Unravelling error sources in miniaturized NIR spectroscopic measurements: The case study of forages

• Published in: Analytica chimica acta Vol. 1211; p. 339900
• Main Authors: Gorla, Giulia, Taiana, Alessandro, Boqué, Ricard, Bani, Paolo, Gachiuta, Olga, Giussani, Barbara
• Format: Journal Article
• Language: English
• Published: Netherlands Elsevier B.V 08.06.2022
• Subjects: Forages; Miniaturized NIR spectrometer; Multivariate measurement error; PLS regression; PLS regression; Miniaturized NIR spectrometer; Multivariate measurement error; Forages
• ISSN: 0003-2670; 1873-4324; 1873-4324
• DOI: 10.1016/j.aca.2022.339900

The use of miniaturized NIR spectrometers is spreading over the scientific literature with a particular focus on developing methods as rapid and easy-to-use as possible and following the philosophy of green analytical chemistry. Several applications and studies are typically presented by comparing results obtained with benchtop instrumentation even when the analytical strategies are substantially different. Indeed, analytical applications that include the use of miniaturized instrumentation are subject to several sources of variability that need to be known at the time of method development. In this study, different statistical strategies were employed to understand the features and limitations of handheld NIR instruments. Because of the high interest in real applications, a common type of hygroscopic powder sample was selected: forages. A step-by-step methodology is presented to statistically address the different issues to consider in order to obtain realistic models when using miniaturized NIR spectrometers. We demonstrate how a careful evaluation of the sources of variability related to an experiment can help in the understanding of the system under study in order to obtain a more reliable development of the method and consciously choose the analytical parameters and strategies of analysis. The results were also compared with those achieved on the same dataset from a benchtop system in order to provide references analogous with those in the literature. [Display omitted] •Statistical methods to estimate the characteristics of miniaturized NIR data.•Study of the sources of errors in miniaturized NIR spectrophotometric measurements.•Multivariate regression models for predicting forage properties using NIR spectra.