An independent, general method for checking consistency between diffraction data and partial radial distribution functions derived from them: the example of liquid water

Abstract: There are various routes for deriving partial radial distribution functions of disordered systems from experimental diffraction (and/or EXAFS) data. Due to limitations and errors of experimental data, as well as to imperfections of the evaluation procedures, it is of primary importance to confirm that the end result (partial radial distribution functions) and the primary information (diffraction data) are consistent with each other. We introduce a simple approach, based on Reverse Monte Carlo modelling, that is capable of assessing this dilemma. As a demonstration, we use the most frequently cited set of "experimental" partial radial distribution functions on liquid water and investigate whether the 3 partials (O-O, O-H and H-H) are consistent with the total structure factor of pure liquid D_2O from neutron diffraction and that of H_2O from X-ray diffraction. We find that while neutron diffraction on heavy water is in full agreement with all the 3 partials, the addition of X-ray diffraction data clearly shows problems with the O-O partial radial distribution function. We suggest that the approach introduced here may also be used to establish whether partial radial distribution functions obtained from statistical theories of the liquid state are consistent with the measured structure factors.