Experimental Determination of U and Th Partitioning Between Clinopyroxene, Garnet, Olivine, and Natural and Synthetic Silicate Melt

Citation

LaTourrette, Thomas (1993) Experimental Determination of U and Th Partitioning Between Clinopyroxene, Garnet, Olivine, and Natural and Synthetic Silicate Melt. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/4ww8-x445. https://resolver.caltech.edu/CaltechTHESIS:01072013-160430261

Abstract

The distribution of U and Th between crystals and quenched silicate melt has been measured in experimentally crystallized natural and synthetic starting compositions. Clinopyroxene was crystallized from two natural basalts and a synthetic composition on the diopside-anorthite join, olivine from one of the natural basalts, and garnet from a synthetic andesite. Clinopyroxene and olivine were crystallized at atmospheric pressure under controlled oxygen fugacity and U and Th distributions were determined by particle track radiography. Garnet crystals were grown at 27 Kbars and actinide distributions were determined by secondary ion mass spectrometry. Crystals were grown by slow cooling in an effort to maintain chemical equilibrium at the crystal-melt interface.

Clinopyroxene- and olivine-melt partition coefficients (D_i^xtl/liq = C_i^xtl/C_i^liq) were calculated from track distributions by correcting for chemical zoning in the crystals (a result of fractional crystallization) and particle range differences between the crystals and glass. Over the range of ƒO₂s and compositions studied, D_U^cpx/liq = 0.0021 - 0.0093, = D_Th^cpx/liq = 0.0047- 0.021, and D_U;Th^ol/liq < 0.00005. Garnet-melt partition coefficients are D_U^gt/liq = 0.0013 and D_Th^gt/liq = 0.0122. In order to study the effect of the U valence state distribution on U-Th fractionation, clinopyroxene crystallization experiments were run at oxygen fugacities corresponding to the Ni-NiO (NNO), Fa-Mt-Qz (FMQ) and 1 log unit more oxidizing than Fe-FeO (IW+ 1) oxygen buffers. All compositions show an increase in D_U^cpx/liq with decreasing ƒO₂, presumably from the increasing proportion of U⁴⁺ in the melt. D_Th^cpx/liq displays little variation with ƒO₂, consistent with the fact that Th is solely tetravalent D_Th^cpx/liq does show an apparent small decrease with decreasing ƒO₂ in one composition, and this is interpreted to be due to compositional changes stemming from variable amounts Na-loss with ƒO₂. A limited compositional analysis indicates that actinides may be incorporated into clinopyroxene by means of a (U,Th)⁴⁺ + 2Na⁺,⇔ 3Ca²⁺ coupled substitution in the M2 site. This mechanism predicts a positive dependence of the actinide partition coefficients on (D_Ca^cpx/liq)³ / (D_Na^cpx/liq)².

While there is evidence for deviations from interfacial equilibrium during crystal growth, D_Th^xtl/liq and D_U^xtl/liq are affected similarly. This results in D_Th^xtl/liq/D_U^xtl/liq, which gives a measure of the degree of U-Th fractionation possible by crystal-melt partitioning, being insensitive to these deviations. D_Th^cpx/liq/D_U^cpx/liq is well behaved, approximately independent of composition, and decreases by a factor of 2 - 3 as the ƒO₂ decreases from NNO to IW+1. This indicates that U-Th fractionation by crystal-melt partitioning is ƒO₂ dependent.

These results indicate that U-Th fractionation by clinopyroxene-melt partitioning during partial melting will result in a melt with Th/U less than the clinopyroxene, U-Th fractionation by olivine will be insignificant for physically realistic melt fractions, and U-Th fractionation by garnet will result in a melt with ThiU greater than the garnet. The observed pattern of ²³⁸U-²³⁰Th disequilibrium in oceanic basalts (MORB and OIB) requires a net UTh fractionation that is large, ubiquitous, and results in (²³⁸U-²³⁰Th) < 1. With these constraints, fractionation by partial melting of spinel lherzolite is ruled out. The sense of U-Th fractionation by garnet is consistent with oceanic basalts, but the magnitude of fractionation is much smaller than the entire range in observed fractionations. Since partial melts of neither spinel nor garnet lherzolite can match the sense and magnitude of ²³⁸U-²³⁰Th disequilibrium in MORB and OIB, partial melting is considered unimportant in generating ²³⁸U-²³⁰Th disequilibrium in oceanic basalts. ²³⁸U-²³⁰Th disequilibrium in MORB and OIB must therefore result from some process other than partial melting, and possible alternative mechanisms are presented.

Thesis Files