Length scales in Brownian yet non-Gaussian dynamics

According to the classical theory of Brownian motion, the mean squared displacement of diffusing particles evolves linearly with time whereas the distribution of their displacements is Gaussian. However, recent experiments on mesoscopic particle systems have discovered Brownian yet non-Gaussian regimes where diffusion coexists with an exponential tail in the distribution of displacements. Here we show that, contrary to the present theoretical understanding, the length scale $\lambda$ associated to this exponential distribution does not necessarily scale in a diffusive way. Simulations of Lennard- Jones systems reveal a behavior $\lambda \sim t^{1/3}$ in three dimensions and $\lambda \sim t^{1/2}$ in two dimensions. We propose a scaling theory based on the idea of hopping motion to explain this result. In contrast, simulations of a tetrahedral gelling system, where particles interact by a non-isotropic potential, yield a temperature-dependent scaling of $\lambda$. We interpret this behavior in terms of an intermittent hopping motion. Our findings link the Brownian yet non-Gaussian phenomenon with generic features of glassy dynamics and open new experimental perspectives on the class of molecular and supramolecular systems whose dynamics is ruled by rare events.
Comment: 17 pages, 11 figures