Three independent techniques are used to separate fine structure from the absorption spectra, the background function in which is approximated by (i) smoothing spline. We propose a new reliable criterion for determination of smoothing parameter and the method for raising of stability with respect to kmin variation; (ii) interpolation spline with the varied knots; (iii) the line obtained from bayesian smoothing. This methods considers various prior information and includes a natural way to determine the errors of XAFS extraction. Particular attention has been given to the estimation of uncertainties in XAFS data. Experimental noise is shown to be essentially smaller than the errors of the background approximation, and it is the latter that determines the variances of structural parameters in subsequent fitting. 61.10.Ht

No abstract

X-ray absorption spectra of the oxide systems Ba1 - x Kx BiO3 and BaPbO3 above the Bi and Pb L3 absorption edges were investigated. It was shown that oxygen ions move in a double-well potential and their oscillations are correlated with the charge-carrier movement. The observed breathing-like oxygen vibration in the double-well potential with large amplitude and low frequency causes the strong electron-phonon coupling and high Tc -values for doped BaBiO3 . Based on the experimental data, a model of the relationship of the electronic and local crystal structures is proposed that is in good agreement with the results from transport measurements, and inelastic neutron and electron scattering, Raman scattering, and photoemission spectroscopy. In the framework of the model, the possible reasons for the superconductivity in perovskite-like oxides are discussed.

The problem of error analysis is considered taking into account all possible correlations and a prior information about the accessible parameter space. Special attention is paid to the correct determination of the relative weight of experimental data and the a priori guess. The applications of statistical χ 2 -and F-tests to the fitting problems are also discussed.

A Bayesian method application to the deconvolution of EXAFS spectra is considered. It is shown that for purposes of EXAFS spectroscopy, from the infinitely large number of Bayesian solutions it is possible to determine an optimal range of solutions, any one from which is appropriate. Since this removes the requirement for the uniqueness of solution, it becomes possible to exclude the instrumental broadening and the lifetime broadening from EXAFS spectra. In addition, we propose several approaches to the determination of optimal Bayesian regularization parameter. The Bayesian deconvolution is compared with the deconvolution which uses the Fourier transform and optimal Wiener filtering. It is shown that XPS spectra could be in principle used for extraction of a oneelectron absorptance. The amplitude correction factors obtained after deconvolution are considered and discussed. 61.10.Ht