Improving the Linearity of Infrared Diffuse Reflection Spectroscopy Data for Quantitative Analysis: An Application in Quantifying Organophosphorus Contamination in Soil
journal contributionposted on 15.01.2006 by Alan C. Samuels, Changjiang Zhu, Barry R. Williams, Avishai Ben-David, Ronald W. Miles,, Melissa Hulet
Any type of content formally published in an academic journal, usually following a peer-review process.
Diffuse reflection data are presented for ethyl methylphosphonate in a fine Utah dirt sample as a model system for organophosphate-contaminated soil. The data revealed a chemometric artifact when the spectra were represented in Kubelka−Munk units that manifests as a linear dependence of spectral peak height on variations in the observed baseline position (i.e., the position of the observed transmission intensity where no absorption features occur in the sample spectrum). We believe that this artifact is the result of the mathematical process by which the raw data are converted into Kubelka−Munk units, and we developed a numerical strategy for compensating for the observed effect and restoring chemometric precision to the diffuse reflection data for quantitative analysis while retaining the benefits of linear calibration afforded by the Kubelka−Munk approach. We validated our Kubelka−Munk correction strategy by repeating the experiment using a simpler systempure caffeine in potassium bromide. The numerical preprocessing includes conventional multiplicative scatter correction coupled with a baseline offset correction that facilitates the use of quantitative diffuse reflection data in the Kubelka−Munk formalism for the quantitation of contaminants in a complex soil matrix, but is also applicable to more fundamental diffuse reflection quantitative analysis experiments.