You are on page 1of 6

Eur. Phys. J.

E (2009)
DOI 10.1140/epje/i2008-10432-2

THE EUROPEAN
PHYSICAL JOURNAL E

Regular Article

Eective potentials of dissipative hard spheres in granular matter


R.A. Bordallo-Favela1 , A. Ramrez-Sato1 , C.A. Pacheco-Molina1 , J.A. Perera-Burgos1,2 , Y. Nahmad-Molinari1 , and
G. Perez2,a
1

Instituto de Fsica, Universidad Aut


onoma de San Luis Potos, Av. Manuel Nava 6, Zona Universitaria, 78290 San Luis Potos,
San Luis Potos, Mexico
Departamento de Fsica Aplicada, CINVESTAV del IPN, A.P. 73 Cordemex, 97310 Merida, Yucat
an, Mexico
Received 5 September 2008 and Received in nal form 5 December 2008
c EDP Sciences / Societ`

a Italiana di Fisica / Springer-Verlag 2009
Abstract. We present an experimental study of the spatial correlations of a quasi-two-dimensional dissipative gas kept in a non-static steady state via vertical shaking. From high temporal resolution images
we obtain the Pair Distribution Function (PDF) for granular species with dierent restitution coecients.
Eective potentials for the interparticle interaction are extracted using the Ornstein-Zernike equation with
the Percus-Yevick closure. From both the PDFs and the corresponding eective potentials, we nd a clear
increase of the spatial correlation at contact with the decreasing values of the restitution coecient.
PACS. 45.70.-n Granular systems 61.20.Ne Structure of simple liquids

1 Introduction
Granular materials under external excitation resemble and
have been used as model systems for atomic, molecular,
and colloidal ensembles in thermal equilibrium [1,2]. However, in granular matter we cannot talk about macroscopic
irreversibility coming from microscopic reversibility since
a granular material has intrinsically dissipative interactions even at its minimal scale (one single grain). The dissipative nature of granular matter means that non-static
assemblies need a permanent injection of energy, and this
energy ow in the system gives rise in turn to the appearance of some types of self-organization, allowing for some
very interesting phenomena such as oscillons and other
wide variety of beautiful patterns, as elegantly shown by
Swinney and coworkers [3].
On the other hand, condensed phases resembling liquids or solids the latter being in crystalline or glassy
form have been reported in granular materials under
shaking excitation. The evolution of these out-of-equilibrium systems towards a steady state with stationary
structures have been described [4], and its jammed-like
system relaxation dynamics has been reported [5]. In particular, crystallization, in the form of a quasi-static phase
with hexagonal order (and, at least in one instance, square
order [6]), has been obtained in two dierent ways: either
by increasing densities or by reducing the amplitude of
the driving. In the rst case the process is basically equal
to the crystallization of hard disks in equilibrium [7,8]:
a

e-mail: gperez@mda.cinvestav.mx

as the granular density is increased the aggregate displays


a transition from uid to hexatic phases, and a posterior
transition to the (hexagonal) crystalline phase. This seems
to be then an instance of the maximization-of-entropy process that originates the crystallization of hard spheres and
disks [9,10] although in the case of ref. [7] a slightly concave lower plate helps the condensation process. On the
other hand, crystallization has been also observed for values of 1 [11,12], but at very low packing fractions
( = 0.42 in the experiment vs. 0.71 for equilibrium
crystallization in a gas of hard disks. Particularly illuminating here are Figs. 1(e) and (f) of Ref. [11] and Fig. 1
of Ref. [12]). There is no clear understanding of the condensation mechanisms underneath the appearance of this
particular crystalline phase, since the low density of the
granulate precludes the standard entropy-driven mechanisms. This crystallization is usually described in terms of
the inelastic collapse phenomenon [13], but the actual process of their formation is not yet experimentally well measured. Therefore, there may be some alternative or complementary mechanisms that need to be considered for this
condensation phenomenon; for instance, it has been suggested that the Prigogines minimum entropy production
rate principle is the origin of the self-organization shown
in the epitaxial granular crystallization process [14].
The techniques and theoretical frameworks developed
for colloidal particles are being translated and applied for
granular materials since almost ten years (torsion balances, light scattering, etc.) [2,15,16]. Among the theoretical tools developed for colloids, a fundamental one has
been the development, starting from more than 60 years

The European Physical Journal E

ago, of an interparticle potential able to describe the interaction between colloidal particles. This interaction is
actually mediated by the solvent and the possible presence of other molecules at several size scales. Part of
this eort is the Dejaurmin, Landau, Vervey and Overvek
(DLVO) [1719] theory. Here, the two main contributions
taken into account are the van der Walls forces and the
pure Coulombic electrostatic forces. Besides these energybased interactions, the presence of intermediate-size elements in a colloidal dispersion elements smaller than
the colloidal particles themselves, but much larger than
the solvent molecules gives origin to subtle entropic effects, known as depletion interactions; these were modeled originally by Asakura and Oosawa [20]. Furthermore,
during the last decade hydrodynamic eects have been
tracked experimentally [21], but it is yet not clear how to
represent them as a new term in the whole picture of the
colloid-colloid interaction.
This article presents an approach closely related to
the DLVO and excluded-volume-forces theories, since it
intends to give a rst step towards the construction of
a granular eective interaction potential suitable for describing the appearance of dierent granular phases in a
steady-state driven granular system. Given the lack of a
comprehensive theoretical framework in which a system
made of many dissipative particles kept in a non-static
steady state via the cyclic injection of energy could be
treated satisfactorily, we try to draft from a phenomenological approach a way to describe the elusive mechanism
by means of which a driven granular system forms condensed phases, treating the system as if it were in thermodynamic equilibrium and extracting from its Pair Distribution Function (PDF) an eective pair interaction potential. In this scenario, it then becomes natural to attribute
the condensation itself to the existence of this eective pair
potential, and this could be a rst step in the process of
predicting under which excitation conditions frequency,
amplitude, and type of excitation and which geometrical and particle parameters lling fraction, restitution
and friction coecients, etc. a condensed phase shall
start to grow.

2 Experimental procedure
The experimental setup is similar to the one used by Olafsen and Urbach [11,22]. We use ve sets of uniform monodisperse spheres made of dierent materials, with diameters ranging from 4.0 to 5.8 mm, chosen in order to
get dierent restitution coecients. These were obtained
letting a sphere fall on top of a plate of the same material. The normal velocities just before and just after the
collision were measured, using a high-speed camera. The
data for diameters and restitution coecients are given
in Table 1. Spheres of one given kind are placed between
two horizontal hexagonal Plexiglas plates, with the side
of the hexagon equal to 10 cm. The distance between the
two plates is set in all cases to 1.6 times the diameter of
the spheres inside. To avoid condensation of the granulate
the lower plate of the cell was made non-at by means of

Table 1. Diameters and restitution coecients e of the granular species used in the experiment.
Species

(mm)

Lead

4.3

0.25

Painted steel

4.4

0.62

Painted glass

4.0

0.69

Steel

4.4

0.75

Plastic

5.8

0.81

Fig. 1. Top view of a section of the lower plate in the experimental set-up. The right upper corner shows an oblique view
(with a relative enlargement factor of 3) of a few bumps.

small protrusions that give a horizontal momentum component to the spheres each time they are compressed by
the shaking cycle against this plate, assuming they are
within the radius of inuence of a protrusion. These bumps
are made touching the Plexiglas with a hot metallic tip,
have roughly the form of an elliptic crater, and have approximate lateral size and height of 1.5 mm and 0.3 mm,
respectively. There are on average close to 16 protrusions
per cm2 of this type in the lower plate, and their distribution is quasi-random, such that the distance between any
of them and its closest neighbor is around 1 to 3 times the
typical bump size (see Fig. 1).
Cells with non-at bottoms have been used in previous experimental set-ups, and their main purpose is to
inject transverse momentum to the granulate. Some very
interesting results have emerged from these experiments.
In particular, Prevost et al. [23] have shown that the correlation between velocity components parallel to the line
between particle centers is strongly negative for cells with
at bottoms, but becomes positive for a cell where smaller
spheres are glued on its lower plate. This means that,
for a at bottom, spheres that undergo a collision have
an average relative velocity larger after the collision than
before. (This is a consequence of the fact that, on average, dissipative collisions with a at plate that does not
move horizontally reduce the horizontal linear momentum
of a grain. Maintaining a stationary situation requires then
that grain-grain collisions increase horizontal momentum).
The opposite phenomenon happens over a rough bottom.

R.A. Bordallo-Favela et al.: Eective potentials of dissipative hard spheres in granular matter

Also, as mentioned in the introduction, Reis et al. [8,24]


nd that, in the presence of a rough bottom (sand-blasted
glass), the density-driven transition to condensation has
essentially the same characteristics of the solidication as
a hard disk system. Finally, a rough moving bottom has
been implemented in one study, laying a close-packed layer
of heavy dimers over the lower plate, with a gas of lighter
spheres on top of this rst layer, and keeping small
enough so that there is no layer mixing [25,26]. Here one
can get a gas that shows molecular chaos, with velocity
distributions that approach the Maxwell-Boltzmann form.
Returning to the present work, the cell is sinusoidally
vibrated in the vertical direction at a frequency f of 60 Hz
by means of a loudspeaker. The working frequency was
chosen at this value to avoid nodes and resonances in the
system that could lead to an inhomogeneous distribution
of particles through the plate. In this regime, a homogeneous stationary distribution of particles appears together
with uniform measurements of the cell acceleration. The
amplitude of oscillations A for the cell was of 0.5 mm,
and kept constant for all the experiments, getting in this
way an adimensional acceleration = (2f )2 A/g = 7.25.
The number of spheres is adjusted so that the 2D packing fraction = N 2 /4S is given by = 0.35 in almost
all experiments and for all tested materials. Here N is
the number of spheres in the system, their diameter,
and S the area of the experimental cell. In this way we
avoid shifts in the measured PDF due to packing fraction
eects. Still, some extra experiments were performed at
= 0.15. Care was taken as well of maintaining the ratio
between linear dimensions of the cell and particle diameters always larger than 50, and of keeping the exploration
region (the area digitally recorded in pictures) far away
from the walls (more than 20 diameters away), and large
enough to allow a fair representation of the main features
of the correlation function for this packing fraction regime.
A high-speed camera (Red-Lake Motion-Meter) operating
at 500 frames per second was placed above the system,
and the cell was obliquely illuminated with diuse light.
In this way, the spheres images were recorded just from
the bright spot shining at their centers, making it easier
for the particle center localization algorithm to process
the frames.
Once under vibration, the system of particles resembled a conned colloidal system driven by Brownian motion [27], since the motion of the spheres in the plane is
constantly aected by their collisions with the scatterers
introduced in the lower plate. In the above experimental
conditions the granular uid neither presented condensed
phases nor stable clusters for any of the tested materials, and there is no evidence of trapping of the spheres
by the bottom plate scatterers. Ten thousand frames per
experiment are analyzed for center particle location using a software build on the IDL platform and provided by
the UASLP-Complex Fluids Laboratory; further calculation of the PDF was performed on the same data analysis
program. Under these experimental conditions, granular
uids for the ve dierent kinds of particles already mentioned were analyzed and their PDFs obtained are shown
in Figure 2.

Fig. 2. Pair distribution functions at a 2D packing fraction


= 0.35, obtained for ve materials with dierent restitution
coecients. The gure shows a very large correlation at contact
for lead beads. The points are experimental data, the lines are
cubic splines interpolations.

3 Experimental results: pair distribution


functions and eective potentials
In equilibrium, the PDF is determined along with temperature and density by the interparticle potential, but
the relationship between these two quantities is not easily
stated. In principle one can use the Ornstein-Zernike (OZ)
equation

h(r12 ) = c(r12 ) + d3 r3 c(r13 ) h(r32 ),
where rij = |ri rj |, is the particle density, h(r) =
g(r) 1 is called the total correlation function, and c(r) is
the direct correlation function, a function that is expected
to have a simpler dependence on the potential than h(r),
and that is actually dened by the OZ equation itself [28].
The relation between h and c can be written in a much
simpler form using Fourier transform, where one nds

h(k)
=

c(k)
.
1 + c(k)

(1)

Still, in the absence of an independent relationship, this


equation serves only as a denition for c. For very diluted
systems [29] a simple approach is to take the direct correlation function equal to the Mayer function
c(r) f (r) = exp[u(r)/kB T ] 1,
and, in the limit of high T , one may simply take the rst
term of the expansion
c(r) u(r)/kB T.
On the other hand, for hard spheres and, in general,
short-range potentials, in either dense or diluted situations a more convenient closure relationship is given by

The European Physical Journal E

Fig. 3. Eective interaction potentials (divided by kB T ) obtained by inverting the PDFs shown in Figure 1 by means of
the OZ equation, using the PY closure. The more attractive
potential corresponds to lead, which has the lowest restitution
coecient.

the Percus-Yevick (PY) approximation [30]


c(r) = [h(r) + 1][1 exp(u(r)/kB T )].

(2)

With this approximation, one can then generate an approximation to the quotient u(r)/kB T , once the g(r) and

are given. This requires the evaluation of h(k),


which in
2D involves the zeroth-order Hankel transform of g(r) 1,
solving (1) for c(k), doing the inverse Hankel transform to
get c(r), and nally solving (2) to get u/kB T .
PDFs are often measured for granular gases in
a stationary state [2,11,22,31], being these in out-ofequilibrium but stationary ensembles. So, it then becomes
an interesting proposal to associate an eective potential
with these structural characteristics. In these cases the
temperature dependence can be either ignored (more exactly, it gets incorporated in the eective potential), or
some convenient denition of T as proportional to the average kinetic energy is implemented. Following this idea,
we have extracted the product ue from the experimentally obtained g(r), a product that from now on we will refer to simply as eective potential; the results are shown
in Figure 3.
Two features are striking on this gure: rst, the very
emergence of eective potentials with just one well-dened
well, and second, the clear assignment of a deeper well
to the softest material, in this case lead. In general the
depth of these wells in ue increases as the restitution
coecient lowers its value. One can conclude then that
smaller restitution coecients softer materials give
rise to stronger (attractive) eective potentials.
One should be aware, however, of some limitations in
the experiments: to begin with, the g(r) curves have continuous derivative at r/ = 1, and look as if coming from
a soft potential, due to the occasional partial overlap of
the spheres in the experiment. This is because the height
of the cell is larger than the spheres diameter. Also, due
to experimental constrains the grains used were in some

Fig. 4. Eective interaction potentials for steel and lead


beads at a packing fraction = 0.15, showing attractive and
hard-sphere behaviors.

cases of dierent diameters (see Tab. 1), and this aects


the interaction with the non-at bottom plate (the same
bottom plate was used in all trials). Still, a clear eect
of the changing restitution coecient is evident in the
graphs.
A clear resemblance of the potentials with those corresponding to a simple liquid and a hard-sphere uid, obtained for lead and steel spheres, respectively, in a more
diluted system (packing fraction = 0.15), is presented in
Figure 4. Here the well depth for the steel spheres assembly is much shallower (less attractive) than in the case of
the denser system depicted in Figure 3, and the eective
potential found seems to be due exclusively to excludedvolume eects. The results for the denser system show that
the measured eective potentials account for dissipative,
depletion and, very likely, for the base plate roughness
eects.
With respect to this last point, there is a possibility
that the observed correlations in position, which give rise
to the proposed eective potentials, may emerge not so
much from the dissipative nature of the collisions and
from entropic eects, but from temporal gravity-assisted
trapping of the spheres an the valleys that form between
bumps in the lower plate. We believe that this trapping
eect represents a very minor, maybe even null, contribution to the dynamics of the system, mostly because, at the
value of used here, the grains can almost be considered
to be moving freely between collisions. That is, the impulse received from the contacts with the horizontal plates
is so large as to make gravity into a minor correction.
The particles y vertically and hit both horizontal plates
with similar momentum, making very unlikely for them to
sit in any given valley. The collision with the lower plate
is for all purposes instantaneous, and the grains cannot
really explore the landscape at the bottom. Notice also
that with a plate separation between 6.9 and 9.3 mm the
bumps at their maximum height ( 0.3 mm) represent
only a 3-4% reduction in the vertical space, meaning that
the grains (with diameters from 4.3 to 5.8 mm) have more

R.A. Bordallo-Favela et al.: Eective potentials of dissipative hard spheres in granular matter

4 Discussion

Fig. 5. Correlation at contact (square marks, left vertical


axis), or well depth (triangular marks, right vertical axis) vs.
restitution coecient of the grains. The lines are provided only
as rough indicators of the behavior. With the exception of one
point, both quantities evolve monotonously. From left to right,
the materials are: lead, painted steel, painted glass, steel, and
plastic.

than enough free space between the plates to move around.


Finally, it is worth restating that Figure 4 indicates clearly
that the restitution coecient is the fundamental parameter that originates non-trivial eective potentials.
Another important feature shown in Figure 4 is a shift
to the right of the potential minimum which appears at
r/ = 1.33 for lead spheres at this lower packing fraction of = 0.15, in comparison with its corresponding
denser case of = 0.35 in which the minimum is located at
r/ = 1.076. This shift is easily explained in terms of the
reduction (increment) in the area available for vagrancy as
the packing fraction is increased (reduced). As the packing fraction is increased the random walks performed by
the particles decrease their mean square displacement and
temporary cages develop while particles become increasingly trapped by their neighbors [5], shifting the rst peak
of the PDF closer to the contact position.
A noticeable increase in the height of the rst peak in
the PDF (Fig. 2) and also in the well depth (Fig. 3) is developed as the restitution coecient decreases. In Figure 5
the rst peak height of the PDF or the well depth of the
corresponding potential are plotted as a function of the
restitution coecient, showing for both quantities with
exception of a single point a monotonous dependence.
These monotonous behaviors can be expected intuitively
since dissipation of energy during each collision event is
an extra factor that should be taken into account in determining not only the relative velocity [23] (which shows
asymmetric correlations), but the position between colliding particles as well. Moreover, dissipation slows down
the entire dynamics of the system allowing for this hightemporal-resolution testing of the granular uid structure
to capture more information of correlated positions before
a new uncorrelating event occur (next shaking cycle).

This paper investigates the structural changes found in


stationary-state granular gas changes with respect to an
ordinary ideal gas in terms of a possible eective interaction potential. This is accomplished within the framework of integral equations, using the PY closure, which is
the most appropriate for very short-range potentials. This
closure relation has been widely used for extracting the effective pair potential in colloids and macromolecules, but
up to our knowledge, it has never been used to get an eective attraction in a granular gases, where the constituent
elements are dissipative hard spheres. As shown in Figures 2 and 3, there is an increase in the pair distribution
at contact and a consequent appearance of a potential
well close to the grains, a well whose depth increases as
the associated restitution coecient is lowered. This correlation increment signals a more ecient clustering for
more dissipative ensembles. The analysis of the PDFs in
terms of eective potentials was partially inspired by the
emergence of condensation (crystallization) in a moderately dense granular gas. However, the data accumulated
up to now is clearly insucient for an actual prediction
of the condensation transition. At the moment we have
only shown that eective potentials with attractive wells
can be extracted from the observed PDFs. Future work
involving the analysis of the eective potential as a function of and is needed to connect to the crystallization
transition.
The system described here is not a simply cooling one,
since energy is fed cyclically via vertical shaking. So the
stationary state corresponds to a situation where web-like
patterns and their constituent clusters start to form between consecutive excitation impulses, but are destroyed
(randomized) with each shaking event. The granular gas
cools down in a characteristic time (relaxation time) that
depends on the particle restitution coecient as well as
on the number of collisions per unit time, and so we will
be more likely to capture cluster features for gases with
smaller relaxation times. The resulting preference of more
dissipative particles to form clusters or to stay at closer
and well determined distances can be interpreted as if
the particle were within a potential well. In fact, if one
takes a pair of particles in a bidisperse hard-sphere system, pairing and cluster formation appear as a result of
the momentum transfer (pressure) exerted by the small
surrounding particles, and this eect is interpreted as coming from a potential (the Asakura-Oosawa potential). This
potential comes actually from the tracing out of the kinematics of the small particles, and therefore disappears for
monodisperse systems (there is nothing to be traced out.)
For dissipative collisions clustering appears even in the
monodisperse case, giving as a result a new type of eective potential. It now comes from the time averaging of
the injection-dissipation cycle in energy. It should be understood that in an actual experimental situation say,
the one described here the eective potential will include not only dissipative eects, but also any inuence
the rough bottom may have. This eect, as argued before,
is expected to be small.

The European Physical Journal E

Most of the research eort done in quasi-2D granular


gases has been oriented towards determining the character of the velocity distribution in order to extend the kinetic theory of gases to these systems. The analysis of the
structure by means of spatial correlations and its dependence on the restitution coecient is quite independent
of these velocity distributions, as shown by van Zon and
McKintosh [32] who state that spatial correlations play
a minor role, if any, in P (v) (P (v) being here the velocity distribution function). The independence of these
two characteristics implies that a more complete description of a granular gas requires also the analysis of spatial
structures and correlations.
As a nal comment, some recent experiments in granular ows are worth noticing, in particular the formation
of a granular water-bell when a granular jet impinges on
a at surface [33], the formation of drops in a falling jet
of grains both in air and in vacuum [34] and the presence of capillary waves in a thick falling ow of grains [35],
seem to imply the presence of a surface tension. The idea
of an eective potential in aggregates of granular matter
can be perhaps used as a principle on which to build these
eective surface tensions.
We thank nancial support from PROMEP under grants
UASLP-PTC-084 and P/CA-23 2007-24-25. We acknowledge
the invaluable support given by Dr. Jose Luis Arauz and his
group at the Complex Fluids Laboratory in UASLP.

References
1. H.M. Jaeger, S.R. Nagel, R.P. Behringer, Rev. Mod. Phys.
68, 1259 (1996).
2. B.V.R. Tata, P.V. Rajamani, J. Chakrabarti, A. Nikolov,
D.T. Wasan, Phys. Rev. Lett. 84, 3626 (2000).
3. C. Bizon, M.D. Shattuck, J.B. Swift, W.D. McCormick,
H.L. Swinney, Phys. Rev. Lett. 80, 57 (1998).
4. B. Knight, E.R. Nowak, E. Ben-Naim, H.M. Jaeger, S.R.
Nagel, Phys. Rev. E 57, 1971 (1998).
5. P.M. Reis, R.A. Ingale, M.D. Shattuck, Phys. Rev. Lett.
98, 188301 (2007).
6. F. Vega-Reyes, J.S. Urbach, arXiv:0803.1158v1 (2008).
7. J.S. Olafsen, J.S. Urbach, Phys. Rev. Lett. 95, 098002
(2005).
8. P.M. Reis, R.A. Ingale, M.D. Shattuck, Phys. Rev. Lett.
96, 258001 (2006).

9. B.J. Alder, T.E. Wainwright, J. Chem. Phys. 27, 1208


(1957).
10. B.J. Alder, T.E. Wainwright, Phys. Rev. 127, 359 (1962).
11. J.S. Olafsen, J.S. Urbach, Phys. Rev. Lett. 81, 4369 (1998).
12. X. Nie, E. Ben-Naim, S.Y. Chen, Europhys. Lett. 51, 679
(2000).
13. S. McNamara, W.R. Young, Phys. Rev. E 53, 5089 (1996).
14. Y. Nahmad-Molinari, J.C. Ruiz-Surez, Phys. Rev. Lett.
89, 264302 (2002).
15. P.K. Dixon, D.J. Durian, Phys. Rev. Lett. 90, 184302
(2003).
16. G. DAnna, P. Mayor, V. Loreto, F. Nori, Nature 424, 909
(2003).
17. B.V. Derjaguin, L. Landau, Acta Physicochim. (URSS)
14, 633 (1941).
18. J.E. Verwey, J.T.G. Overbeek, Theory of the Stability of
Lyophobic Colloids (Elsevier, Amsterdam, 1948).
19. P.C. Hiemenz, M. Rajagopalan, Principles of Colloid and
Surface Chemestry, 3rd edition (Marcel Dekker, New York,
1997).
20. S. Asakura, F. Oosawa, J. Chem. Phys. 22, 1255 (1954).
21. A.E. Larsen, D.G. Grier, Nature 385, 230 (1997).
22. J.S. Olafsen, J.S. Urbach, Phys. Rev. E 60, R2468 (1999).
23. A. Prevost, D.A. Egolf, J.S. Urbach, Phys. Rev. Lett. 89,
084301 (2002).
24. P.M. Reis, R.A. Ingale, M.D. Shattuck, Phys. Rev. E 75,
051311 (2007).
25. G.W. Baxter, J.S. Olafsen, Nature 425, 680 (2003).
26. G.W. Baxter, J.S. Olafsen, Phys. Rev. Lett. 99, 028001
(2007).
27. A. Ramrez-Saito, M. Ch
avez-P
aez, J. Santana-Solano,
J.L. Arauz-Lara, Phys. Rev. E 67, 050403 (2003).
28. M.E. Fisher, J. Math. Phys. 5, 944 (1964).
29. L. Verlet, Phys. Rev. 165, 201 (1968).
30. The PY and other closures for the OZ equation are well
covered in many textbooks, see for instance: D.A. McQuarrie, Statistical Mechanics (University Science Books,
Sausalito, 2000); D.L. Goodstein, States of Matter (Dover
Publications, New York, 1985).
31. A. Burdeau, P. Viot, arXiv:0710.3713v1 (2007).
32. J.S. van Zon, F.C. MacKintosh, Phys. Rev. Lett. 93,
038001 (2004).
33. X. Cheng, G. Varas, D. Citron, H.M. Jaeger, S.R. Nagel,
Phys. Rev. Lett. 99, 188001 (2007).
34. M.E. M
obius, Phys. Rev. E 74, 051304 (2006).
35. Y. Amarouchenei, J.-F. Boudet, H. Kellay, Phys. Rev.
Lett. 100, 218001 (2008).

You might also like