Optimization of instrumentation precision and wavelength resolution for the performance of NIR calibrations of sucrose in a water-cellulose matrix

Instrumentation for near-infrared (NIR) spectroscopic applications should be optimized for the intended application. The influence of signal precision and wavelength resolution was considered for the application of the noninvasive assessment [NIR 700-1050 nm, partial least-squares (PLS) calibration] of the sugar content of fruit, with the use of a model system of sucrose solution on cellulose. The precision (as estimated at the maximum count of the spectrum) of an MMSI Zeiss spectrometer (Carl Zeiss Pty. Ltd.) was varied by altering the number of scans averaged per spectrum, as well as the signal level. Wavelength resolution was varied between 8 and 20 nm (as estimated of the 912 nm Ar peak) by adjustment of the entrance slit of a prototype spectrometer employing a Hamamatsu S4874Q photodiode array as the detector, constructed on an optical bench. PLS calibrations were developed from interactance spectra of 0-20% w/v sucrose solution soaked filter papers, and compared on the basis of standard error of cross-validation (SECV) and coefficient of correlation-validation (R-v). The optimum measurement precision for calibration development was lower than expected at a coefficient of variation (CV) of 0.02 [signal-to-noise ratio (SNR) 5000:1]. Calibration performance was poorer at a resolution of <8 nm full width at half-maximum (FWHM) in the NIR region, but not significantly different at resolutions of between 8 and 20 nm. Further work is required to define the upper threshold of wavelength resolution. We conclude that instrumentation for the application of fruit sorting can have a relatively poor resolution, and can afford to operate at signal-to-noise levels considered low for a photodiode array detector.