Two-dimensional (2D) correlation coefficient analyses of heavily overlapped near-infrared spectra

Abstract

Two-dimensional (2D) correlation coefficient analysis is employed to classify and characterize spectral variations among heavily overlapped near-infrared spectra of pellets and films of three kinds of polyethylene (PE), high-density (HD), low density (LD), and linear low-density (LLD) polyethylene, and five kinds of ivory signature seals. The sample–sample (SS) 2D correlation maps are used for classification while the wavenumber–wavenumber (WW) 2D correlation maps are used for determining spectral variation among the above materials. Both correlation maps are obtained by multiplying the original data with themselves. It is found that the NIR spectra of pellets and films of HD PE are clearly different from those of LD PE and LLD PE, while the NIR spectra of five kinds of ivory seals yield easily discernable squares in the SS correlation maps. The background variation is thought to be behind the differentiation of the PE samples because the WW correlation maps do not indicate appearance of new bands. The correlation results are compared with those of principal component analysis (PCA). This study is a novel application of 2D correlation coefficient analysis which reveals that a comprehensive description of demanding spectral systems is achievable by utterly simple mathematical means because 2D correlation maps are obtained via a single mathematical operation.