Machine Learning-Based Interpretation of Spectroscopy and Diffraction Data for Materials Characterization

A wide variety of spectroscopic and diffraction techniques are commonly used for the identification and characterization of materials in many fields of study. Traditional interpretation of diffractograms or spectral profiles relies on the practitioner’s ability to accurately and systematically featurize these profiles and make comparisons to reference profiles or correlate changes in these measurements to some material characteristic of interest. However, several studies have shown that traditional human-identifiable features such as peak positions and widths can be unreliable metrics depending on the methods used to identify them, particularly when the material deviates significantly from pristine, defect-free states or when the feature of interest is similar in magnitude to experimental noise. To address these challenges, machine learning methods have been used for the classification or featurization of simulated and experimental profiles for the purposes of rapid materials characterization and identification. In this presentation, we propose a reliable protocol based on supervised manifold learning meant for the extraction of meaningful features from simulated vibrational density of states as well as simulated X-ray diffraction data as exemplar data captured via spectroscopic or diffraction techniques, respectively. Using the extracted features (both separately as well as a combined feature set from the simulated vibrational density of states and simulated X-ray diffraction), we present accurate and robust regression models that are able to disentangle complex overlapping material states from individual profiles, demonstrating comprehensive decoding of profiles beyond classical peak analysis. Sandia National Laboratories is a multi-mission laboratory managed and operated by National Technology and Engineering Solutions of Sandia, LLC., a wholly owned subsidiary of Honeywell International, Inc., for the U.S. Department of Energy National Nuclear Security Administration under contract DE-NA0003525.