Anal. Chem. 1906, 58, 2727-2731
2727
Simultaneous Determination of Phenylephrine Hydrochloride and Pheniramine Maleate in Nasal Spray by Solvent Extraction-Flow Injection Analysis Using Two Porous-Membrane Phase Separators and One Photometric Detector Charles A. Lucy and Frederick F. Cantwell* Department of Chemistry, University of Alberta, Edmonton, Alberta, Canada T6G 2G2
Interferlng specles, thimerosal, maleate, and benralkonlum Ion, present In nasal spray are removed from the Injected sample uslng miniature on-line Ion exchange columns. After solvent extraction vla segmented flow, a porous Teflon membrane separates a portlon of the chloroform phase and directs It to the sample flow cell of the photometer. A paper membrane located farther downstream In the segmented flow separates a portion of the aqueous phase and dlrects It to the reference flow cell. The detector electronics are modlfled to allow reasslgnment of the sampWreferencedeslgnatkm of the flow cells, so that posttlve signals can be obialned from both. At pH 13 phenlramlne Is quantltatlvely extracted Into the chloroform phase and phenylephrine remains In the aqueous phase. Injectlons are made at a rate of one every 2 mln. Preclslon and accuracy are 1-2%. Statlstlcal moment analysis Is used to quanttfy the contrlbutlons of the Instrument components to the overall band broadenlng observed.
In recent years analytical solvent extraction has been automated by the use of two-phase segmented flow through narrow tubing. This technique, generally referred to as solvent extraction-flow injection analysis (FIA), has been used for sample preconcentration ( 1 , 2 ) ,for matrix removal (3-5), and as a reactor for postcolumn detection (6). In most cases the species of interest is extracted from the aqueous segments into the organic segments and a portion of the organic phase is separated from the two-phase flow using a phase separator (7-10). The concentration of the sample in this separated organic stream is then monitored by using UV-vis absorption (IO),fluorescence (II),or atomic spectroscopy (3,12,13).The aqueous phase can also be separated from the segmented flow using a hydrophilic membrane. Simultaneous monitoring of both the aqueous and organic phases has been used to analyze diphenhydramine and 8-chlorotheophylline in Dramamine tablets (14)and for the determination of acidity constants (15). In both of these applications two detectors were used, one to monitor the absorbance of the aqueous phase and the other to monitor the organic phase. The use of two detectors increases the cost of such an instrument. It would therefore be advantageous if only a single detector could be used to monitor the two sample streams. Several approaches have been used to make dual measurements with one detector in single phase FIA (16,17). All of the methods have in common the incorporation of a time delay between the two measurements and differ only in the manner in which the two sample streams are presented to the detector. In one method (16) the sample is passed successively through the sample and reference flow cells of the photometer, producing positive and negative peaks. This approach is not
compatible with many modern spectrophotometers and liquid chromatography detectors, which utilize an automatic gain control. In this paper, an extraction-FIA system is presented in which both the organic and aqueous phases are monitored by a single photometric detector. A portion of the organic phase is first separated from the segmented flow by a porous Teflon membrane and passed through the sample flow cell of the detector. Farther downstream a portion of the aqueous phase is separated from the segmented flow by a paper membrane and directed to the reference flow cell. Modification of the detector electronics was necessary to obtain analytically useful signals from both the reference and sample flow cells. With this arrangement any overlap of the peaks from the two flow cells would cause systematic errors because each flow cell acts as the other’s reference. To maximize the frequency with which samples can be injected, it is necessary to minimize the broadening of the sample peaks. Therefore, the contributions of the various components of the instrument to band broadening were measured in order to identify the components that most need improvement. The instrument was used for the assay of phenylephrine hydrochloride and pheniramine maleate, the active components of a commercial nasal spray. Under alkaline conditions pheniramine is quantitatively extracted into chloroform as the neutral free base, while phenylephrine remains unextracted as the anionic phenolate. In order to avoid both spectral and chemical interferences from minor components of the nasal spray, it was necessary to include miniature on-line ion exchange columns in the instrument manifold.
EXPERIMENTAL SECTION Apparatus. A schematic diagram of the extraction-FIA instrument used to analyze a commercial nasal spray is shown in Figure 1. Its design is based on an instrument previously described ( 1 4 ) with the main differences being the sequential arrangement of the phase separators and the use of only one detector. Constant nitrogen pressure is used to produce solvent flow, which can be shut off using valve VI (part no. CAV2031, Laboratory Data Control (LDC), Riviera Beach, FL). Valve Vz (part no. R6031 V6, LDC) is a six-port rotary valve that allows selection of any of a number of solutions contained in the multireagent cylinder (14). In experiments in which the pH was varied, the multireagent cylinder contained six buffer solutions and distilled water, while for the nasal spray analysis it contained only distilled water and 0.1 M hydrochloric acid. All tubing used in the instrument is made of Teflon, with 0.3 mm i.d. tubing being used whenever it is desirable to minimize sample band broadening or to provide increased resistance to flow and 0.8 mm i.d. tubing used in the rest of the system. The sample is injected into a carrier stream of distilled water via a 10-pL slider injection valve, V4 (part no. CSVA-10, LDC). Valve V4 is actuated by an air solenoid valve (part no. SOL-324-VDC, LDC) controlled by an electrical timer that allows
0003-2700/86/0358-2727$01.50/0 0 1986 American Chemical Society
2728
ANALYTICAL CHEMISTRY, VOL. 58. NO. 13. NOVEMBER 1986
Phenimmine
I
1.0
'
m0 c m
f! $
mure 1. Ciagram of exbactbn-FIA l n s m t used fa nasal spray analysis. See ted for details.
variation of fill and injection times. The sample stream flows through a miniature anion exchange column, C, b a d on a design previously described (18). This column consists of a 10 mm long resin bed of 100-140 mesh Amberlyst A-26 macroporous anion exchange resin packed in 1.5 mm i.d. Teflon tubing. The sample stream then merges with the reagent stream, 0.2 M NaOH, at tee-fitting T, and flows through a miniature cation exchange column, C,, consisting of a 10 mm long bed of 100-140 mesh Amberlyst 15 macroporous cation exchange resin. The combined aqueous stream joins the chloroform stream a t the phase segmentor, tee-fitting Tz, and the resulting segmented flow stream passes through the extraction coil, E, in which the solvent extraction occurs between the aqueous and the organic phases and equilibrium is achieved. Various lengths of 0.8 mm i.d. Teflon tubing were used as the extraction coil depending on the experiment. After exiting coil E, the stream enters the first membrane phase separator, M., where a portion of the organic phase is separated. The porous membrane phase separator is similar to one previously dexribd (10). It contains a hydrophobic membrane consisting of two pieces of 4-mil, 10-20 pm pore size Zitex Teflon membrane (no. E249-122, Chemplast, Inc., Wayne, NJ). The portion of the chloroform phase that passes through this membrane flows through the 8-pL sample flow cell of the UV-vis absorbance detector (SF770 Spectroflow, Schoeffel Instrument Corp.). The remainder of the chloroform and all of the aqueous phase pass through a delay coil, D, to the second phase separator, Ma, which is similar to M. except that instead of a hydrophobic membrane it contains a hydrophilic membrane consisting of two pieces of Whatman No. 5 filter paper. This membrane separates a portion of the aqueous phase and passes it through the 8-pL reference flow cell of the detector. The delay coil, D, is 0.8 mm i.d. Teflon tubing of a sufficient length to separate the signals from the two flow cell channels of the detector, as can be seen in Figure 2. The signals from the two channelsare integrated by use of a digital integrator, I (Model 3390A, Hewlett-Packard),to obtain peak areas and heights. The chloroform exiting the sample flow cell, the aqueous solution exiting the reference flow cell, and the combined CHCl,/aqneous stream leaving the second phase separator all pass through Acidflex pump tubes (Technicon Corp.) in a Minipuls variable speed peristaltic pump, P (Gilson Instruments, Ville-le-Belle, France), which is used to provide accurate flow control (10). Acidflex tubes used were as follows: organic filtrate, red-red; aqueous filtrate, white-white; two phase flow, purple-purple. At the time of start up the need to wet each membrane with the appropriate solvent requires the inclusion of some additional plumbing. The system is initially flnshed with methanol from the 'rinse" cylinder. Then the three-port valve V, (part no. CAV3031, LDC) and the four-port valve V, (part no. CAV4031, LDC) are switched to their start up positions and the chloroform and distilled water streams are turned on. The chloroform flow passes through the extraction coil E and wets the Teflon membrane in phase separator M.but is diverted to waste by valve V, before reaching phase separator Ma. The water stream from the multireagent cylinder bypasses the extraction and delay coils and is directed by V, into M. where it wets the paper membrane. Next the reagent flow is initiated by opening the appropriate valve V,. The reagent establishes the segmented flow in the extraction coil. After the segmented flow has reached valve V, switching of valves
z
0.5
0 0
1
2
Time (min) Flgure 2. Instrument response for a nasal spray injection made a1 lime zero. Base line shm is dw, to Switching the f!aw cell dealgnation.
V, and V, restores the instrument to its normal operating mode. The three-port valve V, is included in the system to allow the flushingof air bubbles, which may form in the solvent lines when the instrument is idle. The shut-down procedure a t the end of the day consists of a 5-min flush with 0.1 M HC1 from the mnltireagent cylinder to regenerate the anion exchange column, followed by a 5-min methanol rinse of the full instrument. Detector Modification. The photometer (Schoeffel SF770) was modified to allow electronic switching of the sample/reference designation of its two flow cells. This was done by placing a double-poledoublethrow relay (part no. RA30572051, Elec-Trol) across the two control lines from the chopper to the synchronous demodulator. These lines were used to define the signal from the photomultiplier as coming from either the sample or the reference. The setting of this relay, either for normal or reversed flow cell designation,was controlled with an electronic timer. For the nasal spray analysis, the timer switched the flow cell designation every 60 8. This switching can be identified by the base line shift (Figure 2), which occurs due to small differences in the optical alignment of the two flow cells Further information about the detector modification can be obtained from the authors. The small base line shift that accompanies the flow cell redesignation would be mistaken as a portion of a peak by the integrator. To avoid this problem the integrator is disabled (integration function 9, HP 3390A3 just prior to the shift and reenabled afterward (function -9). In addition all peaks are designated *solvent" peaks (function 3) so that the integration is not terminated prematurely on the tail of the skewed peaks. Reagents. Water was demineralized, distilled, and finally distilled over alkaline permanganate. Reagent grade chloroform (Caledon Laboratories, Ltd.) was distilled and washed with distilled water shortly before use. Reagent grade methanol (Caledon Laboratories,Ltd.) was used as received. Reagent buffers between pH 3.8 and 7.6 were prepared by combining the appropriate volumes of 0.2 M Na2HP0, and 0.1 M citric acid; pH 8.0 and 9.1 buffers were prepared from 0.1 M H,BO, and 0.1 M NaOH pH 9.5 and 10.1 buffers were prepared from 0.025 M Na,B,O, and 0.1 M NaOH, pH 11.3buffer was prepared from 0.05 M NazHPO, and 0.1 M NaOH; buffers with pH >12 were prepared from 0.2 M KCI and 0.2 M NaOH. Amberlyst 15 strong acid cation exchange resin (Rohm and Haas Co.),with an exchange capacity of 4.6 mequiv/g, was ground and sieved. The 1Mt140 mesh portion was slurried in water and the fines were removed by decanting. The resin was then washed in turn with HCI, NaOH, ethanol, and water and air-dried. The Amherlyst A-26 anion exchange resin (Mallinckrodt), with an exchange capacity of 4.2 mequiv/g, was treated in a similar manner.
ANALYTICAL CHEMISTRY, VOL. 58, NO. 13, NOVEMBER 1986
Concentrations of pheniramine maleate and phenylephrine hydrochloride in the standard solutions were calculated by using the assay values supplied by the manufacturer. Evaluation of Band Broadening. The variance, s2, of a peak produced by the detector was taken as a measure of band (peak) broadening. Peak variance, in units of seconds squared, was calculated as the second statistical moment of the peak using the following expression (19, 20):
[
s2 = 1/A
L-(t- tJ2Ci dt]
where A is the peak area, calculated as the nonnormalized zeroth statistical moment, t, is the center of gravity of the peak, calculated as the first statistical moment, and Ci is the concentration (i.e., detector signal) at time t. The integral in eq 1 was evaluated numerically using Simpson’s rule (21) with dt taken as the time interval between successive data points (0.05 8 ) . The base line was estimated by performing linear regression analysis on the data points before and after the peak. The calculationswere performed on an IBM-XT microcomputer using a moment analysis program written in Asyst (Macmillan Software Co.). Data were acquired digitally from the photometric detector using a Lab Master ADC interface board (TM-40-PGL, Tecmar, Cleveland, OH) on the IBM-XT. The accuracy of the zeroth through fourth statistical moments calculated by the program was verified by using Gaussian, Exponential, and Poisson (one and two degrees of freedom) distributions as simulated peaks. The accuracy of the second moment values was further verified by comparison with values calculated by using the Sternberg graphical method (22). The contributions of the individual instrument components to the overall observed peak variance were measured by using configurations of the instrument designed to isolate their individual contributions. The flow rates employed in these studies were approximately the same as those used in the nasal spray analysis described below. Experimental arrangements for measuring the variances were as follows (see Figure 1): sD2(variance due to injector, detector, and associated tubing). Aqueous. The tube from V4 was disconnected from C, and the tube to the detector was disconnected from M,. These two tubes were connected together and aqueous solutions of phenylephrine hydrochloride and pheniramine maleate were injected into the aqueous flow stream. The peak variance observed was defined as sID2(aq). Organic. The instrument configuration used was identical with that used for the determination of sD2(aq),except that flow came from the chloroform cylinder and samples of pheniramine maleate in chloroform were injected into the organic flow stream. The peak variance observed was defined as sD2(org). sc,2 (variance due to cation exchange column): The tube from C, was disconnected from Tzand the tube from the detector was disconnected from M,. C, and the detector tubing were connected together. C, was removed from the system. Aqueous flow was from both the reagent and the multireagent cylinders. S C , ~was calculated as the difference between the observed variance and sD2(aq). The contribution of tee-fitting TI to the peak variance was found to be negligible. sc: (variance due to anion exchange column). The tube from C, was disconnected from TI and the tube from the detector was disconnected from M,. TI and the detector tubing were connected and sc: was determined as the observed variance minus sD2(aq). sp2(variance due to the phase separator and segmentor T2). Aqueous. The tube from V4 was disconnected from C, and was connected to T2in place of the tube from C,. The tube from T2 was disconnected from Vs and the tube from v6 was disconnected from Ma. The tube from T2was connected to Ma. The aqueous phenylephrine hydrochloride sample was injected into 0.1 M NaOH from the multireagent cylinder. The aqueous stream merged with the chloroform stream at Tz. sP2(aq)is the difference between the observed peak variance and sD2(aq). Organic. The same tubing arrangement as that of sp,2(aq)was used. Ma was replaced with M,, and the chloroform and multireagent cylinders traded places. A chloroform solution of pheniramine maleate was injected into the chloroform stream. sp2(org)was calculated as the difference between the observed peak variance and srp2(org). s: (variance due to the extraction coil). The extraction coil E (200 cm X 0.8 mm i.d. Teflon tubing) was inserted between T2
2729
and the membrane phase separator in the arrangements described above for determining sp:. s: for both the aqueous and organic phase samples was calculated as the observed peak variance minus the peak variance observed in the sp: studies. stotalZ(measured) (total peak variance) is the variance observed upon injecting samples into the instrument in the configuration shown in Figure 1. sd2 (variancedue to delay coil and M, for sample in the aqueous phase). The tube from v6 was disconnected from Ma and E was disconnected from M,. E was connected to Ma. sd2 is the difference between the observed variance and the stow2(measured) for phenylephrine hydrochloride. Nasal Spray Assay. Nasal spray samples were injected into the instrument with no prior treatment. Important instrument parameters for the assay were as follows: total chloroform flow rate, 2.8 mL/total aqueous flow rate, 2.8 mL/miq chloroform flow rate through membrane M,, 2.1 mL/min; aqueous flow rate through membrane Ma, 1.1 mL/min; extraction coil length, 200 cm; delay coil length, 700 cm; volume injected, 10 pL; sampling frequency, 1injection/2 min; wavelength, 25& nm; detector cycle, 60 s in the normal flow cell designation followed by 60 s in reverse designation, started at injection; nitrogen pressure, 45 psig; reagent, 0.2 M NaOH. A warmup time of about 30 min is used after solvent flow is established. Two standards were used, which contained pheniramine maleate and phenylephrine hydrochloride concentrations that were 20% above and 20% below the label-claim values in order to “bracket” the expected sample concentrations. The standards contained the same concentrations of benzalkonium chloride and thimerosal as are present in the nasal spray and were made isotonic using NaC1. R E S U L T S AND DISCUSSION The electronics of a dual beam photometric detector were modified so that positive signals could be obtained from samples passing through both its sample and reference flow cells. The instrument shown in Figure 1 was used to characterize the extraction behavior of phenylephrine hydrochloride and pheniramine maleate in terms of pH and other instrument parameters and to determine these two compounds in nasal spray. In addition the band broadening in the instrument was characterized by using statistical moment analysis. Detector Modification. The Schoeffel SF770, like many modern HPLC photometric detectors, utilizes an automatic gain control to maintain the power of the light source a t a constant level over the entire spectral range (23). In such an instrument the dc signal from the reference channel is compared with a fixed voltage and any change in this signal activates the automatic gain control which alters the photomultiplier tube (PMT) sensitivity in the proper direction to restore the reference signal to its original level. Thus, if an absorbing species passes through the reference flow cell, the absorption will be met by an increase in the P M T gain resulting in a truncated negative peak. For this reason, absorbances in the reference flow cell of the unmodified detector were linear with concentration only below 0.25 AU. In order to obtain signals from both of the flow cells which could be measured with a conventional integrator, the electronics of the detector had to be modified to allow electronic switching of the sample/reference designation of the flow cells. In a chopped dual-beam spectrophotometer the incident light passing through the sample and reference channels is mechanically modulated by a chopper and focused onto a single PMT. The resulting output from the P M T is a pulsating electrical signal. Conversion of this signal into absorbance or transmittance information requires the identification of each pulse as coming from either the sample or reference flow cell. Demodulation of these pulses is accomplished using two clocking signals from the chopper drive. Interchanging these clocking signals reverses the portions of the chopper cycle identified with the sample and reference flow cell, resulting
2730
ANALYTICAL CHEMISTRY, VOL. 58, NO. 13, NOVEMBER 1986
Table I. Determination of Active Ingredients in a Commercial Nasal Spray Based on the Use of Peak Area and Peak Height phenylephrine hydrochloride"~c (w/w%)
pheniramine maleatebVc(w/w70) height mfgr assayd
sample
area
height
mfgr assayd
area
lot A lot B
0.477 f 0.007 0.483 f 0.005 0.481 i 0.007
0.472 h 0.002 0.481 f 0.008 0.474 f 0.002
0.48 0.48 0.48
0.198 f 0.001 0.198 f 0.002 0.201 i 0.001
lot c
0.197 f 0.001 0.199 f 0.004 0.200 i 0.001
0.20 0.19 0.20
"Label claim for phenylephrine hydrochloride is 0.50% (w/w). bLabel claim for pheniramine maleate is 0.20% (w/w). standard deviations based on six replicates. Manufacturer's assay values. in a reversal of the flow cell designation, without otherwise affecting the detector performance. If a spectrophotometer without automatic gain control were used, analytically useful signals would be obtained from both flow cells. Thus it would only be necessary to include a timer-controlled inverter in the detector output line in order to produce positive peaks from both flow cells that could be measured using a conventional integrator. The only additional requirement imposed by the modified detector is the need to produce a time difference between the arrival of pheniramine and of phenylephrine at the detector. This time delay was produced by arranging the phase separators in series with a long length of tubing (delay coil, D) between them (Figure 1). Assay Conditions. The active components of the nasal spray are phenylephrine hydrochloride and pheniramine maleate. The pH of the aqueous phase was varied between pH 3.9 and 9.1 in order to study its effect on the extraction of these components. For pheniramine, a plot of the organic phase peak area vs. pH showed the sigmoidal shape (14) which would be predicted for a compound with a pK, of 9.3 and a distribution coefficient of lo3.*(24). Under the same conditions phenylephrine did not extract (log KB =
