Effects of silicate liquid composition on mineral‐liquid element partitioning from Soret diffusion studies

Abstract

Thermal (Soret) diffusion of major, minor, and trace components of naturally occurring silicate melts ranging in composition from tholeiitic basalt to high-silica rhyolite is examined at 1 GPa, mean temperatures of 1380ø-1535øC, and temperature gradients of 50ø-80øC mm-x. Soret diffusion in silicate melts is channeled along network former/network modifier compositional vectors indistinguishable from those observed in silicate liquid immiscibilty studies, but in the former case, gradients are continuous between the compositional (and thermal) extremes. Si, the chief network former, invariably segregates to the hot end of the temperature gradient as do Rb, K, and Na in melts of mafic to intermediate bulk composition. and Ti always show separations antithetic to Si, while Rb, K, and Na fractionate from Si only in silicic melts. Li and A1 display little separation in a temperature gradient. Soret separations of alkali and alkaline earth cations are consistent with their participation in the network as charge compensators for trivalent network formers (i.e., A13+, etc.). These network formers as well as network modifiers show fractionations from Si that depend directly on cation field strength and size, and the polymerization state of the melt. Temperature and absolute concentration have no measurable effect on the fractionation behavior of these components relative to that of SiO 2, the principal determinant of melt structure. The melt compositional dependence of element partitioning is characterized using these data. Twofold to fourfold changes in mineral-liquid partition coefficients for minor and trace elements are expected over the melting range for typical basalts, and similar or greater variabilty is expected during differentiation of silicic magmas. These expectations corroborate reported variability of mineral-liquid partition coefficients from studies of natural igneous rock samples and experimental phase equilibria. INTRODUCTION Basic to our understanding of the behavior of trace elements during magmatic differentiation is a knowledge of the way in which elements distribute between fractionating phases and the variation in element partitioning with changes of temperature, pressure, and phase composition. A quantitative expression for the distribution of an element between two coexisting phases, A and B, is most conveniently represented by the Nernst partition coefficient, defined as the ratio of the element's concentration in respective phases (i.e., Di = Xixtl/xiliq ). During the last decade a number of experimental studies designed to characterize mineral-liquid element partitioning (for example, Watson [1977], Ryerson and Hess [1978], Hart and Davis [1978], Leeman [1978], Takahashi [1978], and Mysen and Virgo [1980], among others) have shown that one of the most important factors influencing the distribution of trace and minor elements between phenocrysts and silicate liquid is the bulk composition and hence structure of the melt phase. Although the actual magnitude of crystal-liquid partition coefficien...

The effect of sodium and titanium on crystal-melt partitioning of trace elements

“…2-liquid partitioning studies and experiments using crystals of fixed composition have shown that melt composition and structure exert an influence over trace element partitioning (e.g., Watson, 1976;Ryerson and Hess, 1978;Lesher, 1986;Kohn and Schofield, 1994). Melt composition may affect trace element partitioning in two ways, determining the degree of melt polymerization and the availability of trace metal complexing agents (Watson, 1976).…”

Microgranitoid enclaves from the Swifts Creek Pluton SE-Australia: textural and physical constraints on the nature of magma mingling processes in the plutonic environment

“…In fact, the chilled margins show a 2-3 fold increase in KeO and a slight decrease in CaO and NaeO relative to the centers of chilled enclaves (EIdERZ, 1987). If this enrichment occurred while both, enclave and host were still partially liquid, probably by diffusion processes (Lt~:SHER, 1986;LESHER et al, 1982;WATSON & JUREWICZ, 1984) or volatile transfer, the reversal of the normal crystallization sequence could be readily explained. The third variant of the chilled enclaves has apparently undergone subsolidus recrystallization.…”

Partitioning of trace elements between rutile and silicate melts: Implications for subduction zones

“…As a number of papers have identified melt composition as a significant factor (Watson, 1976;Ryerson and Hess, 1978;Lesher, 1986;Nielsen et al, 1992;Kohn and Schofield, 1994;Keppler, 1997, 2002;Horng and Hess, 2000;Kushiro and Mysen, 2002;O'Neill and Eggins, 2002;Klemme and Dalpé, 2003;Klemme and Meyer, 2003;Prowatke and Klemme, 2005;Schmidt et al, 2004), we designed four additional experiments along a compositional vector that allowed us to study the effect of melt composition on the partitioning. As the major element composition of rutile is virtually constant in the experiments, we can exclude a significant role of the mineral composition on trace element partition coefficients and examine the effect of melt composition on the partitioning of trace elements.…”

“…It has been proposed that melt composition is a significant factor controlling partition coefficients (Watson, 1976;Ryerson and Hess, 1978;Lesher, 1986;Nielsen et al, 1992;Kohn and Schofield, 1994;Horng and Hess, 2000;Kushiro and Mysen, 2002;Linnen and Keppler, 2002;O'Neill and Eggins, 2002;Klemme and Dalpé, 2003;Prowatke and Klemme, 2005;Schmidt et al, 2004). A recent study that investigated rutile/ melt partition coefficients in haplogranitic compositions found strongly increasing D Nb and D Ta with decreasing K 2 O/(K 2 O ϩ Al 2 O 3 ) of coexisting melts.…”