Search Thermo Fisher Scientific

The goal of almost all ICP-MS analyses is to quantify trace elements within a sample. The output of an ICP-MS is numerical, and usually provided in counts per second. For any quantitative analysis, regardless of the technique, the instrument response is largely meaningless without reference to a known value. Calibration is the process of determining the instrument response for a known concentration of analyte, and in doing so, enabling test sample concentrations to be quantified.

For this reason, almost all analyses include a set of calibration reference samples, or ‘standards’. A range of standards is tested at the start of the analysis to generate a ‘calibration curve.’ In this way, instrument response can be correlated to actual analyte concentration within the samples.

When complete precision is not required, and analyte standards are not readily available, semi-quantitative calibration can be used to generate concentration data. Semi-quantitative calibrations are typically based on the sensitivities of elements surrounding the analyte signal, and the response from this range of elements is already known. These sensitivities are corrected for isotope abundance and a curve is fitted to them. Sensitivity for the analyte is then measured from that curve, and a concentration is calculated.

The response curve is usually a polynomial fit, and therefore requires at least three other elements for effective semi-quantitative analysis to be performed. When elements with similar masses to the analyte are used, semi-quantitative concentration data offer good accuracy.

External calibration uses a range of separate (external) standards of known concentration of the analyte to generate a calibration curve of instrument response. The calibration curve, which is usually a linear regression, is then used to back-calculate the concentration of analyte in unknown samples based on their instrument response.

Pure elemental standards generate highly accurate calibration curves; however, they do not account for matrix-induced effects, such as signal suppression, or instrument drift that can arise due to the matrix present in the sample.

Drift can be corrected by recalibrating throughout the assay, but this can add significant time and expense to most routine assays. Matrix effects can be compensated for by using matrix-matched calibration standards, which means adjusting the matrix of the standards to match that of the samples. Accomplishing this task successfully depends entirely on how well the sample matrix is understood and characterized.

A good way to compensate for matrix effects and instrument drift is to combine external calibration with internal standardization. This is the most widely used method of calibration in ICP-MS and involves adding the same amount of one or more elements to all measured solutions (blanks, calibration standards, quality control standards, unknown samples, etc.). The response from this element is expected to be the same throughout the assay, so that any variation is assumed to be derived from either matrix effects or instrument drift. A mathematical correction factor is calculated from the relative internal standard response that is then applied to the analytes to correct for both matrix and drift effects.

The ideal internal standard for any given analyte is not already present in the sample, has a similar mass and ionization potential as the analyte, and behaves similarly to the analyte, both in solution and the plasma.

In cases where the samples have a very high percentage of total dissolved solids (e.g., 0.3% or higher), internal standardization may not be sufficient to correct for matrix-induced effects and drift. The sample is then divided into several aliquots in order to generate a calibration curve (e.g., three aliquots for a blank and two standards). Into each of these aliquots, external calibration solutions are spiked directly. This yields an ‘offset’ calibration curve, and the negative x-intercept is the analyte concentration.

While the method of standard additions is more robust and reliable than typical calibration, it is costly and time-consuming. Ideally, one might use just one sample to generate a calibration curve, then use that curve to quantify many other samples, provided they all have the same matrix.

Learn how elements and their isotopes provide information about sample origin and identity in our library of applications notes, scientific posters, webinars, and more.

Resources