Determination of Cd and Pb in seawater by graphite furnace atomic absorption spectrometry with the use of hydrofluoric acid as a chemical modifier

Transcription

1 Spectrochimica Acta Part B 57 (00) Determination of Cd and Pb in seawater by graphite furnace atomic absorption spectrometry with the use of hydrofluoric acid as a chemical modifier J.Y. Cabon* UMR CNRS 651-UBO, 6 Avenue Le Gorgeu, BP 809, 985 Brest-Cedex, France Received 1 June 001; accepted 11 December 001 Abstract High concentration of added hydrogen fluoride converted the seawater chloride to the corresponding fluoride matrix, and the liberated hydrochloric acid could be removed during the drying step. The atomization of cadmium and lead could be performed at a relatively low temperature (;1300 8C) at which the vaporization of the fluoride matrix was relatively slow, and the corresponding weak background signals could be separated from the analytical signals in time. Experimental conditions for the determination of Cd and Pb in seawater in the presence of HF were optimized with the use of the a priori calculation of the limit of detection. The experimental limit of detection obtained for Cd and Pb were, respectively, and 0.5 mg l for a 15-ml seawater sample (3s, 0 replicates). The concentrations of Cd determined in a SLEW-1 estuarine water and a CASS- seawater were 0.00"0.00 and 0.016"0.00 mg l Cd, respectively, in good agreement with the 0.018"0.003 and 0.019"0.004 mg l Cd certified values (At the 95% confident level, 10 replicates). 00 Elsevier Science B.V. All rights reserved. Keywords: Atomic spectrometry; Graphite furnace; Seawater; Cd; Pb; HF; Chemical modification 1. Introduction The determination of anthropogenic trace metal pollutants, like Cd and Pb, in natural waters is of great importance because they are involved in biological cycles and present a high toxicity. This paper was presented at Colloquium Spectroscopicum Internationale XXXII, held in Pretoria, South Africa, 8 13 July 001 and is published in the special issue of Spectrochimica Acta Part B, dedicated to that conference. *Tel.: q ; fax: q address: jean-yves.cabon@univbrest.fr (J.Y. Cabon). Graphite furnace atomic absorption spectrometry (GFAAS) has been widely used for trace metal determinations in various types of environmental samples. However, direct determination of Cd and Pb by GFAAS is difficult in seawater as the saline matrix can cause both spectral (high background absorption) and non-spectral interference effects. To overcome these problems, different preconcentrationyseparation procedures have been generally used. However, most of these analytical procedures are generally specific of only one chemical form and, to determine total concentrations, different pretreatment procedures have to be applied. Con /0/$ - see front matter 00 Elsevier Science B.V. All rights reserved. PII: S Ž

2 514 J.Y. Cabon / Spectrochimica Acta Part B 57 (00) sequently, these analytical procedures are generally time-consuming and subject to contamination. In the case of direct determination of trace metals in seawater by GFAAS, it is necessary to minimize both the interference effects and the magnitude of the background absorption signal at the atomization stage to improve the baseline noise and to reduce errors. This can be generally achieved through a chemical modification of the chloride matrix leading to less interfering species having different absorbance spectra and volatilities. To minimize the simultaneous background absorption signal, two ways can be considered: one way, by delaying sufficiently the vaporization of the element in order to eliminate the major part of the salts before atomization (mainly sodium species and volatile decomposition products). In the case of Pb, chemical modifiers like Pd introduced under different forms have been used for this purpose w1 4x. This procedure cannot be used in the case of Cd because the delaying effect is not sufficient to eliminate sodium species without important losses of analyte. Therefore, a second way has to be used by increasing the volatility of the element and selecting a low atomization temperature in order to obtain a good separation of the analyte and the vaporization of sodium species. This can be generally achieved with the use of organic acids like ascorbic or oxalic acid w5 8x; these acids promote a lower atomization temperature of Cd and Pb in seawater and a decrease of the NaCl background absorption signal, particularly with the use of oxalic acid w3,8x. However, organic acids also lead to decomposition products and eventually to the production of carbon residues in the atomizer. Sodium hydroxide has also been used in the case of Cd w9x, but the mass of sodium species in the atomizer is obviously increased and the sodium chloride absorption signal cannot be reduced. A HNO 3 (NH 4) HPO4 modification has been also used, but in this case the optimization of pyrolysis conditions appeared relatively difficult w10 1x. In this work, we examine the use of HF as a new chemical modifier for the seawater matrix. In marine environments, this acid has been currently used for the solubilization of sediments or particulate matter and eventually, for slurry sampling w13x. This acid has been also shown to reduce the background absorption signal of NaCl w14x. Consequently, it appeared interesting to us to examine the modification of the seawater matrix with the use of this chemical modifier. The removal of chloride by HF at the drying step was followed by ion chromatography. Its influence on the atomization signal of Pb and Cd in seawater was then examined. The experimental conditions for the determination of Cd and Pb in seawater have been optimized with the use of the a priori calculation of the detection limit w15,16x.. Experimental.1. Instrument parameters and operation A Perkin-Elmer 4100ZL was used for all atomic absorption measurements (Frequency measurements f s54 Hz). End-capped pyrolytic graphite coated tubes equipped with integrated platforms (Perkin-Elmer) were used. Samples were delivered to the furnace using a Perkin-Elmer AS-70 and stored in acid washed polypropylene cups prior to injection. The light sources were Perkin-Elmer EDL lamps operating at 480 ma for Pb and 30 ma for Cd using the respective resonance lines at 83.3 and 8.8 nm, with a nm spectral bandwidth. The inert gas was argon. Dilutions were carried out with calibrated Gilson Pipetman pneumatic syringes. Typical operating conditions were as follows: 10 ml of sample solution was introduced into the furnace with 10 ml HF solution and heated according to the electrothermal program presented in Table 1. After a drying step at approximately 130 8C, followed by a cooling step at 0 8C for 10 s (1 s ramp) under 50 mlymin argon flow, no pyrolysis was used and the atomization occurred in the max-power mode at the selected atomization temperature with the gas flow interrupted. A final cleaning step with a slow heating rate to 500 8C was used to gently remove the remaining saline matrix after the atomization step. A 5 s (70 points) baseline offset compensation (boc) time was used. One read measurement was performed and the signal was recorded during the atomization step; the atomization time being chosen longer than the duration of the atomic signal. The data recorded on a floppy disk were

3 J.Y. Cabon / Spectrochimica Acta Part B 57 (00) Table 1 Graphite atomizer program Step Drying Cooling Atomization Cleaning step Temp. (8C) Variable 500 Ramp time (s) Hold time (s) Variable 5 Read On Ar flow rate (ml min ) then converted to ASCII format with the peak data reformat Perkin-Elmer software, and then transferred to Microsoft Excel 5 for mathematical post-treatments of the signal: smoothing; determination of the integrated absorbance by determining the beginning and the end of the integration (3s threshold), or by using a moving sum of a fixed number of points; determination of the weighted mean background absorbance; summation of signals, etc... Ion chromatographic study The removal of chloride was followed by analyzing by ion chromatography the residue left on the platform after the drying step. Ten microliters of salt solution were pipetted onto the platform together with 10 ml of HF solution, and then dried. The residue remaining on the platform was dissolved by adding 30 ml of ultrapure water with the use of the autosampler. The resulting solution was then pipetted from the platform with a pneumatic syringe. The operation was repeated three times. The pipetted solutions were diluted to a final volume of 4 ml with ultrapure water. A Dionex DX-100 was used for ion chromatograph measurements. The ion chromatograph was equipped with an AS9-SC column and an ASPRS- II anion self-regenerating suppressor; the eluent being 1.8 mm NaCO3.7 mm NaHCO 3. The concentration of chloride was determined by calibration of the instrument with a chloride standard solution (1 g l Merck)..3. Reagents The standard solutions were prepared by dilution from 1 g l Pb or Cd in 0.5 M HNO Merck 3 standard. Hydrofluoric acid 40% was suprapur grade Merck. Magnesium chloride was pro analysi grade Merck. Seawater was a Mediterranean sample. National Research Council of Canada certified CASS- and SLEW-1 reference materials were analyzed for Cd. Ultrapure water from a Millipore milliro-mq system was used. 3. Results and discussion 3.1. Removal of chloride with the use of HF Seawater is a complex medium and the major elements precipitate at the drying step as various salts (NaCl, MgCl, MgSO 4, CaSO 4, etc.); chloride species representing the major part of the saline matrix. For the determination of trace metals in seawater, different matrix modifiers have been previously used in order to transform the chloride matrix to a less interfering matrix and, depending on the analytical line eventually to a less absorbing matrix. Nitric acid, oxalic acid and ammonium nitrate have similar efficiencies to remove chloride as HCl at the drying or at the pretreatment step. With the use of these chemical modifiers, chloride could be mainly removed out of the furnace at the drying step as HCl for modifierychloride mole concentration ratio of two w17,18x. If total suppression of chloride could not be obtained w19x, a drastic reduction of chloride interference effects and of the background absorption magnitude were generally observed. As shown in Fig. 1, HF could similarly be used to remove seawater chloride out of the furnace at the drying step. However, HF appeared much less efficient than oxalic or nitric acid and much higher acid concentrations had to be used. For a HFy chloride mole concentration ratio higher than 30,

4 516 J.Y. Cabon / Spectrochimica Acta Part B 57 (00) Fig. 1. Relative amount of chloride (%) remaining in the atomizer after addition of HF and drying in 10 ml seawater or 10 ml 0.05 y M MgCl. whfxywcl x represents the molar concentration ratio of the mixed solution. the major part of the seawater chloride was removed out of the furnace at the drying step. However, it could be observed that HF was much more efficient to remove chloride at the drying step from a MgCl solution than from a seawater solution. Indeed, chloride could be efficiently removed out of the furnace for a HFychloride mole concentration ratio of approximately from a MgCl solution. This difference could be attrib- utable to the different solubilities of NaF and MgF in water, respectively, of 4. and % myv w0x. Consequently, Mg was precipitated as MgF at the drying step and chloride was more efficiently removed out of the atomizer as HCl. Consequently, by using HF as a chemical matrix modifier for seawater, Mg is mainly precipitated at the drying step in the atomizer as MgF and MgSO4 salts, instead of MgCl and MgSO4 salts. Therefore, the atomization mechanisms of elements will be subsequently modified, particularly the interference effects induced by the presence of MgCl. 3.. Effect of HF on atomic absorption signals Atomic absorption signals of Pb and Cd and corresponding background signals are shown in Figs. and 3. Pb and Cd signals are delayed in seawater as compared to water and the atomization occurred at the beginning of the background signal, mainly generated by sodium chloride vaporization. For a 10-ml seawater, as also previously observed w3x, a strong chloride interference effect (;90%) was also noted in the case of Pb; a smaller interference effect was noted in the case of Cd (;15%). The interference effect was much more important in seawater than when NaCl was vaporized alone w1x. This could be attributable to the presence of MgCl in seawater. The hydrolysis of MgCl to MgOHCl and MgO at low temperatures has been previously observed w1,x. These oxides could delay the Cd and Pb signals that were consequently vaporized simultaneously to HCl resulting from the decomposition of MgCl or MgOHCl and a higher mass of NaCl, particularly in the case of Pb causing an important chemical interference. Moreover, a simultaneous background absorption signal was observed, more important in the case of cadmium (8.8 nm) than in the case of Pb (83.3 nm), according to the molecular spectrum of NaCl w3x. This background absorption induced a spectral interference effect that was more important at the Cd analytical line

5 J.Y. Cabon / Spectrochimica Acta Part B 57 (00) Fig.. Atomic absorption signal of Cd in 10 ml seawater (left) and in 10 ml seawaterq10 ml 3MHF(right); without spike (a) and with a ng Cd spike (b). T s150 8C. atom Fig. 3. Atomization signal of Pb in 10 ml seawater (left) and in 10 ml seawaterq10 ml 3MHF(right), without spike (a) and with a 100 ng Pb spike (b). T s150 8C. atom

6 518 J.Y. Cabon / Spectrochimica Acta Part B 57 (00) Fig. 4. Influence of addition of HF on the recovery of Pb and Cd in 10 ml seawaterq10 ml HF solution. Tatoms150 8C. y whfxywcl x represents molar concentration ratio of the mixed solution. than at the Pb analytical line; it was dependent on the background absorption magnitude and probably mainly due to the molecular absorption of NaCl w19x. This spectral interference effect induced an important under-correction susceptible of introducing systematic errors on the determination of these elements at low concentration levels. By addition of HF to seawater, chloride was removed at the drying step and for the different added HF concentration, the resulting seawater matrix was modified according to Fig. 1. Consequently, the chloride interference effect observed for Cd and, particularly for Pb decreased when the concentration of HF was increased (Fig. 4). The chloride interference effect was totally suppressed for small amounts of added HF in the case of Cd, but for much higher added HF concentrations in the case of Pb; indeed, a HFychloride mole concentration ratio higher than 10 was necessary to suppress the chloride interference effect observed in the case of Pb. As shown in Figs. and 3, the presence of HF promoted lower atomization temperature for Pb and Cd in seawater. The addition of HF to seawater led to the precipitation of MgF and prevented the hydrolysis of MgCl. As observed when seawater was mixed with HF at room temperature, Pb was co-precipitated from the solution with MgF, indicative of the formation of Pb fluoride compounds. CdF was probably not formed when HF was added to seawater at room temperature because it was not similarly co-precipitated from the solution with MgF. Nevertheless, it could be also formed when chloride was removed at the drying step and co-precipitated or entrapped as CdCl oryand CdF in the modified seawater matrix. This probably led to different atomization mechanisms for Pb and Cd. The trapping effect of MgF (Tmps161 8C, and Tbps 39 8C) was probably less important than the trapping effect of magnesium oxides (Tmps85 8C, and Tbps3600 8C) and consequently, for Cd and Pb a lower atomization temperature that led to the suppression of chloride interference effects was observed. For a high HFychloride mole concentration ratio of approximately 40, well resolved atomic absorption signals and a drastic decrease of the background absorption signal were obtained at the 8.8 nm Cd and the 83.3 nm Pb analytical

7 J.Y. Cabon / Spectrochimica Acta Part B 57 (00) lines, when heated up to 150 8C. The lower background absorption signal was attributable both to the decrease of the absorbance when NaCl was modified to NaF according to their molecular spectra wx, and to the high vaporization temperature of sodium fluoride; the decrease of the background absorption signal being accompanied by a reduction of the error Optimization studies The standard procedure for the determination of the detection limit in GFAAS recommended by IUPAC is time-consuming to optimize experimental conditions, because it requires 0 repeated measurements of a small absorption signal. In practice, for optimization studies in GFAAS, the sensitivity variation has only been monitored, but this parameter is not adequate to optimize experimental conditions because noise limitations are not taken into account, particularly in the presence of a background signal. An interesting tool for the optimization of experimental conditions has been proposed in previous papers, that permits a quantification of the limit of detection with the use of an a priori calculation w15x. The mass detection limit for our Perkin-Elmer apparatus could be expressed by the following relation w16x: Ž. m yey36 ms0.019=10 = QA ÃBG 10 1 = y q (1) t t int boc where QA is the integrated absorbance; m is the mass of analyte; tint is the integration time; tboc is the baseline offset compensation time; the weighted mean background absorption signal A BG is determined using the expression: Ã s B C DABG BG Ž max. qa Q BG E F BGG where Ā BG s t int The baseline noise is dependent on the light flux intensity and related to the E parameter that is displayed on Perkin-Elmer spectrometers and m L Ž 1000yV PM. defined as Es, where VPM is the 10 photomultiplier voltage. The concentration detection limit is expressed as c L s where V is the V sample volume introduced into the furnace. In our experiment, 10 ml of the modifier solution were added into the graphite furnace together with 10 ml of seawater spiked with Pb or Cd. Using fixed spectroscopic parameters (EDL lamps, recommended lamp currents and a nm slit width), i.e. fixed E parameter, no pyrolysis step and constant boc time, the only modifiable atomization parameters were only modifier concentration and atomization temperature. The integrated absorbance, integration time, and simultaneous background absorption magnitude were dependent both on the concentration of HF used, and on the atomization temperature. Therefore, we examined the variation of the limit of detection with the atomization temperature for Cd and Pb for three levels of concentrations: no HF; 5 M HF; and 3 M HF. For the HF concentrations used, the interference effect was minimized (Fig. 4), but different simultaneous background signal magnitudes were obtained depending on the HF concentration and atomization temperature. For this optimization study, the corresponding integration time was calculated after determining the beginning and the end of integration of a 11-point smoothed signal by examining the absorbance values higher than three times the standard deviation of the baseline noise. Q and A A BG were calculated in this integration window at each atomization temperature, and for the three HF concentration levels. The limit of detection was then calculated from Eq. (1) Cadmium The variations of the integrated absorbance of Cd with the atomization temperature are represented in Fig. 5. In unmodified seawater, an increase of the integrated absorbance was noted for atomization temperatures up to C by increasing atomization efficiency. However, due to the dramatic increase of the simultaneous background absorption signal, atomization temperatures higher than C could not practically be used

8 50 J.Y. Cabon / Spectrochimica Acta Part B 57 (00) Fig. 5. Influence of atomization temperature on the integrated absorbance of Cd in seawater, in 10 ml seawaterq10 ml 5 M HF, in 10 ml seawaterq10 ml 3 M HF(left) and on the mass detection limit in the corresponding media (right). in unmodified seawater. In the presence of HF, a higher atomization efficiency of Cd was obtained at low temperatures. It could be noted the presence of a plateau for higher atomization temperatures instead the expected decrease according to the variation diffusion coefficient of Cd w4x. This might be partly due to the non-isothermal conditions of the atomization. In Fig. 5 are represented the variations of the limit of detection with the atomization temperature. In seawater, the detection limit was improved when the atomization temperature was increased from 1150 to C; the improvement due the increase in atomization efficiency and reduction of the integration time (9 3.5 s), was slightly reduced by the increase of the weighted mean background absorption signal ( ). The dramatic increase of the limit of detection for atomization temperatures higher than C was mainly due to the strong increase of the simultaneous background absorption signal. The presence of HF greatly lowered the limit of detection of Cd in seawater. For a 5 or 3 M HF concentration, the decrease of the limit of detection from 1150 to C was mainly due to the reduction of the integration time (;5 s). For higher atomization temperatures, the increase of the limit of detection in the presence of 5 M HF was mainly due to the increase of the simultaneous background absorption signal, generated by the remaining chloride matrix. For a 3 M HF concentration, the increase of the simultaneous background absorption signal was not important and, consequently, the increase of the atomization temperature had no important influence on the detection limit Lead The variations in integrated absorbance of Pb with atomization temperature are represented in Fig. 6. It clearly appeared that the integrated absorbance was highly dependent on both HF concentration and atomization temperature. In unmodified seawater, an increase of the atomization temperature from 150 to C, improved only slightly the atomization efficiency. For a higher atomization temperature, the background

9 J.Y. Cabon / Spectrochimica Acta Part B 57 (00) Fig. 6. Influence of the atomization temperature on the integrated absorbance of Pb in seawater, in 10 ml seawaterq10 ml 5 M HF, in 10 ml seawaterq10 ml 3 M HF(left) and on the mass detection limit in the corresponding media (right). absorption signal was very important and atomization temperatures higher than C could not be used. The presence of a high HF concentration suppressed the chloride interference, and promoted a much higher atomization efficiency at low atomization temperatures. On the other hand, in this medium an increase in atomization temperature induced an important decrease of the integrated absorbance; this variation appeared in relatively good agreement with the variation in diffusion coefficient of Pb with temperature w4x. From Fig. 6, it appeared clearly that the use of HF greatly improves the detection limit of Pb in seawater. In seawater, an important decrease of the limit of detection from 150 to C was observed. The improvement was mainly due both to the increase of the integrated absorbance and the reduction of the integration time (3.5 s). In the presence of 5 M HF, an increase in atomization temperature from 1150 to C led to a higher atomization efficiency and to a shorter integration time (;11 4 s); this explained the important improvement of the detection limit from 1150 to C atomization temperature. For higher atomization temperatures, variations of the integration time were small, and both the decrease of the integrated absorbance and the increase in simultaneous background absorption signal generated by the vaporization of the remaining sodium chloride, led to an increase in the limit of detection. In the presence of 3 M HF, the interference effect was minimized and, both a high atomization efficiency and a very small background absorption signal were obtained at low atomization temperatures. This led to a much lower detection limit. It might be noted that the limit of detection was relatively constant in the atomization temperature range, despite the important decrease of the integrated absorbance with atomization temperature. This was mainly due the reduction in integration time (from ;6 s at C to;1.5 s at C) and the decrease in integrated absorbance, which had opposite effects on the limit of detection. In this medium, the variations of the simultaneous background absorption signal had a relatively small influence on the detection limit.

10 5 J.Y. Cabon / Spectrochimica Acta Part B 57 (00) Fig. 7. Single atomic signal of 100 ng l Cd (a), and averaged atomic signal of 0 replicates of 0 ng l Cd (b) in 15 ml seawaterq15 ml 3 M HF. Single atomic signal of 4 mg l Pb and averaged atomic signal of 0 replicates of 500 ng l Pb (b) in 15 ml seawaterq15 ml 3MHF.Tatoms1300 8C Determination of Cd and Pb in seawater in optimized conditions. From this study, it appeared that the best limits of detection for Cd and Pb were obtained with the use of a 3 M HF concentration. In these conditions, the seawater volume that could be introduced in the atomizer with a single introduction was reduced because an equal volume of modifier had to be introduced and consequently, the limit of detection was increased by approximately twofold. With the use of the Perkin-Elmer spectrometer, a total volume of 30 ml can be routinely introduced that limits the seawater volume to 15 ml. Single atomic signals of 4 mg l Pb and 0.1 mg l Cd, obtained in optimized conditions at C in15mlseawaterq15 ml 3MHF,are shown in Fig. 7. As observed, the atomization of Cd and Pb occurred in the presence of a very low simultaneous background absorption signal. However, it was noted that a relatively small undercorrection appeared simultaneously to the vaporization of the modified seawater matrix, particularly in the case of Cd. This under-correction could represent a significant systematic error for low absorption signals when the integration window was not carefully chosen. The use of a moving integration window (fixed number of integration points) close to the duration of the signal may significantly improve the detection limit, because the beginning and end of integration are difficult to determine for small absorption signals w16x. Using this data treatment, the experimental detection limits of Cd and Pb were determined by using a moving window corresponding to the duration of the respective atomic signals (80 points for Cd; 130 points for Pb). The detection limits obtained for Cd and Pb were and 0.5 mg l (3s, 0 measurements), respectively. In Fig. 7 are also shown atomic signals of Cd and Pb at the detection limit concentration level obtained by averaging 0 individual signals. The concentrations of Cd determined by the standard addition method in a SLEW- 1 estuarine water and a CASS- seawater were 0.00"0.00 and 0.017"0.00 mg l Cd, respectively, in good agreement with the 0.018"0.003 and 0.019"0.004 mg l Cd certi- fied values (At the 95% confident level, 10 meas-

11 J.Y. Cabon / Spectrochimica Acta Part B 57 (00) urements). The detection limit of Pb was too high for its determination in these certified samples at the 0.00 mg l level concentration. 4. Conclusion From this study, it appears clearly that HF may be an interesting modifier for the determination of volatile elements, like Pb and Cd, in seawater and more generally in chloride matrices. Indeed, it promotes an efficient atomization at low temperatures, suppresses chloride interference effects and dramatically reduces the simultaneous background absorption signal. Moreover, because of the absence of decomposition products, no pyrolysis step is necessary. The limit of detection obtained for Cd permits the direct determination of this element in seawater; however, the limit of detection of Pb in seawater is too high to permit the determination of this element in non-polluted seawater. The a priori calculation of the detection limit appears also very useful to optimize experimental conditions; however, systematic errors like miscorrections, that may induce important errors at the limit of detection level, are not included in the calculation and have to be minimized by a careful choice of experimental conditions, and an adequate mathematical data treatment. References w1x G. Schlemmer, B. Welz, Palladium and magnesium nitrates, a more universal modifier for graphite furnace atomic absorption spectrometry, Spectrochim. Acta Part B41(1986) wx P. Bermejo-Barrera, J. Moreda-Pineiro, A. Moreda- Pineiro, A. Bermejo-Barrera, Usefulness of the chemical modification and the multi-injection technique approaches in the electrothermal atomic absorption spectrometric determination of silver, arsenic, cadmium, chromium, mercury, nickel and lead in sea-water, J. Anal. At. Spectrom. 13 (1998) w3x J.Y. Cabon, A. Le Bihan, Determination of lead in seawater by electrothermal atomic absorption spectrometry with transversely heated furnace by using oxalic acid or PdyMg as modifiers, Spectrochim. Acta Part B 51 (1996) w4x E. Alvarez-Cabal Cimadevilla, K. Wrobel, A. Sanz- Medel, Capabilities and limitations of different techniques in electrothermal atomic absorption spectrometry for direct monitoring of arsenic, cadmium and lead contamination of sea-water, J. Anal. At. Spectrom. 10 (1995) w5x R. Guevremont, Organic matrix modifiers for direct graphite furnace atomic absorption determination of cadmium in seawater, Analyst 5 (1980) w6x R.E. Sturgeon, S.S. Berman, J.A.H. Desaulniers, A.P. Mykitiuk, J.W. McLaren, D.S. Russell, Comparison of methods for the determination of trace elements in seawater, Anal. Chem. 5 (1980) w7x M. Hoenig, R. Wollast, Les possibilites et limitations de l atomisation electrothermique en spectrometrie d absorption atomique lors de l analyse directe des metaux lourds dans l eau de mer, Spectrochim. Acta Part B 37 (198) w8x J.Y. Cabon, A. Le Bihan, Direct determination of cadmium in seawater using electrothermal atomization atomic absorption spectrometry with Zeeman-effect background correction and oxalic acid as a chemical modifier, J. Anal. At. Spectrom. 7 (199) w9x C.R. Lan, Direct determination of cadmium in seawater by electrothermal atomic absorption spectrometry with sodium hydroxide as a chemical modifier, Analyst 116 (1993) w10x E. Pruskowska, G.R. Carnrick, W. Slavin, Direct determination of Cd in coastal seawater by atomic absorption spectrometry with the stabilized temperature platform furnace and Zeeman background correction, Anal. Chem. 55 (1983) w11x H. Chuang, S.D. Huang, Direct determination of cadmium in seawater with a graphite atomic absorption spectrometer, Spectrochim. Acta Part B 49 (1994) w1x M.-S. Chan, S.-D. Huang, Direct determination of cadmium and copper in seawater using a transversely heated graphite furnace atomic absorption spectrometer with Zeeman-effect background corrector, Talanta 51 (000) w13x I. Lopez-Garcia, M. Sanchez-Merlos, M. Hernandez- Cordoba, Slurry sampling for the determination of lead, cadmium and thallium in soils and sediments by electrothermal atomic absorption spectrometry with fastheating programs, Anal. Chim. Acta 38 (1996) w14x I. Lopez-Garcia, M. Sanchez-Merlos, M. Hernandez- Cordoba, Use of hydrofluoric acid to decrease the background signal caused by sodium chloride in electrothermal atomic absorption spectrometry, Anal. Chim. Acta 396 (1999) w15x B. L vov, L.K. Polzik, A.V. Borodin, A.O. Dyakov, A.V. Novichikin, Detection limits in Zeeman-effect electrothermal atomic absorption spectrometry, J. Anal. At. Spectrom. 10 (1995) w16x J.Y. Cabon, A. Le Bihan, Influence of experimental parameters in electrothermal atomic absorption spectrometry on a priori calculated instrumental detection limits, Analyst 1 (1997) w17x M.M. Chaudhry, D. Littlejohn, Ion chromatographic study of the effect of ammonium nitrate as a modifier in electrothermal atomic absorption spectrometry, Analyst 117 (199)

12 54 J.Y. Cabon / Spectrochimica Acta Part B 57 (00) w18x J.Y. Cabon, A. Le Bihan, The determination of Cr, Cu and Mn in seawater with transversely heated graphite furnace atomic absorption spectrometry, Spectrochim. Acta Part B 50 (1995) w19x G. Daminelli, D.A. Katskov, P.J.J.G. Marais, P. Tittarelli, Characterization of the vapor-phase molecular and atomic absorption from sea water matrices in electrothermal atomic absorption spectrometry, Spectrochim. Acta Part B53(1998) w0x D.R. Lide (Ed.), CRC Handbook of Chemistry and Physics, 7nd ed., CRC Press, Boca Raton, 199. w1x J.Y. Cabon, A. Le Bihan, Interference of salts on the determination of lead by electrothermal atomic absorption spectrometry, Spectrochim. Acta Part B 51 (1996) wx T. Kantor, On the mechanisms of organic acid modifiers used to eleminate magnesium chloride interferences in graphite furnace atomic absorption spectrometry, Spectrochim. Acta Part B 50 (1995) w3x B.R. Culver, T. Surles, Interference of molecular spectra due to alkali halides in non-flame atomic absorption spectrometry, Anal. Chem. 47 (1975) w4x M. Berglund, D.C. Baxter, Computer program for calculating theoretical characteristic mass values in electrothermal atomic absorption spectrometry, J. Anal. At. Spectrom. 7 (199)