Mitigation of errors in pair distribution function analysis of nanoparticles

Information on the size and structure of nanoparticles can be obtained via analysis of the atomic pair distribution function (PDF), which is calculated as the Fourier transform of X-ray/neutron total scattering. The structural parameters are commonly extracted by fitting a model PDF calculated from atomic coordinates to the experimental data. This paper discusses procedures for minimizing systematic errors in PDF calculations for nanoparticles and also considers the effects of noise due to counting statistics in total scattering data used to obtain the PDF. The results presented here demonstrate that smoothing of statistical noise in reciprocal-space data can improve the precision of parameter estimates obtained from PDF analysis, facilitating identification of the correct model (from multiple plausible choices) from real-space PDF fits.