Anal. Chem. 1996, 68, 2869-2873
Retention Mechanism of Weak Polar Solutes in Reversed Phase Liquid Chromatography Yves Claude Guillaume* and Christiane Guinchard
Laboratoire de Chimie Analytique, Faculte´ de Me´ decine et Pharmacie, Place Saint Jacques, 25030 Besancon Cedex, France
The retention mechanism of a weak polar solute, a series of 10 benzodiazepines in reversed phase liquid chromatography, was investigated over a wide range of mobile phase compositions. The values of enthalpy (∆H°) and entropy (∆S°) of transfer from the mobile to the stationary phases were determined. The method studied each factor (water fraction Φ in the acetonitrile (ACN)/water mixture and column temperature) controlling the retention mechanism. The changes in ∆H° and ∆S° as a function of the water fraction Φ in the ACN/water mixture were examined. These variations are explained using the organization of organic modifier (ACN) in clusters in the ACN/ water mixture. A change in the retention mechanism thus indicated when the ACN/water mixture was used instead of the hydrogen-bonded mobile phase such as CH3OH/ water. Enthalpy-entropy compensation revealed that the retention mechanism was independent of the water fraction Φ but showed that differences between the molecular structures of the benzodiazepines contributed more significantly to changes in the retention process in the CH3OH/water mixture than in the ACN/water mixture. Reversed phase liquid chromatography (RPLC) has been widely studied in recent years,1 yet the retention mechanism in RPLC remains unclear. An explanation of solute retention has been described qualitatively through the hydrophobic effect2-4 or by the use of the partitioning model.5-8 The hydrophobic effect is used particularly to explain solute bonded phase interactions. For weak polar solutes, most entropies of transfer from the mobile phase to the stationary phase are negative. This phenomenon has been attributed to an ordering of water molecules adjacent to the surface of the solute molecules. This hydrophobic effect can be described as the tendency of a weak polar solute to reduce its surface area exposed to water through association with another weak polar solute or its disappearance from the solution by adsorption, thereby increasing the entropy of the system. Cole and Dorsey9 have demonstrated that the hydrophobic effect is a satisfactory explanation of the retention process in RPLC when (1) Majors, R. E. LC-GC 1988, 6, 298. (2) Horvath, Cs.; Melander, W. R.; Molnar, I. J. Chromatogr. 1976, 125, 129. (3) Melander, W. R.; Horvath, Cs. In High Performance Liquid Chromatography, Advances and Perspectives; Horvath, Cs., Ed.; Academic Press: New York, 1980; Vol. 2, p 113. (4) Horvath, Cs.; Melander, W. R. Am. Lab. 1978, 17. (5) Martire, D. E.; Boehm, R. E. J. Phys. Chem. 1983, 87, 1045. (6) Dill, K. A. J. Phys. Chem. 1987, 91, 1980. (7) Dill, K. A.; Naghizadeh, J.; Marquisee, J. A. Annu. Rev. Phys. Chem. 1988, 39, 425. (8) Dorsey, J. G.; Dill, K. A. Chem. Rev. 1989, 89, 33. (9) Cole, L. A.; Dorsey, J. G. Anal. Chem. 1992, 64, 1324. S0003-2700(96)00141-2 CCC: $12.00
© 1996 American Chemical Society
the hydrogen-bonded and highly aqueous mobile phases are used. For the retention process in other situations, this is not an adequate explanation. The liquid-liquid partitioning process postulates the existence of two liquid phases. The solute concentration in each phase is governed by its distribution coefficient. The solute molecule in this model is embedded within the stationary phase.8 In an adsorption, the solute molecule is only in surface contact with this phase.8 The effect of the mobile phase composition on the selectivity and retention mechanism in RPLC has been extensively examined. Empirical modeling approaches for the ternary mobile phase have been published by Kowalska10-12 and Xie et al.13 A thermodynamic treatment was used to model the sorption of solvents into the bonded phases.14,15 Absolute enthalpies and entropies were determined to elucidate the retention behavior of solute molecules. With a narrow fraction of water Φ in the acetonitrile (ACN)/water mixture (0 e Φ e 0.35), Stalcup et al.16 studied the retention behavior of polycyclic aromatic hydrocarbons. They found that the enthalpic contribution is independent of composition in the high fraction of ACN. Stalcup et al.16 showed that the relative entropy contribution to the overall solute retention increases with decreasing water content in an ACN/water mixture. Guillaume and Guinchard17 found a similar increase for less hydrophobic solutes, such as 10 benzodiazepines, when CH3OH was used instead of ACN. It is useful to extend the discussion to the mobile phase containing a high water fraction. Chromatographic systems involving highly aqueous mobile phases, particularly those containing >90% water, have been the most difficult to study. This is probably due to the much greater effect of the mobile phase on the retention mechanism in the water-rich mobile phase, which was attributed to the solvation of the stationary phase by the organic component of the mobile phases. The thermodynamic behavior of 10 benzodiazepines was investigated in the ACN/water mixture over a wide range of water fractions Φ (0.2 e Φ e 0.7). The enthalpies (∆H°) and entropies (∆S°) of transfer of these compounds from the mobile to the stationary phases were determined for the first time for mixtures containing >90% water. To understand the dependence of ∆H° and ∆S° on the water fraction Φ, a model which takes into account the structure (organization) of the solvent mixture was developed. (10) Kowalska, T. Chromatographia 1989, 28 (7-8), 354. (11) Kowalska, T. Chromatographia 1989, 27 (7-8), 389. (12) Kowalska, T. Chromatographia 1991, 31 (3-5), 119. (13) Xie, M.; Zhou, C.; Ren, Z.; Luo, T. J. Chromatogr. 1991, 555 (1-2), 33. (14) Jaroniec, M.; Lin, S.; Gilpin, R. K. Chromatographia 1991, 32 (1-2), 1318. (15) Gilpin, R. K.; Jaroniec, M.; Lin, S. Chromatographia 1990, 30 (7-8), 393. (16) Stalcup, A. M.; Martire, D. E.; Wise, S. A. J. Chromatogr. 1988, 442, 1. (17) Guillaume, Y.; Guinchard, C. J. Liq. Chromatogr. 1994, 17, 2809.
Analytical Chemistry, Vol. 68, No. 17, September 1, 1996 2869
Figure 1. Chemical structures: oxazepam (1), tofisopam (2), diazepam (3), clorazepate dipotassic (4), chlordiazepoxide (5), flunitrazepam (6), clobazam (7), bromazepam (8), nitrazepam (9), and lorazepam (10).
EXPERIMENTAL SECTION Apparatus. The HPLC system consisted of Waters HPLC pump 501 (Saint-Quentin, Yvelines, France), an Interchim rheodyne injection valve, Model 7125 (Montluc¸on, France), fitted with a 20 µL sample loop, a Merck L4000 variable-wavelength UV spectrophotometer detector at 254 nm, and a Merck D2500 chromatointegrator (Nogent-sur-Marne, France). A Waters 150 mm × 3.9 mm i.d. RP18 column (Nova Pak, 5 µm particle size) was used with a controlled temperature in an Interchim Crocil oven, 701. Mobile phase flow rate for most experiments varied from 0.5 to 1.6 mL/min. Solvents and Samples. HPLC grade acetonitrile (Merck) was used without further purification. Water was obtained from an Elgastat option I water purification system (Odil, Talant, France), fitted with a reverse osmosis cartridge. The mobile phase used for these studies was an acetonitrile/water mixture. The range of the water fraction (v/v) was 0.2-0.70. Oxazepam (1), tofisopam (2), diazepam (3), clorazepate dipotassic (4), chlordiazepoxide (5), flunitrazepam (6), clobazam (7), bromazepam (8), nitrazepam (9), and lorazepam (10), obtained from Hoffmann LaRoche (Basel, Switzerland), were prepared using acetonitrile to a final concentration range of 10-80 mg/mL. The chemical structures of these compounds are given in Figure 1. Each solute, or a mixture of these when the 10 compound peaks were well resolved, was injected, and the retention times were measured using a Merck D2500 chromatointegrator. Temperature Studies. Compound retention factors were determined over the temperature range 25-50 °C. The chromatographic system was allowed to equilibrate at each temperature for at least 1 h prior to each experiment. To study this equilibration, the compound retention time of the compound flunitrazepam was measured every hour for 7 h and again after 22, 23, and 24 h. The maximum relative difference in the retention times of this compound between these different measurements was always 0.6%, making the chromatographic system sufficiently equilibrated for use after 1 h. 2870
Analytical Chemistry, Vol. 68, No. 17, September 1, 1996
METHODS Thermodynamic Relationships. Solute retention is usually expressed in terms of the retention factor k′ by the equation
ln k′ ) -
∆H° ∆S° + + ln γ RT R
(1)
where ∆H° (respectively ∆S°) is the enthalpy (respectively entropy) of transfer of the solute from the mobile phase to the stationary phase, T, the temperature, R the gas constant, and γ the phase ratio (volume of the stationary phase divided by the volume of the mobile phase). This equation shows that the plot of ln k′ versus 1/T (called a van’t Hoff plot) has a slope of -∆H°/R and an intercept of ∆S°/R + ln γ. This provides a convenient way of calculating the thermodynamic constants ∆H° and ∆S° for a chromatographic system if the phase ratio is known or can be calculated. Usually, ∆S° is not provided due to the ambiguity in the calculation of the phase ratio for commercial columns. In this work, to estimate the phase ratio γ with a physical constant of the packing material, a physical model was used that links the carbon loading percentage with other properties of the stationary phase.18 RESULTS AND DISCUSSION According to eq 1, van’t Hoff plots were obtained for all solutes. The correlation coefficients for the linear fit were in excess of 0.999. The typical standard deviations of slope and intercept were respectively 0.006 and 0.04. Considering the solute flunitrazepam, with a water fraction equal to 0.40, ∆H° was found to be equal to -3.27 kcal mol-1 and ∆S° was calculated to be -4.23 cal mol-1 K-1. These values for ∆H° and ∆S° again agree with values reported in the literature.19-21 Both ∆H° and ∆S° were negative, as is the case with reversed phase liquid chromatography. As ∆H° became more negative, the solute was retained longer, and (18) Sentell, K. B.; Dorsey, J. G. J. Liq. Chromatogr. 1988, 11, 1875. (19) Gruschka, E.; Colin, H.; Guiochon, G. J. Chromatogr. 1982, 248, 325. (20) Yamamoto, F. M.; Rokushika, S.; Hatano, H. J. Chromatogr. Sci. 1989, 27, 704. (21) Sander, L. C.; Field, L. R. Anal. Chem. 1980, 42, 2009.
Figure 2. Plot of the enthalpy of solute transfer, ∆H° (kcal mol-1), vs the volume fraction of water, Φ, in the ACN/water mixture for the solute flunitrazepam.
Figure 3. Plot of the entropy of solute transfer, ∆S° (cal mol-1 K-1), vs the volume fraction of water, Φ, in the ACN/water mixture for the solute flunitrazepam.
as ∆S° became more negative, the solute was retained less. Evidence for this feature was provided by a thermodynamic study of the retention mechanism in RPLC. The solute molar enthalpy associated with the stationary phase was expected to be lower than that with the mobile phase because of the formation in the mobile phase of strong solvent-solvent interactions. These interactions also gave a higher solute molar entropy associated with the mobile phase relative to the stationary phase. Figures 2 and 3 show ∆H° and ∆S°, respectively, versus water fraction Φ in the ACN/water mixture for the solute flunitrazepam. These variations were the same for all the other benzodiazepines. When Φ was inferior or equal to a critical value Φc, ∆H° was constant, and when Φ g Φc, ∆H° decreased. ∆S° was found to exhibit a parabolic variation, with a maximum for Φc. The plots of ∆H° and ∆S° can be expressed by the following equations:
in relation to Φ, a descriptive model is proposed, based on the particular way that ACN is organized in clusters16 in the ACN/ water mixture or loosely defined clusters containing n ACN molecules, (ACN)n. Evidence for this is provided by temperaturedependent studies of ACN in a variety of solvents, which indicate self-association in a water mixture.22,23 The solute molecule was solvated in microfonds of ACN due to the existence of microheterogeneities in a mixture of ACN/water. The ACN molecules increased the solubility of the solute molecule in aqueous solution. These phenomena can be experimentally explained on the basis of the fact that a solute is, to a large extent, subtracted from the microscopic observation in an ACN/water mixture. The equilibrium between ACN and acetonitrile clusters (ACN)n can be represented by
∆H° ) a1Φ + a2
for Φ g Φc
(2)
nACN h (ACN)n The corresponding equilibrium constant is given by
∆H° ) a3 for Φ e Φc
(3)
K ) [(ACN)n]/[ACN]n
∆S° ) b1Φ2 + b2Φ + b3 for Φ e Φc, Φ g Φc
(4)
If [ACN]0 is the total concentration of ACN in the ACN/water mixture, then
where a1, a2, a3, b1, b2, and b3 are constants given in Table 1. The coefficients of determination, r2, for these fits were always greater than or equal to 0.986. To explain the variations of ∆H° and ∆S°
(5)
(22) Lowenschuss, A.; Yellin, N. Spectrochim. Acta 1975, 31A, 207. (23) Rowlen, K. L.; Harris, J. M. Anal. Chem. 1991, 63, 964.
Analytical Chemistry, Vol. 68, No. 17, September 1, 1996
2871
Table 1. Values of the coefficients from Eqs 2-4
a
eqs 2 and 3
eq 4
compd no.a
a1
a2
a3
r2 b
b1
b2
b3
r2 b
1 2 3 4 5 6 7 8 9 10
-9.80 -10.13 -10.78 -10.83 -10.22 -9.83 -9.99 -9.64 -9.80 -10.00
2.36 2.36 2.11 2.21 2.11 2.62 2.67 2.79 2.69 2.59
-2.30 -2.51 -3.08 -3.00 -2.79 -2.09 -2.13 -1.82 -2.01 -2.21
0.986 0.987 0.990 0.989 0.986 0.989 0.987 0.993 0.991 0.990
-13.24 -15.31 -16.87 -15.84 -13.95 -11.83 -11.54 -13.41 -13.33 -12.56
14.04 15.92 18.22 17.11 15.35 13.02 12.93 14.48 14.53 13.82
-8.10 -8.22 -8.78 -8.72 -8.50 -7.78 -7.84 -7.58 -7.68 -7.84
0.987 0.988 0.989 0.990 0.991 0.997 0.988 0.989 0.990 0.991
See compound numbers and chemical structures in Figure 1. b r ) correlation coefficients of the fits.
[ACN]0 ) [ACN] + n[(ACN)n]
(6)
If y is the fraction of “free” ACN (not in a cluster),
y ) [ACN]/[ACN]0
(7)
Combining eqs 5-7, the y value can be expressed by
y + nK[ACN]0n-1yn ) 1
(8)
d is acetonitrile density, 1 - Φ is the ACN fraction in the ACN/ water mixture, and Mw is the molecular weight of ACN,
[ACN]0 ) d(1 - Φ)/Mw
(9)
Rewriting eq 8, the following is obtained:
y + λn(1 - Φ)n-1yn ) 1
(10)
with
λ ) Kdn-1/Mwn-1 where λ is a constant. n was determined to be equal to 120. Figure 4 shows the variation of y, representing “free” ACN in the ACN/water mixture versus the fraction of water Φ. This curve presents an inflection for a critical value of the water fraction Φc (=0.52). From this model, the following conclusions can be drawn. (i) For a Low Water Fraction Φ (Φ e Φc). In this region, most of the ACN molecules were organized in clusters or solvating the stationary phase. A weak polar solute such a benzodiazepine molecule is solvated by these clusters. The solute mobile phase environment as well as ∆H° remains relatively constant when Φ increase (Figure 2). The entire mobile phase could not be accessible to the solute. The benzodiazepine molecule decreased its surface area exposed to water either through solvation into clusters or through its removal from the solution by adsorption. This would thus explain the observed trend of ∆S° to positively increase when Φ increases (Figure 3). In this water region, ∆H° was constant but ∆S° increased. Therefore, it can be said that the retention mechanism is controlled entropically. (ii) For a High Water Fraction Φ (Φ g Φc). In this region, the number and size of clusters diminished (Figure 4). A weak polar solute such as benzodiazepine molecule comes into less and less contact with ACN and unavoidably into increasingly more contact with water; thus, ∆H° and ∆S° decreased when Φ increased (Figures 2 and 3). This would thus explain that the 2872 Analytical Chemistry, Vol. 68, No. 17, September 1, 1996
Figure 4. Plot of the fraction y ) [ACN]/[ACN]0 representing “free” ACN (not in a cluster) vs the volume fraction of water, Φ, in the ACN/ water mixture.
variation curve of ∆S° in relation to Φ had an optimum value for Φc. It was thought useful to determine thermodynamic data for Φ ) 0.92. The ∆H° and ∆S° values calculated from eqs 2 and 4 are given in Table 2 for each benzodiazepine. The relative differences between data obtained from the van’t Hoff plot were less than 10%. Enthalpy-Entropy Compensation Studies. Investigation of the enthalpy-entropy compensation temperature is an extra thermodynamic approach to the analysis of physicochemical
Table 2. Values Calculated from Eqs 2 and 4 and Experimental As Per a van’t Hoff plot of ∆H° and ∆S° for the 10 Compounds for a Mobile Phase with Φ ) 0.92 compd no.a 1 2 3 4 5 6 7 8 9 10
∆H° (kcal mol-1) calcd exptl eb -6.66 -6.96 -7.81 -7.75 -7.29 -6.42 -6.52 -6.07 -6.32 -6.61
-7.01 -7.21 -8.09 -8.25 -7.71 -6.84 -7.04 -6.57 -6.82 -7.09
∆S° (cal mol-1 K-1) calcd exptl eb
4.99 3.47 3.46 6.06 5.45 6.14 7.39 7.61 7.33 6.77
-6.40 -6.53 -6.29 -6.38 -6.18 -5.82 -5.71 -5.60 -5.59 -5.75
-6.80 -6.71 -6.80 -6.81 -5.97 -6.13 -6.09 -6.06 -6.01 -5.95
5.88 2.68 7.50 6.31 3.32 5.06 6.24 7.59 6.98 3.36
a See compound numbers and chemical structures in Figure 1. b e ) relative difference between calculated and experimental values.
data.21,25-29 Mathematically, enthalpy-entropy compensation can be expressed by the formula
(11)
∆H° ) β∆S° + ∆G°β
where ∆G°β is the Gibbs free energy of a physicochemical interaction at a compensation temperature β (β and ∆G°β are constants). ∆H° and ∆S° are respectively the corresponding standard enthalpy and entropy. According to eq 11, when enthalpy-entropy compensation is observed with a group of compounds in a particular chemical interaction, all of the compounds have the same free energy (∆G°β) at temperature β. If, therefore, enthalpy-entropy compensation is observed for the 10 compounds, all will have the same net retention at the compensation temperature β, although their temperature dependencies may differ. Rewriting eq 11 using eq 1 gives
ln kT′ ) ln kβ′ -
∆H° 1 1 R T β
(
)
(12)
where
ln kβ′ ) -
∆G°β + ln γ Rβ
Equation 12 shows that, if a plot of ln kT′ against -∆H° is linear, then the 10 solutes are retained by an essentially identical interaction mechanism. A plot of ln kT′ (T ) 310 K), calculated for each of the 10 compounds, against -∆H° for two different values of the water fraction Φ was drawn (Φ1 e Φc and Φ2 g Φc). The r2 values plotted were 0.780 for Φ1 ) 0.40 and 0.778 for Φ ) 0.60. This can be considered to be adequate to verify enthalpy-entropy compensation. Therefore, the retention mechanism is the same (24) Marcul, Y.; Migran, Y. J. Phys. Chem. 1994, 95, 400. (25) Tchapla, A.; Heron, S.; Colin, H.; Guiochon, G. Anal. Chem. 1988, 60, 1443. (26) Yamamoto, F. M.; Rokushika, S.; Hatano, H. J. Chromatogr. Sci. 1989, 27, 704. (27) Kucher, M.; Kraus, E.; Rejholec, V.; Miller, V. J. Chromatorgraphy 1988, 449, 391. (28) Melander, W. R.; Campbell, D. E.; Horvath, Cs. J. Chromatogr. 1978, 158, 215. (29) Boots, H. M. J.; De Bokx, P. K. J. Phys. Chem. 1980, 93, 8240. (30) Katz, E. D.; Ogan, K.; Scott, R. P. W. J. Chromatogr. 1986, 352, 67. (31) Snyder, L. R. J. Chromatogr. Sci. 1978, 16, 223. (32) Glajch, J. L.; Kirkland, J. H.; Squire, K. M.; Minor, J. M. J. Chromatogr. 1980, 57, 199.
for these solutes using this chromatographic system. Enthalpyentropy compensation was also used to test the retention mechanism of a benzodiazepine molecule when the water fraction Φ in the ACN/water mixture changed. A plot of ln kT′ (for T ) 310 K) versus -∆H° was tested for each of the 10 compounds when the water fraction varied from Φ g Φc (Φc = 0.52) to 0.70 at a temperature of 310 K. All correlation coefficients were at least equal to 0.988. The high degree of correlation indicates that the retention mechanism for a solute was the same whatever the mobile phase composition. In a previous paper,17 the thermodynamic behavior of these benzodiazepines in a CH3OH/water mixture was studied. The retention mechanism in that mixture was found to be independent of the water fraction Φ. Using the data from that paper,17 a plot of ln kT′ was tested at T ) 310 K for each of the 10 compounds against -∆H° for two Φ values (Φ ) 0.2 and 0.5). The two correlation coefficients, r2, had a maximum equal to 0.210. Thus, in this CH3OH/water mixture, entropy-enthalpy compensation did not exist, which proves the dependence of the benzodiazepine molecular structure in this CH3OH/water chromatographic system. All these results indicate a difference between the ACN/ water and CH3OH/water mixtures. This conclusion is supported by the fact that the methanol solution is dominated by competitive hydrogen bonding and that the availability of “free” methanol for solute solvation decreases rapidly with an increasing water fraction.30 However, ACN solution chemistry is governed by clusters of ACN, where weak polar solutes are preferentially solvated. This fact confirms the usefulness of Snyder’s polarity approach,31 as used by Glajch et al.,32 and the simplified solubility parameter model. These models show that ACN/water and CH3OH/water are isoeluotropic binary mixtures. CONCLUSIONS In this paper, the retention mechanism in RPLC was studied for less hydrophobic solutes, such as 10 benzodiazepines. The thermodynamic property trends were determined over a wide range of water fractions Φ in the ACN/water mixture. The model proposed explains these variations in terms of organization of the organic modifier (ACN) in clusters in the ACN/water mixture. The results obtained and enthalpy-entropy compensation demonstrated that the retention mechanism of these 10 benzodiazepines is independent of the water fraction Φ in the ACN (or CH3OH)/water mixture. It also revealed that differences between their molecular structures contribute more significantly to changes in the retention mechanism in a CH3OH/water mixture than in an ACN/water mixture. ACKNOWLEDGMENT We are indebted to Professors G. Guiochon (University of Tennessee, Knoxville, TN), M. Porthault (Universite´ Claude Bernard, Lyon, France), and A. Siouffi (Universite´ Saint Je´rome, Aix-Marseille, France) for their helpful support. We are grateful to Dr. C. Zhiri for her mathematical advice. Received for review February 14, 1996. Accepted June 6, 1996.X AC960141C
X
Abstract published in Advance ACS Abstracts, July 15, 1996.
Analytical Chemistry, Vol. 68, No. 17, September 1, 1996
2873
