Structure–spectrum relationship in the calculated Raman spectra of silicates

• Published in: Journal of Raman spectroscopy Vol. 55; no. 10; pp. 1113 - 1123
• Main Authors: Bagheri, Mohammad, Komsa, Hannu‐Pekka
• Format: Journal Article
• Language: English
• Published: Bognor Regis Wiley Subscription Services, Inc 01.10.2024
• Subjects: Algorithms; Atomic structure; Cluster analysis; DFT; First principles; Raman spectra; Raman spectroscopy; Raman spectrum; silicate; Silicates; Silicon oxide; Silicon oxides; spectral matching; Trends; Uncertainty analysis
• ISSN: 0377-0486; 1097-4555
• DOI: 10.1002/jrs.6686

Silicate minerals have a rich structural variety consisting of silicon oxide clusters or networks of any dimensionality interdispersed with different types of elements, which is reflected in marked changes in the Raman spectra. Understanding how the changes in the Raman spectra are correlated with the atomic structure would be highly desirable for fast material identification and analysis. Extracting such trends from experimental spectra can be difficult owing to the uncertainties in the structural details of the samples and in ensuring consistency between measurements from different sources. Simulated spectra, however, avoid these problems, making them a good candidate for systematic studies. Here, we study the correlation between the structure and Raman spectral features of 179 silicates derived from a database of Raman spectra simulated using first‐principles calculations. We investigate the spectral similarities with a specific emphasis on materials containing isolated 0D clusters in nesosilicate (SiO4), sorosilicate (Si2O7), and cyclosilicate (Si3O9 and Si4O12) configurations. While trends identified in the previous reports can be confirmed, we find that the variations within each group of similar structural motifs tend to be larger than the changes across groups, and therefore, developing a reliable automated classification algorithm is likely to be challenging. We study the correlation between the structure and Raman spectral features of silicates derived from a database of Raman spectra simulated using first‐principles calculations, with a specific emphasis on materials containing isolated 0D clusters. While trends consistent with previous reports can be identified, we find that the variations within each group of similar structural motifs can be larger than the changes across groups, which makes automated classification challenging.