Portability of near Infrared Spectroscopic Calibrations for Petrochemical Parameters

Abstract

The purpose of this paper is to analyse the origin of the differences in spectral response between different Fourier transform near infrared spectrometers of the same type. The influence of the sample and room temperatures is analysed and the importance of the field of view at the sampling accessory and of the near infrared lamp alignment is demonstrated. The performances of a Shenk–Westerhaus type correction have been demonstrated on seven PLS1 models predicting different physico-chemical properties of hydrotreated gas oils. The bias in the transfer of these models from one system to another was equivalent to the repeatability of analysis for a single system. Nevertheless, it has been shown that this type of standardisation algorithm compensates a frequency shift of 0.32 cm⁻¹ by intensity correction. Spectrometers characterised by a reproducibility in intensity of the order of ± 0.002 absorbance units (in the frequency range 6400–5500 cm⁻¹, with a 2 mm cell) and a wavenumber reproducibility of ± 0.04 cm⁻¹ show no need for calibration transfer with a sample temperature controlled at ± 0.5°C.