Abstract. The influence of hydrogen intercalation on the local structure of rhenium trioxide is studied in-situ by the Re L3-edge EXAFS spectroscopy and analysed using a novel approach, based on the use of evolutionary algorithm and wavelet transform. The proposed method allows us to perform accurate EXAFS analysis within the multiple-scattering approach taking into account contributions from outer coordination shells and to access the information on correlations in atomic thermal motion.
IntroductionThe analysis of extended X-ray absorption fine structure (EXAFS) data from the first coordination shell around the absorbing atom is nowadays a common tool to study the local structure of materials [1]. At the same time, EXAFS spectra, acquired at modern brilliant Xray sources, often contain much more information, especially for crystalline materials, where a contribution of distant coordination shells (up to 10Å and more) is frequently present in the total EXAFS spectrum. The precise analysis of EXAFS spectra beyond the first coordination shell using conventional methods is, however, often impossible since the contributions of outer shells overlap, and the total number of parameters, required to describe the local structure, growths exponentially with the increase of the number of coordination shells, included in the analysis [2]. Moreover, conventional EXAFS analysis even for the first coordination shell can be imprecise for significantly distorted or disordered systems, where the distributions of interatomic distances are often far from the Gaussian one [3].This problem may be treated by applying simulation-based techniques as molecular dynamics (MD) [3,4] and reverse Monte Carlo (RMC) method [5]. For our purposes the latter is especially interesting, since the RMC method does not require the knowledge and calculation of interatomic forces and relies solely on the available experimental EXAFS data. In RMC approach a 3D structure model of the material is constructed using an iterative random process, aimed to minimize the difference between experimental EXAFS and configuration-averaged EXAFS spectra, calculated for the given structure model. Once a good agreement between simulated and experimental spectra is obtained, one may expect that also the constructed model provides a good approximation of the material structure. The existing RMC implementations for EXAFS analysis, however, usually neglect the influence of multiple-scattering (MS) effects or treat them in an approximate way due to concomitant significant computational costs. As a consequence, the information amount, obtainable from EXAFS spectrum, again, is rather limited.