Uncertainty trace analysis

LGC - VAM Publications (i) The Fitness for Purpose of Analytical Methods, A Laboratory Guide to Method Validation and Related Topics, (2) Practical Statistics for the Analytical Scientist A Bench Guide By TJ Farrant, (3) Trace Analysis A structured Approach to Obtaining Reliable Results By E Pritchard, (4) Quantifying Uncertainty in Analytical Measurement, and (5) Quality in the Analytical Chemistry Laboratory. LGC/RSC Publications, London, England. 

Ultra-trace analysis, hydrogen peroxide determination, 638 Ultraviolet see UV Uncatalyzed sulfoxidations, 472-4 Uncertainty, analytical methods, 624 UN Environment Programme (UNEP) chemicals safety, 745, 747 SIDS database, 622 UnfunctionaUzed olefins... 

Precision is particularly important when sample preparation is involved. The variability can also affect accuracy. It is well known that reproducibility of an analysis decreases disproportionately with decreasing concentration [2], A typical relationship is shown in Figure 1.4, which shows that the uncertainty in trace analysis increases exponentially compared to the major and minor component analysis. Additional deviations to this curve are expected if sample preparation steps are added to the process. It may be prudent to assume that uncertainty from sample preparation would also increase with decrease in concentration. Generally speaking, analytical... 

Correct results in trace analysis are not only coimected to a change in the scale of determination or use of a more sensitive detector in the final step of a multistep procedure. These aspects must obviously be taken into account, but they do not decide the success of trace analysis. Each step, and practically each function, could introduce a given uncertainty in the final result. Thus, the final result is connected with much larger uncertainty than the result of a simple measurement. The analyst must be aware that the smaller the concentration of analyte, the greater the uncertainty of the final result. This is usually a result of the following factors ... 

The values of errors and uncertainty strongly depend on the level of analyte content (concentration). High values of these parameters, unacceptable in the case of an analyte content at the percentage level, can be satisfying in the case of trace analysis. 

Added to these constraints are the issues raised by the certainty of a result in terms of both analyte identification and quantitation. The uncertainty of a result is dependent on the analyte concentration. In trace analysis it might be argued that the traceability of a method in identifying a substance would have a traceability chain completely different from that of a method which quantitates the same substance. The traceability chain for a method which both quantitates and identifies a substance could be different again. This differentiation is important in many regulatory analyses in which a zero tolerance for a substance has been set. Under these circumstances, only the presence of the substance has to be established. This problem is discussed in more detail later in this paper. 

BCR studies that substances on the market had purity figures of sometimes less than 70 M) (even 50%) where they were stated to be pure at 98% This demonstrates that in organic and organo-metallic trace analysis impurity, and in general uncertainty due the quality of primary calibrants, cannot be neglected and are sources of important bias in measurements. Therefore, it is of importance for analytical chemists to identify reliable suppliers. If necessary, it may be necessary to purify purchased substances. Collaboration in such studies and exchange of substances is one of the most useful outcomes one can get from colleagues. BCR projects have often initiated such relations. 

The real difficulties remain in the determination of U. It is relatively simple to determine the method uncertainty of nondestructive analysis as repeated measurements can be performed on the same sample [39]. It is far more difficult with destructive methods and in particular in organic trace analysis. In the latter case, all the steps in the procedure rarely allow one to achieve a repeatability with a relative standard deviation of less than several percent. The methods often require a large sample intake as samples of a few milligrams are not easy to handle in extraction systems. 

For trace analysis of cations, the technique that introduces the minimum value of uncertainty is anodic stripping voltammetry. The reason is that the concentration and determination steps take place on the same electrode. The lower value of uncertainty is also a function of the use in the technique of a fully computerized instrument. In this case, the computerization of the instrument is of prime importance for decreasing the uncertainty because it assures a high reproducibility of the parameter characteristics for anodic stripping voltammetry. 

Combining appropriate amounts of the stock solutions of analytical and internal standards, and subsequent dilution, can again be done volumetrically or gravimetrically (via a weighed syringe) the choice can only be determined by considerations of fitness for purpose, but most often careful manipulation of standard flasks and pipets (possibly re-calibrated for the purpose by weighings) is adequate for trace analysis (see Section 9.5.4 for details) as other uncertainties can considerably exceed those introduced in preparing the calibration solutions. 

The uncertainty on the result arises from both random and systematic effects but in trace analysis systematic effects largely determine the uncertainty of an analytical result. The. search for and correction of systematic errors is therefore an important responsibility of every trace analyst. Even after correction for systematic errors the uncertainties on there corrections need to be evaluated and included in the overall uncertainty. Failure to correct for systematic errors leads to the considerable scattering frequently observed with collaborative analyses, and ultimately to inaccurate results. The uncertainty on the result increases di.sproportion-ately with decreasing amounts of analyte in the sample. 

In trace analysis it is not sufficient simply to report a level of reproducibility for the actual determination of an analyte. Evaluating the quality of an analysis requires a knowledge of the reproducibility and the uncertainty arising from systematic effects (- Chemometrics). Errors in sampling and/or sample preparation may be orders of magnitude greater than the standard deviation observed in several repetitions of a determination. 

Moser, J., Wegscheider, W., Meisel, T., and Fellner, N. (2003) An uncertainty budget for trace analysis by isotope dilution ICP-MS with proper consideration of correlation. Anal. Bioanal. Chem., 377, 97-110. 

The sources of uncertainty in NAA analysis are well understood, and can be derived in advance, modelled and assessed experimentally. There are two main kinds of interferences in the calculation of trace-element concentrations by INAA. The first one is formation of the same radionuclide from two different elements. Another kind of interference is from two radionuclides having very close y lines. Whenever interferences occur, the radionuclide of interest can be carried through a post-irradiation radiochemical separation without the danger of contamination. 

A logical approach which serves to minimise such uncertainties is the use of a number of distinctly different analytical methods for the determination of each analyte wherein none of the methods would be expected to suffer identical interferences. In this manner, any correspondence observed between the results of different methods implies that a reliable estimate of the true value for the analyte concentration in the sample has been obtained. To this end Sturgeon et al. [21] carried out the analysis of coastal seawater for the above elements using isotope dilution spark source mass spectrometry. GFA-AS, and ICP-ES following trace metal separation-preconcentration (using ion exchange and chelation-solvent extraction), and direct analysis by GFA-AS. These workers discuss analytical advantages inherent in such an approach. 

One common characteristic of many advanced scientific techniques, as indicated in Table 2, is that they are applied at the measurement frontier, where the net signal (S) is comparable to the residual background or blank (B) effect. The problem is compounded because (a) one or a few measurements are generally relied upon to estimate the blank—especially when samples are costly or difficult to obtain, and (b) the uncertainty associated with the observed blank is assumed normal and random and calculated either from counting statistics or replication with just a few degrees of freedom. (The disastrous consequences which may follow such naive faith in the stability of the blank are nowhere better illustrated than in trace chemical analysis, where S B is often the rule [10].) For radioactivity (or mass spectrometric) counting techniques it can be shown that the smallest detectable non-Poisson random error component is approximately 6, where ... 

Although quantification of the elements present in the y spectrum can in theory be achieved from first principles using the equation given above, in practice uncertainties in the neutron capture cross-section and variations in the neutron flux within the reactor mean that it is better to use standards. These standards must be included in each batch of samples irradiated in order to account for variations in neutron flux inside the reactor. For analysis of minor and trace elements calibration is easier than with other analytical methods provided that the major element composition remains reasonably constant, as the y ray intensity is proportional to concentration over a very wide range of concentrations. However, for analysis of major elements, e.g., silver in silver coins, the relationship between intensity and concentration is more complex, due to progressive absorption of neutrons as they pass through the specimen. In such cases y ray intensity will also depend on the thickness of the sample and therefore specialized calibration methods are required (Tite 1972 277). 

---