Anal. Chem. 2002, 74, 6126-6130

Correspondence

Photoelectrochemical Oxygen Sensor Using Copper-Plated Screen-Printed Carbon Electrodes Jyh-Myng Zen,* Yue-Shian Song, Hsieh-Hsun Chung, Cheng-Teng Hsu, and Annamalai Senthil Kumar

Department of Chemistry, National Chung-Hsing University, Taichung 402, Taiwan

We report here an efficient photocatalytic amperometric sensor for the determination of dissolved oxygen (DO) in phosphate buffer solution using a disposable copperplated screen-printed carbon electrode (CuSPE). The photoelectrochemical activity toward DO of the CuSPE was related to the formation of a p-type semiconductor CuI2O. The solution pH and biased potential (Ebias) were systematically optimized as pH 8 PBS and -0.7 V vs Ag/ AgCl, respectively. Under optimized conditions, the calibration plot was linear in the range of 1-8 ppm with sensitivity and regression coefficient of 23.51 (µA cm2)-1 ppm-1 and 0.9982, respectively. The reproducibility of the system was good with seven successive measurements of DO yielding a RSD value of 1.87%. Real sample assays for groundwater and tap water were also consistent with those measured by a commercial DO meter. The principle used in DO measurement has an opportunity to extend into various research fields. Dissolved oxygen (DO) measurement is important in the fields of biochemical, fermentation control, food production and storage, environmental monitoring, and industrial applications.1-5 Among the various detection routes, electrochemical methods have received much attention in numerous applications since the introduction of the Clark sensor.6 Various detection systems based on oxide catalysts and macrocylic complex (metal-porphyrins or metal-phthalocyanines)-oriented mediators and organic and inorganic dye-based photochemical-quenching systems were reported for oxygen sensing.7-12 Nevertheless, most of the approaches are relatively expensive or have unsatisfactory elec* To whom correspondence should be addressed. Phone: (+886) 4-22854007. Fax: (+886) 4-2286-2547. E-mail: [email protected]. (1) Chaudhury, R. R.; Sorbrinho, J. A. H.; Wright, R. M.; Sreenivas, M. Water Res. 1998, 32, 2400-2412; Abstr. 5565329. (2) Ohashi, M.; Osamu, A. Biotechnol. Adv. 1997, 15, 463. (3) Hua, Q.; Shimizu, K. J. Biotechnology 1999, 68, 135-147. (4) Culberson, S. D.; Peidrahita, R. H. Ecol. Modell. 1996, 89, 231-258. (5) Numata, M.; Funazaki, N.; Ito, S.; Asano, Y.; Yano, Y. Talanta 1996, 43, 2053-2059. (6) Clark, L. C. Electrochemical Device for Chemical Analysis. U.S. Patent 2,913,386, 17 November, 1959. (7) Zen, J.-M.; Wang, C.-B. J. Electroanal. Chem. 1994, 368, 251-256 and references therein. (8) Oritz, J.; Gautierm, J. J. Electroanal. Chem. 1995, 391, 111-118. (9) D’Souza, F.; Hsieh, Y.-Y.; Deviprasad, G. R. J. Electroanal. Chem. 1997, 426, 17-21.

6126 Analytical Chemistry, Vol. 74, No. 23, December 1, 2002

trode modification procedures not to mention the problem associated with extension into practical application under physiological conditions. The instability of the mediator-based modified electrodes in the working matrixes can easily cause error in practical measurements. Even the Clark sensor design needs further improvement in performance to meet the new requirement in DO measurement.13,14 Recently, we noticed a profound electrochemical activity toward the oxygen reduction reaction in the detection assays of H2O2 at a copper-plated screen-printed carbon electrode (designated as CuSPE).15-17 Because of the very insoluble behavior of copper ions with phosphate (Ksp ) 1.39 × 10-37),17,18 the CuSPE was found to be stable and free from electrode fouling. This is indeed a highly desirable condition for application in phosphate buffer at physiological pH.18,19 The specific oxide redox couples of CuI2O/CuIIO (C1/A1) and Cu0/CuI2O (C2/A2) are responsible for effectively mediating the current signals.15-18 It is well known that CuI2O is a p-type semiconductor with an Eg value of 1.9 eV and has good photochemical behavior.20-23 In this work, the CuSPE is utilized for the photoelectrochemical determination of DO, which offers a simple route for the quantitative detection of DO. To our knowledge, so far there is no report for the photochemical quantitative assays of DO using a semiconducting system. (10) Wijnoltz, A. L. B.; Visscher, W.; van Veen, J. A. R. Electrochim. Acta 1998, 43, 3141-3152. (11) Lalande, G.; Cote, R.; Tamilzhmani, G.; Guay, D.; Dobelet, J. P.; DignaruBailey, L.; Wengs, L. T.; Bertrands, P. Electrochim. Acta 1995, 19, 26352646. (12) Gobi, K. V.; Ramaraj, R. J. Electroanal. Chem. 1998, 449, 81-89. (13) Nei, L.; Compton, R. G. Sens. Actuators, B 1996, 30, 83-87 and references therein. (14) Sohn, B.-K.; Kim, C.-S. Sens. Actuators, B 1996, 34, 435-440. (15) Zen, J.-M.; Chung, H.-H.; Senthil Kumar, A. Analyst 2000, 125, 16331637. (16) Zen, J.-M.; Chung, H.-H.; Senthil Kumar, A. Anal. Sci. 2001, 17, i287i290. (17) Senthil Kumar, A.; Zen, J.-M. Electroanalysis 2002, 14, 671-678. (18) Zen, J.-M.; Chung, H.-H.; Senthil Kumar, A. Anal. Chem. 2002, 74, 12021206. (19) CRC Handbook of Chemistry and Physics, 72 ed.; Lide, D. R., Ed.; CRC Press: Boca Raton, FL, 1991. (20) De Jongh, P. E.; Vamaekelbergh, D.; Kelly, J. J. Chem. Commun. 1999, 1069-1067. (21) Tennakone, K.; Kumarsinghe, A. R.; Sirimanne, P. M. J. Photochem. Photobiol. A 1995, 88, 39-41. (22) Richardson, T. J.; Slack, J. L.; Rubin, M. D. Electrochim. Acta 2001, 46, 2281-2284. (23) Nair, M. T. S.; Guerrero, L.; Arenas, O. L.; Nair, P. K. Appl. Surf. Sci. 1999, 150, 143-151. 10.1021/ac020058r CCC: $22.00

© 2002 American Chemical Society Published on Web 10/29/2002

Preparation of the electrode is simple and low cost and thus allows for mass production as disposable electrodes.15-18 Systematic investigation and discussion regarding solution pH and biased potential (Ebias) were carried out to optimize the photocurrent produced. Finally, real sample application was demonstrated for the measurement of DO in water samples, and satisfactory results were obtained in comparison to the values measured by a commercial Clark-type electrode (i.e., DO meter). EXPERIMENTAL SECTION Chemical and Reagents. All chemicals were obtained from Merck (Darmstadt, Germany) in analytical grade. A 1000 ppm Cu(II) solution in 0.1 M nitric acid was used for the platting experiments. The other standard solutions used in interference studies were also obtained from Merck. Unless otherwise stated, a pH 8 phosphate buffer solution (PBS) of ionic strength of 0.1 M was used in all experiments. Working solutions were prepared using double-filtered deionized water. Real water samples were collected from the campus of Chung-Hsing University, Taiwan. Apparatus. Voltammetric and photoelectrochemical measurements were carried out with a CHI model 660 electrochemical workstation (Austin, TX). The three-electrode system consisted of the CuSPE working electrode, an Ag/AgCl reference electrode, and a platinum auxiliary electrode. In situ photochemical experiments were performed using a BAS working cell in combination with a 250-W light source of an overhead projector. All instruments were turned on 10 min before start of the experiments to attain equilibrium with environment. Analytical grade O2/N2 gas was used for the oxygenation/deaeration of the working solution, respectively. For the preparation of a saturated oxygen solution, O2 gas was continuously purged for at least for 30 min and immediately transferred to the working system. Design and Fabrication of the CuSPE. A semiautomatic screen printer was used to prepare disposable SPE as per our earlier report.15 The SPE has a working area of 0.13 cm2 with average film resistance of 85.64 ( 2.10 Ω/cm. As to the preparation of the CuSPE, a Cu layer was electrochemically plated on SPE in 200 mg/L Cu(NO3)2 aqueous solution at -0.7 V versus Ag/AgCl for 300 s. Procedure. The CuSPE was first washed thoroughly with deionized water and then dipped into working solution for subsequence electrochemical and photoelectrochemical experiments. Before the experiments, the CuSPE was pretreated in its base electrolyte in the potential region of -0.8 to +0.5 V at a scan rate of 20 mV/s for five continuous runs. The photocurrent experiments were performed by chronoamperometric (CA) technique. The CuSPE was equilibrated in the blank buffer at an optimized potential of -0.7 V until the current become constant. Normally it took ∼500 s. The quantification of oxygen was achieved by measuring the reduction photocurrent from the CA signals. All experiments were performed at room temperature (25 °C). RESULTS AND DISCUSSION Photoelectrochemical Behavior of DO at the CuSPE. The photoelectrochemical response of the CuSPE at Ebias ) -0.7 V in pH 8 PBS with various amounts of DO under light irradiation is shown in Figure 1. As can be seen, the photocurrents observed are proportional to the amount of DO. Our previous study reported

Figure 1. Photoelectrochemical response of the CuSPE in (a) N2saturated, (b) normal air, and (c) O2-saturated pH 8 PBS with a biased potential (Ebias) of -0.7 V vs Ag/AgCl.

Figure 2. Cyclic voltammetric response of the CuSPE in (a) N2saturated, (b) normal air, and (c) O2-saturated pH 8 PBS at v ) 5 mV/s.

the CuSPE for the determination of H2O2 by flow injection analysis (FIA) at ambient temperature without deoxygenation.15-17 The cyclic voltammogram of the CuSPE in pH 7.4 PBS showed the growth of CuIIO and CuI2O. A well-defined reduction signal corresponding to the mediations of CuIIO and CuI2O appeared in the presence of H2O2. The mechanistic study revealed that the reduction is a coupled chemical reaction mechanism with the operation of pseudo-first-order kinetics on the concentration of Cu2O.15 In this study, cyclic voltammograms, as shown in Figure 2, give evidence for the effect of DO and H2O2 on the CuSPE. For deaerated pH 8 PBS, an anodic shoulder at -0.12 V and a couple of distinct peaks at -0.2 (C1) and -0.3 V (C2) were observed at the CuSPE with a scan rate of 5 mV/s. Similar voltammetric behavior was observed for pH 8 PBS in normal air except with an increase in cathodic current at C1 and C2. A purge with oxygen to pH 8 PBS causes a further increase in current response, indicating the effective DO response on the CuSPE. The increase in current responses at C1 and C2 are due to the electron mediation by the specific redox reactions of CuI2O/CuIIO and Cu0/CuI2O, respectively.15 Meanwhile, the increase in current Analytical Chemistry, Vol. 74, No. 23, December 1, 2002

6127

Figure 3. (A) Photocurrent density (PCD) and anodic peak area against Ebias. (B) Cyclic voltammetrograms of starting light on at different potentials for anodic scan direction (v ) 50 mV/s). Inset: PCD against anodic peak area.

due to oxygen reduction reaction at -0.6 V again clearly indicates the increasing amount of DO in solution purging with oxygen. To understand the photoelectrochemical behavior, the effect of Ebias (-0.3 to -0.9 V) on the photocurrent response of DO at the CuSPE was first studied. Similar to the experiments in Figure 1, the results obtained by individual photoelectrochemical chronoamperometric measurements are given in Figure 3A. As can be seen, the photocurrent signal starts to occur at -0.3 V, increases regularly with the increase in Ebias negatively, and eventually gets saturated around -0.7 V. Meanwhile, it is interesting to note that by turning the light on at different Ebias at a constant anodic scan rate of 50 mV/s, the anodic peak generated on the CuSPE was also found to increase as the light-on potential was <-0.3 V. As shown in Figure 3B, a sharp anodic response starts to appear by turning the light on at Ebias < -0.3 V and reaches a maximum value at -0.7 V. Since potential and time are coupled, the effect may be independent of light-on potential and dependent on time; controlled experiments on increasing the Ebias from -0.7 to 0 V with the same amount of illumination time were thus done by adding a suitable light-on holding time in each Ebias. The fact that no proportional increase in anodic peak area was observed in controlled experiments completely rules out the time factor and further confirms that the photoelectric response was indeed related to the Ebias and in turn to CuI2O formation. Most important, the anodic peak area formed under various Ebias is consistent with the DO photocurrent signal (inset Figure 3A). Based on the above observation, the occurrence of the photoelectrochemical current can be understood from the mechanism as proposed in Scheme 1. Following are the three major steps proposed in the overall photochemical reduction of DO: 1. H2O2 must be produced on the CuSPE from the reduction of DO by a potential negative enough (<-0.3 V vs Ag/AgCl) as shown in eq 1 of Scheme 1A. 6128

Analytical Chemistry, Vol. 74, No. 23, December 1, 2002

Scheme 1a

a (A) Reaction pathway for the photoelectrochemical reduction of oxygen at the CuSPE. (B) Schematic view for the photochemical reduction of oxygen. EVB, ECB, and Eg correspond to the energy levels of valance band, conduction band, and band gap, respectively.

2. The H2O2 produced in the above step can chemically oxidize the Cu0 f CuI2O without the aid of light (eq 2 in Scheme 1A).15 3. Under light-on conditions, the electron-hole pair can be generated in valance band (VB) and conduction band (CB) as CuI2O* and then participates in the DO reduction reaction (eq 2 in Scheme 1A). Under this situation, the CuI2O* can be stabilized

Figure 4. Typical PCD variation (%, with respect to pH 8 response) against solution pH (O2 saturated) at the CuSPE.

with high concentration and further resulted in the increase in anodic peak area as shown in Figure 3B. Scheme 1B shows the schematic mechanism for the photoelectrochemical reduction of DO at p-CuI2O in the CuSPE/ electrolyte interface. Under light irradiation (hv > Eg), the holes in VB move toward the bulk semiconductor to generate a reductive photocurrent and the excited electrons in CB move to the electrode surface to reduce DO. The conduction band edge potential of p-CuI2O is reported to be -1.2 V versus SCE.20 The calculated oxygen reduction potential at pH 8 is -0.01 V versus Ag/AgCl, which is ∼1.17 V more positive than the edge bond potential of the p-CuI2O. Hence, the existed photoelectrons in VB can readily reduce the oxygen to H2O2. Note that the p-CuI2O form of copper has been considered a suitable candidate for the photoelectrochemical reaction than that of passive CuIIO film.20,24 Since H2O2 is essential to the above proposed mechanism, peripheral solution-phase photoelectrochemical experiments in the presence of acidified 0.01 M KMnO4 solution (as an oxidant for H2O2) were done to confirm the formation of H2O2. According to the reaction, 2KMnO4 + 5H2O2 + 3H2SO4 f K2SO4 + 2MnSO4 + 8H2O + 5O2, the disappearance rate of the pink color from KMnO4 is taken as a qualitative parameter for H2O2 formation. In initial experiments with N2-saturated acidified KMnO4 solution at Ebias ) -0.7 V under the light-on condition, the pink color can be sustained for more than 20 min and starts to fade slowly after that. This is as expected since no H2O2 or other oxidant is generated under this condition. For the same experiment in O2saturated solution, the pink color disappeared in just ∼2 min. Controlled experiments with O2-saturated solution in the absence of any applied potential with the light-on condition yielded a response similar to the first case. These observations suggest that the formation of H2O2 through oxygen reduction at the CuSPE can be quickly reduced by MnO4-. It also confirms the operation of eq 1 of Scheme 1A on the oxygen reduction reaction at the CuSPE. Of course, further detailed physical experiments are needed to specify the exact pathways of H2O2 in the working system. Analytical Application. To optimize the experimental conditions for DO measurement, the effect of solution pH (7-12) under (24) Vazquez, M. V.; de Sanchez, S. R.; Calvo, E. J.; Schiffrin, D. J. J. Electroanal. Chem. 1994, 374, 179.

Figure 5. Continuous photocurrent response of CuSPE in pH 8 PBS at normal air condition. Other experimental conditions are as in Figure 1.

Figure 6. (A) Amperometric diagram of PCD at the CuSPE. (B) PCD at the CuSPE and [O2] measured simultaneously from a commercial DO meter against increased oxygenation time of pH 8 PBS. Inset: typical calibration plot at the CuSPE.

a fixed Ebias of -0.7 V was first investigated (Figure 4). A sharp increase in the photocurrent response is observed up to pH; 8 after that, the response decreased. About a 3 times increase in the photocurrent is noticed in pH 8 PBS in comparison to the response at pH 12. Under this condition, electrode regeneration was checked and seven successive measurements of DO in normal air resulted in a small RSD of 1.87% (Figure 5). Extended assays with different CuSPEs yielded a RSD value ∼3% (n ) 5). The calibration graph was constructed by measuring the photocurrent signals at different [DO] and parallel measurements with a Analytical Chemistry, Vol. 74, No. 23, December 1, 2002

6129

Table 1. Interference Studies on the Photoelectrochemical Detection of DO at the CuSPE in pH 8 PBS

Table 2. DO Measurements in Water Samples [O2] (ppm) sample

this study

Cyberscan DO meter

relative errora (%)

groundwater tap water

2.56 3.36

2.68 3.56

4.5 5.6

peak current ratio (%)a

a

compound

with 10 µM

with 100 µM

NaHCO3 NaClO4 Na2SO4 KNO3 KCl SDS Triton X-100

-6.7 -9.7 -6.4 +3.7 +2.2 -1.0 -1.0

-10.3 -14.5 -8.1 +5.3 +3.4 -2.3 -3.0

Experiments with O2-saturated solution.

commercial DO meter were simultaneously checked. Figure 6 shows the amperometric diagram (Figure 6A) and typical photocurrent plot (Figure 6B) for increasing [DO] in pH 8 PBS at Ebias ) -0.7 V. The observed trend between the photocurrent signals and the DO values are in good agreement. The calibration plot (inset Figure 6B) was linear in the range of 1-8 ppm DO with sensitivity and regression coefficient of 23.51 (µA/cm2)/ppm and 0.9982, respectively. Interference effects have been checked with dissolving common ions such as Cl-, HCO3-, ClO4-, SO42-, and NO3- and some surfactants in the working solution. The data obtained are summarized in Table 1. As can be seen, most ions with a concentration as high as 100 µM resulted in acceptable tolerance on the DO-detecting photocurrent signals. The interference from SO42-, HCO3-, and ClO4- can also be largely reduced as concentration is lower than 10 µM. Finally, real sample DO analysis is carried out for groundwater and tap water samples. The CuSPE yielded the DO values of 2.56 and 3.36 ppm for groundwater and tap water samples, respectively. Comparison testing with the DO meter showed acceptable difference of 4.5% and 5.6%, respectively (Table 2). Further lifetime DO measurements under normal air

6130 Analytical Chemistry, Vol. 74, No. 23, December 1, 2002

a

Take the DO value measured by Cyberscan DO meter as standard.

showed a highly stable response in the first hour, and after that, it starts to decrease slowly. Nevertheless, since the working electrodes are disposable in nature, it has a clear advantage for practical applications in diverse fields of research. CONCLUSION The copper-plated screen-printed carbon electrode showed effective photochemical reduction behavior for DO at neutral pH. The p-CuI2O species of the CuSPE is considered to be essential for the photoelectrochemical activity. The analytical applications for DO measurement show good linearity with appreciable regression coefficient comparable with that of recent DO reports. The results obtained from real water samples using the proposed method also showed values close to the commercial DO meter. Overall, this method offers simple and easy DO measurements with good reproducibility. Further work is in progress to combine flow injection-based photochemical analysis with high dead working volume. ACKNOWLEDGMENT The authors gratefully acknowledge financial support from the National Science Council of the Republic of China. Received for review September 17, 2002. AC020058R

January

29,

2002.

Accepted