Detection Limits and Uncertainty
The analyst should know the detection limits for all analytes at several possible wavelengths. Typically, these measurements are obtained during the establishment of the analytical instrument's capabilities.
Modern axial view ICP-OES and
ICP-MS instruments are likely to have detection limits under normal sample introduction modes that will meet or exceed the requirements. It is best to not rely upon the limits published by the manufacturer. In addition, the detection limits will be a function of the sample matrix, in both a physical and spectral sense.
A key point involves the analytical blank. Due to numerous contamination issues, the analytical blank often determines the detection limit capabilities. It is best for the analyst to be conservative when noting the detection limits, making sure not to quote capabilities calculated from published data or determinations made under "ideal" conditions. For the less common elements, an estimate of the "real" detection limit would be a factor several times higher than the limit determined under ideal conditions. Thus, elements like Na, Mg, Ca, Fe, Cr, Cu, Zn, Si, Al, Cl, and S may have a detection limit that is significantly larger than expected, due to the analytical blank.
The uncertainty of an analytical measurement is not limited to the measurement precision of the instrument. Rather, it is a statistical sum of the random and systematic errors that are encountered throughout the entire analytical process. The uncertainty is a combination of errors from sampling, storage, weight and volume manipulations, preparation, calibration, and measurement, during which contamination issues play a major role in trace determinations.
The sampling error can be a major source of uncertainty. In many cases, an estimate of the sampling error can be impossible to judge. The initiator should be aware of this fact. In reality, the uncertainty of a trace measurement will not be known until the project is completed. The accuracy of this value will be only as good as the effort made to identify and measure all of the errors encountered during the entire analytical process. If measurements are being made between 3-5 times the detection limit for the less common elements, then an excellent uncertainty would be ± 30-70%.
Constructing the Sampling Plan
After the problem is defined, the planning process can begin. Analytical text books explain that you must consider the sample collection, sample storage, sample preparation, measurement, and reporting, along with any QA/QC requirements. With so many considerations, where should you start?
A synthetic organic chemist will construct a plan by working backwards from the final product. A similar approach may work well for the trace analyst. Start by examining the following basic information:
The analyte(s) of interest.
The required detection limit(s).
The uncertainty requirement(s).
The chemical composition (matrix) of the sample.
The quantity, availability, and history of the sample.
Much of the above list can be determined based on information gathered while defining the problem. In most cases, analytical resources are available in-house to address the problem. For example:
The basic information listed above is sufficient to determine whether publications or information is available in your reference library. Always start with a search of the literature.
The identity and detection limit requirement of each analyte indicates the analyte measurement technique(s) required and the amount of sample required.
The uncertainty requirement indicates the number of measurements, assuming there is sufficient sample available.
The chemical composition of the sample, together with the identity of the analyte(s), indicates possible sample preparation routes.
The identity of the analyte(s), together with the detection limit requirement(s), indicates the degree that contamination issues should be considered. This determines the need for analytical blanks and special apparatus or a clean area / room.
The sample composition indicates potential interference issues.
The sample composition or type indicates the uncertainty to be expected form the sample collection and/or the need to develop a sampling procedure and to determine sampling uncertainty. For example, the sample may be the only "world's supply", negating the need for a sampling procedure.
The estimated sampling uncertainty can be used to define the analytical measurement precision (i.e. -- reducing the analytical error to less than one third of the sampling error serves no purpose).
The basic information can provide the analyst with potential analytical measurement technique(s), suspected interferences, contamination issues, and the number of sample measurements required per determination (measurement refers to a complete analysis including sampling, preparation, instrumental analysis and reporting the final result and uncertainty). At this stage of the planning process, the analyst can determine if a certified reference material (CRM) should be obtained for method validation. In addition, the chemist can approximate the need for analytical reagents and apparatus and/or calibration standards.
Lastly, estimate the time and cost of the project and base your initial approach on these estimates. Remember, there is always the possibility that more than one iteration may be required before an acceptable approach can be developed