Photocatalytic Sensor for Chemical Oxygen Demand Determination

ments for preparing a calibration curve for the sensor, a DO meter. (Do-25A, TOA ... 2 ) TiO2-loaded oxygen electrode; 3 ) digital multimeter; 4 ) cha...
0 downloads 0 Views 86KB Size
Anal. Chem. 2000, 72, 3379-3382

Photocatalytic Sensor for Chemical Oxygen Demand Determination Based on Oxygen Electrode Yoon-Chang Kim, Kyong-Hoon Lee,† Satoshi Sasaki, Kazuhito Hashimoto, Kazunori Ikebukuro, and Isao Karube*

Research Center for Advanced Science and Technology, The University of Tokyo, 4-6-1 Komaba, Meguro-ku, Tokyo 153-8904, Japan

The construction and performance evaluation of a novel Chemical Oxygen Demand (COD) sensor is described. The sensor measures, using an oxygen electrode, a decrease of dissolved oxygen of a given sample resulting from photocatalytic oxidation of the organic compounds therein. As the photocatalyst, titanium dioxide (TiO2) fine particles adsorbed on a translucent poly(tetrafluoroethylene) (PTFE) membrane was used. The oxygen electrode with the membrane attached on its tip was used as the sensor probe. The operation characteristics of the sensor are demonstrated using an artificial wastewater and real water samples from lakes in Japan. This method is considered to be reliable, in that the observed parameter is close to the theoretical definition of chemical oxygen demand (COD), the amount of oxygen consumed for oxidation of organic compounds. Oxygen demand is an important parameter for assessing the concentration of organic contaminants in the water resources. Because the degradation of organic compounds requires oxygen, their concentrations can be estimated by the amount of oxygen required.1-3 When this oxidation is carried out chemically, the value obtained is called Chemical Oxygen Demand (COD). Biochemical oxygen demand (BOD) is its counterpart cited when the process is carried out using microbes.3 The conventional method for BOD determination has a tedious procedure, and its results are specific to the body of water in question. Thus, COD is preferred for estimating organic pollution,1,2,4-7 but its conven* To whom correspondence should be addressed. Tel: +81-3-5452-5220. Fax: +81-3-5452-5227. E-mail: [email protected]. † On leave from the Research Institute of Innovative Technology for the Earth/New Energy Development Organization (RITE/NEDO). (1) Lee, K.-H.; Ishikawa, T.; McNiven, S.; Nomura, Y.; Sasaki, S.; Arikawa, Y.; Karube, I. Anal. Chim. Acta 1999, 386, 211-220. (2) Balconi, M. L.; Borgarello, M.; Ferraroli, R.; Realini, F. Anal. Chim. Acta 1992, 261, 295-299. (3) Cuesta, A.; Todoli, J. L.; Canals, A. Spectrochim. Acta, Part B 1996, 51, 1791-1800. (4) Jones, B. M.; Sakaji, R. H.; Daughton, C. G. Anal. Chem. 1985, 57, 23342337. (5) Bilanovic, D.; Loewenthal, R. E.; Avnimelech, Y.; Green, M. Water SA 1997, 23, 301-309. (6) Dasgupta, P. K.; Petersen, K. Anal. Chem. 1990, 62, 395-402. (7) Pamplin, K. L.; Johnson, D. C. Electroanalysis 1997, 9, 279-283. 10.1021/ac9911342 CCC: $19.00 Published on Web 07/15/2000

© 2000 American Chemical Society

tional evaluation methods have several disadvantages8-12 such as long analysis time (2-4 h), high probability of errors due to complex procedures dependent upon the operator skill, and consumption of expensive (Ag2SO4) and toxic chemicals (Cr and Hg). Recently, the photocatalytic decomposition of organic pollutants in water has received much attention,13-16 because complete mineralization of the pollutants can ideally be achieved in a simple and efficient manner. In many cases, TiO2 is used as a heterogeneous photocatalyst for this purpose, because it is nonphotocorrosive, nontoxic, and highly effective in its photooxidative destruction of the organic pollutants.13,15-18 The photocatalytic destruction process begins with photogeneration of electrons and holes in the photocatalyst:19 the excitation transfer of an electron from the valence to the conduction band creates an oxidizing site (a “hole”, h+VB) and a reducing site (an “electron”, e-CB). With these holes and electrons, organic compounds are oxidatively degraded, where oxygen is stoichiometrically involved.20,21 Here, we suggest a novel COD sensor using TiO2, with which a change of dissolved oxygen concentration of a given sample is measured during photocatalytic oxidation of the organic compounds therein. It is considered that the value obtained can be reliably correlated with the COD obtained using the conventional methods. (8) Korenaga, T.; Zhou, X.; Okada, K.; Moriwake, T.; Shinoda, S. Anal. Chim. Acta 1993, 272, 237-244. (9) Appleton, J. M. H.; Tyson, J. F.; Mounce, R. P. Anal. Chim. Acta 1986, 179, 269-278. (10) Korenaga, T. Bull. Chem. Soc. Jpn. 1982, 55, 1033-1038. (11) Hejzlar, J.; Kopacek, J. Analyst (Cambridge, U.K.) 1990, 115, 1463-1467. (12) Lloyd, A. Analyst (Cambridge, U.K.) 1982, 107, 1316-1319. (13) Leonard, K. M.; Setiono, D. Adv. Environ. Res. 1999, 3, U8-102. (14) Wang, C.-M.; Heller, A.; Gerischer, H. J. Am. Chem. Soc. 1992, 114, 52305234. (15) O’Shea, K. E.; Garcia, I.; Aguilar, M. Res. Chem. Intermed. 1997, 23, 325339. (16) Alfano, O. M.; Cabrera, M. I.; Cassano, A. E. J. Catal. 1997, 172, 370379. (17) Ilisz, I.; Foglein, K. J. Mol. Catal. A: Chem. 1998, 135, 55-61. (18) Abdel-Wahab, A.-M. A.; Gaber, A. E.-A. M. J. Photochem. Photobiol., A 1998, 114, 213-218. (19) Low, G. K.-C.; McEvoy, S. R. Trends Anal. Chem. 1996, 15, 151-156. (20) Barbeni, M.; Pramauro, E.; Pelizzetti, E.; Borgarello, E.; Gratzel, M.; Serpone, N. Nouv. J. Chim. 1984, 8, 547-550. (21) Martin, S. T.; Herrmann, H.; Choi, W.; Hoffmann, M. R. J. Chem. Soc., Faraday Trans. 1994, 90, 3315-3322.

Analytical Chemistry, Vol. 72, No. 14, July 15, 2000 3379

Table 1. Constituents of an Artificial Wastewater Used in this Work22 Constituents compound

mg/L

nitrohumic acid tannic acid sodium lignisulfonate (NaLS) gum arabic sodium lauryl sulfate

4.25 4.18 2.43 4.70 0.94

Water Quality Analysis method COD (permanganate method) BOD (5-days method) TOC

mg/L 5.89 3.77 9.12

EXPERIMENTAL SECTION Reagents. Titanium dioxide (ST-A01) was obtained from Ishihara Sangyo (Osaka, Japan). Tannic acid and nitrohumic acid were purchased from Wako Pure Chemicals (Osaka, Japan). A translucent PTFE membrane (JHWP02500) with an average pore size of 0.45 µm was purchased from Nihon Millipore (Yonezawa, Japan). Ligninesulfonic acid (sodium salt) and sodium dodecylbenzenesulfonate were purchased from Tokyo Kasei (Tokyo, Japan). Gum arabic was obtained from Sigma (St. Louis, MO). All organic compounds used were of reagent-grade, except nitrohumic acid and gum arabic. These reagents were used without further purification. Lake water usually contains highly stable organic compounds such as humic acid and lignin, which also account for 50% of the composition of the secondary effluents.22 Thus, the composition of an artificial wastewater sample used in this work, shown in Table 1, was based upon that of the secondary effluents from several wastewater treatment plants in Japan.22 Real water samples were collected from lakes all over Japan. Collection conditions were slightly different, but all samples were taken at a depth of 50 cm and kept below 4 °C until use. Experiments were performed within a day or two after collection. They were used without further treatments. All samples were used on the day of preparation. Apparatus. The experimental apparatus was constructed as shown in Figure 1. For immobilization on a translucent PTFE membrane, TiO2 suspension solutions (0.001 g/mL in doubly distilled deionized water) were applied to the translucent PTFE membrane to form a spot of 17 mm in diameter. The spot was aspirated for 1 min such that the TiO2 particles could be entrapped into the pores of the membrane, which was then attached to the oxygen electrode (BO-G, ABLE, Tokyo, Japan) by an O-ring. The amount of loading was 2.49 mg/cm2, unless specified otherwise. Its current output was measured and recorded using a digital multimeter (34401A, Hewlett-Packard, Palo Alto, CA) with a chart recorder (EPR-151A, TOA electronics, Tokyo, Japan). In experiments for preparing a calibration curve for the sensor, a DO meter (Do-25A, TOA Electronics) was immersed, together with the sensor probe, for monitoring a change of dissolved oxygen concentration represented in a part-per-million unit. A UV irradiator with a 6 W lamp (Funakoshi, Tokyo, Japan) and a built-in (22) Tanaka, H.; Nakamura, E.; Minamiyama, Y.; Toyoda, T. Water Sci. Technol. 1994, 30, 215-227.

3380

Analytical Chemistry, Vol. 72, No. 14, July 15, 2000

Figure 1. Schematic diagram of the apparatus. 1 ) magnetic stirrer; 2 ) TiO2-loaded oxygen electrode; 3 ) digital multimeter; 4 ) chart recorder; 5 ) thermostated water bath; 6 ) UV lamp; 7 ) reflector.

reflector was located 0.1 cm away from the side of a Pyrex glass. The temperature was thermostated at 20 °C, unless specified. The sensor probe was kept immersed in water when not in use. Procedure. A single measurement was performed as follows. At first, the probe was arranged in 2 mL of doubly distilled deionized water, and the responses were recorded. When the UV lamp was turned on (λmax ) 365 nm), the current began to decrease until a steady state was reached in 10 min. Then a 0.72mL aliquot of a sample containing organic compounds was added to the doubly distilled deionized water stirred at a constant rotating speed. Upon addition of the substrate, photocatalytic oxidation of the organic compounds began to occur, accompanying a decrease of local oxygen concentration around the TiO2-loaded membrane. The values obtained using the COD sensor were correlated with those determined by the conventional methods using KMnO4 and K2Cr2O7 23 (designated as CODMn and CODCr, respectively): for samples of the artificial wastewater, CODs were evaluated only with the permanganate method, and for real samples both methods were used. All measurements were performed three times. For preparing a calibration curve for the sensor, an appropriate volume of Na2SO3 solution was injected instead of a sample aliquot, after both responses of the DO meter and the COD sensor reached a steady-state value. The response decreases during UV irradiation were recorded. Measurement Principle. It is well-known that the titanium dioxide photocatalysis leads to stoichiometric photomineralization of organic compounds as follows20,21

CxHyOzXk + { x + (y - k - 2z)/4} O2 f xCO2 + kH+ + kX- + {(y - x)/2} H2O

where X represents a halogen atom. Therefore, it is considered that the COD of a given sample can be assessed, by tracing a change of the dissolved oxygen concentration under photocatalytic condition. In the photomineralization, oxygen acts as an acceptor of photogenerated electrons and plays an essential role in degrading organic compounds to CO2 through formation of a superoxide radical ion. In detail, the following processes can be considered

concerning the radical formation.24

Reduction reaction: O2 + 2Haq+ + 2ecb- f H2O2 (1) Oxidation reactions: H2O + 2hvb+ f 2Haq+ + (1/2)O2 (2) 2H2O + 2hvb+ f 2Haq+ + H2O2

(3)

Overall reactions: H2O + (1/2)O2 + 2(ecb- + hvb+) f H2O2 (1 + 2) 2H2O + O2 + 2(ecb- + hvb+) f 2H2O2 (1 + 3) where ecb- and hvb+ are a photogenerated electron in the conduction band and a hole in the valence band, respectively. RESULTS AND DISCUSSION Stability of PTFE Membrane. In constructing the sensor, the responses were significantly variable according to immobilization methods of TiO2. At the first stage of this work, titanium dioxide (TiO2) fine particles were suspended in a sample solution. Using this method, however, 6 h was required for a baseline current to be stabilized, and 20-30 min were required for a single run. Additionally, the data obtained did not show a good reproducibility, which may have been caused by inhomogeneous distribution of TiO2 particles, due to their settling in sample solutions. In contrast, the TiO2-loaded sensor showed a shorter stabilization and response time (10 and 3-5 min, respectively) and a good reproducibility (described later). This time reduction and improved reproducibility can be ascribed to the hydrophobicity of PTFE, concentrating oxygen around a well-defined region of TiO2 fine particles adsorbed on a PTFE membrane.25 In this case, however, one may think that the stability of the PTFE membrane toward UV irradiation becomes a crucial factor determining the overall sensor performance. However, the current decrease was observed to remain constant at 1.73 ( 0.01 µA at least for 170 h (n ) 18), which indicates at least that the state of the PTFE membrane under normal operation conditions of the sensor does not change significantly or is not related with the sensor performance. Optimization of Titanium Dioxide Loading. To optimize the loading amount of TiO2, photochemical reactions were carried out under oxygen-saturated conditions, varying the amount of TiO2 (0.00, 0.25, 0.37, 0.50, 1.49, 2.49, 3.48, and 4.97 mg/cm2). The artificial wastewater was used as a substrate, with its concentration changed in a range considered to be of environmental significance26 (0.12, 0.70, 1.86, 4.10, 9.34, and 18.49 ppm CODMn). Regardless of the amount of TiO2, linear relationships were observed (r ) 0.994-1.000) between the sensor and the conventional method in this range, which covered the range (1-10 ppm (23) Testing Methods for Industrial Wastewater; JIS K 0101; Japanese Industrial Standard Committee: Tokyo, Japan, 1991. (24) Cai, R.; Hashimoto, K.; Fujishima, A.; Kubota, Y. J. Electroanal. Chem. 1992, 326, 345-350. (25) Uchida, H.; Katoh, S.; Watanabe, M. Electrochim. Acta 1998, 43, 21112116. (26) Araki, S.; Numata, M.; Wada, O. Kankyokagakuziten (Dictionary of Environmental Sciences); Kagakudoujin: Tokyo, 1985.

Figure 2. Dependence of the sensor response on the loading amount of TiO2. Measurements were performed at 30 °C. For the other conditions, see the text.

CODMn) for first-rate quality water, directed by the Environmental Quality Standard of Japan.26 Figure 2 shows the dependence of the sensor response upon the amount of TiO2, in terms of the slope of the linear fitting equation (the current decrease vs the concentration of the substrate) obtained at a given amount of TiO2. As shown in this figure, the magnitude of the sensor response was observed to become larger with the increasing amount of TiO2, up to 2.49 mg/cm2, and to fall again at the higher loading amounts (3.48 and 4.97 mg/cm2). This may be explained by an observed fact that it was hardly possible to fully adsorb TiO2 particles onto a PTFE membrane at these amounts. Furthermore, the high loading amount of TiO2 may have caused the light penetration to be retarded. During UV irradiation of a substrate in the presence of TiO2, two kinds of reactions are considered to simultaneously occur:27 (1) homogeneous catalysis, in which the substrates are directly converted to CO2 in the presence of dissolved oxygen and (2) heterogeneous catalysis, in which the TiO2 absorbs light and gets excited to oxidize the substrates. The proportion of homogeneous catalysis in the whole photocatalytic process was estimated by evaluating the ratio of the current decrease in the absence of TiO2 (0.00 mg/cm2) to that in its presence (2.49 mg/cm2) at each concentration of the substrate. Without TiO2, the threshold concentration of the substrate for homogeneous catalysis, i.e., consuming dissolved oxygen, was observed to be 4.10 ppm CODMn. Despite the linear increase of the response magnitude at the higher concentrations (9.34 and 18.49 ppm CODMn), the ratio did not seem to show any great change (5.5-13.4, 4.8-12.3, 5.110.1% for 4.10, 9.34, and 18.49 ppm of CODMn, respectively). Concerning this matter, however, further study is needed because the wide ranges of the ratios observed here was mainly due to the large relative errors of the homocatalytic responses (those obtained without TiO2). Influence of Temperature. The temperature dependence of the sensor responses is shown in Figure 3. Here it is notable that the current decrease of the sensor rose with the increase of temperature, in spite of the usual fact that an increase in temperature leads to a decrease in the total amount of dissolved oxygen. It may be ascribed to the increase of the collision frequency between TiO2 and the substrates. (27) Rajeshwar, K. J. Appl. Electrochem. 1995, 25, 1067-1082.

Analytical Chemistry, Vol. 72, No. 14, July 15, 2000

3381

a range between 0 and 4.7 ppm, and ∆C is the corresponding current decrease. On the basis of this equation, a current decrease observed for oxidizing a given sample was converted to the corresponding oxygen demand of the sample. For real samples from lakes (n ) 6), the COD values obtained using the sensor and the conventional methods were compared. A linear relationship was observed between the values obtained using the sensor and those obtained by conventional methods with a detection limit of 0.25 ppm CODCr () 1.47 ppm CODMn). For the dichromate method, the fitting equation was as follows

COD (sensor method) ) 0.0840 COD (dichromate method) + 0.0188 (r ) 0.989) and for the permanganate method

COD (sensor method) ) Figure 3. Temperature dependence of the sensor response. For the details of experimental conditions, see the text.

Reproducibility and Long-Term Stability. The sensor performance was studied in various ways. At first, for 10 samples of the same concentration (4.10 ppm CODMn of the artificial wastewater), the relative standard deviation of the current changes was 1.25%. In addition, for five different TiO2-loaded membranes (2.49 mg/cm2), the responses to 4.10 ppm CODMn of the artificial wastewater were also reproducible (RSD ) 2.50%). As for the longterm stability of the sensor, daily runs using a single kind of sample (4.10 ppm CODMn of the artificial wastewater, measured three times everyday) for 30 days did not cause any significant change in its response trend (RSD ) 5.7%). Calibration Curve for the COD Sensor and Real Sample Analysis. Using sodium sulfite (Na2SO3) at various concentrations, a calibration graph was prepared based on the following mechanism:28

2SO32- + O2 ) 2SO42A linear relationship was found as follows.

∆C [µA] ) 0.46535 ∆O2 [ppm] + 0.01477 (r ) 0.999) where ∆O2 is the decrease of dissolved oxygen concentration in (28) Zhang, H. R.; Zhang, J.; Wei, Y. S.; Jin, W. J.; Liu, C. S. Anal. Chim. Acta 1997, 357, 119-125.

3382

Analytical Chemistry, Vol. 72, No. 14, July 15, 2000

0.171 COD (permanganate method) - 0.153 (r ) 0.942). CONCLUSION A COD sensor was constructed, and its operation characteristics were described. With the sensor, the COD value of an organic substrate was evaluated by measuring a change of its dissolved oxygen concentration under TiO2 photocatalysis. The observed parameter is close to the theoretical definition of chemical oxygen demand and thus the results obtained are considered to be reliably correlated with those obtained using the conventional methods. The sensor presented good reproducibilities for different samples of the identical concentration (RSD ) 1.25%, n ) 10) and for different TiO2-loaded membranes of the same loading amount (RSD ) 2.50%, n ) 5). The sensor also showed a long-term stability (RSD ) 5.7%, 30 days). Simple instrumentation with no sample pretreatment and the short response time (3-5 min) is another aspect of this sensor system, which can hopefully offer a new possibility as an alternative to the conventional COD evaluation methods. Currently, we are working on modified versions of this sensor, for applications to an in-situ monitoring employing a flow system or an outfield test using a microfabricated oxygen electrode. ACKNOWLEDGMENT The authors would like to thank Dr. Atsunori Hiratsuka for his helpful discussions. Received for review October 1, 1999. Accepted April 3, 2000. AC9911342