Single-scattering calculations of extended x-ray-absorption fine-structure (EXAFS) amplitude and phase were compared with the experimental first-shell data for copper, rhodium, and platinum metal. Theoretical standards used were the tables of Teo and Lee, the tables of McKale et al., and the codes excurv90, mufpot, and feff. The quality of the experimental data was shown to be very high. The experimental first-shell data were obtained by Fourier filtering. The errors introduced in the separation of the first shell from the complete EXAFS spectrum by Fourier filtering were negligible as shown in a model study. The comparison of experimental and theoretical data shows that the accuracy of theoretical standards depends mainly on the treatment of the exchange potential and the energy-dependent losses. The most accurate description of the exchange potential is the energy-dependent Hedin-Lundqvist potential with an energy-dependent self-energy, as used in feff. The use of ground state Xα or energy-independent exchange, as in the tables of McKale et al. or the codes excurv90 and mufpot, is found to be inadequate and leads to large phase and amplitude errors. Addition of an energy-dependent mean free path to the tables of McKale et al. improved the accuracy by 15–25 %. The physical reasons for the differences in accuracy of the theoretical methods examined are discussed.

• Received 25 February 1993

DOI:https://doi.org/10.1103/PhysRevB.50.7872