High pressure liquid chromatographic method for routine analysis of

---

High Pressure Liquid Chromatographic Method for Routine Analysis of Major Parent Polycyclic Aromatic Hydrocarbons in Suspended Particulate Matter Michael Dong and David C. Locke* Department of Chemistry, City University of New York, Graduate Division, Queens College, Flushing, N. Y. 11367

Edward Ferrand Bureau of Technical Services, New York City Depadment of Air Resources, 51, Astor Place, New York, N. Y. 10003

A high pressure liquid chromatographic method has been developed and applied to polycyclic aromatic hydrocarbons (PAH) in New York City suspended particulate matter samples. Particulate matter collected on filters was Soxhlet-extracted using cyclohexane, and extractable matter was prefractionated by thin layer chromatography. The PAH fraction was concentrated and separated by HPLC with an ODS Zorbax column. About 20 PAH were identified in effluent fractions by uv and fluorescence spectrophotometry. Submicrogram quantities of PAH were quantitated by the peak height method. The detection limit of benzo[a]pyrene is about 10 ng. Good crosschecks were obtained by gas chromatography-mass spectrometry (GC-MS).

cy, high sample capacity, and show long term reproducibilit y without bleeding. The collected PAH fractions are directly compatible with uv spectrophotometry with no prior concentration necessary. I t should be pointed out that some PAH isomers can be separated by GC using capillary ( 1 7 ) and long packed ( 1 8 ) columns. Very recently, nematic liquid crystals used as GC stationary phases have been reported ( 1 9 ) to show selectivity similar to HPLC for PAH separations. GC-MS will no doubt continue to be the most effective technique for complete environmental analysis, e.g., for the alkyl and hydro derivatives of PAH. However, the HPLC method developed here does offer an attractive alternative for the routine monitoring of PAH, in terms of cost, accuracy, and ease of operation.

Polycyclic aromatic hydrocarbons (PAH) are the most extensively studied components in airborne particulate matter ( I ) , because many PAH have long been known to be potent carcinogens. Since larger amounts of PAH are produced in the burning of coal than of fuel oil ( 2 ) ,it is important to monitor the PAH level in air if the trend towards increasing coal consumption continues. Analytical methods have been well reviewed ( 2 - 4 ) . Although synthetic PAH mixtures have been separated by high pressure liquid chromatography (HPLC) using reversed-phase (5-7), adsorption ( 8 ) , partition (9-11) and complexation (12-14); no satisfactory quantitative procedure has been published for analysis of PAH in suspended particulate matter. We are reporting a new and rapid method for quantitative analysis of all major parent PAH in air particulate extracts. HPLC is clearly superior to conventional LC in terms of resolution, analysis time, and analytical precision. In the HPLC method described here, a total of approximately 5 h is needed from the extraction of the air filter to the obtaining of the chromatogram. The liquid chromatographic separation takes 100 min, which can be reduced to 20 min if lower resolution is allowed. A conventional LC separation of comparable resolution takes 2-5 days using a long alumina column ( 1 5 ) . HPLC offers many advantages over GC in PAH analysis: it operates a t lower temperatures (below 80 "C); has available a uv detector highly selective for PAH whose molar absorptivities range from lo4 to lo5 a t 254 nm; fraction collection followed by uv and fluorescence spectrophotometry is simple and convenient; reversed-phase packings are highly selective for separating isomeric PAH such as benzo[alpyrene and benzo[e]pyrene which are not normally separated by GC ( 1 6 ) and whose mass spectra are sufficiently similar to make absolute identification with GC-MS difficult. The ODS Zorbax columns used here have good efficien-

EXPERIMENTAL

368 * ANALYTICAL CHEMISTRY, VOL. 48, NO. 2, FEBRUARY 1976

T h e experimental procedure is outlined in Figure 1. Materials. Spectro-grade cyclohexane and methanol were obtained from Burdick and Jackson Inc. PAH standards were purchased from Aldrich, K&K, and Eastman Kodak Co. Some PAH were further purified by zone-refining. Pre-coated silica gel-G T L C plates (500 p thick without fluorescence indicator) were obtained from Analtech Inc. Sampling. Particulate matter samples were provided by the New York City Department of Air Resources. They were collected a t various sampling sites of New York City's aerometric network on glass fiber filters using high volume samplers. The average rate of sampling is 40-50 ft3/min. T h e weight of a 24-hr sample ranges from 90 to 200 mg depending on site, season, and weather conditions. Sample S301 was taken from station 3 (South Bronx) and S303 from station 10 (Manhattan East side) in November and December of 1974. Both locations are in areas of heavy particulate matter pollution and each sample was a combined extract of 6 filters selected randomly across the two months. S201 was a combination of 18 filters collected during March 1974, randomly selected from different stations. Extraction. Although cyclohexane, benzene, and acetone have all been shown to be nearly 100% efficient in Soxhlet extraction of benzo[a]pyrene from filters (20), cyclohexane appears to be the most suitable since it tends to extract fewer extraneous materials and thus eases the burden on the separation method. Material from two or more filters is required when uv characterization of effluent is employed. Filters were Soxhlet extracted for 6 h using 150 ml of cyclohexane. The extract, a clear yellow solution, was concentrated to 5 ml in a rotary evaporator. Sample Pre-Fractionation. Pre-fractionation is necessary to isolate the PAH from the paraffins, olefins, heterocyclics, etc., also present. The pre-fractionation procedure of Brocco e t al. (21) using T L C was employed. A TLC of a typical NYC air particulate extract is shown in Figure 2. Greater than 90% recovery on tested PAH was reported (21) using a similar TLC-GC technique. Separation a n d Identification. A Du Pont 820 Liquid Chromatograph equipped with a fixed wavelength (254 nm) uv detector, refractive index detector, and gradient elution accessory was used with a 25-cm long, 2.1-mm i.d. ODS Zorbax column. Carefully prepared synthetic mixtures of PAH standards were routinely in-

0.4AIR PARTICULATE MATTER

3

-z c

3

r L

EXTRACTION

1

0 3..

'

PAH

z

3

PRE-FRACTIONATION

~

I

W 0

0.2-TLC

1

(0

9

7

e 0 . ' - , 2

PAH FRACTION

STANDARDS

6

47

-

9

___

TIME IN MINUTES P U A N ~ l ~ @ ~ ~ 161l RETENTION ~ ~ TIME a

Figure 3. Liquid chromatogram of synthetic PAH mixture No. 1

I DE NTI F IC AT I ON

Chromatographic conditions: MeOH/H20 65-35, 60 O C , 1200 psi, 0.21 ml/ min, 0.64 full scale absorbance unit (FSAU), 5-pl inject, 1% MeOH/min gradient initiated after the B[a]P peak. Column: ODS Zorbax (25 cm, 2.1-mm i.d.). Peaks: (1) solvent; (2) naphthalene (694 ng); (3) anthracene (139 ng); (4) fluoranthene (421 ng); (5) pyrene (626 ng); (6) triphenylene (120 ng); (7) benz[a]anthracene (216 ng); (8) perylene (137 ng); (9) benzo[a]pyrene (95 ng); (10) benzo[ghi]perylene (129 ng); (1 1) coronene (40 ng)

PEAK HEIGHT

1

$'E;

TFRLdJORESCENCE

Figure 1. Procedure for determination of PAH in air particulate matter

TLC

S I L I C A GEL-G

500 microns

1.5/ 1.0

CYCLOHEXANE/BENZENE

,

3 SOLVENT

7

FRONT

0.2-

r UNDER

UNDER VISIBLE LIGHT

n L - LIGHT D - DARK 8-Blue Br-Brown G -Green P - Purple

Uy

0 05-

6

START

Y-Yellow 8r ( I * ]

-

Brown spot in iodine Chambar

6

GRADIENT/

TIME I N MINUTES

Figure 4. Liquid chromatogram of synthetic PAH mixture No. 2

r

START A I R SAMPLE

Figure 2. Typical thin layer chromatographic pre-fractionation of a NYC air particulate extract jected to check column performance and to optimize operating conditions. One to five pl of pre-fractionated air particulate extract was sufficient for analysis by H P L C peak height method. A 30-50 p1 sample size was necessary when fractions were t o be analyzed by uv. Three-tenth-ml fractions, sufficient to fill a uv microcell of p a t h length 4 mm, were taken. A Cary 14 UV-VIS spectrophotometer and a Perkin-Elmer MFA-PA spectrofluorometer were used to obtain spectra. T h e latter gives spectra uncorrected for source intensity and detector wavelength response variations. Reference spectra were either obtained from PAH standards or from the literature (22-25). GC-MS Crosschecks. T h e PAH fraction of air samples S301 and S303 were analyzed by GC-MS, using a Perkin-Elmer 800 GC coupled to a Nuclide 12-90-G Mass Spectrometer. Mass spectra of each emerging peak were taken at electron energies of 14 and 70 eV. A 12-ft Dexsil-300 GC column was used under conditions similar to those reported by Lao et al. (16). Since there is some uncertainty about the split ratio (FID/MS) and about the efficiency of the Biemann-Watson separator, chromatograms were also obtained using the same column in a Perkin-Elmer 900 GC for quantitative comparison. Blank. Extraction of 2 unexposed filters and subsequent analysis using the identical HPLC procedure showed no PAH.

RESULTS AND DISCUSSION Optimization of Operating Conditions. The ODS Zor-

Chromatographic conditions same as Figure 3. Peaks: (1) solvent; (2) fluorene (276 ng); (3) phenanthrene (291 ng); (4) impurity (anthracene); (5)chrysene (85 ng); (6) benzo[e]pyrene (67 ng); (7) triphenyl benzene (209 ng); (8) anthanthrene (18 ng); (9)dibenzopyrene (10 ng)

bax column, 15% octadecyl silane (ODS) bonded to porous silica microspheres (Zorbax) having a diameter of 5-7 w , is very similar to the ODS Permaphase (Du Pont), 1%ODS bonded to a 37-p diameter pellicular packing (Zipax) ( 5 ) . The optimum operating conditions were chosen empirically by adjusting the amount of MeOH in the mobile phase, temperature, and flow rate singly or in combination for best separation in a reasonable time. Our best separation of 2 synthetic mixtures was obtained a t 60 "C, 65/35 MeOH-H20 and 0.21 ml/min. These are shown in Figures 3 and 4. The calculated number of theoretical plates for the benzo[a]pyrene peak is close to 4000 (similar to the efficiency claimed by the manufacturer). A gradient of 1% MeOH/min was initiated from 65 to 100%after the benzo[alpyrene peak eluted to hasten the elution of the more strongly retained compounds. The base-line shift accompanying the use of gradient elution is caused by the changing refractive index of the effluent. Suspended Particulate Matter Samples. Figure 5 contains 3 liquid chromatograms of 5-pl aliquots of TLC prefractionated air samples (S303, S301, and S201) using the same HPLC conditions as in Figure 3. A gas chromatogram of S303 is shown in Figure 6 for comparison. The envelope under the GC peaks probably represents a conglomeration ANALYTICAL CHEMISTRY, VOL. 48, NO. 2, FEBRUARY 1976

369

0.4

A I R SAMPLES

r

-If

I

10

0 -rd'4"fiJ

o

io

L./

20

30

40 50 €0 70 TIME IN MINUTES

80

90

40

30

20

50

TIME IN MINUTES

,rSTART QRADIENT

100

Figure 6 . Gas chromatogram of air sample

Figure 5. Liquid chromatograms of air samples Chromatographic conditions same as Figure 3. Peaks: (1) solvent; (2) phenanthrene; (3) anthracene; (4) fluoranthene; (5) pyrene; (6) triphenylene; (7) benzo[ghi]fluoranthene; (8) chrysene; (9) benz[a]anthracene; (10) benzo[j]fluoranthene; (1 1) benzo[b]fluoranthene; (12) benzo[e]pyrene; (13) perylene; (14) benzo[k]fluoranthene; (15) benzo[a]pyrene; (16) benzo[gbi]perylene; (17) indeno[l23-cd]pyrene; (18) anthanthrene; (19) coronene

of minor components also present in the TLC PAH fraction. Samples taken from areas of lower particulate matter pollution show significantly less pronounced envelopes. Only major peaks definitely identified by GC retention time and MS are labeled in Figure 6. In Figure 5 , 18 PAH are identified in air samples. Other peaks were only tentatively identified or not identified because of their low concentrations. The major chromatographic peaks group according to the number of aromatic rings. The first pair consists of 3-ring phenanthrene and anthracene; these cannot normally be separated by GC. Similarly, the 4-ring isomers, triphenylene, benz[a]anthracene and chrysene emerge as a single peak in GC (Figure 6). Benz[a]anthracene and chrysene, one of the most difficult pairs to separate, are better separated on ODS Permaphase with 50/50 MeOH-H20 and 40 "C. These conditions cannot be used in the ODS Zorbax column because of the slower flow rate and higher stationary phase loading. GC separates the 5-ring group in 4 peaks Blj]F, B[b]F and B[k]F,

S303

Chromatographic conditions: 165 OC for 2 mln, then programmed to 320 OC at 4 OC/min, 300 OC (injector). 320 OC (detector), sensitivity X256, helium flow rate 30 ml/min, hydrogen pressure 20 psi, air pressure 30 psi. Column: 12 ft, '/*, 6 % Dexsil-300GC on chromosorb W (HP), 80/100 mesh. Peaks: (1) solvent; (2) phenanthrene and anthracene; (3) methylphenanthrene; (4) methylanthracene; (5) aceanthrylene; (6) ethylphenanthrene and ethylanthracene; (7) fluoranthene; (8) pyrene; (9) methylfluoranthene; (10) methylpyrene; (1 1) benzo[c]phenanthrene; (12) benzo[ghi]fluoranthene; (13) triphenylene, chrysene, and benz[a]anthracene; (14) methylbenzo[ghi]fluoranthene; (15) methyltriphenylene, methylchrysene, and methylbenz[a]anthracene: (16) benzo[j]fluoranthene; (17) benzo[b]fluoranthene and benzo[k]fluoranthene; (18) benzo[e]pyrene and benzo[a]pyrene; (19) perylene; (20) methylbenzo [b] fluoranthene and methylbenzo[ k]fluoranthene; (2 1) methylbenzo[ e] pyrene and methylbenzo[a]pyrene: (22) dibenzanthracene; (23) lndeno[ 123cdlpyrene; (24) benzo[gbi]perylene and anthanthrene; (25) methylbenzo[ gbi]perylene; (26) dibenzopyrenes; (27) coronene

B[a]P and B[e]P, and perylene while in HPLC, B[a]P is well resolved from the rest (see Table I for PAH abbreviations). Liquid chromatograms of most air samples studied show remarkably similar features. All the major PAH are always present but usually vary in relative proportions. This reflects the fact that PAH are formed from combustion processes, e.g., automobiles, space heating, incinerators etc., which are ubiquitous. However, the high proportion of pyrene, benzo[ghi]perylene and coronene in S301 is a good indication that the majority of PAH are derived from automobiles (26). The reverse is true for S303, indicating do-

Table I. Comparison of Quantitation Methods on Air Sample S303 Concentration in kg/1000 m 3 Compound names

Phenanthrene Anthracene Fluoranthene Pyrene Triphenylene Benz[a]anthracene Chrysene Benzo[ghi] fluoranthene Benzo[ c ] phenanthrene Benzo[j] fluoranthene Benzo[ b ] fluoranthene Benzo[e]pyrene Perylene Benzo [ h 3 fluoranthene Benzo [ a ] pyrene Benzo[ghi] perylene Indeno[ 1,2,3-cd]pyrene Anthanthrene Coronene Di benzopyrene a Using area. 370

Abbreviation

Phen A F P

(a) HPLCjUV

... ...

... ...

ANALYTICAL CHEMISTRY, VOL. 48, NO.

(c) GC peak area

;::\

0.09 10.80 0'71 1.1 1.05

0.58 B[a']A Ch B[ghi] F B[c]Phen Blil F B[b 1 F B[elP Per B[klF B[alP B[ghi]Per

(b) HPLC peak height

0.6

1.9 2.0

1.8

1

1

1.6 13.0 3'6 0.19 0.04

2.7 0.95

0.1

0.2 0.6 1.3

0.1 0.4

2, FEBRUARY 1976

...

4 1.3-19

... ...

1.0

0.3

...

1.15 0.9 0.3

...

4 1.3-11.6

0.3 0.8 1.4

0.6

Av. U.S. urban concn. from (1)

0.2

I

3.2 for B[alP & B[elP

14.0

t

0.2 3.0 for

B[b]F & B[h]F 1.4 12.3

0.18

0.5

...

.

.

...

i1.5

... 5 0.7 0.5-2.0 5.7 8

...

0.26 2

...

FRACTION 6 PYRENE

I

I

400 WAVELENGTH MU

500

Figure 8. Fluorescence spectrum of the benzo[a] pyrene fraction. Excitation slit width, 4 nm; emission slit width, 4 n m i

FRACTION 21

201 If

1

IO'

B(a) P

+,

300

A I "

l'y

c'

,

400

WAVELENGTH MU

Figure 9. Excitation spectrum of the benzo[a]pyrene fraction. Excitation slit width, 2 nm, emission slit width, 4 n m '.-

0

250

_ <

I

*.-,-___

300

WAVELENGTH

350

MU

Flgure 7. UV spectrum of the pyrene fraction

mestic fuel burning to be the major particulate source. Figure 7 is a uv spectrum of the pyrene fraction isolated from S303 by HPLC. The solid lines give the spectrum of the fraction and the dotted lines give the spectrum of the pure PAH a t a similar concentration. The uv spectra of the isolated PAH are of significantly better quality than most published spectra, because of the higher purity of these compounds. Figures 8 and 9 show fluorescence and excitation spectra of the B[a]P fraction in S303. Collected fractions were first evaporated and samples redissolved in 0.5 ml of cyclohexane for fluorescence measurements. Quantitation. Table I gives comparison data on the determination of 20 PAH by 3 different methods on S303, along with average PAH concentration in U.S. urban areas ( I ) for comparison. A brief description of these methods follows: (a) HPLC/ UV method: The concentrations of each PAH were calculated from peak height of the uv spectra of the collected fractions. Molar absorptivities were taken from the literature (23). The base-line method (27) was used in most cases. (b) HPLC peak height method: The height of each PAH peak in the sample was compared with that of the standard of known concentration. (c) GC peak area method: Fluoranthene, which is always present in air samples and which emerges as a well separated peak, was used as an internal reference compound for detector response factor ( 1 6 ) . The amount of fluoranthene in the sample was first obtained by comparing its peak area with that resulting from an equal volume injection of standard fluoranthene solution. Since the flame ionization detector (FID) response factors of other PAH relative to fluoranthene have been published ( 1 6 ) , their quantities in the sample can be

calculated from the relative peak areas. The HPLC/UV method should be the more accurate method since it is subject to fewer interferences. The GC and HPLC results are in reasonably good agreement. The fact that the HPLC/UV method always gives the lowest result suggests that some of the chromatographic peaks include more than one compound. Quantitation by HPLC peak heights suffers from insufficient resolution while the GC peak area method suffers from the nonselective response of the FID.

ACKNOWLEDGMENT The authors thank B. Aubrey, J. Bove, P. Dalven, V. Kukreja, T. Ntugagu, S. Siebenberg (NYC Dept. of Air Resources), and J. Schmermund (Queens College, CUNY) for helpful discussions and advice.

LITERATURE CITED (1) 0.Hoffmann and E. I. Wynder in "Air Pollution", 2nd ed., Academic Press, New York. 1968, p 187. (2) "Particulate Polycyclic Organic Matter", National Academy of Sciences. Washington, D.C., 1972. (3) 0. Hutzinger, S. Safe, and M. Zanders, "Polycyclic Aromatic Hydrocarbons", Analabs, Inc., Res. Notes. 13, 3 (Dec. 1973). (4) P. K. Mueller and E. L. Kothny, Anal. Chem., 45(5), 1R (1973). (5) J. A. Schmit, R. A. Henry, R. C. Williams, and J. F. Dieckman, J, Chromatogr. Sci., 9, 645 (1971). (6) C. G. Vaughan, B. B. Wheals, and M. J. Whitehouse, J. Chromatogr., 78, 203 (1973). (7) B. B. Wheals C. G. Vaughan, and M. J. Whitehouse, J. Chromatogr., 106, 109 (1975). (8) W. Strubert, Chromatographia. 6, 205 (1973). (9) Kiimisch, J. Chromatogr., 83, 11 (1973). (10) H.-J. Klimisch, Anal. Chem., 45, 11 (1973). (11) R. E. JentoftandT. H. Gouw, Anal. Chem., 40, 1787(1968). (12) B. L. Karger, M. Martin, J. Loheac, and G. Guiochon, Anal. Chem., 45, 496 (1973). (13) R. Vivilecchia, M. Thiebaud, and R. W. Frei, J. Chromatogr. Sci,, 10, 411 (1972). (14) C. H. Lochmuller and C. W. Amoss. J. Chromatogr., 108, 85 (1975). (15) G. J. Cleary, J. Chromatogr., 9, 204 (1962). (16) R. C. Lao, R. S. Thomas, H. Oja. and L. Dubois, Anal. Chem.. 45, 908 (1973).

ANALYTICAL CHEMISTRY, VOL. 48, NO. 2, FEBRUARY 1976

371

(17) M. Novotny, M. L. Lee, and K. D. Bartle, J. Chromatogr. Sci., 12, 606 (1974). (18) B. B. Chakraborty and R. Long, Environ. Sci. Techno/., 1, 829 (1967). (19) G. M. Janini, K. Johnston, and W. L. Zielinski, Jr., Anal. Chsm., 47, 670 (1975). (20) T. W. Stanley, J. W. Meeker, and M. J. Morgan, Environ. Sci. Techno/., I, 927 (1967). (21) D. Brocco. V. Cantuti, and G. P. Cartoni, J. Chromatogr., 49, 66 (1969). (22) E. J. Clar, "Polycyclic Hydrocarbons", 2 vols, Academic Press, New York, 1964. (23) "UV Atlas of Organic Compounds", 5 vols, Plenum Press, New York,

1967-68. (24) R. A. Friedel and M. Orchin, "UV Spectra of Organic Compounds", John Wiley and Sons, New York, 1967. (25) T. J. Porro, R. E. Anacreon, P. S . Flandreau, and I. S. Fagerson, J. Assoc. Off. Anal. Chem., 56, 607 (1973). (26) R. Jeltes, J. Chromatogr. Sci., 12,599 (1974). (27) R. L. Cooper, Analyst (London),79, 573 (1954).

RECEIVEDfor review June 11,1975. Accepted November 7, 1975.

Determination of Theophylline in Plasma Ultrafiltrate by Reversed Phase High Pressure Liquid Chromatography L. C. Franconi and G. L. H a w k * Waters Associates, Maple Street, Milford, Mass. 0 1757

B. J. Sandmann and W. G. Haney University of Missouri-Kansas City, School of Pharmacy, 5 100 Rockhill Road, Kansas City, Mo. 64 1 10

A procedure for the determination of theophylline in plasma by reversed phase hlgh prqssure liquid chromatography is presented. This method uses molecular flltratlon to remove plasma proteins prior to chromatographic analysis. It permits the accurate measurement of plasma levels of theophylline without Interference from the dietary xanthines and their metabolites, caffeine, and 3,7-dlmethylxanthine. No Interference from commonly used drugs or thelr rnetabolltes was found from 75 randomly collected plasma samples. In 55 comparative determinations, the LC method was found to be comparable to the GLC and spectrophotometrlc methods currently being employed.

Theophylline appears to be one of numerous drugs (1) for which monitoring drug plasma concentration is often necessary to ensure effective therapy. Significant variations in theophylline plasma concentrations resulting from a given daily dose have been noted ( 2 ) ,and response to the drug relates to the drug plasma concentration. Therefore, rapid and accurate procedures for the analysis of theophylline in plasma are essential. Because of the strong ultraviolet absorption characteristics of the drug, most procedures designed for its determination in plasma have been based on spectrophotometric analysis (3, 4 ) . Typically, theophylline contained in a volume of plasma is extracted into an organic solvent from which it is back-extracted into an aqueous basic solution. The aqueous solution is neutralized, and the ultraviolet spectrum is determined. Such procedures have two intrinsic deficiencies. First, the partition coefficient of theophylline between organic solvents and water is low, and the initial extraction is, therefore, inefficient. Second, the class of compounds of which theophylline is a member, is often encountered in plasma as a consequence of the metabolism of endogenous biochemicals and of exogenously administered substances, e.g., the ubiquitious caffeine. The currently available ultraviolet procedures are not sufficiently selective to distinguish between members of this class of compounds ( 5 ) . In addition, a number of drugs and their me372

tabolites strongly absorb a t the analytical wavelength, and those present in the final extract (for example, weak acids such as the barbiturates) offer additional complications. This lack of selectivity has led to an interest in chromatographic procedures. Gas-liquid chromatography (GLC) has been used for this determination (5), but derivatives of the drug must be formed prior to analysis. In addition, the chromatographic internal standard used in this GLC method is known (6) to be unstable in the methylating reagent. A GLC procedure offering several improvements has also been recently proposed (7). High speed liquid chromatography has enjoyed some acceptance, but the column packings used thus far have been unstable (8)and retention volumes of metabolites and related substances have been high (9). In addition, these procedures continue to rely on the inefficient extraction of theophylline from plasma or serum. After this study was completed, a procedure for the' determination of theophylline by direct injection of plasma was reported ( I O ) . In light of the above, a non-extractive procedure for the analysis of theophylline in patient plasma by reversed phase HPLC has been developed and evaluated. Results of this procedure have been compared with results from an ultraviolet and a GLC analysis.

EXPERIMENTAL Apparatus and Operating Conditions. A liquid chromatograph equipped with a 6000 psi pump and high pressure injector (ALC Model 202 with Model 6000 pump and U6K injector, Waters Associates, Milford, Mass.) was used for the analyses, and eluent was monitored continuously a t 254 nm. The peak areas were determined by triangulation. The mobile phase was 0.01 M sodium acetate (adjusted to p H 4.0 with acetic acid) and acetonitrile in a ratio by volume of 9:1, and the flow rate was 2.0 ml/min. Plasma samples were filtered a t ambient temperature using a multiple unit system (Pellicon Carrousel and Pellicon PSED 01310, Millipore Corporation, Bedford, Mass.) with 3.0-ml, 13-mm stirred polycarbonate cells. The pressure was maintained a t 50 psi with nitrogen. Membranes. A membrane (Membrane A) consisting of a continuous polymer film supported on a microporous substrate of mixed esters of cellulose and with a nominal molecular weight limit of 25 000 was evaluated in this study. In addition, a molecular

ANALYTICAL CHEMISTRY, VOL. 48, NO. 2, FEBRUARY 1976