Portable Raman spectroscopy and fourier transform near infrared spectroscopy for the quantification of different sinomenine hydrochloride crystal forms

The objective of this paper is to rapidly and accurately quantify the content of the dominant crystal form of Sinomenine hydrochloride (SH) and to evaluate the respective characteristics of Raman spectroscopy and Fourier transform near infrared spectroscopy techniques for rapid quantification of crystalline substances. In this study, we performed an adulterated gradient quantification based on two new crystalline forms of SH prepared in the laboratory in combination with commercially available products. And established 86 samples containing 66 batches of ternary and 20 batches of binary mixtures. We quantified SH possessing polycrystalline forms based on two kinds of spectroscopy techniques combined with chemometric methods. In the process of establishing the partial least squares quantitative model, according to the experimental design thought, we used the ensemble preprocessing method to screen the optimal preprocessing method for the spectral data of crystal samples. Then we optimized four variable selection methods by single factor investigation. Finally, after completing the PLSR model, we found that the regression models established for both Raman spectra (RMSEP = 0.00672) and FT-NIR spectra (RMSEP = 0.00533) have a high ability to rapidly quantify the crystals.