Skip to main content
ARS Home » Research » Publications at this Location » Publication #156997

Title: EFFECT OF WATER AND PHYSICAL STATE ON NEAR- AND MID-INFRARED PARTIAL LEAST SQUARES (PLS) CALIBRATIONS FOR MULTI-COMPONENT CARBOHYDRATE MIXTURES

Author
item Reeves Iii, James

Submitted to: Near Infrared Spectroscopy Journal
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 9/16/2004
Publication Date: 11/5/2004
Citation: Reeves III, J.B. 2004. Effect of water and physical state on near- and mid-infrared partial least squares (pls) calibrations for multi-component carbohydrate mixtures. Near Infrared Spectroscopy Journal. 12:241-250.

Interpretive Summary: Spectroscopy uses light to determine the composition of materials such as animal feedstuffs. Past work has shown that the presence of water in samples can seriously degrade the accuracy of such spectroscopic determinations (calibrations), but has not completely determined the reason behind this degradation. The objective of this research was to determine the effects of water and physical state on near- and mid-infrared calibrations for carbohydrate mixtures. Seventy five mixtures of cellulose, starch, pectin, locust gum, glucose and xylitol were scanned in the near- and mid-infrared in three physical states (wet, simple mixtures of dry constituents, and mixtures dried together). These final samples (mixtures dried together) were prepared by drying mixtures while under constant stirring, to prevent separation of the insoluble polymers by settling, which occurs during normal oven drying, followed by oven drying at 60 C and grinding. The wet samples were produced by adding sufficient water to result in samples of a putty like consistency. Results demonstrated that the effects of water and physical state (matrix effects), seen previously with simple model systems, can be extrapolated to draw some general conclusions about real world calibrations, but that developing model systems to either predict the feasibility of developing a calibration or to determine the reason for the specific results of a calibration must be done with extreme care. Also, from the results, it would appear that factors (e.g., physical state) other than the mere presence of large amounts on water in samples can account for the decrease in calibration accuracy often seen in samples with high moisture contents as compared to their dried counterparts. Finally, the results seen using mid-infrared spectra indicate that mid-infrared diffuse reflectance spectroscopy might be feasible even with samples containing high percentages of water.

Technical Abstract: The objective was to determine the effects of water and physical state on near- and mid-infrared calibrations for carbohydrate mixtures. Seventy five mixtures of cellulose, starch, pectin, locust gum, glucose and xylitol were scanned in the near- and mid-infrared in three physical states (wet, simple mixtures of dry constituents, and mixtures dried together). These final samples (mixtures dried together) were prepared by drying mixtures while under constant stirring to prevent separation of the insoluble polymers by settling, which occurs during normal oven drying. Final drying of the resulting solid mass was carried out in a vacuum oven at 60o C. The dried mass was then ground and scanned in the spectrometer. Finally, all dry samples were ground to a flour like consistency using a A-10 Analytical Mill by Tekmar (Cincinnati, OH) to avoid effects due to gross differences in particle size. The wet samples were produced by adding sufficient water to result in samples of a putty like consistency. Spectra were taken on a Digi-Lab (BIO-RAD, Cambridge, MA) FTS-65 Fourier transform spectrometer equipped with near- and mid-infrared benches by diffuse reflectance. Results demonstrated that the effects of water and physical state (matrix effects), seen previously with simple model systems, can be extrapolated to draw some general conclusions about real world calibrations, but that developing model systems to either predict the feasibility of developing a calibration or to determine the reason for the specific results of a calibration must be done with extreme care. For example, using simple mixtures of components to determine whether a calibration for the same components in a different physical condition (melted together, dried together, etc.) might well lead to erroneous conclusions about its usefulness. Also, from the results, it would appear that factors (e.g., physical state) other than the mere presence of large amounts on water in samples can account for the decrease in calibration accuracy often seen in samples with high moisture contents as compared to their dried counterparts. Finally, the results seen using mid-infrared spectra indicate that mid-infrared diffuse reflectance spectroscopy might be feasible even with samples containing high percentages of water.