Sample diagnosis using indicator elements and non-analyte signals for inductively coupled plasma mass spectrometry

Abstract

A sample diagnosis procedure that uses both non-analyte and analyte signals to estimate matrix effects in inductively coupled plasma-mass spectrometry is presented. Non-analyte signals are those of background species in the plasma (e.g. N +, ArO +), and changes in these signals can indicate changes in plasma conditions. Matrix effects of Al, Ba, Cs, K and Na on 19 non-analyte signals and 15 element signals were monitored. Multiple linear regression was used to build the prediction models, using a genetic algorithm for objective feature selection. Non-analyte elemental signals and non-analyte signals were compared for diagnosing matrix effects, and both were found to be suitable for estimating matrix effects. Individual analyte matrix effect estimation was compared with the overall matrix effect prediction, and models used to diagnose overall matrix effects were more accurate than individual analyte models. In previous work [Spectrochim. Acta Part B 57 (2002) 277], we tested models for analytical decision making. The current models were tested in the same way, and were able to successfully diagnose matrix effects with at least an 80% success rate.