RM certification is the whole process of obtaining the property values and their uncertainties, which includes homogeneity testing, stability testing, and RM characterization [5]. ISO Guide 35 [1] requires one to show that the value of such a certified property does not exhibit a systematic error specific to a method or to a laboratory. By widespread opinion, correctness of analytical results is an obvious prerequisite for the RM characterization in contrast to stability and homogeneity studies in which analytical bias is acceptable [5]. In the present paper we would like to discuss primarily the situation when a systematic error in measurement (analytical) results, specific to a measurement method or to a laboratory developing an IHRM, can be overcome. This is possible when a certification of an IHRM is based on a comparative approach providing a characterization of the value carried by the IHRM in comparison to the value carried by the reference material with higher metrological status and sufficiently similar matrix (a CRM). Papers published in this section do not necessarily reflect the opinion of the Editors, the Editorial Board and the Publisher. Apart from exceptional circumstances, they are not submitted to the usual referee procedure and go essentially unaltered. Accred Qual Assur (2002) 7:122 124 DOI 10.1007/s00769-001-0436-9 ' Springer-Verlag 2002 Ilya Kuselman Alexander Weisman Wolfhard Wegscheider Traceable property values of in-house reference materials Abstract The traceability of in-house reference materials (IHRM) is discussed. It is shown that a systematic error in results of a measured value, specific to a measurement method or to a laboratory developing an IHRM, can be overcome if a comparative approach to IHRM characterization is used. A traceability chain of the value carried by the IHRM to the value carried by the reference material with higher metrological status and sufficiently similar matrix (for example, a certified reference material CRM according to ISO Guide 30) is helpful in such a case. The chain is realized when the IHRM samples are analysed simultaneously with the CRM samples under the same conditions. This and other traceability chains necessary for the IHRM development are examined as the measurement information sources. Keywords Traceability • Reference materials • Uncertainty estimation • Certification • Comparative approach Introduction The traceability of the value carried by a reference material (RM) should be demonstrated by the RM producer [1 3]. The producer shall provide the traceability of results of its measurements to the national or international measurement standards. Where this is not possible, the correlation of results with the values of national or international certified reference materials (CRMs) is required. Ideally, the values of the CRMs should themselves be traceable [2]. More than 220 producers of CRMs throughout the world produce today 12,000 20,000 materials with dif ferent matrixes, analytes and properties [4]. However, many testing (analytical) laboratories cannot find suitable CRMs in the market and develop in-house reference materials (IHRMs) themselves. Often IHRMs are developed in a laboratory to conserve the corresponding expensive CRMs. For example, a pharmaceutical company Chemagis Ltd. produces 30 active pharmaceutical ingredients: steroids, benzodiazepines, antihistamines, hipolipidaemics, blood flow reactants, etc. Only for a few of them Mometasone Furoate, Fluticasone Propionate and Dobutamine Hydrochloride are of ficial reference standards for assay supplied by US, British and European Pharmacopoeias with prices of about $ 180 per unit (50 200 mg). Thus, to support its customers Chemagis is forced to develop IHRMs for assay as well as for impurities and related substances of each produced compound. Therefore, certification of such IHRMs that leads to traceable values is very important. Comparative approach This approach is based on the transmission of the measurement information from a corresponding CRM to the IHRM [6 9]. The following conditions are required for the IHRM characterization: 1) IHRM and CRM are similar materials, 2) the difference in concentrations of the IHRM and CRM matrix components, which are not under characterization, and corresponding properties of the materials (for example, solubility) does not hinder the use of the same measurement (analytical) method for both the IHRM and CRM, 3) the concentrations of a component under characterization in IHRM and CRM do not differ by more than a factor of two [8]. Test portions in pairs one of the IHRM and one of the CRM are analysed by the same analyst and method in the same laboratory and conditions, each pair practically simultaneously. The concentration of the IHRM component under characterization, CIHRM, is compared with its certified concentration, CCRM, in the CRM using the differences in the analysis results Ei=(CIHRM i C CRM i ) for all pairs: i=1, 2, ..., n (n≥20). From this comparison the characterized value is calculated as CIHRM = CCRM + Eavg , where Eavg = Σ Ei/n. (1) Obviously, even if CIHRM i and CCRM i have an additive systematic error, Ei is free from this error by definition. Additivity of bias is a reasonable approximation for nearly iden- 270 I. Kuselman • A. Weisman • W. Wegscheider Table 1 Results of the alcohol determination in the IHRM and CRM LGC5404 i CIHRM i CCRM i Ei 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Average Standard deviation 4.75 4.80 4.78 4.77 4.76 4.77 4.76 4.76 4.75 4.76 4.75 4.80 4.79 4.80 4.78 4.76 4.75 4.78 4.75 4.76 4.769 0.018 5.03 5.04 5.03 5.00 5.00 5.03 5.02 5.03 5.00 4.99 5.02 5.01 5.03 5.02 5.04 5.02 4.99 5.02 5.03 5.01 5.018 0.015 0.28 0.24 0.25 0.23 0.24 0.26 0.26 0.27 -0.25 0.23 0.27 0.21 0.24 0.22 0.26 0.26 0.24 0.24 0.28 0.25 0.249 0.019 tical matrices: a multiplicative bias component is assumed negligible at similar concentrations of the analyte in the CRM and IHRM as limited above. Therefore, Eavg and CIHRM by Eq. (1) are also unbiased. So, the characterization standard uncertainty [10] is u2 (CIHRM) = [u2 (CCRM) + u2 (Eavg )]1/2 u2 (Eavg ) = Σ (Ei E avg )2 / (n 1) n, (2) (3) and u(CCRM) is the standard uncertainty of the value carried by the CRM. If the CRM can be selected according to the criteria of ILAC-G9 Guidelines [11] as satisfactory or acceptable, i.e., if u(CIHRM)/u(CCRM)>4, the uncertainty of the value carried by the CRM is negligible. Otherwise it should be taken into account by Eq. (2). A numerical example of the calculations is shown below. Characterization of an IHRM of ethanol in water. A reference spirit CRM LGC5404 from LGC with the certified value of the ethanol concentration in water of 5.00–0.03% by volume at the 95% level of confidence is used for the alcohol determination in beer [12]. So, CCRM=5.00% and u(CCRM)=0.03/2=0.015% by volume (2 is the coverage factor). To prepare an IHRM a suitable ethanol was diluted gravimetrically with water to produce a solution close to 5% ethanol concentration. The solution when thoroughly mixed was suggested to be homogeneous. Twenty test portions in pairs one of the IHRM and one of the CRM LGC5404 were analysed in the same conditions using a standard gas-chro- It is important that the information on a value carried by a CRM is obtained from measurement (analytical) results in other laboratories (not or not only in the laboratory-producer of the IHRM), probably using different methods. Therefore, if a suitable (adequate) CRM is not available, participation of a second laboratory in an IHRM characterization is desirable, even when an unbiased validated method is used, to evaluate a possible bias specific to the laboratory developing the IHRM. Traceability chains of the method form in this case a traceability scheme of the value carried by the IHRM. Fig. 1 Traceability chains of a value CIHRM carried by an IHRM. Here m is the mass of the IHRM and V is its volume matographic method. Results of the analysis are presented in Table 1. The average difference between the results for IHRM and CRM pairs shown in Table 1 is Eavg= 0.25% by volume. The value carried by the IHRM according to Eq. (1) is CIHRM=5.00+( 0.25)=4.75% by volume. Note, that the average result of the CRM analysis CCRM-avg = Σ CCRM-i /20=5.02% is biased from the certified value CCRM= 5.00% for 0.02% by volume. It is not influencing the value of CIHRM as far as CCRM avg is not used directly in the CIHRM calculation. The standard uncertainty in Eavg calculated by Eq. (3) is u(Eavg)=0.004% by volume. So, the characterization standard uncertainty by Eq. (2) is u(CIHRM)= [0.0152+0.0042]1/2=0.016% by volume. Conclusions It is shown that a systematic error in results of a measured value, specific to a measurement method or to a laboratory developing an IHRM, can be overcome if a comparative approach to IHRM characterization is used. A traceability chain from the value carried by the IHRM to the value carried by the reference material with higher metrological status and sufficiently similar matrix is helpful in such cases. The chain is realized when the IHRM samples are analysed simultaneously with the CRM samples under the same conditions. If a suitable (adequate) CRM is not available, participation of a second laboratory in an IHRM characterization is desirable, even when an unbiased validated method is used, to assess the laboratory bias. Traceability chains of the method form in this case a traceability scheme of the value carried by the IHRM. Traceability chains If the first steps of a procedure as described above are weighing and dissolving, a concentration of the IHRM component under characterization CIHRM (the value carried by the IHRM) expressed in % by volume has three traceability chains: 1) of the IHRM mass to the SI kg, 2) of the IHRM volume after dissolving, also to the SI kg, since calibration of volumetric flasks is performed gravimetrically, and 3) of the IHRM carried value comparison to the CRM carried value. In such a case CIHRM can be shown as a sun with three beams [13] that are metrological pyramids (traceability chains with minimal uncertainty in the top of the pyramid and maximal in the bottom): see Fig. 1. As a rule, uncertainties in traceability chains to the SI kg are negligible in comparison with the ones from the own analytical process. So, the problem is just how narrow the bottom of the third pyramid is, i.e., how much information is lost in the chain IHRM CRM. References 1. ISO Guide 35 (1989) Certification of Reference Materials General and Statistical Principles. 2nd ed, Geneva 2. ILAC-G12 (2000) Guidelines for the Requirements for the Competence of Reference Materials Producers 3. De Bievre P (2000) Accred Qual Assur 5:224 230 4. Zschunke A (ed) (2000) Reference Materials in Analytical Chemistry. A Guide for Selection and Use. Springer, Berlin, p 211 5. Van der Veen AMH, Linsinger TPJ, Schimmel H, Lamberty A, Pauwels J (2001) Accred Qual Assur 6:290 294 6. Shaevich AB (1971) Measurement and Standardization of Chemical Composition of Substances. Publishing House of Standards, Moscow, 280 p (in Russ.) 7. Shaevich AB (1987) Reference Materials for Analytical Purposes. Publishing House Chemistry , Moscow , pp 147 149 (in Russ.) Traceable property values of in-house reference materials 8. Pliner YuL (ed) (1988) Metrological Assurance of Materials Composition in Ferrous Metallurgy. Handbook. 2nd ed, Publishing House Metallur gy , Moscow, pp 157 163 (in Russ.) 9. Pliner YuL, Kuzmin IM (1989) Metrological Problems in Analytical Control of Metallic Products Quality. Publishing House Metallur gy , Moscow , pp 157 165 (in Russ.) 10. EURACHEM/CITAC Guide (2000) Quantifying Uncertainty in Analytical Measurement. 2nd ed, Teddington 11. ILAC-G9 (1996) Guidelines for the Selection and Use of Certified Reference Materials 12. Kuselman I, Anisimov B, Goldfeld I (2001) Accred Qual Assur 6:107 1 10 13. Kuselman I (2000) Proceedings of The Millennium International Conference of The Israel Society for Quality. Vol I, Jerusalem, pp 124 129 MEETING REPORT I. Kuselman (✉) The National Physical Laboratory of Israel (INPL), Givat Ram, Jerusalem 91904, Israel e-mail: [email protected] Tel.: +972-2-6536534 Fax: +972-2-6520797 A. Weisman Chemagis Ltd., P.O.Box 9091, Tel-Aviv 61090, Israel W. Wegscheider Department of General and Analytical Chemistry, University of Leoben, Franz-Josef Strasse 18, 8700 Leoben, Austria 271
1/--страниц