BIOMEDICAL CHROMATOGRAPHY, VOL. 5.47-52 (1991)

Evaluation of Peroxyoxalate Chemiluminescence Postcolumn Detection of Fluphenazine in Urine and Blood Plasma Using High Performance Liquid Chromatography B. Mann and M. L. Grayeski" Department of Chemistry, Seton Hall University, South Orange, NJ 07079, USA

Peroxyoxalate chemiluminescence may be used for sensitive postcolumn detection of phenothiazine analytes separated by high performance liquid chromatography with appropriate optimization of measurement conditions such as solvent, pH and oxalate ester. Detectability of fluorescent analytes by chemical excitation varies greatly, but analytes with low oxidation potentials are generally more readily detected at low levels, as demonstrated for phenothiazines, an important class of fluorescent drugs. Some improvement in detection limits is observed for fluphenazine when chemiluminescence detection is compared to conventional fluorescence detection. Because of the specificity of chemical excitation, fewer interferences from fluorescent impurities in a urine matrix are observed.

INTRODUCTION Peroxyoxalate chemiluminescence has been shown to be a sensitive postcolumn reaction for the detection of many fluorescent compounds (Honda et al., 1983; Kobayashi et al., 1981; Toyoka and Imai, 1983; Watanabe and lmai, 1984; Sigvardson and Birks, 1984; Mann and Grayeski, 1987). With a postcolumn chemiluminescence reaction, the sensitivity and precision of the measurement depends on the nature of the chemiluminescence signal generated when the reagents are mixed. In general, when chemiluminescent reagents are combined, a transient pulse is observed. with a shape that may be characterized by the rate of chemiluminescence decay (reaction rate evaluated with halflife) and the total amount of emitted light (efficiency evaluated with total integrated area) (Orlavic et ul., 1989). Chemiluminescence yield affects sensitivity while reaction kinetics will alter both sensitivity and precision. The chemiluminescent reaction rate and efficiency vary with experimental conditions including the supporting solvent, pH and the fluorescent analyte. In the flow system, the chemiluminescence output is particularly sensitive to experimental conditions since only a portion of the intensity/time curve is being detected. Generally, efficient chemiluminescence and fast reaction rates produce maximum light intensity and low detection limits. For this reason optimization of the detection system with respect to solvent, pH and fluorophore is critical. A few studies on solvent effects (Weinberger, 1984) and pH effects (DeJong et uf., 1986; Honda et al., 1985a) have been reported. Weinberger (1984) investi* Author to whom correspondencc should be addresscd 0269-3879/91/010047-06 $05.00

01991 by John Wiley & Sons, Ltd.

gated the rate and efficiency of peroxyoxalate chemiluminescence in methanol, isopropanol and acetonitrile and concluded that the solvents act sterically to produce the observed effects. Effects of pH have been investigated with bis(2,4,6,trichlorophenyl) oxalate (TCPO) (DeJong et al., 1986; Honda et al., 1985a) and bis(2,Cdinitrophenyl) oxalate (DNPO) (Honda et al., 198Sa). Most analytical systems reported in the literature use the solvents and esters in neutral to basic pH. Not all fluorophores can be sensitively detected by peroxyoxalate chemiluminescence. In fact, some are better detected by conventional photoexcitation (Sigvardson and Birks, 1983). Since different fluorophores produce different intensities of chemiluminescence, successful application of the analytical technique depends on identification of fluorophores which may be detected at low levels of concentration. Among Polycyclic Aromatic Hydrocarbons (PAH), the species with the lowest oxidation potentials are most sensitively detected (Lechtken and Turro, 1974; Sigvardson and Birks, 1984). The reason for the specificity of the reaction to some easily oxidized fluorophores may be explained by the proposed reaction mechanism which includes an electron transfer step. The energy transfer step in recent mechanistic proposals is of the chemically initiated electron exchange luminescence (CIEEL) type, i.e., an excited state is produced by electron transfer from a radical anion/ cation pair. One proposed reaction mechanism is shown in Scheme 1 (Rauhut, 1969; McCapra, 1970). Hydrogen peroxide or peroxide anion react with the oxalic ester to form a peroxy intermediate (1). The intermediate 1 forms a high energy intermediate, purportedly dioxetanedione (2). In the CIEEL mechanism, an encounter complex (3) forms between dioxetanedione and the fluorophore. An electron is then extracted from the fluorophore and simultaneously carReceived 22 December 1989 Accepted 2 July 1990

B. MANN AND M. L. GRAYESKI

48

OOH

00

11 \I

I1 I

k,

H2O2+ ArOCCOAr + ArOCCOAr I OOH

00

OOH

I1 I ArOCCOAr

I

I1 It

4 ArOCCOOH + ArOH (11

OOH

chemicals were reagent grade. TCPO and DNPO were prepared according to the literature and recrystallized from ethyl acetate (Mohan and Turro, 1974). Bis(2,4,5-trichloro-6carbopentoxyphenyl) oxalate (CPPO) was obtained from A. Mohan. Fluphenazine HCl was supplied by L. J. Cline Love. Solutions of TCPO, DNPO and CPPO were prepared daily. Hydrogen peroxide solutions were prepared weekly and stored at 5 "C.

EXPERIMENTAL

Batch chemiluminescence experiments. (1) Solvent study. A TD-20E luminometer (Turner Designs) with stripchart readout on a Fisher series 5000 recorder was used. The sensitivity was decreased by turning the sensitivity knob to obtain onscale readings. Polypropylene test tubes (1.6 mL, 8 X SO rnm) were used to test solvents. The peroxyoxalate ester solution was a 4.0X 1 0 - 3 soh~ tion of TCPO in ethyl acetate. The fluorophore was 1.OX lo-' M perylene in acetonitrile. Hydrogen peroxide (90%) was diluted (3.4 g into 100 mL of the solvent under investigation), except in cases where hydrogen peroxide was dissolved in a cosolvent in order to obtain miscible solutions. A solution of hydrogen peroxide in acetone was used as a cosolvent in the methylene chloride solvent experiment. A mixture of hydrogen peroxide in dimethyl phthalate was used as a cosolvent in the study of toluene solvent. Hydrogen peroxide dissolved in acetonitrile resulted in a miscible solution with dimethyl sulfoxide. For each experiment, a 750 yL aliquot of the solvent under investigation and a 50 pL aliquot of hydrogen peroxide soluM) were mixed. With the tion (final concentration 1.2 x tube in the luminometer, 50pL of TCPO solution (final concentration 2.3 X lo-' M) was injected and the light emission recorded. When dibutyl phthalate was used as a solvent, 50 FL of TCPO solution was injected and the test-tube shaken for about two seconds by hand. (2) pH study. The apparatus described in the solvent study was used with modification of reagent concentrations. A 0.02 M phosphate buffer was made using monobasic sodium phosphate and adjusting the pH with dilute phosphoric acid or sodium hydroxide to the listed pH. Each ester was evaluated separately from pH 2.5 to pH 7.2. A 50 yL portion of 45: 55 acetonitrile: buffer, 10 yL 2.3 X M perylene in acetonitrile and 100 pL of 1.2 M hydrogen peroxide (made from 30% hydrogen peroxide) in. methanol were mixed. Then, 100 yL, 5.0 x 1 0 - ' ~ ,of each ester in ethyl acetate was injected and the emission recorded. The final concentration of reagents were perylene: 8.8X lo-' M , hydrogen peroxide: 0.46 M and ester: 1.9 X lo-' M. ( 3 ) Fluorophore study. The apparatus described in the solvent study was used with modification of reagent concentrations. A 100mL portion of 80% acetonitrile, 20% 4 m ~ phosphate buffer, pH 8.5, was mixed with 100 mL 1 M hydrogen peroxide in acetonitrile and 10mL of a fluorophore at 1.OX ~ O - ' M in acetonitrile. A 10OpL injection of 4 m ~ TCPO in ethyl acetate was made and the emission followed for two minutes. Almost no light was emitted after two minutes. The total integrated chemiluminescence obtained using each fluorophore was divided by the fluorescent quantum yield to give a measure of the chemilurninescence emission efficiency (Seitz, 1981).

Chemicals. The HPLC grade solvents acetonitrile, ethyl acetate, methanol and tetrahydrofuran were used. Aldrich gold label dimethylformamide was used. All other solvents and

Flow experiments. A Gilson Model 302 pump with a Mark I11 pulse damper (Kratos) supplies 1 .0 mLlminute flow through a Rheodyne loop injector (20 pL), a Zorbax 10 mm C-18 25 cm

0

[CO;'FL']FL*+ FL

0

FL* + C02

+ hv

bon dioxide is lost, leaving a radical cationtanion pair. Annihilation of the radical catiodanion pair leads to an excited fluorophore with subsequent fluorescent emission. This paper addresses two questions. First, the effects of solvent and pH are studied to determine how this information can be used to optimize the chemiluminescent measurement. Then the relationship of fluorophore to the intensity of the chemiluminescence is investigated, to evaluate whether or not characteristics of the fluorophore can be used to predict which fluorophores would be sensitively detected by peroxyoxalate chemiluminescence. A Flow Injection Analysis (FTA) system was used to compare detection of fluorescent analytes. Analysis of the data from the above studies may result in identification of useful analytical systems and applications. In this work, the relative chemiluminescent yield and rate were studied by measuring integrated area and reaction times in batch experiments, where the necessary reagents are added to a cell and the chemiluminescence studied as a function of time. The chemiluminescence yield is proportional to the total integrated area of the emission curve obtained, and the relative rates may be compared by measuring the halflife of the first order decay of the signal.

PEROXYOXALAI'E CHEMILUMTNESCENCE POSTCOLUMN DETECTION FOR HPLC

column and a Schoeffel FS 970 fluorometer. A mobile phase of 1 : 1: 0.25 methanol : acetonitrile: 1 % ammonium carbonate adjusted to pH8.0 was used. The fluorometer is operated with the lamp off, no emission filter, a PMT voltage of 1000 V and a four-second time constant. A Kratos URS 051 dual reagent postcolumn mixing device was used to pump 3 . 6 ~ lo-' M TCPO in ethyl acetate and 0.8 M hydrogen peroxide in acetonitrile solution into an SSI mixing tee. A second SSI mixing tee fitting allowed the postcolumn reagents to mix with the analyte flow stream. Simplex optimization was used with variation of each reagent pump until best response was obtained: 0.8 mL/min for TCPO solution and 1.3 mL/min for hydrogen peroxide. The postcolumn pumps and mixing tees were disconnected for fluorescencemeasurements. An excitation wavelength of 270 nm and a 389 nm emission filter were used. To determine recovery, linearity and the limit of detection from plasma and urine samples doped with fluphenazinc,one drop of 10% perchloric acid was added to 1mL of sample, mixed and centrifuged. The clear layer was injected.

RESULTS AND DISCUSSION Comparison of solvents Peroxyoxalate chemiluminescence was investigated using several supporting solvents so that solvents useful for analytical applications could be identified. The chemiluminescence obtained with each solvent is evaluated in terms of the total integrated area (measure of efficiency) and the reaction half-life. Solvents suitable for analysis would afford high chemiluminescence efficiency. Physical properties, such as viscosity, must also be considered in terms of reproducible mixing in a flow system. The chemiluminescence from batch experiments with identical amounts of reagents dissolved in different solvents was investigated. A wide range of chemical functional groups (hydroxy, cyano, ester, etc.) and physical properties such as viscosity are represented in the solvents used (Table 1). First, several practical criteria are necessary for any meaningful results to be obtained: (1) the solvent should not be fluorescent, (2) the solvent should not quench the fluorescence, (3) the solvent should not produce a phase separation Table 1. Effect of solvents on TCPO peroxyoxlate cbemiluminescence Solvent

Acetone Aceton itrile Diobutyl phthalate (DBP) N,N-Dimethylforrnamide (DMF) Dimethyl phthalate (DMP) Dimethyl sulfoxide (DMSO) Ethyl acetate Methanol

Methylene chloride Phenylacetonitrile Tetrahydrofuran (THF) To1ue ne

Area (relative units)

Half-life (s)

(Cp at 20 "C)

620 1630 4500 21 5100 0 3500 890 70 1 97 10

5 113 1200 13 240 72 23 4 10 120 78

0.316 at 25 "C 0.375 9.72 t 37.8 "C 0.802 at 25 "C 17.2 at 25°C 1.98 0.455 0.597 0.449 at 15 "C 1.93at 25°C 0.55 0.590

Viscosity

49

Table 2. Effect of pH on peroxyoxalate chemiluminescence using TCPO, DNPO and CPPO TCPO Half-life

pH

Peak height (cm x 10001

7.22 6.2 5.41 4.28 2.89 2.54

9400 11 100 10 800 2500 430 220

1.3 1.8 1.5 9.6 93 96

(s)

DNPO Peak height Half-life (cm ~ 1 0 0 0 ) (s)

3900 5700 5200 5300 4500 4500

0.6 0.8 0.9 0.9 1.0 1.1

CPPO Peak height Half-life (cm x 1000) (s)

3700 4500 2000 460 160 145

4 4 7 52 230 240

or precipitation of the reagcnts and (4) the solvent must not react with the reagcnts. All solvents used satisfied the above criteria except for DBP which had to be shaken to effect rapid mixing and avoid phase separation. Use of DMSO, phenylacetonitrile, DMF and toluene solvents resulted in minimal signal under these experimental conditions, and use of methylene chloride and THF resulted in moderate emission. The highest total integrated areas are observed with the solvents DMP, DBP, ethyl acetate, acetonitrile, methanol and acetone. The latter solvents might be suitable in chemiluminescence batch experiments. The high viscosity of DBP and DMP could result in inefficient mixing in a flow system. Even though moderate emission is observed using acetone, the relatively short half-life can cause poor precision.

PH

In using peroxyoxalate chemiluminescence as a postcolumn detector, the introduction of buffers is necessary because the reaction is base catalysed. The rate and yield of the reactions are very pH sensitive and an analytical system may be affected greatly. Very fast rates may result in highly variable measurements of the peak area or peak height, since precision is highly dependent on reproducible mixing. Three basecatalysed peroxyoxalate reagents, TCPO, DNPO and CPPO are investigated. The optimum pH for each oxalate ester is determined. From batch experiments, the pH behaviour of reactions with TCPO, DNPO and CPPO is first examined by evaluating the half-life of the reaction and the relative peak height (Table 2). The half-life of the reaction using CPPO is about three times that using TCPO over the entire range of pH 2.5-7.2. The halflife of the reaction using DNPO is shortest of the oxalate esters examined, but does not show much change in half-life in the pH range 2.5-7.2. As expected for base-catalysed reactions, an increase in peak height is observed for increasing pH with the maximum observed around pH 5.4-6.2 for all esters. Conflicting reports about the pH at which maximum emission occurs appear in the literature. DeJong et al. (1986) report the maximum light yield at around p H 4 for TCPO. Other studies suggest the highest emission for TCPO reactions occur at pH 7.5-8 (Weinberger, 1984; Honda et al., 1985a). All the above mentioned studies utilize a flow system. The data is indicative of the fact that the highest emission obtained in a flow system measurement depends on the point along the

50

B . MANN AND M. L. GRAYESKI

intensity/time curve that is measured. In a flow system, the highest chemiluminescence intensity may not occur as the reagents pass through the detector. A change in the reaction rate makes reoptimization of the solvent flow rates, length of connecting tubes, etc. necessary. It should also be noted that the solvent system will also affect the pH optimum (Givens et al., 1988), so the pH optimum is better expressed as a range rather than a single value. All esters would be suitable at pH 5-7, but sufficient emission is observed for DNPO at lower pH to allow adequate sensitivity. Analytical implications

In determination of fluorophores by peroxyoxalate chemiluminescence with a flow system, a detector will typically consist of three flow streams: analyte, oxalate ester and hydrogen peroxide, mixed prior to a light sensor. The composition and flow rate of each stream may be varied independently. Solvents can be selected to allow the highest yield and appropriate kinetics. Three factors are important in considering solvent choice: (1) high efficiency is desirable, (2) a relatively fast rate is desirable so that most of the light can be measured in a short-time frame and ( 3 ) reproducible mixing should occur. To achieve all three is difficult because there is often some tradeoff in the effect of solvent parameters. Using solvents such as water or methanol increases the rate but results in low reaction efficiency. Some viscous solvents increase the efficiency but result in a slow rate. The solvent system must also maintain solubility of reagents and buffers and miscibility of solvents. Solvent studies indicate that the pH "optimum" for analytical purposes depends on a number of factors. In a flow detector system, a pH change can drastically reduce the measured signal. Maximum peak signals are obtained with CPPO and TCPO at pH 6-7. DNPO may be used over the entire pH range 2.5-7. Once the solvents and pH are selected to achieve maximum light emission, the reagent concentrations are adjusted to achieve maximum response. This may be accomplished by using different reagent concentrations or adjusting the flow rate of each reagent in the flow cell. When using a peroxyoxalate postcolumn detection system for HPLC, the mobile phase composition and the flow rate are important factors in determining the postcolumn flow rates. In general, a variety of systems may be chosen to satisfy the factors noted above for reproducible maximum signal, but improvements in detectability are reported to be much more dependent on the nature of the fluorophore than on small changes in the solvent system (Lechtken and Turro, 1974).

fluorophore and the chemiluminescence efficiency is investigated to determine if these properties can be used to predict which fluorophores would be more sensitively detected by chemical excitation. Based on the proposed mechanism, oxidation potentials would be an important parameter in evaluating suitability of fluorophores for chemical excitation (Sigvardson and Birks, 1984; Honda eb al., 1985b). The total efficiency of chemiluminescence consists of three components. First, there is the efficiency of the active intermediate formation. The efficiency of energy transfer is a second factor in the total efficiency. The product of the efficiency of intermediate formation and energy transfer is equal to the excitation efficiency. The fluorescence efficiency is the third component to the total efficiency. The chemiluminescence efficiency is the product of the excitation and fluorescence efficiency. Since only a fraction of the light is collected and detected by the luminometer, only relative excitation and chemiluminescence efficiencies are reported. An evaluation of relative excitation efficiency was made for a variety of fluorescent compounds with oxidation potentials ranging from about 0.4 to 1.5V (vs. SCE). A class of fluorescent drugs, phenothiazines, was identified as having relatively low oxidation potentials and several of these compounds were included in the evaluation. A plot of oxidation potential of fluorophore (Meites and Zuman, 1974; Kabasakalian and McGlatten, 1959) versus the relative excitation efficiency is shown in Fig. 1 . The fluorophores with the lowest oxidation potentials (0.4-0.7 V vs. SCE) show the highest excitation 2.0

X

s" 0

s

-1.0

P

-rn 0

'2.

-2.0 '0.

-3.0

Effect of fluorophore on peroxyoxalate chemiluminescence

In many analytical applications of the reaction, the fluorophore is the analyte. Since the chemiluminescent intensity obtained using different fluorophores is varied, the analytical performance of a detection system is highly dependent on the fluorophore. For this reason, the relationship between specific properties of the

0.0

0.4

0.8

1.2

OXIDATION POTENTIAL(vs

1.6

SCE)

Figure 1. Plot of the relative cherniluminescence efficiency versus the oxidation potential of fluorophores (see Experimental Section for details). 1 : Arninopyrene, 2 : fluphenazine, 3: perphenazine, 4: acetophenazine, 5: atabrine, 6: perylene, 7: 9-rnethylanthracene, 8: bisphenylethynylanthracene (BPEA), 9: 9,10-diphenylanthracene, 10: coronene, 11: anthracene, 12: 9-arninoacridine, 13: pyrene.

51

PEROXYOXALATE CHEMILUMINESCENCE POSTCOLUMN DETECTION FOR HPLC

Table 3. Detection limits of fluorophores by FIA

Fluorophore

Aminopyrene Fluphenazine Perphenazine Acetophenazine Atabrine Perylene 9,lO-Diphenylanth racene Coronene 9-Aminoacridine Pyrene

Detection limit (S/N =3) Fluorescence C hemiluminescence (fmol) (fmol)

21 0 23 000 56 000 13 000 a4

440 710 4000 16 3800

24 11 000 20 000 2700 85 2600 1100 13 000 29 13 000

efficiencies. It should be noted that these relative efficiencies are a measure of chemical excitation only. Fluorescent efficiencies and rate effects have been factored out. Aminopyrene, fluphenazine, perphenazine and acetophenazine show the highest excitation efficiencies of the compounds tested. Efficiency of chemical excitation decreases with increasing oxidation potential of the fluorophore. Pyrene, with an oxidation potential of 1.36 V vs. SCE, is close to the upper limit of oxidation potential which may produce measurable emission. The excitation efficiency is about lo4 times less efficient than aminopyrene. Some of the fluorophores plotted in Fig. 1 were analysed in a flow system to determine whether the oxidation potential is related to analytical detection (Table 3 ) . The FIA experimental results had a relative precision of less than 5% for each fluorophore tested. Only the fluorophores with low oxidation potentials result in lower levels of detection by chemiluminescence than fluorescence. It should be noted that these detection limits are for relative comparison only. Lower absolute detection limits can be observed with an appropriately optimized detection system. This suggests that the oxidation potential may be used to predict potential applications for peroxyoxalate cheiniluminescence detection. Three steps are involved in light production: (1) reaction producing a key intermediate, (2) excitation of the fluorophore via energy transfer from the intermediate and ( 3 ) fluorescent emission from the fluorophore. The solvent and pH effects discussed in the previous section are generally responsible for the efficiency in producing the key intermediate (step l ) , and the nature of the fluorophore generally determines the efficiency of step 2. Both the nature of the fluorophore as well as solvent and pH determine the efficiency of step 3. The lowest detection limits are only obtained when the efficiency of all three steps are maximised.

were identified as a class of compounds which should show efficient energy transfer in peroxyoxalate chemiluminescence. Phenothiazines like fluphenazine are used to treat psychotic disorders (Tjaden e: al., 1976). The drug is usually given orally several times daily. Fluphenazine is often prescribed to outpatients, but half the patients fail to take the prescribed amount of drug after a few months and relapse occurs. For this reason, sensitive blood assays were developed. Typically blood levels of low ppb quantities of fluphenazine are found in medicated outpatients. Thus, a sensitive and specific analytical method is required. Fluphenazine has been determined at low levels by radioimmunologic assay (Hawes et al., 19841, HPLC with coulometric detection (Kabasakalian and McGlatten, 1959) and spectrophotometry (Mahrous and Abdel-Khalek, 1984). Since a published extraction procedure (Tjaden et al., 1976) failed to result in more than 24% recovery, a modified procedure was developed. Precipitation of the protein with perchloric acid allowed determination of the fluphenazine in both urine (Fig. 2) and plasma (Fig. 3). The reversed phase HPLC separation used for fluphenazine requires appreciable amounts of water in the mobile phase (pH 8). Optimization of the reagent flows of TCPO in ethyl acetate and hydrogen peroxide in acetonitrile resulted in an aqueous concentration of about 11% at the point of mixing. Dimethyl phthalate

2

I m

6

-

0

2

4

6

6

2

4

6

Application

One approach for using peroxyoxalate chemiluminescence to obtain low detection limits is to analyse native fluorophores which have low oxidation potentials. Since low oxidation potentials are associated with the most efficient chemical excitation, lower detection limits should be found with chemiluminescent detection than with fluorescent detection. The phenothiazines

- Q

Figure 2. (a) Fluorescence detection of fluphenazine in urine (see Experimental section) and (b) chemiluminescence detection of fluphenazine in urine (see Experimental Section). Chromatograms are (1) fluphenazine standard (100 ppb), (2) blank urine and (3) spiked urine (100 ppb). Fluorescence and chemiluminescence responses measured in relative intensity units.

52

B. MANN AND M. L. GRAYESKI

a.

~

. . 1

0

2

4

6

b. 1

2

0

2

4

6

Figure 3. (a) Fluorescence detection of fluphenazine in blood plasma (see Experimental Section) and (b) cherniluminescence detection of fluphenazine in blood plasma (see Experimental Section). Chromatograms are (1)blank plasma and (2)spiked plasma (100ppb). Fluorescence and chemiluminescence responses measured in relative intensity units.

could not be used instead of ethyl acetate without precipitation of carbonate buffer. Linearity of the chemiluminescence detection system was determined to be at least three orders of magnitude, which is important in this determination since a wide concentration range is expected. The relative standard deviation was less than 10% for 100ppb of fluphenazine in both urine and plasma. It should be noted that for certain fluorophores, absolute detection limits by chemiluminescence detection can be at the attomole level but the absolute limit of detection depends on both the chemiluminescence efficiency and the fluorescent quantum yield. In this application, even though chemiluminescent excitation is efficient in fluphenazine, the low fluorecent quantum yield makes the absolute detection limit by chemiluminescence relatimole), but still better vely high (S/N = 3 , 165 x than detection limits by photoexcitation (S/N = 3, 225 x lo-'* mole). Because of the selectivity of the chemical excitation, an additional advantage is observed with the urine matrix. The specificity of the chemiluminescence versus the fluorescence measurements may be seen in the chromatograms of Fig. 2. The large peak in the chromatogram of fluphenazine in urine using fluorescence detection is due to fluorescent compound(s) in the urine which is(are) not observed with chemiluminescence detection.

The authors wish to acknowledge support from a grant by the Rcsearch Corporation and to thank R. Weinberger and ABUKratos Analytical Instruments for support and equipment.

REFERENCES DeJong, G. J., Lammers, N., Spruit, F. J., Frei, R. W. and Brinkman, V. A. (1986). J. Chromatogr. 353,249. Givens, R. S., Schowen, R. L., Kuwana, T. and Hanaoka, N. (1988).Anal. Chem. 60,2193. Hawes, E . M.. Shetly, H. U.. Conger, J. K., Rauw, C., McKay, C. and Midha, K. K. (1984). J. Pharm. Sci. 73(2),247. Honda, K., Sekino, J. and Imai, K. (1983). Anal. Chem. 55,940. Honda, K., Miyaguchi, K. and Imai, K. (1985a). Anal. Chim. Acta

177,103. Honda, K., Miyaguchi, K. and Imai, K. (1985b). Anal. Chim. Acta 177, 1 1 1 . Kabasakalian, P. and McGlatten, J. (1959).Anal. Chem. 31(3),

431. Kobayashi, S., Sekino, J. and Imai,

K. (1981).Anal. Biochem.

112,99. Lechtken , P. and Turro, N. J. (1974).Mol. Photochem. 6(1),95. Mahrous, M. S . and Abdel-Khalek, M. M. (1984). Talanta 31(4),

289. Mann, 6. and Grayeski, M. L. (1987).J. Chromafogr. 386, 149.

McCapra, F. (1970).Prog. Org. Chem. 8, 231. Meites, L. and Zuman, P. (1974).Handbook o f Organic Electrochemistry, Vol. I. CRC Press. Mohan, A. G. and Turro, H. J. (1974). J. Chem. Ed. 51, 528. Orlovic, M., Schowen, R. L., Givens, R. S., Alvarez, F.. Matuszewski, 6.and Parekh, N. (1989).J. Org. Chem. 54,

3606. Rauhut, M. M. (1969). Acc. Chem. Res. 2, 80. Seitz, R. (1981).CRC Critical Reviews in Analytical Chemistry, December 1981. Sigvardson, K. and Birks, J. W. (1984). Anal. Chem. 56, 1096. Sigvardson, K. and Birks, J. W. (1983). Anal. Chem. 55,432. Tjaden, U. R., Lankplan, J., Poppe, H. and Muusze, R. G. (1976). J. Chromatogr. 125,275. Toyoka. T. and Imai, K. (1983). J. Chromatogr. 282,495. Watanabe, Y. and Imai, K. (1984). J. Chromatogr. 309,279. Weinberger, R. (1984). J. Chromatogr. 314, 155.