Page 1
Statistical mechanics of random packings: from Kepler and
Bernal to Edwards and Coniglio
Hernan A. MakseLevich Institute and Physics Department City College of New York
jamlab.org
... inspired by a Lecture given by Antonio at Boston University in the mid 90’s on unifying concepts of glasses
and grains.
Page 2
Random packings of hard spheres
MathematicsPhysics Engineering
Kepler conjecture
Information theory
Granular matter
Glasses
Pharmaceutical industry
Mining & construction
(III) Polydisperse and non-spherical packings
Kepler (1611)
One of the twenty-three Hilbert's problems (1900).
Solved by Hales using computer-assisted proof (~2000).
Shannon (1948)
Signals → High dimensional spheres
Random close packing (RCP)
Bernal experiments (1960)
(II) High-dimensional packings
(I) Unifying concepts of glasses and grains
Coniglio, Fierro, Herrmann, Nicodemi, Unifying concepts
in granular media and glasses (2004).
Page 3
Parisi and Zamponi, Rev. Mod. Phys. (2010)
Schematic mean-field phase diagram of hard spheres
Theoretical approach I: Theory of hard-sphere glasses (replica theory)
Jammed states (infinite pressure limit)
• Approach jamming from the liquid phase.
• Predict a range of RCP densities
• Mean field theory (only exact in infinite dimensions).
Replica theory: jammed states are the infinite pressure limit of long-lived metastable hard sphere glasses
Page 4
Edwards and Oakeshott, Physica A (1989), Ciamarra, Coniglio, Nicodemi, PRL (2006).
Theoretical approach II: Statistical mechanics (Edwards’ theory)
Statistical mechanicsStatistical mechanics of
jammed matter
Hamiltonian Volume functionEnergy Volume
Microcanonical ensemble
Number of states
EntropyCanonical partition function
Temperature Compactivity
Free energy
Assumption: all stable configurations are equally probable for a given volume.
Page 5
The partition function for hard spheres
1. The Volume Function: W (geometry)
2. Definition of jammed state: force and torque balance
Volume Ensemble + Force Ensemble
Solution under different degrees of approximations
Page 6
Song, Wang, and Makse, Nature (2008)Song, Wang, Jin, Makse, Physica A (2010)
1. Full solution: Constraint optimization problem
2. Approximation: Decouple forces from geometry.
3. Edwards for volume ensemble + Isostaticity
T=0 and X=0 optimization problem: Computer science
4. Cavity method for force ensemble
Bo, Song, Mari, Makse (2012)
Page 7
The volume function is the Voronoi volume
Voronoiparticle
Important: global minimization. Reduce to to one-dimension
Page 8
Coarse-grained volume function
Excluded volume and surface: No particle can be found in:
Similar to a car parking model (Renyi, 1960). Probability to find a spot with in a volume VV
Page 9
Coarse-grained volume function
Particles are in contact and in the bulk:
Bulk term:
Contact term:
z = geometrical coordination number
mean free volume density
mean free surface density
Page 10
Prediction: volume fraction vs Z
Aste, JSTAT 2006X-ray tomography
300,000 grains
Equation of state agrees well with simulations and experiments
Theory
Page 11
Decreasing compactivity X
Isostatic plane
Disordered PackingsForbidden zone no disordered jammed packings can exist
Phase diagram for hard spheresSong, Wang, and Makse, Nature (2008)
0.634
Page 12
Jammed packings of high-dimensional spheres
Page 13
P>(c) in the high-dimensional limit
(I) Theoretical conjecture of g2 in high d (neglect correlations)
Torquato and Stillinger, Exp. Math., 2006
Parisi and Zamponi, Rev. Mod. Phys., 2010
(II) Factorization of P>(c)
Background term Contact term
Large d
3d
(mean-field approximation)
Page 14
Random first order transition theories (glass transition)
(I) Density functional theory (dynamical transition)
(II) Mode-coupling theory:
(III) Replica theory:
Kirkpatrick and Wolynes, PRA (1987).
Kirkpatrick and Wolynes, PRA (1987); Ikeda and Miyazaki, PRL (2010)
Parisi and Zamponi, Rev. Mod. Phys. (2010)
• No unified conclusion at the mean-field level (infinite d). Neither dynamics nor jamming.
• Does RCP in large d have higher-order correlations missed by theory?: Test of replica th.
• Are the densest packings in large dimensions lattices or disordered packings?
Edwards’ theory
Jin, Charbonneau, Meyer, Song, Zamponi, PRE (2010)
Comparison with other theories Isostatic packings (z = 2d) with
unique volume fraction
Isostatic packings (z = 2d) with ranging volume fraction increasing with dimensions
Agree with Minkowski lower bound
Page 16
Beyond packings of monodisperse spheres
Polydisperse packings Non-spherical packings
Clusel et al, Nature (2009)
Donev, et al, Science (2004)
• Higher density?
• New phases (jammed nematic phase)?
Platonic and Archimedean solids
Torquato, Jiao, Nature (2009)
Glotzer et al, Nature (2010).
Ellipses and ellipsoids
A first-order isotropic-to-nematic transition of equilibrium hard rods, Onsager (1949)
Page 17
17
Spheres
Dimers
Triangles
Tetrahedra
Spherocylinders
Ellipses and ellipsoids
Voronoi of non-spherical particles
The Voronoi of any
nonspherical shape can
be treated as interactions
between points and lines
Page 18
Generalizing the theory of monodisperse sphere packings
Theory of monodisperse spheres
Polydisperse (binary) spheres
(dimers, triangles, tetrahedrons, spherocylinders, ellipses, ellipsoids … )
Non-spherical objects
Extra degree of freedom
Distribution of radius P(r) Distribution of angles P( )
Page 19
Result of binary packingsBinary packings
Danisch, Jin, Makse, PRE (2010)
RCP (Z = 6)
Page 20
Results for packings of spherocylinders
Baule, Makse (2012)
Spherocylinder = 2 points + 1 line.
Interactions reduces to 9 regions of
line-points, line-line or point-point interactions.
Prediction of volume fraction versus aspect ratio:
agrees well with simulations
Same technique can be
applied to any shape.
Theory
Page 21
Cavity Method for Force Ensemble
Edwards volume ensemble predicts:
Cavity method predicts Z vs aspect ratio:
Page 25
2525
Solutions exist
No
solutionZ=2d
Page 27
A phase diagram for hard particles of different shapes
Phase diagram for hard spheres
generalizes to different shapes:
Spheres: disordered branch(theory)
Spheres: ordered branch(simulations)
Dimers
Spherocylinders
EllipsoidsFCC
RCP
RLP
Page 28
Conclusions
1. We predict a phase diagram of disordered packings
2. We obtain: RCP and RLP Distribution of volumes and coordination number Entropy and equations of state
3. Theory can be extended to any dimension: Volume function in large dimensions:
Isostatic condition: Same exponential dependence as Minkowski lower bound for lattices.
Page 29
Definition of jammed state: isostatic condition on Z
z = geometrical coordination number. Determined by the geometry of the packing.
Z =mechanical coordination number. Determined by force/torque balance.
Page 30
Sphere packings in high dimensions
Most efficient design of signals (Information theory)
Optimal packing (Sphere packing problem)
Sampling theorem
Question: what’s the density of RCP in high dimensions?
Rigorous bounds Minkowsky lower bound:
Kabatiansky-Levenshtein upper bound:
Signal
High-dimensional point
Sloane