An analysis of existing data and models for point defects in pure (Fe,Mg)-olivine showed that it was necessary to consider thermodynamic non-ideality of mixing in order to describe adequately the concentrations of point defects over the range of measurement. In spite of the various sources of uncertainty, the concentrations of vacancies in octahedral sites in (Fe,Mg)-olivine were of the order of 10−4 per atomic formula unit at 1000 to 1200C, according to 2 studies. The first explicit plots of vacancy concentrations in olivine were presented as a function of temperature and O fugacity, according to the 2 models. It was found that, in contrast to the absolute concentrations at 1100C and the dependence upon fO2, there was considerable uncertainty in the knowledge of the temperature-dependence of the vacancy concentrations. This had to be considered when discussing transport properties such as diffusion coefficients. These defect models in pure (Fe,Mg)-olivine had to be extended by considering aliovalent impurities such as Al and Cr in order to describe the behavior of natural olivine. Such a formulation was used to analyze the large database of diffusion coefficients in olivine. The analysis revealed, for the first time, a change of diffusion mechanism in a silicate mineral from a transition-metal extrinsic domain (TaMED) to a purely extrinsic domain (PED), at fO2 values below 10−10Pa and at temperatures below 900C. The change in diffusion mechanism manifested itself as a change in the fO2-dependence of the diffusivity and as a slight change in the activation energy for diffusion. The activation energy increased at lower temperatures. Defect formation enthalpies in the TaMED regime (distinct from intrinsic defect formation) lay between −66 and 15kJ/mol. The migration energies of octahedral cations were probably 260kJ/mol. Plots were presented for diffusion at various constant fO2 as well as along fO2 buffers, in order to highlight the difference in behavior between them. Considering all the diffusion data and constraints from the point defect models, (Fe–Mg) diffusion in olivine along [001] was best described by, at O fugacities greater than 10−10Pa,

log[DFeMg(m2/s)] = -9.21–[201000+(P-105)(7 x 10-6)]/2.303RT+log[fO2/10-7]/6+3XFe

where P and fO2 were in Pa and XFe was the mole fraction of the fayalite component. At O fugacities less than 10−10Pa:

log[DFeMg(m2/s)] = -8.91–[220000+(P-105)(7 x 10-6)]/2.303RT+3XFe

These equations reproduced all of the 113 experimental data points to within half an order of magnitude. Alternatively, a global equation which averaged out the change in mechanism could be used; but with somewhat larger errors in reproducing the measured diffusion data. It generally underestimated the data at higher temperatures, and over-estimated them at lower temperatures. The fO2 was not explicitly considered here, leading to additional sources of error:

log[DFeMg(m2/s)] = -8.27–[226000+(P-105)(7 x 10-6)]/2.303RT+3XFe

In order to obtain diffusion coefficients along [100] and [010], log[6] had to be subtracted from each of the above equations.

Fe–Mg Diffusion in Olivine II - Point Defect Chemistry, Change of Diffusion Mechanisms and a Model for Calculation of Diffusion Coefficients in Natural Olivine. R.Dohmen, S.Chakraborty: Physics and Chemistry of Minerals, 2007, 34[6], 409-30