Abstract:
Pavese et al. (1999) examined cation partitioning vs. temperature in a synthetic spinel of composition (Mg0.70 Fe0.233+) Al1.97 O4 using structure data obtained from in situ neutron powder diffraction. After imposing assumptions on the site assignment of vacancies and Fe3+ they assigned the remaining cations by applying least-squares minimization to chemical constraints on site-occupancy sums, site-scattering, chemical composition, and thermal expansion of the octahedral site. Their proposed site assignments exhibit a sharp discontinuity in occupancy fractions versus temperature, a necessary consequence of their assumptions on vacancy assignments. In this paper we reexamine the cation partitioning of the same spinel using the constrained least-squares formulation of OccQP (Wright et al. 2000), which optimizes site occupancies without ad hoc assumptions. We obtain strikingly different results, supporting the general view that spinel undergoes a lambda transition at ~1000 K. For all observed parameters, the residuals obtained with the OccQP assignments are lower than those obtained with the Pavese et al. assignments, in some cases by more than 1 order of magnitude.