Granular solids, liquids, and gases

Heinrich M. Jaeger and Sidney R. Nagel

James Franck Institute, The University of Chicago, 5640 S. Ellis Avenue, Chicago,
Illinois 60637

Robert P. Behringer
Department of Physics, Duke University, Durham, North Carolina 27706

Granular materials are ubiquitous in the world around us. They have properties that are different
from those commonly associated with either solids, liquids, or gases. In this review the authors select
some of the special properties of granular materials and describe recent research developments.

[S0034-6861(96)00204-8]

CONTENTS repulsive so that the shape of the material is determined

by external boundaries and gravity. If the grains are dry,
I. Introduction 1259 any interstitial fluid, such as air, can often be neglected
II. An Unusual Solid: Sand at Rest 1261 in determining many, but not all, of the flow and static
III. An Unusual Liquid: Granular Hydrodynamics 1263 properties of the system. Yet despite this seeming sim-
IV. An Unusual Gas: Inelasticity, Clustering, Collapse 1268 plicity, a granular material behaves differently from any
V. Conclusions and Outlook 1270
of the other familiar forms of matter—solids, liquids, or
Acknowledgments 1270
References 1271
gases—and should therefore be considered an additional
state of matter in its own right.
‘‘Who could ever calculate the path of a molecule? In this article, we shall examine in turn the unusual
How do we know that the creations of worlds are not behavior that granular material displays when it is con-
determined by falling grains of sand?’’ (Victor Hugo, Les sidered to be a solid, liquid, or gas. For example, a sand
Miserables). pile at rest with a slope lower than the angle of repose,
as in Fig. 1(a), behaves like a solid: the material remains
I. INTRODUCTION at rest even though gravitational forces create macro-
scopic stresses on its surface. If the pile is tilted several
Victor Hugo suggested the possibility that patterns degrees above the angle of repose, grains start to flow,
created by the movement of grains of sand are in no as seen in Fig. 1(b). However, this flow is clearly not that
small part responsible for the shape and feel of the natu- of an ordinary fluid because it only exists in a boundary
ral world in which we live. No one can seriously doubt layer at the pile’s surface with no movement in the bulk
that granular materials, of which sand is but one ex- at all. (Slurries, where grains are mixed with a liquid,
ample, are ubiquitous in our daily lives. They play an have a phenomenology equally complex as the dry pow-
important role in many of our industries, such as mining, ders we shall describe in this article.)
agriculture, and construction. They clearly are also im- There are two particularly important aspects that con-
portant for geological processes where landslides, ero- tribute to the unique properties of granular materials:
sion, and, on a related but much larger scale, plate tec- ordinary temperature plays no role, and the interactions
tonics determine much of the morphology of Earth. between grains are dissipative because of static friction
Practically everything that we eat started out in a granu- and the inelasticity of collisions. We might at first be
lar form, and all the clutter on our desks is often so close tempted to view any granular flow as that of a dense gas
to the angle of repose that a chance perturbation will since gases, too, consist of discrete particles with negli-
create an avalanche onto the floor. Moreover, Hugo gible cohesive forces between them. In contrast to ordi-
hinted at the extreme sensitivity of the macroscopic nary gases, however, the energy scale k B T is insignifi-
world to the precise motion or packing of the individual cant here. The relevant energy scale is the potential
grains. We may nevertheless think that he has over- energy mgd of a grain of mass m raised by its own di-
stepped the bounds of common sense when he related ameter d in the Earth’s gravity g. For typical sand, this
the creation of worlds to the movement of simple grains energy is at least 1012 times k B T at room temperature.
of sand. By the end of this article, we hope to have Because k B T is irrelevant, ordinary thermodynamic ar-
shown such an enormous richness and complexity to guments become useless. For example, many studies
granular motion that Hugo’s metaphor might no longer have shown (Williams, 1976; Rosato et al., 1987; Fan
appear farfetched and could have a literal meaning: et al., 1990; Jullien et al., 1992; Duran et al., 1993; Knight
what happens to a pile of sand on a table top is relevant et al., 1993; Savage, 1993; Zik et al., 1994; Hill and
to processes taking place on an astrophysical scale. Kakalios, 1994; Metcalfe et al., 1995) that vibrations or
Granular materials are simple: they are large con- rotations of a granular material will induce particles of
glomerations of discrete macroscopic particles. If they different sizes to separate into different regions of the
are noncohesive, then the forces between them are only container. Since there are no attractive forces between

Reviews of Modern Physics, Vol. 68, No. 4, October 1996 0034-6861/96/68(4)/1259(15)/$12.25 © 1996 The American Physical Society 1259
1260 H. M. Jaeger, S. R. Nagel, and R. P. Behringer: Granular solids, liquids, and gases

ture in ordinary gases or liquids is to provide a micro-

scopic velocity scale. Again, in granular materials this
role is completely suppressed, the only velocity scale is
imposed by any macroscopic flow itself. It is possible to
formulate an effective ‘‘granular temperature’’ in terms
of velocity fluctuations around the mean flow velocity
(Ogawa, 1978; Savage, 1984; Walton and Braun, 1986;
Haff, 1986; Campbell, 1990; Ippolito et al., 1995; Warr
and Huntley, 1995; Warr, Huntley, and Jacques, 1995).
Yet, as we shall see, such approaches do not always re-
cover thermodynamics or hydrodynamics because of the
inelastic nature of each granular collision.
The science of granular media has a long history with
much engineering literature devoted to understanding
how to deal with these materials. There are many no-
table names such as Coulomb (1773), who proposed the
ideas of static friction, Faraday (1831), who discovered
the convective instability in a vibrated powder, and Rey-
nolds (1885), who introduced the notion of dilatancy,
which implies that a compacted granular material must
expand in order for it to undergo any shear. Over the
last decade there has been a resurgence of interest in
this field within physics (for overviews see Jaeger and
Nagel, 1992; Behringer, 1993; Bideau and Dodds, 1991;
Bideau and Hansen, 1993; Jaeger et al., 1994; Mehta,
1994; Mehta and Barker, 1994; Behringer, 1995; Hay-
akawa, Nishimori, Sasa, and Taguchi, 1995). Sand piles
have become a fruitful metaphor for describing many
other, and often more microscopic, dissipative dynami-
cal systems. De Gennes (1966), for example, used sand
pile avalanches as a macroscopic picture for the motion
of flux lines in type-II superconductors. A particularly
powerful use of sand as a metaphor lay in the idea of
self-organized criticality (Bak et al., 1988), originally de-
scribed in terms of the avalanches in a sand pile close to
its angle of repose. The self-organization paradigm was
FIG. 1. (a) A pile of mustard seeds that is tilted to an angle postulated to have a wide realm of applicability to a
lower than the angle of repose. (b) The same pile after the variety of natural phenomena. In a similar vein, the
slope has been increased slightly to create an avalanche. Note physics that has been uncovered in granular materials
that the flow occurs only along the surface and that the seeds has clear relevance to what is being done in other areas
deeper within the pile do not participate in the motion. of condensed matter physics. Slow relaxations are found
in vibrated sand piles that bear close similarity to the
slow relaxation found in glasses, spin glasses and flux
the particles, this separation would at first appear to vio-
lattices (Jaeger et al., 1989; Duke et al., 1990; Bo-
late the increase of entropy principle, which normally
guslavskii and Drabkin, 1995; Knight et al., 1995). Fluid-
favors mixing (Zik et al., 1994). In a granular material,
like behavior can be induced in these materials, which
on the other hand, k B T;0 implies that entropy consid-
erations can easily be outweighed by dynamical effects very much resembles similar phenomena exhibited by
that now become of paramount importance. conventional liquids (Douady et al., 1989; Fauve et al.,
An important role of temperature is that it allows a 1989; Zik and Stavans, 1991; Melo et al., 1993; Pak and
system to explore phase space. In a granular material, Behringer, 1993; Jaeger et al., 1994; Pak and Behringer,
k B T;0 precludes such exploration. Unless perturbed 1994; Melo et al., 1995; Pak et al., 1995). Nonlinear dy-
by external disturbances, each metastable configuration namical phenomena are observed that are relevant to
of the material will last indefinitely, and no thermal av- breakdown phenomena in semiconductors (Clauss et al.,
eraging over nearby configurations will take place. Be- 1990), stick-slip friction on a microscopic scale (Reiter
cause each configuration has its unique properties, re- et al., 1994; Radjai et al., 1995), and earthquake dynam-
producibility of granular behavior, even on large scales ics on a macroscopic scale (Carlson et al., 1994).
and certainly near the static limit where friction is im- Despite this interest in granular science, the technol-
portant, is difficult to achieve. Another role of tempera- ogy for handling and controlling granular materials is

Rev. Mod. Phys., Vol. 68, No. 4, October 1996

 H. M. Jaeger, S. R. Nagel, and R. P. Behringer: Granular solids, liquids, and gases 1261

poorly developed. As mentioned above, many of our

industries rely on transporting and storing granular ma-
terials. These include the pharmaceutical industry that
relies on the processing of powders and pills, agriculture
and the food processing industry where seeds, grains,
and foodstuffs are transported and manipulated, as well
as all construction-based industries. Estimates are that
we waste 40% (Ennis et al., 1994; Knowlton et al., 1994)
of the capacity of many of our industrial plants because
of problems related to the transport of these materials.
Additional manufacturing processes, e.g., in the automo-
tive industry, rely on casting large metal parts in care-
fully packed beds of sand. Even a small improvement in
our understanding of granular media behavior could
have a profound impact on industry.
We turn now to examining some of the particular
properties of granular materials that appear under dif-
ferent conditions. The following three sections will ex-
plore their unique behavior, contrasting it to that of or-
dinary solids, liquids, and gases, respectively.

II. AN UNUSUAL SOLID: SAND AT REST

Already in the resting state, granular materials exhibit

a host of unusual behaviors. For example, when the
granular material is held in a tall cylindrical container,
such as a grain elevator or silo, no height-dependent
pressure head occurs as it does with a normal fluid: the
pressure at the base of the container does not increase
indefinitely as the height of the material inside it is in-
creased. Instead, for a sufficiently tall column, the pres-
sure reaches a maximum value independent of the
FIG. 2. (a) A visualization of the forces in a granular medium
height. Owing to contact forces between grains and as viewed between two crossed-circular polarizers. Three mm
static friction with the sides of the container, the con- Pyrex spheres are surrounded by a mixture of water and glyc-
tainer walls support the extra weight (Janssen, 1895). It erol that matches the index of refraction of the Pyrex. A force
is this feature that allows the sand in an hour glass to is exerted on a piston that covers the top surface of the con-
flow through the orifice at a nearly constant rate; a tainer. The stress-induced birefringence makes those beads
nearly linear change in filling height over time makes that are under stress visible as the bright regions. (b) The dis-
this a useful instrument to measure elapsed time. Under- tribution of forces P versus force f measured at the bottom of
lying this simple, time-averaged flow is, however, a com- a cylindrical container filled with spheres. The bottom of the
plicated dynamical behavior; see Gallas et al. (1993); Wu container was lined with carbon paper and a force was applied
et al. (1993); Pöschel (1994); Peng and Herrmann (1995); to the top surface. The line is an exponential fit to the data:
Horikawa et al. (1995); and Sec. III. P(f)5ce 2f/f 0 . After Liu et al. (1995).
We can investigate the network of forces within the
pile in greater detail. One example is shown in Fig. 2(a)
for a three-dimensional arrangement of particles. The the marks left by the forces f exerted by individual
forces within the pile appear to be very heterogeneous, beads. The distribution of forces P(f) is
forming chains along which the stresses are particularly P ~ f ! 5cexp~ 2f/f 0 ! , (1)
intense (Dantu, 1957; Wakabayashi, 1959; Drescher and
de Josselin de Jong, 1972; Ammi et al., 1987; Travers where c and f 0 are constants (Liu et al., 1995). The fluc-
et al., 1987). Shear experiments also attest to the ex- tuations in f are large and scale with depth in the same
treme heterogeneity of granular media. For example, way as the mean force, rather than as its square root as
when a layer of grains is sheared continually in a narrow one might have initially expected. Such behavior has
annulus, the normal stress measured at one of the been explained (Liu et al., 1995; Coppersmith et al.,
boundaries shows large-scale fluctuations with the rms 1996) in terms of a simple model in which masses placed
fluctuations comparable to the mean applied stress on a lattice distribute their weights unevenly and ran-
(Miller et al., 1996). It is not clear from Fig. 2(a) alone domly to the particles on the layer below them. This
what is the distribution of forces within the pile. This can model can be solved exactly in a number of different
be found by simply placing a piece of carbon paper on cases, yielding, in agreement with the experiment and
the bottom of the container and measuring the areas of simulation, an exponential distribution of large forces

Rev. Mod. Phys., Vol. 68, No. 4, October 1996

1262 H. M. Jaeger, S. R. Nagel, and R. P. Behringer: Granular solids, liquids, and gases

perturbation must be comparable to the wavelength.

This sensitivity may again be explained by the extreme
heterogeneity of the material: if the heater lies on or
near a force chain, it can have a much more dramatic
effect on the local transmission of the sound wave than
if it lies in a region away from any of the chains.
A fundamental issue concerns the packing of granular
materials. The random packing of spherical objects was
first studied by Stephen Hales, minister of Teddington,
who used the dimple patterns on peas that had been
expanded in a closed, water-filled vessel to ascertain
their geometrical arrangements (Hales, 1727). Depend-
ing on the procedure for filling the container, a random
assembly of spherical balls can be packed anywhere
from a volume fraction of h 50.55 to h 50.64 (Onoda
and Liniger, 1990; Bideau and Dodds, 1991). Through
FIG. 3. The transmitted rms magnitude of the acceleration (in static friction, force chains can hold the sand pile in a
units of g) versus time showing the effect of a temperature metastable configuration between these limits and keep
pulse (DT51 K) on the sound propagation in a granular me- it from collapsing. How does the system pass between
dium. The inset shows a schematic view of the apparatus. The these states? Since the energy k B T is negligible, the den-
source S was run at 4 kHz. The heater H was 1 cm from the sity can only change from disturbances of the container
driving plate. The data show the response of the detector D 2 by an external source, for instance, by vibrations. For
to two consecutive current pulses separated by 85 seconds. Af- this situation, Mehta and Edwards (1989) have proposed
ter each pulse, which heats only one bead in the container, the a new formalism that replaces conventional thermody-
transmission drops by roughly 25%. After Liu and Nagel,
namics. They neglect energy (since the particles are as-
(1994).
sumed to have no interactions aside from a hard sphere
repulsion) and replace the Hamiltonian with a volume
(see also Radjai et al., 1996). Interestingly, the experi- functional. The entropy is still the logarithm of the num-
ments of Miller et al. (1996) also show nearly exponen- ber of states at a given volume (Monasson and Poul-
tial distributions. iquen, 1996), and the other thermodynamic quantities
The force chains appearing so clearly in Fig. 2(a) are are defined in analogy with ordinary thermodynamics.
also important for many of the properties of the granu- But now, instead of k B T, a new effective temperature
lar material such as the transmission of sound (Liu and emerges that is given by the compactivity of the mate-
Nagel, 1992, 1993; Leibig, 1994; Liu, 1994; Melin, 1994; rial. External vibrations unlock the packing, thereby al-
Sinkovits and Sen, 1995). If one initiates a sound wave at lowing the system to travel slowly through phase space
one point in the material, the transmitted signal at a (Barker and Mehta, 1993; Mehta, 1994).
second position is sensitive to the exact arrangement of Studies (Knight et al., 1995) of granular material set-
all particles in the container. An example of this extraor- tling under vibrations indicate that the relaxation in
dinary sensitivity is shown in Fig. 3. By replacing a single these systems is, in fact, logarithmically slow. Even after
bead in the pile with a small carbon resistor, one can 100 000 vibration cycles, depending on the vibration in-
measure the effect of a small thermal expansion of a tensity, a tube filled with granular material might still
single bead on the transmission of sound (Liu and Na- undergo significant compaction before reaching a steady
gel, 1994). A one-degree change in the resistor produced state. A variety of models have been proposed to ac-
by a short current pulse causes a 100 nm thermal expan- count for this extremely slow settling (Barker and Me-
sion, which is small compared to the size of the particles, hta, 1993; Hong et al., 1994). At present, perhaps the
5 mm, or the wavelength of the sound, 1 cm. Neverthe- most plausible explanation rests on the idea that the rate
less, this perturbation of only a single particle out of the of increase in the granular volume fraction is exponen-
entire bead pack by an amount that is one part in 105 of tially reduced by excluded volume (Ben-Naim et al.,
all the other obvious length scales in the system can cre- 1996). A simple corresponding picture is that of a park-
ate a reproducible 25% change in the transmission of ing lot without assigned slots and with a high density of
sound! This dependence on the microscopic arrange- equal-sized, parked cars (viz. particles). For the person
ment of the particles is again reminiscent of the senti- wishing to park an extra vehicle (or insert an extra par-
ment expressed by Hugo in the quotation at the begin- ticle into the bead pack), the all-too-familiar situation is
ning of this article: macroscopic phenomena can be that there exist large, but not quite large enough, voids
affected by the placement and motion of even a single between the objects already in place. The question is
sand grain. If the heater is placed in another position, it how many other cars (or particles) have to be moved
will sometimes give an equivalent increase in the signal just a bit for the additional one to fit in? If all densifica-
and sometimes produce no discernible change at all tion occurs by random ‘‘parking’’ and ‘‘unparking’’
(Liu, 1994). This sensitivity is far greater than what is events, it takes the cooperative motion of many objects
found in conventional interference effects, where the (exponential in the density) to open up new slots. As a

Rev. Mod. Phys., Vol. 68, No. 4, October 1996

 H. M. Jaeger, S. R. Nagel, and R. P. Behringer: Granular solids, liquids, and gases 1263

result, the approach to the steady-state density is loga- material will only begin to deform for shear stresses
rithmic in time. Experiments on granular materials as above some yield point set by the applied normal
well as simulations of the parking problem indicate that stresses. At the yield point, grains begin to slide past
once the steady state is reached there are large density each other. For a given density, the locus of points at
fluctuations (Ben-Naim et al., 1996). which this occurs in a space of shear and normal stresses
defines the yield surface. Specific models consist of con-
III. AN UNUSUAL LIQUID: GRANULAR HYDRODYNAMICS servation laws in the form of partial differential equa-
tions augmented by constitutive models. Thus there is
Granular materials can flow like liquids, and there are the standard continuity equation for mass conservation,
a variety of theoretical models used to describe such an energy equation, and a momentum equation. The last
flows. We refer to these models as granular hydrody- of these is perhaps the most revealing. It relates the
namics (even though there is nothing wet here), in the stress tensor T i,j and the strain-rate tensor (defined here
sense that they are continuum theories consisting of par- with a minus sign, since granular materials disintegrate
tial differential equations, analogous to the Navier- under tension)
Stokes equations for Newtonian fluids. However, mod-
V i,j 52 ~ ] v i / ] x j 1 ] v j / ] x i ! , (2)
els for granular flow do not have the stature of the
Navier-Stokes equations. Those equations arise out of where v i is the ith component of the velocity field. In
an averaging process over length and time scales that are one of the simplest versions of these models (Jackson,
much larger than typical microscopic scales and much 1983; Schaeffer, 1987),
smaller than macroscopic ones. This separation of scales T i,j 5 s ~ d i,j 1kV i,j / u V u ! . (3)
may not occur in granular flows. Indeed, the issue of
which are the relevant time and space scales is one of Here u V u 2
[SV 2i,j
and k is a constant characteristic for
the most important questions to resolve. Slow granular each material. More specifically, k/ A25sind, where d is
flows of densely packed materials are certainly not er- the angle of internal friction. A comparison of this equa-
godic. More rapid flows are complicated by the phenom- tion with the Navier-Stokes equations reveals that the
ena of clustering or clumping, which can occur when the ordinary viscous terms, proportional to the viscosity and
coefficient of restitution for particle collisions is less the velocity gradients, have been replaced by shear-rate
than unity. Even in commercial settings, such as in the independent terms. This rate-independent feature is
flow of coal in a silo, the largest system size may be only quite remarkable, since it implies that an overall in-
a few thousand grain diameters. Since stress chains can crease in the velocity leaves the stress unchanged. This
easily span 100 grain lengths, there is no compelling rea- feature also means that these equations are more com-
son to believe that the system is homogeneous, and plex mathematically than the Navier-Stokes equations
therefore could be characterized by a continuum model. and apply only when the material is deforming. Models
In an ordinary fluid, an observation of the pressure is like that specified by Eq. (3) are used in soil mechanics
typically carried out over a time spanning an enormous and in the design of materials-handling devices such as
number of collisions so that the individual impacts of hoppers. However, visualization experiments of flow in
single molecules are averaged out. In many granular thin hoppers (Baxter et al., 1989) using continuous x-ray
flows, the amount of temporal averaging is much imaging have revealed a dynamic behavior that is not
weaker. captured by the standard plasticity models. These ex-
Dense slow flows and rapid gaslike flows are useful periments show density waves for rough materials but
idealizations for the development of models. Owing to not in smooth, nearly spherical ones. Figure 4 contrasts
rapid energy dissipation, real systems often exhibit both x-ray images of flow out of a quasi-two-dimensional
flow types simultaneously in different spatial domains (;1-cm-thick) hopper for rough- and smooth-grained
(Drake, 1990); one of the open questions is how to materials. In the case shown here for rough grains, the
model the transition between the two flows accurately. waves propagate upwards (against gravity), but the
Kinetic-theory models (Ogawa, 1978; Jenkins and Sav- propagation direction changes sign if the hopper angle is
age, 1983; Haff, 1983 and 1986; Jenkins, 1987; Lun and made sufficiently steep. These experiments appear to be
Savage, 1987; Savage, 1989; Savage and Jeffrey, 1981; inconsistent with theoretical predictions (Jenike, 1961,
Campbell, 1990) serve to describe granular materials 1964) and indicate that the grain shape plays a crucial
when the density is low. To obtain this state, energy role that requires better understanding. Recent experi-
must be continuously supplied, for instance, by shaking. ments and computer simulations of flow in chute and
Any dynamical quantity, such as the kinetic energy, is hopper geometries, where the flow is quite different
assumed to consist of a slowly varying hydrodynamic from what is found in simple fluids, attest to the role of
part and a fluctuating part. granularity (Lee, 1994b; Pouliquen and Gutfraind, 1996;
The other extreme is treated by models for quasistatic Pouliquen and Savage, 1996; Zheng and Hill, 1996).
plastic deformation (Jackson, 1983), based on Reynolds’ One of the exciting aspects about the present state of
principle of dilatancy (Reynolds, 1885; Bagnold, 1966) the physics of granular media is the vehement debate
and on the idea that deformations in compacted materi- that still exists about the causes for some of the most
als are typically irreversible. Dilatancy occurs because prominent behaviors that these materials exhibit when
the grains interlock under applied normal stress, and the vibrated. We shall briefly discuss two such debates: (i)

Rev. Mod. Phys., Vol. 68, No. 4, October 1996

1264 H. M. Jaeger, S. R. Nagel, and R. P. Behringer: Granular solids, liquids, and gases

FIG. 5. Schematic drawing of convection rolls in a cylindrical

vessel geometry (cross section). (a) Initial configuration. (b)
After just a few taps. (c) After several taps and after the beads
near the cylinder wall have reached the bottom of the convec-
FIG. 4. X-ray images of waves in a hopper of flowing sand, tion roll. (d) Schematic drawing of the situation in a cylindrical
after Baxter et al., (1989). Images (a)–(d) show the waves that cell with smooth frictionless walls except along a vertical stripe
form for rough sand (in this case, sieved construction sand); on the right-hand side that was coated with rough sand. (e)
the darker regions correspond to lower density; the numbers Cross section of a conical cell geometry. Here the rotation
under each image indicate the elapsed time in seconds. Images sense of the convection rolls is reversed, the flow occurring
(e)–(f) show two frames from a run using smooth (Ottowa) upward along the walls. After Knight et al., (1993).
sand. The waves are absent if the material is smooth, i.e., the
particle shape plays a key role. The typical grain sizes of the
two materials were identical: 0.6 mm to 0.7 mm.

G[A v 2 /g (5)
is a bit larger than unity, the material rises above the
floor of the container for some part of each cycle, dilat-
the cause for vibration-induced convection and heaping, ing in the process, so that a macroscopic flow of grains
and the role played by interstitial gas; and (ii) the cause can occur. This flow takes the form of convection rolls
of vibration-induced size separation. that continuously transport grains, as sketched in Fig. 5.
Convective flow in vibrated granular material was first In a typical experiment using cylindrical or rectangular
observed by Faraday 160 years ago (Faraday, 1831), yet vessels the flow is upwards in the center and downwards
its underlying mechanisms are only partially understood. in a thin stream along the side walls, leading to the for-
Both segregation and convection occur when the mate- mation of a central heap with a steady avalanche of
rial is shaken in the vertical direction, typically as grains downward (Evesque and Rajchenbach, 1989;
Fauve et al., 1989; Laroche et al., 1989; Knight et al.,
z5Acos~ v t ! . (4)
1993; Pak and Behringer, 1993; Lee, 1994; Pak and Be-
When hringer, 1994; Ehrichs et al., 1995; Knight et al., 1995;

Rev. Mod. Phys., Vol. 68, No. 4, October 1996

 H. M. Jaeger, S. R. Nagel, and R. P. Behringer: Granular solids, liquids, and gases 1265

Pak et al., 1995). With different boundary conditions, low as P.4 Torr. Pak et al. (1995) have shed light on
such as side walls that are slanted outward, it is possible this conflict through experiments where the pressure was
to reverse the sense of the convection roll, thus inducing held fixed at values between atmospheric pressure and
downward flow in the center (Takahashi et al., 1968; vacuum. The convective heap persisted for P down to
Knight et al., 1993; Jaeger et al., 1994; Bourzutschky and 10 Torr. As P was decreased further, the height L of the
Miller, 1995). More generally, container shape, wall and heap steadily diminished. These results apply for grains
interparticle friction, and internal phase boundaries can of diameter up to about 1 mm, and the effect is more
combine to reverse the direction of the convective flow pronounced for large oscillation amplitude A. A theo-
(Aoki et al., 1996; Knight, 1996; Van Doovn and Be- retical challenge remains to develop a theory that incor-
hringer, 1996). porates both the friction and gas effects.
At least three mechanisms have been proposed to ex- Another key feature of vibrated or flowing granular
plain these states. Savage (1988) considered lateral inho- material is its unique mixing and size-separation (‘‘un-
mogeneities in the shaking and found that inelastic col- mixing’’) behavior (for overviews, particularly also of
lisions of particles lead to upwardly directed pressure the associated industrial processes, see Williams, 1976
gradients that are strongest at the upflow. This mecha- and Fan et al., 1990). When granular materials are
nism may not be relevant to experiments in which the shaken, particles of different sizes tend to separate, with
entire layer is shaken uniformly. A second mechanism the largest particles moving to the top independently of
involves friction with the walls of the container. Several their density (Harwood, 1977; Rosato et al., 1987; Jullien
experiments and numerical simulations have shown a et al., 1992; Duran et al., 1993,1994; Knight et al., 1993;
kind of ratchet effect, which produces a thin, rapidly Cooke et al., 1996). Separation phenomena also occur in
moving boundary layer near the walls and leads to very long, slowly rotating cylinders with the cylinder axis
circulating flow (Gallas et al., 1992a,1992b; Herrmann, horizontal (Savage, 1993; Hill and Kakalios, 1994; Zik
1992; Taguchi, 1992a,1992b; Knight et al., 1993; Thomp- et al., 1994). Here, particles with different dynamical
son, 1993; Luding et al., 1994a,1994b; Bourzutschky and angles of repose aggregate into sharply delineated re-
Miller, 1995; Ehrichs et al., 1995; Hayakawa, Nishimori, gions along the axis. In rotating cylinders or drums with
Sasa, and Taguchi, 1995; Hayakawa, Yue, and Hong, a horizontal axis of rotation, particles flow down the free
1995; Pöschel et al., 1995; Taguchi and Takayasu, 1995). surface in a succession of avalanches (Jaeger et al., 1989;
Recent experiments using magnetic resonance imaging Rajchenbach, 1990; Bretz et al., 1992; Benza et al., 1993;
have been able to probe granular motion noninvasively Evesque, 1993; Morales-Gamboa et al., 1993; Nakagawa
everywhere inside the container (Nakagawa et al., 1993; et al., 1993; Sen et al., 1994; Baumann et al., 1994;
Ehrichs et al., 1995; Kuperman et al., 1995; Knight et al., Bouchaud et al., 1994; Clement et al., 1996; Linz and
1996) [for direct visualization in quasi-two-dimensional Hänggi, 1995; Frette et al., 1996). Particle motion for
containers see, e.g., Ratkai (1976); Tüzün and Nedder- more complicated types of agitation, such as horizontal
man (1982); Duran et al. (1994); Cooke et al. (1996)]. swirling, have also been studied (Scherer et al., 1996).
From such measurements both the depth dependence of An important question, particularly for industry, is how
the convection velocity and the detailed shape of the mixing occurs as a function of the filling fraction of the
velocity profiles have been obtained (Ehrichs et al., drum (Hogg et al., 1974). This question was recently ad-
1995; Knight et al., 1996) (Fig. 6). The experiments show dressed both theoretically and experimentally by Met-
that the fastest flow occurs in the thin boundary layer calfe et al. (1995). These authors found that simple geo-
near the walls. This is very different from what might metrical arguments were sufficient to predict the mixing
occur for a conventional fluid, for which the no-slip con- rate and efficiency. Both mixing and unmixing bear di-
dition applies, and raises a number of issues about the rectly on such technically important processes as the
correct boundary conditions for granular convection and separation of ‘‘fines’’ (which may or may not be desir-
other flows. able) or the mixing of powdered drugs with a binder,
A third mechanism for convection and heaping occurs where a well-controlled and homogeneous mixture is
in the presence of interstitial gas. This effect dominates highly desirable.
when friction with the container walls is eliminated or Several mechanisms have been associated with mixing
reduced (e.g., by choosing periodic boundary conditions and size separation, including sifting (where small par-
that can be realized at least partially in experiments in ticles fall through the gaps between large particles if the
annular containers and/or by choosing relatively small gaps are large enough) and local rearrangements [where
grains). Faraday (1831) was the first to attribute granular large particles will be wedged upwards as smaller grains
convection to the trapping of gas, and analysis of gas avalanche into voids beneath them in the dilation phase
trapping effects has been made by Gutman (1976). More during each shake (Williams, 1976; Rosato et al., 1987;
recently, experimenters have tried to clarify the role of Fan et al., 1990; Jullien and Meakin, 1992; Duran et al.,
gas in granular convection with conflicting results 1993)]. In vertically shaken systems, however, experi-
(Evesque and Rajchenbach, 1989; Fauve et al., 1989; ments (Knight et al., 1993) show a direct link between
Laroche et al., 1989; Pak et al. 1995). One set of experi- convection and size separation: large particles become
ments indicated that the flow stopped when the sur- entrained in the upward convective flow but, once on
rounding pressure was reduced, while another indicated the top surface, remain stranded because they cannot
that convection was virtually unchanged for pressures as follow the smaller grains in the thin layer of downward

Rev. Mod. Phys., Vol. 68, No. 4, October 1996

1266 H. M. Jaeger, S. R. Nagel, and R. P. Behringer: Granular solids, liquids, and gases

FIG. 6. Magnetic resonance (MR) images of a 1 mm slice through the center of a glass cylinder filled with white poppy seeds (the
color table chosen for each of the images is arbitrary). In (a) each grain visible in this image of the stationary cylinder corresponds
to one individual seed (roughly 1 mm across). The coating of seeds glued to the inner surface of the container walls is visible above
the filling level. In (b) the same slice through the center of the container is imaged, except that here a spatial modulation of the
spin polarization in the vertical z direction is applied. The peaks of this modulation appear as bright bands in the image, serving
to label narrow regions in the granular material. In (c), the MR image of the system is prepared in the same way as in (b) but after
a single shake of peak acceleration G58 g. The curvature of the horizontal stripes directly gives the flow profile of the seeds
(displacement per shake). The layer of seeds glued to the wall provides a marker for the position of the bands prior to the shake.
(d) The vertical velocity v (z) of the central region of each band is plotted as a function of depth z below the top surface. A
straight line on this plot indicates an exponential depth dependence of the velocity. Very close to z50 the vertical velocity
component decreases as particles no longer move straight up but turn sideways towards the container walls. Data for several
accelerations, G, are shown. After Ehrichs et al. (1995). (e) A plot of the velocity v (r) at a given depth is plotted as a function of
radial distance from the central axis of the container. The solid line through the points is a fit of the form
v (r)5 v c 1B @ 12I o (r/r 0 ) # , where I 0 is the modified Bessel function of order zero and v c , B, and r c are constants. The data shown
are normalized by the seed diameter d. The three curves pertain to three containers of different diameters, which are indicated by
the horizontal dotted lines. After Knight et al. (1996).

convective flow along the container walls. Recent work exhibit several different wave phenomena (Fauve et al.,
indicates that this mechanism drives size separation also 1989; Melo et al., 1993; Pak et al., 1993; Melo et al., 1995;
in two-dimensional systems (Duran et al., 1994; Cooke Brennen et al., 1996; Clement et al., 1996; Metcalf et al.,
et al., 1996). Thus in granular materials, shaking does 1996; van Doorn and Behringer, 1996; Wassgren et al.,
not induce mixing. In contrast to ordinary liquids, where 1996), as well as more complex, and possibly chaotic,
entropy favors a homogeneously mixed state, dynamics states (Dinkelacker et al., 1987; Douady et al., 1989; Pak
is dominant, and it leads to size separation. Similar be- and Behringer, 1994). The different waves can be either
havior has recently been observed in Couette-like shear traveling (for material with a steeply sloping heap) or
experiments (Khosropour et al., 1996). standing (when heaping is weak or nonexistent). Experi-
In addition to the convection patterns that exist in the ments on the former kind of waves show that not only is
bulk of a vibrated granular material, its free surface can G a relevant parameter but so is the ratio of energies

Rev. Mod. Phys., Vol. 68, No. 4, October 1996

 H. M. Jaeger, S. R. Nagel, and R. P. Behringer: Granular solids, liquids, and gases 1267

FIG. 7. Side views of traveling

waves for shaken sand in an
annular container, after Pak
and Behringer (1993). The
first sequence of images shows
a wave propagating up the
heap and corresponds to times
of 0.0 s, 0.4 s, 0.8 s, and 1.26 s
for (a)–(d), respectively. The
second sequence is similar but
shows the wave pulses forming
near the base of the heap, and
corresponds to times of 0.0 s,
0.4 s, 0.8 s, and 1.13 s for (a)–
(d), respectively.

m v 2 /mgd. We show examples of the traveling waves in shows the striking patterns that evolve when the con-
Fig. 7 for an annular container and of the subharmonic tainer is a large open cylinder. If fine granular materials
standing waves in Fig. 8. In this last figure, well-defined are shaken with large G, bubbling can ensue (Pak and
wave patterns and their superpositions occur that are Behringer, 1994), as in Fig. 8(d), resembling fluidized
strikingly familiar from Faraday instabilities in ordinary beds. At very high G, the resulting state may be consid-
liquids. In the first two parts of this figure, the waves are ered a kind of granular turbulence (Taguchi, 1995; Tagu-
confined to a narrow rectangular container. Part (c) chi and Takayasu, 1995).

Rev. Mod. Phys., Vol. 68, No. 4, October 1996

1268 H. M. Jaeger, S. R. Nagel, and R. P. Behringer: Granular solids, liquids, and gases

FIG. 8. Complex wave behavior observed in thin layers of granular material shaken vertically. (a) Sideview of a subharmonic
standing wave in a quasi one-dimensional rectangular container filled with glass beads at f;20 Hz, G;3.5, after Douady et al.,
(1989). (b) A different standing wave state, in an annular container, after Pak and Behringer (1993). The two snapshots are taken
at successive oscillation periods of the container and demonstrate the subharmonic nature of the response, which repeats after two
drive cycles rather than one. (c) Thin layers organize into a variety of stationary subharmonic two-dimensional patterns depending
on f and G. The pictures show top views of a large cylindrical container filled with brass spheres (165 micron diameter) to a
nominal depth of 8 spheres (photograph courtesy of P. Umbanhowar; see Melo et al., 1993 and 1995). Stripe (upper left) and
square patterns (upper right) correspond to f540 Hz, G53. Highly curved interfaces (lower right) between essentially flat
featureless plateaus coexist 180 degrees out of phase in different parts of the cell (f540 Hz, G55). Hexagonal patterns (lower left)
in this picture are produced by driving the system at two frequencies (16 and 18 Hz) for G52.8. (d) In material consisting of small
diameter grains that are shaken at large amplitudes, bubbling of air pockets can occur, after Pak and Behringer (1994).

IV. AN UNUSUAL GAS: INELASTICITY, CLUSTERING, ergy conservation, such as the theory for ideal gases,
COLLAPSE cannot carry over, with novel features arising for the
statistical mechanics of these systems. It is important to
One crucial difference between ordinary gases or liq- remember that any seemingly fluidlike behavior of a
uids and granular media deserves particular attention: granular material is a purely dynamic phenomenon. For
interactions between grains are inherently inelastic so example, the surface waves do not arise as a linear re-
that in each collision some energy is lost. As a result, all sponse to external energy input but are the consequence
approaches based on purely elastic interactions or en- of a highly nonlinear hysteretic transition out of the sol-

Rev. Mod. Phys., Vol. 68, No. 4, October 1996

 H. M. Jaeger, S. R. Nagel, and R. P. Behringer: Granular solids, liquids, and gases 1269

idlike state. Fluid behavior only sets in above a thresh-

old excitation level, with inelastic collisions bringing the
granular medium to rest almost instantly after the en-
ergy input is stopped. Drop an individual grain, such as a
single marble, onto a glass plate, and it will bounce for
quite a while, whereas a loosely filled sack of the very
same marbles will stop dead on the plate. This strikingly
different collective behavior arises from the exceedingly
large number of rapid inelastic collisions by neighboring
grains. In fact, we often use this curious behavior for
energy absorption in applications including packaging
fillers, recoilless hammers, or the common toy, Hacky
Sacks. At sufficiently high excitation frequencies, practi-
cally all applied energy is dissipated into heat. For ex-
ample, sound propagation in granular materials decays
exponentially for frequencies above 1–2 kHz (Liu,
1994).
The inherent inelasticity of granular collisions leads to
complications if we try to apply Newton’s Laws to indi-
vidual grain-grain interactions because we lack a coher-
ent picture of the dissipative forces involved. Questions
about the correct velocity dependence of friction forces FIG. 9. A typical configuration of 40 000 inelastically colliding
and about the importance of the impact duration or the particles, exhibiting clustering in two dimensions. Here the co-
inclusion of rotational degrees of freedom have been efficient of restitution is 0.6, the time corresponds to 500 colli-
debated for many years (Bagnold, 1954; Maw et al., sions per particle, and the average area fraction occupied by
1981; Campbell, 1990; Jaeger et al., 1990; Herrmann, particles is 0.05. The system is started with a spatially uniform
1992; Walton, 1992; Foerster et al., 1994; Brilliantov distribution of initial velocities and allowed to evolve without
et al., 1996; Hertzsch et al., 1995; Luck and Mehta, 1993; further energy input, after Goldhirsch and Zanetti (1993).
Radjai and Roux, 1995); yet these issues still appear very
much unresolved. The treatment of shearing collisions system of hard discs was started with random initial ve-
that occur off-center and at small relative velocities is locities, in the absence of gravity, and without external
particularly uncertain since it involves a crossover from forcing.
what we usually call ‘‘static’’ to ‘‘dynamic’’ friction. The A particularly exciting development has been the re-
existence of such a crossover also clearly demonstrates cent recognition that a special type of clustering may
that models based on a single fixed parameter like the occur, called ‘‘inelastic collapse.’’ McNamara and Young
coefficient of restitution oversimplify any real collision (1994) showed that inelasticity can lead to an infinite
process. number of collisions in a finite time. (Such a situation of
Since real granular materials are inelastic, energy in- an infinite number of collisions in a finite time occurs
put from the boundaries, as in an ordinary heat bath, whenever a ball bouncing on the ground comes to rest.
may not be sufficient to thermalize the system. If clus- However, in this case there is an attractive force of grav-
tering begins to occur inside the system, this effect may ity pulling the ball back to Earth. In the case of inelastic
indicate a breakdown of Newtonian hydrodynamics, collapse, there is no attractive force between particles to
since such aggregates will not be able to melt away. This cause particle collisions. In this case, it is the many-
clustering in the regime of finite inelasticity has recently particle dynamics that induces the infinite number of
become the focus of much interest (Walton, 1992; collisions.) In one dimension, such a collision sequence
Goldhirsch and Zanetti, 1993). The conditions under leaves the particles ‘‘stuck’’ together in close contact
which clustering should appear have been estimated with no relative motion (McNamara and Young, 1994;
(Goldhirsch and Zanetti, 1993; Esipov and Pöschel, Constantin et al., 1995; Du et al., 1995; Grossman and
1995): if a system of linear extent L is started in a uni- Mungan, 1996). Remarkably, ‘‘inelastic collapse’’ also
form state with grains occupying a volume fraction h , persists in higher dimensions, where it produces dense
the solutions provided by Newtonian hydrodynamics chainlike clusters, as shown in Fig. 10. Apparently, even
will become linearly unstable to cluster formation once in dimensions higher than one, there is a finite region of
the product h L exceeds some constant that depends phase space where such inelastic collapse can occur (Mc-
upon the degree of inelasticity. For large enough L this Namara and Young, 1996; Schörghofer and Zhou, 1996;
result implies that the system always becomes unstable Zhou and Kadanoff, 1996). The precise relationship be-
towards cluster formation, no matter how small the in- tween ‘‘inelastic collapse’’ and the phenomenon of clus-
elastic contribution to each collision. In Fig. 9, a snap- tering, which is the initial signature for a breakdown of
shot taken from a two-dimensional simulation by Gold- ordinary hydrodynamics, needs clarification. One plau-
hirsch and Zanetti (1993), we clearly see the tendency of sible scenario is that once the system forms clusters, the
inelastic collisions to produce particle clustering. Here a occurrence of inelastic collapse in addition requires that

Rev. Mod. Phys., Vol. 68, No. 4, October 1996

1270 H. M. Jaeger, S. R. Nagel, and R. P. Behringer: Granular solids, liquids, and gases

V. CONCLUSIONS AND OUTLOOK

We have only been able to touch on some of the dis-

tinctive properties of granular materials. In the preced-
ing sections we have shown that these materials act as
highly unusual solids, liquids, and gases, depending on
how we prepare and excite them. What should be clear
is that the physics of granular materials spans a wide
variety of phenomena with many possible applications,
ranging from the mundane to the celestial. The experi-
mental techniques used to study these systems likewise
span a wide range of approaches and sophistication—
from the examination of spots left by carbon paper to
high speed videography, magnetic resonance imaging,
and x-ray tomography. Despite their apparent simplicity
these materials display an intriguing range of nonlinear
complex behavior, whose unraveling more often than
not appears to challenge existing physics wisdom. Many
of the new ideas developed within the specific context of
granular materials are applicable to a wider range of
metastable systems whose thermal energy kT is irrel-
evant. Such systems include foams or the tunneling re-
gime for superconducting vortex arrangements. This
emerging new field of research within physics raises
many pressing, and often controversial, issues that need
FIG. 10. A two-dimensional simulation of hard disks colliding to be addressed.
inelastically in a container with periodic boundary conditions. From our personal perspective, we clearly see a chal-
The line of particles that are solid circles consists of those that lenge for new insights from physics to make a strong,
have undergone ‘‘inelastic collapse.’’ These particles have, technologically relevant impact, not achieved thus far, in
within the capacity of the computer simulation, undergone an spite of the tremendous opportunities and a recognized
infinite number of collisions, after McNamara and Young deficiency in our understanding of real granular materi-
(1994). als. In addition, we see a multitude of scientific chal-
lenges. For example, for the case of packing, we do not
know to what extent the packing history is relevant and,
the energy loss per collision must exceed a critical value if relevant, how to include it in theories of compaction
(Esipov and Pöschel, 1995). Yet a different scenario is or stress patterns within the medium. Likewise, when
also conceivable in which all clusters are transients, attempting a hydrodynamic approach to granular flow,
eventually terminating in either inelastic collapse or in we are still at a loss as to how to treat the boundaries
the formation of shear bands, i.e., regions of locally high correctly, while it is obvious that the ordinary hydrody-
shear separating essentially static material (McNamara namic nonslip boundary assumptions are invalid. Of
and Young, 1994 and 1996). course, it remains to be seen to what extent Newtonian
Perhaps the most remarkable aspect of clustering, hydrodynamics needs (or can) be modified to describe
which is also true for the regime of inelastic collapse, is granular media correctly. Certainly amongst the authors
that it leads to long stringlike grain configurations rather of the present article there is debate over this issue. The
than to a shapeless blob of particles. As noted by Gold- debate extends to a related question, namely, whether
hirsch and Zanetti (1993), Fig. 9 in this sense resembles the idea of inelastic collapse is more than a beautiful
qualitative density maps of the visible universe. We theoretical concept or whether it has real experimental
speculate here that the attractive gravitational potential ramifications, and whether the difference between in-
plays the role of a confining container, keeping the den- elastic collapse and the more general case of inelastic
sity high enough for clusters to form. As in simulations clustering is experimentally observable.
of liquids, the most important part of the interparticle Our hope is that the recent surge of interest in the
interaction is the strong repulsive one, and the small at- basic physics of granular media will produce advances
tractive part can often be discarded if a confining con- that can then not only lead to improved applications for
tainer serves to set the density. Hence on very large technological processes but also deepen our understand-
scales, the structures created by repeated inelastic colli- ing of the many related aspects of microscopic and mac-
sions may also be responsible for the coagulation ob- roscopic physics for which sand has been used as a meta-
served in ‘‘gases’’ made up of planets. Thus we get back phor.
to the quotation from Victor Hugo at the beginning of
the article: The motion of grains of sand may indeed be ACKNOWLEDGMENTS
relevant to the creation not just of worlds but of galaxies
and the structure and formation of our astronomical We would like to thank our collaborators for many
landscape. exciting discussions on the topics mentioned in this ar-

Rev. Mod. Phys., Vol. 68, No. 4, October 1996

 H. M. Jaeger, S. R. Nagel, and R. P. Behringer: Granular solids, liquids, and gases 1271

ticle. These include G. W. Baxter, E. Ben-Naim, S. N. Clauss, W., et al., 1990, Europhys. Lett. 12, 423.
Coppersmith, E. van Doorn, E. E. Ehrichs, L. P. Clement, E., J. Rajchenbach, and J. Duran, 1995, Europhys.
Kadanoff, G. S. Karczmar, J. B. Knight, V. Yu. Kuper- Lett. 30, 7.
man, C. H. Liu, B. Miller, E. R. Nowak, C. O’Hern, H. Clement, E., L. Vanel, J. Rajchenbach, and J. Duran, 1996,
K. Pak, D. A. Schecter, and T. A. Witten. We are grate- Phys. Rev. E (in press).
ful to J. Cina for first showing us the quote by Victor Constantin, P., E. Grossman, and M. Mungan, 1995, Physica D
Hugo, to W. Young and S. Esipov for a critical reading 83, 409.
of the manuscript, and to W. Lopes for help with for- Cooke, W., S. Warr, J. M. Huntley, and R. C. Ball, 1996, Phys.
matting. The work at the University of Chicago was sup- Rev. E 53, 2812.
ported by the MRSEC Program of the NSF under Coppersmith, S. N., C. H. Liu, S. Majumdar, O. Narayan, and
Award DMR-9400379 and by U.S. DOE DE-46G02- T. A. Witten, 1996, Phys. Rev. E 53, 4673.
92ER25119, and at Duke University by the NSF and Coulomb, C., 1773, in Memoir de Mathematique et de Phy-
AFOSR under Awards DMS-91034, DMR-9321791, and sique, Vol. 7 (Academie des Sciences, L’Imprimerie Royale,
DMS-9504577. H.M.J. acknowledges fellowship support Paris), p. 343.
from the David and Lucile Packard Foundation and the Dantu, P., 1957, Proc. 4th Int. Conf. Soil Mech. Found. Engng.
Research Corporation. An excerpted version of this ar- 1, 144.
ticle appeared in Phys. Today 49, 32 (1996). de Gennes, P. G., 1966, Superconductivity of Metals and Alloys
(Benjamin, New York), p. 83.
Dinkelacker, F., A. Hübler, and E. Luscher, 1987, Biol. Cy-
REFERENCES
bern. 56, 51.
Douady, S., S. Fauve, and C. Laroche, 1989, Europhys. Lett. 8,
Ammi, M., D. Bideau, and J. P. Troadec, 1987, J. Phys. D 20,
621.
424.
Drake, T. G., 1990, J. Geophys. Res. 95, 8681.
Aoki, K. M., T. Akiyama, Y. Maki, and T. Watanabe, 1996,
Drescher, A., and D. de Josselin de Jong, 1972, J. Mech. Phys.
Phys. Rev. E 54, 874.
Solids 20, 337.
Bagnold, R. A., 1954, Proc. R. Soc. London A 225, 49.
Bagnold, R. A., 1966, Proc. R. Soc. London A 295, 219. Du, Y., H. Li, and L. P. Kadanoff, 1995, Phys. Rev. Lett. 74,
Bak, P., C. Tang, and K. Wiesenfeld, 1988, Phys. Rev. A 38, 1268.
364. Duke, T. A. J., G. C. Barker, and A. Mehta, 1990, Europhys.
Barker, G. C., and A. Mehta, 1993, Phys. Rev. E 4, 184. Lett. 13, 19.
Baumann, G., I. M. Janosi, and D. E. Wolf, 1994, Europhys. Duran, J., J. Rajchenbach, and E. Clement, 1993, Phys. Rev.
Lett. 27, 203. Lett. 70, 2431.
Baxter, G. W., and R. P. Behringer, 1990, in Two Phase Flows Duran, J., T. Mazozi, E. Clement, and J. Rajchenbach, 1994,
and Waves, edited by D. D. Joseph and D. G. Schaefer Phys. Rev. E 50, 5138.
(Springer, New York), pp. 1–29. Ehrichs, E. E., H. M. Jaeger, G. S. Karczmar, J. B. Knight, V.
Baxter, G. W., R. P. Behringer, T. Faggert, and G. A. Johnson, Yu. Kuperman, and S. R. Nagel, 1995, Science 267, 1632.
1989, Phys. Rev. Lett. 62, 2825. Ennis, B. J., J. Green, and R. Davis, 1994, Particle Technology
Baxter, G. W., R. Leone, and R. P. Behringer, 1993, Europhys. 90, 32.
Lett. 21, 569. Esipov, S., and T. Pöschel, 1995, preprint.
Behringer, R. P., 1993, Nonlinear Science Today 3, 1. Evesque, P., 1993, Phys. Rev. E 47, 2326.
Behringer, R. P., 1995, Proc. MRS 367, 461. Evesque, P., and J. Rajchenbach, 1989, Phys. Rev. Lett. 62, 44.
Ben-Naim, E., J. B. Knight, E. Nowak, H. M. Jaeger, and S. R. Fan, L. T., Y. M. Chen, and F. S. Lai, 1990, Powder Technol.
Nagel, 1996, preprint. 61, 255. This is a continuation and update of the overview
Benza, V. G., F. Nori, and O. Pla, 1993, Phys. Rev. E 48, 4095. given by Williams (1976).
Bideau, D., and J. A. Dodds, 1991, Eds., Physics of Granular
Faraday, M., 1831, Phil. Trans. R. Soc. London 52, 299.
Media, Les Houches Series (Nova Science Publishers).
Fauve, S., S. Douady, and C. Laroche, 1989, J. Phys. (Paris) 50,
Bideau, D., and A. Hansen, 1993, Eds., Disorder and Granular
187.
Media, Random Materials and Processes Series (North-
Foerster, S. F., M. Louge, H. Chang, and K. Allia, 1994, Phys.
Holland, New York).
Boguslavskii, Y., and S. Drabkin, 1995, Physica A 222, 75. Fluids 6, 1108.
Bouchaud, J. P., M. E. Cates, J. R. Prakash, and S. F. Edwards, Frette, V., K. Christensen, and P. Meakin, 1996, Nature (Lon-
1994, J. Phys. I (France) 4, 1383. don) 379, 49.
Bourzutschky, M., and J. Miller, 1995, Phys. Rev. Lett. 74, Gallas, J. A. C., H. J. Herrmann, and S. Sokolowski, 1992a,
2216. Phys. Rev. Lett. 69, 1371.
Brennen, C. E., S. Ghosh, and C. R. Wassgren, 1996, Trans. Gallas, J. A. C., H. J. Herrmann, and S. Sokolowski, 1992b,
ASME, J. Appl. Mech. 63, 156. Physica A 189, 437.
Bretz, M., J. B. Cunningham, P. L. Kurcynskii, and F. Nori, Gallas, J. A. C., H. J. Herrmann, and S. Sokolowski, 1993, Int.
1992, Phys. Rev. Lett. 69, 2431. J. Mod. Phys. B 7, 1179.
Brilliantov, N. V., F. Spahn, J. M. Hertzsch, and T. Pöschel, Goldhirsch, I., and G. Zanetti, 1993, Phys. Rev. Lett. 70, 1619.
1996, Phys. Rev. E 53, 5382. Grossman, E. L., and M. Mungan, 1996, Phys. Rev. E 53, 6435.
Campbell, C. S., 1990, Annu. Rev. Fluid Mech. 22, 57. Gutman, R. G., 1975, Trans. Inst. Chem. Eng. 54, 174.
Carlson, J. M., J. S. Langer, and B. E. Shaw, 1994, Rev. Mod. Haff, P. K., 1983, J. Fluid Mech. 134, 401.
Phys. 66, 657. Haff, P. K., 1986, J. Rheol. 30, 931.

Rev. Mod. Phys., Vol. 68, No. 4, October 1996

1272 H. M. Jaeger, S. R. Nagel, and R. P. Behringer: Granular solids, liquids, and gases

Hales, S., 1727, Vegetable Staticks (Innys and Woodward, Lon- Liu, C. H., and S. R. Nagel, 1994, J. Phys. 6, A433.
don); reprinted edition edited by M. A. Hoskin (Oldourne, Liu, C. H., S. R. Nagel, D. A. Schecter, S. N. Coppersmith, S.
London, 1961). Majumdar, O. Narayan, and T. A. Witten, 1995, Science 269,
Harwood, C. F., 1977, Powder Technol. 16, 151. 513.
Hayakawa, H., H. Nishimori, S. Sasa, and Y. H. Taguchi, 1995, Luck, J. M., and A. Mehta, 1993, Phys. Rev. E 43, 3988.
Jpn. J. Appl. Phys. 34, 397. Luding, S., E. Clement, A. Blumen, J. Rajchenbach, and J.
Hayakawa, H., S. Yue, and D. C. Hong, 1995, Phys. Rev. Lett. Duran, 1994a, Phys. Rev. E 49, 1634.
75, 2328. Luding, S., E. Clement, A. Blumen, J. Rajchenbach, and J.
Herrmann, H. J., 1992, Physica A 191, 263. Duran, 1994b, Phys. Rev. E 50, 1762, 4113.
Hertzsch, J. M., F. Spahn, and N. V. Brilliantov, 1995, J. Phys. Luding, S., H. J. Herrmann, and A. Blumen, 1994, Phys. Rev.
5, 1725. E 50, 3100.
Hill, K. M., and J. Kakalios, 1994, Phys. Rev. E 49, 3610. Lun, C. K. K., and S. B. Savage, 1987, Trans. ASME, J. Appl.
Hogg, R., K. Shoij, and L. G. Austin, 1974, Powder Technol. 9, Mech. 54, 47.
99. Maw, N., J. R. Barber, and J. N. Fawcett, 1981, Trans. ASME
Hong, D. C., S. Yue, J. K. Rudra, M. Y. Choi, and Y. W. Kim, J. Lubr. Technol. 103, 12.
1994, Phys. Rev. E 50, 4123. McNamara, S., and W. R. Young, 1994, Phys. Rev. E 50, 28.
Horikawa, S., A. Nakahara, T. Nakayama, and M. Matsushita, McNamara, S., and W. Young, 1996, Phys. Rev. E 53, 5089.
1995, J. Phys. Soc. Jpn. 64, 1870. Mehta, A., 1994, Ed., Granular Matter: An Interdisciplinary
Ippolito, I., C. Annic, J. Lemaitre, L. Oger, and D. Bideau, Approach (Springer, New York).
1995, Phys. Rev. E 52, 2072. Mehta, A., and S. F. Edwards, 1989, Physica A 157, 378.
Jackson, R., 1983, in The Theory of Dispersed Multiphase Mehta, A., and G. C. Barker, 1994, Rep. Progr. Phys. 57, 383.
Flow, edited by R. Meyer (Academic, New York). Melin, S., 1994, Phys. Rev. E 49, 2353.
Jaeger, H. M., J. B. Knight, C. H. Liu, and S. R. Nagel, 1994, Melo, F., P. Umbanhowar, and H. L. Swinney, 1993, Phys.
Mater. Res. Bull. 19, 25. Rev. Lett. 72, 172.
Jaeger, H. M., C. H. Liu, and S. R. Nagel, 1989, Phys. Rev. Melo, F., P. Umbanhowar, and H. L. Swinney, 1995, Phys.
Lett. 62, 40. Rev. Lett. 75, 3838 and preprint.
Jaeger, H. M., C. H. Liu, S. R. Nagel, and T. A. Witten, 1990, Metcalf, T., J. B. Knight, and H. M. Jaeger, 1996, Physica A (in
Europhys. Lett. 11, 619. press).
Jaeger, H. M., and S. R. Nagel, 1992, Science 255, 1523. Metcalfe, G., T. Shinbrot, M. M. McCarthy, and J. M. Ottino,
Janssen, 1895, Z. Verein Deutsch. Ing. 39, 1045. 1995, Nature 374, 39.
Jenike, A. W., 1961, Bulletin of the University of Utah Engi- Miller, B., C. O’Hern, and R. P. Behringer, 1996, Phys. Rev.
neering Experiment Station No. 108. Lett. (in press).
Jenike, A. W., 1964, Bulletin of the University of Utah Engi- Monasson, R., and O. Pouliquen, 1996, preprint.
neering Experiment Station No. 123. Morales-Gamboa, E., V. Romero-Rochin, R. Chicharro-Serra,
Jenkins, J. T., 1987, in Non-classical Continuum Mechanics: and R. Peralta-Fabi, 1993, Phys. Rev. E 47, 2229.
Abstract Techniques and Applications, edited by R. J. Kops Nagel, S. R., 1992, Rev. Mod. Phys. 64, 321.
and A. A. Lacey (Cambridge University, Cambridge), p. 213. Nakagawa, M., S. A. Altobelli, A. Caprihan, E. Fukushima,
Jenkins, J. T., and S. B. Savage, 1983, J. Fluid Mech. 130, 186. and E. K. Jeong, 1993, Exp. Fluids 16, 54.
Jullien, R., P. Meakin, and A. Pavlovich, 1992, Phys. Rev. Lett. Ogawa, S., 1978, in Proceedings US-Japan Seminar on
69, 640. Continuum-Mechanical and Statistical Approaches in the Me-
Khosropour, R., J. Zirinsky, H. K. Pak, and R. P. Behringer, chanics of Granular Materials, edited by S. C. Cowin and M.
1996, Phys. Rev. E (in press). Satake (Gakujutsu Bunker Fukyukai, Tokyo, Japan).
Kim, 1994, Phys. Rev. E 50, 4123 (1994). Onoda, G. Y., and E. G. Liniger, 1990, Phys. Rev. Lett. 64,
Knight, J. B., 1996, preprint. 2727.
Knight, J. B., E. E. Ehrichs, V. Y. Kuperman, J. K. Flint, H. M. Pak, H. K., and R. P. Behringer, 1993, Phys. Rev. Lett. 71,
Jaeger, and S. R. Nagel, 1996, Phys. Rev. E (in press). 1832.
Knight, J. B., C. G. Fandrich, C. N. Lau, H. M. Jaeger, and S. Pak, H. K., and R. P. Behringer, 1994, Nature (London) 371,
R. Nagel, 1995, Phys. Rev. E 51, 3957. 231.
Knight, J. B., H. M. Jaeger, and S. R. Nagel, 1993, Phys. Rev. Pak, H. K., E. van Doorn, and R. P. Behringer, 1995, Phys.
Lett. 70, 3728. Rev. Lett. 74, 4643.
Knowlton, T. M., J. W. Carson, G. E. Klinzing, and W.-C. Peng, G., and H. J. Herrmann, 1995, Phys. Rev. E 51, 1745.
Yang, 1994, Particle Technology 90, 44. Pitman, E. B., and D. G. Schaefer, 1987, Commun. Pure Appl.
Kuperman, V. Yu, E. E. Ehrichs, H. M. Jaeger, and G. S. Math. 40, 421.
Karczmar, 1995, Rev. Sci. Instrum. 66, 4350. Pöschel, T., 1994, J. Phys. I (France) 4, 499.
Laroche, C., S. Douady, and S. Fauve, 1989, J. Phys. (Paris) 50, Pöschel, T., and H. J. Herrmann, 1995, Europhys. Lett. 29, 123.
699. Pouliquen, O., and R. Gutfraind, 1996, Phys. Rev. E. 53, 552.
Lee, J., 1994a, J. Phys. A 27, L257. Pouliquen, O., and S. B. Savage, 1996, preprint.
Lee, J., 1994b, Phys. Rev. E 49, 281. Radjai, F., P. Evesque, D. Bideau, and S. Roux, 1995, Phys.
Leibig, M., 1994, Phys. Rev. E 49, 1647. Rev. E 52, 5555.
Linz, S. J., and P. Hänggi, 1995, Phys. Rev. E 51, 2538. Radjai, F., M. Jean, J.-J. Moreau, and S. Roux, 1996, Phys.
Liu, C. H., 1994, Phys. Rev. B 50, 782. Rev. Lett. 77, 274.
Liu, C. H., and S. R. Nagel, 1992, Phys. Rev. Lett. 68, 2301. Radjai, F., and S. Roux, 1995, Phys. Rev. E 51, 6177.
Liu, C. H., and S. R. Nagel, 1993, Phys. Rev. B 48, 15646. Rajchenbach, J., 1990, Phys. Rev. Lett. 65, 2221.

Rev. Mod. Phys., Vol. 68, No. 4, October 1996

 H. M. Jaeger, S. R. Nagel, and R. P. Behringer: Granular solids, liquids, and gases 1273

Ratkai, G., 1976, Powder Technol. 15, 187. Takahashi, H., A. Suzuki, and T. Tanaka, 1968, Powder Tech-
Reiter, G., A. L. Demirel, and S. Granick, 1994, Science 263, nol. 2, 65.
1741. Thompson, P. A., 1993, in Computer Simulations in Condensed
Reynolds, O., 1885, Philos. Mag. 20, 469. Matter Physics VI, edited by D. P. Landau, K. K. Mond, and
Rosato, A., K. J. Strandburg, F. Prinz, and R. H. Swendsen, H. B. Schuttler (Springer, New York).
1987, Phys. Rev. Lett. 58, 1038. Travers, T., et al., 1987, Europhys. Lett. 4, 329.
Savage, S. B., 1984, Adv. Appl. Mech. 24, 289. Tüzün, U., and R. M. Nedderman, 1982, Powder Technol. 31,
Savage, S. B., 1988, J. Fluid Mech. 194, 457. 27.
Savage, S. B., 1989, in Theoretical and Applied Mechanics, ed- Van Doorn, E., and R. P. Behringer, 1996, preprint.
Wakabayashi, T., 1959, Proc. 9th Jpn. Nat. Congress Appl.
ited by P. Germane, M. Piau, and D. Caillerie (Elsevier, New
Mech., 133.
York), p. 241.
Walton, O. R., 1992, in Particulate Two-Phase Flow, Part I,
Savage, S. B., 1993, in Disorder and Granular Media, edited by
edited by M. C. Roco (Butterworth-Heinemann, Boston), p.
D. Bideau and A. Hansen (North-Holland, Amsterdam), pp.
884.
255. Walton, O. R., and R. L. Braun, 1986, J. Rheology 30, 949.
Savage, S. B., and D. J. Jeffrey, 1981, J. Fluid Mech. 110, 255. Warr, S., and J. M. Huntley, 1995, Phys. Rev. E 52, 5596.
Schaefer, D. G., 1987, J. Differ. Equ. 66, 19. Warr, S., J. M. Huntley, and G. T. H. Jacques, 1995, Phys. Rev.
Scherer, M. A., V. Buchholtz, T. Pöschel, and I. Rehberg, E 52, 5583.
1996, Phys. Rev. E (in press). Wassgren, C. R., C. E. Brennen, and M. L. Hunt, 1996, J.
Schörghofer, N., and T. Zhou, 1996, Phys. Rev. E (in press). Appl. Mech. 63, 712.
Sen, S., N. Mousseau, and G. Overney, 1994, Phys. Rev. E 49, Williams, J. C., 1976, Powder Technol. 15, 245.
4712. Wu, X. I., K. J. Maloy, A. Hansen, M. Ammi, and D. Bideau,
Sinkovits, R. S., and S. Sen, 1995, Phys. Rev. Lett. 74, 2686. 1993, Phys. Rev. Lett. 71, 1363.
Taguchi, Y. H., 1992a, Phys. Rev. Lett. 69, 1367. Zheng, X. M., and J. M. Hill, 1996, Powder Technol. 86, 219.
Taguchi, Y. H., 1992b, J. Phys. II (France) 2, 2103. Zhou, T., and L. P. Kadanoff, 1996, Phys. Rev. E 54, 623.
Taguchi, Y. H., 1995, Physica D 80, 61. Zik, O., D. Levine, S. G. Lipson, S. Shtrikman, and J. Stavans,
Taguchi, Y. H., and H. Takayasu, 1995, Europhys. Lett. 30, 1994, Phys. Rev. Lett. 73, 644.
499. Zik, O., and J. Stavans, 1991, Europhys. Lett. 16, 255.

Rev. Mod. Phys., Vol. 68, No. 4, October 1996