flow injection analysis with

Sep 1, 1982 - Solvent Extraction-Flow Injection without Phase Separation through the Use of .... The first decade of flow injection analysis: from ser...
2 downloads 0 Views 626KB Size
1693

Anal. Chem. 1982, 5 4 , 1693-1697

Characterization of Solvent Extraction/Flow Injection Analysis with Constant Pressure Pumping and Determination of Procyclidine Hydrochloride in Tablets Lynette Fossey (and Frederick F. Cantwell” Department of ChenTistry, University of Alberta, Edmonton, Alberta, Canada T6G 2G2

A flow Injection extractlon apparatus utillzlng a membrane phase separator and constant pressure pumping is described and characterized In terms of extractlon coil length, sample Injection volume, and flow rates. Equations are derived and verlfled that show that under condltlons where the sample component is quantltatlvely extracted into the organic phase, the peak area depends only on the number of moles of sample Injected and the total flow rate of organic solvent. A sampling frequency of 4 samples/min is readily achieved. Procyclldlne hydrochloride Is assayed in tablets with 1% preclslon and accuracy.

In previous papers from this laboratory, membrane phase separators have been described for two-phase titrations (2S31) and for characterizing solvent extraction equilibria (32,33). In the present paper we report the design of an extraction/FIA apparatus employing a membrane phase separator. A constant pressure pumping system is used with a peristaltic pump at the outlet ends of the flow streams to gate the flow. Equations relating peak area of the extracted sample to flow rates are derived and verified. A rapid assay method for procyclidine hydrochloride in pharmaceutical tablets is developed, based on ion-pair extraction of the drug with picrate.

EXPERIMENTAL SECTION Flow injection analysis (FIA), pioneered by Ruzicka and Hansen ( I ) and by Stewart et al. (2),has been applied to many kinds of analytical determinations. The technique has recently been reviewed (3, 4 ) . In some recent studies a solvent extraction step is included in the FIA system. Since the key to rapid, on-stream solvent extraction is an efficient means of phase separation, extraction/FIA systems may be catagorized by the type of phase separator employed: Devices using a chamber rely on gravity to separate the phases (5,6); those with a “tee” separator employ gravity with or without some sort of phase guide made of a material wetted by one phase but not the other (7-22); and those with a membrane phase separator depend upon the selective permeability of a porous membrane to the phase which wets the membrane material (23-27). Some work has been done on monitoring concentration via fluorescence in a two-phase flowing stream without employing a phase separating device (26, 28). Compared to the chamber and “tee” devices, a membrane phase separator can be used with a smaller internal volume, which yields less band broadening, and it is more reliable a t separating the phases at high flow velocities. Consequently, membranes should make possible shorter analysis times. The membrane design can also be used with a larger variety of water-immiscible solvents, since a density difference between the aqueous and organic phases is not required. Usually peristaltic pumps have been used to pump reagents and solvents through the extraction/FIA system. However, rigorous control of flow rates is not always possible with this type of pump alone because of deterioration of the viton rubber (e.g., Acidilex) pump tubes caused by the organic solvent (15). In o w experience, small particles rubbed off the inside of the pump tube tend to clog the porous membrane, further aggravating the problem of changing flow rates. High-pressure liquid chromatography pumps which deliver constant flow rates are one solution to the problem (6,25), though an expensive one when several solvents are to be pumped. Constant pressure pumping offers a less costly alternative (2, 20). Very similar to extraction/FIA in principle are postcolumn extraction detectors that have been developed for liquid chromatography (8, 11-13, 17-20), 0003-2700/82/0354-1693$01.25/0

Apparatus. A schematic diagram of the extraction/FIA system used is shown in Figure 1. The chloroform to be pumped is held in a 1700-mLglass bottle while the water, the aqueous reagent, and the methanol to be pumped are held in 2-L polyethylene bottles. The bottles are placed inside aluminum cylinders which are pressurized with nitrogen. The liquids to be delivered contact only Teflon and either glass or polyethylene. All tubing is Teflon, with 0.3 mm i.d. tubing used whenever it is desirable to minimize sample band broadening or to provide increased resistance to flow and 0.8 mm tubing used in the rest of the system. Two-way Teflon valves, V1 (part no. CAV2031, Laboratory Data Control, Riviera Beach, FL), placed in the solvent delivery lines from the cylinders allow shut-off of each individual flow. A three-way valve V2 (part no. CAV3031, Laboratory Data Control), allows selection of either chloroform or methanol. The latter is used to wash out the system at the beginning and end of the day. The water flows first through an automatic sample injection valve, V3 (part no. SVA-8031, Laboratory Data Control), before joining the aqueous reagent stream at a tee-fitting, TI (part no. CJ-3031, Laboratory Data Control). Valve V3 is actuated by an air solenoid valve (part no. SOL-3-24-VDC, Laboratory Data Control) controlled by an electrical timer which allows variation of fill time and injection time. The combined aqueous stream then joins the chloroform stream at tee-fitting, T2, and the resulting flow stream passars through the extraction coil, C, in which solvent extraction occurs between the aqueous and chloroform phases. The lengths of 0.3 mm i.d. tubing connecting T1to both V3 and T2are made as short as possible to minimize dispersion of the sample zone in tlhese unsegmented-flow regions. A specified length of coiled 0.8 mm i.d. Teflon extraction tube, C, connects T2with the membrane phase separator, M. In M, a fraction of the chloroform phase is separated from the segmented aqueous/chloroform flow stream and passes through the 10-pL flow cell of the spectrophlDtometer,S (UV-50 photometric detector, Varian Instruments, Walnut Creek, CA). The chloroform flow stream exiting the spectrophotometer and the aqueous/chloroform flow stream exiting the membrane phase separator pass through Acidflex pump tubes (Technicon Corp., Tarrytown, NY) in a variable speed peristaltic pump, P (Minipuls 2, Gilson Instrument, Ville-le-Belle, France). Flow rates in both streams are measured by collecting the effluents in burets or graduated cylinders and timing with a stopwatch. The speed of pump P is set so that the total flow rate of both phases is somewhat less than the flow rate observed when the tension is released from the pump rollers. This ensures a slight backpressure and prevents out-gassing from the solvents. The signal 0 1982 American Chemlcai Society

1694

ANALYTICAL CHEMISTRY, VOL. 54, NO. 11, SEPTEMBER 1982

Flgure 1.

details.

Diagram of the extractlon/FIA instrument. See text for FT- FM

A

0

40

80

120

140

200

240

280

320

Extraction Coil Length (crn)

Peak area (A) and peak width at half-height (B) for 90 pprn caffeineinjected vs. length of extraction coil. Areas in A are corrected to F, = 4.6 mL/min by using eq 5. Instrument parameters: F, 3 mL/mln; F , 2 mL/min; injection volume, 44 pL; N, pressure, 40 psig; organic phase, CHCI,; reagent, H20; wavelength, 273 nm. Figure 3.

-

1 crn

4

B

FT

A Flgure 2. Cross section of the membrane phase separator (A). Teflon membrane shown as short dashes. Stainless steel end-plates shown in hashing. Perforated Teflon membrane-support shown in long dashes in A and in front view in B.

from S is fed to a Model 3390A digital recording integrator, I, (Hewlett-Packard Co.) which allows measurement of both peak height and peak area. Figure 2 shows a cross section of the membrane phase separator, M. Two layers of 4-mil, 10-20 fim pore size Teflon membrane (Zitex, No. E249-122, Chemplast Inc., Wayne, NJ) backed by a perforated Teflon support are sandwiched between the two main-body pieces which are made of Kel-F. The main-body pieces are pressed together with four screws and two stainless steel end plates. The volume of the membrane chamber is about 0.06 mL. The three threaded holes accept the standard polypropylene end pieces (part no. TEF 107, Laboratory Data Control) and flaired Teflon tubing. Reagents. Water was demineralized, distilled, and finally distilled from alkaline permanganate. Analytical grade chloroform (B.D.H. Chemicals) was distilled before use, and analytical grade methanol (Terochem Labs Ltd.) was used as received. A 4.92 X M picric acid (Matheson, Coleman and Bell) stock solution was prepared in water and standardized by titration with sodium hydroxide. Picrate reagent solutions were prepared by combining aliquots of the stock solution with NaOH or HC1 to adjust the pH and with NaCl to adjust the ionic strength and were then diluted to volume. Caffeine was U.S.P. grade and both procyclidine hydrochloride and procyclidine hydrochloride tablets (Kemedrin Tablets, 5 mg, Burroughs Wellcome Ltd., La Salle, Quebec) were B.P. grade. Procyclidine Hydrochloride Assay. Twenty tablets were weighed and ground to a f i e powder. One tablet weight of powder was transferred in duplicate into 100-mLvolumetric flasks, diluted to volume with water, and shaken vigorously for 5 min. The solutions were centrifuged and the clear supernatant liquids constituted the sample solutions.

-

Important instrument parameters for the assay were as follows: total chloroform flow rate, 3.2 mL/min; total aqueous flow rate, 3.3 mL/min; chloroform flow rate through the membrane, 1.7 mL/min; extraction coil length, 200 cm; volume injected, 44 pL; sampling frequency (injection rate), 2 samples/min; wavelength, M picrate 400 nm; nitrogen pressure, 40 psig; reagent, 2.46 X with 0.10 M NaCl; pH 2.01. (A warm-up time of about 30 min is used once solvent flow is established.)

RESULTS AND DISCUSSION The characteristics of the extraction/FIA system were studied by using caffeine as a sample component. The variations of sample peak area, width, and height were investigated as a function of extraction coil length, sample injection volume, and liquid flow rates. The following symbols are used for flow rates: Fa is the total flow rate of the aqueous phase, F, is the total flow rate of the organic phase, Fm is the flow rate of organic phase through the membrane, and FT is the sum of F, and Fa. Extraction Coil. The influence of the length of the extraction coil on peak area and peak width a t half-height was studied by changing the length of the coil while keeping constant both the concentration and the volume of the caffeine solution injected. Flow rates F,, Fa, and F,, measured after each run, were found to vary slightly with increasing length of extraction coil, and the effect of their variation was compensated via eq 5 discussed later in this paper. The results are shown in Figure 3A where it is seen that peak area increases with extraction coil length prior to attainment of distribution equilibrium a t about 7 5 cm, beyond which it becomes constant. Peak width at half-height exhibits an initial increase of about 20% with increasing coil length up to the region where distribution equilibrium is attained, and thereafter it increases at only a very slow rate (Figure 3B). While an understanding of the reasons for the faster rate of increase of peak width a t short extraction coil lengths will require further study, the most significant aspect of peak width behavior revealed in Figure 3B is the fact that, for coil lengths in the equilibrium region, peak widths become nearly independent of coil length. Consequently, the extraction coil contributes only very slightly to sample zone dispersion. This is in marked contrast to the extensive laminar flow zone broadening observed in unsegmented flow-through small-bore tubing.

ANALYTICAL CHEMISTRY, VOL. 54, NO. 11, SEPTEMBER 1982

10

r--I

means of the equations presented below. In a concentration-type detector, such as a spectrophotometer, the area of the peak, A, is inversely proportional to the flow rate through the detector F,. In an extraction process involving a fixed number of moles of sample component injected, the peak area is also proportional tal the fraction of sample in the organic phase. At equilibrium this fraction is @’/(I k’)) where k’ is the so-called capacity factor of the sample component (36) and is equal to the ratio of moles of sample component in the organic phase to the moles in the aqueous phase. Peak area is also proportional to the fraction of the organic phase that goes through the membrane and thus goes through the detector (Fm/Fo);and it is proportional to the number of moles of sample injected, n. Combining these relationships and adding the proportionality constant K give an equation for peak area

I

‘ 0

1695

+

L

It ’

/

0 2 oo.0/O

0

200

1 400

800

600

Sample Volume (yL)

-

-

Figure 4. Peak wldth at half-height for 90 ppm caffeine vs. sample 3.0 mL/min; F a volume Injected. Instrument parameters: F , 4.0 mL/mln; F , 1.8 mL/min; extraction coil length, 200 cm; N2 pressure, 40 psig; organlc phase, CHCI,; reagent, H,O; wavelength, 273 nm.

-

To ensure extraction equilibrium, an extraction coil length of 200 cm was used in subsequent studies of system characteristics. Injection Volume. The volume of sample injected was varied by changing the volume of the sample loop in the automatic injection valve while the extraction coil length, caffeine concentration, and flow rates F,, Fa,and F, were held constant. (The volume injected by each loop was determined in a separate calibration experiment using spectrophotometry.) A plot of peak area vs. injection volume (not shown) was linear with a relative standard deviation of 1.4% for the slope. The intercept, in arbitrary integration units, and its 95% confidence limits were -1.7 X IO3 f 4.0 X lo3, showing that the intercept was zero. A plot of peak width at, half-height vs. sample volume injected is presented in Figure 4. The general shape of the curve is understood as follows: when injection volumes are sufficiently small (550 pL), laminar flow of the caffeine zone in the unsegmented flow regions and the mixing-chamber effect in the phase separator are riufficient to affect the concentration profile along the whole length of the caffeine zone, giving it a more or less skewed Gaussian shape (16,34,35). Thus, for small injection volumes the peak width will remain more or less constant. When the injected volume is large enough (2250 pL) so that laminar flow and mixing chamber effects influence only the leading and trailing edges of the caffeine zone, but not the central portion, then the concentration profile of caffeine along the zone exhibits a flat “peak” whose height is independent of injection volume and whose width is directly proportional to injection volume. For injection volumes between 50 and 250 p L intermediate behavior is observed. This interpretation of peak width behavior is borne-out by a plot of peak height vs. volume injected (not shown). Peak height increases more or less linearly for injection volumes 550 pL, is constant for injection volumes 2250 pL, and is increasing nonlinearly a t intermediate injection volumes. A sample volume of 44 p L was used in all other experiments since this volume is in the region where peak widths are nearly independent of volume injected and, thus, it allows a maximum sampling frequency without unnecessary sacrifice of sensitivity (Le., peak area per unit caffeine concentration in the injected solution). Flow Rates. The dependence of peak area on flow rates Fa,F,, and F, can be predicted and experimentally tested by

The capacity factor may be expressed in terms of the distribution ratio, D (the ratio of concentrations of sample component in the two phases), and the ratio of phase volumes (36). Since the phase volume ratio is equal to the flow rate ratio for the phases, k’may be expressed as

Combining eq 1 and 2 and simplifying give the following general equation for peak area:

A=

nDK Fa DF,

+

(3)

This expression shows that peak area is independent of flow rate through the detector, F,. Qualitatively, this is understood by the fact that as F, is, for example, doubled with F, and Fa held constant, the sample component will spend only half as much time in the flow cell but twice as many moles of sample will pass through the cell. Hence, the area will not change. Equation 3 can be rearranged to give 1 1 = --[FT F,(D - l)] (4) A nDK

+

which indicates that a plot of 1 / A vs. [FT+ F,(D - l)] should be linear with a zero intercept. Such a plot for caffeine was constructed in which F, was varied between 2.7 and 8.2 mL/min and D = 20 (37). The plot (not shown) was linear with a relative standard deviation (RSD) of 1.3% for the slope. The intercept and its 95% confidence limits, in arbitrary integration units, were (1.9 f 3.6) X lo4, showing that the intercept was zero. Under conditions of sufficiently large D, such that OF, >> Fa, the denominator in eq 3 can be approximated by OF,. Substituting and rearranging yield

A = Kn/Fo

(5)

This equation shows that when the sample component is quantitatively extracted into the organic phase, then the peak area depends only on the number of moles of sample injected and the total flow rate of organic solvent. A plot of A vs. 1/F, should be linear with zero intercept. Such a plot is presented in Figure 5 using the caffeine data from the case discussed above. Deviation from linearity occurs for low values of F,, where the assumption DF, >> Fa is no longer valid. Equation 3 or 5 can be used where necessary to compensate for the effects of changing flow rates on peak areas. The increase in peak mea with decreasing F, predicted from eq 5 might suggest that a very small ratio of FJF, should be

1696

ANALYTICAL CHEMISTRY, VOL. 54,NO. 11, SEPTEMBER 1982 1

I

I

I

I

l

l

,

l

l

l

l

l

1

,

1

1

1

I

-

ZQOOO

6,000

l6POO -

m p!

a

7 11

4oo -

lZPO0

zoo0

1

1,000

4,000

O

L

0

0

I

1

0

I

300

400

500

600

[ Procyclidine*HCI 1

(mL/min)-’

Figure 5. Caffeine peak area dependence on flow rates plotted according to eq 5. Instrument parameters: extraction coil length, 200 cm; N2 pressure, varied; organic phase, CHCI,; reagent, H,O; wavelength, 273 nm.

t

200

I Picrate I

004 008 012 016 020 024 028 032 036 040

1

2o

IO0

1

I

10

20

Time (min)

Flgure 6. Replicate injections of 90 ppm caffeine solution at a frequency of four per minute with a relative standard deviation of 1.2% for eight injections. Instrument parameters: F, = 4.4 mL/min; F a = 4.6 mL/min; F, = 3.2 mL/min; extraction coil length, 200 cm; injection volume, 44 ELL; N, pressure, 60 psig; organic phase, CHCI,; reagent, H20; wavelength, 273 nm.

used to achieve increased sensitivity. The more rigorous eq 3 shows, however, that there is a theoretically calculable limit to the advantage resulting from this practice. As well, there is a practical limit. As F,/F, is decreased, there comes a point where water will “break through” the Teflon membrane and enter the detector flow cell causing an erratic signal. In our system the “breakthrough” occurred at a value of Fo/Fa = 0.7 (though the value depends on F,). In our analyses we have found that flow rate ratios F,/F, = 1 and F,/F, 0.5 are always reliable. For more direct proof of the lack of dependence of peak area on F,, a study was performed in which F, was held constant while F, was varied. Two separate peristaltic pumps were used in place of pump P in Figure 1. The resulting plot of A vs. F, (not shown) was a horizontal line of slope 31 and intercept 7914, indicating that A is independent of F,. Sampling Frequency and Calibration. A study was performed under optimum flow conditions to determine the sampling frequency that can be achieved with moderate N2 pressure applied to the aluminum pump cylinders. Figure 6 presents a series of replicate injections of a caffeine solution made at the frequency of four per minute. (Nitrogen pressure was 60 psig.) Base line separation is achieved between each peak and the relative standard deviation for the replicate peaks is 1.2%. This sampling rate can be achieved routinely with no special care or precautions. With the present cylinder design, pressures above 100 psig are not recommended for

-

Flgure 7. Peak area of extracted procyciidlnium-picrate as a function of reagent concentration. Instrument parameters: F, 3.5 mL/min; Fa 3.4mL/min; F, 2.0 mL/min; extraction coil length, 200 cm; injection volume, 44 pL; N, pressure, 40 psig; organic phase, CHCI,; reagent, picrate (pH 2.0,ionic strength 0.10);wavelength, 400 nm; sample concentration, 4.90 X lom5 M procyciidine hydrochloride.

-

-

safety reasons. The time required to fill and flush the sample injector loop between injections also provides a practical limitation to sampling frequency in the present design. The dependence of peak area on caffeine concentration in the sample solution (calibration curve) was measured for caffeine concentrations in the range 1-90 ppm, using an injection volume of 44 p L and a sampling frequency of 2 samples/min. A straight line was obtained whose slope had a relative standard deviation of 0.7% and whose intercept and 95% confidence limits were 81 f 162 integration units. Procyclidine Hydrochloride Assay. Procyclidine hydrochloride is a synthetic antispasmodic drug used in the treatment of parkinsonism. Official methods of assay for procyclidine hydrochloride in tablets include ion-pair extraction of the procyclidinium-bromocresol purple ion pair (38). Ion-pair extraction is, in fact, a sensitive and precise technique for the determination of basic drugs (30-32) which is amenable to use in an extraction/FIA system. In our experience, anionic ion-pair forming reagents such as picrate are more suitable for this purpose than azosulfonate dyes because their ion pairs show less tendency to adsorb at interfaces and onto solid surfaces. The assay of procyclidine hydrochloride in commercial pharmaceutical tablets by picrate ion-pair extraction was chosen as an example of a practical application of the extraction/FIA system. Water was used as the aqueous phase, chloroform as the organic phase, and picric acid of optimal pH and of ionic strength 0.1 as the reagent phase. In defining optimum experimental conditions for the assay, we investigated the following three experimental variables: extraction coil length, pH of the reagent, and picrate concentration in the reagent. A plot of peak area vs. extraction coil length, obtained with all other variables fixed, was very similar in appearance to that presented in Figure 3A for the extraction of caffeine. Coil lengths greater than about 100 cm yielded a constant peak area. As previously, a coil length of 200 cm was used for subsequent studies. Previous investigations (32)suggested that maximum extraction of procyclidinium-picrate could be expected at aqueous phase pH in the vicinity of about 2-4. Measured peak area was found to be constant over the pH range 1.6-3.6. Consequently, a reagent pH of 2.0 was used in subsequent experiments. The optimum picrate concentration was identified by plotting peak area vs. the ratio of picrate in the reagent to procyclidene hydrochloride in the sample solution injected.

ANALYTICAL CHEMISTRY, VOL. 54, NO. 11, SEPTEMBER 1982

The plot is shown in Figure 7. Peak area increases until a ratio of about 100:1, above which the area increases only slightly with increasing ratio. In subsequent studies a picrate reagent concentration olf 2.46 X M was used which represents a picrate/procyclidine hydrochloride ratio of about 500:l for the procyclidine hydrochloride concentration used in the study reported in Figure 7. Much higher picrate concentrations are not possible because of the limited solubility of picric acid. A calibration curve was prepared with procyclidine hydrochloride sample concentrations ranging from 2.5 X 10“ M to 2.0 X M. The slope of the resulting straight line plot of peak area vs. sample concentration had a relative standard deviation of 1.3% and mi intercept and 95% confidence limits of -58 f 216. When the procyclidine hydrochloride content of tablets having a label claim of 5 mg/tablet was assayed by the procedure given in the Experimental Section, an assay value of 4.88 mg/tablet was obtained with a standard deviation of 0.06 mg/tablet based on six replicate injections of each standard and each sample solution. This assay value, which corresponds to 97.6% of label claim, is well within the *lo% tolerance limits allowed by the British Pharmacopeia and is in excellent agreement with the value of 98.2% reported by the manufacturer (39).

ACKNOWLEDGMENT The authors thank Burroughs Wellcome Ltd. for supplying the procyclidine hydrochloride tablets and standard, the Chemistry Department, Machine Shop for making the aluminum cylinders and the membrane phase separator, and the Chemistry Department lZle&ronica Shop for building the timer and power supply for the automatic injection valve.

LITERATURE CITED Ruzlcka, J.; Hansen, E. H. “Flow Injection Analysis”; Wiley: New York, 1981. Stewart, K. K.; Beecher; G. R.; Hare, P. E. Anal. Blochem. 1878, 70, 167- 173. Ruzlcka, J.; Hunsen, E. H. Anal. Chim. Acta 1880, 114, 19-44. Ranger, C. B. Anal. Cliem. 1881, 53,20A-32A. Bergamin, F. Id.; Medelros, J. X.; Rels, 8. F.; Zagatto, E. A. 0. Anal. Chlm. Acta 1978, 101, 9-16, Klnkel, J. F. M.; Tomlinson, E. Int. J . Pharm. 1880, 8 , 261-275. Karlberg B.; Thelander, S. Anal. Chim. Acta 1878, 98. 1-7.

1697

TsuJi, K. J. Chromatogr. 1878, 158, 337-348. Karlberg, B.; Johanrson, P. A.; Thelander, S. Anal. Chlm. Acta 1878, 104, 21-28. Kawase, J.; Yamanaka, M. Analyst (London) 1878, 104, 750-755. Lawrence, J. F.; Brinkman, U. A. Th.; Frel, R. W. J . Chromatogr. 1878, 171, 73-80. Lawrence, J. F.; Brlnkman, U. A. Th.; Frel, R. W. J . Chromatogr. 1878, 185, 473-481. Karger, B. L.; Klrby, D. P.; Vouros, P Foltz, R. L.; HMy, B. Anal. Chem. 1878, 51, 1324-2328. Kllnghoffer, 0.;Ruzicka, J.; Hansen, E. H. Talanta 1880, 2 7 , 169-175. Karlberg, B.; Thelander, S. Anal. Chlm. Acta 1880, 114, 129-136. Johansson. P. A.; Karlberg, B.; Thelander, S. Anal. Chim. Acta 1880, 114, 215-226. Werkhoven-Goewle, C. E.; Brlnkman, U. A. Th.; Frel, R. W. Anal. Chim. Acta 1880, 114, 147-154. Van Buuren, c.; Lawrence, J. F.; Brlnkman, U. A. Th.; Honlgberg, I.L.; Frel, R. W. Anal. Chem. 1880, 52,700-704. Reddlnglus, R. J.; UeJong, G. J.; Brinkman, U. A. Th.; Frei, R. W. J . Chromatogr. 1881, 205, 77-84. Kirby, D. P.; Vouros. P.; Karger. B. L.; Hidy, B.; Peterson, B. J . ChroM t O g r . 1881, 203,139-152. TerwelJ-Groen, C. P.; Kraak, J. C.; Niessen. W. M. A.; Lawrence, J. F.; Werkhoven-Goewle, C. E.; Brlnkman, U. A. Th.; Frei, R. W. Int. J . Envlron. Anal. Chem. 1881, 9 , 45-57. Neubert, P.; Relff, K. Fresenius’ Z . Anal. Chem. 1881, 305, 277-284. Kawase, J.; Nakae, A.; Yamanaka, M. Anal. Chem. 1878, 51. 1640-1643. Nord, L.; Karlberg, B. Anal. Chlm. Acta 1880, 118, 285-292. Kawase, J. Anal. Chem. 1880, 5 2 , 2124-2127. Imasaka, T.; Harada, T.; Ishibashi, N. Anal. Chlm. Acta 1881, 129, 195-203. Nord, L.; Karlberg, B. Anal. Chlm. Acta 1881, 125, 199-202. Klna, K.; Shlraishl, K.; Ishlbashi, N. Talanta 1878, 2 5 , 295-297. Cantwell, F. F.; Mohammed, H. Y. Anal. Chem. 1878, 51, 218-223. Mohammed, H. Y.; Cantwell, F. F. Anal. Chem. 1878, 5 1 , 1006-101 2. Mohammed, H. Y.; Cantwell, F. F. Anal. Chem. 1880, 52, 553-557. Cantwell, F. F.; Carmichael, M. Anal. Chem. 1882, 54. 697-702. Cantwell, F. F.; Carmlchael, M. Can. J . Chem.. In press. Ruzlcka, J.; Hansen, E. H. Anal. Chlm. Acta 1878, 99, 37-76. Sternberg, J. C. ”Advances in Chromatography”; Glddings, J. C., Keller, R. A., Eds.; Marcel Dekker: New York, 1966; Chapter 6. Karger, B. L.; Snyder, L. R.; Horvath, C. “An Introduction to SeparatlOn Science”; Wiley-Intersclence: New York, 1973; Chapter 2. Leo, A.; Hansch. C.; Elkins, D. Chem. Rev. 1871, 71, 525. “National Formulary”, 14th Rev.; Mack Printing Co.: Easton, PA, 1975. Burroughs-Wellcome Co., La Salle, Quebec, personal communication.

RECEIVEDfor review February 3,1982. Accepted May 10,1982. This work was supported by an Alberta Heritage Foundation for Medical Research Studentship to L.F., by the Natural Sciences and Engineering Research Council of Canada, and by the University of Alberta.