A Postcolumn Reaction Detection System for Singlet Molecular

Department of Chemistry, Fort Lewis College, Durango, Colorado 81301. A postcolumn photochemical reaction detection scheme, based on the reaction of ...
0 downloads 0 Views 110KB Size
Anal. Chem. 1998, 70, 4081-4085

Photooxidation of 3-Substituted Pyrroles: A Postcolumn Reaction Detection System for Singlet Molecular Oxygen in HPLC K. Denham and R. E. Milofsky*

Department of Chemistry, Fort Lewis College, Durango, Colorado 81301

A postcolumn photochemical reaction detection scheme, based on the reaction of 3-substituted pyrroles with singlet molecular oxygen (1O2), has been developed. The method is selective and sensitive for the determination of a class of organic compounds called 1O2-sensitizers and is readily coupled to HPLC. Following separation by HPLC, analytes (1O2-sensitizers) are excited by a Hg pen-ray lamp. Analytes that are efficient 1O2-sensitizers promote groundstate O2 (3Σg-) to an excited state (1Σg+ or 1∆g), which reacts rapidly with tert-butyl-3,4,5-trimethylpyrrolecarboxylate (BTMPC) or N-benzyl-3-methoxypyrrole-2-tertcarboxylate (BMPC), which is added to the mobile phase. Detection is based on the loss of pyrrole (BTMPC or BMPC). The reaction is catalytic in nature since one analyte molecule may absorb light many times, producing large amounts of 1O2. Detection limits for several 1O2sensitizers were improved by 1-2 orders of magnitude over optimized UV-absorbance detection. This paper discusses the optimization of the reaction conditions for this photochemical reaction detection scheme and its application to the detection of PCBs, nitrogen heterocycles, nitro and chloro aromatics, and other substituted aromatic compounds. Interest in reactive oxygen intermediates has grown significantly over the past two decades due to the importance of these species in redox reactions of organic and inorganic species.1-5 For example, dissolved organic matter in natural waters leads to the photosensitized formation of singlet molecular oxygen (1O2), which assists in the photodegradation of organic pollutants such as phenols and humic matter.6,7 However, the formation of this reactive oxygen species from dyes from textile manufacturing, phenolic derivatives from oil refineries, petrochemical plants, and pesticide plants may effect biota of colored waters, destroying organisms.8 While the formation of 1O2 leads to cell death and (1) Haag, W. R.; Hoigne´, J.; Gassman, E.; Braun, A. M. Chemosphere 1984, 13, 631-640. (2) Canonica, S.; Jans, U.; Stemmler, K.; Hoigne´, J. Environ. Sci. Technol. 1995, 29, 1822-1831. (3) Scully, F. E.; Hoigne´, J. Chemosphere 1987, 16, 681-694. (4) Aguer, J. P.; Richard, C. J. Photochem. Photobiol. A 1996, 93, 193-198. (5) Redmond, R. W. J. Photochem. Photobiol. 1991, 54, 547-556. (6) Tratnyek, P. G.; Elovitz, M. S.; Colverson, P. Environ. Toxicol. Chem. 1994, 13, 27-33. (7) Tratnyek, P. G.; Hoigne´, J. J. Photochem. Photobiol. A 1994, 84, 153-160. (8) Baxter, R. M.; Carey, J. H. Freshwater Biol. 1982, 12, 285-292. S0003-2700(98)00416-8 CCC: $15.00 Published on Web 08/27/1998

© 1998 American Chemical Society

cancer, the controlled generation of 1O2 can have a therapeutic effect. For example, the ability of 1O2 to oxidize nucleic acids and cell membranes, deactivate enzymes, and kill bacteria (e.g., Escherichia coli) has been used in photodynamic therapies (PDT).9,10 Therefore, an important factor in the design of reagents for PDT is their efficiency for the photosensitization of 1O2. Despite the recent surge of interest in photosensitization processes involving reactive aqueous-phase oxygen intermediates, few methods have been reported for the selective and trace-level determination of compounds that sensitize 1O2.11-14 Furthermore, methods for measuring 1O2 photosensitization efficiencies are elaborate and/or expensive.12 A simple, sensitive, and selective procedure for the determination of 1O2-sensitizing organic pollutants at trace levels as well as for evaluation of the efficiency (i.e., quantum yields for 1O2 production) of new drugs used in PDT is needed. The ability of analyte molecules to generate 1O2 has led to the development of luminescence- and absorption-based detection schemes for HPLC.11-14 These approaches take advantage of molecular oxygen’s (3O2) ability to quench the excited triplet state of organic molecules.15-17 In solution, 1O2 is generated through several processes, including photosensitization.18 Briefly, a molecule or photosensitizer (S) absorbs light and transfers energy to dissolved molecular oxygen:18

S + hν f 1S*

absorption

1 *

S f 3S*

intersystem crossing (2)

S + O2(3Σg-) f S + O2(1∆g)

1

3 *

O2 production

(1)

(3)

Equation 3 represents triplet-triplet annihilation, which results in an energy transfer to yield 1O2. 1O2 can relax by several paths, (9) Rywkin, S.; Lenny, L.; Goldstein, J.; Geacintov, N. E.; Margolis-Nunno, H.; Horowitz, B. Photochem. Photobiol. 1992, 56, 463. (10) Nonell, S.; Braslavsky, S. E.; Schaffner, K. Photochem. Photobiol. 1990, 51, 551. (11) Niederla¨nder, H. A. G.; de Jong M. M.; Gooijer, C.; Velthorst, N. H. Anal. Chim. Acta 1994, 290, 201-214. (12) Niederla¨nder, H. A. G.; van Assema, W.; Engelaer, F. W.; Gooijer, C.; Velthorst, N. H. Anal. Chim. Acta 1991, 255, 395-401. (13) Niederla¨nder, H. A. G.; Nuijens, M. J.; Dozy, E. M.; Gooijer, C.; Velthorst, N. H. Anal. Chim. Acta 1994, 297, 349-368. (14) Shellum, C. L.; Birks, J. W. Anal. Chem. 1984, 59, 1834-1841.

Analytical Chemistry, Vol. 70, No. 19, October 1, 1998 4081

including solvent deactivation (eq 4) and quenching by organic and inorganic molecules (A in eq 5).

O2(1∆g) f O2(3Σg-)

solvent deactivation

(4)

A + O2(1∆g) f O2(3Σg-) + A

quenching by A

(5)

1O 2

can also act as a powerful dienophile, readily reacting with a variety of dienes (e.g., A in eq 6) in Diels-Alder-type cycloadditions to form chemically unstable endoperoxides (e.g., products in eq 6).

A + O2(1∆g) f product (p)

reaction with A

(6)

The reaction sequence (eqs 1-6) which leads to photooxidation (eq 6) has been termed type II photooxidation to signify that 1O is an intermediate. The kinetics of the type II photooxidation 2 process have been well characterized.19 Shellum and Birks took advantage of molecular oxygen’s ability to quench the triplet state of organic molecules by developing a detection scheme for HPLC based on 1O2 photooxidation reactions.14 Briefly, a 1O2 trap such as dimethylfuran (DMF) or diphenylfuran (DPF) is added to the mobile phase. Analytes that are poor chromophores but efficient 1O2-sensitizers (see reactions 1-6 above) promote ground-state 3O2(3Σg-) to an excited state (1Σg+ or 1∆g). O2(1∆g) undergoes a variety of photooxidation reactions, including rapid oxidation (photobleaching) of DMF and DPF. Analyte concentration is indirectly measured spectroscopically through a decrease in the trap (furan) concentration or an increase in the concentration of the oxidation product. Because the photooxidation process is catalytic (i.e., analytes can undergo reactions 1-3 several times), significant improvements in detection are observed. Despite the impressive sensitivity of the system, self-photooxidation of DMF and DPF required the use of optical solution filters to reduce absorption by DMF or DPF and thus minimize loss of the 1O2 trap. Unfortunately, the use of optical filters limits the range of analytes that can be detected to those that absorb light at wavelengths longer than DMF or DPF. Niederla¨nder et al. developed a luminescence-based detection scheme for HPLC based on molecular oxygen’s ability to quench the triplet state of organic molecules.11-13 In this system, 3,4diethoxy-1,2-dioxetane is produced from the reaction of 1O2 with 1,2-diethoxyethene. Thermal decomposition of the dioxetane in the presence of 9,10-dibromoanthracene-2-sulfonic acid (DBAS) leads to chemiluminescence.11-13 The chemiluminescence detection scheme has been used in HPLC for the determination of PCBs and substituted aromatic compounds as well as in FIA for the determination of 1O2 quenching efficiencies. Despite the selectivity of the method, the overall efficiency for the detection is only (15) Birks, J. B. Photophysics of Aromatic Molecules; John Wiley: New York, 1970; p 509. (16) Kawaoka, K.; Khan, A. V.; Kearn, D. R. J. Chem. Phys. 1967, 46, 1842. (17) Gijzeman, O. L. J.; Kaufman, F.; Porter, G. Faraday Trans. 2 1973, 69, 708. (18) Turro, N. J. Modern Molecular Photochemistry; University Science Books: Mill Valley, CA, 1991; pp 587-591. (19) Kearns, D. R. In Singlet Oxygen Organic Chemistry. A Series of Monographs; Wasserman, H. H., Murray, R. W., Eds.; Academic: New York, 1979; Vol. 40, p 120.

4082 Analytical Chemistry, Vol. 70, No. 19, October 1, 1998

on the order of 10-5.12 Moreover, the detection scheme requires careful regulation of the detector cell temperature for the thermal decomposition step, the use of reagents that are not commercially available, and the use of cutoff filters to prevent photodecomposition of the dioxetane.11-13 The discovery that 3-substituted pyrroles react rapidly and efficiently with 1O215,20,21 led us to investigate the potential for implementing these compounds in photochemical reaction detection schemes based on type II photooxidation processes (reactions 1-6). This paper describes the novel use of tert-butyl-3,4,5trimethylpyrrolecarboxylate (BTMPC) and N-benzyl-3-methoxyprrole-2-tert-carboxylate (BMPC) as 1O2 traps in a postcolumn photochemical reaction detection scheme for HPLC. Significant improvements in sensitivity over optimized native UV-absorbance detection were achieved for seven model compounds. The method takes advantage of the commercially available pyrrole BTMPC, which has a large rate constant for reaction with 1O2. Furthermore, self-photooxidation of BTMPC and BMPC is significantly lower than that of DPF or DMF, eliminating the need for optical filters in the photochemical reactor. EXPERIMENTAL SECTION Chemicals. tert-Butyl-3,4,5-trimethylpyrrolecarboxylate, quinoline, anthraquinone, 9-acetylanthracene, biphenyl, 9-nitroanthracene, and 2-chloroanthracene were obtained from Aldrich Chemical Co. (Milwaukee, WI). Rose bengal and anthracene were products of Eastman Kodak. The BMPC was a gift from Dr. Harry Wasserman. 9,10-Dibromoanthracene-2-sulfonate (DBAS) was prepared as described by Catalani et al. (ref 22). HPLC-grade methanol, acetonitrile (ACN), and water were obtained from Fisher Chemical (Denver, CO). All samples were made up in 100% ACN. HPLC. A Hewlett-Packard 1100 series HPLC consisting of a quaternary gradient pump and diode array detector, driven by an HP ChemStation, was used for all separations. The flow rate through the column (15 cm × 4.6 mm, 5 µm C-8, Supelco, Supelco Park, PA) was maintained at 1.0 mL/min. Injections were carried out manually using a Rheodyne 7125 injector equipped with a 20µL loop. Photochemical Reactor. The postcolumn photochemical reactor (Figure 1) consisted of a crocheted PTFE tube (10 m length, 0.5 mm i.d., Supelco) coiled around a quartz cylinder (12 cm length, 2 cm o.d.). Two 4.5-mW pen ray low-pressure Hg lamps (JeLight Co., Laguna Hills, CA), inserted in the quartz cylinder, were used to drive the photochemical reaction. The entire reactor was wrapped with aluminum foil to increase the photon flux. Air flowing over the PTFE tubing served to cool the reactor. RESULTS AND DISCUSSION Early investigations of detection schemes based on 1O2 sensitization made use of substituted furans and dioxetanes.11-14 (20) Wasserman, H. H.; Frechette, R.; Rotello, V. M. Tetrahedron Lett. 1991, 32, 7571-74. (21) Wayne, R. P. Principles and Applications of Photochemistry; Oxford University Press: New York, 1988; p 247. (22) Catalani, L. H.; Wilson, T.; Bechara, E. J. H. Photochem. Photobiol. 1987, 45, 273-281.

Figure 1. Schematic of the instrumental setup including the postcolumn reactor used in the detection scheme for singlet molecular oxygen. MP, mobile phase containing pyrrole; P, pump; SL, sample loop; C, C18 column; CR, crocheted reactor; D, absorbance detector; PRL, Hg pen-ray lamp; QC. quartz cylinder; PTFE, crocheted Teflon tubing. The photoreactor was wrapped in aluminum foil to increase photon flux, while air flowing over the PTFE tubing served to cool the reactor.

Figure 2. Absorbance spectrum of a solution containing 1 × 10-4 M BTMPC and 1 × 10-5 M 9-acetylanthracene in 75:25 ACN/H2O following irradiation with a Hg pen-ray lamp. Photolysis times are 30, 60, 90, 120, and 240 s in order of decreasing absorbance at 280 nm.

While these methods led to substantial improvements in detection over UV absorbance for analytes that were efficient 1O2 sensitizers, problems associated with self-photooxidation of furans14 and the poor efficiency of the dioxetane system11-13 limited the analytical utility of the these methods. Moreover, the system described by Niederla¨nder et al. requires a detector cell temperature of 70.0 °C, which can lead to the formation of gas bubbles in the HPLC system.11-13 In addition, the kinetics of the thermally induced dioxetane decomposition are slow (t1/2 ) 60 s), meaning that only a small percentage of the chemiluminescence emission is collected.11-13 Finally, the more efficient 1O2 trap 1,2-diethoxyethylene and the fluorophore DBAS used in the luminescence detection scheme are not commercially available. Recently, Wasserman et al. reported that modifying pyrroles with electron-releasing substituents at the 3 position leads to increased rates and improved yields of photooxidation.20 Rate constants for 1O2 oxidation of 3-substituted pyrroles are on the same order of magnitude as rate constants for the 1O2 oxidation of substituted furans such as DPF and DMF. However, quantum yields for intersystem crossing (ISC) for 3-substituted pyrroles are lower than quantum yields for ISC for substituted furans. This suggests that the degree of self-photooxidation for 3-substituted pyrroles may be less than that for substituted furans. The attractive photophysical properties of modified pyrroles and the fact that they are commercially available led us to study the feasibility of employing BTMPC as a 1O2 trap in a postcolumn reaction detection scheme for compounds that are efficient 1O2sensitizers. Figure 2 illustrates the change in the absorption spectrum of a 10-4 M solution of BTMPC containing 10-5 M 9-acetylanthracene (75:25 ACN/H2O) following irradiation in a cuvette with a Hg penray lamp. The formation of multiple products from the sensitized 1O oxidation of 3-substituted pyrroles has been reported.20 The 2 spectral changes in Figure 2 are the result of analyte (9acetylanthracene)-sensitized photooxidation of BTMPC. Decreases in absorbance at 280 and 252 nm correspond to loss of BTMPC, as found by Wasserman et al. for similar 3-substituted

pyrroles.20 In the absence of 9-acetylanthracene, the loss was significantly less. These results suggest that BTMPC is an efficient 1O2 trap which does not undergo self-photooxidation to the extent that DPF does. In a flowing system, injection of analytes into a mobile phase containing BTMPC, followed by photolysis in a crocheted reactor, results in photooxidation of the pyrrole and a corresponding decrease in absorbance at 280 and 252 nm. Therefore, detection based on a loss of BTMPC results in negative peaks. Since the molar absorptivity of BTMPC at 280 nm is greater than the molar absorptivity at 252 nm, detection of analytes should be carried out most efficiently by monitoring absorption changes at 280 nm. However, at higher analyte concentrations, native absorption by the analytes themselves can affect the magnitude of the signal. For the model compounds used in this work, following the change in absorbance at 252 nm provided the best sensitivity. Several parameters, including solvent composition, pyrrole concentration, photolysis time, and analyte concentration, affect the rate and yield of sensitized BTMPC photooxidation. Therefore, static studies (e.g., Figure 2) were carried out in order to determine the optimum reaction conditions for use in a flowing system. Pyrrole Concentration. Static studies were carried out in order to determine the effect of pyrrole concentration on the efficiency of the detection scheme. Solutions of pyrrole (10-310-6 M) and 9-acetylanthracene (10-5 M, acting as the analyte or 1O -sensitizer) in 100% acetonitrile were photolyzed for 120 s. 2 Following photolysis, an aliquot of the solution containing the pyrrole, oxidized product(s), and 9-acetylanthracene was injected into the HPLC (no column). Peak heights (loss of absorbance at 252 nm) were used to determine the effect of pyrrole concentration on the efficiency of the system. A concentration of 5 × 10-5 M pyrrole yielded the largest change in absorbance at 252 nm (best S/N). However, the S/N did not change by more than 10% over the range of pyrrole concentrations that were studied. Although the degree of self-photooxidation of BTMPC and BMPC is small compared to that of DMF and DPF, the 10% decrease in S/N at higher pyrrole concentrations is probably due to the fact that, at Analytical Chemistry, Vol. 70, No. 19, October 1, 1998

4083

higher concentrations of pyrrole, increased self-photooxidation leads to higher detector noise. These results are in agreement with those reported by Shellum and Birks14 for the photooxidation of substituted furans. Eluent Composition. The lifetime of 1O2 in solvents commonly used in reversed-phase HPLC is relatively short (τ ) 2, 7, and 30 µs for H2O, CH3OH, and CH3CN, respectively).19 In the detection scheme described in this work, BTMPC competes with radiationless deactivation of 1O2 by reacting with 1O2 to form an oxidized product (eq 6). However, it is difficult to predict the effects of solvent composition on the overall efficiency of reactions 1-6 because changing the solvent can increase and decrease the rates of the individual reactions leading to BTMPC photooxidation. To evaluate the effect of water on the overall efficiency of reactions 1-6, we compared the signals produced by several 1O2-sensitizers (rose bengal, DBAS, quinoline, anthraquinone, 9-acetylanthracene, biphenyl, and anthracene) in acetonitrile containing 0, 15, 25, 30, and 50% water using flow injection analysis (FIA). Each analyte was injected into a flowing stream containing 5 × 10-5 M BTMPC and irradiated for 2 min in a crocheted photoreactor. Changes in absorbance at 252 nm for each solvent composition were measured. The results obtained were similar to those of Shellum and Birks.14 The signal (decrease in absorption at 252 nm) did not differ by more than 2% when the aqueous content of the mobile phase was varied from 0 to 50%. Clearly, these data cannot be explained solely by the effects of 1O2 lifetimes alone, and this confirms the results found by Shellum and Birks14 for substituted furanssreaction 6 (and possibly others in eq 1-6) is enhanced in the presence of water. Photolysis Time. The effect of photolysis time on the efficiency of the detection scheme was studied for each analyte in a static system. Solutions of BTMPC and each analyte in 100% acetonitrile were photolyzed in a cuvette for 30, 60, 90, 120, 150, and 180 s. Following photolysis, an aliquot of the solution containing the BTMPC, oxidized product, and analyte was injected into the HPLC (no column). Peak heights (loss of absorbance at 252 nm) were used to evaluate the effect of photolysis time on the efficiency of the system. The optimum photolysis time ranged from 60 to 240 s, depending on the analyte. However, peak heights did not change by more than a factor of 3. In a flowing system, longer photolysis times required the use of longer photochemical reactors, leading to band broadening. Therefore, a crocheted reactor with a residence time of 120 s at a flow rate of 1.0 mL/min was used for the remaining studies. Chromatographic Results. Figure 3 illustrates the application of sensitized BTMPC photooxidation for the detection of seven model compounds. Detection was carried out by adding BTMPC (5 × 10-5 M) to the mobile phase (50:50 ACN/H2O). This mobile phase composition provided sufficient resolution of the analytes and the system peaks (see discussion below). Two low-pressure Hg lamps served as the photolysis source (see Figure 1). With a mobile phase flow rate of 1.0 mL/min, residence time in the 10 m photoreactor was 120 s. Equilibration of the column with BTMPC was complete in approximately 15 min. While adding BTMPC to the mobile phase did not lead to a decrease in the separation efficiency, slight differences in the solvent composition between the analyte solutions and the HPLC mobile phase resulted in two system peaks in the chromatograms. Small 4084 Analytical Chemistry, Vol. 70, No. 19, October 1, 1998

Figure 3. Chromatogram obtained by photochemical reaction detection using BTMPC (5 × 10-5 M). Mobile phase, 50:50 ACN/ H2O; flow rate, 1 mL/min, resulting in a photolysis time of 120 s; detection at 252 nm. Peaks: 1, rose bengal (210 ng); 2, BTMPC; 3, DBAS (67 ng); 4, quinoline (26 ng); 5, anthraquinone (42 ng); 6, 9-acetylanthracene (44 ng); 7, BTMPC; 8, biphenyl (31 ng); and 9, anthracene (36 ng).

Figure 4. Chromatogram obtained by photochemical reaction detection using BTMPC. Conditions are the same as in Figure 3. Peaks: 1, rose bengal (21 pg); 2 BTMPC; 3, DBAS (6.7 pg); 4, quinoline (9 pg); 5, system peak; 6, anthraquinone (4.2 pg); 7, BTMPC; and 8, anthracene (3.6 pg).

changes in solvent composition between the analyte solution and the mobile phase perturb the equilibrium of BTMPC between the column and the mobile phase, leading to a system peak near the void volume of the column. A second system peak, corresponding to the retention of BTMPC, is observed if the analyte solution contains a different concentration of BTMPC than that found in the mobile phase. Even careful preparation of analyte solutions (i.e., by dissolution in the mobile phase) led to system peaks. Despite this challenge, the detection scheme is extremely efficient for those analytes that have retention times different from those of the system peaks. The chromatogram in Figure 4 shows the impressive sensitivity of the system for those analytes that are efficient 1O2-sensitizers. Table 1 summarizes the detection limits for several model compounds using sensitized photooxidation of BTMPC and optimized UV-absorbance detection. With the exception of biphenyl and 9-acetylanthracene, the photochemical reaction detection scheme yielded better sensitivity than optimized UV absorbance. The similar retention times of BTMPC, 9-acetyl-

Table 1. Detection Limit Comparisons for BTMPC and BMPCa

a

analyte

optimized native UV absorbance (pg)

photochemical reaction detection (pg)

enhancement factor

rose bengal DBAS quinoline anthraquinone biphenyl 9-acetylanthracene anthracene 9-nitroanthracene* 2-chloroanthracene*

210 10 129 42 154 16.8 8.2 14.7 12.8

5.3 0.70 3.1 0.43 3080 2050 0.29 0.25 2.8

40× 14× 42× 98× none none 28× 59× 4.6×

Analytes marked with an asterisk are for BMPC.

for the determination of 2-chloroanthracene, sensitive determination of 9-nitroanthracene would not be possible using BTMPC since the retention time of 9-nitroanthracene is too close to the BTMPC system peak.

Figure 5. Chromatogram obtained by photochemical reaction detection using BMPC (5 × 10-5 M). Other conditions are the same as in Figure 3. Peaks: 1, 9-nitroanthracene (446 pg); 2, 2-chloroanthracene (425 pg).

anthracene, and biphenyl limit the sensitivity for these compounds. Challenges associated with interference from system peaks can be overcome by implementing 3-substituted pyrroles with retention times that differ from that of BTMPC. Our most recent efforts have focused on N-benzyl-3-methoxyprrole-2-tert-carboxylate (BMPC), an efficient 1O2 trap with a retention time that is slightly shorter than that of BTMPC. Figure 5 illustrates the impressive sensitivity of the photochemical reaction detection scheme employing BMPC as a 1O2 trap for the determination of 9-nitro- and 2-chloroanthracene. While BTMPC could have been employed

CONCLUSION The use of a postcolumn photochemical reaction detection scheme based on the ability of substituted pyrroles to trap 1O2 (produced from the photolysis of analyte molecules) has been demonstrated to be a simple, sensitive, selective, and efficient means of detection in HPLC. The method reported in this work overcomes problems associated with existing detection schemes based on 1O2 reactions, including self-photooxidation,14 reagent availability,11-13 and poor overall efficiencies, 11-13 while improving detection limits over those of optimized native UV absorption by 1-2 orders of magnitude. We are currently investigating the application of HPLC with postcolumn reaction detection for the determination of PCBs using BTMPC, BMPC, and several other 3-substituted pyrroles. ACKNOWLEDGMENT The authors thank Dr. John Birks and Dr. Curt Shellum for stimulating discussions and encouraging this project. Support for this work was provided by a Cottrell College Science Award of Research Corporation, the Howard Hughes Medical Institute, National Science Foundation ILI Program (Grant No. 9551282), and the Petroleum Research Fund (ACS PRF No. 32776-B4), administered by the ACS. Received for review April 16, 1998. Accepted July 15, 1998. AC980416J

Analytical Chemistry, Vol. 70, No. 19, October 1, 1998

4085