Rapid detection of phenytoin sodium by partial-least squares and linear regression models combined with surface-enhanced Raman spectroscopy.

ABSTRACT

Surface-enhanced Raman spectroscopy (SERS) for quantitative analysis is challenging owing to the unstable enhanced effect. However, it can be improved by combining it with chemometrics. In this study, we established a quantitative analysis method for phenytoin sodium (PS) based on partial least-squares (PLS) and linear regression (LR) models combined with SERS. Gold nanoparticles (AuNPs) were optimally enhanced substrates for PS. 180 PS samples in the concentration range of 0.98 - 980 µg mL-1 were used to establish a quantitative prediction model by PLS regression, and an accurate and robust prediction was achieved. Furthermore, we found that SERS peak intensity showed a good linear correlation with the concentration of PS in the concentration range of 1 - 80 µg mL -1. After using P-mercaptobenzoic acid as an internal standard, the accuracy and precision of the LR model were significantly improved compared with that of the model without an internal standard. In general, PLS chemometrics and LR model with internal standard which were combined with SERS in this paper provide new possible analytical methods for analytes to develop a rapid and sensitive quantitative analysis method.