An optimized spectral reconstruction method for shift excitation Raman differential spectroscopy

Raman spectroscopy is a powerful analytical method, but when the composition of the test sample is intricate, the original spectral data may contain noise and fluorescence background interference, making it more difficult to extract Raman spectral information from the original spectra. Especially the fluorescence background signal, which is typically several orders of magnitude stronger than the Raman signal, can even overwhelm or obscure the Raman signals, thereby impeding the qualitative or quantitative analysis of the Raman spectra. One effective method for removing the fluorescence background is shift excitation Raman differential spectroscopy (SERDS), which typically involves measuring two raw Raman spectra using slightly different excitation wavelengths, combined with reconstruction algorithms, to obtain Raman spectra free from fluorescence interference. For this purpose, a reconstruction method based on Tikhonov regularized least squares (TRLS) was developed in this study, which mitigated the oscillations caused by the direct unconstrained least squares (DULS) reconstruction method. The method was verified and optimized using four groups of artificial datasets with different characteristics. By selecting an appropriate value for parameter α, the relative standard deviation (RSD) of the reconstructed datasets was lower than that of the artificial datasets in most cases. Additionally, we evaluated the performance of the TRLS reconstruction algorithm based on a quantitative model of real Raman spectral datasets, assessing the algorithm's performance from three perspectives: the root mean square error (RMSE), the correlation coefficient (R), and the ratio of prediction to deviation (RPD). The quantitative results indicate that using the TRLS method for reconstruction enhances both prediction accuracy and practicality. In summary, findings from both simulated data and actual experiments demonstrate that the TRLS-based reconstruction method substantially improves the stability and reliability of differential Raman spectra reconstruction.