Direct Determination of Arsenic in Sea-water by Continuous-flow Hydride Generation Atomic Fluorescence Spectrometry JORGE MOREDA-PIN� EIROa, M. LUISA CERVERAb AND MIGUEL DE LA GUARDIA*b aDepartment of Analytical Chemistry, Nutrition and Bromatology, Faculty of Chemistry, University of Santiago de Compostela, Av. de L as Ciencias, s/n 15706-Santiago de Compostela, Spain bDepartment of Analytical Chemistry, Faculty of Chemistry, University of Valencia, Dr. Moliner 50, 96100-Burjassot, Valencia, Spain A highly sensitive and simple procedure was developed for the The use of in situ trapping of the generated hydride vapour direct determination of total As in sea-water samples by hydride in a hot coated graphite furnace (HG-ETAAS) presents higher generation atomic fluorescence spectrometry.The method sensitivity than HGAAS owing to the avoidance of hydride involves the generation of arsenic hydride from sea-water dilution in the Ar flow and the use of atomization temperatures samples, diluted with HCl to a final HCl concentration of higher than 1000 °C.Thus, this technique has been recognized 2 mol l-1, which were merged with a reducing solution, viz., as the most sensitive detection system for trace metal determi- 3% m/v NaBH4. The sample and NaBH4 were pumped at flow nation.6 However, the main disadvantages of this technique rates of 6.0 and 1.3 ml min-1, respectively, and allowed to react are the necessity of using porous graphite structures to obtain in a 200 cm×0.8 mm id reaction coil.The evolved arsenic adequate hydride trapping and the reduced possibility of hydride was removed using an argon flow rate of 400 ml min-1 routine monitoring. and passed to a hydrogen diVusion flame where the atomic The use of atomic fluorescence presents a sensitivity compar- fluorescence of As was measured at 193.7 nm. With the proposed able to the highest sensitivity oVered by HG-ETAAS with a procedure a detection limit of 5.0 ng l-1 was achieved.The reduced cost because graphite tubes are unnecessary, thus oVerrepeatability of the determination varied between 1.5 and 4.0%. ing an attractive detection system for the direct determination of The accuracy was confirmed by the analysis of two sea-water As in liquid samples at trace levels. The literature concerning As reference materials (NASS-4 and CRM-403) and by recovery determination in natural waters and geological and biological studies on natural samples spiked with known concentrations of samples by HGAFS is extensive;12–25 however, only a single AsIII and AsV.The proposed method was successfully applied to paper concerning As determination in sea-water samples by the determination of As in several sea-water samples. The HGAFS was found in the literature,26 which used non-dispersive number of samples that can be analysed is 40 per hour. AFS and radiofrequency-excited electrodeless discharge lamps. The purpose of this paper was the development of a highly Keywords: Arsenic; sea-water; hydride generation; atomic sensitive and accurate method for the direct determination of fluorescence trace amounts of As in sea-water samples using HGAFS.The method should be applicable to routine analysis and Electrothermal atomic absorption spectrometry (ETAAS) has monitoring studies. been extensively used for As determination in several samples owing to its sensitivity and accuracy. However, for complex EXPERIMENTAL samples with a high saline content, such as sea-water, several problems occur owing to the important interference eVects Apparatus from NaCl and K2SO4.1,2 In addition, the As concentration A Unicam VP-90 continuous-flow vapour system equipped in sea-water3 samples is below the detection limit of this with a B-type gas–liquid separator (Cambridge, UK) and a technique, around 1.1–1.9 mg l-1.These levels are reduced Perma pure drier tube (PS Analytical, Sevenoaks, Kent, UK) considerably for unpolluted areas.Thus, the use of diVerent was used. An Excalibur atomic fluorescence detector (PSA preconcentration procedures is necessary, methods being 10033; PS Analytical), equipped with a boosted discharge unavailable for routine analysis. hollow cathode lamp (BDHCL) (Superlamp; Photon, Victoria, The main advantage of analytical methodology based on Australia) for As as the excitation source, a hydrogen diVusion covalent hydride generation (HG) is that it allows the separa- flame as the atom cell, a series of lenses to collect and focus tion of the matrix and aVords increased sensitivity as compared useful radiation, and a specific filter to achieve isolation and with ETAAS, while also oVering automation facilities and the possibility of speciation.4 reduction of flame emission in conjunction with a solar blind Hence, HG combined with atomic absorption spectrometry photomultiplier, was used.Measurements were carried out at (AAS and ETAAS) as the detection system has been used for the resonance wavelength of As (193.7 nm).the direct determination of As in several samples. The application of this technique is well documented and has recently Reagents been reviewed by Fang et al.5 and Dedina.6 However, the All solutions were prepared from analytical-reagent grade sensitivity achieved using HGAAS is often inadequate for nonchemicals using ultrapure water, with a resistivity of 18MV cm, contaminated sea-water samples, making previous preconcenwhich was obtained from a Milli-Q water-purification system tration steps necessary.Thus, diVerent preconcentration (Millipore, Bedford, MA, USA). An AsIII stock standard solu- procedures such as liquid–liquid extraction,7 flotation,8,9 tion (1.000 g l-1) was prepared by dissolving 1.320 g of As2O3 preconcentration on solid resins10 and cryogenic trapping11 (Riedel-de Hae�n, Hannover, Germany) in 25 ml of 20% m/v have been used, increasing the sensitivity but also the analysis time.KOH solution, neutralizing with 20% v/v H2SO4 and diluting Journal of Analytical Atomic Spectrometry, December 1997, Vol. 12 (1377–1380) 1377Table 1 Operating conditions for HGAFS to 1 l with 1% v/v H2SO4. The AsV stock standard solution (Titrisol; 1.000 g l-1) was obtained from Merck (Darmstadt, Parameter Germany). Sodium tetrahydroborate (Fluka, Buchs, Spectrometer operating conditions — Switzerland) dissolved in 0.5% m/v NaOH (Carlo-Erba, Milan, Resonance wavelength/nm 197.3 Italy) was used as the reducing solution.This solution was Bandpass/nm 0.5 prepared daily and filtered before use. Hydrochloric acid Primary current/mA 27.0 solution was prepared from 37% HCl (Merck). NASS-4 Open Boost current/mA 35.0 Ocean Seawater Reference Material for Trace Metals (National Gain ×10 Hydride generation conditions — Research Council of Canada) and CRM-403 Sea Water, sup- [HCl]/mol l-1 2.0 plied by the Commission of the European Communities, were [NaBH4] (%m/v) 3.0 used to evaluate the accuracy of the developed procedure.Reaction coil length/cm 200 Argon C-45 (purity 99.995%) was used as the carrier gas Sample/HCl flow rate/ml min-1 6.0 for the atomizer and as the internal purge gas and was obtained NaBH4 flow rate/ml min-1 1.3 from Carburos Metalicos (Barcelona, Spain). Synthetic air Ar flow rate/ml min-1 400 Atomic fluorescence measurement — (Carburos Metalicos) was used to dry the generated vapour Delay time/s 15 phase in the Perma pure drier tube. Rinse time/s 30 Measurement time/s 40 Memory time/s 30 Procedure for Sample Collection Measurement mode Peak height Sea-water samples were collected from coastal surface water of the Mediterranean Sea near to Valencia in 100 ml polyethyl- Hydrochloric acid concentration ene bottles.The samples were immediately acidified with 100 ml of concentrated HNO3, which provided a pH lower than 2, to The eVect of varying the HCl concentration on the atomic avoid the adsorption of As onto the polyethylene bottle walls.fluorescence from AsIII and AsV is shown in Fig. 2. As can be seen, the signals related to AsIII and AsV increase up to a concentration of 2 mol l-1 HCl. In addition, the response Procedure for Measurement obtained from AsV is around 30% lower than that obtained The method involves the continuous gation of arsenic from AsIII.This is due to the poor eYciency of arsine generation hydride from sea-water samples diluted with HCl to a final obtained using a low NaBH4 concentration (1.5%) and the HCl concentration of 2 mol l-1, which were merged with short length of the reaction coil employed (11 cm), as will be a reducing solution, viz., 3.0% m/v NaBH4. The evolved seen in the following sections. arsenic hydride was transferred, using an argon flow rate of 400 ml min-1, to the atomic fluorescence detector. Reaction coil Fluorescence measurements for samples were interpolated using the calibration line obtained with AsIII standards.A The variation of the eYciency of arsenic hydride generation schematic diagram of the continuous flow injection system is from AsIII and AsV was evaluated using reaction coils of shown in Fig. 1. The operating conditions for HGAFS are diVerent lengths (11, 100, 200 and 300 cm). Results (Fig. 3) shown in Table 1. show that a reaction coil longer than 200 cm is necessary to obtain complete volatilization of AsV.For AsIII no variation of the hydride generation eYciency with the length of the RESULTS AND DISCUSSION reaction coil was observed. Thus, a reaction loop of 200 cm Evaluation of Arsenic Hydride Generation Conditions was selected for further experiments. All experiments were performed on natural sea-water samples and aqueous standard solutions of 1 mg l-1 As expressed as Sodium tetrahydroborate concentration AsIII and AsV.The NaBH4 concentration is an important parameter for arsine generation because arsenic hydride is formed in the presence Fig. 2 EVect of HCl concentration on arsine generation from AsIII and AsV solutions. The concentration of NaBH4 and reaction coil length were 1.5% m/v and 11 cm, respectively. The AsIII and AsV concentrations were 1.0 mg l-1. Error bars indicate the variability of fluorescence measurement as±the standard deviation of three indepen- Fig. 1 Schematic diagram of the continuous-flow system used.dent measurements. 1378 Journal of Analytical Atomic Spectrometry, December 1997, Vol. 12Fig. 3 EVect of the reaction coil length on the integrated signal of AsIII and AsV solutions. The concentrations of HCl and NaBH4 were 2.0 mol l-1 and 1.5% m/v, respectively. The AsIII and AsV concentrations were 1.0 mg l-1. of hydrogen generated by NaBH4 in an acidic medium and because the flame in which the generated hydride is atomized is maintained by the excess of hydrogen produced in this reaction.As can be seen in Fig. 4, an NaBH4 concentration higher than 3.0% m/v is required to obtain complete arsenic hydride generation from both AsIII and AsV. For NaBH4 concentrations lower than 1.5% m/v, the flame is extinguished, while for NaBH4 concentrations higher than 3.5% the instability in the flame caused by the excess of hydrogen gives a poor Fig. 5 EVect of the standard/sample (a) and NaBH4 (b) flow rates on reproducibility and sensitivity of the fluorescence measure- the fluorescence of AsIII and AsV solutions.The HCl and NaBH4 concentrations were 2 mol l-1 and 1.5% m/v, respectively; the reaction ments. Thus, an NaBH4 concentration of 3.0% m/v was coil length was 200 cm. The AsIII and AsV concentrations were selected in order to obtain the best analytical performance. 1.0 mg l-1. As can be seen, the hydride generation from AsV exhibits slow kinetics compared with AsIII, making it necessary to increase the NaBH4 and HCl concentrations to obtain a good mentioned earlier.For an NaBH4 flow rate lower than comparability between the results found for both species.27,28 1.0 ml min-1 the flame is extinguished owing to the low hydrogen concentration obtained. Therefore, a sample flow Flow rate parameters for standard/sample and NaBH4 rate of 6 ml min-1 and an NaBH4 flow rate of 1.3 ml min-1 were selected for further experiments. The eYciency of the arsenic hydride generation was studied for diVerent flow rates of standard/sample in 2.0 mol l-1 HCl and NaBH4.Results shown in Fig. 5(a) indicate an increase in Argon flow rate the As fluorescence signals with an increase in standard/sample An increase in the Ar flow rate produces an increase in the flow rate up to 6.0 ml min-1. However, a decrease in the atomic fluorescence up to an Ar flow rate of 400 ml min-1. hydride generation eYciency was obtained for an NaBH4 flow However, for Ar flows higher than this value, samples were rate higher than 2.0 ml min-1 [see Fig. 5(b)], for the reasons diluted and fluorescence signals reduced. Hence, a 400 ml min-1 Ar flow was selected as the most convenient. Analytical Figures of Merit The calibration and standard additions equations obtained for aqueous standard solutions and natural sea-water samples spiked with AsIII and AsV are shown in Table 2. As can be seen, no matrix eVect was observed; in addition, the slopes obtained for the AsIII and AsV calibration and standard additions equations are similar.The calibration and standard additions graphs Table 2 Calibration and standard additions graphs* AsIII AsV Calibration I=1.5+86.9[As] I=1.7+79.7[As] r=0.9999 r=0.9999 Fig. 4 EVect of NaBH4 concentration on arsine generation from AsIII Standard additions I=85.7+88.9[As] I=84.3+82.2[As] (natural sea-water) r=0.9987 r=0.9999 and AsV solutions. The concentration of HCl and reaction coil length were 2.0 mol l-1 and 200 cm, respectively.The AsIII and AsV concentrations were 1.0 mg l-1. * [As] expressed as mg l-1. Journal of Analytical Atomic Spectrometry, December 1997, Vol. 12 1379obtained were linear up to a concentration of 1.25 mg l-1 for indicative value of 1.461 mg kg-1. Results obtained were 1.28±0.03 mg l-1 and 1.51±0.02 mg kg-1, respectively. These which a signal intensity lower than 200 was obtained. For samples with a high As concentration, the use of a low gain was results are in good agreement with the certified values.selected to obtain an atomic fluorescence intensity below 200. The detection and quantification limits, defined as 3sb/m Application and 10sb/m, where sb is the standard deviation of 11 measurements of a blank and m is the slope of the calibration graphs, The proposed method was applied to the determination of As were 5.0 and 17 ng l-1, respectively. As can be observed, the in diVerent samples from the Mediterranean Sea.Results sensitivity achieved is adequate for As determination in non- ranged from 1.1 to 1.6 mg l-1, with RSDs from 0.5 to 1.7%. polluted sea-water samples and in open ocean sea-water Hence, the developed procedure is suitable for the direct samples. The sensitivity achieved by using this direct procedure determination of As in natural sea-water samples with good for As determination is considerably improved with respect to precision. published literature values concerning the use of HG-ETAAS.Hence, Tsalev et al.29 reported characteristic masses of 31–35 pg for inorganic and organic As species using Ir–Zr- CONCLUSION coated graphite tubes. Ding and Sturgeon30 reported a detec- The use of HGAFS provides adequate sensitivity and accuracy tion limit of 84 ng l-1 using Ir- and Pd-coated graphite tubes for the direct determination of As in coastal and open ocean and electrochemical hydride generation. Willie31 reported a sea-water samples, avoiding the tedious preconcentration pro- detection limit of 140 ng l-1 using an Ir-coated graphite tube, cedures required by using other AAS techniques.The number corresponding to a characteristic mass of 41 pg. of samples that can be analysed is 40 per hour, making the As compared with the previously reported procedures for method suitable for routine analysis and monitoring studies. the simultaneous determination of As, Se, Sn and Hg by nondispersive atomic fluorescence,21 the procedure developed here is simpler, because it does not require the use of expensive REFERENCES radiofrequency-excited discharge lamps.Additionally, the detection limit found by us is four times lower than that 1 Chakraborti, D., de Songhe, W., and Adams, F., Anal. Chim. reported for a 5 ml sample volume. Acta, 1980, 119, 331. 2 Saeed, K., and Thomassen, Y., Anal. Chim. Acta, 1981, 130, 281. The repeatability (relative standard deviation for seven repli- 3 Metals and T heir Compounds in the Environment, ed.Merian, E., cate measurements of the same sample spiked at diVerent VCH, New York, 1991, vol. II.22, p. 1101. concentration levels) [RSD (%)] obtained was lower than 4 Guo, X., and Guo, X., Anal. Chim. Acta, 1995, 310, 377. 4.0% for all concentrations tested, as can be seen in Table 3. 5 Fang, Z., Xu, S., and Tao, G., J. Anal. At. Spectrom., 1996, 11, 1. Analytical recovery values close to 100% were obtained for 6 Dedina, J., and Tsalev, D.L., Hydride Generation Atomic experiments carried out using sea-water samples spiked with Absorption Spectrometry, Wiley, Chichester, 1995. 7 Amankakwash, S. A., and Fasching, J. L., Talanta, 1985, 32, 111. AsIII and AsV. The values obtained, shown in Table 4, were 8 Nakashima, S., Fresenius’ J. Anal Chem., 1991, 341, 570. calculated using the slope of the calibration graph for AsIII. As 9 Nakashima, S., and Yagi, M., Bunseki Kagaku, 1983, 32, 535. can be seen, AsV is reduced eYciently using the selected 10 van Elteren, J.T., Gruter, G. J., Das, H. A., and Brinkman, U. A. experimental conditions. Th., Int. J. Environ. Anal. Chem., 1991, 43, 45. The accuracy of the method was evaluated by analysing 11 van Cleuvenbergen, R. J. A., van Mol, W. E., and Adams, F. C., diVerent reference materials: NASS-4, with a certified inorganic J. Anal. At. Spectrom., 1988, 3, 169. 12 Hueber, D., Smith, B. W., Madden, S., and Winefordner, J. D., As concentration of 1.26±0.04 mg l-1, and CRM-403, with an Appl. Spectrosc., 1994, 48, 1213. 13 Stockwell, P. B., and Corns, W. T., Analyst, 1994, 119, 1641. 14 Heitmann, U., Sy, T., Hese, A., and Schoknecht, G., J. Anal. At. Spectrom., 1994, 9, 437. Table 3 Analytical recovery studies* 15 Stockwell, P. B., and Corns, W. T., J. Autom. Chem., 1993, 15, 79. 16 Waldock, M. J., Mikrochim. Acta, 1992, 109, 23. [As] added/mg l-1 Analytical recovery (%) 17 Corns, W. T., Stockwell, P. B., Ebdon, L., and Hill, S.J., J. Anal. At. Spectrom., 1993, 8, 71. AsIII AsV 18 Guo, T., Liu, M., and Schrader, W., J. Anal. At. Spectrom., 1992, 7, 667. 0.25 101±1 103±3 0.5 107±3 102±2 19 Ebdon, L., and Wilkinson, J. R., Anal. Chim. Acta, 1987, 194, 177. 20 Rigin, V. I., Zh. Anal. Khim., 1986, 41, 46. 0.75 100±4 108±2 1.0 101±1 100±1 21 D’Ulivo, A., PapoV, P., and Festa, C., Talanta, 1983, 30, 907. 22 Rigin, V. I., Zh. Anal. Khim., 1983, 38, 1060. 1.25 99±2 104±2 23 Ebdon, L., Wilkinson, J. R., and Jackson, K. W., Anal. Chim. Acta, 1982, 136, 191. * Natural sea-water samples were spiked with diVerent concentrations of AsIII or AsV and analysed by the proposed procedure using 24 Kuga, K., and Tsujii, K., Anal. L ett., 1982, 15, 47. 25 Azad, J., Kirkbright, G. F., and Snook, R. D., Analyst, 1980, AsIII standards for calibration. 105, 79. 26 D’Ulivo, A., Fuoco, R., and PapoV, P., T alanta, 1985, 32, 103. 27 Lo� pez, A., Torrealba, R., Palacios, M. A., and Ca�mara, C., T alanta, 1992, 39, 1342. Table 4 Repeatability of As determination by HGAFS* 28 Torrealba, R., Bonilla, M., Palacios, M. A., and Ca�mara, C., Analusis, 1994, 22, 478. AsIII concentration/mg l-1 RSD (%) 29 Tsalev, D. L., D’Ulivio, A., Lampugnani, L., Di Marco, M., and 0 3.1 Zamboni, R., J. Anal. At. Spectrom., 1996, 11, 989. 0.25 2.1 30 Ding, W. W., and Sturgeon, R. E., Spectrochim. Acta, Part B, 0.5 3.2 1996, 51, 1325. 0.75 4.0 31 Willie, S. N., Spectrochim. Acta, Part B, 1996, 51, 1781. 1.0 1.5 Paper 7/05264B 1.25 1.7 Received July 22, 1997 * Values reported were found for seven independent measurements Accepted September 22, 1997 of each solution. 1380 Journal of Analytical Atomic Spectrometry, December 1997, V