Leverage analysis allows detection of the reflections that have the greatest influence on the estimate of a refined variable. We have applied leverage analysis in the structure refinement of some rock-forming minerals, in the attempt of settling the best procedures to obtain accurate and reliable results. Attention has been focused on those variables (e.g., the refined site-scatterings) which are fundamental to the determination of site populations, which in turn are the basis for the study of order-disorder processes and petrogenetic modeling. A garnet (pyrope) and an amphibole (tremolite) were used to test the procedure, being illustrative of different symmetries and structural complexity. As the omission of reflections with high leverage on a particular variable may strongly decrease the accuracy in its estimate, the results indicate that systematic procedures for data truncation [either on the basis of (low or high) sin q/l, high DF/sFo or of low I/sI] are potentially dangerous. The choice of a correctly ionized model is shown to be critical to obtain accurate estimates of the refined site-scatterings. The set of reflections with the highest leverage value with respect to a peculiar group of variables was found to be nearly invariant within isomorphous solid-solutions; this implies that the results of this work are valid throughout the amphibole and garnet compositional spaces. The warnings proposed to the treatment and refinement of X-ray diffracted data may be of general utility in the structure refinement of minerals.
Leverage analysis and structure refinement of minerals
2000
Abstract
Leverage analysis allows detection of the reflections that have the greatest influence on the estimate of a refined variable. We have applied leverage analysis in the structure refinement of some rock-forming minerals, in the attempt of settling the best procedures to obtain accurate and reliable results. Attention has been focused on those variables (e.g., the refined site-scatterings) which are fundamental to the determination of site populations, which in turn are the basis for the study of order-disorder processes and petrogenetic modeling. A garnet (pyrope) and an amphibole (tremolite) were used to test the procedure, being illustrative of different symmetries and structural complexity. As the omission of reflections with high leverage on a particular variable may strongly decrease the accuracy in its estimate, the results indicate that systematic procedures for data truncation [either on the basis of (low or high) sin q/l, high DF/sFo or of low I/sI] are potentially dangerous. The choice of a correctly ionized model is shown to be critical to obtain accurate estimates of the refined site-scatterings. The set of reflections with the highest leverage value with respect to a peculiar group of variables was found to be nearly invariant within isomorphous solid-solutions; this implies that the results of this work are valid throughout the amphibole and garnet compositional spaces. The warnings proposed to the treatment and refinement of X-ray diffracted data may be of general utility in the structure refinement of minerals.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.


