Analyzing Raman images of polymer blends by sample–sample two-dimensional correlation spectroscopy

Abstract

A simple method that facilitates analysis of chemical images is presented. It is based on so-called two-dimensional (2D) correlation spectroscopy and its main purpose is to identify the most specific spectra in an imaging set. The selection is achieved by measuring correlation among the spectra via sample–sample (SS) correlation technique. The idea of SS correlation is exemplified through a simple model and then applied to the mapping Raman spectra of compatibilized (a mixture of two polymers is made homogenous thereby improving its useful plastic properties) and uncompatibilized (mechanically blended) blends of Nylon 12/high-density polyethylene (HDPE), and poly(ethyleneterephthalate) (PET)/HDPE. The Raman spectra of the blends are analyzed with and without pre-treatments (baseline correction by an nth order polynomial, and normalization), and it is found that the results are significantly dependent on the pre-treatment. It is concluded that all the analyzed blends do not exhibit band-shape variation despite the compatibilization, but appear to have unique vibrational signatures, i.e. each blend is found to be completely represented by only one spectrum. The sensitivity of the employed SS correlation method is illustrated through making distinction among heavily overlapped spectra differing minutely in the baseline.