ANALYST, JANUARY 1992, VOL. 117 19 Instrumental Comparison for the Determination of Cadmium and Lead in Calcium Supplements and Other Calcium-rich Matrices Bernard P. Bourgoin" Environmental and Resource Studies Program, Trent University, Peterborough, Canada K9J 7B8 Dave Boomer and Mark J. Powell Ontario Ministry of the Environment, laboratory Services Branch, Rexdale, Canada M9W 51 1 Scott W i II ie National Research Council Canada, Institute for Environmental Chemistry, Ottawa, Canada K I A OR6 Duart Edgar Nissei Sang yo Canada Inc., Rexdale, Canada M9 W 6A4 Don Evans Ontario Ministry of the Environment, laboratory Services Branch, Dorset, Canada POA 1 EO Three brands of Ca supplement, a laboratory-reagent grade CaC03 and a certified reference material (International Atomic Energy Agency H-5 Animal Bone) were analysed for Cd and Pb by four different analytical techniques, viz., anodic stripping voltammetry, inductively coupled plasma mass spectrometry, flame atomic absorption spectrometry and electrothermal atomic absorption spectrometry.The Pb levels measured by the four techniques in the bone powder were within the certified Pb level in this certified reference material. Similarly, no significant differences [ p <0.05; analysis of variance (ANOVA)] were observed in samples with Pb concentrations greater than 1 pg g-1. However, the Pb levels in the laboratory-reagent grade CaC03 obtained by flame atomic absorption spectrometry (0.79 pg g-1) averaged about three times higher than those measured by the other three techniques (i.e., 0.25 pg g-1).Although no significant differences ( p <0.05; ANOVA) in Cd levels were observed within any of the samples (intra-sample variability), the Cd concentration measured in the different Ca supplements (inter-sample variability) varied by three orders of magnitude (ranging from 0.07 to 3.59 pg 9-1). Keywords: Cadmium; lead; calcium carbonate; interference; supplements The accurate identification of foodstuffs with high Pb concen- trations is becoming increasingly important because of the substantial decrease in the total Pb content in children's diets over the past 15 years.' It has been recognized for some time that Ca supplements, classified as foodstuffs in the US, contain variable amounts of Pb and that the ingestion of some of these products for extended periods of time could be considered a potential health hazard.2.3 Cadmium, another toxic trace metal, behaves similarly to Pb in the environment and is often associated with Ca-rich material such as bivalve shells.4 The various analytical techniques that have been used to determine heavy metal levels in Ca-rich matrices include anodic stripping voltammetry (ASV) ,576 inductively coupled plasma emission spectrometry,7 flame atomic absorption spectrometry (FAAS) and electrothermal atomic absorption spectrometry (ETAAS) ,&lo and proton-induced X-ray emis- sion.1'3'2 There is no standardized method for determining trace levels in these matrices and the extent of sample pre-treatment ranges from simply analysing untreated sample digests to more complex and time-consuming clean-up proce- dures involving the separation and/or coprecipitation of the trace metals from the major ions in solution.Further, there is no certified reference material (CRM) in which a considerable array of heavy metal levels is certified to verify the accuracy of a particular method. In this work, four different analytical techniques, viz., ASV, inductively coupled plasma mass spectrometry (ICP- MS), FAAS and ETAAS, were used to measure Cd and Pb levels in five different Ca-rich matrices. Sample pre-treatment * Present address: National Water Research Institute, Lakes Research Branch, P.O. Box 5050, 867 Lakeshore Road, Burlington, Ontario, Canada L7R 4A6. was purposely kept to a minimum so as to determine under what circumstances interferences were more problematic. Experimental Sample Types Although there are over 400 types of Ca supplement available on the North American market, based on the form in which elemental Ca occurs, these supplements can be grouped into the following three main categories: CaC03, hydroxyapatite or calcium phosphate [Calo(P04)6(OH)2] and Ca bound to various organic chelates (e.g., gluconate and amino acids).Many of these products are also commonly supplemented with other minerals such as Mg or Zn. The five types of sample selected in this work are listed in Table 1, which also summarizes the various forms and concentrations of Ca and other elements contained in the powders. Three different brands of Ca supplement were included (Brands A-C). The elemental Ca in Brands A and B occurred as CaC03 and consisted of ground oyster shells.Group B was further supplemented with Mg and Zn. Brand C also contained Mg and Zn, but all the minerals in this product occurred as chelates. The other two samples included a laboratory-reagent grade CaC03 powder and a CRM, H-5 Animal Bone, produced by the International Atomic Energy Agency (IAEA), Austria. Participants and Analytical Instrumentation Four independent laboratories: a federal and a provincial agency, a university laboratory and a private firm participated in these analyses. These agencies are listed in Table 2 together with the different types of instrument used and the pertinent analytical conditions under which the analyses were per- formed.20 ANALYST, JANUARY 1992, VOL.117 Procedure The laboratory at Trent University was responsible for the selection of the different samples and for soliciting the participation of the other three laboratories. The identifica- tion of the samples and the identity of the participating laboratories were only disclosed to the other participants after all the analyses had been completed. All of the samples were ashed in a muffle furnace at 425 "C for 24 h at the Trent University laboratory. The temperature was gradually increased at a rate of about 100 "C h-l to avoid combustion of the powders. Sub-samples of the ashed samples were then sent to the National Research Council Canada (NRCC) laboratory for subsequent pre-treatment and analy- sis. The sample digests for the other three methods were prepared at the Trent University laboratory.Table 1 Five different Ca-rich samples analysed, corresponding levels and forms of elements specified by the manufacturer. Brands A-C represent various Ca supplements NRCC sample treatment Approximately 0.5 g of sample was weighed into 120 ml digestion vessels (Teflon PFA) to which were added 6 ml of 16 rnol dm-3 HN03. The vessels were capped and heated in a microwave oven (CEM MDS SlD) at high power for 20 min. The internal pressure was maintained at 310 kPa. The vessels were cooled, opened and heated on a hot-plate under a heat lamp to evaporate the remaining acid to a volume of about 1 ml after which the solutions were diluted to 25 ml with distilled, de-ionized water (DDW). At this point the solutions of the laboratory-reagent grade CaC03 and the IAEA CRM samples were clear and could be analysed.The three brands of Ca supplement had undissolved material present. These solutions were filtered through a pre-washed (1.5 rnol dm-3 HCI) Millipore filter (0.45 pm pore size) housed in a Gelman poly(su1fone) filter-funnel, the filtrate collected and analysed. Guaranteed Sample (mineraVelement) Form specified Laboratory-reagent grade CaC03 powder Ca: 400 mg g-I Zn: (0.005% Pb: <0.005% Cd: <0.005% Ca: 330 mg g- CaC03 precipitated As impurities As impurities As impurities CaC03 from oyster shells CaC03 from oyster shells From oxide From gluconate From amino acid From amino acid From amino acid chelate chelate chelate Brand A Brand B Ca: 170mgg-* Mg: 90 mg g-1 Zn: 30 mg g-1 Ca: 160 mg g-l Brand C Mg: 160 mg g-1 Zn: 8 mg g-1 Trent University sample treatmenl For ASV analyses, approximately 0.25 g of the ashed samples was dissolved in 1 ml of 12 rnol dm-3 HCI.Approximately 5 ml of 0.2 rnol dm-3 sodium acetate (NaOAc) buffer were added and the solutions filtered through pre-washed (1.5 rnol dm-3 HCI) Millipore membrane filters (0.45 pm pore size). The filtrate was adjusted to a pH of 1 S O k 0.05 (NaOAc buffer or 12 rnol dm-3 HCl). The typical volume of the solutions after IAEA CRM H-5 Animal Bone Ca: 212 mg g-1(8)* P: 102 mg g-1(8) Mg: 3.55 mgg-'(0.09) Zn: 89 pg g-* (6) Pb: 3.1 pgg-'(0.6) * 95% confidence interval. Ca10(P04)6(0H)2 Ca10(P04)6(0H)2 Ion substitution Not specified Not specified Table 2 Participants, analytical instruments and settings used for the determination of Cd and Pb in Ca-rich matrices ETAAS FAAS Agency Instrument Tube Lamps Background correction Modifier Wavelengths: Cd Pb Char Atomize Measurement Calibration Dry National Research Council Canada Perkin-Elmer 5000, HGA-500 Platform HCL (Cd); EDL (Pb) Zeeman None Agency Instrument Flame Lamps Background correction Modifier Wavelengths: Cd Pb Nissei Sangyo Canada Hitachi 2-8100 Air-acetylene HCL (Cd and Pb) Zeeman None 228.8 nm 283.3 nm 140 "C for 20 s 400 "C for 20 s 2400 "C for 4 s Peak area Standard additions 228.8 nm 283.3 nm Measurement Calibration Peak area Calibration graph ASV ICP-MS Agency Instrument Mode Electrode Electrolyte Trent University Metrohm VA 646, VA 647 Differential-pulse normal Hanging Hg drop Sodium acetate buffer Agency Instrument Ontario Ministry of the Environment Perkin-Elmer-Sciex Elan mass spectrometer Gas flows: Plasma Auxiliary line Nebulizer Ion lens settings: Bessel box barrel Bessel box stop Einzel lens 1 and 3 Einzel lens 2 Measurement* Calibration 14 1 min-1 0.8 I min-1 0.4 1 min-1 Plating time Plating range Sweep rate U step 90 s -900 to -300 mV 2.5 mV s-1 2 mV +3.67VDC -5.78 VDC - 10.60 VDC -12.08 VDC Peak height External calibration graph Measurement Calibration Peak height Standard additions * The minimum acceptable resolution ( i , e ., low resolution) to be within specifications was 1.0 k 0.1 u.ANALYST, JANUARY 1992, VOL. 117 21 Table 3 Levels of Cd and Pb in the five different types of Ca-rich sample measured by the four different analytical techniques.Samples and instruments are detailed in Tables 1 and 2, respectively Cdyg g-I* Laboratory-reagent IAEA CRM H-5 Technique grade CaCOR powder Brand A Brand B Brand C Animal Bone ICP-MS <0.010f <0.010f 0.54 f 0.05$ 3.31 k 0.14 0.14 f 0.04 ASV <0.024? 0.07 f 0.05 0.63 f 0.06 3.49 k 0.02 0.11 f 0.03 ETAAS <O.O06f 0.07 f 0.02 0.71 f 0.10 3.59 k 0.66 0.017 k 0.003 FAAS <0.035? 0.12 f 0.09 0.71 k 0.06 Pbhg g-I§ <0.035? 3.55 k 0.27 3.83 f 0.25 3.50 k 0.86 1.39 k 0.06 2.77 f 0.26 ICP-MS 0.25 t 0.12 ASV 0.25 f 0.07 2.89 k 0.56 3.26 f 0.31 1.42 f 0.25 2.87 5 0.51 ETAAS 0.24 f 0.05 3.25 f 0.44 3.73 f 0.62 1.33 k 0.13 3.09 f 0.22 FAAS 0.79 k 0.17 3.39 f 0.12 3.56 k 0.34 1.97 f 0.29 4.57 f 1.31 * Detection limits for Cd are 0.010, 0.024, 0.006 and 0.035 yg g-* using ICP-MS, ASV, ETAAS and FAAS, respectively.t Below detection limit. 3 95% confidence intervals. 5 Detection limits for Pb are 0.010, 0.120, 0.120 and 0.5 yg g-1 using ICP-MS, ASV, ETAAS and FAAS, respectively. pH adjustment ranged from 25 to 30 ml. Approximately 20 ml of the solution were transferred into the ASV cell and analysed. Samples analysed by ICP-MS and FAAS were prepared as follows. Approximately 0.5 g of the ashed samples was dissolved in 1 ml of 16 mol dm-3 HN03. The solutions were filtered through pre-washed (1.5 mol dm-3 HN03) Millipore membrane filters and diluted to either 25 or 100 ml with DDW and analysed by FAAS and ICP-MS, respectively. Quality control High-purity, certified reagents (Aristar or Suprapur grade) were used for all the analyses. All the sample types were analysed at least in triplicate together with two procedural blanks.No appreciable amounts of Cd and Pb were measured in the blanks by any of the participants. Results and Discussion The identification and subsequent quantification of heavy metals by the analytical techniques employed in this work is essentially based on the following three distinct physical properties: absorption of discrete wavelengths of electromag- netic energy (i.e., FAAS and ETAAS); half-wave potential (Le., ASV); and atomic mass (i.e., ICP-MS). Although all of these methods are prone to interference problems, it is unlikely that the bias (either signal enhancement or sup- pression) apparent in one method would be exactly the same as in another method.The results of the determination of Cd and Pb are summarized in Table 3. The Pb levels in the IAEA CRM measured by all four analytical techniques were within the range of the certified Pb level for this CRM (Table 1). Similarly, no significant difference [p <0.05, one-way analysis of variance (ANOVA)] was observed in the Pb levels among the three brands of Ca supplement as measured by the four analytical techniques (Table 3). Results from FAAS (dis- cussed below) were, however, at variance with those from the other three techniques when comparing Pb levels in the laboratory-reagent grade CaC03 powder (Table 3). A certi- fied Cd level was not listed for the IAEA CRM and hence it is not possible to comment on the accuracy of the various techniques used for the determination of Cd in Ca-rich matrices.Except for the low Cd level measured in the IAEA CRM by ETAAS, the low variability in Cd levels measured by the different techniques in most of the sample types indicates that the Cd measurements were fairly precise (Table 3). No systematic trend based on instrumentation or any significant differences (p <0.05, one-way ANOVA) in Cd levels within any sample types was observed. The results also indicate that there was a substantial degree of variability in heavy metal concentrations among the three different brands of Ca supplement analysed. This variance was greatest for Cd, which displayed a 40-fold difference between the lowest (Brand A) and the highest (Brand C) concentrations. The Pb concentrations in the Ca supplements averaged higher and varied considerably less than those of Cd.Interference ef€ects associated with AAS analyses can arise from reactions between the analyte and matrix components of the sample, which can affect the formation of the analyte vapour. Although gas-specific interference problems have not been reported for the determination of Cd, several anionic interferences can occur during the determination of Pb using an air-acetylene flame. This could explain why the variability associated with the Pb levels measured in the IAEA CRM by FAAS was four times higher than that associated with the ETAAS results (relative standard deviation = 29 and 7%, respectively). The discrepancy between the Pb levels measured in the laboratory-reagent grade CaC03 by FAAS and the other three techniques might have been associated with 'low end deterioration' of the signal due to the use of a hollow cathode lamp (HCL; Table 2).The signal-to-noise ratio of electrodeless discharge lamps (EDLs) is typically 2-3 times better at lower Pb concentrations. High levels of Ca compounds such as CaO or CaCI2 can cause matrix effects during trace metal determinations by ICP-MS.*3.14 Higher levels of Ca (>500 pg g-1) could cause analyte suppression or enhancement and bias the results. In order to reduce the potential for matrix effects the sample was diluted 100 times, bringing the Ca concentration below 500 pg g-1. There was only one isobaric interference in this work. Cadmium-114, which is the most abundant of the eight stable isotopes of Cd, suffers interference from 114Sn.As the isotopic abundance of the latter is relatively low (0.65%), the contribution from this element to the bias of the results is minimal. In addition, a correction factor is applied to the elemental equation for Cd if Sn is detected above the detection limit in the sample. Interference problems in polarographic measurements can be related to matrix effects or to the overlapping of signals. Matrix interferences that are mainly caused by the presence of organics and surface-active substances were not problematic in this work as these compounds were effectively destroyed during the ashing process. Overlapping signals can occur in analytes with potentials that are very close. Generally, a potential difference of about 100 mV is sufficient to allow22 ANALYST, JANUARY 1992, VOL.117 resolution of the species of interest. For example, the half-wave potentials (E4) of TI+ and As3+ are -0.48 and -0.43 V, respectively. These species, when present at high levels, could conceivably interfere with Pb*+ (Eb = -0.045 V). However, when using a hanging Hg drop electrode, As related interferences would be less problematic as this element does not amalgamate well with Hg. Furthermore, various counter- measures based on chemical means can be used to separate overlapping peaks. For example, the separation of T1 and Pb peaks could be achieved either by altering the pH or the electrolyte ( e . g . , by adding ethylenediaminetetraacetic acid). However, T1- and As-related interferences did not pose any problems in this work as the Pb levels obtained by ASV conformed with those obtained with the other techniques.The Cd and Pb levels measured in the Ca supplements by FAAS (Zeeman-effect background correction system) were not significantly different from the levels obtained using the other techniques. This has particular importance for quality control boards. The analyses performed by FAAS were carried out in a fraction of the time required for the other three techniques. Hence quality assurance groups currently using more complex and laborious techniques could greatly increase the number of samples screened/analysed without substan- tially increasing their analytical time or costs. We are currently investigating whether the FAAS analyses can be further ameliorated, particularly at low trace metal concentrations, by using more complex standard solutions ( i .e . , spiked with CaC03) and also EDL sources. The authors thank S. Lingard and L. Bigelow for analytical assistance with the ASV studies, and Professor R. D. Evans and R. J. Cornett for critical comments during the preparation of the manuscript. This work was funded by grants from the Natural Sciences and Engineering Research Council of Canada to R. D. Evans and R. J. Cornett. References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Goyer, R. A., Environ. Health Perspect., 1990, 86, 177. FDA Drug Bulletin, Food and Drug Administration, Rockville, MD, 1982, April, vol. 12(1), pp. 5-6. Health and Welfare Canada, Health Protection Branch, Notes, 1983, No. 8 (April), Vanier, Ontario. Sturesson, U., Ambio. 1978, 7, 122. Rabinowitz, M. B., Leviton, A., and Bellinger, D. C., Bull. Environ. Contam. Toxicol., 1989, 43,485. Dodge, R. E., and Gilbert, T. R., Mar. Biol., 1984, 82, 9. Dermott, R. M., and Lum, K. R., Environ. Pollur., 1986, 12, 131. Denton, G. R. W., and Burdon-Jones, C . , Mar. Pollut. Bull., 1986, 17, 209. Shen, G. T., and Boyle, E. A., Chem. Geol., 1988,67, 47. Bourgoin, B. P., Mar. Ecol. Prog. Ser., 1990, 61, 253. Carriker, M. R., Swann, C. P., and Ewart, J. W., Mar. Biol., 1982, 69, 235. Carell, B., Forberg, S . , Grundelius, E., Henrikson. L., Johnels, A., Lindh, U., Mutvei, H., Olsson, M., Svardstrom, K., and Westermark, T., Ambio, 1987, 16, 2. Olivares, J. A., and Houk, R. S . , Anal. CIzem.. 1986, 58, 20. Gray, A, L., Spectrochim. Acta, Part B , 1986,41, 151. Paper 1 I01 553 B Received April 3, 1991 Accepted August 14, 1991