材料科学
电子结构
分子物理学
超单元
电子
计算物理学
物理
原子轨道
电子能带结构
能量(信号处理)
电子密度
作者
Wai-Yim Ching,Paul Rulis
标识
DOI:10.1088/0953-8984/21/10/104202
摘要
Over the last eight years, a large number of x-ray absorption near edge structure (XANES) and/or electron energy loss near edge structure (ELNES) spectroscopic calculations for complex oxides and nitrides have been performed using the supercell-OLCAO (orthogonalized linear combination of atomic orbitals) method, obtaining results in very good agreement with experiments. The method takes into account the core-hole effect and includes the dipole matrix elements calculated from ab initio wavefunctions. In this paper, we describe the method in considerable detail, emphasizing the special advantages of this method for large complex systems. Selected results are reviewed and several hitherto unpublished results are also presented. These include the Y K edge of Y ions segregated to the core of a Σ31 grain boundary in alumina, O K edges of water molecules, C K edges in different types of single walled carbon nanotubes, and the Co K edge in the cyanocobalamin (vitamin B(12)) molecule. On the basis of these results, it is argued that the interpretation of specific features of the calculated XANES/ELNES edges is not simple for complex material systems because of the delocalized nature of the conduction band states. The long-standing notion of the 'fingerprinting' technique for spectral interpretation of experimental data is not tenable. A better approach is to fully characterize the structure under study, using either crystalline data or accurate ab initio modeling. Comparison between calculated XANES/ELNES spectra and available measurements enables us to ascertain the validity of the modeled structure. For complex crystals or structures, it is necessary to use the weighted sum of the spectra from structurally nonequivalent sites for comparison with the measured data. Future application of the supercell-OLCAO method to complex biomolecular systems is also discussed.
科研通智能强力驱动
Strongly Powered by AbleSci AI