EFFECTS OF ALUMINUM ON CR DIMERIZATION IN SILICATE MELTS AND IMPLICATIONS FOR CR PARTITIONING AND REDOX EQUILIBRIA

Abstract

The electrochemical measurements reported here provide insight into the activities and structure of individual Cr and Sn species in the melt. This represents an improvement over activity coefficients averaged over multiple species such as can be derived from simple melt/mineral equilibrium experiments. Knowledge of the presence and activities of individual species of Cr in a melt can lead to better understanding and prediction of the complex variations in partitioning of Cr between minerals and melt, which have thus far defied accurate description.Electrochemical experiments in four different silicate melts indicate that, under important magmatic conditions, Cr can be present in the melt as any of three species: Cr2+, Cr3+, and some polymeric form which we represent in this paper as the dimer Cr26+. The presence of two trivalent Cr species, and the oxidation state change between Cr2+ and Cr3+, makes it easy to understand why the partitioning behavior of Cr is complex. The presence of two species (a dimer and a monomer), each containing Cr3+, means that the activity coefficient for a single composite Cr3+ component will vary in a complex way with melt composition, temperature, fO2, and Cr concentration even when mixing in the melt is ideal. For example, if a fixed Cr3+ component is used to model the melt, its activity coefficient will vary by an order of magnitude over the region in which the Cr3+ goes from being present as mostly monomer to mostly dimer. This occurs even if there were no nonideal mixing. Trivalent Cr dimers are particularly sensitive to Al concentration in the melt, consistent with the interpretation that dimers exist as spinel-like structures in the melt.The preceding discussion bears on melts with ideal mixing. But the experiments reported here indicate that mixing in the compositions studied is not ideal. Activity coefficients for the Cr3+ dimer vary by over a factor of 3 as a function of composition for the silicate melts reported here. In addition, we report variations of a factor of 3 in the activity coefficient for divalent Cr as a function of melt composition, meaning that Cr partition coefficients will be significantly sensitive to melt composition even where little trivalent Cr is present. Thus, the effects of nonideal mixing will be added to the effects of the presence of a dimer.The transition from mostly dimer to mostly monomer occurs at concentrations of Cr and Al typical of natural systems, with the dimer becoming less important as Cr3+ concentration decreases. At concentrations of Al of about 8 wt.%, the dimer is an important component down to about 0.25 wt.% Cr3+. At Al concentrations of 11-12 wt.%, the dimer was important to the limit of our experimental resolution, about 0.1 wt.% Cr3+. This means that in the range of natural melts, not only is the Cr not present as a single species, but the proportion that each of the species comprises of the total Cr3+ is not fixed. Because the trivalent Cr exists as more than a single species in the melt, and the proportion of those species is not fixed, use of a single species model to predict variations in activity will work only poorly and require complex variations in activity coefficients. Trivalent Cr activity coefficients, relative to typically-used fixed component models, can be expected to vary as a function not only of melt composition, but of fO2 and total Cr concentration as well. Based on the results presented here, variations in activity of trivalent Cr of more than 100% can be expected over natural ranges of conditions.These results have broader implications for how we view thermodynamic components in magmatic systems. In a thermodynamic treatment of the behavior of an element in a melt, any form of that element can serve as the species of description. However, choosing the principle form of the element actually present in the melt usually minimizes variations of the activity coefficient and improves intuitive understanding of the behavior of the element in the system. What's more, if an element exists in multiple species, no single species will provide a simple or intuitive model for the variation of activity of the element as a whole. Use of a simple ionic model for melts, although successful in some general applications, is not correct in detail as shown by the dimerization of Cr in silicate melts. Other elements may exhibit similar speciation in melts. We demonstrate in this report that Sn also exists as a dimer in silicate melts. At all concentrations of Sn which we studied, the Sn dimer (which we represent as Sn24+) dominated over any monomer present. However, the concentrations we studied were far above natural Sn concentrations, and Sn may exhibit similar behavior to Cr at more natural concentrations of Sn where monomerization would be more likely to occur.