The reliability of the structural parameters obtained from x-ray-absorption fine-structure (XAFS) spectra strongly depends on the quality of the standards used in the data analyses. We have compared about 30 isolated pair standards, extracted from experimental data, with those generated by the FEFF5 code. The absorbing atoms range from Ni (Z=28) to Pb (Z=82) and the backscattering atoms from O (Z=8) to Au (Z=79). In general, good agreement is achieved, with typical errors of 0.005 Angstrom for nearest-neighbor distances R. Significant differences are sometimes found in the backscattering amplitude F(k, R) at low values of the photoelectron wave vector k (< 7 Angstrom(-1)), especially for heavy backscattering atoms such as hu and Pt. We demonstrate that when there are significant differences between experimental and theoretical functions, F(k, R), an experimental standard, shifted from one atom pair to another, provides a better fit to the data than obtained using the theoretical standard and reduces the errors in amplitude and position, often by a factor of 2-3. The effective amplitude reduction factor S-o,eff(2) and the E(o) shift for the calculated XAFS standards are also given; the former can vary by +/- 20% across the Periodic Table. Estimated experimental errors for the amplitudes are < 10%, and similar to 5-6% in many cases. Some important fundamental issues, such as background removal, the influence of the energy resolution on XAFS, and correlations in the fitting parameters, are also addressed.
