Structure analysis by normalized data. I. Increase of relative accuracy of diffraction data

A new method for obtaining physically reliable values of atomic parameters in the refinement of the structure model by the least squares method has been suggested. The method requires the use of experimental data reduced to a scale corresponding to the Bragg scattering from an ideal defect-free crystal. The reduced experimental data can be obtained by interexperimental-minimization (IEM) method—a new algorithm minimizing the difference “experiment 1-experiment 2” (between the data obtained in two experiments, 1 and 2) used in addition to the model-experiment minimization performed in the conventional least square procedure. The use of four different experimental data sets for alexandrite crystals, Al2BeO4: Cr3+, allowed us to show that the application of the IEM algorithm considerably increases the reproducibility level of the experimental data, which, in turn, increases by 30–60% the agreement between the model and the experiment.