Anal. Chem. 2010, 82, 3467–3473
Green Analyzer for the Measurement of Total Arsenic in Drinking Water: Electrochemical Reduction of Arsenate to Arsine and Gas Phase Chemiluminescence with Ozone Mrinal K. Sengupta, Maather F. Sawalha,† Shin-Ichi Ohira,‡ Ademola D. Idowu,§ and Purnendu K. Dasgupta* Department of Chemistry and Biochemistry, University of Texas at Arlington, Planetarium Place, Arlington, Texas 76019-0065 We describe matrix-isolated, reaction chemistry based measurement of arsenic in water down to submicrograms per liter levels in a system that requires only air, water, electricity, and dilute sulfuric acid, the bulk of the latter being recycled. Gas phase chemiluminescence (GPCL) measurement of arsenic is made in an automated batch system with arsenic in situ electroreduced to arsine that is reacted with ozone to emit light. The ozone is generated from oxygen that is simultaneously anodically produced. Of 22 different electrode materials studied, graphite was chosen as the cathode. As(V) is reduced much less efficiently to AsH3 than As(III). Prereducing all As to As(III) is difficult in the field and tedious. Oxidizing all As to As(V) is simple (e.g., with NaOCl) but greatly reduces subsequent conversion to AsH3 and hence sensitivity. The rate of the AsH3-O3 GPCL reaction and hence signal intensity increases with [O3]. Using oxygen to feed the ozonizer produces higher [O3] and substantial signal enhancement. This makes it practical to measure all arsenic as As(V). The system exhibits an LOD (S/N ) 3) for total arsenic as As(V) of 0.36 µg/L (5 mL sample). Comparison of total As results in native and spiked water samples with those from inductively coupled plasma mass spectrometry (ICPMS) and other techniques show high correlation (r2 ) 0.9999) and near unity slopes. Exposure to elevated levels of arsenic, a class I human carcinogen,1 has become a global concern, affecting millions worldwide.2 The currently recommended upper limit of arsenic in drinking water is 10 µg/L.3 Temporal and seasonal changes in arsenic in South Asian well water are well documented,4,5 necessitating frequent measurement. It is desirable that such * Corresponding author. E-mail:
[email protected]. Phone: (817) 272-3171. † Present address: Department of Chemistry, Palestine Technical University, Palestine. ‡ Present address: Department of Chemistry, Kumamoto University, Kumamoto, Japan 860-8555. § Present address: Dow Chemical, Freeport, TX 77541. (1) IARC. Some Drinking-Water Disinfectants and Contaminants, Including Arsenic. IARC Press: Lyon, France, 2004. (2) Mukherjee, A.; Sengupta, M. K.; Ahamed, S.; Hossain, M. A.; Das, B.; Nayak, B.; Rahman, M. M.; Chakraborti, D. J. Health, Popul. Nutr. 2006, 24, 142– 163. 10.1021/ac100604y 2010 American Chemical Society Published on Web 04/09/2010
systems should be reliable, field usable, inexpensive, sensitive and green, not involve unsafe, noxious, or expensive chemicals. A recent review covers the measurement of environmental arsenic.6 Electrochemistry may have the most promise to meet the desired goals but despite innovations,7,8 electrochemistry generally works only well with As(III). Reducing all As to As(III) is not practical in the field and typically involves expensive reagents and/or heating. The reliability of “Field Kit”-results9 near the regulation limits have been questioned,10,11 aside from concerns with arsine leakage and lead/mercury waste generation.12 We have worked on practical field-usable analytical methods for the gas phase chemiluminescence (GPCL)-based determination of arsenic that relies on CL produced by the arsine-ozone reaction.13 This has been used for chromatographic detection,14 validated for soil/dust samples,15 and adapted for manual lowcost use.16 Recently Xue et al.17 have also reported GPCL-based chromatographic measurement of As following photoxidation (3) United States Environmental Protection Agency (U.S. EPA). Federal Register: January 22, 2001, Vol. 66, No. 14, pp 6975-7066. http:// www.epa.gov/fedrgstr/EPA-WATER/2001/January/Day-22/w1668.htm. (4) Chakraborti, D.; Basu, G. K.; Biswas, B. K.; Chowdhury, U. K.; Rahman, M. M.; Paul, K.; Chowdhury, T. R.; Chanda, C. R.; Lodh, D. Arsenic Exposure and Health Effects; Chappell, W. R., Abernathy, C. O., Calderon, R. L., Eds.; Elsevier Science: New York, 2001; pp 27-52. (5) Thundiyil, J. G.; Yuan, Y.; Smith, A. H.; Steinmaus, C. Environ. Res. 2007, 104, 367–373. (6) Luong, J. H. T.; Majid, E; Male, K. B. Open Anal. Chem. J. 2007, 1, 7–14. (7) Male, K. H.; Hrapovic, S.; Santini, J. M.; Luong, J. H. T. Anal. Chem. 2007, 79, 7831–7837. (8) Hrapovic, S.; Liu, Y.; Luong, J. H. T. Anal. Chem. 2007, 79, 500–507. (9) van Geen, A.; Cheng, Z.; Seddique, A. A.; Hoque, M. A.; Gelman, A.; Graziano, J. H.; Ahsan, H.; Parvez, F.; Ahmed, K. M. Environ. Sci. Technol. 2005, 39, 299–303. (10) Rahman, M. M.; Mukherjee, D.; Sengupta, M. K.; Chowdhury, U. K.; Lodh, D.; Chanda, C. R.; Roy, S.; Selim, M.; Quamruzzaman, Q.; Milton, A. H.; Shahidullah, S. M.; Rahman, M. T.; Chakraborti, D. Environ. Sci. Technol. 2002, 36, 5385–5394. (11) Khandaker, N. R. Environ. Sci. Technol. 2004, 38, 479A. (12) Hussam, A.; Alauddin, M.; Khan, A. H.; Rasul, S. B.; Munir, A. K. M. Environ. Sci. Technol. 1999, 33, 3686–3688. (13) Idowu, A. D.; Dasgupta, P. K.; Genfa, Z.; Toda, K.; Garbarino, J. R. Anal. Chem. 2006, 78, 7088–7097. (14) Idowu, A.; Dasgupta, P. K. Anal. Chem. 2007, 79, 9197–9204. (15) 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. (16) Sengupta, M. K.; Hossain, Z. A.; Ohira, S. I.; Dasgupta, P. K. Environ. Pollut. 2010, 158. 252–257. (17) Xue, J. H.; Zhu, Z. L.; Zhang, S. C.; Zhang, X. R. Luminescence 2009, 24, 290–294.
Analytical Chemistry, Vol. 82, No. 9, May 1, 2010
3467
Figure 1. System schematic: (a) liquid handling module, consisting of syringe pump and multiport distribution valve (SP) and various sample/ reagents. (b) Electrochemical arsine generation module: The electrochemical reactor (ECR) consists of a disposable syringe barrel (SB) with a neoprene stopper (NS) housing a Teflon tape (TT) wrapped graphite rod (GR), ceramic tube (CT) that houses a Pt foil anode connected via a Pt wire exiting through the sealed end of tee T2. The power supply (PS) is connected to GR and the Pt anode. Tees, T1 providing waste/wash port WWP and T2 providing anode liquid outlet AO, connects to a three-way isolation valve IV with tee T3 placed in-between. The solid line of solenoid valve IV is the common port, normally connected to the reservoir vessel (RV) that acts as the anolyte and oxygen storage; when turned on, IV is connected to the container TC. One port of SP can access TC and deliver liquid back to RV via T3. The RV liquid is recirculated by pump PP through the anode liquid inlet AI connected to CT. Oxygen from RV exits through glass wool filled liquid trap GT and feeds the ozonizer OZG. (c) Gas phase chemiluminescence detection module. Ozone generated by OZG flows into the chemiluminescence chamber CC while arsine coming from SB via exit tube E enters CC via solenoid valve SV and liquid trap LT, the exit gas flows out through activated Mn oxide catalyst MC. The emitted light is detected by the photomultiplier tube (PMT).
similar to ref 14, while the photoxidation step is much faster, the limits of detection (LODs) obtained do not compare favorably. We have used borohydride reduction throughout to reduce arsenic; the kinetics of this reduction for As(III) and As(V) are very different.18 Electrochemical arsine generation (EAG) has been reproducibly achieved in flow systems with diverse detectors.19-26 In field analysis, discrete batch measurement is more practical (better liquid economy) than flow methods. Except for an early attempt to couple EAG to GPCL in which both sensitivity and reproducibility were poor, described by us in a patent application,27 there are no reports of batch mode EAG. Here we describe a graphite cathode EAG-GPCL system that (a) oxidizes all As to As(V) in situ with NaOCl, (b) uses anodically generated O2 produced during EAG to make ozone, and (c) (18) (19) (20) (21) (22) (23) (24) (25) (26) (27)
Sengupta, M. K.; Dasgupta, P. K. Anal. Chem. 2009, 81, 9737–9743. Arbab-Zavar, M. H.; Hashemi, M. Talanta 2000, 52, 1007–1014. Ding, W.-W.; Sturgeon, R. E. J. Anal. At. Spectrom. 1996, 11, 225–230. Shen-Tu, C.; Fan, Y.; Hou, Y.; Wang, K.; Zhu, Y. J. Chromatogr., A 2008, 1213, 56–61. Li, X.; Jia, J.; Wang, Z. Anal. Chim. Acta 2006, 560, 153–158. Ozmen, B.; Matysik, F. M.; Bings, N. H.; Broekaert, J. A. C. Spectrochim. Acta B 2004, 59, 941–950. Pohl, P.; Zapata, I. J.; Bings, N. H. Anal. Chim. Acta 2008, 606, 9–18. Machado, L. F. R.; Jacintho, A. O.; Menegario, A. A.; Zagatto, E. A. G.; Gine, M. F. J. Anal. At. Spectrom. 1998, 13, 1343–1346. Hueber, D. M.; Winefordner, J. D. Anal. Chim. Acta 1995, 316, 129–144. Dasgupta, P. K.; Idowu, A.; Li, J. Method and Apparatus for Analyzing Arsenic Concentrations Using Gas Phase Ozone Chemiluminescence. International Patent Publication No. WO 2007/081635 A2, July 19, 2007, http://www.wipo.int/pctdb/en/wo.jsp?WO)2007081635 (Accessed March 8, 2010).
3468
Analytical Chemistry, Vol. 82, No. 9, May 1, 2010
utilizes the elevated O3 levels to enhance sensitivity and measures total As as As(V). EXPERIMENTAL SECTION Standards and Reagents. All reagents are commonly available and were obtained from standard suppliers; see Table S1 in the Supporting Information for details of cathode materials. Except as stated, graphite rods used were EDM graphite rods, type EC16, www.graphitestore.com). Instrumentation. The system (Figure 1) has three modules: The sample and reagent handling module a consists of a 48Kstep syringe pump SP with a 10 mL syringe and a multiport distribution valve (used previously, discussed therein13,15,18 and briefly in the Supporting Information). The pump delivers sample and reagents to the electrochemical reactor (ECR), removes waste therefrom, and also supplies deionized wash water. Module b consists of the ECR, a 30 mL plastic syringe barrel SB (P/N 309650, Becton-Dickinson, http://www.bd.com), with a no. 4 neoprene stopper (NS) on the inverted end. Three components enter through the NS, from left: (1) the arsine generation electrode (except as stated, a graphite rod GR, 6.0 mm diameter 150 mm long, wrapped with PTFE tape TT to control the exposed area, only the bottom 2.5 mm was exposed during operation; sufficient length of the electrode protruded through the stopper to be connected to the negative terminal of the power supply PS (Hewlett-Packard 6266B)), (2) a 12 gauge PTFE anolyte inlet line AI (2.1 mm i.d., 3 mm o.d.), push fit into a porous ceramic tube CT (3 mm i.d., 5 mm o.d., 0.2 µm pores). At the bottom of the
syringe, first a larger tee (T1, 3/8 in., polypropylene) is affixed and into it is push fit a second, smaller tee (T2, 3/16 in., both tees: www.arkplas.com). The reactor is washed/drained through the T1 side arm by SP. The bottom of tube CT seals into T2, and the T2 side arm provides the anolyte outlet (AO). A Pt foil electrode (45 mm long, 2.5 mm wide) is connected to a 0.25 mm diameter Pt wire. This electrode assembly is inserted into CT through T2 bottom; the protruding wire is sealed in with hotmelt adhesive and connected to PS(+). The micropores in CT provide bulk flow resistance and prevent catholyte-anolyte mixing but ionic passage under applied voltage (and hence current) is easily established. To minimize effective anode area continuously changing from gas bubbles in a narrow tube, the anolyte (∼0.1 M H2SO4) was recirculated at 6 mL/min by a peristaltic pump PP (Dynamax RP-1, www.Rainin.com). (3) The third component entering the NS is a PTFE tube E that carries the liberated AsH3 to the GPCL detection module (Figure 1c). After electroreduction proceeds for a preset period, the normally closed solenoid valve (SV) opens to let liberated AsH3/H2 enter the chemiluminescence chamber CC (externally silvered inverted test tube) sitting atop a photomultiplier tube (PMT) via a glass wool filled liquid trap LT. The GPCL module has been described;13-16 some details are in the Supporting Information. Ozone is generated with air (5-10 standard cubic centimeters/ min (sccm) pumped by a small miniature air pump) or with electrogenerated oxygen (vide infra). The reactant gases exit CC through a MnOOH catalyst bed MC (Carulite 200, www. caruscorporation.com) to destroy unreacted ozone. Electrogenerated Oxygen-Based Ozone Generation. While oxygen can be independently generated and supplied, oxygen anodically generated during EAG was already available. We used simple gravity-based gas-liquid partitioning (Figure 1b). The reservoir vessel, RV (1-2 L capacity, 75% filled with 0.1 M H2SO4) supplies the anolyte to pump PP that recirculates it through the ECR anode chamber. During electrolysis, the T2 side arm brings out both the anolyte and O2. Through another 3/32 in. tee T3, the stream enters RV where oxygen accumulates in the headspace and passes into the ozonizer (OZG, Figure 1c) through a glass wool trap (GT) that removes any liquid droplets; the flow rate at steady state is ∼4 sccm, the same as that produced electrolytically, enough to supply sufficient ozone flow for the CL reaction. In order to maintain a constant liquid level in RV and constant O2 flow to the OZG, a three-way solenoid valve IV (both SV and IV from www.Nresearch.com) interfaces the ECR via T3 to either RV or SP. During ECR wash, the flowing anolyte goes to a temporary container TC and is later resupplied (as a 3 mL aliquot) to RV by SP through T3. Note that water is lost through both electrolysis and by evaporation from the electrically heated liquid via the evolved gases. The acid concentration in RV thus gradually increases. With 1 L capacity RV 75% filled in the beginning and an EAG current of 1 A and operating at an ambient temperature of 23 °C, we find that the acid level can be maintained constant if instead of the acid, 3 mL of water is supplied after every 19 samples. For an anolyte concentration around 0.1 M H2SO4, minor increases in concentration does not greatly influence the analyte response. The reason for slow concentration increase of the anolyte is discussed in more detail in the Supporting Information.
Operational Sequence. Initially, anolyte flow from RV to the ECR anode chamber is started by pump PP. SP then introduces 5.0 mL of sample, followed by 2.0 mL of 0.5 M H2SO4 as the outer catholyte. Electrolysis begins immediately; a constant current of 1 A is maintained, requiring 15-18 V. Electrolysis gases are allowed to accumulate for 60 s before SV opens for the next 5 min and electrogenerated AsH3 proceeds completely to CC. After measurement of the CL signal, IV connects to SP; SP withdraws the catholyte and sends it to waste. Depending on the desired degree of sample carryover, up to 10 mL of water is pumped into the ECR and kept for 30 s before aspirating off. IV closes, followed by a 3 mL addition of 0.1 M H2SO4 from TC to RV. The entire cycle is ∼8 min/sample. RESULTS AND DISCUSSION System Optimization. Initial optimization of gas flow rates, electrolyte concentration/flow rates was performed using the platinum anode and cathode. Platinum as the anode provides the needed inertness;28 it was also initially selected as the cathode based on its reported use in EAG systems,20,25,26,29,30 albeit some report low efficiency. Key EAG literature findings are summarized in Table S2 in the Supporting Information. After the final cathode choice, electrolyte concentration/flow rates were reoptimized. Glass and polypropylene ECR bodies performed equally well; we retained the plastic disposable syringe. The cathode and anode compartments require a bulk flow barrier; initial comparison of the ceramic tube used with Nafion tubing (4.5/6.6 mm i.d./o.d.) in 1.0/0.2 M H2SO4 catholyte/anolyte showed 30% less resistance for the ceramic tube based cell; it was henceforth used. Ozone and Analyte Flow. The GPCL signal depends in a complex fashion on the ozonizer flow rate. At high flow rates, generated [O3] is reciprocally related to the air/O2 feed rate (constant mass of ozone produced). At very low flow rates, a maximum plateau concentration is seen (this equilibrium concentration is ∼5× higher for O2 than for air). The ASH3-O3 reaction is not instantaneous, higher [O3] leads to a faster reaction and a sharper signal. However, a high flow of O3 reduces the cell residence time and the reactants escape the PMT view before reaction completion. The relevant [O3] value here is that in the cell, after dilution with the AsH3/H2 flow. Consider that electrolytic H2 produced with 1 A current is 7 sccm; accordingly, we found that 4-9 sccm ozonizer flow produces the maximum CL intensity. A further flow increase decreases the signal (details can be seen in Figure S1 in the Supporting Information). Thus, we selected 4 sccm oxygen or 7 sccm air feed flow into the ozonizer. Choice of Cathode. The cathode is critical for EAG and affects hydride generation efficiency and reproducibility; acidic sample compatibility is also important.30 Carbon in different forms and lead have been most commonly used in electrochemical hydride generation (EHG) systems.21,23,28,29 They tolerate acidic media, and proponents report better EHG efficiency than other cathodes. However, carbon electrodes also reportedly show so much lower response for As(V) compared to As(III)20,31 that prereduction is (28) Laborda, F.; Bolea, E.; Castillo, J. R. Anal. Bioanal. Chem. 2007, 388, 743– 751. (29) Ding, W. W.; Sturgeon, R. E. Spectrochim. Acta B 1996, 51B, 1325–1334. (30) Denkhaus, E.; Golloch, A.; Guo, X. M.; Huang, B. J. Anal. At. Spectrom. 2001, 16, 870–878.
Analytical Chemistry, Vol. 82, No. 9, May 1, 2010
3469
Figure 2. Chemiluminescence signal from 50 µg/L As(III) and 50 µg/L As(V) on different cathodes relative to the response of 50 µg/L As(III) on a Pt cathode taken as unity. Metals Nd-Pt are listed in order of their standard reduction potential; carbon, nichrome, and stainless steel are listed thereafter.
essentially mandatory. A Pb cathode reportedly suffers from interferences by high concentrations of other elements.20 It is difficult to compare different literature reports as no two use the same conditions. We compared the performance of 20 metals (Pb, Sn, Zn, Ni, Nb, Cd, Co, C (graphite), Mo, Ti, W, In, Zr, Ta, Pd, V, Nd, Pt, Cu, and Al) and two alloys (Ni-Cr and type 316 stainless steel) as cathodes. Most electrodes were initially tested as 1 cm × 1 cm foils, with a 0.25 mm Pt-wire providing electrical connection; Ni-Cr was tested as a 40-mesh gauze of equivalent electrode area. All cathode materials were tested using 50 µg/L As(III) and As(V) each, with a Pt anode, 0.1/0.5 M H2SO4 as the catholyte/ anolyte (latter recirculated at 6 mL/min), and a constant current of 1.0 A. (These are default conditions for all experiments unless otherwise stated.) A recent review by Laborda et al.28 suggest that EHG takes place in four steps: (a) diffusion of the analyte to the cathode (it is thus essential to convert As(V) to an uncharged species by a strongly acidic medium); (b) reduction to the element on the cathode surface; (c) reduction of water to hydrogen, also on the cathode surface and formation of the hydride thereon; and (d) diffusion of the hydride away from the electrode. Hydride generation efficiencies are the parameters of general interest. Interference by high levels of other metals, sometimes considered a performance parameter, is not relevant in drinking water analysis. In Figure 2, the response of different cathodes toward 50 µg/L of As(III) and As(V) are shown with the As(III) response of a platinum cathode taken as unity. Hydrogen evolution overpotential (HEO) of a metal has often been regarded as an important parameter in governing EAG cathode behavior. With higher HEO values, more of the atomic hydrogen will still be on the electrode surface and thus take part in hydride formation. HEO is a function of the solution composition and current density. Denkhaus et al.30 measured HEO at various current densities for several electrode materials, and they reported that EAG efficien(31) Schaumloffel, D.; Neidhart, B. Fresenius J. Anal. Chem. 1996, 354, 866– 869.
3470
Analytical Chemistry, Vol. 82, No. 9, May 1, 2010
cies did track the HEO values. Further, on electrodes such as Pb, Cd, and amalgamated silver, they reported that As(V) could be reduced as easily as As(III). Our data are clearly not concordant. Neodymium was the only metal that produced no signal; the Nd electrode could not be kept from dissolving even with high negative voltages applied. Electrodes of graphite, Al, V, Zn, In, Sn, Pb, Pd, and stainless steel (316SS) were more effective than Pt for the reduction of As(III). Only Al produced statistically equivalent responses for As(III) and As(V). Appreciable As(V) response was also produced by Zn, In, Pb, 316SS, and to a smaller extent, the graphite electrodes. Al, Zn, and V slowly eroded even with cathodic potential applied; considering the low cost and ready availability of Al foil, it may nevertheless be possible to design a disposable Al electrode based device. In these initial experiments, graphite showed the highest As(III) response but the variability was high. The imprecision came from the fragile nature of the graphite foil and the difficulty of reproducibly establishing electrical connections. Switching to a mechanically robust graphite rod solved reproducibility issues. Cathode Memory and Conditioning. All aspects of EAG cathode behavior are not revealed by Figure 2; these data represent stable signals that are finally obtained when 50 µg/L As is repeatedly injected. Cathodes of Mo, W, In, Zn, Zr, Pd, V, and 316SS exhibited varying degrees of memory; i.e., starting from a blank when the first standard was injected, the response was low. Only after a number of injections was a steady stable response observed. Similarly, returning to blanks did not immediately produce a zero response. This may be related to specific metals forming surface alloys/intermetallic composites with As. The extant literature indicates cathode preconditioning, whether by mechanical scrubbing/polishing, chemical oxidation by brief immersion in nitric/chromic acids, or electrochemical polarity reversal, is often necessary. In our experiments, the need for such conditioning varied considerably: electrodes of Cd, Ni-Cr, Sn, Co, In, and Ni required frequent conditioning. Otherwise after about an hour of operation, the response decreased. Cathode Response and Reproducibility. On the basis of these results, we selected 316SS, Pb, and two different types of carbon (graphite and high-purity spectroscopic grade carbon rods) for further investigation. Reproducibility was checked over 3-6 consecutive days on the same electrode; the 316SS and Pb electrodes were foils while both carbon electrodes were 3-6 mm diameter rods. Response to 50 µg/L As(III) is shown in Figure 3; the response with a 316SS cathode was the lowest and typically exhibited the highest within-day variance. Lead showed great variability between the first and subsequent days. Spectroscopic carbon response increased over days; it is possible with this and the Pb electrodes that deposition of impurity metals (including As) or surface alteration leads to better performance. By far the best performance, in terms of response, within-day and betweenday reproducibilities, was observed with the graphite cathodes. Intraday or interday precision were 0.99 (see Figures S3 and S4 in the Supporting Information). While the difference between As(III) and As(V) signal heights does decrease with oxygen-based ozone generation, the disparity is still too large. Because As(III) is more responsive, we first explored reduction by KI-ascorbic acid23 (for procedure, see the Supporting Information). At 50 µg/L As(V), ∼80% was converted to As(III) as judged by the response; this did not improve by increasing the reduction time from 3 to 5 min. In addition, (a) the blank signal increased, deteriorating LOD to 1 µg/L As, suggesting the presence of As in the relatively concentrated reductant (or CL from some substance derived from the reductant) and (b) the linearity decreased significantly (r2 ) 0.98); data in Figure S5 in the Supporting Information indicate that reduction is less complete at higher [As(V)], characteristic of an equilibrium-driven process. Pre-EAG oxidation of As(III) to As(V) is usually not attempted because of sensitivity loss. We reasoned that even an air-ozonizer GPCL approach may nevertheless provide sufficient sensitivity for drinking water analysis in South Asia and chose NaOCl, an inexpensive strong oxidizing agent. A small amount (0.1 mL of 0.6% w/v NaOCl) was added to the ECR after sample introduction. We observed identical response slopes (26.9 ± 1.3 and 26.8 ± 1.1 mV/(µg/L) As(III) and As(V) originally taken) with intercepts indistinguishable from zero (details in Figures S6 and S7 in the Supporting Information). The reproducibility permitted a S/N ) 3 LOD of 0.65 µg/L, better than that obtained with KI-ascorbate reduction. Oxygen-Based Ozone Generation. We had previously known that if O2 instead of air is used for ozone generation, the CL signal increases markedly.14 However, for a field instrument, carrying an oxygen tank is a burden. Oxygen is, however, generated during EAG and is readily available for ozone generation. Under conditions described in the Experimental Section, we indeed observed 3× better sensitivity (details in Figure S8 in the Supporting Information) and 2× better LOD (0.36 µg As/L) compared to the air-based ozone generation, more than sufficient to meet drinking water As analysis needs. Note that water vapor reduces the CL intensity. It has previously been shown that the use of small CaCl2 packed tubes prior to the CL reaction increases the signal 2.5× (ref 14). However, the LOD here was considered adequate, and the added complexity was not pursued. Analysis of Real Samples and Intercomparison/Validation. We analyzed six tap water samples from Western Texas/ Eastern New Mexico; these regional samples have very high total dissolved solids (TDS) and high total hardness and contain low levels of As (
