Chemometric method to enhance sensitivity and selectivity for surface enhanced raman scattering technique

Abstract

The multivariate curve resolution-alternative least square (MCR-ALS) algorithm was modified with sample insertion constraint. This developed method was proposedly used to deconvolute the overlapping signals in Surface enhance Raman attering (SERS) measurement. The developed method was elucidated with the spectral data simulated by using Gaussian distribution function to generate two independent peaks which correspond to capping agent and analyte, respectively. The spectrum of the two peaks was generated with different overlapping levels (RS = 0 – 1.50) and concentration ratio of analyte and capping agent concentration at 0.01 – 1.00. In MCR-ALS with sample insertion constraint, the number of capping agent spectra were added in the range of 10 – 100 times. After excluding the signal from the capping agent, the calibration model of the analyte was built with R² > 0.92 in all conditions. The obtained calibration model is dramatically improved compared with the model generated using either conventional background subtraction or original MCR-ALS. In the case, the appropriate number of added spectra was automatically optimized. Furthermore, our developed method was performed on a real SERS measurement to quantify carbofuran (analyte) by using azo-coupling reaction with p-ATP (capping agent) on the silver nanoparticles as SERS substrate. The calibration model was generated with R² values = 0.99 and LOD = 28.19 ppm which highly improved with the conventional methods. To access the performance of the calibration model, the model was used to estimate the concentrations of carbofuran in an external validation set. It was found that root mean square error (RMSE) of prediction was only 2.109 and R² = 0.97.