Anal. Chem. 2003, 75, 3153-3160
Effect of Amine Counterion Type on the Retention of Basic Compounds on Octadecyl Silane Bonded Silica-Based and Polybutadiene-Coated Zirconia Phases Xiqin Yang, Jun Dai, and Peter W. Carr*
Department of Chemistry, Smith and Kolthoff Hall, University of Minnesota, 207 Pleasant Street SE, Minneapolis, Minnesota 55455
In a previous paper, we compared the mixed-mode retention characteristics of cationic solutes on octadecyl silane-bonded silica (ODS) and polybutadiene-coated zirconia (PBD-ZrO2) phases. It is well recognized that both reversed-phase and ion-exchange interactions contribute to the retention of cations on ODS phases. The reversedphase interaction results from the bonded hydrocarbon chain; the ion-exchange interaction originates in the ionized residual silanol groups. These two types of interactions also exist on the PBD-ZrO2 phase. The polybutadiene contributes to the reversed-phase interaction and the ionized zirconol, but primarily, the adsorbed Lewis base anions, such as phosphate or fluoride, contribute to the ion-exchange interaction. We have shown that on ODS phases, reversed-phase interactions are much more important, whereas the opposite is true of PBD-ZrO2 phases. In this work, we investigate the effect of several amine mobile phase counterions on the retention of cationic solutes on ODS and PBD-ZrO2 phases. The effects of the chain length and the type of amine (1°, 2°, 3°) counterion on the retention of basic compounds were studied. In contrast to older studies of type A silica-based phases, the results show that the chain length and type of the amine blocker do not have a large effect on the retention of basic compounds with the newer type B silicabased materials. However, on the PBD-ZrO2 phase, very striking differences in retention were observed with different amine counterions. We show that the molecular geometry of the amine counterion has a significant effect on the retention of basic solutes on the PBD-ZrO2 phase. The analysis of basic compounds using reversed-phase liquid chromatography (RPLC) continues to be of great interest to chromatographers, because most pharmaceuticals and biomedically important compounds contain basic groups.1-3 Unfortunately, the separations of these solutes suffer from a few notorious problems on silica-based phases, including irreproducible reten(1) McCalley, D. V. LC-GC 1999, 17, 440-456. (2) Stadalius, M. A.; Berus, J. S.; Snyder, L. R. LC-GC 1988, 6, 494-500. (3) Vervoort, R. J. M.; Maris, F. A.; Hindriks, H. J. Chromatogr. 1992, 623, 207-220. 10.1021/ac034107r CCC: $25.00 Published on Web 05/16/2003
© 2003 American Chemical Society
tion, severely tailed peaks, and low column efficiency.4 Silanol interactions are believed to be the major reason for such behavior.2,5,6 Many methods have been developed to reduce silanophilic retention of basic solutes in RPLC. A common approach is to add amine counterions to the eluent as silanol blockers.2,7-14 Here, the counterion means those that have a charge type opposite that of the stationary phase. Among other methods, adding an ionpairing agent (e.g., SDS) to neutralize the cationic analyte is also very common. The amine counterion competes with the analytes for the silanophilic binding sites and, thus, diminishes the extent of interaction of the basic analyte with the ionized silanols. Structural effects of both the analyte and the amine counterion on the separation have been studied.3,9-12,15-18 The stronger the analyte binds to the surface, the more severe the effect of the interaction is on the separation. For the amine counterion, the stronger it binds, the more effectively will it “block” the silanol sites.2 The interaction between the amine and the surface ionexchange sites is strengthened by reducing steric hindrance around the nitrogen atom, increasing substitution on the nitrogen atom, and increasing hydrophobicity of the amine.2 Kiel et al.10 systematically studied the effect of 15 different amine counterions on the chromatographic behavior of amitriptyline derivatives on a type A silica-based phase. They concluded that longer chain amines were better blockers than shorter amines and that tertiary amines were more effective than were primary and secondary (4) Vervoort, R. J. M.; Debets, A. J. J.; Claessens, H. A.; Cramers, C. A.; Jong, G. I. d. J. Chromatogr., A 2000, 897, 1-22. (5) Sadek, P. C.; Koester, C. J.; Bowers, L. D. J. Chromatogr. 1987, 25, 489. (6) Nawrocki, J. J. Chromatogr., A 1997, 779, 29-71. (7) Snyder, L. R.; Kirkland, J. J.; Glajch, J. L. Practical HPLC Method Development, 2nd ed.; Wiley-Interscience: New York, 1997. (8) Nahum, A.; Horvath, C. J. Chromatogr. 1981, 203, 53-63. (9) Bij, K. E.; Horvath, C.; Melander, W. R.; Nahum, A. J. Chromatogr. 1981, 203, 65-84. (10) Kiel, J. S.; Morgan, S. L.; Abramson, R. K. J. Chromatogr. 1985, 320, 313. (11) Gill, R.; Alexander, S. P.; Moffat, A. C. J. Chromatogr. 1982, 247, 39. (12) Sokolowski, A.; Wahlund, K. G. J. Chromatogr. 1980, 189, 299. (13) Hill, D. W. J. Liq. Chromatogr. 1990, 13, 3147. (14) Hill, D. W.; Kind, A. J. J. Liq. Chromatogr. 1993, 16, 3941-3964. (15) Sadek, P. C.; Carr, P. W. J. Chromatogr. Sci. 1983, 21, 314. (16) McCalley, D. V. J. Chromatogr., A 1994, 664, 139-147. (17) Ascah, T. L.; Feibush, B. J. Chromatogr. 1990, 506, 357. (18) Gasco-Lopez, A. I.; Santos-Montes, A.; Izquierdo-Hornillos, R. J. Chromatogr. Sci. 1997, 35, 525-535.
Analytical Chemistry, Vol. 75, No. 13, July 1, 2003 3153
amines. The same results were also obtained by Stadalius et al.2 In contrast, Sokolowski and Wahlund12 found that bulky amines with several methylenes in each substituent (e.g., tributylamine) did little to improve the chromatography of the basic solutes. Triethylamine (TEA) and n-hexylamine are commonly used as counterions to improve the tailing problem of basic compounds. Dimethyl tertiary amines are found to be even more effective.7 Gill et al.11 extensively studied the importance of the molecular geometry of the amine counterion. Bulky alkyl groups close to the charge center (the nitrogen atom) strongly hinder the amine’s interaction with the silanol groups. Although there are many different opinions as to the efficacy of various amine counterions, tertiary amines (R-N(CH3)2) or quaternary ammonium salts (R(CH3)3N+X-, where R is a long alkyl chain), such as dimethyloctylamine and trimethyloctylammonium ion, are commonly considered to be among the best silanol blockers. Zirconia-based phases are useful alternatives to silica-based phases because of their high chemical and thermal stability.19-21 PBD-ZrO2 is the most widely studied zirconia-based material for RPLC.22 This phase has been proven to be very useful for the separation of nonpolar and polar solutes over a wide range in pHs and at temperatures up to 200 °C.23-25 The main challenge in using zirconia supports is to sequester the hard Lewis acid sites on their surface.26,27 The strong, hard Lewis acid Zr(IV) sites interact with the hard Lewis base functional groups (R - SO3-, R - PO3-, R COO-) of the analytes; these interactions are often very strong. Such interactions usually have slow dissociation kinetics and cause irreproducible retention and poor peak shape. When a hard Lewis base buffer, such as acetate, fluoride, or phosphate is present in the eluent, it interacts very strongly with the surface Lewis acid sites. Adsorbed Lewis base anions impart a negative charge to the surface and cause Coulombic (cation exchange) interactions to occur with positively charged analytes. Consequently, basic (cationic) analytes are retained on PBD-ZrO2 phases by a mixedmode retention process, that is, a strong cation-exchange interaction with the adsorbed buffer anions and a weaker reversed-phase interaction with the PBD coating.25,28,29 The similarity between the mixed-mode retention characteristics of cationic solutes on the ODS and PBD-ZrO2 phases led us to compare them in a previous study.29 Various models of the mixed-mode retention mechanism on the two types of phases were compared and evaluated. We showed that a multisite model is more probable, and the model equations allowed us, for the first time, to quantify the relative amount of hydrophobic and Coulombic contributions to the retention of any solute on any stationary phase.29 In the present work, we first study the effect (19) Dunlap, C. J.; McNeff, C. V.; Stoll, D.; Carr, P. W. Anal. Chem. 2001, 73, 598A-607A. (20) Kawahara, M.; Nakamura, H.; Nakajima, T. J. Chromatogr. 1990, 515, 149158. (21) Trudinger, U.; Mu ¨ ller, G.; Unger, K. K. J. Chromatogr. 1990, 535, 111125. (22) Rigney, M. P.; Weber, T. P.; Carr, P. W. J. Chromatogr. 1989, 484, 273291. (23) Li, J.; Carr, P. W. Anal. Chem. 1997, 69, 837. (24) Li, J.; Carr, P. W. Anal. Chem. 1997, 69, 2202-2206. (25) Hu, Y. Ph.D. Thesis, University of Minnesota, Minneapolis, 1998. (26) Blackwell, J. A.; Carr, P. W. Anal. Chem. 1992, 64, 853-862. (27) Blackwell, J. A.; Carr, P. W. Anal. Chem. 1992, 64, 863-873. (28) Hu, Y.; Yang, X.; Carr, P. W. J. Chromatogr., A 2002, 968, 17-29. (29) Yang, X.; Dai, J.; Carr, P. W. J. Chromatogr., A 2003, in press.
3154
Analytical Chemistry, Vol. 75, No. 13, July 1, 2003
of the degree of substitution on the nitrogen atom of the analyte on its retention on an ODS and a PBD-ZrO2 column. Then, by using a judiciously chosen homologue series (the p-alkylbenzylamines) as probe solutes, the effect of the structure and hydrophobicity of nine alkylamine counterions on the chromatographic behavior of the solutes were studied on the two types of columns. We show that the degree of substitution of the charge center of both the analyte and the amine counterion have a substantial effect on separations on the PBD-ZrO2 phase, although on the specific type B silica ODS phase studied here, the effect of the counterion type was much smaller than on the PBD-ZrO2 phase and not nearly so large as found in the prior studies on type A silica-based phases mentioned above. EXPERIMENTAL SECTION Instruments. All chromatographic experiments were carried out with a Hewlett-Packard 1090 chromatographic system equipped with a binary pump, an autosampler, a temperature controller, and a diode array detector (Hewlett-Packard S.A., Wilmington, DE). Data were collected and processed using Hewlett-Packard Chemstation software. Analytical Columns. The PBD-ZrO2 (batch no. 24-124) particles used in this work were obtained from ZirChrom Separations Inc. (Anoka, MN). The average particle size was 4.1 µm, which was measured by SEM and image analysis. The surface area of the packing was 11.2 m2/g (by nitrogen BET), and the average pore diameter was 500 Å. The carbon content was 2.5% (w/w), which corresponds to 186 µmol/m2 of CH2 groups. The zirconia particles were packed by the downward slurry method at 5000 psi. Stainless steel (316) column blanks with dimensions of 50 mm × 4.6 mm i.d. were obtained from Isolation Technologies (Hopedale, MA). The ODS column used (5-µm particle size, 15 × 0.46 cm i.d.) was an Alltima C18 column obtained from Alltech (Alltech Associates Inc., Deerfield, IL). The surface area of this column was 350 m2/g and the pore size was 100 Å, with a carbon loading of 16% (w/w) (38 µmol/m2 of CH2 groups). Reagents. All chemicals were reagent grade or better. HPLCgrade methanol (MeOH) was from Pharmco (Brookfield, CT). HPLC water was obtained from a Barnsted Nanopure deionizing system (Dubuque, IA) and run through an “organic-free” cartridge followed by a 0.2-µm particle filter. The water was boiled and cooled to room temperature to remove carbon dioxide. The solution was then degassed and stored under helium. The purpose of this step is to prevent the adsorption of carbonate ion onto the stationary phase surface. All solvents were filtered through a 0.2µm filter (Lida Manufacturing Corp., Kenosha, WI) before use. p-Alkylbenzylamines were home-synthesized. The procedure was described elsewhere.29 Other chemicals used in this study were purchased from Aldrich (Aldrich Chemical Co., Milwaukee, WI). Chromatographic Conditions. All measurements were made at a flow rate of 1 mL/min, and detection was at 254 nm. The injection volume was 1 µL with analyte concentrations of 1-2 mg/ mL. The column temperature was controlled to 35 °C with a precision of (0.2 °C. The dead time was determined by injecting uracil and acetone for the ODS and PBD-ZrO2 columns, respectively. The mobile phases were prepared by first dissolving appropriate amounts of phosphoric acid and ammonium hydroxide in water to the desired pH and concentration. The pH of the
aqueous solution was double-checked with a pH meter and adjusted as needed, noting the total amount of all reagents. The same recipe was then used for all mobile phases at the same pH and concentration throughout the study in order to keep the conditions consistent. These solutions were then filtered through a Millipore (type HA) 0.45-µm membrane filter prior to use and, finally, mixed with prefiltered MeOH. RESULTS AND DISCUSSIONS Effect of the Degree of Substitution of the Analyte Cation. In our previous study,28 we found that steric hindrance in the positively charged analyte had a large effect on its retention on the PBD-ZrO2 phase. For example, two analytes, amitriptyline and nortriptyline, have very similar structures. The main difference between them is that there is one more methyl group on amitriptyline (a tertiary amine) than on nortriptyline (a secondary amine). Under most of the pH conditions used in the above study (pH< 12), amitriptyline always eluted from the PBD-ZrO2 column before nortriptyline. A pure reversed-phase retention mechanism requires the more hydrophobic solute (i.e., amitriptyline) to elute later than the less hydrophobic solute (i.e., nortriptyline), which has one fewer methyl group. Only at very high pH (>12), where both amitriptyline and nortriptyline are unprotonated, was amitriptyline more retained on the PBD-ZrO2 phase. However, at low or medium pHs, both analytes are protonated. The extra methyl on the nitrogen of the 3° amine makes the steric hindrance larger or limits how close the cationic center can approach the surface anion site. As a result, on the very negatively charged PBD-ZrO2 column, amitriptyline is less retained than nortriptyline. In this study, we used four simple basic compounds as probes to further compare the effect of solute hydrophobicity and steric hindrance on the retention of basic solutes. These compounds were a “pseudohomologue series” of the primary, secondary, tertiary, and quaternary amines: benzylamine, N-methylbenzylamine, N,N-dimethylbenzylamine, and N,N,N-trimethylbenzylammonium chloride (TMBA). From benzylamine to TMBA, the hydrophobicity of the solute increases, and the steric hindrance around the nitrogen atom also increases. The hydrophobicity helps the amine ion bind to the surface;1,2 however, the larger charge center increases the steric hindrance, which keeps the analyte away from the surface charge. If the retention mechanism is purely reversed-phased, the more methylated solute should be more retained. However, we observe the opposite trend when the analytes are retained by a mixed-mode retention mechanism on the PBD-ZrO2 phase. In Figure 1, we clearly see that on the PBDZrO2 phase, benzylamine is the most and TMBA is the least retained solute at pH 6 when phosphate is present in the eluent. This means that the effect of steric hindrance is more important than is the effect of solute hydrophobicity for retention on the PBDZrO2 phase. The retention factors (k′) of the above four amines on an Alltima ODS column (a type B silica) are also shown in Figure 1. It is very clear that the dependence of the k′ on the number of methyl groups on the ODS phase is totally different from that on the PBD-ZrO2 phase. The variation in k′ on the ODS phase shows no trend with solute structures. Effect of Counterion Type on the Retention of Amines. Ten different counterions were tested, including ammonium, n-butylamine (BA), n-pentylamine (PA), triethylamine (TEA), dimethyl-
Figure 1. Effect of analyte’s degree of substitution by methyl groups on retention under mixed-mode conditions. Mobile phase: 60/40 MeOH/25 mM ammonium phosphate, pH 6, 35 °C. Solutes: (A) benzylamine; (B) N-methylbenzylamine; (C) N,N-dimethylbenzylamine; (D) N,N,N-trimethylbenzylammonium. O, PBD-ZrO2; b, ODS (Alltima).
butylamine (DMBA), dipropylamine (DPA), n-hexylamine (HA), n-octylamine (OA), tributylamine (TBA), and 1,6-hexanediamine (HDA). Among these compounds, there are five primary amines (BA, PA, HA, OA, and HDA), one secondary amine (DPA), and three tertiary amines (DMBA, TEA, and TBA). Four amines (HA, DPA, DMBA, and TEA) have the same number of carbon atoms and should have roughly the same hydrophobicity. We intentionally chose these amines to test the effect of the chain length and the size and type (1°, 2°, 3°) of the counterion on the retention of amines on the PBD-ZrO2 and the ODS columns. To compare the reversed-phase and ion-exchange retention characteristics of the stationary phases when different types of amine counterions are present in the eluent, we used two homologue series as probes, alkylbenzenes and p-alkylbenzylamines (p-R-C6H4-CH2-NH2). The amine homologue series was chosen so that the carbon chain in the R group could be varied without having any significant steric effect on the interaction of the protonated amino group with anionic surface sites. A different homologue series (C6H4-(CH2)n-NH2) did not behave in this fashion. H1 NMR studies not reported here confirmed that >95% of the analytes were protonated under mobile phase conditions similar to those used here. In our previous study,29 we showed that the p-alkylbenzylamine homologue series is an excellent set of probe solutes for estimating the relative contributions of reversed-phase and ion-exchange interactions to the overall retention. Table 1 gives the retention factors of the alkylbenzenes and p-alkylbenzylamines on the PBD-ZrO2 phase as a function of the counterion type. When a secondary or tertiary amine was used as the counterion in the eluent, the cationic analytes did not elute, even though the solute eluted in an ammonium phosphate buffer of the same concentration. The analytes eluted only when the ammonium ion or a primary amine counterion was present in the eluent. In addition, the chain length of the amine counterion has a large effect on the retention of basic solutes. The data in Table 1 make it clear that the chain length of the primary amine counterion Analytical Chemistry, Vol. 75, No. 13, July 1, 2003
3155
Table 1. Retention Factors of Alkylbenzenes and p-Alkylbenzylamines on PBD-ZrO2a k′ solute/counterionb
NH4+
TEA
TBA
DPA
DMBA
BA
PA
HA
OA
HDA
ethylbenzene propylbenzene butylbenzene pentylbenzene benzylamine p-methylbenzylamine p-ethylbenzylamine p-propylbenzylamine p-butylbenzylamine
2.06 3.81 7.24 13.67 6.11 8.24 12.40 20.54 35.67
2.08 3.81 7.23 13.61 NEc NE NE NE NE
2.20 4.06 7.75 14.65 NE NE NE NE NE
2.07 3.79 7.22 13.52 NE NE NE NE NE
2.05 3.79 7.24 13.68 NE NE NE NE NE
2.02 3.72 7.22 NAd 10.06 14.55 21.71 34.60 NA
2.13 3.92 7.62 NA 7.59 11.49 16.97 27.12 NA
2.41 4.44 8.47 16.06 5.35 8.13 12.51 20.55 35.63
2.75 5.07 9.68 17.96 1.73 3.00 4.67 8.42 15.26
1.93 3.54 6.83 NA 1.01 1.44 2.17 3.47 5.97
a Mobile phase: 55/45 MeOH/25 mM phosphate buffer adjusted to pH 6.0. Other experimental conditions: 1 mL/min, 35 °C, 254 nm. b NH + 4 ) ammonium; TEA ) triethylamine; TBA ) tributylamine; DPA ) dipropylamine; DMBA) dimethylbutylamine; BA ) butylamine; PA ) pentylamine; c d HA ) hexylamine; OA ) octylamine; HDA ) 1,6-hexanediamine. No elution observed (k′ . 50). No data available.
Figure 2. Dependence of p-propylbenzylamine’s retention on the number of methyl groups in the primary amine counterions on the PBD-ZrO2 phase. Conditions are the same as in Table 1. The error bars are determined by three replicate injections.
is of utmost importance in controlling analyte retention. The k′ of the probes exhibits a surprising dependence on the counterions hydrophobicity (see Figure 2). We see that n-butyl and npentylamine are less effective counterions, as compared to the simple ammonium ion; however, n-hexyl and, most especially, n-octylamine are far stronger blockers than is the ammonium ion. We attribute these results to the very strong interaction of the counterion and the surface negative charge on the PBD-ZrO2 phase. The bulky substitutents on the nitrogen atom of the secondary or the tertiary amine counterion sterically impede the amine’s interaction with the negative surface charge. As a result, the basic analytes, which are primary amines, bind to the stationary phase very tightly. That is the reason these analytes were not eluted when secondary or tertiary amines were used as the counterion. When we changed the counterion from ammonium to butylamine, two properties of the counterion changed: hydrophobicity and steric hindrance. Butylamine is more hydrophobic and its steric hindrance is larger than that of the ammonium ion. It is well-known that hydrophobic interactions help amines bind to the charged surface;1,2 however, the bulky charge center increases the steric hindrance and inhibits the approach of the counterion to the surface charge. When the chain length is not long (i.e., butylamine and pentylamine), the steric effect is more important 3156
Analytical Chemistry, Vol. 75, No. 13, July 1, 2003
than the hydrophobic effect, which results in weaker binding of the counterion to the surface. This explains why amine solutes have longer retention when one uses butylamine and pentylamine instead of ammonium ion as the counterion. When the chain length increases to six or eight, the hydrophobic effect becomes more important. The amine counterion binds more tightly to the surface with the help of the hydrophobic interaction between the carbon chain on the counterion molecule and the PBD coating. It is also evident from the increase of retention of the neutral alkylbenzenes in the hexylamine and octylamine buffers that the binding of the alkylamine counterion to the stationary phase surface increases the carbon loading of the phase. We also used a diamine (1,6-hexanediamine, pKa’s: 10.2, 11.030) as a counterion. All of the amine analytes had much less retention in the diamine buffer than even in the octylamine buffer (see Table 1). We believe that this indicates the diamine is a better blocker than the singly charged amines. One may argue that the doubly charged counterion made the surface positively charged and, thus, shortened the retention of analyte cations. However, no evidence supports this argument. If the surface were positively charged, the retention factors of the amine analytes would be 1, fronting; 90% in some cases) in controlling retention on the PBD-ZrO2 phase. The results explain the high sensitivity of PBD-ZrO2 phases, as compared to silica-based phases, to the structure and type (primary, etc.) of the amine solute and counterion. Analysis of log k′ vs nCH2 Plots. A very fundamental characteristic of RPLC is its close empirical adherence to a linear relationship between log k′ and the number of methylene groups (nCH2) for a homologous series of solutes.31,32 This relationship is represented in the Martin equation,
log k′RP ) A + BnCH2
(1)
where A is the intercept, which varies with the functional group of the homologous series and is directly related to the phase ratio. The slope, B, is related to the free energy of transferring one methylene group from the mobile to the stationary phase, and to a good approximation, B is independent of the homologous series used. (31) Karger, B.; Gant, J. R.; Hartkopf, A.; Weiner, P. H. J. Chromatogr. 1976, 128, 65. (32) Vigh, G.; Varga-Puchony, Z. J. Chromatogr. 1980, 196, 1.
In Figure 5, the log k′ of alkylbenzenes in eluents with different counterions is plotted against nCH2 of the neutral analytes. The slopes of the plots are almost the same in all of the eluents, which means the amounts of free energy needed to transfer the neutral solute from the mobile to the stationary phase are about the same. The interesting thing is that different intercepts are obtained. This result agrees with our hypothesis that some of the hexylamine and octylamine are adsorbed on the surface and increase the stationary phase carbon loading and, thus, the “effective” phase ratio. The values of log k′ of the p-alkylbenzylamines in eluents with different counterions are plotted against nCH2 of the solutes in Figure 6. In contrast to the above results using the alkylbenzenes, the slopes are different with the various counterions, with the highest slope for octylamine and the lowest for butylamine. The two-site model, which we discussed extensively in our previous paper,29 can be used to explain this result. In the two-site model, the apparent retention factor of a solute, k′, is equal to the sum of the reversed-phase retention factor, k′RP, and the ion-exchange retention factor, k′IEX. So we get
log k′ ) log(k′RP + k′IEX)
(2)
Since k′IEX is independent of the number of methylene units, nCH2, the slope of the plot of log k′ vs nCH2, will vary with the contribution of the k′RP and k′IEX to the apparent retention. Because the ionexchange retentions of the solutes in different counterions are obviously different, the slopes are different. Since octylamine binds to the PBD-ZrO2 phase more tightly than the other singly charged counterions, the ion-exchange interaction between the solutes and the stationary phase is weaker when octylamine is the counterion. In addition, the binding of octylamine to the surface increases the surface hydrophobicity, which causes the reversed-phase retention of the solute to increase. As a result, the relative contribution of k′RP is greater when octylamine versus the other amines is used as the counterion. The slope is the highest under this condition, and it is the closest to the slope of the alkylbenzenes, which are retained by a pure reversed-phase retention mechanism. As seen in Figure 6 (0), the use of the diamine as the counterion gave the lowest intercept and least retention for the basic solutes, as expected. However, the slopes for the p-alkylbenzylamines are lower than anticipated on the basis of the above argument, whereas the slopes for the alkylbenzenes (see Figure 5, 0) are the same as with the other counterions. This suggests that the diamine is modifying the hydrophobicity near Analytical Chemistry, Vol. 75, No. 13, July 1, 2003
3157
Figure 4. Dependence of k′ of p-propylbenzylamine vs counterion type in the eluent. Conditions are the same as in Table 1. (A) ODS and (B) PBD-ZrO2. Counterion type: 1, octylamine; 2, hexylamine; 3, ammonium; 4, triethylamine; 5, dipropylamine. Table 3. Dependence of k′ on the Reciprocal Counterion Concentrationa slope ( s.d.b +
intercept ( s.d.c +
solute/counterion
NH4
OA
BA
NH4
OA
BA
benzylamine p-methylbenzylamine p-ethylbenzylamine p-propylbenzylamine
136 ( 6 185 ( 7 283 ( 9 463 ( 17
37 ( 1 57 ( 2 83 ( 4 137 ( 10
226 ( 10 325 ( 15 483 ( 22 771 ( 36
0.4 ( 0.3 0.6 ( 0.3 0.8 ( 0.4 1.4 ( 0.7
0.4 ( 0.1 0.9 ( 0.3 1.7 ( 0.5 3.8 ( 1.3
1.2 ( 0.3 1.8 ( 0.4 2.7 ( 0.6 4.3 ( 1.0
a k′ is regressed against 1/[NH +] (mM). b Slope and standard error of the slope of the regression described in footnote a, n ) 4. c Intercept 4 and standard error of the intercept of the regression described in footnote a, n ) 4.
Table 4. Analysis of Slope and Intercept of Plots of k′ vs 1/[NH4+] solute
15c
25c
35c
50c
intercept (k′RP)a
15c
% k′IEX/k′ b 25c 35c
50c
benzylamine p-methylbenzylamine p-ethylbenzylamine p-propylbenzylamine
9.41 12.79 19.46 31.93
6.11 8.24 12.40 20.54
4.36 5.90 8.95 14.72
3.00 4.07 6.17 10.15
0.43 0.56 0.78 1.37
95 96 96 96
93 93 94 93
90 91 91 91
86 86 87 87
25d
35d
50d
25d
35d
50d
100d
10.06 14.55 21.71 34.60
7.74 11.23 16.78 26.78
5.88 8.54 12.75 20.34
89 88 88 88
85 84 84 84
80 79 79 79
64 63 62 62
5e
8e
10e
25e
5e
8e
10e
25e
7.60 12.14 18.05 30.46
5.05 8.25 12.47 21.77
4.06 6.77 10.36 18.27
1.73 3.00 4.67 8.42
95 92 91 87
93 88 86 82
92 87 84 79
80 69 63 55
k′ counterion NH4+
BA
OA
benzylamine p-methylbenzylamine p-ethylbenzylamine p-propylbenzylamine
benzylamine p-methylbenzylamine p-ethylbenzylamine p-propylbenzylamine
100d 3.24 4.76 7.15 11.37
1.16 1.77 2.69 4.27
0.35 0.92 1.71 3.82
a Intercept of k′ vs 1/[NH +] (mM). b % k′ c d 4 IEX/k′ ≡ 100 × [(k′-k′RP)/k′]. Concentration (mM) of ammonium ion in the eluent. Concentration (mM) of butylamine in the eluent. e Concentration (mM) of octylamine in the eluent.
the cation-exchange site. This is consistent with the “hydrophobically assisted” ion-exchange mechanism discussed in our prior work.29 When butylamine, a less efficient blocker, is used, the ion-exchange retention dominates, so the slope is the lowest and different from the slope of the alkylbenzenes. Thus, the slope of these plots (see Figure 6) can be used as an estimate of the relative contribution of ion-exchange and reversed-phase retention. Analysis of k′ vs 1/[C+]m Plots. As discussed in our previous paper,29 according to the two-site model a plot of the retention factor, k′, against the reciprocal of the counterion concentration 3158
Analytical Chemistry, Vol. 75, No. 13, July 1, 2003
in the eluent, 1/[C+]m, can provide a quantitative estimation of the contribution of reversed-phase and ion-exchange retention.
k′ ) k′RP + BIEX/[C+]m
(3)
The slope of this plot, BIEX, is directly related to the ion-exchange equilibrium constant and the number of accessible negatively charged sites on the stationary phase and, thus, is proportional to k′IEX. The intercept will be an estimate of the reversed-phase
Figure 5. Plot of log k′ vs nCH2 of alkylbenzenes on the PBD-ZrO2 phase using different mobile phase counterions. Conditions are the same as in Table 1. From the bottom to the top of the figure: 0: 1,6-hexanediamine, log k′ ) 0.275(0.006 nCH2 - 0.27(0.02; r ) 1.000, SD ) 0.009, n ) 3; b: ammonium, log k′ ) 0.274(0.002 nCH2 0.237(0.007; r ) 1.000, SD ) 0.004, n ) 4; 3: butylamine, log k′ ) 0.277(0.006 nCH2 - 0.25(0.02; r ) 1.000, SD ) 0.009, n ) 3; 9: pentylamine, log k′ ) 0.277(0.006 nCH2 - 0.23(0.02; r ) 1.000, SD ) 0.009, n ) 3; O: hexylamine, log k′ ) 0.275(0.002 nCH2 - 0.172(0.008; r ) 1.000, SD ) 0.005, n ) 4; 1: octylamine, log k′ ) 0.273(0.002 nCH2 - 0.109(0.007; r ) 1.000, SD ) 0.004, n ) 4.
Figure 6. Plot of log k′ vs nCH2 of p-alkylbenzylamines on the PBDZrO2 phase using different mobile phase counterions. Conditions are the same as in Table 1. From the bottom to the top of the figure: 0: 1,6-hexanediamine, log k′ ) 0.206(0.006 nCH2 - 0.06(0.02; r ) 0.998, SD ) 0.020, n ) 4; 1: octylamine, log k′ ) 0.238(0.006 nCH2 + 0.22(0.02; r ) 0.999, SD ) 0.019, n ) 5; O: hexylamine, log k′ ) 0.205(0.007 nCH2 + 0.71(0.02; r ) 0.998, SD ) 0.022, n ) 5; b: ammonium, log k′ ) 0.190(0.012 nCH2 + 0.76(0.03; r ) 0.994, SD ) 0.038, n ) 5; 9: pentylamine, log k′ ) 0.183(0.003 nCH2 + 0.88(0.01; r ) 0.999, SD. ) 0.011, n ) 4; 3: butylamine, log k′ ) 0.178(0.005 nCH2 + 0.98(0.01; r ) 0.998, SD ) 0.016, n ) 4.
retention factor, k′RP, at infinite counterion concentration. It is obvious that eq 3 cannot be exact at very low counterion concentration. We have tested that this equation is valid at the concentrations used in this study. Table 3 lists the slopes and intercepts obtained with three counterions. The largest and smallest slopes were acquired with butylamine and octylamine as the counterion, respectively. This result agrees with the above discussion; among the three counterions, ion-exchange contribu-
tion to retention is least with octylamine in the eluent. When butylamine is the counterion, the ion-exchange interaction between the solute and the stationary phase is the largest. On the other hand, the smallest and largest intercepts of the above plots were obtained with ammonium and butylamine as the counterion, respectively. This result seems to disagree with our hypothesis that the adsorption of the octylamine molecules on the stationary phase surface increases the phase ratio and, thus, the reversed-phase retention. However, the three-site model we introduced in our previous paper29 predicts that there are parts of the stationary phase environment where the solutes simultaneously interacts with both the hydrocarbon phase and the surface charge. This means that an increase in the ion-exchange interaction can also lead to an increase in the reversed-phase interaction. Therefore, when a weak blocker (e.g., butylamine) is used as the counterion, both ion-exchange and reversed-phase interactions are greater than when a strong blocker (e.g. octylamine) is used. If we calculate the relative contribution of the reversed-phase and ion-exchange to the overall retention (see Table 4), at the same concentration (25 mM), the lowest percentage of k′IEX/k′ is seen with octylamine as the counterion, which agrees with what the multisite model predicts. It should be noted that in comparison to results on PBD-ZrO2, much lower slopes of k′ vs 1/[C+]m for basic solutes on two different ODS phases were obtained in our previous study.29 On the basis of this, we concluded that for cationic solutes, ionexchange interactions dominate on PBD-ZrO2, whereas reversedphase interactions dominate on type-B ODS columns. CONCLUSIONS In this study, we found that the steric hindrance of the solute has a much greater effect on the retention of basic solutes on PBD-ZrO2 in comparison to ODS phases. As a result, quaternary amines have substantially less retention than the corresponding less hydrophobic primary amine. The same concept also applies to the effect of mobile phase counterions on retention of cationic analytes. That is, in comparison to their effect on ODS phases, the steric hindrance and the hydrophobicity of amine counterions have a very large effect on the retention of basic solutes on PBDZrO2. The effectiveness of the amine counterion results from a balance between hydrophobic and structural effects. The hydrophobicity of the counterion affects the retention of both basic and neutral analytes. The counterion’s hydrophobicity promotes the binding of amine molecules to the negative surface charges, and as a result, the binding of the hydrophobic alkylamine counterion increases the carbon content of the stationary phase. Secondary and tertiary amines, such as dimethylbutylamine and triethylamine, despite their enhanced hydrophobicity relative to linear amines with a smaller number of methylene groups, have higher steric hindrances (unfavorable geometries) and are less effective blockers. Only ammonium ion and primary amine counterions are able to elute basic solutes from PBD-ZrO2, whereas all counterions allow elution on the ODS phase tested. We also note that a diamine counterion was much more effective than were ammonium ions or n-alkylamines as blockers, clearly underscoring the importance of ionic interactions. As a result, use of different primary amine counterions with different charges and hydrophobicities allows adjustment of the retention of the basic solutes on the PBD-ZrO2 Analytical Chemistry, Vol. 75, No. 13, July 1, 2003
3159
phase. In contrast, on modern type B ODS phases, the effect of different types of amine counterion is much smaller and almost, but not quite, insignificant, which is very different from literature indications of the behavior of the older type A ODS phases. In conclusion, cationic solutes are retained on the PBD-ZrO2 phase predominately by ion-exchange interactions, whereas on ODS phases, the reversed-phase interaction is preponderant because, first, the steric hindrance of the analyte ion has a significant effect on the retention on the PBD-ZrO2 phase; second, the hydrophobicity and the type of the amine counterion have a profound effect on the retention of cationic solutes on the PBDZrO2 phase; third, the slope of the k′ vs 1/[C+]m plot, which is a
3160
Analytical Chemistry, Vol. 75, No. 13, July 1, 2003
quantitative estimation of k′IEX, is much larger on PBD-ZrO2 than on ODS phases. ACKNOWLEDGMENT The authors acknowledge the financial support from the National Institutes of Health and thank ZirChrom Separations Inc. for their generous donation of the PBD-ZrO2 particles used in this work. We are also grateful to Alltech Associates Inc. for supplying the Alltima column. Received for review February 3, 2003. Accepted April 3, 2003. AC034107R
