(51h) Limitations of Multivariate Curve Resolution-Alternating Least Squares Analysis for Determining Components in Operando XAS Catalyst Characterization

Operando X-ray Absorption Spectroscopy (XAS) is a powerful tool for elucidating the structure and dynamics of catalysts. The advent of continuous- and quick-scanning capabilities at synchrotron facilities has improved the time resolution of XAS measurements (e.g. ≈3600+ vs. ≈4 spectra/h for quick- and step-scanning XAS, respectively), unlocking new insights into catalyst dynamics with datasets that include the spectroscopy as well as process conditions. To speed up the XAS analysis, statistical tools such as principal component analysis (PCA) and multivariate curve resolution-alternating least squares (MCR-ALS) are now used to determine the number of unique components and their contributions. The eigen spectra from the MCR-ALS are compared to the X-ray absorption near edge structure (XANES) of bulk references to add chemical insights. Attempts to extract and model “extended X-ray absorption fine structure (EXAFS)” from the MCR-ALS results have also been made. However, there has been no study into the validity of using the MCR-ALS results to extract “EXAFS” and how this compares to the XAS data from the experiment.

In this talk we will explore how accurate MCR-ALS eigen spectra “XANES and EXAFS” are for ascribing chemical information by comparing operando XAS results for the reduction of SiO₂ supported Ni nanoparticles (1.7 nm) and bulk NiO. CatXAS, and pTauSpec enabled the processing of the XAS datasets and the PCA/MCR-ALS analysis. Our study shows that the information content in the XANES is better for ascribing chemical insights whereas the “EXAFS” can be modeled, however, the resulting structure models do not always match that extracted from the raw data, Figure 1. This shows that while MCR-ALS is a powerful tool to determine the number of statistically relevant components in a catalytic system, care needs to be taken in interpretation in the results, especially when the EXAFS is attempted to be extracted and modeled.