Choosing internal standards based on a multivariate analysis approach with ICP(TOF)MS

Internal standards are routinely used with inductively coupled plasma mass spectrometry (ICPMS) to minimize the impact of signal instability. Previous studies indicate that internal standard choice should be directed by similarities in mass, but they neglect to address the possibilities of exceptions or what impact a poor internal standard choice may have on quantitative results. A 51 element suite was run under varying instrument and matrix perturbations on an ICPMS instrument equipped with a time of flight mass spectrometer. Each element was ratioed to every other element for perturbed conditions that were intended to simulate variations in matrix composition or drift in operating parameters. These included changing sample delivery rate, matrices (NaCl and acetic acid) and horizontal torch position. The %RSDs of each analyte-to-internal standard ratio were used to rank the relative quality of each analyte-internal standard pair and their response to changing conditions. Multiple ordinary least squares equations were calculated to evaluate various chemical and physical properties that may be predictors of an optimal analyte-internal standard combination. Overall, similarity of masses was found to be the most important predictor of a good internal standard. However, exceptions did exist under various perturbations and among different elements. In a final study, nine analytes were run under perturbed experimental conditions and in two complex matrices (NaCl and acetic acid), but quantitation was attempted using simple aqueous standards. Accuracy and precision were evaluated for several internal standards for each of the nine elements as well for the nine analytes where no internal standard was employed. From this evaluation, it was clear that internal standard choice can have substantial effects on analytical accuracy and the improvement of analytical precision, but there is no single physical or chemical parameter that reliably allows a priori selection of a “good” internal standard for any given analyte.