X-Ray spectroscopy is a premier tool to probe material structures. High-throughput or In-Situ experiment encodes spatial or temporal structural variations in a sequence of spectra. Multivariate Curve Resolution - Alternating Least Squares (MCR-ALS) is an old technology to decompose the sequence. It derives corresponding concentration profiles and individual spectra. However, because of rotational ambiguity, the solution is not unique. We have demonstrated an approach to solve this problem in the context of spectroscopy. We have developed constraints to embed prior knowledge and narrow down the solution space. The deconvolution of a spectral sequence is a non-linear problem. MCR-ALS converts it to two linear problems by performing least square regressions for concentration and spectrum alternatively. We inject the constraints between regressions to enforce conformity to prior knowledge. Figure 1 illustrates the analysis for a synthetic data set using a phase law derived constraint. We have demonstrated this approach in the analysis of a TiO2 thin film growth [Chem. Mater. 2021, 33, 1740]. The process is monitored by In-Situ X-ray absorption near edge spectroscopy (XANES). The surface layer is only a small fraction of the film and is barely visible in any single snapshot. However, MCR isolates surface spectrum successfully with our constraint. In addition, we have developed rank selectivity, smoothness, special shape, and monotonic constraints targeting a broad range of problems.

Spectral sequence deconvolution for in-situ X-ray spectroscopy

ACS Fall 2021

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.