Fig. 20 FTIR-PAS spectra of coal in powder (upper) and single particle form (A,B, and C). Variations in the mineral content of the particles are indicated by bands in the 400-1000 cm and 3600-3700 cm^-1 regions.

C. Quantitative Analyses

The application of factor analysis for processing FTIR-PAS spectra enables quantitative analyses to be readily performed with standard error of prediction (SEP) values below 1% in spite of the truncation of strong absorbance bands that occurs due to photoacoustic signal saturation. Both principal components regression (PCR) and partial least squares (PLS) factor analyses routines tolerate nonlinearities in spectra well and allow concentrations of multicomponent systems to be determined as well as other physical and chemical properties of materials. The quality of the analysis of the FTIR-PAS technique is sensitive to the same considerations as other FTIR sampling methods. These items should be considered when a new quantitative method is being developed for a specific application.

1. The number of learning set samples used and the differences between their physical and chemical property values and those of the unknowns.
2. The provision for several learning set samples with physical or chemical property values above and below those of the unknowns.
3. The accuracy of the learning set.
4. The spectral range or ranges used in the factor analysis.
5. The spectral resolution of spectra.
6. The signal-to-noise ratio of spectra.
7. The system response stability over time and the reproducibility from sample to sample.

The first six considerations are very important to the accuracy of quantitation but are not specific to the sampling method used and will not be discussed in detail here. The last consideration depends on the FTIR, the sampling accessory, the samples, and the