Research
In my group we focus on the material properties of soft matter. We use methods of classical statistical mechanics, in and out of equilibrium. Keywords: Active and passive gels, Rheology, Critical Phenomena, Nonequilibrium statistical mechanics, Coarse graining, Green-Kubo, Fokker-Planck, Stochastic dynamics, Brownian dynamics, Monte Carlo.
Mechanics of fibrous networks
Strain-controlled critical behavior in subisostatic networks
Filamentous polymer networks are ubiquitous in nature. They make up the cytoskeleton of animal cells and form the scaffold of the extracellular matrix. These networks determine the mechanical response of cells and tissues and support elastic forces under external or internal loading. The mechanical stability of networks generally depends on the degree of connectivity: only when the average number of connections between nodes exceeds the isostatic threshold are networks stable. On increasing the connectivity through this point, such networks undergo a mechanical phase transition from a floppy to a rigid phase. However, even subisostatic networks become rigid when subjected to sufficiently large deformations.
We compute the elastic properties of such networks by performing large scale network simulations. The obtained data is analyzed using scaling theory. Using this combination, we have recently shown that the development of rigidity in subisostatic networks is characterized by a strain-controlled continuous phase transition with signatures of criticality. Moreover, the nonlinear mechanics of collagen networks can be quantitatively captured by predictions of scaling theory for controlled critical behavior over a wide range of network concentrations and strains up to failure of the material.
Force distribution in disordered networks
An important design feature of biological materials is the response to large loads, including failure, rupture, damage limitation, and their recovery properties. To understand failure that starts with the rupture of single filaments when the local force exceeds a threshold, it is crucial to understand force distributions in filament networks. It turns out that topology plays a critical role for the distribution of forces in elastic networks, but this topic has received little attention to date.
Our model system considers ensembles of linear random spring networks on a circle (see Fig.). To model a generically forced system, we employ a generation procedure that results in initial configurations that are not in mechanical equilibrium. This is meant to produce a situation equivalent to, say, a cytoskeletal protein network in which molecular motors are turned on that contract the network locally as force dipoles. We then study the resulting force distributions in the relaxed systems using a combination of probabilistic and graph-theoretical techniques.
We show that characteristic quantities, such as mean and variance of force distributions, can be derived explicitly in terms of only two parameters: (1) average connectivity and (2) number of nodes. Our analysis shows that a classical mean- field approach fails to capture these characteristic quantities correctly; the error is particularly pronounced for the biologi- cally most relevant regime of low degrees of connectivity.
Motor driven criticality in Biopolymer networks (Active gels)
Network connectivity plays an important role in active gels. It can be controlled by the number of crosslinks between filaments. In weakly connected systems, motors slide filaments to form static or dynamic clusters. In the opposite limit of a well-connected, elastic network, motors generate contractile stresses as they pull against crosslinks and stiffen the network or cause contraction. The existence of a threshold connectivity that separates these two behaviours has been proposed, because macroscopic contractions are known to occur above certain minimum values of crosslink or actin concentration. We should expect remarkable critical behaviour at the threshold of contraction. Recent theoretical models predict diverging correlation length scales and a strong response to external fields at the threshold of rigidity. Yet the threshold of contraction still remains poorly understood, and experimental evidence of criticality in active gels remains lacking.
In a joint experimental and theoretical study, we showed that the motors actively contract the networks into disjoint clusters that exhibit a power-law size distribution. This behaviour is reminiscent of classical conductivity percolation, for which a power-law size distribution of clusters occurs close to a critical point. However, in sharp contrast to this equilibrium phenomenon, we observe critical behaviour over a wide range of initial network connectivities.
Active Matter
Green-Kubo approach to ABPs
Assemblies of active, interacting Brownian particles (ABPs) are intrinsically nonequilibrium systems. In contrast to equilibrium, for which the statistical mechanics of Boltzmann and Gibbs enables the calculation of average properties, there is no analogous framework out-of-equilibrium. However, useful exact expressions exist, which enable average quantities to be calculated in the nonequilibrium system by integrating an appropriate time correlation function: the Green-Kubo formulae of linear response theory. Transport coefficients, such as the diffusion coefficient or shear viscosity, are thus conveniently related to equilibrium autocorrelation functions. Given the utility of the approach, it is surprising that the application of Green-Kubo-type methods to active Brownian systems has received little attention.
We have extended the Green-Kubo-type methods to treat ABPs. This approach has two appealing features. First, information about the active system can be obtained from equilibrium simulations. Second, the exact expressions derived provide a solid starting point for the development of approximation schemes and first-principles theory. The method we employ is a variation of the integration-through-transients approach, originally developed for treating interacting Brownian particles subject to external flow. A quantity of much interest, particularly for the phenomenon of Motility Induced Phase Transition (MIPS) is the average swim speed which describes how the motion of each particle is obstructed by its neighbours. We demonstrate that this quantity can be obtained from a history integral over the equilibrium autocorrelation of tagged-particle force fluctuations. One can also calculate the average polarization per particle for which the relevant correlation function is the van Hove function, a quantity that features prominently in liquid state theories.
Active Brownian particles subjected to Lorentz force
Lorentz force may appear irrelevant in soft-matter systems which are dominated by overdamped diffusive dynamics. However, it is detectable in, for example, soft tissues where it can be used for imaging applications. In systems with overdamped dynamics, the Lorentz force reduces the diusivity in the plane perpendicular to the magnetic field, which implies that the Fokker-Planck equation (FPE) requires a tensor. This may appear to be trivial, were it not for the recent finding that the tensor has an antisymmetric part, which gives rise to fluxes in the direction perpendicular to density gradients, thereby precluding a diffusive description of the dynamics. Although some aspects of the nondiffusive dynamics have been explored in passive systems, little is known about its effect on active systems.
We perform Brownian dynamics simulations of Active Brownian particles (ABPs) subjected to Lorentz force due to an external magnetic field. The theoretical analysis is performed using a combination of Linear Response theory and gradient expansion. We show that in the presence of a space-dependent magnetic eld, a macroscopic flux emerges from a flux-free system of ABPs. This stands in marked contrast with similar phenomena in condensed matter such as the classical Hall effect, which requires an explicitly broken symmetry: a macroscopic velocity vector in addition to the symmetry breaking due to the magnetic field vector. Figure above shows an example of the effect of the Lorentz force on ABPs. The two signatures of the Lorentz force on ABPs are clearly visible: an inhomogeneous density distribution and fluxes in the direction perpendicular to the gradient of the magnetic field.
Active liquid crystals
While spherical ABPs are ideal for exploring basic concepts, suspensions of anisotropic ABPs are perhaps more relevant, as these better represent the generic types of particles encountered in nature. Self-propelled rods (SPRs), the anisotropic analog of ABPs, for which the self-propulsion along the long axis of the particle breaks the up-down symmetry, exhibit a rich dynamical phase behavior at high (infinite) activity in two dimensions (2D). Simulations of large 2D systems (with rotational diffusion) reveal that at densities below the passive isotropic-nematic (IN) transition, the initially isotropic state begins to destabilize due to the emergence of moving polar clusters, which grow in size upon increasing activity but do not form a global phase.
The theoretical understanding of the phase behavior of SPRs is a difficult problem. According to an early mean-field approach, the density at the IN phase transition is insensitive to activity. In contrast, more general collision based models predict that the transition density decreases with increasing activity. For overdamped (Langevin) dynamics, the current numerical evidence suggesting that activity might stabilize nematic order of SPRs only arises from the observation that, as the density increases, the destabilization of the isotropic phase with respect to polar fluctuations occurs at lower activities. In general, the existence of a (nonpolar) nematic phase and its phase boundary remains an open problem.
We identified in 3D and for small aspect ratios a homogeneous nematic phase, close to equilibrium, which can be clearly distinguished from the isotropic phase, even in a relatively small system. By homogeneous, we mean that there are no appreciable inhomogeneities in the local density. This nonequilibrium nematic phase is gradually destabilized by activity and we observe no evidence for giant number fluctuations (but we cannot exclude the possibility entirely). Our finding is not sensitive to the precise particle shape or the details of the interaction, provided the rods are short. The activity-induced stabilization of the nematic phase predicted by mean-field theory for 2D is thus not universal.
Effective equilibrium approach: Escape of ABPs over potential barrier
The escape of a Brownian particle over a potential barrier is a thermally activated process. Kramers theory accurately describes the the escape process by taking into account the force acting on a particle due to the confining potential and solvent induced Brownian motion. Kramers showed that in the limit of vanishing particle-flux across the barrier, the escape rate decreases exponentially with increasing barrier height. In contrast to Brownian particles, active particles undergo both Brownian-motion and a self-propulsion which requires a continual consumption of energy from the local environment. Due to self-propulsion, active particles are expected to escape a potential barrier at a higher rate than their passive counterparts. However, a quantitative description of their escape rate, explicitly taking the activity was lacking.
We have extended the Kramer's theory of escape over a barrier to ABPs using the effective equilibrium approach. In this approach, one maps the active system to a passive one, with an effective potential. The effective potential is obtained using a first principles approach explicitly taking into account the activity parameters. Using methods from liquid state theories, the structure factor of an active fluid has been calculated and shown to match the results from Brownian dynamics simulations.
Liquid State
Phase transition dynamics
The dynamics of phase transition of a colloidal fluid depends on the diffusion rate of the particles. It is generally assumed that the particles diffuse isotropically in space. However, when a colloidal particle is subject to Lorentz forces arising from an external magnetic field, it diffuses anisotropically in space. Lorentz forces do no work on the system. They only change the underlying dynamics, i.e, the approach to the equilibrium state from a nonequilibrium initial state. This is our main motivation to study the phase transition dynamics of a fluid system in which the particles diffuse anisotropically.
We show that in comparison to the isotropic case, anisotropic diffusion results in qualitatively different dynamics of spinodal decomposition. Using the method of dynamical density functional theory, we predict that the intermediate-stage decomposition dynamics are slowed down significantly by anisotropy; the coupling between different Fourier modes is strongly reduced. We perform numerical calculations for a model (Yukawa) fluid that exhibits gas-liquid phase separation.