Anal. Chem.
2007, 79, 7831
-
7837
†
†
‡
,†
In soluble forms, arsenic occurs as trivalent arsenite [As(III)] and pentavalent arsenate [As(V)]. Besides its wide occurrence in nature as arsenolite and claudetite (As
2
O
3
), arsenic is a byproduct of metal smelting.
1 Arsenite has attracted much attention since recognition in the 1990s of its wide occurrence in well water in
Bangladesh, India, and China.
2a As(III) is ∼ 50% of the total arsenic
(10
-
1000 ppb) in the groundwater of many wells in West Bengal,
India.
2b Groundwater in Taiwan may contain between 400 and 800 ppb.
Accurate measurement of arsenic in drinking water requires sophisticated and expensive techniques (e.g., atomic absorption
* To whom correspondence should be addressed. E-mail: john.luong@cnrcnrc.gc.ca.
† Biotechnology Research Institute.
‡ UCL.
(1) United Nations Synthesis Report on Arsenic in Drinking Water. Chapter 1
(Source and behaviour of arsenic in natural waters). http://www.who.int/ water_sanitation_health/dwq/arsenicun1.pdf.
(2) (a) Rashid, M. H.; Mridha, A.K.
Arsenic contamination in groundwater in
Bangladesh . 24th WEDC Conference, Sanitation and Water for All, Islamabad, Pakistan. 1998; pp 162 165. (b) Chatterjee, A.; Das, D.; Mandal, B. K.;
Chowdhury, T. R.; Samanta, G.; Chakraborti, D.
Analyst 1995 , 120 , 643 -
560. (c) Chen, S. L.; Dzeng, S. R.; Yang, M. H.; Chiu, K. H.; Shieh, G. M.;
Wai, C. M.
Environ. Sci. Technol .
1994 , 28 , 877 881. (d) Abedin, M. J.;
Feldmann, J.; Meharg, A. A.
Plant Physiol.
2002 , 128 , 1120 1128. (e)
Peryea, F. J.; Kammereck, R.
Water, Air, Soil Pollut .
1997 , 93 , 243 254.
(f) Tondel, M.; Rahman, M.; Magnuson, A.; Chowhury, I. A.; Faruquee, M.
H.; Ahmad, S. A.
Environ. Health Perspect.
1999 , 107 , 727 729. (g) Arsenic and Arsenic Compounds , Environmental Health Criteria 224 (WHO): http:// www.inchem.org/documents/ehc/ehc/ehc224.htm.
10.1021/ac070766i CCC: $37.00
© 2007 American Chemical Society
Published on Web 09/18/2007 spectrometry, inductively coupled plasma (ICP), and ICP/mass spectrometry) 6 and facilities as well as trained staff. Field test kits can detect high levels of arsenic but are typically unreliable at lower concentrations of concern for human health. According to the World Health Organization, there is an urgent need to develop simple, reliable, sensitive, and inexpensive equipment for field measurement. Low-cost methods are polarography techniques, 7 cathodic stripping voltammetry, 8 and anodic stripping voltammetry
(ASV) 9 with the latter being most popular due to its low detection capabilities and simple operations. Among different electrode materials, glassy carbon (GC) electrodes have been modified by gold nanoparticles to detect arsenite as low as 10 ppb using ASV
(linear sweep or square wave).
10 However, this technique suffers two major setbacks: lengthy analysis time (15 20 min for the deposition of arsenite) and severe interference caused by copper, another common metal found in natural deposits and groundwater.
Indeed, this interference cannot easily be circumvented since there are several metals and ion species, commonly found in the groundwater, that can be codeposited and stripped off under this operating procedure. Direct oxidative determination of arsenite by anodic linear sweep voltammetry can be realized using a Pt nanoparticle-modified, boron-doped diamond electrode to circumvent well-known interferences encountered in conventional stripping voltammetry.
11a Similarly, iridium-modified, boron-doped
(3) Abernathy, C.; Calderson, R. L.; Chappel, W. R.
Arsenic exposure and health effects ; Elsevier Science Ltd.: London, 1999.
(4) (a) Mukhopadhyay, R.; Rosen, B. P.; Phung, L.; Silver, S.
FEMS Microbiol.
Rev.
2002 , 26 , 311 325. (b)Wysocki, R.; Che´ry, C. C.; Wawrzycka, D.; Van
Hulle, M.; Cornelis, R.; Thevelein, J. M.; Tama´s, M. J.
Mol. Microbiol.
2001 ,
40 , 1391 1401.
(5) Arsenic and clarification to compliance and new source monitoring rule: a quick reference guide ; USEPA 816-F-01-004, Office of Water (4606) 2001; http://www.epa.gov/safewater/arsenic/pdfs/quickguide.pdf.
(6) (a) Story, W. C.; Caruso, J. A.; Heitkemper, D. T.; Perkins, L.
J. Chromatogr.
Sci.
1992 , 30 , 427 432. (b) Shum, S. C. K.; Neddersen, R.; Houk, R. S.
Analyst 1992 , 117 , 577 582.
(7) Kaye, S.
Am. J. Clin. Pathol .
1944 , 14 , 83 85.
(8) (a) Sadana, R. S.
Anal. Chem.
1983 , 55 , 304 307. (b) Li, H.; Smart, R. B.
Anal. Chim. Acta 1996 , 325 , 25 32. (c) Greulach, U.; Henze, G.
Anal. Chim.
Acta 1995 , 306 , 217 223.
(9) (a) Kopanica, M.; Novotny, L.
Anal. Chim. Acta 1998 , 368 , 211 218. (b)
Forsberg, G.; O’Laughlin, J. W.; Megargle, R. G.; Koirtyohann, S. R.
Anal.
Chem.
1975 , 47 , 1586 1592. (c) Sun, Y.-C.; Mierzwa, J.; Yang, M.-H.
Talanta 1997 , 44 , 1379 1387. (d) Davis, P. H.; Dulude, G. R.; Griffin, R.
M.; Matson, W. R.; Zink, E. W.
Anal. Chem .
1978 , 50 , 137 143. (e) Simm,
A. O.; Banks, C. E.; Compton, R. G.
Anal. Chem .
2004 , 76 , 5051 5055.
(10) Dai, X.; Nekrassova, O.; Hyde, M. E.; Compton, R. G.
Anal. Chem.
2004 ,
76 , 5924 5929.
(11) (a) Hrapovic, S.; Liu, Y.; Luong, J. H. T.
Anal. Chem.
2007 , 79 , 500 507.
(b) Ivandini, T. A.; Sato, R.; Makide, Y.; Fujishima, A.; Einaga, Y.
Anal. Chem.
2006 , 78 , 6291 6298.
Analytical Chemistry, Vol. 79, No. 20, October 15, 2007 7831

diamond electrodes fabricated by an ion implantation method can also be used for electrochemical detection of arsenite.
11b
Arsenite can be oxidized to less toxic arsenate chemically or microbially (As III O
2
+
2H
2
O f
As V O
4
3
+
4H
+ +
2e
-
).
12 The arsenite-oxidizing bacteria can either gain energy from arsenite oxidation 13 or have been proposed to do so as part of a detoxification process.
14 Chemolithoautotrophic arsenite oxidation, for which oxygen is used as the terminal electron acceptor, arsenite as the electron donor, and carbon dioxide as the sole carbon source, has been reported for microorganisms isolated from gold mines.
13a,b Aerobic growth with arsenite as the electron donor is exergonic, generating a substantial amount of free energy.
13a
In order to combine the specificity of enzymatic oxidation of arsenite oxidase and the sensitivity of electrochemical detection, this paper describes a biosensor for arsenite using molybdenumcontaining arsenite oxidase, prepared from the chemolithoautotroph NT-26 that oxidizes arsenite to arsenate. The bacterium is resistant to high levels of arsenate, as its growth is not affected by arsenate concentrations up to 400 mM (Clarke, A.; Santini, J.
M., unpublished data). The enzyme is electrochemically deposited on a multiwalled carbon nanotube (MWCNT)-modified electrode to form a sensitive and specific biosensor for analysis of arsenite.
EXPERIMENTAL SECTION
Growth of NT-26 . The Rhizobium sp. str. NT-26 was isolated from a gold mine in the Northern Territory, Australia.
13a NT-26 was grown aerobically at 28 ° C in a minimal salts medium, containing 5 mM arsenite and 0.04% yeast extract (Oxoid) as described by Santini et al.
13a The culture was harvested during late exponential phase at a final absorbance of 0.153 at 600 nm.
For purification of the enzyme, NT-26 was grown in a 7-L batch culture with shaking to provide a total cell yield of 2.5 g (wet weight).
Purification of Arsenite Oxidase (Aro) . The enzyme was purified as described previously 15 except that the gel filtration buffer (50 mM MES, pH 5.5) also contained 100 mM NaCl. The enzyme was stored at 80 ° C as separate 10µ L aliquots at a
(12) (a) Inskeep, W.P.; McDermott, T. R.; Fendorf, S. Arsenic(V)/(III) cycling in soils and natural waters: chemical and microbiological processes compounds. In Environmental chemistry of arsenic; Frankenberger, W. T.,
Ed.; Marcel Dekker Inc.: New York, 2002; pp 183 215. (b) Ehrich, H. L.
Bacterial oxidation of As(III) compounds. In Environmental chemistry of arsenic; Frankenberger, W. T., Ed.; Marcel Dekker Inc.: New York, 2002; pp 313 327.
(13) (a) Santini, J. M.; Sly, L. I.; Schnagl, R. D.; Macy, J. M.
Appl. Environ.
Microbiol .
2000 , 66 , 92 97. (b) Santini, J. M.; Sly, L. I.; Wen, A.; Comrie,
D.; De Wulf-Durand, P.; Macy, J.M.
Geomicrobiol. J .
2002 , 19 , 67 76. (c)
Oremland, R. S.; Hoeft, S. E.; Santini, J. M.; Bano, N.; Hollibaugh, R. A.;
Hollibaugh, J. T.
Appl. Environ. Microbiol.
2002 , 68 , 4795 4802. (d)
Battaglia-Brunet, F.; Joulian, C.; Garrido, F.; Dictor, M.-C.; Morin, D.;
Coupland, K.; Johnson, D. B.; Hallberg, K. B.; Baranger, P.
Antonie van
Leeuwenhoek 2005 , 8 , 1 10. (e) Rhine, E. D.; Phelps, C. D.; Young, L. Y.
Environ. Microbiol .
2006 , 8 , 899 908.
(14) (a) Gihring, T. M.; Banfield, J. F.
FEMS Microbiol. Lett .
2001 , 204 , 335 -
340. (b) Gihring, T. M.; Druschel, G. K.; Mccleskey, R. J.; Hamers, R. J.;
Banfield, J. F.
Environ. Sci. Technol .
2001 , 35 , 3857 3862. (c) Phillips, S.
E.; Taylor, M. L.
Appl. Environ. Microbiol.
1976 , 32 , 392 399. (d) Salmassi,
T. M.; Venkateswaren, K.; Satomi, M.; Nealson, K. H.; Newman, D. K.;
Hering, J. G.
Geomicrobiol. J.
2002 , 19 , 53 66. (e) Weeger, W.; Lievremont,
D.; Perret, M.; Lagarde, F.; Hubert, J.-C.; Leroy, M.; Lett, M.-C.
BioMetals
1999 , 12 , 141 149. (f) Donahue-Christiansen, J.; D’Imperio, S.; Jackswon,
C. R.; Inskeep, W. P.; McDermott, T. R.
Environ. Microbiol.
2004 , 70 , 1865 -
1868. (g) Salmassi, T. M.; Walker, J. J.; Newman, D. K.; Leadbetter, J. R.;
Pace, N. R.; Hering, J. G.
Environ. Microbiol .
2006 , 8 , 50 59.
(15) Santini, J. M.; vanden Hoven, R. N.
J. Bacteriol.
2004 , 186 , 1614 1619.
7832 Analytical Chemistry, Vol. 79, No. 20, October 15, 2007 concentration of 20 µ M. Bradford reagent 16 was used to determine protein concentrations and bovine serum albumin served as the standard. The native molecular mass of Aro was determined by gel filtration chromatography through a Superdex 200
(Amersham Pharmacia Biotech) column as described previously.
17
Materials and Chemicals.
MWCNTs (95% purity, 10
-
20 nm in diameter, 1
-
5 µ m in length) were obtained from NanoLab
(Brighton, MA). Arsenic atomic spectroscopy standard solution
(1000 ppm) in nitric acid (Fluka/Sigma-Aldrich, St. Louis, MO) was used as a stock solution and diluted with buffer to obtain appropriate concentrations of arsenite. All other chemicals were of analytical grade (Sigma-Aldrich) and all solutions were prepared using Milli-Q (Millipore, Bedford, MA) A-10 gradient (18 M Ω
‚ cm) deionized water.
Instrumentation.
Cyclic voltammetry (CV) and amperometric measurements were performed using an electrochemical analyzer
(CHI 601A, CHI Instruments, Austin, TX). Surface characterization was analyzed using CV mode, whereas analyte detection was performed by chronoamperometry with 50 mM phosphate pH 7, containing 10 mM NaCl. A Pt wire (Aldrich, 99.9% purity, 1-mm diameter) and an Ag/AgCl (3 M NaCl) electrode (BAS, West
Lafayette, IN) were used as counter and reference electrodes, respectively. GC (3 mm in diameter, BAS) modified electrodes served as the working electrode. The volume of the electrochemical stir cell was 10 mL. SEM images were obtained using a Hitachi scanning electron microscope (S-2600 N, Tokyo, Japan), operating in high vacuum mode at an acceleration voltage of 8
-
25 kV with a working distance of ∼ 10 mm. The entire GC electrode was inserted into the SEM chamber and grounded with copper adhesive tape (3M, St. Paul, MN) to reduce ionization. Elemental analysis including the total arsenic concentration was performed using an X Series 2 ICPMS system (Thermo Electron Corp.,
Winsford, UK) using the procedure recommended by the manufacturer (Application note 40741).
18 Fluorescence of humic acid and river water was measured using a Gilford Fluoro IV spectrofluorometer (Gilford, Oberlin, OH) with a detector photomultiplier tube voltage set at 750 V.
Electrode Preparation and CNT Film Formation.
GC working electrodes were first cleaned with wet silicon carbide paper, grit 1500 (Hand American Made Hardwood Products, South
Plainfield, NJ), followed by polishing with 0.05µ m alumina slurry
(Buehler, Markham, ON, Canada) on velvet using a model 900 grinder/polisher (South Bay Technol., San Clemente, CA). After being rinsed thoroughly with deionized water, the electrodes were sonicated for 5 min to remove excess alumina followed by a 5-min treatment in piranha solution (70% sulfuric acid and 30% hydrogen peroxide). After sonication for another 5 min in deionized water, the conditioned working electrodes were dried under nitrogen and used for modification. For electrode preparation, MWCNTs
(2 mg/mL) were dissolved in N ′ N -dimethylformamide (A&C
American Chemicals, Montreal, QC, Canada) and sonicated for
∼ 2 h to obtain well-dispersed MWCNTs. The MWCNT slurry
(20 µ L) was applied to the GC electrode surface and air-dried for at least 2 h to form a uniform film.
(16) Bradford, M. M.
Anal. Biochem.
1976 , 72 , 248 254.
(17) Krafft, K.; Macy, J. M.
Eur. J. Biochem.
1998 , 255 , 647 653.
(18) Nash, M.; McSheehy, S. Speciation of arsenic in fish tissues using HPLC coupled with XSeriesII ICP-MS. Thermo Electron Corp., Application note
40741, Winsford, UK.

Deposition of Aro and Analyte Detection.
Aro was deposited
(CHI 760B, CHI Instruments) galvanostatically 19 by applying a constant current of 10 µ A for 10 min to MWCNT-modified GC electrodes. Cyclic voltammograms of the modified electrode surface obtained with and without the enzyme were compared at different scan rates to illustrate the direct electron transfer between the enzyme and the electrode. Optimization of the pH was performed between 5 and 9 using phosphate and 2-( N morpholino)ethanesulfonic acid (MES) buffer. The potential applied to the electrode surface as well as the enzyme concentration (0.2
6.7
µ M in MES buffer containing 10 mM NaCl, pH 6.0) was optimized. Under these optimal conditions, a calibration curve was established for arsenite (1 1000 ppb) to determine the detection limit and response time. Reproducibility and stability of the arsenite oxidase-based biosensor were determined at 25 and 350 ppb arsenite. Data smoothing was performed for low arsenite concentrations ( < 50 ppb) using the software provided by CHI Instruments. For comparison, GC electrode surfaces without arsenite oxidase (MWCNTs only), as well as surfaces without MWCNTs (arsenite oxidase only), were prepared and analyzed for arsenite detection.
Potential Interference by Copper and Analysis of Water
Samples.
The arsenite oxidase biosensor was analyzed in the presence of copper (sulfate or chloride) in the range of 1
-
500 ppb to verify potential interferences. Known arsenite concentrations in a mixture of arsenite and copper were then determined.
Four water samples, tap water, mineral water (Perrier), SLRS-4
Ottawa river water reference material for trace metals and total arsenic (courtesy of Dr. R.E. Sturgeon, Institute for National
Measurement Standards, National Research Council Canada,
Ottawa, Ontario, Canada), and river water from the Lachine Canal of the St. Lawrence River (Montreal, QC, Canada) were analyzed for arsenite using the biosensor. The samples were then spiked with a known concentration of arsenite (40 ppb arsenite to give
20 ppb after admixing) and evaluated. In these studies, 5 mL of the water sample spiked with 5% buffer (20 × , 1 M NaPO
4
,
200 mM NaCl) was added into the stir cell containing 5 mL of buffer. This step maintains the ionic composition in the stirred cell, thus avoiding large current fluctuations.
RESULTS AND DISCUSSION
Purification and Some Characteristics of NT-26 Aro.
NT-
26 Aro was purified from the periplasmic fraction, followed by ammonium sulfate precipitation, hydrophobic interaction, and gel filtration chromatography, which led to a 60-fold enrichment of the enzyme. After the final purification step, with purity of
>
99%, the molecular mass of the enzyme, based on gel filtration chromatography, was estimated to be 219 kDa (data not shown).
According to Santini and vanden Hoven, 15 Aro consists of two heterologous subunits, 98 (AroA) and 14 kDa (AroB), and the subunits are present in a 2:2 stoichiometry, i.e., an
R
2
β
2 configuration. With 2, 4-dichlorophenolindophenol as the artificial electron acceptor, NT-26 Aro exhibited a V max of 2.4
µ mol of arsenite oxidized/min ‚ mg of protein, corresponding to an enzyme turnover of 8.6 s -
1 . The K
M for arsenite was determined to be 61 µ M. The purified Aro protein was reported to possess 2.02
( 0.08 mol of
(19) Li, C.-M.; Sun, C. Q.; Chen, W.; Pan, L.
Surf. Coat. Technol .
2005 , 198 ,
474 477.
Mo and 9.2
(
0.6 mol of Fe per mol of enzyme (
R
2
β
2
).
15 Therefore,
Aro contains molybdenum as a cofactor and iron
sulfur clusters as prosthetic groups.
Electrochemistry and Direct Electron Transfer Using
Carbon Nanotubes.
To date, only Alcaligenes faecalis arsenite oxidase (100 kDa,
R
β configuration) has been crystallographically characterized. The enzyme consists of three distinct electrontransfer centers: a Mo active site and an adjacent high-potential
[3Fe
-
4S] cluster within the
R unit (825 residues) and a Riesketype [2Fe
-
2S] center in the β unit (134 residues).
20 The Mo active sites bear two molybdenum guanine dinucleotide (MGD) ligands, which each chelates the metal ion via a pair of dithiolene S donors.
The Mo active site is in its tetravalent (inactive) oxidation state,
Mo(IV) with a square-based pyramidal coordination geometry. In its active hexavalent state [Mo(VI)], a six-coordinate Mo VI O(X)-
(MGD)
2
(X ) O or OH) has been proposed.
20 In this respect,
Aro falls in the dimethyl sulfoxide reductase family of enzymes.
20
With a special arrangement, electrons egress from the Mo ion after oxidation of arsenite and proceed first via the [3Fe 4S] cluster and then via the Rieske-type [2Fe 2S] cluster.
NT-26, a member of R Proteobacteria , is not related to A.
faecalis , a member of β Proteobacteria .
21
4S] + /0 couple, and the Mo active site.
22
However, the R subunits of the two arsenite oxidases from both microorganisms are ∼ 50% identical, and NT-26 Aro has been suggested to comprise a singleelectron Rieske-type [2Fe
-
2S] 2 + / + couple, a one-electron [3Fe
-
The redox potential of the cofactors in NT-26 Aro could not be determined by cyclic voltammetry since the enzyme is not capable of undergoing direct heterogeneous electron transfer with the pyrolytic graphite electrode.
23 This behavior could be attributed to a poor coverage of the electroactive enzyme or the distribution of orientations of the enzyme on the electrode surface. In their study, the enzyme loading was 0.26 mg
‚ cm
-
2 and the enzyme was codeposited with polymixin, whereas in this study the enzyme loading on the
MWCNT-modified electrode was similar (0.19 mg
‚ cm
-
2 ).
The MWCNT-modified electrode with immobilized Aro displayed an almost symmetrical CV with equal reduction and oxidation peak heights (355 and 415 mV) at increasing scan rates
(Figure 1, curves b e), such that the electrode enabled direct electron transfer, i.e., effecting reduction and then reoxidization of the enzyme without an artificial electron-transfer mediator. The linear relationship between peak oxidation or reduction current and the square root of the scan rate (50 125 mV ‚ s -
1 , bottom right inset) indicated a surface-controlled quasi-reversible process.
Without the enzyme, no redox activity was observed (Figure 1, curve a, top right inset), and this was the case at higher scan rates (75 125 mV ‚ s 1 ). Using a bare GC electrode (no MWCNTs), redox activity was not observed even if the amount of enzyme loading was up to 0.75 mg ‚ cm 2 . The redox potential of [3Fe -
4S] + /0 is 260
-
270 mV, whereas a two-electron voltammetric peak from the Mo center is 292 mV (vs the NHE at pH 6.0).
24 It should be noted that the Rieske center of Aro has a nearby disulfide
(20) Ellis, P. J.; Conrads, T.; Hille, R.; Kuhn, P.
Structure 2001 , 9 , 125 132.
(21) Anderson, G. L.; Williams, J.; Hille, R. J.
Biol. Chem .
1992 , 33 , 23674 -
23682.
(22) vander Hoven, R. N.; Santini, J. M.
Biochim. Biophys. Acta 2004 , 1656 ,
148 155.
(23) Bernhardt, P. V.; Santini, J. M.
Biochemistry 2006 , 45 , 2804 2809.
(24) Hoke, K. R.; Cobb, N.; Armstrong, F. A.; Hille, R.
Biochemistry 2004 , 43 ,
1667 1674.
Analytical Chemistry, Vol. 79, No. 20, October 15, 2007 7833

and facilitating direct electron transfer to the immobilized enzyme.
Assuming the Mo active site and the two Fe
-
S clusters of NT-26
Aro are similar to those of A. faecalis , the oxidation of arsenite by the enzyme coupled with direct electron transfer can be described as follows:
Figure 1.
Cyclic voltammetry at different scan rates for (a) MWCNT/
GC (3 mm) and (b
e) MWCNT/Aro/GC electrodes. (a, b) 50, (c) 75,
(d) 100, and (e) 125 mV
‚ s
-
1 . Aro (6.7
µ M) was deposited at 10 µ A for 10 min, and CVs were run in phosphate buffer pH 7, containing
10 mM NaCl. The upper right inset shows the narrow current range, while the lower right inset shows the relationship between the redox peaks and the square root of the scan rate.
Figure 2.
SEM image of MWCNT/Aro/GC electrode. Acceleration voltage of 8 kV at a working distance of 9.8 mm, 20µ m scale. The inset indicates a film thickness of
∼
400 nm. Acceleration voltage of
25 kV at a working distance of 10.3 mm, 5µ m scale.
bridge holding the cluster binding loop tightly together. Thus, the higher potential Rieske centers are observed for this type of enzyme (100 400 mV) 25a
d as opposed to the Rieske centers found in dioxygenases (typically 100 to 200 mV).
25e
g Direct electrontransfer results have been previously reported for putrescine oxidase.
26
The SEM image (Figure 2) of the enzyme-modified electrode shows the presence of the MWCNTs on the GC surface, while the inset indicates a film thickness of ∼ 400 nm. Apparently, nanoscale “dendrites” of MWCNTs were able to form a network, projecting outward from the electrode surface, acting like bundled ultramicroelectrodes, thereby permitting access to the active site
(25) (a) Zu, Y.; Fee, J. A.; Hirst, J.
Biochemistry 2002 , 41 , 14054 14065. (b)
Iwata, S.; Saynovitis, M.; Link, T. A.; Michel, H.
Structure 1996 , 4 , 567 -
579. (c) Carrell, C. J.; Zhang, H.; Cramer, W. A.; Smith, J. L.
Structure 1997 ,
5 , 1623 1625. (d) Bonisch, H.; Schmidt, C. L.; Schafer, G.; Ladenstein, R.
J. Mol. Biol.
2002 , 319 , 791 805. (e) Link, T. A.
Ad v . Inorg. Chem.
1999 ,
47 , 83 157. (f) Couture, M-, M., J.; Colbert, C. L.; Babini, E.; Rosell, F. I.;
Mauk, A. G.; Bolin, J. T.; Eltis, L. D.
Biochemistry 2001 , 40 , 84 92. (g)
Link, T. A.; Hatzfeld, O. M.; Unalkat, P.; Shergill, J. K.; Cammack, R.; Mason,
J. R.
Biochemistry 1996 , 35 , 7546 7552.
(26) Luong, J. H. T.; Hrapovic, S,; Wang, D.
Electroanalysis 200 5, 17 , 47 53.
7834 Analytical Chemistry, Vol. 79, No. 20, October 15, 2007
Optimization of Arsenite Detection using the MWCNT/
Aro-Modified GC Electrode.
Cyclic voltammetry of the modified electrode performed in the presence of arsenite (1000 ppb) illustrated that the current increased over a wide range of applied potentials (figure not shown). The current signal was similar over a broad potential range (0.3
0.6 V) and only dropped to 50% at
0.1 V (Figure 3A). With respect to the signal-to-noise ratio (S/N
) 3), 0.3 V was used as the applied potential for further experiments to avoid endogenous electroactive interferences commonly observed at higher potentials. The current signal was similar over a broad pH range (6 8) for the immobilized enzyme biosensor and only dropped to 60% at either pH 4.5 or 9.0 (Figure
3B). In contrast, a much sharper optimum at pH 5.5 in MES buffer was reported for the soluble enzyme. MES buffer resulted in much higher background noise; hence, phosphate buffer pH 7 was used for all subsequent studies. The enzyme concentration used for galvanostatic deposition on a bare GC electrode at 10 µ A also affected the response signal for arsenite. As expected, increasing the enzyme loading increased the current signal with the maximum response observed at ∼ 6.7
µ M (Figure 3C). The Aro enzyme electrode did not respond to arsenate (up to 1 ppm), as expected since the enzyme is very specific for arsenite and As-
(V) cannot be oxidized any further. Without the enzyme, the
MWCNT-modified electrode exhibited no response to arsenite, confirming that arsenite per se was not electroactive. Passive enzyme immobilization (soaking modified electrode surface with enzyme solution for 10 min) resulted in an arsenite signal of only
25%, compared to the galvanostatic procedure. Such a result was not completely unexpected since the enzyme could interact with
CNTs via its hydrophobic domains. However, electrodeposition could produce more extensive and compact enzyme deposit with high reproducibility as suggested by Matsumoto et al.
27a As pointed out by Im et al.
27b and later concurred by Matsumoto et al., 27a the driving force for enzyme immobilization was the precipitation of the enzyme on the electrode owing to a decrease in local pH, a result of oxidation of water.
Arsenite Calibration.
A linear relationship between current response and arsenite concentration (Figure 4A) was observed up to 500 ppb, with a sensitivity of 1.43
( 0.08 nA/ppb ( n ) 9,
95% confidence interval, R 2 ) 0.996) and a detection limit of
1 ppb (Figure 4A inset). This detection limit is well within the arsenic standard for drinking water of 0.01 mg/L (10 ppb) set by
(27) (a) Matsumoto, N.; Chen, X.; Wilson, G. S.
Anal. Chem .
2002 , 74 , 362 -
367. (b) Im, D. M.; Jang, D. H.; Oh, H. M.; Striebel, C.; Wiemhoefer, H. D.;
Gaugliz, G.; Goepel, W.
Sens. Actuators, B 1995 , 24 , 149 155.

Figure 4.
(A) Arsenite standard calibration (linear portion, solid line).
The inset shows the calibration plot at low arsenite concentration.
(B) Reproducibility analysis at 25 ppb arsenite. The inset shows the typical current response curve for 25 ppb arsenite.
Figure 3.
Optimization studies for the MWCNT/Aro/GC electrode.
(A) Effect of applied potential (pH 7, 6.7
µ M Aro), (B) pH profile (0.3
V AP, 6.7
µ M Aro), and (C) effect of enzyme concentration for galvanostatic deposition (pH 7, 0.3 V AP).
the U.S. EPA.
5 The response signal for 1000 ppb arsenite was 1.8fold higher than at 500 ppb (dashed line in Figure 4A). Excellent reproducibility (2 5%, 95% confidence interval) was obtained for the detection of 12 repeated analyses with the MWCNT-modified electrode at either 25 ppb (34.0
(
0.70 nA) as shown in Figure
4B or at 350 ppb (466 ( 24 nA). The response time was 10 s when the electrode was provoked by 25 ppb arsenite (Figure 4B, inset). Without MWCNT modification, the Aro deposited GC electrode exhibited very weak response for arsenite, with a significantly higher detection limit ( ∼ 25 ppb). The modified electrodes stored in MES, pH 5.5 at 4 ° C were stable for a few days, and such behavior was reported in the literature.
22
Figure 5.
Influence of copper ion on arsenite detection (a) 20 ppb arsenite, (b) 20 ppb CuSO
4
+
, (c) 500 ppb CuSO
4
20 ppb arsenite, and (e) 500 ppb CuSO
4
+
, (d) 20 ppb CuSO
20 ppb arsenite.
4
Arsenite Determination in the Presence of Copper.
Severe interference caused by copper ion found in natural deposits and groundwater is a major drawback during the detection of arsenite by electrochemical stripping techniques such as linear sweep or square wave. This interference cannot be easily circumvented since several metals and ion species can be codeposited and stripped off under this operating condition.
11 Consequently, the major advantage of the biosensor system is that deposition is not necessary for arsenite detection. Figure 5 shows that copper ion at low levels (20 ppb) provoked no response with the biosensor
(curve b), and in the presence of 20 ppb arsenite (curve d) did
Analytical Chemistry, Vol. 79, No. 20, October 15, 2007 7835

Figure 6.
(A) Biosensor analysis of tap water (a) 20 ppb arsenite,
(b) tap water alone, and (c) tap water spiked with arsenite to give
20 ppb after mixing. (B) Biosensor analysis of Perrier mineral water
(a) 20 ppb arsenite, (b) Perrier mineral water alone, and (c) Perrier mineral water spiked with arsenite to give 20 ppb after mixing. (C)
Biosensor analysis of SLRS-4 river water (a) 20 ppb arsenite, (b)
SLRS-4 river water alone, and (c) SLRS-4 river water spiked with arsenite to give 20 ppb after mixing.
not affect the current signal (28 nA), which resembled the control signal (27 nA) for 20 ppb arsenite (curve a). Only at very high copper concentrations (500 ppb), a small negative current
( 3 nA) was observed (curve c), which slightly affected the current response (25 nA) for arsenite determination (curve e). It was reasoned that water samples would not contain such a high level of copper contamination to cause such interference.
Analysis of Arsenite in Water Samples.
The biosensor was tested for arsenite detection in tap water as shown in Figure 6A.
While no current signal was observed for the tap water alone
(curve b), the signal (27 nA) for the tap water spiked with 20 ppb arsenite (curve c) resembled that of the control signal (28 nA)
7836 Analytical Chemistry, Vol. 79, No. 20, October 15, 2007
Figure 7.
(A) St. Lawrence River water analysis with MWCNT/Aro/
GC electrode (a) 20 ppb arsenite, (b) St. Lawrence River water alone, and (c) St. Lawrence River water spiked with arsenite to give 20 ppb after mixing. (B) St. Lawrence River water analysis with MWCNT/
GC electrode (a) 20 ppb arsenite, (b) St. Lawrence River water alone,
(c) river water spiked with arsenite to give 20 ppb after mixing, and
(d) 10 ppm humic acid. The inset shows a calibration curve for arsenite in the presence of the St. Lawrence River water.
for 20 ppb arsenite (curve a). These results revealed no endogenous components in the tap water interfered with the arsenite detection. Perrier mineral water was then tested for arsenite
(Figure 6B). Similar to tap water, no current signal was observed for the Perrier water alone (curve b), whereas the signal (26 nA) for the Perrier water spiked with 20 ppb arsenite (curve c) resembled that of the control signal (28 nA) for 20 ppb arsenite
(curve a). This carbonated natural spring water has a total mineral salt content of 475 ppm, containing various ions (in ppm: 149
Ca 2
+
, 7 Mg 2
+
, 12 Na
+
, 1 K
+
, 420 HCO
3
-
, 42 SO
4
2
-
, 23 Cl
-
, 1 N,
0.1 F). Therefore, the ions listed above did not interfere with the arsenite detection and the Perrier water is certified to contain no
As, Cu, Pb, or Zn. The biosensor was also tested for arsenite detection in the reference water sample (SLRS-4), as shown in
Figure 6C. This sample contains only 0.68 ppb arsenic (As(III) and As(V)), below the detection limit of the biosensor. No current signal was observed for the SLRS-4 alone as expected (curve b), whereas the signal (27 nA) for the SLRS-4 spiked with 20 ppb arsenite (curve c) resembled that of the control signal (28 nA) for 20 ppb arsenite (curve a). The sample contains significant levels (in ppb) of aluminum (54), barium (12), copper (1.81), iron
(103), manganese (3.37), strontium (26.3), and zinc (0.93); however, such ion species caused no interference in the measurement of arsenite. For this reference river water sample, the

detection time for analysis increased from 10 (for arsenite alone) to ∼ 20
-
30 s.
River water from the St. Lawrence River was analyzed using the arsenite oxidase biosensor. As observed in Figure 7A, this sample gave a positive current response (15 nA) by itself (curve b). The river water sample spiked with 20 ppb arsenite resulted in a signal of 42 nA (curve c), such that the difference (27 nA) between curves b and c resembled that of the control signal
(28 nA) for 20 ppb arsenite (curve a). The river water sample also provoked a positive signal (curve b) of 17 nA for the enzymeless MWCNT-modified electrode (Figure 7B). As expected, the sample spiked with 20 ppb arsenite (curve c) resulted in a signal (17 nA) similar to the signal without the arsenite (curve b), since the enzymeless electrode did not respond to arsenite
(curve a) as described previously. Such results implied that arsenite was not present in the river water, since the current signal
( ∼ 17 nA) for curve b was very similar in the presence or absence of arsenite oxidase. ICPMS analysis confirmed that the level of total arsenic (As(III) and As(V)) was less than 1 ppb, below the detection limit of the biosensor. In addition, the sample contains significant levels (in ppb) of aluminum (150), barium (20), cadmium (3), chromium (2), copper (27), manganese (17) nickel
(114), and zinc (280). All these metals were tested in excess (1 ppm) and provoked no signal response. Humic acid, commonly found in freshwaters, was also tested (Figure 7, curve d), and the signal response (11 nA) observed for 10 ppm humic acid was
(28) http://www.ams.usda.gov/NOSB/cropcommRMR/HumicAcidsTechnical-
Review.pdf.
(29) (a) Lu, F. J.; Huang, T. S.; Lin, Y. S.; Pang, V. F.; Lin, S.Y.
Appl. Organomet.
Chem .
1994 , 8 , 223 228. (b) Yang, H. L.; Tu, S. C.; Lu, F. J.; Chiu, H. C.
Am. J. Hematol .
1994 , 46 , 264 269. (c) Lu, F. J.; Hsieh, H. P.; Yamauchi,
H.; Yamamura, Y.
Appl. Organomet. Chem .
1991 , 5 , 507 512.
(30) Lu, F. J.
Appl. Organomet. Chem .
1990 , 4 , 191 195.
(31) Zeng, K.; Hwang, H. M.; Yu, H.
Int. J. Mol. Sci.
2002 , 3, 1048 1057.
(32) Hirade, M.; Shima, T.; Kawaguchi, H.
Fresenius’s J. Anal. Chem.
1993 , 345 ,
780 783.
similar to the interference observed in Figure 7B (curve b). Humic acid is electroactive since the phenolic and carboxyl groups are the main functional groups of this polyacid.
28 The presence of humic materials and their connection with arsenic in groundwaters has been reported in the literature 29 and the level of total organic carbon in river water is location dependent. Note that humic acid fluoresces at 430 nm (325-nm excitation) 30 and the river water also displayed a similar profile obtained for 20 ppm humic acid
(figure not shown) whereas no fluorescence was observed for the certified river sample. Normal levels of humic acids in river water can range from 20 to 80 ppm 31 and are location dependent since many rivers in Japan contain less than a few ppm humic acid.
32
Nevertheless, the presence of electroactive interferents including humic acid would be circumvented by using a dual-electrode system, one with and one without the enzyme. Based on this approach, arsenite-spiked river water was analyzed using the biosensor, and a linear relationship between current response and arsenite concentration (Figure 7B inset) was observed up to
100 ppb, with a sensitivity of 1.45
( 0.06 nA/ppb ( n
) 7, 95% confidence interval, R 2 ) 0.998) and a detection limit of 2 ppb.
The calibration curve was similar to that reported in Figure 4 for arsenite in buffer.
In brief, a novel biosensor has been developed for arsenite using arsentite oxidase and multiwalled carbon nanotube-modified electrodes. The biosensing scheme is very specific for As(III) with a detection limit of 1 ppb. Other ions commonly found in tap and river water provoke no response. Although electroactive humic acid at ppm levels causes a noticeable background, this interference can be circumvented by using a dual-electrode system, one with and without the enzyme.
Received for review April 17, 2007. Accepted August 8,
2007.
AC070766I
Analytical Chemistry, Vol. 79, No. 20, October 15, 2007 7837