Premium
Unsupervised Screening of Vibrational Spectra by Principal Component Analysis for Identifying Molecular Clusters
Author(s) -
Kiefer Johannes,
Eisen Kristina
Publication year - 2018
Publication title -
chemphyschem
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.016
H-Index - 140
eISSN - 1439-7641
pISSN - 1439-4235
DOI - 10.1002/cphc.201701353
Subject(s) - principal component analysis , chemistry , aqueous solution , component analysis , spectral line , component (thermodynamics) , analytical chemistry (journal) , biological system , computer science , thermodynamics , chromatography , artificial intelligence , physics , astronomy , biology
Vibrational spectra are commonly used to study molecular interactions in solutions. However, the data analysis is often demanding and requires significant experience in order to obtain meaningful results. This study demonstrates that principal component analysis (PCA) can serve as an unsupervised tool for initial screening of non‐ideal mixture systems. Taking the aqueous solutions of dimethyl sulfoxide (DMSO) as an example, PCA reveals—easily and fast—the two prominent stoichiometries at 1:2 and 1:1 molar DMSO:water ratio and significantly outperforms elaborate spectral profile analysis or common algorithms as indirect hard modeling (IHM) or multivariate curve resolution (MCR). The corresponding molecular 1:1 and 1:2 clusters are known to be dominating configurations in the solutions.