1、Designation: E2857 11Standard Guide forValidating Analytical Methods1This standard is issued under the fixed designation E2857; the number immediately following the designation indicates the year oforiginal adoption or, in the case of revision, the year of last revision. A number in parentheses indi
2、cates the year of last reapproval. Asuperscript epsilon () indicates an editorial change since the last revision or reapproval.1. Scope1.1 This guide describes procedures for the validation ofchemical and spectrochemical analytical methods of analysisthat are used by a metals, ores, and related mate
3、rials analysislaboratory.1.2 This guide may be applied to the validation of labora-tory developed (in-house) methods, addition of analytes to anexisting standard test method, variation or scope expansion ofan existing standard method, or the use of new or differentlaboratory equipment.1.3 This guide
4、 may also be used to validate the implemen-tation of standard test methods used routinely by laboratoriesof the mining, ore processing, and metals industry.2. Referenced Documents2.1 ASTM Standards:2E135 Terminology Relating to Analytical Chemistry forMetals, Ores, and Related MaterialsE1601 Practic
5、e for Conducting an Interlaboratory Study toEvaluate the Performance of an Analytical MethodE1763 Guide for Interpretation and Use of Results fromInterlaboratory Testing of Chemical Analysis Methods2.2 ISO Standard:3ISO/IEC 17025 General requirements for the competence oftesting and calibration labo
6、ratories3. Terminology3.1 DefinitionsFor definitions of terms used in this guide,refer to Terminology E135.3.23.2.1 validation (of an analytical method), nconfirmation,by the provision of objective evidence and examination, that amethod meets performance requirements and is suitable for itsintended
7、use.4. Significance and Use4.1 Method validation is a process of demonstrating that themethod meets the required performance capabilities. Interna-tional standards such as ISO/IEC 17025, certifying bodies, andregulatory agencies require evidence that analytical methodsare capable of producing valid
8、results. This applies to labora-tories using published standard test methods, modified standardtest methods, and in-house test methods.4.2 Although a collaborative study is part of this guide, thisguide may be used by a single laboratory for method validationwhen a formal collaboration study is not
9、practical. This guidemay also be applied before a full collaboration study to predictthe reliability of the method.4.3 The use of multiple validation techniques described inthis guide increases confidence in the validity or application ofthe method.4.4 It is beyond the scope of this guide to describ
10、e fully thefundamental considerations in Section 5. For a more descrip-tive definition of these concepts, refer to the InternationalUnion of Pure and Applied Chemistry (IUPAC) technicalreport, “Harmonized Guidelines for Single Laboratory Valida-tion of Methods of Analysis,”4the IUPAC Compendium ofAn
11、alytical Nomenclature (Orange Book),5and the Eurachempublication, The Fitness for Purpose of Analytical Methods, ALaboratory Guide to Method Validation and Related Topics.65. Fundamental Considerations5.1 During the process of method validation, the user of ananalytical method should apply a number
12、of fundamental tenetsof analytical chemistry as they relate to the development andimplementation of test methods. It is important to make thedistinction between the validation of a test method by astandards-developing organization and the implementation ofthat test method by a laboratory. Whether th
13、e test method wasdeveloped by a committee of experts or by one chemist in a1This guide is under the jurisdiction of ASTM Committee E01 on AnalyticalChemistry for Metals, Ores, and Related Materials and is the direct responsibility ofSubcommittee E01.22 on Laboratory Quality.Current edition approved
14、Nov. 1, 2011. Published January 2012. DOI: 10.1520/E285711.2For referenced ASTM standards, visit the ASTM website, www.astm.org, orcontact ASTM Customer Service at serviceastm.org. For Annual Book of ASTMStandards volume information, refer to the standards Document Summary page onthe ASTM website.3A
15、vailable from American National Standards Institute (ANSI), 25 W. 43rd St.,4th Floor, New York, NY 10036, http:/www.ansi.org.4M. Thompson, S. Ellison, and R. Wood, “Harmonized Guidelines for Single-Laboratory Validation of Methods of Analysis,” Pure Appl. Chem., Vol 71, No. 2,2002, pp. 835-855. http
16、:/iupac.org/publications/pac5International Union of Pure and Applied Chemistry Compendium of AnalyticalNomenclature: Definitive Rules 1997, http:/old.iupac.org/publications/analytical_compendium/6EURACHEM Guide, The Fitness for Purpose of Analytical Methods, ALaboratory Guide to Method Validation an
17、d Related Topics, LGC, Teddington,Middlesex, United Kingdom, 1998. www.eurachem.org1Copyright ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959, United Spany laboratory, the laboratory shall implement themethod in the laboratory and shall demonstrate that the m
18、ethodis being performed sufficiently well and that the results meetthe goals for data quality. That is, they should ascertain that themeasurement process provides sufficient levels of performancefit for the purpose of testing the materials at hand. It isadvisable to determine and document performanc
19、e character-istics of the method including repeatability precision, limit ofdetection, limit of quantification, and perhaps other parameters.The laboratory is advised to evaluate the method for bias andfor susceptibility to introduction of bias (namely, ruggedness).A number of important consideratio
20、ns are discussed in 5.1.1-5.1.7, but specific procedures for determination and calculationare beyond the scope of this guide.NOTE 1In the following discussion, the term measurement process istaken to mean the entire process by which a laboratory performs a testincluding sample preparation, measureme
21、nts, and calculation of results.5.1.1 PrecisionThe first step in development and imple-mentation of an analytical method is demonstration thatmeasurements can be made with sufficient repeatability for thepurpose of quantitative analysis. Precision is defined as thedegree of agreement among a set of
22、values. Precision underrepeatability conditions is measured by having a single analystin a single laboratory use a single set of equipment to prepareand analyze portions of a homogeneous material. Precisionunder reproducibility conditions is measured by having anumber of different analysts at differ
23、ent laboratories prepareand analyze portions of a homogeneous material. Any numberof conditions intermediate between repeatability conditionsand reproducibility conditions may be used if the data serves auseful purpose. A good example is having multiple analysts ina single laboratory perform the ana
24、lyses, perhaps on multipledays. In the terminology of Committee E01, repeatability issynonymous with within-laboratory standard deviation, Sr,which is defined as the standard deviation of results collectedon the same material in the same laboratory on different days.In contrast, reproducibility is s
25、ynonymous with between-laboratory standard deviation, SR, which is defined as thestandard deviation of results obtained on the same material indifferent laboratories.5.1.1.1 The most common estimators of precision are stan-dard deviation, relative standard deviation, and variance. Equa-tions and exa
26、mples are available in many texts on statistics.5.1.1.2 The concept of maintenance of the repeatability overa period of time is known as statistical control. The laboratorycan implement tools such as control charts to demonstratestatistical control.5.1.2 Limit of Detection (LD)The detection limit is
27、 de-fined as the lowest amount of analyte that can be distinguishedfrom background by an analytical method. It is important todemonstrate that the measurement process has the capability todetect a significantly lower amount (concentration or massfraction) of the analyte than the laboratory must quan
28、tify. Foradditional information, consult the IUPAC Orange Book andthe Currie paper.75.1.3 Limit of Quantification (LQ)The limit of quantifica-tion is defined as the amount of analyte above which theestimated relative standard deviation (RSD) is #10 %. It isimportant to demonstrate and document that
29、the measurementprocess has the capability to quantify amounts less than orequal to those found in materials to which the test method isapplied. For additional information, consult the IUPAC OrangeBook and the Currie paper.75.1.4 BiasBias is the difference between the obtainedresult for a measurand a
30、nd the true value of the measurand. Ananalytical method may be subject to a known amount of biasthat was estimated when the standard test method was devel-oped and validated by a committee. In an analogous manner, alaboratory developing a new test method or implementing apublished standard test meth
31、od shall perform tests to estimatebias and demonstrate the methods resistance to introduction ofadditional bias, that is, ruggedness. Documentation of thisperformance enables the laboratory to elucidate the scope ofthe method and defend the results obtained using the method.NOTE 2Accuracy is a conce
32、pt related to both bias and precision. It isthe combination of knowledge of both the precision obtainable undervarious conditions and the amount of bias inherent in a given result. Theconcept of accuracy is often used in discussions of the fitness for purposeand the reliability of results from a tes
33、t method. In a published standardtest method, the statements of precision and bias taken together providethe basis for judgments of the accuracy of the test method.5.1.5 SelectivityThe selectivity of a method is its abilityto produce a result that is not subject to change in the presenceof interferi
34、ng constituents. The selectivity of a method can beinvestigated by introducing or varying amounts of substancesand evaluating the results for changes. By understanding theprincipal of measurement, the analyst may be able to define ashort list of suspected interferences and, thereby, limit theamount
35、of effort needed to establish the significant interferenceeffects.5.1.6 Calibration ModelRelative methods require calibra-tion using measurements of suitable reference materials andmathematical fitting of the measured responses to an algorithm,that is, an equation thought to describe adequately the
36、relation-ship between the amount of analyte and the measured response.Algorithms are almost always an approximation of the realworld, and as such, their ability to fit the data has limits that canbe tested by a variety of means including, but not limited to,analyses of certified or other reference m
37、aterials and statisticalevaluation of confidence intervals bracketing the calibrationcurve and extrapolating performance predictions beyond therange of the calibrants.7L. A. Currie, “Nomenclature in Evaluation of Analytical Methods IncludingDetection and Quantification Capabilities,” Pure Appl. Chem
38、., Vol 67, No. 10, 1995,pp. 1699-1723. http:/iupac.org/publications/pacE2857 1125.1.6.1 Working RangeThe term working range is a namegiven to the concept of a portion of a calibration curve thatprovides valid results as opposed to portions that are not fit forpurpose. The range in which the method i
39、s considered to bevalid can be characterized using a number of approaches. Thepreferred methods are those that use objective data for thepurpose of illustrating under which circumstances a calibrationmodel is fit for purpose.5.1.6.2 Calibration PerformanceThere are statisticalmethods for measuring h
40、ow well the chosen calibration algo-rithm, often a line, fits the data consisting of known amounts ofanalyte and measured responses from the analytical instrumentfor the calibrants. For every calibrant, one may calculate thedifference between known and calculated amounts. This infor-mation can be us
41、ed to describe the performance of all or partof the calibration. One can do any of a number of things withthe information, including calculating the standard deviation ofthe differences described above, constructing confidence inter-vals around all of part of the range of amounts, plotting thediffer
42、ence as a function of the amount to look for trends, andspotting any individual calibrant that clearly performs morepoorly than the rest. Documenting behaviors like these, seekingthe causes, and taking corrective actions are suggested meansto validate a test method.NOTE 3The applications of statisti
43、cal tools, for example, confidenceintervals around a calibration, need not be restricted to the region boundedby the lowest and highest calibrants or the lowest and highest validationreference materials measured using the method and a particular calibra-tion. These tools can be extrapolated and stil
44、l provide valid estimates ofmethod performance.5.1.7 RuggednessConsidered in its classical sense, rug-gedness of an analytical method is the resistance of the resultsto change caused by variations in the operational aspects of atest method. Operations characteristics may include substitu-tion of mac
45、hines used to prepare a specimen, substitution ofsources of reagents and ingredients, changes to environmentalconditions, and even changes of personnel. A task group of astandard development committee will perform ruggednesstesting at an early stage in the validation process and at a smallnumber of
46、laboratories before a larger set of laboratories areasked to invest in an interlaboratory study. The laboratoryimplementing a test method is advised to perform their ownruggedness tests at any time during implementation and regularuse of the method to identify and document effects of changesof these
47、 types.6. Means of Method Validation6.1 Once method development following the considerationsof Section 5 has been completed, evidence validating methodincrease the confidence that the method performance is accept-able for meeting measurement quality objectives. The valida-tion methods are described
48、in the following sections.6.2 Analysis of Reference Materials:6.2.1 Select a number of reference materials such that theanalyte amount encompasses the intended scope of the analyti-cal method.6.2.2 Analyze each reference material to determine theanalyte amount present. Replicate determinations may b
49、e madeif these data are to be used to estimate typical methodprecision. If possible, analyze reference materials that areindependent from the calibration. Record all results.6.2.3 Compare the reference material results to the valuesassigned for the material by the developing organization.Assess the acceptability of the test method for generating datain accordance with the laboratorys measurement quality ob-jectives.6.2.4 The following protocol is one approach that has beenfound to be an acceptable means of assessing the acceptabilityof data obtained using
