Determination of Arsenobetaine in Manufactured Seafood Products by Liquid Chromatography, Microwaveassisted Oxidation and Hydride Generation Atomic Absorption Spectrometry DINORAZ VE� LEZ, NIEVES YBA� N� EZ AND ROSA MONTORO* Instituto de Agroquý�mica y T ecnologý�a de Alimentos (CSIC), Apdo. Correos 73, 46100 Burjassot, Valencia, Spain. E-mail: Rmontoro@iata.csic.es A study was carried out to develop and optimize a method for sition. Microwave decomposition of organoarsenicals, separated by HPLC using flow injection systems coupled to determining arsenobetaine (AB) in seafood products by coupling HPLC, microwave-assisted oxidation and HGAAS.HGAAS, has recently been reported for determining AB in urine samples,9,10 water, synthetic fish and sediment extracts,11 Conditions were established for the extraction and instrumental determination of AB in popular seafood products. and water and sediments,12 but there are no reports of the application of this method to seafood products.The analytical features of the method were as follows: the detection limit was 0.68–27.20 ng g-1 As (fresh mass), the The objective of this study was to develop and optimize a method for determining AB in real seafood products by HPLC- relative standard deviation ranged from 0.4 to 6%, and the recovery was 104±5%. The analysis of DORM-1 (Dogfish microwave-assisted oxidation (MO)–HGAAS. AB was extracted quantitatively from commercial seafood Muscle, National Research Council of Canada, Certified Reference Material ) provided an AB value of 16.5±0.6 mg g-1 products in a simple and rapid way.The suitability of the clean-up and chromatographic procedure for separating AB As (dry mass), in agreement with results obtained by other workers using HPLC–ICP-MS. The proposed procedure was from other arsenical species was investigated in extracts of real seafood samples. The detection limit, precision and accuracy used to analyse canned seafood products purchased at local retail market outlets.The contents of total As represented by of the method were also evaluated. The proposed procedure was used to analyse real seafood samples. AB ranged from 5 to 75%. The lowest percentages of AB corresponded, in general, to the bivalve group. Keywords: Arsenobetaine; high-performance liquid EXPERIMENTAL chromatography; microwave-assisted oxidation; hydride Instrumentation generation atomic absorption spectrometry; manufactured The hyphenated HPLC–MO–HGAAS system is shown in seafood Fig. 1 and details of the operating conditions for the system are given in Table 1. The equipment used included a highperformance liquid chromatograph (Hewlett-Packard Model Arsenobetaine (AB) has been reported as being the organoarsenical most frequently found in marine animals,1 and the major 1050), equipped with a Model HP 79852A quaternary pump with on-line de-gassing system; a Rheodyne valve fitted with form of As in fish and crustaceans.1,2 We have reported in a previous study,3 in which total As was determined by HGAAS a 100 ml loop; and data station: viz., a Hewlett-Packard personal computer, Vectra 486/33N Model 170 with 486 micropro- and AB by HPLC–ICP-AES, that AB contents vary over a very broad range, viz., from <0.1 to 5.4 mg g-1 As, fresh mass cessor rated at 33 MHz (Hewlett-Packard Espan�ola, Madrid, Spain).(fm), and consequently they are sometimes below the limit of detection of the method.This shows the need to develop For AB determination, the chromatographic system was connected to a Moulinex Super Crousty domestic micro- methodologies with detection limits for AB lower than those offered by the ICP detector. The total As and the percentage wave oven with a maximum power of 1100 W and an operating frequency of 2450 MHz. A loop of PTFE tubing of total As represented by AB calculated by using these methodologies cannot, alone, evaluate the toxicity or otherwise (1.6 m×0.5 mm id) was placed inside the microwave oven through the ventilation holes.An additional oven load of of seafood products, but both can be indicators of their possible toxicity.3 400 ml of water was placed inside the oven to prevent overheating. The effluent from the microwave oven passed through HGAAS coupled to HPLC as a post-column derivatization method is one of the most inexpensive and convenient ways PTFE tubing (0.5 m×0.5 mm id) and was cooled in an icebath before it reached the hydride generator, to avoid over- of improving the sensitivity of As determination.One major limitation to using HGAAS as part of an HPLC system for pressure and decomposition of the sodium tetrahydroborate. The Perkin-Elmer (Norwalk, CT, USA) Model 5000 atomic determining AB is that this arsenical compound does not form volatile hydrides. This problem can be solved by two different absorption spectrometer was equipped with a Perkin-Elmer FIAS-400 system operating as a hydride generator in approaches, viz., UV4–7 or microwave on-line decomposition of organoarsenicals to forms suitable for generating arsines.continuous-flow mode. A drainage system (flow 9 ml min-1) was incorporated for the waste solution from the hydride Both systems provide appropriate detection power but the UV-HG attachment is time consuming. The efficient heating generation, working at constant pressure through a peristaltic pump.An electrothermally heated quartz cell was employed. of the microwave oven, which can rapidly decompose the organoarsenicals,8 possesses advantages over the UVdecompo- The HGAAS system was controlled by the software of a Journal of Analytical Atomic Spectrometry, January 1997, Vol. 12 (91–96) 91Fig. 1 Hyphenated HPLC–MO–HGAAS system. Table 1 Operating conditions for HPLC–MO–HGAAS P-Selecta Vibromatic-340 mechanical arm shaker (Selecta, Barcelona, Spain). High-performance liquid chromatography— Column Hamilton PRP X-100, 10 mm polymer-based Reagents and Samples anion-exchange column (25.0 cm×4.1 mm Analytical-reagent grade water (QRG), 18 MV cm, obtained id) (Teknokroma, Barcelona, Spain) Guard column Hamilton PRP X-100, 10–20 mm polymer- with a Milli-Q water-purification system (Millipore Ibe�rica, based anion-exchange guard column Madrid, Spain), was used for the preparation of reagents and (25×2.3 mm id) (Teknokroma) standards.All chemicals including standards and solutions Mobile phase Phosphate buffer (Na2HPO4–H3PO4), were of pro analysi quality or better, viz., hydrochloric acid, 3 mmol l-1 at pH 5.0.Flow rate, 1 ml nitric acid, ammonia solution 32% extra pure, sodium hydrox- min-1 ide and HPLC phosphate buffers (99.5% Na2HPO4 ·2H2O and Injection volume 100 ml Temperature 28 °C 85% H3PO4 ). The stock standard solutions were: arsenite solution Hydride generation— (1000 mg l-1 AsIII), prepared by dissolving 1.320 g of arsenic Cell temperature: 900 °C trioxide (Riedel-de Hae�n, Hanover, Germany) in 25 ml of 20% Sample solution: Phosphate buffer, 3 mmol l-1 at pH 5.0; m/v KOH solution, neutralizing with 20% v/v H2SO4 and 1 ml min-1 flow rate diluting to 1 l with 1% v/v H2SO4; arsenate solution (1000 mg Reducing agent: 2.0% m/v NaBH4 in 0.7% m/v NaOH; l-1 AsV) (Titrisol, Merck, Darmstadt, Germany).Solutions of 1.9 ml min-1 flow rate HCl solution: 3 mol l-1; 1.9 ml min-1 flow rate monomethylarsonic acid (1000 mg l-1 MMA) and dimethylar- Carrier gas: Argon; 45 ml min-1 flow rate sinic acid (1000 mg l-1 DMA) were prepared by dissolving appropriate amounts of CH3AsO(ONa)2 ·6H2O (Farmitalia Atomic absorption spectrometer— Carlo Erba, Milan, Italy) and (CH3 )2AsNaO2·3H2O (Fluka Wavelength 193.7 nm Chemika Biochemika, Alcobendas, Madrid, Spain) in water.Spectral bandpass 0.7 nm AB solution (973 mg l-1) and arsenocholine (AC) solution Lamp power 8.5 W (electrodeless discharge lamp) (1000 mg l-1) were obtained from the Service Central Microwave oxidation— d’Analyse du CNRS-SCA, Vernaisson, France.The CRM Power 1100 W DORM-1 (Dogfish Muscle) was obtained from the National Oven load 400 ml water Research Council of Canada, Institute for Environmental DiE tubing (1.6 m×0.5 mm id) Chemistry (Ottawa, Canada). Refrigeration coil PTFE tubing (0.5 m×0.5 mm id) For the clean-up of the methanol–water extracts, columns Oxidizing solution 1% m/v K2S2O8 in 2.5% m/v NaOH; packed with Dowex 50W-X8 cation-exchange resin (H+ form, 0.6 ml min-1 flow rate 1×6 cm) (Bio-Rad, Barcelona, Spain) were employed.Oxidizing potassium persulfate (Probus, Barcelona, Spain) solutions (1, 3, and 5% m/v) were prepared daily in 2.5% m/v separate programmable PC system. The spectrometer signal NaOH. As reducing solutions for hydride generation coupled was acquired by a Hewlett-Packard Model 35900 C analogueto HPLC, sodium tetrahydroborate(III ) (Probus) solutions (1, to-digital converter, using the chromatograph software.Peak 2, 3, 4 and 5% m/v) were prepared daily by dissolving NaBH4 area signals were recorded. powder in 0.7% m/v NaOH solution and filtering through Determination of total As in previously dry-ashed samples Whatman No. 42 filter-paper. was performed with an atomic absorption spectrometer All glassware was treated with 10% v/v HNO3 for 24 h and (Perkin-Elmer Model 5000) equipped with a flow injection then rinsed three times with QRG water before use.When not system (Perkin-Elmer FIAS-400) operating as a hydride generin use, glassware was placed in 10% v/v HNO3 for 24 h. ator in continuous-flow mode. Various canned seafood products were purchased at local A lyophilizer equipped with a microprocessor controlling retail outlets. Descriptions are given in Table 2. the lyophilization process (FTS Systems, New York, USA) was employed. The microprocessor was connected to an Epson Sample Preparation Equity I+ computer.Other equipment used included a Janke & Kunkel A10 The brine or sauce in the canned seafood products was removed by the method for determining the drained mass of water-cooled mill (Schott Glaswerke, Mainz, Germany) and a 92 Journal of Analytical Atomic Spectrometry, January 1997, Vol. 12Table 2 Seafood products analysed for AB extraction process was repeated three times and the extracts were combined.Sample Description (source) Seafood product Clean-up Fish— The methanol–water extract of the lyophilized samples was Salmon 01 Smoked salmon, skinless and boneless, water, salt, 150 g drained mass evaporated (T=55°C) to dryness, redissolved in 25 ml of 0.1 (Chile) mol l-1 HCl and adjusted to a pH less than 2 (1.8±0.2) with L amellibranchs— 4 mol l-1 HCl. The acidified solution was passed through a Razor 02 Pacific razor clams, water, salt, strong cation exchanger (Dowex 50W-X8, 1×6 cm).The Clams stabilizer (E-450) and antioxidant column was washed with 25 ml of reagent grade water and the (H-3246), 65 g drained mass (Chile) AB sorbed was eluted with 75 ml of 4 mol l-1 ammonia. The 03 Pacific razor clams, water, salt, 78 g drained mass (Chile) eluate was evaporated to dryness and the residue was redis- Cockles 04 Cockles, water, salt, 130 g drained solved in 3 ml of QRG water, filtered through Whatman No. 1 mass (The Netherlands) and No. 42 filter-paper and finally through a 0.45 mm filter.Langostillo 05 Langostillo, water, salt, 60 g drained mass (Pontevedra, Spain) AB Determination by HPLC–MO–HGAAS Mussel 06 Mussel in brine, water, salt, 70 g drained mass (Spain) For chromatographic separation, samples and standard solu- Clams 07 Clams, water, salt, 65 g drained mass tions were loaded into a 100 ml sample loop and injected onto (Spain) the chromatographic column. The eluate from the column was Gastropods— mixed with the persulfate solution before entering the micro- Snails 08 Sea snails, water, salt, stabilizer (E-450) and antioxidant (E-300, wave oven.The thermo-oxidized effluent, cooled in an ice- H-3246), 110 g drained mass (Chile) bath, was in turn interfaced via PTFE tubing to the continuous Crustaceans— HGAAS system. The HPLC separation was performed by Shrimp 09 Peeled shrimp, water, salt, citric acid, isocratic elution with 3 mmol l-1 phosphate buffer of pH 5. 120 g drained mass (Thailand) Quantifications were made by the method of standard 10 Peeled shrimp, water, salt, citric acid, additions and peak area signals were measured.The peak 120 g drained mass (Thailand) Crab 11 Fancy pacific crab, water, salt, areas recorded were the average of at least two injections of stabilizer (E-450) and antioxidant each solution. Spikes of AB were added to samples prior to (H-3246, E-300), 120 g drained mass injection. The amounts of AB added were approximately equal (Chile) to two and three times the AB contents determined previously by comparison with the calibration graph.In order to correct the data for reagent contamination, reagent blanks were canned foods.13 The samples obtained were pressed between taken through the total procedure (extraction, clean-up, two sheets of filter-paper, cut into pieces and frozen at-20 °C; HPLC–MO–HGAAS). The instrumental conditions and ana- they were then freeze-dried for 20 h. Sublimation heat was lytical parameters used are listed in Table 1.supplied by conduction from heating plates at 20°C. The lyophilized samples were crushed and homogenized to a fine RESULTS AND DISCUSSION powder in a water-cooled mill. The resulting powder was stored in previously decontaminated twist-off flasks and stored Extraction at 4°C until analysis. The AB recovery efficiency of the methanol–water extraction was evaluated by spiking three sub-samples of lyophilized Determination of Total As squid samples (mass of sample 2.00±0.01 g; As content 0.19 mg g-1, fm) with 1 ml of 2 mg ml-1 AB as As (0.28 mg g-1 As) The seafood products were dry-ashed, applying the dry minbefore performing the extraction process and determining AB eralization methodology developed previously,14 and total As as As by HGAAS after an ashing step.The mean recovery of was determined by HGAAS. The ash from the mineralized AB was 101±3%. samples was dissolved in 5 ml of 50% v/v HCl, washed with water and filtered through Whatman No. 1 filter-paper into a 25 ml calibrated flask. Determinations of total As were also Clean-up made in dry-ashed solid residues resulting from the methanol– In order to avoid the overlapping of AB and AsIII on the water extraction procedure. Calibration graphs made with AsV chromatographic column and to fractionate the arsenical were employed in each instance. The instrumental conditions species, a clean-up procedure is necessary. Preliminary assays used for the determination of As by HGAAS in continuous- were made with a Hypersep IC-OH anionic cartridge, as flow mode were as follows: atomic absorption spectrometer: employed by Lo�pez-Gonza�lvez et al.11 With this cartridge, wavelength, 193.7 nm; spectral bandpass, 0.7 nm; lamp power, AsIII, AsV, MMA and DMA were retained, while AB and AC 8.5 W (electrodeless discharge lamp).Hydride generation: cell passed through. It was observed that the cartridge does not temperature, 900 °C; sample solution, 1 ml min-1 flow rate; permit on-line separation of AB in samples with a high AB reducing agent, 1.5% m/v NaBH4 in 0.7% m/v NaOH, 1 ml content since the amount of phase employed is low.The need min-1 flow rate; HCl solution, 1.5 mol l-1, 2.5 ml min-1 flow to assay highly diluted samples in order to avoid overloading rate; carrier gas, argon, 45 ml min-1 flow rate. the cartridge would involve greater complexity in the treatment of the sample.Consequently, it was decided not to use the Methanol–Water Extraction cartridge. The quantitative recovery of an aqueous standard of AB The lyophilized seafood products were extracted by applying the methodology developed previously.15 The lyophilized after a Dowex clean-up procedure was described in a previous study.16 In this work, 32.185 ng of AB, expressed as As, sample (2.00±0.01 g) was weighed into a 50 ml centrifuge tube with screw top and conical base. A 40 ml volume of methanol– obtained from an extract of a sample of DORM-1 was quantitatively recovered after passing the extract through the water (1+1 v/v) was added and the tube was agitated for in in a mechanical arm shaker.The extract was collected Dowex resin. This was shown by a comparison of the AB contents found in DORM-1 samples without clean-up after centrifugation at 2000 rev min-1 (378g) for 10 min. The Journal of Analytical Atomic Spectrometry, January 1997, Vol. 12 93(16.10±0.82 mg g-1 As, dm) and with clean-up (16.52±0.62 mg were selected on the basis of giving a good separation for AB and DMA while providing the shortest retention times for g-1 As, dm) (dm=dry mass).After the Dowex clean-up, of the six species analysed (AsIII, both species. Fig. 2 shows a typical chromatogram, obtained under AsV, MMA, DMA, AB and AC), only AB and DMA were present in the extracts. Consequently, the overlapping of AsIII the optimum conditions described above and using HPLC–MO–HGAAS, of (a) a standard mixture of AsIII, AsV, and AB, tested in this work and frequently described in the literature, was avoided.9,17,18 The AsIII content in DORM-1 is MMA, DMA, AB and AC (10 ng As, respectively) spiked in aqueous solution, and chromatograms of a cleaned aqueous <50 ng g-1.19 Because of this, and given the high dilution (1+716) employed for the DORM-1 lyophilized sample analy- clam extract (b) and a cleaned aqueous clam extract spiked with standards of AB and DMA (c) (5 ng As both).As can be sis, there will be no overlap of AB with AsIII in the uncleaned sample. The Dowex resin employed in this work has a higher seen, AB is well separated from DMA, which co-elutes with AB in the clean-up step. exchange capacity than the anionic cartridge and can separate AB from AsIII in samples with a wide range of AB concentrations (0.5–16.5 mg g-1 As, dm). The clean-up procedure MO–HGAAS makes possible the isolation of AB from other arsenical species Preliminary MO was performed with an aqueous AB standard (MMA and AsV) with a high retention time (MMA, tR= (100 mg l-1 As), connected on-line with the HGAAS detector. 26 min; AsV, tR=26 min), thereby decreasing the time of analy- Initially, 1% persulfate and a 1.6 m coil (residence time 12 s) sis for each sample.were employed in order to establish the working power and oven load necessary for MO. The maximum power of the Evaporation Temperature Study for the AB Solutions microwave oven (1100 W) was selected as the working power.In order to speed up the process of evaporating the extract to Lower power levels led to intermittent suspension of microwave dryness after extraction and clean-up, recovery assays were radiation, giving rise to chromatographic peak splitting. An carried out with an AB standard (100 mg l-1 As), spiked in additional oven load was necessary to avoid overheating the methanol or ammonia solutions and evaporated at differ- system.A load of 400 ml of water was the minimum volume ent temperatures. AB was subsequently determined using that prevented the water from boiling during the time required HPLC–MO–HGAAS. At a temperature of 80°C, the mean (4 min) for determining AB by HPLC–MO–HGAAS. recovery of AB was 77±15%. AB can readily be converted to With a 1.6 m coil and 1% persulfate in 2.5% NaOH, signals trimethylarsine oxide and/or DMA by heating in the presence of aqueous standards (of 100 mg l-1 As of AsIII, AsV, MMA, of a base.20,21 A temperature of 55°C did not cause losses DMA, AB, AC) were compared by MO–HGAAS.The during the process of evaporating to dryness and the recovery absorbance signals obtained were of the same order for all the of AB was 100±9%. species. Le et al.8 obtained the same result with a 3 m coil and 2.7% potassium persulfate in 1.2% NaOH. The AsIII, AsV and HGAAS Conditions AB signals in 3 mmol l-1 phosphate buffer of pH 5.0 were compared, and it was observed that they provided the same The HCl concentration (3 mol l-1) and flow rate (1.9 ml absorbance readings as those obtained for the aqueous stan- min-1) and the NaBH4 flow rate (1.9 ml min-1) used here dards.Thus, under the conditions selected, complete conversion were the same as those proposed by Lo�pez-Gonza�lvez et al.11 of AB to hydride-forming arsenicals was achieved. in a previous work. Subsequently, different concentrations of NaBH4 (1, 2, 3, 4 and 5%) were tested for aqueous standards HPLC–MO–HGAAS of AsV and DMA (100 mg l-1 of each arsenical compound as As), in order to study the efficiency of AsH3 generation.For The effect of coil length (residence time) and persulfate concen- both arsenical species the signal increased until the NaBH4 tration on MO was studied with aqueous standards of AB and concentration reached 3%, and then remained constant up to concentrations of 4 and 5%.However, the use of NaBH4 concentrations above 2% involves difficult and tedious cleaning of the gas–liquid separator system after analysing each sample, owing to the excessive hydrogen bubbling produced during hydride generation. This makes it impractical to use NaBH4 concentrations above 2% for on-line AB determination. HPLC–MO–HGAAS Conditions HPLC separation of AB The ionic nature of organic and inorganic As compounds makes them amenable to anion-exchange HPLC. This type of chromatography uses buffers compatible with the HGAAS detection system.After the Dowex clean-up used in this work, only AB and DMA were present in the extracts injected onto the chromatographic column. In order to separate AB and DMA, a phosphate buffer and a polymer-based column (PRPX100) were selected which were also used in a previous study to determine MMA and DMA.15 In order to optimize the chromatographic separation of AB and DMA in isocratic mode, the phosphate buffer concentration was varied from 3 Fig. 2 HPLC–MO–HGAAS chromatograms: (a) standard mixture of to 20 mmol l-1 and the pH from 5 to 7. Low-concentration AsIII, AsV, MMA, DMA, AB and AC (10 ng As, respectively) spiked phosphate buffers produced a better resolution for AB and in aqueous solution; (b) cleaned aqueous clam extract; (c) cleaned DMA. A higher pH also improved the resolution of AB and aqueous clam extract spiked with standards of AB and DMA (5 ng DMA. Optimum chromatographic conditions were obtained As both).Peak identification: 1: AC; 2: AB; 3: AsIII; 4: DMA; 5: MMA; and 6: AsV. with a 3 mmol l-1 phosphate buffer of pH 5.0. These conditions 94 Journal of Analytical Atomic Spectrometry, January 1997, Vol. 12AsIII (100 mg l-1 As) injected separately onto the chromato- The detection limit, established as the AB concentration in the seafood product that provides an absorbance area reading graphic column and subsequently oxidized in the microwave oven.The AB MO efficiency, expressed in per cent., was statistically different from that of the blank, was calculated by dividing three times the standard deviation of the absorbance obtained by comparing peak area signals from AB with those obtained with AsIII. The data obtained are shown in Fig. 3. area readings, at the retention time of AB, of nine reagent blanks by the slope of the standard additions graph and taking For comparison with the AB signals, AsIII signals were chosen rather than AsV because the former has a shorter retention into account the sample mass and dilution employed.The dilutions used for the analysis of the AB water sample extracts time (3.4 min) than the latter (26 min) and because, as stated above, AsIII and AsV give the same absorbance signals with that provided good peak separation ranged from 1+1 to 1+79, hence, the detection limits for AB vary in accordance MO–HGAAS. Tests with coil lengths of 0.8, 1.0 and 1.3 m showed that with the dilution employed (from 0.68 to 27.20 ng g-1 As, fm).The detection limits for AB on a fresh mass basis were when the persulfate concentration was increased from 1 to 3 or 5% there was an increase in the peak area signals obtained calculated by taking the average moisture content of seafood products to be 75%.3 The precision of the method, calculated for AB, but the signals obtained were always lower than the signals obtained for AsIII. Using a 1.6 m coil, for all the by independent analysis of 11 canned seafood products, analysing each in trcate, ranged from 0.4 to 6%.For DORM-1, persulfate concentrations tested, complete decomposition of AB to arsine-forming species was observed, as shown by the the precision obtained from three sub-samples was 4%. The mean recovery, evaluated by spiking three canned seafood sub- equality of the AB and AsIII peak area signals. In order to reduce consumption of reagents, the definitive working con- samples of shrimps, snails and mussels with AB, was 104±5%.ditions selected were a coil length of 1.6 m (residence time 12 s) and a persulfate concentration of 1%. AB Determination in Real Samples The levels of total As and AB expressed in terms of fresh mass Analytical Features of the Method for all the seafood samples analysed are given in Table 4. For The analytical characteristics (Table 3), such as detection limit, total As, the range found was 0.55–2.65 mg g-1 As (fm). precision and recovery, were evaluated in samples of seafood Samples of cockles, mussels, crab and clams gave values over products prepared as described under Experimental. 2 mg g-1 As (fm). For AB, the levels found ranged from 0.03 to 1.61 mg g-1 As (fm). As can be seen, in all instances AB levels in the samples analysed were above the detection limit of the method. The percentages of total As represented by these contents are also shown in Table 4. The highest percentages of AB (75 and 61%) were found in crab and salmon, respectively.The remaining samples showed an AB content ranging from 5 to 49% of total As. The lowest percentages of AB corresponded, in general, to the bivalve group, with contents of AB ranging from 8 to 36%. These results are similar to those reported by us for percentages of total As represented by AB in canned lamellibranchs in a previous study.3 Other workers have also reported low percentages of As represented by AB: Shibata and Morita22 in NIES (National Institute of Environmental Studies) CRM No. 6 Mussel (13%), and Larsen et al.19 in NIST SRM 1566 Oyster Tissue and NIES CRM No. 6 Mussel Fig. 3 Microwave oxidation efficiency versus residence time of AsIII and AB (100 ng ml-1 As both). A, 1% Potassium persulfate; B, 3% (11 and 14%, respectively). potassium persulfate; C, 5% potassium persulfate; D, arsenite. By considering the levels of AB and total water-soluble As we were able to obtain the As levels that would correspond to water-soluble species other than AB.These levels were around Table 3 Analytical characteristics of the method 1 mg g-1 As (fm) for razor clams and mussels. A proportion of AB as As this As could be due to AsIII, AsV, MMA, DMA, AC, TMAO (trimethylarsine oxide) and TMA+ (tetramethylarsonium ion), Detection limit*/ng g-1 2.72 (dm)#0.68 (fm) 108.80 (dm)#27.20 (fm) and also to arsenosugars. Shibata and Morita22 and Larsen23 have stated that bivalves Precision (RSD) (%)— generally contain not only AB but also As-containing ribofur- Seafood† 0.4–6 anosides as major As species.The relatively large content of DORM-1‡ 4 arsenosugars found in the mussel and oyster samples can be explained by the fact that shellfish feed on marine algae, which Recovery§ (%)— Shrimps 102±7 (0.03, 0.04) are the arsenosugar-synthesizing organisms in the marine food Snails 109±3 (0.62, 0.34) chain.23 There is good agreement for the arsenosugar levels Mussel 100±5 (0.48, 0.22) obtained by the above-mentioned workers in NIES CRM Mean 104±5 No. 6 Mussel, employing a water–methanol–chloroform extraction system23 or a methanol–water system.22 The latter * Nine reagent blanks were employed; the first set of values corre- extraction technique, as modified by us,15 i.e., by changing the spond to a 1+1 dilution and the second set to a 1+79 dilution; results expressed as As; dm=dry mass; fm=fresh mass. † RSD interval volume of methanol–water and reducing the number of extracobtained from independent analysis of 11 canned seafood products.tions, was also employed in this work for the quantitative Each seafood was analysed in triplicate. ‡ Mean RSD obtained from extraction of AB. three independent analyses of the DORM-1 sample. § Percentage The mean As concentration corresponding to species other recoveries expressed as mean±s from three independent analyses. than AB in the canned seafood samples analysed was 1.02 Values in parentheses are the species average concentration of the mg g-1 As (fm), which represents 70% of the mean total As unspiked samples (first value) and concentration of added species (second value) in mg g-1 As (fresh mass).(1.45 mg g-1 As, fm). Obviously, more data on the arsenical Journal of Analytical Atomic Spectrometry, January 1997, Vol. 12 95Table 4 Total As, total water-soluble As, water-soluble As corresponding to species other than AB, AB contents and percentages of total As represented by AB in canned seafood samples Total water-soluble Water-soluble As Seafood product Sample Total As* AB* AB (%)† As*,‡ other than AB* Fish— Salmon 01 0.23 0.14 61 0.17 0.03 L amellibranchs— Razor clams 02 1.55 0.13 8 0.96 0.83 03 0.62 0.07 11 0.35 0.28 Cockles 04 2.06 0.10 5 0.50 0.40 Langostillo 05 1.92 0.69 36 1.35 0.66 Mussel 06 2.65 0.48 18 1.22 0.74 Clams 07 2.06 0.44 21 1.02 0.58 Gastropods— Snails 08 1.27 0.62 49 0.78 0.16 Crustaceans— Shrimps 09 0.87 0.40 46 0.62 0.22 10 0.55 0.03 5 0.33 0.30 Crab 11 2.15 1.61 75 2.01 0.40 * Results expressed in mg g-1 As, fresh mass.† Percentages of total As. ‡ Total As in samples-As in solid residue resulting from the methanol– water extraction. species contained in seafood are needed.24 Also, studies of the REFERENCES possible effects of these compounds on humans are required. 1 Morita, M., and Edmonds, J. S., Pure Appl. Chem., 1992, 64, 575. The AB content detected in DORM-1 (16.5±0.6 mg g-1 As, 2 Phillips, D.J. H., Aquat. T oxicol., 1990, 16, 151. dm) using HPLC–MO–HGAAS is compared in Table 5 with 3 Ve� lez, D., Yba�n� ez, N., and Montoro, R., J. Agric. Food Chem., results reported in the literature. The value obtained here was 1995, 43, 1289. 4 Atallah, R. H., and Kalman, D. A., T alanta, 1991, 38, 167. very close to those obtained by other workers using HPLC– 5 Violante, N., Petrucci, F., La Torre, F., and Caroli, S., ICP–MS,19,20,25,26 and also to the result obtained by us in a Spectroscopy, 1992, 7, 36. previous study to determine AB by HPLC–ICP.16 6 Howard, A.G., and Hunt, L. E., Anal. Chem., 1993, 65, 2995. 7 Albertý�, J., Rubio, R., and Rauret, G., Fresenius’ J. Anal. Chem., 1995, 351, 415. CONCLUSIONS 8 Le, X.-C., Cullen, W. R., and Reimer, K. J., Appl. Organomet. Chem., 1992, 6, 161. The proposed method permits the sensitive, precise and accu- 9 Le, X.-C., Cullen, W. R., and Reimer, K.J., T alanta, 1994, 41, 495. rate determination of AB in real samples representative of 10 Lo�pez-Gonzalvez, M. A., Go�mez, M. M., Ca�mara, C., and seafood products. Moreover, the methodology described here Palacios, M. A., Mikrochim. Acta, 1995, 120, 301. provides the performance required for use with the instrumen- 11 Lo�pez-Gonza�lvez, M. A., Go�mez, M. M, Ca�mara, C., and tation available in many control laboratories. The high level Palacios, M. A., J. Anal. At. Spectrom., 1994, 9, 291.of arsenical species other than AB found in some samples 12 Martin, I., Lo�pez-Gonza�lvez, M. A., Go�mez, M. M., Ca�mara, C., and Palacios, M. A., J. Chromatogr. B, Biomed. Appl., 1995, emphasizes the need to obtain a better understanding of the 666, 101. As species composition of seafood. The hyphenated technique 13 Presidencia de Gobierno, Orden 21 de noviembre de 1984 por la proposed here could make a substantial contribution to this que se aprueban las normas de calidad para conservas, BOE aim.(Boletý�n Oficial del Estado), 30/11, 1–3/12/1984, 287, 34574. 14 Yba�n� ez, N., Cervera, M. L., and Montoro, R., Anal. Chim. Acta, Funds to carry out this work were provided by the Comisio�n 1992, 258, 61. 15 Ve� lez, D., Yba�n� ez, N., and Montoro, R., J.encia y Tecnologý�a (CICyT), Project 1996, 11, 271. ALI92-0147. D.V. received a Research Personnel Training 16 Yba�n� ez N., Ve�lez, D., Tejedor, W., and Montoro, R., J. Anal. At. Grant from the Ministerio de Educacio�n y Ciencia. We are Spectrom., 1995, 10, 459. grateful to Maite de la Flor for her assistance in the perform- 17 Lo�pez-Gonza�lvez, M. A., Go�mez, M. M., Palacios, M., and ance of analytical work. We also thank the Service Central Ca�mara, C., Fresenius’ J. Anal. Chem., 1993, 346, 643. d’Analyse du CNRS (SCA) of Vernaisson (France) for provid- 18 Caroli, S., La Torre, F., Petrucci, F., and Violante, N., Environ. Sci. Pollut. Res., 1994, 4, 205. ing the arsenobetaine and arsenocholine standards. 19 Larsen, E. H., Pritzl, G., and Hansen, S. H., J. Anal. At. Spectrom., 1993, 8, 1075. 20 Beauchemin, D., Bednas, M. E., Berman, S. S., McLaren, J. W., Siu, K. W. M., and Sturgeon, R. E., Anal. Chem., 1988, 60, 2209. Table 5 Determination of AB in DORM-1 sample; results in mg g-1 21 Kaise, T., Yamauchi, H., Hirayama, T., and Fukui, S., Appl. As dry mass Organomet. Chem., 1988, 2, 339. 22 Shibata, Y., and Morita, M., Appl. Organomet. Chem., 1992, 6, 343. AB Analytical technique ref. 23 Larsen, E. H., Fresenius’ J. Anal. Chem., 1995, 351, 582. 15.7±0.8 HPLC–ICP-MS 20 24 Vather, M., Clin. Chem., 1994, 40, 679. 15.7 HPLC–ICP-MS 25 25 Shibata, Y., and Morita, M., Anal. Chem., 1989, 61, 2116. 14.2 HPLC–ICP-MS 19 26 Branch, S., Ebdon, L., and O’Neill, P., J. Anal. At. Spectrom., 15.1±0.6* HPLC–ICP-MS 26 1994, 9, 33. 16.1±0.4* HPLC–ICP-MS 26 16.5±0.9 HPLC–ICP-AES 16 Paper 6/02769E 16.5±0.6 HPLC–MO–HGAAS This work Received April 22, 1996 Accepted October 2, 1996 * Values obtained by employing two extraction procedures. 96 Journal of Analytical Atomic Spectrometry, Jan