Partitioning behavior of trace elements between dacitic melt and plagioclase, orthopyroxene, and clinopyroxene based on laser ablation ICPMS analysis of silicate melt inclusions

Partitioning behavior of Sc, Ti, V, Mn, Sr, Y, Zr, Nb, Ba, La, Nd, Sm, Eu, Gd, Dy, Ho, Yb, Hf, and Pb between dacitic silicate melt and clinopyroxene, orthopyroxene, and plagioclase has been determined based on laser ablation-inductively coupled plasma mass spectrometric (LA-ICPMS) analysis of melt inclusions and the immediately adjacent host mineral. Samples from the 1988 eruption of White Island, New Zealand were selected because petrographic evidence suggests that all three mineral phases are in equilibrium with each other and with the melt inclusions. All three phenocryst types are found as mineral inclusions within each of the other phases, and mineral inclusions often coexist with melt inclusions in growth-zone assemblages. Compositions of melt inclusions do not vary between the different host minerals, suggesting that boundary layer processes did not affect compositions of melt inclusions and that post-trapping modifications have not occurred. Partition coefficients were calculated from the host and melt inclusion compositions and results were compared to published values. All trace elements examined in this study except Sr are incompatible in plagioclase, and all measured trace elements except for Mn are incompatible in orthopyroxene. In clinopyroxene, Sc, V, and Mn are compatible, and Y, Ti, HREE, and the MREE are only slightly incompatible. Most partition coefficients overlap the wide range of values reported in the literature, but the White Island data are consistently at the lower end of the range in published values. Results from the literature obtained using modern microanalytical techniques such as secondary ion mass spectrometry (SIMS) or proton induced X-ray emission spectroscopy (PIXE) also fall at the lower end of the published values, whereas partition coefficients determined from bulk analysis of glass and crystals separated from volcanic rocks typically extend to higher values. Rapid crystal growth-rates, crystal zonation, or the presence of accessory mineral inclusions in phenocrysts likely accounts for the wide range and generally higher partition coefficients obtained using bulk sampling techniques. The results for 3+ cations from this study are consistent with theoretical predictions based on a lattice strain model for site occupancy. The results also confirm that the melt inclusion-mineral (MIM) technique is a reliable method for determining partition coefficients, as long as the melt inclusions have not experienced post-entrapment reequilibration