Characterizing dynamic length scales in glass-forming liquids
Reply to Comment by Flenner and Szamel on our paper in Nature Physics 8, 164 (2012).
cond-mat.dis-nn↗
arXiv subjects
Walter Kob
Publications and source records attributed to Walter Kob.
Spatial correlations in glass-forming liquids across the mode-coupling crossover
We discuss a novel approach, the point-to-set correlation functions, that allows to determine relevant static and dynamic length scales in glass-forming liquids. We find that static length scales increase monotonically when the temperature is lowered, whereas the measured dynamic length scale shows a maximum around the critical temperature of mode-coupling theory. We show that a similar non-monotonicity is found in the temperature evolution of certain finite size effects in the relaxation dynamics. These two independent sets of results demonstrate the existence of a change in the transport mechanism when the glass-former is cooled from moderately to deeply supercooled states across the mode-coupling crossover and clarify the status of the theoretical calculations done at the mean field level.
cond-mat.dis-nn↗
Static point-to-set correlations in glass-forming liquids
We analyze static point-to-set correlations in glass-forming liquids. The generic idea is to freeze the position of a set of particles in an equilibrium configuration and to perform sampling in the presence of this additional constraint. Qualitatively different geometries for the confining set of particles are considered and a detailed comparison of resulting static and dynamic correlation functions is performed. Our results reveal the existence of static spatial correlations not detected by conventional two-body correlators, which appear to be decoupled from, and shorter-ranged than, dynamical length scales characterizing dynamic heterogeneity. We find that the dynamics slows down dramatically under confinement, which suggests new ways to investigate the glass transition. Our results indicate that the geometry in which particles are randomly pinned is the best candidate to study static correlations.
cond-mat.stat-mech↗
The vibrational density of states of a disordered gel model
We investigate the vibrational density of states (vDOS) in harmonic approximation of a binary mixture of colloidal patchy particles with two and three patches for different relative compositions $x_2$. At low temperature, this system forms a thermo-reversible gel, i.e. a fully bonded network of chains of two-patches particles, in which the branching points are provided by three-patches particles. For all the compositions we find in the vDOS a pronounced peak at low frequency whose height grows on increasing the fraction of two-functional particles, or equivalently with the average length of the chains. To identify the various spectral features, we compare the vDOS of the whole system with the one of small representative structures of the network and with the vDOS of a long linear chain of two-patches particles and we find that these structures are indeed able to rationalize the various peaks in the vDOS of the full system. At large $x_2$ the vDOS of the gel and of the long chain show remarkable similarities. Analyzing the dispersion relations and the spectrum of the linear chain we show that the excess of low frequency modes, the analog of the boson peak in glassy disordered systems, arises from the strong coupling between rotations and translations.
cond-mat.soft↗
Non-monotonic temperature evolution of dynamic correlations in glass-forming liquids
The viscosity of glass-forming liquids increases by many orders of magnitude if their temperature is lowered by a mere factor of 2-3 [1,2]. Recent studies suggest that this widespread phenomenon is accompanied by spatially heterogeneous dynamics [3,4], and a growing dynamic correlation length quantifying the extent of correlated particle motion [5-7]. Here we use a novel numerical method to detect and quantify spatial correlations which reveal a surprising non-monotonic temperature evolution of spatial dynamical correlations, accompanied by a second length scale that grows monotonically and has a very different nature. Our results directly unveil a dramatic qualitative change in atomic motions near the mode-coupling crossover temperature [8] which involves no fitting or indirect theoretical interpretation. Our results impose severe new constraints on the theoretical description of the glass transition, and open several research perspectives, in particular for experiments, to confirm and quantify our observations in real materials.
cond-mat.dis-nn↗
Predicting the Structure of Alloys using Genetic Algorithms
We discuss a novel genetic algorithm that can be used to find global minima on the potential energy surface of disordered ceramics and alloys using a real-space symmetry adapted crossover. Due to a high number of symmetrically equivalent solutions of many alloys a conventional genetic algorithms using reasonable population sizes are unable to locate the global minima for even the smallest systems. We demonstrate the superior performance of the use of symmetry adapted crossover by the comparison of that of a conventional GA for finding global minima of two binary Ising-type alloys that either order or phase separate at low temperature. Comparison of different representations and crossover operations show that the use of real-space crossover outperforms crossover operators working on binary representations by several orders of magnitude.
cond-mat.mtrl-sci↗
Computer Simulation Study of the Phase Behavior and Structural Relaxation in a Gel-Former Modeled by Three Body Interactions
We report a computer simulation study of a model gel-former obtained by modifying the three-body interactions of the Stillinger-Weber potential for silicon. This modification reduces the average coordination number and consequently shifts the liquid-gas phase coexistence curve to low densities, thus facilitating the formation of gels without phase separation. At low temperatures and densities, the structure of the system is characterized by the presence of long linear chains interconnected by a small number of three coordinated junctions at random locations. At small wave-vectors the static structure factor shows a non-monotonic dependence on temperature, a behavior which is due to the competition between the percolation transition of the particles and the stiffening of the formed chains. We compare in detail the relaxation dynamics of the system as obtained from molecular dynamics with the one obtained from Monte Carlo dynamics. We find that the bond correlation function displays stretched exponential behavior at moderately low temperatures and densities, but exponential relaxation at low temperatures. The bond lifetime shows an Arrhenius behavior, independent of the microscopic dynamics. For the molecular dynamics at low temperatures, the mean squared displacement and the (coherent and incoherent) intermediate scattering function display at intermediate times a dynamics with ballistic character and we show that this leads to compressed exponential relaxation. For the Monte Carlo dynamics we find always an exponential or stretched exponential relaxation. Thus we conclude that the compressed exponential relaxation observed in experiments is due to the out-of-equilibrium dynamics.
cond-mat.dis-nn↗
Influence of the glass transition on the liquid-gas spinodal decomposition
We use large-scale molecular dynamics simulations to study the kinetics of the liquid-gas phase separation if the temperature is lowered across the glass transition of the dense phase. We observe a gradual change from phase separated systems at high temperatures to nonequilibrium, gel-like structures that evolve very slowly at low temperatures. The microscopic mechanisms responsible for the coarsening strongly depend on temperature, and change from diffusive motion at high temperature to a strongly intermittent, heterogeneous and thermally activated dynamics at low temperature, leading to logarithmically slow growth of the typical domain size.
cond-mat.stat-mech↗
A microscopic model for colloidal gels with directional effective interactions: Network induced glassy dynamics
By means of molecular dynamics, we study the structure and the dynamics of a microscopic model for colloidal gels at low volume fractions. The presence of directional interactions leads to the formation of a persistent interconnected network at temperatures where phase separation does not occur. We find that large scale spatial correlations strongly depend on the volume fraction and characterize the formation of the persistent network. We observe a pre-peak in the static structure factor and relate it to the network structure. The slow dynamics at gelation is characterized by the coexistence of fast collective motion of the mobile parts of the network structure (chains) with large scale rearrangements producing stretched exponential relaxations. We show that, once the network is sufficiently persistent, it induces slow, cooperative processes related to the network nodes. We suggest that this peculiar glassy dynamics is a hallmark of the physics of colloidal gels at low volume fractions.
cond-mat.soft↗
When gel and glass meet: A mechanism for multistep relaxation
We use computer simulations to study the dynamics of a physical gel at high densities where gelation and the glass transition interfere. We report and provide detailed physical understanding of complex relaxation patterns for time correlation functions which generically decay in a three-step process. For certain combinations of parameters we find logarithmic decays of the correlators and subdiffusive particle motion.
cond-mat.soft↗
Structural Relaxation of a Gel Modeled by Three Body Interactions
We report a molecular dynamics simulation study of a model gel whose interaction potential is obtained by modifying the three body Stillinger-Weber model potential for silicon. The modification reduces the average coordination number, and suppresses the liquid-gas phase coexistence curve. The low density, low temperature equilibrium gel that can thus form exhibits interesting dynamical behavior, including compressed exponential relaxation of density correlations. We show that motion responsible for such relaxation has ballistic character, and arises from the motion of chain segments in the gel without the restructuring of the gel network.
cond-mat.soft↗
Coupling/decoupling between translational and rotational dynamics in a supercooled molecular liquid
We use molecular dynamics computer simulations to investigate the coupling/decoupling between translational and rotational dynamics in a glass-forming liquid of dumbbells. This is done via a careful analysis of the $α$-relaxation time $τ_{q^{*}}^{\rm C}$ of the incoherent center-of-mass density correlator at the structure factor peak, the $α$-relaxation time $τ_{2}$ of the reorientational correlator, and the translational ($D_{t}$) and rotational ($D_{r}$) diffusion constants. We find that the coupling between the relaxation times $τ_{q^{*}}^{\rm C}$ and $τ_{2}$ increases with decreasing temperature $T$, whereas the coupling decreases between the diffusivities $D_{t}$ and $D_{r}$. In addition, the $T$-dependence of $D_{t}$ decouples from that of $1/τ_{2}$, which is consistent with previous experiments and has been interpreted as a signature of the "translation-rotation decoupling." We trace back these apparently contradicting observations to the dynamical heterogeneities in the system. We show that the decreasing coupling in the diffusivities $D_{t}$ and $D_{r}$ is only apparent due to the inadequacy of the concept of the rotational diffusion constant for describing the reorientational dynamics in the supercooled state. We also argue that the coupling between $τ_{q^{*}}^{\rm C}$ and $τ_{2}$ and the decoupling between $D_{t}$ and $1/τ_{2}$, both of which strengthen upon cooling, can be consistently understood in terms of the growing dynamic length scale.
cond-mat.soft↗
Glass transitions in 1, 2, 3, and 4 dimensional binary Lennard-Jones systems
We investigate the calorimetric liquid-glass transition by performing simulations of a binary Lennard-Jones mixture in one through four dimensions. Starting at a high temperature, the systems are cooled to T=0 and heated back to the ergodic liquid state at constant rates. Glass transitions are observed in two, three and four dimensions as a hysteresis between the cooling and heating curves. This hysteresis appears in the energy and pressure diagrams, and the scanning-rate dependence of the area and height of the hysteresis can be described by power laws. The one dimensional system does not experience a glass transition but its specific heat curve resembles the shape of the $D\geq 2$ results in the supercooled liquid regime above the glass transition. As $D$ increases, the radial distribution functions reflect reduced geometric constraints. Nearest-neighbor distances become smaller with increasing $D$ due to interactions between nearest and next-nearest neighbors. Simulation data for the glasses are compared with crystal and melting data obtained with a Lennard-Jones system with only one type of particle and we find that with increasing $D$ crystallization becomes increasingly more difficult.
cond-mat.dis-nn↗
Static and dynamic properties of a reversible gel
We study a microscopically realistic model of a physical gel and use computer simulations to investigate its static and dynamic properties at thermal equilibrium. The phase diagram comprises a sol phase, a coexistence region ending at a critical point, a gelation line, and an equilibrium gel phase unrelated to phase separation. The global structure of the gel is homogeneous, but the stress is supported by a fractal network. Gelation results in a dramatic slowing down of the dynamics, which can be used to locate the transition, which otherwise shows no structural signatures. Moreover, the equilibrium gel dynamics is highly heterogeneous as a result of the presence of particle families with different mobilities. An analysis of gel dynamics in terms of mobile and arrested particles allows us to elucidate several differences between the dynamics of equilibrium gels and that of glass-formers.
cond-mat.soft↗
Tracking heterogeneous dynamics during the alpha-relaxation of a simple glass-former
We study the relaxation process in a simple glass-former - the KA lattice gas model. We show that, for this model, structural relaxation is due to slow percolation of regions of co-operatively moving particles, which leads to heterogeneous dynamics of the system. We find that the size distribution of these regions is given by a power-law and that their formation is encoded in the initial structure of the particles, with the memory of initial configuration increasingly retained with increasing density.
cond-mat.stat-mech↗
On the relaxation dynamics of glass-forming systems: Insights from computer simulations
We discuss the relaxation dynamics of a simple lattice gas model for glass-forming systems and show that with increasing density of particles this dynamics slows down very quickly. By monitoring the trajectory of tagged particles we find that their motion is very heterogeneous in space and time, leading to regions in space in which there is a fast dynamics and others in which it is slow. We determine how the geometric properties of these quickly relaxing regions depend on density and time. Motivated by this heterogeneous hopping dynamics, we use a simple model, a variant of a continuous time random walk, to characterize the relaxation dynamics. In particular we find from this model that for large displacements the self part of the van Hove function shows an exponential tail, in agreement with recent findings from experiments and simulations of glass-forming systems.
cond-mat.soft↗
A genetic algorithm for the atomistic design and global optimisation of substitutionally disordered materials
We present a genetic algorithm for the atomistic design and global optimisation of substitutionally disordered bulk materials and surfaces. Premature convergence which hamper conventional genetic algorithms due to problems with synchronisation is avoided using a symmetry adapted crossover. The algorithm outperforms previously reported Monte Carlo and genetic algorithm simulations for finding low energy minima of two simple alloy models without the need for any redesign.
cond-mat.mtrl-sci↗
New fitting scheme to obtain effective potential from Car-Parrinello molecular dynamics simulations: Application to silica
A fitting scheme is proposed to obtain effective potentials from Car-Parrinello molecular dynamics (CPMD) simulations. It is used to parameterize a new pair potential for silica. MD simulations with this new potential are done to determine structural and dynamic properties and to compare these properties to those obtained from CPMD and a MD simulation using the so-called BKS potential. The new potential reproduces accurately the liquid structure generated by the CPMD trajectories, the experimental activation energies for the self-diffusion constants and the experimental density of amorphous silica. Also lattice parameters and elastic constants of alpha-quartz are well-reproduced, showing the transferability of the new potential.
cond-mat.dis-nn↗