Anal. Chem. 1994,66, 327-334
Influence of pH on Retention and Selectivity in Micellar Liquid Chromatography: Consequences of Miceilar-I nduced Shifts of Ionization Constants Andrew H. Rodgerst Chemistry Department, University of New Orleans, Lakefront, Louisiana 70 148 Morteza 0. Khaledi' Department of Chemistry, North Carolina State University, P.0. Box 8204, Ralebh, North Carolina 27695
The retention and selectivity of ionizable solutes in the twosurfactant-mediatedreversed-phase LC techniques (micellar and ion pair) are compared through the use of a generalretention equation for mono- and zwitterionic solutes. In RPLC with different mobilephase modirers (organicsolvents,surfactants), the influence of pH can be quantitatively described by one general equation. The existing theory for the secondary chemical equilibria has been applied to compare selectivity effects in micellar and ion-pair chromatography. In order to predict the influence of different mobilephase compositions on the selectivity of ionizable compounds,one should determine how a mobile-phase parameter influences the ionization equilibria (Le., selective shifts of pK,) and self-selectivity (defined as the ratio of retention factors of the acid/conjugate base). For example, the first ionization constants of amino acids and peptides in aqueous mobile phase are between pH 2.3 and 3.4. Consequently, the pH required to maximize the retention of these solutes by ion suppression is less than the operational pH range of silica-based columns. In addition, since the pK.1 values of these solutes are similar, adjustment of pH has little effect on separation selectivity. In contrast, ion-pairing and micellar mobile phases with SDS surfactant increase the magnitude and range of the ionization constants. These trends are more pronounced with micellar mobile phase. The displacement of solute ionization constants to higher pH with micellar mobile phase allows the maximal, limiting, retention of zwitterionic solutes to be observed within the pH limits of silica-based columns. It is also observed that the range of the apparent ionization constants is greater in the micellar mobile phase. This wider range allows for a more selective distribution of the constants, to the extent that pH becomes an important parameter to influence selectivity when micellar mobile phases are used. Reversed-phase liquid chromatography (RPLC) has been the most popular HPLC technique for the separation of uncharged solutes in the past decade. The limitations of ionexchange columns and the high performance of the alkylsilane bonded phases has prompted workers to investigate the usefulness of conventional RPLC with hydroorganic eluents for the separation of ionizable solutes. This has become possible through the incorporation of secondary chemical equilibria (SCE) such as protropic and ion interaction 'Present address: Witco Corp., P.O. Box 310, Hahnville, LA 70057. 0003-2700/94/0366-0327$04.50/0 0 1994 American Chemical Society
equilibria. It has been shown that control of mobile-phase pH can lead to impressive selectivities.' However, a major constraint of the silica bonded phases is the limited pH range of between 2.5 and 1.5. This is troublesome in the separation of poorly retained strong acids and bases whose dissociation constants are either outside or close to the boundaries of this range. In such cases, introduction of electrostatic interactions through the addition of a small amount of moderately hydrophobic surfactant (with a charge opposite to that of the solute) to the eluent provides adequate retention and selectivity. This technique, widely known as ion pair chromatography (IPC), has greatly extended the capabilities of RPLC for the separation of charged compounds. Nevertheless, IPC suffers from several shortcomings, one of which is poor reproducibility of retention. This is primarily because the operating concentration of ion-pairing reagents in the mobile phase is located on the steepest part of the surfactant isotherm on the alkyl bonded stationary phase. Consequently, small fluctuations in mobiIe-phase surfactant concentrations proportionally change the amount of adsorbed surfactant on the stationary phase. This greatly influences retention, which is directly related to the stationary-phase charge density. The amount of adsorbed surfactant is also a function of the composition of the mobile phase, especially the concentration of organic modifier. Consequently, changes in mobile-phase composition may lead to long equilibration times of the stationary phase. This translates into difficulties in the application of gradient elution in IPC and makes method development difficult as long reequilibration times are required. Micellar liquid chromatography (MLC) is also capable of separating charged and uncharged compounds. In MLC, a longer chain surfactant (as compared to IPC where usually nc I 12) is used at higher concentrations (above the critical micelle concentration, cmc) to ensure the formation of micellar aggregates. The stationary phase is modified with an approximately constant concentration of ionic surfactant. Due to the unique characteristics of micelles in the mobile phase and constant composition of the stationary phase (which is independent of micelle concentration in the mobile phase), MLC offers several advantages over IPC. For example, (1) Karger, B. L.;Le Page, J. N.; Tanaka, N. In High-Performance Liquid Chromatography: Advances and Perspectives;Horvath, C . S., Ed.;Academic Press: New York, 1980; Vol. 1.
Ana&ticalChemistty, Voi. 00, No. 3, February 1, 1994 327
simultaneous enhancement of solvent strength and selectivity, feasibility of optimizing eluent composition due to a more predictable and robust behavior, short equilibration times after changes in micelle concentrations or even organic modifier, and alleviation of the general elution problem have been These capabilities, combined with the previously reported advantages such as the possibility of direct on-column injection of physiological fluids, rapid gradient capability, and enhanced detection providecompellingreasons to consider MLC as an alternative to IPC for the separation of charged compounds or mixtures of charged and uncharged solutes.“1° A shortcoming of MLC is poor chromatographic efficiency as compared to conventional RPLC with hydroorganic mobile phases. The chromatographic efficiency in MLC, however, is probably comparable to that of IPC. In both IPC and MLC, the alkyl stationary phase is modified with hydrophobic surfactant monomers, which causes a slow resistance to mass transfer from the surfactant-modified stationary However, the negative effect of broad peaks on separation can be adequately compensated for in MLC by the simultaneous optimization of selectivity and solvent ~ t r e n g t h . ~ . ~ Unfortunately, the capabilities of MLC in separating charged compounds have been mostly overlooked in the past. To the best of our knowledge, there has only been one systematic study on the retention behavior of monoprotic solutes in MLC by Arunyanart and Cline-Love,14where they studied the effect of pH and ionic strength on the retention of weak acids and bases. They reported the derivation of a mathematical model todescribe theobserved behavior in MLC in terms of pH at fixed micelle concentration. In this paper, retention equations are reported for monoand diprotic solutes in MLC and it is demonstrated that the dependence of retention on pH in different HPLC techniques (RPLC with purely aqueous eluent, ion-pairing eluent, and micellar mobile phases) can be described by one general equation. The main difference between these methods is the apparent dissociation constants and limiting retention factors (at high and low pH) in the differing mobile phases.
EXPERIMENTAL SECTION A Waters Assoc. (Milford, MA) liquid chromatographic system, consisting of a 6000A pump, a Varian UV 50 detector (200 nm), and a U6K injector, was used to collect the chromatographic measurements. The following analytical columns were used and protected with a silica precolumn and a dry-packed C-18 column: (1) LichroCart, ODS (Merk,
Darmstadt, Germany); 12.5 X 0.4 cm, dp = 5 pm, VO= 0.66 mL. (2) PRP-1, poly(styrene4ivinylbenzene); 5 X 0.46 cm, dp = 5 pm, VO= 0.23 mL. (3) Ultrasphere, ODS (Alex); 15 X 0.46 cm, dp = 5 pm, VO = 0.92 mL. Stock solutions of sodium dodecyl sulfate (SDS; puriss grade, Fluka) were prepared in doubly distilled, deionized water. Phosphoric acid (HPLC grade) and its mono- and divalent sodium salts were used to buffer the mobile phase (Fisher Scientific Co.). In the ion-pairing mobile phase, 2-propanol (Fisher Scientific Co.) was used as the organic modifier. The mobile phase was prepared by combining the surfactant solution/2-propanol/buffer/doublydistilled deionized water, adjusting the pH, and filtering through a 0.45-pm Nylon-66 membrane (Rainin Instruments). Amino acids and peptides were obtained from Sigma Chemicals (St. Louis, MO). Column void volumes were determined in the absence of surfactant from the first baseline deviation resulting from the injection of NaN03 (10 measurements). Safety Considerations. Sodium dodecyl sulfate is a strong chemical irritant and potential allergen. Appropriate handling precautions should be used including the avoidance of inhalation.
RESULTS AND DISCUSSION Monoprotic Species. Equations describing the retention of ionizable species in RPLC have been described in the literature,I”O and in MLC the retention of monoprotic species can be described by the following equilibria:14
+ + + - +
HA + L,
HAL, (KsHA)
(i)
HA
M
HAM (K,HA)
(ii)
A-
L,
A-L, (KsA-)
(iii)
A-
M
A-M (K, A-)
(iv)
HA
A-
H+ (K,)
(VI where and K mare~the~respective binding constants of the weak acid to the stationary-phase ligands, L,, and micelle, M. K s ~and - KmA- are the binding constants for the conjugate base form, and Ka is the acid dissociation constant in aqueous solution. Defining the retention factor as
and using the constants i-v, then the retention for monoprotic acids in MLC is given by
(2) Khaledi, M. G.;Strasters, J. K.; Rodgers, A. H.; Breyer, E. D. Anal. Chem. 1990, 62, 130. (3) Strasters, J. K.; Breyer, E. D.; Rodgers, A. H.; Khaledi, M. G.J. Chromatogr. 1990,511, 17. (4) Kord, A. S.;Khaledi, M. G.Anal. Chem. 1992,64, 1901. ( 5 ) Madamba, L.S.;Khaledi, M. G.,submitted for publication in J. Chromatogr. (6) Armstrong, D.W. Sep. Purif. Methods 1985, 14, 213. (7) Yarmchuck, P.;Weinberger, R.; Hirsch, R. F.; Cline-Love, L. J. Anal. Chem. 1982, 54, 2233. (8) Hinze, W. L.; Singh, H. N.; Baba, Y.; Harvey, N. G. TrAC, Trends Anal. Chem. 1984, 3, 193. (9) Dorsey, J. G.Adu. Chromatogr. 1987, 27, 167. (IO) Hinze, W. L. In Organized Media in Chemical Separation; Hinze, W. L., Armstrong, D. W., Eds.;ACS Symposium Series 342; American Chemical Society: Washington, DC, 1987. (11) Dorsey, 3. G.; De Echegaray, M. T.; Landy, J. S.Anal. Chem. 1983,55,924. (12) Armstrong, D. W.; Ward, T. J.; Berthod, A. Anal. Chem. 1986, 58, 579. (13) Berthod, A.; Borgerding, M. F.;Hinzc, W. L. J . Chromatogr. 1991,556,263. (14) Arunyanart, M.; Cline-Love, L. J. Anal. Chem. 1985, 57, 263.
328
Analytical Chemistry, Vol. 66, No. 3, February 1, 1994
where k’nA and kk- are the limiting capacity factors of the acid and conjugate base. The micelle concentration, [MI, is (15) Knox, J. H.; Hartwick, R. A. J . Chromatogr. 1981, 204, 3 . (16) Melander, W.; Horvath, C . In Zon Pair Chromatography: Theory with Biological and Pharmaceutical Applications; Hearn, M. T. W., Ed.; Chromatographic Science Series; Marcelle Dekker: New York, 1985; Vol. 31. (17) Horvath. C . Anal. Chem. 1977, 49, 142. (18) Pietrzyk, D. J.; Kroeff, E. P.; Rotsch, T. D. Anal. Chem. 1978, 50, 497. (19) Pietrzyk, D. J.; Kroeff, E. P.; Rotsch, T. Anal. Chem. 1978, 50, 502. (20) Foley, J. P.; May, E. M. Anal. Chem. 1987, 59, 102.
(7 Micelle
"1
\
PH
Flgurs 1. Predlctlon of retentlon and pK.,, for Trp In MLC. Retentlon at 0.05 (open squares with cross) and 0.1 M SDS (solld squares) both fltted with eq 3 (known values Kw = 7210, K , - = 6.5, and pK., = 2.35) to give $ L , K , = 17 800 and $L,KA- = 2.59 (solld line). Lower line predicted from eq 3 at 0.20 M SDS compared to experlmentally determined retentionat 0.20 M SDS (crosses). Notethat the derivatives of the curves (not shown) give the pK.,, values directly from eq 3. Column 3, 50 mM Na2H!W4adjusted to pH with concentrated H,PO4.
the stoichiometric surfactant concentration minus the cmc. Equation 2 predicts retention as a function of micelle concentration and pH. This equation can be modified so that retention can be predicted at any value of micelle and proton concentration:21
(3) Verification of Equation 3. Figure 1 shows eq 3 fitted to the MLC retention of the amino acid Trp at 0.05 and 0.10 M SDS. The equation is used to obtain 4hKsHA and ~&K,Afrom the retention of 0.05 and 0.10 M SDS with the PKal = 2.35;K mand~ K d~ - values are known.22 Equation 3 is then used to predict retention, at 0.20 M SDS. The figure shows that the predicted and experimental retentions at 0.20 M SDS are in good agreement. Note that eq 3 correctly predicts observed values of pKam (the observed, or apparent ionization constant in micellar solution) at pH values in excess of 4.73 from the pKal value of 2.35. Relationship between K, and Km. The phenomenon of micelle-induced shifts of ionization constants is well documented.23-34 One equation to describe this effect is given by (21) Rodgers, A. H.; Starsters, J. K.; Khaledi, M. G. J. Chromatogr. 1993,638, 203. (22) Rodgers, A. H.Ph.D. Dissertation, University of New Orleans, 1990. (23) Berizin, I. V.;Martiek, K.; Yatsiriskii, A. K. Russ.Chem.Reu. (Engl. Traml.) 1973, 42,187. (24) Evans, D. F.; Ninhamd, B. W. J. Phys. Chem. 1983,87, 5025. (25) El Seoud, 0. A. Adv. Colloid Interface Sci. 1989, 30, 1. (26) Underwood, A. L. Anal. Chim. Acta 1982, 140, 89. (27) Pramramauro, E.; Pelizzctti, E. AMI. Chim. Acto 1981, 126, 253. (28) Fernhdcz, M. S.; Fromherz, P. J. Phys. Chem. 1977, 81, 1755. (29) Funasaki, N. J. Phys. Chem. 1979,83. 1998. (30) Chaimovich, H.; Alexio, R. M. V.; Cuccovia, I. M.; Zannctte, D.; Quina, F. H. In Solution Behavior of Surfactants-Theoretical and Applied Aspects; Mittal, K., Fendler, E.J., Eds.;Plenum Press: New York, 1982; Vol. 2, p 949. (31) Bonilha, J. B. S.; Chiericato, G.; Martins-Franchetti, S.M.; Ribaldo, E. J.; Quina, F. H.J . Phys. Chem. 1982, 86, 4941. (32) Garnett, C. J.; Lambie, A. J.; Beck, W. H.; Liler, L. J. Chem. Soc., Faraday Trans. 1 1983, 79,953. (33) He, Z. M.; OConnor, P. J.; Romsted, L.S.;Zanettc, D. J . Phys. Chem. 1989, 93, 4219. (34) Khaledi, M. G.; Rodgers, A. H. Anal. Chim. Acta 1990, 239, 121.
dynamically modified stationary phase Flgurs 2. MLC of a zwltterlon wlth anionlc surfactant.
Berizin et al.:23 (4)
This equation predicts that the apparent ionization constant in micellar solution Kam,is a product of the aqueous ionization constant, Ka, and the ratio of the solute binding constants. Equation 2 is simplified by dividing numerator and denominator by 1 K ~ H A [ Mand ] substituting eq 4 to give
+
Equation 5 is similar to that reported by Horvath et al. for RPLC:16J7
Thedifference between eqs 5 and 6 is thevalueof the ionization constant; Ka is the value measured in aqueous media and Kam is that in micellar media. Diprotic Species. For a zwitterionic compound in MLC, the retention is controlled by the equilibria shown in Figure 2. (Note that the possibility of direct transfer of the solute from micelle to stationary phase is not shown.) Using the equilibria in Figure 2 and the method used to derive eq 2 gives
+ Kmc[MI)+ k:(l + Km,[Ml)Kal/[H+l + k'a(1 + Kma[MI)KalKa2/[H+12)/(1+ &,[MI + (1 + Km,[MI)Ka1/[H+I + (1 + Kma[M1)KalKa2/[H'12)
k'= (k'#
(7) The limiting retention factors of the cation (HABH+), zwitterion (-ABH+) and anion (-AB) are represented by k i , k i , and k i , respectively, Kmc, Km, and Kmaare the corresponding solute/micelle equlibrium constants, [MI is the micelle concentration, and Kal and Ka2are the deprotonation constants for the solute in aqueous media. AS Ka1,m = Ka,I( 1 Analytcal Chemkby, Vol. 66, No. 3,February 1, lQQ4
329
+ Kmz[MI)/(l + Kmc[MI) and Ka2,m = Ka,2(1 + Kma[Ml)/ (1 t &,,[MI), then eq 7 can be simplified to k‘= k i + ~ ’ A , , ~ / [ H ++I k’a~al,m~a2,m/[~+I’ (8) 1 + ~al,m/[H+I+ ~ a l , m ~ a 2 , m / [ ~ + 1 ~ When the micelle concentration is zero in eq 7, then the equation simplifiesto that previously reported for zwitterionic solutes in RPLC:’J6
Again, eqs 8 and 9 differ only in the value of the ionization constants. Equation 9 predicts that the retention of zwitterionic compounds in RPLC passes through a minimum at intermediate pH due to the greater polarity of the zwitterionic solute. At the extremes of pH, the solute will be singlycharged and less polar and show greater retention. This form of retention has been observed for peptides in i ~ n - p a i r i n gand ~~ hydroorganic mobile phases.’* Panels a and b of Figure 3 show the simulated retention of zwitterionic compounds in MLC based upon eq 7 with nonionic and anionic micelles. The relative values for binding and dissociation constants are selected on the basis of the chromatographic and potentiometric results. Retention behavior with nonionic micelles (Figure 3a) is similar to conventionalRPLC, with the retention passing through a minimum. For anionic surfactant (Figure 3b), electrostatic repulsions between the solute and surfactant will result in low retention at elevated pH. In both cases, retention decreases with increasing micelle concentration. The figures predict a micellar-induced shift of ionization constant with the apparent pKa values increasing with anionic micelles. The magnitude of the pKa shift is a function of micelle concentration as predicted by eq 4. Verification of Zwitterionic Retention in MLC. Figure 4 shows the results of the nonlinear regression of the retention of zwitterionic histidine fitted with a modified form of eq 8.36 This amino acid has a low pKa2value of 6 , which is observable within the pH range of a silica-based column. The figure confirms the behavior predicted in Figure 3b: decreasing retention at elevated pH, decreasing retention with increasing micelle concentration, and increasing PKa with increasing micelle concentration to the extent that maximal retention is observed at pH >2. This is significant as pKa1 for this solute is 1.82 and maximum retention in aqueous mobile phases would be expected at pH ) at the optimum hydrogen ion concentration ([H+],,,) from eq 14. This allows the optimum selectivity (aopt) to be defined in terms of the mean self-selectivity (SS,) and the respective values of the ionization constants of solutes i and j throughZo PHoptimum
- 2ss, + SS,~/~(SS,K*~+ K*~)/(K*~K*~)‘/~ aopt - 2ss, + SS,~/~(SS,K; + K*J/(K*~K;)’/~ (15) Equation 15 is shown in Figure 8, where aOpt is plotted as a function of SS, and DpK*. The figure shows that as the difference between the ionization constants, DpK*, tends to zero, large differences in SS, have little effect on cyopt; conversely, with SS, as small as five, a DpK* of 1 pH unit gives a value of aopt approaching 2.74. It is apparent that the DpK* term is a significant factor in dictating the magnitude of cyopt. Hence the figure shows that the MLC solute pairs (7-9 from Table 4) have selectivities comparable and higher than those for the ion-pair mobile phase (4-6) even though the ion-pair mobile phase is characterized by the higher mean self-selectivities.
Table 4 shows the data workup for Figure 8. The table shows that the mean pK* is highest for the micellar case at 0.39 compared to 0.06 for aqueous and 0.12 for ion pairing. However, the ion pairing mean SS, value of 48.68 is higher than for the micellar value as 12.42 and aqueous value of 6.46. These terms combined give the selectivity at optimum pH from eq 15. The mean selectivity for the aqueous system is low at 1.06, for ion pairing the value is 1.23,and the micellar mean is highest at 1.66. The table also shows that the mean optimum pH for maximum selectivityfor the micellar examples occurs at pH 4.79, for ion pairing at 4.15, and for aqueous at 3.08. Caution should be taken when means for analytespecific data such as these are interpreted, but the data are presented in this way to illustrate the contributions of the difference in pK and the mean self-selectivity to the maximum selectivity at optimum pH. The data clearly show that even though the ion-pair mobile phase demonstrates the greatest self-selectivity,the micellar mobile phase can produce greater selectivities due to greater differences in solute ionization constants. One should becautious about generalizing the results shown in Figure 8 to all MLC separations as compared to those with IPC. This is because eqs 14 and 15 are valid for a pair of soluteswith identical limiting retention factors. This condition is not met for all solutes examined in the three different techniques. A more exact approach would be to derive a general selectivityequation for any pair of solutes and compare the capabilities of different techniques under their own optimum pH. The main problem with this method would be the complexity of the equation. In addition, eq 14 predicts the optimum pH value, assuming that it is not dependent upon other parameters such as ion-pairing concentration, micelle concentration, and organic modifier composition.This is not true for MLC (and perhaps for IPC) where the DpK values and probably the self-selectivity term depend upon the micelle concentration. In other words, one would obtain a different optimum value for pH at various micelle concentrations. A better approach would be the simultaneous optimization of these two parameters in MLC through interpretive method^.^.^^ Anatytical Chemlstry, Vol. 66, No. 3, February 1, 1994
333
CONCLUSION Through appropriate modeling, the comparative effect of mobile-phase composition on the variation of solute ionization constant has been demonstrated. In micellar liquid chromatography, these effects can have a favorableselective nature on the ionization constants of zwitterionic compounds with similar aqueous ionization constants. However, it should be noted that the micellar-induced pK shifts do not necessarily lead to an enhancement of selectivity. In fact, for a group of compounds whose aqueous pKa and hydrophobicities are very different, micelles might play a leveling role as shown for chl~rophenols.~~ It has also been demonstrated that micellarinduced pK shifts in anionic micelles result in apparent pK values for peptides, and particularly amino acids, that are (38) Khaledi, M. G.; Strasters, J. K.; Smith, S.C. Anal. Chern. 1991, 63, 1820.
394
Ana&ticaiChemistty, Vol. 66, No. 3, February 1, 1994
more compatible with the pH range of alkylsilane stationary phases. Therefore, MLC on alkylsilanestationary phases can be considered as an alternate to other forms of RPLC utilizing specialty, pH-stable columns operating at the extreme ranges of pH in an attempt to maximize the retention of solutes (via ion suppression) that are both strongly acidic (or basic) and hydrophilic in nature.
ACKNOWLEDGMENT We are grateful to the National Institutes of Health for thesupport of this research through a grant t0N.S.C.U. (First Award GM38738). We also thank Joost Strasters for useful discussions. Received for review May 13, 1993. Accepted October 1, 1993." Abstract published in Advance ACS Abstracrs, November IS, 1993.
