Siderophile (metal-seeking) element concentrations in the earths's upper mantle are indicators of how the earth accreted and differentiated. A long-standing problem in geology is that, although the earth has a large metallic core, mantle siderophile elements appear to be in much higher concentrations than can be explained by simple metal-silicate equilibrium. Murthy suggests that the abundances of siderophile elements in the earth's mantle can be explained by equilibration at higher temperatures than those achieved in most laboratory experiments. Murthy finds that metal/silicate partition coefficients (D), measured in the laboratory at temperatures of 1200 [degrees] to 1600 [degrees] C, can account for high abundances of mantle siderophile elements if these coefficients are extrapolated to higher temperatures, 3000 to 4000 K. This method of extrapolation allows one to predict mantle siderophile element abundances that agree reasonably well with observed abundances, which argues for an extensive magma ocean early in the history of the earth. However, we believe that the extrapolation technique used by Murthy is incorrect.
First, Murthy's method of extrapolation is made without reference to a governing chemical reaction. Such reference is important when dealing with metal/silicate partitioning because of the change in valence that accompanies the transfer of the siderophile element from metal to silicate (Fig. 1). Murthy's extrapolation equates the chemical potential of the siderophile element in the metal to that in the silicate with no account taken of an oxidizing or reducing agent. This would require that the partitioning siderophile element does not undergo a change in valence during metal/silicate partitioning.
Second, Murthy's method implicitly assumes that the Gibbs free energy change for the partitioning reaction, [Delta][G.sup.0], is constant, independent of pressure and temperature. This assumption, however, is generally incorrect. One could argue that pressure effects may fortuitously cancel those of temperature and allow [Delta][G.sup.0] to remain constant. However, the proper calculation cannot be performed because speciations and partial molar volumes of the relevant siderophile elements in silicate systems are unknown.
Third, Murthy's method does not seem to agree with most experimental evidence. For the great majority of siderophile elements, higher temperatures promote siderophile behavior and the values of D should increase, not decrease, as Murthy calculates. Nickel and, possibility, cobalt are exceptions to this rule, presumably because their stoichiometry parallels that of FeO, and because iron-oxygen reactions control oxygen fugacity (f[O.sub.2]). However, the values of D for Ga and P (Fig. 1, A and B), W, Ge, and Mo increase with increasing temperature[3,4]. Generally, metal/silicate liquid partition coefficients tend to either increase or remain constant as temperature increases along paths of buffered oxygen fugacity, such as that of iron-wustite (IW). Because [Mathematical Expression Omitted] = [Delta] H/[RT.sup.2], the IW path is expected to flatten at higher temperatures. In addition there may be complications from the phase change of metallic iron, which melts between 1300 and 1600 [degrees] C. Even so, based on the available data, slopes are either positive or zero (Fig. 1, A and B).
Finally there is an example of a naturally occurring, extremely reduced...