Achieving Valid Trace Analysis

1.1.1 What is Trace Analysis?

The term 'trace analysis' is widely used to describe the application of analytical chemistry (the measurement of amount of substance) under circumstances where the amount of analyte is very small. As such, its scope is as broad as that of analytical chemistry. The range of inorganic analytes is relatively small and comprises around 100 elements together with organometallic compounds and the common anions. However, in the field of organic chemistry several million compounds are known to exist and many are of interest at trace levels. Furthermore, materials presented for analysis of inorganic or organic species span an enormous range of composition and properties.

In the past, almost all analysis was undertaken using the so-called 'classical' techniques which involved dissolution of the sample, removal of any interfering species by precipitation and/or complexing agents, followed by determination using titrimetry, gravimetry or colorimetry. Such procedures required good manipulative skills and a deep understanding of the basic chemistry involved, even when used to determine relatively high concentrations or amounts of analytes. Many of these 'classical' techniques are capable of trace analysis but each analysis is generally time consuming and difficult. Hence in the heyday of 'classical' analysis the number of routine trace analysis measurements was quite small. However, from around the 1930s a range of physico-chemical tools was developed and applied to the solution of analytical problems. Techniques based on spectroscopy gained quite wide acceptance from the 1930s, particularly for trace element analysis in fields such as metallurgy, and were further developed in the 1950s and 1960s. Similarly, the chromatographic separation of organic compounds was developed in the 1950s and allowed sensitive determination of many important species using spectroscopic, electrochemical, or other detectors.

These developments in instrumental analysis enabled the analyst to routinely determine lower and lower concentrations of analytes and to resolve or separate very complex mixtures. This ability stimulated demands for trace analysis from industry and from those interested in applications such as environmental and consumer protection, forensic science, and clinical analysis. It is probably fair to say that the apparant ease with which these instrumental techniques could detect and measure analytes at ever lower concentrations led to an unwarranted confidence in the ability of the analyst to produce trace analysis data cheaply and reliably. Only as experience with each new or improved technique accumulated was it realized that many such trace analysis applications are reliable only if the instrumental determination is preceded by quite elaborate chemical manipulations to overcome interference effects. This need increases the cost, time, and expertise required and introduces the problems of contamination and loss.

In spite of the fact that the term 'trace analysis' is widely used by analytical chemists and their clients it does not actually have a clear or unambiguous definition. Many analysts would apply the term to determinations made at or below the part per million level, i.e. 1 ppm [equivalent to] 1 µg g-1 [equivalent to] 0.0001%, or 1 mg 1-1 for liquids. Other analysts would define the term more generally as applying to an analysis where the concentration of the analyte is low enough to cause difficulty in obtaining reliable results. This may be caused solely by the low concentration of analyte in the matrix but other factors may also be important. Factors such as analyte losses, contamination, or interference may influence the perceived difficulty of analysis at lower concentrations and it may not be possible, or useful, to assign a numerical limit to trace analysis. Generally, the amount of sample available for analysis is plentiful but in applications such as clinical or forensic analysis there may be only small portions for use. This in turn will limit the mass of analyte presented to the detector, even though the initial concentration might be quite high. All such applications requiring special precautions to be taken are considered to fall within the scope of trace analysis and will be included in this book.

1.1.2 The Importance of Trace Analysis in Today's World

Measurements based on analytical chemistry are important to almost every aspect of daily life. They are critical to the success of many business sectors, the effectiveness of many public services depends on them, and everyone benefits from the use of such measurements to safeguard health, safety, and the quality of the products consumed or used. Some examples of the applications of analytical chemistry are given in Table 1.1.1. Such applications cover an enormous range of concentrations, from major constituents of materials down to contaminants present at parts per billion or below. Nevertheless, it is true to say that an ever increasing proportion of all analytical measurements can be described as trace analysis.

Trace analysis measurements play a key role in many areas of interest to industry and commerce, to governments, and to individuals. For example, the development and production of many new materials, of microelectronic devices, and of safe pesticides has been dependent on the availability of specific trace analysis techniques. Similarly, trace analysis is used in the first instance by governments to set many regulatory limits for purposes such as protection of the environment, or protection of the consumer, or to protect the health and safety of the workforce. Subsequently, trace analysis must be used by both industry and government to monitor or enforce these limits. Trace analysis is also essential to ensure the smooth flow of trade between companies or countries. For example, a manufacturing company purchasing materials or components will need to know that its suppliers are meeting an agreed specification. Obviously, checking such specifications may require a wide variety of physical or chemical measurements but trace analysis data are often vital, particularly with high technology products or materials intended for human consumption or application. Similarly, international trade is subject to extensive controls and regulations, many of which depend on trace analysis data.

1.1.3 Difficulty of Achieving Reliable Trace Analysis

In general, trace analysis is an extremely demanding activity requiring extensive knowledge, skill, and experience. The particular problems presented by trace analysis can be summarized as follows.

(i) The concentration of the analyte to be determined is much lower than that of the other constituents present in the matrix,

(ii) Contamination arising from reagents, apparatus, or the laboratory environment is more critical and may lead to false results.

(iii) Losses of analyte by adsorption, degradation, or during analytical operations are more critical at very low concentrations and may even result in failure to detect substances actually present in the matrix at concentrations well above the anticipated detection limit.

(iv) Constituents of the matrix may interfere with the detection system used, leading to falsely high values, resulting in the need for more extensive purification and/or more selective detectors.

(v) The results obtained with the commonly used instrumental techniques are less precise than those obtained using classical procedures.

(vi) Generally, it is difficult to check the reliability of methods because there are relatively few reference materials available for the enormous range of trace analysis applications.

It must be emphasised that the factors in (i) to (vi) cause problems for the majority of laboratories. Several authors have published papers reporting the results of collaborative studies of trace analysis methods. Frequently the values reported on a supposedly homogeneous and well-characterised material show an enormous range, despite the fact that analysts taking part in such studies are usually experts in the specific application or method and often treat the samples more carefully than most routine determinations. For example, Sherlock et al. published the results of a UK inter-comparison exercise for cadmium and lead in selected foods. The spread of results, from 30 laboratories, covered one and a half orders of magnitude and this paper did much to highlight the lack of agreement between laboratories and the need for quality assurance in analytical laboratories. The poor agreement arose in part because the levels chosen were close to the limits of detection of the routine methods used by many of the laboratories. Nevertheless, this is a common state of affairs for many 'real' samples. Even the results achieved by a sub-group of large, expert laboratories were spread over a much wider range than many analysts would have expected (Table 1.1.2). For lead in liver and cadmium in fish food, for example, the range was greater than 100% of the mean value. At the time the paper was one of the first widely read reports to raise doubts on the reliability of analytical data at low concentrations and to point out that users of analytical data must be aware that 'the customer gets what he pays for'.

Another source of examples highlighting the difficulty of trace analysis is the European Union's 'Community Bureau of Reference' (BCR) which has carried out an extensive programme to prepare certified reference materials. These products are carefully prepared and checked for homogeneity and stability. The analyte concentration is then usually determined by a number of specialist laboratories using several different techniques. In this way a consensus value for the concentration of the analyte is obtained with a statistical estimate of the measurement uncertainty. The results reported by the BCR programme show that initially there is a large spread of results, even from expert laboratories, the situation improves as the study proceeds and analysts have the opportunity to meet to discuss their problems. One such study is shown in Figure 1.1.1. The first collaborative study for the determination of a chlorinated biphenyl in fish oil showed a variation in results from 300 to 1400 ng g-1. In the second study carried out 2 years later, after the group of laboratories had studied the reasons for the discrepancy and consequently removed sources of bias, the variation had been reduced to a range of 300 to 700 ng g-1, with the majority of participants in a range spanning only 100 ng g-1. Both of these studies also show a wide variation of estimated uncertainty between different laboratories.

A graphic illustration of the decreasing reliability of chemical analyses at lower levels was given by Horwitz et al. in 1980. Evaluation of the results from over 200 collaborative studies conducted by the United States Association of Official Analytical Chemists (AOAC), showed empirically that the coefficient of variation between laboratories was a function of the concentration. These studies covered a wide range of analytes and matrices with analyte concentrations ranging from 10% down to 1 ppb (1 ng g-1). The relationship is shown diagrammatically in Figure 1.1.2. Clearly, reproducibility is a major problem in trace analysis. Individual laboratories will be able to achieve better within-laboratory precision than that shown in Figure 1.1.2 but in today's world comparability between laboratories is all important. The data reported in Figure 1.1.2 demonstrate only the achievable precision and not the bias of a method. This latter factor will usually introduce further uncertainty into analytical results.

These examples clearly demonstrate the need for all laboratories to evaluate and improve the quality of analytical data, particularly at trace levels. They also show the benefits that can be achieved when analysts work together to improve methodology and techniques.

1.1.4 How Can the Reliability of Trace Analysis be Improved?

This book highlights areas where problems might occur and possible ways of dealing with them, It is, however, worthwhile to reflect briefly on why, apparently, so many trace analysis measurements go wrong. There may be laboratories that knowingly and unscrupulously produce poor quality data but, in general, this seems an unlikely source of such widespread problems. The laboratories participating in the collaborative studies mentioned earlier knew that their results would be evaluated at length and closely compared with those of other organizations. If anything, they would have carried out their analyses more carefully.

There do, however, seem to be two other much more likely reasons for incorrect results. Firstly, and as mentioned earlier, trace analysis is often extremely difficult. It presents problems which push the abilities of analytical chemists to their limits. An analysis referred to as 'routine' may well be founded on complex chemical principles, requiring difficult practical manipulations and involving measurements using very expensive, state-of-the-art scientific instruments. Yet all of this will frequently be entrusted to a relatively inexperienced operator with little opportunity to turn to more experienced or knowledgeable colleagues for help. Thus it is important to raise the level of understanding of those carrying out trace analysis. In particular, the easy availability of extremely sensitive instruments means that many such measurements are made by personnel who do not consider themselves as 'analysts'. In some cases they may not be chemists or have no relevant scientific training and may regard the problems of trace analysis as solved once the instruction manual of the instrument has been mastered. Hence the reliability of trace analysis measurements should improve if those using these instruments have ready access to clear, understandable advice on the pitfalls of the common techniques and ways to overcome them. This advice needs to be complemented by information on handling and preparing samples for trace analysis prior to the instrumental measurement itself.

The other reasons for poor trace analysis data seem to stem from the economic and other pressures now facing most laboratories. There is the temptation to neglect well-known precautions designed to reduce contamination, or to apply a method to determine analytes at levels below those for which the method has been validated. Carrying out measurements in an area saturated with analyte will inevitably cause errors in trace analysis. The need to deal with large numbers of samples in a short period may lead to mistakes caused by human error or by inexperienced staff. Sample pre-treatment may be undertaken by different analysts from those making the end determination leading to potential problems unless there is close collaboration. These problems are, of course, quite common and not unique to trace analysis. As might be expected, there is also a common approach to avoiding them. This is based on the introduction of quality assurance principles into the workplace. This approach requires agreement on what needs to be done to ensure that the analytical service provided is fit for its intended purpose. It is then possible to set up a quality system which sets out clearly what must be done and makes provision to monitor day-to-day operations to ensure that the specified work has been carried out.

1.1.5 The Purpose and Structure of This Book

The book covers all aspects of trace analysis from receiving the sample at the laboratory to the end determination and suggests ways of dealing with both organic and inorganic analytes in a variety of matrices. The aim in preparing this publication has been to bring together many of the practical considerations which have been found to be important over a wide range of trace analysis applications. Thus it should be useful both to analysts entering the area for the first time and to those undertaking a new type of determination. Equally, analysts who already regularly undertake trace analysis in a specific application area may find the book a useful source of new ideas or alternative approaches. The ultimate aim has been to provide all analysts with at least some of the practical advice which may be available to those fortunate enough to work alongside an experienced and knowledgeable trace analyst to whom they can turn when problems arise. The book does not attempt to describe detailed methods or protocols for particular analytes in specified matrices, although useful references can be found in several of the chapters. Similarly, individual analytical techniques have not been discussed in detail as there are many publications in which this has been done already. Considerable attention is given, however, to guidance on factors that should be taken into account when selecting appropriate methodologies and techniques for a particular application. We have attempted to show the relative merits of many methodologies and techniques; although practical details are constantly changing, the relative positions often remain the same. No doubt areas are omitted and many hints and tips are left out. These can always be added in the next edition, if you let us know.