Speciation of chromium in mineral waters and salinas by solid-phase extraction and graphite furnace atomic absorption spectrometry

A simple GF-AAS method for speciation analysis of chromium in mineral waters and salinas was developed. Cr(VI) species were separated from Cr(III) by solid-phase extraction with APDC (ammonium pyrrolidinedithiocarbamate). The APDC complexes were formed in the sample solution under proper conditions, adsorbed on Diaion HP-2MG resin and the resin was separated from the sample. After elution with concentrated nitric acid Cr(VI) was determined by GF-AAS. Total chromium was determined by GF-AAS directly in the sample and Cr(III) concentration was calculated as the difference between those results.

The detection limit of the method defined as 3 s of background variation was 0.03 μg l⁻¹ for Cr(VI) and 0.3 μg l⁻¹ for total chromium. RSD for Cr(VI) determination at the concentration of 0.14 μg l⁻¹ was 9%, and for total chromium at the concentration of 5.6 μg l⁻¹ was 5%. The recovery of Cr(VI) was in the range of 94–100%, dependently on type of the sample.

The investigation of recovery of the spiked Cr(VI) showed that at concentration levels near 1 μg l⁻¹ and lower recovery may be reduced significantly even by pure reagents that seem to be free from any reductants.     

Interference from arsenate when determining phosphate by the malachite green spectrophotometric method

The effect of arsenate on phosphate determination by the malachite green spectrophotometric method was investigated. The molar absorptivities of the molybdophosphate and malachite green–molybdoarsenate species at 625 nm and a final acidity of 0.38 M were calculated as 10.4±0.13×10⁴ and 7.2±0.17×10⁴ l mol⁻¹ cm⁻¹ respectively, indicating that arsenate could interfere in phosphate measurement. Arsenate concentrations as low as 23 μg l⁻¹ caused increase in colour development in phosphate solutions. However, the extent of colour development for both anions depended on the final acid concentration of the solution. An acidified sodium sulphite solution (0.83 M NaSO₃, 0.83 M H₂SO₄) quantitatively prevented arsenate colour development up to 300 μg l⁻¹ As(V). It was also demonstrated that the method removed As(V) interferences in mixed As/P solutions and therefore can be used to treat natural water samples with elevated arsenate concentrations before phosphate measurement.     

Biosorption of heavy metals on Aspergillus fumigatus immobilized Diaion HP-2MG resin for their atomic absorption spectrometric determinations

A solid phase extraction procedure based on biosorption of copper(II), lead(II), zinc(II), iron(III), nickel(II) and cobalt(II) ions on Aspergillus fumigatus immobilized Diaion HP-2MG has been investigated. The analytical conditions including amounts of A. fumigatus, eluent type, flow rates of sample and eluent solutions were examined. Good recoveries were obtained to the spiked natural waters. The influences of the concomitant ions on the retentions of the analytes were also examined. The detection limits (3sigma, N = 11) were 0.30 μg l⁻¹ for copper, 0.32 μg l⁻¹ for iron, 0.41 μg l⁻¹ for zinc, 0.52 μg l⁻¹ for lead, 0.59 μg l⁻¹ for nickel and 0.72 μg l⁻¹ for cobalt. The relative standard deviations of the procedure were below 7%. The validation of the presented procedure is performed by the analysis of three standard reference materials (NRCC-SLRS 4 Riverine Water, SRM 1515 Apple leaves and GBW 07605 Tea). The procedure was successfully applied for the determination of analyte ions in natural waters microwave digested samples including street dust, tomato paste, black tea, etc.     

Differential-pulse voltammetric determination of low μg l cyanide levels using EDTA, Cu(II) and a hanging mercury drop electrode

The proposed method for cyanide determination at the ultratrace level by differential pulse voltammetry is based in the sensitivity enhancement obtained when both Cu(II) and EDTA are present in the background electrolyte. Comparison of the detection limits and linear dynamic ranges using the conventional borate (pH 9.75), and the proposed borate-EDTA–Cu(II) background electrolytes was carried out. Best results have been obtained with the addition of 0.5 mmol l⁻¹ EDTA and 0.02 mmol l⁻¹ of Cu(II), which allow a detection limit of 1.7 μg l⁻¹ CN⁻ (65 nmol l⁻¹ — absolute detection limit 34 ng) with a precision better than ±2% for a 40 μg l⁻¹ level. Calibration range extended from detection limit up to 100 μg l⁻¹. Cyclic voltammetry indicates that the measured cyanide peak is obtained when the electrogenerated CuCN adsorbed onto the hanging mercury drop electrode surface, is oxidised at positive going potential scan. The method has been successfully applied to various industrial waste waters such as metal-finishing waste waters, water/sand mixtures from cleaning processes of coke production, leachates from wastes obtained from electrolytic cells of aluminium production, and liquors from gold extraction industry. Results obtained by the proposed method showed good agreement with those obtained by the standard methods (ion-selective potentiometry and the spectrophotometric pyridine method).     

Continuous flow system for lead determination by faas in spirituous beverages with solid phase extraction and on-line copper removal

A continuous flow system for the determination of lead in home made spirituous beverages was developed. The determination was based on the formation of a neutral chelate of the element with ammonium pyrrolidine dithiocarbamate, its adsorption onto a minicolumn packed with sodium faujasite type Y synthetic zeolite, followed by elution with methyl isobutyl ketone and determination by flame atomic absorption spectrometry. Ethanol and copper interfere strongly in the determination and therefore, must be separated prior to the analysis. Copper is removed by precipitation with rubeanic acid, while ethanol is eliminated by rotaevaporation. Sample solutions containing Pb²⁺ in the concentration range from 5 to 120 μg l⁻¹ at pH 2.5 could be analyzed, by using a preconcentration time of 3 min. Preconcentration factors from 80 to 140 were achieved for a sample volume of 6 ml and the detection limit varied from 1.4 to 3.5 μg l⁻¹, depending on the matrix composition. The relative standard deviations for 60 μg l⁻¹ Pb was 3.2% (n = 10) and the recovery of spikes (20, 40, 60 and 80 μg l⁻¹) added to the samples was estimated within 92–105% range, suggesting that lead can be quantitatively determined in such samples. Determining lead in several samples by an alternative technique further checked the accuracy. Finally, the concentrations of Pb²⁺ determined in 28 samples of Venezuelan spirituous beverages were in 12.6–370.0 μg l⁻¹ range, depending on the fermenting material based on different mixtures of agave, raw sugar cane and white sugar.     

Determination of nanomolar concentrations of phosphate in freshwaters using flow injection with luminol chemiluminescence detection

A simple and rapid flow injection (FI) method is reported for the determination of phosphate (as molybdate reactive P) in freshwaters based on luminol chemiluminescence (CL) detection. The molybdophosphoric heteropoly acid formed by phosphate and ammonium molybdate in acidic conditions generated chemiluminescence emission via the oxidation of luminol. The detection limit (3× standard deviation of blank) was 0.03 μg P l⁻¹ (1.0 nM), with a sample throughput of 180 h⁻¹. The calibration graph was linear over the range 0.032–3.26 μg P l⁻¹ (r²=0.9880) with relative standard deviations (n=4) in the range 1.2–4.7%. Interfering cations (Ca(II), Mg(II), Ni(II), Zn(II), Cu(II), Co(II), Fe(II) and Fe(III)) were removed by passing the sample through an in-line iminodiacetate chelating column. Silicate interference (at 5 mg Si l⁻¹) was effectively masked by the addition of tartaric acid and other common anions (Cl⁻, SO₄²⁻, HCO₃⁻, NO₃⁻ and NO₂⁻) did not interfere at their maximum admissible concentrations in freshwaters. The method was applied to freshwater samples and the results (26.1±1.1–62.0±0.4 μg P l⁻¹) were not significantly different (P=0.05) from results obtained using a segmented flow analyser method with spectrophotometric detection (24.4±4.45–84.0±16.0 μg P l⁻¹).     

Analysis of acrylamide in coffee and dietary exposure to acrylamide from coffee

An analytical method for analysing acrylamide in coffee was validated. The analysis of prepared coffee includes a comprehensive clean-up using multimode solid-phase extraction (SPE) by automatic SPE equipment and detection by liquid chromatography tandem mass spectrometry using electrospray in the positive mode. The recoveries of acrylamide in ready-to-drink coffee spiked with 5 and 10 μg l⁻¹ were 96±14% and 100±8%, respectively. Within laboratory reproducibility for the same spiking levels were 14% and 9%, respectively. Coffee samples (n = 25) prepared twice by coffee machines and twice by a French Press Cafetière coffee maker contained 8±3 μg l⁻¹ and 9±3 μg l⁻¹ acrylamide. Five ready-to-drink instant coffee prepared twice contained 8±2 μg l⁻¹. Hence, the results do not show significant differences in the acrylamide contents in ready-to-drink coffee prepared by coffee machine, French Press or from instant coffee. Medium roasted coffee contained more acrylamide (10 μg l⁻¹) than dark roasted coffee (5 μg l⁻¹). Males aged 35–45 years, drinking on average 1.1 l coffee per day are exposed to the highest doses of acrylamide from coffee. The dietary intake of acrylamide from coffee comprises, on an average, 10 μg day⁻¹ for males and 9 μg day⁻¹ for females aged 35–45 years. Probabilistic modelling of the exposure of Danish consumers (all adults) to acrylamide from coffee shows a mean exposure of 6.5 μg day⁻¹ and a 95 percentile of 18 μg day⁻¹.     

Detection of hidden hazelnut protein in food by IgY-based indirect competitive enzyme-immunoassay

The development of an indirect competitive enzyme-immunoassay for the detection of hidden hazelnut protein in complex food matrices is described. A sensitive and selective polyclonal antibody was raised by immunisation of laying hens with protein extracts from roasted hazelnuts. In contrast to traditional antibody generation in mammals, the antibody was not isolated from the blood of immunised mammals but from the egg yolk of immunised chickens. A standard calibration curve was optimised using immunoaffinity purified antibody extract and a coating antigen concentration of 10 μg ml⁻¹. One percent skim milk powder was chosen for blocking. The assay has a minimum detection limit of 10 μg l⁻¹, with an IC₅₀ of 618 μg l⁻¹ when a 50 mM phosphate buffer at pH 7.5 and 10 mM sodium chloride is used as assay buffer. The cross reactivity testing shows a high specificity for hazelnut proteins and various foods and food additives were found to be non reactive except beans, sunflower seed or poppy seed.     

Ion chromatographic determination of lithium at trace level concentrations: Application to a tracer experiment in a high-mountain lake

The water residence time of a high-mountain seepage lake in the Austrian Alps was derived from the flushing rate of a tracer substance. A diluted lithium chloride solution was injected into the lake during holomictic conditions in order to favour the homogeneous distribution of the tracer. The exponential decline of the mass of lithium in the lake revealed a water residence time of 1.5 to 3 months for summer and almost no lake water exchange during winter. Lithium concentrations ranged from background values of 0.06 μg l⁻¹ to about 3 μg l⁻¹ immediately after the tracer injection. Lake water samples were analyzed with ion-exchange chromatography using a Dionex device with a CS 12A separation column. The method detection limit determined according to the definition of the US Envirinmental Protection Agency amounted to 0.009 μg l⁻¹.     

Field speciation of chromium with a sequential injection lab-on-valve incorporating a bismuthate immobilized micro-column

A fully automated and portable analyzer for field speciation of inorganic chromium in wastewater was developed. The instrument consists of a micro-sequential injection lab-on-valve (LOV) system and a miniature USB2000 spectrophotometer. A multi-purpose flow cell was incorporated on one side of the main body of the LOV, which offers vast potentials and versatilities in its compatibility with various detection modes. On-line oxidation of trivalent chromium was performed on a bismuthate immobilized silica micro-column reactor integrated in the LOV. When determining Cr(VI), its chromogenic reaction with 1,5-diphenylcarbazide (DPC) was facilitated in the flow cell and the absorbance was monitored in situ at 548 nm via optical fibers. While for the quantification of total chromium, Cr(III) was oxidized on-line by aspirating sample solution through the oxidizing column reactor, followed by chromogenic reaction with DPC and the absorbance was monitored in the flow cell. With a sampling volume of 200 μl, the detection limits of 5.6 μg l⁻¹ for Cr(VI) and 6.8 μg l⁻¹ for total chromium were achieved along with a sampling frequency of 60 h⁻¹. A R.S.D. value of 2.0% was recorded at 32 μg l⁻¹ of Cr(VI). The practical applicability of the speciation analyzer was validated by analyzing Cr(VI) and total chromium contents in two certified reference materials. The feasibility of performing rapid field speciation of chromium in wastewater samples was also demonstrated.     

Simultaneous determination of total arsenic and total selenium in Chinese medicinal herbs by hydride generation atomic fluorescence spectrometry in tartaric acid medium

By HG-AFS, a new method was proposed for simultaneous determination of total arsenic and total selenium existed in the Chinese medicinal herbs in tartaric acid medium. The effects of analytical conditions and coexisting ions on the fluorescence signal intensity of analytes were investigated. The proposed method was provided with linear response ranges above 22 μg l⁻¹ for As and 44 μg l⁻¹ for Se, and the detection limits of 0.13 and 0.12 μg l⁻¹ were obtained for As and Se respectively. The recoveries of 93.8–96.1% for As and 95.3–99.1% for Se, and the precision of 1.2–3.8% and 2.4–5.3% (R.S.D., n = 8) respectively, were obtained via simultaneous determined four samples of Chinese medicinal herbs and three certified botanic reference materials successfully. The proposed method has the advantages of simple operation, high sensitivity and high efficiency.     

Atomic absorption spectrophotometric determination of trace copper in waters, aluminium foil and tea samples after preconcentration with 1-nitroso-2-naphthol-3,6-disulfonic acid on Ambersorb 572

An atomic absorption spectrophotometric method for the determination of trace copper after adsorption of its 1-nitroso-2-naphthol-3,6-disulfonic acid chelate on Ambersorb 572 has been developed. This chelate is adsorbed on the adsorbent in the pH range 1–8. The copper chelate is eluted with 5 ml of 0.1 mol l⁻¹ potassium cyanide and determined by flame atomic absorption spectrometry (FAAS). The selectivity of the proposed procedure was also evaluated. Results show that iron(III), zinc(II), manganese(II) and cobalt(II) at the 50 μg l⁻¹ level and sodium(I), potassium(I), magnesium(II), calcium(II) and aluminium(III) at the 1000 μg l⁻¹ level did not interfere. A high enrichment factor, 200, was obtained. The detection limit (3σ) of copper was 0.34 μg l⁻¹. The precision of the method, evaluated by seven replicate analyses of solutions containing 5 μg of copper was satisfactory and the relative standard deviation was 1.7%. The adsorption of copper onto Ambersorb 572 can formally be described by a Langmuir equation with a maximum adsorption capacity of 14.3 mg g⁻¹ and a binding constant of 0.00444 l mg⁻¹. The accuracy of the method is confirmed by analysing tomatoes leaves (NIST 1573a) and lead base alloy (NBS 53e). The results demonstrated good agreement with the certified values. This procedure was applied to the determination of copper in waters (tap, river and thermal waters), aluminium foil and tea samples.     

Simultaneous determination of bromide and nitrate in contaminated waters by ion chromatography using amperometry and absorbance detectors

This study describes a new ion chromatography method using a low-capacity anion exchange column with amperometric and absorbance detection for rapid and simultaneous determination of Br⁻ and NO₃⁻ in contaminated waters where one of these ions is present in excess compared to other. The use of two detectors overcomes the problem of baseline separation for Br⁻ and NO₃⁻ for accurate quantification, which was commonly encountered when using a low-capacity anion exchange column and suppressed conductivity detection mode. The method achieved accurate quantification of these two ions without requirement of baseline separation. The accuracy of 2.8% for NO₃⁻ was determined using a quality control sample obtained from UN GEMS/Water PE Study No. 6. The detection limits for Br⁻ and NO₃⁻ were 20 and 6 μg l⁻¹ (25 μl sample), respectively. Linearity of these two ions was over three orders of magnitude with a correlation coefficient >0.998. The influence of potential interfering ions was also studied followed by the determination of Br⁻ and NO₃⁻ in seawater, unsaturated zone water, soil extract and groundwater.     

Application of diffusive gradient in thin films technique (DGT) to measurement of mercury in aquatic systems

The diffusive gradient in thin films (DGT) technique was investigated and used to measure mercury concentration in river water. Mercury ions are covalently bound to amide nitrogen groups of commonly used polyacrylamide, which makes this gel unsuitable as a diffusive medium. In contrast, agarose gel was found as the diffusive gel for mercury measurements. Basic performance tests of agarose DGT verified the applicability of Fick's first law for DGT measurements. Two selective resins, Chelex-100 with iminodiacetic groups and Spheron-Thiol with thiol groups were used. The measured diffusion coefficient in agarose gel was close to that in water. The concentration of mercury in Svitava river measured by DGT with Speron-Thiol resin gel was higher (0.0116 ± 0.0009 μg l⁻¹) than those obtained by Chelex-100 (0.0042 ± 0.0005 μg l⁻¹). Different capture efficiencies of two adsorbents enable to estimate fractions of mercury bonded in different complexes in the river water. The concentrations of mercury found by DGT both Chelex-100 and Speron-Thiol resin gels are much lower than that measured directly in the river water (0.088 ± 0.012 μg l⁻¹). This difference indicates that DGT concerns inorganic ions and labile species only, and that it is not able to include inert organic species and colloids.     

Determination of naringin in grapefruit juice by cathodic stripping differential pulse voltammetry at the hanging mercury drop electrode

A highly sensitive cathodic stripping voltammetric method for the determination of naringin is presented. It is based on the formation and accumulation of two naringin–mercury complexes at the electrode surface, followed by reduction of the surface species during a differential pulse voltammetric scan. The cathodic stripping responses at −0.25 V and −0.42 V, are evaluated with respect to various experimental conditions, such as composition and pH of the supporting electrolyte, naringin concentration, accumulation potential and preconcentration time. The new method is suitable for the determination of naringin concentrations between 0.1 mg l⁻¹ (1.72×10⁻⁷ mol l⁻¹) and 40 mg l⁻¹ (6.88×10⁻⁵ mol l⁻¹). A 3σ limit of detection of 32 μg l⁻¹ (55 nmol l⁻¹) can be reached. The relative standard deviation (r.s.d.) is <1.5%. Recovery experiments yielded a mean recovery of 97% (r.s.d.=4.1%). The application of the procedure to the selective determination of naringin in grapefruit juice is demonstrated.     

Comparison of chemical modifiers for selenium determination in soil aqua regia extracts by ZETAAS

The effect of various chemical modifiers 0.5 g l⁻¹ Pd, 1% (w/v) Ni, 0.5 g l⁻¹ Pd + 1% (w/v) Ni and 1.0 g l⁻¹ Pd on the measurement of selenium in soil aqua regia extracts, by ZETAAS, is described. Two Certified Reference Materials (CRMs) RTC-CRM 023-050 and RTC-CRM 025-050 were used for this study. Pyrolysis and atomization curves were obtained for each chemical modifier and their optimal values were assessed. By using standard addition as calibration method, accurate results were obtained for all the chemical modifiers studied. The precision was similar for both CRMs, with a maximum value of 7.5% R.S.D. The limits of detection and quantification for selenium in the soil extracts (n = 10) were 3.0 and 6.0 μg l⁻¹, respectively. The characteristic mass of selenium is assessed as 10 pg. The use of aqua regia as extractant gave quantitative results for selenium in the CRMs assayed.     

Spectrophotometric simultaneous determination of nitrite, nitrate and ammonium in soils by flow injection analysis

A new rapid flow injection procedure for the simultaneous determination of nitrate, nitrite and ammonium in single flow injection analysis system is proposed. The procedure combines on-line reduction of nitrate to nitrite and oxidation of ammonium to nitrite with spectrophotometric detection of nitrite by using the Griess-llosvay reaction. The formed azo dye was measured at 543 nm. The influence of reagent concentration and manifold parameters were studied. Nitrite, nitrate and ammonium can be determined within the range of 0.02–1.60 μg mL⁻¹, 0.02–1.60 μg mL⁻¹ and 0.05–1.40 μg mL⁻¹, respectively. R.S.D. values (n = 10) were 2.66; 1.41 and 3.58 for nitrate, nitrite and ammonium, respectively. This procedure allows the determination and speciation of inorganic nitrogen species in soils with a single injection in a simple way, and high sampling rate (18 h⁻¹). Detection limits of 0.013, 0.046 and 0.047 μg mL⁻¹were achieved for nitrate, nitrite and ammonium, respectively. In comparison with others methods, the proposed one is more simple, it uses as single chromogenic reagent less injection volume (250 mL in stead of 350 mL) and it has a higher sampling rate.     

Determination of lead in water samples by graphite furnace atomic absorption spectrometry after cloud point extraction

Cloud point extraction (CPE) has been used for the pre-concentration of lead, after the formation of a complex with 2-(5-bromo-2-pyridylazo)-5-(diethylamino)-phenol (5-Br-PADAP), and later analysis by graphite furnace atomic absorption spectrometry (GFAAS) using octylphenoxypolyethoxyethanol (TritonX-114) as surfactant. The chemical variables affecting the separation phase were optimized. Separation of the two phases was accomplished by centrifugation for 15 min at 4000 rpm. Under the optimum conditions i.e., pH 8.0, cloud point temperature 40 °C, [5-Br-PADAP] = 2.5 × 10⁻⁵ mol l⁻¹, [Triton X-114] = 0.05%, added methanol volume = 0.15 ml, pre-concentration of only 10 ml sample permitted an enhancement factor of 50-fold. The lower limit of detection (LOD) obtained under the optimal conditions was 0.08 μg l⁻¹. The precision for 10 replicate determinations at 5 μg l⁻¹ Pb was 2.8% relative standard deviation (R.S.D.). The calibration graph using the pre-concentration system for lead was linear with a correlation coefficient of 0.9984 at levels near the detection limits up to at least 30 μg l⁻¹. The method was successfully applied to the determination of lead in water samples.     

Flow injection determination of bismuth in urine by successive retention of Bi(III) and tetrahydroborate(III) on an anion-exchange resin and hydride generation atomic absorption spectrometry

Bismuth as BiCl₄⁻ and BH₄⁻ ware successively retained in a column (150 mm × 4 mm, length × i.d.) packed with Amberlite IRA-410 (strong anion-exchange resin). This was followed by passage of an injected slug of hydrochloric acid resulting in bismuthine generation (BiH₃). BiH₃ was stripped from the eluent solution by the addition of a nitrogen flow and the bulk phases were separated in a gas–liquid separator. Finally, bismutine was atomized in a quartz tube for the subsequent detection of bismuth by atomic absorption spectrometry. Different halide complexes of bismuth (namely, BiBr₄⁻, BiI₄⁻ and BiCl₄⁻) were tested for its pre-concentration, being the chloride complexes which produced the best results. Therefore, a concentration of 0.3 mol l⁻¹ of HCl was added to the samples and calibration solutions. A linear response was obtained between the detection limit (3σ) of 0.225 and 80 μg l⁻¹. The R.S.D.% (n = 10) for a solution containing 50 μg l⁻¹ of Bi was 0.85%. The tolerance of the system to interferences was evaluated by investigating the effect of the following ions: Cu²⁺, Co²⁺, Ni²⁺, Fe³⁺, Cd²⁺, Pb²⁺, Hg²⁺, Zn²⁺, and Mg²⁺. The most severe depression was caused by Hg²⁺, which at 60 mg l⁻¹ caused a 5% depression on the signal. For the other cations, concentrations between 1000 and 10,000 mg l⁻¹ could be tolerated. The system was applied to the determination of Bi in urine of patients under therapy with bismuth subcitrate. The recovery of spikes of 5 and 50 μg l⁻¹ of Bi added to the samples prior to digestion with HNO₃ and H₂O₂ was in satisfactory ranges from 95.0 to 101.0%. The concentrations of bismuth found in six selected samples using this procedure were in good agreement with those obtained by an alternative technique (ETAAS). Finally, the concentration of Bi determined in urine before and after 3 days of treatment were 1.94 ± 1.26 and 9.02 ± 5.82 μg l⁻¹, respectively.     

Determination of As(III) and total inorganic arsenic by flow injection hydride generation atomic absorption spectrometry

A simple procedure was developed for the direct determination of As(III) and As(V) in water samples by flow injection hydride generation atomic absorption spectrometry (FI–HG–AAS), without pre-reduction of As(V). The flow injection system was operated in the merging zones configuration, where sample and NaBH₄ are simultaneously injected into two carrier streams, HCl and H₂O, respectively. Sample and reagent injected volumes were of 250 μl and flow rate of 3.6 ml min⁻¹ for hydrochloric acid and de-ionised water. The NaBH₄ concentration was maintained at 0.1% (w/v), it would be possible to perform arsine selective generation from As(III) and on-line arsine generation with 3.0% (w/v) NaBH₄ to obtain total arsenic concentration. As(V) was calculated as the difference between total As and As(III). Both procedures were tolerant to potential interference. So, interference such as Fe(III), Cu(II), Ni(II), Sb(III), Sn(II) and Se(IV) could, at an As(III) level of 0.1 mg l⁻¹, be tolerated at a weight excess of 5000, 5000, 500, 100, 10 and 5 times, respectively. With the proposed procedure, detection limits of 0.3 ng ml⁻¹ for As(III) and 0.5 ng ml⁻¹ for As(V) were achieved. The relative standard deviations were of 2.3% for 0.1 mg l⁻¹ As(III) and 2.0% for 0.1 mg l⁻¹ As(V). A sampling rate of about 120 determinations per hour was achieved, requiring 30 ml of NaBH₄ and waste generation in order of 450 ml. The method was shown to be satisfactory for determination of traces arsenic in water samples. The assay of a certified drinking water sample was 81.7±1.7 μg l⁻¹ (certified value 80.0±0.5 μg l⁻¹).