Chemiluminescent Reaction of Fluorescent Organic Compounds with

The decomposition of peroxomonosulfate (HSO5-) has been investigated by .... Heterogeneous Activation of Oxone Using Co3O4 ... Luminol Chemiluminescen...
0 downloads 0 Views 123KB Size
Anal. Chem. 2000, 72, 1148-1155

Chemiluminescent Reaction of Fluorescent Organic Compounds with KHSO5 Using Cobalt(II) as Catalyst and Its First Application to Molecular Imprinting Jin-Ming Lin* and Masaaki Yamada

Department of Applied Chemistry, Graduate School of Engineering, Tokyo Metropolitan University, 1-1 Minami-Ohsawa, Hachioji, Tokyo 192-0397, Japan The decomposition of peroxomonosulfate (HSO5-) has been investigated by chemiluminescence (CL). A weak CL was observed during mixing the HSO5- solution with the Co2+ solution in unbuffered conditions. An appropriate amount of fluorescent organic compounds (FOCs), such as dansyl amino acids and pyrene, was added to the KHSO5/Co2+ solution, a strong CL was recorded. A possible CL mechanism, based on studies of the fluorescence, CL, and UV-visible spectra and comparison of Co3+ oxidation ability with the SO4•- radical ion, was discussed. The CL from HSO5-/Co2+ is the emission of singlet oxygen produced from the catalytic decomposition of HSO5-. It was suggested that the decomposition of HSO5- in aqueous solution with Co2+ proceeds via oneelectron transfer to yield SO4•- radical ion. The FOC was attacked by SO4•- radical ion and oxidized to decompose into small molecules. During this proceeding, CL emission was given out. The present CL system has been developed as a flow injection analysis for FOCs. The detection limits (S/N ) 3) were in the concentration range 10-9-10-7 M for FOCs. Oxidation decomposition and CL emission of the analytes have been used in the molecular imprinting recognition. As an initial attempt, dansyl-Lphenylalanine was used as a template molecule and methacrylic acid and 2-vinylpyridine were used as functional monomers. The network copolymer imprinted with dansyl-L-phenylalanine exhibits an affinity for the template molecule. When the flowing streams of HSO5- and Co2+ solutions mixing through the molecularly imprinted polymer particles filled the flow cell, the template molecule, dansyl-L-phenylalanine reacted with the HSO5-/Co2+ solution and CL was emitted. The dansyl-L-phenylalanine was decomposed during the CL process, and the cavities of a defined shape and an arrangement of functional groups complementary to the template in the polymer were left for the next sample analysis. Although the first chemiluminescent (CL) phenomenon was reported more than 100 years ago.1 development of this research field was slow before the latter half of the 20th century. Application of the CL method in analysis became possible with appearance 1148 Analytical Chemistry, Vol. 72, No. 6, March 15, 2000

of the highly sensitive photomultiplier tube in the 1950s. The high sensitivity and the simple instrumentation of modern CL analysis have made many chemists interested in this field during the last 40 years. The most visible fact in the rapid development is reflected in the great numbers of publications devoted entirely to the discussion of CL.2-6 However, the high sensitivity of the CL method also faced a challenge of selectivity; most CL systems could not be used to determine the analytes directly. Many analysts made great efforts to improve the CL selectivity.7,8 Their efforts have been successful in the CL immunoassay for some analytes,9 but most CL systems were still used as the postcolumn detectors for, for example, high-performance liquid chromatography,10,11 gas chromatography,12 or capillary electrophoresis.13,14 The CL systems used as postcolumn detectors have also met their problems; e.g., liquid chromatography needs the eluent, which often is a mixture of organic solvent and salts. This eluent not only affects the separation but also influences the CL emission. Faced with the shortcoming of selectivity and the decomposition of some reactants, the research and uses of CL in analysis had * Corresponding author: (tel) +81-426-77-1111; (fax) +81-426-772821; (e-mail) [email protected]. (1) Radziszewski, B. Chem. Ber. 10, 1877, 70, 321-332. (2) Gundetrmann, K.-D. Chemilumineszenz Organischer Verbindungen; SpringerVerlag: Berlin, 1968. (3) Birks, J. W. Chemiluminescence and Photochemical Reaction Detection in Chromatography; VCH Publishers: New York, 1989. (4) Kricka, L. J. Anal. Chem. 1995, 67, 499R-502R. (5) Lin, J.-M.; Ishii, M.; Yamada, M. Bunseki 1998, 865-872 (6) Roda, A.; Pazzagli, M.; Kricka, L. J.; Stanley, P. E. Bioluminescence and Chemiluminescence Perspectiures for the 21st Century; John Wiley & Sons: Chichester, U.K., 1999. (7) (a) Robards, K.; Worsfold, J. P. Anal. Chim. Acta 1992, 266, 147-173. (b) Lewis, S. W.; Price, D.; Worsfold, P. J. J. Biolumin. Chemilumin. 1993, 8, 183-199. (c) Bowie, A. R.; Sanders, M. G.; Worsfold, P. J. J. Biolumin. Chemilumin. 1996, 11, 61-90. (8) Nieman, T. A. Handbook of Instrumental Techniques for Analytical Chemistry; Settle, F. A., Ed.; Prentice Hall: Upper Saddle River, NJ, 1997; pp 541-559. (9) Tsuji, A.; Kanno, T. Chemiluminescent Immunoassay; Life Sciences: Tokyo, 1992 (in Japanese). (10) Appelblad, P.; Jonsson, T.; Backstrom, T.; Irgum, K. Anal. Chem. 1998, 70, 5002-5009. (11) Dapkevicius, A.; Beek, T. A.; Niederlander, H. A. G.; Groot, A. Anal. Chem. 1999, 71, 736-740. (12) Hao, C.; Shepson, P. B.; Drummond, J. W.; Muthuramu, K. Anal. Chem. 1994, 66, 3737-43. (13) Garcia Campana, A. M.; Baeyens, W. R. G.; Zhao, Y. Anal. Chem. 1997, 69, 83A-88A. (14) Lin, J.-M.; Goto, H.; Yamada, M. J. Chromatogr., A 1999, 844, 341-348. 10.1021/ac9911140 CCC: $19.00

© 2000 American Chemical Society Published on Web 02/16/2000

decreased in the recent years. Endowing CL with high selectivity and making use of the reactant-destroying character will give this highly sensitive method a new activity in analytical field. Two effective research methods could be considered. One is to set up CL systems without special CL reagents, such as luminol and lucigenin. Even though the special CL reagents have high CL quantum yields, their CL emissions vary with the metal ions, buffer conditions, and other additives. A few relatively highly selective CL systems, such as the BrO-/OH- system for ammonium ion,15 Na2CO3/NaHCO3/Cu2+ system for sulfate,16 MnO4-/H+ system for ascorbic acid17 and KIO4/CO32+ system for H2O218,19 determinations, have been reported. Another consideration is to prepare a material that has molecular recognition and CL characters, that is, a molecular recognition CL sensor. To our knowledge, there is no report based on this idea and it may be due to the lack of a suitable CL system. Of interest is a report of the decomposition of HSO5-,20 a not-well-known oxidant. The HSO5- ion is highly reactive; in the presence of trace amounts of transition-metal ions, it decomposes rapidly. The transition-metal catalytic decomposition of HSO5- has been reported to initiate radical polymerization,21-23 and the investigations of HSO5- decomposition can be traced back to 40 years ago.20 It has been used to determine trace amounts of Co(II) by spectrophotometry24 or Co(II) and Fe(II) by a CL method.25 It is interesting to note that a strong CL emitted when a trace amount of fluorescent organic compounds (FOCs) was added to HSO5-/Co2+ solution.26 The FOCs include dansyl amino acids, fluorescent dyes, and many polycyclic aromatic hydrocarbons. But until now, there has been no detailed report concerned with the CL mechanism of the HSO5-/Co2+/ FOC reaction even though a rough CL mechanism has been considered.26 Our study showed that the FOCs were oxidized and decomposed during the reaction. It differs from the usual CL systems using FOCs as CL enhancers. This result encourages us to make use of the HSO5-/Co2+/FOC CL system in molecular imprinting. An unconventional combination of one of the highly sensitive detection methods with a new molecular recognition technique raises a new molecular recognition CL sensor. Molecular imprinting is a rapidly developing technique for the preparation of polymers having a high affinity for a target molecule. The molecularly imprinted polymer (MIP) is synthesized by polymerizing a “template” molecule with a functional monomer and a highly cross-linking monomer. The template is then removed, leading to a material containing cavities with threedimensional structure complementary in both shape and chemical (15) Hu, X.; Takenaka, N.; Takasuna, S.; Kitano, M.; Bandow, H.; Maeda, Y. Anal. Chem. 1993, 65, 3489-3492. (16) Lin, J.-M.; Hobo, T. Anal. Chim. Acta 1996, 323, 69-74. (17) Agater, I. B.; Jewsbury, R. A. Anal. Chim. Acta 1997, 356, 289-294. (18) Lin, J.-M.; Arakawa, H.; Yamada, M. Anal. Chim. Acta 1998, 371, 171176. (19) Lin, J.-M.; Yamada, M. Anal. Chem. 1999, 71, 1760-1766. (20) (a) Ball, D. L.; Edwards, J. O. J. Am. Chem. Soc. 1956, 78, 1125-1129. (b) Berliner, E.; Chen, M. M. J. Am. Chem. Soc. 1958, 80, 343-347. (21) Samal, R. K.; Sahoo, P. K.; Bhattercharjee, S. P. J. Mol. Catal. 1985, 33, 225-239. (22) Manivannan, G.; Maruthamuthu, P. Eur. Polym. J. 1987, 23, 311. (23) Vivekanandam, T. S.; Gopalan, A.; Vasudevan, T.; Umapathy, S. J. Polym. Sci., Part A: Polym. Chem. 1998, 36, 2715-2719. (24) Endo, M.; Ishihara, M.; Yotsuyanagi, T. Analyst 1996, 121, 391-394. (25) Makita, Y.; Suzuki, T.; Yamada, M.; Hobo, T. Nippon Kagaku Kaishi 1994, 701-706. (26) Tsukada, S.; Miki, H.; Lin, J.-M.; Suzuki, T.; Yamada, M. Anal. Chim. Acta 1998, 371, 163-170.

functionality to that of the template. This method was used for chromatographic separations,27 solid-phase extractions,28 membranes,29 antibody mimics and sensors,30 etc.31 Recent work with model systems has led to a number of receptors for amino acid derivatives.32,33 Using dansyl amino acid as the template molecule, the separation of dansyl D,L-amino acids was carried out. In this work, as an initial attempt, a polymer imprinted by dansyl-Lphenylalanine (dns-L-Phe) was prepared. This MIP can be used to recognize dns-L-Phe. When the HSO5-/Co2+ solution flows through the polymer, dns-L-Phe adsorbed on the MIP is oxidized to decompose and give out CL emission. After the CL reaction, the cavities for dns-L-Phe were reserved in the polymer. The detection limit of dns-L-Phe with the present CL sensor is 4 × 10-7 M. EXPERIMENTAL SECTION Reagents All of chemicals were of analytical grade and used as received. Water was obtained from a Milli-Q purification system (18.3 MΩ‚cm-1, Japan Millipore, Tokyo, Japan). Solutions of KHSO5 available in the form of a triple salt (2KHSO5‚KHSO4‚K2SO4) as Oxone (Aldrich Chemie, Steinheim, Germany) and cobalt(II) sulfate (CoSO4) (Kanto Chemical) were prepared daily. Stock solutions of pyrene, naphthalene, anthrancene, and their derivatives were prepared with water/acetonitrle solvent and standard solutions by diluting with water so that the content of the organic solvent was as low as possible. 2-Vinylpyridine (2-VPy), methacrylic acid (MAA), ethylene glycol dimethacrylate (EDMA), 2,2′azobis(2,4-dimethylvaleronitrile) (ABDV), dns-L-Phe, and other dansyl amino acids were obtained from Tokyo Kasei Kogyo (Tokyo, Japan). Stock solutions of dansyl amino acids were prepared with water. Apparatus The batch method for the CL profile was carried at a Lumicounter 600 (Microtec NITI-ON, Funabashi, Japan). A schematic diagram of the CL flow sensor is shown in Figure 1. A 0.3-0.4-g sample of MIP particles was packed into a 5.0 mm i.d. × 3.0 cm length glass tube. This glass tube was placed in front of a R585 photomultiplier tube (Hamamatsu Photonics, Shizuoka, Japan). The flowing streams of KHSO5 and CoSO4 solutions were controlled by an automatic switching valve (Sanuki Industry Co., Ltd, Tokyo, Japan). A PC recorder system (Tokken, Chiba, Japan) was used for recording and data treating. KHSO5 solution, CoSO4 solution, and carrier (H2O) were delivered to the flow sensor by minipumps (P1-P3). The solutions of samples were injected by means of a 100-µL loop valve injector placed close to the cell. Black Teflon tube (1 mm i.d.) was used for the flow lines. When the (27) Remcho, V. T.; Tan, Z. J. Anal. Chem. 1999, 71, 248A-255A. (28) Bjarnason, B.; Chimuka, L.; Ramstrom, O. Anal. Chem. 1999, 71, 21522156.. (29) Pilestsky, S. A.; Panasyuk, T. L.; Piletskaya, E. V.; Nicholls, I. A.; Ulbricht, M. J. Membr. Sci. 1999, 157, 263-278. (30) (a) Kriz, D.; Ramstrom, O.; Svensson, A.; Mosbach, K. Anal. Chem. 1995, 67, 2142-2144. (b). Masbach, K.; Ramstrom, O. Bio/Technology 1996, 14, 163-170. (c) Kriz, D.; Ramstrom, O.; Masbach, K. Anal. Chem. 1997, 69, 345A-349A. (31) (a) Wulff, G. Angew. Chem., Int. Engl. 1995, 34, 1812-1832. (b) Mosbach, K.; Haupt, K. J. Mol. Recognit. 1998, 11, 62-68. (c) Bartsch, R. A., Maeda, M., Eds. Molecular and Ionic Recognition with Imprinted Polymers; ACS Symposium Series 703; American Chemical Society: Washington, DC, 1998. (32) Sellergren, B.; Lepisto, M.; Mosbach, K. J. Am. Chem. Soc. 1988, 110, 58535860. (33) Lin, J.-M.; Nakagama, T.; Uchiyama, K.; Hobo, T. Chromatographia 1996, 43, 585-591.

Analytical Chemistry, Vol. 72, No. 6, March 15, 2000

1149

Figure 2. CL profiles in the batch system. The concentrations of HSO5-, Co2+, and dns-L-Phe are 2.0 × 10-2, 1.0 × 10-3, and 1.0 × 10-5 M, respectively. The volume of each reagent solution is 50 µL.

borosilicate glass test tube and dissolved in 20 mL of acetonitrile. Then 75 mg of initiator (ABDV) was added, and the tubes were purged with nitrogen before polymerization at 45 °C overnight (24 h). The resulting polymers were crushed and ground in a mechanical mortar and wet-sieved with water to a particle size of 100-200 µm. A reference polymer was prepared without dns-LPhe based on the same procedure.

Figure 1. Schematic diagram of the CL flow sensor prepared by molecularly imprinted polymer particles. P1, P2, and P3 are peristaltic pumps. V1 and V2 are the autoswitching valves used to control the flow states and sample injection. (a) CL detection: CoSO4 solution and KHSO5 solution were mixed in a 20-cm-length mixing coil and then flowed through the CL sensor. When the CL sensor is replaced by a spiral-type flow cell, it becomes a flow injection system for FOCs. (b) Sample injection and sensor washing. Photomutiplier tube (PMT, V ) -750 V) was used to receive the CL emission.

MIP particle-filled glass tube is replaced by a glass tube (15 cm length × 1.0 mm i.d.) made into a spiral-type flow cell, the detection system as shown in Figure 1a can be used for FOC flow injection analysis (FIA). The signals were recorded by a Shimadzu U-125MN recorder (Tokyo, Japan). A F-4010 fluorescence spectrophotometer (Hitachi, Tokyo, Japan) and a Shimadzu UV-2200 UV-visible spectrophotometer were used. The electrolytic cell for Co3+ ion preparation was the same as that used by Tanaka.34 Direct current for the electrolysis was supplied from a model 5248 regulated power supply (Metronix Corp., Tokyo, Japan). It was kept constant at 1.5 A for 2 h, which was sufficient for reaching a steady state, and then maintained at 0.2 A throughout the experiments in order to make up for the decrease of Co3+ concentration through its reaction with water. MIP Preparation. Polymers were prepared according to the method described by Ramstrom et al.35 MAA/2-VPy and EDMA were used as a functional monomer and cross-linker, respectively. A 1.64-mmol sample of dns-L-Phe, 65.6 mmol of EDMA, 3.28 mmol of MAA, and 3.28 mmol of 2-VPy were weighed in a 50-mL (34) Tanaka, H.; Morita, H.; Shimomura, S.; Okamoto, K. Anal. Sci. 1997, 13, 607-612. (35) Ramstrom, O.; Andersson, L. I.; Mosbach, K. J. Org. Chem. 1993, 58, 75627564.

1150

Analytical Chemistry, Vol. 72, No. 6, March 15, 2000

RESULTS AND DISCUSSION CL-FIA System for FOCs. Under unbuffered conditions, HSO5- is decomposed with Co2+ to yield reactive species (SO4•-, SO5•-, Co3+ , etc.), which are strong oxidizing agents leading to CL from FOCs. The CL profiles in the batch method for the mixtures of HSO5-, Co2+, and dns-L-Phe solution are shown in Figure 2. The results indicated that mixing of HSO5- solution with Co2+ solution gives evidence of CL emission, which will be the blank CL emission in the CL flow system. When the dns-L-Phe solution or pyrene solution was added to the HSO5-/Co2+ mixed solution, a stronger and sharper CL signal appeared. This result means that the reaction of FOCs with the reactive species is a fast CL process. The present CL system can be developed as a FIA for dansyl amino acids and other FOCs. The optimum conditions and the detection limits of some FOCs by CL-FIA are listed in Table 1, indicating that these compounds may be determined down to 1 × 10-9-1 × 10-7 M. The log-log plots of signal against FOC concentration were linear over a concentration range of 2-3 orders of magnitude. Explanation of the CL from HSO5- Decomposition over Co2+. There have been several reports of the metal ion-catalyzed decomposition of HSO5-,36 the reaction between HSO5- with Co2+ is believed to take two different courses:37

(36) (a) Thompson, R. C. Inorg. Chem. 1981, 20, 1005-1010. (b) Gilbert, B. C.; Stell, J. K. J. Chem. Soc., Perkin Trans. 2 1990, 1281-1288. (c) Cammarota, L.; Campestrini, S.; Carrieri, M.; Furia, F. D.; Ghiotti, P. J. Mol. Catal. A 1999, 137, 155-160. (37) Maruthamuthu, P.; Neta, P. J. Phy. Chem. 1977, 81, 937-940.

Table 1. Optimum Conditions and Analytical Figures of Merit of the CL-FIA System

compd

Optimum Conditions sample carrier (water), 3.0 mL/min flow rate 2 × 10-2 M KHSO5, 1.0 mL/min flow rate 1 × 10-3 M CoSO4, 1.0 mL/min flow rate

P1: P2: P3:

H2O NaN3

Dynamic Linear Range (M) dns-Phe 3 × 10-7∼1 × 10-5 dns-Gly 1 × 10-7∼2 × 10-5 flavin mononucleotide 2 × 10-8-1 × 10-5 pyrene 9 × 10-8-5 × 10-5 1-aminonaphathalene 6 × 10-8-7 × 10-6 1-aminoanthracene 1 × 10-8-1 × 10-5 dns-Phe dns-Gly dns-Val dns-Trp dns-Asp pyrene 1-aminopyrene 1-nitropyrene

Detection Limit (S/N ) 3)/M 3 × 10-7 phenanthrene 1 × 10-7 1-aminonaphathalene 1 × 10-7 1-chloronaphatharene 3 × 10-7 1-aminoantharene -8 8 × 10 flavin mononucleotide -8 9 × 10 Rhodamin 6G -8 4 × 10 Methylene blue -8 5 × 10 thionin

Relative Standard Deviation (%) Dns-Phe (5 × 10-7 M, n ) 15) pyrene (5 × 10-8 M, n ) 15) flavin mononucleotide (1 × 10-7 M, n ) 15) sample frequency/h

concn (M)

rel CL intensitya

10-7 10-6 10-5 10-4

100 98 90 52 10

compd DABCO

concn (M)

rel CL intensitya

10-7 10-6 10-5 10-4 10-3

99 94 71 42 9

a Batch method: 50 µL of 2 × 10-2 M KHSO solution added to 50 5 µL of 1 × 10-3 M CoSO4 solution containing NaN3 or DABCO.

8 × 10-8 6 × 10-8 8 × 10-8 1 × 10-7 2 × 10-8 1 × 10-8 4 × 10-9 6 × 10-9 3.1 3.7 2.2 ∼90

The oxidation potential of peroxomonosulfate (1.82 V)38 is slightly higher than that of Co3+ (ECo3+/Co2+ ) 1.802 V). The CL spectra obtained with cutoff filters showed that there was only a CL peak band (471-478 nm) for the HSO5-/Co2+ system. The CL observed in the HSO5-/Co2+ system may be due to formation of a singlet oxygen molecular pair, (1O2)2*. It is well known that the singlet oxygen, being of higher energy than the ground-state triplet oxygen, can lose its excess energy via luminescence.39 During the last 30 years, numberous CL investigations40 concerned with (1O2)2* have been reported. In our previous work,16,41,42 the CL wavelength near 480 nm was also recorded. All of them could be well explained by the formation of the (1O2)2* species. In this work, the formation of (1O2)2* may be take place as the follow reactions:

HSO5- + SO4•- f SO5•- + HSO4-43

(3)

HSO5- + OH• f SO5•- + H2O36a

(4)

2SO5•- + H2O f 2HSO4- + 3/21O2 44

(5)

1

Table 2. Effect of NaN3 and DABCO on the CL Signal

O2 + 1O2 f (1O2)2* f 2O2 + hν45

(6)

The catalytic decomposition of HSO5- can be summarized as

2HSO5- f 2H+ + 2SO42- + O2 + hν 36a,43

Figure 3. Effect of D2O and H2O on the CL signals. Mixing order: 2 × 10-2 M KHSO5 + 1 × 10-3 M CoSO4, The volume of each solution was 50 µL.

from HSO5- has been identified by optical pulse radiolysis.37 Evolution of O2 during the reaction was confirmed by addition of 20 mL of 0.2 M KHSO5 to 10 mL of 1.0 × 10-3 M CoSO4 solution. About 40 mL of gas was generated after the complete decomposition. It has been confirmed that this gas is oxygen. Two compounds, 1,4-diazabicyclo[2,2,2,2]octane (DABCO)46 and NaN3,47 known to be quenchers of 1O2 were also used in this experiment. As shown in Table 2, the CL intensity of HSO5- reacting with Co2+ was decreased in the presence of NaN3 or DABCO. The CL emission from mixing HSO5- solution with Co2+ solution decreased with the increase of NaN3 or DABCO concentration. Another experiment to confirm the generation of 1O2 was carried out by using deuterium oxide (D2O) instead of water. In D2O, the lifetime of 1O2 is ∼10-fold longer than that in water.48 The use of this phenomenon in detecting the 1O2 reaction has been reported in a number of investigations.49,50 As shown in Figure 3, the CL profiles of HSO5-/Co2+ reacting in H2O and D2O media are different. Both peak height and area of the CL signals from mixing HSO5- and Co2+ in D2O solutions are bigger than those results in water. This phenomenon evidenced the fact that 1O was produced during the decomposition of HSO -. 2 5

(7)

The formation of SO5•-, SO4•-, and OH• as the radical products (38) Steele, W. V.; Appelman, E. H. J. Chem. Thermodyn. 1982, 14, 337-344. (39) Khan, A. U.; Kasha, M. J. Chem. Phys. 1963, 39, 2105-2106. (40) Wasserman, H. H.; Murray, R. W. Signlet Oxygen; Academic Press: New York, 1979. (41) Wu, X. Z.; Yamada, M.; Hobo, T.; Suzuki, S. Anal. Chem. 1989, 61, 15051510. (42) Lin, J.-M.; Hobo, T. Talanta 1995, 42, 1619-1623.

(43) Zhang, Z.; Edwards, J. O. Inorg. Chem. 1992, 31, 3514-3517. (44) Mariano, M. H. Anal. Chem. 1968, 40, 1662-1667. (45) Stauff, J.; Schmidkunz, H.; Hartmann, G. Nature (London) 1963, 198, 281282. (46) Ouannes, C.; Wilson, T. J. Am. Chem. Soc. 1968, 90, 6527-6528. (47) Deneke, C. F.; Krinsky, N. I. Photochem. Photobiol. 1977, 25, 299. (48) Merkel, P. B.; Nilsson, R.; Kearns, D. R. J. Am. Chem. Soc. 1972, 94, 10301031. (49) Shellum, C. L.; Birks, J. W. In Chemiluminescence and Photochemical Reaction Detection in Chromatography; Birks, J. W., Ed.; VCH: New York, 1989; p 236. (50) Sawyer, D. T. Oxygen Chemistry; Oxford University Press: New York, 1991.

Analytical Chemistry, Vol. 72, No. 6, March 15, 2000

1151

The change of pH value of the KHSO5 solution with decomposition time was also measured. Without Co2+ ion, the decomposition of HSO5- is very slow. Therefore, there was almost no pH change during 1 h. With Co2+ as catalyst, the pH decreased quickly. The pH of a 20 mL of 0.01 M KHSO5 solution containing 5 × 10-4 M CoSO4 was .∼ 2.25, after 1 h of decomposition the pH became 1.85. This means that the H+ ions were generated during the decomposition reaction. These data supported reaction 7, although this reaction had has been suggested for a long time.36a,43 CL Mechanism of the HSO5-/Co2+/FOCs System. Although the CL of HSO5-/Co2+ can be explained by the above experiments, the CL from the FOCs induced by Co2+-catalyzed decomposition of HSO5- is more difficult to interpret. CL spectra of pyrene and dns-L-Phe were recorded with the continuing flow system using a Hitachi F-4010 fluorescence spectrometer. The spectra had two peaks (λmax ) 460 and 560 nm) for pyrene and another two peaks (λmax ) 465 and 516 nm) for dns-L-Phe, respectively. The fluorescent spectra for dns-L-Phe and pyrene have also been recorded. As shown in parts a and b of Figure 4, the maximum fluorescent wavelengths of dns-L-Phe and pyrene are 396 and 553 nm, respectively. These wavelengths differ from the CL wavelengths of dns-L-Phe and pyrene. It means that neither the dns-L-Phe nor pyrene is a CL emitter during the CL reaction. The CL wavelength at 460 nm may be assigned to the transition from (1O2)2*, which was produced from the chain reactions 1-6. Another CL wavelength cannot be ascribed to any component of the reaction system. Furthermore, the reaction of dns-L-Phe (or pyrene) with Co2+ and HSO5- mixed solution was also monitored by the fluorescent detection. As shown in Figure 4, the fluorescent intensity decreased with the mixing time of dns-L-Phe (or pyrene) with HSO5- and Co2+ solution. The fluorescence disappeared quickly after adding the CoSO4 and HSO5- solutions into the dnsL-Phe or pyrene solution. This phenomenon also appeared for other FOCs, e.g., fluorescent dyes, naphthalen, anthrancene, and their derivatives. The UV-visible spectra of dns-L-Phe and pyrene were also recorded (Figure 5). Similar to the fluorescence results, the absorptions were decreased or disappeared after the dns-LPhe (or pyrene) reacted with the HSO5-/Co2+ solution. From these results, it is believed that the chemical construct of FOC was destroyed during the CL reaction. We also noticed that all FOCs used were aromatic or polycyclic aromatic rings. Dansyl amino acids have bright emission in HSO5-/Co2+ solution; however, CL has not been observed from the amino acids although the reaction conditions were the same. These results indicate that the aromatic ring plays an important role in the CL reaction, because aromatic and polycyclic aromatic hydrocarbons are electron-rich molecules which can be attacked by the high oxidizing power intermediate chemical generators. The reaction products were nonfluorescent and the emission light was generated from the intermediates. In the present CL system, two reactive species are considered, one is Co3+ ion and the other is SO4•- radical ion. The Co3+ ion is a strong oxidant, it has an absorption band at maximum wavelength 610 nm.34 The effect of the mixing time of HSO5- and Co2+ on the dns-L-Phe CL emission is plotted in Figure 6 (curve 1). The CL intensity of dns-L-Phe exhibited a maximum at ∼3 min after mixing HSO5- solution with Co2+ solution and was observed even after 40 min. The catalytic decomposition of HSO5- is fast 1152

Analytical Chemistry, Vol. 72, No. 6, March 15, 2000

in neutral and weakly basic solutions, but much slower at lower pH solution.43 From Figure 6, it can be seen that production of reactive species leading to CL is considered to be slow under the unbuffered condition (pH ∼2.5). The time dependency of the Co3+ concentration was also measured in the absence of FOC and HSO5-. The Co3+ solution was prepared by an electrochemical method.34 The concentration of Co3+ was monitored by an UVvisible spectrophotometer at the absorption of 610 nm. As shown in Figure 6 (curve 2), the curve patterns of CL intensity and Co3+ absorption at 610 nm are almost the same. This phenomenon made us consider that the FOCs possibly reacted with Co3+ directly. To make clear the reactivity of Co3+ ion in the present CL system, the CL batch method was carried out and the results are listed in Table 3. The mixing of electrochemically generated Co3+ solution with dns-L-Phe (or pyrene) did not cause any CL. Therefore, it is evident that Co3+ ion is not a reactive species that can react with FOCs directly. When using Co3+ as catalyst, from Table 3 we find that the CL intensity of pyrene or dns-L-Phe is much stronger in HSO5-/Co3+ solution than in HSO5-/Co2+ solution. It may be due to the increased SO4•- or HSO5 radicals produced when HSO5- reacts with Co3+.

Co3+ + HSO5- f Co2+ + HSO5

(8)

This result supports the suggestion that HSO5 or SO4•- radical ion plays an important role in the CL process. On the basis of the above experiments, we believed that HSO5- was decomposed into HSO5 or SO4•-. The FOC molecule was attacked by the HSO5 or SO4•- radical ions and decomposed into small molecules. CL Molecular Recognition Sensor Based on the Proposed CL System and Molecular Imprinting. The MIP is often prepared from functional monomers, cross-linking monomer, and template molecule. The template molecule is also the analyte. Usually, the analytes selectively adsorbed in the cavities of the MIP were eluted by organic solvent and detected with an UV or fluorescence detector.31 Therefore, if the affinity is strong it needs a long time and a large volume of organic solvent to wash the polymer. In the present work, we used dns-L-Phe as template molecule, which can be selectively absorbed on the MIP prepared. When the MIP containing dns-L-Phe molecule contacts the HSO5-/Co2+ mixed solution, CL emission is given out and dnsL-Phe is decomposed into small molecules. These small molecules can be washed easily. Moreover, the nonfluorescence decomposition products have no CL emission, which also makes for higher recognition for the proposed sensor. On the basis of this considereation, a CL molecular recognition sensor was assembled as in Figure 1. The procedure of polymer preparation was the same as the report of Ramstrom.35 This polymer has been used as a fiber-optic fluorescence sensing device30a and applied in capillary electrochromatography.33 Therefore, we have not discussed the preparation method here. However, the characteristics of the polymer are worth pointing out. Using the mixture of MAA and 2-VPy as monomers, the polymer is more transparent than that using only MAA as monomer.30a It is only slightly softer than the MAA MIP, and destruction of the polymer was not observed during the CL reaction. The more important point is that the MAA/2-VPy MIP is of higher recognition ability than the MAA MIP.33,35 The transparency is beneficial to the CL observation.

Figure 4. Fluorescent spectra of dns-L-Phe solution (a) and pyrene solution (b) with and without the addition of KHSO5 and CoSO4 mixed solution. Conditions: (a) (1) 1 × 10-4 M dns-L-Phe solution; (2) 3.0 mL of 1 × 10-4 M dns-L-Phe solution + 30 µL of 5 × 10-3 M CoSO4 and 60 µL of 0.01 M KHSO5. The mixing time of the three reagents was 2 min; (3) (2) with 5-min mixing time; (4) (2) with 10-min mixing time; (5) (2) with 30-min mixing time; (6) 3.0 mL of 1 × 10-4 M dns-L-Phe solution + 30 µL of 5 × 10-3 M CoSO4 and 200 µL of 0.01 M KHSO5 after 1-h mixing time. (b) (1) 3.0 mL of 1 × 10-6 M pyrene solution. From spectra (2-4) the concentrations of KHSO5 and CoSO4 were the same as in (a) except the mixing times were 1, 3, and 5 min, respectively. Conditions for spectrum b-5 were the same as in spectrum a-6.

Owing to the higher recognition, transparent character, and simple polymerization, the MAA/2-VPy MIP is especially suitable for the initial study of the MIP CL. The setup of the polymer-packed CL flow sensor is shown in Figure 1. The 100-200-µm polymer particles were packed into a

glass tube. For a new sensor, to “extract” the print molecules from the polymer frame, the mixture of HSO5- and Co2+ solutions flowed through the sensor for ∼10 min with the flow status at Figure 1a. Then water was passed through the sensor to wash the residual HSO5- and Co2+ solutions and the decomposed Analytical Chemistry, Vol. 72, No. 6, March 15, 2000

1153

Table 3. Comparison of the Chemiluminescence of Pyrene and dns-L-Phe with Co2+ and Co3+ Ions as Catalysis by the Batch Method CL systema

relative CL intensity

+ pyrene (or dns-L-Phe) Co3+ + pyrene (or dns-L-Phe) Co2+ + HSO5- + pyrene Co2+ + HSO5- + dns-L-Phe Co3+ + HSO5- + pyrene Co3+ + HSO5- + dns-L-Phe

0 0 16 18 90 88

Co2+

a The concentrations of HSO -, Co2+, pyrene, and dns-L-Phe were 5 2 × 10-2, 1 × 10-3, 1 × 10-6, and 1 × 10-6 M, respectively. The volume for each reagent solution was 50 µL. The Co3+ solution was obtained by electrochemical generation from 0.2 M CoSO4 (in 2 M H2SO4) at 2 A current for 15 min.

Figure 5. Absorption spectra of dns-L-Phe solution (a) and pyrene solution (b) before and after reacting with KHSO5/CoSO4 solution. Reference solution, water. (a) (1) 1 × 10-4 M dns-L-Phe solution; (2) 3.0 mL of 1 × 10-4 M dns-L-Phe solution + 30 µL of 5 × 10-3 M CoSO4 and 60 µL of 0.01 M KHSO5. The mixing time of the three reagent solutions was 3 min. (b) (1) 1 × 10-4 M pyrene solution; (2) 3.0 mL of 1 × 10-6 M pyrene solution + 30 µL of 5 × 10-3 M CoSO4 and 60 µL of 0.01 M KHSO5. The mixing time of the three reagent solutions was 3 min. Spectra 3 of both (a) and (b) correspond to the absorption of 3.0 mL of water + 30 µL of 5 × 10-3 M CoSO4 and 60 µL of 0.01 M KHSO5.

Figure 6. Effect of time after mixing of HSO5- and Co2+ for the addition of dns-L-Phe (curve 1) and time course (curve 2) of the absorbance (610 nm) of Co3+ formed. The concentrations of KHSO5 and CoSO4 were 0.02 and 0.05 M, respectively. The volume of HSO5and Co2+ was 50 µL each.

products of dns-L-Phe. Being different from the usual molecular imprinting recognition systems, here is unnecessary to use the organic solvent to extract the print molecule. The template molecules react with HSO5-/Co2+ solution directly. Water was 1154 Analytical Chemistry, Vol. 72, No. 6, March 15, 2000

Figure 7. CL signals in the present flow sensor. The polymer was imprinted with dns-L-Phe. Peak 1: the CL background of KHSO5 and CoSO4 mixed solution. Peak 2: 1 × 10-6 M dns-L-Phe. Peak 3: 5 × 10-6 M dns-L-Phe. Peak 4: 5 × 10-6 M dns-D-Phe. The flow rates of 2 × 10-2 M KHSO5 solution, 1 × 10-3 M CoSO4 solution, and carrier (water) were 1.0, 1.0, and 1.5 mL/min, respectively.

used as the sample carrier and the polymer washing solvent. The proceeding of molecule recognition and detection with the proposed CL molecular recognition sensor could be summarized as four steps. Step 1, the polymer particles in cell were cleaned by water for ∼10 min (sensor washing). Step 2, a 100 µL of sample was injected into the cell with the water carrier. Step 3, to wash the mixed compounds, except the template, the water flow was continued through the polymer particles in the cell. This is an important step, and the flow rate and flow time need to be controlled. The flow time was controlled by an automatic switching valve. When the flow rate of water (carrier) was 1.5 mL/min, the suitable flow time for this step was 4 min. A too-long flow time caused the loss of template molecule caught by the MIP, but sufficient washing time was necessary for the elimination of components other than the target molecule. Step 4, the mixed solution of KHSO5 and CoSO4 the flowed through the polymer particle-packed cell (CL detection). The reaction of template molecule caught in the MIP with the mixed solution of KHSO5 and CoSO4 gave out a bright CL emission. The CL signals of dnsL-Phe and dns-D-Phe are shown in Figure 7. The plateau (peak 1) is a background CL emission from the KHSO5 and CoSO4 mixed solution. The plateau is stable and the width of the plateau

line was defined as noise. Peaks 2 and 3 correspond to 1 × 10-6 and 5 × 10-6 M dns-L-Phe solutions, respectively. Their peak heights were measured from the plateau line to the peak. The ratio of the peak height to the noise was defined as S/N here. When 5 × 10-6 M dns-D-Phe was added into the sensor, only a small peak (peak 4) as high as the CL background was appeared. It means that the dns-D-phe was not caught in the dns-L-Pheimprinted polymer. The detection limit of dns-L-Phe is 4 × 10-7 M, near the result of CL-FIA. We also found that the CL signals of the FIA-CL system and the CL flow sensor were similar. They are sharp and stable. Therefore, it can be though that the bound force between the MIP and dns-L-Phe is not strong enough to influence the CL reaction. The calibration graph of logistic CL intensity vs logistic dns-L-Phe concentration was linear in the 4 × 10-7-5 × 10-5 M range. CONCLUSIONS This preliminary work opens a new field of research on the combination of the molecular imprinting with the high-sensitivity

CL method. The imprinted cavities of a defined shape and functional groups in the molecularly imprinted polymer are expected to be developed not only with the molecule recognition function but also as a special CL reaction medium, for example, as a solid catalysis in the heterogeneous-phase CL reactions. In the present work, a main shortcoming of the CL reaction, the decomposition of some reactants during the CL reaction, was made to good use for the proposed molecular imprinting CL sensor. ACKNOWLEDGMENT This work was partially supported by a grant-in-aid for scientific research from the Ministry of Education, Science and Culture of Japan, which we gratefully acknowledge.

Received for review September 27, 1999. Accepted January 3, 2000. AC9911140

Analytical Chemistry, Vol. 72, No. 6, March 15, 2000

1155