Electrochemiluminescence of Luminol in Alkaline Solution at a

Reagents and Solutions. Luminol was obtained from Merck. Spectroscopic graphite rods (SES-6) with a 0.6-mm diameter were obtained from Shanghai Carbon...
0 downloads 0 Views 150KB Size
Download PDF
Anal. Chem. 2003, 75, 324-331

Electrochemiluminescence of Luminol in Alkaline Solution at a Paraffin-Impregnated Graphite Electrode Hua Cui,* Gui-Zheng Zou, and Xiang-Qin Lin*

Department of Chemistry, University of Science and Technology of China, Hefei, Anhui 230026, People’s Republic of China

The behavior of luminol electrochemiluminescence (ECL) at a paraffin-impregnated graphite electrode (PIGE) at different applied potentials was studied. Five ECL peaks were observed at 0.31, 0.59, 1.09, 1.54, and -0.58 V versus SCE, respectively, being related to potential scan direction and ranges, N2, O2, pH of the solution, and KCl concentration. The emission spectra of various ECL peaks at different potentials showed that all ECL peaks were initiated by luminol reactions. X-ray diffraction demonstrated that a simple mixture was formed between graphite and paraffin. The fluorescence spectra on the surface of the PIGE suggested that certain groups on the graphite were oxidized when the positive potential was applied to the electrode. In the presence of O2, three main ECL peaks were obtained in 0.1 mol/L KCl at pH 12.2. The ECL peak at 0.59 V with a shoulder is likely due to the reaction of luminol radicals with O2 and further electrooxidation of luminol radicals. The ECL peak at 1.54 V was suggested to be due to the electrooxidation of OH- to HO2- at higher potential and then to O2•-, which reacted with luminol to produce light emission. Moreover, the oxygen-containing functional groups formed by the oxidation of the surface of the graphite electrode might enhance the ECL. At -0.58 V, the dissolved oxygen in solution was reduced to HO2-, resulting in light emission. At a potential higher than 1.64 V, ClO- was formed, leading to a broad emission wave and enhancement of the ECL peak at -0.58 V upon the reversal scan. Under nitrogen atmosphere, an ECL peak appeared at 1.09 V. At this potential, OH- was oxidized to O2, followed by the reaction with luminol to generate light emission. At pH 13.2 or 0.5 mol/L KCl, the shoulder of the ECL peak at 0.59 V became an ECL peak at 0.31 V. The conversion of luminol radicals into excited 3-aminophthalate may undergo two routes. Under these conditions, two routes might proceed at a different rate to form another ECL peak. It is concluded that luminol ECL could be readily excited by various oxygen-containing species electrogenerated at different applied potentials. Three strong ECL peaks obtained at different potentials on the PIGE might be of a potential to improve analytical * Corresponding authors. E-mail: [email protected]; [email protected].

324 Analytical Chemistry, Vol. 75, No. 2, January 15, 2003

selectivity and sensitivity for the detection of some analytes. Electrogenerated chemiluminescence (ECL) has become an important and valuable detection method in analytical chemistry in recent years.1 ECL is conventionally initiated by using various electropulse signals such as symmetric double-step potential,2,3 triangle pulse,4 and low-frequency5 and high-frequency6 squarewave potentials in order to obtain high sensitivity. The dependence of ECL on the applied potential has not received much attention although Kuwana7 and Rubinstein and Bard8 measured ECL at different potentials. Recently, some studies on ECL using slowly scanning potentials have been carried out. ECL was found to depend on applied potential, electrode material, and surface state of the electrode. In our previous work,9,10 luminol ECL was studied at glassy carbon (GC) and platinum electrodes under conventional cyclic voltammetric (CV) conditions. Four ECL peaks were resolved on both Pt and GCE electrodes, and the later included a cathodic ECL peak. Lucigenin ECL was also examined under CV conditions, and two ECL peaks were observed on GC electrode.11 In Bard’s group,12 two ECL waves were found on Pt, Au, and GC electrodes when the ECL of Ru(bpy)32+/tripropylamine (TPrA) was studied under CV conditions. Although the ECL waves were obtained by simply varying the potential, they are rarely studied in the literature. The curves of ECL intensity (I) versus potential (E) (I-E curves) are comparable with cyclic voltammograms, but more sensitive than cyclic voltammograms, showing rich and complex information about the electrodeelectrolyte system. They are very useful for mechanistic studies of ECL reactions and electrode reactions. Carbon in its various forms has become a popular material for the construction of solid electrodes used in electroanalytical chemistry. Spectroscopic graphite is particularly useful for the study of the oxidation of organic species due to the large positive potential range over which it can be used in aqueous solutions. It (1) Fa¨hnrich, K. A.; Pravda, M.; Guibault, G. G. Talanta 2001, 54, 531-559. (2) Haapakka, K. E.; Kankare, J. J. Anal. Chim. Acta 1982, 138, 263-275. (3) Haapakka, K. E. Anal. Chim. Acta 1982, 141, 263-268. (4) An, J. R.; Chen, X.; Chen, H. Chin. J. Anal. Chem. 1988, 16, 127-132. (5) Fleet, B.; Keliher, P. N.; Kirkbright, G. F.; Pickford, C. J. Analyst 1969, 94, 847-854. (6) Pastore, P.; Magno, F.; Collinson, M. M.; Wightman, R. M. J. Electroanal. Chem. 1995, 397, 19-26. (7) Kuwana, T. J. Electroanal. Chem. 1963, 6, 164-167. (8) Rubinstein, I.; Bard, A. J. J. Am. Chem. Soc. 1980, 102, 6641-6642. (9) Lin, X. Q.; Sun, Y. G.; Cui, H. Chin. J. Anal. Chem. 1999, 27 (5), 497-503. (10) Sun, Y. G.; Cui, H.; Lin, X. Q. Acta Chim. Sin. 2000, 58 (5), 567-571. (11) Sun, Y. G.; Cui, H.; Lin, X. Q. J. Luminesc. 2001, 92, 205-211. (12) Zu, Y. B.; Bard, A. J. Anal. Chem. 2000, 72, 3223-3232. 10.1021/ac0201631 CCC: $25.00

© 2003 American Chemical Society Published on Web 01/14/2003

Figure 1. Electrochemiluminescence cell assembly. WE, PIGE; RE, Ag reference electrode; CE, platinum wire; B, porous glass filter.

is easily modified with polymer films to improve the selectivity of electroanalysis due to its porousness.13 The aim of this work is to explore the behavior of luminol ECL at the paraffin-impregnated graphite electrode (PIGE) at different applied potentials. It might be of potential to improve the sensitivity and selectivity of luminol ECL for the detection of some compounds by use of different electrode potentials and electrode materials. At least five ECL peaks were observed at different applied potentials. The effects of potential scan direction and range, N2, O2, pH, and KCl concentration on luminol CL were examined. The chemiluminescence (CL) spectra of various ECL waves at different potentials were analyzed. The surface status of the PIGE was characterized with X-ray diffraction and fluorescence spectra. The mechanism for luminol ECL at the PIGE was proposed. EXPERIMENTAL SECTION Instrumentation. ECL was conducted by a homemade ECL system, including a model CHI832 electrochemical working station (Chenhua Inc., Shanghai), an H-type electrochemical cell (homemade), a model 1P21 photomultiplier tube (PMT) (Beijing, China), a model GD-1 luminometer (Xi’an, China), and a computer. During measurement, the potential was applied to the working electrode via the electrochemical working station and ECL was generated. The observation window was placed in front of the PMT biased at -600 V. The output current of the PMT was amplified by the luminometer and was recorded through a 1-kΩ resister to the second WE port of the CHI832 with “potential” off. The H-type ECL cell was constructed as shown in Figure 1.11 One arm of the H-cell served as the working compartment, containing a Teflon electrode rod with a side PIGE working electrode (28.3-mm2 surface area) and a closely accompanying Ag reference electrode. Another arm of the H-cell served as the auxiliary chamber, in which a platinum counter electrode was inserted. The working compartment was separated from the auxiliary compartment by a piece of porous glass filter, which was used to prevent solution mixing. To prevent interference from the (13) Cuory, L. A.; Heineman, W. R. J. Electroanal. Chem. 1988, 256, 327-341.

diffusion of redox products at the counter electrode, the auxiliary compartment was filled with blank solution for experiments. The auxiliary arm was also covered by black tape in order to prevent any light emission from the auxiliary processes. The silver wire reference electrode placed in a reference channel in the Teflon electrode rod had an open end close to the edge of the working electrode in order to have precise potential control. The advantage of the use of an Ag quasi-reference electrode (AgQRE) is predominately the simplicity for cell construction and quick potential response. Although the potential of the AgQRE was found essentially stable during an experiment, it had to be measured under each experimental condition before and after the experiment. ∆E ) EAg/Ag+ - ESCE in different solutions was given for potential calibrations. The potentials reported in this work are all calibrated as the potentials versus SCE. A model MXPAHF rotating anode X-ray diffractometer (MAC) and a model RF-5301PC spectrofluorophotometer (Shimadzu) were used for characterization of the electrode surface. Reagents and Solutions. Luminol was obtained from Merck. Spectroscopic graphite rods (SES-6) with a 0.6-mm diameter were obtained from Shanghai Carbon Works. Paraffin wax (mp 55-59 °C) was obtained from Shanghai Specimen Model Inc. A 99.999% pure nitrogen and oxygen supply was used. A 1.0 × 10-2 mol/L stock solution of luminol was prepared by dissolving luminol in 0.1 mol/L sodium hydroxide solution without purification. Working solutions of luminol were prepared by dilution of the stock solution. All reagents were of analytical grade, and redistilled water was used throughout. Procedures. The graphite working electrode was impregnated with paraffin until it was saturated under infrared light prior to use. The paraffin-impregnated graphite electrode was polished with filter paper (GB/T1914-93 ASPME 832-81, Fuyang Special Paper Co., Hangzhou, China), rinsed with water, and dried with filter paper. A 2.0-mL portion of the sample solution and the same size blank solution without luminol were added to the working arm and the auxiliary arm of the ECL cell, respectively. When the potential was applied to the working electrode, ECL was generated. The curves of ECL intensity versus the applied potential (IE) and the curves of current versus the applied potential (i-E) were recorded simultaneously. However, if the blank solution without luminol was added to the working arm, and the sample solution to the auxiliary arm, no light emission was observed from the working electrode. This confirmed that all recorded light emissions were from the working processes. The experiments were carried out under air-saturated atmosphere, nitrogen atmosphere, and oxygen atmosphere, respectively, for comparison. Under nitrogen or oxygen atmosphere conditions, nitrogen or oxygen was bubbled through the solutions for 15 min in both arms of the cell and keep flowing over the solution during experiments. CL spectra of various ECL peaks at different potentials were measured by inserting the filters at wavelengths of 350, 390, 430, 450, 470, 500, and 590 nm (light cannot pass at wavelengths lower than these wavelengths). ∆Ifλ (the difference between ECL intensity with the filter at λl and ECL intensity with the filter at λh; λl is a lower wavelength, and λh is a higher wavelength) was calculated as shown: e.g., ∆If350 ) Iblank(without filter) - I350, ∆If390 ) I350 - I390, ∆If430 ) I390 - I430, etc. The curves of ∆Ifλ versus λ are consistent with CL spectra. Analytical Chemistry, Vol. 75, No. 2, January 15, 2003

325

Figure 2. I-E curves and CV curves of luminol under air-saturated atmosphere. Conditions: luminol, 1 × 10-4 mol/L; KCl, 0.1 mol/L; pH 12.2. CV (A, C) and I-E (B, D) curves with initial anodic (A, B) and cathodic (C, D) scan directions; first-order differential voltammograms in the luminol solution (a) and in blank solution (b). EAg/Ag+ - ESCE ) 35 mV. 1-5 represent ECL-1-5; Circled 1 and 3 represent shoulder and tail, respectively.

RESULTS AND DISCUSSION Cyclic Voltammograms and ECL of Luminol under Air Saturation. Cyclic voltammograms and I-E curves of luminol in alkaline solution under air-saturated conditions are shown in Figure 2. In CVs (Figure 2A), an anodic cyclic voltammetric peak (cvp4) was observed at 1.54 V (vs SCE) on the positive scan. Upon reversal of the potential scan from 1.84 V, a cathodic peak (cvp5) was obtained at -0.87 V. Moreover, on the positive scan, other three anodic peaks (dvp1, dvp2, dvp3) were found at 0.31, 0.59, and 1.09 V in the first-order differential voltammograms. Among them, dvp1 and dvp2 did not appear in blank solution without luminol. This implied that dvp1 and dvp2 correlated to the oxidation of luminol, whereas dvp3 and cvp4 were hypothesized to be the oxidation of OH- and the PIGE. Cvp5 is the reduction of the dissolved oxygen in solution to HOO- according to our previous studies on luminol ECL at a glass carbon electrode.9 In I-E curves (Figure 2B), corresponding to three of five voltammetric peaks (dvp2, cvp4, cvp5), two ECL peaks (ECL-2, ECL-4) were observed at 0.59 and 1.50 V, respectively, on the positive scan, and one ECL peak (ECL-5) at -0.58 V upon the reversal scan. Moreover, corresponding to two other voltammetric peaks (dvp1, dvp3), ECL-2 at 0.59 V had a shoulder and tail at 0.31 (position 1) and 1.09 V (position 3), respectively. When the potential scan direction was changed, i.e., from 0.04 f -1.57 f 0.04 f 1.84 f 0.04 V (Figure 2C and D), the same three ECL peaks appeared. However, ECL-2 increased, ECL-4 did not change, and ECL-5 decreased. The results revealed that ECL-2 might be enhanced by species produced via cvp5, whereas the compound produced during the positive potential scan could increase ECL-5. 326 Analytical Chemistry, Vol. 75, No. 2, January 15, 2003

Luminol ECL under Nitrogen Atmosphere. Nitrogen was used to remove the dissolved oxygen in the solution, and the I-E and CV curves of luminol under nitrogen atmosphere obtained are shown in Figure 3. In this case, besides three ECL peaks under air atmosphere (Figure 2B and D), the tail (position 3) of ECL-2 became a separated peak ECL-3. ECL-2 decreased remarkably, which suggested that ECL-2 attributed to the electrooxidation of luminol could be significantly enhanced in the presence of dissolved oxygen. ECL-4 did not change in the intensity. ECL-5 still occurred, which is consistent with the fact that the oxidation of OH- to O2 at the higher positive potentials was not affected by the removal of O2. When the potential scan started with cathodic direction, both of ECL-5 and cvp5 disappeared, confirming that ECL-5 depended on the dissolved oxygen in the solution. Luminol ECL under Oxygen Atmosphere. ECL behavior of luminol under oxygen atmosphere was also studied. Under oxygen atmosphere, ECL-2 increased and ECL-4 did not change, indicating oxygen could enhance ECL-2. When the potential scan started with a cathodic direction, ECL-5 enhanced slightly owing to the increase of the concentration of oxygen in the solution. Effect of the Potential Scan Range. If the reversal of the potential scan was from 1.64 V after cvp4 under air-saturated condition, ECL-5 became weaker and the counterpeaks of ECL-2 and ECL-4 appeared clearly. The results suggested that electrochemical or chemical reactions at the cvp4 step led to the disappearance of the counterpeaks and the enhancement of ECL-5. Under nitrogen atmosphere, the switching potential was selected as 0.84, 1.24, and 1.64 V, respectively, for a comparison. It was found that no cvp5 was observed for the scan switched

Figure 3. I-E curves and CV curves of luminol under N2 atmosphere. Conditions: luminol, 1 × 10-4 mol/L; KCl, 0.1 mol/L; pH 12.2. CV (A, C) and I-E (B, D) curves with initial anodic (A, B) and cathodic (C, D) scan directions. EAg/Ag+ - ESCE ) 35 mV.

Figure 4. I-E curves and CV curves of luminol at different concentrations of KCl under air-saturated atmosphere. Conditions: luminol, 1 × 10-4 mol/L; pH 12.2. KCl, 0.0 mol/L (A, B), EAg/Ag+ - ESCE ) 137 mV; 0.5 mol/L (C, D), EAg/Ag+ - ESCE ) 24 mV. CV (A, C) and I-E (B, D) curves with initial anodic scan direction.

from 0.84 V; however, cvp5 was obtained from 1.24 or 1.64 V. Therefore, it is deduced that dvp3 at 1.09 V was due to the oxidation of OH- to O2. Effect of KCl Concentration. Figures 2 and 4 show the I-E and CV curves of luminol at 0.1, 0.0, and 0.5 mol/L KCl,

respectively. ECL-2 decreased with the increase of KCl concentration. Without KCl, the background current after cvp4 almost disappeared, the corresponding counterpeaks of ECL-2 and ECL-4 appeared, and the intensity of ECL-5 was the same as the case initiating the potential scan cathodically (Figure 2D). The peak Analytical Chemistry, Vol. 75, No. 2, January 15, 2003

327

potential of ECL-2 and ECL-4 shifted ∼+100 mV and that of ECL-5 ∼-100 mV, which was caused by the change of ion strength in the solution. With 0.5 mol/L KCl, the shoulder (position 1) of ECL-2 in Figure 2B and D split to another peak ECL-1 from ECL2. ECL-2 and ECL-4 decreased obviously. With switching potential of 1.75 V, a strong broad light emission peak was observed. It was deduced that Cl- was oxidized to ClO- at higher positive potentials,14 followed by the reaction with luminol to generate the light emission. This is the reason the broad light emission wave was observed upon the reversal scan and the intensity of light emission increased with the increase of KCl concentration. ClOis likely to be produced beyond the potential of 1.64 V because no broad light emission wave was found with switching potential of 1.64 V or with switching potential of 2.14 V in solutions without KCl. In the case of ClO- generation, ECL-5 was also observed to be enhanced, which is likely due to ClO- reacting with HO2electrogenerated at negative potentials to produce fresh atomic oxygen [O], resulting in the enhancement of ECL-5.15 Halide ions are well-known fluorescence quenchers. The reason Cl- decreased ECL-2 and ECL-4 could be that, in the presence of Cl-, excited emitter was transferred into excited triplet state, leading to quenching of ECL-2 and ECL-4. Effect of pH. I-E and CV curves of luminol at pH 12.02, 11.20, and 13.20 were also studied. At pH 11.20, all ECL peaks decreased. ECL-4 and cvp4 almost disappeared, indicating that the electrochemical reactions corresponding to cvp4 were related to the concentration of OH-. At pH 13.20, all ECL peaks also decreased, but ECL-1 and ECL-2 were clearly resolved. After ECL-4 appeared, the light intensity continued to increase and the corresponding cvp4 became indistinct, postulating that other subsequent electrochemical reactions occurred. Effect of Luminol Concentration. The luminol concentrations of 0.5 × 10-4, 1.0 × 10-4, and 2.0 × 10-4 mol/L were tested as shown in Figure 5. It was found that the intensity of all ECL peaks increased with the increase in concentration of luminol. Therefore, the light emission observed at different potentials depended on luminol. CL Spectra of Various ECL Peaks. The CL spectra of various ECL peaks in I-E curves were analyzed as shown in Figure 6. The CL spectra of ECL-2, ECL-4, ECL-5, and ECL-6 (broad light emission wave at 0.84 V upon reversal scan) were obtained under Figure 2 conditions. ECL-1 and ECL-3 occurred at pH 13.0 and under nitrogen atmosphere (Figure 3), respectively, and thus their CL spectra were measured under these conditions. Although ECL-1 and ECL-3 partly overlapped with ECL-2, their CL spectra at selected potentials were mainly contributed by ECL-1 and ECL3. The results show that the maximum emission of all peaks 1-6 is at 425 nm, corresponding to the light emission of 3-aminophthalate.1 It was concluded that all ECL peaks were initiated by luminol reactions. Characterization of Electrode. The electrode was characterized by X-ray diffraction and fluorescence spectra as shown in Figure 7 and Figure 8, respectively. X-ray diffraction results demonstrated that two new peaks appeared at 2θ of 21.3200 and 23.6600, respectively, after adding paraffin, corresponding to that (14) Flis, I. E.; Mishchenko, K. P.; Troitskaya, N. V. Zh. Fiz. Khim. 1959, 33, 1744-1749. (15) Lu, X. H.; Lu, M. G. Chin. J. Spectrosc. Spectral Anal. 1994, 14 (1), 123127.

328 Analytical Chemistry, Vol. 75, No. 2, January 15, 2003

Figure 5. Effect of luminol concentration on ECL peaks under airsaturated atmosphere. Conditions: luminol, (a) 1.0 × 10-4, (b) 0.5 × 10-4, and (c) 2.0 × 10-4 mol/L; KCl, 0.1 mol/L; pH 12.2. EAg/Ag+ ESCE ) 35 mV.

Figure 6. CL spectra of ECL peaks at different potentials. Under Figure 2 conditions, ECL-2, 0.60 V; ECL- 4, 1.56 V; ECL-5, -0.58 V; ECL-6 (broad light emission wave upon reversal scan), 0.84 V. At pH 13.2, ECL-1, 0.29 V. Under N2 atmosphere, ECL-3, 0.96 V. For ECL-1 and ECL-3, high voltage used for the photomultiplier was -800 V. For other ECL peaks, it was -600 V.

of paraffin. This indicated that a simple mixture was formed between graphite and paraffin. When the potential was scanned from 0 to 1.84 V, no significant changes were observed in X-ray diffraction spectra. The fluorescence spectrum of PIGE was consistent with that of graphite, and only one peak was observed at 363.6 nm, implying that the fluorescence was from a certain group in the graphite molecule. However, if the potential was scanned from 0 to 1.84 V, the fluorescent peak shifted to 369.3 nm. Graphite was reported to be easily electrooxidized to form oxygen-containing functional groups such as carboxyl, hydroxyl, carbonyl, lactone, and quinone-like groups.16 It is possible that the group with fluorescence was oxidized at positive potentials, resulting in the shift of the fluorescent peak. (16) Evans, J. F.; Kuwana, T. Anal. Chem. 1977, 49 (11), 1632-1635.

peak because these two oxidation peaks could be seen even at very high pH. It was hypothesized that dvp1 and dvp2 correspond to the oxidation of luminol to the luminol radical (L•-) and the oxidation of L•- to 3-aminophthalate (AP2-*), respectively. Previous studies suggested that oxidative species such as O2 or OOH- were necessary for the conversion of luminol radicals into AP2-*.2,18,19 However, in this work, it seems that luminol radical L•- could be converted into AP2-* in the absence of O2 or OOH- because ECL-2 could still be observed under nitrogen atmosphere. It is evident that luminol radical L•- can be converted into AP2-* via direct electrooxidation. In the presence of O2, the mechanism for ECL-2 was suggested to be that luminol anion was oxidized to luminol radical (L•-) at 0.31 V( eq 1),

ECL-2: Figure 7. X-ray diffraction pattern of graphite, PIGE, and electrooxidized PIGE.

Figure 8. Fluorescence spectra of graphite, PIGE, and electrooxidized PIGE. Excitation wavelength, 220 nm.

Mechanism of Luminol ECL at PIGE. CL spectra demonstrated that the emitter of all ECL peaks was 3-aminophthalate. Therefore, all ECL peaks were initiated by the reactions of luminol with the species electrogenerated at different potentials. Luminol ECL at Anode. The experimental results showed that ECL-1 was invisible and ECL-2 happened with a shoulder in most of cases. ECL-1 only appeared at pH 13.20 and at 0.5 mol/L KCl and overlapped partly with ECL-2. The peak potential of ECL-1 and ECL-2 was consistent with that of dvp1 and dvp2 in CVs. Therefore, ECL-1 and ECL-2 were related to the oxidation of luminol. Moreover, ECL-2 could be enhanced by oxidative species such as O2 and O2•-. Sakura17 studied the oxidation behavior of luminol in a pH 7.4 aqueous solution at a glassy carbon electrode. Two distinct oxidation peaks were seen at 0.48 and 0.60 V versus SCE, corresponding to a one-electron oxidation of luminol monoanion to diazasemiquinone radical (L•-). The first oxidation peak was claimed as a prepeak due to the oxidation of luminol absorbed on the electrode surface, the second one was the diffusion peak. He indicated that the prepeak disappeared at pH >9.0 due to more hydroxide ions existing on the electrode surface, interfering with the absorption of luminol. Therefore, in our case, dvp1 and dvp2 were difficult to explain as the prepeak and diffusion (17) Sakura, S. Anal. Chim. Acta 1992, 262, 49-57.

LH- - e f LH• f L•- + H+ (Ep ) 0.31 V)

(1)

L•- + O2 f O2•- + L

(2)

L•- + O2•- f LO22-

(3)

LO22- f AP2-* + N2

(4)

AP2-* f AP2- + hν

(5)

followed by further oxidation of L•- to AP2-*. The conversion of luminol radicals into AP2-* may occur by two routes: (a) luminol radicals were oxidized to AP2-* by O2 dissolved in the solution (eqs 2-5);2,18,19 (b) luminol radicals were oxidized to AP2-* at 0.59 V electrochemically (eq 6), which has also been proposed for luminol ECL on Pt electrode.4 In most of cases, routes a and b might almost happen simultaneously and thus ECL-2 appeared with a tiny shoulder. At pH 13.20 and at 0.5 mol/L KCl, routes a and b might proceed with different rates, and thus ECL-1 overlapped partly with ECL-2 appearance. If the initial scan direction was cathodic, OOH- was produced at negative potentials.

O2 + H2O + 2e f OOH- + OH- (Ep ) -0.87 V) (7) Upon the reversal scan, OOH- was oxidized to O2•- (eq 8),2 which could enhance ECL-2 via eqs 3-5.

OOH- - e f HOO• / O2•-

(8)

In most of cases, ECL-3 was invisible and only showed as a tail of ECL-2. However, ECL-3 could be clearly seen under nitrogen atmosphere although it overlapped partly with ECL-2. Under this condition, the experiments controlling the switching potential (18) Klingler, W.; Strasburger, C. J.; Wood, W. G. Trends Anal. Chem. 1983, 2 (6), 132-136. (19) Shevlin, P. B.; Neufeld, H. A. J. Org. Chem. 1970, 35 (7), 2178-2182.

Analytical Chemistry, Vol. 75, No. 2, January 15, 2003

329

showed that the reduction wave cvp5, corresponding to reduction of dissolved oxygen to HO2-, was only observed in the switching potential between 0.84 and 1.24 V. There was no oxygen in the solution under nitrogen atmosphere. Oxygen was from oxidation of OH- at positive potential. Therefore, dvp3 at 1.09 V was due to the oxidation of OH- to O2. Earlier studies also demonstrated that OH- was readily oxidized to O2 at positive potential.20 ECL-3 is likely due to the oxidation of OH- to O2, followed by the reaction with luminol.

4OH- - 4e f O2 + 2H2O (Ep ) +1.09 V)

(9)

LH- + O2 ff AP2-*

(10)

from 2.14 V in solutions without KCl. ClO- reacted with luminol to generate a broad emission wave upon the reversal scan.

Cl- + OH- - 2e f ClO- + H2O (Ep g 1.64 V) (14) LH- + ClO- ff AP2-*

Luminol ECL at Cathode. It is believed that cvp5 is the reduction of the dissolved oxygen in solution to OOH-.9 The pathways for ECL-5 could be

O2 + H2O + 2e f OOH- + OH-

(Ep ) -0.87 V) (16)

LH- + OOH- f f AP2-* It is well known that the reaction rate of luminol with O2 or hydrogen peroxide is slow in the absence of catalysts, which gives weak chemiluminescence. Indeed, ECL-3 was weak. The experiments indicated the ECL-3 could be clearly resolved only under nitrogen atmosphere. The reason could be that the balance in eq 9 shifted to the right under nitrogen atmosphere. It seems that cvp4 at 1.59 V was not related to the potential scan direction, N2, O2, and Cl-. When KNO3 was used for the experiments instead of KCl or the concentration of KCl was zero, cvp4 still occurred. Moreover, cvp4 did not depend on electrode materials either because the oxidation wave at 1.44 V similar to cvp4 was observed on a gold electrode (It will be published elsewhere.). The results from the effect of pH showed cvp4 disappeared at lower pH. It is reported that OH- was readily oxidized to HO2- at higher positive potential (eq 11)20 and HO2-

3OH- - 2e f HO2- + H2O

(11)

HO2- + OH- - e / O2•- + H2O

(12)

LH- + O2•- f f AP2-*

(13)

could be further oxidized to O2•- (eq 12).2,20 Therefore, ECL-4 is likely due to the reaction of luminol with O2•-. The ECL, corresponding to the oxidation wave at 1.44 V, was weak on the gold electrode, whereas ECL-4 was strong on the PIGE. The fluorescence spectra of PIGE showed that the fluorescence spectrum shifted from 363.6 to 369.3 nm after the positive scan. It was deduced that the graphite was electrooxidized to form oxygen-containing functional groups such as carboxyl, hydroxyl, carbonyl, lactone, and quinone-like groups.16 Our recent work showed that polyphenolic compounds could enhance the luminol-induced ECL.21 The oxygen-containing functional groups formed by the oxidation of the surface of the graphite electrode might enhance the ECL. However, no such oxygen-containing functional groups were formed on the surface of gold electrode and thus weak emission was observed. If the potential was higher than 1.64 V, it is believed that Clwas oxidized to ClO- because no broad light emission wave was found with potential scans reversed from 1.64 V or scans reversed (20) Wu, W. C.; Fong, H. G.; Wu, K. Z. Handbook for Standard Electrode Potentials (Chinese); Science Press (China): Bejing, 1991; pp 158-159. (21) Sun, Y. G.; Cui, H.; Lin, X. Q.; Li, Y. H.; Zhao, H. Z. Anal. Chim. Acta 2000, 423, 247-253.

330

Analytical Chemistry, Vol. 75, No. 2, January 15, 2003

(15)

(17)

If the initial scan direction started anodically, ClO- was produced at higher positive potential. It is well known that ClO- might react with the HO2- obtained by reduction of dissolved O2 to form singlet oxygen, emitting light at a wavelength above 600 nm. However, no light emission was observed in the blank solution without luminol. It was evident that the enhanced ECL-5 was not related to the emission of singlet oxygen. Moreover, no light emission was observed in luminol solution if the band-pass filter of 590 nm was used. Lu15 studied the CL from the luminol-H2O2ClO- system. Much stronger light emission was obtained from the system than that from luminol-H2O2. Therefore, we believed that ECL-5 was enhanced by the following reaction

LH- + OOH- + ClO- ff AP2-*

(18)

CONCLUSIONS At the PIGE, at least five ECL peaks corresponding to five chemical or electrochemical reaction routes were found under CV conditions. The mechanism for ECL peaks was proposed due to electrooxidation of luminol and the reactions of luminol with various electrogenerated oxygen-containing species such as O2, O2•-, OOH-, and ClO- at different applied potentials. The behavior of luminol ECL under CV conditions observed on the PIGE is different from those on Pt and glass carbon electrodes. For luminol ECL, the PIGE can provide much stronger light emission than Pt, Au, and glass carbon electrodes, implying that the detection sensitivity by using luminol ECL can be improved on the PIGE. In most of cases, three strong ECL peaks can be obtained on the PIGE. Some organic compounds can be detected based on their enhancement and inhibition of luminol ECL.21 The correlation between electrode potential and ECL enhancement and inhibition was observed.21 The enhancement and inhibition of such three ECL peaks at different potentials by various organic compounds are probably different because they follow different luminol ECL reactions, which provides a possibility for the selective detection of some organic compounds by controlling the potential applied at the electrode. The effects of some polyphenols on luminol ECL at different potentials and analytical potential are under investigation. Five ECL peaks found in I-E curves on the PIGE corresponded to five redox peaks in CVs. However, only two redox peaks were visible in conventional CVs and the other three only in the first-order differential voltammograms. Moreover, ECL

peaks in I-E curves were much more sensitive than cyclic voltammetric peaks. Similar results were obtained in our previous studies on Pt and glass carbon electrodes. This implies that ECL peaks in I-E curves might be used as a new indicator of electrode reactions. Well-defined ECL counterpeaks, comparable to the counterpeaks of cyclic voltammetric peaks, were observed upon the reversal scan in some cases at the PIGE. The reason and the kinetics are not fully clear and further work is needed. ACKNOWLEDGMENT The support of this research by the National Natural Science Foundation of P.R. China (Grant 29875025), the National Education Committee of P.R. China (Grant 2000035816), and Oversea

Outstanding Young Scientist Program Of China Academy of Sciences are gratefully acknowledged. SUPPORTING INFORMATION AVAILABLE I-E curves and CV curves of luminol at the 0.04 f 1.64 f 0.04 f -1.57 f 0.04 V range, at different concentration of KCl with initial cathodic scan direction, and at different solution pH. This material is available free of charge via the Internet at http:// pubs.acs.org.

Received for review March 13, 2002. Accepted October 29, 2002. AC0201631

Analytical Chemistry, Vol. 75, No. 2, January 15, 2003

331