You are on page 1of 9

Philosophical Magazine

ISSN: 1478-6435 (Print) 1478-6443 (Online) Journal homepage: https://www.tandfonline.com/loi/tphm20

Slow dynamics of a confined supercooled binary


mixture in comparison with the bulk phase

P. Gallo, A. Attili & M. Rovere

To cite this article: P. Gallo, A. Attili & M. Rovere (2004) Slow dynamics of a confined
supercooled binary mixture in comparison with the bulk phase, Philosophical Magazine,
84:13-16, 1397-1404, DOI: 10.1080/14786430310001644143

To link to this article: https://doi.org/10.1080/14786430310001644143

Published online: 02 Sep 2006.

Submit your article to this journal

Article views: 25

View related articles

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=tphm20
Philosophical Magazine, 1 May–1 June 2004
Vol. 84, Nos. 13–16, 1397–1404

Slow dynamics of a confined supercooled binary mixture


in comparison with the bulk phase

P. Galloy, A. Attili and M. Rovere


Dipartimento di Fisica, Università di Roma Tre, Istituto Nazionale per la Fisica
della Materia Unità di Ricerca Roma Tre, Istituto Nazionale per la Fisica della
Materia Democritos National Simulation Center, Via della Vasca Navale 84,
I-00146 Roma, Italy

Abstract
Dynamic properties of a liquid Lennard-Jones binary mixture embedded in an
off-lattice matrix of soft spheres and the corresponding bulk phase are studied
upon supercooling close to the crossover temperature of mode-coupling theory
by molecular dynamics simulations. The model confining system is designed
to mimic the properties of silica xerogels. In spite of the strong confinement the
dynamic behaviour of the mixture is consistent with the mode-coupling
predictions. The parameters of the theory are compared for bulk and confined
phases. The range of validity of the theory shrinks in confinement. Consistently
with the hypothesis of a growing length scale in supercooled liquids, the dynamics
of the confined mixture are faster than those of the equivalent bulk phase.

} 1. Introduction
In many relevant situations, liquids are confined in different environments.
Experiments clearly show that the physical properties of the bulk phase can change
drastically upon confinement. It is therefore of paramount importance both to assess
the extent of the modifications of these properties in the specific systems and to
extract general trends out of the vast phenomenology (Frick et al. 2000).
In the present study we focus in particular on the modification of the dynamic
properties of a model liquid upon supercooling when confined, addressing the impor-
tant issue of the approach to the glass transition. Modification of the glass transition
temperature have been reported in literature to be strongly dependent on the kind of
confining potential (Frick et al. 2000).
One of the most successful theories in describing the behaviour of a liquid under-
going a glass transition is the mode-coupling theory (MCT) (Götze and Sjögren
1992). The MCT is able to describe the modifications of the dynamic scenario
of bulk supercooled liquids on approaching a crossover temperature Tc. At Tc the
system passes from a regime where ergodicity is attained through structural relaxa-
tions to a regime where this mechanism is frozen and only activated processes allow
the system to explore the configurational space. Only upon further supercooling do
hopping processes freeze, and then the liquid becomes a glass. Tc can be estimated
through experiments and computer simulations or predicted by the theory.

y Author for correspondence. Email: gallop@fis.uniroma3.it.

Philosophical Magazine ISSN 1478–6435 print/ISSN 1478–6443 online # 2004 Taylor & Francis Ltd
http://www.tandf.co.uk/journals
DOI: 10.1080/14786430310001644143
1398 P. Gallo et al.

Molecular dynamics studies of model liquids in restricted geometries, intended to


assess the applicability of MCT, can be an important contribution to the character-
ization of the glass transition scenario in confinement. In this field, recent progress
has been made for water confined in a silica nanopore (Gallo et al. 2000a,b, Gallo
and Rovere 2003) and for confined thin polymer films (Varnik et al. 2002).
In this paper we present recent results from an extensive study on a Lennard-
Jones binary mixture (LJBM) embedded in a disordered array of soft spheres (Gallo
et al. 2002, 2003, Attili et al. 2003) designed so as to mimic the properties of silica
xerogels (Page and Monson 1996). The aim of the present work is to perform a close
comparison of the dynamic properties of the confined mixture with those of an
equivalent bulk.

} 2. Models and simulation details


The confining medium consists of a rigid disordered array of soft spheres. The
simulation box contains 16 soft spheres labelled in the following with the letter M.
The interspaces of this structure host a liquid LJBM of 1000 particles, composed of
800 particles of type A and 200 particles of type B. The interaction pair potential can
be written in the following general form:
   6 
 12 
V ðrÞ ¼ 4  ; ð1Þ
r r
where the indices  and  indicate the particle types A,B,M. The parameters for the
three components are listed in table 1. The parameters of the LJBM have been
chosen as in the work by Kob and Andersen (1995a,b) where a full test of the
MCT properties of the bulk was performed. The shifted-potential technique with
a cut-off of 2.5  has been used for the simulations. Periodic boundary conditions
are applied. In the following, Lennard-Jones units are used. The box length of the
confined system is L ¼ 12:6. The LJBM is in a strongly confined environment as only
a few layers of particles can accommodate the soft spheres. We have conducted
molecular dynamics simulations of this confined system in the NVE ensemble. The
system was equilibrated at different reduced temperatures via a velocity rescaling
procedure. Owing to the soft-spheres potential the local number density of the
mixture strongly depends on temperature. In fact, in a previous study (Gallo et al.
2003) by means of an estimate of accessible free volume through the Voronoi tes-
sellation, we verified that the number density of the confined system undergoes a
drastic variation upon decreasing temperature, changing from   0:7 to   1:2.
We therefore decided to compare the behaviour of the confined system along the
isochore with a bulk phase at the same pressures and temperatures. For the bulk

Table 1. Parameters of the Lennard-Jones and soft-sphere potentials as


defined in equation (1). In the table, A and B refer to the particle of
the binary mixture while M refers to the confining soft spheres.
Values are expressed in Lennard-Jones units.
  
AA 1.0 1.0 1
BB 0.88 0.5 1
AB 0.8 1.5 1
MA 3.0 0.32 0
MB 2.94 0.22 0
Slow dynamics of a confined supercooled binary mixture 1399

0 0.5 1 1.5 2 2.5 3


1.25
16 a)
1.2
14
1.15
12
1.1
P (ε/σ )

ρ (σ )
3

P bulk

-3
10 P conf. 1.05
ρ bulk
8
1

6 0.95

4 0.9

b)
10

E bulk
5 E conf.
E (ε)

-5

0 0.5 1 1.5 2 2.5 3


1/T (ε/kB)

Figure 1. (a) Pressure as a function of inverse temperature for both bulk (f) and confined
(*) systems. The bulk density corresponding to the simulated state points is also
shown (m): (——), (– – – –), (—  —), polynomial interpolations. (b) Total energy
per atom versus inverse temperature for bulk (g) and confined (œ) systems: (——),
(– – – –), polynomial interpolations.

system, equilibration at the desired values of T and P was reached through the use of
the Berendsen weak-coupling scheme performing rescaling of both velocity and par-
ticle positions. Production runs are carried out in the microcanonical ensemble for
the bulk system as well. State points of both systems are reported in figure 1 together
with the densities of the bulk LJBM. These range from  ¼ 0:9 to  ¼ 1:2 and
therefore appear quite similar to the estimated corresponding values for the confined
system. For the bulk LJBM the temperatures investigated range from T ¼ 5:0 to
T ¼ 0:48 while for the confined LJBM they range from T ¼ 5:0 to T ¼ 0:37.
T ¼ 0:48 and T ¼ 0:37 are, for the bulk and the confined systems respectively, the
lowest temperatures at which it is possible to equilibrate. In figure 1 the total energies
per particle are also reported for both systems. We found at higher temperatures an
1400 P. Gallo et al.

average increase in the internal energy for the confined liquid of about 5.0 A with
respect to the bulk value; this trend reduces asymptotically to 3.0 A as the systems
are supercooled. We further note that no thermodynamic instabilities are observed
for the state points investigated as no abrupt changes in thermodynamic quantities
are evident.

} 3. Single-particle dynamics of the bulk and


confined Lennard-Jones binary mixture
In order to highlight analogies and differences in the MCT behaviours of the
bulk and confined LJBMs we studied the single-particle dynamics of species A and B
for both systems in detail.
The van Hove self-correlation function (VHSCF) can give an important contri-
bution to this characterization. We verified that at high temperatures the two systems
exhibit very similar correlators (not shown here). For these temperatures both bulk
and confined distributions are seen to reproduce reasonably well the expected Gaussian
shape. As supercooling progresses in both systems the van Hove correlators deviate
from Gaussian behaviour and differences between corresponding curves are enhanced.
In particular, atoms in the confined system are more mobile than those in the bulk
system at the same temperature. This is shown in figure 2 where the correlators of the

Confined Bulk
0.01 0.1 1 0.01 0.1 1
10 10
a) b)
8 8
4πr G (r,t)

6 6
(s)
2

4 4

2 2

0 0
4 4
10 10

2 2
10 10
t

0 0
10 10
c) d)

-2 -1 0 -2 -1 0
10 10 10 10 10 10
rmax rmax

Figure 2. VHSCF evaluated at the same temperature T ¼ 0.48 for both (a) confined and
(b) bulk systems. The time range spanned by the curves is 2  102 4 t 4 1  105 for
the bulk and 2  102 4 t 4 2  104 for the confined systems. (c), (d) The positions of
the corresponding peaks as functions of time in a log–log plot.
Slow dynamics of a confined supercooled binary mixture 1401

two systems are reported for T ¼ 0.48 for both bulk and confined systems. We see
that, while for the bulk system the phenomenon of the clustering of the curves at
intermediate times is already well visible, for the confined system it is much less
marked. This is best evidenced by the position of the peaks as a function of space
and time, also shown in figure 2. The curve clustering is connected with the rattling
of the particles in the transient caging of the nearest neighbours.
We note that the clustering phenomenon is partially masked in confinement by
the existence of important hopping processes already at high temperatures. These
processes start to transfer part of the intensity of the main peak of the VHSCF to
secondary hopping peaks (not shown) while the main peak is still in the clustering
region (Gallo et al. 2003).
From the analysis of the VHSCF we can deduce that at the same values of
pressure and temperature and roughly the same local density the confined system
moves much more rapidly than the bulk and its dynamics, consistent with the caging
foreseen by the MCT, and is also characterized by important hopping effects.
An analysis of the self-intermediate scattering function (SISF), that is the space
Fourier transform of the VHSCF, has also been carried out for both systems as a
function of temperature. The SISF is expected to follow, for the time window cor-
responding to the breaking of the cage, the Kolrausch–William–Watts stretched-
exponential function
   
t
fQ ðtÞ ¼ fQ exp  ; ð2Þ

where  is the relaxation time,  the stretching exponent and fQ the non-ergodicity
parameter.
In figure 3 we report the parameters extracted from the fit to the stretched-
exponential function for both A and B particle correlators in bulk and confined
systems at the peaks of the structure factor, where the MCT features are best evident.
The positions of the peaks (not shown) almost coincide for bulk and confined
systems and they do not substantially depend on temperature. For A and B particles
they are QMAX  7:06 and QMAX  5:90 respectively.
As expected from the analysis of the VHSCF, we see that relaxation times
increase dramatically upon decreasing temperature for both systems and that the
values for the confined system are significantly lower than the those of the corre-
sponding bulk at the same temperature. Similar trends are also observed for the non-
ergodicity parameters. This last quantity depends only mildly on temperature for
both systems. While both  and fQ behave consistently with the MCT, the stretching
parameter shows a different behaviour in confinement. The characteristic flattening
of the values of  on approaching Tc, generally observed in supercooled bulk liquids,
is not found in our confined system. This could possibly be connected to hopping
and to the related issue of dynamic heterogeneities that can in turn determine the
stretching of the exponential function. In order to clarify this point, further addi-
tional studies are needed and are now in progress.
The MCT predicts a power-law behaviour for the relaxation time
 1 / ðT  Tc Þ / ðc  Þ : ð3Þ
From these equations both the crossover temperature and the crossover density of
the theory can be estimated together with the parameter . The values obtained for
both species in the bulk and confined systems are reported in table 2. From the
1402 P. Gallo et al.
5
10
a)

τA(T), τB(T)
4 A part.
10 Bulk B part.
3
10 A part.
Conf. B part.
2
10
1
10
1 b)
βA(T), βB(T)

0.8
0.6
0.4

0.85 c)
fQ (T), fQ (T)

0.8
B

0.75
0.7
A

0.65
0.3 0.4 0.5 0.6 0.7 0.8 0.9
T
Figure 3. Parameters extracted from the fit to the stretched exponential of the SISFs of A
(f, g) and B (*, œ) particles at the peak of the structure factor as functions of
temperature for bulk (f, *) and confined (g, œ) systems: (a) relaxation times ;
(b) stretching exponents ; (c) non-ergodicity parameters fQ.

Table 2. Critical exponents and parameters of MCT for A and B particles of bulk and
confined LJBM. Results are from equations (3), (4) and (5).
Bulk mixture Confined mixture
A B A B
Tc 0.445 0.445 0.356 0.356
c 1.228 1.228 — —
2.849 2.987 2.8 2.8
fQc 0.81 0.81 0.72 0.80
hQ 0.52 0.53 0.35 0.43
b 0.45 0.531 0.355 0.34
0.816 0.767 0.871 0.880

values reported we note that the prediction of the MCT that the parameters should
be the same for both A and B particles is sufficiently satisfied by our systems. The
main finding is that the glass transition temperature is shifted downward by 20%
with respect to the bulk value. The parameter is less affected by the presence of the
soft spheres instead. We note in passing that the lowest temperatures that could be
Slow dynamics of a confined supercooled binary mixture 1403

reached in bulk and confined systems correspond approximatively to the value of the
small parameter of the theory:  ¼ ðT  Tc Þ=Tc  0:05.
One of the most significant tests of the MCT is that the SISF obeys the time
temperature superposition principle in the time region of the caging of the particle.
This means that, once the SISF curves at different temperatures are rescaled by a
characteristic time of the system, they must all fall on to the same master curve
that is to be fitted by the following analytic expression also known as the von
Schweidler law
 t b
GðtÞ ¼ fQc  hQ : ð4Þ

In figure 4 the SISFs are reported versus t=. The  values are those extracted from
the stretched-exponential fit (see figure 3). The master curve is evident for both the
bulk and the confined systems and in both cases it follows the von Schweidler law.
The values of the parameter extracted from the fit are also reported in table 2. The
anomalous exponent b is lower in confinement. From b the system-specific exponent
parameter l can also be extracted
Gð1  bÞ2
l¼ : ð5Þ
Gð1 þ 2bÞ
Values of l are reported in table 2. An increase in l is evident upon confinement.

-5 -4 -3 -2 -1 0 1
10 10 10 10 10 10 10
1.0 a)
T = 0.8
0.8
T = 0.48
0.6
0.4
Bulk
0.2 QMAX = 7.06
(QMAX,t)

0.0 von Schweidler fit

1.0 b)
A
(s)
F

0.8 T = 0.48
0.6
T = 0.37
0.4
Confined
0.2 QMAX = 7.06

0.0 von Schweidler fit


-5 -4 -3 -2 -1 0 1
10 10 10 10 10 10 10
t/τα

Figure 4. Master plot of the SISF of A particles for (a) bulk and (b) confined systems at the
peak of the structure factor QMAX for A particles, where each curve, corresponding to
a different temperature, is rescaled by the relaxation time  reported in figure 3: (– – –),
fitted to the von Schweidler law. The range of validity of the master plot is
0:48 4 T 4 0:8 for the bulk LJBM and 0:38 4 T 4 0:48 for the confined LJBM.
The fit parameters are reported in table 2.
1404 Slow dynamics of a confined supercooled binary mixture

We note that the range of validity of the von Schweidler law is much more
limited for the confined system. At the same  both SISFs display similar plateaux;
therefore both systems appear to approach Tc in similar fashions but for the confined
system the MCT features show up in a more limited range of temperatures:
T ¼ 0:11. For the bulk LJBM the range for which the von Schweidler law is
evident is T ¼ 0:32, that is three times larger.

} 4. Concluding remarks
We reported the results achieved in the computer simulation of the single-particle
dynamics of a LJBM in the bulk phase and confined in an off-lattice matrix of soft
spheres upon supercooling. The confining medium is modelled to reproduce the main
features of silica xerogels. Comparison has been performed between the two systems
at the same state points. The LJBM reproduces the MCT features upon this kind of
confinement. We note that the same mixture in a different confining medium does
not reproduce the MCT features (Scheidler et al. 2000, 2002).
Our MCT system specific parameter change for the LJBM upon confinement, in
particular the values of , and b, decrease. Correspondingly, l increases. In the
confined system the crossover temperature Tc is shifted downwards, reflecting an
acceleration of dynamics. Phenomenological arguments based on the existence of
domains of cooperative dynamics of growing length upon supercooling are consis-
tent with a decrease in Tc in confinement (Adam and Gibbs 1965).
The fact that the range of validity of the MCT in confinement is more limited
might pose an important restriction on the experimental possibility of detecting
signatures of MCT behaviour. Simulations as a function of the size of the soft
spheres are now in progress to further investigate the modifications of the MCT
properties upon confinement.

References
Adam, G., and Gibbs, J., 1965, J. chem. Phys., 43, 139.
Attili, A., Gallo, P., and Rovere, M., 2003 (to be published).
Frick, B., Zorn, R., and Bu« ttner, H., (editors), 2000, J. Phys., Paris, IV, 10.
Gallo, P., and Rovere, M., 2003, J. Phys.: condens. Matter, 15, 1521.
Gallo, P., Pellarin, R., and Rovere, M., 2002, Europhys. Lett., 57, 212; 2003, Phys. Rev. E
(to be published).
Gallo, P., Rovere, M., and Spohr, E., 2000a, Phys. Rev. Lett., 85, 4317; 2000b, J. chem.
Phys., 113, 11 324.
Go« tze, W., and Sjo« gren, L., 1992, Rep. Prog. Phys., 55, 241.
Kob, W., and Andersen, H. C., 1995a, Phys. Rev. E, 51, 4626; 1995b, ibid., 52, 4134.
Page, K. S., and Monson, P. A., 1996, Phys. Rev. E, 54, R29.
Scheidler, P., Kob, W., and Binder, K., 2000, Europhys. Lett., 52, 277; 2002, ibid., 59, 701.
Varnik, F., Baschnagel, J., and Binder, K., 2002, Phys. Rev. E, 65, 021 507.

You might also like