JOURNAL OF CHEMICAL PHYSICS

VOLUME 121, NUMBER 4

22 JULY 2004

Relationship between cohesive energy density and hydrophobicity

Giuseppe Grazianoa)
del Sannio, Via PortArsa, Dipartimento di Scienze Biologiche ed Ambientali, Universita 11-82100 Benevento, Italy

Received 1 April 2004; accepted 5 May 2004 It has been recently claimed that the large cohesive energy density of water is the ultimate cause of the poor solubility of nonpolar compounds in water. In order to test the validity of this idea, we analyze the difference in solubility between light water and heavy water of several nonpolar compounds. Even though the cohesive energy density of D2 O is larger than that of H2 O, nonpolar compounds are slightly more soluble in D2 O than H2 O. In such case experimental data do not support the correctness of the large cohesive energy density as the ultimate cause of hydrophobicity. We show that D2 O is a slightly better solvent than H2 O for nonpolar compounds because it is slightly less costly to create a cavity in the former liquid. This is because there is slightly more void volume in heavy water than in light water. 2004 American Institute of Physics. DOI: 10.1063/1.1766291

I. INTRODUCTION

Hydrophobicity is the term used to indicate the poor solubility of nonpolar compounds in water. Even though this phenomenon is well recognized, its molecular origin is still strongly debated in the scientic community.1 4 In the last few years, independently, Lazaridis5 and Kodaka,6 by means of different arguments, proposed that the large cohesive energy density of water is the real cause of the poor solubility of nonpolar compounds. Actually, this explanation was rst advanced by Hildebrand7 in a small note published in 1968, and subsequently supported by Privalov and Gill,8 and by Honig and colleagues.9 The cohesive energy density10 is dened by ced ( vapH RT )/ v , where vapH is the molar vaporization enthalpy change at the temperature T and v is the molar volume of the liquid at the same temperature. The ced of water is by far the largest among all common solvents as a consequence of both the strength of hydrogen bonds and the smallness of the molar volume of water. On this basis Hildebrand stated: The fact that octane is nearly insoluble in water is merely the result of the fact that van der Waals octane-water attraction is not strong enough to penetrate the high cohesion of water. In order to be considered right, this claim should be tested in a clear-cut way. This is the aim of the present work. We would like to show that the ced argument is not of general validity because it fails to rationalize the well-known experimental nding that simple nonpolar compounds are more soluble in heavy water than in light water.11
II. RESULTS

D2 O, 12,13 and v 18.073 cm3 mol 1 for H2 O and 18.136 cm3 mol1 for D2 O. 14 Thus, at 25 C, ced(D2 O) 2365 J cm 3 and ced(H2 O) 2297 J cm 3 . These numbers indicate that D2 O has a ced larger than that of H2 O: ced(D2 O)/ced(H2 O) 1.0296. This is not a surprise since it is generally recognized that the hydrogen bonds are enthalpically stronger in D2 O than H2 O: by 0.96 kJ mol1 according to Nemethy and Scheraga,15 0.93 kJ mol1 according to Marcus and Ben-Naim,11 and 0.84 kJ mol1 according to quantum mechanical calculations by Scheiner and Cuma.16 The larger ced of D2 O, on the basis of the claim by Hildebrand, Lazaridis, and Kodaka, should imply that D2 O is a poorer solvent for nonpolar compounds than H2 O. In other words, the solubility of a nonpolar compound should be lower in D2 O than H2 O. However, experimental solubility measurements state the contrary: the solubility of simple nonpolar molecules is larger in D2 O than H2 O. 11,1719 The values of the Ben-Naim20 standard Gibbs energy change associated with the transfer from light to heavy water, G "(H2 O D2 O), listed in Table I, are all negative at room temperature. The numbers are small in view of the strict similarity of the two solvents, but the difference goes systematically into one direction. If hydrophobicity were proportional to the ced of water, the Ben-Naim standard Gibbs energy change associated with the transfer from the gas phase to heavy water could be calculated by means of the following relationship: G " D2 O calc G " H2 O ced D2 O /ced H2 O , 1 where G "(H2 O) is the experimental Ben-Naim standard Gibbs energy of hydration, whose values at 25 C for the nonpolar solutes considered are listed in the fourth column of Table I. According to Eq. 1, the values of G "(D2 O) calc are larger than those of G "(H2 O), so that their difference is always a positive quantity see the numbers listed in brackets in the fth column of Table I, in complete contrast with experimental data compare the open squares with the lled
1878 2004 American Institute of Physics

In order to clarify the role played by ced in the hydrophobicity phenomenon, we compare the solubility of several nonpolar compounds in H2 O with that in D2 O, because light and heavy water have different values of ced. At 25 C, vapH 44.0 kJ mol 1 for H2 O and 45.4 kJ mol1 for
a

FAX: 39/0824/23013; Electronic mail address: graziano@unisannio.it

0021-9606/2004/121(4)/1878/5/$22.00

Downloaded 24 Dec 2012 to 163.152.19.67. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

J. Chem. Phys., Vol. 121, No. 4, 22 July 2004

Cohesive energy density and hydrophobicity

1879

TABLE I. Hard sphere diameter, polarizability, experimental Ben-Naim standard Gibbs energy of hydration at 25 C, experimental Ben-Naim standard Gibbs energy of transfer from H2 O to D2 O at 25 C for several nonpolar compounds, and SPT estimates of G c (H2 O D2 O). The numbers reported in brackets in the fth column are the G "(H2 O D2 O) estimates obtained via Eq. 1. The values of and come from Ref. 33, except those of C3 H8 and n -C4 H10 , which come from Ref. 34. The values of G "(H2 O) and G "(H2 O D2 O) come from Table II of Ref. 11.

Ne Ar Kr CH4 C2 H6 C3 H8 n -C4 H10 O2 N2 CH4 SF6
a

3
0.393 1.63 2.46 2.70 4.33 6.29 8.12 1.56 1.73 2.86 4.48

G "(H2 O) kJ mol1 11 191a 8400a 8373b 6922b 8408b 7682b 8196c 8715c 8639b 10 265b 13 056b 12 585b

G "(H2 O D2 O) J mol1 203a 331 245a 248 144b 248 138b 204 155b 249 129b 227 96c 243 84c 258 196b 256 308b 304 259b 386 207b 375

G c (H2 O D2 O) J mol1 104 145 145 160 168 225 292 357 149 168 251 340

2.78 3.40 3.60 3.70 4.38 5.06 5.65 3.46 3.70 4.66 5.51

Reference 19. Reference 18. c Reference 17.

b

ones in Fig. 1. Experimental G "(H2 O D2 O) data do not support the claim that the large ced of water is the ultimate cause of hydrophobicity. According to well-founded statistical mechanical approaches,2123 the Ben-Naim standard Gibbs energy change associated with the transfer from light to heavy water of a nonpolar compound is given by G " H2 O D2 O G c D2 O G c H2 O E a D2 O E a H2 O , 2 where G c is the work of cavity creation and E a is the energy of turning on solute-water van der Waals interactions. As a rst approximation, it is reliable to assume that the E a term is nearly the same in D2 O as well as H2 O, because the

two isotopic species of water have very close values of the dielectric constant, refractive index, and molecular dipole moment.15 Thus, one obtains G " H2 O D2 O G c D2 O G c H2 O G c H2 O D2 O . 3

FIG. 1. Experimental Ben-Naim standard Gibbs energy of transfer from H2 O to D2 O at 25 C lled squares of several nonpolar compounds from left to right: Ne, Ar, O2 , Kr, CH4 , N2 , C2 H6 , CF4 , C3 H8 , SF6 , and n -C4 H10) vs their effective diameter; SPT estimates of G c (H2 O D2 O) at 25 C as a function of the cavity diameter continuous line; calculated values of the Ben-Naim standard Gibbs energy of transfer open squares, where G (D2 O) calc has been obtained by means of Eq. 1.

Equation 3 means that the different solubility of a nonpolar compound in D2 O with respect to H2 O should be due to the difference in the work of cavity creation. The latter quantity can be reliably calculated by means of the analytical formula provided by scaled particle theory, SPT.24,25 In performing calculations we use the experimental density at 25 C of H2 O, 0.997 05 g cm3, and D2 O, 1.104 45 g cm3, as reported by Kell;14 the pressure is xed at 1 atm, as suggested by several authors.25,26 For the size of both H2 O and D2 O molecules, we use 2.80 because this value is close to the location of the rst peak in the oxygen-oxygen radial distribution function of both light and heavy water.2729 Moreover, we30 have shown that, by assigning a diameter of 2.80 to water molecules, SPT is able to reproduce satisfactorily both the cavity size distribution and the G c values obtained from molecular dynamics simulations in TIP4P water31 by Pohorille and Pratt.32 The calculated G c values are always large and positive but, more importantly, prove to be slightly smaller in D2 O than H2 O see the numbers listed in the last column of Table I. The difference increases with the cavity size, passing from 0.12 kJ mol1 for a cavity of 3 diameter to 0.40 kJ mol1 for a cavity of 6 diameter. By plotting the experimental G "(H2 O D2 O) data versus the effective size33,34 of the nonpolar compounds listed in Table I, and the SPT estimates of G c (H2 O D2 O) versus the cavity diameter, it is evident that the latter quantity more than qualitatively accounts for the experimental data compare the lled squares with the continuous line in Fig. 1.

Downloaded 24 Dec 2012 to 163.152.19.67. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

1880

J. Chem. Phys., Vol. 121, No. 4, 22 July 2004

Giuseppe Graziano

The SPT formula shows that the magnitude of G c , at a given temperature for a given cavity diameter, is determined by two geometric properties of the solvent:35 the size of its molecules , and its volume packing density 3 N Av/6v , i.e., the ratio of the physical volume of a mole of solvent molecules over the molar volume of the solvent. The two variables work in opposite directions: for a given , G c increases on increasing , and for a given , G c increases on decreasing . This behavior can be readily explained. The quantity (1 ) is the fraction of the total volume not occupied by solvent molecules. Thus, when increases, the fraction of void volume diminishes and G c increases. On the other hand, the diameter of the solvent molecules determines the size of the pieces in which the void volume is partitioned. Keeping xed , if the diameter of the solvent molecules decreases, it diminishes the average size of the pieces in which the void volume is partitioned. This implies that the number of molecular-sized cavities, those important for the solvation of real solutes, becomes smaller and the corresponding G c values increase. It is the smallness of water molecules to render G c extremely large and positive in water with respect to the other common liquids, and eventually to cause hydrophobicity.22,23,3538 For the pair H2 O-D2 O there is no difference in the size of the molecules, but there is a small difference in the volume packing density: at 25 C, (D2 O) 0.381 66 while (H2 O) 0.382 98, because the molar volume of D2 O is slightly larger. As a consequence G c is smaller in D2 O than H2 O, and the solubility of nonpolar compounds is slightly larger in heavy water than in light water. A similar conclusion emerged in previous SPT analyses39 41 of solubility differences between the two isotopic species of water. Since we are interested in the difference between the work of cavity creation in two liquids having molecules of identical size, the SPT results have to be considered robust and reliable. According to Fig. 1, the difference in the work of cavity creation fully accounts for the experimental G "(H2 O D2 O) data in the case of solutes having a small polarizability. On the other hand, Fig. 1 shows that the experimental G "(H2 O D2 O) values become less negative on increasing the size of the solute molecule and so its polarizability, at odds with the trend of G c (H2 O D2 O). This should be reliably caused by the attractive solute-water van der Waals interactions. The volume packing density, the polarizability, and the dielectric constant of D2 O are slightly smaller than those of H2 O i.e., 1.26 3 for D2 O and 1.45 3 for H2 O in Ref. 42; see also Ref. 15, causing a reduction of the strength of van der Waals interactions. This interpretation is supported by the nding11 that, at 25 C, G "(H2 O D2 O) 129 J mol 1 for ethane ( 4.3 3 ), 0 for uoromethane ( 3.0 3 and 2.08 D), 70 J mol1 for chloromethane ( 5.4 3 and 1.89 D), and 250 J mol1 for bromomethane ( 6.0 3 and 1.82 D). Clearly, the contribution of van der Waals interactions becomes more important in the Gibbs energy balance of Eq. 2 on increasing both the polarizability and the dipole moment of the solute molecule.

III. DISCUSSION

The experimental nding that simple nonpolar compounds are slightly more soluble in D2 O than H2 O cannot be reconciled with the idea that the large ced of water is the ultimate cause of hydrophobicity. The ced of D2 O is larger than that of H2 O, so one should expect that the former liquid would be a poorer solvent for nonpolar compounds than the latter. This expectation, however, is wrong. We already pointed out that the ced argument cannot explain the trend of xenon solubility in n-alkanols at room temperature.43 In addition, experimental data show that the volume packing density decreases while the ced increases on passing from light to heavy water. Therefore, Kodakas suggestion6 of a proportionality between the volume packing density of a solvent and its ced does not seem to hold. A qualitative rationalization of G "(H2 O D2 O) data is provided by approaches that dissect the hydration process in two consecutive steps: creation of the cavity and turning on solute-water van der Waals interactions. The latter contribution, as a rst approximation, can be considered to be the same in light and heavy water, at least for solutes having a small polarizability. Assuming that there is no difference in size between D2 O and H2 O molecules, SPT calculations show that the work of cavity creation is slightly smaller in D2 O than H2 O because the former liquid is characterized by a slightly smaller value of the volume packing density. Freely speaking, there is slightly more void volume in D2 O than H2 O and this favors the insertion of nonpolar solutes. The assumption that there is no difference in size between H2 O and D2 O molecules is an important point. According to crystallographic data, the molecular dimensions and the length of a hydrogen bond are identical in H2 O and D2 O ice within a few thousandths of an angstrom unit.13,15 Similarly, the rst peak of the oxygen-oxygen radial distribution function in H2 O and D2 O occurs at the same position.27 From a theoretical point of view a difference in size between H2 O and D2 O could be expected,13,44 because the amplitude of vibrations involving deuterium is smaller than that of vibrations involving protium, as a result of the larger mass of deuterium. However, the difference in size caused by this effect is very small13 i.e., it is a second order effect affecting the vibrational uctuation about the average length of C-D and C-H bonds. On the basis of information theory results, Pratt and colleagues45,46 suggested the existence of a proportionality between the work of cavity creation in water and the reciprocal of its isothermal compressibility, G c 1/ T . Since it is generally assumed that G c , where is the surface tension of the liquid,47 and it has been recently shown that the product T is a constant for several liquids at 25 C, including water,48 the relationship G c 1/ T is expected. On this ground, Pratt and colleagues49 explained the slightly larger solubility of nonpolar compounds in D2 O with respect to H2 O by recognizing that the former liquid is slightly more compressible than the latter: at 25 C, T 46.50 10 6 bar 1 for D2 O, and 45.25 10 6 bar 1 for H2 O. 50 Clearly, T is a macroscopic thermodynamic quantity and a microscopic interpretation of the difference between D2 O and H2 O should be provided.

Downloaded 24 Dec 2012 to 163.152.19.67. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions

J. Chem. Phys., Vol. 121, No. 4, 22 July 2004

Cohesive energy density and hydrophobicity

1881

Pratt and colleagues45 suggested that the small value of T for H2 O with respect to the other common solvents should be a reection of the stiffness of its hydrogen bonding network i.e., the strength of hydrogen bonds. This interpretation, however, does not seem to hold. Even though the hydrogen bonds in D2 O are enthalpically stronger than those in H2 O11,15,16 see also Appendix A, T is larger for D2 O than for H2 O. The isothermal compressibility of a system is a measure of the ensemble uctuations in local number density:51 T ( v / RT ) ( 2 2 )/ 2 , where the number density N Av / v , the angled brackets denote ensemble average values, v is the molar volume, and R is the universal gas constant. The small value of T in the case of H2 O should be a reection of the fact that the molar volume of H2 O is signicantly smaller than that of common solvents: at 25 C, v (in cm3 mol 1 ) 18.07 for water, 89.41 for benzene, 108.75 for c-hexane, 131.62 for n-hexane, and 97.09 for carbon tetrachloride.14,33 The molecular origin of this is the fact that the size of water molecules is the smallest among all other solvents.25,33 We suggest that the isothermal compressibility of D2 O is slightly larger than that of H2 O mainly because the molar volume of D2 O is slightly larger than that of H2 O see also Appendix B. Assuming that the molecular sizes of H2 O and D2 O are the same, this means that the volume packing density of D2 O is slightly smaller than that of H2 O. In other words, D2 O has a slightly larger void volume than H2 O and so it proves to be more compressible. It is evident that this explanation corresponds to that used to rationalize the nding that G c is slightly smaller in D2 O than H2 O. The present analysis, even though rudimentary in some respects, is able to provide a rationalization of the experimental G "(H2 O D2 O) data for several nonpolar compounds. The latter represent a robust test to distinguish a physically correct interpretation of hydrophobicity from the noncorrect ones. Since the G "(H2 O D2 O) data cannot be rationalized using the idea that the large ced of water is the cause of hydrophobicity, this idea should not be considered correct. The large unfavorable entropy change determining hydrophobicity around room temperature has nothing to do with the large ced of water. The process of cavity formation provides the correct explanation: the negative entropy change due to the excluded volume effect is exaggerated in water by the small size of individual water molecules.
ACKNOWLEDGMENTS

At 25 C, P int 141.8 J cm 3 for D2 O ( 0.221 10 3 K 1 ), and 169.5 J cm3 for H2 O ( 0.257 10 3 K 1 ). Thus, even though ced(D2 O) ced(H2 O), it results P int(D2 O) P int(H2 O), conrming that ced and P int account for different properties of the liquids, especially when hydrogen bonds are present.52 Following Dacks proposal, one obtains ced(D2 O) P int(D2 O) 2223.2 J cm 3 , and ced(H2 O) P int(H2 O) 2127.5 J cm 3 ; these numbers indicate that hydrogen bonds are stronger in D2 O than H2 O by 95.7 J cm3. If the latter number is multiplied by the mean molar volume, 18.1 cm3 mol1, and divided by 2, the mean number of hydrogen bonds per water molecule, one obtains that the difference in strength amounts to 866 J per mole of hydrogen bond at room temperature. This estimate agrees with those obtained by more rigorous treatments.11,15,16 It is worth noting that, at 25 C, a fraction of hydrogen bonds is broken;53 this fraction should be different in the two isotopic species of water,41,54 but such difference should be so small to be neglected in a simple calculation as the present one.
APPENDIX B: RELATIONSHIP BETWEEN MOLAR VOLUME AND STRENGTH OF HYDROGEN BONDS

This work was supported by Grant COFIN-2002 from the Italian Ministry for Instruction, University and Research M.I.U.R., Rome.
APPENDIX A: AN ESTIMATE OF THE DIFFERENT STRENGTH OF HYDROGEN BONDS IN D2 O AND H2 O

In general, the molar volume of a liquid is determined by the size of its molecules and the strength of intermolecular interactions. In the case of water, however, characterized by the presence of strongly directional interactions such as the hydrogen bonds, one has to make different considerations. A perfect tetrahedral arrangement leads to an increase of molar volume. The transition from liquid water to ice at 0 C is accompanied by an increase of molar volume,13 because there is an increase in the number of molecules with a perfect tetrahedral coordination. The coordination number is 4 in ice whereas its average value in liquid water is 5, due to the presence of an interstitial molecule that is not hydrogen bonded to the central one.2729 On this basis, the increase in molar volume passing from H2 O to D2 O and T2 O13 should reect the increase in the strength of hydrogen bonds that, in turn, leads to an increase in the number of molecules with perfect tetrahedral coordination. In other words, the structural arrangement of D2 O and T2 O molecules in the liquid state is more icelike than that of H2 O molecules. Thus, the increased strength of hydrogen bonds in D2 O with respect to H2 O manifests itself not only in the increase of ced but also in the increase of molar volume. Since the molecular dimensions are identical for H2 O and D2 O molecules at room temperature, the slightly larger molar volume of D2 O means that there is slightly more void volume in D2 O than in H2 O. This fact renders less costly the cavity creation in heavy water, favoring the solubility of nonpolar compounds.
1

Dack52 proposed that, in the case of liquids characterized by the presence of hydrogen bonds, the quantitative difference between ced and P int ( U / V ) T T / T , where U is the internal energy and is the coefcient of thermal expansion, should be a measure of the strength of hydrogen bonds.

W. Blokzijl and J. B. F. N. Engberts, Angew. Chem., Int. Ed. Engl. 32, 1545 1993. 2 L. R. Pratt, Annu. Rev. Phys. Chem. 53, 409 2002. 3 N. T. Southall, K. A. Dill, and A. D. J. Haymet, J. Phys. Chem. B 106, 521 2002. 4 B. Widom, P. Bhimalapuram, and K. Koga, Phys. Chem. Chem. Phys. 5, 3085 2003. 5 T. Lazaridis, Acc. Chem. Res. 34, 931 2001. 6 M. Kodaka, J. Phys. Chem. B 105, 5592 2001; 108, 1160 2004.

Downloaded 24 Dec 2012 to 163.152.19.67. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions