CHAPTER 4 ON THE IMPORTANCE OF METROLOGY IN CHEMISTRY 1

E.Bulska1,2 and P.D.P. Taylor2 Faculty of Chemistry, University of Warsaw, Pasteura 1, 02-093 Warszawa, Poland 2 EC-JRC Institute for Reference Materials and Measurements, Retiesweg, B-2440, Geel, Belgium

ABSTRACT With growing globalisation of various human activities, also the results of chemical measurements become directly or indirectly involve in this process. There is no doubt that quality of chemical measurements is an important issue in modern society influencing quality of life and border-crossing trade. On an international scale, the world of chemical measurements is undergoing major changes. Over the last decade, initiatives have been taken at the international level and across measurement sectors to ensure that measurement science issues are applied in a systematic way. This is done to improve the quality of chemical measurement results and thus make them acceptable everywhere. In the past two major routes towards quality in chemical measurements have been applied: use of quality management systems and use of accreditation. Only recently, have the principles of measurement science (metrology) received the attention they should. This does not replace the need for most aspects of QA (Quality Assurance), but complements this, i.e. bringing a solid foundation to build on. Firstly applied in physics, they are now recognised as a necessary tool in chemistry as well.

Chapter 4

1 INTRODUCTION The growing demand of the public for reliability of the measurement results in analytical chemistry has been recognised over the last decade [1]. Following the paper by de Bievre [2] the importance of this can be demonstrated by the following examples: (i) In a food sector the quality of products often required to establish the content of toxic substances in order not to exceed the legal limits. So the results of the chemical measurement should allows to answer the question: Have the formal limits of particular toxic substance if food or drink been exceeded (e.g. content of Cd in milk) ? (ii) In a clinical sector the diagnosis is often based on the results of the content of particular substance in blood or urine. So the results of the chemical measurements should allowed to answer the question: Have correct decision been made on whether a patient must go on medication or just on a diet (e.g. cholesterol in blood) ? (iii) In environmental sector the results influence the value of the particular region in respect of living conditions or tourists attraction. So the results of the chemical measurements should allow answering the question: Was the reduction in value of land reasonably estimated (e.g. because of the toxic pollutant in the soil) ? Metrology is a science of measurements and obviously, measurements of different kinds are carried out in analytical and bioanalytical chemistry. Chemical measurements are essential in different fields (environmental, geology, medicine, biology etc.). Important decision are often based on those (e.g. either food can be eaten, goods can be sold, patient should be treated), in support of legislation (related to health care, trade), production process and social problems [3]. There is strong evidence from different ILC’s (Interlaboratory Comparisons) that the quality of such measurements results are still unsatisfactory, probably because the metrological aspects of those measurements is insufficiently stressed [4]. The international today’s society relies on a proper measurement infrastructure, which in much extend depend on the properly trained people. This requires the knowledge dissemination, concerning the chemical measurements based on the metrological principles, which are considered to be most important in this respect [5]. Still it is beliefs in various sectors that the implementation of metrology in analytical chemistry means mainly the replacement of old terms with new ones, for example the uncertainty replace the ‘precision’, and traceability replace the ‘accuracy’. It should be therefore stressed that it is not just a replacement of wording but this cover the major changes in general concept of understanding the measurement process. The proper understanding of the metrological principle is also required by those who need to work under the requirement of ISO 17025 [6]. This chapter contains a set of modules covering fundamental information concerning the principle of metrology in chemistry. This covers a general introduction to metrology in chemistry, uncertainty and traceability of measurement results and the issue of validation of the measurement procedure as well as statistical tools used for uncertainty evaluation are covered respectively. Also the topic related to the use of CRM (Certified Reference Materials), types and aims of ILC’s (Interlaboratory Comparisons). As it was already stressed above, today’s society relies on a proper measurement infrastructure, e.g. to realise the international trade, to implement regulation, to guarantee consumer protection and at least but not least to support scientific research. Key players in such an infrastructure are measurement service provides (national metrology institutes, national and sector reference laboratories, control laboratories etc), 43

Chapter 4

national accreditation bodies and organisation responsible for education and training [7]. This is realised via the European Programm TrainMiC (Training in Metrology in Chemistry). The trainMiC (www.trainmic.org) platform intends to be open to all organisations and people accepting basic principles of the metrology [8]. The status of the metrology in chemistry in Poland is described in the European Commission Report EUR 199915 EN [9]. This Report describes the activity performed by respective governmental bodies: Central Office of Measurements; Polish Centre for Accreditation; Polish Committee for Standardisation as well as various activities performed under the responsibility of the ministries of the: (i) Environment; (ii) Agriculture and Rural Development; (iii) Health; (iii) Economy; (iv) Labour and Social Affairs. Also the activities of non-governmental bodies (Committee of Analytical Chemistry of Polish Academy of Sciences; POLLAB; REFMAT) as well as selected education activity were described. This exemplified in direct way how important is the metrological issue in the activity of our society. The important conclusion form the Status Report is that in Poland always the significant activity in respect of organisation and/or participation in interlaboratories comparisons as well as in production of reference materials was perform over the many years. Several examples are listed in the Appendix of that Status Report. 2 METROLOGY IN CHEMISTRY The aim of this chapter is to discusses the topic of what is metrology, why it is needed as well as what is its position in analytical chemistry. The similarities and differences between the metrology in physics and in chemistry will be stressed by focusing on the fact that in physical measurement the issue is to compare quantities (e.g. lengths of different tables) traceable to a unit (e.g. metre) as in chemical measurement the issue is to compare an amounts of analyte (e.g. content of DDT in meat) traceable to a unit (e.g. mol/kg). As a consequence of this the major impact in physical measurements comes from the calibration of instrument, where as in chemical measurement additionally to that, the measurement procedure calibration should be considered. According to the metrological principles reliable measurements depend on having defined standards for analytes, demonstrable traceability of results to the defined standards and an understanding of the uncertainties of those processes. In trading relationships, both the sides often repeat measurements and then, regulatory agencies usually required their own independent check. This replication of effort well reflects the current inability of chemical measurement to produce consistent results over distance and time. Metrology has been developed from physical measurement and emphasises results traceable to defined reference standards, normally the International System of Units (SI), and fully analysed uncertainty budgets based on the processes set out in the Guide to the Expression of the Uncertainty of Measurement (GUM) [10]. This process involved identifying each component of the measurement that contributes to uncertainty, estimating the contribution of each component of uncertainty, than combining these estimations to calculate the total uncertainty. Much of the improvement in consistency of physical measurements has been achieved by use of the uncertainty budget to better define and controls the measurement environment. The situation with respect of chemical measurements is much more complex and difficult. The comparison of the application of metrology in physic and chemistry is summarised in Table 1.

44

Chapter 4

TABLE 1. The comparison of the metrological issue in physical and chemical measurements

Metrology in Physics vs. Chemistry Physics Chemistry Measurement Units Influence by.. Major impact Dependent on.. Example

Comparing a quantity; e.g. temperature m, s, K Often relies on direct measurements Equipment calibration

Comparing a quantity of analyte; e.g. DDT in milk mol/kg, mg/kg Various factors affect the results

Chemical treatment (e.g. extraction, digestion); reference materials used; …and equipment calibration To large extend “sample Strongly “sample dependent” independent” Length of the table Concentration of Pb in: Seawater; Soils; Blood etc.

It is clear that chemical measurement has a fundamental difference from physical measurement in that it does not take place under controlled and defined conditions. Almost always, the primary objective of chemical measurement is to determine the amount of components of interest, not the total composition of the samples. Total composition will almost always remain unknown and therefore the total environment under which the measurement is taking place cannot be defined and controlled. Unknowns will always increase the uncertainty associated with any measurement. In the last years much effort has been applied to introduce the metrology concepts of physical measurement into the chemical measurement. For example: - The Bureau International des Poids et Mesures (BIPM) has put in place a consultative committee, the Consultative Committee on the Quality of Material (CCQM), to strengthen the relationship of chemical measurements to its SI unit, the mole (www.bipm.fr). - EURACHEM and CITAC have developed a guide for quantifying uncertainty in chemical analysis based on metrological principles and GUM to quantifying uncertainty of measurement [11]. - ISO/IEC 17025:1999 is replacing ISO 25 guide as the standard against which laboratories are accredited and supports these moves by having an increase emphasis on this metrological approach [6]. In every days laboratory performance a chemical measurements depends on a combination of the separation of the sub-samples from the material intend to be analysed, physical measurements as well as chemical separation of the species of interest. Therefore, understanding of the uncertainty of the chemical measurements could not be possible without an understanding of the whole process. This also leads to the fact that laboratory should document the whole procedure in a transparent way. However, it is still possible to find the ‘traditional;’ beliefs that: The result is correct, but there is no need to show why; It is not necessary to state and demonstrate traceability; It is not possible to write model equation; The smaller the number behind ± the better laboratory works; I did this for long time and I know my business.

45

Chapter 4

They are several evidence (e.g. Interlaboratory Comparisons results) that claim is not the same as demonstration proofs [4]. Therefore a lot of attempt has been done in order to shift the beliefs of analytical chemists into that based on the use of the metrology. The way to improve the understanding of the advantages of the metrology when applied into the chemical measurements can be summarise as follows: Truth-value: the available informations are always limited; the ‘truth value’ only exists theoretically, as it can only be approximated; Realism: just do the best what you can in available infrastructure and conditions; accept that it will never be perfect; Transparency: document your work in open way, leaving nothing out; Critical review: there are never problems, unless you look critically; Communication: it is necessary and strongly recommended to use standardised/unified language and practice across discipline and sectors. The implementation of the metrology in chemical measurements, which is nowadays strongly stressed in the ISO/IEC 17025:1999 leads to the points, which always have to taken into consideration: 1. Choose a correct measurement procedure – look for validation and confirm that; 2. Described correctly the measurement procedure (measurement equation); 3. State reference to which results are traceable and demonstrate it; 4. Make an evaluation of the uncertainty of the results; 5. Choose suitable Certified Reference Materials (CRMs) or reference standards and use them correctly. 3 TRACEABILITY IN CHEMICAL MEASUREMENTS Good quality of analytical results is essential when important decisions have to be made. It is well accepted that a key property of a reliable result is its traceability to stated reference. The ability to compare results wherever they originate is an important metrological principle. Traceability is the principal tool to test established comparability of the results. This is achieved by linking the individual result of measurement to a common, tabled reference or standard. Therefore the results can be compared through their relation to that reference. This strategy of linking results to a reference is termed ‘traceability’. The International System of Units (SI) is at the top of the system. According the VIM [12] definition (VIM, 6.10) “traceability is a property of the result of a measurement or the value of a standard whereby it can be related to stated reference, usually national or international standards, through an unbroken chain of comparisons all having stated uncertainties”. A measurement is a process, in the course of which the measurand is compared to a standard. For practical measurements, usually a working standard, not a primary standard is used. To state the uncertainty of the measurement results, the uncertainty of the value assigned to the working standard must be known. In analytical chemistry, traceability of measurement results to SI units is not always possible and the traceability hierarchy ends below the level of the SI units. Even in the case of reference materials, when the values are fixed by mutual agreement, the comparability of measurement results is limited.

46

Chapter 4

TABLE 2. Calibration hierarchy Metrological traceability

Standards

Institution

Uncertainty

Primary std International std. BIPM*) National std. NMI**) Reference std. Accredited Calib. Labs Transfer std. Company Travelling std. Calibration Centre Working std. Test labs *) Bureau International des Poids et Mesures **) National Metrology Institutes Traceability in chemistry is difficult because of the proper definition of the measurand; matrix interferences; difference in composition and inhomogenity of the samples; instability; complex sample treatment required; reliable results. Incorporating traceability to the mole and uncertainty budgets into chemical analysis is more complex than is their application to physical measurement. However the traceability to the same stated reference is of essential importance for comparability of the results. Meaningful comparison between measurements is only possible of the results is expressed in the same units (measurement scale). Mostly used units belongs to the SI system (m, kg, s, A, K, mol) and can be used separately or being their combination. But also traceability can be perform to best internationally agreed reference scale (if no SI) such us: • delta scale for isotopic measurements; • pH scale; • scale of octane numbers for petroleum fuel. Several steps should be used in order to establish and demonstrate traceability in chemical measurements. This can be undertake via: (i) specifying the measurand and model equation; (ii) choosing the suitable measurement procedure; (iii) demonstration through validation that the calculation and measurement conditions include all the influence quantities that affect significantly the result; (iv) choosing reference standard; (v) estimation uncertainty associated with the measurement. To conclude it is clear that every link in the traceability chain must consist of comparisons of an unknown value with a known value. In chemical measurements this has to be done during the validation of the measurement procedure. The stated reference might be a (SI) unit or a conventional reference scale (e.g. pH;delta scale for isotopic measurements; the scale of octane numbers for petroleum fuel). A traceability chain should be designed and then demonstrated with reliable uncertainty. In practice, traceability in chemicalmeasurements is established either via a link to values obtained by reference measurements or via link to values carried by reference standards, which are themselves linked to values obtained by reference measurements.

47

Chapter 4

4 UNCERTAINTY OF MEASUREMENT RESULTS Uncertainty is a metrological term, which is defined as follows: parameter, associated with the result of a measurement that characterised the dispersion of the values that could be reasonably is attributed to the measured. Knowledge of the uncertainty of measurement results is fundamentally important for laboratories, their clients and all institutions using these results for comparative purposes. Competent laboratories known the performance of their testing methods and the uncertainty associated with the results. The level of uncertainty that is accepted has to be decided on the basis of fitness for purposes, the decision having been reached in consultation with the client. Sometimes a large uncertainty may be acceptable; sometimes a small uncertainty is required. There are lots of mysteries about the uncertainty of results in the analytical chemistry society. Also in most academic curricula of analytical chemistry, the random and systematic error concept is still considered to be useful in order to evaluate the quality of the measurement result. In most cases an uncertainty is considered to be a very complicated (complex) approach. “It has to be done because it is required.” Indeed, the uncertainty evaluation is required by the ISO 17025 but it is also a perfect and in fact simply way to improve the knowledge about the measurement procedure used. The analytical chemist should be convinced that evaluating uncertainty of the measurement results yields an improvement of the quality of a measurement procedure. Repeating the measurement 2, 10 or 100 times do not give all information to have reliable results!

Measurement results are reliable only if their uncertainty is known and quantified.

There are several reasons why the uncertainty become so important: 1. Firstly because it is required by ISO/IEC 17025:1999 and for accreditation; 2. The uncertainty of the result demonstrates the quality of the measurements; 3. It improve the knowledge about the measured 4. Inside laboratory the uncertainty can be used in order to document in transparent way the measurement procedure; 5. For the end-user the uncertainty give the result with proper confidence; 6. The uncertainty allows the comparison of the results. It is important to understand that a well-documented uncertainty statement underpins the results and provides transparency of the applied procedure. This could be the best defence in discussion, when the comparison of the results is required as well as when the decision around the legal limits have to be made. The proper evaluation of the measurement uncertainty of the result in routine laboratories is getting more attention. It is generally accepted that the use of analytical methods requires estimating the measurement uncertainty in order to compare the results with the confidence requirement. Well-characterised uncertainties are also fundamental to the implementation of traceability. The “Guide to the Expression of Uncertainty in Measurement” (GUM) published by ISO [10] established general rules for evaluation and expressing uncertainty for various kind of measurements. This approaches requires the identification of all possible sources of uncertainty associated with the applied procedure; the estimation of their magnitude either from experimental or published data; finally the combination of all individuals to give standard and

48

Chapter 4

expanded uncertainties for the whole measurement procedure. It is commonly accepted that uncertainty of measurement results evaluation according to the GUM is a useful and acceptable concept to evaluate results of a measurement. It allows other (e.g. assessors) to understand what and how things were done. It allows also the analyte to combine prior knowledge and observation in a consistent and well-defined way and at least but not lest it doesn’t requires measuring with smallest achievable uncertainty It is also important to stress that according to GUM § 3.4.8. “…The evaluation of uncertainty is neither a routine task nor a purely mathematical one; it depends on detailed knowledge of the nature of the measured and of measurement…”. Uncertainty arising from the repeatability of chemical measurement is a characteristic of the method. Its calculation has similarities to the calculation of uncertainty in physical measurement and many components are identical to those involved in physical measurement, components such as uncertainty in mass and volume. However, others such as purity of reference materials and recovery are rather unique to chemistry, but once determined, can still be incorporated into the uncertainty budget using the standard techniques developed for physical metrology and described in the publication, GUM. The interpretation of this guide for analytical chemistry was done by Eurachem in 1995 [8]. It is always important to know when should the evaluation of uncertainty be perform: When a procedure is introduced inside the laboratory; When a critical factor changes in the procedure (instrument, operator..) During or together with analytical procedure validation This means that an individual evaluation process is not needed for every individual result produced!! The GUM approaches cover respective 10-steps, which leads to the proper evaluation of the uncertainty: 1. Defined the measurand 2. Describe the model equation (for the measurement procedure) 3. Identify all possible sources of uncertainty; Possible sources of uncertainty: recovery of analyte from a complex matrix; storage conditions; reagent purity; assumed stoichiometry; measurement conditions; instrument response; bias of instrument; instrument resolution; uncertainty of standards and CRM’s; variation in repeated observation. 4. Estimate values for all input quantities using the type A and type B uncertainty; Input quantities uncertainty (type A / B) repeated observation (A) validation (A and/or B) manufacturers specification (B) calibration certificates (B) literature data (B) 5. Evaluate the standard uncertainty (1 standard deviation) of each input quantity; Before combining, all uncertainty must be expressed/covered as “estimated” standard uncertainty. When available as: Standard deviation: use as is Confidence intervals: convert Stated range: convert Expanded uncertainty: convert 49

Chapter 4

6. Calculate the value of the measurand (using the model equation); 7. Calculate the combined standard uncertainty of the result; Assuming no correlation between input quantities the law of uncertainty propagation can be used: 8. Calculate the expanded uncertainty (with a selected k); The coverage factor usually used is k = 2, representing a coverage of about 95%, if the distribution is normal. Expanded uncertainty give a more realistic range of possible values. 9. Analyse the uncertainty contribution index 10. Document all steps in a Report We expect that this module will convince that the uncertainty of measurement results, evaluated according to the GUM, is a very useful approach since: • it enables the analyst to combine prior knowledge and observations in a consistent and well defined way; • it does not require measuring with smallest achievable uncertainty, but rather is focused on realistic evaluation of the analytical performance of the laboratory; • Measurement procedure is a logical sequence of various steps done in a certain time and space; according to the law of uncertainty propagation, only input quantities contribute to the uncertainty result. Reproducibility should be considered as a part of the total uncertainty. 5 BASIC STATISTIC FOR UNCERTAINTY EVALUATION The analytical results provided by the laboratory is always based on the statistical evaluation of the raw data. This covers mainly average of the set of data and standard deviation. In fact what is needed to apply the GUM uncertainty, is based also on some basic statistics. The important aspect of making the uncertainty budget is to use the law of propagation, the tool which enable to combine the uncertainty of type A and type B in a consist way. At last but not least the understanding of distribution patterns of the set of data is required for uncertainty evaluation. It is important to be aware that different kind of distribution (normal, rectangular or triangular) occurs and this influences the conversion of the input quantities of type B (previous experiments, literature data, manufactures information etc.) in a form of standard uncertainty. Examples of distributions I. Normal distribution is used when an estimate is made from repeated observation of a randomly varying process. An uncertainty is then given in a form of a standard deviation s, a relative standard deviation s/xmean or a coefficient of variance CV%. An uncertainty is given in the form of a 95% (or other) confidence interval. U(x) = c / 2 for c at 95% U(x) = c / 3 for c at 99.7% II. Rectangular distribution is used when the informations are taken from a certificate or other specification, which gives limits without specifying a level of confidence e.g. Concentration of the calibration standard is quoted as 1000 ± 2 (mg/l) The purity of the cadmium is given on the certificate as 99.99 ± 0.01 (%)

50

Chapter 4

2 a (= ± a)

X The value is between the limits: [- a ….+ a ]. An estimate is made in the form of a maximum range (± a) whit no knowledge of the shape of the distribution. Assumed standard uncertainty is according to the equation:

s = u ( x) = a / 3 III. Triangular distribution is used when the available information concerning x is less limited that for a rectangular distribution. Values close to the mean value x are more likely than near to the extremes. 2 a (= ± a)

X The manufacture quoted volume for the flask of 100 ± 0.1 (ml) at t = 20 oC. A nominal value is most probable. An estimate is made in the form of a maximum range (± a) described by a symmetric distribution.

s = u ( x) = a / 6 6 VALIDATION OF A MEASUREMENT PROCEDURE Validation of a measurement procedure can be regarded as one of the most important part of the every day laboratory work. When choosing the most promising candidate method for intended use, one should consider the experience of own laboratory, as well as equipment ready to be applying, the time and cost involved in particular measurements. Validation of the measurement procedure increases confidence for users of the measurement procedure and measurement results, and provides information on expected procedure performance characteristics.

51

Chapter 4

According to ISO/IEC 17025 validation is the confirmation by the examination and provision of objective evidence that the particular requirements for a specific intended use are fulfilled. Among much objective evidence expected, performance characteristics of the measurement procedure are the most important. Performance characteristics of the measurement procedure, e.g. working range, linearity (of the procedure), sensitivity, detection limit, belong to one group of the validation parameters. The second group of the validation parameters comprises the property of the results obtained with this procedure, like traceability and uncertainty of the measurement result. TABLE 3. Validation checklist Performance parameters of a procedure (quantitative) - working range; - detection / determination limits; - sensitivity Properties of the results obtained with a validated procedure - traceability - uncertainty

recovery; robustness; selectivity; specificity; repeatability; reproducibility

It should be pointed out that in various documents different terminology for the validation is used. For example ISO/IEC 17025 uses ‘method validation’; VIM uses ‘procedure validation’; and GLP uses ‘standard operation procedure’. Besides the wording being use in all cases validation is consider as a study of the procedure, not of the analyst or of the laboratory performance. Validation provides information on the procedure performance. The intended use of the validation is Compliance with regulations; Maintain quality and process control; Support national and international trade; Support research. In all cases the validation includes: Specification of requirements; Determination of the procedure characteristic; Checking whether the requirements can be fulfilled by the procedure Statement that the chosen measurement procedure is valid for the intended use. Which methods (procedures) should be validated? Non-standard methods; In-house developed methods; Standard methods used outside their intended scope; Modified standard methods. 52

Chapter 4

There are many validation techniques recommended by international standards, regulatory bodies, and other authorities. For scientific laboratories the systematic assessment of the quantities influencing the result is highly recommended. Some parts of the whole uncertainty budget should be compared to the validation study: precision of the input quantities, recovery of the measurement procedure (bias!), robustness of the input quantities. TABLE 4. Validation techniques recommended I II

Evaluation of uncertainty = systematic assessment of the quantities influencing the result Using of CRM for calibration and for evaluation of the accuracy of the results

III

Participation in inter-laboratory comparisons

IV

Comparison of results with that obtained with other procedure

V

Systematic assessment of the factors influencing the result

There is a very tight connection between the development of the measurement procedure (input parameters optimisation) and validation of the measurement procedure for intended use. In all cases it is important to cover the whole analytical procedure within the whole range of its application. Therefore three "golden" rules should be fulfilled: a) validate whole measurement procedure (from sampling to measured signal); b) validate procedure for use in all intended matrices; c) validate procedure for use in intended concentration range. The analyst should record the results obtained within the validation and the final conclusion whether the applied procedure is fit for the intended use. The validation should be as extensive as it is necessary to meet the need of the given application or field of application. In many cases all procedure need to be validated but with different degree of validation! Examples 1. Determination of cholesterol in serum Limit of determination is not important, especially the uncertainty is important, so the attention should be put on that. 2. Survey of environmental contamination (towards finding the hot spots) LOD and uncertainty size is not important. Range and linearity is important 3. Doping control against the limit LOD is important, uncertainty is extremely important Range and linearity are not important.

53

Chapter 4

7 ON THE USE OF CERTIFIED REFERENCE MATERIALS Certified reference materials (CRMs) and reference materials (RMs) are widely used in analytical chemistry. Their proper application provides the most direct information about the quality of and confidence in the obtained measurement results. CRMs are used for calibration of an apparatus; for the assessment of the measurement method; for establishing traceability of the measurement results; for the determining the uncertainty of these results. According to VIM the definitions are as follows: Reference Material: “material or substance one or more of whose properties are sufficiently homogeneous and well established to be used for the calibration of an apparatus, the assessment of the measurement method, or for assigning values to materials” Matrix Reference Material: “a ‘natural’ substance more representative of laboratory samples that has been chemically characterised for one or more elements, constituents etc., with a known uncertainty” Certified Reference Material: “reference material, accompanied by a certificate, one or more of whose properties values are certified by a procedure which established traceability to an accurate realisation of the unit in which the property values are expressed, and for which each certified values is accompanied by an uncertainty at a stated level of confidence.” Certified reference materials (CRMs) are generally recognised as an excellent means to check analytical accuracy. Users of these CRMs can compare their analytical results for those materials with the certified value taking into consideration both the uncertainty of the CRN and the uncertainty of the measurement. Therefore, reliable estimates for CRM uncertainty are necessary. Types of the CRM according to their use: Pure substances for calibration (e.g. solution of Cd to prepare calibration solution for AAS measurements) Pure substance for matrix matching (e.g. high purity Cu to make a Zn/Cu calibration series for ICP-OES) Matrix CRMs (e.g. cholesterol in serum) When looking for the quality of chemical measurement results today it is clear that still the same “traditional” simple concepts exist in the analytical chemistry society. In many cases it is still assumed that if one uses a quality management system, written standards or certified reference materials (CRM), one automatically gets better quality results? There is considerable evidence that this is not the case and that the selection of appropriate CRMs by the user with respect to sample matrix, concentration range and uncertainty of certified properties is essential. The production of RM and CRM is not a trivial task and it should be perform according to the ISO 35 document, where the integrated process of correct preparation, homogeneity and stability demonstration, and accurate and traceable characterisation of the material is described. For the high quality of RM or CRM the traceability and uncertainty of certified value should be state and demonstrate. The selection of the appropriate CRM’s by the user should be done with respect to sample matrix, concentration range and uncertainty of the certified properties.

54

Chapter 4

Information and catalogues can be available via web: IRMM www.irmm.jrc.be BAM www.bam.de (also COMAR dbase) NIST www.nist.gov LGC www.lgc.co.uk REFMAT www.refmat.org.pl 8 INTERLABORATORY COMPARISONS The results from Interlaboratory Comparisons (ILC’s) are of crucial interest for a laboratory as these provide clear information of its ability to demonstrate reliable results to its customers. Participating to ILC enable to demonstrate ability to measure and in case it is necessary should lead to improve the quality of results. Laboratories either participate voluntarily or are forced by external requirements (e.g. legal, accreditation, control bodies). In case of unsatisfactory performance, the result from participating to the ILC either indicates that there was an unaccounted bias in the result or/and that the uncertainty was underestimated. Therefore, participation to an ILC can help to make a decision of the necessary action perform bias correction or/and reassess the magnitude of the uncertainty estimates. Most ILC’s schemes involved comparison of participant results with an assigned value, which has been derived from a reference laboratory, a sub-group of participants, consensus from the overall population of test results or by some other means. The advantages and disadvantages in the use of consensus values and reference values have been discussed recently by Van der Veen et all [13]. The organiser of ILC’s should preferably provide an expert commentary on overall laboratory measurement performance against prior expectation or prior requirement, taking uncertainty into account. If the laboratory reports the measurement results with the corresponding uncertainty (which should always be done for ILC’s), its performance can be evaluated based on both the reported numerical values and its uncertainty. The organiser of ILC shall assign a value with an appropriately small measurement uncertainty, while a participating laboratory can measure the same with greater “fit-for-purpose” uncertainty, often customer and/or cost driven. It is expected that each laboratory taking part in the interlaboratory certification studies should return a GUM compliant uncertainty budget along with its results for the measurand. In order to do this effectively participating laboratory must use wellcharacterised methods, which they are familiar with.

55

Chapter 4

COMMON LANGUAGE Terms Accuracy of measurement Calibration

Definition Closeness of the agreement between the result of a measurement and a true value of measurand Set of operation that establish, under specified conditions, the relationship between values of quantities indicted by a measuring instrument or measuring system, or values represented by a material measure or a reference materials, and the corresponding values realised by standards [VIM 6.11] Measurand Particular quantity subjected to measurement [VIM 2.1] Measurement Set of operations having the object of determining a value of a quantity [VIM 2.1] Method of Logical sequence of operations, described generically, measurement used in the performance of measurements [VIM 2.4] Measurement procedure Set of operations, described specifically, used in the performance of particular measurements according to a given methods [VIM 2.5} Model equation The equation used to calculate the result of a measurement Quantity Attribute of a phenomenon, body or substance that may be distinguished quantitatively and determined quantitatively [VIM 1.1] Result of a Value attributed to a measurand, obtained by measurement measurement [VIM 3.1] Value (of a quantity) Magnitude of a particular quantity generally express as a unit of measurement multiplied by a number [VIM 1.18] EXAMPLES Quality

Analyte

Concentration DDT Content Activity

Cd Amylase

pH

H+

Measurand

Unit

Concentration of DDT Content of Pb Activity of amylase Concentration (activity) of H+

ng/l

Stated reference SI

ng/kg Katal

SI SI

pH unit

pH scale

Octane index

Octane number Octane number scale

56

Chapter 4

9 CONCLUSIONS As the importance of the results of chemical measurements increased in various field of public activity, a reliability of those results can be achieved only by tracing the results back to a standard realising the unit in which the measurement result is expressed. Therefore, analytical chemist should also use metrological way of thinking and terms like traceability and uncertainty of the measurement results. The uncertainty can be stated only if the traceability of the measurement result to a system of units is guaranteed. REFERENCES [1]. Dube G., Accred Qual Assur, 6, 3 (2001) [2]. De Bievre P., Fresenius J Anal Chem, 366, 1 (2000) [3.] King B., Accred Qual Assur, 6, 236 (2001) [4.] De Bievre P., and Taylor P.D.P, Fresenius J Anal Chem, 368, 567 (2000) [5.] Bulska E., Taylor P., Anal. and Bioanal. Chem., in print (2003) [6.] ISO/IEC 17025 (1999) General requirements for the competence of testing and calibration laboratories, ISO, Geneva [7.] Majcen N., Bulska E., Leiti I., Vassileva E., Papadakis I., Taylor P., Accred. Qual. Assur., 7, 419 (2002) [8.] Taylor P., Bulska E., Vassileva E., Majcen N., Suchanek M., Accred. Qual. Assur., in print (2003) [9.] Bulska E., Lipinski J., Papadakis I., De Bievre P., Taylor P., Meterology in Chemistry: Status report of Poland, EUR 19915 EN (2001) [10.] ISO (1995) Guide to the Expression of Uncertainty in Measurement, ISO, Geneva [11.] Quantifying Uncertainty in Analytical Measurements, EURACHEM (1995) [12.] International Vocabulary of Basic and General Terms in Metrology, ISO (1993), Geneva [13]. Van der Veen A.M.H., Horvat M., Milacic R., Bucar T., Repinc U., Scancar J., and Jacimovic R., Accred Qual Assur, 6, 264 (2001)

57