Qualitative discriminant analysis applications

NIR instruments may not only be calibrated for quantitative analysis, but they may also be trained for qualitative purposes. This process is usually termed discriminant analysis. The criteria noted previously for establishing quantitative calibrations with minimum prediction error are equally applicable to discriminant calibration sets. 

As we argued earlier, application in chemistry and materials science usually requires quantitative information. When considering methods of analysis, one may discriminate between very fast but less accurate methods to get qualitative answers, and slower but more accurate methods yielding true quantitative information. The former one is the class of methods that can be speeded up to a high extent, i.e., thousands of samples a day. This may apply to techniques... 

The phenomenon of discrimination induced by matrix effects is quite disadvantageous in qualitative analysis, but in the quantification procedure its influence is devastating. Several techniques applying MS or MS-MS, however, can help to overcome this crucial influence. The most common way out of this dilemma is the application of column separation to minimise or even exclude matrix effects entailed with the disadvantage of long-lasting separations. In parallel, the difference between the time needed for the performance of LC examinations and the time needed for FIA becomes obvious and will be pointed out later on. 

Rodger et al. (1998) demonstrated the use of surface-enhanced resonance Raman scattering (SERRS) spectroscopy, without any separation procedure, to analyze dyes and pigments in lipsticks. Lipsticks smeared on glass and cotton surfaces required treatment with a surfactant, for example, poly(L-lysine), and silver colloid prior to the analysis. This in situ SERRS method was applied to six commercial lipstick samples. Discrimination between the samples and identification of some of the pigments present were achieved. The method is qualitative in nature and was suggested to have potential for forensic and quality-control applications. 

---