JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 199 1 VOL. 6 477 Platform in Furnace Zeeman-effect Atomic Absorption Spectrometric Determination of Arsenic in Beer by Atomization of Slurries of Sample Ash Maria Luisa Cervera Ascensio Navarro and Rosa Montoro lnstituto de Agroquimica y Tecnologia de Alimentos (CSIC) Jaime Roig 7 7 4601 0 Valencia Spain Miguel de la Guardia and Amparo Salvador Departamento de Quimica Analitica Universidad de Valencia Dr. Moliner 50 467 00 Burjassot Valencia Spain A precise and accurate procedure is proposed for the determination of As in beer samples at the ng g-I level by electrothermal atomic absorption spectrometry. The sensitivity of this technique is enhanced by a factor of 2.5 by introducing a preconcentration step consisting of preliminary dry ashing of the samples and direct injection into the L'vov platform of a slurry of the ashes.The high background caused by the beer sample matrix is compensated for by the use of Zeeman correction. The arsenic is atomized under stabilized temperature platform furnace conditions with the addition of a nickel-ascorbic acid chemical modifier. The effect that each of these compounds has on the determination of As was studied both separately and in combination. The methodology developed has a characteristic mass of 35 pg which corresponds to a concentration of 0.7 ng of As per g of beer a recovery of 102 f 2% and a relative standard deviation of 4.5% for six independent analyses of a sample containing 8.5 ng g-' of As. The accuracy of the method was confirmed by the analysis of a certified reference material sample.The proposed procedure was used to analyse real samples of beer and the results are comparable to those found by hydride generation atomic absorption spectrometry. Keywords Arsenic determination; beer sample; platform in furnace Zeeman-effect atomic absorption spectrometry; slurry atomization; chemical modifier The determination of arsenic in vegetable matrices by electrothermal atomic absorption spectrometry is hindered by severe suppression of the analyte signal owing to the sample matrix constituents.' In addition to this chemical interference spectral interference has been reported in the determination of arsenic when deuterium background correction is used.2 Use of the stabilized temperature platform furnace (STPF) concept3 in combination with Zeeman-effect background correction can overcome these limitation^.^.^ Nevertheless for direct determination of arsenic in previously digested food samples a preconcen- tration step is required when the arsenic content in the original sample is at the ng g-' level.Solvent extraction6 or the coprecipitation of arsenic with an ammonium pyrrolidine dithiocarbamate-nickel com- plex have been used to preconcentrate arsenic.' These procedures also allow the elimination of matrix effects but they are time consuming and laborious. Slurry atomization is an alternative technique for the introduction of solid samples in atomic spectrometry.8-10 Direct injection of slurries of sample ash makes it possible to use a lower dilution than would be required to dissolve the ash totally,ll thus improving the detection limits. However the suppres- sion effect caused by the sample matrix is enhanced by the introduction of the slurry and it is necessary to use chemical modification to minimize this interference Zeeman-effect background absorption correction is also required.The addition of nickel avoids the loss of arsenic during the ashing stage and permits higher pyrolysis temperatures so that the removal of matrix components is easier and more complete.12 Moreover nickel has a supplementary role in preventing the formation of phosphorus molecular species which cause a structured background.* The effectiveness of ascorbic acid in minimizing signal suppression in the determination of other elements using electrothermal atom- ization is also well known.I3J4 No method has been published in the literature for the direct determination of low levels of arsenic in food samples. In this work a method has been developed for determining arsenic in beer at the ng g-' level.Slurries of the ashes obtained from the sample were used to preconcentrate the arsenic with a combina- tion of STPF conditions with Zeeman-effect background correction with nickel-ascorbic acid as the chemical modifier. The analytical parameters of the method were established and a series of real beer samples analysed. The results were compared with those found by hydride generation atomic absorption spe~trometry.~~ Experimental Equipment A Perkin-Elmer Zeeman/3030 atomic absorption spectro- meter equipped with an HGA-600 graphite furnace and an AS 60 autosampler was used throughout this study.The instrument includes a graphics display. The highly time- resolved signals were plotted with a Perkin-Elmer PR- 100 printer. Pyrolytic graphic coated graphite tubes with an inserted pyrolytic graphite L'vov platform Perkin-Elmer Part No. 112660 were used exclusively. A Heraeus Model 1 100/3 muMe furnace fitted with a Jumo DPG-4411 digital microprocessor was used to ash the samples. Reagents Analytical-reagent grade water (1 8 MR cm-l specific resistivity) was used throughout. All reagents used were of the highest purity available and at least of analytical-reagent grade. An aqueous stock solution of As111 was prepared from arsenic 111 oxide (Riedel de Haen).The ashing aid was prepared by stirring 2.5 g of Mg(N03)2-6H,0 and 0.25 g of MgO in 100 ml of water until homogeneous. The chemical modifier used was Ni in the form of Ni(N03)2 and ascorbic acid. The ascorbic acid together with nitric acid was used as the suspending agent. A National Institute of Standards and Technology (NIST) Standard Reference Material (SRM) NIST SRM 1573 Tomato Leaves was also used. Optimization of Analytical Parameters Slurries of the sample ashes stabilized with nitric acid and the surfactant Gandax SX (Molins KAO Barcelona Spain) or alternatively with nitric and ascorbic acids were478 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 199 1 VOL. 6 injected onto the L'vov platform. Nickel and mixtures of nickel and ascorbic acid were assayed as chemical modifiers.Pyrolysis and atomization parameters were optimized for both samples and standards in order to obtain the best sensitivity and highest precision and accuracy. In all instances the integrated peak area absorbances were used in accordance with STPF conditions. Recommended Procedure Dry ashing of the samples Add 2.5 ml of ashing aid containing 2.5% m/v of Mg(N0,)2 and 0.25% m/v of MgO to 25.00 Ifr 0.0 1 g of previously de- gassed beer samples and mix well. Evaporate in a sand- bath until totally dry controlling initial foam formation. Ash in a muffle furnace as described in an earlier report,15 at a temperature of less than 450 "C until white ash is obtained. Sometimes it may be necessary to wet the ash with HN03 (1 0% v/v) and repeat the ashing step.Slurry formation and determination of arsenic Add 100 pl of HN03 to the white ash and place in an ultrasonic bath to remove any solids adhering to the glass wall. Add 10 ml of the chemical modifier solution of ascorbic acid (2% m/v) and nickel (0.5% m/v) and homogenize the sample slurries shaking vigorously in the ultrasonic bath. Take volumes of 2 ml of slurry and add 20 pl of standard aqueous solutions containing 0 4 6 8 and 10 pg m1-I of As in order to obtain the various standard additions levels. Analyse by platform in furnace Zeeman-effect atomic absorption spectrometry using the analytical parameters indicated in Table 1 and a slurry injection of 20 pl. Results and Discussion Stabilization of Beer Ash Slurry Two alternative methods were employed to stabilize the beer ash slurry.Firstly the use of nitric acid and the surfactant Gandax SX permits the dispersal of the ash from 25 g of beer in 5 ml and secondly the use of nitric and ascorbic acids provides ash slurries of 25 g of beer in 10 ml. Both types of slurries are homogeneous and stable for more than 5 h without the use of any mechanical or ultrasonic stirrer and so the use of a conventional autosampler to inject the slurries into the graphite furnace is possible. In order to test the stability of the slurry small amounts of a sample were placed in the autosampler rack and injected into the graphite furnace at fixed time intervals. The absorbance values obtained were compared with those found for samples stirred just prior to injection.Selection of the Chemical Modifier The high background correction capacity and the accuracy of the Zeeman effect provides great flexibility in the selection of chemical modifiers and in the concentrations of Table 1 Analytical parameters for the determination of arsenic in beer. Wavelength 1 93.7 nm; arsenic electrodeless discharge lamp 8.5 W; slit width 0.7 nm Internal Time/s Ar flow Step "C Ramp Hold ml min-' Temperature/ rate/ 1 Drying 90 10 20 300 2 Drying 120 10 20 300 3 Ashing 800 10 10 300 4 Ashing 1400 10 60 300 5 Atomization 2300 0 5 0 6 Cleaning 2650 1 5 300 7 Cooling 20 10 10 300 those used. Accordingly nickel nitrate and 'mixtures of nickel nitrate and ascorbic acid were assayed in this work for use as chemical modifiers. The Zeeman-effect corrected absorbance of stabilized ash slurry samples and standards is plotted in Fig. 1 as a function of the amount of nickel added.In the standards nickel nitrate prevents pre-atomization analyte loss but also reduces the absorbance of arsenic. Increased amounts of nickel nitrate have a supplementary effect on samples the suppression effect and background are reduced by using between 100 and 200 pg of nickel in the form of Ni(N03)2. In the analysis of real samples using slurries containing 100 pg of Ni poor accuracy (results not shown) were obtained and the results found were different to those obtained by hydride generation atomic absorption spectro- metry.15 Consequently the use of mixtures of nickel nitrate and ascorbic acid as chemical modifiers was assessed. The effect of the ascorbic acid concentration on background and also on the sensitivity and accuracy of the arsenic determi- nation is shown in Table 2.The best results were obtained with a 2% m/v ascorbic acid concentration because it provides a higher absorbance and a lower background. Samples containing nickel nitrate and ascorbic acid provide better results than those containing only nickel. This can be explained by the relative availability of active carbon provided by ascorbic acid in the atomization process,I4 which causes a reduction of suppressive interfer- ence and enhancement of the arsenic signal. 2'2 I 2.1 - 2.0 - 1.9 - 1.8 - 1.7 - al c 4 1.6 - 0.5 1 ZAA 0.4 0.3 0.2 0.1 ~ 100 200 300 400 ! 0 Mass of Ni/pg Fig. 1 Effect of mass of nickel on the arsenic and background response in slurries of 25 g of beer ash stabilized with Gandax SX ashing temperature 1100 "C; ashing time 30 s; atomization temperature 2 300 "C; and injection volume 20 pl.A Real sample with 100 ng ml-l of As; and B 100 ng ml-' As standardJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 199 1 VOL. 6 479 Table 2 Effect of ascorbic acid concentration on the arsenic determination carried out in 10 ml of a nitric acid slurry of 25 g of beer ash also containing 100 p g of nickel Ascorbic acid concentration (Yo m/v) Parameter 1 2 3 Absorbance* 0.1 17 0.174 0.169 Background* 0.922 0.715 0.841 As concentration/ng g-' 22.6 24.8 23.8 Sensitivity (absorbance*/ng ml-I) 1.99 x 10-3 2.88 x 10-3 2.79 x 10-3 Relative differencet (O/O) - 6.2 2.9 - 1.2 *Integrated peak area absorbance units.t Between values found by Zeeman-effect corrected atomic absorption spectrometry and hydride generation atomic absorption spectrometry1s (24.1 ng g-I). Background Correction The atomization of slurries of ashed food samples provides high background readings. For arsenic determination the use of a magnesium salt as the ashing aid also increases background. In order to measure and compensate for the background effect the use of Zeeman correction is neces- sary. In this instance a Perkin-Elmer Zeeman/3030 with the magnetic field perpendicular to the optical beam enabled good background correction to be achieved and thus reproducible measurements of the arsenic samples and standards were obtained. A typical background absorbance value for an aqueous arsenic standard is of the order of 0.160.However reagent blank solutions containing magnesium provide background absorbance readings of about 0.450. The atomization of real samples presents background absorbance values of 0.600. Optimization of the Temperature-Time Programme The ashing and atomization temperature-time programme was optimized in order to provide maximum matrix decomposition (without loss of arsenic); minimum back- ground; and maximum sensitivity. 1.4 1.2 1 .o 0.8 8 0.6 m 0.5 C fJ 2 2 0.4 0.3 0.2 0.1 BG 800 1200 1600 2000 2 400 Temperatu rePC Fig. 2 Optimization of the temperature-time programme. Ashing temperature was established for a real sample using an atomiza- tion temperature of 2300 "C. The atomization temperature was determined after an ashing step of 60 s at I400 "C.A Real sample; and B 100 ng ml-I As standard The effects of ashing and atomization temperatures on the absorbance of an aqueous standard solution of arsenic and a real sample of ash slurry are indicated in Fig. 2. It can be observed that the optimum ashing and atomization temperatures correspond to 1400 "C and 2300 "C respec- tively. The background was reduced between 1000 and 1400 "C in the ashing step. On the other hand it was observed that an ashing time of 60 s provided better results than the use of 40 s. Comparison Between Direct Calibration and Standard Addi- tions Method A typical comparison between the analytical and standard additions graphs is shown in Fig. 3. It can be seen that the matrix provides a negative effect of 40% on the sensitivity obtained by electrothermal atomization.In order to com- pensate for this effect it is necessary to use the standard additions method in order to obtain accurate results. Besides the use of spiked samples provides high absor- bance readings which are more reproducible. Analytical Characteristics of the Proposed Method The analytical characteristics such as sensitivity detection limit precision and accuracy for the determination of arsenic in beer samples were evaluated from a study of the calibration graphs the recovery assays and from the analysis of the certified vegetable sample. 0.6 0.5 0.4 a) C 0.3 s D 6 0.2 0.1 (20) 0 20 40 60 80 100 [Asl/ng ml-' Fig. 3 A Aqueous arsenic calibration graph; and B standard additions graph for arsenic in a sample of beer ash slurry (stabilized with nitric and ascorbic acids and containing 1 OOpg of Ni).A 2Opl sample volume was used throughout480 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1991 VOL. 6 Sensitivity The sensitivity was established using the mean value of the slopes of 13 standard additions graphs. It corresponded to 2.7 x loe3 (50.1 x integrated area absorbance units for each ng ml-I of As. obtained at this level. For this reason and as the character- istic mass is always very similar to the absolute detection limit this can be used for the estimation of the lowest concentration that can be routinely measured in real samples. 16* Detection limit The detection limit corresponded to 2.7 ng g-l this value was evaluated using three times the standard deviation of the blank measurements divided by the sensitivity.The characteristic mass for the aqueous standard (the arsenic mass that provides an absorbance value of 0.0044) using 20 pg of Ni was 18 pg (0.4 ng g-l) and the characteristic mass for the procedure (using 100 pg of Ni 2% ascorbic acid and applying the analyte additions technique) was 35 pg (0.7 ng g-'). These parameters are expressed in nanograms of arsenic per gram of beer sample taking into account sample size and the dilution carried out in the procedure recommended. The detection limit calculated from the standard devia- tion of the blanks which are almost indetectable in all instances is too high owing to the low reproducibility Precision The precision expressed as the relative standard deviations (standard deviationlmean) of six independent analyses of a sample containing 8.5 ng g-' of arsenic is 4.5%.In order to evaluate the accuracy of the method recovery assays were performed on a matrix of beer and a certified sample of vegetable origin was analysed. Percentage recovery A recovery of 102 k 2% was obtained for a known amount of 25 ng g-I of arsenic added to a real sample containing 8.5 ng g-'. The good recovery obtained proves that the loss of arsenic and contamination do not occur during the various stages of analysis. Table 3 Determination of arsenic in real samples by hydride generati~n'~ and electrothermal atomic absorption spectrometry Hydride generation Electrothermal As found ng g-' 73.7 66.2 As found/ 65.0 67.9 61.0 47.3 49.7 6.5 6.1 67.1 57.8 61.8 7.3 7.3 10.5 13.3 11.7 12.9 9.3 6.5 8.5 4.4 6.1 5.3 4.0 2.0 4.0 8.9 8.5 7.8 8.8 8.5 8.5 26.2 24.8 22.2 ng g-* Sample 1 Mean k SD* 7 0 a 5 Mean f SD* 65+3 46.0 48.8 6.3 6.9 59.9 62.4 4 7 f 2 6.6 f 0.4 61 a 2 4 8 k 2 6.3 k0.3 62+5 5 8.5 7.9 8.2 2 0.4 8 2 2 6 11.9 12.5 12.2 f 0.4 12.6 f 0.8 7 7.2 7.5 7.3 a 0.2 8 + I 8 9 10 5.5 5.5 5.5 k 0.0 5.3 k 0.8 3.9 3.5 3.7 k 0 .3 8.5 k0.5 3 5 1 8.0 8.2 8.8 9.0 8.5k0.4 25.1 24.0 24.9 22.5 24.0 2 4 k I 11 24+2 * SD = standard deviation.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1991 VOL. 6 48 1 Accuracy The NIST SRM 1573 Tomato Leaves (containing 0.27 k 0.05 pg g-lof arsenic) was analysed by the proposed method in order to determine the accuracy of this method for the determination of arsenic in vegetable samples.The results obtained 0.28 f 0.04 pg g-l are in close agreement with the certified values and this testifies to the accuracy of the method and its applicability to the analysis of other food samples of vegetable origin. Analysis of Real Samples The determination of arsenic in real beer samples was carried out by both quartz tube atomization of arsenic hydride' and electrothermal atomic absorption spectrome- try using the recommended procedure for preliminary dry ashing of the samples in all instances. The results obtained by the two methods assayed are comparable as can be seen in Table 3. The regression equation value calculated between the concentration values1*J9 obtained by electrothermal atomi- zation and those found by hydride generation demon- strates that the proposed method does not require the use of blank corrections and constant relative errors are not obtained.Conclusions The proposed method allows the determination of arsenic at the ng g-l level by electrothermal atomic absorption spectrometry. This is the first time that such a low concentration of arsenic has been determined accurately by electrothermal atomic absorption spectrometry with direct injection of digested food matrices. The use of the STPF concept Zeeman-effect background correction and the nickel-ascor- bic acid chemical modifier to avoid the high background effect and matrix interference allows arsenic atomization in digested food samples. The use of the slurry approach to inject beer sample ash also allows preconcentration of the samples which improves the detection limit of the method. The results obtained in the analysis of real and certified samples show the suitability of the proposed method for this type of determination.Funds to carry out this work were provided by the Cornission Interministerial de Ciencia y Tecnologia (CI- CyT) Project AL189-0521 for which we are deeply indebted. References 1 Hoenig M. and Van Hoeyweghen P. Spectrochim. Acta Part B 1982,37 81 7. 2 Hoenig M. and Van Hoeyweghen P. Intern. J. Environ. Anal. Chem. 1986 24 193. 3 Slavin W. Manning D. C. and Carnrick G. R. At. Spectrosc. 1981 2. 137. 4 Fernandez F. J. Bohler W. Beaty. M. M. and Barnett W. B. A4t. Spectrosc. 1981 2 73. 5 Fernandez F. J. and Giddings R. At. Spectrosc. 1982 3 61. 6 Ishizaki M. Buseki Kagaku 1977 26 667. 7 Dabeka R. W. and Lacroix G. M. A. Can. J. Spectrosc. 1985 30 154. 8 Langmyhr F. J. and Wibetoe. G. Prog. Anal. At. Spectrosc. 1985 8 193. 9 Ebdon L. Foulkes M. E. and Hill S. Microchem. J. 1989 40 30. 10 Littlejohn D. Stephen S. C. and Ottaway J. M. Anal. Proc. 1985 22 376. 1 1 de Benzo Z. A. Fernandez M. Carrion N. Eljuri E. At. Spectrosc. 1988 9 87. 12 Zhumanova K. M. Beilina A. Z. and Muldakhmetov Z. M. Deposited Doc. 1977 VINITI 37 13-77 13 pp. (Russ); Chem. Abstr. 1979 91 203435~. 13 Regan J. G. T. and Warren J. Analyst 1978 103 447. 14 Hoenig M. Scokart P. O. Van Hoeyweghen P. Anal. Lett. 1984 17 1947. 15 Cervera M. L. Navarro A. Montoro R. Catala R. J. Assoc. Off Anal. Chem. 1989,72 282. 16 Grobenski Z. Lehmann R. Radziuk B. and Voellkopf U. At. Spectrosc. 1986 7 61. 17 de Loos-Vollebregt M. T. C. Koot J. P. and Padmos J. J. Anal. At. Spectrom. 1989 4 387. 18 de la Guardia M. Salvador A. and Berenguer V. An. Quim. Ser. B 1981 77 129. 19 de la Guardia M. Salvador A. and Berenguer V. An. Quim. Ser. B 1983 79 446. Paper 0/04 732E Received October 22nd I990 Accepted May 23rd 1991