Kombination von optimaler Versuchsplanung und einigen Matrixmethoden zur Mehrkomponentenkalibrierung und — analyse

Summary

Multicomponent analysis often suffers from matrix effects. In the present paper several methods are employed to cope with those effects and to keep experimental expense as low as possible and hence to reduce analysis time. Based on the determination of Na, K, and Ca by flame atomic emission spectroscopy experimental design in combination with multivariate calibration (matrix methods) has been tested for suitability in multicomponent analysis. Good results are achieved by statistically designed experiments. The experimental domain should be kept small to allow for linear models. Comparable results are obtained using the P method. As expected the K approach (Beer model) yields the worst results.