Solvation Entropy Made Simple Arxiv:1901.11128V1 [Physics.Chem

Home , Molecularity, Solvation, Solvent effects

Alejandro J. Garza∗

The Dow Chemical Company, 1776 Building Midland, Michigan 48674, United States

Abstract The entropies of molecules in solution are routinely calculated using gas phase formulae. It is assumed that, because implicit solvation models are fitted to reproduce free energies, this is sufficient for modeling reactions in solution. However, this procedure exaggerates entropic effects in processes that change molecularity. Here, computationally efficient (i.e., having similar cost as gas phase entropy calculations) approximations for determining solvation entropy are proposed to address this issue. The Sω, S, and Sα models are nonempirical and rely only on physical arguments and elementary properties of the medium (e.g., density and relative permittivity). For all three methods, average errors as compared to experiment are within chemical accuracy for 110 solvation entropies, 11 activation entropies in solution, and 32 vaporization enthalpies. The models also make predictions regarding microscopic and bulk properties of liquids which prove to be accurate. These results imply that ∆Hsol and ∆Ssol can be described separately and with less reliance on parametrization by a combination of the methods presented here with existing, reparametrized implicit solvation models.

Introduction isfactory as it inevitably underestimates ∆Sreac effects and can result in unphysical negative It is standard practice in computational chem- free energy barriers (a recent publication10 dis- istry to calculate the free energy of molecules cusses this and related ad hoc methods to com- in solution utilizing gas phase entropies.1,2 Gas pute entropies in solution from entropies in the phase entropies are well described by analyti- gas phase). Another workaround is to scale gas cal formulas of statistical mechanics, but there phase entropies by a rule-of-thumb factor of ≈ is no similarly efficient way of estimating en- 0.65.1,11–14 While scaling gas phase entropies tropies in solution rigorously. However, the may be reasonable for small molecules, it is not use of gas phase entropies to model processes justified for larger molecules for which the vi- that change molecularity (e.g., adsorption and brational and cavity, rather than translational binding) in solution often exaggerates entropy and rotational, entropy terms dominate (there changes during the reaction, ∆S .1–8 In con- reac is no reason for which vibrations should change densed media, the translational and rotational arXiv:1901.11128v1 [physics.chem-ph] 30 Jan 2019 drastically upon solvation; numerical results motions of a molecule are hindered, reducing support this view15). Implicit solvation mod- the entropy as compared to the gas phase. Re- els are often fitted to reproduce experimental arrangement of the solvent to form a cavity for free energies under standard conditions,2,16–20 the solute further lowers the entropy of the sys- so one could expect them to improve ∆G . tem. The typical error in ∆S for bimolec- reac reac However, unless a model designed to account ular reactions in solution is so large that some for temperature effects is used and appropriate authors have suggested to completely neglect derivatives are taken (see, e.g., refs. 21,22), the translational and rotational entropy5,6 (or just fact that the T ∆S term in ∆G relies on the former9). This approach is, however, unsat- reac reac gas phase entropies will result in an incorrect

1 temperature dependence of the reaction. Fur- Theory thermore, in actual applications, implicit solva- The total entropy of an atom or molecule in tion models often do not improve the too-high solution is decomposed into contributions from binding free energies caused by the use of gas vibrations, translations, rotations, and the sol- phase entropies.1–9 Molecular dynamics and al- vent cavity: chemical free energy methods can in principle determine free energies in solution without re- S = Sv + St + Sr + Sc. (1) lying on gas phase formulas.23–25 Nonetheless, such calculations require significant additional The vibrational entropy can be computed from work from the user and are computationally de- the harmonic oscillator approximation in the manding, which makes them unsuitable for rou- same way as in the gas phase. It is well known, tine or high-throughput calculations. They also however, that this approximation yields un- suffer from inherent reproducibility issues due physically large contributions to the entropy to the sensitivity of Newtonian dynamics to ini- from low-frequency modes. To avoid this is- tial conditions and the lingering effects of such sue, Sv is calculated with the method proposed conditions if sampling is insufficient.24,25 by Grimme,26 which can be seen as a quasi hin- The purpose of this work is to formulate an dered rotor that interpolates between the har- efficient approximation for calculating molecu- monic oscillator and free rotor entropy formu- lar entropy in solution based only on physical las. One can, however, compute Sv with any ap- and geometric arguments. Three models, Sω, proximation deemed appropriate as Sv does not S, and Sα, are derived that rely only on such influence solvation for the methods presented arguments—no empirically fitted parameters— here. and elementary solvent properties (e.g., mass density). The methods differ only in how the Translational Entropy cavitation entropy is calculated: Sω, S, and Sα utilize, respectively, the Pitzer acentric fac- The contributions from St in terms of the trans- 27,28 tor (ω), the relative permittivity (r), and r lational partition function are as well as the isobaric thermal expansion coef- ∂ ln (q ) ficient (α). This establishes a connection be- S = k ln (q ) + k + kT t , (2) t t ∂T tween ω (a microscopic property), r, and α V (macroscopic properties). The cost of evaluat- where q is approximated by the familiar ex- ing entropy with these models is comparable to t pression obtained from the eigenenergies of a the cost of calculating gas phase entropies with particle of mass m confined in a volume V : the ideal gas/rigid rotor/harmonic approximation. Their accuracy is tested by constructing 2πmkT 3/2 a database of 110 experimental solvation en- qt = V. (3) h2 tropies and 11 activation entropies in solution; the resulting average errors are in the range of All quantities in eq. 3 are unambiguously de- 2–3 cal/mol-K, which is comparable to the ac- fined except for V . For an ideal gas V = kT/P , curacy of gas phase entropies and within what is but in condensed media V will depend on prop- considered chemical accuracy (≤ 1 kcal/mol at erties of the medium such as, e.g., its density 300 K). Additionally, the models make testable and particle volume. Here, we define V in terms predictions regarding molecular and bulk prop- of the volume of the solute cavity, vc, as well as erties of liquids that prove to be in agreement the average number of accessible cavities Nc, with experiment. V = Ncvc, (4)

so that we can evaluate V based on a physical interpretation of Nc and vc. In our model, vc is

2 the volume of a sphere with a radius equal to Typically, Nc ≈ 1, but the hopping terms can the sum of the spherical equivalent radii of the be important in cases of small solutes in bulky solute and the volume of free space per solvent or low density solvents. Note that previous particle. This definition is equivalent to works have also utilized a cavity volume to 30 determine St. However, the definition of vc 3 1/3 1/3 and consideration of the possibility of hopping vc = VM + Vfree (5) distinguish the present approach from previous with ones. S Mw Although we have written in eq. 1 separate Vfree = − VS, (6) NAρ terms for St, Sr, and Sc, these are actually in- tertwined. As we see next, the definition of St where VS/M is the volume of a solvent/solute molecule, ρ the mass density of the medium, is important to determine Sr, and Sr is in turn used to derive an approximation for Sc. The NA Avogadro’s number, and Mw the molecular weight (throughout this document, sub- way to think about St to more easily under- scripts/superscripts “M” and “S” denote solute stand Sr as conceptualized here is simply as the entropy of a point particle in a box, as opposed and solvent, respectively). Here, VM and VS are determined from van der Waals radii,29 though to the entropy of an object that has rotations. vc is relatively insensitive to how molecular volumes are defined (vide infra). Rotational Entropy To determine Nc, we define the probability We define Sr in terms of the contributions from per solvent particle of “hopping” to an adjacent the rigid rotor approximation and the transla- cavity, x, based on the solvent, solute, and free tional entropy lost by virtue of acquiring a gyra- volumes as tion radius while being confined to V = Ncvc. 2/3 2/3 Assuming that rotation is fast, the radius rc max{Vfree − VM , 0} x = 2/3 2/3 . (7) of the cavity in which the centroid of a linear Vfree + VS or spherically symmetric rigid rotor can move That is, the solute can only hop if the cross freely is effectively reduced by its radius of gy- ration rg, sectional area of VM is smaller than Vfree

(given that all cross sectional areas are deﬁned Natoms identically in terms of volume, regardless of 2 1 X 2 rg = (rk − rmean) . (10) Natoms shape). Furthermore, assuming eﬀective spher- k=1 ical shapes for each volume and introducing 1/3 For nonsymmetric rotors, the reduction by of rc = [3vc/(4π)] as the radius of the cavity, there will be rc by rg is also assumed as an averaged radius is necessary to preserve rotational invariance. 4π 2/3 r2 Hence, the rotational entropy is N = 4 c (8) x 3 2/3 2/3 Vfree + VS ∂ ln (qr) Sr =k ln (qr) + kT + sites for hopping per cavity. Because there will ∂T V be at least one cavity available for the solute, St(T, rc − rg) − St(T, rc), (11) and considering that the probability of hopping n n times is x , Nc may be approximated as where St(T, r) is the translational entropy at 3 temperature T and volume V = Nc4πr /3, and k=∞ X k qr the rigid rotor rotational partition function. Nc = 1 + Nx x For a nonlinear molecule27,28 k=1 1 π1/2 8π2IkT 3/2 = 1 + Nx − 1 . (9) qr = 2 , (12) 1 − x σr h

3 1/3 31 where I = (IxIyIz) is the average moment of acentric factor, ω, of the solvent as inertia and σr the rotational symmetry number. ideal A possible issue with eq. 11 is that, in the case ideal real ∆Hvap ∆Svap − ∆Svap = − ω (13) of an extremely large and nonspherical solute in Tc a dense solvent, we could have rc < rg. Such = −5.365ωk. a situation does not occur in any of the systems studied here, but this issue is resolved by The 5.365 factor arises due to the fact that an a physical interpretation of the model: when ideal liquid obeying the standard corresponding 2/3 1/3 states theorem will have a constant ∆H /T = rc − rg < (3/4π) Vfree , one should replace vap c 2/3 1/3 5.365k independent of temperature.33,34 Acen- rc − rg in eq. 11 with (3/4π) Vfree . The reason for this being that the solute in this situ- tric factors for common solvents are readily ation will be surrounded in each direction by available; they relate molecular shape and po- larity to non-ideal behavior. The more polar molecules with an associated free volume Vfree. and nonspherical a molecule is, the larger its Furthermore, if rc < rg rotation inside the cav- acentric factor is expected to be. ity will not be free and qr should be revised accordingly. In the Appendix, we provide an There are two issues with equating Sc to eq. 13. The first one is that is that, because an approximate expression for qr for use in this situation. ideal liquid experiences changes only in transla- One may ask why have we not chosen to define tional motions during a phase transition, eq. 13 gas contains losses to rotational entropy upon con- St → St(T, rc − rg) and Sr → Sr given that densation. One must therefore disentangle from such a choice would allow to write St and Sr eq. 13 the loss in rotational entropy from the en- in terms of qt and qr in a more straightforward manner. The reason for our choice here is that, tropy changes due to rearrangement of solvent molecules around the solute. We write thus a as shown next, the definition of Sr in eq. 11 first draft for S that subtracts the loss of rota- allows for a convenient way to evaluate Sc from c the Pitzer acentric factor of the solvent. tional entropy from eq. 13: Sω = −5.365ωk − ∆Sgas→sol (14) Cavity Entropy: Acentric Factor Approx- c,draft r,S S S S imation = −5.365ωk − St(T, rc − rg ) + St(T, rc ). Dionis´ıo et al.31 provide an interpretation of Eq. 14 applies to a pure substance; it accounts Sc based on a reference ideal liquid (i.e., one for entropy lost due to correlations between the that is spherical and nonpolar). Because there the dissolved and surrounding molecules. For are no correlations between the solute and the a mixture, we must consider the differences in surrounding molecules, Sc is set to zero for shape and size of the solute and solvent as the the reference ideal liquid. The cavity term is entropy loss will be greater the more solvent then identified with the difference in vaporiza- molecules coordinate around the solute. Thus, tion entropy of the ideal and real versions of we write a pure liquid. For substances in a standard ω gas→sol state that obey a three-parameter correspond- Sc,draft = −(5.365ωk + ∆Sr,S )G(RM, RS), ing states principle32 (i.e., those well described (15) by introducing an acentric factor, in addition where G(RM , RS) is a function of the molec- to reduced temperature and pressure, to cor- ular geometries of the solute RM and the sol- rect for nonideal behavior), this difference can vent RS satisfying G(RX , RX ) = 1. The func- be conveniently evaluated in terms of the Pitzer tion G must account for the number of solvent molecules that coordinate around the solute relative to the solvent. This is a packing problem and such problems are, in general, combinato- rial NP-hard and have no analytical solution

4 (much work relevant to liquids has been done where W (VM, nS) is the reversible work required on the packing problem, see, e.g., refs. 35,36 to create the cavity in the fluid. Since we are for reviews). For the sake of having a practical dealing with an ideal version of the solvent, we method to compute Sc, we introduce an approx- neglect surface tension terms and write W as imation here. Suppose we are packing cubes in volume work only a box-shaped cavity. An analytical solution is then straightforward: G(RM , RS) = AM /AS, W (VM, nS) = VMP (VS, nS). (19) with AX being the surface area of X. However, this expression does not take into account cur- To determine the pressure in the medium, vature, which may lead to inaccurate estimates P (VS, nS), we use the exact relation for hard 38 of the packing. For example, if we interchanged spheres the cubes packed around the box by spheres of p (V ≤ V ) = 1 − V n , (20) diameter equal to the length of a side of the 0 M S M S cubes, we would overestimate the G by a factor so that the pressure in the solvent is of 6/π because of the larger volume to surface area ratio of the sphere. If we introduce the kT P (VS, nS) = − ln (1 − VSnS) . (21) following shape factor VS box Hence, the cavity entropy in the reference ideal φX = AX /AX (16) liquid is box where AX is the surface area of X and AX the surface area a box that exactly encloses the 0 kVM Sc = ln (1 − VSnS) . (22) volume of X (i.e., the minimum bounding box VS of VX ). Then we can construct G to be exact for And the final expression for Sω becomes cubes and spheres (simply packed on a surface) c as ω 0 gas→sol Sc = Sc − (5.365ωk + ∆Sr,S )G. (23) AM φS G(RM , RS) = . (17) ASφM All of the terms in eq. 1 are now defined and While approximate, this choice for G also we can see that the acentric factor approxima- provides better estimates than the area ratio tion Sω has the following characteristics: (1) AM /AS for packing of other dissimilar geomet- It does not require significantly more compu- ric objects (e.g., ellipsoids and cuboids). Like tational resources than the calculation of gas volumes, surface areas are here calculated based phase entropies; (2) it is nonempirical in the on van der Waals radii.29 sense that it does not employ adjustable pa- The second issue that needs to be addressed rameters based on experimental data; and (3) to calculate Sc from the relation in eq. 13 is it only requires knowledge of the mass density of that the entropy of cavity formation of an ideal the solvent and its acentric factor, both of which liquid is not zero. The probability of finding are readily available for common solvents. It an empty site that can fit and ideal solute in a is worth noting that ω can be estimated from medium of hard spheres is less than one, and the boiling point (Tb), critical temperature (Tc), 31 thus we must have Sc < 0 even for the ideal and critical pressure (Pc) of a substance as liquid. The entropy contributions arising from T P this situation can be estimated as follows: the ω = b ln c − 1. (24) probability arising from statistical fluctuations 5.365(Tc − Tb) 1 atm of finding an unoccupied volume V in a fluid M Quantitative structure-activity relationships of number density n = N ρ/M S is37 S A w can also be used to estimate ω for certain classes W (V , n ) of compounds.39 p = exp − M S , (18) 0 kT

5 Cavity Entropy: Scaled Particle Theory 0(r − 1)/(r + 2), which yields An alternative way to describe the thermody- 3 r − 1 namics of cavity formation is provided by scaled y = . (28) 4π r + 2 particle theory; a statistical mechanical model based on hard spheres of radii defined such that In other words, we chose the scaled radius of macroscopic properties are reproduced (for re- the solvent RS to be consistent with its relative views on the subject, see refs. 19,37). The permittivity. free energy of cavity formation in scaled par- The cavity entropy can now be obtained via ticle theory is determined from the probability Maxwell’s relations, of inserting a cavity center in a liquid composed of spheres of certain volume and number den- α ∂Gc Sc = − , (29) sity. The resulting free energy expression that ∂T P is often used in polarizable continuum models evaluating the partial derivative by application is19,37 of the chain rule, remembering that nS depends 3 on the temperature. If we let f = G /(kT ), G =kT − ln(1 − y) + R+ c c 1 − y then " # # 3y 9 y 2 ∂G G ∂f ∂y + R2 , (25) c = c + kT (30) 1 − y 2 1 − y ∂T P T ∂y ∂T P where R = RM/RS is the ratio of the scaled with radii of the solute and solvent and y is reduced ∂y ∂y ∂nS number density of the solvent: = = −αy (31) ∂T P ∂nS ∂T P y = (4π/3)R3n (26) S S where α = (1/V )(∂V/∂T ) is the isobaric volu- ρ P = (4π/3)R3N S . metric thermal expansion coefficient of the sol- S A M S w vent. Frequently, |Gc/T | |αkT y(∂f/∂y)p| so Eq. 25 is highly sensitive to the value of y,40 and we explore the possibility of computing Sc from hence the effective radii are normally treated as nS and r only as parameters19 adjusted (“scaled”) to reproduce S = G /T. (32) known solvent properties. Here, we estimate c c them based on the same kind of information The rest of the contributions to the entropy used to develop the acentric factor approxima- can be computed as described before. Thus, tion and commonly available solvent properties. the scaled particle theory (SPT) approxima- The ratio R is thus calculated as tions Sα and S, differ from Sω only in the definition of S . Like S , S and S do not employ V 1/3 c ω α R = M . (27) empirical parameters but require knowledge of VS elementary properties of the medium: dielectric Since R is a ratio quantity, we can expect eq. 27 constant, mass density, and the thermal expan- to be reasonably accurate as long as the vol- sion coefficient in the case of Sα. Thus, a connection between a molecular property, ω, and umes VM and VM are computed in a consistent manner (here using van der Waals volumes). To two macroscopic properties, α and r, is estab- define y in a way that is suitable for scaled par- lished through the different ways proposed here ticle theory, we employ the polarizability-based to calculate Sc. Indeed, as shown later, reasonable estimates of ω can be obtained from r by definition of molecular volume, Vp = αp/(4π0), solving for ω such that S = Sω and vice versa. and the Clausius–Mossotti equation, nSαp/3 = c c

6 Standard States and Concentration (2.2 cal/mol-K) are comparable to the MAE for the calculated gas phase entropy (1.8 cal/mol- In calculating absolute and solvation entropies, K). At 300 K, these errors (≈ 0.7 kcal/mol) we adopt the following conventions: standard are within what is normally considered chem- states of gases are defined based on the ideal ical accuracy (≤ 1 kcal/mol). The errors for gas equation at 1 bar and 298.15 K (standard S◦ = S◦ + ∆S are also within chemical conditions); for pure liquids, it is the state of sol gas accuracy for all three approximations S◦ (3.0 the substance at standard conditions; for mix- ω cal/mol-K), S◦ (2.8 cal/mol-K), and S◦ (2.5 tures the solute concentration is 1 M. Entropy α cal/mol-K) at 300 K. Figure 1 shows the ex- changes due to changes in concentration are es- perimental vs calculated S◦ for the complete timated with the usual relation sol dataset in the SI; there is excellent correlation between the experimental data and both the ∆Sconc = k ln (ci/cf ) , (33) acentric factor and SPT methods. For clarity, where ci and cf are, respectively, the concen- Sα entropies are omitted in this figure as they trations in the initial and final state. Thus, the are very similar to S entropies (slope = 0.981, entropy penalty for bringing a gas at standard intercept = 0.1 cal/mol-K, R2 = 0.975). conditions to a 1 M concentration is ∆Sconc =

−6.4 cal/mol-K. o o 2

100 ● + ( ) Sω = 0.956 Sexpt 0.3 R = 0.969 o o 2 ● Sε = 0.979 S − 0.1 (R = 0.974) ● expt ● ● ● Benchmarks ● ● ●● ● ● 80

Experimental and computed gas phase and so- ●●● ● lution entropies for 110 pure substances and bi- ●● ●●● ● ● ● ● ● ●●●● (cal/mol−K) ●●● nary mixtures are given in the Supporting Infor- ● 60 ● ● l ●● ●●●

o ●● ● ● ●● ● ● o s ● ●●● ● ● ● ●● mation (SI). The reference data were compiled S ● ● ● ●● ●● ● ●●● ● ● ●● ●●●●● ●● ● from the NIST Standard Reference Database ●●● ● ● ●●● ● ● ● ● ● 41 42 ● ● ● ● ●● ●●

40 ● Number 69, published Henry Law data, and ●● ● ●● ● ● ●●●● ● ● cross-referencing enthalpies from the Acree En- ● ● ● ● ● ● Calculated

● 43 ●● ● ● thalpy of Solvation Dataset and free energies ● ● 44 ●● from the Minnesota Solvation Database. A 20 ●●● ●●● ●●● ●●● 31,45,46 ●● few other sources were also used. Geome- ● ● ● ● ● tries and vibrational frequencies were computed ● 47 20 40 60 80 100 with the GFN-xTB method. The calculated Experimental So (cal/mol−K) entropies assume that the gas-phase geometries sol and vibrational entropies do not change upon Figure 1: Experimental vs calculated standard solvation (an assumption that has been used in entropies in solution for a set of 110 pure sub- 18 solvation free energy models and is supported stances and mixtures. Results from Sα are very 15 by simulations ). For molecules such as n- similar to S and are omitted for clarity. octane and n-hexanol for which conﬁgurational entropy becomes important, a conﬁgurational When dissolving a noble gas in itself, ∆S ≈ entropy term of 1.8 cal/mol-K per non-terminal ∆St − kT ln(csol/cgas). The good accuracy with carbon48 is included in the total entropy. It is which the models predict ∆S for the Ne–Xe also assumed that this term is identical for the series (error ≈ 1 cal/mol-K) thus indicates gas-phase and dissolved species. that the way in which St is calculated is well- Table 1 shows representative ∆Ssol data from grounded. Entropies in solvents that can form the SI as well as error (calculated − experi- hydrogen bonds such as water are also accu- mental) statistics for the full database. The rate, even though the models make no spe- ◦ mean absolute errors (MAEs) for ∆Sω (2.4 cial consideration for such interactions. This ◦ ◦ cal/mol-K), ∆S (2.3 cal/mol-K), and ∆Sα suggests that, in most cases, the information

7 Table 1: Experimental and calculated gas phase and solvation entropies (cal/mol-K) for various substances and mixtures.

◦ ◦ ◦ ◦ ◦ ◦ Solute Solvent Sgas,exp Sgas,calc ∆Ssol,exp ∆Sω ∆S ∆Sα Helium Helium 8.1 9.0 -4.7† 0.1 -3.4 -3.4 Argon Argon 37.0 37.0 -18.4‡ -21.0 -19.8 -19.6 Benzene Benzene 64.3 64.7 -22.9 -23.5 -23.6 -23.2 n-Pentane n-Pentane 83.5 80.1 -20.5 -22.5 -21.8 -21.7 n-Octane n-Octane 111.6 111.5 -25.3 -23.0 -22.3 -22.0 Chloroform Chloroform 70.6 73.6 -25.2 -24.3 -24.9 -23.6 Acetaldehye Acetaldehyde 59.8 60.1 -31.8 -26.1 -27.9 -24.9 Acetone Acetone 70.6 66.6 -22.7 -25.1 -27.1 -24.6 Acetic Acid Acetic Acid 67.6 69.1 -29.8 -27.9 -26.7 -25.6 Water Water 45.1 45.1 -28.4 -31.1 -32.4 -31.9 Acetic Acid Water 67.6 69.1 -25.4 -30.2 -29.0 -28.1 Ethanol Water 67.6 64.9 -31.6 -30.3 -28.9 -28.0 Butanol Water 86.5 83.0 -35.1 -35.5 -32.2 -31.0 Argon Water 37.0 37.0 -23.0 -22.5 -23.9 -23.3 Carbon Dioxide Water 51.1 50.3 -26.5 -24.9 -26.1 -25.4 Cyclohexane Water 71.2 70.9 -31.6 -36.1 -32.4 -31.1 1,4-Dioxane Ethanol 71.6 64.9 -17.1 -25.1 -23.5 -20.9 n-Octane Ethanol 111.6 111.5 -19.7 -32.8 -27.0 -23.6 Nitromethane Ethanol 71.7 67.4 -17.5 -23.3 -22.2 -20.2 Methanol Butanol 57.3 56.6 -20.8 -18.2 -18.7 -17.9 n-Octane Butanol 111.6 111.5 -22.9 -26.5 -24.3 -22.6 Pentanol Benzene 95.9 92.2 -20.8 -19.3 -19.5 -19.0 Cyclohexane Benzene 71.2 70.9 -16.4 -18.6 -18.6 -18.1 Ethanol Toluene 67.6 64.9 -17.6 -16.8 -17.2 -16.8 n-Octane Toluene 111.6 111.5 -16.8 -20.5 -20.1 -19.6 Acetonitrile Cyclohexane 62.4 59.7 -14.5 -16.2 -16.4 -16.1 Hexanol Cyclohexane 105.0 100.9 -17.7 -19.7 -19.3 -18.9 Toluene Cyclohexane 76.7 78.2 -18.0 -18.8 -18.6 -18.2 Ethanol n-Hexane 67.6 64.9 -12.7 -15.5 -15.7 -15.4 Hexanol n-Hexane 105.0 100.9 -16.7 -18.9 -18.6 -18.2 Nitromethane n-Hexane 71.7 67.5 -14.8 -15.2 -15.4 -15.1 Benzene n-Hexane 64.3 64.7 -17.7 -16.8 -16.7 -16.4 Ethanol Chloroform 67.6 64.9 -19.4 -18.3 -19.0 -18.0 n-Octane Chloroform 111.6 111.5 -18.9 -22.7 -22.4 -20.8 Mean Error§ -1.1 -0.7 -0.6 0.2 Mean Absolute Error§ 1.8 2.4 2.3 2.2 Median Error§ 0.0 -0.5 -0.6 0.1 Median Absolute Error§ 1.5 2.1 1.9 1.7 Mean Absolute Deviation§ 1.8 2.8 2.7 2.2 Minimum Error§ -7.1 -13.1 -8.5 -6.7 Maximum Error§ 4.9 5.7 6.6 7.7 †4.2 K. ‡87.3 K. SErrors are for the 110 solvation entropies in the SI.

8 Table 2: Experimental and calculated entropies of activation (cal/mol-K) at 1 M concentration for various reactions in the gas phase and aqueous solution.

‡ ‡ ‡ ‡ ‡ ‡ Reaction ∆Sgas,exp ∆Sgas,calc ∆Saq,exp ∆Sω ∆S ∆Sα H + CH3OH H2 + CH2OH -15.4 -16.7 -7.3 -6.0 -3.6 -3.8 H + CH3CH2OH H2 + CH3CHOH -16.0 -16.5 -7.8 -4.8 -3.0 -3.3 H + (CH3)2CHOH H2 + (CH3)2COH -17.3 -16.1 -6.3 -4.5 -2.6 -2.9 H + CD3CD2OH HD + CD3CDOH -16.0 -16.4 -5.5 -4.8 -3.0 -3.3 H + (CD3)2CDOH HD + (CD3)2COH -17.2 -16.1 -5.6 -4.5 -2.6 -2.8 H + CH2(OH)2 H2 + CH(OH)2 -12.4 -14.9 -3.1 -3.9 -1.4 -1.7 H + (CH2OH)2 H2 + HOCH2CHOH -16.5 -17.5 -10.4 -6.0 -3.9 -4.2 CH3SH + CH3 CH3S + CH4 -22.1 -18.6 -13.9 -7.6 -4.7 -5.0 CH3SH + CH2OH CH3S + CH3OH -23.9 -26.4 -18.8 -14.6 -12.0 -12.3 CH3SH + (CH3)2COH CH3S + (CH3)2CHOH -25.5 -29.7 -16.0 -17.8 -14.6 -15.1 Mean Error -0.6 2.0 4.3 4.0 Mean Absolute Error 1.8 2.5 4.3 4.0 necessary to determine solvation entropy is en- typical errors are within the uncertainty of the coded in ρ, ω, and r. However, the largest experimental techniques employed to determine error occurs for octane in ethanol with the Sω ∆Ssol, we warn that the Sω and SPT approx- method. This method estimates the loss of en- imations are based on a physical picture that tropy due to solute-solvent correlations based does not consider complexation. Hence, the ap- on the acentric factor of the solvent. Thus, proximations may fail for ions or molecules that we expect larger, negative errors in cases of form coordination complexes with the solvent. nonpolar solutes in solvents having strong in- Table 2 provides additional benchmarks in the teractions (such errors will be exacerbated as form of entropies of activation (∆S‡) for ten the size of the solute increases; however, er- bimolecular reactions in gas phase and aque- rors will increase with system size for any ap- ous solution. These reactions have been stud- proximation when calculating properties which ied before in the context of solvation entropy 49 are not intensive ). Therefore, Sω is better and the experimental data are reasonably ac- suited for describing solutes that have interac- curate.2 The initial and transition state geome- tions with the solvent that are similar to the tries were computed in the gas phase at the solvent-solvent interactions. Note, for example, ωB97X-D50/6-31G(d) level in Gaussian;51 fre- ◦ that the ∆Sω error for octane is dramatically quencies were scaled by the recommended fac- reduced in butanol or toluene (≈ −3.6 cal/mol- tor for this level of theory (0.949).52 The av- ‡ K) as compared to ethanol (−13.1 cal/mol-K). erage errors in ∆Saq are similar to those for ‡ Ethanol also has the largest ω = 0.644 in the ∆Sgas. The ≈ 1.5 cal/mol-K larger average er- set of solvents studied, which suggests greater rors in the SPT methods as compared to Sω deviations from ideal behavior and the three- arise due to error in the entropy of the hydro- parameter corresponding states principle.32 We gen atom, which is involved in seven of the ten ◦ also caution that one should not draw strong reactions. The experimental Saq for the hydro- conclusions based on errors of ≈ 4 cal/mol-K gen atom is 10.5 cal/mol-K, whereas Sω, S, or less. Particularly when it comes to mix- and Sα give 8.8, 7.0, and 7.3 cal/mol-K, re- ‡ tures, experimental solvation entropies are not spectively. Note that, in average, ∆Sgas,calc un- ‡ as precise as free energies (presumably due to derestimates ∆Saq,exp by −10.1 cal/mol-K even the use of extrapolation to determine the for- though we have adjusted for 1 M concentra- mer). To give an example, for butanol in aque- tion and the molecularity of the initial and ﬁnal ous solution, the standard deviation in ∆Ssol states is the same for all reactions in Table 2. determined from various sources of Henry Law This is so because the transition state still expe- data42 is 3.8 cal/mol-K. Despite the fact that riences a loss of (mostly translational) entropy.

9 The ≈ −10 cal/mol-K of the gas phase formu- VS larger also make Vfree smaller, partially oﬀ- las is also seen for the Diels–Alder reaction of setting changes in vc. Thus, our approxima- cyclpentadiene with methyl acrylate in toluene tions are relatively insensitive to the deﬁnition (Fig. 2). All three Sω, S, and Sα methods esti- of molecular volumes. This means that one mate ∆S‡ within 1 cal/mol-K of its experimen- could use, e.g., Bader volumes 54 or volumes tal value53 (−29.7 cal/mol-K) for this archety- from coupled cluster calculations55 to apply the pal Diels–Alder reaction. model ab initio to species for which the van der Waals radius is not available (e.g., ions, heavy elements).

∆Sexp = -29.7 300 ∆S = -40.6 gas CO2Me A) + 280 Xe )

∆Sω = -30.1 3 CO Me 2 260 ∆Sε = -29.3 ∆ Sεα = -29.6 240 Kr Figure 2: Experimental and calculated activa- 220 tion entropies (cal/mol-K at 298 K and 1M con- Cavity Volume ( Å 200 Ar centrations) for the Diels–Alder reaction of cy- 180 clpentadiene with methyl acrylate in toluene. 15 20 25 30 35 40

32 Additional Remarks and Testable 30 Predictions 28 The entropy models proposed here make testable predictions that go beyond solvation 26 Entropy ( cal / mol - K ) entropies. Some of these predictions are dis- 24 cussed here along with other features of the B) models such as their dependence on the defini- 22 15 20 25 30 35 40 tion of molecular volume. Atomic Volume(Å 3) As is the case for other solvation models, the solvent cavity is central to our approximations. Figure 3: Dependence of the (A) cavity volume Here, vc depends on the solvent density and on and (B) translational entropy on VS for liquid the volumes VM and VS. There are multiple noble gases at their boiling point. ways to define molecular volumes apart from the union van der Waals atomic volumes used Note also from Fig. 3B that the model pre- here (e.g., isodensity surfaces and Bader vol- dicts a discontinuity in the derivative of S with umes54). However, the dependence of S on how respect to the volume. This discontinuity cor- volumes are defined is small as long as these are responds to the point at which VS = Vfree and reasonable and used consistently. This is illus- arises due to the hopping term Nc, which gives trated in Fig. 3 using noble gases as an exam- the solute a nonzero probability of escaping its ple. A change in VS as large as 270% does not cavity if VS < Vfree. Since ρ determines Vfree change vc by more than about 20% (Fig. 3A). and depends on the temperature, (∂G/∂T )P Likewise, the change in translational entropy— will be exhibit a discontinuity as the liquid ex- the largest component to the entropy for small pands and reaches VS = Vfree. The discontinuity molecules—is modest: only about 3 cal/mol- may thus be associated with a liquid-gas phase K differences in the same range of VS varia- transition below Tc. Under this interpretation, tions (Fig. 3B). Volume definitions that make VS corresponds to the discontinuity point at Tb. Thus, our model predicts the radii of the Ne,

10 Ar, Kr, and Xe to be 1.50, 1.81, 1.88, and 2.07 and ∆Svap ≈ ∆St so that we must have A,˚ respectively. This compares well with the ∂∆S ∂ ln(kT/P v ) van der Waals radii of these elements: 1.54, vap = k c (35) 1.88, 2.02, and 2.16 A,˚ in the same order as ∂T ∂T above.29 That is, we have obtained molecular 1 = k − α(1 + VS/Vfree) = 0. volumes from the density of a substance at their T boiling point. Relatedly, ∆H (T ) can be determined from Per the above discussion, VS = Vfree at Tb and vap b hence ∆Svap if Tb is known. A simple way of 1 α(T ) = . (36) estimating ∆Hvap is provided by Trouton’s b 2Tb rule,56 which states that ∆H = 10.5kT , or, vap b −3 −1 more accurately, by the Trouton–Hildebrandt– This relation predicts α = 18.5 × 10 K −3 −1 Everett’s (THE) rule57,58 for neon and α = 5.7 × 10 K for argon, in agreement with their respective experimen- THE tal59,60 values of 15.4 and 4.8 × 10−3K−1. More ∆Hvap = [4 + ln(Tb/K)]kTb, (34) generally, we can write which works well for simple liquids that do not −1 −1 from strong interactions such as H-bonds. Fig- α(T, ρ) = T (1 + VS/Vfree) . (37) ure 4 compares experimental and calculated va- ω α porization enthalpies for 32 liquids with eq. 34. Furthermore, by equating Sc to Sc , ω can be estimated analytically from r and α as Eq. 34 gives a MAE of 1.4 kcal/mol for ∆Hvap; Sω, S, and Sα bring the error down to chem- 1 ω = S0 − S ( , α) − ∆Sgas→sol , ical accuracy with MAEs of 0.7, 0.9, and 0.7 5.365k c c r r kcal/mol, respectively (see SI). (38) ω or just from r by making Sc = Sc = Sc(r, 0). ●■ ■ ● ● ◆ ◆ ◆ ◆ ●■ ● ● Expt. ◆■ ◆ ● Similarly, r can be estimated from ω by this ■◆● ●■■ 8 ◆ ■■ ◆ ● ■ ■ ◆ same relation, but solving a nonlinear equation Sω ◆ ● ● ● ●■ ◆■ ◆ Sε ◆■● to obtain the former instead. Table 3 com- ◆ ● 6 ■ ● ● pares experimental α, ω, and r values with

4 ● those predicted from eqns. 37 and 38. Reason- ◆■◆■ THE ● able estimates of α, ω, and r are obtained that ■ ◆● 2 ■ ◆● correlate well with their experimental values as

●■ measured by the Pearson correlation coeﬃcient Vaporization Enthalpy ( kcal / mol ) ◆ 0 ◆●■ (> 0.85). As could be expected, the calculated 0 100 200 300 400 constants are more precise for weakly-polar sol- Temperature(K) vents that do not form H-bonds. In fact, alco- Figure 4: Experimental and calculated vapor- hols that have large ω values are excluded be- ω ization enthalpies of 32 compounds compared cause Sc = Sc() does not have a numerically ω to the THE rule (eq. 34). stable solution in these cases (|Sc | is signiﬁ- cantly larger than |Sc|, which makes r → ∞). From here other predictions of the model can The relative errors in predicted ω values are be derived. Consider an ideal liquid that obeys large for certain substances due to the fact that Trouton’s rule. For such a liquid, the vaporiza- for pure liquids Sc is typically in the range of tion entropy does not change with temperature 0–4 cal/mol-K. Thus, a discrepancy of only 1 ω cal/mol-K between Sc and Sc results in a large relative error in ω. Despite these caveats, the fact that the predicted constants are reasonable and correlate to experiment is further evidence that the theories presented here are well-

11 Table 3: Experimental and predicted thermal expansion coeﬃcients α (10−3K−1), acentric factors ω, and dielectric constants r at standard conditions. Mean absolute errors, median absolute errors, and Pearson correlation coeﬃcients are also given.

expt Substance αexpt α(T, ρS) ωexpt ω(r) ω(r, α) r r(ω) Neon† 15.40 16.59 -0.029 -0.099 -0.121 1.5 2.5 Argon† 4.80 4.72 0.001 -0.118 -0.139 1.5 3.4 Ethylene† 2.40 3.13 0.089 0.089 0.088 1.0 1.0 Cyclohexane 1.21 1.47 0.212 0.188 0.153 2.0 2.4 Benzene 1.25 1.48 0.212 0.223 0.179 2.3 2.1 Toluene 1.08 1.46 0.263 0.257 0.216 2.4 2.5 m-Xylene 0.99 1.46 0.325 0.28 0.243 2.4 3.2 n-Pentane 1.58 1.69 0.251 0.185 0.168 1.4 2.2 n-Hexane 1.41 1.64 0.299 0.277 0.242 1.9 2.2 n-Octane 1.14 1.55 0.398 0.334 0.302 2.0 3.1 Chloroform 1.27 1.43 0.218 0.272 0.155 4.8 3.4 Dioxane 1.12 1.81 0.307 0.218 0.179 2.3 4.0 Ethyl-ether 1.60 1.66 0.281 0.378 0.26 4.3 2.3 Acetaldehyde 1.69 1.64 0.303 0.469 0.184 21.1 5.1 Acetone 1.43 1.60 0.304 0.495 0.256 20.7 4.3 Acetonitrile 1.36 1.60 0.278 0.498 0.243 37.5 4.5 Water 0.21 1.65 0.344 0.466 0.424 78.5 11.3 MAE 0.41 0.080 0.069 3.3 Median AE 0.24 0.066 0.063 2.8 Pearson Correlation 0.994 0.852 0.927 0.933 † At T = Tb. grounded. to estimate ∆G values in solution in cases when We have thus provided ample evidence that— ∆Hsol ≈ ∆Hgas (not an uncommon occurrence, without the need of empirical parameters—the especially in nonpolar solvents). Therefore, the Sω and SPT approximations make sound pre- methods presented here oﬀer a practical alter- dictions regarding properties of liquids and pro- native to drastically reduce the problem of inac- vide solvation entropies with average errors that curate ∆Sreac terms in solution. This can be of are within chemical accuracy. The fact that substantial value in catalyst and drug discov- solvation entropies can be determined in a fast ery, where processes that change molecularity and accurate manner with these methods has are ubiquitous, accurate free energies are criti- implications on the development of implicit sol- cal in determining activity, and a high volume vation models. If the entropy is computed with of calculations is often inevitable. any of the Sω, S, or Sα models, then the enthalpy can be calculated with a complementary Acknowledgements implicit solvation method such as, e.g., polarizable continuum models19 or joint density func- I would like to thank Prof. Stefan Grimme tional theory.61 Since the enthalpy contribu- (Universit¨atBonn) for providing Dow Chemi- tions to solvation are largely electrostatic, such cal with a test license the XTB program devel- techniques should be able to provide an accu- oped by his research group. I would also like to rate ∆Hsol. Thus, we would have a model for thank Dr. Peter Margl, Dr. Steven Arturo, Dr. ∆Gsol that correctly describes both ∆Hsol and Ivan Konstantinov, and Dr. Marc Coons (Dow ∆Ssol, and that at the same time is less reliant Chemical) for helpful discussions. on parametrization than most existing solvation methods. The Sω and SPT approximations can also be used in a standalone manner

12 Supporting Information Available above analysis suggests that a reasonable partition function for a prolate or oblate, nonlinear Solvent constants and all of the experimental solute with r < r may be written as and calculated entropies and enthalpies. c g θ 2 8πIkT 3/2 q = qgas 0 = θ2, (45) Appendix r r π h2 0

Here we provide an educated guess for the ro- 2 which recovers Eq. 12 when θ0 = π except for tational partition function qr of extremely non- the 1/σr factor. Whether or not one should spherical solutes for which rc < rg. Let us begin divide q by a symmetry factor would depend with the Schr¨odingerequation for the quantum r 62 on θ0 and the rotational symmetry of the solute. pendulum A reasonable choice for θ based on our model h¯2 d2ψ would be − − U(θ)ψ(θ) = Eψ(θ). (39) 2I dθ2   rg θ0 = 2 arccos q  , (46) For U = 0 one recovers the free√ rotor limit with 2 2 0 imθ rg + rfree eigenfunctions ψm(θ) = e / 2π and eigenenergies E (m) =h ¯2m2/2I. The partition func- 0 1/3 tion under the usual integral approximation is with rfree = [3Vfree/(4π)] . References on how to determine symmetry numbers in various sit- Z ∞ 63,64 −¯h2m2/2I uations are available in the literature. q0 = e dm (40) 0 πIkT 1/2 = . (41) References 2¯h2 (1) Soroush, M. Computational Quantum Suppose now that we constrain the pendulum Chemistry: Insights into Polymerization inside a “cavity” by a potential Reactions. Elsevier Science, 2018, 43–45. ( 0, if |θ| ≤ θ0/2 (2) Leung, B.O.; Reid, D.L.; Armstrong, U(θ) = (42) ∞, otherwise. D.A.; Rauk, A. Entropies in Solution from Entropies in the Gas Phase. J. Phys. This is simply the particle in a box problem. Chem. A 2004, 108, 2720-2725. Acceptable eigenfunctions is this situation are U p (3) Wolfe, S.; Kim, C.K.; Yang, K.; Weinberg, ψm(θ) = 2/θ0 cos(mπθ/θ0) and the energy 2 2 2 2 N.; Shi, Z. Hydration of the Carbonyl spectrum is EU (m) =h ¯ m π /2Iθ0. The partition function from integration is therefore Group: A Theoretical Study of the Co- operative Mechanism. J. Am. Chem. Soc. IkT 1/2 1995, 117, 4240–4260. qU = 2 θ0. (43) 2πh¯ (4) Wolfe, S.; Shi, Z.; Yang, K.; Ro, S.; Weinberg, N.; Kim, C.K.; Hydration of The ratio of qU /q0 is the Carbonyl Group. Further Evidence

qU /q0 = θ0/π, (44) for a Cooperative Mechanism from Ex- perimental and Theoretical Studies of the in agreement with the interpretation of the par- Hydrations of Formaldehyde, Acetalde- tition function as the available phase space vol- hyde, Acetone, and Cyclohexanone. Can. ume of the system. It is not possible to derive J. Chem. 1998, 76, 114–124. analytically a similar relation for a general rotation (the rigid rotor does not have analyti- (5) Sumimoto, M.; Iwane, N.; Takahama, T.; Sakaki, S. Theoretical Study of cal solutions for Ix 6= Iy 6= Iz). However, the

13 Trans-Metalation Process in Palladium- the Kinetics of PhosphineCatalyzed [3+2] Catalyzed Borylation of Iodobenzene with Cycloaddition Reactions of Allenoates and Diboron. J. Am. Chem. Soc. 2004, 126, ElectronDeﬁcient Alkenes. Chem. Eur. J. 10457–10471. 2008, 14, 4361–4373.

(6) Liu, C. T.; Maxwell, C. I.; Edwards, D. R.; (14) Plata, R. E.; Singleton, D. A. A Case Neverov, A. A.; Mosey, N. J.; Brown, R. Study of the Mechanism of Alcohol- S. Mechanistic and Computational Study Mediated Morita Baylis–Hillman Reac- of a Palladacycle-Catalyzed Decomposi- tions. The Importance of Experimental tion of a Series of Neutral Phosphoroth- Observations. J. Am. Chem. Soc. 2015, ioate Triesters in Methanol. J. Am. Chem. 137, 3811–3826. Soc. 2010, 132, 16599–16609. (15) Ribeiro, R.F.; Marenich, A.V.; Cramer, (7) Garza, A.J., Bell, A.T. and Head-Gordon, C.J.; Truhlar, D.G. Use of Solution- M., Is Subsurface Oxygen Necessary for Phase Vibrational Frequencies in Contin- the Electrochemical Reduction of CO2 on uum Models for the Free Energy of Solva- Copper? J. Phys. Chem. Lett. 2018, 9, tion. J. Phys. Chem. B 2011, 115, 14556– 601–606. 14562.

(8) Garza, A.J.; Pakhira, S.; Bell, A.T.; (16) Barone, V.; Bencini, A.; Cossi, M.; Mat- Mendoza-Cortes, J.L.; Head-Gordon, M. teo, A.D.; Mattesini, M.; Totti, F. Assess- Reaction Mechanism of the Selective Re- ment of a Combined QM/MM Approach duction of CO2 to CO by a Tetraaza for the Study of Large Nitroxide Systems II 2+ [Co N4H] Complex in the Presence of in Vacuo and in Condensed Phases. J. Am. Protons. Phys. Chem. Chem. Phys. 2018, Chem. Soc. 1998, 120, 7069–7078. 20, 24058–24064. (17) Andzelm, J.; Kölmel,C.; Klamt, A. In- (9) Tanaka, R.; Yamashita, M.; Chung, L.W.; corporation of Solvent Effects Into Den- Morokuma, K.; Nozaki, K. Mechanistic sity Functional Calculations of Molecular Studies on the Reversible Hydrogenation Energies and Geometries. J. Chem. Phys. of Carbon Dioxide Catalyzed by an Ir- 1995, 103, 9312–9320. PNP Complex. Organometallics 2011, 30, 6742–6750. (18) Kelly, C.P.; Cramer, C.J.; Truhlar, D.G. SM6: A Density Functional Theory Con- (10) Besora, M.; Vidossich, P.; Lledós, A.; tinuum Solvation Model for Calculating Ujaque, G.; Maseras, F. Calculation of Re- Aqueous Solvation Free Energies of Neu- action Free Energies in Solution: A Com- trals, Ions, and Solute-Water Clusters. J. parison of Current Approaches. J. Phys. Chem. Theory Comput. 2005, 1, 1133– Chem. A 2018, 122, 1392–1399. 1152.

(11) Spickermann, C. Entropies of Condensed (19) Tomasi, J.; Mennucci, B.; Cammi, R. Phases and Complex Systems: A First Quantum Mechanical Continuum Solva- Principles Approach. Springer, 2011. tion Models. Chem. Rev. 2005, 105, 2999– 3094. (12) Yu, Z.X.; Houk, K.N. Intramolecu- lar 1, 3-Dipolar Ene Reactions of Ni- (20) Skyner, R.E.; McDonagh, J.L.; Groom, trile Oxides Occur by Stepwise 1, 1- C.R.; Van Mourik, T.; Mitchell, J.B.O. Cycloaddition/Retro-Ene Mechanisms. J. A Review of Methods for the Calculation Am. Chem. Soc. 2003, 125, 13825–13830. of Solution Free Energies and the Mod- (13) Liang, Y.; Liu, S.; Xia, Y.; Li, Y.; Yu, elling of Systems in Solution. Phys. Chem. Z.X.; Mechanism, Regioselectivity, and Chem. Phys. 2015, 17, 6174–6191.

14 (21) Klamt, A.; Eckert, F. COSMO-RS: A Continuum Models: Eﬀect of the Loss Novel and Eﬃcient Method for the A Pri- of Translational Degrees of Freedom in ori Prediction of Thermophysical Data of Bimolecular Reactions on Gibbs Energy Liquids. Fluid Ph. Equilibria 2000, 172, Barriers. J. Phys. Chem. B 2005, 09, 43–72. 23618–23623.

(22) Eckert, F.; Klamt, A. Fast Solvent Screen- (31) Dion´ısio,M.S.; Ramos, J.J.M.; Gon¸calves, ing via Quantum Chemistry: COSMORS R.M. The Enthalpy and Entropy of Cavity Approach. AIChE J. 2002, 48, 369–385. Formation in Liquids and Corresponding States Principle. Can. J. Chem. 1990, 68, (23) Ratkova, E.L.; Palmer, D.S.; Fedorov, 1937–1949. M.V. Solvation Thermodynamics of Or- ganic Molecules by the Molecular Integral (32) Mansoori, G.A.; Patel, V.K.; Edalat, Equation Theory: Approaching Chemical M. The Three-Parameter Corresponding Accuracy. Chem. Rev. 2015, 115, 6312– States Principle. Int. J. Thermophys. 6356. 1980, 1, 285–298.

(24) Bhati, A.P.; Wan, S.; Hu, Y.; Sherborne, (33) Moura-Ramos, J.J.; Dion´ısio, M.S.; B.; Coveney, P.V. Uncertainty Quantifica- Gon¸calves, R.M.; Diogo, H.P., 1988. A tion in Alchemical Free Energy Methods. Further View on the Calculation of the J. Chem. Theory Comput. 2018, DOI: Enthalpy of Cavity Formation in Liq- 10.1021/acs.jctc.7b01143. uids. The Influence of the Cavity Size and Shape. Can. J. Chem. 1988, 66, 2894– (25) Loeffler, H.H.; Bosisio, S.; Duarte Ramos 2902. Matos, G.; Suh, D.; Roux, B.; Mob- ley, D.L.; Michel, J. Reproducibility of (34) J. S. Rowlinson. Liquids and Liquid Mix- Free Energy Calculations Across Dif- tures. 2nd ed. Butterworths, London. ferent Molecular Simulation Software. 1971. J. Chem. Theory Comput. 2018, DOI: 10.1021/acs.jctc.8b00544. (35) Torquato, S.; Stillinger, F.H. Jammed Hard-Particle Packings: From Kepler to (26) Grimme, S. Supramolecular Binding Bernal and Beyond. Rev. Mod. Phys. Thermodynamics by DispersionCorrected 2010, 82, 2633. Density Functional Theory. Chem. Eur. J. 2012, 18, 9955–9964. (36) Finney, J.L. Bernals Road to Random Packing and the Structure of Liquids. Phi- (27) McQuarrie, D. Statistical Mechanics. Uni- los. Mag. 2013, 93, 3940–3969. versity Science Books, 2000. (37) Pierotti, R.A. A Scaled Particle Theory (28) Jinnouchi, R.; Anderson, A. B. Elec- of Aqueous and Nonaqueous Solutions. tronic Structure Calculations of liquid- Chem. Rev. 1976, 76, 717–726. solid interfaces: Combination of Density Functional Theory and Modified Poisson– (38) Reiss, H.; Frisch, H.L.; Lebowitz, J.L. Boltzmann Theory. Phys. Rev. B: Con- Statistical Mechanics of Rigid Spheres. J. dens. Matter Mater. Phys. 2008, 77, Chem. Phys., 1050, 31, 369–380. 245417. (39) Gharagheizi, F.; Mehrpooya, M. Predic- (29) Bondi, A. Van der Waals Volumes and tion of Some Important Physical Proper- Radii. J. Phys. Chem., 1964, 68, 441–451. ties of Sulfur Compounds Using Quanti- tative StructureProperties Relationships. (30) Ardura, D.; López, R.; Sordo, T.L. Rela- Mol. Divers. 2008, 12, 143. tive Gibbs Energies in Solution through

15 (40) Tang, K.E.; Bloomﬁeld, V.A. Excluded (48) Taylor, W.J. Average Length and Ra- Volume in Solvation: Sensitivity of Scaled- dius of Normal Paraﬄn Hydrocarbon Particle Theory to Solvent Size and Den- Molecules. J. Chem. Phys. 1948, 16, 257– sity. Biophys. J., 2000, 79, 2222–2234. 267.

(41) P.J. Linstrom and W.G. Mallard, Eds., (49) Perdew, J. P.; Sun, J.; Garza, A. NIST Chemistry WebBook, NIST Stan- J.; Scuseria, G. E. Intensive Atomiza- dard Reference Database Number 69, tion Energy: Re-Thinking a Metric for National Institute of Standards and Electronic Structure Theory Methods. Z. Technology, Gaithersburg MD, 20899, Phys. Chem. 2016, 230, 737–742. https://doi.org/10.18434/T4D303, (re- trieved October 29, 2018). (50) Chai, J.D.; Head-Gordon, M. Long-Range Corrected Hybrid Density Functionals (42) Sander, R. Compilation of Henry’s Law with Damped AtomAtom Dispersion Cor- Constants (version 4.0) for Water as Sol- rections. Phys. Chem. Chem. Phys. 2008, vent. Atmospheric Chem. Phys. 2015 15, 10, 6615–6620. 4399–4981. (51) Gaussian 16, Revision B.01, Frisch, M. J.; (43) Chickos, J.S.; Acree Jr, W.E. En- Trucks, G. W.; Schlegel, H. B.; Scuse- thalpies of Vaporization of Organic and ria, G. E.; Robb, M. A.; Cheeseman, J. Organometallic Compounds, 1880–2002. R.; Scalmani, G.; Barone, V.; Petersson, J. Phys. Chem. Ref. Data 2003 32, 519– G. A.; Nakatsuji, H.; Li, X.; Caricato, 878. M.; Marenich, A. V.; Bloino, J.; Janesko, B. G.; Gomperts, R.; Mennucci, B.; (44) Marenich, A.V.; Cramer, C.J.; Truhlar, Hratchian, H. P.; Ortiz, J. V.; Izmaylov, D.G. Universal Solvation Model Based on A. F.; Sonnenberg, J. L.; Williams-Young, the Generalized Born Approximation with D.; Ding, F.; Lipparini, F.; Egidi, F.; Go- Asymmetric Descreening. J. Chem. The- ings, J.; Peng, B.; Petrone, A.; Henderson, ory Comput. 2009 5, 2447-2464. T.; Ranasinghe, D.; Zakrzewski, V. G.; (45) Pollack, G.L.; Himm, J.F., 1982. Solubil- Gao, J.; Rega, N.; Zheng, G.; Liang, W.; ity of Xenon in Liquid nAlkanes: Temper- Hada, M.; Ehara, M.; Toyota, K.; Fukuda, ature Dependence and Thermodynamic R.; Hasegawa, J.; Ishida, M.; Nakajima, Functions. J. Chem. Phys. 1982, 77, T.; Honda, Y.; Kitao, O.; Nakai, H.; 3221–3229. Vreven, T.; Throssell, K.; Montgomery, J. A., Jr.; Peralta, J. E.; Ogliaro, F.; (46) Pollack, G.L.; Himm, J.F.; Enyeart, J.J. Bearpark, M. J.; Heyd, J. J.; Brothers, Solubility of Xenon in Liquid nAlkanols: E. N.; Kudin, K. N.; Staroverov, V. N.; Thermodynamic Functions in Simple Po- Keith, T. A.; Kobayashi, R.; Normand, J.; lar Liquids. J. Chem. Phys. 1984, 81, Raghavachari, K.; Rendell, A. P.; Burant, 3239–3246. J. C.; Iyengar, S. S.; Tomasi, J.; Cossi, M.; Millam, J. M.; Klene, M.; Adamo, C.; (47) Grimme, S.; Bannwarth, C.; Shushkov, Cammi, R.; Ochterski, J. W.; Martin, R. P. A Robust and Accurate Tight-Binding L.; Morokuma, K.; Farkas, O.; Foresman, Quantum Chemical Method for Struc- J. B.; Fox, D. J. Gaussian, Inc., Walling- tures, Vibrational Frequencies, and Non- ford CT, 2016. covalent Interactions of Large Molecular Systems Parametrized for All spd-Block (52) NIST Computational Chemistry Compar- Elements (Z = 186). J. Chem. Theory ison and Benchmark Database; NIST Comput. 2017, 13, 1989–2009. Standard Reference Database Number 101 Release 19, April 2018, Editor: Russell D.

16 Johnson III http://cccbdb.nist.gov/ (re- (63) Gilson, M.K.; Irikura, K.K. Symmetry trieved December 17, 2018). Numbers for Rigid, Flexible, and Flux- ional Molecules: Theory and Applica- (53) Ruiz-López, M.F.; Assfeld, X.; Garc´ıa, tions. J. Phys. Chem. B 2010, 114 16304– J.I.; Mayoral, J.A.; Salvatella, L., Solvent 16317. Effects on the Mechanism and Selectivities of Asymmetric Diels–Alder Reactions. J. (64) Fernández-Ramos, A.; Ellingson, B.A.; Am. Chem. Soc. 1993, 115 8780–8787. Meana-Pañeda,R.; Marques, J.M.; Truh- lar, D.G. Symmetry Numbers and Chem- (54) Bader, R.F.; Carroll, M.T.; Cheeseman, ical Reaction Rates. Theor. Chem. Acc. J.R.; Chang, C. Properties of Atoms 2006, 118, 813–826. in Molecules: Atomic Volumes. J. Am. Chem. Soc. 1987, 109, 7968–7979.

(55) Mantina, M.; Chamberlin, A.C.; Valero, R.; Cramer, C.J.; Truhlar, D.G. Consis- tent van der Waals Radii for the Whole Main Group. J. Phys. Chem. A 2009, 113, 5806–5812.

(56) F. Trouton. On Molecular Latent Heat Philos. Mag. 1884, 18, 54.

(57) Hildebrand, J.H. The Entropy of Vapor- ization as a Means of Distinguishing Nor- mal Liquids. J. Am. Chem. Soc. 1915, 37 970–978.

(58) D. H. Everett. J. Chem. Soc. 1960, 2566.

(59) Rabinovich, A.; Yasserman, A.A.; Ne- dostup, Y.I.; Veksler, L.S. Thermophys- ical Properties of Neon, Argon, Kryp- ton, and Xenon. Hemisphere, Washington, D.C., 1988.

(60) Streett, W.B.; Staveley, L.A.K. Experi- mental Study of the Equation of State of Liquid Argon. J. Chem. Phys. 1969, 50, 2302–2307.

(61) Letchworth-Weaver, K.; Arias, T.A. Joint Density Functional Theory of the Electrode-Electrolyte Interface: Applica- tion to Fixed Electrode Potentials, Inter- facial Capacitances, and Potentials of Zero Charge. Phys. Rev. B 2012, 86, 075140.

(62) Doncheski, M.A.; Robinett, R.W. Wave Packet Revivals and the Energy Eigen- value Spectrum of the Quantum Pendu- lum. Ann. Phys. 2003, 308, 578–598.