Abstract
A method is presented which allows the FT-IR spectrum of a material to be related to many of the significant practical properties of that material. The method uses factor analysis of all or part of the spectra of a calibration set of well-characterized samples to determine the minimum set of factors and the associated factor loadings required to reproduce the spectra. Multiple linear regression of the factor loadings against measured properties gives close correlations which are used to predict the properties of unknown samples similar to those constituting the base set. The method is often fast compared with conventional techniques and, by suitable choice, of calibration sets, can give semiquantitative results for a widely varying set of samples, or quantitative results of accuracy comparable with conventional techniques for a calibration set and unknowns from a narrower range of samples. Examples of the use of the method are given for coal, bauxite, manganese dioxide ore, and diesel fuel.
Get full access to this article
View all access options for this article.