Anal. Chem. 1980, 52, 553-557
553
Two-Phase Photometric Ion-Pair Titrations of Drugs and Surfactants Hussain
Y. Mohammed
and Frederick F. Cantwell'
Department of Chemistry, University of Alberta, Edmonton, Alberta, Canada T6G 2G2
Two-phase ion-pair titrations are performed on cationic and anionic samples in chloroform/aqueous buffer mixtures by continuously measuring the absorbance of the organic phase while the titrant picrate or methylene blue is added. Titration curve shapes are quantitatively described by theoretical titration equations. Precision and accuracy are better than lYO. The relative merits of monitoring the organic or aqueous phase are discussed.
I n a previous paper ( I ) , it was shown t h a t a recently described titration apparatus (2) can be used t o continuously monitor t h e absorbance of t h e aqueous phase during t h e course of a two-phase photometric ion-pair titration. T h e resulting titration curves a r e quantitatively described by a theoretical titration equation. T h e utility of t h e method was demonstrated by titrating several cationic drugs with picrate ion. T h i s approach holds considerable promise as a general method for t h e volumetric determination of ionic species. An obvious outgrowth of t h e previous work is t h e use of photometric titrations for the analysis of ionic detergents, since two-phase ion-pair titrations employing indicators for visual end-point detection are already a well established technique in t h a t field (3-5). Also of value would be a theory to predict titration curve shapes and optimum titration conditions when t h e absorbance of t h e organic phase is monitored. In t h e present work, two-phase photometric ion-pair titrations a r e extended to situations in which t h e absorbance of the organic phase is measured. T h e filter probe is modified t o pass only t h e organic phase, and theoretical titration equations are derived which take into account several common side reactions. T h e validity of t h e approach is demonstrated by titrating anionic and cationic surfactants and a cationic drug compound with titrants such as methylene blue and picrate. T h e relative merits of monitoring t h e aqueous and organic phase absorbances are discussed. Earlier studies relevant t o this work have already been summarized ( I ) .
EXPERIMENTAL Apparatus. The titration apparatus previously described ( 2 ) was modified as follows. In order to selectively pump the immiscible organic phase through the spectrophotometer flow cell, the triple layer of filter paper on the filter probe was replaced with a double layer of porous Teflon membrane (Zitex Filter Membrane, 10-20 pm pore size, 4 mils thick, No E249-122, Chemplast Inc., Wayne, N.J.). Because the membrane was easily perforated by the sharp edges of the coarse Teflon mesh preLiously used to support the filter paper, the mesh was replaced with a 0.8-mm thick wafer of Teflon which was cut to a diameter equal to the bottom of the filter probe and was extensively perforated and lapped smooth. Thus constructed, and with due caution against striking the membrane surface of the probe against sharp objects, the filter probe has been used for over a hundred titrations without the need to change the Teflon membrane. In some of the titrations the Mini Micro 2/6 peristaltic pump described earlier ( 2 )was replaced with a variable speed Minipuls-2 peristaltic pump (Gilson France S.A., Villiers-le-Bel, France) and the 7-cm long, 1.65-mm i.d. Acidflex pump tube was replaced with an 18-cm long, 1.14-mm i.d. tube. A porous Teflon filter was placed between the pump and flow cell to trap the occasional 0003-2700/80/0352-0553$01 O O / O
particle of rubber abraded off the inside of the pump tube. This was constructed by sandwiching a small circle of 28-mil thick 3G50 Fm pore size Zitex Teflon filter membrane (No. K1064-222) between two small Teflon rings which were, in turn, held between the flared ends of the Teflon tubing in a Cheminert tube end fitting (No. T E F 107 and 107A3, Laboratory Data Control, Riviera Beach, Fla.). As a final modification, a 10-ft length of 0.3-mm i.d. Teflon tubing was used as a return line to connect the spectrophotometer outlet back to the titration vessel in order to suppress bubble formation in the organic phase a t high pumping rates. Reagents. The phenothiazine derivative promethazine hydrochloride was from the same batch used previously ( I ) . The surfactants benzethonium chloride (K + K, Plainview, N.Y.), hexadecyltrimethylammonium bromide (Analyzed Reagent, J. T. Baker Chemical Co., Phillipsburg, N.J.), sodium lauryl sulfate (U.S.P. Equivalent, Fisher Scientific Co.), and sodium dodecylbenzene sulfonate (K + K, Plainview, N.Y.) were used as received. The first two were assayed by the met.hod of the New and Nonofficial Remedies ( 6 ) ,which involves precipitation of the ferricyanide salt of the quaternary ammonium ion, filtration, and iodometric determination of excess ferricyanide. The two anionic surfactants were assayed by two-phase ion-pair titration to a visual end point with methylene blue as indicator, according to the method of Blank (7) using benzethonium chloride as titrant in place of cetylpyridinium bromide (8). Potassium ferricyanide and other reagents required for the surfact.ant assays were analytical reagent grade. Picrate titrant (0.039 M) was prepared and standardized as previously described ( I ) . Methylene blue titrant was prepared by dissolving methylene blue (U.S.P. Equivalent, Fisher Scientific Co.) in 0.0'75 M phosphate buffer (pH '7.5), repetitively extracting this solution with chloroform until the chloroform phase was colorless, and diluting to volume with phosphate buffer. The chloroform extractions are required to remove the methylene azures and dimethylthionoline, oxidation products of methylene blue which are present in all commercial samples (9). The methylene blue titrant solution was standardized by titrating, in triplicate, aliquots of the picric acid titrant in pH '7.50 buffer using the present ion-pair photometric method. Sharp titration curves were obtained. The average and average deviation of the methylene blue molarity was 0.04946 f 0.00015. Water was demineralized, distilled, and finally distilled from alkaline permanganate. Chloroform was either analyzed reagent grade (J. T. Baker Chemical Co.) or ACS reagent grade (Caledon Laboratories, Georgetown, Ontario). In all applications, chloroform was shaken immediately before use with an equal volume of distilled water and filtered through dry Whatman No. 2 filter paper ( 2 ) . Distribution Coefficients. Measurement of the distribution coefficient of promethazinium ion and the distribution coefficient of promethazinium picrate between chloroform and 0.052 M phosphate buffer (pH 2.g0)a t 25 f 1 "C have been described (I). The molar absorptivity of promethazinium picrate at 400 nm was measured in chloroform by adding an excess of picrate to aqueous solutions containing several different concentrations of promethazine hydrochloride, equilibrating with an accurately known volume of chloroform, allowing the phases to separate, and pumping the chloroform phase through the Teflon filter membrane into the 1-cm flow cell. A linear Eleer's law plot was obtained for promethazinium picrate. The distribution coefficient of benzethonium picrate between chloroform and 0.063 M aqueous phosphate buffer (pH 7.50) a t 25 f 1 "C, and the molar absorptivity of benzethonium picrate were determined by procedures analogous to those used above -C 1980
American Chemical Society
554
ANALYTICAL CHEMISTRY, VOL. 52, NO. 3, MARCH 1980
Table I. Assay Values ( a )and Average Deviations for Cations Titrated with Picrate compound benzethonium.Cl hexadecyltrimethylammonium.Br promethazine.HC1
photometrica 92.1 95.6
:
99.8
lowing two equations which describe the titration of a sample cation BH+ u’ith titrant anion A-:
other
I
0.4 0.1
92.0 96.0
i
0.4b
:
0.6
99.8
z
0.05L‘
0.lb
I
Resent method. Sample concentration about 2.5 x M , assuming that it is all in the aqueous phase. Ferricyanide method ( 6 ) . Average and average deviation of Fajan’s, nonaqueous, and photometric ion-pair titration in which aqueous phase absorbance was monitored ( I ) . for promethazinium picrate. The distribution coefficient of benzethonium ion between chloroform and 0.063 M phosphate buffer (pH 7.5,J was determined in a manner similar to that previously used for promethazinium ion ( I ) . The molar absorptivity of picric acid in chloroform was determined after a single extraction of a 0.1 M HC1 solution of picric acid-the amount of picric acid extracted into chloroform being calculated from the known pK, and distribution coefficient of the neutral picric acid species (10). Titration Procedure. A 25.00-mL volume of phosphate buffer (pH 7.50 or 2.g0) and a 70.00-mL volume of water-washed chloroform are pipetted into the thermostated titration vessel and the spectrophotometer is set to 100% T with the chloroform phase pumping through the flow cell. Then 5.00 mL of aqueous sample solution is added. The titration proceeds as previously described (1, 2 ) .
RESULTS AND DISCUSSION Titration Equations. The chemical and phase distribution equilibria, including hydrolysis and ion pairing side reactions, associated with the ion pairing titration reaction between the cation BH’ and the anion A- are shown in Equation 1. Species without subscripts are in the aqueous phase and those with the script “ 0 ” are in the organic phase.
T h e equilibrium constants describing these equilibria have been defined under conditions of constant ionic strength and buffering of the aqueous phase ( I ) . T h e species BH’ represents either a weak acid such as promethazinium ion or an aprotic ion such as benzethonium or methylene blue ion. Likewise, A- represents either a weak base such as picrate or a very weak base such as lauryl sulfate or dodecylbenzene sulfonate. Either BH+ or A- may be the titrant and the other the sample. As before ( I ) , the treatment neglects ion pairing and selfassociation of ions in the aqueous phase and it neglects dissociation and aggregation of ion pairs in the organic phase. Of consequence in the present case is the fact that interfacial adsorption is neglected and all chemical species are assumed to be in one or another of the bulk liquid phases. Combining the equilibrium constant expressions (right hand side of Table I in ref. 1)with mass balance expressions and with Beer‘s law expressions for all absorbing species in the organic phase (i.e,, B, B H X , BHA, HA, MA) yields the fol-
and
(3) In these equations nA and nBHare moles of titrant and t g H X , e;, tHA,and sample added t o the titration vessel; t i f A are the products of molar absorptivities a t the wavelength used and cell pathlength, for the subscript species in the organic phase; AoM is the observed absorbance of the organic phase, corrected for dilution if necessary; V and V , are volumes of aqueous and organic phase in liters; a H is the activity of hydrogen ion in the aqueous phase. Equation 2 is solved for [BH] by substituting the experimentally measured values of all of the constants and an assumed value of Aobd. Only one root of Equation 2 will have a real, positive value of [BH] which is then substituted into Equation 3 to calculate the corresponding value of nA. T h e process is repeated with new assumed values of Aohsd until the entire theoretical titration curve has been calculated. For the case in which a sample A- is titrated with titrant BH+, the titration equations are:
ANALYTICAL CHEMISTRY, VOL. 52, NO. 3, MARCH 1980
555
and
I t is evident from previous considerations ( I ) that a titrant should be chosen which gives as large a value of Ks as possible, all other things being equal. A smaller value of Ks results in greater curvature in the vicinity of the equivalence point in the plot of Aobd vs. moles of added titrant and, consequently, in a less precise endpoint evaluation. In general, i t can be shown that larger values of Kb, larger values of K,, and more dilute sample solutions (larger V or smaller ~ B H will ) have an effect similar t o smaller values of K,. A larger volume of organic solvent ( Vo), however, will improve the curve shape, though the effect is not pronounced. An increase in any of the side-reaction equilibrium constants Kg, KI,BHx,Km, and KI,MA will yield poorer titration curves. Figure 1 shows that the correct choice of p H can be very important when BH+ is a weak acid and/or A- is a weak base. In this example, with K, = lo-' a n d Kb = lo-'', it is only a t intermediate p H values in the vicinity of (Ka(Kb/K,))'I2that sharp titration curves are obtained. At low p H , protonation of A- competes successfully with the ion pair reaction between B H + and A-, and a t high p H , deprotonation of B H + is too competitive. It is therefore desirable not only that the titrant yield a large value of Ks with the sample, but also that it be a very weak base (small Kb) if it is the anion, or a very weak acid (small K,) if it is the cation. In such circumstances it will be possible to select a p H low enought or high enough to repress the hydrolysis reaction of the sample ion without incurring excessive hydrolysis of the titrant. Equations 2-5 can be used to predict titration curves, under various experimental conditions, for cases in which any or all of the chemical species in the organic phase absorb radiation at the wavelength of interest. However. it is anticipated that the technique will be most attractive for the titration of noncolored sample ions (e.g., surfactants) with colored titrant ions, since such a titration can be performed in a colorimeter. For this reason, only titrant-containing species are assumed t o absorb in the theoretical curves shown in Figure 1. Considering the more general titration case in which all of the species in the organic phase absorb radiation at the wavelength of interest, it can be shown that the linear segments of the titration curve well before a n d well after the equivalence point will have the following properties. Before the equivalence point, both the slope of the line and its ordinate intercept can have zero or positive values. After the equivalence point, the slope of the line can be zero or positive and its intercept can be negative, zero, or positive. When the titration is performed a t a wavelength a t which only titrant-containing species absorb (e.g., BHA, HA, and MA for titrant A-) and the titrant is added as an aqueous solution, there is no dilution correction in the linear region of the curve before the equivalence point. This is in contrast to the case in which aqueous phase absorbance is measured ( 1 ) . Before the equivalence point in the titration of B H + with A-, the linear region is given by the expression:
which shows that Aobsd is independent of V, the volume of
T 4 >
25
Figure 1. Theoretical titration curves Influence of pH. n,, = 1.00 X mol; V = 0.03 L; V, = 0.07 L, K, = lo-'; K, = loVib;K , BHx = t, = o; = io*; = 1.0; K,,,, = 0.0;K, = ios; K,, = IC? HA = 500; CBHA = lo4; K, = 10". p H indicated by e a c h curve I
-
&
aqueous phase. After the equivalence point, the linear region is given by the expression: Aobsd
=
~BHA~BH ~
V,
+
(nA
-
nBH)
In titrations where the p H is high enough that the species HA is negligible, and where the species MA is negligible, the numerator term in the bracketed quantity of Equation 7 is essentially zero and the linear portion of the titration curve beyond the equivalence point is a line with slope zero and ordinate intercept ( t g n~B H~/ which has no dilution correction. In cases where the numerator term has a nonzero value so that the linear region beyond the equivalence point has a positive slope, a dilution correction is necessary; but it applies only to the second term on the right of Equation 7. In many cases this correction is small enough to be neglected as it was in all of the titrations described in this paper. Discussions of theoretical titration behavior in this section have emphasized the titration of sarnple BH' with titrant A-. Analogous conclusions can readily be drawn concerning the reverse titration. Titration of Cations. Two quaternary ammonium surfactants and the ammonium form of a tertiary amine drug were titrated with picrate ion in order to test Equations 2 and 3 and to demonstrate the precision and accuracy of the method. T h e theoretical titration curve for benzethonium chloride, calculated using the independently measured values of the constants given in the caption, is shown as a solid line in Figure 2 . T h e points, which were obtained in an actual titration, are seen to lie on the predicted curve suggesting the validity of Equations 2 and 3. Since the pH is high ( p H 7.5') so that the amount of the neutral picric acid species (HA) is negligible and since sodium picrate (MA) does not extract ( I j, the linear region beyond the equivalence point has a zero slope, and no dilution correction is necessary. The assay value and precision, based on end points obtained by linear extrapolation of the titration curves, are presented in Table I along with results obtained for the determination of benzethonium chloride by the ferricyanide method (6). Similar d a t a are shown for the determination of hexadecyltrimethyl ammonium bromide. In general. the agreement between the titration and the ferricyanide method is several parts per thousand which, for an ion, pair titration carried out
v,)
556
ANALYTICAL CHEMISTRY, VOL. 52, NO. 3, MARCH 1980 .
Table 11. Assay Values (%) and Average Deviations for Anions Titrated with Methylene Blue compound
phot o m et r i c
sodium lauryl sulfate sodium dodecylbenzene sulfonate
80.1 i 0.20
o t h e Pi
74.0
z
O.Jb
81.2 72.3
58.3
i
0.03c
57.9 i 0.3
z
i
0.2 0.1
Present method. Sample concentration about 3 x R.3 in aqueous phase. Present method. Sample concentration about 3 X 1 0 - j Tv3 in aqueous phase. Present method, but aqueous phase was 0 . 2 M in Na,SO, and absorbance was measured after shutting o f f stirrer and allowing phases t o separate (see text). Sample concentration about 1 . 5 X M in aqueous phase. ii Method of Blank (see text). a
.c/
~~~
.
~
-~ ~~~
.
ICC
" & I
Figure 2. Titration of benzethonium chloride with picrate. The line is calculated from Equations 2 and 3 using the values below. The points are experimental. nBn = 6.82 X lod mol; V = 0.0300 L; V, = 0.0706 L; K, = 0; K, = 2.14 X K I , B H X 7 6.5; K J , M A = O.O;, KB = 0; 5, = 29.5; K, = 2.34 X 10'; tBHX= 0; tg = 0;cMA = lo4; tHA= 430; tBHA= 1.02, X lo4; pH 7.5,; wavelength, 400 nm
, ,.-
Figure 4. Experimental titration curve of sodium lauryl sulfate with methylene blue. V = 0.0300 L; V, = 0.0700 L; pH 7.5,; wavelength
= 658 nm (see Table 11)
Figure 3. Titration of promethazine hydrochloride with picrate. The line is calculated from Equations 2 and 3 using the values below. The points are experimental. n, - 8.55 X mol;,K, = 1.0 X lo-'; K,,,, = 0.9; K, = 7.9 X lor;>, = 7.24 X lo6; tBnA= 8.20 X lo3; pH 2.9,; (all other parameters as in Figure 2) at a sample concentration of about 104 M, is quite acceptable. Figure 3 shows a comparison between the theoretical and observed titration curves for promethazine hydrochloride, and Table I includes a comparison between the assay value and that obtained previously ( I ) . It will be noted in Figure 3 that the experimental points in the vicinity of the equivalence point fall below the theoretical curve. A possible explanation of this phenomenon is nonequilibrium in the ion pair extraction process a t the very low concentrations of BH' and A- encountered near the equivalence point. A similar effect was not observed when the aqueous phase absorbance was monitored ( 1 ) . In other words, the departure of ions from the bulk aqueous phase may be more rapid than their appearance in the bulk organic phase, suggesting that the slow process may be related to the rate of interfacial desorption. At any rate, the phenomenon is not evident outside the immediate region of the equivalence point, and end points obtained by linear extrapolation yield assay values in good agreement with other methods (Table I). Since the slope of the titration curve beyond the equivalence point is only slightly greater than zero, dilution corrections are negligible (vide supra). Titration of Anions. The surfactant anions lauryl sulfate a n d dodecylbenzene sulfonate were titrated with methylene
blue cation. The resulting assay values are compared in Table I1 with those obtained by the visual indicator titration. A typical titration curve for sodium lauryl sulfate is shown in Figure 4. T h e sodium dodecylbenzene sulfonate sample was an inhomogeneous syrup so that it was necessary to prepare a dilute stock solution of it and to assay aliquots of the stock solution by both the phobometric and visual methods. Because of the sample inhomogeneity, stock solutions prepared a t different times contained different amounts of surfactant (e.g., 58% and 73%) but both assay methods yielded the same value on a given stock solution (Table 11). Even with a sample concentration of 4 X M the results obtained by the proposed method agree within 2 % with those obtained by the recognized indicator method (in which the sample concentration is 6 X M). Emulsions and Adsorption. When the titration of sodium dodecylbenzene sulfonate was originally performed at a sample concentration of M with continuous stirring, an emulsion formed and small amounts of the aqueous phase passed through the Teflon filter probe membrane. Performing the titration at 4 x 10-5M sample concentration eliminated this problem. Although satisfactory precision and accuracy were obtained at this lower sample concentration (Table II), there is, as predicted, greater curvature in the vicinity of the equivalence point. It was found that increasing the ionic strength of the aqueous phase (e.g., 0.2 M Na,SO,) eliminated emulsification a t sample concentrations of M but, at the same time, caused low assay values (e.g., 10% relative). T h a t is, the "plateau" after the equivalence point occurred at a lower absorbance value t h a n predicted. However, when the same titration was repeated, but the absorbance of the organic phase
ANALYTICAL CHEMISTRY, VOL. 52, NO. 3, MARCH 1980
was measured after shutting off the stirrer a n d allowing the phases to separate, an accurate end point was obtained (Table
11). T h e dependence of organic phase absorbance on stirring was a reversible phenomenon-i.e., turning the stirrer off increased the absorbance a n d turning it on again decreased the absorbance. Also, the absorbance decrease caused by stirring was greater for higher sample concentrations. It is evident that this phenomenon is a consequence of interfacial adsorption of the ion pair formed between the sulfonate and methylene blue. During vigorous stirring, the ratio of t h e interfacial area between the two phases to their bulk solution volume is increased by several orders of magnitude over that for the unstirred system. Thus, while a barely detectable fraction of t h e ion pairs might be adsorbed at the interface between 70 mL of quiescent chloroform and 30 mL of water, this fraction could become quite significant in a stirred solution. Of the systems reported, only methylene blueedodecylbenzene sulfonate exhibited a detectable amount of interfacial adsorption.
CONCLUSIONS When titrant is added as an aqueous solution a n d the organic phase absorbance is monitored, it is often possible t o neglect dilution corrections, which represents an advantage over using the absorbance of the aqueous phase (I). However, interfacial adsorption, which was seen a t higher sample concentrations in the titration of dodecylbenzene sulfonate with methylene blue, a n d slow transfer of the ion-pair from the interface into the bulk organic phase, which was possibly seen in the titration of promethazinium ion with picrate, are potential sources of titration error when the absorbance of the organic phase is used to follow the titration. On the other hand, when the aqueous phase absorbance is used, the error in the linear regions before a n d after the equivalence point, which are used to locate the end point by extrapolation, is likely to be negligible. This is because an ion-pair present at the interface has already left the bulk aqueous phase a n d produced the corresponding decrease in absorbance. This is an advantage to using aqueous phase absorbance. However, it is evident from this study and the previous one ( 1 ) that accurate a n d precise determinations can be performed by monitoring t h e absorbance of either phase in cases where interfacial influences are absent.
557
T h e absence of adsorption effects can readily be demonstrated in a proposed titration system by comparing absorbances observed during stirring with those observed after phase separation, or alternatively by comparing assay values obtained in titrations of several different sample concentrations. I t is also recommended, as with all photometric measurements, that photometric linearity of the spectrophotometer be verified using the same flow cell and over t h e same absorbance range encountered during a titration. A variety of water immiscible solvents can be used as the organic phase in ion-pair titrations. Although t h e strong ion-pair extracting solvent methylisobutyl ketone (MIBK) is incompatible with Acidflex peristaltic pump tubing, mixtures of MIBK with chloroform up to at least a 1:l ratio can be used. Nonpolar solvents such as carbon tetrachloride are also compatible with the tubing. Provided that values of K , are large enough to yield sharp titration curves, it is preferable to use a less polar organic solvent, since the likelihood of aqueous phase "carryover" into the flow cell is reduced. Finally, it should be mentioned that properties of the phases other than absorbance can be used to continuously follow the course of a two-phase titration. In particular, the titration apparatus employed in this work should be well suited to radiometric titrations (11, 12) which can be performed only in systems of more t h a n one phase.
LITERATURE CITED Mohammed, H. Y.; Cantwell, F. F. Anal. Chem. 1979, 5 1 , 1006. Cantwell, F. F.; Mohammed, H. Y. Anal. Chem. 1979, 5 1 , 218. Heinerth, E. In "Anionic Surfactants-Chemical Analysis", Cross, J. T., Ed.; Dekker: New York, 1977; Chapter 6. Cross, J. T. I n "Cationic Surfactants ', Jungermann, E., Ed.; Dekker: New York, 1970; Chapter 13. Hummel, D. "Identification and Analysis of Surface-Active Agents"; Interscience: New York, 1962; Vol. 1. "Tests and Standards for New and Nonofficial Remedies"; Lippincott: Philadelphia, 1953; p 28. Reference 5; p 228. Reference, 3; p 224. Abbot, D. C. Analyst (London) 1962, 87,286. Gustavii, K.: Schill, G. Acta fharm. Suec. 1966, 3, 241. Braun, T.; Tolgyessy, J. "Radiometric Titrations"; Pergamon: New York, 1967. Zolotov, Y. A. "Extraction of Chelate Compounds"; Ann-Arbor Humphry: Ann-Arbor, Mich., 1970; Chapter 6.
RECEI\TDfor review July 16, 1979. Accepted October 26, 1979. Work supported by the National Sciences and Engineering Research Council of Canada and the University of Alberta.
