Sie sind auf Seite 1von 7

Chemical Physics Letters 481 (2009) 28

Contents lists available at ScienceDirect

Chemical Physics Letters


journal homepage: www.elsevier.com/locate/cplett

FRONTIERS ARTICLE

Hydroxide anion at the airwater interface


Christopher J. Mundy a, I-Feng W. Kuo b, Mark E. Tuckerman c, Hee-Seung Lee d, Douglas J. Tobias e,*
a

Chemical and Materials Sciences Division, Pacic Northwest National Laboratory, Richland, WA 99352, USA Chemical Sciences Division, Lawrence Livermore National Laboratory, Livermore, CA 94550, USA c Department of Chemistry, Courant Institute of Mathematical Sciences, New York University, New York, NY 10003, USA d Department of Chemistry and Biochemistry, University of North Carolina, Wilmington, NC 28403, USA e Department of Chemistry, AirUCI, University of California, Irvine, CA 92697-2025, USA
b

a r t i c l e

i n f o

a b s t r a c t
Whether aqueous interfaces are acidic or basic has implications for interfacial chemistry, but the question remains open. We employ rst-principles molecular dynamics simulations to determine the intrinsic propensity of OH for the airwater interface and nd that OH is stabilized by roughly kBT at the interface vs. the bulk. We predict, therefore, that the surface population OH is slightly enhanced. Our simulations suggest that the solvation of OH at the interface is similar to that observed in small water clusters, and they reveal changes in the orientation of solvating water molecules that are consistent with surface-sensitive vibrational spectra. 2009 Published by Elsevier B.V.

Article history: Received 20 August 2009 In nal form 2 September 2009 Available online 4 September 2009

1. Introduction Reactive ions at airwater interfaces are implicated in numerous heterogeneous chemical processes involving aqueous aerosols and thin lms in the atmosphere. Although a consensus appears to have emerged between various theoretical calculations and experiments on the distributions of several inorganic ions in the interfacial region of aqueous solutions [1], there is considerable disagreement on the interfacial propensities of the water ions, i.e., excess protons and hydroxide anions, formed by water autodissociation [25]. The relative populations of the water ions at an aqueous interface vs. in bulk water determine whether the interfacial region is neutral, acidic, or basic. The acidity/basicity of aqueous interfaces could have implications for interfacial chemistry, e.g., by modifying the relative populations of reactive acids such as HNO3 and H2SO4 and their conjugate bases in atmospheric aerosols. Empirical force eld-based and rst-principles MD simulations consistently agree that the excess proton (H+), in the form of hydronium (H3O+) and Zundel (H5 O 2 ) cations, has a propensity for the interface [69]. MD simulations based on polarizable force elds predict that the hydroxide (OH) anion is repelled from the airwater interface [7], while a non-polarizable empirical valence bond model suggests that the population of OH at the interface is the same as in the bulk [10]. The theoretically predicted H+ enhancement has recently led to the assertion that the surface of

* Corresponding author. Address: Department of Chemistry, 1102 Natural Sciences 2, University of California, Irvine, CA 92697-2025, USA. Fax: +1 (949) 824 9920. E-mail address: dtobias@uci.edu (D.J. Tobias). 0009-2614/$ - see front matter 2009 Published by Elsevier B.V. doi:10.1016/j.cplett.2009.09.003

neat water is acidic, with pH < 4.8 [2], which is substantially lower than the bulk value, pH 7. There is considerable disagreement amongst experiments probing the interfacial propensities of H+ and OH. Macroscopic measurements of the transport of air bubbles and oil droplets in water have been interpreted in terms of a negative surface charge, which is attributed to excess surface OH [4]. Surface-sensitive spectroscopic experiments suggest that excess protons are present in the interfacial region [1114], but corresponding data for OH could be interpreted in terms of a wide range of behavior, from slight enhancement to strong repulsion, but not strong enhancement [5,14]. Recent X-ray photoemission measurements, supported by force eld-based MD simulations, appear to be incompatible with strong adsorption of OH [15], while Cs+ reactive ion scattering and low energy sputtering experiments have revealed that OD anions migrate to the surface of ice lms as the temperature is raised following the hydrolysis of sodium atoms by D2O [16]. It is clear that the acidity/basicity of the water surface is still an open question from both the experimental and theoretical perspectives, and that the extent of adsorption or depletion of OH at the airwater interface is a major source of uncertainty. Here we report an investigation of the behavior of OH at the airwater interface based on rst-principles molecular dynamics (FPMD) simulations. Accurate simulation of the behavior of OH in water requires an approach that accounts for its chemical reactivity, and its exotic electronic structure and solvation shell [17]. FPMD simulation, in which the nite-temperature dynamics of a system is driven by forces computed from on the y electronic structure calculations, is a powerful approach for treating the delicate energetics at play when OH interacts with water. FPMD

C.J. Mundy et al. / Chemical Physics Letters 481 (2009) 28

simulations allow chemical bond-breaking and bond-forming events to be studied naturally, without prior knowledge of the underlying electronic processes, and they also explicitly include the many-body polarization effects associated with ion solvation. Indeed, the microscopic behavior of OH in bulk water was rst predicted by FPMD simulations [17,18]. The calculations suggested that OH in bulk water can accept up to four hydrogen bonds through the hydroxyl oxygen (O*) in a roughly square-planar arrangement. This hypercoordination of O* is unexpected, but can be explained by a delocalization of the three lone pairs of electrons [17]. The simulations also showed that OH has an anomalous transport mechanism involving proton transfer reactions from water molecules in the rst solvation shell of O*, which are initiated by a change in coordination number from four to three, and they predicted that OH is able to donate a weak hydrogen bond that is stabilized by the reduction in coordination number [17]. It is the combination of coordination number reduction and hydrogen-bond donation that drives the anomalous transport [17,19] and determines the corresponding kinetics [20]. The FPMD predictions were subsequently conrmed experimentally via neutron diffraction [21,22], X-ray absorption spectroscopy [23], X-ray diffraction [24,25], Fourier-transform infrared spectroscopy [26], and, most recently, in novel corehole spectroscopic measurements [27], as well as by additional FPMD calculations [17,20]. In an accompanying Frontiers article, Sun and Dang report a classical MD simulation study of the reorientational dynamics of OH in water based on an empirical polarizable potential [28]. This study underscores the connection between the peculiarities of the solvation shell of the hydroxide anion and its dynamics and transport in aqueous solution. In this Letter we report free energy calculations based on FPMD simulations that suggest that OH has a weak intrinsic propensity for the airwater interface. We analyze the solvation of OH at the interface and nd that it is similar to that observed in small water clusters. We also observe changes in the orientation of solvating water molecules that are consistent with surface-sensitive vibrational spectra. Our extensive FPMD calculations provide benchmark data for the further development of a molecular-scale understanding of acidbase chemistry at aqueous interfaces.

71.44 box with periodic boundary conditions applied in three dimensions, resulting in a slab of aqueous solution 31 thick with a bulk concentration [OH]  0.25 M, or pH  13. No counterions were included in the simulation, and the net charge of the system was compensated by a neutralizing background charge. The system size is large by current standards of FPMD simulations, but necessary to support stable airwater interfaces [35,36]. The simulations were performed in the canonical ensemble at 300 K by coupling NosHoover chain thermostats with characteristic frequencies of 2000 cm1 to every degree of freedom [37]. The timestep of the simulations was 0.5 fs. At each timestep of the simulations, the wavefunction was minimized to a tolerance of 106 H. 2.3. Free energy calculations A complication that arises in FPMD simulations of H+ and OH in water is that these species can change positions discontinuously as a result of proton donation to/acceptance from a neighboring water molecule. To circumvent this problem in our thermodynamic calculations, we restrained a coordinate, d = r(O*H) r(O0 H), where r(O*H) and r(O0 H) are the distances between the hydroxide oxygen (O*) and the oxygen (O0 ) of the closest water molecule to the hydroxide ion and the proton (H) shared between them, respectively; d  0.5 for a well-dened OH ion with O* solvated by four water molecules, and d  0.0 for a species (H3 O 2 ) in which a proton is being shared by O* and O0 [17,19]. Two series of restrained FPMD simulations were performed to calculate the free energies as either a function of interfacial depth (z) or the extent of proton transfer (d) of the hydroxide anion. We used one harmonic restraining potential,

U z K z z z0 2 ;

to keep the hydroxide oxygen atom positioned in the vicinity of, z0, the position of the ion relative to the total center of mass of the slab, and another

U d2 K d d2 d2 0 2 ;

2. Methods 2.1. First-principles molecular dynamics simulations The simulations were carried out using the CP2K software suite (http://cp2k.berlios.de). The forces were computed via the QuickStep module, which contains an accurate and efcient implementation of density functional theory using dual basis sets of Gaussian type orbitals (TZV2P) and plane waves (expanded to Ecut = 280 Ry) for the electron density [29]. Only the valence electrons were considered explicitly, and the core electron states were represented via GoedeckerTeterHutter pseudopotentials [30]. The Grimme dispersion correction was used in conjunction with the Becke exchange and LeeYangParr correlation (BLYP) functional [3133]. As will be demonstrated below, the BLYP/TZV2P theoretical model provides a reasonable description of the effects of solvation on the proton transfer energetics compared to the higher level MP2/TZV2P model. The dispersion correction has recently been shown to vastly improve the density and structure of the BLYP description of water at near ambient conditions [34]. 2.2. System setup and simulation protocols All of the systems simulated using FPMD consisted of one hydroxide anion and 215 water molecules in a 15 15

to hold the value of d2 near a reference value, (d2)0. To map the PMF as a function of z, a series of FPMD simulations was performed in which the hydroxide anion was subjected to the restraint potential U(z) (Eq. (1)) at a set of interfacial depths, z0, spaced every 1.06 apart, starting from the bulk region and extending beyond the Gibbs dividing surface (GDS, dened here as the point at which the liquid and vapor densities are equal) of the slab. In this rst series of simulations proton transfer was suppressed by employing the restraint potential U(d2) (Eq. (2)) with d0 = 0.45. To quantify the stabilization/destabilization of the hydroxide ion due to delocalization toward H3 O 2 (i.e., d < 0.45), a second series of simulations was performed to map the PMF as a function of d using umbrella sampling and the restraint potential U(d2) (Eq. (2)), both in the bulk (z0 = 0.0 ) and at the GDS (z0 = 15.4). The interfacial depth (z) and proton transfer coordinate (d) were harmonically restrained using Eqs. (1) and (2) with force constants of Kd = 0.1 H bohr4 and Kz = 0.1 H bohr2, respectively. Each of the FPMD simulations was at least 15 ps long, of which 45 ps was considered equilibration. The PMF as a function of z was mapped using 14 simulations, and each of the two PMFs as a function of d required 7 simulations. Thus, a total of >400 ps of FPMD trajectories were generated in this investigation. 2.4. Quantum chemical calculations Quantum chemical calculations were carried out to assess the accuracy of the TZV2P basis set for the study of solvated hydroxide ions. The calculations were carried out at the MP2 or DFT level of

C.J. Mundy et al. / Chemical Physics Letters 481 (2009) 28

theory with the TZV2P basis set using the GAMESS quantum chemistry package [38,39]. Unlike the FPMD simulations, pseudopotentials were not used, i.e., the quantum chemical calculations are all-electron. For the DFT calculations, the BLYP [32,33] exchange correlation functional was used. Since solvated hydroxide ions have not been studied yet with the TZV2P basis set, a test calculation of the optimized geometry of H3 O 2 was performed using the MP2/aug-cc-pVDZ basis set, and the result was compared with the literature. The resulting optimized structure agrees with the previous work of Xantheas [40]; all interatomic distances are within 0.002 and angles are within 0.3 except the dihedral angle, which differs by 0.6.

3. Results 3.1. Calibration of the FPMD simulation methodology with quantum chemical calculations In order to compute the proton transfer barrier of gas phase H3 O 2 , the structure of H3 O2 was rst optimized and the equilibeq rium OO distance (r OO ) was obtained; we found that r eq OO = 2.46 using MP2/TZV2P and 2.49 using BLYP/TZV2P. Note that the shared proton (H*) and two oxygen atoms are not collinear in the equilibrium structure. In subsequent calculations, the OO distances were xed at r eq OO , and the shared proton was dragged along the line connecting the two oxygen atoms. At each location of the shared proton, all of the remaining coordinates were relaxed. The resulting potential energy curves for proton transfer are plotted in Fig. 1A for H3 O 2 . We rst note that the BLYP/TZV2P level of theory shows a single well potential with no proton transfer barrier. The MP2/TZV2P calculation also leads to a nearly at potential energy curve, but there is a very small (0.03 kcal/mol) barrier. A few high level ab initio calculations have been reported on the proton transfer barrier of H3 O 2 , and the reported barrier heights are quite sensitive to the level of theory and the size of the basis set [4144], ranging from 0.08 kcal/mol with the MP2/aug-cc-pVTZ to 7.19 kcal/mol with the MP2/6-311++G** models. Although the proton transfer barrier in the gas phase is insignicant with the TZV2P basis, microsolvation of the H3 O 2 ion changes this picture drastically. To investigate the effect of solvation, we computed the proton transfer barrier of H3 O 2 (H2O)4, where each oxygen atom in H3 O 2 receives hydrogen bonds from two neighboring water molecules. A local minimum with a linear OH*O structure was obtained, and the equilibrium OO distances were found at r eq OO = 2.51 for MP2/TZV2P and req OO = 2.52 for BLYP/TZV2P. Constrained minimizations, in which the OO distances are xed at req OO , were performed with the location of the shared proton xed at particular distances from the mid point of the two oxygen atoms, ranging from 0 to 0.3 with an increment of 0.02 . A few extra calculations were carried out around the saddle point and the minimum. The potential energy curves thus obtained are shown in Fig. 1B. It is evident that microsolvation creates a proton transfer barrier (0.8 kcal/mol) in the BLYP/TVZ2P basis set calculation. The presence of a barrier upon solvation also agrees with the results obtained from plane-wave DFT calculations [17,42]. The location of the minimum is around d  0.35 , which is similar to what we observed for the undersolvated H3 O 2 ion on the interface in the FPMD simulations reported below. This result implies that solvation pushes the location of the shared proton to one of the oxygen atoms in the BLYP/TZV2P calculation. MP2 calculations with a TZV2P basis set predict similar behavior, and the barrier height increases from 0.03 kcal/mol to 1.5 kcal/ mol upon solvation with four water molecules. According to our prior experience with the hydrated proton [42,45] and hydroxide ion [17,42], proton transfer occurs in a short period of time. The rate-limiting step is the preorganization of sol-

Fig. 1. Dependence of proton transfer energetics on electronic structure method in H3 O 2 and H3 O2 (H2O)4. Proton transfer barriers obtained from BLYP/TZV2P (black) and MP2/TZV2P (red) calculations for (A) gas phase H3 O 2 , (B) H3 O2 (H2O)4 with the water structure relaxed, and (C) H3 O 2 (H2O)4 with a frozen water structure. The coordinate d, dened in the main article, is a coordinate measuring the extent of proton transfer: When d = 0, a proton is shared between OH and the nearest water molecule; d  0.4 corresponds to a hydrogen bond between OH and the nearest water molecule.

vent molecules, rather than the proton transfer itself. What we normally observed in bulk simulations is that the proton is transferred rather quickly once the solvation shell structure is reorganized and properly prepared for the proton transfer (referred to as the presolvation concept [17]). As a result, the proton transfer barriers reported in Fig. 1B are probably underestimated. Therefore, it is useful to estimate the proton transfer barrier from calculations with all atoms except the shared proton frozen at the equilibrium structure. Single point calculations with the shared proton dragged along the line between the two oxygen atoms were performed with all other atoms at their equilibrium positions. The resulting potential energy curve is reported in Fig. 1C. Because the structure is not relaxed, the potential energy curve is not smooth in the middle. The barrier heights obtained from these calculations are 1.7 kcal/mol for the BLYP/TZV2P basis set and 2.3 kcal/mol for MP2/TZV2P basis set. MP2 calculations consistently lead to a higher proton transfer barrier, but the general trend is identical in the MP2 and DFT calculations. The proton transfer barriers shown in Fig. 1C are also in the same range as what we observed in bulk simulations of the hydrated hydroxide ion using plane-wave DFT methods [17]. 3.2. Free energy prole for OH adsorption at the airwater interface We determined the propensity of OH for the neat airwater interface by mapping the free energy prole (potential of mean

C.J. Mundy et al. / Chemical Physics Letters 481 (2009) 28

force, PMF) for bringing the ion from the bulk to the airwater interface using density functional-based FPMD simulations. We calculated the PMF, DF(z), from the integral of the d- and z-biased average force on the z-coordinate, hF ziz0 , using a series of simulations with OH restrained at selected values of z0 ranging from the bulk to the interface, and d restrained near 0.45, which is the minimum free energy value for OH in bulk water [17]. To quantify the stabilization/destabilization due to delocalization of OH toward H3 O 2 (i.e., d < 0.45), we also mapped the PMF as a function of d using umbrella sampling (see below). The PMF for the transport of OH through the water slab is plotted along with the water density prole in Fig. 2. The latter shows that bulk water density predicted by our theoretical model, 0.92 g/ cm3, is in reasonable agreement with the experimental value of 1.00 g/cm3 at 300 K and 1 atm pressure, and that the location of the GDS is z = 15.4 . The PMF is at in the interior of the slab up to z  12 , exhibits a shallow minimum centered at the GDS, and begins to rise as the ion enters the vapor (i.e., z > 16 ). Thus, we predict that OH is equally likely to be anywhere in the slab up to just below the GDS, and is slightly more stable (by about 0.5 kcal/mol) on the surface vs. in the bulk. Although our PMF is based on simulations that are relatively short compared to those obtained with empirical potentials, the mean force and corresponding PMF are clearly indicative of weak adsorption. 3.3. Potential of mean force along the proton transfer coordinate The PMFs computed by umbrella sampling as a function of the proton transfer coordinate d in the bulk and at the GDS are shown in Fig. 3. The bulk result has a minimum at d = 0.47, and is essentially the same as a PMF derived previously from a histogram of d in an unrestrained FPMD simulation of OH in bulk water [17]. At the GDS the minimum is shifted to d = 0.38, indicating that the ion is somewhat delocalized toward H3 O 2 in its most stable

Fig. 3. Proton transfer energetics of the hydroxide anion. Potentials of mean force computed as a function of the proton transfer coordinate d for the hydroxide anion in bulk water (red circles) and at the airwater interface (blue circles). The continuous lines represent ts to fourth-order polynomials meant to provide guides to the eye.

conguration at the airwater interface. Moreover, the barrier to proton transfer appears to be reduced at the GDS vs. in the bulk. This is consistent with quantum chemical calculations reported above, which show that the proton transfer barrier decreases with desolvation (Fig. 1). Thus, we expect that nuclear quantum effects [17] would further delocalize the hydroxide anion at the airwater interface and increase its H3 O 2 character. The difference between the two PMFs at d = 0.47, 0.1 kcal/mol, is the additional stabilization imparted to the ion by shifting the free energy minimum to d = 0.38 at the GDS. Combining the information contained in the PMFs as functions of z and d, our FPMD simulations predict that the hydroxide anion is stabilized by 0.6 kcal/mol at the interface vs. in the bulk, with a corresponding enhancement in the population at the interface of e0:6 kcal=mol=kB T  2:7 at 300 K. 3.4. Bulk vs. interfacial solvation of OH The radial distributions of water oxygen (Ow) and hydrogen (H) atoms around the hydroxide oxygen (O*) and hydrogen (H*) atoms plotted in Fig. 4A and B for the ion in the bulk show that our FPMD simulations reproduce the square-planar solvation of O* and the average coordination number of 4.4 obtained in previous FPMD studies [17,20]. This peculiar solvation structure arises from the ring-shaped distribution of electrons on O* (Fig. 4C), which is a departure from the three lone pairs of electrons expected based on conventional valence bond concepts (i.e., Lewis structures). As the ion approaches the airwater interface (GDS), it loses the water molecule hydrogen-bonded to H*, and one of the water molecules solvating O* (Fig. 4D and F), so that the average coordination number of the ion is 3.0. The remaining three water molecules in the rst solvation shell are in a roughly trigonal-planar arrangement around O* (Fig. 4F). The sharpening of the peaks of the O*Ow and O*H g(r) indicates that the solvation shell is more ordered at the GDS vs. in the bulk (Fig. 4A and D). The arrangement of water molecules around the ion at the GDS, depicted in Fig. 4F, is remarkably similar to the structures of OH(H2O)3 and OH(H2O)4 clusters inferred from vibrational spectra and quantum chemical calculations [46]. It is also consistent with a FPMD study of a OH(H2O)6 cluster, in which an initial threefold-coordinated hydroxide core reaches a state in which two threefold-coordinated OH moieties at the cluster edge frequently shuttle a proton between them across the connecting hydrogen bond [47]. This study shows that, at the cluster edge, coordination number and the proton transfer barrier are both reduced, as is seen at the GDS in the present study.

Fig. 2. Free energy prole for hydroxide anion adsorption at the airwater interface. (A) Snapshot from a rst-principles MD simulation of the hydroxide anion (blue) at the airwater interface. (B) Water density prole (black circles, left axis) and potential of mean force, DF(z), (blue circles and line, right axis) for moving the hydroxide anion through the water slab. The Gibbs dividing surface, determined from a t of the density prole to a hyperbolic tangent function (black line), is indicated by the vertical dashed line at z = 15.4 .

C.J. Mundy et al. / Chemical Physics Letters 481 (2009) 28

Fig. 4. Aqueous solvation and electronic structure of the hydroxide anion. (A) Radial distribution functions, g(r), of water oxygen (Ow) and hydrogen (H) atoms around the oxygen atom (O*) of the hydroxide anion in bulk water. (B) Radial distribution function of water oxygen atom around the hydrogen atom (H*) of the hydroxide anion in bulk water. (C) Electron localization functions [52] (gray surfaces) depicting the location of the valence electrons of the hydroxide anion (blue) and water molecules (red) in its solvation shell in bulk water. (D)(F) same as (A)(C), but for the ion at the interface. In panels (A, B) and (D, E), the dashed lines depict the running coordination numbers derived from the corresponding radial distribution functions.

broad, the most probable dipole orientation points slightly toward the bulk liquid (cos(h) < 0), in agreement with SHG data [48]. At the ion-containing interface the distribution shifts toward cos(h) > 0, signaling a reorientation such that more of the water dipoles are pointing out of the liquid, consistent with the interpretation of hydroxide-induced changes in phase-sensitive SFG spectra [14]. Fig. 5 also shows that, while the OH bond samples a broad range of orientations at the interface, it mostly points out of the solution, as expected based on the weak hydrogen bond donor character of the hydroxide hydrogen. 4. Discussion
Fig. 5. Water and hydroxide anion orientational probability distributions at the air water interface. Distributions of the cosine of the angle, h, between the water dipole or hydroxide anion OH bond and a vector normal to the airwater interface. Black: water dipole at the ion-free interface of the slab; red: water dipole at the ioncontaining interface; blue: hydroxide anion OH bond.

3.5. Comparison with spectroscopic measurements Our calculations are supported by surface-sensitive nonlinear spectroscopies, particularly second harmonic generation (SHG) and vibrational sum frequency generation (SFG). SHG has been used to probe the presence of hydroxide at the interface via the electronic charge transfer to solvent transition [5]. The concentration dependence of the SHG intensity could be tted by a Langmuir adsorption isotherm with a range of free energies of adsorption (1 to +10 kcal/mol) consistent with a slight enhancement to a strong depletion, but inconsistent with a strong adsorption of OH to the interface. SFG spectra have revealed changes in the vibrational spectra of water molecules at the interface of hydroxide solutions that imply the presence of OH in the interfacial region [1214], but the measurements do not enable the relative populations of ions at the interface and in the bulk to be determined. Phase-sensitive SFG spectra have been interpreted in terms of a change of water orientation in the vicinity of OH at the interface [14]. Our simulations are consistent with this interpretation, as illustrated by the water dipole orientational distributions plotted in Fig. 5. Although the distribution at the ion-free interface is

The thermodynamic calculations reported here, which were based on rst-principles MD simulations, predict a small stabilization of the hydroxide anion at the airwater interface relative to bulk water by kBT at ambient temperature. Moreover, the calculations suggest a slight propensity for the hydroxide anion to delocalize toward the H3 O 2 species at the interface. Although the composition of the system considered here, which consisted of a single OH anion and 215 water molecules, has a nominal OH concentration of 0.25 M, our calculations address the intrinsic propensity of the OH for the neat water surface because they are carried out with periodic boundary conditions that generate semi-innite slabs of water (in the direction parallel to the interface), in the absence of counterions (e.g., excess protons in the case of neat water, or other cations in the case of strong basic solutions). The connection of our results to the concentration proles of the OH anion in the presence of counterions will require the solution of the problem of interacting ions [49], and it is expected that the propensity of OH for the airwater interface will be modied by interactions with counterions. Nonetheless, the single ion potential of mean force for hydroxide adsorption to the airwater interface obtained in this investigation is dramatically different than the corresponding results obtained with empirical, polarizable force eld models, but it appears to be consistent with a prediction based on a non-polarizable empirical valence bond model [10]. Moreover, the present study reveals novel solvation of the OH anion in the vicinity of the interface, and suggests that the OH anion

C.J. Mundy et al. / Chemical Physics Letters 481 (2009) 28

can be stabilized slightly at the airwater interface by changing its solvation structure in a way that is reminiscent of the microsolvation of OH in small water clusters [46]. Our results constitute an important step forward in the microscopic understanding of acidbase chemistry at aqueous interfaces [24]. Potentials of mean force computed from MD simulations based on empirical polarizable potentials [3,50], empirical valence bond models [8], and FPMD simulations employing the BLYP functional [9] predict that the excess proton is stabilized by 13 kcal/ mol at the airwater interface vs. bulk water, which is substantially greater than the stabilization that we predict for OH. Thus, in agreement with a recent thermodynamic analysis of surface tension data [51], the consensus of theoretical studies to date is that the excess proton is accumulated more than OH at the interface. However, the calculations reported here, which employ an FPMD protocol that has successfully predicted the bulk solvation and transport mechanism of OH, suggest that the surface enrichment of protons over hydroxide ions, and hence the surface acidity, is substantially less than was previously suggested [2,3]. Acknowledgements D.J.T. and M.E.T. acknowledge support from the National Science Foundation (Grants CHE-0431512 and CHE-0704036). H.S.L. was supported by a summer research initiative and the Cahill research fund at UNCW. The work of I.F.W.K. was performed under the auspices of the US Department of Energy (DOE) at Lawrence Livermore National Laboratory (LLNL) under Contract DE-AC5207NA27344. C.J.M. is supported by the DOE Ofce of Basic Energy Sciences Chemical, Geosciences, and Biosciences division. All of the bulk calculations were performed using the computing resource NWice located in the Environmental Molecular Sciences Laboratory at Pacic Northwest National Laboratory (PNNL). The interface calculations were enabled by a 20082009 INCITE award to C.J.M. on the CRAY XT4 (using resources of the National Center for Computational Sciences at Oak Ridge National Laboratory (ORNL), which is supported by the Ofce of Science of the U.S. DOE under Contract No. DE-AC05-00OR22725) and the BlueGene/ P at Argonne National Laboratory (resources of the Argonne Leadership Computing Facility at Argonne National Laboratory, which is supported by the Ofce of Science of the U.S. DOE under contract DE-AC02-06CH11357). C.J.M. and D.J.T. also acknowledge NERSC 2008 Early Use awards on the CRAY XT5 at ORNL. I.F.W.K. acknowledges computer time allocated via the Computational Grand Challenge awards at LLNL. D.J.T. acknowledges a grant of computer time from the Molecular Sciences Computing Facility at PNNL. References
[1] P. Jungwirth, D.J. Tobias, Chem. Rev. 106 (2006) 1259. [2] V. Buch, A. Milet, R. Vcha, P. Jungwirth, J.P. Devlin, Proc. Natl. Acad. Sci. USA 104 (2007) 7342. [3] R. Vcha, V. Buch, A. Milet, J.P. Devlin, P. Jungwirth, Phys. Chem. Chem. Phys. 9 (2007) 4736. [4] J.K. Beattie, A.M. Djerdjev, G.G. Warr, Faraday Discuss. 141 (2009) 31. [5] P.B. Petersen, R.J. Saykally, Chem. Phys. Lett. 458 (2008) 255. [6] L.X. Dang, J. Chem. Phys. 119 (2003) 6351. [7] M. Mucha, T. Frigato, L.M. Levering, H.C. Allen, D.J. Tobias, L.X. Dang, P. Jungwirth, J. Phys. Chem. B 109 (2005) 7617. [8] S. Iuchi, H. Chen, F. Paesani, G.A. Voth, J. Phys. Chem. B 113 (2009) 4017. [9] H.-S. Lee, M.E. Tuckerman, J. Phys. Chem. A 113 (2009) 2144. [10] C.D. Wick, L.X. Dang, J. Phys. Chem. A 113 (2009) 6356. [11] P.B. Petersen, R.J. Saykally, J. Phys. Chem. B 109 (2005) 7976. [12] T.L. Tarbuck, S.T. Ota, G.L. Richmond, J. Am. Chem. Soc. 128 (2006) 14519. [13] L.M. Levering, M.R. Sierra-Hernndez, H.C. Allen, J. Phys. Chem. C 111 (2007) 8814. [14] C. Tian, N. Ji, G.A. Waychunas, Y.R. Shen, J. Am. Chem. Soc. 130 (2008) 13033. [15] B. Winter, M. Faubel, R. Vcha, P. Jungwirth, Chem. Phys. Lett. 474 (2009) 241. [16] J.-H. Kim, Y.-K. Kim, H. Kang, J. Phys. Chem. C 113 (2009) 321.

[17] M.E. Tuckerman, D. Marx, M. Parrinello, Nature 417 (2002) 925. [18] M.E. Tuckerman, K. Laasonen, M. Sprik, M. Parrinello, J. Chem. Phys. 103 (1995) 150. [19] M.E. Tuckerman, C.A.D. Marx, Acc. Chem. Res. 39 (2006). [20] A. Chandra, M.E. Tuckerman, D. Marx, Phys. Rev. Lett. 99 (2007) 145901. [21] A. Botti, F. Bruni, S. Imberti, M.A. Ricci, A.K. Soper, J. Chem. Phys. 119 (2003) 5001. [22] S. Imberti, A. Botti, F. Bruni, G. Cappa, M.A. Ricci, A.K. Soper, J. Chem. Phys. 122 (2005) 194509. [23] C.D. Cappa, J.D. Smith, B.M. Messer, R.C. Cohen, R.J. Saykally, J. Phys. Chem. A 111 (2007) 4776. [24] T. Megyes, S. Balint, T. Grosz, T. Radnai, I. Bako, P. Sipos, J. Chem. Phys. 128 (2008) 044501. [25] R. Vcha, T. Megyes, I. Bak, L. Pusztai, P. Jungwirth, J. Phys. Chem. A 113 (2009) 4022. [26] M. Smiechowski, J. Stangret, J. Phys. Chem. A 111 (2007) 2889. [27] E.F. Aziz, N. Ottosson, M. Faubel, I.V. Hertel, B. Winter, Nature 455 (2008) 89. [28] Sun, L.X. Dang, Chem. Phys. Lett. XXX (in press). [29] J. VandeVondele, M. Krack, F. Mohamed, M. Parrinello, T. Chassaing, J. Hutter, Comput. Phys. Commun. 167 (2005) 103. [30] S. Goedecker, M. Teter, J. Hutter, Phys. Rev. B 54 (1996) 1703. [31] S. Grimme, J. Comput. Chem. 27 (2006) 1787. [32] A.D. Becke, Phys. Rev. A 38 (1988) 3098. [33] C. Lee, W. Yang, R.C. Parr, Phys. Rev. B 37 (1988) 785. [34] J. Schmidt, J. VandeVondele, I-F.W. Kuo, D. Sebastiani, J.I. Siepmann, J. Hutter, C.J. Mundy, J. Phys. Chem. B (in press). [35] I-F.W. Kuo, C.J. Mundy, Science 303 (2004) 658. [36] C.J. Mundy, I-F.W. Kuo, Chem. Rev. 106 (2006) 1282. [37] G.J. Martyna, M.E. Tuckerman, M.L. Klein, J. Chem. Phys. 97 (1992) 2635. [38] M.S. Gordon, M.W. Schmidt, in: C.E. Dykstra, G. Frenking, K.S. Kim, G.E. Scuseria (Eds.), Theory and Applications of Computational Chemistry: The First Forty Years, Elsevier, Amsterdam, 2005, p. 1167. [39] <http://www.msg.chem.iastate.edu/GAMESS/GAMESS.html>. [40] S.S. Xantheas, J. Am. Chem. Soc. 117 (1995) 10373. [41] J.A. Platts, K.E. Laidig, J. Phys. Chem. 100 (1996) 13445. [42] M.E. Tuckerman, D. Marx, M.L. Klein, M. Parrinello, Science 275 (1997) 817. [43] C.C.M. Samson, W. Klopper, J. Mol. Struct. (Theochem) 586 (2002) 201. [44] A. Ponti, M. Mella, J. Phys. Chem. A 107 (2003) 7589. [45] D. Marx, M.E. Tuckerman, J. Hutter, M. Parrinello, Nature 397 (1999) 601. [46] W.H. Robertson, E.G. Diken, E.A. Price, J.-W. Shin, M.A. Johnson, Science 299 (2003) 1367. [47] X. Li, V.E. Teige, S.S. Iyengar, J. Phys. Chem. A 111 (2007) 4815. [48] M.C. Goh, J.M. Hicks, K. Kemnitz, G.R. Pinto, K. Bhattacharyya, K.B. Eisenthal, J. Phys. Chem. 92 (1988) 5074. [49] Y. Levin, Phys. Rev. Lett. 102 (2009) 147803. [50] C.D. Wick, I-F.W. Kuo, C.J. Mundy, L.X. Dang, J. Chem. Theory Comp. 3 (2007) 2002. [51] L.M. Pegram, M.T. Record, Chem. Phys. Lett. 467 (2008) 1. [52] A.D. Becke, K.E. Edgecombe, J. Chem. Phys. 92 (1990) 5397.

Chris Mundy received a B.S. in Chemistry from Montana State University in 1988, and a Ph.D. in Physical Chemistry from the University of California, Berkeley in 1992. From 1993 to 1998 he was a post-doctoral fellow at the University of Pennsylvania under the direction of Prof. M.L. Klein. Thereafter, he was a post-doc in the group of Prof. Michele Parrinello at the Max-PlanckInstitute for Solid State Research in Stuttgart, Germany until 2000. From 2001 to 2006 he was a Staff Scientist and Group Leader in Computational Chemistry and Chemical Biology at Lawrence Livermore National Laboratory. Since 2006, he is a Chief Scientist in the Fundamental Science Directorate a Pacic Northwest National Laboratory. His research interests are in computational statistical mechanics using density functional theory (DFT) based interaction potentials.

I-Feng William Kuo obtained his B.S. in Chemistry from the University of California, San Diego in 1997, and went on to obtain his Ph.D. in Chemistry in 2002 at the University of California, Irvine. From 2002 to 2005 he was a post-doctoral fellow at the Lawrence Livermore National Laboratory. Currently, he is a staff scientist at Lawrence Livermore National Laboratory in the Chemical Sciences Division.

C.J. Mundy et al. / Chemical Physics Letters 481 (2009) 28 proton transport in condensed phases, catalytic effect of metal-doped nanotube, and antimicrobial activity of small amphipathic peptides, along with the development of ab initio molecular dynamics technique using real-space basis sets.

Mark Tuckerman obtained his B.S. in physics from U.C. Berkeley in 1986 and his Ph.D. in Physics from Columbia University in 1993. From 1993 to 1994, he held an IBM postdoctoral fellowship at the IBM Forschungslaboratorium in Rschlikon, Switzerland, and from 1995 to 1996, he was an NSF postdoctoral fellowship in Advanced Scientic Computing at the University of Pennsylvania in Philadelphia. He is currently Professor of Chemistry and Mathematics at New York University. In 2005, he received the Friedrich Wilhelm Bessel Research Award from the Alexander von Humboldt Foundation, which he spent in 2006 and 2007 at the Ruhr-Universitt in Bochum, Germany. His research interests include reactions in solution, organic reactions on semiconductor surfaces, and development of the methodology of molecular dynamics, including novel techniques for enhancing conformational sampling and prediction of free energies in biological systems, and the development of new approaches to electronic structure and ab initio molecular dynamics calculations.

Hee-Seung Lee received his B.S. degree in Chemistry from Seoul National University in 1992, and his Ph.D. degree in Chemistry from the Ohio State University in 2001. During graduate studies, he applied diffusion Monte Carlo method and variational approach to study small open-shell radical complexes with rare gas atoms. He spent the next two years at the University of Chicago as a postdoc and continued his work on developing variational techniques to compute highly excited rovibrational states of polyatomic molecules. In 2004, he moved to NYU and expanded his research interest to condensed phases, such as liquid water, aqueous interfaces, and solid acids. Since 2007, he is an assistant professor at University of North Carolina at Wilmington. His current theoretical/computational research focuses on

Doug Tobias obtained his B.S. and M.S. in Chemistry at the University of California, Riverside, in 1984 and 1985, respectively, and his Ph.D. in Chemistry and Biophysics at Carnegie Mellon University in 1991. He did postdoctoral research in the Department of Chemistry at the University of Pennsylvania from 1991 to 1995, and was a guest researcher at the NIST Center for Neutron Research at the National Institute of Standards and Technology from 1995 to 1997. He was appointed to the faculty in the Department of Chemistry at UC Irvine in 1997, promoted to Associate Professor in 2003 and to Full Professor in 2005. He was elected Fellow of the American Association of the Advancement of Science in 2006. His research interests include the development and application of molecular simulation techniques to the study of membrane biophysics, heterogeneous atmospheric chemistry, and the chemical dynamics of aqueous interfaces.

Das könnte Ihnen auch gefallen