No abstract

No abstract

Theoretical X-ray absorption fine structure (XAFS) standards are developed for arbitrary pairs of atoms throughout the periodic table (Z < 94). These standard XAFS spectra are obtained from ab initio single-scattering XAFS calculations, using an automated code, feff, which takes into account the most important features in current theories: (i) an exact treatment of curved-wave effects; (ii) approximate molecular potentials derived from relativistic atoms, (iii) a complex, energy-dependent self-energy; (iv) a well defined energy reference, feff also yields tables of XAFS phases and amplitudes as well as mean-free paths. Sample results are presented and compared with experimental results and with earlier work. We find that these theoretical standards are competitive with experimental standards, permitting XAFS analysis at lower wavenumbers and yielding distance determinations typically better than 0.02 Á and coordination numbers typically better than 20%. These standards also provide theoretical tests of chemical transferability in XAFS.

High-order scattering is found to be essential for the convergence of the multiple-scattering (MS) theory of x-ray-absorption fine structure, both in the near-edge and the extended regimes. These contributions are calculated using an ah initio curved-wave scattering-matrix formalism. Convergence to full MS accuracy is demonstrated for fee Cu, as well as for molecular O2 and N2, where our approach provides a high-order MS interpretation of the a* shape resonances. PACS numbers: 78.70.Dm, Curved-wave multiple-scattering (MS) theory [1] provides a unified theory for x-ray-absorption fine structure (XAFS) that encompasses both the extended (EXAFS) and near-edge (NEXAFS) regimes. When carried to all orders, this theory is equivalent to exact treatments based on wave functions and Hamiltonian diagonalizations. However, because present computational methods have limited applicability (such as to low-order MS [2], full MS at low energies only [3], or MS in small clusters [4]), there has been considerable speculation and controversy [1-4] about the nature and extent of MS in XAFS (e.g., on the need for full MS in NEXAFS or the importance of nonshadowing MS in EXAFS). In this Letter we introduce the first unified high-order MS treatment of XAFS with sufficient speed and accuracy to treat extended systems at both low and high energies. We find that neither fuU-MS nor low-order-MS theories are fully satisfactory; low-order theories generally contain too little MS, while much of that in fuU-MS theories is smeared out by inelastic losses and thermal disorder. We show this using high-order MS calculations for fee Cu and for molecular O2 and N2. We find that the MS expansion for Cu with up to 7 scatterers and about 10^ paths converges to broadened band-structure results, while calculations with about 10^ paths yield agreement with experiment out to 8.5 A. Similarly, the MS expansion for O2 and N2 with up to 13 backscatterings converges to full MS Xa-scattered-wave (Xa-SW) calculations. Our approach also leads to a MS interpretation of the a* shape resonances [5] that are observed in NEXAFS and greatly simplifies their calculation. We demonstrate that such resonances are special XAFS peaks, whose asymmetry and location result from coherent high-order MS.Our XAFS calculations are based on an automated implementation of (i) the curved-wave MS formalism of Rehr and Albers (RA) [6], (ii) an efficient method for enumerating MS paths, (iii) ah initio scattering potentials and phase shifts that include inelastic losses [7], and (iv) MS Debye-Waller factors. We calculate the normalized XAFS X = (A* -/xo)//io, where /x is the x-ray-absorption coefficient and /io the smooth atomiclike background.The absorption ji is proportional to the projected photoelectron density of states or, equivalently, to the imaginary part of a Green's-function matrix element. More precisely, the MS expansion for the polarization-averaged Jf-shell XAFS is given by a sum over all scattering paths r, X = T.rlm{e^^^^i:m{'^m\GtN"'Gt2GtiG\lm)) (in matrix notation). He...