© Copyright 1996 American Chemical Society
AUGUST 7, 1996 VOLUME 12, NUMBER 16
Letters Molecular Dynamics Study of Water next to Electrified Ag(111) Surfaces Karl J. Schweighofer, Xinfu Xia, and Max L. Berkowitz* Department of Chemistry, University of North Carolina, Chapel Hill 27599 Received November 21, 1995. In Final Form: May 31, 1996X The aim of this report is to compare the density profiles, orientation, and structure of water near charged and uncharged Ag(111) interfaces obtained using molecular dynamics calculations with those suggested from a recent experiment. In the simulations, water contained between two parallel Ag(111) surfaces with charge densities of 0.0, 8.85, and 26.55 µC/cm2 on the left wall and opposite values on the right wall experiences external electric fields of 0.0, 1.0, and 3.0 V/Å. The predicted orientation and density of water for the zero field case is consistent with previous predictions for water near metal surfaces and indicates that water is present in layers and has an orientational structure similar to that of hexagonal ice. For simulations with charged surfaces, there is a reordering of water throughout the simulation cell in response to the polarization of the water dipoles. We show that the spacing between the peaks in the liquid density profiles obtained from our MD calculations matches the results of the experiment but that, contrary to the experimental results, we have not observed a dramatic increase of water density near the charged metal.
Introduction The study of the molecular properties of water near metal surfaces is one of the most interesting and relevant subjects in the fields of electrochemistry, organic chemistry, and materials science. The microscopic environment at the water monolayer/metal interface has been examined using a variety of experimental techniques such as surface infrared spectroscopy,1,2 X-ray scattering,3 LEED,4 photoelectron spectroscopy,5 and other methods.6,7 However, it has been difficult to obtain accurate molecular scale * Author to whom correspondence should be directed. E-mail address:
[email protected]. X Abstract published in Advance ACS Abstracts, July 1, 1996. (1) In Spectroscopic and Diffraction Techniques in Interfacial Electrochemistry; Gutierrez, C., Melendres, C., Eds.; Kluwer: Dordrecht, 1990. (2) The Advancing Frontier in the Knowledge of the Structure of the Interphases; Bockris, J. M., Gonzalez-Martin, A., Eds.; Kluwer: Dordrecht, 1990. (3) Toney, M.; Howard, J.; Richer, J.; Borges, G.; Gordon, J.; Melroy, O.; Weisler, D.; Yee, D.; Sorensen, L. Nature 1994, 368, 444-446. (4) Doering, D. L.; Madey, T. E. Surf. Sci. 1982, 123, 305. (5) Kiskinova, M.; Pirug, G.; Bonzel, H. P. Surf. Sci. 1985, 150, 319.
S0743-7463(95)01061-4 CCC: $12.00
information from experiments on buried water/metal interfaces. A particularly interesting quantity is the distance dependent density profile for water next to metal surfaces. Recently, Toney et al. obtained such a profile from an X-ray scattering experiment performed on water next to a charged Ag(111) surface.3,8 The reported surface charge densities in the experiment were 25.0 and -10.0 µC/cm2. The profiles inferred from the experiment display layers of water extending at least three molecular diameters from the metal surface. Also, the profile of water density adjacent to the positively charged plate indicates a very large areal density for the layer of water immediately next to the surface, implying that the hydrogen bonding in water is disrupted and the properties of water in the layers next to the metal are very different from those in the bulk. The results of Toney et al.3 are partly in agreement with, and partly contradictory to, (6) Structure of Electrified Interfaces; Lipkowski, P. N. R., Ed.; VCH: New York, 1993. (7) Thiel, P. A.; Madey, T. E. Surf. Sci. Rep. 1987, 7, 211. (8) Toney, M.; Howard, J.; Richer, J.; Borges, G.; Gordon, J.; Melroy, O.; Weisler, D.; Yee, D.; Sorensen, L. Surf. Sci. 1995, 335, 326-332.
© 1996 American Chemical Society
3748 Langmuir, Vol. 12, No. 16, 1996
recent computer simulations on water/metal interfaces.9-11 One important result on which there is agreement is that water molecules form layers next to the metal surface, and this layering decreases as one moves toward the bulk region. For example, from recent simulations on the uncharged Pt/water interface,10 one can distinguish two to three such layers. For the same system it was also observed that the first layer of water situated on the surface is nearly immobile due to a weak chemisorption. In contrast to the conclusions drawn by Toney et al. on the water/Ag(111) surface, the simulation results from the charged water/Pt(100) interface did not show a very large enhancement of water density for the layer adjacent to the metal.9 Since the experimental and simulation results differ significantly with respect to the magnitude of the density fluctuations, we performed a molecular dynamics study of water next to a surface which is closer in its geometry to Ag(111). This may allow us to perform a better comparison between the results from recent experiments of Toney et al. and simulation. A word of explanation on the presence of ions is appropriate, since the experiment was performed in a 0.1 M solution of NaF and the simulations contained no ions. According to Toney et al. the ions do not significantly adsorb on the surface; thus, the concentration of the ions at the interface is assumed to be small. In addition, because of the small system size in the simulation, a comparable concentration would result in only a single ion pair in the simulation cell. Also due to the approximate nature of the potentials, we cannot make the claim that our simulation reproduces the exact conditions of the experiment, yet this is an issue whenever simulations are used to gain insight about molecular phenomena. Methods The simulations consisted of 512 SPC/E12 water molecules confined between two parallel Ag(111) surfaces that lie in the x-y plane. The box dimensions (Lx ) 23.12 Å, Ly ) 20.02 Å) were chosen to be commensurate with the Ag(111) lattice, where the length in the z direction (Lz ) 35.20 Å) was chosen to reproduce the bulk density for water in the center of the box when the Ag surfaces are uncharged. Three individual simulations were performed with the Ag surface on the left charged to 0.0, 8.85, and 26.55 µC/cm2, respectively, whereas the surface on the right carried an equal but opposite charge. These correspond to external electric field strengths of 0.0, 1.0, and 3.0 V/Å. We do not know the exact value of the dielectric constant for the SPC/E water at high field, but estimates show that it is close to 10 when the external field is 3.0 V/Å.13 That means that the total field is ∼0.3 V/Å, of the same order of magnitude as in the experiment. For the simulations with the high field, the size of the box in the z direction was increased by ∼10% in order to keep the density of water in the middle of the box at 1.0 g/cc. To describe the interaction between the Ag and water, we use a crude assumption that the variation in the energetics of adsorbtion on transition metals such as Pt, Ni, Cu, and Ag is only very minor.14 Therefore, to describe the water/Ag interaction, we use the same potential as in our previous studies of water/Pt; however, the lattice constant for the metal was changed to 4.09 Å, which is appropriate for Ag(111). The details of the potential parameters and form have been described previously.10 At this point it is proper to remind the reader that the potential used in ref 10 to describe the Pt/water interaction does not contain the image charge term, (9) Xia, X.; Berkowitz, M. Phys. Rev. Lett. 1995, 74, 3193-3196. (10) Raghavan, K.; Foster, K.; Motakabbir, K.; Berkowitz, M. J. Chem. Phys. 1990, 94, 2110-2117. (11) Nagy, G.; Heinzinger, K. J. Electroanal. Chem. 1990, 296, 549. (12) Berendsen, H. J. C.; Grigera, J. R.; Straatsma, T. P. J. Phys. Chem. 1987, 91, 6269. (13) Xia, X.; Berkowitz, M. In preparation. (14) Siepmann, J. I.; Sprik, M. J. Chem. Phys. 1995, 102, 511-517.
Letters
Figure 1. Density profiles F(z) for different electric field strengths: solid lines, oxygen profiles; dotted lines, hydrogen profiles. which is most often associated with the induction effect in the metal. This image term is also absent in the Spohr-Heinzinger potential,15 on which our potential is based. The image charge description is an oversimplified representation of the induction effect which becomes more accurate when the charge is far away from the surface. For water molecules directly adsorbed on the surface, the metal induction is partly taken into account by the quantum chemical calculations performed on water/metal clusters which form the basis for the Spohr-Heinzinger model and our own. The image charge interaction can be included to describe the metal/water interaction for water beyond the adsorbtion layer of water. Since the image force decays rather quickly, this correction is expected to be small compared to the water/water interaction.16 Long range forces were taken into account using the prescription of Aloisi et al.17 Molecular dynamics trajectories were performed using the Verlet algorithm and the SHAKE routine with a 2.5 fs time step for 100 ps. Equilibration times varied for the different electric fields, the highest field requiring 300 ps and all others requiring only 100 ps. A constant temperature of 300 K was maintained by periodically rescaling the velocities.
Results The time-averaged density profiles for each value of the electric field are shown in Figure 1. These profiles are calculated by dividing the simulation box into 0.1 Å slabs parallel to the x-y plane and counting the number of water molecules in each slab. A water molecule is counted if its oxygen is inside the slab. The positions of the oxygens and hydrogens are binned in different distributions, and these number densities are multiplied (15) Spohr, E.; Heinzinger, K. Ber. Bunsen-Ges. Phys. Chem. 1988, 92, 1358. (16) Foster, K. Unpublished communication. (17) Aloisi, G.; Foresti, M. L.; Guidelli, R.; Barnes, P. J. Chem. Phys. 1989, 91, 5592.
Letters
Figure 2. Orientational probability distribution functions for Ag(111) with zero field: solid lines, distributions for angle θ (between water dipole and -z); dotted lines, distributions for angle R (between O-H vector and -z).
by the appropriate conversion factor to yield the water density profile based on either hydrogen or oxygen in units of grams per cubic centimeter. The oxygen density profile for the zero field case shows two distinct water layers propagating from the Ag surfaces followed by a third, less distinct layer. The transition to the bulk density occurs within 10 Å from the surface. The layering of water near metal surfaces and hydrophobic walls is well-known.10,18-20 Both the oxygen and hydrogen density profiles are symmetric with respect to the center of the box, as expected, and the main peaks in the hydrogen profile roughly coincide with the oxygen peaks but are slightly shifted. In addition there are two small peaks in the hydrogen density profiles that are found between the main peaks near each surface. As shown in a previous paper on the water/Pt(111) interface,10 the first of these small peaks (closest to the surface) is part of the water in the first layer whose dipole vectors point slightly away from the surface, whereas the second small peak represents hydrogens which are part of water molecules in the second layer pointing slightly back toward the Ag(111) surface. The orientational distributions in Figure 2 clearly show the water orientation in these layers and will be discussed. In comparing the E ) 0.0 V/Å and E ) 1.0 V/Å density profiles, one may see that there is little change in the oxygen density profile, indicating that the water molecules have not undergone a significant rearrangement with respect to their positions along the z axis. The hydrogen density profile however has become asymmetric. On the right, the two small peaks discussed above have become one whereas, on the left, these peaks have almost completely disappeared and become part of the larger peaks. In addition the peak height has increased on the left and decreased on the right. Although the shape of the hydrogen density profile has undergone a significant change upon application of the field, the positions of the (18) Derjaguin, B. V.; Churaev, N. V. In Fluid Interfacial Phenomena; Croxton, C. A., Ed.; Wiley: New York, 1986; p 523. (19) Lee, C. Y.; McCammon, J. A.; Rossky, P. J. J. Chem. Phys. 1984, 80, 4448. (20) Spohr, E. J. Phys. Chem. 1989, 93, 6171.
Langmuir, Vol. 12, No. 16, 1996 3749
major peaks have not shifted. This is indicative of the reorientation of the water molecules in response to the field but implies that a large scale restructuring of the liquid has not occurred. As the field is increased to 3.0 V/Å, the layering of the water is apparent. The biasing to the right of the hydrogen density profile relative to the profile for oxygen demonstrates that the water is significantly polarized, with the dipoles pointing from the positively charged plate on the left to the negatively charged one on the right. There is a significant redistribution of oxygen density in response to the electric field as a result of the orientational polarization of the water. To better quantitate this and to help us compare with the recent results of Toney et al.,3 we have calculated the number of water molecules in the first water layer next to the positively charged plate. These layers correspond to the leftmost oxygen peaks in Figure 1. The data are shown in Table 1. In Table 1 the following quantities are tabulated: the number of water molecules (Nw) in the first peak of the oxygen density profile of Figure 1, the number of water molecules expected if the region contained bulk water (Nb), the length of the region containing the water molecules in angstroms, and the ratio Nw/Nb. These are shown for the three electric field strengths. A noticeable feature is that the number of water molecules in the first peak decreases by 34% upon increasing the field from 0.0 to 3.0 V/Å. The change is much less for the 1.0 V/Å case. This shows that the water is restructuring under the high field. This can be seen by the increased number of peaks in the density profile in Figure 1. The density of water in these regions is also of interest. From the results of Toney et al.3 one predicts a density of approximately 2.25 g cm-3 for the 3.0 V/Å case. In contrast, we observed the highest density was for the zero field case (∼1.55 g cm-3). The increase in density for these layers is due to chemisorption of water onto the metal. It is also of interest to compare the areal densities from the simulation to those of experiment. While the experimental results indicate that the areal water density next to a positively charged plate carrying a surface charge of 25 µC/cm2 is very large and has the value 2.6 × 1015 cm-2, our simulations predict a density of 0.78 × 1015 cm-2. In order to have a clear picture of the water structure however, we must examine the orientational distributions for the water dipole and the O-H vector. The orientation of the water in nine consecutive layers of thickness 1.0 Å extending from the metal surface was determined by calculating orientational distribution functions. Distributions were calculated for the angles θ and R, which are defined by the angle between the water dipole µ and the negative z axis (perpendicular to the interface) and by the angle between the O-H bond vector rOH and the negative z axis, respectively. These distributions are shown in Figure 2 and are for the zero field case only. The centers of the slabs are labeled in each plot. The solid lines correspond to the θ distributions, and the dotted lines, to R. The upper left plot shows that the dipoles and O-H bonds lie nearly parallel to the metal surface for water immediately next to the metal (this corresponds to water occupying the leftmost half of the leftmost peak in the oxygen density profile of Figure 1 (top plot)). We shall call this sublayer of water L1a. The second sublayer (L1b) consists of water that is next to and hydrogen bonded with water in L1a, and it occupies the rightmost half of the leftmost peak in the oxygen density profile. The orientation of L1b water is bimodal with respect to the dipole orientation, demonstrating that there are two preferred orientations, one pointing slightly toward the metal and one pointing slightly away from the metal. The
3750 Langmuir, Vol. 12, No. 16, 1996
Letters
Table 1. Statistics for Water in the First Layer next to the Positive Plate Nw Nb region, Å Nw/Nb
E ) 0.0 V/Å
E ) 1.0 V/Å
E ) 3.0 V/Å
55.1 35.6 -16.00 e z e -13.70 1.55
54.9 37.1 -16.10 e z e -13.70 1.48
36.2 34.0 -18.08 e z e -15.98 1.06
orientational distribution of water in layer L1 next to Ag(111) is very similar to that observed for the layer of water next to Pt(111).10 This means that the adsorbed layer of water has an orientational structure characteristic of hexagonal ice. The next two layers of water (L2 and L3) display orientational structure somewhat similar to that observed for the water/Pt(111) system, although the mirror image symmetry present in the latter system (for layers L2b and L2c) is absent in the Ag(111) system. This means that, next to the Ag(111) surface, the orientational ordering characteristic of hexagonal ice does not propagate as far as three layers of water. The water orientation near the metal surface is a direct result of the interplay between forces arising from hydrogen bonding, O-Ag adsorbtion on top sites, H-Ag repulsion, and alignment of the dipoles with the electric field. When no field is present, O-Ag adsorbtion and H-Ag repulsion favor a planar orientation of water near both plates with the hydrogens pointing slightly away. This also allows the hydrogens to participate in hydrogen bonding with other water molecules in the first layer. The attempt to maximize the hydrogen bonding between adsorbed waters and other water molecules in the first layer results in the reversal of the water dipoles from pointing slightly away from the metal to pointing toward the metal. This small adjustment allows a hydrogen bond network to be maintained throughout the liquid, and near the metal this conspires with O-Ag adsorbtion and H-Ag repulsion, causing hexagonal ice-like domains. In the center of the box (-5.0 Å < z < 5.0 Å), the dipole orientation shows a tendency to remain parallel to the x-y plane and is typified by the distribution shown in the last plot of Figure 2. This effect is still under investigation, but it is thought to arise from the effects of the long range forces, which are absent in our previous calculations of water/ metal interfaces. Adding an electric field disturbs the balance of forces such that, at the positive plate, O-Ag adsorbtion and H-Ag repulsion are aided by the field. The opposite effect occurs at the negative plate. The 1.0 V/Å field is not enough to significantly affect the spatial water structure. The flipping of water molecules in response to the 3.0 V/Å field, however, causes some of the water near the positive surface to be desorbed in order to preserve the hydrogenbonding network. Some of these water molecules become part of the second water layer, and this restructuring of water propagates from the positive surface toward the bulk. Near the center of the box, water is still strongly oriented but without the ordering effect brought on by O-Ag adsorbtion. Desorbtion near the negatively charged plate is more dramatic, since, in addition to the polarizing effect of the field, oxygen is repelled by the negatively charged plate and hydrogen is attracted. We examine more closely the water structure as a function of distance from the surface by looking at the oxygen-oxygen pair correlation functions (Figure 3). These are calculated in each of the 1.0 Å slabs by counting the number of oxygens inside a spherical shell of radius r centered on a particular oxygen atom and dividing by the number expected from a homogeneous liquid with the bulk density. This compares the number of neighbors around an interfacial water oxygen to that expected from a homogeneous system with the bulk density. For a given
Figure 3. O-O pair correlation functions for water in 1.0 Å slabs which are centered at the positions indicated in the legends. The plots to the right correspond to water near the negatively charged plate, whereas those on the left correspond to water near the positively charged plate.
pair of atoms, both must be inside the slab to contribute. Therefore the pair correlation function depends not only on r but also on the positions zi and zj of the two oxygens. The pair correlation functions g(rOO), g(rOH), and g(rHH) for the zero electric field case inside the 10 Å slab defined by -5.0 < z < 5.0 Å yielded results identical to that of an independent calculation of g(r) on bulk SPC/E water (data not shown). The top two plots in Figure 3 correspond to the g(r) at the left and right plate, respectively, for the zero field case. The fact that these plots are identical again illustrates the symmetry first observed in the density profile. The four curves within a plot correspond to g(r) for the four 1 Å slabs closest to the (111) surface. There is a striking difference between the first layer and subsequent layers. The second, third, fourth, and fifth peaks of the O-O g(r) for the first layer correspond very closely with the respective peaks of the Ag-Ag g(r) for sites on the (111) surface (not shown). The position of the first peak of the O-O g(r) (2.78 Å), however, is still close to the 2.77 Å of bulk water but not near the 2.89 Å between adjacent Ag sites on the 111 surface. Thus near the metal surface, there is a competition between hydrogen bonding (favoring the smaller O-O spacing) and adsorption of the oxygens on the top sites of the metal (inducing long range order on the water). Periodicity in the O-O correlation has been shown for water monolayers on metal
Letters
Langmuir, Vol. 12, No. 16, 1996 3751
Figure 5. Comparison of the oxygen density profiles from the experimental data of Toney et al. versus our MD calculations. The left panel corresponds to the E ) 3.0 V/Å case and shows the distribution at the positively charged plate. The right panel corresponds to the E ) 1.0 V/Å case and shows the distribution at the negatively charged plate. Since the position of the wall is uncertain, the MD data were linearly translated along the z axis to best match the positions of the experimental peaks: solid lines, MD data; dotted lines, experimental data. Figure 4. (A-C) Stroboscopic projections of the oxygens (dark dots) onto the Ag(111) lattice (circles) for water in the first layer next to the positive plate. Frames A, B, and C refer to E ) 0.0, 1.0, and 3.0 V/Å, respectively. Frame D shows secondlayer water for the 3.0 V/Å case for comparison with frame C.
surfaces.14 These results indicated that the first peak in the O-O g(r) matched closely the Ag-Ag spacing of the metal. As the electric field is turned up, changes in g(r) are a sensitive indicator of structural changes in the various water layers. Most notably, the peaks of the adsorbed oxygens on the positively charged plate become sharper, and the minima become flatter, much like the g(r) for a crystal lattice. For water next to the negative plate, however, this structure begins to disappear as the field is increased, indicating that at the negatively charged plate water is desorbed from the surface and loses its periodicity. The water next to the positive plate retains a strongly periodic structure as the field is increased, indicative of the formation of icelike water layers. The restructuring of water at the positive plate in response to the 3.0 V/Å field is clearly shown in Figure 4, which is a stroboscopic picture of water in layers next to the (111) surface. These were made by marking the position of water oxygens in the layers for 40 consecutive time steps separated by time periods of 25 fs. The large circles in Figure 4 are the positions of the Ag(111) sites. Panels A-C show L1 water for the 0.0, 1.0, and 3.0 V/Å fields, respectively. Panel D shows water in layer L2 for the 3.0 V/Å field and demonstrates that the water in this layer shows substantially more translational motion than water in direct contact with the surface (panel C). In comparing panels A and B, almost all of the top sites are occupied and some patches of water are arranged in a structure commensurate with hexagonal ice. Figure 4B shows that the 1.0 V/Å field is not enough to disrupt the structure near the metal (a fact that is borne out of the experimental results as well and can be seen by examining the density profiles in Figure 5, panel 2). In panel C the 3.0 V/Å electric field resulting from the positive surface charge on the plate has released some oxygens from their adsorbtion sites. The density of these layers of water does not show the dramatic increase that has been proposed;3 rather, the density decreases in relation to the zero field case for water in the first layer. This is a graphic example of the data presented in Table 1. This behavior is the result of water polarization accompanied by reorganization in an attempt to maintain a hydrogen-bonded network. These forces dominate over the electrostatic interactions
between negatively charged oxygens and the positively charged plate. When the electric field was increased to 4.0 V/Å, we observed a very strong layering of water propagating from the surface of Ag(111) and creation of cubic ice-like domains in the middle of the water lamella (not shown). This behavior is similar to that observed for water between Pt(100) surfaces.9 Figure 5 shows a comparison of the density profiles from our simulations with the results from Toney et al.3 and is the major focus of this letter. The left panel shows the profiles for the 3.0 V/Å field, and although the layering observed by Toney et al.3 is comparable with our findings, the integration of the peaks interpreted from the X-ray data yields a density of approximately 2.25 g/cc for the first layer water. Such a large deviation from the bulk density has never been reported nor predicted from simulations. The right panel shows the comparative results for the 1.0 V/Å case for water next to the negatively charged wall. The agreement between the data is much better for this case; however, the experiments showed a very well defined third layer, in contrast to our results. It is important to be reminded here that the experiments were performed in a 0.1 M NaF solution. Although we perform our simulations without ions, it was argued by Toney et al. that the ions do not significantly adsorb to the surface, and thus their concentration at the interfaces is small. The effects of ions on the water/(charged metal) interface will be a topic for further study. Discussion The results presented in this letter on water layering and structure near the Ag(111) surface are consistent with previous theoretical studies by other groups on the microscopic environment near a water/metal interface.14,20-22 We have shown that the spacing between water layers observed in the density profiles agrees well with the results of Toney et al.3 Our results describing the orientation of water near the metal surface are also in agreement with these experiments. There is a strong discrepancy, however, in the magnitude of the density fluctuations near the positively charged surface when the field is large. Whereas Toney et al.3 have indicated a greater than twofold increase in the density of water near the highly charged positive surface, we have reported a 1.1-fold increase relative to bulk water. Our observations have also shown that the water density in the first layer next to the positively charged surface (21) Rose, D. A.; Benjamin, I. J. Chem. Phys. 1991, 95, 6856. (22) Glosli, J. N.; Philpott, M. R. J. Chem. Phys. 1992, 96, 6962.
3752 Langmuir, Vol. 12, No. 16, 1996
decreases with the application of a 3.0 V/Å electric field rather than increases. This is because the field strongly polarizes the water molecules, causing some of the oxygens to desorb from the positive Ag sites and move into the subsequent water layer in an attempt to keep intact some of its hydrogen bond network. Orientational ordering due to the field and oxygen ordering due to the positive surface conspire to produce an icelike structure which propagates from the positive Ag surface toward the bulk. In addition, we believe that hydrogen bonding is the dominant factor in determining water structure even under very harsh conditions, as previous computational results on the effects of an electric field on bulk water,23 on a liquid/liquid interface,24 and at the solid/liquid interface9 have indicated. Recent experimental findings also indicate that hydrogen bonding plays an important role in the response of water to a change in surface charge.25 Finally, we would like to make some remarks about the comparison between the experiment and the simulations presented in Figure 5. The simulation is not without shortcomings; namely, the Ag/water potential is somewhat ad-hoc, the induction of the metal is not treated consistently (since, in the Spohr-Heinzinger model, effective induction is built into the potential at short ranges), electronic polarization of the water is not included, and the long range forces are treated in an average way. The interpretation of the experimental data may also present significant challenges. The assumption that electrochemical reactions are absent and that ions do not contribute to the scattering data may be tenuous. If we accept that these possibilities are inconsequential, then, on the basis of the results of our simulations and the (23) Svishchev, I. M.; Kusalik, P. G. Phys. Rev. Lett. 1994, 73, 975. (24) Schweighofer, K. J.; Benjamin, I. J. Electroanal. Chem. 1995, 391, 1-10. (25) Quan Du, E. F.; Shen, Y. R. Phys. Rev. Lett. 1994, 72, 238-241.
Letters
previous success of simulations in interpreting the structure of bulk aqueous systems, we are confident that the electrostatic effect alone cannot produce the hydrogen bond disruption described by Toney et al.3 In a later analysis of the experiment8 Toney et al. used a structure of highpressure ice (ice-VIII) as a conceptual guide to speculate on a possible reason for the high areal densities they observed. In this case, a water hydrogen-bonded network and large areal densities are not necessarily mutually exclusive. This is because high-pressure ices have interpenetrating hydrogen-bonded networks, implying that the first layer is actually a bilayer. We do not observe the signature of high-pressure ice in our simulations for the field strength of 3.0 V/Å. When we increased the field to 4.0 V/Å, however, there was a restructuring of water into a cubic ice-like formation, similar to what we observed in our Pt/water simulations.9 An important conclusion, which may be drawn from the results of this simulation and the experimental results of Toney et al., is that water undergoes a restructuring in the presence of a strong electric field. Continued experimental efforts will be certain to improve our understanding of the structure of water near metal surfaces and our ability to generate more realistic potential models for these systems. Acknowledgment. This work was supported by a grant from the Office of Naval Research. We gratefully acknowledge the assistance of our colleagues: Dr. Uli Essmann, Dr. Lalith Perera, and Laura Sremaniak. Some of the calculations were performed at the North Carolina Supercomputer Center and on the Hewlett-Packard Cluster at the University of North Carolina. We also express our gratitude to the reviewers for their careful and thoughtful criticism. LA951061R
