Investigation of phase transitions in liquid crystal 12BBAA using window clustering of infrared spectra

• Published in: Journal of molecular liquids Vol. 341; p. 117233
• Main Authors: Osiecka-Drewniak, Natalia, Czarnecki, Mirosław A., Galewski, Zbigniew
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.11.2021
• Subjects: 12BBAA; Clustering analysis; Infrared spectroscopy; Liquid crystals; Machine learning; Phase transition; Window analysis; Infrared spectroscopy; Machine learning; Clustering analysis; Window analysis; 12BBAA; Liquid crystals; Phase transition
• ISSN: 0167-7322; 1873-3166
• DOI: 10.1016/j.molliq.2021.117233

•Window clustering algorithm interprets temperature-depended infrared spectra.•The method makes it possible to reduce the dimensionality of large data set.•Algorithm shows spectral characteristic of phase associated with spectral regions.•Method is sensitive to melting process of alkyl chains in liquid crystal phases. The k-means cluster analysis has been applied to the temperature-dependent infrared spectra of 4-bromobenzylidene-4′-dodecyloxyaniline (12BBAA). This method appears to be particularly sensitive to the tracking of the melting process of alkyl chains. 12BBAA may form a variety of phases including: isotropic liquid, smectic A, crystalline smectic B and crystal phase. Clustering methods are helpful tool for detection of phase transitions, but do not provide information on the structural changes due to the phase transitions. For this reason, window clustering analysis method has been proposed. Multivariate data has been divided into small spectral windows, and then cluster analysis has been applied for each data window. This approach allows to reduce the dimensionality of large data set, while reveals the spectral characteristic of given phase associated with various regions of spectra.