Luminol Chemiluminescence in Unbuffered Solutions with a Cobalt(II

Using a heterogeneous catalyst, Co(II)−ethanolamine complex sorbed on Dowex-50W resin, the chemiluminescence (CL) of luminol in unbuffered or weakly...
1 downloads 3 Views 182KB Size
Download PDF
Anal. Chem. 2001, 73, 5043-5051

Luminol Chemiluminescence in Unbuffered Solutions with a Cobalt(II)-Ethanolamine Complex Immobilized on Resin as Catalyst and Its Application to Analysis Jin-Ming Lin,*,†,‡ Xiaoquan Shan,† Soichi Hanaoka,‡ and Masaaki Yamada‡

Research Center for Eco-Environmental Science, Chinese Academy of Sciences, P.O. Box 2871, Beijing 100085, China, and Department of Applied Chemistry, Graduate School of Engineering, Tokyo Metropolitan University, 1-1 Minami-Ohsawa, Hachioji, Tokyo 192-0397, Japan

Using a heterogeneous catalyst, Co(II)-ethanolamine complex sorbed on Dowex-50W resin, the chemiluminescence (CL) of luminol in unbuffered or weakly acidic solution was studied in the presence of H2O2. The maximum luminol CL wavelength at pH 5.7 was 448 nm, 23 nm longer than that in a basic solution (pH 10.5). Three different ligands, mono-, di-, and triethanolamine, and six transition metal ions, Co(II), Cu(II), Ni(II), Mn(II), Fe(II), and Fe(III) were compared by CL measurements. The CL intensity decreased in the order mono- > di- > triethanolamine and Co(II) > Cu(II) > Ni(II) > Fe(III) > Mn(II) > Fe(II). This heterogeneous CL system was developed as H2O2 and glucose flow-through sensors. Detection limits (S/N ) 3) of H2O2 and glucose using Dowex-50W-X4-Co(II)-monoethanolamine as catalyst are 1 × 10-7 M and 1 × 10-6 M, respectively. On the basis of the studies of the CL, fluorescence, UV-vis and ESCA spectra and the effect of dissolved oxygen in luminol solution, a mechanism for CL emission in unbuffered solution was considered as the formation of a superoxide radical ion during the decomposition of H2O2 catalyzed by the Co(II)-ethanolamine immobilized resin. Then the superoxide radical ion acted on luminol and the CL was emitted. The applications of the proposed method to determine H2O2 in rainwater without any special pretreatment and glucose in human urine and orange juice samples give satisfactory results. Since the chemiluminescence (CL) phenomenon of luminol was first reported by Albrecht 1 in 1928, CL resulting from the reaction of this reagent with H2O2 in a strongly basic medium has been extensively studied.2-6 The high sensitivity and conven* Corresponding author. Phone: +81-426-77-1111. Fax: +81-426-77-2821. E-mail: [email protected]. † Chinese Academy of Sciences. ‡ Tokyo Metropolitan University. (1) Albrecht, H. O. Z. Phys. Chem. 1928, 136, 321. (2) Behrens, H.; Totter, J. R.; Philbrook, G. E. Nature (London) 1963, 199, 595-596. (3) White, E. H.; Bursey, M. M. J. Am. Chem. Soc. 1964, 86, 941-942. (4) Mere´nyi, G.; Lind, J. S. J. Am. Chem. Soc. 1980, 102, 5830-5835. (5) Lin, J.-M.; Ishii, M.; Yamada, M. Bunseki 1998, 865-872. 10.1021/ac010573+ CCC: $20.00 Published on Web 09/22/2001

© 2001 American Chemical Society

ience of luminol CL methods have led to widespread use of this reagent and its derivatives in many fields. The most visible fact in the applications of luminol as a CL reagent is reflected in the great number of publications.5-9 But until now, to our knowledge, almost all luminol (or its derivative)-H2O2-metal ion CL systems have involved a basic solution (pH > 8.0), except for a few special microheterogeneous systems, for example, microemulsion10 or water-soluble polymer media.11 The fact that the luminol-H2O2 system using metal ions as catalyst could not emit in neutral or weakly acidic homogeneous solution could be explained by the relative stability of H2O2 in a low-pH solution. Generally, the decomposition of H2O2 begins with the hydroperoxide ion (HO2-), which is easily formed in basic solution with the acid-base equilibrium (pKa ) 11.7) of H2O2.12 It is well-known that many substances influence the CL systems of luminol in basic solution; therefore, many luminol CL methods have high sensitivity, but their selectivities were unsatisfactory. Because of the importance of the luminol-H2O2 CL reaction, we wanted to develop an unbuffered luminol-H2O2 CL system. However, because of the chemical characters of luminol and H2O2, it is difficult to establish a homogeneous luminol CL in neutral or weak acidic solution, even though many chemists have made a lot of effort to do so. There are reports of transition metal complexes immobilized on Dowex-50W resin that are of high catalytic activity for H2O2 decomposition.13-16 The kinetic investigations have indicated that the transition metal complexes immobilized on Dowex-50W resin (6) Easton, P. M.; Simmonds, A. C.; Rakishev, A.; Egorov, A. M.; Candeias, L. J. Am. Chem. Soc. 1996, 118, 6619-6624. (7) Gundermann, K.-D.; McCapra, F. Chemiluminescence in Organic Chemistry; Springer-Verlag: Berlin, 1987. (8) Birks, J. W. Chemiluminescence and Photochemical Reaction Detection in Chromatography; VCH: New York, 1989. (9) Blum, L. J. Bio- and Chemi-Luminescent Sensors, World Scientific: Singapore, 1997. (10) Ishimaru, N.; Lin, J.-M.; Yamada, M. Anal. Commun. 1998, 35, 67-69. (11) Karatani, H. Bull. Chem. Soc. Jpn. 1987, 60, 2023-2029. (12) Nosaka, Y.; Yamashita, Y.; Fukuyama, H. J. Phys. Chem. B 1997, 101, 58225827. (13) El-Sheikh, M. Y.; Habib, A. M.; Ashmawy, F. M.; Gemeay, A. H.; Zaki, A. B. J. Mol. Catal. 1989, 55, 396-405. (14) El-Sheikh, M. Y.; Habib, A. M.; Gemeay, A. H.; Zaki, A. B.; Bargon, J. J. Mol. Catal. 1992, 77, 15-22. (15) Salem, I. A. J. Mol. Catal. 1993, 80, 11-19. (16) Salem, I. A. J. Chem. Kinet. 1995, 27, 449-505.

Analytical Chemistry, Vol. 73, No. 21, November 1, 2001 5043

Figure 1. Schematic diagram of the CL flow-through sensor of the determinations of H2O2 and glucose: P, peristaltic pump; PMT, photomultiplier tube.

remain stable throughout the decomposition reaction.15 It has been suggested that at the start of the reaction, the transition metal complex reacts with a molecule of H2O2 to form the peroxo-metal complex. Then this complex reacts with another molecule of H 2O2 to yield the active intermediate products, and the complex undergoes self-decomposition, with the evolution of O2. Unfortunately, the intermediates produced from the decomposition of H2O2 with metal complex-resin as catalyst have not been clarified until now, even though the kinetics of H2O2 decomposition have been well-studied, and many applications in analytical chemistry have been reported.17 As an initial attempt, a well-known resin, Dowex-50W resin treated with Co(III)-enthanolamine complex, was used in the present work. In an unbuffered solution, the luminol-H2O2 system has a bright CL emission when this resin is used as the catalyst. A simple, highly selective and sensitive CL flow-through sensor for determinations of H2O2 and glucose was assembled on the basis of this observation. EXPERIMENTAL SECTION Reagents. All of the chemicals were of analytical-reagent grade and were used as received. Water was obtained from a Milli-Q purification system (Japan Millipore, Tokyo). Luminol, glucose oxidase, D-(+)-glucose, and 30% (v/v) H2O2 solution are the products of Tokyo Kasei (Tokyo, Japan). A 1 × 10-5 M unbuffered luminol stock solution was prepared by dissolving 5.31 mg of luminol in 3.00 L of water. This solution (pH ) 5.7) can be stored at least one month. A 0.1 M H2O2 solution was prepared by volumetric dilution of 30% H2O2. The exact concentration of the stock solution was determined by titration. Dowex-50W resin in the hydrogen form was used as a strongly acid cation exchanger. It is available as spherical beads of sulfonated styrene divinylbenzene copolymers in the hydrogen form from Dow Chemical (Midland, MI). Amberlite IRC-50 weak acid cation-exchange resin was purchased from Organo Co. Ltd. (Tokyo). Monoethanolamine (MEA), diethanolamine (DEA), triethanolamine (TEA), and other reagents were from Nakalai Tesque Inc. (Kyoto). Apparatus and Procedures. The batch studies of CL were carried out using a Lumicounter 1000 (Microtec NITI-ON, Funabashi). The flow sensor was an flow line system (Figure 1) consisting of a peristaltic pump (SJ1211, Atto, Tokyo), a CL detector (Lumiflow LF-800, Microtec NITI-ON, Funabashi, Japan), (17) Mottola, H. A.; Pe´rez-Bendito, D. Anal. Chem. 1996, 68, 257R-289R.

5044 Analytical Chemistry, Vol. 73, No. 21, November 1, 2001

a flow cell (3-cm length, 0.5-cm i.d. glass tube filled with resin) that was placed at the front of photomultiplier tube, and a 90-µL loop injector placed near the flow cell. The carrier was a 1 × 10-5 M unbuffered luminol solution. H2O2 solution was injected directly into the carrier. When this flow system is used to determine glucose, a PTFE tube (5.0-mm i.d., 10-cm length) filled with glucose oxidase immobilized Amberlite IRC-50 resin is placed in the flow line just before the CL flow sensor. The CL and fluorescence spectra were obtained using a F-4010 fluorescence spectrophotometer (Hitachi, Tokyo). For the CL spectra, an unbuffered 1 × 10-3 M luminol (or 0.01 M acetate at pH 5.7, 0.05 M NaHCO3/NaOH solution at pH 10.5) and 1 × 10-2 M H2O2 solutions were fed through separate lines into a cell filled with Co(II)-MEA immobilized resin particles and placed inside the cell holder of the fluorescence spectrophotometer. The flow rates for both the luminol and the H2O2 solutions were 1.5 mL/ min. The excitation lamp was off during the CL spectra recording. ESCA and UV-vis spectra were recorded using ESCA-3400 (Shimadzu, Kyoto, Japan) and UV-2200 (Shimadzu) spectrophotometers, respectively. The ‚O2- radical that was formed during the decomposition of H2O2 was detected by the reduction of nitroblue-tetrazolium (NBT),18 cytochrome c reaction,19 and ESR spectroscopy20,21 (JES-RE3X spectrometer, JEOL, Tokyo). The spin-trap 5,5-dimethylpyrroline-N-oxide (DMPO) was used to test for the existence of radicals, and it was prepared as described by Floyd.22 The electrochemical generation of Co3+ ion was based on the reference,23 which has been successfully used for our previous work.24 The method for the kinetic study was based on the measurement of O2 from the decomposition of H2O2.25 A weighed quantity of the resin was shaken with 10.0 mL of H2O2 in a closed reaction vessel that was kept in a thermostatic water bath, and the volume of O2 evolved was measured using a gasometric technique. The (18) Bielski, B. H. J.; Shiue, G. G.; Bajuk, S. J. Phys. Chem. 1980, 84, 830-833. (19) Lisdat, F.; Ge, B.; Ehrentreich-Fo ¨rster, E.; Reszka, R.; Scheller, F. W. Anal. Chem. 1999, 71, 1359-1365. (20) Nagy, L.; Galba´cs, Z. M.; Csa´nyi, L. J.; Horva´th, L. J. Chem. Soc., Dalton Trans. 1982, 859-863. (21) Ono, Y.; Matsumura, T.; Kitajima, N.; Fukuzumi, S. J. Phys. Chem. 1977, 81, 1307-1311. (22) Floyd, R. A. Biochim. Biophys. Acta 1983, 756, 204-216. (23) Tanaka, H.; Morita, H.; Shimomura, S.; Okamoto, K. Anal. Sci. 1997, 13, 607-612. (24) Lin, J.-M.; Yamada, M. Anal. Chem. 2000, 72, 1148-1155. (25) Kanungo, S. B.; Parida, K. M.; Sant, B. R. Elctrochim. Acta 1981, 26, 11571167.

Figure 2. Procedures of MEA-Co(II)-resin and MEA-Co(III)-resin preparation and the color changes of the resin at different states.

rate of H2O2 decomposition was determined at four different temperatures in the range 25-45 °C. The actual change in concentration during the catalytic reaction was determined from the following formula

C ) [C0 - (PVg)/(RTV1)]

(1)

where C0 and C are the concentrations of H2O2 at beginning and at time t, respectively. Vg is the volume of O2 gas liberated at t and V1 is the total volume of H2O2 solution. P, R, and T correspond to the atmospheric pressure (∼1 atm), gas constant (8.3145 J‚mol-1‚K-1) and temperature (K), respectively. Preparations of Co(II)-MEA Immobilized Resin and Glucose Oxidase Immobilized Resin. A 10-mL portion of resin was treated with 20 mL of 1 M HCl solution and vibrated by an SS-8 shaker (Tokyo Rikakikai Co. Ltd., Tokyo) for 1 h. After washing with H2O, the resin was converted into the Co(II) form by mixing it with 20 mL of 1.0 M CoSO4 solution for 1 h. The resin in the Co(II) form was collected and washed with H2O until it was free of any excess of Co(II) ions. A 20-mL portion of 1.0 M MEA solution was mixed with the Co(II) form resin and vibrated for 1 h. After equilibrium was attained, the resin was collected and washed with H2O until it was free of any excess of the ligand. The formation of Co(III)-MEA from the Co(II)-MEA complex form on the resin was treated with 3.0 mL of 1.0 M H2O2 solution. After washing with water, ∼0.4 mL of the resin was placed into the flow cell. The immobilization of glucose oxidase on Amberlite IRC-50 resin was based on the reference.26 After the resin was regenerated by 1.0 M HCl and washed with water, 1.0 mL of resin was mixed with 20 mL of 2.0 mg/mL glucose oxidase solution and vibrated for 1 h. Last, the resin was washed with water, ∼0.5 mL of the resin was placed into the PTFE tube. Collection of Rainwater Sample and Preparation of Glucose Sample. Rainwater samples were collected at the campus (26) Chitohata, I. Immobilized Enzyme; Kodansha: Tokyo, 1975.

of Tokyo Metropolitan University, Tokyo. The rain was collected in beakers maintained at 0 °C. The analysis for H2O2 was conducted immediately after collection. The samples of glucose were from orange juice (Qoo, Nippon CocaCola Company, Tokyo) and urine. The orange juice was diluted 100-fold by water. Both rainwater and the diluted juice solution were analyzed directly. Because of the high concentration of sodium ions and protein in the human urine sample, pretreatment of the urine sample was necessary. After the urine sample was heated at 90 °C for 15 min, the sample solution was filtered through filter paper (Advantec, Tokyo, Japan). The filtered solution was then diluted 10-fold by water. To decrease the effect of cationic ions on urine determination, as shown in Figure 1, an ion-exchange column (5-mm i.d., 5-cm length, filled with Dowex-50-X8, 50-100 mesh) was assembled with the sampling injector. RESULTS AND DISCUSSION Preparations of Co(II)-MEA Complex Immobilized Resin and the Batch Studies of CL. As shown in Figure 2, the treatment of Dowex-50W-2X with the Co(III)-MEA complex has four steps: (a) resin regeneration, (b) formation of resin in Co(II) form, (c) formation of the resin in Co(II)-MEA form, and (d) formation of resin in Co(III)-MEA form. The colors of the resin after the different steps were yellow (A), pink (B), green (C) and brownish green (D), respectively. When the electrochemically generated Co3+ ion was immobilized on the resin, the color was orange (F). On reacting with the MEA ligands, this color also changed to brownish green (G) without the H2O2 treatment. It must be pointed out that the Co3+ ion is unstable in the solution, so this treatment should be completed within 30 min. Another phenomenon was also observed: when the resin in the Co(II)-MEA form dried in air for one week, its color was changed slowly from green to deep brownish green (E), which could be explained by slow oxidation of the Co(II) on the resin with oxygen in the air. On addition of H2O2 to the resin in Co(II)-MEA form, however, the color changed from green to brownish green immediately, which means that the reaction of resin-Co(II)-MEA Analytical Chemistry, Vol. 73, No. 21, November 1, 2001

5045

Table 1. Optimum Conditions and Analytical Figures of Merit of the CL Flow-through Sensor for the Determination of H2O2 in Rainwater optimum conditions resin: MEA-Co(II) complex immobilized Dowex-50W × 2 carrier: 1 × 10-5 M luminol in unbuffered solution (pH ) 5.7) carrier flow rate: 2.0 mL/min sample injection interval time: 60 s calibration curve

ICL ) 2.46 + 5.53 × 106 [H2O2]; (r2 ) 0.9990, n ) 12)

dynamic liner range detection limit

samplea 1 2 Figure 3. Possible chemical combination model of monoethanolamine to cobalt(II) sorbed on Dowex-50W resin.

with H2O2 is a fast process. These results indicated that the Co(II)-MEA could be oxidized to Co(III)-MEA by H2O2 or O2 on the resin surface. With the batch method, injection of 100 µL of 1 × 10-3 M H2O2 solution into 10 mg MEA-Co(III)-resin led to a weak CL signal. When 100 µL of 1 × 10-5 M luminol solution was added into the H2O2-resin heterogeneous system, a very strong CL was recorded. These phenomena were the same for all three of the Co(III)-MEA immobilized resins (D, E, and G). When a 100 µL solution of 1 × 10-4 M Co(MEA)22+ (pH ) 7.2) was mixed with 100 µL of 1 × 10-3 M H2O2 solution, no CL was observed. Also, addition of 100 µL of 1 × 10-5 M luminol solution to the Co(MEA)22+-H2O2 mixed solution led to no CL emission with our instrument. It is a well-known fact that the CL of the luminol-H2O2 system catalyzed by a metal ion or metal complex takes place only in a basic solution; however, in this work, with the resin as catalyst, the luminol-H2O2 CL reaction took place under neutral or even weakly acidic conditions. The ratios of metal ion to ligand in the resin were also determined by titration and back-titration. The result showed that the [cobalt]/[MEA] is 1:2. The [cobalt]/ligand ratios of the complexes before and after measuring CL 100 times with 1 × 10-5 M luminol/1 × 10-5 M H2O2 solution did not change. The chemical structure calculated by CS ChemOffice program of the Co(II)-MEA complex sorbed on resin is shown in Figure 3. Both MEA molecules are at the same side of the cobalt atom when oxygen and nitrogen atoms of the MEA donate only a lone electron pair to the Co(II) ion. This structure allowws the Co(II)-MEA complex to combine with the resin easily, because the space resistance of the MEA is small. The capacity and the moisture content of the resin were determined at the end of the experiment, and no change was found. Therefore, the resin treated 5046 Analytical Chemistry, Vol. 73, No. 21, November 1, 2001

3

1 × 10-7-2 × 10-5 M 1 × 10-7 M (S/N ) 3)

Results of Analysis of H2O2 in Rainwater H2O2 ref. added found recovery in sample method36 (×10-6 M) (×10-6 M) (×10-6 M) (×10-6 M) (%) 0 1 5 0 2 4 0 2 5

3.06 3.95 8.45 3.21 5.24 7.12 15.2 17.2 20.2

89 107 102 98

3.1 ( 0.2

3.2

3.4 ( 0.1

3.4

15.2 ( 0.3

14.9

100 100

a Each sample was analyzed five times. The results are the averages. Samples 1, 2, and 3 were collected at the campus of Tokyo Metropolitan University on April 20, July 10, and November 2, 2000, respectively.

Table 2. Determinations of Glucose in Orange Juice and Human Urine Samples Recovery results founda

sample

amt (mg/mL)

added (mg/mL)

founda (mg/mL)

recovery (%)

juice 1 juice 2 urine 1 urine 2

56.8 ( 0.8 57.3 ( 0.5 0.039 ( 0.005 0.027 ( 0.007

10.0 10.0 0.03 0.03

0.93 0.95 0.026 0.027

93 95 87 91

a

Mean of three determinations.

with Co(II)-MEA is stable using the low concentration of luminol solution as carrier. Optimization of the CL Flow-through Sensor for H2O2 and Glucose. The accurate estimation of H2O2 in natural waters, in the atmosphere, and enzymatically formed in the assay of oxidoreductase is becoming of great importance for environmental and clinical analysis. Both in environmental analysis of H2O2 and clinical analysis of glucose, the determinations of H2O2 and glucose in unbuffered conditions are desirable. Glucose analysis based on the inherent specifity of an enzymatic reaction has provided the most accurate means for quantitation of glucose. The glucose oxidase method, originally described Keston,27 is the most (27) (a). Keston, A. S. Abstract of Papers, 129th Meeting of the American Chemical Society, Dallas, Texas, April, 1956; p 31C. (b). Bostick, D. T.; Hercules, D. M. Anal. Chem. 1975, 47, 447-452.

Figure 4. Effect of the injection interval time on the CL intensity. Conditions: flow system is the same in Figure 1; MEA-Co(III)Dowex-50W was used as catalyst; 1 × 10-5 M luminol solution at flow rate of 2.0 mL/min; injection volume of H2O2 solution, 90 µL. Table 3. Effect on the CL Intensity of Dissolving O2 and N2 in Luminol Carrier Solution CL intensity (S/N)a 5× usual N2 (10 min) O2 (10 min) a

10-6

M H2O2

27 16 25

1 × 10-5 M H2O2 59 42 61

S/N: ratio of the CL signal to noise.

commonly employed enzymatic technique for glucose analysis for biological fluids. The method is based on the following reaction. glucose oxidase, pH 5.6

β-D-glucose + O2 98 D-gluconic acid + H2O2 (2)

The H2O2 generated from this reaction is measured by many analytical methods, for example, spectrophotometry,28 fluorimetry,29 CL,30,31 and electrochemical32,33 techniques. Soluble glucose oxidase has a broad pH profile with a reported optimum at pH 5.6.34 Therefore, the present unbuffered luminol (pH 5.7)-H2O2Co(II)-MEA CL system is the most suitable as a monitor for H2O2 determination and reaction 1. The flow system assembled as in Figure 1 is a simple and highly sensitive sensor for the determination of H2O2 in the rainwater without sample pretreatment and glucose in juice and urine. To establish optimal conditions for the determination of H2O2, the CL intensity emitted from CL flowthrough sensor was measured as a function of luminol carrier solution, metal ion, ligand, resin, and the flow rate. Luminol Carrier Solution. The concentration and the flow rate of luminol carrier solution were examined. The CL intensity (S/N, ratio of CL signal-to-noise, 1 × 10-5 M H2O2) corresponding to the concentrations of luminol at 1 × 10-6, 5 × 10-6, 1 × 10-5, (28) Katsumata, H.; Sekine, T.; Teshima, N.; Kurihara, M.; Kawashima, T. Talanta 2000, 51, 1197-1204. (29) Schubert, F.; Wang, F.; Rinnerberg, H. Mikrochim. Acta 1995, 121, 237. (30) Marshall, R. W.; Gibson, T. D. Anal. Chim. Acta 1992, 266, 309. (31) Spohn, U.; Preuschoff, F.; Blankenstein, G.; Janasek, D.; Kula, M. R.; Hacker, A. Anal. Chim. Acta 1995, 303, 109. (32) Xu, J.-J.; Chen, H.-Y. Anal. Biochem. 2000, 280, 221-226. (33) Qian, J. H.; Liu, Y. C.; Liu, H. Y.; Yu, T. Y.; Deng J. Q. Anal. Biochem. 1996, 236, 208. (34) Brigh, H. J.; Appleby, M. J. Biol. Chem. 1969, 244, 3625.

2 × 10-4, and 5 × 10-4 M were 20, 40, 55, 54, 50, and 48, respectively. The higher the concentration of luminol, the stronger the CL that was observed. But the dissolution of luminol in water is low, and the high concentration of luminol caused a high CL noise. Therefore, a luminol solution of 1 × 10-5 M is suitable as the carrier. The pH value of this solution is 5.7. This so-called unbuffered luminol solution is also a weak acid; as a result of the chemical structure of the luminol molecule, it only means that luminol solution was prepared in pure water. Therefore, the luminol carrier solution is of weakly buffering ability for the sample solution, which is beneficial for maintaining the CL reaction stability when the sample pH varies slightly. The flow rates of the luminol carrier solution from 1.0 to 4.0 mL/min were compared. There was no evidenc for a difference in CL intensity (S/N) for the determination of 1 × 10-5 M H2O2 at flow rates of 1.0, 1.5, 2.0, 2.5, 3.0, 3.5, and 4.0 mL/min. The biggest (56) and the smallest (50) CL S/N were only 6 units different. The low flow rate caused a broad CL signal, which decreased the CL intensity, but too high a flow rate of luminol solution will cause a high pressure in the flow line and the noise will be increased. A flow rate in the range of 1.5-2.5 mL/min is suitable; therefore, 2.0 mL/min was used. Center Metal Ions and Ligands of Complexes. Using MEA as a ligand, the effects of six different transition metal ions, Co(II), Cu(II), Ni(II), Mn(II), Fe(II) and Fe(III), on the CL intensity were compared. The CL intensities for 1 × 10-4 M H2O2 using resins treated with these six metal ions, Co(II), Cu(II), Ni(II), Fe(III), Mn(II) and Fe(II) were 520, 200, 154, 6, 5, and 5, respectively. The kinetic study of the H2O2 decomposition was also carried out by changing the metal ion on the resin. The rate constant of the decomposition reaction was in the order Co(II) > Cu(II) > Ni(II) > Fe(III) > Mn(II) > Fe(II), which is the same as the order of CL intensity. Similar to the metal ion experiments, using Co(II) as the center metal ion of the complex, three ligands, MEA, DEA, and TEA were compared. The CL intensities in the ratio of signal-to-noise with Co(II)-MEA, -DEA and -TEA complex were 300, 125, and 28, respectively. On the basis of these results, a Co(II)-MEA complex was used. Cross-Linkage Degree and Mesh Size of Resin. Three Dowex-50W resins with different cross-linkage degrees (DVD%), 8, 4, and 2 were compared. Their exchange capacities are 1.7, 1.1, and 0.6 meq/mL, respectively. Generally, the CL intensity should be stronger with the resin higher exchange capacity. But in this work, using Dowex-50-X2 (DVB% ) 2), the CL intensity was the strongest. This result could be explained by the greater the degree of resin cross-linkage, the fewer H2O2 molecules that can arrive at the resin surface. This phenomenon is called the salting-out effect.35 The mesh sizes of the resin were also studied. When the mesh size of the resin particle changed from 50 to 100, 100 to 200, to 200-400 mesh, the CL intensity increased. The result could be considered as the result of the increase of the effective surface area with the decrease of the mesh size of the resin particles. But we also found that the lifetime of the CL flowthrough sensor depended on the exchange capacity of the resin. By using an 8% DVB resin, the lifetime of the sensor is ca. 2-fold longer than that resin of 2% DVB. Therefore, considering the (35) Helfferich, F. Ion Exchange; McGraw-Hill: New York, 1962.

Analytical Chemistry, Vol. 73, No. 21, November 1, 2001

5047

Figure 5. Absorption spectra of NBT in H2O2 solution containing resin. Conditions: MEA-Co(III)-resin, 10 mg; 10-4 M NBT, 2.0 mL; 0.1 M H2O2, 1.0 mL; references, water. Times in the figure are the mixing time of NBT with resin and H2O2 solution.

sensitivity of H2O2 determination and the lifetime of the sensor, a resin of 4% DVB and 100-200 mesh was used in this work. Determination of H2O2 in Rainwater. On the basis of the above experiments, the optimum conditions and the analytical results of H2O2 in rainwater are listed in Table 1. The samples collected on November 2, 2000, had a high concentration of H2O2, which may be due to the effect of the volcano eruption on Miyake Island of Tokyo in September 2000. Addition of standard H2O2 solutions to the rainwater samples showed that there was a good linear relationship between the H2O2 concentration and the CL intensity for all of the samples. The H2O2 recoveries for the rainwater samples were near 100%, which means that the interference of substances affecting the H2O2 determination were relatively few. The influences of other foreign species were also studied. The transition metal ions, for example, 1000-fold Zn2+, Ni2+, Co2+, and Mn2+ and 100-fold Cu2+, Cr3+, and Fe3+ have no effect on the determination of 5 × 10-7 M H2O2. These metal ions were often used as the catalyst for luminol-H2O2 CL reaction in a basic solution. However, in neutral or weak acidic solution, their catalytic character disappeared, which caused the present CL flowthrough sensor to be highly selective. Other substances, such as 500-fold Cl-, HCO3-, CO32-, NH4+, K+, Na+, Ca2+, Mg2+, SO42-, HPO42-, SO32-, acetate, and vitamin C, 10-fold NH2NH2 and hydroxymethyl hydroperoxide and 1-fold OCl- and BrO- have no influence on the H2O2 determination. These data showed that the selectivity and sensitivity of the present sensor are high enough to be used in practical applications. Determinations of Glucose in Orange Juice and Human Urine. The proposed H2O2 flow-through sensor was combined with a glucose oxidase-resin column placed between the injector and sensor cell and a pretreated resins column (5-mm i.d., 10-cm length) before the injector (Figure 1). The detection conditions, 5048 Analytical Chemistry, Vol. 73, No. 21, November 1, 2001

including the luminol carrier solution, flow rate, and sensor unit, are the same as these conditions for H2O2 determination. The glucose oxidase column is of 5.0-mm i.d. and 10-cm length. Although a longer enzyme column can give out a high sensitivity for the glucose determination, the pressure of the flow system is increased with the column length. The suitable column length is 10 cm. The calibration graph of CL intensity (S/N) vs glucose concentration was linear in the 1.0 × 10-6 to 1 × 10-4 M range (ICL(S/N) ) 7.2 + 6.5 [glucose]/µM; r2 ) 0.9858, n ) 12). The detection limit was 1 × 10-6 M H2O2 at a signal-to-noise ratio of 3. The reproducibility in peak height was determined by measuring 10 replicates of 1 × 10-5 M glucose; the relative standard deviation was 2.3%. With the pretreated column, interference studies showed that the foreign species present in orange juice and urine were of no influence on the determination of glucose at their normal concentration level. The determination results of glucose in juice and urine samples are summarized in Table 2. The relative standard deviation was 3-6% (n ) 5) for each sample. The recoveries for each sample were 89-106%, which are good enough for practical use. Possible Mechanism of Luminol-H2O2-Resin Heterogeneous CL System. There are several reports13-16,37,38 of the decomposition of H2O2 with a transition metal-complex-immobilized resin as the catalyst. Although the kinetic and thermodynamic parameters were evaluated, the decomposition mechanism has not been clarified. It has been proposed that the H2O2 decomposition reaction in the presence of resin results in the formation of a peroxo-metal complex. This peroxo-metal complex then undergoes self-decomposition with the evolution of O2. Unfortunately, until now, there has been no report of CL with the metal-complex-immobilized resin as catalyst. The CL emission of luminol-H2O2 in unbuffered or weak acidic solution also merits

The complete decomposition of 1.0 mole of H2O2 with the resin as catalyst generated ∼0.5 mole O2 gas. Therefore, a cyclic electron-transfer process39 was proposed for the decomposition of H2O2 on MEA-Co(II)-resin. Such a process is initiated either by transfer of an electron from the peroxide to an oxidizing site on the surface to produce a superoxide radical ion (‚O2-), or by transfer of an electron from a reducing site to the dissolved oxygen in the carrier, also to yield ‚O2-. When the carrier is H2O2 or the O2-containing solution, it can be considered that the reaction for the transfer of an electron from the reducing site to the peroxide to H2O2 or O2 to yield the hydroxyl radical (‚OH) or superoxide radical ion (‚O2-)

[CoII(MEA)2]2+ + H2O2 f [CoIII(MEA)2]3+ + ‚OH + OH(3) [CoII(MEA)2]2+ + O2 f [CoIII(MEA)2]3+ + ‚O2- (4) is fast.14,40 The OH- ion formed from this reaction can catch a proton from H2O2 to form HO2- in neutral or weakly acidic medium.

H2O2 + OH- f H2O + HO2-

(5)

Then the HO2- ion combines with [CoIII(MEA)2]3+ to form a peroxo-metal complex [CoIII(MEA)2(HO2)]2+, which undergoes self-decomposition.

[CoIII(MEA)2(HO2)]2+ f [CoII(MEA)2]2+ + HO2. (6) Figure 6. ESCA spectra of resin in MEA-Co(II) (upper) and MEACo(III) (lower) forms.

study. To study the CL mechanism of the present heterogeneous system, the following experiments were carried out. Effects of O2 and N2 in Carrier on CL. We noticed that the CL intensity was affected by the injection interval time (Figure 4). The longer the luminol carrier solution flowed through the sensor, the stronger the CL emission was. This phenomenon may be due to the slow oxidation of the MEA-Co(II)-resin by the dissolved oxygen. The same as the resin in the Co(II)-MEA form oxidized by air, the formation of resin from the Co(II)-MEA form to the Co(III)-MEA form by the dissolved O2 in the carrier solution also took a relatively long time. From Table 3, when the luminol carrier solution was degassed with bubbling N2 for 10 min, the CL intensity was decreased by about 1/3. When O2 was bubbled into the degassed carrier solution for 10 min, the CL was brought back up to the usual level. These results indicated that the dissolved O2 takes part in the reaction. Decomposition of H2O2 into H2O and O2. When a 0.1M H2O2 solution was added to the resin in the Co(II)-MEA form by the batch method, the decomposition of H2O2 was evident, and a weak CL emission was observed. In the flow system, when a 0.1 M H2O2 solution flowed through the resin phase, the waste solution contained a lot of gas bubbles, and a background CL emission was also recorded. The gas was certified as O2. The volume of O2 produced was measured by a gasometric technique.25

When the ‚OH radical reacts with the HO2‚ radical, the singlet oxygen (1O2) is formed, which is the emitter of the CL emission of the decomposition of H2O2.

HO2‚ + ‚OH f H2O + 1O2 (7)40 1

O2 f 3O2 + hυ (8)41

The formation of the free ‚OH radicals in the heterogeneous decomposition of H2O2 has been mentioned by several reports.40,42,43 Reaction 4 is slow. Only a few of the ‚O2- radical ion from the dissolved O2 were formed, but it also caused a small background CL when luminol solution flowed through the resin phase. The formation of [Co(III)(MEA)2]3+ is an important procedure in the CL reaction. When H2O2 was injected into the Co(III)-MEA(36) Lin, J.-M.; Arakawa, H.; Yamada, M. Anal. Chim. Acta 1998, 371, 171176. (37) El-Sheikh, M. Y.; Ashmawy, F. M.; Salem, I. A.; Zaki, A. B.; Nickel, U. Transition Met. Chem. (London) 1991, 16, 319-23. (38) Salem, I. A. J. Mol. Catal. 1994, 87, 25-32. (39) Kanungo, S. B.; Parida, K. M.; Sant, B. R. Electrochim. Acta 1981, 26, 11571167. (40) Gemeay, A. H. Colloid Surf., A 1996, 116, 277-284. (41) Wassmerman, H. H.; Murray, R. W. Singlet Oxygen; Academic Press: New York, 1979. (42) Selvaraj, P. C.; Mahadevan, V. J. Mol. Catal., A 1997, 120, 47-54. (43) Gemeay, A. H.; Salem, M. A.; Salem, I. A. Colloids Surf., A 1996, 117, 245252.

Analytical Chemistry, Vol. 73, No. 21, November 1, 2001

5049

Figure 7. CL spectra of luminol-H2O2 reaction in unbuffered (A) and buffer (B) solutions. The luminol (1 × 10-3 M, 1.5 mL/min) and H2O2 (0.01 M, 1.5 mL/min) solutions were separately pumped to a mixing jointer and passed through the resin-filled glass tube, which was placed in the cell holder of the fluorescence spectrophotometer. Spectra B and A corresponded to the luminol dissolved in water (unbuffered solution, pH 5.7) or in a 0.01 M acetate buffer (pH 5.7) and 0.05 M NaHCO3/NaOH buffer solution (pH 10.5), respectively.

resin solid phase, a superoxide radical was formed, and the resin in the Co(III)-MEA form was reduced to the Co(II)-MEA form.

[CoIII(MEA)2]3+ + H2O2 f [CoII(MEA)2]2+ + ‚O2- + 2H+ (9)

This step is fast, which was evidenced by the batch CL profiles (not shown here). In the present work, the evidence for the generation of the ‚O2- radical in H2O2 decomposition was carried out with NBT reduction, cytochrome c reaction, and ESR reaction. NBT Reduction, Cytochrome c Reaction, and ESR Detection for the ‚O2- Radical. The reaction of ‚O2- with NBT has frequently been used for detecting ‚O2- radicals.18,44-47 The ‚O2radical produced from the chemical reactions can reduce NBT to its blue diformazan pigment.18 In the present CL system, when NBT was added to the H2O2 solution containing the resin, the color change of NBT solution from yellow to blue was very evident. The absorption spectra of NBT in H2O2 solution containing resin are shown in Figure 5. Without the resin, the absorption of NBT in H2O2 solution at 552 nm undergoes almost no change. After adding 10 mg of the MEA-Co(III)-resin to the H2O2 (44) Afanas’ev, I. B. Superoxide Ion: Chemistry and Biological Implications; CRC Press: Boca Raton, 1989. (45) Bielski, B. H. J.; Richter, H. W. J. Am. Chem. Soc. 1977, 99, 3019-3023. (46) Goto, H.; Lin, J.-M.; Yamada, M. Bunseki Kagaku 1998, 47, 417-422. (47) Lin, J.-M.; Yamada, M. Anal. Chem. 1999, 71, 1760-1766.

5050

Analytical Chemistry, Vol. 73, No. 21, November 1, 2001

solution, the absorption at 552 nm is increased quickly. Like NBT, cytochrome c is also often used for the measurement of the superoxide dismutase activity.19,44,46,48 Cytochrome c has a maximum absorption at 410 nm. When cytochrome was reduced by the ‚O2- radical ion, the absorption was decreased. This phenomenon was observed during the H2O2 decomposition using the resin as the catalyst. Without the catalyst, the absorption of cytochrome c in H2O2 solution was almost unchanged. Addition of 10 mg of catalyst to the solution led to a rapid decrease in the absorption at 410 nm. These results indicated that superoxide radical ions were produced during the decomposition of H2O2. The ‚O2- radical formed during the decomposition of H2O2 was also successfully detected by ESR spectroscopy. The g⊥ value is not influenced by the pH of the solution;20,21 therefore, it is possible to determine ‚O2- at g⊥ ) 2.001. After 10 min, ∼10-5 M ‚O2- radical could be found in a 10 mL of the 0.1 M H2O2 solution containing 10 mg of resin. The speed of the radical formation decreased as the reaction time increased, which could be due to the decreasing concentration of H2O2. Cobalt States Existing on the Resin Surface. The cobalt state is a key point of this solid catalyst. In Figure 6A,B, two ESCA (electrospectroscopy for chemical analysis) spectra were compared before and after the MEA-Co(II)-resin was treated with H2O2 solution. From the data book,49 CoF3 and CoSO4 have their 2p3/2 binding energy at ranges of 782.1-782.8 and 783.7-784.4 (48) Asai, R.; Matsukawa, R.; Ikebukuro, K.; Karube, I. Anal. Chim. Acta 1999, 390, 237-244.

eV, respectively. Although these two data do not directly correspond to Co(II)-MEA or Co(III)-MEA complexes, they are related to the states of Co(II) and Co(III). When the electronically generated Co3+ ion23,24 was immobilized on the resin, the color of the resin was orange. After it combined with the MEA, the color changed to brownish green without the H2O2 treatment (see the color change in Figure 2). The ESCA spectrum of the brownish green product was the same as that in Figure 6B. Combining the ESCA spectra with the color changes of the resin when it was treated with CoSO4 and MEA solutions, we believe that the peaks at binding energies 784.3 and 782.8 eV correspond to Co(II)MEA and Co(III)-MEA, respectively, on the surface of the resin. The two peaks at Figure 6A could be explained by partial oxidation of Co(II)-MEA by oxygen in the air. After the resin was treated with H2O2 solution, all of the Co(II)-MEA complexes were oxidized to the Co(III)-MEA state, which gives only one peak in the 780-785 eV band (Figure 6B). CL Spectra and the Emitter. The CL spectra of the present luminol-H2O2 system were shown in Figure 7. Curves A and B correspond to luminol in unbuffered solution (or at a pH 5.7 for 0.01 M acetate buffer) and in 0.05 M NaHCO3-NaOH buffer (pH ) 10.5), respectively. This figure indicated that the maximum CL emission wavelengths of luminol are 448.2 nm under unbuffered or weak acidic conditions and 425.6 nm under basic conditions. It is well-known that the maximum CL wavelength of luminol in basic solution is near 425 nm,2-4,7-9 but the maximum CL spectrum (λmax ) 448.2 nm) of luminol in unbuffered solution is presented here for the first time. To identify the luminol CL emitter in unbuffered or in weak acidic solution, the fluorescent spectra of 3-aminophthalate, which has been confirmed as the emitter3,7 of luminol CL in basic solution, were also determined in media of two different pHs. The maximum fluorescent wavelengths (λex ) 304 nm) of 3-aminophthalate in buffered (pH 10.5) and unbuffered solutions (pH 5.7) were 421.8 and 453.6 nm, respectively; therefore, it is certain that the emitter of the present heterogeneous luminol-H2O2 system is 3-aminophthalate, the same as in the usual luminol CL reaction. Luminol CL Emission in Unbuffered Solution. On the basis of the above results, a possible mechanism for the CL of the luminol-H2O2 reaction with the resin as catalyst was proposed in Figure 8. When the unbuffered luminol carrier solution flows through the resin phase, the resin in Co(II)-MEA form is oxidized to Co(III)-MEA form by the dissolved oxygen, and a very small amount of ‚O2- is generated. This is a slow reaction, and the formation of ‚O2- radical causes a small background CL emission. The support for this proposition was the effects of the dissolved O2 and N2 in the carrier and the H2O2 injection interval on the CL intensities. After injection of H2O2 solution to the carrier, H2O2 is oxidized by the MEA-Co(III)-resin and the superoxide radicals (49) Moulder, J. F.; Stickle, W. F.; Sobol, P. E.; Bomben, K. D. Handbook of X-ray Photoelectron Spectroscopy; Physical Electronics Inc.: 1995, p 81.

Figure 8. Possible mechanism of unbuffered luminol-H2O2 CL system with MEA-Co(II)-resin as catalyst.

are generated. This is a fast reaction. The ‚O2- radical attacks the luminol molecule and brings out CL. The emitter of the CL reaction is the excited state of 3-aminophalate, which is the same as the lumniol CL reaction in basic solution. CONCLUSIONS This is the first time that the Co(II)-MEA-immobilized resin has been used as a catalyst in the chemiluminescence of the luminol-H2O2 system. Because of the unbuffered conditions, high selectivity, and sensitivity, the proposed method was developed as a CL flow-through sensor for the determination of H2O2 in rainwater and glucose in urine and juice. The above experiments strongly supported the proposed CL mechanism of luminol in neutral or weakly acidic solution as shown in Figure 8. The effects of dissolved O2, ‚O2-, H2O2, and the catalyst on the CL were studied. The present work told us that luminol could emit not only in basic but also in weak acidic solutions. Applications of the present method in environmental and clinical analyses are possible. ACKNOWLEDGMENT This work was supported in part by a Grant-in-Aid for Scientific Research from the Japan Society for Promotion of Science (13640608) and National Natural Science Foundation of China, which we gratefully acknowledge. Received for review May 24, 2001. Accepted August 3, 2001. AC010573+

Analytical Chemistry, Vol. 73, No. 21, November 1, 2001

5051