You are on page 1of 6

mqlU

ELSEVIER Physica A 231 (1996) 117-122

Molecular dynamics investigation of an ethanol-water solution


M o u n i r T a r e k * , D o u g l a s J. T o b i a s , M i c h a e l L. K l e i n Department of Chemistry, University of Pennsylvania, Philadelphia, PA 19104-6323, USA

Abstract A molecular dynamics simulation of a 0.1 M ethanol-water solution with an air/solution interface was performed. Redistribution of ethanol molecules was observed during the simulation, which was initiated from a bulk solution. The results of the simulation show good agreement with surface tension measurements and the number density profiles of the ethanol excess from neutron reflectivity experiments. A depletion layer beneath the ethanol surface excess was revealed by the simulation. Ethanol molecules are oriented at the surface such that the alkyl group points out of the solution. The number of water molecules involved in the hydrogen bonding with the ethanol molecules decreases by a factor of 2 between the surface and the bulk.

1. Introduction Despite the importance of ethanol-water mixtures in surface chemistry and industrial applications, very little is known about their interface structure. In the early 1930s, these mixtures were thermodynamically characterized by means of surface tension and surface-excess measurements [1-4]. The application of the Gibbs equations to the mixture has been discussed, but there has been some confusion with regard to the structure of the dividing surface. It has been suggested, but not confirmed, that a water layer exists beneath the alcohol monolayer at the surface. Recently, some insight into the structure of the interface has been provided by neutron reflectivity measurements [5]. The authors have applied this technique to determine the density profiles of the different components of the mixture along the normal to the interface. In this paper we report the preliminary results of a Molecular Dynamics simulation of a 0.1 M aqueous ethanol solution. The simulation reproduced the observed segregation of the ethanol molecules from the bulk mixture to the interface, and permitted a more * Corresponding author. 0378-4371/96/$15.00 Copyright (~) 1996 Elsevier Science B.V. All fights reserved SSD! 0378-4371(95)00450-5

118

M. Tarek et al. IPhysica A 231 (1996) 117-122

detailed characterization of the structure of the system. After presenting the models and methods used for the calculation, we will compare simulation and experimental results and provide some additional details of the structure of the solution.

2. Models and MD simulation An intermolecular potential consisting of site-site Coulomb and Lennard-Jones interactions was employed. The SPC/E model was used for the water molecules [6], and a united atom model was used for the ethanol molecules [7]. A torsional potential for the C-O bond was included [7]. The system simulated contained 95 ethanol and 854 water molecules (i.e. 0.1 M ethanol solution). An initial configuration was set up starting from a well equilibrated water bulk system (33.21x33.21 x31.75 A3), and the appropriate number of water molecules (chosen randomly) were replaced by ethanol molecules. Following energy minimization of the solution, a MD simulation was carried out for 15 ps at constant temperature (T = 298K) and 70ps at constant temperature and pressure (P = 0) during which time the density of the system converged near the experimental value [8]. The final configuration of the constant pressure run was used to initiate the simulation of the inhomogeneous system. After extending the z dimension of the simulation box equally at both ends to a length of 100 A, 343 ps of dynamics was performed at constant volume and temperature. The constant temperature and pressure dynamics were carried out using an extended system method [9] in which the equations of motions were integrated using a Verlet-like algorithm [10] with a time step of time of 1.5fs. Bonds involving hydrogen atoms were constrained using the SHAKE algorithm. The Ewald method was used to compute the electrostatic interactions and the minimum image convention was employed to calculate the real-space part of the Ewald sum and the van der Waals interactions [ 11].

3. Results and discussion


3.1. A 9 9 r e g a t i o n at the interface

To illustrate the rate of aggregation of the ethanol molecules at the interface from the bulk solution, density profiles of the components of the mixture are displayed at different stages of the run in Fig. 1. It appears that monolayers of ethanol form at both sides of the water bulk in less than 50ps. Fluctuations of the densities of the bulk and interface regions occur on longer timescales. When averaged over the whole simulation run (Fig. 2), the profiles for the surface L excess ethanol have a symmetric shape. A snapshot of the last configuration, shown in Fig. 3, and the overall shape of the density profiles in Fig. 2, reveal the existence

M. Tarek et al. IPhysica A 231 (1996) 117-122

119

P(Z)l:'" o.o3~

.... ' .... ' .... ' .... '"~ / ~ a l

I:'" .... ' .... ' .... ' .... "':1 o.o3~~'~..~ bl

o.o3~-I:'" ....]~v~c~' ' .... '"l .... ' ....

o.o~ ~{ _:~ I I o.o~ :/.... ~- ,1 t o.o~ r~ ,. t~ 1 '7 j/".....i ".,,."t t 'r O",,...""<-it t 'L~//--~-t....,.."t 1 .....
-2o-to o ~o ~o -2o-~o o ,o ~o -2o-~o o ~o 20

(.~)

Fig. 1. Number density profiles for ethanol (dashed lines) and water (solid lines) averaged over 15 ps: (a) 4 5 ~ 0 ps; (b) 90-105 ps; (c) 145-160 ps.

0.03

~,,~0.02
0.01

o
-20-10
z

10

20

(1)

Fig. 2. Number density profiles averaged over the entire simulation for ethanol (dashed lines) and water (solid lines). Note the ethanol depletion just inside the surfaces.

Fig. 3. Final configuration of the system from the simulation. The ethanol molecules are drawn as spheres with their van der Waals radii and the water molecules as sticks.

120

M. Tarek et al. IPhysica A 231 (1996) 117-122 150

100

Z
v

50

0
~" - 5 0
-

100 ~ 280

i J

,,,I,,,17

300

320

340

t(ps)
Fig. 4. Surface tension as a function of simulation time (solid line). The dashed line corresponds to the experimental value.

of a region of enhanced water density and depleted ethanol density (depletion layer) between the surface and the interior. The water density in this layer approaches, but is lower than, the pure water density (0.033 molecule//k3).

3.2. Comparison to experiment


The surface tension of the system under study has been calculated for the last 60ps of the run (see Fig. 6). It is well-known that simulation estimates of the surface tension of systems containing hydrogen-bonded molecules are subject to large errors. With this in mind, at the surface area/molecule (concentration) studied here, the calculated surface tension, 54 + 35 mN/m, is in reasonable agreement with the experimental value of 35 mN/m. In order to compare the number density profiles to the results of neutron reflectivity measurements [5], we point out first that the authors have performed their analysis by assuming shape distributions which do not take into account any depletion layer beneath the ethanol monolayer. In fact, the experimental data reduction was carried out assuming a Gaussian distribution (Pc) for the excess ethanol at the surface, and a tanh distribution (Ps) for the 0.1 M bulk solution, i.e., Ps=~Ps l+tanh , (1)

pe = pOeexp( -41n2(z- ze)z) ~

(2)

It is clear that the simulation results can be compared to the experiment only to a limited extent according to the same assumption. For this purpose, the ethanol density profile has been divided in two parts: the first is the excess, which appears to have a Gaussian shape as assumed by the experimentalists, and a second part, in which the density increases until reaching a constant value at the bulk solution (Fig. 5). Capillary waves appear to be appreciably longer than the simulation cell and therefore are not observable in the simulation [12]. Taking into account the effect of the thermal

M. Tarek et al./Physica .4 231 (1996) 117-122


0.025p .... I .... i .... J .... I ....

121

0.02 ~-~-0.015
N

0.01 0.005 ~,, , ~ , ~ ~,, ,,,I,,,, I.,I 0 5 10 15 20 25


z

Fig. 5. Number density profiles of all ethanol molecules (solid line), and division into Gaussian profiles for excess and bulk ethanol molecules (broken lines).
0.1 0.08 0.06 0.04 0.02 0 ........ I ....
I''

t:L

50

100 150

0 (deg)
Fig. 6. Probability distributions of the angle with respect to the surface normal for the ethanol HO (solid line) and OC (dotted line) bonds. roughness [5], the width o f the ethanol excess at the interface determined by experiment is ae = 4-I- 1 A, which agrees very well with the value obtained directly from the simulation, 4.6 0.2 A.
3.3. Orientation o f the molecules

The local orientation o f the ethanol molecules at the surface has been analyzed in terms o f angular distributions o f the HO and OC bonds with respect to the normal to the interface (Fig. 6). The hydroxyl group displays a broad, asymmetric distribution with a maximum around 60 , while the OC bond is oriented such as the alkyl group projects out from the interface.
3.4. Association with water molecules

Ethanol-oxygen/water-oxygen radial distribution functions were calculated for the ethanol excess and for the bulk solution. Integrating up to the first minima in g00(r),

122

M. Tarek et al./Physica A 231 (1996) 117-122

the coordination numbers are 2.8 for the bulk and 1.3 for the excess. This indicates, at least in the first approximation, that the hydrogen bonding network is quite different in the two regions. This observation is confirmed by a detailed analysis of the hydrogen bonding to be presented elsewhere [13]. In summary, a molecular dynamics simulation of a 0.1 M ethanol-water solution was performed with an air/solution interface. The results compare favorably with the available structural and thermodynamic measurements. Moreover, the simulation confirms the proposed existence of a depletion layer between the surface and the interior. Orientational ordering of the ethanol molecules at the interface and the hydrogen bonding network around the molecules have been briefly analyzed. A more detailed analysis of the structure and dynamics of the system is presently in Ref. [13]. Furthermore, the system size dependence of the results and the properties of the bulk region are being investigated in a simulation of a system twice as large as the one studied here.

Acknowledgements
This research was supported by the National Institutes of Health under GM 40712.

References
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] R.K. Schofield and E.K. Rideal, Phil. Mag. 13 (1932) 807. J.A.V. Butler and A. Wightman, J. Chem. Soc. (1932) 2089. W.F. Kenrick Wynne-Jones, Phil. Mag. 12 (1931) 907. E.A. Guggenheim and N.K. Adam, Proc. R. Soc. Lond. A 139 (1933) 218. Z.X. Li, J.R. Lu, D.A. Styrkas, R.K. Thomas, A.R. Rennie and J. Penfold, Mol. Phys. 80 (1993) 925. H.J.C. Berendsen, J.R. Grigera and T.P. Straatsma, J. Phys. Chem. 91 (1987) 6269. W.L. Jorgensen, J. Phys. Chem. 90 (1986) 1276. J. Timmermans, Physico-Chemical Constants of Binary Systems in Concentrated Solutions, Vol. 2 (Interscience, New York, 1960). G.L. Martyna, D.J. Tobias and M.L. Klein, J. Chem. Phys. 101 (1994) 4197. G. Ciccotti and J.P. Ryckaert, Comput. Phys. Rep. 4 (1986) 345. M.P. Allen and D.J. Tildesley, Computer Simulation of Liquids (Oxford University Press, Oxford, 1989). M. Tarek, D.J. Tobias and M.L. Klein, J. Phys. Chem. 99 (1995) 1393. M. Tarek, D.J. Tobias and M.L. Klein, J. Chem. Soc. Faraday Trans. 92 (1996) 559.

You might also like