Determination of Ethenzamide and Acetoaminophen in Quaternary Powdered Samples by near Infrared Spectroscopy and Searching Combination Moving Window Partial Least Squares Algorithm: Effects of Spectral Resolution and Signal-to-Noise Ratio

The present study reports an application of the searching combination moving window partial least squares (SCMWPLS) algorithm to the determination of ethenzamide and acetoaminophen in quaternary powdered samples by near infrared (NIR) spectroscopy. Another purpose of the study was to examine the instrumentation effects of spectral resolution and signal-to-noise ratio of the Büchi NIRLab N-200 FT-NIR spectrometer equipped with an InGaAs detector. The informative spectral intervals of NIR spectra of a series of quaternary powdered mixture samples were first located for ethenzamide and acetoaminophen by use of moving window partial least squares regression (MWPLSR). Then, these located spectral intervals were further optimised by SCMWPLS for subsequent partial least squares (PLS) model development. The improved results are attributed to both the less complex PLS models and to higher accuracy of predicted concentrations of ethenzamide and acetoaminophen in the optimised informative spectral intervals that are featured by NIR bands. At the same time, SCMWPLS is also demonstrated as a viable route for wavelength selection. The systematic comparison of different spectral resolutions and signal-to-noise ratios of NIR spectra has shown that the spectral resolution 8 cm⁻¹ and 64 co-added scans can lead to better calibration models and satisfactory modelling performance.