partitioning of platinum-group elements (pge) and ... · mss/iss partitioning experiments reveal...
TRANSCRIPT
Partitioning of Platinum-Group Elements (PGE) and
chalcogens (Se, As, Te, Sb, Bi)
at controlled fO2-fS2 conditions
by
Yanan Liu
A thesis submitted in conformity with the requirements for the degree of Doctor of
Philosophy
Department of Earth Science
University of Toronto
@Copyright by Yanan Liu 2015
ii
Partitioning of Platinum-Group Elements (PGE) and chalcogens (Se,
As, Te, Sb, Bi) at controlled fO2-fS2 conditions
Yanan Liu
Doctor of Philosophy
Department of Earth Sciences
University of Toronto
2015
Abstract
A more quantitative understanding of how highly siderophile elements (HSEs: Os, Ir, Ru, Rh,
Pt, Pd, Au, Re), Ag, Pb and chalcogens (As, Se, Sb, Te and Bi) behave during magmatic
sulfide solidification in nature builds on several important aspects: 1) accurate measurements
of their concentrations in natural samples; 2) quantitative constraints on their partitioning
behaviors; 3) a crystallization model that can reasonably simulate the ore deposits of interest.
In order to better measure the concentrations of these elements during the solidification of
sulfide melts, we have developed a new Laser Ablation Inductively Coupled Plasma Mass
Spectrometry (LA-ICPMS) standard (Ge-Sb-S glass; Ge6) for improving the trace level
measurements of these elements in both sulfide and silicate glass matrices. The Ge6
composition has excellent glass forming capability and can incorporate a variety of dopants
up to the 100 ppm level, except for Ru, Os, Re, W and Mo, which can be doped to 5-10 ppm
iii
without sacrificing homogeneity. Reference materials (JBSulfide, NiS4, JGb-1, JG-1a, JB-2,
BIR-1, BHVO-1) were tested using Ge6 as the standard, and the results agree with accepted
values. To better understand and simulate the partitioning behaviors of these elements, a
series of experiments to measure partition coefficients (D values) between monosulfide solid
solution (MSS) and sulfide melt, as well as MSS and intermediate solid solution (ISS) were
also conducted, at 0.1 MPa and 860–926 °C (where these phases can be stabilized), log fS2 -
3.0 to -2.2 (similar to the Pt–PtS buffer), with fO2 controlled at the fayalite–magnetite–quartz
(FMQ) buffer. The IPGEs (Os, Ir, Ru), Rh and Re are found to be compatible in MSS
relative to sulfide melt with D values ranging from ~20 to ~5, and DRe/DOs of ~0.5. Pd, Pt,
Au, Ag, Zn, Pb, as well as the chalcogens, are incompatible in MSS, with D values ranging
from ~0.1 to ~1 x 10-3. For the same metal/sulfur ratio, D values for the IPGEs, Rh and Re
are systematically larger than most previous studies and correlate with higher oxygen content
in the sulfide liquid, reflecting the significant effect of oxygen on increasing the activity
coefficients for these elements in the melt phase. MSS/ISS partitioning experiments reveal
that Ru, Os, Ir, Rh and Re are partitioned into MSS by a factor of >50, whereas Pd, Pt, Ag,
Au and the chalcogens partition from weakly (Se, As) to strongly (Ag, Au) into ISS.
Uniformly low MSS- and ISS- melt partition coefficients for the chalcogens, Pt, Pd, Ag and
Au will lead to enrichment in the residual sulfide liquid, but D values are generally too large
to reach early saturation in Pt–Pd-chalcogen-rich accessory minerals, based on current
solubility estimates. Instead, these phases likely precipitate at the last dregs of crystallization.
Modeled evolution curves for the PGEs and chalcogens are in reasonably good agreement
with whole-rock sulfide compositions for the McCreedy East deposit (Sudbury, Ontario),
consistent with an origin by crystallization of MSS, then MSS + ISS from sulfide magma.
iv
Acknowledgments
I am extremely grateful to James Brenan, who provided this precious opportunity for me to
be a graduate student again, who patiently taught and guided me through the past 6 years,
who encouraged, inspired and motivated me when I was nearly desperate for lacking of any
progress with the project. All your advice will not only benefit this thesis, but will also be
invaluable for my future research!
I am also in great debt to Mike Gorton, who allowed me with the freedom and space needed
to complete this degree from work, who provided helpful advice on analytical related topics,
who always encouraged and comforted me for all the ups and downs in my life during the
past years.
I am grateful to The University of Toronto for providing such an opportunity, which allowed
me to further my education while being employed fulltime. Great thanks also go to Society
of Economic Geologists (SEG) who provided part of the funding supporting this project.
Sincere appreciation goes to everyone who has helped me with lab-work and analysis: Colin
Bray, Neil Bennett, Boris Foursenko, Duane Smythe, Carol Cheyne, George Kretchmann,
Sandra Kamo just to name a few.
Last but not least, my family! Thank you for your great tolerance, patience, support and
sacrifices. This thesis would not be possible without you!
v
Table of Contents
Acknowledgments .............................................................................................................IV
Table of Contents ................................................................................................................ V
List of Figures ................................................................................................................. VIII
List of Tables .....................................................................................................................XI
I. An Introduction to common magmatic Ni-Cu- Platinum Group Elements (PGE)
ore deposits ...................................................................................................... 1
1.1 Significance of magmatic Ni-Cu-PGE ore deposits ............................................. 1
1.2 Saturation of a sulfide liquid ................................................................................ 2
1.3 Crystallization of a sulfide liquid ......................................................................... 5
1.4 Role of chalcogens in magmatic Ni-Cu-PGE deposits ........................................ 6
1.5 Contributions of this thesis ................................................................................... 9
1.5.1 Author contributions .............................................................................. 10
1.5.2 Measurement of PGE and chalcogens at trace concentrations level----
Synthesis of a chalcogenide glass standard for laser-ablation inductively
coupled plasma mass spectrometry........................................................ 10
1.5.3 Partitioning of PGE and chalcogens within sulfides under controlled
fO2, fS2 conditions ---- Experimental measurements and origin of MSS-
melt fractionation ................................................................................... 11
1.5.4 Compositional evolution of magmatic sulfide melt: models from
partitioning experiments applied to the McCreedy East Deposit,
Sudbury, Ontario .................................................................................... 12
II. Measurement of PGE and chalcogens at trace concentrations level ---- Synthesis of
a chalcogenide glass standard for laser-ablation inductively coupled plasma
mass spectrometry (LA–ICPMS) ................................................................... 18
2.1 Introduction ...................................................................................................... 18
2.2. Synthesis methods ............................................................................................ 21
2.2.1 Doped sulfide synthesis ......................................................................... 21
2.2.2 Dopants addition .................................................................................... 22
2.3. Characterization of the synthesized material .................................................... 24
vi
2.3.1 General Aspect ....................................................................................... 24
2.3.2 Major and Minor element distributions ................................................. 25
2.3.3 Trace element distributions .................................................................... 26
2.4 Discussion and application ................................................................................. 32
2.4.1 Assessment of homogeneity and factors affecting homogeneity ........... 32
2.4.2 Application ----- Testing reference materials ........................................ 35
2.5 Conclusion .......................................................................................................... 37
III. Partitioning of PGE and chalcogens within sulfides under controlled fO2,
fS2 conditions ---- Experimental measurements and origin of MSS-melt
fractionation .................................................................................................... 66
3.1. Introduction ....................................................................................................... 66
3.2. Experimental Technique ................................................................................... 67
3.2.1 General Strategy .................................................................................... 67
3.2.2 Starting materials ................................................................................... 68
3.2.3 Verification of experiment fO2 and fS2.................................................. 69
3.2.4 Partitioning experiments ........................................................................ 71
3.3. Analytical Techniques ....................................................................................... 72
3.3.1. Major element analysis ......................................................................... 72
3.3.2. Trace element analysis .......................................................................... 73
3.4. Results ............................................................................................................... 75
3.4.1 General aspects ...................................................................................... 75
3.4.2 Attainment of equilibrium ..................................................................... 76
3.4.3 MSS-melt partitioning ........................................................................... 76
3.4.4 MSS-ISS partitioning ............................................................................. 80
3.4.5 ISS-melt partitioning.............................................................................. 81
3.5. Discussion ......................................................................................................... 82
3.5.1 Origin of the PGE and chalcogen partitioning systematics ................... 82
3.6 Summary and Conclusions ................................................................................ 86
IV. Compositional evolution of magmatic sulfide melt: models from
partitioning experiments applied to the McCreedy East Deposit, Sudbury,
Ontario ........................................................................................................... 117
vii
4.1. Elemental evolutionary models for McCreedy East Ore Body, Sudbury (Canada)
................................................................................................................................ 117
4.2. Timing and sequence of certain magmatic PGMs .......................................... 123
4.2.1 Solubility of Pt-Pd-Te-As-bearing PGMs in sulfide liquid ................. 123
4.2.2 Estimates for the timing of crystallization of Bi-, Sn-, Zn- and
Pb-bearing phases ............................................................................... 125
4.2.3 Crystallization sequence and factors affecting the accuracy of the
model ................................................................................................... 128
4.3. Conclusions ..................................................................................................... 130
V. Summary and Conclusions .................................................................................... 141
viii
List of Figures
Figure 1-1 Summary of sulfide/silicate partition coefficients for PGEs ......................... 14
Figure 1-2 Illustration of sulfide melt crystallization sequence ...................................... 15
Figure 1-3 Preliminary statistics compiled for PGEs presenting in alloys, sulfide and
chalcogen related compounds ....................................................................... 16
Figure 1-4 Summary of sulfide/silicate partition coefficients for PGEs ......................... 17
Figure 2-1 Photo and illustration of experimental tube assembly for Ge-Sb-S synthesis39
Figure 2-2 XRD scan for Ge glass products .................................................................... 40
Figure 2-3 BSE example of Ge glass products ................................................................ 41
Figure 2-4 Examples demonstrating proper/improper spiked sample for standard addition
analyses .......................................................................................................... 42
Figure 2-5 Results from HCl based solution ICPMS ...................................................... 43
Figure 2-6 Results from aqua regia + HNO3 based solution ICPMS .............................. 44
Figure 2-7 Part of the results from HF + HNO3 + Parr bomb based solution ICPMS,
using standard addition protocol .................................................................... 45
Figure 2-8 Complete set of results from HF+HNO3+Parr bomb solution ICPMS .......... 46
Figure 2-9 TRA spectra for homogeneous/heterogeneous scenarios .............................. 47
Figure 2-10 Homogeneity tests for different batches (RSD% vs sigma) ......................... 48
Figure 2-11 RSD% vs concentration in different batches ................................................. 49
Figure 2-12 Reference material testing on NiS4 and JBSulfide, standardized on Ge6...... 50
Figure 2-13 Reference material testing on silicate samples, standardized on Ge6 ........... 51
Figure 3-1 Capsule design for partitioning experiments ................................................. 89
ix
Figure 3-2 BSE demonstrating the preservation of FMQP in the run products .............. 90
Figure 3-3 Sulfur fugacity calibration curve ................................................................... 91
Figure 3-4 Fugacities and temperatures for each partitioning experiment ...................... 92
Figure 3-5 BSE of typical experimental products (MSS, melt, ISS) .............................. 93
Figure 3-6 Identification of ISS (a: by oxygen content; b: by major compositions) ...... 94
Figure 3-7 Measured MSS/melt partition coefficients of PGE compared with literature data
........................................................................................................................................... 95
Figure 3-8 Measured MSS/melt partition coefficients of chalcogens compared with
literature data ................................................................................................. 96
Figure 3-9 Evaluating the effect of having As in melt on PGE MSS/melt partitioning .. 97
Figure 3-10 Results of MSS/ISS partitioning experiments and factors affecting MSS/ISS
partition coefficients ...................................................................................... 98
Figure 3-11 Calculated ISS/melt partition coefficients for both PGE and chalcogens ... 100
Figure 3-12 Compilation of PGE solubility data in MSS ................................................ 101
Figure 3-13 MSS/melt partition coefficients for PGE vs M/S in MSS ........................... 102
Figure 3-14 MSS/melt partition coefficients for chalcogens vs M/S in MSS ................. 104
Figure 4-1 Figure from Mungall (2007) demonstrating previous models on Sudbury
crystallization process .................................................................................. 132
Figure 4-2 MIR vs fraction of liquid remaining for available experimental data ......... 133
Figure 4-3 Modeled trajectory compared with field measurements for selected elements
......................................................................................................................................... 134
Figure 4-4 Extrapolation of Pt and As solubility in sulfide melt for sperrylite to crystallize
......................................................................................................................................... 135
x
Figure 4-5 Examples form Dare et al. (2014) showing PGE zoning ............................. 136
Figure 4-6 Modeled trajectory for Sn compared with field data from Dare et al. (2014)
......................................................................................................................................... 137
Figure 4-7 Modeled trajectory for Zn compared with field data from Dare et al. (2014)
......................................................................................................................................... 138
Figure 4-8 Zoning texture examples from Dare et al. (2009) ....................................... 139
xi
List of Tables
Tabe 2-1 Synthesis history and major element compositions of different batches of
chalcogenide glasses. ............................................................................................... 52
Table 2-2 NIST610 and MSS5 reference values used to reduce data with LA-ICPMS
analyses. .................................................................................................................. 53
Table 2-3 Summary of LA-ICPMS results for Ge2............................................................... 54
Table 2-4 Summary of LA-ICPMS results for Ge3................................................................ 55
Table 2-5 Summary of LA-ICPMSresults for Ge4................................................................. 56
Table 2-6 Summary and Comparison of LA-ICPMS and solution ICPMS (HF + HNO3 + Parr
bomb) for Ge6........................................................................................................ 57
Table 2-7 Summary of HCl based solution ICPMS, for elements that standard addition can
be applied (ppm) on Ge6. ...................................................................................... 58
Table 2-8 Summary of HNO3 based solution ICPMS results for elements that standard
addition can be applied (ppm) on Ge6. .................................................................. 59
Table 2-9 Summary of recommended values of Ge6 ............................................................. 60
Table 2-10 Testing JBSulfide using Ge6 as the standard. ...................................................... 61
Table 2-11 Testing silicate reference materials using Ge6 as the standard. ........................... 62
Table 3-1. Nominal Composition of starting materials ........................................................ 105
Table 3-2. Composition of MSS5 ......................................................................................... 106
Table 3-3. Major element analysis by microprobe, in elemental weight percent ................. 107
Table 3-4. Trace analysis and partition coefficients for precious metals (ppm) .................. 110
Table 3-5. Trace analysis and partition coefficients for chalcogens (ppm) .......................... 113
Table 3-6. Trace analysis and partition coefficients for Zn, Sn and Pb (ppm) ..................... 116
xii
Table 4-1. Summary of parameters to calculate the initial sulfide melt composition at
McCreedy East ................................................................................................... 140
1
Chapter I. An Introduction to common magmatic Ni-Cu-
Platinum Group Elements (PGE) ore deposits
1.1 Significance of magmatic Ni-Cu-PGE ore deposits
Due to their high resistance to chemical attack, wear and tarnish, high melting temperature
and stable electrical properties, PGEs are becoming essential in several aspects of our
modern life: such as chemical reagents, jewelry, autocatalysts, electronics, dental restorative
materials etc. Sudbury Canada is the third largest Pt producer (~8930 kg/year) and Pd
producer (~26365 kg/year) worldwide (Johnson Matthey Platinum Market Report 2013). The
estimated reserve of PGE in Canada is 310,000 kg in 2007, which would last another 20
years if no more reserves can be discovered and the same productivity was maintained.
Especially in recent years, as demand arises for cleaner and more sustainable resources, fuel
cell related developments are imposing even more demand for PGEs. By comparison with
these increasing demands, our knowledge of these PGE ores appears quite limited. Although
2
small amounts of PGE can be found as alloys or metals in alluvial and placer deposits, PGE
are often commercially produced along with Ni-Cu deposits (e.g., Bushveld, Stillwater,
Sudbury, Noril’sk, Voisey’s Bay). Some of these PGE deposits contain less than 1–2%
sulfide minerals and tend to form laterally relatively persistent stratiform horizons in large
layered intrusions that are often relatively easy to trace once they have been intersected such
as the case of Bushveld and Stillwater. Other deposits containing more than 10% sulfide
minerals (e.g., Sudbury, Noril’sk, Voisey’s Bay) often tend to occur as melt sheet or irregular
orebodies in relatively small intrusions or at the base of komatiite, basaltic komatiite and
picrite lava channels, which makes them relatively difficult to locate (Naldrett 1981; Maier
2005). Therefore, a better understanding of the genesis of these deposits is becoming more
crucial for future exploration and is also the motivation for this study.
1.2 Saturation of a sulfide liquid
Researchers have in general agreed upon the scheme (Naldrett 1969; Naldrett 2004), where
exsolution of immiscible sulfide liquids from mafic-to-ultramafic magmas, once formed, can
be a "primary collector" for PGE as the first step. During Mid-Ocean-Ridge Basalt (MORB)
genesis when silicate melt is extracted from the mantle, if magma separates at 1~2.5 GPa, the
temperature along the adiabat would correspond to 1100 ~ 1400 °C. Over this range of P-T
conditions, silicate magma with a similar composition to MORB would be able to dissolve as
much as ~1750 ppm S (Liu et al. 2007). As this magma evolves in P, T and bulk composition,
the capacity to dissolve S would change accordingly. When the concentration of S remaining
in the silicate magma reaches the Sulfide Concentration at Sulfide Saturation (SCSS), a
3
separate sulfide phase will then exsolve. SCSS can be affected by several factors:
thermodynamics predict that temperature has a positive impact on SCSS, thus lower
temperature would result in lower SCSS and favor sulfide liquid immiscibility. This is also
confirmed by numerous experimental studies (Mavrogenes and O’Neil 1999; Jugo et al. 2005;
Liu et al. 2007; Li and Ripley 2009). As magma ascends, pressure will decrease as well,
which serves to increase SCSS (Mavrogenes and O’Neil 1999). Meanwhile, SCSS can also
be affected by the composition of the melt. Under the same P-T, magma with a more basaltic
composition would have SCSS ~10 times higher than the rhyolitic endmember, with FeO
being the main factor (Haughton 1974; Wendlandt 1982; Mavrogenes and O’Neil 1999; Liu
et al. 2007; Baker and Moretti 2011; Klimm et al. 2012). The presence of hydrous
components will also greatly enhance the capability of the silicate melt to accommodate
sulfur (Liu et al. 2007; Fortin et al. 2015). For a rough comparison, at 1250 °C 1GPa, the
SCSS would drop to 155 ppm for an anhydrous, more rhyolitic magma, instead of 1750 ppm
as suggested for a MORB like composition (Liu et al. 2007). The great decrease in SCSS
would eventually result in the sulfide saturation in the silicate magma. Once formed, droplets
of immiscible sulfide liquid would then tend to settle through less dense silicate magma.
Current knowledge on the partitioning behaviors of PGE and chalcogens between sulfide
liquid and silicate melt comes from three sources: 1) in situ measurements between silicate
magma and the sulfide droplet inclusions in natural samples (Peach et al. 1990; Patten et al.
2013). The uncertainties in these measurements usually arise from the small size of the
sulfide droplet and the difficulty of ensuring a “clean”, sulfide free silicate reference. 2)
4
Experimental studies measuring partition coefficients of PGE between sulfide melt and
silicate melt directly (Fleet et al. 1991; Bezmen et al.1994; Fleet et al.1996, 1999; Crocket et
al.1997; Peach et al. 1994; Li and Audetat. 2012). 3) calculated from metal solubility
experiments in sulfide and silicate melts respectively (Andrews and Brenan 2002; Fonseca et
al. 2009; Mungall and Brenan 2014). A brief summary of these experimentally-measured
partition coefficients is illustrated in Fig 1-1. The results span quite a wide range of more
than 100-fold between different studies. Apart from the difference in experimental conditions
and melt compositions, part of the origin of these differences in these experimental
measurements arises from the nano nugget effect, since small particles of metals or alloys at
nanometer scales can form in the experimental products, as suggested by Fortenfant et al.
(2006) for Os, and Bennett et al. (2014) for Pt. This was demonstrated by Mungall and
Brenan (2014), in which centrifuged experiments were found to have a lower Pt content (by
24% and 70%) compared with the static runs. This may also explain the discrepancy between
the calculated partition coefficients (Andrews and Brenan 2002; Fonseca et al. 2009;
Mungall and Brenan 2014) with some of the experimental measurements (Fleet et al. 1991;
Bezmen et al.1994; Fleet et al.1996, 1999; Crocket et al.1997). Despite the large difference
in the exact values of these partition coefficients, all these studies confirmed that PGE have a
strong preference for the sulfide melt over the coexisting silicate melt, on the order of 1000
~10,000,000. The combination of physical removal of dense sulfide liquid and chemical
concentration of the PGE in the sulfide liquid is considered to be responsible for forming
most economically viable, magmatic-sulfide deposits (Naldrett 1969; Naldrett 2004) as the
first step.
5
1.3 Crystallization of a sulfide liquid
Phase equilibria experiments on typical magmatic sulfide compositions predict the formation
of a crystalline Fe-rich monosulfide solid solution (MSS; [Fe,Ni]1-xS), which at 0.1 MPa
occurs at a Tmax of 1190 oC, corresponding to Fe0.917S (Jensen, 1942). The exact liquidus will
depend on pressure, Ni and Cu content and fS2/fO2 (Ebel and Naldrett, 1996; Bockrath et al.,
2004; Fleet and Pan, 1994; Naldrett, 1969). With cooling to a temperature less than 900 oC,
MSS is followed by a Cu-rich Intermediate Solid Solution ([Cu,Fe]1-xS; ISS; Dutrizac, 1976),
and magnetite. As temperature further decreases, the MSS-ISS assemblage would undergo
sub-solidus recrystallization to mostly pyrrhotite (Fe1-xS), pentlandite ((Fe,Ni)9S8) and
chalcopyrite (CuFeS2), as evidenced in natural magmatic sulfide samples. Within the Earth’s
crust, field studies provide evidence for efficient magmatic sulfide differentiation in
association with relatively large igneous bodies, as for the case of the Sudbury (Canada) and
Norilsk-Talnakh (Russia) Districts. Ores rich in Fe and Ni are interpreted as MSS cumulates,
and those which are Cu-rich are postulated as mixtures of evolved sulfide liquid and
cumulate ISS (e.g., Naldrett et al., 1982; 1992; Li et al., 1992; Zientek et al., 1994; Ballhaus
et al., 2001). A brief summary of this crystallization sequence is summarized in a simple
sketch of Fig 1-2.
Magmatic sulfide crystallization results in a significant separation of the HSE (highly
siderophile elements), with the IPGE (Ru, Os, Ir) and Re concentrated in the MSS cumulates,
and the PPGE (Pd, Pt, Au, Naldrett 2013) following the evolved liquid. In addition to the
evidence for large-scale magmatic sulfide differentiation, past studies of sulfide in upper
mantle peridotites and diamonds have also identified sulfide liquid fractionation, albeit on a
6
much smaller scale (Guo et al., 1999; Szabo´ and Bodnar, 1995; Alard et al., 2000; Lorand
and Alard, 2001; Luguet et al., 2001; 2004; Jenner et al. 2012). In that context, a distinction
is made between so-called Type 1 and Type 2 sulfides, using criteria and nomenclature from
Luguet et al (2001). Type 1 sulfides are characterized by high Ni relative to Cu abundances,
and primitive upper mantle (PUM)-normalized depletions in Rh and Pd relative to Ir (as well
as Ru and Os), and interpreted to be residual MSS. Type 2 sulfides have variable Ni/Cu, and
similar PUM-normalized abundances of Ir (Ru, Os), Rh and Pd, considered to be consistent
with trapped immiscible sulfide liquid. Hence, a better understanding of the partitioning
preference of the PGE will provide insight into the petrogenesis of sulfides both in crustal
and upper mantle environments.
1.4 Role of chalcogens in magmatic Ni-Cu-PGE deposits
The chalcogens considered in this thesis include elements in Group 15 and 16 of the periodic
table: As, Sb, Bi, Se and Te. Like the PGE, chalcogens generally have high metal/silicate
enrichment factors (greater than 10, McDonough 2003, Rose-Weston et al. 2009). As such,
more than 90% of the planetary inventories of the chalcogens are expected to be in the core,
leaving very low abundances in the silicate part of the Earth. Furthermore, chalcogens are
also considered moderately volatile, with the 50% condensation temperature at 10-4
atm to be
around 700 K (McDonough 2003). The estimated concentrations of chalcogens in the bulk
silicate earth are therefore quite low, with Se around 0.075 ppm, As around 0.05 ppm, Te
around 0.012 ppm, Sb around 0.0055 ppm, Bi around 0.0025 ppm (McDonough and Sun
1995, McDonough 2003). Despite their low concentrations, chalcogen compounds comprise
an important fraction of the world’s PGE ore deposits. Preliminary statistics (Figure 1-3)
7
indicate that, for reef type PGE deposits, chalcogen compounds can account for at least 12
vol% of all the PGE ores; while for magmatic breccia type deposits, chalcogen compounds
can account for up to 60 vol% of all the PGE bearing minerals.
Even when coexisting with sulfides, chalcogen-compounds are sometimes found to be more
capable of enriching the PGE. For example, Gervilla et al., (1996; 1998), Hanley (2007) and
Godel et al (2012) have reported a close textural relationship between relatively PGE poor-
base metal sulfide and coexisting PGE-rich arsenide phases (NiAs, nickeline; Ni11As8,
maucherite; NiAsS, gersdorffite) in the magmatic sulfide segregations within the Ronda and
Beni Besoura peridotite bodies, the Kylmakoski (Finland) Ni-Cu deposit and komatiite-
hosted base metal sulfide mineralization (Dundonald Beach South, Ontario; Rosie Ni
Prospect, Western Australia). Textures in these occurrences imply the presence of an
immiscible arsenide melt at the magmatic stage. Recent work on samples from Creighton
Mine (Sudbury, Ontario; Dare et al., 2010b) have shown that the base metal sulfides are not
the dominant hosts for some PGE, and that Ir, Rh, Pt (and Au) occur as chalcogen-bearing
discrete platinum group minerals (PGMs; i.e., IrAsS - RhAsS, PtAs2), possibly crystallizing
before or with MSS. Chalcogen-bearing phases are also associated with late-stage low sulfur
precious metal haloes around massive sulfide bodies as has been documented at various
locations around Sudbury, Ontario (Farrow and Watkinson, 1997), suggesting a role for these
elements as ligands for metal transport. There is also evidence for remobilized chalcogen-
rich melts associated with high grade metamorphism of base metal sulfides, such as the
Challenger gold deposit of South Australia (Tomkins and Mavrogenes, 2001). Thus
8
although the sulfide-PGE association is strong, the chalcogens may affect the distribution of
PGEs within a magmatic ore system as well.
In addition to the empirical studies mentioned above, there is also some limited high
temperature experimental work which has contributed to the understanding of the behaviors
of these elements in magmas. For example, some data are now available for the partitioning
of these elements between sulfide and silicate (Yi et al. 2000; Jenner et al. 2012; Li and
Audetat 2012; Patten et al. 2013; Brenan 2015), as summarized in Figure 1-4. As for the
PGE, the measured partition coefficients of the chalcogens also vary over a wide range.
Group 16 elements (Se and Te) appear to be strongly partitioning into the sulfide phase, with
D(sulfide/silicate) of ~1000, similar to Au, whereas As and Sb only exhibit slight preference
in the sulfide, and Bi appears to be mildly partitioned into the sulfide phase.
Relatively few experimental studies have investigated the behaviors of the chalcogens during
the internal differentiation of sulfide melts. Helmy et al. (2007) investigated the solubility of
Pt-Pd-tellurides in different sulfide phases and studied the partitioning behavior of Te
between MSS and sulfide melt, which was found to decrease with increasing Te content. It
was then suggested that Pt and Pd were strongly complexed by Te and thus stabilized in the
sulfide melt phase instead of MSS. Similarly, Helmy et al. (2013a) discussed the possible
formation of Pt-As nanoparticles in the sulfide liquid before a discrete Pt-As mineral
becomes a stable phase. Thus chalcogens may affect PGE distributions before forming any
discrete phases. Helmy et al. (2013b) focused on measuring the partition coefficients of PGE
between the sulfide and arsenide melt phases. D(sulfde/arsenide) for Pt was suggested to be
9
smaller than 2×10-5
, and for Pd less than 0.01. However, in order to ensure that As was
available as an anion for association with the other metals added to experiments, the fS2 in all
experiments was kept low (3 log units above Fe-FeS equilibrium). The possibility of a
magmatic arsenide melt phase and the applicability of these partition coefficients to natural
samples have not been verified yet. Helmy et al. (2010) focused on experimental
measurement of the MSS-sulfide melt partition coefficients for a suite of chalcogens (Se, Te,
As, Sb, Bi). All the chalcogens were found to be incompatible in MSS relative to the sulfide
melt, with Se being the relatively more compatible end member, Bi being the least
compatible one. Similar to the sulfide-arsenide melt partitioning work, these experiments
were also conducted at very low sulfur and oxygen fugacities (Fe-FeS, and Fe-FeO
equilibrium, respectively). The applicability of these data to natural systems is unclear. A
very fundamental question as to whether the PGE-chalcogen compounds observed in natural
samples are possible products of magmatic crystallization or evidence of post-magmatic
imprints is still not fully answered, as yet. Therefore, the primary goal of this study is to
further our knowledge of PGE partitioning during sulfide crystallization, quantitatively
assess the roles of chalcogens during PGE-ore petrogenesis, and refine our current models on
how the PGE ores form, all of which will unfold in the following chapters.
1.5 Contributions of this thesis
1.5.1 Author contributions
The experiments and run product analyses in Chapter II were completed by the author. The
original idea of the Ge-Sb-S glass in Chapter II was inspired by Ding et al. (2011). Carol
Cheyne helped with setting up the analytical software for solution ICPMS analyses. The
10
experiments, run product analyses, model construction and the programing in Chapter III and
IV were completed by the author. J.M. Brenan and the author both contributed to the design
of the study. Chapter III and part of Chapter IV have been published in Geochimica et
Cosmochimica Acta (Liu and Brenan 2015).
1.5.2 Measurement of PGE and chalcogens at trace concentration level ---- Synthesis of
a chalcogenide glass standard for laser-ablation inductively coupled plasma mass
spectrometry (LA–ICPMS)
A potential LA-ICPMS standard (Ge6), based on a chalcogenide glass, was developed and
tested against a wide range of reference materials. Compared with the original proposal of
Ding et al. (2011), the synthesis method developed here further extends the dopants to
include a complete set of PGEs (Ru, Os, Ir, Rh, Re, Pt, Pd, Ag, Au) and chalcogens (As, Sb,
Bi, Se, Te), together with some transition metals (Cr, Mn, Fe, Co, Ni, Cu, Zn) and some
refractory elements (Mo, W, Ga, Sn, Pb). Aside from Ru, Os, Re, Mo, W, all other dopants
could be homogenized up to the 100 ppm level, compared to the sub-ppm to ppm level
achieved by Ding et al. (2011). Ru, Os and Re can be homogenized up to 10 ppm, whereas
Mo and W can only be doped up to 5 ppm and remain homogeneous within the standard. The
digestion protocol using aqua regia as originally proposed by Ding et al. (2011) was found to
be inefficient for almost all the dopants studied in this project. Alternatively, a combined HF
+ HNO3 + Parr bomb digestion produced significantly improved results. Solution ICPMS
results agreed with values measured by LA-ICPMS within analytical errors, except for Os,
Te, Pt, Pd, Bi and Ir, likely due to the oxidizing effect of the HNO3 and the Te – co-
precipitation effect. A variety of standard reference materials were analyzed using Ge6 as the
11
standard, including those with sulfide (JBSulfide, NiS4) and silicate matrices (BIR-1, BHVO-
1, JGb-1a, JG-1a, JB-2). The returned concentrations were in most cases within error of the
accepted values, thus we conclude that any matrix mismatching effects are negligible.
1.5.3 Partitioning of PGE and chalcogens within sulfides under controlled fO2, fS2
conditions ---- Experimental measurements and origin of MSS-melt fractionation
Partitioning experiments were conducted for PGEs and chalcogens between MSS- sulfide
melt, and MSS-ISS under controlled conditions. Experiments were done at 0.1 MPa, over a
range of 850 °C to 930 °C. The oxygen fugacity was buffered at fayalite-magnetite-quartz
(FMQ), and the sulfur fugacity was monitored by the composition of coexisting pyrrhotite,
and calculated to be close to the Pt-PtS buffer. These conditions would correspond to a
environment similar to a sulfide melt coexisting with a silicate melt containing 10 wt% FeO.
D(MSS/melt) for PGE and chalcogens have been measured by previous studies (Fleet et al.
1993; Li et al. 1996; Brenan 2002; Mungall et al. 2005; Ballhaus et al. 2001, Li and Audetat
2012, Helmy et al. 2010). However, with the exception of Mungall et al. (2005), no other
studies buffered oxygen and sulfur fugacities. For the MSS-melt partitioning part of the study,
the PGE partition coefficients measured agreed well with Mungall et al. (2005) and
chalcogen partition coefficients measured agreed with Helmy et al. (2010) in general, despite
some differences in experimental conditions. The effect of chalcogens on PGE partitioning
was evaluated so as to better understand the finding of Helmy et al. (2013a), that the MSS-
melt partition coefficient for Pt is reduced significantly by arsenic complexing in the sulfide
liquid. We found that PGE partitioning was unaffected by the presence of As, or any other
chalcogens, which may be related to a lack of chalcogen anion species under the conditions
12
of our experiments. The difference in D(MSS/melt) amongst the PGE (IPGE, Rh, Re
compatible in MSS, Pt, Pt, Ag, Au incompatible in MSS) is interpreted to arise due to both a
decrease in metal activity coefficients with increased metal/sulfur in the MSS and
coordination complexes in the sulfide melt.
The D(MSS/ISS) for both PGEs and chalcogens are new data not previously reported, except
for the case for Au (Jugo et al. 1999). Our D(MSS/ISS) for Au agreed within error with the
pyrrhotite/ISS partition coefficient for Au reported in Jugo et al. (1999). The D(MSS/ISS) of
Group 16 chalcogens (S, Se, Te) do not display significant differences from their
D(MSS/melt), while the Group 15 pnictogens (As, Sb, Bi) displayed a nearly 10-fold
increase when compared with their D(MSS/melt). Combined with D(MSS/melt), a new set of
D(ISS/melt) were calculated for both PGE and chalcogens, which have not previously been
reported. All the PGE, Te, Sb and Bi have a similar preference for the sulfide melt phase
over the coexisting ISS, while Se and As do not appear to be fractionated by ISS
crystallization.
1.5.4 Compositional evolution of magmatic sulfide melt: models from partitioning
experiments applied to the McCreedy East Deposit, Sudbury, Ontario
Mungall (2007) attempted to model Ni vs Cu for the whole Sudbury Igneous Complex using
available D(MSS/melt). The high Ni (> 10 wt% Ni) and relatively low Cu samples (5~22 wt%
Cu) were interpreted as lying along a mixing line between ISS cumulate and pentlandite. In
this study, based on the newly measured partition coefficients, especially the new D(ISS/melt)
13
data, a 3-stage evolutionary model for sulfide melt crystallization is presented. The steps
involved are: I) MSS crystallization; II) MSS - ISS co-crystalllization; III) ISS only
crystallization. In Stage II where both MSS and ISS are coexisting, a new parameter defined
as the MSS ISS Ratio (MIR), defined as the weight ratio between MSS and ISS, is proposed
to simplify the calculations. A correlation between MIR and fraction of the residual sulfide
liquid was established based on the regression of the available experimental data of Fleet and
Pan (1994) and this study. Once the model was constructed, the magmatic sulfide ores from
the McCreedy East Deposit (Sudbury) were selected as a test of the model, since a relatively
comprehensive dataset is available in literature for both PGE and chalcogens (Dare et al.
2011, 2014). The initial sulfide composition was selected by adopting the approach of
Mungall (2004) based on estimates for the composition of the Sudbury Impact melt sheet and
silicate/sulfide mass ratios (R-factor). The boundary points between adjacent stages were
selected in reference to the other available phase equilibrium studies (Fleet et al. 1993, Ebel
and Naldrett 1996, Naldrett et al. 1999). Modeled evolutionary curves match the raw data
reasonably well, capturing both the concentration level of MSS cumulates and the relatively
more Cu rich samples (ISS cumulates and liquids). Combined with the available literature
data on PGE and chalcogen solubilities in sulfide melt, a tentative crystallization sequence
for PGE- and chalcogen-rich accessory minerals is also proposed. This sequence agrees with
the textural evidence provided in Dare et al. (2014).
14
Figure 1-1 Summary of partition coefficients between sulfide melt and silicate melt for
PGE from the literature. Data from Peach et al. (1990) and Patten et al.
(2013) are based on in situ measurements of natural sulfide droplets and
silicate matrix, the rest are from laboratory studies at controlled conditions.
15
Figure 1-2 Illustration of sulfide liquid crystallization: a) after sulfide melt separated,
monosulfide solid solution (MSS) first crystallizes; b) as temperature
decreases, MSS and intermediate solid solution (ISS) both crystallize as
solid phases, together with some magnetites; c) as temperature further cools,
the crystallized solid cumulates will further break down into chalcopyrite,
pentlandite and pyrrhotite, more details in the text.
a) T < 1190 °C
b) T < 900 °C
c) T < 600 °C
16
Figure 1-3. Preliminary statistics compiled from literature for the compositions of PGMs
in various types of ore deposits. Merensky Reef and J-M Reef represent the
“reef type” PGE ore deposits; Platreef Sandsloot, Platreef Potgletesrus and
Lac des Iles represent the “magmatic breccia type” PGE ore deposits.
Data (in vol%) are compiled from: Merensky Reef -- Kinloch & Peyerl (1990),
average of Rustenburg, Union and Amandelbult sections; J-M Reef -- Godel
and Barnes (2008); Platreef-- Sandsloot data from Howell et al. (2006);
Potgietesrus data from Kinloch (1982); Lac des Iles -- Edgar and Sweeny
(1990). Alloys include Pt-Fe alloys, Pd alloys, Ag-Au electrum, etc. Sulfides
include PGE-S compounds; Chalcogen compounds include tellurides,
arsenides, sulfasenides, and other compounds containing chalcogens as major
elements.
17
Fig 1-4 Summary of literature data on partition coefficients between sulfide melt
and silicate melt for chalcogens. Values vary over a factor of 100. Compared
to the PGE, Se, Te and Bi, the elements As and Sb do not appear to be
significantly favoring sulfide melt over the silicate melt, as partition
coefficients range from <1 to ~10.
18
Chapter II. Measurement of PGE and chalcogens at trace
concentration level
---- Synthesis of a chalcogenide glass standard for laser-ablation
inductively coupled plasma mass spectrometry (LA–ICPMS)
2.1 Introduction
The concentrations of platinum group elements (PGE) and chalcogens (including Se, Te, As,
Sb, Bi) in common base metal sulfides (pyrrhotite, pentlandite, chalcopyrite, etc.) are often
too low for analysis by common in situ methods, such as Electron Probe Micro-analyzer
(EPMA). The advent of LA-ICPMS has significantly improved the analytical capability for
these elements, with detection limits down to ppb levels. Thus it is now feasible to
investigate a great variety of geological materials at a much finer scale without too much
sacrifice in spatial resolution. However, the ablative nature of this technique results in the
continuous consumption of standards which are essential for calibration. A sustainable
19
supply of homogeneous and laser-stable standards with ppm concentrations has thus become
highly sought by the geological community. Several approaches have been utilized to
synthesize such standard materials. Ballhaus and Sylvester (2000) synthesized Fe1-xS
monosulfide polycrystalline aggregates doped with Ir, Ru, Rh, Pt and Pd added as chloride
solutions to FeS + S powders at concentrations of around 5~10 ppm. The doped starting
material was annealed in silica glass capsules at 950 ºC, 1 GPa using a piston cylinder
apparatus. Cabri et al. (2003) synthesized Ru, Rh and Pd-doped sulfide standards using
evacuated silica glass ampoules at 0.1 MPa by initially melting starting materials at 1205oC
followed by slow cooling to 1000oC and prolonged annealing. Run products were cross
calibrated against the Ballhaus-Sylvester standards. A similar synthesis strategy was
employed by Sylvester et al. (2005) and Barnes et al. (2006). A potential shortcoming of this
method is that it is limited to elements that form a homogeneous solution with Fe-sulfide at
concentrations producing count rates suitable for precise calibration.
As an alternative, the possibility of cold pressing powders was explored by other researchers.
Perkins et al. (1997) doped sulfide powders with multi element solutions and pressed the
powders to pellets using polyvinyl alcohol as a binder. Wilson et al. (2002) describe a similar
approach, with starting powders produced by precipitating an amorphous Fe-Cu-Zn sulfide
by reducing metal-bearing multi-element sulfate solutions. However, the standards pressed
in these methods are likely to have different ablation characteristics than single phase
materials, and it is difficult to guarantee the homogeneity. Wohlgemuth Ueberwasser et al.
(2007) revisited the pellet synthesis strategy using a binder-free, high pressure-temperature
synthesis method. The dopants (up to 60 ppm) were added as solutions by micro syringe to
20
reagent powders, then transformed into sulfide by synthesis in a silica tube at 0.1 MPa by
progressive heating to 900oC. The resulting material was powdered, then sintered at 900-
1200oC and 1.5 to 2 GPa to obtain pellets with the theoretical density. Although this
approach produces homogeneous, dense material, it still requires hot pressing in a piston
cylinder, which thus limits the amount of material that can be produced.
Recently, Ding et al. (2011) investigated the use of a chalcogenide glass (atomic proportions
Ge28Sb12S60) as a standard for a large number of minor and trace elements, including the
suite of interest here. Whereas base metal sulfide melts quench to an inhomogeneous array of
dendritic crystals, the addition of Ge and Sb results in a glass-forming composition, but with
similar total sulfur as in naturally-occurring sulfide minerals. Moreover, the synthesis
method involves melting and cooling in evacuated silica ampoules at 0.1 MPa and relatively
low temperature (~ 950oC), which significantly simplifies the experimental procedure and
equipment requirements, allowing for large amounts (e.g., gram levels) of material to be
produced. Although the doped element suite investigated by Ding et al (2011) was extensive
(Na, As, Ba, Bi, Br, Cr, Cd, Cl, Co, Ga, Au, Mn, In, I, Ir, Pb, Hg, Mo, Ni, Os, Pd, Pt, K, Cs,
Re, Rh, Se, Ag, Sr, Te, Tl, Sn, V, W and U), absolute concentrations of some elements in the
final material were low (< 1 ppm in most cases), making it a poor standard for elements with
low sensitivity, and some elements (e.g., Ir, Os, K, Cs, Re, Sn, and U) were inhomogeneous
(>40% variation amongst multiple analyses). In the current study, we have further explored
the possibility of this material as a standard by investigating the appropriate concentrations
and synthesis conditions for achieving sample homogeneity. As well, we have assessed the
utility of the chalcogenide glass as the standard by measuring the concentration of a suite of
21
trace elements in reference materials including silicates and sulfides whose composition has
been previously determined.
2.2. Synthesis methods
As per the approach of Ding et al (2011), our general strategy was to initially synthesize a
base metal sulfide which incorporates all the elements of interest at relativey high
concentrations. The rationale behind this step is to make use of the generally high diffusion
rates of the dopants in crystalline and molten sulfide, which produces a reasonably
homogeneous starting material. Small amounts of the enriched sulfide could then be mixed
together with a large quantity of Ge, Sb and S powders, and eventually melted into a glass,
producing final dopant concentrations at the trace level. Element concentrations can be easily
adjusted by changing the ratio between the amounts of doped sulfide and the dopant free Ge-
Sb-S mixture.
2.2.1 Doped sulfide synthesis
The doped sulfide composition employed by Ding et al. (2011) contained Fe-Cu-Zn and
sulfur (bulk composition not reported), and was synthesized by precipitation from trace
element doped hydrated sulfate salts (following Wilson et al., 2002). A problem with this
method is that relatively low levels of the other trace element dopants were precipitated with
the sulfide, limiting the amount that could be added to the chalcogenide glass starting
material. In the work reported here, synthesis of the doped sulfide was done by melting in a
22
sealed silica ampoule, which avoids these mass balance issues. The bulk sulfide composition
chosen was guided by the Fe - Ni+Cu - S phase diagram of Li et al. (1996), staying well
within the phase field for monosulfide solid solution (MSS) at the final run temperature.
Elemental Fe, Ni, Cu, Mn and S powders were carefully weighed, mixed and ground under
ethanol for 30 minutes Once dried, the mixture was loaded into a 6×10 mm silica tube,
tightly packed to approximately 15 ~20 mm high, and a silica spacer was then laid on top.
The whole assembly was connected to a vacuum line for about one hour, and sealed above
the spacer with an Acetylene-oxygen torch. The charge was then placed into a box furnace at
600 ºC, for at least 12 hours to anneal. The box furnace was independently calibrated for hot
zones in 3 dimensions before use. The measured temperature variation within the hot zone is
±5 ºC. After annealing, the furnace was gradually raised to 1000 ºC over 5 hours, kept at
1000 ºC for 2 hours to ensure complete melting, then cooled down to 600 ºC over 6 hours.
The assemblage was then taken out of the furnace and cooled in air. The Cu content of the
sulfide synthesized initially (MSSA) was found to be above the desired level, so we
synthesized a Cu-free sulfide mixture (MSSB), following exactly the same procedure
described above, which was then added to the Cu-rich MSSA, in order to dilute the Cu to the
target concentration as MSSC.
2.2.2 Dopants addition
Once this sulfide matrix was made, a suite of dopants could then be added as elemental
powders, including Ru, Os, Re, Ir, Rh, Pd, Pt, Au, Ag, Se, As, Bi, Te, Cr, Mn, Co, Ga, W,
Mo, Pb. The goal of this first round of doping was to mix approximately 5 mg of each dopant
23
together with 900 mg MSSC, corresponding to around 0.5 wt % of each dopant. The doped
MSSC mixture was again loaded into a 6×10 mm silica tube packed to approximately 30 ~
40 mm high. A tightly fit, high purity silica rod with similar length was positioned on top of
the mixture and evacuated for one hour until sealed with the Acetylene torch. The sealed
silica tube was put in a clean alumina crucible, carefully positioned into the hot zone of the
fisher box furnace and then melted following the same procedure as MSSA and cooled in air.
The sample retrieved from the silica tube was ground under ethanol for another 30 minutes,
stored as MSSD.
The chalcogenide glass proposed by Ding et al. (2011) has a stoichiometric composition of
Ge:Sb:S = 28:12:60, which was suggested to have an excellent glass forming capability. In
order to make 10 g of this glass, 3.752 g Ge, 2.697 g Sb and 3.551 g S elemental powders
were weighed and mixed with 0.2 mg MSSD. The whole mixture was ground and mixed
under ethanol for 45 min and dried under a heating lamp to serve as the starting material.
This starting material was again loaded into a 6×10 mm silica tube, evacuated and put into
the Fisher box furnace at a temperature of less than 200 ºC. Once the temperature stabilized,
the furnace was then heated up in incremental steps first to 250 ºC at a rate of 2.86 ºC/min,
held at 250 ºC for 30 min, and heated to 450 ºC at a slower rate of 1.67 ºC/min. At 450 ºC the
furnace was held for another 30 min, and heated to 650 ºC at the same rate of 1.67 ºC/min,
and held at 650 ºC for 30 min. From 650 ºC and above, the temperature was raised at a rate
of 1.2 ºC/min and held at the melting temperature over a fixed duration to allow the doped
elements to diffuse throughout the whole sample. Once the sample was completely melted
and homogenized, the furnace was then cooled to 850 °C at a rate of 0.5 ºC/min and held at
24
850 ºC for another 2 hours to prevent bubble formations (Ding et al. 2011). The whole
charge was then taken out of the furnace and quenched in cold salty water. After quenching,
the charge was returned to the furnace around 220 ºC, to anneal for 2 h to remove the inner
stress induced by the quenching. In total 4 rounds of synthesis were conducted. The
differences in the heating protocols among different runs are listed in Table 2-1. A typical
quenched charge is shown in Figure 2-1. The silica tube was cut open by a diamond saw and
the dark brown, glassy boule was then retrieved. Due to its brittle nature, the glass samples
from our experiments usually broke into large pieces (0.5~1 cm long, 0.5 mm wide), instead
of a whole piece as shown in Ding et al. (2011). Random chips of the retrieved glass were
then handpicked and mounted in a one inch diameter epoxy mold, ground on 600 mesh SiC
sand paper, and then directly polished with 0.3 um alumina powder using H2O as the
lubricant.
2.3. Characterization of the synthesized material
2.3.1 General Aspect
Part of the unmounted glass chips were ground into fine powders and characterized by X-ray
diffraction (XRD) to confirm the structure. The XRD scan was conducted at the Department
of Earth Sciences, University of Toronto. A traditional Cu X-ray tube was used and coupled
with a Ni filter between the X-ray source and the sample. A zero background silica holder
was employed to minimize the interference from the holder material. Scanning angles ranged
from 15 ° to 45 ° (measured in 2θ), with 0.02 ° per step and 1.5 sec dwell time at each step,
since all the major reflection peaks of the Ge-Sb-S compounds are within this range. The
25
resulting XRD pattern is shown in Figure 2-2, consistent with a normal distribution, short
range order, and amorphous structure. This is also in accordance with the previous structure
study by Li et al. (2014) using Raman spectroscopy.
The glass chips mounted into a one inch diameter epoxy puck was later carbon coated and
examined using the JEOL JSM-6610 LV scanning electron microscope at Department of
Earth Sciences, University of Toronto. The backscattered electron image (BSE) in Figure 2-3
visually verified the homogeneity of the sample to a first approximation. The BSE image was
taken at magnification of 160 ×; with an accelerating voltage of 15 kV, spot size of 50 nm
and a working distance of 16 mm. Increased magnification up to 10000 revealed neither
exsolution nor crystalline textures.
2.3.2 Major and Minor elements distributions
To quantitatively assess the homogeneity, major and minor elements were analyzed by
Cameca SX50 Electron Microprobe at Department of Geology, University of Toronto. An
accelerating voltage of 20 kV, beam current of 35 nA, and a defocused beam of 10 um were
used to analyze elements of Ge, Sb, S, Fe, Cu and Ni. Counting time was set to be 20 sec on
peak, 10 sec on each side of the background for major elements such as Ge, Sb, and S. For
minor elements such as Fe, Cu and Ni, counting times of 40 sec on peak and 20 sec on
background were used. Elemental Ge, a synthetic pentlandite, synthetic CuSbS2 and a natural
tetrahedrite were used as the standards for Ge, Fe, Ni, Cu, Sb and S. To better assess the
spatial distributions of these major and minor elements, linear traverses were conducted on
26
different glass shards, with 50 point analyses along each line. Sb and S are assessed to be
homogeneous within 0.1% precision, Ge is homogeneous within 0.2%, and Fe is
homogeneous within 0.4% precision (Probe for EPMA, Donovan 2013). The major element
compositions of all the run products are listed in Table 2-1.
2.3.3 Trace element distributions
The mounted glass shards were then analyzed by LA-ICPMS to test for homogeneity at the
trace element level. Analysis was performed at the Department of Earth Sciences, University
of Toronto. The system employs a VG PQ-EXCELL ICP-MS and New Wave UP-213 laser
for high spatial resolution sampling. Helium was used as the carrier gas to transport the
ablation aerosol from the sampling cell to the plasma. Factory-supplied time resolved
software was utilized for the acquisition of individual analyses. A typical analysis involved
20 seconds of background acquisition with the ablation cell being flushed with He, followed
by laser ablation for 60 seconds. A frequency of 10 Hz, beam size of 75 um, 55% output
power were used on the laser. Multiple linescans, each about 150 um long, were collected on
these randomly picked glass shards. Data reduction was done off-line using the GLITTER
version 5.3 software package, supplied by Macquarie Research, Ltd. Cr, Co, Ga, Mo, W
were reduced using NIST 610 as the standard, while others were reduced using an in-house
synthesized and cross-calibrated mono-sulfide (MSS5) as the standard (Table 2-2). 57
Fe was
selected as the reference element. The LA-ICPMS results of different runs are listed in Table
2-3 (Ge2), Table 2-4 (Ge3), Table 2-5 (Ge4), Table 2-6 (Ge6). The standard deviations (STD)
were calculated for multiple linescans measured in each sample using the relation:
27
1
)( 2
n
XXSTD i (Eq. 2-1)
where Xi is the measured concentration in each linescan for a given element; X is the
averaged concentration of all these multiple linescans in the same sample for the same given
element; n is the number of linescans conducted for the same sample. The 1σs were
calculated by GLITTER based on the propagation of the counting statistics uncertainties. In
GLITTER, the concentration of each element i (Conci) is calculated according to the
following equation:
jij
ij
iyieldAbundance
CPSConc
1 (Eq.2-2)
where CPS represents the mean count rate, counts per second of the measured isotope j of
element i (background subtracted); abundanceij represents the natural abundance of isotope j
of element i; yield represents the CPS per ppm for the measured isotope j, and is calculated
as:
std
is
j
isjyield
yieldINTyieldyield
(Eq. 2-3)
where yieldis is the yield of the element used as internal standard, also referred to as the
reference element;
std
is
j
yield
yieldINT
is the ratio of the yield of isotope j to the yield of the
internal standard element is, interpolated over the standard analyses. To calculate the one
sigma error, GLITTER uses counts to estimate the uncertainties on the signal and
background counts, propagated through the equations. An assumed 1% uncertainty (relative)
28
was used on all the elemental concentrations of the reference material, and a 3% uncertainty
(relative) on the values of the internal standard element in the unknown sample, and then
propagated throughout the calculations.
Once the homogeneity of the synthesized glass was established, the sample was then
analyzed by solution ICPMS to verify the concentrations of each dopant. A clean Savillex
container was first dried and weighed with a Mettler Toledo Analytical balance with a
resolution of 0.01 mg. A glass shard weighing around 37 mg was rinsed with trace level
clean HNO3 (Alfa Aesar) and then deionized H2O (17.8 MΩ). Once dried, the chip was then
digested within the Savillex container. Different digestion protocols were assessed in order to
find the optimal approach:
A) aqua regia + HCl
Freshly made aqua regia was added to the Savillex container together with the weighed
sample glass chip. A hotplate was positioned in a pre-cleaned fume hood. The whole
container was then placed on the hot plate with the cap tightened at around 100 ~120 °C for 3
days. An orange-brown colored solution was achieved with no obvious signs of residues left
at the bottom of the container. The container cap was then loosened but kept covered, and the
whole container was kept on the hotplate at around 80 °C until dried. Diluted HCl (1 mol/L)
was used to dissolve the dried residue at the bottom of the container. The resulting solution
was then further diluted in a pre-weighed 25 ml volumetric flask with the same diluted HCl
to produce the sample used in solution ICPMS analysis (hereafter, the unknown solution).
29
A standard addition protocol was employed whenever the standard solutions with matching
matrix were available. For the HCl based unknown solution, “Precious Metals, plasma
standard solution, Specpure®, Au, Ir, Os, Pd, Pt, Re, Rh, Ru @ 100µg/ml, Matrix 20% HCl”,
“Semi Metals, plasma standard solution, Specpure®, As, Bi, Ga, Ge, In, Pb, Sb, Se, Sn, Te,
Tl @ 100µg/ml, Matrix 20% HCl”, “Refractory Metals, plasma standard solution,
Specpure®, Al, B, Cr, Hf, Mo, Nb, Si, Ta, Ti, V, W, Zr @ 100µg/ml, Matrix 5% HCl” were
used for making the spikes. Each of the standard solutions was first diluted to 1 ppm using
the same HCl used for making the unknown solution. The same proportion of each of these
diluted standard solutions was then weighed and mixed into the same container and stored as
the spike solution. For each set of samples, 5 falcon tube solutions were prepared:
1) Tube 1: 5 mg 1 mol/L HCl, acting as the blank sample;
2) Tube 2: 5 mg unknown solution with no spike;
3) Tube 3: 5 mg unknown solution with 6 ul spike solution;
4) Tube 4: 5 mg unknown solution with 12 ul spike solution;
5) Tube 5: 5 mg unknown solution with 18 ul spike solution.
To compensate for the possible concentration range among different dopants, part of the
unknown solution was further diluted 100 times with the same 1 mol/L HCl and then spiked
with the same procedure as described above. All dilutions were conducted by weight to
better ensure the consistency and reduce measuring errors.
30
The resulting solutions were then analyzed using a Thermo X Series II Solution ICPMS in
the Earth Sciences Department, University of Toronto. The same 1 mol/L HCl used to take
up the unknown solution was also used as the washing solution between different falcon
tubes. Each solution was measured 3 times and only the averaged results were used. During
the analysis, the count rate for both the spiked and unspiked solutions were monitored
against the calculated additional concentrations (Figure 2-4a). Anomalies falling outside of
the regressed line were excluded, as demonstrated in Figure 2-4b. Concentration data were
then reduced using the factory supplied software.
The solution made using this Approach A, however, was found to be unstable. Flocky
precipitates can form one day after the solution was made. Solution ICPMS results of these
solutions (Table 2-7) were compared with laser ablation ICPMS results, as shown in Figure
2-5, which indeed exhibited consistent underestimation for almost all the dopants. Therefore,
we considered this digestion approach to be ineffective.
B) aqua regia + HNO3
The flocky precipitates in HCl based solutions were suspected to be related to the insoluble
nature of certain metal chlorides (AgCl, PtCl2, etc.). The comparison in Figure 2-5 seems to
suggest that such phases not only precipitate out of the solution themselves, but may also
absorb other dopants. To avoid having HCl as the matrix, we modified the previous approach
by using aqua regia and diluted nitric acid instead. A carefully cleaned and weighed glass
chip was first digested in a Savillex container on a hotplate using aqua regia. Once dissolved,
31
the container was dried on a hotplate, and the residue taken up by 2 wt% HNO3 (trace level
clean) into a 25 ml volume of solution. The resulting solution appeared to be clear in color,
and no significant precipitates were observed after 2~3 days. For the HNO3 based solutions,
VWR multi standard 10 ppm solution (Mn, Ag, Cu, Ni, Ga, As, Co, Cr, Se) was employed in
making spikes. The standard solution was first diluted into a one ppm solution using the
same 2 wt% HNO3 mentioned above, and then the same spiking recipe was employed as the
HCl based sample series described above. The same 2 wt% HNO3 was also used as the blank
sample and washing solution between different solution tubes.
This digestion procedure was also found to be unsatisfactory. When compared with the LA-
ICPMS results as in Figure 2-6, solution ICPMS results (Table 2-8) still appeared to be
consistently underestimating all the spiked elements, which was considered as an indication
of incomplete digestion. The potential problem was then speculated to be the HCl involved
in the aqua regia digestion step, which means that the insoluble PGE-chlorides may not have
been transferred completely into the unknown solution initially.
C) HNO3 + HF + Parr bomb
Avoiding the HCl in aqua regia, the best digestion protocol was found to be a combination of
HNO3 + HF + Parr bomb. For this approach, the sample glass chip was first digested using
HNO3 : HF (volume ratio 4:1) in a Parr bomb at 200 oC for two days. A clear colorless
solution was achieved. The solution was subsequently dried on a hotplate at ~ 80 oC to
completely drive off the HF, and then taken up in 2 wt% HNO3. The same HNO3 based spike
32
solution was used as in Approach B. This method turned out to be most efficient in digesting
the Ge-Sb-S glass. The results for the spiked elements were again compared with laser
ICPMS results, as illustrated in Figure 2-7. All the measured elements appear to be consistent
with laser ICPMS measurements within analytical uncertainties. For the elements that
standard addition protocol can not be applied to due to the lack of a matrix matched standard
solution, a calibration curve was established instead for each analyte. The spiking solution
employed in Approach A was further diluted to the concentrations of 0.5 ppb, 1 ppb, 5 ppb,
10 ppb, 50 ppb respectively, with deionized water. Analyzed results from both the standard
addition protocol and the calibration curve protocol are summarized in Table 2-6. All the
solution ICPMS results agree with the LA-ICPMS results within analytical errors, except for
Os, Ir, Pt, Pd, Te, Bi (Figure 2-8). Os is notorious for its volatile character (Sun et al. 2001),
while the other mismatched elements in the list might be in part due to the oxidizing effect of
HNO3 and the co-precipitation effect of Te (Savard et al. 2010). Thus the concentrations of
these elements are not yet as well constrained by solution ICPMS.
2.4 Discussion and application
2.4.1 Assessment of homogeneity and factors affecting homogeneity
The Ge-Sb-S glass matrix has been widely used in various fields, such as spectroscopy,
microscopy, astronomy, biology, and sensing (Yi et al. 2014). The glass forming capability is
relatively robust and insensitive to the major element compositions. The exact formula in our
synthesis experiments varied among different batches from 24 to 33 atoms for Ge, and 10 to
14 atoms for Sb, calculated based on 60 S atoms (Table 2-1). Yet no significant impact can
33
be observed as to exsolution texture or crystalline features. Electron microprobe analyses
revealed no gradient in major element distributions along the linear traverses done in samples
from all the batches.
As to the trace element distributions, homogeneity can be evaluated in two ways: 1) Time
resolved analysis during a long laser ablation traverse can provide first order filtering for
evaluating the homogeneity of a sample. Examples of time resolved raw counts plots are
shown on a log scale as in Figure 2-9. Whereas 75
As in both Figure 2-9a and Figure 2-9b
appear homogeneous, 185
Re appears heterogeneous in Figure 2-9a. 2) A more rigorous test of
the homogeneity was done by comparing the standard deviations of multiple analyses
(RSD%, relative standard deviation, Eq. 2-1) with the uncertainties in signals (1σ %) reduced
directly by GLITTER (Eq. 2-2), as introduced in section 2.3.3. A sample would thus be
regarded as homogeneous by definition if RSD% is no larger than the 1σ %, as demonstrated
in Gilbert et al. (2013).
The homogeneities of each different run assessed this way are illustrated in Figure 2-10 a-d.
Most of the dopants appear to be homogeneous in all four runs, except for Ru, Os, Re, Mo,
W which are only homogeneous in Ge6. A closer look at these 5 elements among different
runs can be found in Figure 2-11, with relative standard deviation (RSD%) calculated based
on multiple analyses, versus the concentration of that element in each run. Figure 2-11
enabled us to speculate on the possible factors affecting the homogeneity of these 5 elements.
All the 5 elements displayed no significant difference in RSD% between Ge2 and Ge3. As
34
listed in Table 2-1, the only difference between these two runs is the melting durations (10 h
vs 30 h). Lack of significant improvements in homogeneity indicates that increasing melting
time is not effective in promoting homogeneity while synthesizing this standard material.
On the other hand, a significant drop in the RSD% was found for Ru and Os from Ge3 to
Ge4. The main differences between Ge3 and Ge4 include: 1) Melting temperature was
increased from 950 ºC (Ge3) up to 975 ºC (Ge4), since Ru and Os homogeneity may have
been hampered by their relatively slow diffusion rates (Brenan 2002); 2) After melting for
10h, the Ge4 charge was immediately quenched. Additionally, the bubble prevention step
involving first cooling to 850 ºC and remaining at 850 ºC for 2 h as originally proposed in
Ding et al. (2011) was skipped for Ge4; and also 3) No annealing was done after quenching.
The original motivation of having the bubble prevention step and annealing step in Ding et al.
(2011) was to reduce the internal stress of the product glass and achieve a single, intact glass
rod as long as 2 cm. Since the standard is mainly developed for in-situ LA-ICPMS
applications, which is not particularly sensitive to the sample size at the centimeter scale, we
chose to skip the stress reducing steps so as to shorten the overall synthesis duration and
reduce the risk of introducing local crystallization at low temperatures. Importantly,
increasing the melting temperature appears to be effective in improving the homogeneity for
Ru and Os.
Despite these efforts, the inhomogeneity of Re, W and Mo did not significantly improve by
either prolonged duration or higher melting temperature, which indicated that the observed
35
heterogeneity may not be controlled by a diffusion related mechanism. For comparison, Ding
et al. (2011) diluted Mo to 0.2 ppm and W to 0.08 ppm. Their reported RSD% is 18.07% for
Mo, and 20.05% for W. Re was also reported to be heterogeneous by Ding et al. (2011) with
RSD% of 40%, although the doping concentration for Re was not reported. But based on the
other dopants level, it might be well below 1 ppm. Since Ge4 has 1.6 ppm Mo, 3.8 ppm Re
and W, there is a possibility of oversaturating these dopants in the matrix. This result is
similar to the findings of Diliberto et al. (2002), who found that the use of Ru as a sintering
agent can induce heterogeneity in a MoW alloy, and longer sintering time or higher sintering
temperature does not seem to improve homogeneity. Although the cause of this phenomenon
is not well understood yet, their description seems similar to the observations from the
synthesis reported here. Therefore, to improve the homogeneity for Mo and W, we suspected
that we might have to control the Ru concentration, as well as reducing the dopant level for
Mo, W and Re. We thus further diluted Ge4 approximately 8 times to produce Ge6, and Re,
Mo and W all responded favourably to this change (Figure 2-11). Although Ru and Os can
be homogenized at concentrations close to 30 ppm, lower Ru concentration may have played
a role in promoting the homogeneity of Mo, W and/or Re. Thus the dopant level for these
elements (Ru, Os, Re, W, Mo) should be limited to levels of less than 10 ppm.
2.4.2 Application ------ Testing reference materials
To better evaluate the applicability of the GeSbS glass as a standard material for materials
with different matrices we measured the trace element content of a series of reference
materials using Ge6 as the standard (best estimated values listed in Table 2-9). These include
36
sulfides (NiS4, JBSulfide), as well as silicate glasses (JB-2, JGb-1, JG-1a, BIR-1 and BHVO-
1) prepared by melting rock standards. NiS4 and JBSulfide are both in-house synthesized
sulfide standards. NiS4 contains 1988 ppm Se, 1930 ppm Sb, 2140 ppm Te, 1349 ppm Pb,
1.26 wt% As, 1349 ppm Bi and 1.26 wt% Cu (calibrated by electron microprobe, Brenan
2015). JBSulfide is a Fe-Cu-doped NiS bead containing 260 ppm Ru, 247 ppm Pd, 302 ppm
Os, 315 ppm Ir, 237 ppm Rh, 95 ppm Re and 294 ppm Pt (calibrated as described in Mungall
and Brenan 2014). Results for NiS4 and JBSulfide are shown in Figure 2-12, using 57
Fe as
the reference element, and the original data can be found in Table 2-10a - b. BIR-1 and
BHVO-1 are part of a suite of USGS reference materials. BIR-1 came from one of the
interglacial lava flows often referred to as Reykjavik dolerites, containing trace amounts of
REE and transitional metals (BIR-1: 368 ppm Cr, 51.1 ppm Co, 115 ppm Cu, 72 ppm Zn).
BHVO-1 came from a basaltic lava from Kilauea caldera, Kilauea volcano, Hawaii. It is also
a basaltic glass containing 127 ppm Cu, 103 ppm Zn. JGb-1, JB-2, JG-1a are part of a suite
of Geological Survey of Japan (GSJ) reference materials (Imai et al. 1995). JGb-1 is a gabbro
containing 59.3 ppm Cr, 111 ppm Zn, 1.11 ppm As, 1.92 ppm Pb. JB-2 is a basaltic glass
from O-shima volcano in Japan, containing 27.4 ppm Cr, 110 ppm Zn, 227 ppm Cu, 2.98
ppm As, 5.4 ppm Pb. JG-1a is from granitic rock, containing 38.8 ppm Zn, 1.3 ppm Cu, 0.39
ppm As, 27.0 ppm Pb. All these rock standards originally obtained as powders have been
melted at 1400 °C, 1 GPa in high purity graphite to homogenize for LA-ICPMS. For these
silicate glass reference materials, the comparison results between measured values and
accepted values based on different elements are illustrated in Figure 2-13, and the original
data can be found in Table 2-11. Although the matrix of these silicate glasses can be quite
different from the Ge6, the test results are in general in good agreement with the accepted
37
values. Therefore, based on these test comparisons, it appears that Ge6 could be used as a
standard to measure both silicate and sulfide samples.
2.5 Conclusion
These synthesis experiments confirmed that Ge-Sb-S mixture has excellent glass forming
characteristics and the capacity to dissolve a range of trace elements. The synthesis
procedure itself is straight forward and can be completed using relatively unspecialized lab
equipment. The whole glass synthesis procedure can be optimised to no more than 28 hours
in total. Synthesizing the doping sulfide matrix in silica tubes instead of precipitation from
doped solutions as in Ding et al. (2011) enables a larger variety of elements to be doped into
the glass matrix at concentrations much closer to natural samples. The dopant level and
sample size can be easily adjusted as needed in most cases. Most of the doped elements can
be homogeneously distributed up to 100 ppm level. Ru, Os can be homogeneously doped up
to 30 ppm, but to minimize the possible effect of Ru on Mo, W and Re, a concentration close
to 10 ppm is preferred. Re, Mo and W can be homogeneously doped up to 10 ppm and 5 ppm
respectively. The absolute concentrations of all the dopants were verified by solution ICPMS.
A digestion protocol involving HNO3 + HF + Parr bomb was found to be most efficient. The
solution ICPMS results agreed with LA-ICPMS results within analytical errors for most of
the dopants, except for Os, Ir, Pt, Pd, Te, Bi, which might be attributed to the Te co-
precipitation effect. To better verify the concentrations of the dopants and the capability of
this new standard, a variety of reference materials including both sulfide matrix (JBSulfide
and NiS4) and silicate matrix (BIR-1, JGb-1, BHVO-1, JG-1a, JB-2) were tested using Ge6
38
as the standard. The measured results agree reasonably well with the accepted values. The
matrix effects hence do not appear to be significant for different reference materials.
Therefore, Ge6 has great potential as a standard for trace level measurements for transitional
metals, PGE and chalcogens in both sulfide and silicate samples.
39
Figure 2-1 Photo of an experimental product and the capsule design for the synthesis
experiment. Sample size can easily go up to 3~4 cm high in a 6×10 mm silica
tube, given the large hot zone afforded by the box furnace at low synthesis
temperatures.
40
Figure 2-2 Examples of XRD patterns of synthesis product (Ge2), consistent with glassy
structure, no significant crystalline phases formed.
41
Figure 2-3 BSE of the synthesized material (Ge2), confirming the textural homogeneity.
42
a)
b)
Figure 2-4 Examples of properly spiked (a) and improperly spiked (b) samples using
solution ICPMS. Dashed line represents the linear regression through the
individual points. The data point in red (Figure 2-4b) was excluded from the
data reduction.
43
Figure 2-5 Solution ICPMS results of aqua regia based digestion protocol taken up in 1
mol/L HCl matrix versus laser ablation ICPMS results for Ge6 (data of PGE
and chalcogens reduced using MSS5, the rest reduced using NIST610, data
are listed in Table 2-6). Except for Au, almost all the dopants appeared to be
underestimated. Dashed line represents 1:1 ratio.
44
Figure 2-6 Solution ICPMS results of aqua regia based digestion, taken up in in 2%
HNO3 matrix, versus laser ablation ICPMS results (Ge6, data of PGE and
chalcogens reduced using MSS5, the rest reduced using NIST610, data are
listed in Table 2-6). Underestimation persists for these elements measured by
standard addition protocol, which may indicate an incomplete digestion.
Dashed line represents 1:1 ratio.
45
Figure 2-7 HF + HNO3 + parr bomb digestion taken up in 2% HNO3 matrix, solution
ICPMS results versus laser ablation ICPMS results (Ge6, data of PGE and
chalcogens reduced using MSS5, the rest reduced using NIST610, data are
listed in Table 2-6) for spiked elements, dashed line representing 1:1 ratio.
46
a)
b)
Figure 2-8 Summary of solution ICPMS results (HF + HNO3 + Parr bomb digestion
taken up in 2% HNO3 matrix) for ALL the dopants, compared with LA-
ICPMS (Ge6, data of PGE and chalcogens reduced using MSS5, the rest
reduced using NIST610, data are listed in Table 2-6). Except for Os, Ir, Te, Pt,
Pd, everything else is consistent between the two sets of analyses. Dashed line
represents 1:1 ratio line. Fig 2-8a illustrated the higher concentration dopants
while Fig 2-8b illustrated the lower concentration ones.
47
a)
b)
Figure 2-9 Examples of Time Resolved Analysis (TRA) for homogeneity test at trace
element level. Arsenic acts as the homogeneous example in both scans (2-9a
Ge4 and 2-9b Ge6), while Re is considered heterogeneous in 2-9a.
48
a) Ge2
b) Ge3
c) Ge4
d) Ge6
Figure 2-10 Summary of the LA-ICPMS data for all the runs, with relative standard
deviations (RSD%) based on multiples analyses as the Y axis, and the
averaged relative uncertainties in the individual analysis as the X-axis.
Dashed line represents 1:1 ratio. All the dopants can be regarded as
homogeneous in Ge6, as the RSD% is always smaller than the uncertainty due
to counting statistics.
49
Figure 2-11
Relative standard deviation (RSD%) versus
the measured concentrations by LA-ICPMS
in the case of Ru and Os, and versus the
added concentration by weighing in each run
for Re, Mo and W. Mo and W data were also
compared with Ding et al. (2011). Dashed
line represents 10% RSD, acting as a
reference line only. The data show an overall
improvement in homogeneity (low RSD%)
with decreasing concentration. This suggests
that the heterogeneity may be due to the
presence of phases that concentrate these
elements.
50
Figure 2-12 Using Ge6 as the standard, measured results were compared with the accepted
concentration values for JBSulfide and NiS4, for all the related dopants.
Dashed line represents 1:1 ratio.
51
Figure 2-13 Measured results were compared
with accepted values on an elemental basis, for
BIR-1, BHVO-1, JGb-1, JG-1a, JB-2. Dashed
line represents 1:1 ratio. Labels in italic represent
where database values are not available. Instead,
the measured values reduced using NIST610
were used as a reference.
a b
c d
e
52
Tabe 2-1 Synthesis history and major element compositions of different batches of
chalcogenide glasses.
Run #
Melting
Temperature
(°C)
Melting
Duration
(hours)
Bubble
Growth
Prevention
at 850 °C
Anneal
at
220 °C
Ge wt% Sb wt% S wt% Fe wt% Total
Ding et al.
(2011) 950 10 Y Y 36.55 26.92 35.66 0.28 99.41
error 0.25 0.22 0.28 0.0059
Ge2 950 10 Y Y 36.49 25.70 36.82 0.82 99.83
error 0.24 0.13 0.12 0.02
Ge3 950 30 Y Y 35.44 25.16 38.53 0.83 99.96
error 0.22 0.13 0.11 0.011
Ge4 975 10 N N 36.24 25.88 38.23 0.28 100.63
error 0.22 0.16 0.19 0.010
Ge6 975 10 N N 35.87 24.84 39.98 0.20 100.89
error 0.27 0.18 0.18 0.012
53
Table 2-2 NIST610 and MSS5 referenced values used to reduce data with LA-ICPMS
analyses.
Elements MSS5*
Conc. (ppm)
NIST610*
Conc. (ppm)
Cr 405.2
Mn 6800 412.96
Fe 579300 505.7
Co 405
Ni 10526 444
Cu 240.13 430.3
As 67.4 294.73
Se 76.33 109
Mo 376.8
Ru 29
Rh 80
Pd 64
Ag 60.75 239.4
Sn 396.3
Sb 60.55
Te 45.83
W 445.3
Re 32.4 103.7
* Reference data of MSS5 are from Brenan (2015), and data of NIST610 are from Pearce et
al. (1997).
54
Table 2-3 Summary of LA-ICPMS results of Ge2, using the same time-temperature
synthesis protocol of Ding et al. (2011). Data reduced using 57
Fe as the
reference element.
Conc.
(ppm) STD
1 RSD%
2 1σ
3 σ%
4
Expected5
conc. (ppm)
Cr53 1.18E+02 5.35E+00 4.55E+00 1.23E+01 1.05E+01 1.08E+02
Mn55 1.12E+02 2.31E+00 2.06E+00 8.14E+00 7.25E+00 1.16E+02
Co59 1.13E+02 4.27E+00 3.78E+00 8.04E+00 7.12E+00 1.35E+02
Ni60 2.75E+02 8.96E+00 3.26E+00 1.04E+01 3.79E+00 2.09E+02
Cu63 2.63E+01 6.80E-01 2.57E+00 1.33E+00 5.05E+00 2.85E+01
Ga71 9.22E+01 3.34E+00 3.62E+00 6.41E+00 6.96E+00 8.58E+01
As75 3.38E+02 1.25E+01 3.68E+00 2.30E+01 6.81E+00 3.41E+02
Se82 9.56E+01 4.79E+00 5.01E+00 6.26E+00 6.56E+00 9.77E+01
Mo97 1.28E+01 2.84E+01 2.21E+02 1.27E+00 9.90E+00 5.39E+01
Ru101 5.87E+01 4.85E+01 8.26E+01 6.64E+00 1.13E+01 6.00E+01
Rh103 1.19E+02 4.78E+00 4.03E+00 3.79E+00 3.19E+00 1.20E+02
Pd108 1.07E+02 2.16E+00 2.02E+00 5.83E+00 5.45E+00 1.08E+02
Ag109 1.11E+02 2.65E+00 2.38E+00 7.83E+00 7.03E+00 1.05E+02
Te128 1.15E+02 2.80E+00 2.43E+00 9.04E+00 7.84E+00 1.16E+02
W184 9.77E+00 1.75E+01 1.79E+02 7.10E-01 7.22E+00 4.29E+01
Re185 2.38E+01 5.01E+01 2.10E+02 2.81E+00 1.18E+01 9.91E+01
Os189 6.59E+01 4.44E+01 6.74E+01 6.62E+00 1.00E+01 7.22E+01
Ir193 1.25E+02 1.02E+01 8.21E+00 1.54E+01 1.23E+01 1.44E+02
Pt195 1.50E+02 2.93E+00 1.96E+00 6.72E+00 4.48E+00 1.31E+02
Au197 8.45E+01 8.23E+00 9.74E+00 1.22E+01 1.44E+01 8.58E+01
Pb208 6.08E+02 1.21E+01 1.99E+00 2.99E+01 4.92E+00 5.90E+02
Bi209 1.09E+02 2.84E+00 2.60E+00 5.05E+00 4.62E+00 1.10E+02
Note: 1). STD is calculated based on the standard deviation of multiple analyses on the same
sample; 2). 100.
% conc
STDRSD ; 3). 1σ is calculated by Glitter software based on the
counting statistics uncertainty propagation, it represents the uncertainty of each individual
measurement; 4). .100.
1%1
conc
; 5) concentration added based on weight
measurements.
55
Table 2-4 Summary of LA_ICPMS results for Ge3, with 30 h melting time instead of 10 h,
labels same as in Table 2-3. Data reduced using 57
Fe as the reference element.
Conc.
(ppm) STD RSD% 1σ σ%
Expected
conc. (ppm)
Cr53 1.43E+02 2.92E+00 2.04E+00 1.04E+01 7.23E+00 1.16E+02
Mn55 1.19E+02 2.22E+00 1.87E+00 4.86E+00 4.09E+00 1.21E+02
Co59 1.38E+02 2.27E+00 1.65E+00 8.37E+00 6.07E+00 1.39E+02
Ni60 7.40E+02 1.97E+01 2.66E+00 3.29E+01 4.45E+00 7.77E+02
Cu63 3.04E+02 5.78E+00 1.90E+00 2.20E+01 7.22E+00 3.28E+02
Ga71 1.10E+02 1.87E+00 1.70E+00 6.58E+00 5.97E+00 1.12E+02
As75 3.67E+02 7.00E+00 1.91E+00 2.71E+01 7.39E+00 3.41E+02
Se82 1.10E+02 6.23E+00 5.65E+00 7.47E+00 6.77E+00 9.76E+01
Mo97 1.66E+00 1.73E+00 1.04E+02 1.40E-01 8.30E+00 5.41E+01
Ru101 1.53E+01 7.04E+00 4.60E+01 3.19E+00 2.09E+01 6.00E+01
Rh103 1.44E+02 7.69E+00 5.36E+00 2.60E+01 1.81E+01 1.56E+02
Pd108 1.30E+02 4.76E+00 3.66E+00 4.24E+00 3.25E+00 1.40E+02
Ag109 1.29E+02 4.56E+00 3.52E+00 5.41E+00 4.18E+00 1.37E+02
Te128 1.43E+02 7.87E+00 5.52E+00 6.98E+00 4.90E+00 1.51E+02
W182 3.91E+00 1.23E+00 3.14E+01 2.10E-01 5.46E+00 4.31E+01
Re185 1.41E+01 2.90E+01 2.06E+02 1.66E+00 1.18E+01 1.14E+02
Os189 1.67E+01 6.40E+00 3.83E+01 3.23E+00 1.93E+01 7.22E+01
Ir193 1.48E+02 1.38E+01 9.35E+00 2.68E+01 1.81E+01 1.44E+02
Pt195 2.04E+02 1.61E+01 7.90E+00 1.63E+01 8.01E+00 1.97E+02
Au197 1.13E+02 6.76E+00 5.96E+00 1.63E+01 1.44E+01 1.12E+02
Pb208 7.74E+02 6.85E+01 8.85E+00 4.65E+01 6.01E+00 7.67E+02
Bi209 1.45E+02 1.24E+01 8.58E+00 8.09E+00 5.58E+00 1.42E+02
56
Table 2-5 Summary of results Ge4, which was melted at 975 °C instead of 950 °C, labels same
as in Table 2-3. Data reduced using 57
Fe as the reference element.
Conc.
(ppm) STD RSD% 1σ σ%
Expected
conc. (ppm)
Cr52 1.48E+02 2.39E+00 1.62E+00 5.50E+00 3.72E+00 1.57E+02
Mn55 5.52E+01 4.26E+00 7.72E+00 1.16E+01 2.09E+01 5.80E+01
Co59 1.38E+02 2.00E+00 1.44E+00 4.80E+00 3.47E+00 1.35E+02
Ni60 2.08E+02 1.18E+01 5.67E+00 3.39E+01 1.63E+01 2.47E+02
Cu63 4.64E+02 1.48E+02 3.19E+01 1.55E+02 3.34E+01 3.43E+02
Ga71 1.11E+02 1.74E+00 1.57E+00 4.02E+00 3.63E+00 1.44E+02
As75 2.74E+02 7.07E+00 2.58E+00 7.93E+01 2.89E+01 3.06E+02
Se82 4.42E+01 6.19E+00 1.40E+01 1.55E+01 3.51E+01 4.40E+01
Mo97 1.60E+00 1.74E+00 1.09E+02 3.40E-01 2.12E+01 2.69E+01
Ru101 2.25E+01 1.12E+00 4.99E+00 5.09E+00 2.27E+01 3.00E+01
Rh103 4.90E+01 1.22E+00 2.48E+00 1.90E+00 3.88E+00 5.76E+01
Pd108 4.26E+01 1.50E+00 3.51E+00 1.82E+00 4.27E+00 5.20E+01
Ag109 3.71E+01 1.43E+00 3.87E+00 2.77E+00 7.46E+00 4.30E+01
Te128 4.45E+01 2.03E+00 4.56E+00 4.05E+00 9.11E+00 4.77E+01
W182 3.81E+00 1.19E+00 3.13E+01 4.30E-01 1.13E+01 2.14E+01
Re185 3.88E+00 6.63E+00 1.71E+02 9.70E-01 2.50E+01 5.69E+01
Os189 2.63E+01 1.46E+00 5.55E+00 4.24E+00 1.61E+01 3.61E+01
Ir193 6.34E+01 3.48E+00 5.50E+00 4.46E+00 7.03E+00 7.40E+01
Pt195 6.68E+01 4.73E+00 7.09E+00 3.42E+00 5.12E+00 8.16E+01
Au197 2.89E+01 1.73E+00 6.01E+00 2.40E+00 8.30E+00 3.23E+01
Pb208 4.18E+02 6.78E+01 1.62E+01 1.29E+02 3.09E+01 4.42E+02
Bi209 4.86E+01 2.72E+00 5.60E+00 2.14E+00 4.39E+00 5.80E+01
57
Table 2-6 Summary and comparison of LA-ICPMS and solution ICPMS (HF + HNO3 + Parr
bomb) for Ge6, units in ppm. LA-ICPMS data reduced using 57
Fe as the reference
element, while solution ICPMS data were reduced using standard addition protocol
for Mn, Ag, Cu, Ni, Ga, As, Co, Cr, Se, and calibration curve protocol for the
rest (see text for details).
Element Laser
ICPMS STD 1σ
Detection
Limit
Solution
ICPMS STD
Detection
Limit
Expected
conc.
(ppm)
Cr53 8.88E+00 1.71E-01 5.06E-01 2.89E-01 8.71E+00 1.31E-01 8.39E-02 9.83E+00
Mn55 7.01E+00 5.52E-02 4.06E-01 9.51E-02 7.06E+00 1.14E-01 1.97E-02 7.24E+00
Co59 5.15E+00 1.09E-01 2.80E-01 4.60E-02 4.73E+00 9.27E-02 3.30E-03 5.63E+00
Ni60 6.98E+01 1.55E+00 4.82E+00 9.44E+00 4.27E+01 8.91E-01 9.98E-02 6.17E+01
Cu63/65 2.42E+01 1.21E+00 2.25E+00 1.49E-01 2.83E+01 5.55E-01 8.10E-02 2.15E+01
Ga71 5.25E+00 1.14E-02 2.94E-01 9.48E-02 4.83E+00 6.64E-02 9.68E-03 6.00E+00
As75 1.98E+02 2.82E+00 2.03E+01 5.70E-01 1.45E+02 4.44E-01 9.52E-02 2.04E+02
Se82 9.67E+00 1.18E+00 4.22E+00 1.42E+00 9.86E+00 3.26E-01 4.57E-01 8.80E+00
Mo97 3.81E+00 1.35E-01 2.00E-01 3.81E-02 4.33E+00 1.24E-01 1.93E-01 3.37E+00
Ru101 3.31E+00 5.89E-02 3.36E-01 2.18E-02 2.63E+00 1.05E-01 4.87E-01 3.74E+00
Rh103 3.61E+00 4.34E-02 2.46E-01 6.05E-03 3.13E+00 6.86E-02 4.87E-01 3.60E+00
Pd108 3.32E+00 1.41E-01 2.54E-01 6.02E-03 2.23E+00 6.58E-02 4.59E-01 3.25E+00
Ag109 3.48E+00 3.63E-02 4.62E-01 2.21E-02 3.41E+00 6.89E-02 1.58E-03 3.07E+00
Te128 4.62E+00 3.96E-01 4.96E-01 3.87E-01 2.21E+00 2.12E-02 4.91E-01 5.96E+00
W184 4.22E+00 8.79E-02 2.44E-01 1.51E-02 5.08E+00 3.18E-01 4.34E-01 3.91E+00
Re185 1.02E+01 4.47E-01 8.36E-01 7.51E-03 9.93E+00 3.18E-01 4.39E-01 1.07E+01
Os189 4.01E+00 7.89E-02 1.68E-01 1.19E-02 1.07E+00 4.95E-03 ?? 4.51E+00
Ir193 5.28E+00 1.83E-01 3.36E-01 6.91E-03 3.54E+00 6.86E-02 4.84E-01 6.17E+00
Pt195 8.15E+00 2.14E-01 5.52E-01 1.73E-02 5.27E+00 1.33E-01 2.26E-01 1.02E+01
Au197 2.53E+00 9.04E-02 1.44E-01 8.88E-03 2.09E+00 5.30E-02 7.39E-01 2.69E+00
Pb208 4.38E+02 1.16E+01 5.71E+01 2.61E-02 3.22E+02 7.35E+00 2.90E-01 4.17E+02
Bi209 9.30E+00 3.30E-01 4.54E-01 7.55E-03 7.39E+00 2.69E-01 5.76E-01 7.25E+00
58
Table 2-7 Summary of HCl based solution ICPMS, for elements that standard addition can be
applied (ppm) in Ge6.
HCl based
Solution ICPMS STD
Ga71 1.52E+00 9.64E-03
Se82 4.69E+00 8.05E-02
Mo97 2.01E+00 3.72E-02
Ru101 2.63E+00 2.47E-01
Rh103 3.13E+00 8.23E-02
Pd108 1.73E+00 1.65E-01
Te128 6.58E-01 1.65E-01
W184 3.62E+00 3.29E-01
Re185 8.89E+00 8.23E-02
Os189 1.07E+00 3.82E-02
Ir193 3.54E+00 8.23E-02
Pt195 5.27E+00 1.65E-01
Au197 9.19E+00 2.34E-02
Bi209 2.88E+00 3.44E-03
Table 2-8 Summary of HNO3 based solution ICPMS results for elements that standard addition
can be applied (ppm) in Ge6.
Aqua Regia + HNO3 STD HF + HNO3 + parr bomb STD
solution ICPMS
solution ICPMS
Cr53 4.04E+00 2.47E-02 8.71E+00 1.31E-01
Mn55 7.06E+00 3.95E-02 8.05E+00 1.14E-01
Co59 3.60E+00 4.20E-02 4.73E+00 9.27E-02
Ni60 2.90E+01 2.08E-01 4.27E+01 8.91E-01
Cu63 2.83E+01 3.52E-01 3.03E+01 5.55E-01
Ga71 2.76E+00 5.90E-02 6.84E+00 6.64E-02
As75 1.45E+02 1.02E+00 3.57E+01 4.44E-01
Se82 5.84E+00 2.84E-02 9.86E+00 3.26E-01
Ag109 9.21E-01 1.85E-02 3.41E+00 6.89E-02
59
Table 2-9 Summary of the preferred values for Ge6 (in ppm) together with the source of
the data.
Element Preferred Uncertainty Method
Cr53 8.71E+00 1.30E-01 Solution ICPMS
Mn55 7.06E+00 1.10E-01 Solution ICPMS
Co59 4.73E+00 9.00E-02 Solution ICPMS
Ni60 4.27E+01 8.90E-01 Solution ICPMS
Cu63 2.83E+01 5.60E-01 Solution ICPMS
Ga71 4.83E+00 7.00E-02 Solution ICPMS
As75 1.45E+02 4.40E-01 Solution ICPMS
Se82 9.86E+00 3.26E-01 Solution ICPMS
Mo97 4.33E+00 1.24E-01 Solution ICPMS
Ru101 2.63E+00 1.05E-01 Solution ICPMS
Rh103 3.13E+00 6.86E-02 Solution ICPMS
Pd108 3.32E+00 2.54E-01 LA ICPMS
Ag109 3.41E+00 6.89E-02 Solution ICPMS
Sn118 6.01E+02 1.63E+01 Solution ICPMS
Te128 4.62E+00 4.96E-01 LA ICPMS
W184 5.08E+00 3.18E-01 Solution ICPMS
Re185 9.93E+00 3.18E-01 Solution ICPMS
Os189 4.01E+00 7.89E-02 LA ICPMS
Ir193 5.28E+00 3.36E-01 LA ICPMS
Pt195 8.15E+00 5.52E-01 LA ICPMS
Au197 2.09E+00 5.30E-02 Solution ICPMS
Pb208 3.22E+02 7.35E+00 Solution ICPMS
Bi209 9.30E+00 4.54E-01 LA ICPMS
60
Table 2-10a Testing JBSulfide using Ge6 as the standard, using Fe57 as the reference
element, 1σ calculated based on counting statistics uncertainty propagations.
Element Database* Measured (ppm) 1σ
Si29
Cr52 3.05E+01 5.28E+00
Mn55 1132 ± 239 1.33E+03 6.51E+01
Co59
Cu65 2176 ± 500 2.25E+03 1.17E+02
Zn66 1.00E+02 5.59E+00
Zn68 9.55E+01 7.25E+00
Ga71
As75
Se82
Mo95 2.55 ± 0.213 2.83E+00 6.40E-01
Ru101 260 ± 7 2.98E+02 1.26E+01
Rh103 237 ± 5 2.81E+02 1.66E+01
Pd108 247 ± 4 2.57E+02 9.26E+00
Ag109 6.60E-01 2.10E-01
Sn122
Te128 7.43E+00 2.09E+00
W182
Re185 95 ± 2 1.19E+02 6.39E+00
Os189 302 ± 6 3.94E+02 2.56E+01
Ir191 315 ± 12 3.65E+02 1.95E+01
Pt195 294 ± 8 3.65E+02 1.54E+01
Au197
Pb208 8.4 ± 0.84 8.94E+00 6.00E-01
Bi209
* Reference data of JBSulfide are from Mungall and Brenan (2014).
61
Table 2-10b Testing NiS4 using Ge6 as the standard, using Fe57 as the reference element,
1σ calculated based on counting statistics uncertainty propagations.
.Element Database* Measured (ppm) 1σ
Si29
Cr52
Mn55
Co59
Cu65 12592 ± 514 1.17E+04 5.25E+02
Zn66
1.72E+01 2.50E+00
Zn68
1.83E+01 5.60E+00
Ga71
As75 1027 ± 586 9.79E+02 2.10E+02
Se82 1988 ± 21 1.55E+03 1.11E+02
Mo95
1.01E+01 9.00E-01
Ru101
2.64E+00 5.10E-01
Rh103
1.41E+00 3.90E-01
Pd108
8.20E-01 2.40E-01
Ag109
1.56E+00 2.30E-01
Sn118
Sn122
4.16E+02 1.84E+01
Te128 2140 ± 58 2.37E+03 5.80E+02
W182
Re185
Os189
Ir191
1.23E+00 1.30E-01
Pt195
1.82E+00 2.80E-01
Au197
6.52E+00 3.10E-01
Pb208 1349 ± 23 1.43E+03 3.27E+02
Bi209 1349 ± 23 2.00E+03 9.20E+01
* Reference data of NiS4 are from Brenan (2015).
62
Table 2-11a Testing BIR-1 using Ge6 as the standard, using Fe57 as the reference element,
1σ calculated based on counting statistics uncertainty propagations.
.Element Database* Measured (ppm) 1σ
Si29
Cr52 368 ± 8.25 3.67E+02 2.90E+01
Mn55
7.51E+02 4.20E+01
Co59 51.1 ± 1.88 5.93E+01 3.60E+00
Ni61 165 ± 5.88 1.31E+02 1.70E+01
Cu65 115 ± 3.71 1.24E+02 4.95E+00
Zn66 72 ± 18 5.83E+01 2.70E+00
Zn68 72 ± 18 6.17E+01 2.80E+00
Ga71
2.35E+01 1.19E+00
As75 0.44 ± ? 5.70E-01 1.00E-01
Se82
2.53E+00 1.27E+00
Mo95
4.90E-01 8.00E-02
Ru101
Rh103
Pd108
Ag109
Sn118
9.50E-01 9.00E-02
Sn122
6.40E-01 1.20E-01
Te128
W182
Re185
Os189
Ir191
Pt195
Au197
Pb208 3.1 ± ? 2.79E+00 1.95E-01
Bi209
* Reference data are compiled from Flanagan (1984), Govindaraju (1994) and Plumee
(1998).
63
Table 2-11b Testing JGb-1-1 using Ge6 as the standard, using Fe57 as the reference
element, 1σ calculated based on counting statistics uncertainty propagations.
Element Database* Measured (ppm) 1σ
Si29
Cr52 59.3 ± 13.5 5.71E+01 3.30E+00
Mn55 1325 ± ? 1.17E+03 4.67E+01
Co59 61.6 ± 5.8 5.66E+01 2.13E+00
Ni61 25.4 ± 5.7 3.71E+01 7.79E+00
Cu65 86.8 ± 5.6 1.73E+01 1.13E+00
Zn66 111 ± 8.0 1.11E+02 4.54E+00
Zn68 111 ± 8.0 1.18E+02 4.89E+00
Ga71 18.9 ± 3.3 1.98E+01 8.20E-01
As75 1.09 ± 0.058 1.02E+00 2.60E-01
Se82 0.17 ± 0.017 2.07E+00 8.50E-01
Mo95 0.45 ± 0.23 3.90E-01 1.10E-01
Ru101
Rh103
Pd108
Ag109
2.03E+00 1.20E-01
Sn118 0.36 ± 0.13 5.10E-01 1.50E-01
Sn122 0.36 ± 0.13 5.30E-01 2.30E-01
Te128
W182 0.81 ± ? 9.20E-01 1.90E-01
Re185
Os189
Ir191
Pt195
8.20E-01 7.00E-02
Au197
Pb208 1.9 ± 0.74 1.37E+00 8.00E-02
Bi209 0.014 ± ? 2.00E-02 1.00E-02
* Reference data are compiled from Imai et al. (1995), Govindaraju (1994) and Plumee
(1998).
64
Table 2-11c Testing JG-1a using Ge6 as the standard, using Fe57 as the reference element,
1σ calculated based on counting statistics uncertainty propagations.
.Element Database* Measured (ppm) 1σ
Si29
Cr52 18.6 ± 4.4 1.80E+01 1.17E+00
Mn55
2.82E+02 1.17E+01
Co59 5.7 ± 1.54 4.56E+00 2.00E-01
Cu65 1.3 ± 0.59 9.10E-01 2.00E-01
Zn66 38.8 ± 2.2
Zn68 38.8 ± 2.2 3.56E+01 1.69E+00
Ga71 17 ± 0.70 1.79E+01 6.90E-01
As75 0.39 ± 0.12 4.70E-01 2.50E-01
Se82
Mo95 0.67 ± 0.29 1.93E+00 1.20E-01
Ru101
Rh103
Pd108
Ag109
2.60E-01 8.00E-02
Sn118 4.2 ± 0.56 4.50E+00 2.20E-01
Sn122 4.2 ± 0.56 4.54E+00 3.40E-01
Te128
W182
3.32E+00 1.41E+00
Re185
Os189
Ir191
Pt195
Au197
Pb208 27 ± 2.8 2.50E+01 1.08E+00
Bi209
* Reference data are compiled from Imai et al. (1995), Govindaraju (1994) and Plumee
(1998).
65
Table 2-11d Testing JB-2 using Ge6 as the standard, using Fe57 as the reference element,
1σ calculated based on counting statistics uncertainty propagations.
Element Database* Measured (ppm) 1σ
Si29
Cr52 27.4 ± 5.3 2.58E+01 1.72E+00
Mn55
9.60E+02 3.11E+01
Co59 39.8 ± 6.6 4.00E+01 1.58E+00
Cu65 227 ± 16 7.29E+01 5.07E+00
Zn66 110 ± 11 1.09E+02 4.58E+00
Zn68 110 ± 11 1.14E+02 4.84E+00
Ga71 17 ± 2.7 1.72E+01 6.60E-01
As75 2.98 ± 0.85 3.02E+00 3.20E-01
Se82
3.43E+00 1.12E+00
Mo95 0.54 ± 0.47 4.80E-01 8.00E-02
Ru101
Rh103
Pd108
Ag109
1.81E+00 1.00E-01
Sn118 0.56 ± 0.44 4.80E-01 7.00E-02
Sn122 0.56 ± 0.44 4.20E-01 1.10E-01
Te128
7.00E-01 2.20E-01
W182 0.26 ± ? 3.00E-01 3.00E-02
Re185
Os189
Ir191
Pt195
Au197
Pb208 5.4 ± 1.08 4.19E+00 2.00E-01
Bi209 0.033 ± ? 2.00E-02 1.00E-02
* Reference data are compiled from Imai et al. (1995), Govindaraju (1994) and Plumee
(1998).
66
Chapter III: Partitioning of PGE and chalcogens within
sulfides under controlled fO2, fS2 conditions
---- Experimental measurements and origin of MSS-melt
fractionation
3.1. Introduction
Phase equilibrium information exists to predict the low temperature melting behavior of
chalcogen-rich compositions (see review by Frost et al., 2002), but there is only limited
information to assess the role of these elements in the early sequestration of PGE, or their
effect on PGE distribution during solidification of massive sulfide bodies (Helmy et al., 2007,
2010, 2013a,b). Thus, the importance of the chalcogens to mass balance assessments in
massive sulfide deposits remains to be fully explored. Knowledge of MSS-sulfide melt
67
partitioning serves as a first order constraint on the behavior of the chalcogens, as it provides
a means to assess if early enrichment or depletion of these elements is likely. As such,
experiments have been done to measure the partitioning of PGE and chalcogens between
MSS and sulfide melt. Results are also presented for the partitioning of this element suite
between intermediate solid solution (ISS) and MSS, information which was previously
unknown. Such data provide the essential information for predicting the composition of MSS
and ISS cumulates, the trajectory of sulfide liquid compositions, as well as the likelihood of
chalcogen-rich PGM saturation or separation of an immiscible chalcogen melt phase. An
important attribute of this study is that experiments were done at conditions of sulfur and
oxygen fugacity similar to mildly oxidized terrestrial magmas; at such conditions the oxygen
content of the sulfide liquid can reach several wt% (e.g., Naldrett 1969; Kress 1997; Mungall
et al., 2005; Fonseca et al., 2008). As shown for the PGEs, dissolved oxygen in the sulfide
melt can exert an influence on metal activity coefficients (e.g., Andrews and Brenan, 2002;
Fonseca et al., 2007, 2011), so is clearly an important parameter to be explored in the context
of MSS- and ISS-melt partitioning.
3.2. Experimental Technique
3.2.1 General Strategy
The goal of the experiments was to measure MSS-melt and MSS-ISS partitioning of PGE
and chalcogens at conditions in which fO2 and fS2 could be controlled and/or estimated.
Experiments were done with the PGE and chalcogens doped separately, and with a combined
element suite to determine the extent to which partitioning of PGEs would be affected by the
68
presence of chalcogens in the system. Experiments used evacuated silica tubes so as to
contain the volatile PGE and chalcogens (Figure 3-1). Oxygen fugacity was buffered using
FMQ, with values of fO2 calculated from the calibration of O’Neill (1987). Although solid
noble-metal sulfide buffers offer a straightforward means of fS2 control, reconnaissance
experiments revealed significant loss of chalcogens to those materials by volatile transfer.
Instead, the fS2 was monitored in experiments by measuring the Fe/S ratio of pyrrhotite
added to the FMQ assemblage, which allowed the fS2 to be calculated using the calibration
of Toulmin and Barton (1964).
3.2.2 Starting materials
A PGE- and chalcogens- free starting material (MSSY2, listed in Table 3-1) was made by
first melting a powdered mixture of 45.2 wt% Fe, 5.87 wt% Ni, 16.56 wt% Cu and 32.3 wt%
S together in a 4×6 mm (ID×OD) vacuum-sealed silica tube. This “base” sulfide is designed
to produce a large liquid fraction with coexisting MSS, and is guided by the composition of
MSS and melt produced in the experiments of Li et al. (1996). The sample was first
annealed overnight at 600 °C then melted at 1000 °C for 2 hours. The furnace temperature
was then decreased to 600 °C, and the sample removed and cooled in air. The run product
was ground thoroughly under ethanol for 30 min, dried and stored in a desiccator.
This dopant-free starting material was subsequently divided into 3 parts: part A was only
mixed with a suite of high-purity PGE and precious metal powders (Ru, Rh, Pd, Ag, Re, Os,
Ir, Pt, Au); part B was only mixed with a suite of chalcogen powders (As, Se, Sb, Te, Bi);
69
part C remained dopant free. Each of the doped mixtures were then loaded into separate
silica tubes and fused following the same procedure as previously described. After fusing,
part A and B served as the PGE-dopant rich sulfide composition and a chalcogen-dopant rich
sulfide composition, respectively, with trace element concentrations of around 5000 ppm.
These compositions were then fused with more of the dopant free starting material (part C),
to further dilute the concentrations (Table 3-1), then stored for the subsequent MSS-melt
partition experiments.
A more Cu-rich base composition (ISS#2, listed in Table 3-1) was employed to achieve
saturation in subequal proportions of MSS and ISS in experiments done at 850-875oC.
Powdered mixtures of 35.6 wt% Fe, 6.72 wt% Ni, 26.1 wt% Cu and 31.7 wt% S containing
PGE and chalcogen dopants were melted following the same procedure as the MSS-melt
starting material synthesis. Both of the MSS-melt starting materials and MSS-ISS starting
materials were periodically replaced with freshly-synthesized material, as the sulfides slowly
oxidize during storage.
3.2.3 Verification of experiment fO2 and fS2
The oxygen fugacities in experiments were buffered at FMQ, which was verified by the
successful preservation of all three buffering phases at the end of each experiment. Figure 3-
2 provides an example of the textural development in the buffer portion of experiments. As
can be seen here, and in all other run products, the fayalite-magnetite-quartz assemblage is
preserved, along with abundant pyrrhotite (so-called, FMQP, Shi 1992).
70
The sulfur fugacity was calculated based on the composition of this coexisting pyrrhotite
according to the Toulmin and Barton (1964) calibration. As they demonstrated, the fS2 of a
pyrrhotite-bearing system can be calculated from the mole fraction of FeS in pyrrhotite (N)
and the temperature (T) through equation 3-1:
1)-3 (Eq. 91.119981.0130.39)1/1000)(83.8503.70(log 210 NTNfS
Whereas Toulmin and Barton (1964) measured the composition of pyrrhotite by x-ray
diffraction, we chose to estimate N based on electron microprobe analysis. In order to
confirm the accuracy of this approach, three experiments were done in which synthetic
pyrrhotite was equilibrated at an fS2 imposed using solid sulfide buffers (Ru-RuS2, Pt-PtS,
Ir2S3-IrS2). In these test experiments, sulfide buffer pairs were packed into the bottom of a
silica tube, then a layer of coarse grained silica, then the pyrrhotite powder. The silica layer
serves to physically separate the samples, yet allow the vapor phase to permeate through.
Tubes were then evacuated, sealed, and equilibrated in the furnace at 950 °C for 3 days, then
quenched. The sulfide buffer was removed and analyzed by X-ray diffraction to confirm that
both phases were still present. The pyrrhotite powder was mounted separately in epoxy and
analyzed by electron microprobe. The fS2 for each sulfide buffer pair at the temperature of
the experiment was calculated using the thermodynamic data of Barin (1995). The fS2
recorded by the pyrrhotite composition determined by electron microprobe analysis can also
be calculated accordingly by Equation 3-1. The comparison of the two is illustrated in Figure
3-3. All the measured fS2 values are in good agreement with those predicted by the Toulmin
and Barton (1964) calibration, confirming the accuracy of the fS2 estimated in our
experiments.
71
Values of fS2 determined from the Fe/S ratio of the pyrrhotite are relatively consistent among
different experiments in this study and are on average 0.5 (+/- 0.2) log units higher than the
Pt-PtS buffer (calculated from the thermodynamic data in Barin, 1995; Figure 3-4). These
sulfur fugacities are slightly higher than those generated in the experiments of Mungall et al.
(2005), which were buffered at Pt-PtS. The fO2 of experiments reported by Mungall et al.
(2005) were also buffered at FMQ, consistent with comparable oxygen content of the sulfide
melts produced in the two studies. The fO2-fS2 conditions of this study are essentially
identical to values expected for sulfide saturation of a silicate melt having 10 mole % FeO at
the FMQ buffer (Figure 3-4), emphasizing the applicability of our experimental data to
natural conditions.
3.2.4 Partitioning experiments
Partitioning experiments were conducted using evacuated silica tubes, as illustrated
previously in Figure 3-1. Approximately 60 mg of the doped sulfide sample powder was
loaded and tightly packed at the bottom of a 3×5 mm silica tube. A layer (~5 mm thick) of
coarse-grained silica (~1 mm in size) was packed above the sulfide sample, then the FMQP
mixture. Above this was inserted a well-fit silica rod spacer (0.7 ~1.5 cm long). The loaded
silica tube was then evacuated for 60 minutes or more, and fused shut with a torch. Samples
were placed upright on the hearthplate of a glass-melting furnace, and held at temperatures
and durations as listed in Table 3-3. The temperature within the furnace was monitored by
an S-type thermocouple (Pt-PtRh10%) calibrated against the melting point of gold.
72
Calibration experiments monitored both the cold-junction compensated emf directly, as well
as the emf-converted temperature display output. Emf was converted to temperature using
manufacturer-supplied conversion tables. Both temperature measurements yielded identical
values, and were found to be accurate to within 2oC. Experiments were terminated by
dropping the sample into a bath of salty water + ice. Once retrieved, the quenched charge
was mounted in epoxy, ground open, vacuum impregnated with epoxy, then ground and
polished for further analysis.
3.3. Analytical Techniques
3.3.1. Major element analysis
Major elements (Fe, Ni, Cu, S, and O) were analyzed using the Cameca SX50 electron
microprobe in the Department of Earth Sciences, University of Toronto. Analytical
conditions were 20 kV, 50 nA, and a defocused beam of 30 um in diameter, to compensate
for the quenched interlaced texture of ISS and dendritic texture of sulfide melt. Fe and S
were counted for 20 sec on peak and 10 sec on each side of the background. Cu and Ni were
counted for 30 sec on peak, 15 sec on each side of the background and O was counted for 60
sec on peak, 30 sec for each side of the background. A newly synthesized stoichiometric FeS
(equilibrated with metallic Fe) was used as the standard for Fe and S. Copper was calibrated
on synthetic chalcopyrite, and Ni was calibrated on a natural pentlandite. To reduce the
potential effects induced by surface oxidation, the synthetic FeS standard was always re-
polished and carbon coated again before each analytical session. Oxygen was analyzed with
a synthetic PC1 crystal (2d = 60 A), using hematite as the standard. To ensure the
73
consistency in carbon coating thickness, the hematite standard was also re-polished before
each analytical session and carbon coated with the unknown samples at the same time. For
each individual phase, at least 5 analyses were acquired and only those totals between 98.5
and 101.5 were used for further data processing. Major element compositions of the run
products are reported in Table 3-3.
3.3.2. Trace element analysis
Trace elements were analyzed by laser ablation inductively-coupled plasma mass
spectrometry (LA-ICPMS) in the Department of Earth Sciences at University of Toronto.
Sulfides were analyzed using a laser repetition rate of 4 Hz, a 0.05 mm spot size, and with
the analysis area moving back and forth during ablation. Beam irradiance was optimized
depending on photon-coupling characteristics of the sulfide. Helium was used as the carrier
gas to transport the ablation aerosol from the sampling cell to the plasma. Factory-supplied
time resolved software was utilized for the acquisition of individual analyses. A typical
analysis involved 20 seconds of background acquisition with the ablation cell being flushed
with He, followed by laser ablation for 60 seconds. Analyses were collected in a sequence,
with the first and last four spectra acquired on the reference material. For each different
phase, five individual analyses were acquired and the results averaged. Data were reduced
off-line using the GLITTER version 5.3 software package, supplied by Macquarie Research,
Ltd. Trace element concentrations were quantified using a synthetic MSS (in house reference
MSS5, composition listed in Table 3-2) doped with PGE, Ag, Au and chalcogens. The PGE
content of MSS5 was determined using our in-house JB-sulfide standard (see Mungall and
Brenan 2014 for details of this material), whereas Au, Ag and chalcogens were estimated
74
using NIST610. In our experience, the difference in abundance determinations using silicate
or sulfide standards is <20% relative (see Mungall and Brenan 2014 for demonstration of
this). Although absolute abundances may be subject to this uncertainty, calculated partition
coefficients will be more accurate than this, as they are based on yield-corrected count rates
in the different sulfide phases. Ablation yields in sulfides were corrected by referencing to
the known concentration of Ni as determined by electron microprobe analyses. Where
possible, multiple isotopes were measured to assess molecular and isobaric interferences;
with trace element concentrations reported using the following isotopes: 75
As, 82
Se, 101
Ru,
103Rh,
108Pd,
109Ag,
121Sb,
128Te,
185Re,
189Os,
193Ir,
194Pt,
197Au and
209Bi. Pure Cu and Ni
metal were analyzed to correct for the argide interferences on 103
Rh and 101
Ru. The so-
corrected 101
Ru and 103
Rh were also compared with the results reduced following the
correction procedure listed in Sylvester (2001). The difference in concentration calculated
using the two correction procedures is <16% relative.
Once the concentration of a certain element i is obtained, the partition coefficient D is
calculated by the definition:
melti
MSSi
C
C(MSS/Melt) D
(Eq. 3-2)
where MSSiC is the concentration of this element in MSS, and Melt
iC is the concentration of i in
melt phase. D (MSS/ISS) and D (ISS/melt) are defined in the same manner. The uncertainty
of these partition coefficients [dD (MSS/melt)] can be calculated through Eq 3-3:
75
3)-(Eq.3 )()()/()/( 22
melt
i
melt
i
MSS
i
MSS
i
C
dC
C
dCmeltMSSDmeltMSSdD
where d MSSiC is the uncertainty in MSS
iC measurement, d Melt
iC is the uncertainty in Melt
iC
measurement, estimated from the standard deviation of multiple analyses of the same phase.
Trace element composition of run product phases and calculated partition coefficients are
reported in Tables 3-4, 3-5 and 3-6.
3.4. Results
3.4.1 General aspects
For the first bulk composition employed, MSSY2 (listed in Table 3-1), a series of
reconnaissance experiments were conducted to establish the liquidus temperature, and first
observed stable MSS at 930°C. Subsequent partitioning experiments conducted below this
temperature produced large, homogeneous MSS crystals segregated to the bottom of the
charge, coexisting with sulfide melt (Figure 3-5a). Experiments carried out at temperatures
between 850oC and 875
oC, using the Cu-rich bulk composition (ISS#2 listed in Table 3-1),
produced the stable phase assemblage of MSS and ISS (Figure 3-5b). Stable ISS is
distinguished by the presence of a trellis-like pattern of Cu-rich (bornite-like) exsolution
lamellae. MSS and ISS appear intimately intergrown in these experiments. Although the ISS
exhibits some textural similarities to quenched sulfide melt, we interpret it as a stable,
crystalline phase for three reasons. First, the quench texture is distinctly different from the
melt produced in higher T experiments, as the Cu-rich phase appears to have exsolved in a
crystallographically-controlled manner (i.e., orthogonal trellis pattern). Second, the oxygen
76
content of the quenched ISS is low (<0.6 wt%), and more similar to the MSS, in contrast to
the more oxygen-rich character of the quenched sulfide liquid (~2 wt%) produced at only
slightly higher T (Figure 3-6a). Third, as shown in Figure 3-6b, the ISS composition is
comparable to that produced in the experiments of Fleet and Pan (1994) and Helmy et al.
(2007) and is consistent with the compositional bounds defined by phase equilibria in the Cu-
Fe-S system (Raghavan, 2004).
3.4.2 Attainment of equilibrium
MSS-melt equilibrium was assessed in a number of ways. First, multiple electron
microprobe analyses across single grains revealed a lack of any systematic zoning of the
major elements, and similarly, the time-resolved spectra for trace elements was found to be
uniform. Second, experiments done for different durations at the lowest temperatures for
each element suite (900 ºC and 1 to 7 days for PGE-doped runs, 885 ºC and 3 to 8 days for
chalcogen-doped runs) yielded reproducible partition coefficients (Tables 3-4 and 3-5). Such
results are consistent with the rapid diffusion of PGE in MSS, as documented for Os by
Brenan et al. (2000). Using the temperature dependence for the Os diffusivity from that
work indicates that at 900 ºC, it should take less than 22 hours for an Os atom to diffuse over
a mm-scale distance, indicating that run durations exceeding one day, as in this study, should
be sufficient to attain equilibrium.
3.4.3 MSS-melt partitioning
77
Three experiments were conducted with PGE-Ag-Au-doped starting materials at 930 °C,
915°C and 900 °C respectively. MSS/melt partition coefficients for Ru, Os, Ir, Re and Rh
are >1, whereas Cu, Pd, Pt, Ag and Au are incompatible in MSS. The averaged values of all
the PGE-Ag-Au doped runs are plotted together with literature data in Figure 3-7, sorted in
the order of compatibility in MSS. The sulfur fugacity of these runs is broadly similar (logfS2
= -2.3 to -2.6), and our measurements agree with literature data, in most cases within error.
Detailed differences in partitioning behavior amongst studies are discussed in section 3.5.1.
The partitioning behavior of the chalcogens was investigated through another three
experiments done at temperatures ranging from 885 °C to 915 °C. All the chalcogens studied
are incompatible in MSS, as illustrated in Figure 3-8. Partition coefficients are consistent in
magnitude with past work (Helmy et al., 2010; Li and Audetat, 2012; Brenan 2014), although
the large variation shown for As partitioning reported by Helmy et al (2010) is interpreted to
result from a change in As speciation over the range of experimental conditions investigated.
Also, we have found that Sb, Bi and Pb are more incompatible in MSS than measured in the
work of Li and Audetat (2012).
Helmy et al (2013a) measured the effect of As on the partitioning of Pt between MSS and
sulfide melt at the fS2 imposed by the Fe-FeS buffer at 950oC and 0.1 MPa. Their results
showed a ~10-fold decrease in D(MSS/melt) as the As content of the melt increased from 0
to 40 ppm. The decrease in D(MSS/melt) was attributed to the formation of Pt-As
nanoclusters in the sulfide liquid, as evidenced by the presence of various combinations of
78
nanometer-sized crystalline PtAs2, and amorphous Pt-As phases, present as inclusions
trapped in run-product MSS crystals. Importantly, experiments in which these phases were
found were done at arsenic concentrations well below macroscopic saturation in immiscible
Pt-arsenide melt or sperrylite. Helmy et al (2013a) use these results to argue that the
behavior of the PGE in the presence of such ligands as As (and possibly other chalcogens)
could be controlled by the surface properties of precrystalline nanoclusters. Although the
PGE-doped experiments were not specifically designed to verify this effect, they contained a
range of arsenic concentrations, and therefore may be relevant in this context. Experiments
yn50, 51, 52 and 71 were not intentionally doped with As, but nonetheless contain ~40-80
ppm as a contaminant in the starting material. As an As-free control, experiment yn90 was
synthesized with special care to avoid contamination, and was found to contain < 5.4 ppm As
in the resulting melt. A high As-containing sample (yn60) was synthesized by mixing the
PGE-Ag-Au-doped starting material with approximately the same amount of the chalcogen-
doped mixture, resulting in ~140 ppm As in the melt. The As/Pt molar ratio ranged from ~1
to 1.8 for the experiments with ~40-70 ppm As (As/∑PGE molar ratio of ~0.25 to 0.4) to
~6.4 (As/∑PGE molar ratio of ~2) for experiment yn60 doped with ~140 ppm As. For
comparison, the experiments of Helmy et al. (2013a) were doped with ~20 ppm Pt, and
contained ~7 to ~40 ppm As, corresponding to a range in molar As/Pt of ~1 to ~5. The
variation in the MSS/sulfide liquid partition coefficient with As content of the liquid from
this study is plotted in Figure 3-9a-b. For the range of As concentrations investigated, there
is no significant impact on partitioning of the PGE-Au-Ag suite, including arsenic on Pt
partitioning. Whereas Helmy et al. (2013a) observed a relatively large effect of As on Pt
partitioning, this behavior could not be reproduced here. The seemingly inert behavior of As
79
was also demonstrated in the experiments of Fleet et al (1993) who measured MSS-sulfide
melt partitioning for samples doped with up to 3500 ppm As, 4600 ppm Bi and 3300 ppm Te,
compared to 40-50 ppm of the PGE (molar As/Pt of ~230; and As/∑PGE of ~50), and found
that D(MSS/melt) for Pt (and other PGE) was identical to equivalent chalcogen-free
experiments. A notable difference between studies is sulfur fugacity; whereas Fleet et al.
(1993) used sulfur-excess starting materials, as in this study with log fS2 in the range of -2 to
-2.6 (i.e., near values for natural silicate magmas), the fS2 of experiments done by Helmy et
al (2013a) was buffered at much lower values near Fe-FeS (approximate log fS2 of -7).
Assuming both As- and S-related Pt species in the sulfide liquid, their relative abundance can
be expressed as a homogeneous exchange reaction of the form:
2222
Asx
PtSSy
PtAs yx (Eq. 3-4)
This implies that an increase in the sulfur fugacity will shift the equilibrium to the right,
promoting the formation of the Pt-S species. Therefore, it seems plausible that the general
lack of any effect of arsenic on Pt (or other PGE) partitioning observed at high fS2 indicates
that the Pt-As species existing at low fS2 has been consumed by reaction (3-4). Results
would therefore suggest that arsenic might not be an important complexing agent at the much
higher fS2 required to stabilize sulfide liquid in an FeO-bearing silicate magma, due to the
effect of fS2 on As speciation.
80
3.4.4 MSS-ISS partitioning
Three PGE-Au-Ag- doped experiments were conducted at 875 °C, 860 °C and 850 °C in
order to measure the MSS/ISS partition coefficients for these elements. The averaged results
are plotted in Figure 3-10a. Partition coefficients for Ru, Os and Ir are relatively imprecise,
as these elements are present near detection limits in the ISS phase. We also encountered
anomalously elevated concentrations of the more incompatible elements (As, Sb, Bi) when
measuring some ISS grains. Abundances are enriched in relative proportion to the order of
compatibility, obtained from “normal” analyses, which show lower concentrations. These
anomalous values were attributed to the presence of a small amount of trapped liquid within
the ISS, which could not be identified, or avoided during LA-ICPMS analysis. Such
anomalous values were excluded from the calculated average concentrations reported.
Experiment yn80 was conducted with a more Cu rich starting material, ISS#2, yet the results
for all three experiments are consistent: Ru, Os, Ir, Rh and Re are more compatible in MSS
than in ISS, whereas Pd, Pt, Ag and Au partition preferentially into ISS. Similar MSS/ISS
partition coefficients for the element pairs Pt-Pd and Au-Ag indicates that the onset of ISS
crystallization will not change the relative fractionation between these elements. Jugo et al.
(1999) measured the partitioning of Au between pyrrhotite and ISS at 850 °C, 100 MPa,
obtaining a value of 0.0184±0.0016, which is in very good agreement with the partition
coefficient of 0.0127±0.0055 determined in this study (Figure 3-10a).
Three chalcogen-doped MSS/ISS partitioning experiments were conducted at 860 to ~866 °C.
Results are provided in Figure 3-10b, including data for experiment yn73 measured in two
81
different analytical sessions by laser ICP-MS, done 6 months apart. Those results are self-
consistent within analytical uncertainties. Partition coefficients for Se and As between MSS
and ISS are near unity, while Te, Sb and Bi all prefer the ISS phase relative to MSS. Values
of MSS/ISS partitioning of Te measured by Helmy et al (2007) are ~10-fold larger than those
determined in this study (Figure 3-10c). Other than the difference in experimental conditions
and concentration levels, we suspect this may also be partially related with our compositional
difference in the ISS phase between the two studies, as demonstrated in Figure 3-6b and
Figure 3-10c. The ISS produced in this study contains relatively higher Cu concentrations
compared with the ISS in Helmy et al. (2007). We speculate that higher Cu may result in
increased distortion of the ISS lattice, providing more tolerance for larger ions, such as Te, to
be accommodated, resulting in a lower D(MSS/ISS).
3.4.5 ISS-melt partitioning
We have estimated ISS-sulfide melt partition coefficients by combining average values for
MSS-sulfide melt, and MSS/ISS partitioning determined in this study. The results are
illustrated in Figure 3-11a and 3-11b. The calculated partition coefficients indicate that all
the precious metals should behave similarly to each other when partitioning between ISS and
melt, with each weakly preferring melt relative to ISS. In terms of the chalcogens, the
calculated partition coefficient for Se is near unity, whereas Te, As, Sb and Bi are moderately
incompatible. Given the potential for trapped melt in the ISS analyses, the true partition
coefficients for the most incompatible elements between ISS and sulfide melt might be
somewhat lower, so the incompatible elements may enrich in the residual liquid at an even
faster rate than predicted in this study.
82
3.5. Discussion
3.5.1 Origin of the PGE and chalcogen partitioning systematics
Fleet et al. (1993) noted that the partition coefficients between MSS and sulfide liquid
change progressively between different chemical subgroups of PGE, being higher for the
iron triad (Ru, Os) and lower for the nickel triad (Pd, Pt). Subsequent experiments, including
results of this study, have confirmed this observation, and extended it to the copper triad (Ag,
Au), which are the most incompatible PGE in MSS. All of the chalcogens are found to be
incompatible in MSS. In addition to these inter-element fractionations, there are some
differences in partition coefficients measured in the current work compared with past results.
Here attempts are made to rationalize these observations in the context of ligand field theory,
and the composition of MSS and melt.
Monosulfide solid solution has a NiAs-type structure, with Fe in six-fold coordination with S,
incorporating vacancies on the Fe sites and Fe3+
holes to satisfy the charge imbalance in
metal deficient MSS (see review by Wang and Salveson, 2005). Ballhaus and Ulmer (1995)
showed that Pt and Pd (and by extension, the other PGE) substitute for Fe in MSS on a one-
for-one basis. Insight into the possible mineral structure control on PGE incorporation into
MSS can be gained by considering the relative solubilities of the PGE in MSS compared to a
fixed standard state (pure metal or pure metal sulfide). Figure 3-12 provides a summary in
terms of the solubility of Os, Rh, Pt and Pd in MSS as a function of the metal/sulfur ratio.
All of the PGE solubilities increase with decreasing metal/sulfur, indicating a decrease in
83
metal activity coefficients, and that PGE substitution is enhanced by the presence of Fe
vacancies (Ballhaus and Ulmer, 1995). In this context, MSS/sulfide melt partitioning for the
PGE-Au, including this and past results, are plotted as a function of the M/S in the MSS in
Figure 3-13. Where there are data for a significant range in MSS composition (Rh, Ir, Pd, Pt),
values of D(MSS/melt) show a weak increase with decreasing M/S ratio – a trend that
follows the increase in metal solubility, and hence consistent with decreasing metal activity
coefficients in MSS.
Although the general trends in partitioning with MSS composition seem consistent with
expectations, the sense of PGE fractionation by MSS-melt partitioning is not reflected in the
metal solubility data. Specifically, 1) Rh is found to be more soluble than Pt and Pd, 2) the
solubility of Pd is significantly higher than Pt, and 3) both Os and Pt are least soluble. These
latter two differences are inconsistent with the overall incompatibility of Pt and Pd relative to
Os and Rh, and the generally similar D(MSS/melt) for the pairs Pt-Pd and Os-Rh. Reasons
for the difference in metal solubility in MSS may be related to steric effects, and their role in
charge delocalization, as none of the common PGE sulfides (PtS, PdS, Ir2S3, etc) have the
NiAs structure type (e.g., Raybaud et al., 1997), although a detailed discussion of this is
beyond the scope of this thesis. The important point here is that these inconsistencies in the
solubility data imply that MSS-melt partitioning of the PGE must also be controlled by
coordination complexes formed in the sulfide melt phase. Whereas Ru, Rh, Ir, and Os (and
Re) are in VI-fold coordination in their known sulfides, both Pd and Pt are in IV-fold
coordination (see summary by Raybaud et al., 1997). Furthermore, the likely oxidation state
for Pt and Pd is 2+ in molten sulfide under the conditions of our experiments (Fonseca et al.,
84
2009), which has a d8 electronic configuration, and hence stabilized by square planar
coordination. In the absence of such sites in the NiAs-type structure, it therefore seems
reasonable that both Pt and Pd are stabilized in square planar coordination by the more
“permissive” sulfide liquid structure. A similar argument may also hold for Au and Ag,
which, assuming a 1+ oxidation state, are stabilized in low coordination number (II-fold to
IV-fold) complexes (Carvajal et al., 2004). Therefore, although the PGE may be soluble in
the MSS, their relative preference for the melt or solid phase appears to depend on which
coordination environment is most energetically favored.
Although the data from this study are generally consistent with previous results, our partition
coefficients for the compatible PGE, Ru, Re, Os, Ir and Rh, seem to be systematically higher
for a given MSS composition. Results from Mungall et al (2005) are similarly offset. Both
this study and Mungall et al (2005) employed buffering techniques to fix fO2 at the FMQ
buffer, resulting in sulfide melt with oxygen contents of 1-2 wt%. This is in contrast to past
experiments which were unbuffered, and nominally oxygen free. There is a most
pronounced influence on the solubility of the PGE in the presence of dissolved oxygen.
Results of previous solubility studies have documented a sharp decrease in Re, Os, Ir, Ru and
Pt solubility in sulfide melt at fO2 of ~FMQ-2 to -3 (depending on the metal, and the fS2;
Fonseca et al.; 2007; 2009; 2011; Andrews and Brenan, 2002), corresponding to a sharp rise
in the oxygen content of the sulfide liquid from nil to ~1-5 wt%. The solubility decrease
over this interval is ~10-fold, and implies a complementary increase in the activity
coefficient for these metals in the melt. The effect of an increase in the activity coefficient
for a metal in the melt phase would be to increase D(MSS/melt), which is the sense of the
85
offset noted above. In this context, it is also worth mentioning that the addition of Cu and Ni
to an Fe-S melt composition has been shown to change the solubility of Ru, Ir and Os
(Fonseca et al. 2007; 2009; 2011; Andrews and Brenan, 2002; Brenan 2008) with these
additives acting in opposite ways. Whereas Ni increases the solubility of these metals (e.g.,
0-23 wt% Ni results in ~2-fold increase in Os solubility), Cu results in a decrease (e.g., 0-26
wt% Cu results in a 3-fold drop in Os solubility; Fonseca et al., 2011), implying sympathetic
changes in the activity coefficients for these PGE in the melt phase. Hence, the relatively
high copper content of the melts produced in this study (~30 wt%) compared to previous
work (~4 to 13 wt%) would also result in a modest increase in partition coefficients.
From this previous discussion, the variation in D with M/S ratio for the compatible PGE
seems consistent with known activity-composition relations in the sulfide melt, however the
origins of the significant differences in partitioning seen for the incompatible PGE, Au, and
to lesser extent Pt and Pd are less clear. For MSS with a similar range in M/S, values of
D(MSS/melt) for Au are found to vary by ~10-fold, with results from Li and Audetat (2013)
and Fleet et al (1993) recording higher values than past determinations. Unlike past
experiments, in which the PGE were added at ppm to low wt% levels, experiments done by
Li and Audetat (2013) were at saturation in pure Au, corresponding to 9 to ~15 wt% Au in
the sulfide liquid. The effect of such high metal loading on partitioning is unknown, but
could very well be outside the concentration limits of Henryian behavior, and is certainly
beyond natural abundance levels. Fleet et al (1993) also measured elevated values of
D(MSS/melt) for Au, as well as Pt and Pd. PGE dopant levels were low, so Henryian
behavior is not likely to be an issue, and the composition of MSS and sulfide melt are similar
86
to previous work. The only difference in method was the use of SIMS for sample analysis,
with partition coefficients for Au determined using the ratio of sulfur-normalized count rates
in the MSS and sulfide melt. As documented by Fleet et al (1993), this is a robust technique
for measuring Au in sulfides. However, it is possible that the rather small spot employed
(20-30 microns), and small number of analyses acquired (2) might not have fully captured
the true variation in the Au content of the texturally inhomogeneous quenched sulfide melt.
In terms of the behavior of the chalcogens, like sulfur, Se, Te and As show partitioning that
is nearly invariant with M/S in the MSS (Figure 14), suggesting a one-to-one substitution for
sulfur. Helmy et al. (2010) document a decrease in the partition coefficients for Te and As
with decreasing M/S, which might reflect a change in metal speciation over the conditions
investigated. MSS-melt partition coefficients for these elements measured in this work,
Helmy et al (2010) and Brenan (2015) are generally in agreement, however. Sn, Sb, Bi and
Pb are the most incompatible elements (Table 3-6) in MSS (D<0.01), which possibly reflects
the significant steric adjustment in the sulfide lattice (Makovicky, 2006) required to
accommodate the non-bonding lone s2 electron pair, assuming the typical oxidation states for
these elements (i.e., Sn2+
, Sb3+
, Bi3+
, and Pb2+
).
3.6 Summary and Conclusions
1 ) Partitioning experiments for PGE and chalcogens have been conducted in sealed
silica tubes at controlled fO2 (FMQ) and fS2 (similar level to Pt-PtS) between MSS-
87
melt and MSS-ISS, temperatures ranging from 850 °C to 930 °C. Measured MSS-
melt partitioning coefficients of PGEs were compared and found not significantly
different from available literature data, except for the IPGEs, which are in general
larger, suggesting the effects of the dissolved oxygen in the melt.
2) The studied chalcogens were found to be incompatible between MSS and melt,
except for Se which showed almost no significant preference in one over the other.
The effect of chalcogens on PGE partition coefficients was evaluated, and no
significant impact was observed for the experimental conditions investigated as
opposed to the work of Helmy et al. (2013a). The discrepancy between these results
is suspected to be due to the large difference in oxygen and sulfur fugacities between
the two studies.
3) The differences in D(MSS/melt) among these PGEs are not consistent with their
relative solubilities in MSS, and no simple systematics can be regressed based on the
D(MSS/melt) values vs the M/S of the MSS. The origin of these differences was then
speculated to be related not only to the required steric adjustment in MSS crystal
structure, but also to the coexisting melt’s composition, especially for those elements
whose proper coordination site is missing in MSS lattice (e.g. Pt, Pd, preferred planar
coordinated).
88
4) ISS was successfully stabilized in the experiments together with MSS, featured by
lamellae texture and similar oxygen content as MSS. D(MSS/ISS) was measured for
PGEs and chalcogens under controlled fO2 and fS2 conditions, and the first set of
D(ISS/melt) was calculated combining D(MSS/melt) and D(MSS/ISS). All the PGEs
are consistently incompatible between ISS and melt, while for the chalcogens, S
Group elements( S, Se, Te) showed no preference between ISS and melt, whereas the
As group elements (As, Sb, Bi) all appear incompatible in ISS.
89
Figure 3-1. Illustration of the sample configuration used in this study. Pre-melted Fe-Ni-
Cu-S powder doped with PGE/chalcogens was packed down at the bottom of
the tube, followed by a layer of high purity silica powder. Buffers were
positioned directly above the silica powder, then a tight-fitting silica rod
spacer to eliminate headspace.
90
Figure 3-2. Backscattered electron image showing the buffer phases from experiment
yn51, done at 915 ºC, for 3 days. EDS and XRD analysis confirmed that
fayalite (Fa) – magnetite (Mt) – pyrrhotite (Po) – quartz (Qz) are still
preserved.
91
Figure 3-3. Comparison between sulfur fugacities imposed by the solid buffer assemblage
(as indicated) with values calculated from the composition of coexisting
pyrrhotite (method of Toulmin and Barton, 1964).
92
Figure 3-4. Summary of logfO2, logfS2 and temperature for experiments conducted in this
study. Values of FMQ are calculated from O’Neill (1987), and the Pt-PtS
reference curve from thermodynamic data in Barin (1987). Measured values
of fS2 plot above the Pt-PtS buffer, and are consistent with the fS2 requried to
saturate a silicate melt with 10 mole FeO in sulfide liquid (curve labeled XFeO
= 0.1).
93
a)
b)
Figure 3-5. Backscattered electronic images showing different phase textures present in
experiments. a) Clear phase separation can be observed for experiment yn51,
performed to measure the partitioning between MSS and sulfide melt at 915
ºC, for duration of 3 days. MSS sinks to the bottom of the tube, with a
homogeneous appearance. Sulfide melt lies above MSS, with a dendritic
quench texture, together with abundant gas bubbles. b) Homogeneous MSS
crystals coexists with ISS, which is distinguished by bright Cu-rich stripes
(bornite-like composition), exhibiting a trellis-like pattern with a Cu-poor
(chalcopyrite like) matrix. Upper right corner shows an example of the melt
phase texture from experiment yn51, in which the sulfide liquid contains 26
wt% Cu.
melt
94
a)
b)
Figure 3-6. Identification of ISS. a) Oxygen content in MSS, ISS and melt versus the
temperature of the experiments. The sulfide liquid is distinguished from MSS
and ISS as it contains significant amounts of oxygen, whereas MSS and ISS
tend to have similar and low levels of oxygen regardless of temperature. b)
Variation in Cu as a function of Fe (at%) comparing the compositional
variation of ISS, MSS, and sulfide melt produced in this study with those
from past work. The limits for the composition of ISS are taken from
Raghavan (2004) from the Cu-Fe-S ternary system. The composition of ISS
produced in the experiments of this study agree well with the range defined by
the phase diagram, and the previous experimental work of Fleet and Pan
(1994) and Helmy et al (2007).
95
Figure 3-7. D(MSS/melt) for all the PGEs studied, sorted from the most to the least
compatible in MSS. Averaged values of all the MSS/melt partitioning
experiments are plotted in blue squares, while other symbols represent
previous literature data, for comparison.
96
Figure 3-8. Comparison of partition coefficients for the chalcogens and other chalcophile
elements measured in this study, with previous results from Helmy et al.
(2010) and Li and Audetat (2012).
97
a)
b)
Figure 3-9. MSS/melt partition coefficients of all the PGEs studied, versus the As content
of the coexisting melt phase. a) compatible PGEs; b) incompatible PGEs. The
sample with the lowest arsenic content contains undetectable As (<5 ppm), so
the value plotted is a maximum.
98
a)
b)
99
c)
Figure 3-10. a) Experimentally determined D(MSS/ISS) for all the PGEs studied,
compared with the averaged D(MSS/melt) measured in this study, and the
values from Jugo et al. (1999) on Au partitioning between pyrrhotite and ISS.
10b) Experimentally determined D(MSS/ISS) for all the chalcogens from
different runs, compared with the averaged D(MSS/melt) measured in this
study. Experiment yn73 was analyzed twice at 6 months apart to evaluate the
reproducibility between different analytical sessions. Our results are self-
consistent between different experimental durations and analytical sessions.
10c) Difference in D(MSS/ISS) of Te between this study and Helmy et al.
(2007) versus the Cu/(Cu+Fe+Ni) of ISS.
100
a)
b)
Figure 3-11. a) Calculated D(ISS/melt) based on the experimentally determined
D(MSS/melt) and D(MSS/ISS) for all the PGEs studied. Values of
D(MSS/melt) are included for comparison. b) Calculated D(ISS/melt) based
on the experimentally determined D(MSS/melt) and D(MSS/ISS) for the
chalcogens
101
a)
b)
c)
Figure 3-12. Summary of the solubility for Rh,
Pd, Pt and Os in MSS as a function of Metal/sulfur.
Results in a) and c) are saturated in the pure metal
phase (corrected to unit activity, where appropriate,
for alloys containing Fe), whereas in b), MSS
coexists with PdS or PtS.
102
a)
b)
c)
d)
e)
f)
103
g)
h)
i)
j)
Figure 3-13. Summary of MSS/sulfide melt partition coefficients for the PGE and select
chalcophile elements plotted as a function of the metal/sulfur (M/S) ratio of
the coexisting MSS. With the exception of the studies by Mungall et al.
(2005), Li and Audetat (2012; 2013), and this work, all other experiments
were done unbuffered, with the sulfide liquid nominally oxygen-free. Note
that elements which comprise the Cu triad (Cu, Au) do not seem to be
affected by the composition of the MSS. Figure 13j for Au is in log scale
instead of linear scale as for the other elements.
104
Figure 3-14. Summary of MSS/sulfide melt partitioning of the chalcogens as a function of
the M/S ratio of the MSS from this and past studies. Filled symbols represent
data from this study and Brenan (2015), while half-filled symbols represent
data from Helmy et al. (2010). S and Se appear unaffected by MSS
composition, consistent with their incorporation as anions.
105
Table 3-1. Nominal Composition of starting materials
Fe Cu Ni S
MSSY2 (wt%) 45.0 16.5 5.7 32.1
Au Ag Pt Ir Re Rh Os Pd Ru
+ PGE doped (ppm) 87.3 58.4 46.9 66.2 54.7 47.1 48.8 70.6 53.6
As Te Se Sb Bi
+ chalcogen doped (ppm) 143.0 117.7 137.7 155.5 292.9
Fe Cu Ni S
ISS#2 (wt%) 35.6 26.1 6.7 31.7
Au Ag Pt Ir Re Rh Os Pd Ru
+ PGE doped (ppm) 179.2 75.1 52.7 123.3 122.9 69.3 170.7 83.4 154.0
As Te Se Sb Bi
+ chalcogen doped (ppm) 145.8 120.1 140.4 158.5 298.7
Zn Sn Pb
+ Zn, Sn, Pb doped (ppm) 91.0 97.6 89.0
106
Table 3-2. Composition of MSS5
Elements Conc. (ppm)
Fe 579300
S 397600
Ni 10526
Cu 240.13
Mn 6800
Ru 29
Rh 80
Pd 64
Re 32.4
Os 72
Ir 56
Pt 56
Au 28.8
Ag 60.75
As 67.4
Se 76.33
Sb 60.55
Te 45.83
Pb 71.63
Bi 79.9
107
Table 3-3. Major element analysis by microprobe with run conditions, in elemental weight percent
Run T in C Days logfo2 logfs2 Phases Fe Ni Cu S O Totals M/S
yn50 926 5 -12.3 -2.54 MSS+melt yn50 MSS 53.99 5.05 3.75 36.71 0.08 99.59 0.97
std 0.09 0.05 0.19 0.24 0.08 0.17
yn50 melt 37.56 6.19 24.57 29.89 2.05 100.27 1.10
std 0.61 0.12 0.26 0.15 0.12 0.44
yn51 915 3 -12.5 -2.41 MSS+melt yn51 MSS 53.54 5.11 4.17 36.53 0.11 99.41 0.97
std 0.17 0.05 0.25 0.14 0.10 0.23
yn51 melt 36.23 6.23 26.37 29.87 1.74 100.44 1.13
std 0.52 0.25 0.69 0.14 0.17 0.19
yn52 903 4 -12.7 -2.57 MSS+melt yn52 MSS 53.40 5.33 4.24 36.46 0.14 99.56 0.97
std 0.12 0.04 0.27 0.20 0.13 0.28
yn52 melt 35.42 6.49 27.05 29.61 1.77 100.35 1.13
std 0.76 0.28 0.83 0.12 0.15 0.11
yn53 901 3 -12.8 -2.41 MSS+melt yn53 MSS 53.30 5.23 3.98 36.71 0.25 99.57 0.95
std 0.04 0.05 0.05 0.07 0.07 0.13
yn53 melt 33.30 6.01 29.67 29.32 2.41 100.84 1.09
std 0.88 0.44 1.10 0.42 0.30 0.39
yn55 915 5 -12.5 -2.18 MSS+melt yn55 MSS 53.78 5.07 3.91 36.15 0.44 99.46 0.96
std 0.12 0.07 0.14 0.29 0.25 0.39
yn55 melt 35.24 6.19 27.72 28.92 2.61 100.79 1.10
std 0.12 0.17 0.28 0.20 0.29 0.19
yn56 885 4 -13 -2.77 MSS+melt yn56 MSS 53.51 5.36 4.12 37.66 < D.L.* 100.74 0.95
std 0.17 0.02 0.09 0.21 0.33
yn56 melt 32.90 6.19 29.52 29.37 2.11 100.10 1.11
std 1.26 0.22 1.97 0.77 0.31 0.41
yn58a 900 4 -12.8 -2.77 MSS+melt yn58a MSS 53.92 5.19 3.94 37.24 < D.L. 100.39 0.96
std 0.28 0.04 0.04 0.20 0.32
yn58a melt 34.97 5.84 28.18 30.06 2.08 101.28 1.10
std 1.12 0.41 1.39 0.35 0.17 0.15
yn58b 900 4 -12.8 -2.24 MSS+melt yn58b MSS 52.87 4.85 3.74 37.44 < D.L. 99.06 0.93
108
std 0.49 0.04 0.04 0.14 0.53
yn58b melt 33.92 5.76 28.13 30.11 1.81 99.87 1.09
std 0.47 0.19 0.28 0.41 0.19 0.41
yn58c 900 4 -12.8 -2.26 MSS+melt yn58c MSS 53.67 4.98 3.77 36.82 < D.L. 99.33 0.96
std 0.41 0.06 0.04 0.06 0.53
yn58c melt 35.69 5.69 26.23 30.18 1.88 99.80 1.09
std 1.78 0.16 1.54 0.66 0.33 0.83
yn60 900 3 -12.8 -2.3 MSS+melt yn60 MSS 53.14 5.18 4.57 36.30 < D.L. 99.43 0.98
std 0.50 0.06 0.58 0.27 0.25
yn60 melt 34.94 5.76 27.39 29.72 2.16 100.04 1.09
std 1.73 0.72 2.60 0.48 0.22 0.35
yn57 850 3 -13.7 -2.73 MSS+ISS yn57 MSS 52.41 6.53 4.32 36.29 0.34 99.89 0.97
Std 0.20 0.08 0.16 0.25 0.10 0.30
yn57 ISS 30.28 5.50 32.94 31.17 0.37 100.27 1.16
std 0.49 0.27 0.35 0.11 0.02 0.34
yn54 875 3 -13.2 -2.83 MSS+ISS yn54 MSS 51.74 6.34 5.07 36.54 0.18 99.86 0.97
Std 0.42 0.03 0.39 0.16 0.06 0.12
yn54 ISS 29.12 4.40 34.92 30.87 0.39 99.69 1.16
Std 0.71 0.42 1.03 0.48 0.07 0.59
yn62 900 3 -12.8 -2.48 MSS+melt yn62 MSS 54.04 4.66 4.09 37.33 < D.L. 100.33 0.95
Std 0.23 0.08 0.07 0.16 0.39
yn62 melt 35.25 5.09 28.62 29.88 1.77 100.77 1.12
Std 0.56 0.23 0.62 0.12 0.14 0.24
yn63 885 8 -13 -2.38 MSS+melt yn63 MSS 52.80 5.38 4.47 36.77 < D.L. 99.53 0.96
std 0.81 0.03 0.10 0.58 0.79
yn63 melt 31.29 5.16 31.73 28.31 2.09 99.19 1.13
std 2.36 0.36 3.37 1.07 0.50 0.43
yn71 900 1 -12.8 -2.57 MSS+melt yn71 MSS 52.70 5.34 4.49 36.29 < D.L. 99.02 0.98
std 0.24 0.06 0.24 0.33 0.37
yn71 melt 34.95 5.93 27.67 29.61 1.51 99.91 1.14
std 0.71 0.62 1.21 0.34 0.14 0.53
yn73 866 3 -13.4 -2.77 MSS+ISS yn73 MSS 51.61 6.27 4.84 36.22 0.07 99.01 0.98
std 0.32 0.14 0.23 0.28 0.03 0.31
yn73 ISS 28.25 4.16 36.58 29.92 0.31 99.23 1.21
109
std 0.46 0.58 0.74 0.28 0.05 0.33
yn80 860 1 -13.5 -2.95 MSS+ISS yn80 MSS 48.61 8.71 6.50 35.61 0.47 99.90 0.98
std 0.47 0.17 0.83 0.26 0.08 0.23
yn80 ISS 27.54 6.39 36.18 29.87 0.59 100.57 1.21
std 0.72 0.22 0.97 0.29 0.12 0.28
yn81 860 1 -13.5 -2.56 MSS+ISS yn81 MSS 49.43 9.01 5.62 35.42 0.34 99.82 1.00
std 0.04 0.11 0.52 0.06 0.10 0.27
yn81 ISS 28.20 6.66 35.27 29.98 0.49 100.60 1.22
std 0.12 0.32 0.07 0.10 0.07 0.24
yn83 860 7 -13.5 -2.33 MSS+ISS yn83 MSS 49.73 8.63 5.04 36.14 0.06 99.60 0.99
std 0.09 0.09 0.25 0.13 0.02 0.10
yn83 ISS 28.21 6.37 34.87 30.21 0.15 99.79 1.22
std 0.59 0.21 0.66 0.27 0.02 0.22
yn88 860 3 -13.5 -2.60 MSS+ISS yn88 MSS 49.81 8.06 6.17 34.92 0.59 99.55 1.00
std 0.87 0.14 0.93 0.22 0.20 0.41
yn88 ISS 28.47 5.53 35.92 29.82 0.86 100.60 1.19
std 0.56 0.50 1.36 0.24 0.05 0.23
yn89 902 2 -12.7 -3.02 MSS+melt yn89 MSS 53.93 5.05 4.47 35.78 < D.L. 99.24 1.00
std 0.16 0.03 0.22 0.17 0.24
yn89 melt 36.94 5.98 26.01 30.11 1.30 100.34 1.15
std 2.17 0.31 2.83 0.66 0.07 0.19
yn90 900 1 -12.8 -2.46 MSS+melt yn90 MSS 52.37 5.44 5.17 35.87 < D.L. 98.85 0.99
std 0.32 0.15 0.35 0.28 0.22
yn90 melt 34.22 5.00 29.67 29.31 1.55 99.75 1.15
std 1.36 0.47 1.96 0.29 0.11 0.35
* < D.L.= below detection limit
110
Table 3-4. Trace elemental analyses and partition coefficients for precious metals (ppm)
Line Ru101 Rh103 Pd108 Ag109 Re185 Os189 Ir193 Pt194 Au197
yn50 MSS 2.01E+02 9.35E+01 6.44E+00 1.24E+00 1.40E+02 1.70E+02 1.66E+02 3.36E+00 6.82E-01
std 4.20E+00 1.39E+00 5.15E-01 3.26E-01 2.68E+00 6.88E+00 1.75E+00 5.45E-01 1.12E-01
yn50 melt 8.39E+00 2.20E+01 8.49E+01 1.03E+02 2.25E+01 1.23E+01 2.30E+01 1.08E+02 1.28E+02
std 1.83E+00 1.72E+00 3.48E+00 4.24E+00 1.62E+00 1.88E+00 2.44E+00 5.06E+00 3.78E+00
D(MSS/melt) 2.40E+01 4.25E+00 7.59E-02 1.20E-02 6.22E+00 1.38E+01 7.22E+00 3.11E-02 5.33E-03
error in D**
5.25E+00 3.38E-01 6.82E-03 3.20E-03 4.64E-01 2.19E+00 7.69E-01 5.25E-03 8.89E-04
yn51 MSS 1.48E+02 8.07E+01 6.80E+00 1.11E+00 1.19E+02 1.14E+02 1.48E+02 3.22E+00 6.02E-01
std 1.26E+01 1.65E-01 4.52E-01 1.42E-01 3.65E+00 7.86E+00 4.56E+00 7.35E-02 7.46E-02
yn51 melt 8.64E+00 1.93E+01 9.07E+01 1.16E+02 2.11E+01 1.11E+01 2.21E+01 1.03E+02 1.19E+02
std 2.02E+00 2.62E+00 7.91E+00 9.23E+00 3.71E+00 2.15E+00 3.92E+00 7.99E+00 7.42E+00
D(MSS/melt) 1.71E+01 4.18E+00 7.49E-02 9.58E-03 5.62E+00 1.02E+01 6.70E+00 3.11E-02 5.07E-03
error in D 4.25E+00 5.68E-01 8.21E-03 1.44E-03 1.00E+00 2.09E+00 1.20E+00 2.51E-03 7.03E-04
yn52 MSS 1.18E+02 7.45E+01 6.64E+00 1.09E+00 1.22E+02 1.09E+02 1.20E+02 3.42E+00 5.94E-01
std 8.41E+00 9.12E-01 3.40E-01 1.25E-01 3.69E+00 6.88E+00 4.32E+00 3.88E-01 1.75E-02
yn52 melt 4.87E+00 1.37E+01 1.07E+02 1.29E+02 1.71E+01 8.14E+00 1.27E+01 1.24E+02 1.35E+02
std 1.00E+00 8.74E-01 4.68E+00 5.35E+00 1.32E+00 7.17E-01 1.30E+00 4.80E+00 8.55E+00
D(MSS/melt) 2.41E+01 5.44E+00 6.21E-02 8.45E-03 7.13E+00 1.34E+01 9.45E+00 2.76E-02 4.40E-03
error in D 5.27E+00 3.53E-01 4.18E-03 1.03E-03 5.92E-01 1.45E+00 1.03E+00 3.31E-03 3.07E-04
111
yn60 MSS 4.32E+01 4.03E+01 3.70E+00 5.78E-01 5.98E+01 4.29E+01 6.78E+01 1.69E+00 3.62E-01
std 7.03E+00 1.16E+00 6.10E-01 5.74E-02 6.11E+00 7.23E+00 9.42E+00 1.84E-01 7.32E-02
yn60 melt 2.58E+00 8.11E+00 5.03E+01 7.00E+01 9.99E+00 4.03E+00 1.02E+01 5.74E+01 6.59E+01
std 6.41E-01 9.77E-01 4.93E+00 7.37E+00 6.77E-01 9.68E-01 1.15E+00 6.10E+00 4.27E+00
D(MSS/melt) 1.67E+01 4.97E+00 7.36E-02 8.25E-03 5.99E+00 1.06E+01 6.68E+00 2.95E-02 5.49E-03
error in D 4.97E+00 6.15E-01 1.41E-02 1.19E-03 7.34E-01 3.12E+00 1.02E+00 4.49E-03 1.17E-03
yn57 MSS 1.38E+02 1.20E+02 2.91E+01 4.22E+00 1.69E+02 1.13E+02 1.64E+02 2.35E+01 2.53E+00
Std 4.34E+00 4.85E+00 5.30E-01 8.03E-01 5.48E+00 4.60E+00 6.22E+00 1.30E+00 2.93E-01
yn57 ISS 2.55E+00 1.58E+00 1.40E+02 1.95E+02 1.64E+00 1.97E+00 1.26E+00 1.50E+02 1.38E+02
std 6.77E-01 6.25E-01 1.13E+01 8.75E+00 1.22E+00 5.62E-01 6.86E-01 4.00E+01 5.20E+01
D(MSS/ISS) 5.41E+01 7.58E+01 2.08E-01 2.16E-02 1.03E+02 5.74E+01 1.30E+02 1.57E-01 1.83E-02
error in D 1.45E+01 3.01E+01 1.72E-02 4.23E-03 7.67E+01 1.65E+01 7.10E+01 4.27E-02 7.23E-03
yn54 MSS 8.41E+01 6.79E+01 2.50E+01 4.99E+00 8.71E+01 5.77E+01 9.83E+01 2.56E+01 2.59E+00
Std 9.30E+00 3.97E+00 4.63E-01 1.42E+00 6.11E+00 6.49E+00 9.32E+00 9.93E-01 1.41E+00
yn54 ISS 1.76E+00 1.61E+00 1.30E+02 1.83E+02 1.09E+00 1.33E+00 1.57E+00 1.91E+02 2.46E+02
Std 3.25E-01 2.91E-01 1.87E+01 9.41E+00 3.18E-01 4.28E-01 1.77E-01 5.09E+01 1.03E+02
D(MSS/ISS) 4.77E+01 4.22E+01 1.92E-01 2.73E-02 7.99E+01 4.34E+01 6.26E+01 1.34E-01 1.05E-02
error in D 1.02E+01 8.01E+00 2.79E-02 7.89E-03 2.40E+01 1.48E+01 9.22E+00 3.61E-02 7.23E-03
yn62 MSS 1.04E+02 6.49E+01 8.09E+00 1.21E+00 9.61E+01 9.67E+01 1.15E+02 2.78E+00 5.45E-01
Std 2.92E+00 1.71E+00 6.07E-01 1.55E-01 6.35E+00 4.39E+00 5.97E+00 3.31E-01 7.52E-02
112
yn62 melt 2.58E+00 1.32E+01 1.08E+02 1.22E+02 1.17E+01 4.79E+00 1.24E+01 1.15E+02 1.23E+02
Std 6.38E-01 1.73E+00 1.34E+01 9.91E+00 1.41E+00 1.50E+00 2.57E+00 1.09E+01 1.29E+01
D(MSS/melt) 4.05E+01 4.92E+00 7.49E-02 9.92E-03 8.21E+00 2.02E+01 9.27E+00 2.42E-02 4.43E-03
error in D 1.01E+01 6.57E-01 1.09E-02 1.50E-03 1.13E+00 6.39E+00 1.98E+00 3.68E-03 7.68E-04
yn71 MSS 1.56E+02 9.19E+01 7.97E+00 1.85E+00 1.42E+02 1.48E+02 1.68E+02 3.83E+00 6.68E-01
std 2.51E+01 1.62E+00 2.95E-01 3.67E-01 8.85E+00 2.19E+01 8.79E+00 4.09E-01 1.46E-01
yn71 melt 5.31E+00 1.40E+01 9.24E+01 1.45E+02 1.61E+01 6.46E+00 1.35E+01 9.54E+01 1.28E+02
std 1.49E+00 1.87E+00 9.52E+00 4.67E+00 2.13E+00 9.83E-01 2.19E+00 1.70E+01 1.19E+01
D(MSS/melt) 2.95E+01 6.56E+00 8.63E-02 1.28E-02 8.82E+00 2.29E+01 1.24E+01 4.01E-02 5.22E-03
error in D 9.51E+00 8.84E-01 9.45E-03 2.57E-03 1.29E+00 4.86E+00 2.11E+00 8.33E-03 1.24E-03
yn80 MSS 6.44E+02 2.85E+02 2.13E+01 6.16E+00 5.38E+02 3.18E+02 5.34E+02 1.89E+01 1.65E+00
std 3.62E+01 9.93E+00 5.84E-01 9.28E-01 5.50E+01 1.85E+01 4.26E+01 8.14E-01 8.59E-02
yn80 ISS 3.24E+00 6.09E+00 1.11E+02 1.47E+02 6.43E+00 1.92E+00 6.55E+00 9.86E+01 1.80E+02
std 8.13E-01 8.99E-01 6.19E+00 8.91E+00 1.33E-01 3.13E-01 5.82E-01 1.25E+01 3.83E+01
D(MSS/ISS) 1.99E+02 4.68E+01 1.92E-01 4.19E-02 8.37E+01 1.66E+02 8.15E+01 1.92E-01 9.17E-03
error in D 5.12E+01 7.10E+00 1.19E-02 6.80E-03 8.73E+00 2.87E+01 9.74E+00 2.57E-02 2.01E-03
yn90 MSS 5.43E+02 3.58E+02 1.48E+01 4.88E+00 4.59E+02 1.25E+03 2.54E+03 7.17E+00 8.47E-01
std 8.21E+01 1.08E+01 4.19E+00 5.07E+00 2.29E+01 1.69E+02 2.34E+02 3.65E+00 4.04E-02
yn90 melt 2.33E+01 5.53E+01 1.12E+02 1.84E+02 6.27E+01 6.92E+01 1.79E+02 1.10E+02 2.23E+02
std 5.51E+00 7.13E+00 1.44E+01 2.76E+01 9.64E+00 1.66E+01 3.67E+01 1.53E+01 4.40E+01
113
D(MSS/melt) 2.33E+01 6.48E+00 1.33E-01 2.66E-02 7.32E+00 1.81E+01 1.42E+01 6.53E-02 3.79E-03
error in D 6.55E+00 8.58E-01 4.12E-02 2.79E-02 1.18E+00 4.99E+00 3.21E+00 3.45E-02 7.68E-04
Detection Limit 5.11E-01 1.52E-01 4.61E-01 3.30E-01 1.26E-01 2.27E-01 1.22E-01 3.49E-01 6.77E-02
** error in D was calculated through :
where dMSSiC is the uncertainty in
MSSiC measurement, d
Melt
iC is the uncertainty in Melt
iC measurement,
estimated from the standard deviation of multiple analyses of the same phase.
)()()/()/( 22
melt
i
melt
i
MSS
i
MSS
i
C
dC
C
dCmeltMSSDmeltMSSdD
114
Table 3-5. Trace analysis and partition coefficients for chalcogens (ppm)
Line As75 Se82 Sb121 Te128 Bi209
yn53MSS 3.21E+01 1.02E+02 6.10E-01 7.16E+00 6.62E-01
std 7.89E-01 1.14E+01 1.38E-01 1.20E+00 1.88E-01
yn53melt 2.43E+02 1.67E+02 2.60E+02 2.20E+02 5.42E+02
std 3.01E+01 1.20E+01 6.13E+01 2.75E+01 1.43E+02
D(MSS/melt) 1.32E-01 6.11E-01 2.35E-03 3.25E-02 1.22E-03
error in D 1.67E-02 8.12E-02 7.68E-04 6.80E-03 4.73E-04
yn55MSS 2.97E+01 9.14E+01 5.92E-01 5.96E+00 8.66E-01
std 1.38E+00 5.01E+00 1.06E-01 9.67E-01 3.89E-01
yn55melt 1.89E+02 1.53E+02 1.97E+02 1.84E+02 3.91E+02
std 1.03E+01 1.07E+01 2.21E+01 1.01E+01 5.50E+01
D(MSS/melt) 1.57E-01 5.97E-01 3.01E-03 3.24E-02 2.21E-03
error in D 1.12E-02 5.30E-02 6.36E-04 5.55E-03 1.04E-03
yn56MSS 3.39E+01 9.61E+01 7.54E-01 6.90E+00 6.35E-01
std 9.99E-01 4.88E+00 1.01E-01 5.97E-01 4.09E-02
yn56melt 2.35E+02 1.71E+02 2.89E+02 2.70E+02 5.01E+02
std 6.60E+00 7.20E+00 2.01E+01 1.06E+01 5.02E+01
D(MSS/melt) 1.44E-01 5.62E-01 2.61E-03 2.56E-02 1.27E-03
115
error in D 5.86E-03 3.71E-02 3.94E-04 2.43E-03 1.51E-04
yn58aMSS 3.14E+01 9.95E+01 6.06E-01 7.31E+00 7.02E-01
std 2.00E+00 5.10E+00 1.55E-01 1.63E+00 1.15E-01
yn58amelt 3.26E+02 1.74E+02 3.40E+02 2.97E+02 7.08E+02
std 7.72E+01 1.02E+01 1.03E+02 7.86E+01 2.04E+02
D(MSS/melt) 9.63E-02 5.72E-01 1.78E-03 2.46E-02 9.92E-04
error in D 2.36E-02 4.45E-02 7.06E-04 8.51E-03 3.29E-04
yn58bMSS 2.79E+01 8.73E+01 6.80E-01 6.56E+00 5.90E-01
std 1.38E+00 4.06E+00 8.21E-02 1.29E+00 9.67E-02
yn58bmelt 2.20E+02 1.47E+02 2.38E+02 2.04E+02 4.78E+02
std 1.26E+01 1.57E+01 2.07E+01 1.37E+01 5.46E+01
D(MSS/melt) 1.27E-01 5.94E-01 2.86E-03 3.22E-02 1.23E-03
error in D 9.61E-03 6.92E-02 4.26E-04 6.69E-03 2.46E-04
yn58cMSS 2.97E+01 9.32E+01 6.28E-01 7.87E+00 6.89E-01
std 1.65E+00 5.40E+00 1.54E-01 1.70E+00 1.58E-01
yn58cmelt 1.51E+02 1.44E+02 1.29E+02 1.75E+02 2.32E+02
std 2.89E+01 1.21E+01 6.07E+01 2.10E+01 1.21E+02
D(MSS/melt) 1.97E-01 6.47E-01 4.87E-03 4.50E-02 2.97E-03
error in D 3.93E-02 6.60E-02 2.58E-03 1.11E-02 1.69E-03
116
yn60MSS 2.04E+01 4.91E+01 4.70E+00 9.26E+01 2.87E+00
std 5.07E+00 1.31E+01 2.31E-01 2.99E+00 3.39E-01
yn60melt 1.35E+02 7.94E+01 1.57E+03 2.49E+03 1.72E+03
std 9.39E+00 1.72E+01 1.42E+02 1.53E+02 1.43E+02
D(MSS/melt) 1.51E-01 6.18E-01 2.99E-03 3.72E-02 1.67E-03
error in D 3.09E-02 2.12E-01 3.08E-04 2.58E-03 2.41E-04
yn63MSS 3.15E+01 8.06E+01 8.72E+00 1.68E+02 5.48E+00
std 5.28E+00 4.36E+00 5.91E-01 6.85E+00 3.25E-01
yn63melt 1.89E+02 1.11E+02 2.58E+03 4.83E+03 3.14E+03
std 4.46E+01 7.80E+00 9.82E+02 1.07E+03 1.27E+03
D(MSS/melt) 1.67E-01 7.26E-01 3.38E-03 3.48E-02 1.75E-03
error in D 4.83E-02 6.44E-02 1.31E-03 7.84E-03 7.15E-04
yn73MSS 6.20E+01 8.10E+01 2.43E+01 2.31E+02 1.70E+01
std 5.68E+00 5.73E+00 2.11E+00 3.61E+01 3.88E+00
yn73ISS 7.48E+01 1.24E+02 7.10E+02 4.22E+03 7.71E+02
std 1.47E+01 5.55E+00 1.10E+02 2.66E+02 1.27E+02
D(MSS/ISS) 8.29E-01 6.53E-01 3.42E-02 5.47E-02 2.20E-02
error in D 1.80E-01 5.47E-02 6.07E-03 9.22E-03 6.19E-03
yn81MSS 1.20E+02 9.04E+01 8.17E+00 1.27E+01 1.39E+01
std 9.19E+00 1.35E+01 1.34E+00 1.46E+00 5.74E-01
117
yn81ISS 1.38E+02 1.64E+02 1.33E+02 2.02E+02 4.33E+02
std 1.32E+01 3.32E+01 2.89E+01 2.81E+01 1.22E+02
D(MSS/ISS) 8.70E-01 5.51E-01 6.14E-02 6.29E-02 3.21E-02
error in D 1.07E-01 1.39E-01 1.67E-02 1.14E-02 9.14E-03
yn83MSS 1.07E+02 9.44E+01 6.19E+00 1.37E+01 1.60E+01
std 6.51E+00 1.10E+01 8.29E-01 5.91E+00 7.74E-01
yn83ISS 1.28E+02 1.40E+02 1.04E+02 1.74E+02 4.32E+02
std 7.00E+00 2.76E+01 6.35E+00 2.54E+01 3.13E+01
D(MSS/ISS) 8.36E-01 6.74E-01 5.95E-02 7.87E-02 3.70E-02
error in D 6.84E-02 1.54E-01 8.76E-03 3.58E-02 3.22E-03
yn50 melt 4.05E+01 < D.L. 9.90E+00 < D.L. 7.48E+01
std 1.07E+01
1.01E+00
7.66E+00
yn51melt 4.59E+01 < D.L. 9.78E+00 < D.L. 3.35E+01
std 1.06E+01
9.20E-01
2.98E+00
yn52 melt 5.39E+01 < D.L. 1.25E+01 < D.L. 4.37E+01
std 1.53E+01
1.48E+00
5.41E+00
yn71melt 6.84E+01 < D.L. 3.38E+01 < D.L. 1.13E+02
std 1.20E+01
2.11E+00
1.09E+01
yn90 melt < D.L. < D.L. 2.14E+00 < D.L. 2.00E+00
std
1.33E+00
8.76E-01
Detection Limits 5.41E+00 8.23E+00 3.47E-01 1.93E+00 6.18E-02
118
Table 3-6. Trace analysis and partition coefficients for Zn, Sn and Pb (ppm)
Line Zn 66 Sn 118 Pb 208
yn88 MSS 4.49E+01 2.31E+00 3.57E+00
std 1.53E+01 8.50E-01 2.98E-01
yn88 ISS 1.54E+02 4.04E+01 1.78E+02
std 5.31E+00 3.02E+00 1.42E+01
D(MSS/ISS) 2.91E-01 5.70E-02 2.01E-02
error in D 9.94E-02 2.10E-02 2.30E-03
yn89 MSS 2.15E+01 1.55E+00 3.88E-01
std 7.06E+00 5.80E-01 9.70E-02
yn89 melt 3.43E+01 1.72E+02 3.13E+02
std 3.59E+00 8.53E+00 1.93E+01
D(MSS/melt) 6.27E-01 9.01E-03 1.24E-03
error in D 2.16E-01 3.40E-03 3.19E-04
Detection Limit 3.27E+00 1.55E+00 2.97E-01
119
Chapter IV. Compositional evolution of magmatic sulfide melt:
models from partitioning experiments applied to the McCreedy
East Deposit, Sudbury, Ontario
4.1. Compositional evolutionary models for McCreedy East Ore Body,
Sudbury (Canada)
An important application of the partition coefficients for the PGE and chalcogens, whose
measurement is described in Chapter 3, is to understand the compositional evolution of
magmatic sulfide melt. To this end, quantitative models can be developed, then tested against
the trends exhibited by data from natural magmatic sulfide occurrences. In this context,
Mungall (2007) modeled the compositional evolution of orebodies from the 1.85 Ga Sudbury
Igneous Complex (SIC) using parameterizations for MSS-melt partition coefficients for Ni
and Cu available at the time (Fig 4-1). Under the framework of MSS-melt equilibrium
crystallization, Mungall (2007) was able to reproduce the Ni and Cu concentration levels for
MSS cumulates. The compositional array for the most Ni rich samples (> 10 wt% Ni), was
120
postulated to be due to mixing between pentlandite and ISS, while the intermediate, Cu rich
samples (5~22 wt% Cu), were suggested to lie on a mixing line between MSS and residual
sulfide liquid or between MSS and ISS cumulates. This model requires that if the Cu rich
samples were due to the mixing between MSS and liquid, there must have been continuous
equilibration of solids and liquids during cooling to low temperatures close to the sulfide
solidus; or if the Cu rich samples were due to the mixing between MSS and ISS cumulates, a
significant mass of highly Ni- and Cu-enriched residual sulfide liquid must have left the MSS
– ISS cumulates behind during a late-stage migration process at temperatures well below the
solidus temperature of the enclosing silicate rocks. An important additional test of this sort of
model is the nature of PGE and chalcogen fractionation once ISS begins to crystallize, which
could not be done previously due to a lack of past experimental constraints. In this chapter,
the newly-acquired ISS/melt partition coefficients were combined with the MSS/melt
partition coefficients measured under controlled fO2 and fS2 conditions, to develop a
relatively simple sulfide melt evolutionary model that takes into account the liquid path once
ISS has begun to crystallize.
The model proposed in this chapter is tested against the natural data from one of the Sudbury
ore bodies, McCreedy East, which is located along the northern margin of the SIC. The body
consists of a pyrrhotite-rich contact ore, and a chalcopyrite-rich footwall ore. Like other
orebodies associated with the SIC, the compositional zoning is interpreted to arise by
separation of MSS cumulates to form a more evolved, copper-rich liquid at the magmatic
stage (Li et al., 1992; Naldrett et al., 1999). The chalcopyrite-rich footwall ore is interpreted
by Dare et al (2010; 2014) to have undergone further differentiation to produce ISS
121
cumulates, and a late-stage, highly fractionated sulfide liquid. Along with the PGE, and
other precious metals, Dare et al (2011; 2014) also report concentrations of the chalcogens in
both ore types, which offers the opportunity to explore the magmatic differentiation model
using the partitioning data measured in this thesis. In detail, Dare et al (2014) propose a
multi-stage model in which MSS cumulates form initially to produce the pyrrhotite-rich
contact ore, then the more Cu-rich liquid is drained into footwall fractures to undergo
internal differentiation by predominantly ISS crystallization, followed by growth of a variety
of PGMs initiated by declining temperature and enriched abundances in the residual liquid.
Modeling the intricacies of such a complex crystallization history is beyond the scope of this
project, but instead attempts were made to apply the partitioning data to determine if initial
MSS removal, followed by closed system crystallizationof ISS, captures the broad trends in
element variation.
The initial sulfide liquid composition for McCreedy East was estimated according to the
approach of Mungall et al. (2004), where the bulk composition of the SIC is believed to be
dominated by melt from the lower crust, and sulfide liquid PGE concentrations are consistent
with equilibration with a moderately large silicate reservoir (silicate/sulfide mass ratio, R-
factor, of 700; Mungall et al., 2005). In this exercise, a suite of elements were modeled
including the chalcogens (Se, As and Te), as well as compatible (Ir) and incompatible (Pt, Pd)
PGE for which abundance and partitioning data are available. As in past work, the variation
in Cu abundance is the chosen metric for extent of differentiation, given the incompatible
behavior of this element in MSS. The concentrations of Cu, Ir and Pd in the SIC are taken
from Mungall et al (2004), whereas estimates for the Pt, Se and As contents of the SIC are
122
taken from the average lower crustal abundances of Rudnick and Gao (2003) and Gao et al
(1998). For Te a lower crustal abundance of 15 ppb was adopted, which is 3x the “order of
magnitude” estimate from Wedepohl (1995), and found consistent with the Sudbury ore
compositions. The composition of the coexisting sulfide liquid is then calculated using the
sulfide liquid/silicate liquid partition coefficients from Mungall and Brenan (2014) for Cu, Ir,
Pt and Pd, Li and Audetat (2012) for As, and Brenan (2015) for Se and Te. A summary of
these parameters, as well as the calculated initial sulfide liquid compositions, is provided in
Table 4-1.
Differentiation models were constructed in three major stages: Stage I, MSS-only
crystallization; Stage II, MSS-ISS co-crystallization; Stage III, ISS-only crystallization. In
the context of the model proposed by Dare et al. (2014), we first assumed fractional
crystallization of MSS in Stage I and Stage II, then equilibrium crystallization in Stage III.
For the case of fractional crystallization, the concentration of the trace element in the liquid,
CL is calculated from:
1
0
D
L FCC (Eq 4-1)
In which Co is the initial concentration, F is the fraction of liquid remaining and D is the bulk
solid/liquid partition coefficient. As temperature falls, ISS will start to appear together with
MSS and Stage II starts from this point on. The Cu content of the sulfide liquid
corresponding to the onset of ISS crystallization is estimated from the phase equilibrium
experiments of Ebel and Naldrett (1996). The estimated Cu/(Cu + Ni + Fe) of the parental
sulfide liquid for the SIC is ~0.06, which is similar to the value of 0.078 for bulk
123
composition c2b2 from Ebel and Naldrett (1996), which produced ISS when the liquid phase
contained 21.68 wt% Cu. Thus a Cu concentration of 22 wt% was adopted as the boundary
point between Stage I and Stage II. Although this is not well constrained in the natural
system, the exact value of this boundary does not have a significant impact on our model as a
first order approximation. The bulk partition coefficient during Stage II crystallization is
defined as:
MSSISS XmeltMSSDXmeltISSDmeltsolidD )/()/()/( (Eq 4-2)
In which XISS and XMSS are the mass fractions of ISS and MSS in the crystallizing
assemblage. Values of XISS and XMSS are calculated from their relative proportions by weight
(MSS to ISS Ratio; MIR) from the experimental data of Fleet and Pan (1994) and this study.
Values of MIR have been parameterized as a function of F based on experiment mass
balance, with the data and regression function shown in Figure 4-2. During Stage II, the
effect of increasing the amount of crystallizing ISS (MIR decreasing, as shown by
experiment) is to lower the bulk partition coefficients for Ru, Os, Re, Ir and Rh, but raise
values for Pd, Pt, Au and Ag, as well as Te, but not Se and As, in which ISS and MSS-melt
partitioning are similar. The onset of Stage III is taken to occur when Cu in the liquid
reaches 32 wt%, as proposed by Naldrett et al. (1999). Stage III involves internal
equilibrium crystallization between ISS and liquid (Dare et al. 2014); phase concentrations
are calculated according to mass balance for the fixed bulk composition at the onset of this
stage.
124
A comparison between these modeled compositions and those reported by Dare et al. (2014),
normalized to 100% sulfide, is shown in Figure 4-3a-f. Cu poor samples (Cu < 22wt%,
average of MCR2, 3, 4, 6 and MCR5 in Dare et al. 2014) are in reasonable agreement with
predictions for the MSS cumulates for all the elements considered. For the case of Ir, which
is the most compatible element modeled, calculated liquid curves are orders of magnitude
lower than concentrations in the natural samples (Figure 4-3a). Whereas the detection limit
for Ir reported by Dare et al. (2014) is ~0.05 ppb, and the highest Ir concentration in these Cu
rich samples is less than 1 ppb, the mismatch between our model and their measurement for
Ir may be due to the increasing analytical uncertainty for values close to the detection limit.
Despite this shortcoming, the model predicts residual liquids which are essentially devoid of
Ir, consistent with observations. MSS- and ISS-melt partition coefficients for Se are similar
and near unity, so the model predicts only a relatively small variation in the Se content of
liquids and solids throughout the crystallization interval, which is consistent with the limited
range in Se concentrations in both Cu-poor and Cu-rich samples (Figure 4-3b). For the case
of the other elements modeled, the Cu-rich samples from McCreedy East appear to cluster
into two groups (Group I: MCR9 4700L, MCR10 4810L, MCR11-13 4810L, MCR11-23
4810L; Group II: MCR7A 4550L, MCR12 4810L, MCR13, MCR14), with the spread in
between roughly consistent with the offset between calculated liquids (Group I) and ISS
(Group II) over the Stage II crystallization interval. The magnitude of the compositional
shift between initial MSS cumulates, and Stage II liquid and solids is also predicted
reasonably well for elements of different compatibility. Some of the more Cu-rich
compositions are also consistent with a limited evolution along a Stage III liquid trajectory.
125
Hence to a first order, the results are consistent with the magmatic sulfide evolution model
proposed by Dare et al (2014).
4.2. Timing and sequence of certain magmatic PGMs
4.2.1 Solubility of Pt-Pd-Te-As-bearing PGMs in sulfide liquid
As described by Dare et al (2014), the PGM assemblage at McCreedy East is dominated by
phases in the system Pt-Pd-Bi-Te-(As), including such minerals as michenerite ([Pt,Pd]
BiTe), froodite (PdBi2) and minor sperrylite (PtAs2). Texturally, the PGM assemblage
suggests late-stage crystallization from a compositionally-evolved sulfide liquid. Taking the
Pt-As-S system as an example, the conditions for PGM saturation in a sulfide melt can be
expressed according to a modification of reaction (3-2):
2
0
42
20
2
2
2
ln4
lnlnln2
ln
ln
24
22
2
2
SAsPtAsSPt
x
S
x
AsPt
x
SPtAs
x
x
fx
xx
RT
G
f
RTG
Sx
PtAsSx
xAsPt
xx
X
x
(Eq 4-3)
in which, ΔG0
is the net Gibbs free energy of the reaction, T is the temperature in Kelvin, α
represents the activity of each component, and 𝑓𝑆2 represents the sulfur fugacity. The
conditions for saturation in sperrylite, PtAs2, will thus depend on temperature, concentration
of As, sulfur fugacity and the proportion of Pt2+
(hence the valence states of the PGEs). To
date, only the studies of Helmy and co-workers (Helmy et al., 2007; 2013b) have
investigated PGM solubility in this level of detail (although fS2 has not been controlled). For
126
the case of sperrylite, Helmy et al. (2013b) measured the concentrations of Pt and As in
sulfide liquid required for saturation at 0.1 MPa and 1150-770oC. Measured Pt and As
concentrations in the sulfide liquid range from 12200 ppm Pt and 9400 ppm As at 1150oC to
3100 ppm Pt and 6200 ppm As at 770oC, indicating solubility decreases with temperature.
Such elevated concentrations for these elements are not obtained in liquids calculated for the
Stage I and II crystallization intervals, indicating early saturation in sperrylite is unlikely.
During Stage III crystallization, the Pt content of the calculated liquid only reaches ~180
ppm even when the fraction of remaining liquid is quite small (0.0001); this is still
significantly lower than the saturation value of 3100 ppm at 770oC. The modeled As in the
liquid is ~220 ppm at the same F, which is also far below the experimentally-determined
value. However, if it is assumed that the solubility of sperrylite decreases linearly with
temperature, or 1/T, as in Eq. 4-3, then by 650 oC, the Pt and As concentrations required for
saturation have dropped to ~180 ppm and ~493 ppm respectively (Fig 4-4) , which are more
similar to the calculated concentrations in the highly evolved liquid. Decreasing temperature
will also drive crystallization in the system, resulting in a concomitant decrease in F. Hence,
results suggest saturation of sperrylite at McCreedy East is plausible for a highly fractionated
sulfide liquid with an initial composition similar to the values in Table 4-1.
In terms of the timing of formation of Pt-Pd tellurides such as michenerite, data for specific
solubility experiments on this phase are lacking, so here we use values measured by Helmy
et al (2007) for moncheite (PtTe2) and kotulskite ([Pd,Ni]Te) as a rough guide. Whereas data
for the sulfide liquid in the experiments of Helmy et al (2007) are not available, the
composition of a hypothetical liquid can be calculated from that of MSS coexisting with the
127
moncheite + kotulskite assemblage. Using our measured values of D(MSS/melt), the Pd and
Te content in the sulfide melt is expected to be ~34500 and ~62000 ppm, respectively, and
varies little with temperature over the range from 1015 to 500oC. In contrast, the modeled
concentrations of Pd and Te in Stage III sulfide liquid when F~0.0001 is ~73 and ~305 ppm,
respectively, which is insufficient for Pd-telluride to saturate. Hence, at an F when saturation
in sperrylite seems possible, the melt would be significantly undersaturated in Pd telluride,
suggesting that if these phases are magmatic, they would form from an even more evolved
melt.
4.2.2 Estimates for the timing of crystallization of Bi-, Sn-, Zn- and Pb-bearing phases
In addition to the above-described phases, minerals containing Bi, Sn, Zn and Pb also
comprise the accessory phase suite in some ore samples from McCreedy East. Although the
constraints on their saturation behaviour are sparse, here the relative timing of these phases
was speculated in the context of the current sulfide melt evolution model.
Bi-bearing PGMs at McCreedy East, include Froodite (PdBi2), Insizwaite (PtBi2) and some
Bi dissolved in Michenerite (PdBiTe) as well. But information on their solubility in sulfide
melt are lacking. However, some rough constraints can be placed on the timing of saturation
to compare with model results. To estimate the Bi solubility, we use the average whole rock
Bi content of 81.9 ppm, 65 % of which can be accounted for by PGM. The mass fraction of
PGMs is 0.00010267 taken for all the samples studied by Dare et al. (2014), which would
translate to 519000 ppm Bi if all Bi-bearing PGM were coming from the trapped residual
128
liquid. If Bi minerals are the first to saturate from the trapped sulfide liquid, this could act as
a very rough estimate for the lower limit of Bi solubility. Assuming the Bi content of the
initial sulfide melt to be around 10 ppm, which results in melt and solid trajectories in
reasonable agreement with observations, at F ~ 0.0001 the modeled concentration of Bi in a
Stage III sulfide liquid would be around 3231 ppm. Compared with the estimated solubility,
it does not seem likely for Bi to saturate before Pt or As. This conclusion is consistent with
textural observations reported by Dare et al (2014; here reproduced in Figure 4-5) in which
sperrylite cores are encapsulated by PdBiTe overgrowths.
A similar exercise can also be applied to the timing of saturation of Sn-rich phases. In terms
of the liquid evolution curve, the composition of the initial sulfide can be determined from
estimates for Sn in the lower continental crust, which is around 2.1 ppm (Wedepohl 1995).
The partition coefficient for Sn between sulfide and silicate is between 4.43 (Li and Audetat
2012) and 10.7 (Pattern et al. 2013). Assuming an R-factor of 700, the concentration of Sn
in the initial sulfide melt before crystallization would be around 2.06 ppm. However, this
initial value results in liquid compositions that consistently underestimate the natural samples.
To better match the natural data, an initial Sn concentration approximately 10 times the value
mentioned above is required (~20.6 ppm), resulting in the model evolution shown in Figure
4-6. We thus suspect that the initial concentration of Sn for McCreedy East was
compromised by other sources rather than being controlled by the lower continental crust
alone. This change must have happened prior to sulfide melt crystallization started, since the
concentration span between MSS cumulates and the Cu rich samples (Group I liquid
dominant samples) is still consistent with liquid evolution from a single initial bulk
129
composition. Exploring the potential source for Sn addition would be beyond the scope of
the project, however, if Sn is also controlled by magmatic processes for these samples, as
well as the case of PGEs and other chalcogens as discussed above, a successful reproduction
of the concentrations in both the MSS cumulates and the Cu-rich samples (including both
liquid and ISS dominant samples) can be regarded as a constraint for a properly estimated
initial concentration. With this revised initial Sn concentration, we can thus estimate the
timing of formation of Sn-rich PGMs within the same framework as for sperrylite. The
experimental data for Sn solubility in sulfide melt is scarce. After extensive searching, we
found a single experiment (Run# 62s) from Righter and Drake (2000) in which the Sn
concentration in sulfide melt is 0.75 wt%, although in this case the saturating phase is
metallic Sn. Accepting this value as a first order estimate for Sn solubility in sulfide melt,
compared with the trajectory in Figure 4-6, Sn does not reach saturation at the same F at
which sperrylite crystallization is likely; the point at which the As content of the melt is ~1/2
that required for sperrylite saturation, Sn has reached only 1/3 of the required saturation level.
Therefore, with the current estimations, we predict that Pt-arsenide would form prior to PtSn.
This conclusion is also consistent with the textural relationships provided in Figure 4-5a and
4-5d.
Applying a similar methodology to predict spalerite or galena saturation, both of which are
observed at McCreedy East, is complex, since ZnS – PbS – FeS -(Ag2S) form solid solutions
(Mavrogenes et al. 2001). Stevens et al. (2000) did solubility experiments involving sulfide
melt, galena, chalcopyrite, sphalerite and pyrrhotite at temperatures between 750 °C to
1000 °C. The Zn content in the sulfide melt is positively correlated with temperature. At
130
750 °C, the Zn saturation value was measured to be 0.6 wt% in sulfide melt coexisting with
sphalerite. If we consider this possible temperature effect alone on Zn solubility, by 650 °C
when Pt saturation from the melt is possible, the Zn saturation value would become as low as
3192 ppm. Similar to the Sn scenario, an estimate of the initial Zn content of the sulfide melt
based on values for the lower continental crust (~102 ppm) consistently underestimates the
Zn contents in all the samples reported in Dare et al. (2014). A Zn model fitting all these data
requires a starting concentration of around 500 ppm. The model results are illustrated in
Figure 4-7. Due to the compatible nature of Zn in ISS, the concentration of Zn in the residual
melt is predicted to decrease as more ISS crystalized. Thus saturation of sphalerite directly
from the sulfide melt before other accessory phases does not seem likely. We expect that
sphalerite crystallization would only be possible at a very late stage, occurring interstitially,
or possibly forming at subsolidus conditions. An important result is that ISS cumulates
should be relatively more Zn rich compared with the residual liquid. It follows that a higher
percentage of sphalerite should then be expected for the Group II samples compared with
Group I, which is consistent with the fraction of sphalerite estimated by Dare et al. (2014)
between the two groups (0.017 vs 0.013). In contrast to Zn, the Pb content of the sulfide melt
at galena saturation is enormous (25 wt%; Stevens et al., 2000), indicating that Pb-enriched
compounds are unlikely to form at the magmatic stage.
4.2.3 Crystallization sequence and factors affecting the accuracy of the model
In Stage III, an estimate of the starting concentration combined with the solubility will
enable us to calculate how much fractionation is required for saturation to occur:
131
)/(1
)/(0
meltISSD
meltISSDC
C
Fliq
(Eq. 4-4)
where F represents the fraction of liquid remaining compared with the starting bulk of Stage
III, C0 is the starting concentration of the interested element, Cliq is the concentration of the
element in liquid, D(ISS/melt) is the experimentally determined partition coefficient between
ISS and melt. Based on the above calculations, for a sulfide melt similar to the assumed
starting composition of McCreedy East, a possible crystallization sequence due to saturation
in the sulfide melt might follow Pt > As > Sn > Bi >Ag > Pd > Te, with Zn and Pb only
possible in the very late magmatic stage or even post-magmatic stage. This sequence is in
general accord with the “Center to Edge” histogram in Dare et al. (2014, Fig 8B), and also
consistent with the thermal stability of these PGMs, with Pt-As or Pt-Sn stable at high
temperature (1400 ~ 850 °C), Bi-Te compounds stable only at lower temperature (600
~500 °C).
The occurrence and relative crystallization sequence of PGE and chalcogen-bearing
accessory minerals will certainly hinge on estimates for the initial sulfide liquid composition,
and the mineral solubility, the latter being poorly known at this time. At other deposits, for
example, if As levels in the initial sulfide melt are higher, the timing of arsenide or
sulfarsenide minerals such as sperrylite, irarsite (IrAsS), hollingworthite (RhAsS), and PGE-
rich cobaltite (CoAsS) etc., might occur significantly earlier, as suggested by Dare et al.
(2010) based on textural evidence of the inclusions in base metal sulfide cumulate phases at
the Creighton Mine (Sudbury, Ontario, Figure 4-8). Another complicating factor is the
132
solubility of PGEs and chalcogens in other PGM phases, such as Merenskyite
((Pd,Pt)(Te,Bi)2), volynskite (AgBiTe2), etc, which might also affect the crystallization
sequence. It is also important to note that the liquid and solid evolution curves are calculated
with the assumption that partition coefficients are not sensitive to other parameters, such as
temperature, melt compositions, sulfur and oxygen fugacities, etc. As mentioned in Chapter 3,
previous studies have demonstrated that MSS/melt partition coefficients could be affected by
the metal/sulfur ratio of MSS (Li et al. 1996, Helmy et al. 2007, Ballhaus et al. 2001). The
MSS-ISS mixing line in Mungall (2007) could also be interpreted as changing D (MSS/melt)
for Ni as crystallization continues.
4.3 Conclusions
The primary conclusions of this chapter are:
1) A magmatic sulfide crystallization model was developed which assumes three major
stages: 1) MSS-only crystallization; 2) MSS-ISS-co-crystallization; 3) ISS-only
crystallization, with each as fractional or equilibrium. A MATLAB GUI was created
to help visualize such a model for element pairs (as in the appendix).
2) Using the Cu content in the residual melt as an index for sulfide melt crystallization,
the evolutionary trajectories of all the investigated elements can be modelled. The
modeled curves are consistent with the field data from McCreedy East (Dare et al.
2014). For some elements, model curves can also be used to estimate a deposit-
dependent starting composition, or evaluate any potential post-magmatic imprints.
133
3) Combined with solubility measurements, these modelled trajectories can also be used
to explain or predict the magmatic crystallization sequence for common PGMs (PtAs2,
PdTe, PtSn, etc.). Our model-based predictions match the textural observations
provided in Dare et al. (2014).
4) The model can not yet be extended to elements such as Zn and Pb, to explain the
occurrence and enrichment of sphalerite or galena often associated with these Ni-Cu-
PGE deposits. Other factors that may alter the modeling results include the estimates
for starting compositions (McCreedy vs Ceighton), variations of partition coefficients
during the crystallization, solubility approximations in sulfide melt, etc.
134
Figure 4-1. Equilibrium crystallization model for MSS/melt fractionation, figure from
Mungall (2007) for easier reading. White squares represent the modeled MSS
cumulates, while the low Ni, intermediate Cu samples were interpreted as the
mixing line between MSS and ISS.
135
Figure 4-2. MIR vs fraction of liquid in the system (F). Data compiled from Fleet and Pan
(1994) and this study, with labels corresponding to the identity of specific
experiments. Dashed line represents the best fit curves throughout the
investigated range, with the regression equation and R-value as labeled on the
plot.
136
a)
b)
c)
d)
e)
f)
Figure 4-3. Modeled evolution curves for sulfide liquid, MSS and ISS based on the experimentally determined partition
coefficients from this study, tested against the field data from Dare et al. (2014) for the McCreedy East deposit,
Sudbury. The model assumes fractional crystallization in Stage I and II, and equilibrium crystallization in Stage III
(see text for definition of stages). The initial sulfide composition is calculated based on D(sulfide/silicate) and
estimates for lower continental crust, as listed in Table 4-1. Copper in weight percent, Ir, Pd and Pt in ppb, while
other concentrations in ppm. Data are reported as whole rock analysis, normalized to 100% sulfide. Green curves:
MSS cumulates; red curves: ISS cumulates; blue curves: liquid in all stages. Tick marks represent the percentages
of liquid remaining in the system.
137
Figure 4-4 Extrapolated solubility for Pt and As in sulfide melt, based on the
experimental data from Helmy et al. (2013b). Blue and red dashed lines
represent regression either without, or including, the superliquidus data,
respectively. At 650 °C, the solubility of Pt and As in sulfide melt is estimated
by the average value calculated from the two regression methods.
138
Figure 4-5 Examples of textural relationships among PGMs at McCreedy East taken
from Dare et al. (2014), which is consistent with the model prediction that Pt-
arsenide would crystallize first from late-stage melts, thus forming the crystal
core, surrounded by Pd-rich phases; for the anions, As-Sn minerals would
crystallize before Te minerals.
139
Figure 4-6 Comparison between modeled and measured Sn concentrations at McCreedy
East (data from Dare et al., 2014). The model assumes an initial silicate melt
composition with a Sn content ~10x higher than estimates for the lower
continental crust, as required to produce the ore compositions. The red dashed
line represents the estimated saturation level for Sn in sulfide melt (Righter
and Drake, 2000) to be around 7500 ppm.
140
Figure 4-7 Comparison between modeled and measured Zn concentrations at McCreedie
East (data from Dare et al., 2014). Dashed line represents the calculated
possible sphalerite saturation of 3192 ppm (after correcting for the
temperature effect on solubility from Stevens et al., 2000), which is not
achieved by the model liquid composition.
141
Figure 4-8 Textural evidence from Dare et al. (2009) for samples from the Creighton
Mine Sudbury which have been used to suggest early saturation in arsenide
phases at the magmatic stage.
142
Table 4-1. Summary of parameters to calculate the initial sulfide melt composition at McCreedy East.
Concentration in
SIC Unit
Source of data Sulfide/silicate
partition coefficient Source of data
Concentration
in initial
sulfide
Cu 0.0075 ± 0.00016 wt%
Mungall et al. (2007) 1473 Mungall and Brenan (2014) 3.56 ± 0.08
Ir 0.11 ± 0.03 ppb
Mungall et al. (2004) 458000 Mungall and Brenan (2014) 77.0 ± 21.0
Co 36 ± 0.1 ppm
Gao et al. (1998) 80 Rajmani and Naldrett (1978) 2588 ± 7.5
Pd 2.78 ± 0.044 ppb
Gao et al. (1998) 209000 Mungall and Brenan (2014) 1942 ± 31.0
Pt 2.87 ± 0.364 ppb
Gao et al. (1998) 845000 Mungall and Brenan (2014) 2010 ± 255
Se 0.166 ± 0.0022 ppm
Gao et al. (1998) 1443 Brenan (2015) 78.4 ± 1.08
Bi 0.2 ± 0.01 ppm
Gao et al. (1998) 213 Li and Audetat (2012) 33 ± 1.93
As 1.6 ± 0.09 ppm
Gao et al. (1998) 1.9 Li and Audetat (2012) 3.04 ± 0.175
Te 0.0151 ± 0.0121 ppm Wedepohl (1995)
Salters and Stracke (2004) 13447 Brenan (2015) 10.1 ± 8.10
143
Chapter V. Summary and Conclusions
How the PGE and chalcogens get distributed, transferred and enriched prior to their final
deposition as ore grade deposits in nature, is a complicated topic. While qualitative
information is available through textural studies of the natural samples, a comprehensive,
quantitative framework has not yet been fully established, mapping their evolutionary history
during the crystallization journey. With this project, we attempted to step further toward this
goal based on the foundation provided by previous work.
Accurate measurements of PGE and chalcogens using LA-ICPMS greatly rely on the quality
of the selected standard. In Chapter II, the synthesis and characterization of a newly
improved LA-ICPMS standard (Ge6) was reported as a potential candidate for both sulfide
and silicate glass samples. The synthesis can be accomplished with ordinary lab equipment
and within a short period of time (~28 hours). The Ge-Sb-S matrix has an excellent glass
forming capability and is relatively insensitive to the exact proportions among major
144
elements. Most the PGEs and chalcogens can be doped homogeneously up to 100 ppm,
except for Ru, Os, Re, Mo and W which can only be doped up to 10 and 5 ppm respectively,
before sample heterogeneity is encountered. The synthesized material was calibrated using
both LA-ICPMS and solution ICPMS, and tested using a variety of sulfide and silicate
reference materials (JBSulfide, NiS4, JB-2, JGb-1, JG-1a, BIR-1, BHVO-1). The test results
in general agree with the documented values, indicating an insignificant matrix effect for
these measurements. Further possible development of the standard would be to include
lithophile elements, such as the REE, which would expand the range of information gained
from an individual analysis.
The partition coefficient is a key parameter in tracking the element of interest during any
fractionation processes. In Chapter III, we reported the results of MSS- and ISS-sulfide melt
partitioning experiments, which provided an internally-consistent set of partition coefficients
for PGE, chalcogens and other select chalcophile elements (Pb, Zn, Sn). Results showed that
MSS crystallization will significantly fractionate this element suite, as the Ru, Os, Ir, Rh and
Re are found to be compatible in MSS relative to sulfide melt with D values ranging from
~20 to ~5, whereas Pd, Pt, Au, Ag, Pb, Zn, Sn as well as the chalcogens, are incompatible in
MSS, with D values ranging from ~0.1 to ~1 x 10-3
. Partition coefficients for Ru, Os, Ir, Rh
and Re are systematically larger than most past studies (excepting the buffered experiments
of Mungall et al., 2005), correlating with a higher oxygen content in the sulfide liquid and
reflecting the significant effect of oxygen on increasing the activity coefficients for these
elements in the melt phase. The relative partitioning between MSS/ISS measured in this
study indicates that the onset of ISS crystallization will decrease the overall compatibility of
145
the Ru, Os, Ir, Rh and Re in the crystallizing assemblage, whereas Pd, Pt, Au and Ag become
less incompatible, so the relative fractionation imparted by MSS crystallization is subdued.
Combined with available solubility data in sulfide melt, the timing and sequence of
saturation of PGE and chalcogens are discussed in Chapter IV. Modeled evolution curves for
the PGE and chalcogens are in reasonably good agreement with whole-rock sulfide
compositions for the McCreedy East deposit, consistent with an origin by crystallization of
MSS, then MSS + ISS from sulfide melt. The uniformly low MSS-melt and ISS-melt
partition coefficients for the chalcogens, Pt, Pd, Ag and Au will lead to continuous
enrichment in the residual sulfide liquid, but only as the amount of residual melt becomes
small; D values are generally too large for Pt, Pd, As and Te to reach early saturation in
accessory minerals rich in these elements. Among the studied elements, a relative saturation
sequence was obtained based on comparison between modeled concentration in sulfide melt
and the corresponding solubility data from the literature. Pt is most likely to reach early
saturation, followed by phases that contain As, Sn, Pd, Te, Bi, Zn. These predictions are in
qualitative agreement with the textural evidence provided in Dare et al. (2014). These models
may also have the potential of constraining the starting compositions, assessing the timing of
possible post-magmatic mechanisms, distinguishing ISS cumulates and residual liquids, etc.
The direct products of this thesis include: 1) a newly improved LA-ICPMS standard for
silicate and sulfide analyses for PGEs, chalcogens and transitional metals; 2) a new set of
experimentally determined, internally-consistent partition coefficients between MSS-sulfide
melt, MSS-ISS and ISS-sulfide melt under controlled fO2 and fS2 conditions; 3) an
146
evolutionary model for PGE and chalcogens with a graphic user interface (GUI) for future
modeling purpose, which proved to be capable of reproducing field data from the McCreedy
East at Sudbury.
147
References
Alard, O., Griffin, W. L., Lorand, J. P., Jackson, S. E., and O'Reilly, S. Y. (2000). Non-
chondritic distribution of the highly siderophile elements in mantle sulphides. Nature,
407(6806), 891-894.
Andrews, D.A. and Brenan, J.M. (2002) The solubility of ruthenium in sulfide liquid:
implications for platinum group mineral stability and sulfide melt–silicate melt
partitioning. Chem. Geol. 192, 163-181.
Baker, D.R. and Moretti, R. (2011) Modeling the solubility of sulfur in magmas: a 50-year
old geochemical challenge. Reviews in Mineralogy and Geochemistry, 73,
167-213.
Ballhaus, C. and Ulmer, P. (1995) Platinum-group elements in the Merensky Reef. 2.
Experimental solubilities of Platinum and Palladium in Fe1-x S from 950 ° to 450 °C
under controlled fS2 and fH2. Geochim. Cosmochim. Acta 59, 4881-4888.
Ballhaus, C. and Sylvester, P. (2000): Noble metal enrichment processes in the Merensky
Reef, Bushveld Complex. J. Petro. 41, 545-561.
Ballhaus, C. Tredoux, M., Spath, A. (2001) Phase relations in the Fe-Ni-Cu-PGE-S system
at magmatic temperature and application to massive sulfide ores of the Sudbury
Igneous Complex. J. Petro. 42, 1911-1926.
Barin, I (1995) Thermochemical data of pure substances. Weinheim, New York, 1885 pp.
148
Barnes, S.-J., Cox, R.A. and Zienteck, M.L. (2006) Platinum-group element, Gold, Silver
and Base Metal distribution in compositionally zoned sulfide droplets from the
Medvezky Creek Mine, Noril’sk, Russia. Contrib. Mineral. Petro. 152: 187-200.
Bennett, N., Brenan, J.M. and Koga, K.T. (2014) The solubility of platinum in silicate melt
under reducing conditions: Results from experiments without metal inclusions.
Geochim. Cosmochim. Acta, 133, 422-442.
Bezmen, N.I., Asif, M., Brugmann, G.E., Romanenko, I.M., and Naldrett, A.J. (1994)
Distribution of Pd, Rh, Ru, Ir, Os, and Au between sulfide and silicate metals.
Geochim. Cosmochim Acta, 58, 1251–1260.
Bockrath, C., Ballhaus, C., & Holzheid, A. (2004). Stabilities of laurite RuS 2 and
monosulfide liquid solution at magmatic temperature. Chem. Geol., 208(1), 265-271.
Brenan, J.M., Cherniak, D.J. and Rose, L.A. (2000) Diffusion of Osmium in Pyrrhotite and
Pyrite: Implications for Closure of the Re-Os Isotopic System. Earth Planet. Sci.
Lett., 180, 399-413.
Brenan, J.M.(2002) Re-Os Fractionation in Magmatic Sulfide Melt by Monosulfide Solid
Solution, Earth Planet. Sci. Lett., 199: 257-268.
Brenan, J.M. (2008) Re-Os fractionation by sulfide-silicate partitioning: A new spin.
Chemical Geology, Special Issue on Highly Siderophile Elements 248, 140-165.
Brenan, J. M. (2015). Se–Te fractionation by sulfide–silicate melt partitioning: Implications
for the composition of mantle-derived magmas and their melting residues. Earth
Planet. Sci. Lett., 422, 45-57.
Cabri, L.J., Sylvester, P., Tubrett, M.N., Peregoedova, A. and Laflamme J.H.G. (2003)
Comparison of LAM-ICP-MS and Micro-PIXE results for Palladium and Rhodium in
selected sampels of Noril’skl and Talnakh sulfides. Can. Mineral. 41, 321-329.
Carmichael ISE, Ghiorso MS (1986) Oxidation-reduction relations in basic magma: a case
for homogeneous equilibria. Earth Planet. Sci. Lett. 78, 200-210.
149
Carvajal, M. A., Alvarez, S. and Novoa, J. J. (2004) The Nature of Intermolecular CuI-Cu
I
Interactions: A Combined Theoretical and Structural Database Analysis. Chem. Eur.
J., 10, 2117–2132. doi: 10.1002/chem.200305249
Crocket, J. H., and M. E. Fleet (1997) Implications of composition for experimental
partitioning of platinum-group elements and gold between sulfide liquid and basalt
melt: the significance of nickel content. Geochim. Cosmochim. Acta 61, 4139-4149.
Dare, S.A.S., Barnes, S-J. and Prichard H.M. (2010a) The distribution of platinum group
elements (PGE) and other chalcophile elements among sulfides from the Creighton
Ni-Cu-PGE sulfide deposit, Sudbury, Canada, and the origin of palladium in
pentlandite. Mineral Deposita DOI 10.1007/s00126-010-0295-6.
Dare, Sarah A. S., Barnes, S-J., Prichard, H. M. (2010b) The Timing and Formation of
Platinum-Group Minerals from the Creighton Ni-Cu-Platinum-Group Element
Sulfide Deposit, Sudbury, Canada: Early Crystallization of PGE-Rich Sulfarsenides.
Econ. Geol. 105, 1071-1096.
Dare, S.A.S., Barnes, S-J., Prichard H.M. and Fisher, P. (2011) Chalcophile and platinum -
group element (PGE) concentrations in the sulfide minerals from the McCreedy East
deposit, Sudbury, Canada, and the origin of PGE in pyrite. Mineral Deposita 46, 381-
407.
Dare, S.A.S., Barnes, S-J., Prichard H.M. and Fisher, P. (2014) Mineralogy and
Geochemistry of Cu-Rich Ores from the McCreedy East Ni-Cu-PGE Deposit
(Sudbury, Canada): Implications for the Behavior of Platinum Group and Chalcophile
Elements at the End of Crystallization of a Sulfide Liquid. Econ. Geol. 109, 343-366.
Ding, L., Yang, G., Xia, F., Lenehan, C.E., Qiang, G., McFadden, A., Brugger, J., Zhang, X.,
Chen, G. and Pring, A. (2011) A LA-ICP-MS sulphide calibration standard based on
a chalcogenide glass. Mineral. Maga. 75, 279-287.
Diliberto, S., Kessler, O., Rapin, C., Steinmetz, P., & Berthod, P. (2002). Development of
chromia forming Mo-W-Cr alloys: synthesis and characterization. J. of materials
science, 37(15), 3277-3284.
Dutrizac, J.E. (1976) Reactions in cubanite and chalcopyrite. Can. Mineral., 14:172–181.
150
Ebel, D.S., and Naldrett, A.J. (1996) Fractional crystallization of sulfide ore liquids at high
temperature. Econ. Geol. 91, 607–621.
Farrow, C.E.G. and Watkinson, D.H. (1997). Diversity of precious metal mineralization
in footwall Cu-Ni-PGE deposits, Sudbury, Ontario: implications for hydrothermal
models of formation; Can. Mineral 35, 817-839.
Flanagan, F.J. (1984) Three USGS mafic rock reference samples, W-2, DNC-1, and BIR-1:
U.S. Geological Survey Bulletin 1623, 54.
Fleet, M. E., Stone, W. E., & Crocket, J. H. (1991). Partitioning of palladium, iridium, and
platinum between sulfide liquid and basalt melt: Effects of melt composition,
concentration, and oxygen fugacity. Geochim. Cosmochim. Acta, 55(9), 2545-
2554.
Fleet, M.E., Chryssiykus, S.L., Stone W.E., and Weisener, C.G. (1993) Partitioning of
platinum-group elements and Au in the Fe-Ni-Cu-S system: experiments on the
fractional crystallization of sulfide melt. Contrib. Mineral. Petro. 115, 36-44.
Fleet, M.E. and Pan, Y.M. (1994) Fractional crystallization of anhydrous sulfide liquid in the
system Fe-Ni-Cu-S, with application to magmatic sulfide deposits. Geochim.
Cosmochim. Acta 58, 3369-3377.
Fleet, D. J., Wagner, H., & Heeger, D. J. (1996). Neural encoding of binocular disparity:
energy models, position shifts and phase shifts. Vision research, 36(12), 1839-1857.
Fleet, M. E., Crocket, J. H., Liu, M., & Stone, W. E. (1999). Laboratory partitioning of
platinum-group elements (PGE) and gold with application to magmatic sulfide–PGE
deposits. Lithos, 47(1), 127-142.
Fortenfant, S. S., Dingwell, D. B., Ertel-Ingrisch, W., Capmas, F., Birck, J. L., & Dalpe, C.
(2006). Oxygen fugacity dependence of Os solubility in haplobasaltic melt. Geochim.
Cosmochim. Acta, 70(3), 742-756.
Fonseca, R.O.C., Mallman, G., O’Neil, H.C., Campell, I.H. (2007) How chalcophile is
rhenium? An experimental study of the solubility of Re in sulphide mattes. Earth
Planet Sci Lett 260, 537-548.7
Fonseca R. O. C., Campbell I. H., O’Neill H. S. C., Fitz Gerald J. D. (2008) Oxygen
solubility and speciation in sulphide-rich mattes. Geochim. Cosmochim. Acta
72, 2619-2635.
151
Fonseca R. O. C., Campbell I. H., O’Neill H. S. C., Allen, C.M. (2009) Solubility of Pt in
sulphide mattes: Implications for the genesis of PGE-rich horizons in layered
intrusions. Geochim. Cosmochim. Acta 73, 5764-5777.
Fonseca, R.O.C., Mallman, G., O’Neil, H.C., Campell, I.H., Laurenz, V. (2011)
Solubility of Os and Ir in sulfide melt: Implications for Re/Os fractionation during
mantle melting. Earth Planet Sci Lett 311: 339-350.7
Fortin, M.-A., Riddle, J., Desjardins-Langlais, Y., and Baker, D.R, (2015). The effect of
water on the sulfur concentration at sulfide saturation (SCSS) in natural melts.
Geochim. Cosmochim. Acta 160, 100-116.
Frost, B. R., Mavrogenes, J. A., & Tomkins, A. G. (2002). Partial melting of sulfide ore
deposits during medium-and high-grade metamorphism. Can. Mineral 40(1), 1-18.
Gao, S., Luo, T. C., Zhang, B.R., Zhang, H.F., Han, Y.W., Hu, Y.K. and Zhao, Z.D.et al.
(1998) Chemical composition of the continental crust as revealed by studies in east
China. Geochim. Cosmochim. Acta 62, 1959-1975.
Gervilla F., Leblanc M., Torres-Ruiz, J., and Fenoll H.P. (1996) Immiscibility between
arsenide and sulfide melts: a mechanism for the concentration of noble metals. Can.
Mineral 34, 485-502.
Gervilla, F., Papunen, H., Kojonen, K. and Johanson, B. (1998). Platinum-, palladium- and
gold-rich arsenide ores from the Kylmäkoski Ni-Cu deposit (Vammala Nickel Belt,
SW Finland); Mineralogy and Petrology 64, 163-185.
Gilbert, S., Danyushevsky, L., Robinson, P., Wohlgemuth-Ueberwasser, C. Pearson, N.,
Savard, D., Noman, M. and Hanley, J. (2013) A Comparative Study of Five
Reference Materials and the Lombard Meteorite for the Determination of the
Platinum-Group Elements and Gold by LA-ICP-MS. Geostandards & Geoanalytical
Research 37, 51-64.
Godel, B., Barnes, S.-J. and Maier, W. D. (2007) Platinum-group elements in sulphide
minerals, platinum-group minerals, and the whole rock of the Merensky Reef
152
(Bushveld Complex, South Africa): Implication for the formation of the reef. J.
Petro. 48, 1569-1604.
Godel, B., González-Álvarez, I., Barnes, S.-J., Barnes, S. J., Parker, P., & Day, J. (2012).
Sulfides and sulfarsenides from the rosie nickel prospect, Duketon Greenstone Belt,
Western Australia. Econ. Geol., 107(2), 275-294.
Govindaraju, K. (1994). 1994 compilation of working values and sample description for 383
geostandards. Geostandards newsletter, 18(S1), 1-158.
Guo, J., Griffin, W. L., & O'Reilly, S. Y. (1999). Geochemistry and origin of sulphide
minerals in mantle xenoliths: Qilin, Southeastern China. J. Petro., 40(7), 1125-1149.
Hanley, J. J. (2007). The role of arsenic-rich melts and mineral phases in the development of
high-grade Pt-Pd mineralization within komatiite-associated magmatic Ni-Cu sulfide
horizons at Dundonald Beach South, Abitibi subprovince, Ontario, Canada. Econ.
Geol., 102(2), 305-317.
Haughton, D. R., Roeder, P. L., & Skinner, B. J. (1974). Solubility of sulfur in mafic
magmas. Econ. Geol. 69(4), 451-467.
Helmy, H.M., Ballhaus, C., Berndt, J., Bockrath, C. and Wohlgemuth-Ueberwasser, C. (2007)
Formation of Pt, Pd, Ni tellurides: experiments in sulfide-telluride systems. Contrib.
Mineral. Petrol. 153, 577-591.
Helmy, H.M., Ballhaus, C., Wohlgemuth-Ueberwasser, C., Fonseca, R.O.C., and Laurenz, V.
(2010) Partitioning of Se, As, Sb, Te and Bi between monosulfide solid solution and
sulfide melt-Application to magmatic sulfide deposits. Geochim. Cosmochim. Acta
74, 6174-6179.
Helmy, H. M., Ballhaus, C., Fonseca, R. O.C., Wirth, R., Nagel, T., & Tredoux, M. (2013a).
Noble metal nanoclusters and nanoparticles precede mineral formation in magmatic
sulphide melts. Nature communications, 4.
153
Helmy, H.M., Ballhaus, C., Fonseca, R.O.C. and Nagel, T.J. (2013b) Fractionation of
platinum, palladium, nickel, and copper in sulfide–arsenide systems at magmatic
temperature. Contrib Mineral Petrol 166, 1725–1737.
Imai, N., Terashima, S., Itoh, S., & Ando, A. (1995). 1994 compilation values for GSJ
reference samples," Igneous rock series". Geochemical Journal 29(1), 91-95.
Jenner, F. E., & O'Neill, H. S. C. (2012). Major and trace analysis of basaltic glasses by
laser‐ablation ICP‐MS. Geochemistry, Geophysics, Geosystems: 13(3).
Jensen, E. (1942) Pyrrhotite: melting relations and composition: American Journal of
Science, 240, p.695-709.
Jugo, P.J., Candela, P.A. and Piccoli, P.M. (1999) Magmatic sulfides and Au:Cu ratios in
porphyry deposits: an experimental study of copper and gold partitioning at 8508C,
100 MPa in a haplogranitic melt–pyrrhotite–intermediate solid solution–gold metal
assemblage, at gas saturation. Lithos 46, 573–589.
Jugo, P.J., Luth, R.W. and Rochards, J.P. (2005) Experimental data on the speciation of
sulfur as a function of oxygen fugacity in basaltic melts. Geochim. Cosmochim. Acta.
69: 497-503.
Karup-Moller, S. and Makovicky, E. (2002) The system Fe-Os-S at 1180 °C, 1100 °C and
900 °C. Can. Mineral. 40, 499-507.
Klimm, K., Kohn, S. C., & Botcharnikov, R. E. (2012). The dissolution mechanism of
sulphur in hydrous silicate melts. II: Solubility and speciation of sulphur in hydrous
silicate melts as a function of fO 2. Chem. Geol. 322, 250-267.
Kress, V. (1997). Magma mixing as a source for Pinatubo sulphur. Nature, 389(6651), 591-
593.
154
Li, C., Naldrett, A.J., Coats, C.J.A., and Johannessen, P. (1992) Platinum, palladium, gold,
and copper-rich stringers at the Strathcona mine, Sudbury: Their enrichment by
fractionation of a sulfide liquid: Econ. Geol. 87, 1584–1598.
Li, C. and Ripley, E.M. (2009) Sulfur contents at sulfide-liquid or anhydrite saturation in
silicate melts: Empirical equations and example applications. Econ. Geol. 104, 405–
412.
Li, Y. and Andetat, A. (2012) Partitioning of V, Mn, Co, Ni, Cu, Zn, As, Mo, Ag, Sn, Sb,
W, Au, Pb, and Bi between sulfide phases and hydrous basanite melt at upper mantle
conditions. Earth Planet Sci Lett 355, 327-340.
Li, Y. and Andetat, A. (2013) Gold solubility and partitioning between sulfide liquid,
monosulfide solid solution and hydrous mantle melts: Implications for the formation
of Au-rich magmas and crust–mantle differentiation. Geochim. Cosmochim. Acta 118,
247-262.
Li, Z., Lin, C., Qu, G., Nie, Q., Xu, T., & Dai, S. (2014). Phase Separation in
Nonstoichiometry Ge–Sb–S Chalcogenide Glasses. Journal of the American Ceramic
Society, 97(3), 793-797.
Liu, Y., Samaha, N. T., & Baker, D. R. (2007). Sulfur concentration at sulfide saturation
(SCSS) in magmatic silicate melts. Geochim. Cosmochim. Acta 71(7), 1783-1799.
Liu, Y., & Brenan, J. (2015). Partitioning of platinum-group elements (PGE) and chalcogens
(Se, Te, As, Sb, Bi) between monosulfide-solid solution (MSS), intermediate solid
solution (ISS) and sulfide liquid at controlled fO 2–fS 2 conditions. Geochim.
Cosmochim. Act 159, 139-161. doi:10.1016/j.gca.2015.03.021.
Lorand, J. P., & Alard, O. (2001). Platinum-group element abundances in the upper mantle:
new constraints from in situ and whole-rock analyses of Massif Central xenoliths
(France). Geochim. Cosmochim. Acta 65(16), 2789-2806.
Luguet, A., Alard, O., Lorand, J. P., Pearson, N. J., Ryan, C., & O’Reilly, S. Y. (2001).
Laser- ablation microprobe (LAM)-ICPMS unravels the highly siderophile element
geochemistry of the oceanic mantle. Earth Planet. Sci. Lett. 189(3), 285-294.
155
Luguet, A., Lorand, J. P., Alard, O., & Cottin, J. Y. (2004). A multi-technique study of
platinum group element systematic in some Ligurian ophiolitic peridotites, Italy.
Chem. Geol. 208(1), 175-194.
Maier, W. D. (2005). Platinum-group element (PGE) deposits and occurrences:
mineralization styles, genetic concepts, and exploration criteria. Journal of African
Earth Sciences, 41(3), 165-191.
Makovicky, E. and Karup-Mollers, S. (1993): The system Pd–Fe–S at 900°, 725°, 550°, and
400°C. Econ. Geol. 88: 1269 –1278.
Makovicky, E., Makovicky, M., and Rose-Hansen, J. (2002) The system Fe-Rh-S at 900 °C
and 500 °C. Can. Mineral. 40: 519-526.
Makovicky, E. (2006) Crystal Structures of Sulfides and Other Chalcogenides. Reviews in
Mineralogy and Geochemistry 61: 7-125.
Mavrogenes, J. A., & O’Neill, H. S. C. (1999). The relative effects of pressure, temperature
and oxygen fugacity on the solubility of sulfide in mafic magmas. Geochim.
Cosmochim. Acta 63(7), 1173-1180.
Mavrogenes, J. A., MacIntosh, I. W., & Ellis, D. J. (2001). Partial melting of the Broken Hill
galena-sphalerite ore: Experimental studies in the system PbS-FeS-ZnS-(Ag2S).
Econ. Geol. 96(1), 205-210.
McDonough, W.F. and Sun, S-S. (1995) The Composition of the Earth. Chem. Geol. 120,
223-253.
McDonough, W. F. (2003). Compositional model for the Earth's core. Treatise on
geochemistry, 2, 547-568.
156
Mungall, J.E., Ames, d.E. and Hanley, J.J. (2004) Geochemical evidence from the Sudbury
structure for crustal redistribution by large bolide impacts. Nature 429, 546-548.
Mungall, J.E., Andrews, D.R.A., Cabri, L.J., Sylvester, P.J. and Tubrett M. (2005)
Partitioning of Cu, Ni, Au, and platinum-group elements between monosulfide solid
solution and sulfide melt under controlled oxygen and sulfur fugacities. Geochim.
Cosmochim. Acta 69, 4349- 4360.
Mungall, J.E. (2007) Crystallization of magmatic sulfides: An empirical model and
application to Sudbury ores. Geochim. Cosmochim. Acta 71, 2809-2819.
Mungall, J.E. and Brenan, J.M. (2014) Partitioning of platinum-group elements and Au
between sulfide liquid and basalt and the origins of mantle-crust fractionation of the
chalcophile elements. Geochim. Cosmochim. Acta, 125, 265-289.
Naldrett, A. J. (1969). A portion of the system Fe–S–O between 900 and 1080 C and its
application to sulfide ore magmas. J. Petro. 10(2), 171-201.
Naldrett, A. J. (1981). Nickel sulfide deposits: classification, composition and genesis. Econ.
Geol, 75, 628-655.
Naldrett, A. J., Innes, D. G., Sowa, J., & Gorton, M. P. (1982). Compositional variations
within and between five Sudbury ore deposits. Econ. Geol. 77(6), 1519-1534.
Naldrett, A. J., Lightfoot, P. C., Fedorenko, V., Doherty, W., & Gorbachev, N. S. (1992).
Geology and geochemistry of intrusions and flood basalts of the Noril'sk region,
USSR, with implications for the origin of the Ni-Cu ores. Econ. Geol. 87(4), 975-
1004.
Naldrett, A.J., Asif, M., Schandl, E., Searcy, T., Morrison, G., Binney, P., and Moore, C.,
(1999) PGE in the Sudbury ores: Significance with respect to the origin of different
ore zones and the exploration for footwall orebodies: Econ. Geol. 94,: 185–
210.
157
Naldrett, A. J. (2004). Magmatic sulfide deposits: geology, geochemistry and exploration.
Springer Science & Business Media.
Naldrett, A. J. (2013). The lithospheric mantle plays no active role in the formation of
orthomagmatic ore deposits. Econ. Geol., 108(8), 1953-1970.
O'Neill, H. S. C., & Wall, V. J. (1987). The Olivine—Orthopyroxene—Spinel oxygen
geobarometer, the nickel precipitation curve, and the oxygen fugacity of the Earth's
Upper Mantle. J. Petro, 28(6), 1169-1191.
Patten, C., Barnes, S.-J., Mathez, E. A., & Jenner, F. E. (2013). Partition coefficients of
chalcophile elements between sulfide and silicate melts and the early crystallization
history of sulfide liquid: LA-ICP-MS analysis of MORB sulfide droplets. Chem.
Geol., 358, 170-188.
Perkins, T. T., Smith, D. E., & Chu, S. (1997). Single polymer dynamics in an elongational
flow. Science, 276(5321), 2016-2021.
Peach, C. L., Mathez, E. A., & Keays, R. R. (1990). Sulfide melt-silicate melt distribution
coefficients for noble metals and other chalcophile elements as deduced from MORB:
implications for partial melting. Geochim. Cosmochim. Acta, 54(12), 3379-3389.
Peach, C. L., Mathez, E. A., Keays, R. R., & Reeves, S. J. (1994). Experimentally
determined sulfide melt-silicate melt partition coefficients for iridium and palladium.
Chem. Geol., 117(1), 361-377.
Raghavan, V. (2004) Cu-Fe-S (Copper-Iron-Sulfur). J. Phase Equi. Diff. 25(5): 450-457.
Raghavan, V. (2006) Cu-Fe-S (Copper-Iron-Sulfur). J. Phase Equi. Diff. 27(3): 290. DOI:
10.1361/154770306X109872.
158
Rajamani, V. and Naldrett, A.J. (1978) Partitioning of Fe, Co, Ni, and Cu between
Sulfide Liquid and Basaltic Melts and the Composition of Ni-Cu Sulfide Deposits.
Econ. Geol. 73, 82-93.
Raybaud, P., Kresse, G., Hafner, J and Toulhoat, H. (1997) Ab initio density functional
studies of transition-metal sulphides: I. Crystal structure and cohesive properties. J.
Phys.: Condens. Matter 9, 11085–11106.
Righter, K., & Drake, M. J. (2000). Metal/silicate equilibrium in the early Earth—new
constraints from the volatile moderately siderophile elements Ga, Cu, P, and Sn.
Geochim. Cosmochim. Acta, 64(20), 3581-3597.
Rose-Weston, L., Brenan, J.M., Fei, Y.W., Secco, R.A. and Frost, D.J. (2009) Effect of
pressure, temperature and oxygen fugacity on the metal-silciate partitiong of Te, Se
and S: Implications for earth differentiation. Geochim. Cosmochim. Acta 73: 4598-
4615.
Rudnick, R.L. and Gao, S. (2003) Composition of the Continental Crust. Treatise on
Geochemistry, Volume 3. (Editor: Roberta L. Rudnick. Executive Editors: Heinrich
D. Holland and Karl K. Turekian). ISBN 0-08-043751-6. Elsevier, 2003: 1-64.
Salters, V.J.M. and Stracke, A. (2004). Composition of the depleted mantle. Geochem.
Geophys. Geosys. 5: 1525-2027. doi: 10.1029/2003GC000597.
Savard, D., Barnes, S.- J., & Meisel, T. (2010). Comparison between Nickel‐Sulfur Fire
Assay Te Co‐precipitation and Isotope Dilution with High‐Pressure Asher Acid
Digestion for the Determination of Platinum‐Group Elements, Rhenium and Gold.
Geostand. Geoanal. Res., 34(3), 281-291.
Stevens, G., Prinz, S., & Rozendaal, A. (2005). Partial melting of the assemblage sphalerite+
galena+ pyrrhotite+ chalcopyrite+ sulfur: implications for high-grade metamorphosed
massive sulfide deposits. Econ. Geol. 100(4), 781-786.
Sun, Y. L., Zhou, M. F. and Sun, M. (2001) Routine Os analysis by isotope dilution-
inductively coupled plasma mass spectrometry: OsO4 in water solution gives high
sensitivity. J. Anal. At. Spectro.16, 345–349.
Sylvester, P.J. (2001) A practical guide to platinum-group element analysis of sulfides by
laser ablation ICPMS. P. Sylvester (Ed.), Laser-Ablation-ICPMS in the Earth
Sciences, Principles and Applications, Min. Assoc, Canada, 203–211 Short Course 29,
Chap. 13.
159
Sylvester, P. J. (2005) Laser ablation ICP-MS developments and trends for 2003. Geostand.
Geoanal. Res. 29, 41–52.
Szabó, C., & Bodnar, R. J. (1995). Chemistry and origin of mantle sulfides in spinel
peridotite xenoliths from alkaline basaltic lavas, Nógraád-Gomor Volcanic Field,
northern Hungary and southern Slovakia. Geochim. Cosmochim. Acta, 59(19), 3917-
3927.
Tomkins, A.G. and Maverogenes, J.A. (2001) Redistribution of gold within arsenopyrite and
lollingite during Pro-and retrograde metamorphism: Application to timing of
mineralization. Econ. Geol. 96, 525-534.
Toulmin, P. and Barton, P.B. (1964) A Thermodynamic study of pyrite and pyrrhotite.
Geochim. Cosmochim. Acta 28, 641-671.
Wang, H.P. and Salveson, I. (2005) A review on the mineral chemistry of the non-
stoichiometric iron sulphide, Fe1-x S (0 <= x <= 0.125): polymorphs, phase relations
and transitions, electronic and magnetic structures. Phase Transitions 78, 547-567.
Wedepohl, K.H. (1995) The composition of the continental crust. Geochim. Cosmochim.
Acta 59, 1217-1239.
Wendlandt, R.F. (1982) Sulfide saturation of basalt and andesite melts at high pressures and
temperatures. Am. Mineral. 67, 877–885.
Wilson, S.A., Ridley, W.I. and Koenig, A.E. (2002) Development of sulfide calibration
standards for the laser ablation inductively-coupled plasma mass spectrometry
technique. J. Anal. At. Spectrom. 17: 406-409.
Wohlgemuth-Ueberwasser C.C., Ballhaus, C., Berndt, J., Stotter nee Paliulionyte, V. and
Meisel, T. (2007) Synthesis of PGE sulfide standards for laser ablation inductively
coupled plasma mass spectrometry (LA-ICP-MS). Contrib Mineral Petrol 154, 607–
617.
Yi, W., Halliday, A.N., Alt, J.C., Lee, D., Rehkämper, M., Garcia, M.O. and Su, Y. (2000).
Cadmium, indium, tin, tellurium, and sulfur in oceanic basalts: Implications for
160
chalcophile element fractionation in the Earth. J. of Geophysical Research, 105, doi:
10.1029/2000JB900152. issn: 0148-0227.
Yi, C., Zhang, P., Chen, F., Dai, S., Wang, X., Xu, T. and Nie, Q. (2014) Fabrication and
characterization of GeSbS chalcogenide glass for photonic crystal fibers. Applied
Physics B: Lasers & Optics 116: 653.
Zientek, M.L., Likhachev, A.P., Kunilov, V.E., Barnes, S.J., Meier, A.L., Carlson, R.R.,
Briggs, P.H., Fries, T.L. and B.M. Adrian (1994) Cumulus processes and the
composition of magmatic ore deposits: Examples from the Talnakh District, Russia.
Ont. Geol. Surv. Spec. 5, 373–392.
161
Appendix
To facilitate the modeling, a MATLAB program with GUI was created. The user can choose
any two element pairs for the X-and y-axes, and monitor their evolutionary curves during the
crystallization process of the sulfide liquid. Stage I and Stage II adopt fractional
crystallization by default, while Stage III has the option to choose between either fractional
or equilibrium crystallization. A snapshot of the software is shown in Fig A-1. The upper left
panel allows the user to input the estimated initial concentrations for elements on the X- and
Y-axes individually. The pull-down menu enables the user to choose “fractional
crystallization mode” or “equilibrium crystallization mode” for Stage III. The panel in the
upper middle position allows the user to input the partition coefficients between MSS/sulfide
melt and ISS/sulfide melt for the elements chosen. The upper right panel allows the user to
vary the "boundary point", which is defined by the concentration of X-axis element at which
crystallization enters Stage II, and then a second boundary point (“Starting X for Stage III”)
defines when crystallization enters Stage III. This same panel also allows the user to define
model reference points, such as the detection limit of the Y-axis element or its estimated
saturation level. This feature is designed to help the user to visualize whether the modeled
concentration of the Y-axis element makes practical sense. For example, in the case of the
compatible elements (Ru, Ir, etc), the modeled concentration would not have any
significance if it falls below the analytical detection limit. For incompatible elements,
continuous crystallization will lead to their accumulation in the residual liquid. However,
162
once the saturation of that element is reached, the concentration in the coexisting sulfide
liquid and ISS may/may not follow the simple evolutionary trend predicted by the model.
The lower left panel allows the user to specify the plotting ranges for both X and Y axis.
Once the "Plot" button is clicked, the panel in the lower right corner will display the values
of the corresponding "F" (fraction of liquid remaining compared with the original bulk) at
each boundary point. If a saturation concentration is specified, the corresponding F to this
saturation will also be displayed in this panel. At the same time, a new MATLAB figure
window will pop up with the corresponding plots. The user can then make use of the
inherited features of MATLAB figure window to rescale, label, save or print the modeled
curves. In the menu bar zone of this GUI, there is also the option to print preview the main
window to record the parameter settings as a PDF file. Two snap shots of the generated plots
are illustrated in Figure A-2 to present the different scenarios of fractional and equilibrium
crystallizations in Stage III, and the detailed MATLAB program is attached as follows:
popup_sel_index = get(handles.popupmenu1, 'Value');
switch popup_sel_index
case 1 %fractional crystallization
Dcu_mss = str2double(get(handles.D1x, 'String'));
%read in MSS/melt partitioning data for element on X axis
Dpd_mss = str2double(get(handles.D1y, 'String'));
%read in MSS/melt partitioning data for element on Y axis
163
C0cu = str2double(get(handles.C0x, 'String'));
%read in estimated initial concentration for element on X axis
C0pd = str2double(get(handles.C0y, 'String'));
%read in estimated initial concentration for element on Y axis
CuX1_Max = str2double(get(handles.Cx1max, 'String'));
%read in user defined boundary point between Stage I and Stage II
figure();
%generage a new figure to make use of MATLAB build in plotting tool
%Stage I fractional crystallization
Fmax = 1.0;
%initial fraction of liquid 100%, taken as 1
Fmin1 = power(CuX1_Max/C0cu,1.0/(Dcu_mss-1));
%calculate the fraction of liquid remaining when X-axis element reached the user-defined
boundary point
164
F = [Fmin1:0.01:Fmax];
%Step size set to be relative 1% for Stage I
Cx1 = C0cu*power(F,Dcu_mss-1);
%Reyleigh fractionation equation to calcualte the liquid composition
Cy1 = C0pd*power(F,Dpd_mss-1);
semilogy(Cx1,Cy1,'b-','LineWidth',2);hold on;
%plot liquid curve for Stage I
semilogy(Cx1*Dcu_mss,Cy1*Dpd_mss,'g-', 'LineWidth',2);
%plot corresponding MSS curve for Stage I
set(handles.Fmin1, 'String', Fmin1);
%record and display the fraction of liquid at then end of Stage I
Fpts1 = [0.2,0.3,0.6,1.0];
%set places where tick markers are intended using F as index
Cxpds1 = C0cu*power(Fpts1,Dcu_mss-1);
%corresponding liquid concentrations at these tick marked positions can be calculated by
Reyleigh
165
Cypds1 = C0pd*power(Fpts1,Dpd_mss-1);
dx = 1; dy = 0;
% set up the length of ticks
for i=1:4
semilogy([Cxpds1(i)*Dcu_mss - dx,Cxpds1(i)*Dcu_mss + dx],
[Cypds1(i)*Dpd_mss, Cypds1(i)*Dpd_mss],'g-', 'LineWidth',2);
semilogy([Cxpds1(i)*Dcu_mss,Cxpds1(i)*Dcu_mss], [Cypds1(i)*(1-
dy)*Dpd_mss, Cypds1(i)*(1+dy)*Dpd_mss],'g-', 'LineWidth',2);
end
% plotting tick marks at these specified locations on MSS curve
dx = 0; dy = 0.2; % set up the length of ticks
for i=1:4
semilogy([Cxpds1(i) - dx,Cxpds1(i)+ dx], [Cypds1(i),
Cypds1(i)],'b-', 'LineWidth',2);
166
semilogy([Cxpds1(i),Cxpds1(i)], [Cypds1(i)*(1+dy), Cypds1(i)*(1-
dy)],'b-', 'LineWidth',2);
end
% plotting tick marks at these specified locations on liquid curve
PdX1_Max= C0pd*power(Fmin1,Dpd_mss-1);
%record the corresponding Y-axis element concentration when Stage I ends
% Stage 2 fractional crystallization
C0cu2 = CuX1_Max;
%the initial concentration in stage II should be the same as the ending concentration in Stage
I
C0pd2 = PdX1_Max;
Dcu_iss = str2double(get(handles.D2x, 'String'));
Dpd_iss = str2double(get(handles.D2y, 'String'));
%read in ISS/melt partition coefficients
167
CuX2_Max= str2double(get(handles.D2x, 'String'));
%User define boundary point between Stage II and III
F(1) = 1.0;
Cx(1) = C0cu2;
Cy(1) = C0pd2;
i = 1;
%initializing for the loop to calculate the liquid concentration
while(Cx(i) >=22 & Cx(i)<=CuX2_Max & F>0)
%conditional loop
i = i+1;
F(i) = 0.95;
%fixed step size, model should be independent of step size, we are using 5% fractionation
each time
mir(i) = 1.92*power(F(i),0.33);
%regression function yield by Fleet et al. (1993) experimental data and the data in this
project
168
Cx(i) = Cx(i-1)/(F(i) + Dcu_iss*(1-F(i))/(mir(i)+1) +
Dcu_mss*mir(i)*(1-F(i))/(mir(i)+1));
Cy(i) = Cy(i-1)/(F(i) + Dpd_iss*(1-F(i))/(mir(i)+1) +
Dpd_mss*mir(i)*(1-F(i))/(mir(i)+1));
%liquid concentrations calculated at each step
Fmin2 = power (F(i), i-1);
%trying to keep track of the fraction of liquid remaining in the system compared with the
beginning of Stage II
PdX2_Max = Cy(i);
%keep track of the final concentration of Y_axis element at the end of Stage II
if Fmin2<1 && Fmin2>0.8
CxH(1)= Cx(i);
CyH(1)= Cy(i);
F_H(1)=Fmin2;
elseif Fmin2>0.6 && Fmin2<0.8
CxH(2)= Cx(i);
CyH(2)= Cy(i);
F_H(2) = Fmin2;
169
elseif Fmin2>0.3 && Fmin2<0.5
CxH(3)= Cx(i);
CyH(3)= Cy(i);
F_H(3) = Fmin2;
end
%pass out the positions where tick marks are preferred
end
semilogy(Cx,Cy,'b-','LineWidth',2);hold on
% liquid curve in stage II
dx = 0;dy = 0.2;
%set up lengths for tick marks on liquid curves
for j=1:3
semilogy([CxH(j)+dx,CxH(j)-dx], [CyH(j), CyH(j)],'b-',
'LineWidth',2);
170
semilogy([CxH(j),CxH(j)], [CyH(j)*(1+dy), CyH(j)*(1-dy)],'b-',
'LineWidth',2);
end
semilogy(Cx*Dcu_iss,Cy*Dpd_iss,'m-', 'LineWidth',2);
% ISS curve in Stage II
dx = 0;dy = 0.2;
%set up lengths for tick marks on ISS curves
for j=1:3
semilogy([CxH(j)*Dcu_iss+dx,CxH(j)*Dcu_iss-dx], [CyH(j)*Dpd_iss,
CyH(j)*Dpd_iss],'m-', 'LineWidth',2);
semilogy([CxH(j)*Dcu_iss,CxH(j)*Dcu_iss], [CyH(j)*Dpd_iss*(1+dy),
CyH(j)*Dpd_iss*(1-dy)],'m-', 'LineWidth',2);
end
semilogy(Cx*Dcu_mss,Cy*Dpd_mss,'r-', 'LineWidth',2);
% MSS curve in Stage II
171
dx = 1;dy = 0;
%set up lengths for tick marks on MSS curves
for j=1:3
semilogy([CxH(j)*Dcu_mss+dx,CxH(j)*Dcu_mss-dx], [CyH(j)*Dpd_mss,
CyH(j)*Dpd_mss],'r-', 'LineWidth',2);
semilogy([CxH(j)*Dcu_mss,CxH(j)*Dcu_mss], [CyH(j)*Dpd_mss*(1+dy),
CyH(j)*Dpd_mss*(1-dy)],'r-', 'LineWidth',2);
end
set(handles.Fmin2, 'String', Fmin2*Fmin1);
%fraction of liquid remaining at the end of stage II compared with original bulk is displayed
in the main interface
%stage III fractional Crystallization
172
C0_Cu3 = CuX2_Max;
C0_Pd3 = PdX2_Max;
%initial concentration in Stage III should be the same as the maximm in Stage II
F3 = [0.002,0.3,0.6,1];
%setting up the place to have tick marks
Cx = C0_Cu3*power(F3,Dcu_iss-1);
Cy = C0_Pd3*power(F3,Dpd_iss-1);
%calculate the liquid composition with Rayleigh fractionation
F3pts = [0.002,0.3,0.6,1];
Cxpd3 = C0_Cu3.*power(F3pts,Dcu_iss-1);
Cypd3 = C0_Pd3.*power(F3pts,Dpd_iss-1);
%calculate the corresponding coordinates for tick marks.
semilogy(Cx,Cy,'b-','LineWidth',2);hold on
% liquid curve in stage III
semilogy(Cx*Dcu_iss,Cy*Dpd_iss,'m-', 'LineWidth',2);
%iss curve in Stage III
173
dx = 1;dy = 0;
% adding markers on liquid curves
for i=1:4
semilogy([Cxpd3(i)+dx,Cxpd3(i)-dx], [Cypd3(i), Cypd3(i)],'b-',
'LineWidth',2);
semilogy([Cxpd3(i),Cxpd3(i)], [Cypd3(i)*(1+dy), Cypd3(i)*(1-
dy)],'b-','LineWidth',2);
end
dx = 1; dy=0;
%adding markers on ISS curves
for i=1:4
semilogy([Cxpd3(i)*Dcu_iss + dx,Cxpd3(i)*Dcu_iss-dx],
[Cypd3(i)*Dpd_iss, Cypd3(i)*Dpd_iss],'m-', 'LineWidth',2);
semilogy([Cxpd3(i)*Dcu_iss,Cxpd3(i)*Dcu_iss], [Cypd3(i)*(1-
dy)*Dpd_iss, Cypd3(i)*(1+dy)*Dpd_iss],'m-', 'LineWidth',2);
174
end
xmin = str2double(get(handles.xmin, 'String'));
xmax = str2double(get(handles.xmax, 'String'));
ymin = str2double(get(handles.ymin, 'String'));
ymax = str2double(get(handles.ymax, 'String'));
axis([xmin xmax ymin ymax]);
%read in the plot range options specified by user
Y_max = str2double(get(handles.saturation, 'String'));
% read in estimated saturation values
X_range = [xmin,xmax];
semilogy(X_range,[Y_max,Y_max],'r-.','LineWidth',1);hold on
%plot the saturation line for Y-axis element
F_relative = exp((log(Y_max)-log(C0_Pd3))/(Dpd_iss-
1))/(Fmin1*Fmin2);
%calculate the fraction of liquid compared with original bulk
set(handles.F_relative, 'String', F_relative);
175
%Display the F when saturation is reached compared with original bulk
case 2 % Equilibrium Crystallization
%Stage I fractional crystallization
…………………….
%Stage II fractional crystallization
…………………..
%stage III Equilibrium Crystallization
C0_Cu3 = CuX2_Max;
C0_Pd3 = PdX2_Max;
%initial concentration at the beginning of Stage III
F3 = [0.002,0.3,0.6,1];
Cxpd3 = C0_Cu3./(Dcu_iss - Dcu_iss*F3 + F3);
176
Cypd3 = C0_Pd3./(Dpd_iss - Dpd_iss*F3 + F3);
% liquid concentration calculated using mass balance
semilogy(Cxpd3,Cypd3,'b-','LineWidth',2);hold on
% liquid curve in stage III
dx = 1;dy = 0;
%setting up lengths for tick marks
for i=1:4
semilogy([Cxpd3(i)+dx,Cxpd3(i)-dx], [Cypd3(i), Cypd3(i)],'b-',
'LineWidth',2);
semilogy([Cxpd3(i),Cxpd3(i)], [Cypd3(i)*(1+dy), Cypd3(i)*(1-
dy)],'b-', 'LineWidth',2);
end
%draw tick marks at preferred locations
semilogy(Cxpd3*Dcu_iss,Cypd3*Dpd_iss,'m-', 'LineWidth',2); %
ISS curve in Stage III
177
dx = 1; dy=0;
for i=1:4
semilogy([Cxpd3(i)*Dcu_iss + dx,Cxpd3(i)*Dcu_iss-dx],
[Cypd3(i)*Dpd_iss, Cypd3(i)*Dpd_iss],'m-', 'LineWidth',2);
semilogy([Cxpd3(i)*Dcu_iss,Cxpd3(i)*Dcu_iss], [Cypd3(i)*(1-
dy)*Dpd_iss, Cypd3(i)*(1+dy)*Dpd_iss],'m-', 'LineWidth',2);
end
xmin = str2double(get(handles.xmin, 'String'));
xmax = str2double(get(handles.xmax, 'String'));
ymin = str2double(get(handles.ymin, 'String'));
ymax = str2double(get(handles.ymax, 'String'));
axis([xmin xmax ymin ymax]);
%setting up plot options specified by user
Y_max = str2double(get(handles.saturation, 'String'));
178
% estimated saturation level
X_range = [xmin,xmax];
semilogy(X_range,[Y_max,Y_max],'r-.','LineWidth',2);hold on
%plot up the saturation line
Y_min = str2double(get(handles.Y_DL, 'String'));
%detection limit for Y-axis element
semilogy(X_range,[Y_min,Y_min],'b-.','LineWidth',2);hold on
%plot up the detection limit line
end
179
Figure A-1 Snapshot of the Smelt interface, which can be convenient saved to remember
the parameters settings for the calculation.
180
Figure A-2 Examples of fractional crystallization (upper) and equilibrium crystallization
(lower) in stage III.