Sei sulla pagina 1di 7

This article was downloaded by: [Texas A&M University Libraries]

On: 08 August 2014, At: 14:51


Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

Molecular Physics: An International Journal at the


Interface Between Chemistry and Physics
Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/tmph20

The behaviour of liquid alkanes near interfaces


a

P. SMITH , R. M. LYNDEN-BELL & W. SMITH

Atomistic Simulation Group and Irish Centre for Colloid Science , School of Mathematics
and Physics, The Queen's University , Belfast, BT7 1NN, UK
b

CCLRC, Daresbury Laboratory , Daresbury, Warrington, WA4 4AD, UK


Published online: 01 Sep 2009.

To cite this article: P. SMITH , R. M. LYNDEN-BELL & W. SMITH (2000) The behaviour of liquid alkanes near interfaces,
Molecular Physics: An International Journal at the Interface Between Chemistry and Physics, 98:4, 255-260, DOI:
10.1080/00268970009483289
To link to this article: http://dx.doi.org/10.1080/00268970009483289

PLEASE SCROLL DOWN FOR ARTICLE


Taylor & Francis makes every effort to ensure the accuracy of all the information (the Content) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions

MOLECULAR
PHYSICS,
2000, VOL.98, No. 4, 255-260

The behaviour of liquid alkanes near interfaces


P. SMITH,' R. M. LYNDEN-BELL'* and W. SMITH2
'Atomistic Simulation Group and Irish Centre for Colloid Science, School of
Mathematics and Physics, The Queen's University, Belfast BT7 INN, UK
2CCLRC, Daresbury Laboratory, Daresbury, Warrington WA4 4AD, UK

Downloaded by [Texas A&M University Libraries] at 14:51 08 August 2014

(Received 7 September 1999; revised version accepted 26 October 1999)


Simulations of thick films of liquid alkanes supported on a wax-like substrate were carried out
at a number of temperatures in order to investigate the structure and dynamics of molecules
near the solid-liquid and the liquid-vapour interfaces. Films of butane, octane and a mixture
were investigated. Near the solid surface the liquids were found to be structured and molecular
diffusion slowed. However, there was no evidence of a frozen layer at this interface even near
the bulk freezing temperature. The mixed liquid showed considerable segregation at both
interfaces with preferential absorption of butane at the liquid-vapour interface and octane
at the liquid-solid interface.

1. Introduction
The behaviour of liquids near solid surfaces is important for a range of phenomena, particularly rheology and
lubrication. The introduction of the surface force apparatus [l] and later the atomic force microscope [2] has
allowed the properties of liquids near solids to be probed
on a molecular length scale while atomistic simulations
can be used to relate these properties to the structure of
the liquid near the interface [3, 41. Although the liquidvapour interface is technologically less important,
changes in structure and composition are reflected in
the surface tension.
The simplest liquids are composed of nearly spherical
molecules such as OCTMS and cyclohexane and their
properties can be modelled by spherical Lennard-Jones
particles. The next step in complexity is provided by the
homologous series of normal alkanes, which are particularly important as they are used as lubricants. The
alkanes are a series of similar molecules which differ
only in their length and which have weak and non-specific intermolecular forces.
The results of both simulations and experiments have
demonstrated the presence of layering of liquid density
and a reduction of mobility near a solid surface. Simulations have also provided some examples of preferential
adsorption near the solid surface in liquid mixtures. The
liquid-vapour interface of long chain alkanes shows a
first order surface ordering phase transition just above
the bulk melting transition [5, 61.

* Author for correspondence. e-mail: r.lynden-bell@qub.


ac.uk

Our interest in this topic was rekindled by recent


experiments using neutrons [7, 81 which have shown
the existence of a persistent solid monolayer on a graphite surface under a liquid film at temperatures up to
about 20" above the bulk melting point for a number of
liquids including methane, n-heptane and n-dodecane as
well as alcohols. Among the questions which come to
mind are whether this is a universal phenomenon or
merely the result of specific substrate-liquid interaction.
We were also interested in the liquid-vapour interface [6]
and in simple liquid mixtures. The results described in
this paper suggest that the solid monolayer is probably
the result of the strength of the substrate-liquid interaction and give some interesting insights into the behaviour of mixtures of alkanes near the solid-liquid and
liquid-vapour interfaces.

2. Potentials and simulations


The alkanes were modelled by a united atom model in
which each methylene and methyl group is represented
by a Lennard-Jones centre. The CC bond lengths are
fixed but bond angles are allowed to vary and a torsional potential allows for conformational changes.
The parameters used for the alkanes are the same as
we used in our earlier work on liquid alkanes [6] and
come from the work by Jorgensen et al. [9] with the
bond angle potential given by Weber [lo].
In order to provide a non-specific solid substrate we
constructed a potential with the same average properties
as the liquid. The potential was constructed by averaging the potential of a semi-infinite slab of LennardJones atoms at a uniform density and is given by

Moleculur Physics ISSN 002C8976 print/ISSN 1362-3028 online 0 2000 Taylor & Francis Ltd
http://www.tandf.co.uk/journaIs/tf/00268976.html

Downloaded by [Texas A&M University Libraries] at 14:51 08 August 2014

256

P. Smith et al.

where z is the distance from the substrate, p is the atomic


density of the substrate and u and E are the Lennard
Jones parameters for the atom in question. This potential
has no cut-off and acts on all the atoms. The strength of
the potential is merely a function of the z position and the
density. p was chosen to be equal to the average liquid
density in each run, and the sigma and epsilon values used
were the same as for methylene and methyl groups.
We used the molecular simulation program
DLPOLY [ l l ] with an added hexagonal prism
boundary condition. For all our simulations we used a
Berendsen thermostat with a time constant of 0.1 ps and
a simulation time step of 2fs. We simulated octane and
butane separately and mixed together. The separate
systems contained 240 octane molecules and 480
butane molecules respectively while the mixed system
contained 112 octane and 224 butane molecules giving
the same number of carbon atoms in each type of molecule. The resulting films were 55-70A thick. Initially
the molecules were in the all trans state with their heads
distributed on a face-centred cubic (fcc) lattice. The
systems were then equilibrated at 300K for 400ps;
then cooled and further equilibrated for 400ps at each
temperature before collecting data over 200 ps. Simulations were run for octane at 198 K, 223 K and 293 K and
for butane at 143K, 198K and 223K. The melting

points of octane and butane are 216K and 135K respectively, while the boiling points are 399K and
273 K [12]. These temperatures were chosen to give
some runs near the freezing point and some at higher
temperatures for comparison. The simulation of the
mixed system was performed at 223 K.
3. Results
The structure of the films of neat liquids near the substrate showed density oscillations extending to about 4 or
5 layers (20-25 above the substrate. Above this region
there was a region of bulk liquid about 30-40 wide with
uniform density below the liquid-gas interface region.
Thus the two interfaces in the system were far enough
apart that any interaction between them was negligible.
A more detailed study of the properties of the molecules
in the layered region near the solid substrate showed that
for both substances individual molecules tended to lie
parallel to the surface and the percentage of gauche
defects was less than in the bulk.
The density profile of the mixture of butane and
octane is shown in figure 1, which shows interesting
features at both the solid-liquid and the liquid-vapour
interfaces. The liquid-vapour surface is almost pure
butane, with an excess of octane in the region below.
The situation near the solid surface is somewhat more
complex. Figure 2 is a snapshot showing the first layer
above the substrate. The most major feature to notice is

A)

0.8 m

0)

0.5

Figure 1. The partial densities of butane (-) and


octane (---) in a mixed
film at 223K averaged
over 200 ps. The solid substrate is on the left and the
vapour on the right, The
solid substrate causes
layering in the liquid out
to 25A. The inset shows
the density near the substrate in greater detail.
Note two peaks between
the first and second layers
which arise from bridging
butane molecules.

2.0

10.0

20.0

30.0

40

40.0

Distance from substrate in 8,

6.0

50.0

8.0

10.0

60.0

Downloaded by [Texas A&M University Libraries] at 14:51 08 August 2014

The behaviour of liquid alkanes near interfaces

Figure 2. A slice through a configuration of the system showing tke molecules in the first layer nearest the substrate
( M A ) . Note that there are more butane than octane
molecules in this layer, and some of the butane molecules
are normal to the view and contain less than four atoms in
this slice. These are the molecules that bridge between the
first and second layers.

that there is more butane than octane in this layer and


that this cannot be due to packing frustrations between
regions of ordered octane molecules since there are
regions where octane molecules could replace butanes.
We have to seek other explanations for this. The lowest
layer contains about 25% by mass more butane than

257

octane, but the next two layers are strongly dominated


by octane molecules. Thus overall octane is preferentially adsorbed at the solid-liquid interface while
butane is preferentially adsorbed at the liquid-vapour
interface. Figure 2 gives an idea of the size of the
cross-section of the simulation cell in relation to the
size of the molecules. It also shows some gauche defects
in both butane and octane molecules and molecules
bridging between the first and second layer which can
also be seen by the sharp peaks between the first and
second layer in the inset of figure 1 .
The mobility of molecules in the layer nearest the
solid substrate provides a good measure of the nature
(solid or liquid) of this layer. We measured the in-plane
diffusion constant from the slope of the mean square
displacement parallel to the surface as a function of
time for molecules: in the first layer; in the second
layer; for a region with a similar number of molecules
in the central part of the film for both neat liquids and
for the mixture; and in the surface region of the mixture.
The boundaries used to define these regions wert:
bottom layer ct6
second layer 6-10
bulk 25-35 A
and surface 45-55
The mean square displacement was
calculated separately in both x and y directions and the
two sets of data used to estimate the uncertainty in our
results. For the mixture two independent runs were performed yielding four sets of data which were used to
calculate the uncertainty.
Figures 3 and 4 show the in-plane diffusion constants
for different regions of the films of butane and of octane
at a number of temperatures. We can see in every case

A;
A.

A;

1.5

Butane

4C

1.0

.-

.-0
u)

/#

=I

E
U

(I)

m
-

5 0.5

Bulk like region

/#

//

f---------------------~

Figure 3. In-plane diffusion


coefficients in butane
films as a function of ternperature for the first order
layer, the second ordered

Second layer

I--------------d
0.0

I
I

I
I

24 1.0

258

P. Smith et al.

Octane

-4 1.0
C
.-

0
.v)

E
U
al
C

m
Q

Downloaded by [Texas A&M University Libraries] at 14:51 08 August 2014

= 0.5
Figure 4. In-plane diffusion
coefficient in octane films
as a function of temperature for the first order
layer, the second ordered
layer and a bulk-like
region.

0.0

190.0

21 0.0

230.0
250.0
270.0
Temperature in degrees Kelvin

21

Mixture 223K

1.5

.
ul
Q

290.0

.-K
K
ul
.-0
ul

E
U

x-------_-_{
-,
,

1 -

2-

Octane

al
C

m
Q

-C
Figure 5. The diffusion coefficient for a mixture of
octane (- - -) and butane
(-) in the first layer (A),
the second layer (B), the
bulk ( C ) and the surface
region (D). The lines are
drawn as a guide to the
eye.

that the diffusion in the two layers nearest the solid


substrate is substantially less than in the bulk and that
the diffusion rates decrease with temperature. However,
the diffusion constant in the layer next to the solid does
not vanish even at the lowest temperatures. We deduce
that the lowest layer is behaving as a liquid rather than
being frozen, even near the bulk freezing point. In runs

1
D

B
Layer in System

at the same temperature (223K) butane diffuses more


rapidly than octane as one would expect.
Diffusion coefficients for films of mixed alkanes at
223 K are shown in figure 5. In the second layer and in
the bulk the diffusion constants of both types of molecules are similar although in pure liquids they differ by
a factor of two (bulk) or more (second layer). In these

The behaviour of liquid alkanes near interfaces

Downloaded by [Texas A&M University Libraries] at 14:51 08 August 2014

regions large molecules diffuse faster, suggesting that the


viscosity is less than in pure octane. The mobility of
butane in the surface (where it is favoured) is substantially higher than in both the bulk mixture and in pure
butane.
4. Discussion
Layering near the liquid-solid interface is a universal
feature of liquids, which has been demonstrated experimentally [2, 131 and in many simulations. The phenomenon is observed both in long chain fluids and in fluids
of spherical molecules. The first question we wish to
address is how far the increased melting point of the
lowest layer observed by Castro et al. [7] is also a universal phenomenon or whether it depends on the
strength and degree of corrugation of the substrateliquid potential.
The cause of layering is the packing of the molecules
against the surface. Part of this is an entropic effect due to
the increase of free volume of individual molecules when
the density is layered compared to that in a uniform density. This is shown by the existence of layering in simulations of hard particles against hard walls [14]. However
the importance of energetic effects is demonstrated by
simulations with varying surface potentials. Winkler
et al. [15] compared the layering in hexadecane on a corrugated model of graphite with a purely repulsive flat surface and Gupta et al. [I61 studied both the structure and
dynamics of liquid octane on a series of corrugated Lennard-Jones surfaces with varying degrees of attraction for
the liquid. These studies show that the magnitude of
attraction and to some degree the extent of corrugation
of the surface have a large effect on the amplitude of the
density oscillations near the surface and the mobility of
molecules in various layers. However, Balasubramanian
et al. [17], who performed Monte Carlo simulations of
n-hexadecane films ranging from l 4 n m on a featureless
flat metal (Au) surface over a range of temperatures from
350K to 650K, found that the only effect of using a flat
substrate is the loss of correlations in the plane structure
in the first layer above the substrate between molecules in
the layer and the substrate.
In our simulations we have endeavoured to cover the
temperature range down to that of the bulk melting
point in order to investigate variations in mobility.
There are two difficulties: first we do not know the
melting points of our model alkanes (as opposed to
the experimental melting points); and secondly the
time scale of our simulations are limited. However we
are using comparatively small molecules and simulating
for times longer than 0.5 ns so that there should be time
for equilibration. Although the liquid mobility near the
substrate is greatly reduced at low temperatures, we
always find a measurable diffusion constant. Our sub-

259

strate has the same average attractive properties as wax,


which is less attractive than either graphite or metals.
The only demonstration of an ordered unmelted monolayer of alkanes at a solid-liquid interface in a simulation that we are aware of is in the simulations of Winkler
et al. [I51 of hexadecane on graphite. When they
replaced the graphite by a purely repulsive surface this
monolayer melted. On the other hand a decrease in
mobility near the surface has been observed in many
simulations [18, 191 as well as in our own. We believe
that the persistence of mobility in the lowest layers, even
at the lowest temperatures, is good evidence that the
large increases in melting temperature (about 10%)
seen by Castro et al. for a wide range of liquids on
graphite are the result of the stronger attraction of molecules by the substrate than by the bulk liquid.
Segregation of molecules near interfaces is the result
of the change in balance between energy and entropy for
different types of molecules; it has been observed in
previous simulations, for example in a study of benzene-heptane mixtures in a graphite slit (with no
liquid-vapour interface) [ 191 and in simulations of
films of hexane-hexadecane mixtures on a gold surface
[20] (with both solid-liquid and liquid-vapour interfaces). Considering first the solid-liquid interface, we
note that in Xia and Landmans [20] simulations the
preferential adsorption of the long molecules was strongest in the bottom layer and decreased as theobulk was
approached, reaching the bulk ratio by 20A. In the
second and third layers our results (which are at a
much lower temperature, 223 K in place of 31 5 K) are
similar, showing a high proportion of the longer octane
molecules, but in the first layer the proportion of octane
molecules is lower. A similar difference between the first
and higher layers was seen by Kotelylanskii and
Hentschke [I91 in their study of benzene-heptane. However in that case the explanation is probably that the
interaction between benzene molecules does not favour
a planar arrangement so that the energy in the first layer
can be reduced by adsorbing heptane molecules in preference to benzene molecules. As these considerations do
not apply to alkanes, we must look elsewhere for the
explanation of our observations.
We believe that the segregation that we observe is due
to the relative stabilization of octane versus butane
when the solvent becomes layered. Within the layers
the molecules are confined to two dimensions rather
than three. The increase in free energy on the introduction of a gauche defect is larger in two dimensions than
three as the energetic terms increase and the entropic
stabilization decreases, so the probability of gauche
defects is reduced compared to bulk liquid and the
energy is lowered. This stabilization is greater for
longer molecules as there are more possible sites for

Downloaded by [Texas A&M University Libraries] at 14:51 08 August 2014

260

The behaviour of liquid alkanes near interfaces

gauche defects in a long chain. This is a similar effect to


that found in our previous work [6] where the liquid
vapour interface encouraged a surface ordered layer.
The lowering of the energy due to the intermolecular
interactions is also greater for longer molecules, which
can pack more densely than shorter ones as the shortest
intermolecular CC distances are greater than the intramolecular CC bond lengths. Pursuing this argument we
would expect octane to be preferred over butane. Why
does this not happen in the lowest layer? Although it is
possible that in this layer, which is even flatter than the
overlayer, the optimum packing corresponds to butane
molecules filling up interstices between the octane molecules, the snapshot shown in figure 2 suggests that this
cannot be the sole explanation.
The segregation of small molecules at the liquidvapour interface can be understood from the thermodynamic point of view as being the result of the lower
surface tension (which is equal to the surface free
energy for a liquid) of smaller alkanes compared to
longer ones. This still leaves the question as to why
the surface tension at a given temperature decreases
with chain length. To take an example, the difference
of surface free energy between octane and butane at
273 K is 8.86 mJ mp2 which is made up of contributions
of 2.3 mJm-2 of energy and 6.5 mJ mp2 from the -TAS
term [21]. The positive surface entropy is probably the
result of the decrease in density across the interface, and
the consequent increase in free volume for molecules in
this region. We note that there is no significant ordering
of the molecular orientations or change in the percentage of gauche defects near the liquid-vapour interface.
Thus these short chain alkanes do not show any surface
ordering even near the bulk melting transition. Experimentally [ 5 ] , it is also true that surface ordering only
occurs in alkanes whose chain length is greater than 15.
In these simulations we have found no evidence for an
immobile layer near the solid substrate, which we have
interpreted as suggesting that the strength of the solidliquid interaction is responsible for the experimental
observations. Preferential absorption occurs at both
the solid-liquid and the liquid-vapour interfaces with
longer molecules adsorbed at the solid and shorter
ones at the vapour interface.

We thank S. M. Clarke and R. K. Thomas for


drawing attention to this problem and providing
copies of their work before publication, and the referee
for drawing our attention to [21]. We acknowledge
financial support from EPSRC (grants GR/K2065 1
and GR/L08427), the IF1 (grant to the Irish Centre
for Colloids and Biomaterials), DEN1 and the Daresbury Laboratory (CAST award to PS).
References
[l] ISRAELACHVILI,
J. N., 1987, Arc. Chem. Res., 20, 415.
[2] OSHEA,S. J., WELLAND,
M . E., and RAYMENT,
T., 1992,
Appl. Phys. Lett., 60, 2356.
[3] GELB,L. D., and LYNDEN-BELL,
R. M . , 1993, Chem.
Phys. Lett., 24, 328; 1994, Phys. Rev. B, 49, 2058.
[4] PATRICK,
D. L., and LYNDEN-BELL,
R. M., 1997, Surf.
Sci., 380, 224.
[5] EARNSHAW,
J. C . , and HUGHES,
C. J., 1992, Phys. Rev. A ,
46. 4494.
SMITH,P., LYNDEN-BELL,
R . M., and SMITH,W., 1999,
Molec. Phys., 94, 249.
CASTRO,M. A., CLARKE,
S. M., INABA,A., and THOMAS,
R. K . , 1997, J. phys. Chem.. B101, 8878.
CASTRO,
M . A . , CLARKE,
S. M., INABA, A,, ARNOLD,T . ,
and THOMAS,
R . K . , 1998, J. phys. Chem., B102, 10528.
JORGENSEN, W. L., MADURA,J. D., and SWENSON,
C. J.,
1984, J . Am. chem. Soc., 106, 6638.
WEBER,T. A . , 1979, J . chem. Phys., 70, 4277.
FORESTER
T. R . , and SMITH,
W . , 1996, J . molec. Graphics,
14, 136
LIDE,D. R . , 1994, Handbook of Chemistry and Physics
(Boca Raton: CRC Press).
CHRISTENSON,
H. K., GRUEN,D. W. R., HORN,R. G . ,
J. N., 1987, J. chem. Phys., 87, 1834.
and ISRAELACHVILI,
GROOT,R . D., FABER,N. M., and VAN DE ERDE,J. P.,
1987, J. chem. Phys., 62, 861.
WINKLER,R . G., SCHMID,R. H., GERSRMAIR,
A., and
REINEKER,
P., 1996, J . chem. Phys., 104, 8103.
GUPTA,S., KOOPMAN,D. C . , WESTERMANN-CLARKE,
I . A . , 1994, J . chem. Phys., 100,8444.
G . B., and BITSANIS,
BALASUBRAMANIAN,
S., KLEIN,M. L., and SIEPMANN,
J. I., 1995, J. chem. Phys., 103, 3184.
XIA, T. K . , OUYANG,J . , RIBARSKY,M. W., and
U., 1992, Phys. Rev. Lett., 69, 1967.
LANDMAN,
KOTELYANSKII,
M . J., and HENTSCHKE,
R . , 1995, Phys.
Rev. E. 49, 910.
XIA,T. K . , and LANDMAN,
U., 1993, Science, 261, 1310.
R OSSINI,D., 1952, Selected Values of Physical and
Thermodynamic Properties of Hydrocarbons and Related
Compounds (Washington: Carnegie Press).

Potrebbero piacerti anche