Confirmation of standard error analysis techniques applied to EXAFS using simulations

Abstract

Proper application of error analysis techniques remains uncommon in most EXAFS analyses. Consequently, many researchers in the community remain distrustful of parameter-error estimates. Here, we demonstrate the accuracy of conventional methods through r-space fits to simulated data. Error estimates are determined as a function of r by averaging many scan simulations in r-space. The statistical-χ2 value can then be calculated. Since χ2 corresponds to the degrees of freedom in a fit, we check Stern's rule for the number of independent data points in an EXAFS spectra. Finally, we apply these simple methods to real data from a Cu foil, highlighting the overwhelming role of systematic errors in theoretical backscattering functions and pointing to the ultimate power of the EXAFS technique if such errors could be removed. © 2009 IOP Publishing Ltd.