Calibration errors due to variations in peak characteristics in the measurement of transient signals by inductively coupled plasma-scanning mass spectrometry

The impact of variations in peak characteristics on the fidelity of transient signal measurement by inductively coupled plasma-quadrupole mass spectrometry (ICP-QMS) was investigated. Specifically, the question as to whether the multi-element capabilities or the accuracy in determined analyte amounts were deteriorated compared to what has been reported previously when not considering peak variations was addressed. The peak characteristics considered were the time of the signal maximum (t(peak)), the standard deviation of the assumed Gaussian input function generated by the sample introduction system (sigma(G)), and the time constant for signal decay (tau). Investigations of simulated exponentially-modified Gaussian peaks revealed that, for variations of peak characteristics within reasonable ranges, measurement noise and variations in t(peak), sigma(G) and tau all contributed to calibration uncertainty. Electrothermal vaporisation (ETV) and flow injection (FI) systems were used to experimentally generate transient signals of varying peak characteristics. Removing data points from the raw signals simulated the monitoring of up to 100 mass-to-charge ratios, allowing calibration data and analyte amounts to be determined from the processed signals. To obtain calibration graph slopes with relative standard deviations below 1% for the ETV-ICP-QMS system, it was found necessary to acquire 7-24 data points per peak for 50-5 ms dwell times. On this basis, the maximum number of mass-to-charge ratios that could be monitored in a typical ETV-ICP-QMS analysis was 4-10 using dwell times of 50-5 ms. With the FI-ICP-QMS system, variations in the peak characteristics between calibration standards and samples meant that, to obtain less than 3% error in determined analyte amounts, at least 7 or 10 points per peak were required for external and internal standardisation, respectively. It was found that variations in peak characteristics contributed more than measurement noise to the error in determined analyte amounts. In recent studies it has been reported that 3-4 data points per peak are sufficient to accurately monitor a transient if peak variations are not considered, which, for typical ETV signals, would allow the monitoring of 20 mass-to-charge ratios during a single measurement cycle. Thus the results obtained here show that the multi-element capability of ICP-QMS when monitoring transient signals can be severely compromised by such variations.