Dependence of Selectivity on Eluent Composition and Temperature in

Department of Chemistry and Biochemistry, Queens College, City University of New York, Flushing, New York 11367. Optimized conditions of aqueous ...
0 downloads 0 Views 376KB Size
Download PDF
Anal. Chem. 2003, 75, 1355-1364

Dependence of Selectivity on Eluent Composition and Temperature in the HPLC Separation of Taxanes Using Fluorinated and Hydrocarbon Phases Ralf Dolfinger and David C. Locke*

Department of Chemistry and Biochemistry, Queens College, City University of New York, Flushing, New York 11367

Optimized conditions of aqueous acetonitrile (ACN) eluent composition and temperature are established for the rapid separation of a standard mixture of 15 taxanes, on each of five different fluorinated phases and one C8 hydrocarbonaceous phase. On both types of stationary phase, the retention factors (k′) of most of the taxanes decrease at the same rate with increasing ACN concentration. However, the taxanes containing a xylosyl group show a higher rate of decrease, which necessitates careful control of eluent composition to achieve separation of all the taxanes. Temperature can have a remarkable and counterintuitive effect on retention and selectivity. For the C8 phase with eluent compositions in the 40%-60% ACN range, the k′ values of the xylosyl taxanes show an increase with increasing temperature over the range from 25 to 55 °C; the k′ values for 10-deacetyl baccatin III and 10-deacetyl taxol go through a maximum over the same ranges. The other taxanes behave normally. The same pattern is observed on the propyl(perfluorophenyl) phase, although this and the other fluorinated phases are less retentive. This accounts for the common belief that fluorinated stationary phases offer resolution of taxanes superior to that on hydrocarbon phases. The higher retention on the latter requires eluent compositions near 50% ACN, where careful temperature optimization is required, which in practice is rarely performed. The lesser retention on the fluorinated phases allows use of lower ACN concentrations where the aberrant temperature effect is not found, so good separations without temperature optimization can be achieved. Further evidence of the lack of any fundamental difference in the selectivity of the two types of stationary phase is the similarity of the surface excess isotherms measured for ACN/H2O on both the fluorinated and hydrocarbon phases. The class of natural products named taxanes include paclitaxel, the cytotoxic and antimitotic cancer drug.1 Paclitaxel (the common name, Taxol, is a Bristol-Myers Squibb trademark) is found at * Corresponding author. Phone: 718-997-3271. Fax: 718-997-3349. E-mail: [email protected]. (1) Danishefsky, S. J.; Masters, J. J.; Young, W. B.; Link, J. T.; Snyder, L. B.; Thomas, V. J. Am. Chem. Soc. 1996, 118, 2843-2849. 10.1021/ac020558k CCC: $25.00 Published on Web 01/30/2003

© 2003 American Chemical Society

only low concentrations in Taxus spp., its natural source,2 and complete synthesis is not commercially feasible.3 More naturally abundant taxanes such as baccatin III and 10-deacetyl baccatin III can serve as synthetic precursors to paclitaxel.3-7 Since natural sources contain a variety of taxanes, analytical methods for their complete separation are required. The major taxanes are represented in a standard mixture of 15 compounds provided by Hauser Chemical Co. (Boulder, CO). We used this same mixture in this work as we used earlier,8 to study in detail the mechanism of selectivity and to optimize the reversed-phase HPLC (RPLC) separation. Most HPLC work on taxanes has been concerned with mixtures containing only two to eight compounds,9-16 a significantly easier task. For these simpler mixtures, octadecyl,11,12,14,16 phenyl,9-11,13,15,16 cyano,13 and fluorinated phases such as perfluorophenyl (PFP)8,9,11,15,16 have been used as RPLC stationary phases. The reported methods are time-consuming, the fastest requiring 35 min to separate the 15 taxanes.8 The literature, and especially the column manufacturers’ technical literature, would have it that fluorinated phases offer superior selectivity over hydrocarbonaceous RPLC phases for taxane separations.8,9,15-18 Apart from this assertion, there has been no study of how or why a mechanism of retention and selectivity (2) Wani, M. C.; Taylor, H. L.; Wall, M. E.; Coggan, P.; McPhail, A. T. J. Am. Chem. Soc. 1971, 93, 2325-2327. (3) Vazquez, A.; Williams, R. M. J. Org. Chem. 2000, 65, 7865-7869. (4) He, L.; Jagtap, P. G.; Kingston, D. G. I.; Shen, H. J.; Orr, G. A.; Horwitz, S. Biochemistry 2000, 39, 3972-3978. (5) Baloglu, E.; Kingston, D. G. I. J. Am. Chem. Soc. 1999, 62, 1068-1071. (6) Denis, J. N.; Greene, A. E.; Guenard, D.; Gueritte-Voegelein, F.; Mangatal, L.; Potier, P. J. Am. Chem. Soc. 1988, 110, 5917-5919. (7) Gennari, C.; Carcano, M.; Donghi, M.; Mongelli, M.; Vanotti, E.; Vulpetti, A. J. Am. Chem. Soc. 1997, 62, 4746-4755. (8) Shao, L. K.; Locke, D. C. Anal. Chem. 1997, 69, 2008-2016. (9) Reichheimer, S. L.; Tinnermeier, D. M.; Timmons, D. W. Anal. Chem. 1992, 64, 2323-2326. (10) Liu, J.; Volk, K. J.; Mata, M. J.; Kerns, E. H.; Lee, M. S. J. Pharm. Biomed. Anal. 1997, 18, 1729-1739. (11) Theodoridis, G.; Laskaris, G.; de Jong, C. F.; Hofte, A. J. P.; Verpoorte, R. J. Chromatogr., A 1998, 802, 297-305. (12) Wu, Y.; Zhu, W. J. Liq. Chromatogr., Relat. Technol. 1997, 20, 3147-3154. (13) Witherup, K. M.; Look, S. A.; Michael, W. S.; McCloud, T. G.; Issaq, H. J.; Muschik, G. M. J. Liq. Chromatogr. 1989, 12, 2117-2132. (14) Harvey, S. D.; Campbell, J. A.; Kelsey, R. G.; Vance, N. C. J. Chromatogr. 1991, 587, 300-305. (15) Kopycki, W. J.; Elsohly, H. N.; McChesney, J. D. J. Liq. Chromatogr. 1994, 17, 2569-2591. (16) Ketchum, R. E. B.; Gibson, D. M. J. Liq. Chromatogr. 1993, 16, 25192530.

Analytical Chemistry, Vol. 75, No. 6, March 15, 2003 1355

should differ between hydrocarbonaceous and fluorinated phases. In addition, in none of this work was column temperature reported as an experimental variable in optimizing the separation. Thus, the initial goal of this work was to compare the retention behavior of the taxanes on a linear octyl bonded phase with that on several fluorinated bonded phases over a range of aqueous acetonitrile eluent compositions and column temperatures and to compare surface excess isotherms of acetonitrile on these phases. We found that especially on the more retentive hydrocarbon phase, at eluent compositions near the separation optimum, temperature can have a remarkable and counterintuitive effect on retention and selectivity. Our further mechanistic work on attempting to explain the observations reported here will be published separately. EXPERIMENTAL SECTION The liquid chromatograph (LC) used was a Hewlett-Packard (HP) 1090 Series II (Agilent Technologies, Avondale, PA) with autosampler and photodiode array detector, running under Chemstation. Column temperature was controlled to within (0.1 °C using a laboratory-constructed 3-in.-o.d. polyacrylic tubular water jacket capped on both ends with rubber stoppers through which connections were made for the column, water inlet and outlet, and temperature probe. A Lauda RM20 water thermostat (Brinkmann Instruments, Westbury, NY) circulated water through the jacket. Temperature was monitored with a Fisher Scientific temperature probe connected to a Fisher Accumet pH meter (Fisher Scientific, Springfield, NJ). For surface excess isotherm measurements, a Waters 410 refractive index (RI) detector (Waters Chromatography, Milford, MA) was connected to the LC and controlling software using a HP 35900E A/D interface. The RI cell temperature was set at 40.0 °C. The mobile-phase flow rate was measured using an analytical buret immersed in a large graduated cylinder through which water was circulated from a second Lauda RM20 water thermostat. The flow rate measurement accuracy was (0.01 mL/min. Six RPLC columns were kindly donated by Keystone Scientific Inc. (now ThermoHypersil-Keystone, Bellefonte, PA): Betasil C8 (linear octyl), surface area 325 m2/g; Ethyl-PFH (linear perfluorohexyl), surface area 115 m2/g; RP-100 (linear perfluorohexyl), surface area 322 m2/g; Fluofix 120E (branched propyl perfluorohexyl), surface area 300 m2/g; Propyl-PFP (perfluorophenyl), surface area 333 m2/g; and PFP-100 (perfluorophenyl), surface area 311 m2/g. All were 150 mm long × 4.6 mm i.d., 5-µm particle size, and 100-Å pore size except the Ethyl-PFP, which had 300-Å pore size. The Propyl-PFP and PFP-100 columns were different production batches of the same stationary phase. All columns were manufactured by Keystone Scientific Inc. except the Fluofix 120E column, which was manufactured by Neos Co. (Kobe, Japan) and distributed by Keystone. Surface area data were provided by Keystone Scientific Inc. The structures of the bonded phases are shown in Figure 1. HPLC grade acetonitrile (ACN) and glacial acetic acid were purchased from Fisher Scientific. Water was glass-distilled and passed through a Milli-Q purification system (Millipore Corp., (17) Catalog 2002;ThermoHypersil-Keystone, Bellefonte, PA, 2002; pp 53 and 112. (18) Chromatography Columns and Supplies Catalog, Phenomenex, Torrance, CA, 2002/03; pp 117-118.

1356 Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

Figure 1. Structures of the bonded phases studied.

Bedford, MA). Deuterated water and deuterated ACN used for the determination of surface excess isotherms were purchased from Aldrich (Milwaukee, WI). Taxane standards were kindly donated by Hauser Chemical Co. Their structures are shown in Figure 2. All taxane standards were dissolved in ACN/H2O/acetic acid (70/30/0.1) (v/v/v) at a concentration of ∼40 µg/mL. Mobile phases (other than those used in the surface excess isotherm determinations) were generated using the gradient mixer of the HP 1090. The accuracy of this device was verified by generating mixtures of 1% aqueous acetone and ACN stepwise over the entire concentration range and measuring the response of the diode array UV detector to each mixture (this procedure is described in the LC Resources DryLab Manual, section on System Check). Eluents for the isotherm determinations were prepared gravimetrically in batches of ∼500 g with an accuracy of (0.01 g. All solvents and premixed eluents were degassed by helium purging for 90 s prior to use. RESULTS AND DISCUSSION Separation of Taxanes on Hydrocarbonaceous and Fluorinated Stationary Phases. Optimized conditions for the separation of the mixture of 15 taxanes using aqueous ACN eluents were found on the six stationary phases described above. The methodology used was as follows. The column temperature was set initially at 21 °C, the column equilibrated with 95/5 ACN/ H2O (v/v) eluent, and 5 µL of sample injected. Consecutive runs at 21 °C were then performed, each time reducing the ACN concentration in 5% decrements. This was continued until the peak capacity (the number of completely resolved peaks that could be fitted spatially between the first- and last-eluting peaks) was sufficient for complete separation of the 15 taxanes. At this point, the eluent composition and column temperature were changed independently in small increments to evaluate their effect on the separation of poorly resolved peak pairs. The resulting information was used to determine conditions for the optimum separation. A solvent gradient step was then added to shorten the separation time. The optimum conditions developed for each column are given in Table 1. These are not, of course, necessarily unique sets of conditions. Sample chromatograms are shown in Figure 3. We should note that we attempted optimization using a commercial optimization software package, DryLab II version 2.04, kindly donated by LC Resources. Unfortunately, the complexity of this

Figure 2. Structures of the taxane standards.

Table 1. Optimum Conditions for the Separation of the Standard 15 Taxane Mixture on Various Columns Propyl-PFP and PFP-100 Fluofix 120E

RP-100 Ethyl-PFH Betasil C8

35/65 ACN/ H2O, isocratic for 12.5 min followed by gradient to 58/42 ACN/ H2O in 13.0 min; temperature, 22.0 °C; flow rate 2.0 mL/min; analysis time, 22 min (a) 33/67 ACN/ H2O, isocratic for 5.5 min followed by gradient to 70/30 ACN/ H2O in 10.0 min; temperature, 40.0 °C; flow rate 3.0 mL/min; analysis time, 9.5 min (b) 33/67 ACN/ H2O, isocratic for 9.0 min followed by gradient to 70/30 ACN/ H2O in 10.0 min; temperature, 60.0 °C; flow rate 2.0 mL/min; analysis time, 13 min 32/68 ACN/ H2O, isocratic for 7.0 min followed by gradient to 50/50 ACN/ H2O in 6.5 min; temperature, 50.0 °C; flow rate 3.0 mL/min; analysis time, 11 min 27/73 ACN/ H2O, isocratic for 9.0 min followed by gradient to 70/30 ACN/ H2O in 10.0 min; temperature, 40.0 °C; flow rate 3.0 mL/min; analysis time, 12 min (a) 35/65 ACN/ H2O, isocratic for 19.0 min followed by gradient to 85/15 ACN/ H2O in 8.0 min; temperature, 21.5 °C; flow rate 2.0 mL/min; analysis time for all 15 taxanes, 24 min (b) 47.5/52.5 ACN/ H2O, isocratic for 6.5 min followed by gradient to 80/20 ACN/ H2O in 10.0 min; temperature, 21.0 °C; flow rate 1.0 mL/min; 14 taxanes separated in 13.5 min. (c) 45/55 ACN/ H2O, isocratic for 8.0 min followed by gradient to 80/20 ACN/ H2O in 10.0 min; temperature, 21.0 °C; flow rate 1.0 mL/min; 14 taxanes separated in 15.5 min

Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

1357

1358

Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

Figure 3. Sample chromatograms of the taxane standards on various columns: (A) Propyl-PFP; (B) RP-100; (C) Fluofix; (D) Betasil C8.

separation caused by the nonlinear temperature dependence and noncollinear eluent composition dependence of retention of the various solutes eluded successful application of DryLab. As shown in Table 1, the same HPLC conditions were used for the two perfluorophenyl phases, which are different production batches of the same material; this is a good indication of the repeatability of manufacture. The two methods used with the Fluofix 120E differ mainly in temperature; at the higher temperature, 7-epitaxol and taxol C reverse elution order, and the resolution of other pairs shift, e.g., for 7-xylosyl taxol and 10-deacetyl-7-xylosyl taxol C. The Ethyl-PFH and RP-100 differ only in pore size, but the two columns show significant differences in elution order; in particular, taxinine M and 10-deacetyl-7-xylosyl taxol C migrate differentially. RP-100 has a larger surface area resulting in higher retentivity, which necessitates increasing the ACN concentration in the eluent, which in turn changes the selectivity. For the Betasil C8 column, method a in Table 1 provides separation of all 15 taxanes but is not entirely satisfactory because the lengthy isocratic part produces peaks that are retained longer and are consequently broad. Methods b and c are faster and give sharper peaks, but are able to separate only 14 taxanes. Using method b, baccatin III and 10-deacetyl-7-xylosyl taxol coelute; with method c, taxinine M and 7-xylosyl taxol are not separated. These latter two methods are, however, economical of solvent because of the lower percentage of ACN, shorter analysis times, and lower flow rate than any of the other methods introduced here or published elsewhere. Thus, for the separation of the taxanes, the hydrocarbonaceous phase can in fact compete with any of the fluorinated phases. The main difference between them is the somewhat higher retentivity of the C8 phase, as has been noted for other types of compounds.19-24 Among the fluorinated phases, the PFPs showed the highest retentivity and the PFH and Fluofix the lowest retentivity. General Features of the Taxane Separation. Based on our previous8 and current work, there are several common features in the separations on any of the stationary phases that enable the 15 taxanes to be classified into several groups of compounds. The

first is composed of baccatin III and its derivatives (taxanes 1-3 in Figure 2), which have the basic taxane ring structure but lack the C-13 side chain. The second group consists of the taxanes with the xylosyl group attached to the C-7 position (4, 6-8). The third includes paclitaxel (12) and its 10-deactyl derivative 9. The fourth contains 7-epitaxol (11) and its 10-deacetyl derivative 15. The last group we call the “independent” taxanes, consisting of taxinine M (5), cephalomannine (10), benzyl analogue 13, and taxol C (14). The elution order is generally that of increasing group number; the major differences among different columns is the degree of overlap between the groups and the placement of the independents within the overlaps. We find it is the precise column temperature and eluent composition that governs these differences and thus the resolution of the 15 taxanes. Temperature and composition must also be carefully controlled to ensure efficiency and ruggedness of the methods. Effect of Eluent Composition. The elution behavior of the 15 taxanes was studied on the octyl and the fluorinated stationary phases. The results using Betasil C8 and its fluorinated analogue Ethyl-PFH are representative; in Figure 4 is shown plots of ln k′ (at 21 °C) versus percent ACN for the two columns. The plots are quite similar for each column and clearly reveal the necessity of close control over eluent composition to achieve complete resolution. The plots for most of the taxanes are essentially parallel. However, there are four compounds with distinctly steeper slopes. These four are the xylosyl taxane group. At any eluent composition, the elution order within the xylosyl taxane group and within the remaining taxanes is the same on both stationary phases. But increasing the percentage of ACN causes (19) Billiet, H. A. H.; Schoenmakers, P. J.; de Galan, L. D. J. Chromatogr. 1981, 218, 443-454. (20) Reta, M.; Carr, P. W.; Sadek, P. C.; Rutan, S. C. Anal. Chem. 1999, 71, 3484-3496. (21) Yamamoto, F. M.; Rokashika, S. J. Chromatogr., A 2000, 898, 141-151. (22) Geng, X.; Carr, P. W. J. Chromatogr. 1983, 269, 96-102. (23) DeMiguel, I.; Roueche, A.; Betbeder, D. J. Chromatogr., A 1999, 840, 3138. (24) Danielson, N. D.; Beaver, L. G.; Wangsa, J. J. Chromatogr. 1991, 544, 187199.

Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

1359

Figure 4. Variation of ln k′ with eluent composition on Betasil C8 and Ethyl-PFH, 25 °C.

a larger change in ln k′ values of the xylosyl group than of the others, which changes selectivity. For example, on the Betasil C8 column, 10-deacetyl-7-xylosyl taxol C is well resolved from taxinine M at 40% ACN, but they coelute at 42.5% ACN. As the ACN concentration is raised incrementally to 60%, 10-deacetyl-7-xylosyl taxol C crosses in front of two other taxanes. The same behavior is found on the Ethyl-PFH column. These observed differences in elution order on both phases are clearly caused by differences in mobile-phase composition. The difference between the two phases is retentivity, not selectivity. The practical consequence of this is that eluent composition must be carefully adjusted on the basis of experimentation and controlled precisely to achieve complete separation. Effect of Column Temperature. Although temperatures above ambient have long been advocated to improve LC efficiency,25 temperature has been a neglected factor in RPLC selectivity. In general, the effects of temperature have been regarded as small and to decrease retention by increasing (25) Meyer, V. R. Practical HPLC, 2nd ed.; Wiley: New York, 1998.

1360 Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

temperature. Only recently has attention been paid to the use of temperature as a secondary variable after eluent composition to optimize separations.26-35 In the case of the taxanes, we have found that temperature can have a remarkable and counterintuitive effect on retention and selectivity. In Figure 5 are shown plots of ln k′ versus 1/T (K) on Betasil C8, Ethyl-PFH, and Propyl PFP, at eluent compositions known to provide adequate peak separation. On Ethyl-PFH (Figure 5A), the effect of temperature on k′ is as generally expected; retention decreases essentially linearly with increasing temperature. Since the lines are approximately parallel, selectivity is not dependent on T. On Propyl PFP (Figure 5B), the k′s of the 10-deacetyl-7-xylosyl taxols B and C vary little with T and seem to pass through a maximum. Similar but more pronounced behavior is observed on Fluofix 120E. The most remarkable result is found on the Betasil C8 column (Figure 5C). Here, the k′s of the four xylosyl taxanes actually increase with increasing temperature over the range studied. In addition, the curves for 10 deacetyl baccatin III and 10-deacetyl taxol go through a maximum. The important practical consequence of this aberrant behavior is the change in elution order with temperature. The nine taxanes studied here show no fewer than seven changes in elution order between 25 and 55 °C. Thus, if one neglected to include temperature as a variable in the separation optimization, as was the case in all previous studies of the taxane separation, the result will be insufficient resolution. To verify that this unusual temperature behavior is a mobilephase rather than a stationary-phase effect, we studied the effect of T at various eluent compositions on Betasil C8 and Propyl PFP. As is shown in Figure 6, at 90/10 ACN/H2O (v/v). retention decreases with increasing T, i.e., behavior that is ordinarily considered normal. At 70/30 ACN/H2O, the four xylosyl derivatives start to show an increase in k′ with T, which is more pronounced at 47.5/52.5 ACN/H2O and still evident at 35/65 ACN/H2O. For compositions more water-rich than this, the normal pattern returns. The same pattern of changes is found for the Propyl PFP column, as shown in Figure 7. Thus, here too there seem to be no significant basic differences in the behavior of the taxanes on the hydrocarbonaceous and the fluorinated stationary phases. However, herein lies the explanation for the reported apparent advantage of the fluorinated columns for taxane separations. The hydrocarbonaceous phase shows a higher retentivity, which to ensure sufficient peak capacity necessitates the use of eluent compositions near 50% ACN. It is around this composition that the aberrant temperature behavior of the xylosyl taxanes is (26) Snyder, L. R.; Kirkland, J. J.; Glajch, J. L. Practical HPLC Method Development, 2nd ed.; Wiley-Interscience: New York, 1998. (27) Snyder, L. R. J. Chromatogr., B 1997, 689, 105-115. (28) Zhu, P. L.; Dolan, J. W.; Snyder, L. R.; Djordjevic, N. M.; Hill, D. W.; Sander, L. C.; Waeghe, T. J. J. Chromatogr., A 1996, 756, 21-39. (29) Zhu, P. L.; Dolan, J. W.; Snyder, L. R. J. Chromatogr., A 1996, 756, 41-50. (30) Zhu, P. L.; Dolan, J. W.; Snyder, L. R.; Hill, D. W.; VanHeukelem, L.; Waeghe, T. J. J. Chromatogr., A 1996, 756, 51-62. (31) Zhu, P. L.; Dolan, J. W.; Snyder, L. R.; Djordjevic, N. M.; Hill, D. W.; Lin, J. T.; Sander, L. C.; VanHeukelem, L. J. Chromatogr., A 1996, 756, 63-72. (32) Dolan, J. W.; Snyder, L. R.; Djordjevic, N. M.; Hill, D. W.; Saunders: D. L.; VanHeukelem, L.; Waeghe, T. J. J. Chromatogr., A 1998, 803, 1-31. (33) Dolan, J. W.; Snyder, L. R.; Saunders: D. L.; VanHeukelem, L J. Chromatogr., A 1998, 803, 33-50. (34) Wolcott, R. G.; Dolan, J. W.; Snyder, L. R.; Bakalyer, S. R.; Arnold, M. A.; Nichols, J. A. J. Chromatogr., A 2000, 869, 211-230. (35) Dolan, J. W. LCGC North Am. 2002, 20, 524-530.

Figure 5. Variation of ln k′ with column temperature: (A) Ethyl-PFH, 30% ACN; (B) Propyl PFP, 30% ACN; (C) Betasil C8, 47.5% ACN.

most pronounced, and under these conditions, the separation is unlikely to succeed without careful temperature optimization, which has rarely been done. On the other hand, the fluorinated phases have lower retentivities that allow use of lower ACN concentrations. Here the aberrant temperature effect is not found, so good separations without temperature optimization are possible. But this advantage of the fluorinated phases is clearly caused by solute/mobile-phase effects, not by the nature of the bonded stationary phase. Surface Excess Isotherms. To compare further the two types of stationary phase, we determined the surface excess isotherms of ACN using the method described by Kovats et al.36 The surface excess refers to the higher concentration of ACN in the adsorbed layer of mobile-phase components on the surface of the stationary phase relative to the ACN concentration in the bulk eluent. The method is based on the measurement of the retention volumes VR of deuterated ACN, CD3CN (ACN*), and heavy water, (36) Ha, N. L.; Ungvaral, J.; Kovats, E. S. Anal. Chem. 1982, 54, 2410-2421.

D2O, which are monitored in the column effluent using a refractive index detector. For mobile-phase volume fractions of ACN and H2O, ΦACN and ΦH2O, the surface excess (µL/m2) of the ACN is given by

ΨACN ) (VR,ACN* - VR,D2O)ΦACNΦH2O/S

(1)

where S is the stationary-phase surface area. The retention volumes were measured from the retention times of the deuterated compounds and the eluent flow rate corrected for the column temperature and mean column pressure. The retention times at each eluent composition were determined using 5-µL injections each of ACN, H2O, CD3CN, D2O, and mixtures of ACN and D2O, CD3CN and H2O, CD3CN and D2O, and ACN and H2O, each mixture of the same composition as the eluent, as recommended by Kovats et al.36 Surface excess isotherms were measured at 25 and 45 °C for all the stationary phases. Representative examples are shown in Figure 8 for the Betasil C8, Propyl PFP, and Fluofix 120E columns. Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

1361

Figure 6. Effect of temperature on ln k′ on Betasil C8 at different eluent compositions. Percent ACN (v/v): A, 90; B, 70; C, 47.5; D, 35; E, 30; F, 27. Symbols: same as Figure 5.

The experimental data points are fitted with a sixth-degree polynomial function. All the isotherms show a maximum excess ACN adsorption at eluent compositions of approximately 0.4 volume fraction of ACN at 25 °C and 0.35 volume fraction of ACN at 45 °C, consistent with exothermic adsorption. The Betasil C8 1362 Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

isotherm is in very good agreement with published surface excess isotherms for hydrocarbonaceous phases.36,37 There are no published isotherms for any fluorinated phases. (37) Kazakevich, Y. V.; McNair, H. M. J. Chromatogr. Sci 1995, 33, 321-327.

Figure 7. Effect of temperature on ln k′ on Propyl-PFP at different eluent compositions. Percent ACN (v/v): A, 90; B, 80; C, 70; D, 47.5; E, 35; F, 30. Symbols: same as Figure 5.

The forms of the isotherms are the same on all the columns; the surface excess of ACN increases rapidly to ∼0.3 volume fraction, indicating preferential solvation of the bonded phase with the organic modifier in water-rich eluents. After passing through the maximum, the surface excess decreases almost linearly to

0.86-0.90 volume fraction of ACN. It becomes slightly negative at ∼0.95 volume fraction before terminating at zero in pure ACN. The negative excess can be interpreted in terms of water in the nearly pure ACN eluent being hydrophobically expelled onto the stationary phase,36-38 where it can hydrogen bond to residual Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

1363

Figure 8. Surface excess isotherms of ACN on three phases: (A) Betasil C8; (B) Propyl-PFP; (C) Fluofix 120E. Upper curves, 25 °C; lower curves, 45 °C.

silanol groups.36,38 This could account for the only significant difference between the Keystone phases and the Fluofix 120E, for which the negative excesses were about 0.01 and 0.05 µL/m2, respectively. The former were all produced from the same base silica while the imported Fluofix column is based on a silica that apparently has a larger number of residual silanols. The maximum surface excess values at 25 °C for the Keystone columns range from 0.60 µL/m2 for the Betasil C8 to 0.67 µL/m2 for the EthylPFH. Given the uncertainty in the surface areas (reported to us by Keystone39 as (10%) and variations in the packing densities of the columns, the differences are surprisingly small. The Fluofix 120E maximum surface excess was 0.53 µL/m2, which can also be attributed to a larger surface concentration of silanol groups. (38) Koch, C. S.; Ko¨ster, F.; Findenegg, G. H. J. Chromatogr. 1987, 206, 257273. (39) Henry, R. M., Keystone Scientific, private communication, 2000.

1364

Analytical Chemistry, Vol. 75, No. 6, March 15, 2003

The similarity of the surface excess isotherms on the hydrocarbonaceous and the fluorinated phases further indicates the fundamental similarity of the chromatographic behavior of these chemically different types of stationary phases. ACKNOWLEDGMENT We thank Richard M. Henry of Keystone Scientific for donating the columns and providing technical information about them, Tom Warden and Brian Starbuck of Hauser Chemical Research for providing the taxanes, and Lloyd R. Snyder for the DryLab II software.

Received for review December 31, 2002. AC020558K

September

9,

2002.

Accepted