Analysis method of vibrational spectra

摘要

A plurality of Raman spectral datasets of an unknown sample acquired under the same measurement conditions is selected in a plurality of Raman spectral data items having different spectral profiles acquired by setting different measurement conditions in a plurality of unknown mixed samples, a spectrum acquired under the same measurement conditions as those of the Raman spectral dataset of the unknown sample is selected in the spectra of the benchmark molecule to be classified, a PCA loading is calculated from the Raman spectral dataset of the unknown sample, the correlation between the calculated PCA loading and the selected spectrum of the benchmark molecule to be classified is calculated, and the PCA loading having a highest correlation with the spectrum of the benchmark molecule to be classified is determined in the PCA loadings calculated from a plurality of Raman spectral datasets of the unknown sample having different measurement conditions.

主权项

1. A computer-implemented method of vibrational spectrum analysis of an unknown mixed sample, the computer comprising a processor, the method comprising steps of:
a first measuring step of measuring, by a measuring apparatus controlled by the computer, under a first measurement condition a first Raman spectrum of a benchmark molecule, the measuring apparatus at least having a light source and a Raman spectrum detector; a second measuring step of measuring, by the measuring apparatus controlled by the computer, under a second measurement condition, different from the first measurement condition, a second Raman spectrum of the benchmark molecule, wherein the first measurement condition differs from the second measurement condition with respect to at least one of: (a) a different light irradiation condition and (b) a different detection condition and wherein a measurement obtained in the first measurement step and a measurement obtained in the second measurement step are stored in a storage of the computer; a dataset selection step of selecting, by the processor, a plurality of small datasets of a Raman spectral dataset, each small dataset including Raman spectral data of a plurality of unknown mixed samples acquired under the same measurement condition in reference to the measurements stored in the storage of the computer, selected from the first measurement condition and the second measurement condition, wherein the Raman spectral dataset includes a plurality of Raman spectral data items having different spectral profiles acquired based on physical properties of the samples including at least one of a degree of depolarization and a vibrational phase relaxation time with respect to the plurality of unknown mixed samples including a plurality of unknown components at an unknown mixture ratio; a calculation step of calculating, by the processor, PCA loadings from the small datasets selected in the dataset selection step, the PCA loadings being loadings of principal component analysis; a spectrum selection step of selecting, by the processor, a respective Raman spectrum that is preliminarily acquired under the same measurement condition as those of each small dataset selected in the dataset selection step, in the Raman spectra of the benchmark molecule to be classified which is a component intended to be classified; a correlation calculation step of calculating, by the processor, a correlation between the PCA loadings calculated in the calculation step and each Raman spectrum of the benchmark molecule to be classified selected in the spectrum selection step; a determination step of determining, by the processor, a PCA loading which has the highest correlation with the Raman spectrum of the benchmark molecule to be classified, in the PCA loadings calculated from the plurality of small datasets having different measurement conditions; and a classification step of classifying, by the processor, the unknown mixed samples with respect to the benchmark molecule to be classified by plotting a PCA score with respect to the PCA loading determined in the determination step.