Transcript Jamming
Jamming Andrea J. Liu Department of Physics & Astronomy University of Pennsylvania Corey S. O’Hern Mechanical Engineering, Yale Univ. Leo E. Silbert Ning Xu Vincenzo Vitelli Matthieu Wyart Sidney R. Nagel Physics, S. Illinois U., Carbondale Physics, UPenn, JFI, U. Chicago Physics, UPenn Janelia Farms; Physics, NYU James Franck Inst., U. Chicago Jamming Umbrella concept that aims to tie together – two of oldest unsolved problems in condensed-matter physics • Glass transition • Colloidal glass transition – systems only recently studied by physicists • Granular materials • Foams and emulsions ¿Is there common behavior in these systems so that we can benefit by studying them in a broader context? Stress Relaxation Time • Behavior of glassforming liquids depends on how long you wait – At short time scales, silly putty behaves like a solid – At long time scales, silly putty behaves like a liquid QuickTime™ and a Sorenson Video 3 decompressor are needed to see this picture. Speeded up by x80 Stress relaxation time t: how long you need to wait for system to behave like liquid Glass Transition Earliest glassmaking 3000BC QuickTi me™ a nd a TIFF (Uncompre ssed ) decomp resso r are need ed to se e th is p icture. QuickTime™ and a TIFF (Uncompressed) decompressor are needed to see this picture. Glass vessels from around 1500BC When liquid is cooled through glass transition – Particles remain disordered – Stress relaxation time increases continuously – Can get 10 orders of magnitude increase in 1020 K range Colloidal Glass Transition Suspensions of small (nm-10mm) particles include – Ink, paint – McDonald’s milk shakes, ….. – Blood Micron-sized plastic spheres suspended in water form crystals glasses Stress relaxation time increases with packing fraction Granular Materials Materials made up of many distinct grains include – Pharmaceutical powders – Cereal, coffee grounds, …. – Gravel, landfill, …. San Francisco Marina District after Loma Prieta earthquake QuickTime™ and a TIFF (Uncompressed) decompressor are needed to see this picture. Static granular packing behaves like a solid Shaken granular packing behaves like a liquid Foams and Emulsions Suspension of gas bubbles or liquid droplets – Shaving cream – mayonnaise QuickTime™ and a TIFF (Uncompressed) decompressor are needed to see this picture. QuickTime™ and a TIFF (Uncompressed) decompressor are needed to see this picture. Courtesy of D. J. Durian • Foams flow like liquids when sheared • Stress relaxation time increases as shear stress decreases Phenomena look similar in all these systems • No obvious structural signature of jamming • Dramatic increase of relaxation time near jamming • Kinetic heterogeneities QuickTime™ and a BMP decompressor are needed to see this picture. Supercooled liquids Courtesy of S. C. Glotzer Colloidal suspensions Granular materials Courtesy of E. R. Weeks and D. A. Weitz Courtesy of A. S. Keys, A. R. Abate, S. C. Glotzer, and D. J. Durian These Transitions Are Not Understood • We understand crystallization and a lot of other phase transitions – Liquid-vapor criticality, liquid crystal transitions – Superconductivity, superfluidity, Bose-Einstein cond… – Many exotic quantum transitions, etc. • But glass transition, etc. remain mysterious – Are they really phase transitions or are they just examples of kinetic arrest? • Why are these systems so difficult? – They are disordered – They are not in equilibrium Jamming Jam ( jam), v. i. 1 To develop a yield stress in a disordered system 2 To have a stress relaxation time that exceeds 103 s in a disordered system E.g. Supercooled liquids jam as temperature drops Colloidal suspensions jam as density-1 drops Granular materials jam as driving force drops Foams, emulsions jam as shear stress drops ¿Can we unify these systems within one framework? Jamming Phase Diagram A. J. Liu and S. R. Nagel, Nature 396 (N6706) 21 (1998). Glass transition Granular matter Foams and emulsions Exp’tal Jamming Phase Diagram V. Trappe, V. Prasad, L. Cipelletti, P. N. Segre, D. A. Weitz, Nature, 411(N6839) 772 (2001). 1.0 0.8 0.6 0.4 10 kBT/U 0.2 0.0 20 1 2 Pa Colloids with depletion attractions 3 1/ Point J C. S. O’Hern, S. A. Langer, A. J. Liu and S. R. Nagel, Phys. Rev. Lett. 88, 075507 (2002). C. S. O’Hern, L. E. Silbert, A. J. Liu, S. R. Nagel, Phys. Rev. E 68, 011306 (2003). Problem: Jamming surface is fuzzy Temperature unjammed soft, repulsive, finite-range spherically-symmetric potentials jammed 1/Density J • Point J is special – Random close-packing – Isostatic – Mixed first/second order zero T transition – Connections to glasses and glass transition Shear stress How we study Point J • Generate configurations near J – e.g. Start w/ random initial positions 1 r / ij V (r) 0 r ij Ti=∞ r ij – Conjugate gradient energy minimization (inherent structures, Stillinger & Weber) • Classify resulting configurations non-overlapped V=0 p=0 overlapped V>0 p>0 or Tf=0 T =0 Onset of Jamming is Onset of Overlap -2 D=2 D=3 (a) -4 =2 -6 =5/2 p p0(f f c ) 1 -8 •Shear modulus and pressure vanish at the same fc 0 (b) -2 -4 -6 1 =2 =5/2 G G0(f fc ) 1.5 0 (c) -4 •Pressures for different states collapse on a single curve -3 3D log(f- fc) •Good ensemble is fixed f - fc -2 2D 2 3 Durian, PRL 75, 4780 (1995); D. J. 5 4 3 2 C. S. O’Hern, S. A. Langer, A. J. Liu, S. R. Nagel, PRL 88, 075507 (2002). log (f- f ) c Dense Sphere Packings ¿What is densest packing of monodisperse hard spheres? Johannes Kepler (1571-1630) Conjecture (1611) Thomas Hales Fejes Tóth 2D Proof (1953) 3D Proof (1998) 2D 3D triangular is densest possible packing f / 12 0.906 FCC/HCP is densest possible packing f / 18 0.740 Disordered Sphere Packings Stephen Hales (1677-1761) Vegetable Staticks (1727) J. D. Bernal (1901-1971) 2D 3D frcp 0.84 frcp 0.64 < < f cp 0.906 f cp 0.740 •Random close-packing is not well-defined mathematically –One can always make a closer-packed structure that is less random S. Torquato, T. M. Truskett, P. Debenedetti, PRL 84, 2064 (2000). –But it is highly reproducible. Why? Kamien, Liu, PRL 99, 155501 (2007). How Much Does fc Vary Among States? • Distribution of fc values narrows as system size grows 1 N=16 N=32 N=64 N=256 N=1024 N=4096 2 1.5 2 w N 0.55 3 1 w 0.58 0.6 f0 0.62 fc 0.64 1 234 log N • Distribution approaches delta-function as N • Essentially all configurations jam at one packing density • J is a “POINT” Point J is at Random Close-Packing 1 f * f 0 N 1/ d 2 1.5 log(f*- f0) N=16 N=32 N=64 N=256 N=1024 N=4096 2 3 1 w 0.58 0.6 f0 0.62 fc 0.64 0.7 1 234 log N •Where do virtually all states jam in infinite system limit? f * 0.842 0.001 2d (bidisperse) RCP! * f 0.639 0.003 3d (monodisperse) Most of phase space belongs to basins of attraction of hard sphere states that have their jamming thresholds at RCP Point J Temperature soft, repulsive, finite-range spherically-symmetric potentials 1/Density unjammed jammed Shear stress J • Point J is special – Random close-packing – Isostatic – Mixed first/second order zero T transition – Connections to glasses and glass transition 4 =2 Number of Overlaps/Particle Z 6 =5/2 Just below fc, no 8 particles overlap Just above fc there are Zc overlapping neighbors per particle 0 (b) 2 =2 =5/2 4 6 0 3D (c) -1 -2 -3 2D -5 -4 Z Zc Z0 (f fc ) 0.5 -3 Zc 3.99 0.01 (2D) Zc 5.97 0.03 (3D) -2 log(flog (f-fcf)c) Durian, PRL 75, 4780 (1995). O’Hern, Langer, Liu, Nagel, PRL 88, 075507 (2002). Verified experimentally: Majmudar, Sperl, Luding, Behringer, PRL 98, 058001 (2007). Isostaticity • What is the minimum number of interparticle contacts needed for mechanical equilibrium? •No friction, spherical particles, D dimensions –Match unknowns (number of interparticle normal forces) equations (force balance for mechanical stability) –Number of unknowns per particle=Z/2 –Number of equations per particle = D James Clerk Maxwell Z 2D • Same for hard spheres at RCP Donev, Torquato, Stillinger, PRE 71, 011105 (‘05) • Point J is purely geometrical! Doesn’t depend on potential Marginally Jammed Solid is Unusual L. E. Silbert, A. J. Liu, S. R. Nagel, PRL 95, 098301 (‘05) Density of Vibrational Modes f fc • Excess low-w modes swamp w2 Debye behavior: boson peak • g(w) approaches constant as f fc • Result of isostaticity M. Wyart, S.R. Nagel, T.A. Witten, EPL 72, 486 (05) Isostaticity and Boundary Sensitivity M. Wyart, S.R. Nagel, T.A. Witten, EPL 72, 486 (05) •For system at fc, Z=2d •Removal of one bond makes entire system unstable by introducing one soft mode •This implies diverging length as f-> fc + For f> fc, cut bonds at boundary of circle of size L Count number of soft modes within circle Ns Ld 1 Z Zc Ld Define length scale at which soft modes just appear 1 0.5 f fc Z Zc Diverging Length Scale Ellenbroek, Somfai, van Hecke, van Saarloos, PRL 97, 258001 (2006) Look at response to small particle displacement Define h(r) f r (r) 2 f r (r) 2 0.5 Diverging Time and Length Scales / 20.5 w* f f c 0.26 f f c w* QuickTime™ and a TIFF (Uncompressed) decompressor are needed to see this picture. w • For each f-fc, extract w* where g(w) begins to drop off • Below w* , modes approach those of ordinary elastic solid • Decompose corresponding eigenmode in plane waves • Dominant wavevector contribution is k*=w*/cT • We also expect k # w * /c L with # 0.5 2 /k * Point J Temperature soft, repulsive, finite-range spherically-symmetric potentials 1/Density unjammed jammed Shear stress J • Point J is special – Random close-packing – Isostatic – Mixed first/second order zero T transition – Connection to glasses and glass transition Summary of Jamming Transition 0 if r V (r) 1 r / if r • Mixed first-order/second-order transition (random first-order phase transition) • Number of overlapping neighbors per particle 0 Z Z Z ( f f ) c 0 c • Static shear modulus 2 G G0 f f c f fc f fc f f c • Two diverging length scales f f 0 c • Vanishing frequency scale w w 0 f fc * 1 0.49 0.03 0.48 0.03 0.26 0.05 # 0.5 Similarity to Other Models • In jamming transition we find – Jump discontinuity & =1/2 power-law in order parameter – Divergences in susceptibility/correlation length with =1/2 and =1/4 • This behavior has only been found in a few models – Mean-field p-spin interaction spin glass Kirkpatrick, Wolynes – Mean-field compressible frustrated Ising antiferromagnet – Mean-field kinetically-constrained Ising models – Mean-field k-core percolation and variants – Mode-coupling approximation of glasses – Replica solution of hard spheres • Yin, Chakraborty Sellitto, Toninelli, Biroli, Fisher Schwarz, Liu, Chayes Biroli, Bouchaud Zamponi, Parisi These other models all exhibit glassy dynamics!! First hint of quantitative connection between sphere packings and glass transition Point J Temperature soft, repulsive, finite-range spherically-symmetric potentials 1/Density unjammed jammed Shear stress J • Point J is special – Random close-packing – Isostatic – Mixed first/second order zero T transition – Connection to glasses and glass transition Low Temperature Properties of Glasses • Distinct from crystals • Common to all amorphous solids • Still mysterious – Excess vibrational modes compared to Debye (boson peak) – Cv~T instead of T3 (two-level systems) ~T2 instead of T3 at low T (TLS) – K has plateau – K increases monotonically crystal / 20.5 w* f f c w* amorphous T ffc Energy Transport 1 Ci (T )di V i thermal heat carried conductivity by mode i P. B. Allen and J. L. Feldman, PRB 48,12581 (1993). diffusivity of mode i Kubo formulation 2 di 2 2 Sij w i w j 3h w i i j Kittel’s 1949 hypothesis: rise in above plateau due to regime of freq-independent diffusivity N. Xu, V. Vitelli, M. Wyart, A. J. Liu, S. R. Nagel (2008). Ioffe-Regel Crossover • Crossover from weak to strong scattering at wIR wIR ~ w* Ioffe-Regel crossover at boson peak • Unambiguous evidence of freq-indep diffusivity as hypothesized for glasses • Freq-indep diffusivity originates from soft modes at J! Quasilocalized Modes QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. • Modes become quasilocalized near IoffeRegel crossover • Quasilocalization due to disorder in coordination z • Harmonic precursors of two-level systems? Relevance to Glasses • Point J only exists for repulsive, finite-range potentials • Real liquids have attractions U Repulsion vanishes at Attractions serve to hold system at high enough density that repulsions come into play (WCA) finite distance r • Excess vibrational modes (boson peak) believed responsible for unusual low temp properties of glasses • These modes derive from the excess modes near Point J N. Xu, M. Wyart, A. J. Liu, S. R. Nagel, PRL 98, 175502 (2007). Glass Transition Would expect : exp(A / T ) Arrhenius behavior But most glassforming liquids obey something like : exp A / T T0 L.-M. Martinez and C. A. Angell, Nature 410, 663 (2001). T0 measures “fragility” x T0(p) is Linear • • • 3 different types of Temperature trajectories to glass transition unjammed – Decrease T at fixed f – Decrease T at fixed p jammed – Increase p at fixed T Shear stress 4 different potentials – Harmonic repulsion J 1/Density – Hertzian repulsion – Repulsive Lennard-Jones (WCA) – Lennard-Jones All results fall on consistent curve! QuickTime™ and a TIFF (LZW) decompressor • T0 -> 0 at Point J! are needed to see this picture. Experimental Data for Glycerol K. Z. Win and N. Menon 350 Tm 300 5 10 Hz 250 10 3 10 T 200 Pg P0 Tg 150 T0 0 5 10 15 20 P/kb 25 30 Conclusions • Point J is a special point T • First hint of universality in jamming transitions xy • Tantalizing connections to glasses and glass transition 1/r J • Looking for commonalities can yield insight • Physics is not just about the exotic; it is all around you Hope you like jammin’, too!--Bob Marley Bread for Jam: NSF-DMR-0605044 DOE DE-FG02-03ER46087 Imry-Ma-Type Argument M. Wyart, Ann. de Phys. 30 (3), 1 (2005). • Upper critical dimension for jamming transition may be 2 • Recall soft-mode-counting argument Ns Ld 1 Z Zc Ld • Now include fluctuations in Z Ns Ld 1 Z Zc Ld ZLd / 2 • This would explain – Observed exponents same in d=2 and d=3 – Similarity to mean-field k-core exponents* *k-core percolation has different behavior in d=2 J. M. Schwarz, A. J. Liu, L. Chayes, EPL 73, 560 (2006) C. Toninelli, G. Biroli, D. S. Fisher, PRL 96,035702 (2006) Nature of Vibrational Modes Visualize 2D displacement vectors in mode i at atom Participation ratio Nature of Vibrational Modes localized Nature of Vibrational Modes localized disturbances merging Nature of Vibrational Modes extended Nature of Vibrational Modes wave-like Nature of Vibrational Modes resonant Characterize modes in different portions of spectrum. Mode Analysis of 3D Jammed packings Stressed Unstressed replace compressed bonds with relaxed springs. coordination number Low resonant modes have high displacements on under-coordinated particles. at low increases weakly as How Localized is the Lowest Frequency Mode? • Mode is more and more localized with increasing f • Two-level systems and STZ’s Strong Anharmonicity at Low Frequency Gruneisen parameters is O(1) for ordinary solids at low Compression causes increase in stress consistent with scaling of Stronger anharmonicity at lower frequencies; two-level systems? Thermal Diffusivity Gedanken experiment energy diffusivity Thermal conductivity sum over all modes heat capacity low T Kinetic theory of phonons in crystals: speed of sound mean free path eg.waves of frequency incident on a random distribution of scatterers (ignore phonon interactions) Rayleigh scattering Calculation of Diffusivity linear response theory & harmonic approx. Kubo formula for amorphous solids: AC conductivity Extract Diffusivity from double limit: N 0 Energy flux matrix elements calculated directly from eigenmodes and eigenvalues Diffusivity an intrinsic property of the modes and their coupling independent of temperature J. D. Thouless, Phys. Rep. 13, 3, 93 (1974). P. B. Allen and J. L. Feldman, PRB 48,12581 (1993). Diffusivity of Jammed Packings Stressed Unstressed replace compressed bonds with relaxed springs • Above wL modes are localized and diffusivity vanishes • Most modes are extended with low diffusivity • Plateau extends all the way to low w in DC limit Scaling Collapse P. Olsson and S. Teitel, cond-mat/07041806 T • Measure shear viscosity, length scale vs. shear stress • Scaling collapse of all data xy • Two branches: one below J, one above • Power-law scaling at Pt. J 1/r J K-Core Percolation in Finite Dimensions There appear to be at least 3 different types of k-core percolation transitions in finite dimensions 1. Continuous transition (Charybdis) 2. No percolation until p=1 (Scylla) 3. Mixed transition? Continuous K-Core Percolation • Appears to be associated with self-sustaining clusters • For example, k=3 on triangular lattice Self-sustaining clusters don’t exist in sphere packings p=0.4, before culling p=0.6, after culling • p=0.4, after culling p=0.65, after culling pc=0.6921±0.0005, M. C. Madeiros, C. M. Chaves, Physica A (1997). No Transition Until p=1 • E.g. k=3 on square lattice There is a positive probability that there is a large empty square whose boundary is not completely occupied Voids unstable to shrinkage, not growth in sphere packings After culling process, the whole lattice will be empty Straley, van Enter J. Stat. Phys. 48, 943 (1987). M. Aizenmann, J. L. Lebowitz, J. Phys. A 21, 3801 (1988). R. H. Schonmann, Ann. Prob. 20, 174 (1992). C. Toninelli, G. Biroli, D. S. Fisher, Phys. Rev. Lett. 92, 185504 (2004). A k-Core Variant •We introduce “force-balance” constraint to eliminate self-sustaining clusters k=3 24 possible neighbors per site Cannot have all neighbors in upper/lower/right/left half •Cull if k<3 or if all neighbors are on the same side Fraction of sites in spanning cluster Discontinuous Transition? Yes • The discontinuity c increases with system size L • If transition were continuous, c would decrease with L Pc<1? Yes Finite-size scaling If pc = 1, expect pc(L) = 1-Ae-BL Aizenman, Lebowitz, J. Phys. A 21, 3801 (1988) We find pc (L ) 0.396(1) We actually have a proof now that pc<1 (Jeng, Schwarz) Diverging Correlation Length? Yes 1 1.5 • This value of 1 1.5 collapses the order parameter data with 1.0 • For ordinary 1st-order transition, 1/d 0.5 BUT • Exponents for k-core variants in d=2 are different from those in mean-field! Mean field 1/2 1/4 d=2 1.0 1.5 Why does Point J show mean-field behavior? • Point J may have critical dimension of dc=2 due to isostaticity (Wyart, Nagel, Witten) • Isostaticity is a global condition not captured by local kcore requirement of k neighbors Henkes, Chakraborty, PRL 95, 198002 (2005).