Mark Cave
A chemometric modelling procedure is described using principal component analysis (PCA) that is capable of improving short-term precision in simultaneous ICP-AES. The PCA model is set up from mean centered and standard deviation scaled emission data, allowing direct comparison of noise from emission signals with widely varying intensities. The procedure does not require the use of an internal standard but uses a combination of the noise measured on all analytical lines to predict the noise on any one emission line. Two PCA models have been tested: the first using training data from one multi-element solution at five different integration times and the second using data from a test solution analysed at five different dilutions but at a single integration time. Improvements in coefficients of variation for the 18 elements studied ranged from a factor of ≈1 to ≈18. Reduction of the noise found on the blank allowed reductions in detection limit ranging from a factor ≈2 to ≈18.