Multiple-perturbation two-dimensional (2D) correlation analysis for spectroscopic imaging data

Abstract A series of data analysis techniques, including multiple-perturbation two-dimensional (2D) correlation spectroscopy and kernel analysis, were used to demonstrate how these techniques can sort out convoluted information content underlying spectroscopic imaging data. A set of Raman spectra of polymer blends consisting of poly(methyl methacrylate) (PMMA) and polyethylene glycol (PEG) were collected under varying spatial coordinates and subjected to multiple-perturbation 2D correlation analysis and kernel analysis by using the coordinates as perturbation variables. Cross-peaks appearing in asynchronous correlation spectra indicated that the change in the spectral intensity of the free C O band of the PMMA band occurs before that of the C O⋯H O band arising from the molecular interaction between PMMA and PEG. Kernel matrices, generated by carrying out 2D correlation analysis on principal component analysis (PCA) score images, revealed subtle but important discrepancy between the patterns of the images, providing additional interpretation to the PCA in an intuitively understandable manner. Consequently, the results provided apparent spectroscopic evidence that PMMA and PEG in the blends are partially miscible at the molecular level, allowing the PMMAs to respond to the perturbations in different manner.