Linearity and Detection Limits

Confirm Basic Performance Criteria

The following excerpt was taken from Part 17: Method Validation of our Trace Analysis series. This secsion discusses performance criteria confirmation during the method validation process. Please note that the validation process is more detailed and specific.

The method must 'fit the purpose' as agreed upon between the client and the analyst. In the case of trace analysis, the following criteria are typically evaluated as part of the method development process:

• Specificity involves the process of line selection and confirmation that interferences (of the types discussed in part 15 and part 16) for the ICP-OES or ICP-MS measurement process are not significant. A comparison of results obtained using a straight calibration curve (without internal standardization to that of internal standardization and/or to the technique of standard additions) will give information concerning matrix effects, drift, stability, and the factors that influence the stability. The various types of spectral interferences encountered using ICP-MS and ICP-OES (see above links) should be explored.
• Accuracy or Bias can be best established through the analysis of a certified reference material (CRM, or SRM if obtained from NIST). If a CRM is not available, then a comparison to data obtained by an independent validated method is the next best approach. If an alternate method is not available, then an inter-laboratory comparison, whereby the laboratories involved are accredited (ISO 17025 with the analysis on the scope of accreditation) is a third choice. The last resort is an attempt to establish accuracy through spike recovery experiments and/or the use of standard additions.
• Repeatability (single laboratory precision) can be initially based upon one homogeneous sample and is measured by the laboratory developing the method. The repeatability is expressed as standard deviation.
• Limit of Detection (LOD) is a criterion that can be difficult to establish. The detection limit of the method is defined as 3*SD₀, where SD₀ is the value of the standard deviation as the concentration of the analyte approaches 0. The value of SD₀ can be obtained by extrapolation from a plot of standard deviation (y axis) versus concentration (x axis) where three concentrations are analyzed ~ 11 times each that are at the low, mid, and high regions of interest. This determination should be made using a matrix that matches the sample matrix.
• Sensitivity or delta C = 2 (2)^1/2 SD_c, where SD_c is the standard deviation at the mid point of the region of interest. This represents the minimum difference in two samples of concentration C that can be distinguished at the 95% confidence level.
• Limit of Quantitation (LOQ) is defined as 10 SD₀ and will have an uncertainty of ~ 30% at the 95% confidence level.
• Linearity or Range is a property that is between the limit of quantitation and the point where a plot of concentration versus response goes non-linear.

Was this helpful?  Tell a Friend | Let us know