Department of Chemistry and Biochemistry

 Texas Tech University

 Abstract

Extensive arsenic contamination of drinking water has been reported in many parts of the world. Trivalent As (As (III)) is regarded to be more toxic than pentavalent As (As(V)) and is slowly oxidized by dissolved oxygen to the latter. Collecting samples in the field and bringing them back to the laboratory is generally unsatisfactory for reliable analysis of As(III) and As(V), due to ongoing oxidative change of the sample. Even simple measurements of total As in the field is problematic due to the inadequate sensitivity of the available colorimetry based field kits. Anodic stripping analysis can be a good field technique to perform total and speciated measurements of dissolved inorganic As down to the one ìg/l level. We offer an inexpensive portable stripping voltammetric instrument with a simple procedure for the field analysis of As(III) and As(V).

Influences from foreign ions, organic compounds, pre-electrolysis potential and time and stripping medium have been investigated. The instrument is operated off a desktop or a notebook computer, equipped with an A/D-D/A card. The attainable limit of detection (LOD) is dependent on the deposition time, an LOD of 0.5 ìg/l As(III) or As(V) is obtained for a deposition time of 80 s.

Extensive arsenic contamination of drinking water has been reported in many parts of the world. Early reports of serious As contamination of groundwater and its relationship to skin cancer came from Taiwan 30 years ago [hftn ]. Similar reports have since come from many other regions, including the US []. However, until recently, arsenic issues in the US have not received much media attention []. Analysis of cancer risk from inorganic As in drinking water have been reported [,]; the US Environmental Protection Agency (USEPA) is presently considering new lowered standards for As in drinking water to be effective in this country by the turn of the century.

The World Health Organization recommends a maximum of 10 ìg/l As in drinking water. In Bangladesh, only 10 out of 65 districts have groundwater that contains As below this level. In some 45 districts, the arsenic concentrations are greater than 50 ìg/l; many samples in the several mg/l range have been reported. In general, in natural water samples, arsenic may be present as inorganic arsenite, arsenate, and monomethylarsonic acid (MMAA) and dimethylarsinic acid (DMAA), the toxicity is believed to decrease in that order [7]. In the groundwater samples from Bangladesh and India, organic As species are undetectable while Arsenite comprises often 50% or more of the total As [6-10].

Present USEPA approved methods for the determination of As in water are all atomic spectrometric methods: Inductively coupled plasma - atomic emission spectrometry (ICP-AES), ICP- mass spectrometry, platform graphite furnace - atomic absorption (PGF-AA), graphite furnace - AA and hydride generation – flame AA []. Notably, simple flame AA, the most affordable of the atomic spectrometric instruments, is incapable of determining As at the requisite levels without hydride generation. In addition, with atomic spectrometric detection, some form of front-end separation technique is needed if speciation is desired. More importantly, the economic outlay in capital cost (and consumables, especially for ICP) are such that it deters widespread application in developing countries.

In the affected areas in Bangladesh and India, when one well is shut down due to high As levels (often painted red for ready identification) and consumption increases at another well where As concentrations have previously been determined to be safe, the As levels in the water from the latter well often go up to unacceptable level within a few days []. Under the circumstances, routine widespread field-testing is obviously necessary and is being advocated by many organizations. The most widely used field kit is based on the Gutzheit test: Several hundred ml of a water sample is taken, rendered strongly acid with HCl and zinc dust added. Nascent hydrogen reduces all As to arsine, which is purged out by the evolving H₂. After passage through a plug soaked in lead acetate to remove H₂S, the liberated gas passes through a HgBr₂-impregnated filter that turns yellow upon exposure to arsine. As concentration is determined by comparison with a color chart. The method is in wide use although the manufacturer of the most popular kit suggests that the detection limit is 100 ìg/l. Even with modifications designed to improve the method, field studies show that it reads too low below 100 ìg/l and too high above 100 ìg/l As, compared to reference measurements by atomic spectrometry []. Spectrophotometric versions of the same method can be more sensitive and arsenic analysis in water is becoming a growing cottage industry in the region. Simple inexpensive light emitting diode-based spectrometers have been designed that can be easily taken to the field []. All of these overlooks the fact that not only the cost of the reagents (which must obviously be arsenic free) are not insignificant, it is the prudence of promulgating tests involving the use of other toxic, hazardous and caustic chemicals by unskilled or semi-skilled personnel for a method that can produce an extremely toxic gas, that is in question.

Arsenic standard solution (1000 mg/l, Alfa/Aesar) was used for total As measurements. A stock solution containing 1000 mg/l As(V) was made using Na₂HAsO₄ 7H₂O (Baker). A stock solution of 200 mg/l As(III) was made by dissolving As₂O₃ (Fisher) in 6 M HCl and stored at 4° C. It was stable for at least 1 month when checked against a freshly prepared standard solution. The required low concentration standards were prepared daily by dilution of the stock solution. A plating solution containing 50 ppm Au(III) was made by dissolving AuCl₃ (Alfa/Aesar) in 6 M HCl and diluted to attain a final concentration of 0.5 M HCl. 4.5 M HCl was used as the stripping medium.

Figure 1 shows the general instrumental arrangement. A desktop or notebook laptop personal computer (PC, both types were used in this study) equipped with an A/D-D/A card controls a set of four polytetrafluoroethylene (PTFE) solenoid valves (V1-V4) all of which when open, connects to a common port (Biochem Valve Corp., Hanover, NJ) and allows the selected liquid to flow through the flow cell FC to a waste container. The flow occurs under gravity, thus the liquid storage bottles and the waste containers are maintained in a relatively constant configuration. The electrical connections to the electrodes in FC are made through the P&CC interface, a simple, PC-controlled potentiostat and constant current source. Details of FC are shown in Figure 2. It is simply a 1/16 in. barbed miniature tee (Ark-Plas, Flippin, AR) with 1.4 mm i. d. polyvinyl chloride (PVC) pump tubing connected to two of the horizontal arms. Using a fine bore hypodermic needle, a platinum wire (50 ìm diameter, unless otherwise stated) is inserted through one of the horizontal arms and soldered to a more sturdy lead. The solder joint and the area of penetration through the PVC wall is then protected with epoxy adhesive. In operation this is coated with a gold film and serves as the working electrode W. The counter electrode C is also a wire electrode (any of a variety of metals, in a variety of sizes, can be used for this purpose, we chose a 250 ìm diameter Pt wire) inserted in the opposite arm of the tee and secured in the same fashion. The reference electrode R is a chloridized 250 ìm dia. silver wire secured in the vertical arm of the tee in the same fashion as the other electrodes. A segment of a 0.2 mm i.d. PTFE tube is inserted in the vertical arm atop R, to restrict flow through the vertical arm (<10% of flow relative to other arms). In practice, the vertical arm is connected to a container with 1 M HCl and a small rate of gravity-induced flow of the HCl solution through this arm, out to waste through the counter electrode port, serves to maintain the reference electrode at a constant potential.

1. zzAu film is deposited on the Pt wire working electrode by turning V4 on and flushing the cell with Au solution (all flow rates ~ 1 ml/min). First the electrode is cleansed and any old film removed; the electrode is held at +0.95 V (all potentials vs. Ag/AgCl ref. electrode) for 5 s. The electrode is then held at – 0.1 V for 9 s to plate Au and then cycled to +0.6 V for 1 s to remove easily oxidizable co-deposited impurities. This is repeated 5 more times. Little more than 1 ml Au solution (ca. 55 ìg gold) is used.

V1 is turned on. The sample flows through the cell for 5 s then electrodeposition is started at –0.2 V. Deposition periods of 20, 5, and 1 s are respectively used for anticipated As(III) values of 0-50, 50-300 and >300 ìg/l.

V2 is turned on and the stripping medium (4.5 M HCl) flows through the cell for 30 s while the electrode is held at –0.1 V to remove easily oxidizable co-deposited impurities. A linear staircase sweep (5 mV, 16.7 ms per step) to 0.55 V is then applied, requiring 2.2 s for the sweep time. The computer calculates and displays di/dV vs. V and takes the peak-peak height difference (in ìA/V) as the signal. (The stripping analysis may be performed by either voltammetry or potentiometry, depending on the position of switches K11 and K12, which are simultaneously actuated. These switches are actuated by a fifth digital output line. For carrying out the analysis in the suggested stripping voltammetry mode, K11 and K12 are simply in the off position).

Upon completion of the above steps, the computer prompts one to add an oxidant (250 ìl of 2 mM KMnO₄ or 100 ìl of saturated bromine-water per 100 ml sample) to oxidize all inorganic As to As(V) to each of the sample and the spiked sample bottle and swirl it to effect mixing.

V1 is turned on to allow solution to flush the connecting tubing and cell for 30 s. The electrode is held at –1.6 V for 5 s. When the applied voltage is below -0.5 V, pre-electrolysis current is significantly increased due to evolution of hydrogen and A1 can be saturated. In this case, the feedback resistor associated with A1 is automatically reduced to 1 KÙ via a relay that is turned on by a sixth digital output whenever the applied voltage to the working electrode is less than –0.5 V.

1.Preoxidation at –0.1 V is carried out for 1 s to remove easily oxidizable impurities. This is enough for 50-300 ìg/L total As. For 0-50 ìg/L total As, repeat the deposition procedure in step 7 (except that the flush period is reduced to 5s, the connecting tube is filled with the same sample and do not need to be flushed) and this preoxidation step up to 3 more times.

 Copies of the software and other relevant assistance to build the instrument for noncommercial purposes are available from the authors. A complete notebook computer based instrument is also commercially available from AnalTech (Lubbock, TX).

Electrochemical techniques provide a simple and sensitive means not only to measure total arsenite and arsenate but also to make a differential measurement since As(III) and As(V) exhibit different electrochemical behavior. Conventionally, As(III) is electrochemically reducible to the element while As(V) is generally considered not reducible electrochemically. Most approaches thus reduce As(V) to As(III) by a chemical reductant such as iodide [] mannitol [], sulfur dioxide [, , ], hydrazine-hydrobromic acid [, ], cysteine or sulfite [] prior to the electrochemical measurement of total As. More often than not, the reduction procedure is carried out in strongly acidic media at an elevated temperature, hardly a simple procedure to carry out in the field. Moreover, all of the above approaches [21-28] as well as other reported procedures for determining As(III) [, ] uses cathodic stripping analysis, typically with a hanging mercury drop electrode; this requires the removal of oxygen. Again, this is not a field-friendly procedure.

Although relatively less effort has been placed on the determination of As by anodic stripping techniques, it is well known that this is possible on a gold or gold-coated electrode [24, 25, 31-35]. Moreover, it is known that As(V) can be directly electroreduced on a gold electrode if the applied potential is made sufficiently cathodic [31-33]. We have therefore concentrated on this approach to develop a field-deployable instrument.

Figure 4 shows the dependence of the arsenic stripping signal with the deposition potential. The maximum signal is reached at - 0.2 V. At greater cathodic potentials, down to -1.2 V, the stripping signal continues to decrease, probably due to increased evolution of hydrogen. At still greater reduction potentials, the signal no longer changes, probably because hydrogen evolution becomes constant. After addition of 5 ìM permanganate to the sample, no deposition of As occurs until a deposition potential of -0.8 V. This means As(III) can be totally oxidized by permanganate to As(V), which is not reduced at potentials sufficient to reduce As (III). However, it can be seen that As(V) can be reduced if the deposition potential is made sufficiently negative. The stripping signal increases with decreasing deposition potential down to -2 V but even at such potentials, it does not equal the same signal obtained from the same concentration of As(III). Presumably, there are kinetic barriers to the reduction step that makes it a less efficient process than the reduction of As(III). For this reason, it is not possible to accurately determine total As in a sample by simply using a sufficiently cathodic deposition potential. All of the As must be in one form for an appropriate interpretation of the results. It is for this reason that an oxidation step to convert all As to As(V) is used before the total As determination step.

Unlike deposition in Hg, which forms an amalgam, arsenic does not dissolve significantly in the gold film. Furthermore, elemental arsenic is not conductive. Extensive deposition of As on the film changes the film resistance and can alter the deposition rate. As a result, it is necessary to choose an appropriate deposition period. In our experimental conditions, the stripping signal of 50 ìg/l As(III) and 50 ìg/l As(V) increases linearly with the deposition period, up to deposition times of 30 s (r² = 0.9955) and 40 s (r² = 1.0000), respectively. Higher deposition periods result in a gradual decrease of the stripping signal per unit deposition time.

Arsenic oxidation and reduction are reversible in higher concentration of hydrochloric acid. This may be due to the formation of AsCl₃; the formation of AsCl₃, instead of AsH₃s24 , has also been used in atomic spectrometry and reportedly is subject to less interferences than that involved with AsH₃ generation []. The As stripping peak gets narrower with increasing HCl concentration from 0.5 to 6 M HCl. The stripping peaks for all the metals shift to more cathodic potentials, but to different extent. Mercury forms an even stronger association with chloride relative to copper and arsenic. Consequently, the mercury stripping peak appears after the copper stripping peak in 0.5 M HCl while the situation is reversed in 6 M HCl. However, addition of large volumes of acid to significant volumes of water samples create disposal and transport problems in the field and is impractical. It is sufficient to deposit the sample from a pH 1- 2 medium and perform the stripping while flowing a more acidic medium through the cell. In fact, this flow can be adjusted to be considerably less than 1 ml/min, if desired, without affecting results adversely. Figure 5 shows the stripping curves from a sample of tap water spiked with 50 ìg/l each of Hg(II), Cu(II) and As(V) and acidified to a pH of ~1 after 20 s deposition at -1.6 V and stripping in different concentrations of HCl. The As stripping signals increase with HCl concentration from 0.5 to 6 M HCl. While operating at high acid concentration is clearly better, it is also obviously desirable to minimize the generation of acidic waste. Based on these results, a stripping medium of 4.5 M HCl was used in all experiments.

The analytical problem in Bangladesh and India is primarily concerned with identifying samples with As levels of 50 ìg/l and higher. In this case, a deposition time of 5 s is sufficient for measuring either As (III) or total As. At this deposition time, the stripping signal increases linearly with As concentration up to 350 ìg/l As (r² = 0.9984 for As(III), 0.9999 for As(V)). With samples containing > 300 ìg/l As, the deposition time should be reduced to 1s. The system is, of course readily amenable to measuring lower concentrations of As. With a deposition time of 20 s at -0.2 V and –1.6 V, respectively, the signal is linear with As(III) concentration up to at least 60 ìg/l As (r² = 0.9986 for As(III) and 1.0000 for As(V)). Addition of As to our local tap water samples (no As is detectable in the sample as such and residual chlorine present in the sample converts added As, in whatever form, to As(V), the total dissolved solids concentration (over 1000 mg/l) is much higher than the samples typically encountered in Bangladesh and Eastern India) in 5 ppb increments to 60 ppb and stripping analysis after a deposition time of 20 s at -1.6 V, shows excellent linear correlation of the stripping signals (r² = 0.9983).

The LOD is related to the deposition time and decreases with increasing deposition time. In the application targeted by this work, extraordinarily low LODs are not needed. The longest deposition time studied was 80 s. Under this condition, based on a S/N=3 criterion, an LOD of 0.5 ìg/l As (III) or As(V) was observed (Figure 6). A smaller diameter working electrode (0.025 mm) yields even lower noise levels and thus lowers the LOD further (to about 0.2 ppb). For analysis below the ìg/l level, stripping potentiometry is preferable. This can be carried out with this instrument but is not discussed here.

The most likely interfering species may be Cu, Hg, Zn and Bi because the stripping peaks of Hg and Bi are close to the stripping peak of As and intermetallic compounds of Cu-Zn and Cu-As can be formed at the working electrode. Figure 7 shows an anodic stripping voltammogram for Bi and As, the two are reasonably separated and given the paucity of Bi in drinking water, no realistic interference is expected. There is no significant change upon adding 100 ìg/l of Pb or Zn (that for Zn is not separately shown) but addition of Zn at the 1200 ìg/l level causes the As stripping peak to decrease by ~30%. Figure 8 shows an anodic stripping voltammogram for As in the presence of both Bi and Hg. Addition of 100 ppb Hg(II), a very high concentration relevant to natural waters, results in a Hg peak well separated from the As response and does not significantly affect the peak height of the latter. However, the As signal is decreased with increasing concentration of Cu. Typically, the peak-peak height difference in the derivative plot (not shown) can show a decrease of 50% in the As signal when Cu is present in 20-fold excess. This probably occurs because of the deposition of the intermetallic compound Cu₃As [29].

Given realistic surfactant and other trace metal concentrations in drinking water, it is possible that reliable values for As can be obtained without further ado. However, we feel that a much greater degree of reliability is conferred upon implementing standard addition, as outlined in the Experimental section. Standard addition of 10-60 ìg/l As in 10 ìg/l steps to local municipal water samples (containing no detectable As otherwise) indicate recoveries of 95, 107, 101, 103, 101, and 97%, respectively. Replicate analysis of the same sample (n=8) on the same gold film showed a relative standard deviation of 2.2%.