化学测量结果最后使用人的溯源性

trends Traceability in chemical measurements for the end usersMiguel Valcaèrcel*,Angel R|èosDivision of Analytical Chemistry,Faculty of Sciences,University of Coèrdoba,Cordoba,SpainTraceability is the foundation of any metrolog-ical science.Traditionally this term has been associated with the results of a particular measure as synonymous with the accuracy or reliability of such results(i.e.,a certain level of quality).It ensures a clear(unbroken) relationship between results and the corre-sponding samples by using appropriate standards and calibrated equipment.This simple principle can be easily implemented in physical measurements,in which^com-monly^the measurement process is mainly performed by the simple use of a(single)cali-brated equipment.The case of chemical measurements is quite different because only the measurement step of the analytical process can be calibrated,but not the other steps.As measurements have a clear sample dependence,standards are not available in many situations.Therefore,the term metrol-ogy in chemistry needs to relate to a wider meaning of traceability,more addressed to a practical approach which is useful for end users(chemists and clients).Some adapta-tions are needed if a rational use of the term is to be applied to routine analytical work.In this article we propose a more practical and £exible view of term,`traceability'in the¢eld of metrology in chemistry.z1999Elsevier Science B.V.All rights reserved. Keywords:Traceability;Quality;Standards;Metrology1.IntroductionTraceability is a basic(inherent)aspect of a metro-logical science such as analytical chemistry.This dis-cipline^the analysis^is the third basic component of chemistry,beside theory and synthesis[1].Traceabil-ity in chemistry appeared simultaneously with the¢rst analytical measurement carried out,many centuries ago,but the term was never used before the last two decades,in contrast to its wide use in physical meas-urements.The interest on this topic has grown in the last few years,and a number of articles has been pub-lished(e.g.,[2^8]).As pointed out by Adams[7],it is also the source of strong polemics(e.g.,Alexandrov [9],and Ackermann et al.[10]vs Haësselbarth [11,12],and Alexandrov[13]vs Valcaèrcel and Rios[3])as a consequence of the fact that the trace-ability concept has not been de¢nitively settled in met-rology in chemistry.Almost all analysts understand what traceability means globally.The differences start when it is applied in daily practice.There is a need to approach traceability in chemical measure-ments in a de¢nitive way,not from a dogmatic view but from a wide consensus^taking into account the opinions of the end users(analysts and chemists). Nevertheless,it must be recognised that it is not an easy task to establish an approach to traceability with a suf¢cient degree of£exibility to satisfy both theoreticians and analysts.In this article we try to develop a new view of traceability in metrology in chemistry,based on our previous articles[2,3].This new approach uses the main conclusions of a docu-ment supported by the European Commission(SMT4-CT96-6505)[14,15],which can be summarised by the sentences:``The direct application of the existing concepts,norms and guides to metrology in chemistry and biology is far from adequate to the real routine work T'',``In many cases it is contrary to the common sense acquired through years of experience T''and ``Speci¢c developments coherent with the bench level work would be welcome''.0165-9936/99/$^see front matterß1999Elsevier Science B.V.All rights reserved. PII:S0165-9936(99)00157-02.The two sides of analytical chemistryMany of the problems related to traceability in chemical measurements arise from the divorce between the basic and applied sides of analytical chemistry [1],which are shown in Fig.1.On the one hand,analytical chemistry deals with metrology in chemistry,which is directly related to the quality of measurements,and thus with the measurement stand-ards used in the comparisons implicit in measure-ments.Accuracy and uncertainty are two key analyt-ical properties in this context [16],in which the establishment of a metrological hierarchy is manda-tory in order to avoid frequent misunderstandings [17].The practical side of analytical chemistry is charac-terised by the concept of `¢tness for purpose'[18],which is consistent with the planning and solving of the so-called analytical problem [19].Besides other properties,representativeness [20]is one of the cor-nerstones of the quality of this aspect.The conventional (orthodox)approach to traceabil-ity can be ascribed to the basic side of analytical chem-istry.But,as shown in Fig.1,it is and must be related to the applied facet,in order to make this concept understandable,accessible and useful for the end users [14,15].The extended view of traceability [3]is thus essential in Analytical Chemistry.3.From a classical to a metrological approach to analytical chemistryAs can be seen in Fig.2,switching from a classical to a metrological approach to analytical chemistry entails some changes,namely:(a)Conventional analytical properties,such as accuracy and precision,will increasingly be replaced by traceability and uncertainty.(b)Of¢cial or standard methods will be progres-sively substituted by traceable methods.(c)The general interest in analytical processes will be completed by a particular emphasis on the quality of both the measurement standards involved and the quality of the measurements.(d)The generic non-systematic interest in qual-ity will be shown in an increasing interest in the application of quality-systems,both inside and outside the analytical laboratory,under quality assurance principles,which are implemented through quality control and quality assessment activities [21].(e)The importance of relationships among lab-oratories will increase as a consequence of the need to put emphasis on comparability and har-monisation in order to achieve mutual recogni-tion of results,which is a substantial milestone in social and economic international relationships.The evolution of analytical chemistry in such topics implies an enrichment of the basic side of the disci-pline,although no relevant innovations arereallyFig.1.The basic and applied facets of analytical chemistry and its relationships with conventional and extended approaches totraceability.Fig.2.Extension of the metrological character of analytical chemistry from the classical approach.introduced.The old principles do not disappear but, rather,they are improved.4.Orthodox versus heterodox concepts of traceabilityThe only way of precisely de¢ning traceability in a global manner entails the integration of two nuances of meanings^namely(see Fig.3):(a)the establishment of one or more clear-cut unequivocal relationships to well-stated references(standards);and(b)the docu-mented`history'of a product(e.g.,a result,a sample, measuring standard,and instrument)or a system(e.g., an analytical method,a laboratory).The two meanings are not independent:rather,they are indispensably related to de¢ne traceability in a complete and correct way.Many of the problems in the de¢nition arise from a partial view of the concept.These two meanings are well re£ected when the word traceability is translated from English into other European languages.Accord-ing to some prestigious scientists in this¢eld[4^6], traceability is an exclusive characteristic of a result, which is coherent with the de¢nition of such term in the metrological dictionary[22].In this orthodox focus,the same de¢nition is used for metrology both in physics and in chemistry.Nevertheless,there are crucial differences from a practical point of view [15,17],which make it impossible to use such a`dog-matic'de¢nition,originally developed in the physical domain.In the ISO8402vocabulary of terms used in quality assurance and quality management domains,trace-ability is approached in a simple and broad way [23]based on the second meaning of the concept, shown in Fig.1:i.e.,the``ability to trace the history,application or location of an entity by means of recorded identi¢cations''.This approach is consistent with the etymology of the word[24,25]and is one of the bases for proposing the extension of the traceability concept in metrology in chemistry to dif-ferent analytical facets[3].Some authors[9]con-sider this extended meaning to be a heterodox approach but,in our opinion,it is the only way to make the concept more practical and useful in analyt-ical chemistry.5.Traceability as a characteristic of different analytical facetsA new quasi-heterodox approach to traceability has been proposed[3],taking into account the following aspects:(a)the integral de¢nition of traceability shown in Fig.1;(b)the broad de¢nition contained in the Quality Management and Quality Assurance Vocabulary[23];(c)the inherent features of metrol-ogy in chemistry[15,17];and(d)the bench-top level in which`traceability',`traceable',and`to trace' are used frequently in ordinary language to char-acterise not only results but also other analytical aspects.Below(Fig.4),we describe the use of trace-ability as a property of results,standards,samples, tools and processes which are not independent but are strongly related through a well-established hier-archy[3].Fig.4.Systematic use of traceability as a property of differ-ent facets of analyticalchemistry. Fig.3.Integral traceability concept in chemical analysisand its relationships with conventional and extendedapproaches,which are related to different norms andguides.6.Traceability of a resultThe of¢cial and orthodox approach to traceability re£ects a crucial property of a result of a measurement process.In the second edition of the metrological dic-tionary [22]traceability is de¢ned as ``a property of a result of a measurement or the value of a standard whereby it can be related to stated references,usually national or international standards,through an unbro-ken chain of comparisons all having stated uncertain-ties''.There are three signi¢cant differences between this de¢nition and that of the ¢rst edition (1984)of the same dictionary.First,it considers the value of a stand-ard as an additional subject to which the traceability attribute can be assigned.Secondly,there is the intro-duction of the term `reference',which has a broader meaning than `standard'.Finally,the statement that uncertainty is a substantial aspect without traceability cannot be de¢ned properly.A broader view of the traceability of a result can be achieved if the following three essential aspects are considered systematically (see Fig.5):(a)The ability to relate the result to well-estab-lished standards,which is consistent with the of¢cial de¢nition.The `unbroken chain of com-parisons'is a crucial facet of this ¢rst meaning.The comparisons are those involved in chemical measurements based on calibration,because `tomeasure'is `to compare'.This chain must be as short as possible,in order to reduce the uncer-tainty of the result.The `strength'of each link of the chain depends strongly on the quality of the measurement standards used,and on methodo-logical quality,which should be checked by using both internal and external control /assess-ment systems under the umbrella of quality assurance.(b)The documented `history'of the production of the result,which implies de¢ning,in a clear way,the human,material,instrumental,meth-odological,environmental and temporal ele-ments contributing to its generation.The docu-ments must answer basic questions concerning with the results ^who?,what?,how?and when?This facet is not stated in the metrological de¢nition;it is intrinsic with the traceability con-cept and consistent with quality systems (e.g.,EN 45000,GLPs)applied in chemical laborato-ries.(c)The practical connotation of traceability of the results is based on the establishment of unbroken chains of comparisons of the results provided by many laboratories with a common standard of high metrological quality.Two immediate consequences appear.First,all the standards used in the independent chain of com-parisons will be traceable to one another;and second,all the results produced by the laborato-ries will be traceable among them and,as a con-sequence,all laboratories will be reliably related,comparable and harmonised.The traceability of a result is at the top of a hierarchy of traceability concepts [3]in such a way that it is based on the importance and extension of all of them.7.Traceability among standardsThe de¢nition of traceability given in 1993[22]considers this term as an attribute of the data of a measurement standard.Therefore,this approach of traceability is within the orthodox de¢nition.Never-theless,there is urgency in establishing an (almost)de¢nitive classi¢cation among the standards used in metrology in chemistry using proper words and under-standable de¢nitions for the end users,instead of try-ing to adapt that well-established in metrology in physics,in which the traceability chain is simple,short,and easily setup.Fig.5.Integral de¢nition of the traceability of a result of a measurement on metrology in chemistry.In the authors'opinion[2],the best system for establishing a classi¢cation among standards and the traceability links among them,it is by considering several key aspects,which can be integrated.First,it is necessary to design a general hierarchy of the main types of standards:(1)basic standards(SI units);(2) chemical standards(mass of12C,atomic weights, Avogadro number,Faraday);and(3)analytical(tan-gible)standards used in chemical laboratories,which can,in turn,be primary and secondary(working) standards.Chemical standards can be considered as the traceability link between basic and analytical standards.Secondly,the difference between reference materials(RM)and certi¢ed reference materials (CRM),based on the body responsible for the asso-ciated data(and uncertainties),must be established among analytical standards.Thirdly,it is necessary to distinguish analytical standards according to their nature and their use.They can be physical standards, pure substances and their mixtures and matrix stand-ards.A detailed description of these classi¢cations integrated into a single one is beyond the scope ofthis paper.Traceability among standards is the most relevant foundation of traceability of the results[3].This is well illustrated when the quality of the well-known standards does not meet the minimum requirements [26,27].8.Traceability of equipmentTwo main kinds of equipment are used as tools in analytical processes:(a)instruments which produce information related to the presence or concentration of analytes in the sample;and(b)apparatus which per-forms a function(e.g.,cooling,extraction,centrifuga-tion,distillation),but produce no information.The traceability of equipment can be de¢ned as the detailed,timely,and customised recording of installa-tion,malfunctioning and repairs,periodic calibration and corrections(if needed),hours of use,samples processed,standard used,etc.,in such a way that all questions(what?,how?,who?,etc.)should have a detailed answer in the pertinent documents.Both meanings of traceability(see Fig.3)are involved in the traceability of equipment.On the one hand,the relationship to standards through calibration and,on the other hand,the`history'of its behaviour. This traceability concept is coherent with its generic de¢nition in the ISO vocabulary of QM and QA[23]. It is also considered indirectly in the two most relevant laboratory quality systems,namely:(a)EN45000 norms,in which`equipment'is an essential entry in the quality manuals;and(b)good laboratory practices (GLPs),in which each equipment must have an SOP (standard operational procedure).The relationships to standards of equipment are materialised in the calibration operations.In metrol-ogy in chemistry there are two different but comple-mentary notions of calibration that do not exist in met-rology in physics,except in a few cases.In Fig.6we depict the differential characteristics of equipment and methodological calibration in terms of de¢nition,aim, and standards involved.In the¢eld of UV-visible absorption spectroscopy,the target of instrumental calibration is the spectrophotometer itself,which can be checked both quantitatively(by using absorbance standards such as potassium dichromate solution)and qualitatively(by using wavelength standards such as holmium and didymium¢lters).The target of meth-odological(analytical)calibration is the chemical measurement process that uses the calibrated spectro-photometer as an instrument:the standards used are solutions of measured concentration of the analyte that provide a calibration curve.Methodological calibra-tion is rather unusual in physical measurements. The traceability of equipment is also a foundation of the traceability of a result.It is also based on the traceability of the standards involved in its calibration [3].Fig.6.Differences between instrumental and methodolog-ical(analytical)calibration.9.Traceability of sample aliquotsThe portion(aliquot)of samples initially subjected to the analytical process must be traced unequivocally to two crucial references.On the one hand,it must be related to the chemical information needs,through the analytical problem[19],i.e.,to the¢tness for purpose [18].On the other hand,it is necessary to assure a reliable relationship between the sample aliquot and the corresponding results through the so-called`sam-ple custody chain':this aspect was described as`track-ability'by Fleming et al.[28].This traceability con-cept is quite heterodox relative to the metrological de¢nition[22]and the fact that it is not based on measurement references,but it is coherent with the generic de¢nition of traceability[23],and takes into account the two facets of analytical chemistry depicted in Fig.1.It corresponds to the representativeness(a top analytical property)[20].If both traceability links arising from the sample aliquot are well established,the results provided willbe traceable to the chemical information needs posed by clients,which is the¢nal goal of analytical chem-istry.It is interesting to note that analytical quality systems start by considering,implicitly,this traceabil-ity concept(e.g.,in pro¢ciency testing schemes for sampling[29,30]).10.Traceability of analytical methodsThere is a tendency to qualify analytical processes as traceable instead of using the traditional attributes such as`standard',`of¢cial',`reference',etc.,which do not assure the quality of results.This attribute can be incorporated in the framework of validation of ana-lytical methodologies because it refers to a substantial part of it(relationship to standards).A traceable method can be de¢ned as a method,both in its development and in its routine application,that produces results(with their uncertainties)character-ised by a de¢ned traceability to well stated references. This de¢nition implies the need to demonstrate this traceability experimentally,which is based on a che-mometric comparison of the results provided and the date of a CRM,or the results provided by a primary method applied to aliquots of the same sample. The traceability of an analytical method is another foundation of the traceability of the results it provides, but it is also based on the traceability of the equip-ment and the standards involved in calibration opera-tions.11.Traceable to what?All the traceability concepts in analytical chemistry rely on tangible or intangible references to de¢ne a relevant characteristic of results,standards,equip-ment,samples and methods.However,the crucial question in this context is to state what is the last(ulti-mate)reference,or the reference of other references. By tradition,in metrology in physics the last refer-ence of the unbroken chain of comparisons involved in the traceability of a result is an SI unit.This situation is quite in£exible and dif¢cult to maintain in metrology in chemistry.A more£exible approach to answering this question,taking into account the extended mean-ing of traceability[3]in chemical measurements,is depicted in Fig.7.The use of an SI unit(i.e.,a basic standard,e.g.,kg and/or the mole as the alternative reference)in met-rology in chemistry is quite unrealistic,as stated by Thompson[8],and is only applicable if primary meth-ods(gravimetry,titrimetry,coulometry,IDMS,etc.) are involved,because they have short and simple chains of comparisons among standards,as well as the fact that tangible analytical standards are not involved.Any reference material with appropriate characteristics that is well known and well-de¢ned can be considered as the ultimate standard.There are other alternatives,such as traceability to a primary method,to a laboratory of excellence,and to a group of laboratories which have participated in apro¢-Fig.7.Ultimate references to establish traceability on met-rology in chemistry as a characteristic of results,stand-ards,equipment,methods and sample aliquots.ciency testing scheme under the management of an independent,accredited body.As regards the broadest meaning of traceability in analytical chemistry,the references should be the results and chemical information needs for the trace-ability of sample aliquots.12.A new de¢nition of traceability in chemical measurements?Taking into account the substantial differences between metrology in physics and chemistry [15,17],the tendency of existing norms and guides to cover both sides in a general way,and the need to make the traceability concept more accessible and practical in chemistry,there is a need to develop a new,broader de¢nition of traceability in analytical chemistry,based on the arguments expressed in this paper.The variety of analytical facets to which the traceability property can be assigned makes it impos-sible to outline a single de¢nition of traceability. The following de¢nitions are proposals to be consid-eredasstartingpointsforreachingaconsensusbetween theoreticians and end users(analysts and clients).12.1.Traceability of analytical resultsA property of the result of a chemical measurement process linked unambiguously to the sample aliquots and to the chemical information needs,whereby it can be related to stated references,through an unbroken chain of comparisons all having stated uncertainties, and the ability to trace the documented history of its generation.It is based on the traceability of standards, equipment,and the methodologies involved in its gen-eration.12.2.Traceability of a measuring standardA property of the value of such a reference whereby it can be related to stated references through an unbro-ken chain of comparisons,all having stated uncertain-ties,as well as the ability to trace the documented history of its production in which crucial character-istics such as homogeneity,stability,origin,etc., must be clearly stated.12.3.Traceability of an equipmentA detailed,timely and customised recording of the installation,breakdowns,hours of use,samples proc-essed,and any other feature of interest relating to a particular equipment,with special emphasis on the performance of the maintenance,calibration and cor-rection activities.12.4.Traceability of methodologiesThe ability of a methodology to produce traceable results after the appropriate validation.References[1]M.Valcaèrcel,Trends Anal.Chem.16(1997)124.[2]M.Valcaèrcel,A.R|èos,Analyst120(1995)2291.[3]M.Valcaèrcel,A.R|èos,Fresenius'J.Anal.Chem.359(1997)473.[4]B.King,Analyst122(1997)197.[5]Y.I.Alexandrov,Analyst121(1996)1137.[6]M.Thompson,Analyst121(1996)285.[7]F.Adams,Accred.Qual.Assur.3(1998)308.[8]M.Thompson,Analyst121(1996)285.[9]Y.I.Alexandrov,Fresenius'J.Anal.Chem.357(1997)563.[10]M.Ackerman,W.Britschuh,H.Buëchler,G.Kuëppers,C.Luëhrs,B.J.Schlothmann,G.Staats,R.Zens,Fresenius' J.Anal.Chem.357(1997)562.[11]W.Haësselbarth,Fresenius'J.Anal.Chem.352(1995)400.[12]W.Haësselbarth,Fresenius'J.Anal.Chem.354(1996)263.[13]Y.I.Alexandrov,Fresenius'J.Anal.Chem.359(1997)263.[14]Metrology in Chemistry and Biology:A PracticalApproach,Report EUR18405EN,Directorate General Science,Research and Development,European Commis-sion,Luxembourg,1998.[15]M.Valcaèrcel,A.R|èos,E.Maier,Accred.Qual.Assur.4(1999)143.[16]M.Valcaèrcel,A.R|èos,Anal.Chem.65(1993)78A.[17]M.Valcaèrcel,A.R|èos,Trends Anal.Chem.18(1999)68.[18]M.Thompson,T.Fearn,Analyst121(1996)275.[19]M.Valcaèrcel,A.R|èos,TrendsAnal.Chem.16(1997)385.[20]A.R|èos,M.Valcaèrcel,Analyst119(1994)109.[21]M.Valcaèrcel,A.R|èos,Trends Anal.Chem.13(1994)17.[22]International Vocabulary of Basic General Terms in Met-rology(1993)2nd Edition,ISO,Geneva.[23]ISO8402(1994)Quality Management and QualityAssurance Vocabulary,ISO,Geneva.[24]Oxford English Dictionary,Oxford University Press,Oxford,2nd Edition,1989.[25]Webster's Third New International Dictionary,Merriam-Webster,Chicago,IL,1981.[26]L.H.Keith,Environ.Sci.Technol.28(1994)590A.[27]D.Nobie,Anal.Chem.66(1994)868A.[28]J.Fleming,B.Neidhart,Ch.Tausch,W.Wegscheider,Accred.Qual.Assur.1(1996)43.[29]M.H.Ramsey,A.Argyraki,M.Thompson,Analyst120(1995)2309.[30]A.Argyraki,M.H.Ramsey,M.Thompson,Analyst120(1995)2799.。