A DANREF Certified Reference Material for Chromate in Cement Jesper Kristiansen*, Jytte Molin Christensen and Kirsten Byrialsen National Institute of Occupational Health, Lersø Parkall�e 105, DK-2100 Copenhagen, Denmark. E-mail: jkr@ami.dk Two candidate reference materials for chromate in cement were produced in the DANREF network and certified in an interlaboratory study. Fifteen laboratories participated in the interlaboratory study and six different analytical methods were used.The certified values were estimated as the consensus mean of laboratory mean values (outliers excluded). Only results from laboratories using methods relying on chromate speciation were accepted. The certified values (±95% confidence limits) were 0.678 (±0.075) mg CrVI kg21 dry cement for the low level and 6.04 (±0.28) mg CrVI kg21 dry cement for the high level. Methods based on total chromium determination gave on average results that were 6.5% higher (both levels). However, the difference between speciation and non-speciation results was significant at the high concentration level only.Keywords: Hexavalent chromium; cement; interlaboratory study; certified reference material Hexavalent chromium is toxic to humans, in whom it may cause bronchial cancer,1–3 skin allergy4,5 and probably also asthma1 and renal injury.6,7 Portland cement contains 1–40 mg of watersoluble hexavalent chromium per kilogram of dry cement8 and the potential risk of working with cement has been recognized by occupational health services in several countries.In Denmark, a Working Environment Service Order9 has decreed a maximum content of 2 mg of extractable chromate per kilogram of dry cement, and this limit has also been established in other countries, including Sweden, Norway and Germany. Chromate† dissolves when cement is mixed with water, and hand contact with wet cement may lead to a skin diseases such as ‘cement eczema’ or ‘chromate ulcers’, which are serious, often debilitating, allergic skin reactions.1,10–13 Unfortunately, chromate cannot be removed from the raw cement, but must be reduced to chromium(iii) after dissolution.Chromium(iii) is insoluble in the strongly alkaline solution formed by adding water to cement.10 Reducing agents such as iron(ii) are therefore added to the cement, but sometimes other, and unfortunately less effective, reducing agents are used, e.g., sulfite. Sulfite does not reduce chromium(vi) in alkaline solution, but when the cement extract is acidified during standard routine analysis, chromium(vi) is reduced, giving rise to a false-negative result.In view of this, the Danish Working Environment Service demanded a revision of the Danish standard method for the determination of extractable chromate in cement. In the course of this work, the Danish National Institute of Occupational Health initiated an intercomparison between three Danish laboratories. The results of this intercomparison indicated systematic differences between the laboratories, indicating a need to improve the quality of the analysis. Hence candidate certified reference materials for chromate in cement were produced within the auspices of DANREF, which is a network of Danish laboratories that produces and supports the use of certified reference materials.14 In order to establish the certified values, an interlaboratory collaborative study was launched with the participation of 15 laboratories from seven European countries.This paper presents the results of this interlaboratory study. Experimental Reagents and Materials Cement was prepared by Aalborg Portland (Aalborg, Denmark). Potassium dichromate (pro analysi), 1,5-diphenylcarbohydrazide (pro analysi) and sulfuric acid (pro analysi) were purchased from Merck (Darmstadt, Germany). The indicator solution was prepared by dissolving 0.125 g of 1,5-diphenylcarbohydrazide in 25 ml of ethanol (r = 0.79 g ml21) in a 50 ml calibrated flask and diluting to volume with water.A chromate stock standard solution (50 mg l21 Cr6+) was prepared by dissolving 0.1414 g of dried potassium dichromate (K2Cr2O7) in water in a 1000 ml calibrated flask and diluting to volume with water. A chromate working standard solution (5 mg l21 Cr6+) was prepared by diluting 50.0 ml of the stock standard solution with water in a 500 ml calibrated flask. All water used was Milli-Q filtered water (Millipore, Molsheim, France).Cement was stored in polypropylene flasks fitted with screwcaps (100 ml or 1 l), that were acid cleaned before use. Extraction of cement with water took place in glass test-tubes (40 mm diameter). A filter funnel mounted on a 100 ml suction flask connected to a suction pump was used to separate aqueous extract from cement. Polypropylene filters (pore diameter 5 mm) were used. Pipettes (5 and 25 ml) and calibrated flasks (50 ml) were used for dilutions. Preparation of the candidate reference material The candidate reference material was produced by mixing two cements with low and high chromate levels.The total amount prepared at each level was approximately 1 kg. The cement was mixed in a ball mill with rubber pebbles for 30 min, and thereafter transferred to a 1 l polypropylene flask and stored under nitrogen. This stock cement was thoroughly remixed immediately before the preparation of candidate reference material samples.The samples were prepared by filling 100 ml polypropylene flasks with approximately 30 g of cement and flushing with nitrogen. The flasks were kept in tightly closed plastic bags to prevent uptake of water by the cement. The homogeneity was checked afterwards by chemical analysis. Apparatus The absorption of the diphenylcarbohydrazide–chromate complex at 540 nm was measured using a Hewlett-Packard (Avondale, PA, USA) Model 8450A UV/VIS spectrophotometer. Total chromium in the extracts was determined by † In the following, the expression ‘chromate in cement’ will be used synonymously with ‘extractable chromate in cement’, well knowing that not all chromate of the cement is extracted under normal working conditions.10 Analyst, October 1997, Vol. 122 (1155–1159) 1155atomic absorption spectrometry using a Perkin-Elmer (Norwalk, CT, USA) Z-5100 atomic absorption spectrometer with Zeeman-effect background correction and equipped with a PE- 60 autosampler.Determination of CrVI The Danish Standard DS 1020 describes a method for determination of chromate in cement based on the formation of a colored complex between the chromate ion and 1,5-diphenylcarbohydrazide. 15 The following is a brief summary of DS 1020. Several standard methods for chromate determinations issued in other countries, e.g., in Sweden and Germany,16 are similar in their principles to DS 1020. Extraction Cement (25 g, weighed exactly) was mixed with 25.0 ml of water.The mixture was shaken or stirred vigorously for 15 min. The aqueous extract was separated from cement by filtering the suspension through a dry filter funnel placed on a dry suction flask. Measurement A 5.00 ml volume of the filtrate was transferred into a 50 ml calibrated flask, 5 ml of sulfuric acid (1.8 mol l21) were added and the solution was diluted to approximately 40 ml with water. The solution was allowed to cool to room temperature, then 5.0 ml of indicator solution were added, water was added to the mark and the solution was thoroughly mixed.The absorbance was measured 15–30 min after addition of the indicator solution at 540 nm. A 5 ml volume of water treated as the filtrate served as blank. Calibration Volumes of 1.0, 2.0, 5.0, 10.0 and 15.0 ml of chromate standard solution were transferred into 50 ml calibrated flasks and 5 ml of sulfuric acid (1.8 mol l21) and water were added as described above.The calibration solutions contained 5, 10, 25, 50 and 75 mg of Cr6+, respectively. Analytical variation The between-laboratory standard deviation of the DS 1020 standard method was 0.5 mg of chromium(vi) per kilogram of dry cement at a concentration of 5.3 mg kg21 in an interlaboratory comparison.16 Homogeneity and Stability Homogeneity of the candidate reference material was coirmed before the material was distributed to laboratories in the intercomparison study by analyzing three samples at each level.The homogeneity was investigated in more detail in the intercomparison study17 by testing the pooled repeatability standard deviation (expressing the between-vial variability) versus the analytical standard deviation (expressing the analytical variability). The stability of the material at room temperature was indicated by a preliminary study of another cement batch. The stability of the candidate reference material will be monitored at regular 6 month intervals.Initial data from two study occasions showed no indications of instability. Intercomparison Study Two concentration levels were distributed to the laboratories participating in the study. In measuring the samples, the laboratories were advised to follow a common extraction procedure (as described above). All laboratories received written instructions for handling the materials and for extraction and analytical procedures. However, the laboratories were free to choose the analytical method for the determination of chromate in the extract.The analytical methods used are described briefly in Table 1. Three samples at each concentration level were sent to all laboratories. After determining the chromium(vi) content the laboratories returned the results on a standardized report sheet together with information about the analytical method, including extraction and calibration. Data Evaluation Data from the intercomparison study were evaluated using the DANREF PC program.14 This computer program is designed specifically for statistical evaluation of reference material data and documentation of the quality of reference materials in accordance with international guidelines.14,17,18 The steps in the statistical evaluation of the intercomparison results were as follows (in brief).First, the goodness of fit of the laboratory mean values to a normal distribution was tested by the Kolmogorov–Smirnov–Lilliefors test.19 Failure to fit a normal distribution may be due to the presence of outlying values among the laboratory results.Therefore, the second step in the evaluaiton was to detect extreme values among the laboratory mean values and variances by the Grubbs and Cochran test20 according to the IUPAC 1994 harmonized procedure for outlier removal.21 The significance level was 99%. Laboratories with extreme values were excluded from the data sets before further processing. Third, each data set was analyzed by a one-way analysis of variance (ANOVA) to test if all laboratory mean values estimate the same (consensus) mean value, and to estimate the within- and between-laboratory variation.The fourth and final step in the evaluation was testing for normal distribution of laboratory means with laboratories with extreme values excluded. Repeatability and reproducibility were estimated according to ISO 5725.20 For more general statistical analysis of data, the Minitab statistical software was used.Results and Discussion Interlaboratory Study A brief description of the analytical principles used in the intercalibration study is given in Table 1. Individual results, given as laboratory mean values with 95% confidence limits, are given in Fig. 1. All laboratories used 10–25 g of cement, a cement : water ratio of 1.00 and an extraction time of 15 min. As can be seen Table 1 Description of the methods used in the intercomparison study on chromium(vi) in cement Method ID Analytical principle Color Direct determination of chromate color DPC Reaction with 1,5-diphenylcarboxyhydrazide and determination of colored complex ETAAS Determination of total chromium in cement extracts by ETAAS FAAS Determination of total chromium in cement extracts by FAAS ICP1 Separation of chromium species on an aluminium oxide column and determination of chromate by ICP-AES ICP2 Determination of total chromium in cement extracts by ICP-AES 1156 Analyst, October 1997, Vol. 122from Fig. 1, 13 laboratories used a method identical with or similar to, the Danish standard method (method DPC). Two laboratories determined chromate spectrophotometrically by directly measuring the chromate color at 372.5 nm in the extract (method Color). One laboratory used an aluminum oxide column to separate chromate from other anionic and cationic species and detection of chromate by ICP-MS (method ICP1).22 Three methods did not include speciation of CrVI.Instead they relied on the assumption that CrIII precipitates in the alkaline Fig. 1 Results from the laboratory intercomparison study (a) Low level; (b) high level. Laboratory means and 95% confidence intervals. Methods: see Table 1. Methods marked with asterisks indicate total chromium determinations; results marked Cochran indicate Cochran outlier; results marked Grubbs indicate Grubbs outlier (see text). Summary of results [mean value ± half-width of the 95% confidence interval (number of accepted results)]: (a) low level, all methods, 0.704 ± 0.069 mg kg21 (n = 17); chromate-speciating methods, 0.678 ± 0.075 mg kg21 (n = 13); total chromium methods, 0.79 ± 0.22 mg kg21 (n = 4); (b) high level, all methods, 6.21 ± 0.27 mg kg21 (n = 17); chromate-speciating method, 6.04 ± 0.28 mg kg21 (n = 13); total chromium methods, 6.77 ± 0.25 mg kg21 (n = 4); the last value is significantly different from the mean of chromate-speciating methods, P < 0.01.Analyst, October 1997, Vol. 122 1157cement–water suspension. Two laboratories used FAAS, one laboratory used ETAAS and one laboratory ICP-AES (method ICP2) to determine the chromium content in the cement extract. All laboratories that participated in the study used aqueous standards prepared from potassium dichromate (purity > 99.9%) for calibration. Six sets of results from four laboratories did not pass the outlier test described in the Experimental section (Fig. 1). In five cases the reason was a too high analytical standard deviation (Cochran outlier), i.e., these laboratories reported significantly less precise results than the other laboratories. Laboratory L6 (low level) was identified as having an extreme mean value by the Grubbs test, but also in this case the variance was fairly large (Fig. 1). Consensus mean values were estimated after outlier exclusion for the whole data set and for subsets hereof, namely methods based on speciation (DPC, Color, ICP1) and nonspeciation methods (FAAS, ETAAS, ICP2) (Fig. 1). Although the difference between consensus mean values of speciating and non-speciating methods is significant only at the high concentration level, it could not be excluded that methods relying on the determination of total chromium in cement extracts in general yield higher results compared with methods that rely on speciation. Four laboratories reported results by both speciating and non-speciating methods, and the average relative deviation between their speciation and non-speciation results was 6.5% (SEM 8.2%) at both levels.Homogeneity The laboratory repeatability standard deviation contains variability from both analytical uncertainty and between-vial differences. In Table 2, the pooled repeatability standard deviation obtained in the intercomparison study is compared with the analytical repeatability standard deviation obtained by replicate analysis of cement extracts. The ratio of the variances is not significantly different from unity (F-test, P > 0.05).Hence the between-vial differences are not significant in comparison with the analytical standard deviation. Certified Values and Their Uncertainties As differences between speciation methods and total chromium methods could not be excluded, it was decided to apply the consensus means of speciation methods as the certified values. Hence the certified values should be 0.678 mg kg21 at the low level and 6.04 mg kg21 at the high level.For the candidate materials to be accepted as certified reference materials, the certified values must be relevant to the users, and established with an uncertainty estimate that encompasses all relevant uncertainty contributions.23 The ‘naturally’ occurring values of extractable chromate in cement are often in the range 1–15 mg CrVI kg21 dry cement, but occasionally cement may contain up to 40 mg kg21.8 Hence the consensus means of the two materials, 0.678 and 6.04 mg CrVI kg21 dry cement, are at the low end of the naturally occurring range.However, they fall on each side of the value 2 mg CrVI kg21 cement, which is a statutory limit value in several countries. Moreover, mixing of the two materials allows the evaluation of an analytical method in the range between 0.678 and 6.04 mg kg21. Hence the values are relevant with respect both to naturally occurring contents of chromate in cement and to the need in occupational health measurements. The uncertainties of the certified values were estimated as the half-width of the 95% confidence interval of the consensus value.Since the variability includes contributions from different vials and from the passage of time, the uncertainty encompasses both stability and homogeneity uncertainty components, in addition to analytical variability. The uncertainty is 0.075 mg kg21 (relative uncertainty 11%) at the low level and 0.28 mg kg21 (relative uncertainty 4.6%) at the high level.These values are based on speciation methods only, and do not include uncertainty from method-dependent variations. As mentioned above, methods measuring total chromium have a tendency to give slightly higher results (+6.5%). To give an acceptable range of the certified value for these methods, the uncertainty must be recalculated in order to account for methoddependent effects. Because method-dependent differences constitute an uncorrected systematic effect, the systematic effect is added linearly to the uncertainty.23 As the systematic effect (+6.5%) amounts to 0.044 mg kg21 at the low level and 0.39 mg kg21, at the high level, the recalculated uncertainties are 0.12 mg kg21 (low level) and 0.67 mg kg21 (high level).These uncertainties should be applied only when a total chromium method is used for the determination of chromate in cement and when the correction for systematic effects due to non-speciation is unknown.The certified values and the uncertainties apply only to the extraction procedure described in this paper. In conclusion, the certified reference materials may be useful in validating analytical methods for determination of chromate in cement and in demonstrating compliance with occupational exposure norms. The availability of certified reference materials is important in harmonizing chromate regulations in the occupational health field.Financial support from the Danish Agency for Trade and Industry is gratefully acknowledged. We are grateful to Aalborg Portland (Aalborg Denmark) for providing the cement material. M. Toxværd and G. S. Nielsen are acknowledged for carrying out chemical analyses. We are greatly indebted to the following laboratories and persons for their participation in this interlaboratory collaborative study: Arbeitssicherheit und Umweltschutz, Betriebsorganisation und Arbeitsstudium eV, Berlin, Germany (Dr.Kieburg), Berufsgenossenschaftliches Institut f�ur Arbeitssicherheit, Sankt Augustin, Germany (D. Breuer), Euroc Research AB, Slite, Sweden (K. Nyberg), Finncement AB, Pargas, Finland (P. Tuohiniemi), Force Institute, Brøndby, Denmark (O. Petersen), Henkel Bautechnik GmbH, Unna, Germany (K. Ehlermann), Murværkscentret DTI, Hasselager, Denmark (H. Gram Pedersen), Norcem AS, Brevik, Norway (E. Stoltenberg-Hansson), Sheffield Hallam University, UK (C. McCleod), SINTEF, Trondheim, Norway (I.Meland), Swedish National Testing and Research Institute, Borås, Sweden (J. Winblad), University for Horticulture and Food Industry, Hungary (P. Fodor), Aalborg Portland A/S, Aalborg, Denmark (J. Almeborg) and National Institute of Occupational Health, Oslo, Norway (Y. Thomassen). References 1 Environmental Health Criteria 61, World Health Organization, Geneva, 1988. Tale 2 Testing of homogeneity of the cement materials. sr = Pooled repeatability standard deviation obtained in the intercomparison; s0 = analytical standard deviation; df = degrees of freedom; NS = between-vial differences not significant (P > 0.05) sr, pooled s0, pooled repeatability SD analytical SD Cement (between vials)/ (within vial)/ F-statistic material mg kg21 mg kg21 ( = sr,2/s0 2) Low level 0.040 (df = 21) 0.031 (df = 16) 1.655 (NS) High level 0.26 (df = 18) 0.172 (df = 16) 2.285 (NS) 1158 Analyst, October 1997, Vol. 1222 IARC Monographs on the Evaluation of Carcinogenic Risks to Humans, IARC Monographs, Vol. 49, International Agency for Research on Cancer, Lyon, 1990. 3 Aitio, A., and Tomatis, L., in Trace Elements in Health and Disease, ed. Aitio, A., Aro, A., J�arvisalo, J., and Vainio, H., Royal Society of Chemistry, Cambridge, 1991. 4 Basketter, D. A., Bratico-Vangosa, G., Kaestner, W., Lally, C., and Bontinck, W. J., Contact Dermatitis, 1993, 28, 15. 5 Menn�e, T., Veien, N., Sjølin, K. E., and Maibach, H. I., Am.J. Contact Dermatitis, 1994, 5, 1. 6 Environmental Health Criteria 119, World Health Organization, Geneva, 1991. 7 Petersen, R., Mikkelsen, S., and Thomsen, O. F., Occup. Environ. Med., 1994, 51, 259. 8 Fregert, S., and Gruvberger, B., Berufsdermatosen, 1972, 20, 238. 9 Danish Working Environment Service Order, Water-Soluble Chromate in Cement, The Danish Working Environment Service, Copenhagen, 1983 (in Danish). 10 Fregert, S., Gruvberger, B., and Sandahl, E., Contact Dermatitis, 1979, 3, 39. 11 Goh, C. L., Gan, S. L., and Ngui, S. J., Contact Dermatitis, 1986, 15, 235. 12 Avnstorp, C., Contact Dermatitis, 1991, 25, 81. 13 Irvine, C., Pugh, C. E., Hansen, E. J., and Rycroft, R. J. G., Occup. Med., 1994, 44, 17. 14 Kristiansen, J., Christensen, J. M., Lillemark, L., Linde, S. A., Merry, J., Nyeland, B., and Petersen, O., Fresenius’ J. Anal. Chem., 1955, 352, 157. 15 Cement—Water-Soluble Chromate—Test Method, DS 1020, Danish Standard, Copenhagen, 1st edn., 1984 (in Danish). 16 Technische Regeln f�ur Gefahrstoffe 613, Bundesarbeitsblatt, 1993, 4, 63. 17 ISO Guide 35, Certification of Reference Materials, General and Statistical Principles, International Organization for Standardization, Geneva, 1985. 18 European Commission, Guidelines for the Production and Certification of BCR Reference Materials, Doc. BCR/48/93, Commission of the European Community, Brussels, 1994. 19 Conover, W. J., Practical Nonparametric Statistics, Wiley, New York, 2nd edn., 1981, p. 357. 20 ISO Standard 5725-2, Accuracy (Trueness and Precision) of Measurement Methods and Results—Part 2: Basic Method for the Determination of the Repeatability and Reproducibility of a Standard Measurement Method, International Organization for Standardization, Geneva, 1994. 21 Horwitz, W., Pure Appl. Chem., 1995, 67, 331. 22 Cox, A. G., Cook, I. G., and McLeod, C. W., Analyst, 1985, 110, 331. 23 Bureau International des Poid et Mesures, International Electrotechnical Commission, International Federation of Clinical Chemistry, International Organization for Standardization, International Union of Pure and Applied Chemistry, International Union of Pure and Applied Physics and International Organization of Legal Metrology, Guide to the Expression of Uncertainty in Measurement, International Organization for Standardization, Geneva, 1993.Paper 7/01619K Received March 7, 1997 Accepted June 23, 1997 Analyst, October 1997, Vol. 122 1159 A DANREF Certified Reference Material for Chromate in Cement Jesper Kristiansen*, Jytte Molin Christensen and Kirsten Byrialsen National Institute of Occupational Health, Lersø Parkall�e 105, DK-2100 Copenhagen, Denmark. E-mail: jkr@ami.dk Two candidate reference materials for chromate in cement were produced in the DANREF network and certified in an interlaboratory study. Fifteen laboratories participated in the interlaboratory study and six different analytical methods were used.The certified values were estimated as the consensus mean of laboratory mean values (outliers excluded). Only results from laboratories using methods relying on chromate speciation were accepted. The certified values (&plusfidence limits) were 0.678 (±0.075) mg CrVI kg21 dry cement for the low level and 6.04 (±0.28) mg CrVI kg21 dry cement for the high level. Methods based on total chromium determination gave on average results that were 6.5% higher (both levels).However, the difference between speciation and non-speciation results was significant at the high concentration level only. Keywords: Hexavalent chromium; cement; interlaboratory study; certified reference material Hexavalent chromium is toxic to humans, in whom it may cause bronchial cancer,1–3 skin allergy4,5 and probably also asthma1 and renal injury.6,7 Portland cement contains 1–40 mg of watersoluble hexavalent chromium per kilogram of dry cement8 and the potential risk of working with cement has been recognized by occupational health services in several countries.In Denmark, a Working Environment Service Order9 has decreed a maximum content of 2 mg of extractable chromate per kilogram of dry cement, and this limit has also been established in other countries, including Sweden, Norway and Germany. Chromate† dissolves when cement is mixed with water, and hand contact with wet cement may lead to a skin diseases such as ‘cement eczema’ or ‘chromate ulcers’, which are serious, often debilitating, allergic skin reactions.1,10–13 Unfortunately, chromate cannot be removed from the raw cement, but must be reduced to chromium(iii) after dissolution.Chromium(iii) is insoluble in the strongly alkaline solution formed by adding water to cement.10 Reducing agents such as iron(ii) are therefore added to the cement, but sometimes other, and unfortunately less effective, reducing agents are used, e.g., sulfite. Sulfite does not reduce chromium(vi) in alkaline solution, but when the cement extract is acidified during standard routine analysis, chromium(vi) is reduced, giving rise to a false-negative result.In view of this, the Danish Working Environment Service demanded a revision of the Danish standard method for the determination of extractable chromate in cement. In the course of this work, the Danish National Institute of Occupational Health initiated an intercomparison between three Danish laboratories.The results of this intercomparison indicated systematic differences between the laboratories, indicating a need to improve the quality of the analysis. Hence candidate certified reference materials for chromate in cement were produced within the auspices of DANREF, which is a network of Danish laboratories that produces and supports the use of certified reference materials.14 In order to establish the certified values, an interlaboratory collaborative study was launched with the participation of 15 laboratories from seven European countries.This paper presents the results of this interlaboratory study. Experimental Reagents and Materials Cement was prepared by Aalborg Portland (Aalborg, Denmark). Potassium dichromate (pro analysi), 1,5-diphenylcarbohydrazide (pro analysi) and sulfuric acid (pro analysi) were purchased from Merck (Darmstadt, Germany). The indicator solution was prepared by dissolving 0.125 g of 1,5-diphenylcarbohydrazide in 25 ml of ethanol (r = 0.79 g ml21) in a 50 ml calibrated flask and diluting to volume with water.A chromate stock standard solution (50 mg l21 Cr6+) was prepared by dissolving 0.1414 g of dried potassium dichromate (K2Cr2O7) in water in a 1000 ml calibrated flask and diluting to volume with water. A chromate working standard solution (5 mg l21 Cr6+) was prepared by diluting 50.0 ml of the stock standard solution with water in a 500 ml calibrated flask. All water used was Milli-Q filtered water (Millipore, Molsheim, France).Cement was stored in polypropylene flasks fitted with screwcaps (100 ml or 1 l), that were acid cleaned before use. Extraction of cement with water took place in glass test-tubes (40 mm diameter). A filter funnel mounted on a 100 ml suction flask connected to a suction pump was used to separate aqueous extract from cement. Polypropylene filters (pore diameter 5 mm) were used. Pipettes (5 and 25 ml) and calibrated flasks (50 ml) were used for dilutions.Preparation of the candidate reference material The candidate reference material was produced by mixing two cements with low and high chromate levels. The total amount prepared at each level was approximately 1 kg. The cement was mixed in a ball mill with rubber pebbles for 30 min, and thereafter transferred to a 1 l polypropylene flask and stored under nitrogen. This stock cement was thoroughly remixed immediately before the preparation of candidate reference material samples.The samples were prepared by filling 100 ml polypropylene flasks with approximately 30 g of cement and flushing with nitrogen. The flasks were kept in tightly closed plastic bags to prevent uptake of water by the cement. The homogeneity was checked afterwards by chemical analysis. Apparatus The absorption of the diphenylcarbohydrazide–chromate complex at 540 nm was measured using a Hewlett-Packard (Avondale, PA, USA) Model 8450A UV/VIS spectrophotometer.Total chromium in the extracts was determined by † In the following, the expression ‘chromate in cement’ will be used synonymously with ‘extractable chromate in cement’, well knowing that not all chromate of the cement is extracted under normal working conditions.10 Analyst, October 1997, Vol. 122 (1155–1159) 1155atomic absorption spectrometry using a Perkin-Elmer (Norwalk, CT, USA) Z-5100 atomic absorption spectrometer with Zeeman-effect background correction and equipped with a PE- 60 autosampler.Determination of CrVI The Danish Standard DS 1020 describes a method for determination of chromate in cement based on the formation of a colored complex between the chromate ion and 1,5-diphenylcarbohydrazide. 15 The following is a brief summary of DS 1020. Several standard methods for chromate determinations issued in other countries, e.g., in Sweden and Germany,16 are similar in their principles to DS 1020. Extraction Cement (25 g, weighed exactly) was mixed with 25.0 ml of water. The mixture was shaken or stirred vigorously for 15 min.The aqueous extract was separated from cement by filtering the suspension through a dry filter funnel placed on a dry suction flask. Measurement A 5.00 ml volume of the filtrate was transferred into a 50 ml calibrated flask, 5 ml of sulfuric acid (1.8 mol l21) were added and the solution was diluted to approximately 40 ml with water. The solution was allowed to cool to room temperature, then 5.0 ml of indicator solution were added, water was added to the mark and the solution was thoroughly mixed.The absorbance was measured 15–30 min after addition of the indicator solution at 540 nm. A 5 ml volume of water treated as the filtrate served as blank. Calibration Volumes of 1.0, 2.0, 5.0, 10.0 and 15.0 ml of chromate standard solution were transferred into 50 ml calibrated flasks and 5 ml of sulfuric acid (1.8 mol l21) and water were added as described above.The calibration solutions contained 5, 10, 25, 50 and 75 mg of Cr6+, respectively. Analytical variation The between-laboratory standard deviation of the DS 1020 standard method was 0.5 mg of chromium(vi) per kilogram of dry cement at a concentration of 5.3 mg kg21 in an interlaboratory comparison.16 Homogeneity and Stability Homogeneity of the candidate reference material was confirmed before the material was distributed to laboratories in the intercomparison study by analyzing three samples at each level.The homogeneity was investigated in more detail in the intercomparison study17 by testing the pooled repeatability standard deviation (expressing the between-vial variability) versus the analytical standard deviation (expressing the analytical variability). The stability of the material at room temperature was indicated by a preliminary study of another cement batch. The stability of the candidate reference material will be monitored at regular 6 month intervals.Initial data from two study occasions showed no indications of instability. Intercomparison Study Two concentration levels were distributed to the laboratories participating in the study. In measuring the samples, the laboratories were advised to follow a common extraction procedure (as described above). All laboratories received written instructions for handling the materials and for extraction and analytical procedures. However, the laboratories were free to choose the analytical method for the determination of chromate in the extract.The analytical methods used are described briefly in Table 1. Three samples at each concentration level were sent to all laboratories. After determining the chromium(vi) content the laboratories returned the results on a standardized report sheet together with information about the analytical method, including extraction and calibration. Data Evaluation Data from the intercomparison study were evaluated using the DANREF PC program.14 This computer program is designed specifically for statistical evaluation of reference material data and documentation of the quality of reference materials in accordance with international guidelines.14,17,18 The steps in the statistical evaluation of the intercomparison results were as follows (in brief). First, the goodness of fit of the laboratory mean values to a normal distribution was tested by the Kolmogorov–Smirnov–Lilliefors test.19 Failure to fit a normal distribution may be due to the presence of outlying values among the laboratory results.Therefore, the second step in the evaluaiton was to detect extreme values among the laboratory mean values and variances by the Grubbs and Cochran test20 according to the IUPAC 1994 harmonized procedure for outlier removal.21 The significance level was 99%. Laboratories with extreme values were excluded from the data sets before further processing.Third, each data set was analyzed by a one-way analysis of variance (ANOVA) to test if all laboratory mean values estimate the same (consensus) mean value, and to estimate the within- and between-laboratory variation. The fourth and final step in the evaluation was testing for normal distribution of laboratory means with laboratories with extreme values excluded. Repeatability and reproducibility were estimated according to ISO 5725.20 For more general statistical analysis of data, the Minitab statistical software was used.Results and Discussion Interlaboratory Study A brief description of the analytical principles used in the intercalibration study is given in Table 1. Individual results, given as laboratory mean values with 95% confidence limits, are given in Fig. 1. All laboratories used 10–25 g of cement, a cement : water ratio of 1.00 and an extraction time of 15 min. As can be seen Table 1 Description of the methods used in the intercomparison study on chromium(vi) in cement Method ID Analytical principle Color Direct determination of chromate color DPC Reaction with 1,5-diphenylcarboxyhydrazide and determination of colored complex ETAAS Determination of total chromium in cement extracts by ETAAS FAAS Determination of total chromium in cement extracts by FAAS ICP1 Separation of chromium species on an aluminium oxide column and determination of chromate by ICP-AES ICP2 Determination of total chromium in cement extracts by ICP-AES 1156 Analyst, October 1997, Vol. 122from Fig. 1, 13 laboratories used a method identical with or similar to, the Danish standard method (method DPC). Two laboratories determined chromate spectrophotometrically by directly measuring the chromate color at 372.5 nm in the extract (method Color). One laboratory used an aluminum oxide column to separate chromate from other anionic and cationic species and detection of chromate by ICP-MS (method ICP1).22 Three methods did not include speciation of CrVI. Instead they relied on the assumption that CrIII precipitates in the alkaline Fig. 1 Results from the laboratory intercomparison study (a) Low level; (b) high level. Laboratory means and 95% confidence intervals. Methods: see Table 1. Methods marked with asterisks indicate total chromium determinations; results marked Cochran indicate Cochran outlier; results marked Grubbs indicate Grubbs outlier (see text). Summary of results [mean value ± half-width of the 95% confidence interval (number of accepted results)]: (a) low level, all methods, 0.704 ± 0.069 mg kg21 (n = 17); chromate-speciating methods, 0.678 ± 0.075 mg kg21 (n = 13); total chromium methods, 0.79 ± 0.22 mg kg21 (n = 4); (b) high level, all methods, 6.21 ± 0.27 mg kg21 (n = 17); chromate-speciating method, 6.04 ± 0.28 mg kg21 (n = 13); total chromium methods, 6.77 ± 0.25 mg kg21 (n = 4); the last value is significantly different from the mean of chromate-speciating methods, P < 0.01.Analyst, October 1997, Vol. 122 1157cement–water suspension. Two laboratories used FAAS, one laboratory used ETAAS and one laboratory ICP-AES (method ICP2) to determine the chromium content in the cement extract. All laboratories that participated in the study used aqueous standards prepared from potassium dichromate (purity > 99.9%) for calibration. Six sets of results from four laboratories did not pass the outlier test described in the Experimental section (Fig. 1). In five cases the reason was a too high analytical standard deviation (Cochran outlier), i.e., these laboratories reported significantly less precise results than the other laboratories. Laboratory L6 (low level) was identified as having an extreme mean value by the Grubbs test, but also in this case the variance was fairly large (Fig. 1). Consensus mean values were estimated after outlier exclusion for the whole data set and for subsets hereof, namely methods based on speciation (DPC, Color, ICP1) and nonspeciation methods (FAAS, ETAAS, ICP2) (Fig. 1). Although the difference between consensus mean values of speciating and non-speciating methods is significant only at the high concentration level, it could not be excluded that methods relying on the determination of total chromium in cement extracts in general yield higher results compared with methods that rely on speciation. Four laboratories reported results by both speciating and non-speciating methods, and the average relative deviation between their speciation and non-speciation results was 6.5% (SEM 8.2%) at both levels.Homogeneity The laboratory repeatability standard deviation contains variability from both analytical uncertainty and between-vial differences. In Table 2, the pooled repeatability standard deviation obtained in the intercomparison study is compared with the analytical repeatability standard deviation obtained by replicate analysis of cement extracts.The ratio of the variances is not significantly different from unity (F-test, P > 0.05). Hence the between-vial differences are not significant in comparison with the analytical standard deviation. Certified Values and Their Uncertainties As differences between speciation methods and total chromium methods could not be excluded, it was decided to apply the consensus means of speciation methods as the certified values. Hence the certified values should be 0.678 mg kg21 at the low level and 6.04 mg kg21 at the high level.For the candidate materials to be accepted as certified reference materials, the certified values must be relevant to the users, and established with an uncertainty estimate that encompasses all relevant uncertainty contributions.23 The ‘naturally’ occurring values of extractable chromate in cement are often in the range 1–15 mg CrVI kg21 dry cement, but occasionally cement may contain up to 40 mg kg21.8 Hence the consensus means of the two materials, 0.678 and 6.04 mg CrVI kg21 dry cement, are at the low end of the naturally occurring range.However, they fall on each side of the value 2 mg CrVI kg21 cement, which is a statutory limit value in several countries. Moreover, mixing of the two materials allows the evaluation of an analytical method in the range between 0.678 and 6.04 mg kg21. Hence the values are relevant with respect both to naturally occurring contents of chromate in cement and to the need in occupational health measurements.The uncertainties of the certified values were estimated as the half-width of the 95% confidence interval of the consensus value. Since the variability includes contributions from different vials and from the passage of time, the uncertainty encompasses both stability and homogeneity uncertainty components, in addition to analytical variability. The uncertainty is 0.075 mg kg21 (relative uncertainty 11%) at the low level and 0.28 mg kg21 (relative uncertainty 4.6%) at the high level.These values are based on speciation methods only, and do not include uncertainty from method-dependent variations. As mentioned above, methods measuring total chromium have a tendency to give slightly higher results (+6.5%). To give an acceptable range of the certified value for these methods, the uncertainty must be recalculated in order to account for methoddependent effects. Because method-dependent differences constitute an uncorrected systematic effect, the systematic effect is added linearly to the uncertainty.23 As the systematic effect (+6.5%) amounts to 0.044 mg kg21 at the low level and 0.39 mg kg21, at the high level, the recalculated uncertainties are 0.12 mg kg21 (low level) and 0.67 mg kg21 (high level).These uncertainties should be applied only when a total chromium method is used for the determination of chromate in cement and when the correction for systematic effects due to non-speciation is unknown.The certified values and the uncertainties apply only to the extraction procedure described in this paper. In conclusion, the certified reference materials may be useful in validating analytical methods for determination of chromate in cement and in demonstrating compliance with occupational exposure norms. The availability of certified reference materials is important in harmonizing chromate regulations in the occupational health field.Financial support from the Danish Agency for Trade and Industry is gratefully acknowledged. We are grateful to Aalborg Portland (Aalborg Denmark) for providing the cement material. M. Toxværd and G. S. Nielsen are acknowledged for carrying out chemical analyses. We are greatly indebted to the following laboratories and persons for their participation in this interlaboratory collaborative study: Arbeitssicherheit und Umweltschutz, Betriebsorganisation und Arbeitsstudium eV, Berlin, Germany (Dr.Kieburg), Berufsgenossenschaftliches Institut f�ur Arbeitssicherheit, Sankt Augustin, Germany (D. Breuer), Euroc Research AB, Slite, Sweden (K. Nyberg), Finncement AB, Pargas, Finland (P. Tuohiniemi), Force Institute, Brøndby, Denmark (O. Petersen), Henkel Bautechnik GmbH, Unna, Germany (K. Ehlermann), Murværkscentret DTI, Hasselager, Denmark (H. Gram Pedersen), Norcem AS, Brevik, Norway (E. Stoltenberg-Hansson), Sheffield Hallam University, UK (C.McCleod), SINTEF, Trondheim, Norway (I. Meland), Swedish National Testing and Research Institute, Borås, Sweden (J. Winblad), University for Horticulture and Food Industry, Hungary (P. Fodor), Aalborg Portland A/S, Aalborg, Denmark (J. Almeborg) and National Institute of Occupational Health, Oslo, Norway (Y. Thomassen). References 1 Environmental Health Criteria 61, World Health Organization, Geneva, 1988. Tale 2 Testing of homogeneity of the cement materials. sr = Pooled repeatability standard deviation obtained in the intercomparison; s0 = analytical standard deviation; df = degrees of freedom; NS = between-vial differences not significant (P > 0.05) sr, pooled s0, pooled repeatability SD analytical SD Cement (between vials)/ (within vial)/ F-statistic material mg kg21 mg kg21 ( = sr,2/s0 2) Low level 0.040 (df = 21) 0.031 (df = 16) 1.655 (NS) High level 0.26 (df = 18) 0.172 (df = 16) 2.285 (NS) 1158 Analyst, October 1997, Vol. 1222 IARC Monographs on the Evaluation of Carcinogenic Risks to Humans, IARC Monographs, Vol. 49, International Agency for Research on Cancer, Lyon, 1990. 3 Aitio, A., and Tomatis, L., in Trace Elements in Health and Disease, ed. Aitio, A., Aro, A., J�arvisalo, J., and Vainio, H., Royal Society of Chemistry, Cambridge, 1991. 4 Basketter, D. A., Bratico-Vangosa, G., Kaestner, W., Lally, C., and Bontinck, W. J., Contact Dermatitis, 1993, 28, 15. 5 Menn�e, T., Veien, N., Sjølin, K. E., and Maibach, H. I., Am. J. Contact Dermatitis, 1994, 5, 1. 6 Environmental Health Criteria 119, World Health Organization, Geneva, 1991. 7 Petersen, R., Mikkelsen, S., and Thomsen, O. F., Occup. Environ. Med., 1994, 51, 259. 8 Fregert, S., and Gruvberger, B., Berufsdermatosen, 1972, 20, 238. 9 Danish Working Environment Service Order, Water-Soluble Chromate in Cement, The Danish Working Environment Service, Copenhagen, 1983 (in Danish). 10 Fregert, S., Gruvberger, B., and Sandahl, E., Contact Dermatitis, 1979, 3, 39. 11 Goh, C. L., Gan, S. L., and Ngui, S. J., Contact Dermatitis, 1986, 15, 235. 12 Avnstorp, C., Contact Dermatitis, 1991, 25, 81. 13 Irvine, C., Pugh, C. E., Hansen, E. J., and Rycroft, R. J. G., Occup. Med., 1994, 44, 17. 14 Kristiansen, J., Christensen, J. M., Lillemark, L., Linde, S. A., Merry, J., Nyeland, B., and Petersen, O., Fresenius’ J. Anal. Chem., 1955, 352, 157. 15 Cement—Water-Soluble Chromate—Test Method, DS 1020, Danish Standard, Copenhagen, 1st edn., 1984 (in Danish). 16 Technische Regeln f�ur Gefahrstoffe 613, Bundesarbeitsblatt, 1993, 4, 63. 17 ISO Guide 35, Certification of Reference Materials, General and Statistical Principles, International Organization for Standardization, Geneva, 1985. 18 European Commission, Guidelines for the Production and Certification of BCR Reference Materials, Doc. BCR/48/93, Commission of the European Community, Brussels, 1994. 19 Conover, W. J., Practical Nonparametric Statistics, Wiley, New York, 2nd edn., 1981, p. 357. 20 ISO Standard 5725-2, Accuracy (Trueness and Precision) of Measurement Methods and Results—Part 2: Basic Method for the Determination of the Repeatability and Reproducibility of a Standard Measurement Method, International Organization for Standardization, Geneva, 1994. 21 Horwitz, W., Pure Appl. Chem., 1995, 67, 331. 22 Cox, A. G., Cook, I. G., and McLeod, C. W., Analyst, 1985, 110, 331. 23 Bureau International des Poid et Mesures, International Electrotechnical Commission, International Federation of Clinical Chemistry, International Organization for Standardization, International Union of Pure and Applied Chemistry, International Union of Pure and Applied Physics and International Organization of Legal Metrology, Guide to the Expression of Uncertainty in Measurement, International Organization for Standardization, Geneva, 1993. Paper 7/01619K Received March 7, 1997 Accepted June 23, 1997 Analyst, October 1997, Vol.