Article pubs.acs.org/JPCC
Solvent-Controlled Shuttling in a Molecular Switch Peng Liu,† Christophe Chipot,‡,§ Xueguang Shao,† and Wensheng Cai*,† †
College of Chemistry, Nankai University, Tianjin, 300071, People’s Republic of China Équipe de Dynamique des Assemblages Membranaires, UMR 7565, Nancy Université, BP 239, 54506 Vandoeuvre-lès-Nancy cedex, France § Theoretical and Computational Biophysics Group, Beckman Institute, University of Illinois at Urbana−Champaign, Urbana, Illinois 61801, United States ‡
S Supporting Information *
ABSTRACT: Rotaxanes driven by solvents have been shown to facilitate translocation of drugs into cells. Shuttling is critical to fulfill this function. Despite the importance of this solventdriven motion, the mechanism that underlies shuttling remains unclear. In the present contribution, a molecular shuttle controlled by solvent, and formed of α-cyclodextrin (α-CD), dodecamethylene, and bipyridinium moieties, has been studied by means of microsecond time scale molecular dynamics simulations combined with free-energy calculations. Shuttling driven by both solvent and temperature has been investigated by determining the potentials of mean force (PMF) that delineate the process of moving the α-CD along the thread in DMSO and water, at 300 and 400 K. In DMSO, the barriers of the PMFs at both temperatures appear to be virtually the same. At low temperature, however, site exchange of the CD is slowed down. In contrast, the barrier in water is shown to be 4.0 kcal/mol higher than in DMSO, thwarting site exchange. Partitioning the PMFs into free-energy components suggests, in contrast with DMSO, that water interacts favorably with the bipyridium moieties, but less so with the alkyl chain, hence yielding a higher free-energy barrier. This observation is supported by the analysis of the structural features of the rotaxanes from the molecular dynamics trajectories.
■
workers8 designed a rotaxane-based carrier. The macrocycle component of the rotaxane can shuttle in response to the environmental change. Further experiments show that shuttling is critical for the permeation of drug-carrier complexes through membranes. Yet the mechanism that underlies shuttling is still not well understood. A cyclodextrin-based molecular shuttle designed by Harada and co-workers6 can be considered as a paradigm for solventdriven rotaxanes. This molecular shuttle is formed by an αcyclodextrin (α-CD), a dumbbell-shaped thread composed of two dodecamethylene units, or stations, three 4,4′-bipyridinium units, or linkers, and two 2,4-dinitrophenyl groups, or stoppers (see Scheme 1). In dimethylsulfoxide (DMSO) at 400 K, shuttling was confirmed by 1H NMR spectra.6 The free-energy change characterizing the shuttling process under the latter condition has been estimated in a preliminary investigation,9 and was found to agree with experiment. Experiment also reveals that cooling the system to 300 K abolishes shuttling.6 Furthermore, changing the solvent to water also disables shuttling.6 The main thrust of the present contribution is to
INTRODUCTION Rotaxanes are mechanically interlocked molecular complexes composed by a linear molecule with stoppers at both termini and a macrocycle threaded onto the latter. This macrocycle can shuttle between two or more stations connected by linkers under external stimuli. This unique property makes rotaxanes promising candidates for molecular machines and devices. In the design of molecular-scale microelectronic devices, quick external triggers and fast response of rotaxanes are important concerns. Much effort has been, therefore, devoted to investigate light- and electro-driven rotaxanes.1,2 Over the past decade, an increasing number of rotaxanes have been designed, synthesized, and characterized. The field of applications has been extended to drug carriers3 and membrane transporters,4 for which a quick response is not required as it would in microelectronics. External triggers are, therefore, not restricted to photons and electrons, but can be other stimuli, such as pH gradient,5 temperature,6,7 or solvent.4,6 Rotaxanes driven by solvent can play an important role to help transport material into biological cells. When a drug molecule enters a cell, it experiences a decay in polarity as the environment changes from the bulk water to the hydrophobic interior of the membrane. This abrupt change of the surroundings creates an obstacle toward drug permeation. In a pioneering effort to tackle this issue, Smithrud and co© 2012 American Chemical Society
Received: November 27, 2011 Revised: January 13, 2012 Published: January 17, 2012 4471
dx.doi.org/10.1021/jp2114169 | J. Phys. Chem. C 2012, 116, 4471−4476
The Journal of Physical Chemistry C
Article
module25 of NAMD. The model reaction coordinate, ξ, was chosen as the distance separating the centroid of the CD from the left linker, as shown in Scheme 1. For numerical efficiency, the pathway, extending from 8 to 36 Å, was broken down into 28 consecutive, 1-Å wide windows. Instantaneous values of the force were accrued in bins, 0.1 Å wide. The variation of the free energy, ΔG(ξ), was determined by integrating the average force acting on ξ. In each window, a trajectory of at least 10 ns was generated. The simulation time was extended incrementally to probe the convergence of the free energy in those regions featuring a barrier. Quasi nonergodicity scenarios prone to occur in multistage approaches21 were probed by running additional ABF simulation using 2.5-Å wide windows embracing the model reaction coordinate. Three ABF simulations, corresponding to the rotaxane immersed in DMSO at 400 K, in DMSO at 300 K, and in water at 300 K, were carried out. The total simulation time amounted to 1.296, 1.014, and 0.368 μs, respectively. Block average regression was applied to estimate the standard error of the free-energy change.22,23,26 To demonstrate that the model reaction coordinate, ξ, used herein corresponds to an appropriate choice, the concept of committor was utilized.26,27 The corresponding calculation and discussion are provided in the Supporting Information.
Scheme 1. [2]Rotaxane Molecule Formed by an α-CD, Two Dodecamethylene Moieties (Stations), Three 4,4′Bipyridinium Moieties (Linkers), and Two 2,4Dinitrophenyl Moieties (Stoppers)
examine the effect of both temperature and solvent on the shuttling process. Toward this end, shuttling of the rotaxane driven by temperature and solvent was investigated by unconstrained molecular dynamics (MD) simulations combined with freeenergy calculations. The free-energy profiles characterizing the shuttling process under three distinct conditions, viz. DMSO at 300 K, DMSO at 400 K, and water at 300 K, were determined. To interpret the physical origin of shuttling, the different contributions extracted from the potentials of mean force (PMFs) were analyzed critically to shed light onto environmental effects.
■
■
SIMULATION DETAILS System Construction. The structure of the rotaxane presented in Scheme 1 was obtained from a previous work.9 The rotaxane was immersed in a box of DMSO and in a box of water, resulting in two molecular assemblies. The initial size was 70.8 × 69.8 × 69.7 Å3 for the DMSO cell, involving 2402 DMSO molecules, and 63.9 × 63.9 × 63.9 Å3 for the water cell, containing 8130 water molecules. The backbone of the rotaxane was softly restrained to its extended conformation to avoid spurious folding of the alkyl chains. Six chloride ions were placed in the solvent box, 10 Å away from the rotaxane to ensure electric neutrality. A soft harmonic potential was used to restrain the position of the counterions. Molecular Dynamics Simulations. All MD simulations were conducted employing the NAMD 2.7b2 program10 with the CHARMM 27 force field.11 A number of parameters describing the linkers and the stoppers of the chain, which are absent in this force field, were optimized in a previous contribution.9 The carbohydrate solution force field (CSFF) parameters12 have been used to represent the α-CD. The TIP3P model13 was used for water. The parameters of DMSO were taken from ref 14. Langevin dynamics was used to control the temperature at 400 and 300 K for the two environments, respectively. The pressure was maintained at 1 atm, employing the Langevin piston method15 for the three assays. Covalent bonds involving hydrogen atom were constrained using the SHAKE/RATTLE algorithms,16,17 except for water, for which the SETTLE algorithm was applied.16 The Lennard-Jones interactions were calculated using a smooth cutoff (10.0/12.0 Å). Long-range electrostatic interactions were determined using the Particle Mesh Ewald method.18 The r-RESPA multiple time step algorithm was applied to integrate the equations of motion with a time step of 2 and 4 fs for short- and long-range interactions, respectively. Visualization and analysis of the MD trajectories were performed with VMD.19 Free-Energy Calculations. The free-energy profiles delineating the shuttling process under the three different conditions were generated using the adaptive biasing force (ABF)20−24 method implemented with the collective variables
RESULTS AND DISCUSSION Free-Energy Profiles. The three free-energy profiles characterizing the shuttling process of the α-CD along the thread in the two environments, at different temperatures, have been gathered in Figure 1. These profiles reveal that (i) each PMF possesses two stable states separated by a significant barrier, (ii) the height of the barrier in DMSO at 400 is identical to that in DMSO at 300 K, and (iii) the barrier, relative to the minima of the free-energy landscape in water at 300 K, is significantly higher than that in DMSO. For each solvent and temperature condition, two relatively flat regions, namely, 8 ≤ ξ ≤ 14 and 30 ≤ ξ ≤ 36 Å, are separated by a broad and high-energy barrier spanning 15 ≤ ξ ≤ 29 Å. These two regions correspond to stable states, wherein the α-CD overlaps with the dodecamethylene chain section of the thread. The barrier reflects an unstable state, wherein the αCD is found on the central bipyridinium moiety. The freeenergy differences between the central linker and the station for the forward (ΔGforward) and the backward (ΔGbackward) transfer processes are reported in Table 1. The free-energy barrier (ΔG‡) is defined as their mean value. The slight asymmetry of the PMF mirrors that of the α-CD. A detailed analysis of the observed asymmetry of the PMF can be found in the Supporting Information. The free-energy barrier, ΔG‡, determined in DMSO at 400 K is equal to 23.3 ± 0.2 kcal/mol, and agrees with the free energy of activation for the site-exchange process determined experimentally.6 This value is essentially the same as the mean value found in DMSO at 300 K. The values of ΔG‡ were introduced into the Eyring equation28 to estimate the characteristic time for shuttling (Table 1). Not unexpectedly, the time in DMSO at 300 K is much longer than that at 400 K. It indicates that site exchange is unlikely to occur or to be observed over a reasonable time scale. This analysis rationalizes the experimental observation that cooling the system from 400 to 300 K abolishes site exchange. The free-energy barrier measured in water at 300 K is 4.0 kcal/mol higher than that determined in DMSO at the same temperature. The extra free 4472
dx.doi.org/10.1021/jp2114169 | J. Phys. Chem. C 2012, 116, 4471−4476
The Journal of Physical Chemistry C
Article
Figure 2. Decomposition of the total free-energy profile into van der Waals CD−thread, electrostatic CD−thread, and CD−solvent contributions for the shuttling process (A) in DMSO at 400 K, (B) in DMSO at 300 K, and (C) in water at 300 K.
Figure 1. Free-energy profiles delineating the shuttling process along ξ: (A) in DMSO at 400 K, (B) in DMSO at 300 K, and (C) in water at 300 K. The error bars represent the standard error of the free-energy difference.23
interaction favors the shuttling process. A closer look at Figure 2 reveals that the depths of the valley for all three conditions are virtually the same. This result is suggestive of a marginal influence of the solvent on the interaction of the thread with the CD. The CD−thread interaction was further decomposed into van der Waals and electrostatics contributions. The former component features for each condition two valleys separated by a barrier. The two valleys stem from favorable van der Waals interactions of the CD with the dodecamethylene chain. The barrier can be ascribed to steric hindrances arising from the overlap of the CD with the central linker. Not unexpectedly, the van der Waals terms in Figure 2A,B are comparable, representing the primary contribution to the complete PMFs. The same cannot be said for Figure 2C, where the height of the barrier is approximately half of that displayed in parts A and B of Figure 2. To delve further into this observation, the geometric change of the α-CD cavity was studied by measuring the average area of the central disk spanned by the six glycosidic oxygen atoms of the CD, as shown in Figure 3A−C. Expansion of the cavity can be visualized in the pronounced barrier, which corresponds to the α-CD passing through the central linker. Comparing the heights of the barriers, it can be inferred that the temperature has a negligible effect on the deformation of the α-CD. Deformation in DMSO is, however, much larger than that in water. The difference in the rigidity may be ascribed to distinct intramolecular hydrogen bonds in the α-CD (see Figure 3D−
Table 1. Activation Free Energiesa system DMSO, 400 K DMSO, 300 K water, 300 K
ΔGforward (kcal/mol)
ΔGbackward (kcal/mol)
ΔG‡ (kcal/mol)
Tshuttling (s)
24.4 ± 0.2
22.1 ± 0.1
23.3 ± 0.2
0.034
25.0 ± 0.2
22.0 ± 0.1
23.5 ± 0.2
2.7 × 102
29.0 ± 0.1
26.5 ± 0.1
27.5 ± 0.1
≥107
a
The experimental free energy of activation for the site-exchange process in DMSO and 400 K is ca. 20 kcal/mol.6
energy leads to a significant increase of the calculated shuttling time in water at 300 K with respect to DMSO at the same temperature (Table 1). Consequently, no shuttling under the former condition was observed by experiment. Driving Force Responsible for Shuttling. To identify the free-energy components that contribute to the higher freeenergy barrier measured in water at 300 K, a number of intermolecular interactions were monitored as a function of time. This was achieved by (i) partitioning the instantaneous force acting along the model reaction coordinate into CD− thread and CD−solvent contributions and (ii) binning and integrating the latter independently. The resulting free-energy components are gathered in Figure 2. CD−Thread Interactions. The CD−thread contributions for all three temperature and solvent conditions are similar. These free-energy components possess a broad and shallow valley spanning 16 ≤ ξ ≤ 30 Å, which indicates that the CD−thread 4473
dx.doi.org/10.1021/jp2114169 | J. Phys. Chem. C 2012, 116, 4471−4476
The Journal of Physical Chemistry C
Article
Figure 3. Fluctuation of the area of the central plane formed by the six glycosidic oxygen atoms of the α-CD for the shuttling process (A) in DMSO at 400 K, (B) in DMSO at 300 K, and (C) in water at 300 K. Evolution of the number of disrupted hydrogen bonds within the α-CD during shuttling, (D) in DMSO at 400 K, (E) in DMSO at 300 K, and (F) in water at 300 K.
F). As can be observed, the number of broken hydrogen bonds during shuttling is the smallest for the rotaxane immersed in water. Analysis of the molecular geometry and the number of disrupted hydrogen bonds confirm that differences in van der Waals CD−chain interactions result from different degrees of α-CD’s deformation. Although van der Waals CD−chain interactions differ, which can be seen from the comparison of the CD-thread(vdw) terms in parts A, B, and C of Figure 2, the sum of van der Waals and electrostatic CD−chain interactions are essentially similar reflected by the CD−thread terms in these figures, hence suggesting that differences in van der Waals interactions can be counterbalanced by the electrostatic term. This result implies that DMSO favors electrostatic interactions of the α-CD with the positively charged bipyridinium moieties. CD−Solvent Interactions. As illustrated in Figure 2, the CD−solvent free-energy component for each solvent and temperature condition possesses a broad, high barrier constituting the main contribution to the barrier of the PMF. The average height of the barrier, relative to the left-hand and right-hand sides of the curves, has been measured as ca. 35, 35, and 40 kcal/mol in DMSO at 400 K, in DMSO at 300 K, and in water at 300 K, respectively. Not too surprisingly, the difference in the height of the barrier primarily stems from the nature of the solvent. To appreciate the effect of the solvent on the CD-solvent interactions, the distribution of the solvent molecules around the central bipyridinium moiety and the left alkyl chain was monitored. This was achieved by computing the bidimensional radial distribution functions, g(r; ξ), of the sulfur atom of DMSO or the oxygen atom of water with respect to the centroid of the central linker (Figure 4A−C) and the left alkyl chain (Figure 4D−F). The maximum of the density found in the region spanning 4 ≤ r ≤ 6 Å corresponds to the first solvation shell of the central linker or the left alkyl chain. In Figure 4A−C, the solvation shell of the central linker is discontinuous along ξ, for 17 ≤ ξ ≤ 29 Å, which means that it is disrupted as the CD moves toward the central linker. In Figure 4D−F, no solvation shell is observed for the left alkyl chain when 8 ≤ ξ ≤ 16 Å. When ξ is comprised between 16 and 36
Figure 4. Evolution of the radial distribution function of the center of mass (central bipyridinium moiety)-S (DMSO)/O (water) pair as a function of r, the distance separating the pair of atoms, for the shuttling process (A) in DMSO at 400 K, (B) in DMSO at 300 K, and (C) in water at 300 K. Evolution of the radial distribution function of the center of mass (left alkyl chain)-S (DMSO)/O (water) pair as a function of r for shuttling (D) in DMSO at 400 K, (E) in DMSO at 300 K, and (F) in water at 300 K.
Å, a solvation shell formsin other words, the solvation shell of the left alkyl chain forms as the CD leaves the left station. In addition, the orientational anisotropy of the solvent molecules near the central linker was measured. In Figure 5A−C, the peak in the distribution of ⟨cos θ⟩ for a radial range characterizing the first solvation shell indicates that the electrostatic 4474
dx.doi.org/10.1021/jp2114169 | J. Phys. Chem. C 2012, 116, 4471−4476
The Journal of Physical Chemistry C
Article
moiety and the depth of the valley is the free energy gained by the solvation of the alkyl chain. No obvious difference was found between the profiles determined in DMSO at 300 and 400 K, in line with the above observation. Disrupting the solvent shell of the thread is, therefore, similar in both conditions. Compared to water, however, shuttling in DMSO requires less energy to desolvate the bipyridinium moiety and results in an energetic gain to solvate one of the alkyl chains, thus rationalizing that the freeenergy barrier arising from unfavorable CD−solvent interactions in DMSO is lower than that in water.
■
CONCLUSION
■
ASSOCIATED CONTENT
In this contribution, the influence of the temperature and the solvent on the shuttling process of rotaxanes has been explored. Shuttling of the α-CD in DMSO and water at different temperatures has been studied by means of free-energy calculations. The barriers of the PMFs measured quantitatively agree well with the activation free energy determined experimentally. The free-energy barriers are virtually identical in DMSO, at two different temperatures, viz. 400 and 300 K. High temperature, however, favors shuttling kinetically. Changing the solvent from DMSO to water increases the free-energy barrier by 4.0 kcal/mol, thereby rationalizing why shuttling of rotaxanes immersed in water is not observed. Partitioning the PMFs into different components reveals that the hydrophobic nature of the alkyl chain and the hydrophilic nature of the bipyridinium moiety contribute predominantly to the increment of the free-energy barrier. Compared with polar solvents, the interactions of these two moieties with nonpolar solvents favor shuttling. Shuttling of this rotaxane could, therefore, be controlled by regulating the chemical properties of the different moieties. Put together, these results provide new insights into the shuttling of rotaxanes and are envisioned to help understand the mechanism underlying rotaxane-aided drug penetration.
Figure 5. Orientational anisotropy of the solvent molecules neighboring the central bipyridinium moiety expressed in terms of the average of cos θ as a function of |R|, where θ is the angle formed between the dipole moment borne by the solvent molecules and R, the vector that connects the center of mass of a solvent molecule to the central bipyridinium moiety for the shuttling process (A) in DMSO at 400 K, (B) in DMSO at 300 K, and (C) in water at 300 K. Orientational anisotropy of the solvent molecules lying near the left alkyl chain for the shuttling process (D) in DMSO at 400 K, (E) in DMSO at 300 K, and (F) in water at 300 K.
interaction of the positively charged moiety and the environment is appreciable. In Figure 5D−F, the absence of a peak implies that no solvation shell is formed. Thread−Solvent Interactions. To ascertain quantitatively changes in the solvation shell around the thread, the central bipyridinium moiety−solvent and the alkane−solvent interactions were computed independently. The corresponding freeenergy components are depicted in Figure 6. Within the region
S Supporting Information *
Calculation and discussion of the committor distribution, exploration of the asymmetry between two stationary states, and comparison of the PMFs of ref 9 and of the present work. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Notes
The authors declare no competing financial interest.
■
Figure 6. Interaction of the central bipyridinium moiety with the solvent and of the alkyl chains with the solvent (DMSO at 400 K, DMSO at 300 K, and water at 300 K).
ACKNOWLEDGMENTS
The study was supported by National Natural Science Foundation of China (Nos. 20873066 and 20835002), and National Basic Research Program of China (No. 2011CB935904). The CINES, Montpellier, France, is gratefully acknowledged for provision of generous amounts of CPU time on their SGI Origin 2000.
spanning ca. 10 ≤ ξ ≤ 32 Å, a high barrier emerges for the bipyridinium−solvent and a shallow valley for the alkane− solvent counterpart. The height of the barrier represents the free energy needed to desolvate the central bipyridinium 4475
dx.doi.org/10.1021/jp2114169 | J. Phys. Chem. C 2012, 116, 4471−4476
The Journal of Physical Chemistry C
■
Article
REFERENCES
(1) Balzani, V.; Clemente-León, M.; Credi, A.; Ferrer, B.; Venturi, M.; Flood, A. H.; Stoddart, J. F. Proc. Natl. Acad. Sci. U. S. A. 2006, 103, 1178−1183. (2) Bissell, R. A.; Córdova, E.; Kaifer, A. E.; Stoddart, J. F. Nature 1994, 369, 133−137. (3) Hernandez, R.; Tseng, H.-R.; Wong, J. W.; Stoddart, J. F.; Zink, J. I. J. Am. Chem. Soc. 2004, 126, 3370−3371. (4) Dvornikovs, V.; House, B. E.; Kaetzel, M.; Dedman, J. R.; Smithrud, D. B. J. Am. Chem. Soc. 2003, 125, 8290−8301. (5) Leung, K. C.-F.; Chak, C.-P.; Lo, C.-M.; Wong, W.-Y.; Xuan, S.; Cheng, C. H. K. Chem. Asian J. 2009, 4, 364−381. (6) Kawaguchi, Y.; Harada, A. Org. Lett. 2000, 2, 1353−1356. (7) Umehara, T.; Kawai, H.; Fujiwara, K.; Suzuki, T. J. Am. Chem. Soc. 2008, 130, 13981−13988. (8) Bao, X.; Isaacsohn, I.; Drew, A. F.; Smithrud, D. B. J. Am. Chem. Soc. 2006, 128, 12229−12238. (9) Liu, P.; Cai, W.; Chipot, C.; Shao, X. J. Phys. Chem. Lett. 2010, 1, 1776−1780. (10) Phillips, J. C.; Braun, R.; Wang, W.; Gumbart, J.; Tajkhorshid, E.; Villa, E.; Chipot, C.; Skeel, R. D.; Kalé, L.; Schulten, K. J. Comput. Chem. 2005, 26, 1781−1802. (11) MacKerell, A. D.; Bashford, D.; Bellott, M.; Dunbrack, R. L.; Evanseck, J. D.; Field, M. J.; Fischer, S.; Gao, J.; Guo, H.; Ha, S.; et al. J. Phys. Chem. B 1998, 102, 3586−3616. (12) Kuttel, M.; Brady, J. W.; Naidoo, K. J. J. Comput. Chem. 2002, 23, 1236−1243. (13) Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R. W.; Klein, M. L. J. Chem. Phys. 1983, 79, 926−935. (14) Strader, M. L.; Feller, S. E. J. Phys. Chem. A 2002, 106, 1074− 1080. (15) Feller, S. E.; Zhang, Y. H.; Pastor, R. W.; Brooks, B. R. J. Chem. Phys. 1995, 103, 4613−4621. (16) Ryckaert, J.-P.; Ciccotti, G.; Berendsen, H. J. C. J. Comput. Phys. 1977, 23, 327−341. (17) Andersen, H. C. J. Comput. Phys. 1983, 52, 24−34. (18) Darden, T.; York, D.; Pedersen, L. J. Chem. Phys. 1993, 98, 10089−10092. (19) Humphrey, W.; Dalke, A.; Schulten, K. J. Mol. Graph. 1996, 14, 33−38. (20) Darve, E.; Pohorille, A. J. Chem. Phys. 2001, 115, 9169−9183. (21) Darve, E.; Rodríguez-Gómez, D.; Pohorille, A. J. Chem. Phys. 2008, 128, 144120. (22) Hénin, J.; Chipot, C. J. Chem. Phys. 2004, 121, 2904−2914. (23) Rodriguez-Gomez, D.; Darve, E.; Pohorille, A. J. Chem. Phys. 2004, 120, 3563−3578. (24) Chipot, C.; Hénin, J. J. Chem. Phys. 2005, 123, 244906. (25) Hénin, J.; Fiorin, G.; Chipot, C.; Klein, M. L. J. Chem. Theory Comput. 2010, 6, 35−47. (26) Bolhuis, P. G.; Chandler, D.; Dellago, C.; Geissler, P. L. Annu. Rev. Phys. Chem. 2002, 53, 291−318. (27) Pan, A. C.; Sezer, D.; Roux, B. J. Phys. Chem. B 2008, 112, 3432−3440. (28) Eyring, H. J. Chem. Phys. 1935, 3, 107−115.
4476
dx.doi.org/10.1021/jp2114169 | J. Phys. Chem. C 2012, 116, 4471−4476