
4Q rustertourgite (inci.)
\ 60
jj|80
\ioo20
Pd at%
ooD•O*
Figure 5. Compositional variation of Pt-Pd-Sn phases. Rustenbuigite, (Pt,Pd)3Sn, attached to gold (open circle) shows compositional
overlap with rustenbuqiite inclusions in gold (stars).
SOUTH AFRICAN JOURNAL OF GEOLOGY
ROLAND MERKLE, WIEBKE GROTE AND PETER GRÄSER 181
40y
20/
• ^ ^ ^ ^ ^ ^ ^ ^ ^ ^
100
Sn at%
Au
80y
60 /
/
80
+ Ag
0 \
60
at%
V 20
140
atokiteniggliitepaolovitepaolovitewithPtrustenburgite
40 rifitenburgite (incl.)
\ 60
20
R + Pd at"
00n•0*
Figure 6. Distinction beiwcfn p;iok)vilc, (Fd,Pt)2Sn, and atokite / aihtunlnirgilc, (l'd,Pt)3Sn. dcspiic inicrfercncc trom gold-rich
RH
2 0 /
/
100
hPdat
40y
/
/ o
60
/
80
Au
80y
/
+ Ag100 A
60
at%
V 20
\
V40
sobolevskite ?kotJskite ?- Pt2BiTe3
40
\\ 60
\ 80
D
\ioo
20
Te + Biat%
Figure 7, (Pt + Pd) / (Te + Bi) raiioh of .some inclusions.
SOUTH AFRICAN lOÜRNAL OF GEOLOGY
182 COMPOSITIONAL AND TEXTURAL PECULIARITIES OF GOLD-RICH ALLOYS FROM THE MERENSKY REEF
100 80 60 40 20
sobolevskite ? DkotJskite ? *~ Pt2BiTe3 *
60
80
Figure 8. Anaiy.scs at koliilskitc, l'd(Tf.iii), should plot on ihe line of 50 al % Pd,
R
4 0 /
20/
/100 80
at%
Sat100*
80/ '
/ .
60
/o
1
spenylite O) laurite *
mixed analyses n
\40
\ 80
, \') 20
As
100
at%
Figure 9. Analyses representing sperrylite, laiirite. and mixed analyses between the two.
SOL'TH AFRICAN JOURNAL OF GEOLOGY
ROLAND MERKLE, WIEBKE GROTE AND PETER G R A S E R 183
R
40y
20/
100
at%
60y
/
80
Sat100 A
80/
/
60
/o
\20
\
\
40
umamed
,40
\ 60
\
20
PtSnS 0
L 80
\
Sn
1100
at%
Figure 10. Projection of six analyses of unniiintd PtSnS.
ratios of rustenburgite attached to the gold-rich alloys(Figure 5). Tlie ratio of Sn to (Pd + Pt) of atokite andrustenburgite seems not to be vastly affected by theamount of interference from surrounding gold-rich alloy,while most analyses of paolovite deviate from theexpected ratio by a surplus of (Pd + Pt) (Figure 6).An isothermal section of the system Au-Pd-Sn at roomtemperature (Anhock et al., 1998) implies that syntheticPd¿Sn (equivalent to the natural paolovite) can exhibitan extensive range of solid solution with up to ca15 aromic % Au, if coexisting with a gold-rich alloy. Mostof our mixed analyses (Table 1) calculate to (Pd,Pt)2,xiinand give no indication of Au substituting for Pd in thispaolovite. A surplus of Pt + Pd appears to be a commonfeature of paolovite (Cabri, 2002).
In some grains, inclusions containing Te and Bi wereencountered (Figure 7). There is little doutit about theidentity of sobolevskite (Pdi.nBi), but the analysis whichwe interpret to be kotulskite, Pd(Te,Bi), seems to beslightly Pd-deficient (Figure 8). We attribute this toanalytical problems. Tlie third type of inclusion inFigures 7 and 8 appears to be moncheite (generalformula (Pt.PdKTe.Bi):, but in this case no Pd wasfound), although the composition is very close toPt BiTe? and may suggest a distinct phase. Too Uttieinformation is available to warrant speculation.
Spenylite (PtAsj) and laurite (RuS ) were found asisolated and small combined inclusions, which resultedin sotne mixed analyses (Figure 9).
Not all analyzed inclusions could be aíkx:ated toknown PGM. Even if severe analytical problems areassumed, some grains appear to represent unnamedminerals. We observed three grains consisting of Pd, Sn,and As, but were not able to obtain acceptable analyses.The analysis with the highest total of 93.5 weight %calculates to ~Pd7.8Sn2As.
Several grains of a phase containing only Pt, Sn. andS were encountered (Figure 10). Although all analyseshave low totals and are considered semi-quantitative,there is little doubt that the composition of this phase isPtSnS.
We did not observe any inclusions containing Sb andcan therefore not confimi the presence of inclusionswhich tnay be considered to be mertieite II (Pd^Sbj).
DiscussionMineralogyThere are clearly similarities between the compositionsof the gold found in our study and the analyses reponedpreviously frtjm the Merensky Reef. Gold from Atokappears to contain small amounts of Pt instead of Pd(Schwellnus et al.., 1976). while none of our analyses ofinclusion-free gold contained platinum in detectableamounts. We are inclined to believe that tiny inclusionsof Pt-containing minerals may be responsil^le and thatthe reported Pt is not present as a solid solution in gold.
Palladian gold has been reported from manylocalities. Judging from the available literature it seems
SOUTH AFRICAN JOURNAL OF GEOLOGY
184 COMPOSITIONAL AND TEXTL^RAL PECULIARITIES OF GOLD-RICH ALLOYS FROM THF MERENSKY REEF
that even grains with -12 weight % Pd from theMiessijoki placer in Finland (Tornroos and Vuorelainen,1987), or up to 19 weight % Pd in the Cauê Mine inBrasil (Olivo et al., 1994; Olivo et al, 1995) arehomogeneous. However, this aspect will require carefulre-evaluation because palladian gold from Cauê Minecertainly can be compositionally and mineralogicallyheterogeneous (J.H.D. Schorscher, personal commun-ication, 2004), The gold from the Meren.sky Reef, with aconsiderable lower Pd-content of up to about 6 weight%, undeniably has Pd in solid solution and containsparticulate Pd-rich phases as inclusions.
Although inclusions of platinum-group minerals ingold-rich alloys are rarely reported, the observatitm ofsuch a case in the alluvium of the Durance River, France(Johan et at.. 1990), where the gold contains -2 (ini-sized inclusitin.s enriched in Ni, Pt and Sn and up to-3 weight ' 1 of Sn, shows thai the Merensky Reef is notunique and that a systematic search for such inclusionsmight reveal more examples.
Unfortunately, the gold grains in this study areisolated from their original setting and it is therefore notpossible to relate the dominance of specific mineralsobserved as inclusions (i.e., platinum group mineralscontaining Bi, Te, Sn, and As) in a gold grain to theproportions of platinum-group minerals in the mineralassemblage in a given locality (Kinloch. 1982).
The unnamed Pt-Sn-S phase is occasionallyencountered in very small grains in the UG-2, thePlatreef. and the Merensky Reef (Kinloch, 1982;Louwrens, 1996). but normally grain sizes are too smallto obtain quantitative information about the proportionsof the elements. The only quantitative analyses in theavailable literature ( Barkov t?/a/., 2001) show very littlevariation in the composition compared to our analyses.Whether this is purely an anaKtical artefact remainsunresolved.
GenesisThe •liydrothermal" concept for the formation of theMerensky Reef has been advocated by some scientists,i.e.. that the Merensky Reef is a mineralization in whichthe PGE were remobilisée!, if not introduced, by flLiids(Stumpfl and Tarkian, 1976; Ballhaus and Stumpfl, 1986;Barnes and Campbell, 1988). Evidence for thehydrothermal modification of platinum group mineralsin the Merensky Reef has been presented (Merkle andVerryn, 2003). The formation of metallic gold grainsof the sizes which form the bulk of this investigationis very unlikely if only magmatic processes areconsidered.
Tlie tenor of 2 to 3 weight % base metal sulphides(Merkle and McKenzie. 2002) in the Merensky Reef isthought to have accumulated under magmaticconditions as a sulphide melt (Naldrett et al.. 1986;Barnes and Campbell, 1988; Cawthorn and Boerst, 2006;Wilson and Chunnett, 2006). The partition coefficient ofAu into such sulpliide melts (i.e.. the ratio of theconcentration in the sulphide melt divided by
the concentration in the silicate melt under equilibriumconditions) has been estimated Lo be up lo 18600 (Peachet al., 1990; Stone et al.., 1990), depending on intrinsicconditions like oxygen fugacity, cation ratios in thesulphide melt, or metal/sulphur ratio of the sulphidemelt (Crocket et ai, 1992; 1997; Fleet et ai, 1999). Suchhigh partition coefficients suggest that the element inquestion is homogeneously distributed throughout thesulphide melt.
Gold is alloyed with variable amounts of silver,copper, and other minor components (like Pd) in ametallic phase of very high density, whereas most of thePGE form compounds (with sulphur, tellurium, andother ligands) of lower density, A variable proportion ofPd is present as ,solid solution in base melal sulphides(Paktunc et al. 1990; Cabri, 1992; Ballhaus and Ryan,199S) and except for .selected localities, Pt-Fe alloys arerare (Kinloch. 1982). i.e.. the volume proportion ofmetallic goltl should be considerably lower than theweight proportion (1-S to 6.6% of 5 PGF,+Ati) suggt-sts.U is also to be expected that the bulk of the low goldconcentration in the sulphide melt would result in small,finely dispersed grains on cooling. Reports of higherproportions of metallic gold in Merensky Reef (Kinloch,1982) therefore imply remobilisation and concentrationof metallic gold. Large grains of up to 500 |xm length asstudied here are therefore considered to have formedby hydrothermal remobilisation from magmaticsulphides and the co-precipitation of (he gold and itsinclusions.
Individual grains contain distinct zones in which thedusting of PGM occurs. In our opinion these zonesrepresent growth zones of the gold. Our observationsimply that in the coarse gold from the Merensky Reef,Pd-rich inclusions vastly outnumber the presence ofPt-rich inclusions.
If tliL'se inclusions of PGM represent exsolutions, thiswould imply that the high temperature alloy contained ahigher amount of Pd in solid solution, which exceededsolubility on cooling, while the concentration of Pt waslower and was more efficiently exsolved. It also wouldimply that the Sn. Bi. and Te in the PGM had to bedissolved in the gold as well. We were not succes.sful inour attempts to kx:ate phase diagrams whicli wouldallow us to quantify the maximum sohiliility of Sn, Bi,and Te in gold as a function of temperature underhydrothermal conditions. However, phase diagrams ofbinary systems relevant to our study show that there isan extended solid .solution between Au and Pd, Pt, or Snat elevated temperatures, by far exceeding the valuesobserved in the gold grains under investigation(Massalski. 1990), although solubility might be reducedat low temperature. There is no indication that thesolubility of Pt in gold is so distinctly lower than of Pdat the concentration levels in this gold that it could causetlie observed dominance of Pd-rich PGM, This impliesthat the element content of the inclusions are unlikely tohave been in solid solution and unmixed on cooling.The finely dispersed PGM in gold-rich alloys are
SOUTH AFRICAN JOURNAL OF GEOLOGY
ROLAND MERKLE. WIEBKE GROTE AND PETER GRÄSER 185
therefore considered to be physical inclusions, notcxsolutions, ;md tlie additional presence of Sn, Bi, Te,and As in the fluid must be seen as the reason for thepresence of paniculate i GM,
The palladium dominatue must therefore be afunction of the composition of the hydrothermal (luidfrom which the gold precipitated.
The differences in the aqueous geochemistry of Ptand Pd are small but distinct enough to cause a smallerstability field of PdS (vysotskite) compared to PtS(cooperite), depending on temperature, pH, chlorineactivity, and oxygen fugacity (Wood et ai, 1992;Gammons et al., 1992), It is therefore possible topreferentially leach Pd from (Pd,Pt,Ni)S (braggite).This is the reason why braggite, the most likely primaryPGM forming from the sulphide melt, may loose Pdpreferentially through interaction with hydrothermalfluids (Merkle and Verryn, 2003). Examples for suchPd-depleted braggite are known, amongst others, fromthe Merensky Reef (Schwellnus et al., 1976; Cabri et al.,1978; Kingston and El-Dosuky. 1982; Mostert e/«/., 1982)and Lukkulaisvaara (Barkov el ai, 1995).
The precipitation of PGM and inclusion in the goldgrains would have to happen below the upper thermalstability limits of the inclusions. Unfortunately, the upperthertnal stability limits for all the minerals found asiticlusions are rather high. K<.>tulskiie has a minitnumstability temperature of 720°C (Elliot. 1965; Shunk,1969). while atokite, moncheite, niggliite. andrustenburgite were synthesized at 1000°C (Shelton el al.,1981; Kim, 1990). Laurite and sperrylite can be found asprimary inclusions in chromite grains, which imply evenhigher stability temper.uures. All the phases that occur asinclusions in gold, and for which thermal stability couldbe obtained, are stable at temperatures which are to thehigh end of temperatures expected for any hydrolhertnalactivity that may have redeposited gold.
Suinniary and ConclusionsWe conclude that the assemblage of iticlusions inthe large gold-rich alloys of the Merensky Reef reflectsthe variability of compositions and intrinsic conditions ofhydrothermal fluids, which occasionally remobilised andprecipitated Au and Ag in the Merensky Reef into largegrains. The rarity of these large gold-rich alloy grainsitiiplies that the imlk of the low Au concentration isfinely dispersed in small grains. Inclusions of PGM areinterpreted to be co-precipitates from the samehydrothermal fluids. It is at present not po,ssible toquantify the conditions under which these gold grainswere ptecipitated.
AcknowledgmentsWe were requested t(i keep the source of the materialconfidential, which does not diminish our gratitude forthe opportunity to report on our findings. We thankLouis Cabri and the anonymous reviewers for theirconstructive comments.
ReferencesAnliock, S,, OpiJcnnann, 11,, Kallmayer, C, A-stlifnhrenner, R., Thomas, L,
and Reichl, H. 11998), in Investi nations of Au-Sn alloys on ditTi-rtni end-
metallizatitins for high teinperyiure applicaiion.s. In.- Electronics
Manufacturing Technology Synipo.sium, 1998, lEMT-Eiirope 1W8, Twenty-
Second lEEE/CPMT International, 156-1Ö5,
Ballhaus, C, and Ryan, C,G, (1995). i'lütinuni-group elements in
tht' Mert-nsky RL'ef I, PGE in solid solution in ba.su metiil .sulphides and the
down-tempe rature ec|uilibrium hi.story of Meren.sky OR'S. Contributions to
Mineralogy and Pi'irolog\\ 122, 122-211,
Ballhaus, CG, and Stumpll. E.K <19H(i), Sulfide and i>kinnuiii iiiinerali,sali»m
in the Merensky Reef: evideme from hytinius silkate.s and tluid inclusions.
Contributions to Mineralogv and Petrology, 94, 193-204.
Bwrkov, A,Y.. Martin, R.R, Kaukonen, R,J, and Alapieti. T,T, (2001),
The occurrence of Pb-Cl-(OH) and Pt-Sn-S compounds in the Merensky
Ki'ef, iîushveld layered complex. St^uth Africa. Canadian .iiineralogvst.
39, 1397-140.Í,
Barkov, A,Y„ Pakhomovskii. Y,A, and Menshiko\\ Y,t> (1995), Zoning in the
platinum-group sulfide minerals from the Lukkulaisvaara and Inwndrov.'iky
layered intrusion,s, Russia, ,\'eui's Jahrbuch für Mineralogie. Abliandlungen,
169,97-117,
liâmes. S,.I, and (.Campbell, I,H. (1988). Role of late magnutic fluids in,Merensky-type platinum deptwiLs: a discussion. Geology. 16, 4ÍW-191.
Barnes, S,-J. and Maier, W,D, 12002), Platinum-group Elements andMicrostructures of Normal Merensky Reef from Impala Platinum Mines,Hiishveld Complex. Journal of Petrology. 43, 103-128,
linnard, H.J,, De Villiers, .Í,P,R, and Vij)oen, E,A, (1976). A mineralogicalinvestigation of ttie Merensky Reef at the Western Platinum Mine, nearMarikana, Si>uth Africa, Economic Geology, 71 . 1299-1307.
Cabri, LJ, (1980), Determination of ideal formulae for new mineral.s of thejilatinum group, fm A.V. Sidorenko, V,S, Sobolev, D,V, Riind()uist,I.Y, Nekrasov, T.N, Shadlun, A,D. Genkln. N,N, Mozgova and.N,S, Borinikov (Editors), Siilpho.salts. Platinum Minerals and OreMicroscopy, Proceedings of the XJ.th General Mœtina ofthi- International
Mincrttlogital AsxiK'iation. Novosibirsk. 1978. Institute of the Ck-olog)- oJ'Ore
Lk-posits, Akad. Nauk SSSR, Nauka. Moscow. USSR. TiT-lóS,
Cabri, L,J, (1992), The distribution of trace precious metals in minerals andmineral products, Minemhgical Magazine. 56, 289-308.
Cabri. L,J, 12002). Tlie Platinum-Group Minerals, /ti.- LJ. Cabri (Editor),"llic Geology, Geochemisiry, Mineralogy and Mineral Reneficiation ofI'laiininn-Grotip Elemenls, Canitclian Institute of .Mining. .Metallurg}' and
Petroleum. Special Volume. 54, 13-129,
Cabri. l.,J.. Laflamme, }.\\.G.. .Stewan, J.M,. Turner, K, and Skinner, B,J,(1978), On ctKïperite, limggile and vyMitskite, American Mineralogist,
63, 832-839,
Cawthorn, R.G, and Boerst. K, (2006). Origin of the peipnatitic pyroxenite inthe Merensky Unit, Bushveld Complex, South Afvica. Journal of Petrology,
47, 1509-1530.
CnKket, J,H.. Fleei. M,li, and Stone, W,K. (1992). E.\|5erimental partitioningof osniiuiu, iridium and gold between basall melt and sulphide lit[uid at13OO°C, Australian Journal of Earth Science. 39, 427-432.
Crocket, ,1,H,. Plt-et, M,E, and Stone. W,F., (1997), Implications of compositionfor experimental partitioning of p latin um-group elements and goldbetween suiiWe liquid and basalt melt: Tlie significance of nickel a>ment,Geochimica et Cosmoçhimica Acta, 61 , 4t39'4t49,
Elliot, R,P. (1965) Constitution of Binary Alloys, 1st Supplement, McGrau--
Hill, New York. 877p.
Farquhar, J, (1986), l l ie WeMcrn Platinum Mine. In: CR. Anhaeusser and S,Maske (Editors). Mineral Deposits of Southern Africa, Geological Society of
.South Africa. 113S-l!-i2,
Fleet. M,E,, Crwket, J,H,. Uu, M. and Stone, W,E, (1999). Lalxiratorypartitioning of platinum-group c-itrnenLs (PGE) and gold with applicationto magmatic suifide-PGE dejxisits. Lithos, 47, 127-l'l2,
Franktyn, C.B, and Merkle, R,K,W, (1999). Milli-PKE of coexisting cooperiteand braggite - a comparison with eiearon micropnabe analysis. Nuclear
Instnimi'nts and .Metl.uids in Phy.-:ics Research B. 158, 'i'iO-'i'i'i,
Gammons, C,H,. BÍ(Him. M,S,, and Yu. Y, (1992), Experimental investigationof the hydrothermal ge(M.~hemistr\- of platinum and palladiunv 1. Solui)iliiyof platinum and palladium sulfidc minerals in NaCl/H>SO, solutions al300°C. Geochimicit et Cosmwhimica Ada, 56, 3881-3894,
SOUTH AFRICAN JOtlRNAl, OF GEOLOGY
186 COMPOStTIONAL AND TEXTURAL PECULIARITIES OF GOLD-RICH ALLOYS FROM THE MERENSKY REEF
Johan, Z., Ohnenstetler, M., Fischer, W., and Amassé. J, (1990). Platinum-
group minerals trom the Durance River alluvium, France, Mineralogy and
Petrology. 42, 2a7-306,
Kim, W.S, {1990>. Phase reiation.s in the s>'sleni Pt-Sb-Te. Canadian
Mineralogist. 28. (iT-i-MS,
Kingston, G.A. and Hl-Dosuky, B.T. U982), A contribution on the platinum-
group mineralogy of the Meren.sky Reef at the Ru.stenbutg platinum mine.
Economic Geolc^. 77. 1367-1384.
Kinloch, ED. (1982), Regional trends in tht- plaLinum-group mineralogy of
the Critical Zone of ihe Bushveid Complex, South Africa, Economic
Geology, 77, 1328-1347,
l.eeb-Du Toit, A. (19H()). The Impala Platinum Mines, In: C.R. Anliaeus.ser and
S. Ma.-îke (Edit<.)rs), Mineral Depo.fiLs of Southern Africa. Geological Sociely
ofSoutb Africa. 109M10(i,
Louwrens, E.L. (1996), Mineralogical investigations of piatinum-group
minerals in UG-2 plant products (UlOl; U1Û2), Conßäetitial Report.
R5V96-IMR.27. Impala Platinum. V> p.
Mas.sal.ski. TB. (Editor) (1990). Binary alloy phase diagrams. American
Society for Metals. Metals Park. Obio. United States of America. 2224p.
Merkle. R.K.W. and McKenzie, A,D, (2002). The Mining and Beneficiation of
South African PGE Ores - An overview. In: LJ. Cabri (Editor),
Tile Geology. Geochemistr>'. Mineralogy and Mineral Beneficiation of
Platinum-Group Elements. Canadian Institute of Mining. Metallurgy and
Petroleum. Special Volume. 54, 793-809.
Merklf, R.K.W. and Verryn, S.M.C. (1991)- Coexisting cooperite and braggite
- new data. ICAM '91, International Congre.^*: on Applied Alineralogy.
Papers. Minerakigica) ARSi>ciaiion of South Africa. 2 Chapter 61, 1-1C>.
Mfrkle, K,K.W. and Verryn. S,M.C. (200.i), Coopt'rite and braggiie from the
liushveld Compiex: implicalions for ihe miscibiliiy gap and icientilication.
Mineralium Deposita. 38, 581-3H«.
Mihálik, P., Hiemslra, S.A.. and De Villiers, J.P.R, (1975). Rustenburgite and
atokite, two new platinum-group minerals from the Merensky Reef,
Bushveid igneou.s complex, Canadian Mineralogist. 146-150.
Mos.-«»!!, R.J. (1986). The Atok Pblinuni Mine, in: C.R. Anhaeus.ser and
S, Maske (Editors), Mineral Deposits of Southern Africa. Geological Scx-iety
of South Africa, 1143-1154.
Mo.stert, A.B., Hofmeyr, PK. and Potgieter, G.A. (1982). The platinum group
mineralogy of the Merensky Reef at the Impala platinum mines,
Bophuthatswana. Economic Geology. 77, 1385-1394.
Naldrett, AJ,, Gasparrini, E.C., Bames, SJ., Von Gruenewaldt, G. and Sharpe,
M.R. (198ÍÍ). The upper critical zone of the Bushveld Complex ¿md a mcxJd
for the origin of Merensky-type ore.s. Economic Geology.
81, I105-I117,
Oberthür, T., Melcher. F„ Gast, L, Wöhrl, C, and L<x!ziak, j . (2004>, Detrital
piatinum-group minerais ln rivers draining the eastern Bushveid Complex,
South Africa. Canadian Mineralogía. 42, 563-582.
Olivo, G.R,, Gauthier, M. and Bardoux, M. (1994), Palladian gold from the
Caue iron mine, Itabira District, Minas Gerais, Brazil, Mineralogical
Magazine. 58, 579-587.
Olivo. G.R., Gauthier, M., Bardoux, M., Desa, E,L, Foaseca, J.T.F, and
Santana, F.C, Í1995}. Paliadium-liearing gold deptxsit ho.sfed by Proterozoic
lake superior-typt- iron-formation al the Caue iron mine, Itabira district,
southern Sao Francisco craton. Brazil: Geologic and stnicttiml controls.
Economic Geology. 90, 118-13^,
Paktunc, A.D., Hulbert, LJ. and Harris, D.C. (1990). Partitioning of the
platinum-grotip and other trace elemenLs in sulfides from the Bushveid
Complex and Canadian occurrences of Nickel-Copper sulRdes. Canadian
Mineralogist. 28, 47'i-488.
I'each, C.L, Mathez, E.A. and Keays, R.R. (1»/X)). Sulfide melt - .silicate meit
distribution coefficients for noble metals and other chalcophile elemenis a.s
deduced from M ORB: implications for partial melting. Geocbimica
el Cosmochimica Acta, 54, 3379-3389,
Reed, SJ.B, (1975) Electron Microprobe Ana!y,sis, 1 Edition, Cambridge
Unifersity Press, united Kingdom. 400p.
Reid, A,M., Le Roex, A,P and Minter, W,E,L, (1988). Comixxsiiion of gold
grains in the Vaal Placer. Klerlcsdorp, Scjuth Africa, Mineralium Deposita.
23, 211-217,
Schwetlnus, J,SJ,, Hiemstra, S,A, and Gasparrini. E. (1976), 'Ihe Merensky
reef at the Atok platinum minu and ils environs. Economic Geology,
71, 249-260,
Shelton, K,L, Merewether, PA., and Skinner, BJ. (1981), Phases and phaserelations in the .system Pd-Pt-Sn. Canadian Mitteralogist. 19, 'i99-(J05,
Shunk. F.A, (1969). Constitution of Binary Alloy.s. 2nd Supplement, McGraw-
Hill, Netv York, 720 p.
Steele. TW,. Levin J., and Copelowitz. 1, (1975), The preparation and
ccdificulion of a reference sample of a precious metal ore. NIM Report,
1696, 50 p.
.Stone, W.E., Crocket, J H,. and Fleet, M.E, (1990), Panitioning of palladium,
iridium, platinum, and gold tx^tween sulfide liquid and ba.sal( melt al
1200°C. Getxhimica et Cosmocbimica Ada. 54, 2341-2344,
Stumpfl, E.F, and Tarkian, M, (1976), Platinum genesLs; New mineralogical
evidence. Economic Geology, 71 , 1451-1460,
Tömroos, R, and Vuorelainen,Y, (1987). Platinum-gKiup metals and their
alloys in nuggets from alluvial deposits in Finnish Lapland, Lithos,
20, 491-500.
Vermaak, C,F. (1976). The Merensky Reel' - thoughts on its environment and
genesis, , Economic Geology, 71 , 1270-1298.
Vermaak, C F and Hendriks, LP, ( 1976), A review of the mineralogy of the
Merensky Reef, witli specific reference to new data on precious metal
mineralogy. Economic Geology. 7 1 , 1244-1269,
Viljoen. MJ, (1994), A review of regional variations in facie.s and grade
distribution of the Mereasky Reef. Western Bushveid Complex with some
mining implicatloas. In: CR. Anhaeusser (Editor), Proceedings, XVth CMMl
Qjngress, South African Institute of Mining and Metallurgy / Geological
Society of S<,)uth Africa, 3, Gcolc^ty, 183-194.
Viljr^n, MJ. (1999), The nature and origin of the Merensky Reef of the
western Bushveiil Complex based on geological faciès and geophysical
data, Soutb African Journal of Geology.102, 221-239.
VÜjoen. MJ,, De Klerk, WJ,. Coetzer, RM,, Hatch. NP,, Kinloch, E, and
Peyer!, W, (1986a). The Union Section of Rustenburg Platinum Mine.s
Limited with reference to the Merensky Reef, In: C. R. Anhaeu.sser and
S, Maske (Editors), Mineral Depcxsits of Southern Africa, Geological S<Kiety
of South Africa. 10(jl-1090,
Vitjoen, MJ. and Hielier, R. (1986), The Rustenburg Section of Rustenburg
Platinum Mines Limited, with reference to the Merensky Reef. /». CR,
Anhaeusser and S, Maske (Editors), Mineral Deposits of Southern Africa,
Geological Society of Soutb Africa. 1107-1134.
Vilfoen. M.J., Theron, J., Underwood. B., Walters. B.M.. Weaver, J. andPeyerl, W. (1986b), The Amandelbult Section of the Ru.stenberg Platinum
Mines Limited, with reference to the Merensky Reef, hi: CR, Anhaeu.sser
and S, Maske (Editors), Mineral Deposits of Siiuthern Africa. Geological
Society of South Africa. 1041-1060,
Von Gruenewaldt, G,, Dicks, D,, De Wet, J, and Horsch, H, 11990), PGE
mineralization in the western sector of the eastern Bushveid Complex,
Mineralogy ami Petrology. 42, 71-95,
Wilson, A, and Chunnett, G, (2006), Trace element and platinum group
element distributions and the genesLs of the Merensky Reef, We.stern
Bushveid Complex, South Africa. Journal of Petrology. 47, 2369-2403,
Wood. S.A., Mountain. B,W,, and Pan, P, (1992), The aqueous geochemisiry
of platinum, palladium and gold: Recent experimental constraints and a
re-evaluation of theoretical predictions, Canadian Mineralogist,
30, 955-9H2,
Editorial handling: A, Bumby
SOtJTH AFRICAN JOtlRNAL OF GEOLOGY