Rapid online determination of feed concentration in nitroguanidine spray drying process by near infrared spectroscopy

This work investigated the feasibility of using near-infrared (NIR) spectroscopy to rapidly online detect the feed concentration of nitroguanidine (NQ) in order to ensure the uniformity and stability of the NQ feed during the spray drying process. A Fourier transform NIR spectrometer with a detector installed on the NQ slurry pipeline was used to collect original spectra of NQ samples with different concentrations. Standard normal variate was used to preprocess the original spectra, and a quantitative model with 7 factors of feed concentration was established by partial least squares regression in the spectral wavenumber ranges of 6890-6444 cm(-1) and 5126-4236 cm(-1). The determination coefficients of calibration and prediction (R-c(2),R-p(2)) were 0.9945 and 0.9880, the root mean square errors of calibration and prediction (RMSEC, RMSEP) were 0.111 and 0.157, and the residual predictive deviation (RPD) values of calibration and prediction were 13.5 and 9.18, respectively. The established model was applied to detect the feed concentration of 8 other samples, and the difference between the near-infrared prediction values and the traditional measured values was only 0.29%. The results indicate the excellent predictive performance and repeatability of the model and provide new insights into the NIR spectroscopy rapid online detection of the feed concentration of NQ during spray drying process.