922
Anal. Chem. 1985, 57,922-926
Molter, D. Armos. Environ. 1980, 14, 1067-1076. Schone, E. Z . Anal. Chern. 1984, 3 3 , 127. Matsui, H. J . Mefeorol. SOC.Jpn. 1949, 2, 380-381. Kok, G. L.;Darnaii, K. R.: Winer, A. M.; Pitts, J. N., Jr.; Gay, B. W., Jr. Environ. Sci. Technol. 1978, 12, 1077. (6) Kok, G. L. Atmos. Environ. 1980, 14, 656. (7) Yoshizumi, K.; Aokikazuyuki, I. N.; Toshichi, 0.; Toshimi, K.; Shujkamakura; Tajima, M. Atrnos. Environ, 1984, 18,395-401. (8) Zika, R.; Saitzman, E.; Chameides, W. L.; Davis, D. D. J . Geophys. Res. 1982, 87,5015-5017. (9) Heikes, B. G.; Lazrus, A. L.; Kok. G. L.; Kunen, S. M.; Gandrud, B. W.; Gitiin, S. N.; Sperry, P. D. J . Geophys. Res. 1982, 87,3045. ( I O ) Guiibauit, G. G.; Brignac, P.; Juneau, M. Anal. Chem. 1988, 40, 1256. (11) Lind, J.; Kok, G. L., submitted for publication, (12) Richards, L. W.; Anderson, J. A.: Blumenthai, D. L.; McDonaids, J. A,: Kok, G. L.; Lazrus, A. L. Atmos. Environ. 1983, 17, 911. (13) Kok, G. L.; Gitiin, S. N.; Lazrus, A. L., submitted for publication, (14) Rieche, A,; Hitz, F. Ber. Dtsch. Chem. Ges. 1920, 62, 2458. (15) Johnson, R. M.; Siddigu, I . W. “The Determination of Organic Peroxides”; Pergammon Press: London, 1970. (16) Aubar, M.; Taube, H. J . Am. Chem. SOC. 1954, 76,6243-6247. (17) Bhattacharyya, P. K.; Veeraraghavan, R. J . Cbem. Kinetics 1977, 6 0 , 629-640. (2) (3) (4) (5)
RECEIVED for review October 15, 1984. Accepted December 26, 1984. The development of the analytical method was funded by the Electric Power Research Institute under Contract 1630-12. The intercomparison between the peroxidase and luminol methods was funded by the Environmental Protection Agency. Although the research described in this report has been funded in part by the United States Environmental Protection Agency through interagency agreement EPA-AD49F2A182 to the National Science Foundation, it has not been subjected to the agency’s required peer and policy review and therefore does not necessarily reflect the views of the agency, nor does mention of trade names or commercial products constitute endorsement or recommendation for use. The National Center for Atmospheric Research is sponsored by the National Science Foundation.
Determination of Acidity Constants by Solvent ExtractionlFlow Injection Analysis Using a Dual-Membrane Phase Separator Lynette Fossey and Frederick F. Cantwell* Department of Chemistry, University of Alberta, Edmonton, Alberta, Canada T6G 2G2
The absorbances of both the organic and aqueous phases In a solvent extraction/flow Injection analysis system are slmultaneously monitored. Acidity constants are determlned from straight llne plots relating the ratlo of peak areas In the aqueous and organlc phases, A,/A,, to the hydrogen ion actlvlty of the aqueous phase. Theoretical equatlons descrlblng this relatlonshlp for both HA and BH’ charge type acids are derived and verified experimentally uslng 3,5-dimethylphenol (pK, = 10.09 f 0.01) and p-toluldlnlum Ion (pK, = 5.28 f 0.01). The dlstrlbutlon coefficlent of the neutral conjugate species is also obtained durlng the experiment. Some distinct practlcal advantages to using the dual-membrane device over the slngle-membrane devlce are dlscussed.
The use of solvent extraction for determining acid-base dissociation constants (1-7) is particularly attractive for compounds that have low solubility in water and whose two conjugate species have the same absorption spectrum. For such compounds, the low solubility precludes accurate pK, determinations by potentiometric titration in water, and the similarity of spectra precludes the use of the spectrophotometric technique. However, the labor and time involved diminish the attractiveness of batchwise solvent extraction for this purpose. Continuous extraction systems employing rapid phase separation make it possible to perform solvent extraction measurements much more rapidly and conveniently and therefore make pKa determination by solvent extraction more attractive (8). Solvent extraction/flow injection analysis (FIA) employing either one or two porous membranes as phase separators has been shown to give precise, accurate, and very rapid analytical determinations of drugs in pharmaceutical
tablets (9, 10). In the present paper we use a solvent extraction/FIA technique employing two membrane phase separators which allows simultaneous spectrophotometric monitoring of concentration in both the aqueous and organic phases. Equations are derived which relate peak areas in the organic and aqueous phases to hydrogen ion activities in the aqueous phase and which permit the determination of acidity constants of both HA and BH+ charge type acids. Validity of the equations is experimentally demonstrated using 3,5dimethylphenol and p-toluidinium ion as test acids.
THEORY The relationship between sample peak area in the organic phase A,,, and the hydrogen ion concentration in a solvent extraction/FIA system has been derived for a BH’ type acid in earlier papers from this laboratory (9,10). In the present paper, we are concerned with determining a “mixed” acidity constant which incorporates the concentrations of the protonated and deprotonated sample species and the activity of the hydrogen ion. The equation relating peak areas in the organic phase with hydrogen ion activity in the aqueous phase can be written as
Here b is the path length of the spectrophotometer flow cell, f is a response factor which relates the absorbance from the detector to a count rate on the integrator, n is the moles of sample injected, tB,, is the molar absorptivity of the sample in the organic phase, KB is the distribution coefficient of the conjugate base B, K, is the acidity constant of BH+, Faand F, are flow rates of the aqueous and organic phases, respectively, and aH is the hydrogen ion activity in the aqueous phase. This equation is similar to eq 4 in ref 10 if the “system constant” in that paper is defined as K = fbtB.,.
0003-2700/85/0357-0922$01.50/00 1985 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 57, NO. 4, APRIL 1985 923 A similar expression can be derived for sample peak area in the aqueous phase, A,, taking into account both absorbing species, B and BH+
b ’f’n(cB,aKa
A, =
f cBH+,aaH)
+
FaaH
+ F&BKa
(2)
where b’ is the path length of the flow cell in the spectrophotometer used to monitor the aqueous phase, f ’is the response factor for the aqueous phase detector and integrator, and c B , ~and EBH+,~are the molar absorptivities for the neutral and protonated sample species, respectively, in the aqueous phase. The denominators of eq 1 and 2 are the same so that dividing eq 2 by eq 1,yields the following simple expression:
A,
- =-
b If‘cBH+,a
IftcB,a
bf cB,$B
-k aH
(3)
(4) The distribution coefficient of the neutral conjugate species
B can also be calculated by using Ka and the data collected for the organic phase. As discussed previously (IO), eq 1 can be linearized by taking the reciprocal of both sides so that a plot of l / A o vs. aHshould be linear. The distribution coefficient K B can be calculated from the following expression:
(5) where I2and S2represent the intercept and slope, respectively, of the plot of l / A o VS. a ~ . An equation can be derived also for an HA type acid which relates sample peak area in the organic phase, A,, t o the aqueous phase pH, assuming that HA is the only extractable species bfncHA,$HAaH FaaH
+
+ F&HAaH
(6)
where is the molar absorptivity for the neutral form of the sample species in the organic phase and KHA is the distribution coefficient of the neutral species HA. Likewise for the aqueous phase blf‘n(EHA,aaH f EA-,aKa)
A, =
FaaH
+
+ F&HAaH
(7)
where and C A - ,are ~ the molar absorptivities of the protonated and deprotonated sample species, respectively, in the aqueous phase. Dividing eq 7 by eq 6 yields the following:
_Aa
bIf’EHA,a bftHA,aHA
+ -a1H
b’f’cA-,aKa
(8) bfcHA,&HA
A plot of Aa/Ao vs. l / a H should yield a straight line from which can be calculated the acidity constant, Ka
where S3 and I , represent the slope and y intercept, respec-
a F]
w&’
Solenoid -AI,
Figure 1. Diagram of the solvent extraction/FIA instrument used for pK, determinations; see text for details.
tively, of the plot of Aa/Ao vs. l/aH. Finally, to measure K H A , eq 6 can be linearized by taking the reciprocal of both sides, viz.
bfcB,&aKB
When the sample is injected into reagent buffers of various pH, and peak areas are simultaneously measured in the organic and aqueous pahses, a plot of Aa/Ao vs. a H should yield a straight line. I t is then possible to determine the acidity of the constant for BH+ from the slope, SI,and intercept, 11, plot as well as the molar absorptivity ratio of the protonated and deprotonated sample species in the aqueous phase
A, =
1’
-1-- F a +F&HA+ -1 A0
bfnEHA,&HA
FaKa
(10)
a H bfncHA,oKHA
A plot of l / A o vs. l / a H should yield a stright line with slope, S4,and y intercept, 14,from which the distribution coefficient can be calculated using the expression
KHA=
I8aKa
- S4Fa
S4FO
EXPERIMENTAL SECTION Apparatus. A diagram of the solvent extraction/FIA system used in the pK, determination is shown in Figure 1. Its design is similar to that of a previously described instrument (10) with the main difference being that there is only one aqueous solvent. Solvent flows are maintained via constant N2 pressure pumping. The organic phase is cyclohexane which is contained in a glass bottle inside an aluminum pressure cylinder. The six reagent buffers and water are held in seven glass containers inside a multireagent aluminum pressure cylinder described previously (10).
Valve V4 is a six-port rotary valve (part no. R6031V6, Laboratory Data Control) used to select any one of six reagent buffers. Valve V2 is a three-port slider valve (part no. CAV 3031, LDC) which allows selection of either buffer or a water wash. All tubing is made of Teflon, whith 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 i.d. tubing used in the rest of the system. The sample is injected via an automatic sample injection valve V3 (part no. SVA-8031,LDC) into the reagent stream, which is a buffer of known pH. This injection valve contains a “dummy” loop of equal size to the injection loop, so that the flow rate of aqueous reagent is the same in both the load and inject positions. The reagent stream is split at TI into two parallel branches which are reunited at T2to facilitate mixing between the sample and the buffer and to reduce refractive index effects (see Results and Discussion). The aqueous phase joins the cyclohexane stream at tee-fitting T, (part no. CJ-3031, LDC) and the resulting twophase flow passes through the extraction coil, C. The aqueous phase is separated from the organic phase at the end of the extraction coil via a dual-membrane phase separator as described in a previous paper (IO). The extraction coil and phase separator are immersed in a constant temperature bath, shown as dashed lines in Figure 1. The absorbance of the organic phase is monitored with spectrophotometer SI (Spectroflow Monitor 757, Kratos Analytical Instruments) while the absorbance of the aqueous phase is monitored with spectrophotometer S2 (Varichrom photometric detector, Varian). The signals from S1 and S2 are fed to two channels of a digital integrator (VISTA CDS 401, Varian) to obtain peak areas. The signals are also monitored as peaks on recorders (not shown). A peristaltic pump, P (Gilson Instruments, Ville-Belle, France), is used on the outlet lines to ensure accurate flow control (9, 10). The apparatus was modified for determination of molar absorptivity ratios of the protonated and deprotonated sample
924
ANALYTICAL CHEMISTRY, VOL. 57, NO. 4, APRIL 1985
species in the aqueous phase by disconnecting Tz from T3 and disconnecting the membrane phase separator, M, from the aqueous phase detector, S2,and then directly connecting Tz to
SZ.
All pH measurements were made with a glass and calomel electrode pair using an Accumet Model 525 pH meter (Fisher Scientific Co.). Reagents. Water was demineralized, distilled, and finally distilled from alkaline permanganate. Analytical grade cyclohexane was purified by passage through a silica gel column containing a sintered glass frit at the outlet. Reagent buffers of pH 9.60,9.70,9.80, 9.90, 10.00, and 10.11 and ionic strength 0.10 were prepared by adding enough NH4C1to yield a final concentration of 0.10 M and enough concentrated aqueous ammonia to yield the desired pH in a final volume of lo00 mL. Reagent buffers of pH 4.61, 4.78, 5.02, 5.22, 5.42, and 5.63 and ionic strength of 0.10 were prepared by adding enough sodium acetate to yield a final concentration of 0.10 M and enough glacial acetic acid to yield the desired pH in a final volume of 500 mL. 3,5-Dimethylphenol (Aldrich Chemical Co.) and p-toluidine (B.D.H. Chemicals) were both 99+ % pure as reported by the manufacturers. Sample solutions of 3,5-dimethylphenol and p-toluidine contained 0.10 M NaCl. For the determination of molar absorptivity ratios, reagent buffers of pH 2.0, 12.0, and 12.6 were prepared by appropriate dilution of a solution of concentrated HCl or NaOH with HzO until the desired pH was reached. Molar Absorptivity Determination. Molar absorptivity ratios for the protonated and deprotonated sample species in the aqueous phase were measured with the same spectrophotometer and wavelength setting as those used for the acidity constant determination. The molar absorptivity ratio, cHA,,/tA-,, for 3,5-dimethylphenol was determined by preparing two sample solutions, both 3.0 X M in 3,5-dimethylphenol but one adjusted to pH 2.0 with HC1 and the other adjusted to pH 12.6 with NaOH. These samples were injected into an HCl or NaOH reagent of the same pH and peak areas were measured for six replicate injections. Important instrument parameters were as follows: flow rate, 0.7 mL/min; wavelength, 281 nm; volume injected, 44 pL; sampling frequency, one injection per minute; Nz pressure, 20 psig. The flow rates differed slightly (0.01 mL/min) for the two reagents so that peak areas were corrected to a constant flow rate of 0.690 mL/min as previously described (9). The molar absorptivity ratio, ~ g H + , ~ / t gfor , ~ , p-toluidine was determined similarly by using 5.0 X M p-toluidine sample solutions-one adjusted to pH 2.0 with HC1 and the other adjusted to pH 12.0 with NaOH. Changes in the instrumental parameters were as follows: flow rate, 0.5 mL/min; wavelength, 260 nm. Here too a small correction was made to peak areas as a result of slightly different flow rates of the two reagents (9). CalibrationCurves. These were measured with the apparatus shown in Figure 1. Samples were all 0.10 M in NaCl to reduce refractive index effects (see Results and Discussion). Calibration curves for 3,5-dimethylphenol were obtained at a wavelength of 281 nm by injecting samples ranging in concentration from 8 X M into an ionic strength 0.10, pH 9.80 M to 8 X NH,/NH,Cl buffer. Calibration curves for p-toluidine were obtained at a wavelength of 260 nm by injecting samples ranging in concentration from 4 x M to 1.4 X M into an ionic strength 0.10, pH 5.11 acetic acid/sodium acetate buffer. The organic and aqueous phases were monitored simultaneously and peak areas were measured for each sample injected. Acidity Constant Determination. The acidity constant of 3,5-dimethylphenolwas determined by injecting a sample solution that was 4.0 X lo4 M in 3,5-dimethylphenoland 0.10 M in NaCl into reagent buffers of various pH. The reagents were ionic strength 0.10 NH,/NH4C1 buffers that ranged in 0.1 increments of pH from pH 9.6 t o pH 10.1. Additionally, a sample that was 0.10 M in NaCl was injected into each reagent buffer to serve as a blank. The extraction coil was made long enough to ensure that extraction equilibrium was attained, and the extraction coil and phase separator were thermostated to 25.0 i 0.1 OC. Both the organic and aqueous phases were monitored simultaneously, and peak areas were measured for each sample injected. The pH of the aqueous effluent was measured for each buffer used to ensure
that no change in pH occurred during the extraction/FIA procedure. Important instrument parameters for the acidity constant determination of 3,5-dimethylphenolwere as follows: wavelength for both detectors, 281 nm; extraction coil, 200 cm; volume injected, 44 p L ; sampling frequency, one injection per minute; Nz pressure, 40 psig; total cyclohexane flow rate, F,, 2.5 mL/min; total aqueous flow rate, Fa,2.2 mL/min; cyclohexane flow through the membrane, F,,,, 0.9 mL/min; aqueous flow through the membrane, F,,,, 1.1mL/min. The acidity constant of the p-toluidinium ion was determined in a similar manner using a sample solution that was 1.0 x M in p-toluidine and 0.10 M in NaCl. The reagents were ionic strength 0.10 acetic acid/sodium acetate buffers that ranged in 0.2 pH increments from pH 4.6 to pH 5.6. The phase separator and extraction coil were thermostated at 20.0 0.1 OC. Instrument parameters that differed from those listed above for 3,5-dimethylphenol were as follows: wavelength for both detectors, 260 nm; F,, 2.1 mL/min; F,,,, 1.0 mL/min.
RESULTS AND DISCUSSION Two compounds, 3,5-dimethylphenol and p-toluidine, were chosen to test the method for HA and BH+ charge type acids, respectively. Although our discussion will pertain mainly to the use of a dual-membrane phase separator to monitor both the organic and aqueous phases, two single-membrane phase separators each designed as previously (9) and placed in series could be used instead. Experimental Conditions. In the design of the experiment for the determination of the acidity constant of a particular compound, several factors should be taken into account. The first is the choice of the organic solvent. In order to accurately measure peak areas in both the organic and aqueous phases, for reagent pHs in the vicinity of the expected pK, of the sample, it is necessary that the distribution coefficient for the sample be neither too large nor too small. The value of K H A or KB should be between approximately 1 and 10, although this will depend on the molar absorptivities of the sample species in the two phases. The derivation presented in the theory section assumes that dimerization of the sample in the organic phase and ion-pair extraction of the sample with components present in the buffer are negligible. If they are not, the equations must be modified to take them into account. Dimerization of the sample in the organic phase is more likely to be encountered a t higher concentrations and when the sample is polar and the organic phase is nonpolar. This problem may be avoided by keeping the sample concentration low. In the present case a preliminary check for dimerization was made via Beer's law plots for 3,5-dimethylphenol and p-toluidine in cyclohexane using the Cary 118 spectrophotometer over the concentration ranges of interest. For the former compound a plot of absorbance M to 8 X M a t 281 nm vs. concentrations from 4 X was linear with a zero intercept. For the latter compound a plot over the concentration range 4 X M to 1 X M made a t 287 nm was linear with zero intercept. The absence of ion-pair extraction of 3,5-dimethylphenolate (A-) and of p-toluidinium (BH') was checked by seeing if detectable concentrations of these are extracted into cyclohexane from 0.10 M NaCl solutions adjusted to pHs where either A- or BH+ were the only species present in significant amounts. In batch extractions, with absorbances of the cyclohexane phases measured on the Cary 118 spectrophotometer, no detectable 3,5-dinitrophenol was extracted from an aqueous phase a t pH 12.6 and no detectable p-toluidine was extracted from an aqueous phase a t pH 2.0. When mutual solubility of the aqueous and organic solvents is significant, the phases should be preequilibrated before they are used in the solvent extraction/FIA system. Preequilibration of the phases will not affect the value of the acidity constant determined but may affect the value of the measured
ANALYTICAL CHEMISTRY, VOL. 57, NO. 4, APRIL 1985
925
Table I. Acidity Constant Determination by Solvent Extraction/FIA compound
trial no.
3,5-dimethylphenol
1 2 1 2
p-toluidinium
slope (f% RSD)'
4.49 X (f1.3%) 4.49 x 10-10 ( f l . l % ) 4.31 X lo5 (f0.73%) 4.10 X lo5 (10.77%)
y intercept*
PK,e
2.71 f 0.10 2.84 i 0.08 6.17 f 0.08 6.16 f 0.09
10.08 f O.01dc 10.10 f 0.01,c 5.30 f 0.013d 5.27 f O.Olad
'Percent relative standard deviation. *Uncertaintiesare 95% confidence limits. CTemperature= 25.0 f 0.1 "C and ionic strength = 0.10. dTemperature = 20.0 f 0.1 "C and ionic strength = 0.10. eUncertaintiesare one standard deviation. Table 11. Distribution Coefficient Determination by Solvent Extraction/FIA compound
trial no.
slope (i% RSD)n
3,5-dimethylphenol
1 2
4.75 X (f2.1%) 4.60 X (f4.3%) 4.94 x 10-16 ( ~ 2 . 2 % ) 0.700 (f0.47%) 0.648 (i0.63%)
3 p-toluidine
1 2
y interceptb (2.14 f 0.02) X (2.16 f 0.03) X lov5 (2.16 f 0.02) x 10-5 (1.669 f 0.009) X (1.72 f 0.01) x 10-5
Fo, mL/min
Fa,mL/min
2.50
2.18
2.49 2.57
2.22 2.26 2.12 2.12
2.50 2.60
distribution coefficiente 2.4 2.4 2.2 3.2
f 0.2' f 0.2' f 0.2'
f 0.2d 3.3 f 0.2d
"Percent relative standard deviation. buncertainties are 95% confidence limits. CTemperature= 25.0 f 0.1 "C and ionic strength = 0.10. dTemuerature = 20.0 f 0.1 "C and ionic strength = 0.10. eUncertaintiesare one standard deviation. distribution coefficient (11). In the present case the phases were not preequilibrated since the solubility of cyclohexane in water is only 0.006% at 25 "C and the solubility of water in cyclohexane is 0.01% a t 20 "C (12). Salt was added to the sample solutions injected for two reasons. Most importantly, an inert electrolyte is added to ensure constant ionic strength throughout the concentration profile of the sample zone and to match the ionic strength of the sample zone to the surrounding buffer. I t additionally served to reduce refractive index effects, as discussed in the next section. Unlike a spectrophotometric determination of acidity constants, it is not necessary that the molar absorptivities of the protonated and deprotonated. sample species in the aqueous phase be different a t the wavelength chosen. It is, of course, also not necessary to have both spectrophotometers set to the same wavelength. It is desirable to choose the pH of the buffers to be in the vicinity of the inflection point of the peak area vs. reagent pH plot for the sample (10). For distribution coefficients in the suggested range of 1 to 10, this inflection point will occur within about one pH unit of the sample pK,. Refractive Index Peaks. Peaks can occur in flow injection analysis as a consequence of a difference in refractive index between the sample plug injected and the surrounding reagent stream. The effect is characterized by adjacent positive and negative peaks as the sample passes through the flow cell. The refractive index peak is superimposed upon the absorbance peak for the sample and may affect peak heights and shapes, especially for samples of low absorbance. Various authors have noted this effect (13-15) and Betteridge et al. (15) have given a detailed analysis of the phenomenon as it pertains to flow injection analysis. In the present study refractive index peaks occurred in the aqueous phase. It was found, in general that the area of the positive portion of the refractive index peak is equal to the area of the negative portion. Thus, there is no net contribution to the peak area for the absorbance peak of the sample. If, however, the sample absorbance peak is small compared to the refractive index peak, the observed peak shape will be distorted, preventing proper integration by an electronic integrator. This imposes a limit on sample detection. One way to get around this problem is to match the refractive index of the sample and reagent carrier stream. In the present case, the ionic strength of the sample solution was matched to that of the reagent stream by adding salt (NaCl)
to the sample. While this did not completely match the refractive indexes, it brought them much closer together and greatly reduced the size of the refractive index peaks. The size and shape of the refractive index peaks were routinely monitored by injecting 0.10 M NaCl into each reagent, in place of the sample. Their small size and the fact that the positive and negative portions were of equal area resulted in no problems in measuring sample peak areas for the sample concentrations employed. Molar Absorptivities. Accurate ratios for the molar absorptivities of the two conjugate species in the aqueous phase are required in eq 4 and 9. These were measured by a flowinjection technique, without solvent extraction, as described earlier, in which the ratio of peak areas obtained at low pH and a t high pH is equal to the ratio of molar absorptivities. For 3,5-dimethylphenol tm,,/t~-,, at 281 nm was 0.500 f 0.005, while for p-toluidine EBH+,~/CB,,a t 260 nm was 0.353 f 0.006. Calibration Curves. Equations 1 through 11 assume a linear relationship between integrator signal and concentration. This was checked for 3,5-dimethylphenol and p-toluidine for both the organic and aqueous phases using the solvent extraction/FIA system. The plot of peak area for the organic phase vs. concentration of 3,5-dimethylphenol injected was linear with relative standard deviation (RSD) for the slope of 0.28%. The y intercept and its 95% confidence limits were 64.0 f 64.8 in arbitrary integration units. A similar plot for the aqueous phase was linear also with a RSD for the slope of 0.46%. T h e y intercept and its 95% confidence limits were -3038.0 f 4218.6. The plot of peak area for the organic phase vs. concentration of p-toluidine injected was linear with a RSD for the slope of 0.74%. T h e y intercept and its 95% confidence limits were 9.4 f 881.4 in arbitrary integration units. The corresponding plot for the aqueous phase was a straight line with a RSD for the slope of 1.2%. The y intercept and its 95% confidence limits were 1556.8 f 6631.5. The zero y intercepts of the calibration plots for both compounds in the aqueous phase shows that the underlying refractive index peaks do not have a net effect on sample peak areas. Acidity Constants. The acidity constant for 3,5-dimethylphenol was determined from plots of A , / A , vs. l/aH. The experiment was run twice and peak areas obtained were based on an average of six replicate injections of sample into each buffer. The slope and y intercept values of the plots along with the calculated pKa's and computed errors are re-
926
ANALYTICAL CHEMISTRY, VOL. 57, NO. 4, APRIL 1985
ported in Table I. The linearity of the plots is evidence for the validity of the equations for HA charge type acids. The average value of the pK, for 3,5-dimethylphenol at an ionic strength of 0.10 and temperature of 25 "C was 10.09 f 0.Ol4. The stated uncertainty is one standard deviation and includes the computed error in determining the slope, y intercept, and molar absorptivity ratio as well as an estimated error due to the calibration of the pH meter used to measure the pHs of the reagent buffer solutions. Literature values for the acidity constant of 3,5-dimethylphenol a t 25.0 "C determined spectrophotometrically and corrected for activity coefficient effects to zero ionic strength are 10.20 (16) and 10.19 (17). For comparison purposes if we correct our pKa value to zero ionic strength by calculating the activity coefficient via the Davies equation (18),we obtain a pKa0 value of 10.20. The acidity constant for the p-toluidinium ion was determined from duplicate plots of Aa/A, vs. aH. The results are reported in Table I. The linearity of the plots is proof of the validity of the equations for BH+ charge type acids. The average value of the pKa for the p-toluidinium ion at an ionic strength of 0.10 and a temperature of 20.0 "C was 5.28 f 0.Ol3. Literature values determined at 20 "C and an ionic strength of 0.1 are 5.44 (19), 5.21 (20),and 5.159 (21). Since the literature values show considerable variation, we can only say that our value is well within the range of reported values. Distribution Coefficients. The distribution coefficients for B and HA between cyclohexane and the aqueous buffers can be calculated from eq 5 and 11,respectively. However, since the experiment was optimized for determination of acidity constants, the pH region examined is not the best for accurate measurement of KB and KHk The slopes and y intercepts of the plots, the total flow rates of the aqueous and organic phases, and the calculated distribution coefficients and computed errors are reported in Table I1 for replicate trials. The uncertainties are one standard deviation and they include the error in the slope and y intercept, in the acidity constant, in the total flow rates of the organic and aqueous phases, and in the calibration of the pH meter. Comments. Although this discussion has dealt with the determination of acidity constants using data collected from both the organic and aqueous phases, i t is possible to determine them by monitoring peak areas in only the organic phase. Equation 10 indicates that for an HA acid, a plot of l/Ao vs. l/aHshould give a straight line. If the system constant, eHA,,bf, is measured in a separate experiment and accurate values are obtained for Fa, F,, and n, a value for KHA can be calculated from the intercept and K, can be calculated from the slope of that plot. The system constant is determined by using a single phase FIA system by injecting a sample of known concentration (dissolved in the organic phase) into a stream of the same organic phase. The injection valve is placed close to the detector of interest to minimize band broadening, and peak areas are measured for each sample injected. It is of
course important to use the same detector and integrator settings as for the acidity constant determination. The system constant is determined from the relation A = nK/Fo (9). Similarly for a B base, eq 1can be rearranged to describe a linear relationship between l/Ao and aH(10). The product KBK, can be calculated from the intercept and K, from the slope of that plot, if accurate values are measured for tB,,bf, Fa,F,, and n. The disadvantage of the two-membrane method described in this paper, compared to the one membrane method is that two spectrophotometric detectors are necessary in the former. However there are some distinct advantages to measuring peak areas in both phases. It is not necessary to know the flow rates, the number of moles of sample injected, the extraction constant, or the system constant. It is also not necessary that the phases be preequilibrated or that the flow rates be constant from one injection to the next. These conclusions become evident upon examination of eq 3, 4, 8, and 9 in which none of the above appears, either explicitly or implicitly. The net effect is that very accurate values for acidity constants can be obtained. Work is presently under way to allow both the aqueous and organic phases to be monitored using the reference and sample channels of a single spectrophotometric detector.
LITERATURE CITED Sekine, T.; Hasegawa, Y.; Ihara, N. J . Inorg. Nucl. Chem. 1973, 3 5 , 3968-3970. Golumbic, C.; Goldbach, G. J . A m . Chem. SOC. 1951, 7 3 , 3966-3967. Golumbic, C.; Orchin, M.; Weller, S. J . Am. Chem. SOC. 1949, 7 1 , 2624-2627. Golumblc, C.; Orchin, M. J . Am. Chem. SOC.1950, 72, 4145-4147. Ezumi, K . ; Kubota, T. Chem. Pharm. Bull. 1980, 28, 85-91. Kubota, T.; Ezumi, K. Chem. Pharm. Bull. 1980, 28, 3673-3678. Xie, T. M.; Dryssen, D. Anal. Chim. Acta 1984, 160, 21-30. Kinkel, J. F. M.; Tomllnson, E. Int. J . Pharm. 1980, 6 , 261-275. Fossey, L.; Cantwell, F. F. Anal. Chem. 1982, 5 4 , 1693-1697. Fossey, L.; Cantwell, F. F. Anal. Chem. 1983, 5 5 , 1882-1885. Carmicheal, M.; Cantwell, F. F. Can. J . Chem. 1982, 60, 1286-1290. High-Purity Solvent Guide"; Burdlck & Jackson Laboratories Inc.: Muskegon, MI, 1980. Ham, G. Anal. Proc. 1981, 18, 69-70. Anderson, L. Anal. Chim. Acta 1979, 110, 123-128. Betterldge, D.; Dagless, E. L.; Fields, B.; Graves, N. F. Ana/@ (London) 1978, 103, 897-908. Chen, D. T. Y.; Laldler, K. J. Trans. Faraday SOC. 1962, 5 8 , 480-489. Herington, E. F. G.; Kynaston, W. Trans. Faraday SOC. 1957, 53, 138-1 42. Davies, C. W. J . Chem. SOC. 1938, 2093-2092. Bolton, P. D.; Hall, F. M. Aust. J . Chem. 1967, 2 0 , 1797-1804. Willl, A. V.; Meir, H. Helv. Chim. Acta 1956, 3 9 , 318-322. Bernasconi, C.;Koch, W.; Zolllnger, Hch. Helv. Chim. Acta 1963, 4 6 , 1 184-1 190.
RECEIVED for review October 19, 1984. Accepted December 19, 1984. 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.
