ANALYST, JUNE 1993, VOL. 118 587 Development of an International Chemical Measurement System" Plenary Lecture Bernard King Laboratory of the Government Chemist, Queens Road, Teddington, Middlesex, UK W7 7 OLY Systems are required to ensure that analyses carried out in different laboratories are comparable, and as the requirements become more stringent the task becomes more difficult. We already have many of the components of an international system, there are, however, gaps and there is no adequate overall framework t o facilitate collaboration. This paper reviews some of the driving forces that will shape future requirements and proposes an outline system covering both conceptual issues and the organizational infrastructure. The following underpinning concepts and requirements are discussed: clearly defined targets in the form of requirement specifications; knowledge of trueness or measurement uncertainty; provision of traceability through an unbroken chain of calibration t o primary standards; and the use of quality systems t o provide transparency.Some of the key organizations operating at the UK, European and international levels are considered in terms of their contribution t o learned and professional activities, metrology, the provision of a quality focus, accreditation and the production of standards. The main new requirements are: an international organization t o act as a focus for analytical quality and more strategic planning of collaborative work programmes. Keywords: Measurement system; quality; measurement uncertainty; european organizations; traceability In his opening address at the SAC '92 Conference, the Government Chemist, Richard Worswick, spoke about the historic developments and the forces shaping the future of analytical chemistry.In this paper it is proposed that if we are to respond to the challenges of the 21st century we will need to develop a more formal and comprehensive international chemical measurement system. Although it is not formally recognized or even well known in analytical circles, we already have national and international systems but these are more suited to physical measurements than chemical analysis. Nonetheless, we already have many of the necessary com- ponents, ranging from the SI unit, the mole, to chemical reference materials and accreditation schemes complying with international quality standards.Put simply, the system, consists of activities that help eiisure that chemical analysis is valid and that an analysis carried out in one laboratory is comparable with similar analyses carried out in other parts of the world. In recent years, there has been growing recognition that we need to do more to ensure the quality of our work and a great deal has and is being done to develop some of the components of the measurement system. What we need to do now is to fit the components together to form a coherent system that will be adequate to serve our future needs. This paper discusses recent developments and indicates where more action is required. An attempt is also made to describe an overall structure that ties the strands together.My specification for the system is that it should address both conceptual issues such as measurement traceability and the organizational infrastructure, in terms of geographical region, type of activity and sectoral interest. The system must also be based on the quality principle 'fitness for purpose' and be sufficiently flexible to cope with state-of-the-art measurements and simple routine analysis. Finally, it must place emphasis on international rather than national activities and structures. Driving Forces The driving forces to which we need to respond are increasing complexity of requirement, automation/remote sensing, * Presented at SAC '92, an International Conference on Analytical Chemistry, Reading, UK, September 20-26, 1992.demand for quality and value for money, and international- ism. Society will continue to demand more analyses and more complex analysis. Advances in technology and information technology will help us to respond to the demands by facilitating automation, on-line analysis and remote sensing. However, these developments will further increase customer expectations for instant on-the-spot analyses. These develop- ments represent opportunities and threats to the analytical chemist. One issue will be our ability to provide quality data at an acceptable price. Customers will also increasingly expect a better dialogue with the analyst to agree the requirement and visible evidence that the data and their interpretation are valid. The simultaneous pressures on quality and price will require economies of scale that will lead to greater specializa- tion and present business difficulties for some laboratories.The final driving force is internationalism. The single European market is part of a continuing trend to eliminate barriers to international trade. As our world effectively shrinks we need to harmonize our systems to ensure that analyses carried out in different countries are comparable. The control of disease and of environmental pollution are further examples of issues that are critically dependent on sound chemical analysis and international cooperation. All of these issues will have a major impact on our work and we need to reassess and, where necessary, upgrade our support systems. Historically, analytical chemistry has served society well but there are clear signs that some chemical analyses are not fit for purpose.Unless we act now the situation will become worse and could lead to loss of public confidence in our work. The Quality Problem The quality problem can be illustrated by the results of interlaboratory comparisons such as the Community Bureau of Reference (BCR) study of trace toxic elements in milk powder. The summary data in Table 1 show that the spread of results obtained by laboratories across Europe was initially very large and, incidentally, it was much greater than expected. However, when the scientists concerned collabor- ated it proved possible to identify the causes of the differences and subsequently to obtain good agreement.588 ANALYST, JUNE 1993, VOL.118 The quality problem is being increasingly recognized and much is being done to improve matters. The challenge before us is to build a system that is sophisticated enough and robust enough to serve the needs of the 21st century. Key Organizations Some of the organizations that contribute to the international chemical measurement system are listed in Table 2. These organizations cover different activities such as professional matters or accreditation and also different geographical regions ranging from national groups to European regional groups and in some cases the interest is world-wide. The European and UK organizations can be replaced by other equivalent national and regional organizations. Also, a number of other organizations could be added to the table, particularly if it were to be extended to include sectoral activities such as the environment or health.The learned and professional societies provide the bed-rock of technical support. They are showing increasing interest in quality matters, particularly in areas such as training and technology transfer. There are growing national and regional programmes, such as the Valid Analytical Measurement (VAM) programme in the UK and the BCR programme in Europe. Focus groups sych as the Chemical Measurement Advisory Committee (CHEMAC) in the UK and the European Focus for Analy- tical Chemistry (EURACHEM) in Europe are helping to coordinate activities but as yet there is no truly international organization covering all aspects of the chemical measurement system.Accreditation is reasonably well served in that we have internationally agreed quality standards and both national bodies and international coordination in place. The challenges here are to rationalize the requirements of the different schemes and to develop criteria that cover both routine and more research-based work. Standards making is Table 1 Ranges of results (ng g-’) for trace elements in milk First Second Element intercomparison intercomparison Cd 0.4-4500 1-5.6 Hg 0.6-42 0.73-1.27 Pb 68-5500 92.4-1 12.5 c u 470-9257 475-700 Table 2 Some key organizations involved in the international chemical measurement system Area of activity UIC Europe? International* professional RSC FECS IUPAC Learned and Metrology VAM BCR BIPM Quality focus CHEMAC EURACHEM ? Accreditation NAMAS WELAC ILAC Standards BSI CEN I S 0 * RSC = The Royal Society of Chemistry; VAM = Valid Analytical Measurements; CHEMAC = Chemical Measurement Advisory Committee; NAMAS = National Measurement Accreditation Ser- vice; BSI = British Standards Institution.t FECS = Federation of European Chemical Societies; BCR = Community Bureau of Reference; EURACHEM = A European Focus for Analytical Chemistry; WELAC = Western European Laboratory Cooperation; CEN = European Committee for Standard- ization. t IUPAC = International Union of Pure and Applied Chemistry; BIPM = International Bureau of Weights and Measures; ILAC = International Laboratory Accreditation Conference; I S 0 = Inter- national Organization for Standardization. again well served by organizations. What is required is a reduction in duplication and a harmonized approach that is more transparent.The geographical cooperation within specific activities, such as standards making, is generally effective but communication between the activity groups is less well developed and this is an important function for groups such as CHEMAC, EURA- CHEM and a possible new international organization. A serious limitation to the effectiveness of many of these groups is the fact that much of the work is unfunded and carried out in the margins by enthusiasts. One of the keys to progress is the provision of national and international funding and where appropriate top-down strategic planning of the work. Before expanding on some of the recent developments, the conceptual issues involved in the measurement system will be discussed, Underpinning Concepts For the measurement system to work effectively we need a clear understanding of the conceptual issues and how they fit together.Targets First we need more clearly defined targets in the form of requirement specifications, study designs and sampling proto- cols. When engineers build a bridge or construct a car, they work to detailed specifications and plans and by comparing what is produced with what was specified it is possible to assess quality. All too often the specification for chemical analysis is ‘what is this?’ or ‘how much of X is in this sample?’. Our approach is too simplistic and we must devise a protocol that helps customers specify their requirements in detailed terms appropriate to the task.For example, accuracy is an expensive commodity and too much can be almost as bad as too little. With more complex investigations, a study design is needed to evaluate the feasibility of options and prepare a work plan. The answer will often depend on the way the sample is taken and how it is processed. Sampling is often considered to be the weakest link in the quality chain and is often the major contribution to measurement uncertainty. Studies of surfaces, the examination of forensic specimens and the characteriza- tion of contaminated land, for example, present special sampling problems. At the Laboratory of the Government Chemist we have long recognized the importace of sampling and it is therefore particularly pleasing that we have recently won a contract from the UK government to help improve this part of the measurement system.Trueness For a measurement to be of value we need to know how reliable or true it is. The traditional approach has been to determine the precision by carrying out repeat measurements. Unfortunately, this approach fails to take account of some systematic errors and as a result it is common to overstate the accuracy or trueness of measurement data. An alternative approach, which is used by physicists and is beginning to gain acceptance with chemists, is the concept of measurement uncertainty.’ Uncertainty is an estimate of what the error might be and involves professional judgement. Another definition of measuremcnt uncertainty is what is unknown about the error. Some of the key sources of error in chemical analyses are listed in Table 3.The assessment and control of thcse errors requires detailed validation of each step of the process. In some cases errors can be eliminated or corrected for, but there will always be some uncertainty concerning each source of error and this con- tributes to the uncertainty of the measurement.ANALYST, JUNE 1993, VOL. 118 589 Table 3 Sources of error in chemical analysis Activity Errors Sampling Heterogeneity, decay/loss, contamination Calibration Analysis Certified value, mismatch between sample and Weighing, dissolution, ashing, digestion, dilution, standard separation, derivatization, contamination, chemical interference, measurement, data manipulation, reporting The overall uncertainty U is the root mean square of the sum of the component uncertainties ( U1-U,): u = (up + u,2 + u32 + .. . .)& where U,-U,, are estimated as standard deviations. Some errors will be of a random or pseudo-random nature and can be estimated from the standard deviation of repeat measurements. Others will be systematic in nature, F.g., incomplete extraction or peak overlap. It is sometimes possible to assess the uncertainty experimentally, for ex- ample, different extraction procedures can be used and spiking can often help assess the level of recovery; sometimes it will be necessary to rely on professional judgement. Knowledge of bias can also be obtained from the use of alternative, preferably primary, methods and the use of certified reference materials when these are available.Of course, known systematic bias can be corrected for provided that the uncertainty associated with the bias is significantly less than the bias. In practice, only two or three factors can be expected to dominate the overall uncertainty because in the root mean square equation small components make negligible contributions to the answer. The overall uncertainty should be expressed as a band describing the range within which the true value lies with a certain probability. The magnitude of acceptable uncertainty will vary with the application. If the estimated uncertainty is too large for a particular purpose, it is necessary to eliminate some of the sources. This approach is relatively new but there is now a draft International Organization for Standardization (ISO) guide1 that explains the underlying theory and indicates how uncertainties can be assessed in different fields of measurement.At the Laboratory of the Government Chemist we have a project that is attempting to assess the overall uncertainty of some typical analytical measurements. We hope to be able to publish some fully worked examples that will help others assess their measurement uncertainties. Proper attention to measurement uncertainty, in addition to providing technical benefit, will provide market differentia- tion between high- and low-grade results. Traceability Traceability exists when there is an unbroken chain of calibration linking measurements made in one laboratory with measurements made in other places. Also, there should be an unbroken chain linking measurements made at the working level to primary standards.The links in the chemical traceability chain are: SI units such as the kilogram and mole; primary standards such as the kilogram; relative atomic masses; pure chemical reference materials; primary methods; matrix reference materials; secondary methods and reference materials; and working methods and reference materials. The custodian of the SI units and certain primary standards is the International Bureau of Weights and Measures (BIPM). Although there are primary standards for physical measures such as mass, length and time, there is no primary standard for the chemical measure, the mole, and that is one of our fundamental problems. Because of the vast number of different types of chemical analyses it is not feasible, of course, to have a single chemical primary standard, but nonetheless we need to establish a link between working analysis and primary standards.This link is partly provided by the relative atomic masses, which relate the mole to the kilogram. Unfortunately, chemical factors such as matrix effects constitute a break in the traceability chain and therefore chemical standards or reference materials are also required. Chemical standards in turn must be characterized and certified by chemical analysis, thus creating a closed circle. To break out of the circle we need absolute or primary methods. Isotope dilution mass spectrometry, which relies on measuring only isotopic ratios, a unitless measure, and mass is an example of such a method.In principle this process can be continued down to the secondary standards and working levels, but in practice there are many missing links in the traceability chain. Although there are about 20000 chemical reference mater- ials available world-wide, these are still not enough to calibrate the vast range of chemical measurements. Also, there is no system for distinguishing the top level of primary reference materials from lesser secondary and working level materials. It is hoped that two recent developments will help improve matters. First, the BIPM has established a working group to examine what can be done to improve the traceability of chemical analysis. This group is examining the feasibility of establishing a small number of primary chemical reference materials and reference methods to form the top of a chemical traceability system.Although it will take some years before any tangible benefit can be achieved, it is an essential part of the process of improving the quality of chemical analysis. Another recent development is the establishment of an IS0 working group to prepare a guide for the accreditation of reference material producers. Although this will not guaran- tee the reliability of products, it will help raise standards and could be a first step towards the international certification of reference materials themselves. Transparency Transparency exists when there are systems in place that ensure the reliability of data and provide the customer and user with clear evidence of reliability. Having a formal quality assurance (QA) system and preferably accreditation goes a long way to achieving transparency.There are three main types of quality standards: (i) I S 0 Guide 25; Euronorm (EN) 45000; National Measurement Accreditation Service (NAMAS) MlO for testing and calibration laboratories; (ii) I S 0 9000; EN 29000; British Standard (BS) 5750 for manufacturing and service industries; and (iii) Good Labora- tory Practice (GLP) for non-clinical testing of chemicals for health and environmental purposes. These standards are published under a variety of numbers by both national and international standards bodies. Although there are small, but for some purposes significant, differences between the standards, the underlying principles and require- ments are broadly the same. The standards are interpreted in different ways, however, by the accreditation bodies who place different emphasis on different aspects of the require- ments. For example, in the UK NAMAS prefer to accredit laboratories for specific tests and their system does not yet fit comfortably with non-routine analysis or research and devel- opment (R & D) work.In GLP, particular emphasis is placed on study design, record keeping and the role of the Study Director. On the other hand, BS 5750 takes a broader approach to quality but applies to all activities within the laboratory. Organizations are required to define their quality objectives and then to install and monitor systems that ensure they are achieved. In the UK, over 300 laboratories are NAMAS accredited for some aspect of chemical analysis and many other countries590 ANALYST, JUNE 1993, VOL.118 are progressing in a similar way in response to customer demand. There are two major problems to address, however. First, the multiplicity of standards and approaches need to be harmonized into a single system that covers all aspects of chemical analysis and that is internationally recognized. Second, we need to find ways of persuading the majority of laboratories who have no formal QA to improve their practices. A significant step towards the harmonization of accredita- tion standards is the preparation of a detailed technical guide describing best practice. This document is at an advanced draft stage and has been prepared by a joint working group of the Western European Laboratory Cooperation (WELAC) and EURACHEM.Although prepared specifically for the EN 45000 standard, the guide is equally applicable to other systems. Transparency is also provided by participation in profi- ciency testing schemes in the form of routine interlaboratory comparison exercises. A growing number of schemes are now available and some operate on an international basis. The burden of QA on laboratories is considerable and we must ensure that a cost-effective balance is achieved. It is important that proficiency testing is kept to a minimum and that it is effectively organized. An IS0 protocol that describes good practice in the organization of proficiency testing schemes is in the course of preparation and this will help to ensure that schemes comply with best practice. The following section returns to organizational matters and explains some of the developments that have occurred in recent years, starting with the UK. Developments in the UK In 1989, a government White Paper was published entitled ‘Measuring up to the Competition.’ It set out the govern- ment’s policy and plans for measurement support and announced a new programme to help improve the validity of analytical measurement.The first VAM programme ran for 3 years and in 1991 it was extended for a further 3 years. The Laboratory of the Government Chemist is the main contractor but a number of other organizations are also involved. There are 11 projects and the work is of three types: promotion and technology transfer; development of the organizational infra- structure; and technical projects concerned with topics such as trace analysis.The largest project is ‘Calibration and Tracea- bility,’ which covers work on reference materials and the development of primary methods for their certification. Other technical projects such as trace analysis, sampling and statistics range from R & D concerned with solving measure- ment-related problems to the development of protocols that describe good practice. The work is always aimed at providing analytical chemists with the basic tools needed to ensure valid measurement. The government spending on the VAM pro- gramme represents the core of a larger activity supported by other institutes and industry. A great deal of emphasis is being placed on promotion and technology transfer.These activities have two aims. The first is to raise awareness of the importance of valid analytical measurement, particularly with the customers of measure- ments. Only when the customers recognize the importance of quality will we be able to make real progress. The second aim is to communicate new developments and good practices that already exist in particular sectors to the broader analytical community. The second major thrust of the VAM programme is to develop the organizational infrastructure both in the UK and internationally; CHEMAC has been established to provide a UK focus for analytical quality and to provide a mechanism for input to international affair$. Senior level representatives from government, industry, academe and professional organ- izations such as The Royal Socie1,T of Chemistry ensure effective cooperation between the many interested groups and oversee the activities of working groups on reference mater- ials, proficiency testing, and education and training.European Scene There is a great deal of new European activity largely stimulated by the drive to complete the single European market. Of the many organizations there are two that are central to the harmonization of analytical measurement practice. The BCR organizes and helps fund projects needed to solve community-wide measurement problems. Much of the work is concerned with the organization of collaborative projects that lead to the certification of reference materials. A recent evaluation of the BCR programme by a group led by Dr.T. Quinn, the Director of BTPM, concluded that much good work has been done but the resources were insufficient to tackle the task. It is recommended that future programmes are much larger and that a strategic plan be developed to ensure that maximum benefit is obtained. The BCR is central to improving measurement quality in Europe and many of us look forward to it playing a larger role. EURACHEM There is also a need for a broadly based focus for analytical quality matters in Europe and this is provided by EURA- CHEM. Since EURACHEM was established in 1989 in Frankfurt, it now has members from 18 countries. The aims are to provide a framework for cooperation between chemical laboratories across Europe and to promote good analytical quality practices.Membership is open to European Community and Euro- pean Free Trade Association countries and observer status can be afforded to other countries in Europe and to European- based organizations such as BCR and the Federation of European Chemical Societies. The main work of the organiza- tion will increasingly be done by specialist working groups such as the Education and Training Working Group. Some of the activities undertaken by EURACHEM during the last two years are organization of collaborative projects, preparation of guide for the accreditation of chemical laboratories, organization of workshops for specialists from across Europe on education and training in analytical QA and traceability of chemical measurements, preparation of a protocol on profi- ciency testing, and organization of symposium on Good Automated Laboratory Practice.One of the roles of EURACHEM is to provide liaison between horizontal activities such as metrology, accreditation and standards making. To complete the network, member countries are encour- aged to establish national EURACHEM groups to provide 2- way dialogue between European-level activities and the broader community of working analysts. The aims are to identify priority issues by consultation at the working level, to identify partners for collaborative projects and to keep analytical chemists informed about developments. International Organizations Analytical chemistry is well served by international organiza- tions except that there is no single organization to act as a focus for all the various standards that go to make up analytical quality.An international meeting, in Atlanta, GA, USA, ir March 1993 was planned in order to establish ways of improving collaboration between the many organizations and countries that are active in the field. It was expected that either a new organization would be formed or an existing organiza- tion would expand to provide a focus.ANALYST, JUNE 1993, VOL. 118 59 1 Sectors The final aspect of the infrastructure matrix is the sectoral one. There are some activities such as high-level metrology or the operation of accreditation schemes that are generic in that they serve all or a number of sectors. Other activities are of relevance to specific sectors such as the pharmaceutical industry or forensic science. This aspect of the infrastructure can be viewed either as a sphere with generic activities at the core or as a metrology pyramid with generic activities at the apex. Either way, it is important to see all the activities as part of an over-all system so that sectors learn from each other and duplication of effort is minimized. Conclusion This paper has examined a number of areas of activity and indicated how they fit into an over-all measurement system. To summarize, we need: ( i ) greater clarity about underpinning concepts; (ii) an international organization to act as a focus for analytical quality; (iii) more strategic planning and some coordination of work programmes; and ( i v ) governments and industry to be prepared to pay for quality. There is a good deal of activity and analytical chemists are working hard to respond to the needs of society. The author is confident that with the help of our customers we can rise to the challenge of the 21st century. Reference 1 Guide to the Expression of Uncertainty In Measurement, IS01 TAG 4lWG3, International Organization for Standardization, Geneva, June 1992. Paper 2105557K Received October 19, 1992