Electrochemical Arsine Generators for Arsenic Determination

Jul 18, 2014 - It has recently become affordable thanks to its large scale production for extensive use in mobile phones and so on for heat dissipatio...
0 downloads 5 Views 3MB Size
Article pubs.acs.org/ac

Electrochemical Arsine Generators for Arsenic Determination Hong Shen† and Purnendu K. Dasgupta* Department of Chemistry and Biochemistry, The University of Texas at Arlington, Arlington, Texas 76019-0065, United States S Supporting Information *

ABSTRACT: Arsine generation is the gateway for several sensitive and selective methods of As determination. An electrochemical arsine generator (EAG) is especially green: we report here the use of two electrode materials, aluminum and highly oriented (ordered) pyrolytic graphite (HOPG) never before used for this purpose. The first is operated on a novel constant voltage mode: current flows only when the sample, deliberately made highly conductive with acid, is injected. As a result, the cathode, despite being a highly active metal that will self-corrode in acid, lasts a long time. This EAG can be made to respond to As(III) and As(V) in an equivalent fashion and is fabricated with two readily available chromatographic T-fittings. It permits the use of a wire roll as the cathode, permitting rapid renewal of the electrode. The HOPG-based EAG is easily constructed from ion chromatography suppressor shells and can convert As(III) to AsH3 quantitatively but has significantly lower response to As(V); this difference can be exploited for speciation. The success of Al, an active metal, also dispels the maxim that metals with high hydrogen overpotential are best for electrochemical hydride generation. We report construction, operation, and performance details of these EAGs. Using gas phase chemiluminescence (GPCL) with ozone as a complementary green analytical technique, we demonstrate attractive limits of detection (LODs) (S/N = 3) of 1.9 and 1.0 μg/L As(V) and As(III) for the HOPG-based EAG and 1.4 μg/L As(V) or As(III) for the Al-based EAG, respectively. Precision at the ∼20 μg/L As(V) level was 2.4% and 2.1% relative standard deviation (RSD) for HOPG- and Al-based EAGs, respectively. Both HOPG- and Al-based EAGs permitted a sample throughput of 12/h. For groundwater samples from West Texas and West Bengal, India, very comparable results were obtained with parallel measurements by induction coupled plasma-mass spectrometry.

T

reaction of AsH3 thus generated with O3 to produce chemiluminescence (CL). It is possible to detect inorganic As down to sub-μg/L levels; As(III) and As(V) both respond. The reduction can be accomplished chemically,5−8,11 with NaBH4, or electrochemically.9,10 While initially we used NaBH4 and devised methods to improve the stability of this reagent (the cost of NaBH4 is not insignificant relative to developing economies and stability of the solid is poor at tropical humidities), electroreduction is more attractive. Laborda et al. reviewed various cell designs, electrode materials, interferences, etc. for hydride generation in 2007,13 and some 22 papers already specifically described electrochemical arsine generators (EAGs). They classified cell designs based on whether the electrodes and catholyte/anolyte flows were parallel or concentric, and all designs since described fall in one of these categories. In one unique design described later,10 part of the catholyte flow leaks into the anode chamber to constitute the anolyte flow. Laborda et al. also made a valiant attempt to identify the best electrode materials based on the extant literature. However, operating conditions, including cell

he presence of arsenic (As) in groundwater in many parts of the world, especially in the Gangetic delta, and South Asia in general, is widely recognized as one of the greatest environmental calamities of our time.1 It is also well-known that virtually all of this groundwater arsenic is in the more toxic inorganic form.2 Although atomic spectrometry in its various forms and induction coupled plasma-mass spectrometry (ICPMS) are the instruments of choice for arsenic determination in the developed world, these options are too expensive where the need is most acute. Further, they cannot be used onsite. Recent efforts on nonatomic spectrometric methods of arsenic analysis have been reviewed;3 by far, the most extensive are electrochemical efforts as the requisite sensitivity can be attained at a low cost. In practice, however, electrochemical techniques are not routinely used; as Kinniburgh and Kosmus4 put it, electrodes are notoriously f ickle; standard addition and frequent electrode maintenance/rejuvenation are needed to get reliable results. We have spent a significant amount of effort in recent years in developing and applying a gas phase chemiluminescence (GPCL) system for measuring waterborne arsenic;5−10 the system has been modified and further developed by others,11,12 and the instrument is now in use in Bangladesh. The basic principle is the reduction of inorganic As to AsH3, and the © 2014 American Chemical Society

Received: May 4, 2014 Accepted: July 18, 2014 Published: July 18, 2014 7705

dx.doi.org/10.1021/ac501636u | Anal. Chem. 2014, 86, 7705−7711

Analytical Chemistry

Article

Figure 1. The FI-EAG-GPCL analyzer, schematically shown. Left module: pumping and sample delivery; blue indicates Al cathode; central module: hydride generation and gas−liquid separation; dotted line items are used for Al cathode for auxiliary purging; right module: ozone generation and GPCL detection.

pyrolytic graphite (HOPG) planar cathode directly put in the housing for an anion chromatography suppressor and (b) a concentric electrode design where an Al wire is used as an easily renewable cathode. Their performance with a flow injectionGPCL-based analyzer is described.

designs, current densities, and contact times, differed widely among the different reports. Largely on the basis of their previous work14 and Denkhaus et al.,15,16 they suggested that efficiency of arsine generation is directly dependent on the hydrogen overpotential (HOP) of the cathode material. If H+ reduction to H2 is the limiting process, the HOP controls the cathode potential. A higher HOP may mean higher energy availability in the charge transfer reaction and thus increased efficiency of AsH3 generation. Arsine generation efficiency was suggested to follow the HOP order: Pb > Hg ≈ Ag > Cd > glassy carbon > Ag > Au > Pt. As some of our initial results were not concordant, we studied 22 electrode materials under identical conditions (current density of 1 A cm−2, 0.1 M H2SO4 catholyte) with both As(III) and As(V). Of the electrode materials mentioned above, Pb, Cd, graphite, and Pt were included.10 For As(V), the order was Pb > graphite > Cd > Pt, and for As(III), the order was graphite > Pb > Pt > Cd. We hasten to add that we also observed that the relative efficiencies can vary with the current density and the acidity of the solution, but clearly, the overall behavior does not follow the HOP. In our experiments, Zn, where hydrogen overpotential is irrelevant, produced among the highest responses for As(III). On most cathodes, the difference between As(III) and As(V) response was marked, Al being an exception where the two responded identically under these operating conditions. However, even with significant cathodic potential applied to it, Al dissolved in an acidic medium. An active metal that spontaneously dissolves in an acidic solution has never been used as an EAG cathode. Regardless of the specific electrodes and the cell design, EAG cells are custom-made. For coupling to GPCL in particular, our previous EAG design2 utilized the anodically generated O2 to make O3; this increases [O3] and improves limits of detection (LODs). It did, however, complicate overall system design as well as EAG design. With a regulation limit of 10 μg/L, sub-μg/ L LODs are not really needed and simpler (albeit less sensitive) systems are more practical for field use. Similarly, speciation of inorganic As(III) and As(V) is needed only in a minority of studies, and regulations are based on total [As]; field analysis systems can do without such speciation (there is in fact no evidence that the chronic toxicity from μg/L levels of inorganic As is affected by its oxidation state). Herein, we demonstrate two EAG designs: (a) a parallel electrode type design that uses a highly oriented (ordered)



EXPERIMENTAL SECTION Standards, Reagents, and Components. Analytical grade reagents were used throughout whenever available. Details of reagents and other components and sources are given in the Supporting Information. Samples and Comparative Measurements. Groundwater samples containing low levels of As from West Texas and higher levels from West Bengal, India, were supplied by collaborators. Comparative measurements were done by the previously described graphite cathode EAG10 and by a Thermo Fisher Scientific X-series II ICPMS.17 System Setup. As shown in Figure 1, the system consisted of three modules: (left module) the pumping and injection module composed of a 4-channel (three actually used) peristaltic pump (Dynamax RP-1, www.Rainin.com) and a 6port injector (V-450, http://webstore.idex-hs.com/). In the following where the described parameter differs between the HOPG and the Al cathode system, the value for the latter is given in parentheses. The injector was equipped with a 100 (400) μL loop. The anolyte was 0.25 (0.075) M H2SO4; the catholyte was 0.01 M H2SO4 (water). Both anolyte and catholyte were pumped at 1 (1.6) mL/min; the third pump channel was operated at the same flow rate in the aspiration mode to aspirate the liquid effluent from the gas−liquid separator (GLS) connected to the cathode outlet. The central module consisted of the EAG. For the Al system, an auxiliary gas purging arrangement (shown and discussed in detail later) was also present. The central module also contained the GLS, a 1 mL disposable polypropylene pipet tip. The EAG was powered in the constant current (voltage) mode at 500 mA (8.00 V) by the power supply (E3616A, www.hp.com). The gaseous effluent from the GLS, bearing AsH3 and H2, entered the third module (assisted by air purging), the GPCL detector, consisting of an externally silvered chemiluminescence (CL) chamber5 with two inlets for arsine and O3, respectively, and an outlet. A miniature photosensor (photomultiplier module H5784, www.hamamatsu.com) monitored the CL intensity. A 7706

dx.doi.org/10.1021/ac501636u | Anal. Chem. 2014, 86, 7705−7711

Analytical Chemistry

Article

replacement with (dis)assembly involving a large number (18) of screws was onerous. It is possible that, had the device been operated in the constant voltage mode (vide inf ra) with a water carrier, the electrode corrosion rate would have been negligible. In hindsight, assembly with spring-loaded screws or a more easily removed clamp would have permitted the exposure of a new portion of a small width Al foil roll. We opted, however, for a concentric design easily made from inexpensively available components. A similar design was originally used in the first electrodialytic suppressors18 and is shown in Figure 3. The

secondary 2-stage operational amplifier provided further gain (typ. 950×) and offset functions. Unless otherwise stated, the photomultiplier tube (PMT) was operated at 85% of the maximum voltage that provided ∼1/3 of the maximum gain. A miniature air pump (Sensidyne, 6 V DC) pumped ambient air (filtered at the inlet through an activated carbon cartridge) via a pulse dampener buffer (a 300 mL thin-wall polyethylene bottle) through a miniature ozone generator (EOZ-300Y, www.enaly. com) into the CL chamber. Small sections of a capillary restrictor tubing were placed after the ozone generator (0.1 mm i.d., ∼25 mm long) and the exit of the CL chamber (0.3 mm i.d., ∼25 mm long), to make the ozone production and the CL reaction occur at higher than ambient pressure and provide greater ozone concentrations and hence greater CL intensity. The instrument output was collected by a 12-bit A/D card directly into Microsoft Excel. Both peak height and area-based quantitation was used; the uncertainties are reported as ±1 sd (n ≥ 3). EAG Design. HOPG-Based Parallel Electrode EAG. Highly ordered pyrolytic graphite displays extraordinarily high directional thermal and electrical conductivity (HOPG has ∼4× the thermal conductivity of copper). It has recently become affordable thanks to its large scale production for extensive use in mobile phones and so on for heat dissipation). It is denser than standard pyrolytic or molded graphite and is not porous (that leads to memory effects/tailing responses). The HOPG EAG, made from a discarded electrodialytic suppressor for anion chromatography, is shown in Figure 2. Details of its construction are given in Supporting Information. The active electrode area is 140 mm × 10 mm, and the cathode chamber volume is ∼125 μL. Renewable Aluminum Cathode EAG. We initially experimented with standard kitchen Al foil instead of the HOPG sheet as the cathode in the design of Figure 2 under the same conditions. The consumption rate of the foil was too rapid, and

Figure 3. Concentric design for a replaceable Al wire cathode. A Nafion cation exchange membrane tube houses an Al wire cathode. The catholyte/sample flows through the Nafion tube to the GLS. A Pt wire wrapped around the Nafion serves as the anode; the anolyte flows countercurrent through an outer PTFE jacket. The device can also be operated without a jacket, and the entire Nafion tube is immersed in a beaker of the dilute anolyte acid.

construction details are given in the Supporting Information. An important feature of this design is the ability to loosen two terminal nuts, expose a new portion of the Al wire (terminus ending in a spool) functioning as the cathode, and reseal the device in less than a minute. Sample Pretreatment. Total Arsenic. Ten mL of water was put through the H+-form cation exchange resin bed to remove potential interferences. Twenty μL of NaOCl (6% w/v) was added, vortex mixed, and allowed to stand for 1 min; 2.5 mL of 0.05 M H2SO4 was added to make the solution acidic, thus increasing its conductivity and promoting the decomposition of excess hypochlorite. Brief ultrasonication can be used to remove the excess chlorine in lieu of or in addition to the next step that involves the addition of 30 μL of 1 M NH2OH·HCl. The pretreated samples were loaded into an autosampler or manually injected. After initial analysis, the sample with the highest concentration from West Bengal was diluted 10-fold prior to sample treatment to keep it within calibration range. For the Al EAG, only acidification was sufficient. Speciated Measurement. In the HOPG system, As(III) responded ∼4× greater than As(V) (see below). As such, making the measurement twice, with and without the peroxidation step, could provide a speciated value. This possibility was tested with a synthetic mixture containing As(III) and As(V) in known ratios.

Figure 2. Planar HOPG cathode electrochemical arsine generator assembly configuration. 7707

dx.doi.org/10.1021/ac501636u | Anal. Chem. 2014, 86, 7705−7711

Analytical Chemistry



Article

RESULTS AND DISCUSSION Operational Differences and Rationale. Parametric optimization details are presented in the Supporting

Electroreduction to arsine requires an acidic pH. A combined solution is preaddition of sufficient acid to the sample, rendering a uniformly low pH and high conductivity, independent of initial sample composition. This is similar to total ionic strength adjustment buffer addition prior to potentiometric fluoride determinations. While this approach would be impractical for samples of variable and high salinity, e.g., estuarine waters, most arsenic analyses involve groundwater/drinking water. For the majority of such samples, addition of 15−50 mM H+ results in a constant (high) conductivity. We spiked a small quantity of 2.5 M H2SO4 to the sample to attain a concentration of 10−25 mM H2SO4 in the treated sample; the acid neutralizing capacity (or ionic strength) of most groundwater samples is negligible in comparison. To put conductivities in perspective, 10 mM H2SO4 is ∼4000 mg/L total dissolved solids (TDS), measured in terms of NaCl. The current that results when the sample passes through the system at the particular voltage applied is in the optimum range for efficient arsine production. If the catholyte is pure water throughout, the background current for the Al cathode device will be zero. In practice, there is a small Donnan-forbidden penetration of the anolyte acid into the catholyte stream,19 resulting in low single digit mA background currents at 8 V and minimal erosion. The geometry of the Al EAG is more prone toward bubble entrapment; the optimum flow rate for better reproducibility is therefore higher in the Al EAG. For the Al vs HOPG device, the respective cathode areas, cathode chamber volume, and the residence time in the cathode chamber at the optimum flow rates were ∼280 vs 1400 mm2, ∼80 vs 125 μL, and 7.5 vs 3 s. In addition, the temporal current profile follows the concentration profile of the dispersed sample: as the sample disperses in the stream, the conductivity and the attendant current also decreases. As AsH3 formation is current dependent, the AsH3

Figure 4. (a) Response to a 5 mg/L As(III) standard on two different types of graphite cathodes in the same cell. (b) Response to a 0.2 mg/ L As(III) standard before and after a 5 mg/L As(III) injection on a standard graphite cathode; the latter has a net peak height of ∼28% higher.

Information. Here, we discuss some key operational differences between the two devices. All reported EAG devices use constant current reduction in an acidic solution. Active metal electrodes cannot be used in such systems; an Al electrode will rapidly dissolve. Even if the cathode is readily renewed, frequent replacement is inconvenient. In a continuous flowthrough system, however, analyte reduction actually takes place only a fraction of the time. We reasoned that a combination of a relatively nonconductive catholyte carrier and constant voltage operation can effectively eliminate erosion when the electrode is not actually being used. On the other hand, when it contacts the cathode, the sample must be sufficiently conductive for enough current to flow. To avoid sample to sample current variations, the sample conductivity needs to be constant.

Figure 5. Effect of current on CL response of As(III) and As(V) CL intensity. (a) HOPG EAG. 100 μL sample. Catholyte: 0.01 M H2SO4; anolyte: 0.25 M H2SO4; electrolyte flow rate: 1.0 mL min−1; O3 capillary: 2.5 cm. Note that the As(V) concentration is twice the As(III) concentration; the peak As(III) response is ∼4× the peak As(V) response. (b) Al EAG. 400 μL sample. Catholyte: water, sample acidification level 0.01 M H2SO4; anolyte: 0.01 M H2SO4; catholyte and anolyte flow rate: 1.6 mL min−1; O3 capillary: 2.5 cm. The inset shows the response to sequentially injected standards of 25 μg/L As(III). 7708

dx.doi.org/10.1021/ac501636u | Anal. Chem. 2014, 86, 7705−7711

Analytical Chemistry

Article

Figure 6. (a, b) HOPG system. 100 μL injections; i = 400 mA; anolyte, 0.25 M H2SO4; catholyte, 0.01 M H2SO4. (a) As(III) samples in 0.01 M H2SO4, 1−100 μg/L. Blue lower inset shows magnified view of 1−10 μg/L injections. Brown upper inset shows repeatability at 10 μg/L. (b) As(V) samples in 0.01 M H2SO4, 1−200 μg/L. Blue lower inset shows magnified view of 1−20 μg/L injections. Brown upper inset shows repeatability at 20 μg/L. (c) Al EAG system. As(V) samples in 0.01 M H2SO4, 2−100 μg/L. Blue lower inset shows magnified view of 2−10 μg/L injections. Brown upper inset shows repeatability at 20 μg/L. V = 8 V; anolyte, 0.075 M H2SO4; catholyte, water.

supplies air to the ozone generator) is brought to the upper part of the GLS via a normally open solenoid valve. In optimized operation, the air flow is turned off 1 min after injection; the arsine is allowed to accumulate for 2.5 min, and then, the valve is opened again to rapidly transport the AsH3. This was also observed to improve the precision of the measurements. Memory Effects. Under many conditions, EAG devices exhibit memory effects that can be severe.20−23 In our initial experiments with the planar device, we experimented with standard graphite plates as the cathode. These are porous and cause tailing and exhibit memory effects. Figure 4a demonstrates the behavior of a planar cell using the standard graphite vs an HOPG cathode; ordinary pyrolytic graphite is only marginally better. Of greater concern beyond the increased baseline with the standard graphite electrode is the increase in peak height of a lower level sample after a high level sample; see Figure 4b. Admittedly, the effects in Figure 4 are exaggerated in that most real groundwater samples are not likely to contain 5 mg/L As, but the effects are proportionate at lower concentrations. The memory effect suggests that arsine/As somehow remains trapped on the electrode and some of this is released when a second impulse of arsine is produced. The corresponding effect on HOPG cathodes is at least 3−5× lower and is not seen at all on an Al cathode. Response to As(III) vs As(V). HOPG Cathode. Figure 5a shows the response of the system to As(III) and As(V) as a function of the current. These results suggest that the HOPG EAG will not be as attractive if total arsenic measurement is the primary objective because one will have to convert all the arsenic into either As(III) or As(V). On the other hand, precisely because the responses are different and there is considerable dependence of the exact behavior on the current, it is particularly conducive to speciated measurement by measuring the sample twice either (a) at two different current levels (e.g., 200 and 500 mA) or (b) before and after oxidation of all As to As(V) (or reduction of all As to As(III)). Al Cathode. In contrast, the response of the Al cathode to As(III) and As(V) was much closer. The exact response depended on the peak current; this in turn was controlled by the level of the acidification of the sample and the applied

Figure 7. Intercomparison results for different groundwater samples. The low level (W-series) samples were analyzed by HOPG EAG after NaOCl oxidation both by a conventional calibration method (external standard, red bars) and by standard addition (blue bars) whenever sufficient sample was available. These results were linearly correlated with a correlation coefficient of 0.99 and a slope of 0.95, with an intercept statistically indistinguishable from zero, indicating no matrix effects, despite the high salinity and low As levels. Also shown are measurements by a previously described GPCL method10 (green bar) and hydride generation ICPMS (white).16

generation efficiency is not equal across the sample band and maximum conversion efficiency is attained only at the peak. All of these factors contribute to lower average AsH3 conversion in the Al device; a 4× larger sample volume is used to compensate for this. Because total current is much less in the Al EAG system, much less electrolytic gas is produced, and if it is operated in the same manner as the HOPG EAG, the evolved arsine will be transferred to the detection cell too slowly. To swiftly purge the generated arsine, an additional line from the air pump (that 7709

dx.doi.org/10.1021/ac501636u | Anal. Chem. 2014, 86, 7705−7711

Analytical Chemistry

Article

showed no effect on As(III) determination; the two do not interact. Adding Fe(III) to As(III) results in conversion of the As (III) to As(V), at least in part. With a solubility product of 10−23 for ferric arsenate,25 the arsenic is effectively removed from solution. The addition of Fe(III) even at 1 mg/L to any standard containing As(III) and/or As(V) results not only in severe negative interference, on the HOPG cathode, but also it irreversibly decreases the response to subsequent samples. On the Al cathode, there is severe interference on that sample but there are no memory effects. Paradoxically, actual groundwater samples from South Asia containing Fe at the mg/L level (Hussam and Munir26 report concentrations exceeding 20 mg/ L in Bangladesh groundwater) showed no electrode poisoning with the HOPG electrode. In fresh anoxic groundwater, the dominant forms of both Fe and As are the lower oxidation states but they do oxidize over time. While the Fe levels are still easily measurable by most analytical techniques, these may not represent soluble iron. With the solubility product as given above, at a typical groundwater pH of 8, it will be readily computed that no more than ∼10−17 M As(V) can exist if 20 mg/L Fe3+ is present. Not really a paradox, it is well-known that in the absence of other complexing ligands the dominant forms of Fe(III) in natural waters are oxo- and hydroxo-bridged polynuclear Fe(III) clusters;27 the solubility product constant of Fe(OH)3 is ∼10−41.25 Obviously, adding Fe(III) to As standards cannot mimic the chemistry in groundwater. In any case, to avoid any possible complications, we adopted the placement of an H+-form small cation exchange containing cartridge either in the sample loading line or immediately following the injector. It is to be noted that the affinity of strong acid form cation exchange resins for nonalkali/alkaline-earth metals is so high that they cannot be displaced by low concentrations of Na+ or Ca2+, as they last a very long time for any application involving analysis of potable water samples and can be readily regenerated with 2 M H2SO4. Electrode Lifespan. Except when Fe(III) was freshly added to a sample, the HOPG electrodes lasted at least 500 h (duration of our experiment) in continuous use with synthetic standards or real groundwater samples. We could not rejuvenate a Fe(III)-poisoned electrode by anodic oxidation at 100 mA or reduction with acid and 1% NaBH4. Even mechanical abrasion did not fully restore the original performance. We deliberately designed the Al cathode to be easily renewable. However, it lasted a minimum of 2 weeks during which 600−800 injections were made. At this time, the electrode was replaced to check the ease of exposing a new portion rather than a real replacement need. Response to Organic Arsenic Species. We previously observed that, while NaBH4 reduces the two most common organic As species, MMA and DMA, to the corresponding arsines, these produce relatively low CL. For the HOPG EAG, 100 μg/L As as MMA produced a signal equivalent to ∼40% of 10 μg/L As(V) and DMA was even less sensitive; it produced a response 0.4% that of As(V). The Al EAG behaved in a similar fashion. For all purposes, given the preponderance of inorganic As over organic As in groundwater, these systems will essentially exclusively measure inorganic As. Note that, while NaOCl readily oxidizes inorganic As(III) to As(V), we observed no conversion of MMA or DMA to inorganic As. Application to Real Samples and Intercomparison. Nine relatively low arsenic content, high salinity (1000−5000 mg/L TDS) West Texas groundwater samples (W1−W9) and four much higher level Gangetic delta groundwater samples

voltage. We show here the voltage dependent data for acidification at the 10 mM H2SO4 level; very similar results are obtained at the 25 mM H2SO4 level except that the corresponding voltages are lower. As can be seen in Figure 5b, the signal reached a plateau level for As(V) with 8 V applied and did not change thereafter. For As(III), the same plateau was reached with 7 V applied. We operated at 8 V to obtain an equivalent response. Analytical Figures of Merit. Results of parametric optimization are reported in the Supporting Information. Figure 6a,b shows typical calibration output and indications of reproducibility, respectively, for As(III) and As(V) for the HOPG EAG. Area-based calibration plots are shown for As(III) and As(V) in Figures S1 and S2 in the Supporting Information; the behavior is qualitatively the same for height or area-based calibration: In both quantitation modes and for both As(III) and As(V), the behavior is linear at higher levels but a distinct departure from linearity is readily perceptible for both As (III) and As(V). The data are well represented by quadratic equations with zero intercepts: HOPG As(III) signal area (V × s) = (9.72 ± 0.24) × 10−3[As(III)]2 + 2.25 ± 0.02[As(III)]; r 2 = 0.9940

(1)

HOPG As(V) signal area (V × s) = (2.32 ± 0.10) × 10−3[As(V)]2 + 0.952 ± 0.020[As(V)]; r 2 = 0.9972

(2)

where the applicable concentration ranges are 1−100 and 2.5− 200 μg As/L, respectively, for a 100 μL sample. The S/N = 3 LODs were 1.0 and 1.9 μg/L for As(III) and As(V), respectively, corresponding to mass LODs of 70 and 130 pg. At 10 μg/L As(III) and 25 μg/L As(V), the respective relative standard deviations were 5.0% for As(III) and 2.4% for As(V), respectively (n = 7 each). The peak width at 1% of peak height is 5 min, resulting in a maximum sample throughput (1% carryover) of 12/h. Representative output for the Al EAG with a 400 μL sample (either As(III) or As(V)) injected and 8 V applied is shown in Figure 6c. The height-based calibration equation was: Al EAG signal height (mV) = (4.59 ± 0.20) × 10−3[As]2 + 1.67 ± 0.04[As]; r 2 = 0.9967

(3)

Precision for the 20 μg/L As(V) solution was 2.1% relative standard deviation (RSD). The LOD (S/N = 3) was 1.4 μg/L corresponding to a mass LOD of 560 pg. The peak width at 1% of peak height was 5 min here as well, corresponding to a sample throughput of 12/h. Interferences. As may be expected, in accordance with previous studies, Na+, K+, Ca2+, Mg2+, Cl−, SO42−, NO3−, and HCO3− had no statistically significant effect at a 100 mg/L level on 10 μg/L As in either electrode system. Sulfide and silicate, tested at the 1 mg/L level, also had no effect on the signal from 10 μg/L As. Sn and Sb are known to form hydrides that react with ozone to produce CL, but the signals are ∼2 orders of magnitude weaker.5 Of the other metals, Cu(II) was of interest in that it is known to present a severe interference in EAGs,21 as well as in voltammetry.24 Up to a concentration of 2 mg/L, substantially greater than levels that can be present in drinking water, Cu2+ showed no statistically significant effect on 10 μg/L As in either system. Iron is present in many water samples in significant concentrations. However, it is very difficult to mimic the effect of iron on measuring As using laboratory standards relative to how Fe and As actually occur in real samples. Fe(II) 7710

dx.doi.org/10.1021/ac501636u | Anal. Chem. 2014, 86, 7705−7711

Analytical Chemistry

Article

(7) Sawalha, M. F.; Sengupta, M. K.; Ohira, S.-I.; Idowu, A. D.; Gill, T. E.; Rojo, L.; Barnes, M.; Dasgupta, P. K. Talanta 2008, 77, 372− 379. (8) Sengupta, M. K.; Hossain, Z. A.; Ohira, S.-I.; Dasgupta, P. K. Environ. Pollut. 2010, 158, 252−257. (9) Sengupta, M. K.; Sawalha, M. F.; Ohira, S.-I.; Idowu, A. D.; Dasgupta, P. K. Anal. Chem. 2010, 82, 3467−3473. (10) Sengupta, M. K.; Dasgupta, P. K. Anal. Chem. 2011, 83, 9378− 9383. (11) Assegid, K.; Ahmed, F.; Ahamed, S.; Hussam, A. Anal. Methods 2011, 3, 2921−2928. (12) Hashem, M. A.; Jodai, T.; Ohira, S.; Wakuda, K.; Toda, K. Anal. Sci. 2011, 27, 733−738. (13) Laborda, F.; Bolea, E.; Castillo, J. R. Anal. Bioanal. Chem. 2007, 388, 743−751. (14) Bolea, E.; Laborda, F.; Castillo, J. R. Anal. Sci. 2003, 19, 367− 372. (15) Denkhaus, E.; Beck, F.; Bueschler, P.; Gerhard, R.; Golloch, A. Fresenius’ J. Anal. Chem. 2001, 370, 735−743. (16) Denkhaus, E.; Golloch, A.; Guo, X. M.; Huang, B. J. Anal. At. Spectrom. 2001, 16, 870−878. (17) Sengupta, M. K.; Dasgupta, P. K. Anal. Chem. 2009, 81, 9737− 9743. (18) Strong, D. L.; Dasgupta, P. K. Anal. Chem. 1989, 61, 939−945. (19) Dasgupta, P. K.; Bligh, R. Q.; Lee, J.; D’Agostino, V. Anal. Chem. 1985, 57, 253−257. (20) Hueber, D. M.; Winefordner, J. D. Anal. Chim. Acta 1995, 316, 129−144. (21) Ding, W. W.; Sturgeon, R. E. J. Anal. At. Spectrom. 1996, 11, 421−425. (22) Bolea, E.; Laborda, F.; Castillo, J. R.; Sturgeona, R. E. Spectrochim. Acta, Part B 2004, 59, 505−513. (23) Hashemi, M.; Arbab-Zavar, M. H.; Sarafraz-Yazdi, A. Talanta 2004, 64, 644−649. (24) Dai, X.; Compton, R. G. Electroanalysis 2005, 17, 1325−1330. (25) Langmuir, D.; Mahoney, J.; Rowson, J. Geochim. Cosmochim. Acta 2006, 70, 2942−2956. (26) Hussam, A. K.; Munir, A. K. M. J. Environ. Sci. Health, Part A 2007, 42, 1869−1878. (27) Powell, A. K.; Heath, S. L. Comments Inorg. Chem. 1994, 15, 255−296.

(B1−B4) were analyzed for the total arsenic content by different methods. Of the present EAGs, the data for the Al EAG (similar data were obtained with the HOPG EAG after NaOCl oxidation) are shown in Figure 7 along with the results from other methods (original data are shown in Table S1 in the Supporting Information). It is to be noted that all analyses were done without cation exchange. The results are very comparable. As(III) and As(V) Speciation. Although there is no regulatory need, speciation is often of interest. The HOPG EAG has very different responses for As(III) and As(V) and is therefore well suited for speciation. An alternative to measuring at two different current regimes10 is measuring the sample before and after oxidative conversion of all As to As(V). This does not require changing instrumental conditions; it was chosen because baseline stabilization time between modes is eliminated. Table S2 in the Supporting Information shows results of speciation of low level As(III) and As(V) mixtures from analysis before and after hypochlorite oxidation and interpretation by individual calibration equations and demonstrates that the approach can be successfully practiced, even at low levels.



ASSOCIATED CONTENT

S Supporting Information *

Additional information regarding caveats on absolute magnitudes of the signals, descriptions of standards, reagents, and components, device construction, and figures and data on areabased As(III) and As(V) calibration, relative responses of organic As species vs As(V) (all using the HOPG EAG), intercomparison results of groundwater samples, speciation results of As(III) and As(V) in a mixture, and details of parametric optimization. This material is available free of charge via the Internet at http://pubs.acs.org.



AUTHOR INFORMATION

Corresponding Author

*Phone: 817-272-3171. Fax: 817-272-3808. E-mail: Dasgupta@ uta.edu. Present Address †

H.S.: Institute of Microanalytical Systems, Department of Chemistry, Zhejiang University, Hangzhou, China, 310058. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was supported in part by the National Science Foundation Grant PFI: AIR-TT 1414383, University of Texas at Arlington Jenkins Garrett Professorship to P.K.D., and by the Interdisciplinary Seed Research Fund of Zhejiang University (No. JCZZ-2013010). H.S. acknowledges the helpful assistance of Jian Ma, Mrinal K. Sengupta, and C. Phillip Shelor.



REFERENCES

(1) Smith, A. H.; Lingas, E. O.; Rahman, M. Bull. W. H. O. 2000, 78, 1092−1103. (2) Fazal, M. A.; Kawachi, T.; Ichion, E. Water Int. 2001, 26, 370− 370. (3) Ma, J.; Sengupta, M. K.; Yuan, D.; Dasgupta, P. K. Anal. Chim. Acta 2014, 831, 1−23. (4) Kinniburgh, D. G.; Kosmus, W. Talanta 2002, 58, 165−180. (5) Idowu, A. D.; Dasgupta, P. K.; Genfa, Z.; Toda, K.; Garbarino, J. R. Anal. Chem. 2006, 78, 7088−7097. (6) Idowu, A.; Dasgupta, P. K. Anal. Chem. 2007, 79, 9197−9204. 7711

dx.doi.org/10.1021/ac501636u | Anal. Chem. 2014, 86, 7705−7711