Reliability of pair distribution function analysis in <i>in situ</i> experiments

<i>In situ</i> and <i>operando</i> pair distribution function (PDF) studies are becoming commonly used to study chemical reactions, nucleation and growth of nanoparticles, or structural changes during the operation of batteries, catalysts, thermoelectric devices <i>etc.</i> However, repeated time-resolved total scattering experiments and subsequent PDF analysis are often not prioritized due to the scarce synchrotron beam time available. This means that the experimental uncertainty and reproducibility of the experimental methods are unknown, and the full potential of <i>in situ</i> PDF experiments may not be exploited. Here, we quantify the experimental uncertainty of the PDF technique in an <i>in situ</i> study of the hydro-thermal synthesis of ZrO₂ nanoparticles. Systematic variation of the parameters used to obtain the PDF shows that the user-defined parameters can potentially affect the chemical conclusions obtained from the time-resolved experiment. We found that comparable results are best obtained using the same input parameters across different experiments. We also compare different PDF algorithms to examine whether the processing algorithm influences the chemical analysis.