Liquid Water Confined in Carbon Nanochannels at High Temperatures

Oct 10, 2007 - Departamento de Sistemas Fı´sicos, Quı´micos y Naturales, Facultad de Ciencias Experimentales, UniVersidad. Pablo de OlaVide, Carre...
0 downloads 0 Views 178KB Size
12524

J. Phys. Chem. B 2007, 111, 12524-12530

Liquid Water Confined in Carbon Nanochannels at High Temperatures G. Nagy,† M. C. Gordillo,‡ E. Gua` rdia,§ and J. Martı´*,§ Materials Department, KFKI-Atomic Energy Research Institute, H-1525 Budapest, P.O.B. 49, Hungary, Departamento de Sistemas Fı´sicos, Quı´micos y Naturales, Facultad de Ciencias Experimentales, UniVersidad Pablo de OlaVide, Carretera de Utrera, km 1, 41013 SeVilla, Spain, and Departament de Fı´sica i Enginyeria Nuclear, UniVersitat Polite` cnica de Catalunya, B4-B5 Campus Nord, 08034 Barcelona, Catalonia, Spain ReceiVed: April 25, 2007; In Final Form: July 18, 2007

Structure, hydrogen bonding, electrostatics, dielectric, and dynamical properties of liquid water confined in flat graphene nanochannels are investigated by molecular dynamics simulations. A wide range of temperatures (between 20 and 360 °C) have been considered. Molecular structure suffers substantial changes when the system is heated, with a significant loss of structure and hydrogen bonding. In such case, the interface between adsorbed and bulk-like water has a marked tendency to disappear, and the two preferential orientations of water nearby the graphite layers at room temperature are essentially merging above the boiling point. The general trend for the static dielectric constant is its reduction at high temperature states, as compared to ambient conditions. Similarly, residence times of water molecules in adsorbed and bulk-like regions are significantly influenced by temperature, as well. Finally, we observed relevant changes in water diffusion and spectroscopy along the range of temperatures analyzed.

Introduction Our understanding of the electrochemical properties of the graphite/water interface at room temperature is well established, even at the atomic level.1-4 The structure and dynamics of water close to hydrophilic and hydrophobic surfaces has been sufficiently well described in a significant number of works,5-10 and the effect of surface defects has also been addressed as well.11 A review of water in slab geometries was recently reported by Zangi.12 Much less is known, however, about the properties of the solid/liquid interfaces of water at higher temperatures, especially above its boiling point. Although a large number of very important industrial processes involving surfaces work under extreme conditions, the tools to understand their details, especially on the nanoscale, are very limited. From the experimental point of view, electrochemical methods, with or without surface analytical tools, are being developed rapidly, but the theoretical approach has been limited to a phenomenological description of the graphite/water interface contribution to certain processes. With this work, we make a first step toward a detailed description of the solid/liquid interfaces at high temperatures. To accomplish this aim, we fall back on our experience in simulation of water under extreme conditions13 and of interfaces between both hydrophilic14 and hydrophobic surfaces15-17 and water. Carbon surfaces are of great interest because (i) they are used in everyday life devices, such as electrodes in batteries or in other electrolytic processes, (ii) nanotubes and graphene sheets have unique properties, and (iii) the C(0001) plane serves as a model for hydrophobic surfaces. Thus, we have chosen the interface between the C(0001) surface and water to study the effect of temperature on the interfacial properties of water at the atomic level. As our previous experience shows, molecular * Corresponding author. E-mail: [email protected]. † KFKI-Atomic Energy Research Institute. ‡ Universidad Pablo de Olavide. § Universitat Polite ` cnica de Catalunya.

dynamics (MD) simulation is a suitable tool to investigate the nanoscale properties of water nearby solid surfaces, such as platinum, graphite, and carbon nanotubes. On the basis of these results, we performed a series of MD simulations of water between two graphene planes as a function of temperature in the range 20-360 °C. The next section contains the model and simulation details, then the results are given and discussed, and, finally, conclusions are drawn. Model and Simulation Details We confined liquid water between two parallel graphene planes 3.1 nm apart from each other. A simulation box of ∆x ) 3.44 nm, ∆y ) 3.41 nm, and ∆z ) 3.1 nm was set up, where the z direction is taken to be perpendicular to both surfaces. We performed a series of molecular dynamics simulations in the NVT ensemble in the temperature range from 20 to 360 °C, with a homogeneous temperature variation of 30-40 K between consecutive states, to analyze the evolution of structure, electric properties, and dynamics of water between ambient and supercritical high-density conditions. Water density in the middle of the simulation box was adjusted to be close to the liquid density of the equilibrium liquid/gas system at the corresponding temperature.18 Some characteristic parameters are given in Table 1. A relevant point was the slight reduction (5%) of the atomic effective charges at the three highest temperatures, to provide stability to the system by approaching the molecular dipole to the gas-phase value. Strictly speaking, we changed the model because the flexible SPC potential employed in this work (originally parametrized for ambient conditions) becomes unstable at high temperatures, as it was observed in a previous study at supercritical conditions.19 The water-carbon potential was assumed to be a 12-6 Lennard-Jones type, with the same parametrization employed in previous works (see, e.g., ref 16). An alternative dependence (9-3 Lennard-Jones potential) was used by Brovchenko et al. to study the liquid-vapor coexistence curve in the case of water

10.1021/jp073193m CCC: $37.00 © 2007 American Chemical Society Published on Web 10/10/2007

Liquid Water Confined in Carbon Nanochannels

J. Phys. Chem. B, Vol. 111, No. 43, 2007 12525

TABLE 1: Characteristics of the MD Simulationsa T/°C

T/K

Nwater

25 55 90 127 160 200 250 280 322 363

298 328 363 400 433 473 523 553 595 633

1089 1040 1000 975 950 900 850 770 680 550

Ftheo/g cm 1.000 0.985 0.962 0.935 0.910 0.863 0.800 0.750 0.666 0.526

-3

Fbulk/g cm

-3

1.028 ( 0.051 0.991 ( 0.027 0.958 ( 0.030 0.944 ( 0.026 0.923 ( 0.024 0.882 ( 0.019 0.831 ( 0.022 0.755 ( 0.013 0.669 ( 0.015 0.534 ( 0.015

qox/e -0.82 -0.82 -0.82 -0.82 -0.82 -0.82 -0.82 -0.78 -0.78 -0.78

a Nwater, number of water molecules in the simulation box; Ftheo, theoretical bulk density; Fbulk, density in the central region of the simulation box; qox, partial charge of oxygen in terms of the electron charge (e).

in slit-like and cylindrical nanopores.20 These authors consider that the 12-6 form is not quite appropriate for water-surface interactions near the critical point of water. However, our simulations have been performed safely far enough from the critical point of our model, to minimize shortcomings of the potential model employed. The water-water inter- and intramolecular interactions have been modeled by means of a flexible SPC potential, which was specifically reparametrized to correctly reproduce the main features of the infrared spectrum of water at ambient conditions.21 The critical point of our model (Tc ) 370 °C, Fc ) 0.32 g/cm3)22 is located very close to the experimental value (Tc ) 374 °C, Fc ) 0.322 g/cm3).22 Long-ranged electrostatic interactions have been included by a 3D Ewald sum procedure introduced by Spohr23 by artificially enlarging the z-coordinate of the simulation box to 5 times the x-coordinate. A leapfrog Verlet integration algorithm with coupling to a thermal bath has also been employed.24 The integration time-step was 0.5 fs. Stabilization runs of at least 30 ps in all cases were followed by production runs of lengths between 50 and 100 ps. Results and Discussion Structure, Hydrogen Bonding, and Orientational Order. The effect of temperature on liquid water confined in a graphite slab manifests itself in changes of the structure of interfacial water. To follow this, we computed the changes in the density of water as a function of distance from the C(0001) surface. In Figure 1a and b, the density profiles of oxygen and hydrogen are plotted separately for a series of temperatures in the distance range 0-1 nm. As the peaks on the profiles indicate, at room temperature the water interface consists of two layers: the adsorbed water at the wall in the range 0.2-0.45 nm and a second, intermediate layer between 0.45 and 0.7 nm. Farther from the surface, the water is bulk-like. As the temperature increases, continuous changes may be observed. The peaks related to the interfacial structure become smaller, and that of the intermediate water layer gradually disappears. At 633 K, water inside a nanometric graphite slab consists of a single layer of adsorbed water and, farther away from the surface, a wide bulk-like region. Good overall agreement of oxygen density profiles is found between the present work and Brovchenko et al.20 for the state at 595 K and 0.666 g/cm3 (this work) and 580 K and similar density (Figure 18 in ref 20). These changes may be understood by taking into account that water molecules are more mobile at higher temperatures. Regarding hydrogen bonding, we can compare the average value of hydrogen bonds per water molecule in bulk (3.7 for the present model19), in confined water at room temperature16 (3.6

in the central bulk-like region with a clear tendency to decrease as interfaces are approached), and at high temperatures (present work). The effects of confinement and hydrophobicity were also reported by Choudhury and Pettitt10 for finite carbon plates leading to similar conclusions, that is, reduction of hydrogen bonding in the interfaces. In Figure 2, percentages of water molecules forming a given number of hydrogen bonds are shown. The hydrogen-bond definition has been assumed in the geometrical form.19 With increasing temperature, the number of water molecules with 4 and 3 hydrogen bonds decreases, while those with 0, 1, and 2 hydrogen bonds increase. At 633 K, even the number of water molecules forming 2 hydrogen bonds starts to decrease, and there is no water with 5 or 4 hydrogen bonds. The increase of the number of broken hydrogen bonds indicates that the interfacial structure becomes more and more distorted, and this fact leads to the gradual disappearance of the layered structure. Marked changes in all hydrogen-bond profiles are observed at the three highest temperature states (553-633 K). This fact might be due to the reduction of the molecular dipole of water in these particular states indicated above. Another factor to be considered is that the density of water decreases as temperature increases. As it can be seen from Table 1, Ftheo ≈ 0.5 g/cm3 at 633 K. Obviously, the lower bulk density has an effect on the interfacial structure as well. From our earlier results,16 it is clear that there is no lateral ordering beyond the adsorbed layer, even at room temperature, if a low-density phase of water between graphite walls is considered. Thus, we may attribute the disappearance of the intermediate water layer when the system moves nearby the critical point mostly to the decreasing bulk density. The fact that the peak related to the hydrogen of the adsorbed water is broader than the oxygen peak (Figure 1a and b) indicates that water orientation is not homogeneous in this layer. The findings of Choudhury and Pettitt10 indicate a loss of orientational order when water is located far from the outside surface of the graphite walls. This is in good agreement with our results at room temperature. To analyze the water orientational order when temperature increases, we have computed the probability of finding a molecule in a given orientation. In Figure 3, the probability distributions of the cosine of the angle between the molecular dipole vector and the surface normal vector, θdip, are given for the bulk (a), intermediate (b), and adsorbed water (c) sets. Figure 4 shows similar distributions for the cosine of the angle between the molecular OH vectors and the surface normal vector, θOH. Figures 3a and 4a clearly indicate that there is no preferential orientation in the bulk region, as both distributions are homogeneous at all temperatures. We can observe that at the intermediate region some degree of orientational order exists at room temperature. From Figures 3b and 4b, one may deduce that one preferential orientation is with φdip ≈ 140°, θOH1 ≈ 180°, and θOH2 ≈ 75°, showing that the water molecule turns with one OH bond toward the bulk, while the dipole vector is tilted toward the C(0001) surface. In the second preferential orientation, both OH bonds point toward the bulk as θdip f 0° and θOH ≈ 75°. By increasing the temperature, the peaks in the distributions become less and less pronounced and, close to the critical point, the distributions are homogeneous, indicating that water molecules are bulk-like in this region. There are two preferential orientations of the adsorbed water at room temperature (Figures 3c and 4c). One type of molecule is aligned with the instantaneous molecular plane almost parallel to the surface, with both OH bonds pointing slightly toward

12526 J. Phys. Chem. B, Vol. 111, No. 43, 2007

Nagy et al.

Figure 1. HOPG/water temperature-dependent profiles. Top curves correspond to the lowest temperature state (298 K), as indicated by arrows. The curves have been shifted along the density axis for clarity purposes. (a) Oxygens, (b) hydrogens.

Figure 2. Percentage of water molecules forming n hydrogen bonds (HB) for different temperatures.

Figure 3. Orientational distribution of molecular dipole moments in different environments: (a) bulk, (b) intermediate, (c) adsorbed. Temperature dependence is indicated in the plots.

the carbon atoms, because θdip ≈ 100° and θOH ≈ 100°. For the rest of the molecules, one OH bond points to the intermediate layer, with the dipole vector pointing away from the surface as θdip ≈ 60°, θOH1 ≈ 100°, and θOH2 ≈ 0°. These orientations persist at higher temperatures, although the peaks become smaller as temperature increases and the two peaks in θµ merge above the boiling point. Clearly, as the thermal energy increases, interfacial water loses structure and the preferential orientations become less pronounced. Electrostatics. The interfacial orientational inhomogeneities lead to nonuniform charge distribution, and thus to a finite

surface potential drop, ∆χ, across the graphite/water interface. Because we found marked changes in the structural properties of water with increasing temperature, it is expected that the electrostatic properties at the liquid interface will be also temperature dependent. To visualize the effect, we calculated the charge density, F(z), and the electrostatic potential profiles, χ(z), perpendicular to the C(0001) surface.7 The results are shown in Figure 5a and b. As it is already known for ambient conditions,17 the charge density was found to fluctuate within the two interfacial water layers (Figure 5a), leading to fluctuations of the potential (Figure 5b). In the bulk phase, however,

Liquid Water Confined in Carbon Nanochannels

J. Phys. Chem. B, Vol. 111, No. 43, 2007 12527

Figure 4. Orientational distribution of molecular OH directions in different environments: (a) bulk, (b) intermediate, (c) adsorbed. Temperature dependence is indicated in the plots.

Figure 5. Electric properties at different temperatures: (a) charge density, (b) electrostatic potential, (c) potential drop. Temperature dependence is indicated in the plots.

the charge density is zero, as it is expected for a homogeneous dielectrics, and the potential is constant, with a drop of ∆χ ≈ -0.28 V. With increasing temperature, the amplitude of oscillations decreases, and they become confined to the adsorbed layer as the second water layer behaves bulk-like at high temperatures. The corresponding electrostatic potential profiles, understandably, show smaller oscillations confined also to the adsorbed layer as the critical point is approached. The total surface potential drop was found to decrease with increasing temperature; in Figure 5c, the potential drop is displayed as a function of temperature. An exponential decay is observed, and the corresponding exponential fit is plotted. The largest changes may be observed below the boiling point. By further increasing the temperature, the potential drop does not change significantly: ∆χ ≈ -0.45 V. The overall effect of increasing the temperature manifests itself in a relatively large negative potential shift of -0.25 V. Dielectric Properties. The usual way to present the dielectric properties of liquids is by means of the static dielectric constant . For systems with long-range interactions computed with the Ewald sum rule and conducting boundary conditions, the dielectric constant can be computed as:17

 ) 1 + 4π〈M 2〉/3VkBT where V stands for the volume, kB is the Boltzmann factor, and 〈M 2〉 is the averaged total dipole moment of the system. We considered two high-temperature systems (473, 633 K) and compared them to previous results obtained at ambient conditions.17 A cumulative average of the static dielectric constant at 633 K as a function of time is presented in Figure 6. The values of  due to water molecules in adsorbed layers and bulklike water are reported, together with the overall value. There we can observe that all three curves, after a transient regime of

Figure 6. Cumulative average of the static dielectric constant at 633 K. The value obtained for the full system (-) is compared to those from bulk-like (- - -) and adsorbed water (- - -).

around 25 ps, have fully converged to stable values after 40 ps. In Table 2, numerical values for several relevant quantities such as occupation percentages, , and the capacitance per surface unit17 are reported, distinguishing between the overall value and those computed for water molecules located in each region. We observe that errors in  are large at room conditions but they are being reduced as temperature increases, as expected. Interestingly, the percentage of water molecules located in the adsorbed regions is close to 20% in all cases. The big difference arises in the intermediate region, which tends to disappear as temperature increases, as it can be observed from Figure 1. Such region is totally absent at 633 K. In all cases,

12528 J. Phys. Chem. B, Vol. 111, No. 43, 2007

Nagy et al.

TABLE 2: Static Dielectric Constant of Confined Water in Each Region at Three Temperaturesa T/K

region

% occupation



C (µF/cm2)

τresI (ps)

τresC (ps)

298

adsorbed intermediate bulk total adsorbed intermediate bulk total adsorbed bulk total

23 17.0 60 100 20 19 61 100 20 80 100

172 ( 95 30 ( 8 35 ( 5 85 ( 25 34 ( 10 19 ( 5 30 ( 3 27 ( 3 8.3 ( 0.4 5.8 ( 0.2 5.8 ( 0.2

630.9 109.8 18.7 28.5 131.0 64.6 16.6 9.3 31.9 2.4 2.0

54.0 14.0 127.0

12.0 2.8 57.5

16.0 7.5 58.0

5.0 2.0 25.0

5.2 74.4

1.3 10.4

473

633

a The equivalent capacitance per surface unit (C) and residence times of water molecules τres (intermittent, continuous) have been also included.

the bulk-like zone includes most of the molecules of the system, from 60% at 298 and 473 K to 80% at 633 K. This indicates that properties such as dielectric constant or residence times will be strongly influenced by the existence of such large bulk zone. As in ambient conditions, the values for  are bigger in the adsorbed regions and decrease in the intermediate and bulk zones. This fact was already observed by Ballenegger and Hansen25 in slab and spherical geometries. However, at high temperatures, the overall values of  are much closer to those from the bulk-like zone, basically due to the increase of molecular disorder induced by the high thermal energies, which produce a loss of orientational order of the molecular dipoles. Following a previous work,17 we included an estimation of the water capacitance per surface unit26 (C). In all cases, the values are in the range normally observed for capacitances of electrochemical systems in aqueous environments (between 10 and 100 µF/cm2). As it is expected, the reduction of C with increasing temperature can be directly associated with the reduction in  together with the influence of the relative size d of the considered region. The remarkable fact is that at high temperatures the overall value of C is lower than those of the bulk-like regions, whereas at ambient conditions the trend was exactly reversed. This suggests that the low capacitances at high temperatures (as compared to ambient conditions) are mainly due to the large degree of disorder at the molecular level. Residence Times of Water Molecules. The residence time of a water molecule in a given region (τres) may be described as the average time spent by the water molecule in such region before moving away. Such times can be defined assuming that water is not allowed to re-enter the region after its first leave, that is, the continuous residence time (τresC). Conversely, if we allow the multiple re-entrance of water molecules until they definitely leave the given region for a period of time of infinite length, we will deal with the intermittent residence time (τresI). For practical purposes, such “infinite’’ length in time is taken to be of the order of a nanosecond. We collected both residence times at T ) 473 and 633 K and compared them to the values obtained at ambient conditions.17 Our results are reported in Table 2. The general trend is, in all cases, that intermittent values are significantly larger (between 2.3 and 7 times) than continuous ones. The order of magnitude of τresI can be compared to the “occupation time” defined by Choudhury and Pettitt,10 indicating qualitative good agreement with our findings at room temperature. At all temperatures, water at the central bulk-like regions is the most stable, while water in the interfaces tends to be more mobile. The intermediate zone (when it exists) shows the lower values. The non-monotonic behavior of τresI is remarkable: it

Figure 7. Oxygen self-diffusion coefficients at several temperatures. A comparison between values obtained from adsorbed and bulk water is included.

has a large value at 298 K in the bulk region, and it decreases at 473 K and surprisingly rises again at 633 K. We think that this is due to the balance between the strength of graphitewater interactions and molecular motions due to thermal energy. In the case of T ) 633 K, the last is of larger importance than the former, leading to short residence times in the interfaces. Conversely, continuous times decrease with increasing temperatures in a monotonic way. Dynamics. It is well known that diffusion of water becomes faster with increasing temperature, at least for bulk unconstrained water.27,28 In our confined system, such behavior was analyzed by calculating the oxygen self-diffusion coefficients DO by means of the slope of oxygen mean square displacements and, equivalently, through the integration of atomic velocity autocorrelation functions. In Figure 7, we plotted DO separately for bulk and adsorbed water as a function of temperature. As it is expected, DO increases with increasing temperature, and the overall gain is about a factor of 10. The values do not follow an Arrhenius-type behavior; at high temperatures DO grows faster than it is expected. In our previous work,17 we pointed out that water diffusion is hindered in the interfacial region at room temperature. We found this behavior to prevail in the whole temperature region as DO was always found to be larger for the bulk than for the adsorbed layer. This is in good agreement with the findings of Liu et al.29 To more specifically analyze the influence of confinement, we computed the ratio Dbulk/Dads. As it can be observed from Figure 7, this ratio shows a decreasing trend with increasing temperature, indicating a more enhanced change in the water mobility in the adsorbed layer than in the bulk phase.

Liquid Water Confined in Carbon Nanochannels

Figure 8. Maxima of the molecular stretching bands at different temperatures.

Figure 9. Maxima of the molecular bending bands at different temperatures.

One possible way to check the validity of our results is to calculate spectral densities, which may be directly compared to experimental results of infrared or Raman spectroscopy, at least concerning the location of the spectral bands. To do this, we computed spectral densities through the Fourier transforms of the hydrogen and oxygen velocity autocorrelation functions following a standard procedure already employed in previous works (see ref 21, for instance). We extracted the peak positions for the bending and stretching modes for bulk and adsorbed water. In Figure 8, maxima of the stretching band are plotted as a function of temperature. Considering bulk-like water as the reference, a blue shift is observed for adsorbed water along the whole temperature region. This is consistent with our previous knowledge from room-temperature simulations.17 With increasing temperature, the value of both maxima for bulk and adsorbed water tends to increase, with an overall change of about 150 cm-1. This trend may be understood by taking into account the breaking of the hydrogen-bonded water structure with increasing temperature (see Figure 2), which produces a growing number of free OH bonds, which are directly related to large stretching frequencies, gradually approaching the gas-phase values (3650 cm-1 for symmetric stretch and 3750 cm-1 for antisymmetric stretch30). What is new here is the fact that adsorbed water has stretching frequencies always larger than bulk water, which can probably be attributed to the preferential orientations of water in the adsorbed layer. Opposite trends were observed for the bending mode. We plotted the corresponding band maxima in Figure 9 as a function of temperature, both for bulk and for adsorbed water. We consider again the bulk-like phase as the reference, and, in this

J. Phys. Chem. B, Vol. 111, No. 43, 2007 12529 case, we observe a red shift of the adsorbed water bending for the whole temperature region. This is again in good agreement with our previous knowledge from room-temperature simulations.17 On the other hand, the difference between bending values at room temperature and those corresponding to the highest temperature case is about 100 cm-1 for both adsorbed and bulk water. In our opinion, the main reason for this significant frequency shift is again due to the fact that a considerable number of hydrogen bonds are destroyed at high temperatures, with dominance of water molecules forming scarcely two hydrogen bonds. Such predominance of linear water chains is common in vapor, and it is related to the fact that the bending mode tends to reduce its value with respect to liquid water. So, in vapor water the typical bending frequencies are30 of about 1590 cm-1, whereas the value of roomtemperature liquid water is31 1650 cm-1. Conclusions A series of MD simulations of water between two graphene planes as a function of temperature in the range 20-360 °C were performed to determine the properties of solid/liquid interfaces at higher temperatures, especially above the boiling point of water. The water structure changes substantially with increasing temperature. While the water interface consists of two water layers at room temperature, the intermediate water layer gradually disappears by increasing the temperature; at 633 K, the water structure at the graphite/water interface consists of a single layer of adsorbed water. In parallel, the increase of the number of broken hydrogen bonds indicates that the interfacial structure becomes more and more distorted, and this fact leads to the gradual disappearance of the layered structure. The two existing preferential orientations of the adsorbed water at room temperature persist at higher temperatures, although they can be less and less distinguished as temperature increases as the two orientations merge above the boiling point. In accordance with the structural changes, the electrostatic potential through the interface shows smaller and smaller oscillations confined only to the adsorbed layer as the critical point is approached. The total surface potential drop decreases with increasing temperature, with the largest changes below the boiling point. The overall effect results in a relatively large negative potential shift of -0.25 V. Under ambient conditions, the dielectric constant is bigger in the adsorbed regions and decreases in the intermediate and bulk zones. At higher temperatures, the overall values of  are much closer to those from the bulk-like zone, basically due to the increase of molecular disorder induced by the high thermal energies, which produce a loss of orientational order in the molecular dipoles. The water capacitance decreases when the temperature increases; at high temperatures, the overall value of C is lower than those of the bulk-like regions, whereas at ambient conditions the trend was exactly reversed. This suggests that the low capacitances at high temperatures (as compared to ambient conditions) are mainly due to the large degree of disorder at the molecular level. Similarly, the residence time of a water molecule in each of relevant regions considered in this work (adsorbed, bulk) is also influenced by the temperature. The general trend is that τres diminishes with increasing temperature. However, the intermittent residence time in the bulk region has non-monotonic behavior: it has a large value at room temperature, and it decreases with increasing temperature and surprisingly rises again close to the critical point, which can be attributed to the balance between the strength of graphite-water interactions and molecular motions due to thermal energy.

12530 J. Phys. Chem. B, Vol. 111, No. 43, 2007 As it is expected, water diffusivity increases with increasing temperature. The ratio Dbulk/Dads shows a decreasing trend with increasing temperature, indicating a more enhanced change in the water mobility in the adsorbed layer than in the bulk phase. Water vibrations are also influenced by temperature and confinement. Both stretching and bending frequencies are affected by two kinds of frequency shifts, one of them associated with temperature changes and the second due to confinement effects. Acknowledgment. A joint Hungary-Spain “Integrated Action” with references E-36/04 (Hungary) and HH2004-0006 (Spain) is acknowledged. G.N. is indebted for the financial support of the National Science Research Fund, Hungary (OTKA), under Contract No. T049202. M.C.G. has received support from the “Grupo FQM-205” of “Plan Andaluz de Investigacio´n” (PAI) and from the “Ministerio de Educacio´n y Ciencia” of Spain (Grant FIS2006-02356/FEDER funds). E.G. and J.M. gratefully acknowledge financial support from the “Generalitat de Catalunya” (Grant 2005SGR-00779), from the “Ministerio de Educacio´n y Ciencia” of Spain (Grant FIS200612436-C02-01), and from European Union “FEDER” funds (UNPC-E015). References and Notes (1) Schmickler, W. Interfacial Electrochemistry; Oxford University Press: Oxford, 1995. (2) Benjamin, I. Chem. ReV. 1996, 96, 1449. (3) Berard, D. R.; Kinoshita, M.; Cann, N. M.; Patey, G. N. J. Chem. Phys. 1997, 107, 4719. (4) Guidelli, R.; Schmickler, W. Electrochim. Acta 2000, 45, 2317. (5) Heinzinger, K. Molecular dynamics of water at interfaces. In Structure of Electrified Interfaces, Frontiers in Electrochemistry; Lipkowski, J., Ross, P. N., Eds.; VCH Publishing: New York, 1993; p 239. (6) Gallo, P.; Rovere, M.; Spohr, E. J. Chem. Phys. 2000, 113, 11324. (7) Henderson, M. A. Surf. Sci. Rep. 2002, 46, 5.

Nagy et al. (8) Spohr, E. Electrochim. Acta 2003, 49, 23. (9) Kumar, P.; Buldyrev, S. V.; Starr, F. W.; Giovambattista, N.; Stanley, H. E. Phys. ReV. E 2005, 72, 051503. Giovambattista, N.; Debenedetti, P. G.; Rossky, P. J. J. Phys. Chem. C 2007, 111, 1323. (10) Choudhury, N.; Pettitt, B. M. J. Am. Chem. Soc. 2005, 127, 3556. Choudhury, N.; Pettitt, B. M. J. Phys. Chem. B 2005, 109, 6422. (11) Nagy, G.; Denuault, G. J. Eletroanal. Chem. 1997, 433, 153. Nagy, G.; Denuault, G. J. Electroanal. Chem. 1997, 433, 161. Nagy, G.; Denuault, G. J. Electroanal. Chem. 1998, 450, 159. (12) Zangi, R. J. Phys.: Condens. Matter 2004, 16, S5371. (13) Martı´, J.; Gordillo, M. C. Phys. ReV. B 2007, 75, 085406. (14) Nagy, G.; Heinzinger, K. J. Electroanal. Chem. 1990, 296, 549. Nagy, G.; Heinzinger, K. J. Electroanal. Chem. 1992, 327, 25. (15) Gordillo, M. C.; Nagy, G.; Martı´, J. J. Chem. Phys. 2006, 124, 054707. (16) Martı´, J.; Nagy, G.; Gordillo, M. C.; Gua`rdia, E. J. Chem. Phys. 2006, 124, 094703. (17) Martı´, J.; Nagy, G.; Gua`rdia, E.; Gordillo, M. C. J. Phys. Chem. B 2006, 110, 23987. (18) Schmidt, E. In Properties of Water and Steam in SI Units; Grigull, U., Ed.; Springer-Verlag: Du¨sseldorf, 1969. (19) Martı´, J. J. Chem. Phys. 1999, 110, 6876. (20) Brovchenko, I.; Geiger, A.; Oleinikova, A. J. Phys.: Condens. Matter 2004, 16, S5345. (21) Martı´, J.; Padro´, J. A.; Gua`rdia, E. J. Mol. Liq. 1994, 62, 17. (22) Liew, C. C.; Inomata, H.; Arai, K. Fluid Phase Equilib. 1998, 144, 287. (23) Spohr, E. J. Chem. Phys. 1997, 107, 6342. (24) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; DiNola, A.; Haak, J. R. J. Chem. Phys. 1984, 81, 3684. (25) Ballenegger, V.; Hansen, J.-P. J. Chem. Phys. 2005, 122, 114711. (26) C is roughly equivalent to 0/d, where d stands for the length of the considered region, defined by the limits of each region given by the minima of the oxygen density profiles of Figure 1. 0 is the vacuum electrical permittivity. (27) Krynicki, K.; Green, C. D.; Sawyer, D. W. Faraday Discuss. Chem. Soc. 1978, 66, 199. (28) Hausser, R.; Maier, G.; Noack, F. Z. Naturforsch. 1996, A21, 1410. (29) Liu, Y.-C.; Wang, Q.; Lu, L.-H. Chem. Phys. Lett. 2003, 381, 210. (30) Chen, S. H.; Toukan, K.; Loong, C. K.; Price, D. L.; Teixeira, J. Phys. ReV. Lett. 1984, 53, 1360. (31) Bertie, J. E.; Ahmed, M. K.; Eysel, H. E. J. Phys. Chem. 1989, 93, 2210.