Improvement of the ab initio embedded cluster method for luminescence properties of doped materials by taking into account impurity induced distortions: The example of Y[sub 2]O[sub 3]:Bi[sup 3+]

New ab initio embedded-cluster calculations devoted to simulating the electronic spectroscopy of Bi(3+) impurities in Y(2)O(3) sesquioxide for substitutions in either S(6) or C(2) cationic sites have been carried out taking special care of the quality of the environment. A considerable quantitative improvement with respect to previous studies [F. Real et al. J. Chem. Phys. 125, 174709 (2006); F. Real et al. J. Chem. Phys. 127, 104705 (2007)] is brought by using environments of the impurities obtained via supercell techniques that allow the whole (pseudo) crystal to relax (WCR geometries) instead of environments obtained from local relaxation of the first coordination shell only (FSR geometries) within the embedded cluster approach, as was done previously. In particular the uniform 0.4 eV discrepancy of absorption energies found previously with FSR environments disappears completely when the new WCR environments of the impurities are employed. Moreover emission energies and hence Stokes shifts are in much better agreement with experiment. These decisive improvements are mainly due to a lowering of the local point-group symmetry (S(6)--C(3) and C(2)--C(1)) when relaxing the geometry of the emitting (lowest) triplet state. This symmetry lowering was not observed in FSR embedded cluster relaxations because the crystal field of the embedding frozen at the genuine pure crystal positions seems to be a more important driving force than the interactions within the cluster, thus constraining the overall symmetry of the system. Variations of the doping rate are found to have negligible influence on the spectra. In conclusion, the use of WCR environments may be crucial to render the structural distortions occurring in a doped crystal and it may help to significantly improve the embedded-cluster methodology to reach the quantitative accuracy necessary to interpret and predict luminescence properties of doped materials of this type.