Abstracts and Publications


Bennett, N. and Brenan, J.M.

Controls on the solubility of rhenium in silicate melt: Implications for the osmium isotopic composition of Earth's mantle. Earth and Planetary Science Letters, In press.

Brenan, J.M., Finnigan, C.F., McDonough, W.F. and Homolova, V. 

Experimental constraints on the partitioning of the Ru, Rh, Ir, Pt and Pd between chromite and silicate melt: The importance of ferric iron. Chemical Geology, 302303, pp 1632, 2012.

Caciagli-Warman, N.C., Brenan, J.M., McDonough, W.F., and Phinney, W.C. 

Experimental constraints on Li partitioning and Li isotope fractionation during subduction zone dehydration.  Chemical Geology, 280, pp 384398, 2011.

Brenan, J.M. and Bennett, N. Soret

Separation of highly siderophile elements in Fe-Ni-S melts: Implications for solid metal-liquid metal partitioning.  Earth and Planetary Science Letters, vol 298, pp 299-305, 2010

Brenan, J.M. and McDonough, W.F.

Core formation and metal-silicate fractionation of osmium and iridium from gold. Nature Geoscience, vol 2, pp 798-801, 2009.

Rose, L.A., Brenan, J.M., Fei, Y., Secco, R.A. and Frost, D.

Perspectives on Earth Differentitation from Metal-Silicate Partitioning of Te, Se, and S.  Geochimica et Cosmochimica Acta., vol 73, 4598-4615, 2009.

Brenan, J.M., Haider, N. and Andrews, D.

Experimental evaluation of liquid immiscibility in a portion of the system Fe-Ni-Cu-S using high gravitational acceleration. Economic Geology, vol 103, pp 1563-1570, 2008

Finnigan, C.S., Brenan, J.M., Mungall, J.E. and McDonough, W.F.

Experiments and models bearing on the role of chromite as a collector of platinum group minerals by local reduction.  Journal of Petrology, vol 49, pp 1647-1665, 2008.

Brenan, J.M.

Re-Os Ffractionation by sulfide-silicate partitioning: A new spin. Chemical Geology, Special Issue on Highly Siderophile Elements, vol 248, pp 140-165, 2008


Presented at the 2006 GAC-MAC Meeting, Montreal, Canada

Sulfide-Silicate Partitioning of Mo, Sb, Pb, Se AND Te: Inital Results

Brenan, J.M., brenan@geology.utoronto.ca, Dept of Geology, University of Toronto,

McDonogh, W.F., Dept of Geology, University of Maryland

There are a variety of trace elements whose geochemical affinity (chalcophile, siderophile, lithophile) in magmatic systems depends on the conditions of P, T and the fugacities of oxygen and sulfur. This is particularly the case for metals and semi-metals, like Mo, Sb, Pb, Se and Te, and a full interpretation of their behaviour requires knowledge of partitioning amongst oxide, sulfide and metal phases. To assess the role of sulfide in controlling these elements, we are performing experiments to measure partitioning between molten sulfide and silicate. Our initial experiments have been performed at 1 atm with samples consisting of doped Fe sulfide (approx. 1000 ppm each) + synthetic basalt contained in crucibles made from San Carlos olivine megacrysts. The crucible, along with a sulfide buffer (i.e., Pt-PtS) is vacuum-sealed in a fused quartz ampoule, then equilibrated under static conditions at 1200ºC for 2-4 days. To enhance sulfide-silicate phase separation, we have also done an experiment in which, after the static soak, the sample is subject to an acceleration of about 400g, at 1200ºC, using a modified large capacity centrifuge. High pressure experiments are also in progress, in this case samples are contained in graphite, which fixes the oxygen fugacity near the CCO buffer. By buffering the fugacity of one gas species, the fugacity of the other can be estimated through the heterogeneous equilibrium: FeO(silicate) + S2 = FeS(sulfide) + O2. The accuracy of this method was confirmed using the Fe content of Ir wires added to separate experiments. Coexisting sulfide and silicate are analysed for trace elements by LA-ICPMS. Preliminary results obtained for 1 atm experiments at Pt-PtS/FMQ 0.5 yielded sulfide-silicate partition coefficients of 3, 7 and 40 for Mo, Sb and Pb, respectively. Minimum partition coefficients for Te and Se are ~1000 and ~100, as these elements were below detection in coexisting silicate. Available mineral-silicate melt partitioning of Mo, Sb and Pb suggests these elements should be more incompatible then the LREEs during mantle melting. In contrast, the constancy of Pr/Mo, Pr/Sb and Ce/Pb in oceanic lavas indicates similar partitioning. Our preliminary sulfide-silicate partitioning data resolves this paradox, and points to mantle sulfide as an additional host for apparently “lithophile” elements in the source for MORB and OIB.


Presented at the 2006 AGU Joint-Assembly, Baltimore, USA

Is a 'Late Veneer' Necessary? Answers From Metal-Silicate Partitioning of Te, Se, and S

* Rose, L A (rose@geology.utoronto.ca) , University of Toronto, 22 Russell Street, Toronto, ON M5S 3B1 Canada
Brenan, J M (brenan@geology.utoronto.ca) , University of Toronto, 22 Russell Street, Toronto, ON M5S 3B1 Canada

Chondrite-normalized mantle abundances of Te, Se, and S show depletions from the Planetary Volatility Trend (McDonough and Sun, 1995). This is likely due to their siderophile behavior under reducing conditions (Kilburn and Wood, 1997), resulting in strong partitioning into the core-forming Fe-Ni liquid. A mass balance calculation using Te, Se, and S concentrations in C1-chondrites and a modeled primitive mantle shows similar depletions for each element, with core/mantle "partitioning" of about 100 (McDonough and Sun, 1995). Values for Se and Te relative to S used in this estimate are consistent with those measured in sulfide globules from the upper mantle (Hattori et al., 2001). Experiments have been done to investigate whether this depletion could be achieved solely by equilibrium metal-silicate partitioning at the P-T-{\it fO$_{2}$ conditions relevant to core formation. If experimental results match the estimated D$^{core-mantle}$, this could reflect either a single P-T condition at the base of a magma ocean where metallic liquid ponded, or an average P-T condition covering a depth range within which metallic liquid droplets rained through semi-crystalline silicate. If the results are discrepant, the abundance of Te, Se, and S were disrupted after initial equilibrium partitioning. Higher experimental D$_{Te,Se,S}$ necessitates the addition of chalcophile-rich material to the mantle after core formation (i.e. a 'late veneer') and lower experimental D$_{Te,Se,S}$ requires removal of chalcophiles from the mantle, perhaps in a late 'Hadean sulfide matte' (Wood and Halliday, 2003). Liquid metal-liquid silicate partitioning experiments were performed using both piston-cylinder and multi-anvil presses over a range of pressure (1-12 GPa) and temperature (1835-$2135\deg$C). Powdered oxide and metal added in 50:50 proportion was contained within MgO crucibles. Oxygen fugacity was varied by adding different amounts of Si to the starting material. Experiments ranged from -7 to -0.5 log units below the iron-w\"{u}stite buffer (IW). Major elements in both quenched metal and silicate were determined by EMPA while Te and Se abundance in the quenched silicate portion was measured using LA-ICP-MS at the University of Toronto. At constant P and T, all three metal-silicate partition coefficients increase with increasing {\it fO$_{2}$. Te and S are the most siderophile, followed by Se. The presence of 5-10 wt% S increased both D$_{Te}$ and D$_{Se}$ by a factor of three. At 2.9 GPa, $1960\deg$C and log {\it fO$_{2}$ between IW-4 and IW-1.5, metal- silicate partitioning of the three elements converge to 100, matching the observed core-mantle partitioning. Previously, it was shown that both D$_{S}$ and D$_{Se}$ are positively correlated with P while increased T lowers D$_{S}$ (Li and Agee, 1996; Li, 2000). After applying these corrections and extrapolating our data to the estimated conditions of core formation (P=25-40 GPa; T=2000-$3500\deg$C; e.g. Li and Agee, 1996; Righter et al., 1997), the point of equal metal- silicate partitioning of Te, Se, and S is only raised by a factor of 3. Consequently, the observed core-mantle partitioning of Te, Se, and S is consistent with high temperature equilibrium within a magma ocean, obviating the need for a 'late veneer'. Chalcophile element (including Pb) loss to the core with the separation of a Hadean sulfide 'matte' would have further depleted the mantle. It would be quite fortuitous if a 'late veneer' refertilized the mantle to match the expected equilibrium abundances prescribed by metal-silicate partitioning. This result is consistent with conclusions based on the metal-silicate partitioning behavior of other elements, for instance Ni, Co, and Pt.


Presented at the 2006 Workshop on Highly Siderophile Elements, Durham, U.K.

Sulfide phase equilibrium and HSE partitioning near the mantle solidus
James M. Brenan, Dept of Geology, University of Toronto, 22 Russell St, Toronto, Canada
William F. McDonough, Dept of Geology, University of Maryland, College Park, USA

We have done a series of experiments at 1.5 GPa to determine the stability of MSS relative to sulfide liquid in the presence of silicate melt, and to measure interphase partitioning of Re and the PGEs.  Experiments were performed on mixtures of basalt + Re-PGE-doped sulfide (approx. Fe36Ni12Cu1S52 atomic) + H2O encapsulated in graphite-lined Pt capsules.  The FeO content of the silicate melt was varied by addition of Fe2O3, which by imposing graphite and sulfide saturation, changes the relative sulfur fugacity through the heterogenous equilibrium: FeOsil + 1/2S2 = FeSsulf + 1/2O2.  Experiments have been performed at 1200-1250 °C with samples containing silicate melt with FeO contents of ~5, 10 and 18 wt%.  Samples with 10 wt% FeO produce MSS + sulfide liquid at 1200 °C (see Figure 1), and sulfide liquid at 1250 °C, whereas those with 18 wt% FeO produced sulfide liquid-only over the same T range.  An experiment at 1250 °C containing silicate melt with 5 wt% FeO yielded MSS-only as the stable sulfide.   MSS-sulfide melt partitioning was measured for the experiment done at 1200 °C, and sulfide melt-silicate melt partitioning for the experiment done at 1250 °C, each with 10 wt% FeO in the silicate melt.  The PGE and Re content of coexisting phases was determined by LA-ICPMS with calibration against an in-house sulfide standard.  MSS/sulfide melt partition coefficients are: Os 1.8(±0.5);  Re 1.5(±0.5); Ir 0.9(±0.3); Pt 0.05(±0.03); Pd 0.06(±0.02).  Sulfide-silicate partition coefficients for Re and Os are ~20,000 and >100,000, respectively, with the latter a minimum value, as Os in the coexisting silicate was undetectable (< 10 ppb).  Relative MSS-sulfide melt partition coefficients are consistent with previous work although absolute values are somewhat lower, which may be the result of the higher temperature of our experiments.  The relative MSS-sulfide liquid partitioning of the PGEs is consistent with their fractionation in mantle-derived basalts, supporting the original hypothesis of Bockrath et al (1).  Re and Os are not fractionated by MSS-sulfide melt partitioning, although mantle-derived magmas have significantly higher Re/Os then their source.  However, the 5-fold lower value of DRe relative to DOs for sulfide-silicate partitioning accounts for this behaviour. 

Figure 1.  Backscattered electron image of a portion of an experiment run at 1200oC showing coexisting MSS + sulfide liquid (quenched silicate glass is dark material surrounding sulfide).

Another application of these results is to the origin of the compositional distinction between “enclosed” and “interstitial” sulfides in mantle samples.  Owing to their relatively high I-PGE content, “enclosed” sulfides have been suggested to represent residual MSS, wheras relatively P-PGE-rich “interstitial” sulfide as the conjugate sulfide melt.  Our data indicate that although the sense of fractionation is correct, the conjugate sulfide liquid should show significant absolute enrichments in Pt, Pd abundances, relative to MSS, and similar Ir and Os, which is not observed.

References 1. C. Bockrath, C. Ballhaus, C., A. Holzheid, Science 305, 1953 (2004).


Presented at the 2005 AGU Fall Meeting San Francisco, USA

Fractionation of Highly Siderophile Elements (HSEs) by Sulfide-Silicate Partitioning: A New Spin
AU: *
Brenan, J M

EM: brenan@geology.utoronto.ca
AF: University of Toronto, Dept of Geology 22 Russell St, Toronto, ON M5S 3B1 Canada

AU: McDonough, W F
EM: mcdonoug@geol.umd.edu
AF: University of Maryland, Dept of Geology, College Park, MD 20742-4211 United States

AB: Existing sulfide-silicate partitioning data show that the HSEs are strongly sequestered by molten sulfide during mantle melting. However, partition coefficients for specific elements are either highly variable, or non-existent, so the specific role of sulfide-silicate partitioning in fractionating the HSEs is poorly known. Previous studies have relied on bulk analytical techniques for HSE analysis, so precise measurement of individual partition coefficients has been hampered by extreme sulfide-silicate partitioning and the likely presence of trace sulfide in the silicate separate. We report on a new technique to measure sulfide-silicate partitioning that involves subjecting sulfide-silicate mixtures to high acceleration at superliquidus conditions to promote phase separation. Time-resolved laser ablation ICP-MS analysis is used to document phase homogeneity and establish elemental concentrations. HSE-doped (approx. 500 ppm each) samples of Fe sulfide + synthetic basalt are contained in olivine crucibles. The crucible and sulfide buffer (either Pt-PtS or Ru-RuS2) are vacuum-sealed in a fused quartz ampoule and equilibated under static conditions at 1200 C for about 48 hours. Following the static soak, the sample is quenched, sealed into a fresh ampoule, and subject to about 400g, at 1200 C, using a modified large capacity centrifuge. By buffering the sulfur fugacity, the oxygen fugacity can be estimated through the heterogeneous equilibrium: FeO(silicate) + S2 = FeS(sulfide) + O2. Sulfide-silicate partition coefficients at the Pt-PtS/FMQ buffers are greater than 10,000 for the PGEs and 230 for Re. At Pt-PtS/FMQ-1, the measured values of Dsulfide/silicate are 1000-3000 for the PGEs and 700 for Re. Our data supports the model of Bockrath et al (2004), requiring residual mss (mono-sulfide solid solution) to produce the observed PGE fractionation in mantle-derived magmas. The observed Re-Os fractionation during crust-mantle differentation is adequately modeled with our new partitioning data, obviating the need for control by residual alloy, silicate or oxide phases.