Exploration of quantitative kinetic models for the evaporation of silicate melts in vacuum and in hydrogen

Two basic approaches (pure component reference (PCR) and equilibrium reference (EQR)) to modeling silicate melt evaporation are explored. The PCR model calculates the maximum possible evaporation rates of the pure oxides from their equilibrium vapor pressures and rescales these rates according to the activities of the oxides in the silicate melts and the melt densities. The EQR model calculates the maximum possible evaporation rates based on the equilibrium vapor pressures of the melts. Differences between the calculated and experimentally determined evaporation rates are accounted for with evaporation (alpha (evap)) coefficients that are only dependent on temperature. Two versions of the PCR model, Cases 1 and 2, are explored to try to resolve apparently contradictory conclusions about the composition of the evaporating species based on Mg and Si isotope fractionation during evaporation (species are not in thermodynamic equilibrium proportions) and direct measurements of gas species in Langmuir experiments (species are in roughly equilibrium proportions). The Case 2 and EQR models cannot explain the observed isotope fractionations unless evaporation occurred under non-Rayleigh conditions, either because there was significant recondensation during the experiments or because diffusion was playing a limiting role. Whether or not the role of diffusion is included, the PCR and EQR models are able to reproduce the elemental results of evaporation experiments of "chondritic" melts from temperatures of 1700 to 2000 degreesC, and up to mass losses of about 95%. However, the models underestimate absolute evaporation rates in very Ca- and Al-rich melts. This may reflect errors in the model used to estimate oxide activities. The EQR model can only reproduce the observed evaporation behavior of Na if, unlike the other oxides, its alpha (evap) coefficient is close to unity. Based on available diffusion data, diffusion is not slow enough in "chondritic" or forsteritic melts to explain the isotopic fractionations of Mg and O in the evaporation experiments, but it may play a role in limiting Si isotope fractionation. Provided recondensation was not a significant factor in the experiments, at present PCR Case 1 appears to be the best model if both the Langmuir and the isotopic fractionation experiments are to be explained.