Surface structure determination by SEXAFS: The reliability of bond lengths and coordination numbers from multi-shell analyses

Abstract

Any structural determination from measurements of surface extended X-ray absorption fine structure (SEXAFS) is dependent on deriving sufficient, reliable structural parameters from experimental spectra. The reliability with which a multi-shell analysis (based on computation of the EXAFS spectrum and comparison with experimental data) can produce bond lengths and coordination numbers has been tested. Furthermore, the limitations placed on such an analysis by superimposing typical experimental levels of random noise onto the input data set has been investigated, both in terms of the number of parameters which can be reproduced and their associated errors. Significance tests have also been applied to a multi-shell analysis of experimental SEXAFS measurements of chlorine adsorbed on silver {110}. These tests on both real and simulated data show that a number of (reproducible) bond lengths may be expected from a multi-shell analysis and with a sufficient accuracy to distinguish between shells separated by as little as ≈ 0.3 A ̊ , for a maximum k of 8.7 Å −1. The effective coordination number of the nearest-neighbour shell can be reproduced within an error limit of ±0.2. Large correlations between Debye-Waller factors, coordination numbers and inelastic terms in the EXAFS amplitude place a high degree of inaccuracy, in some cases as high as ±3, on the values obtained for the effective coordination number of higher shells, which cannot by themselves be used to infer the adsorption site geometry of an adatom.