The dynamical structure factor in topologically disordered systems

V. Martı́n-Mayor, Marc Mézard, Giorgio Parisi, P. VerrocchioHAL 2000 · DOI

TL;DR

This paper presents a theoretical framework using random matrix theory to calculate the dynamical structure factor of topologically disordered systems, like glasses and supercooled liquids. The authors' analytical results, validated by numerical simulations, explain the observed high-frequency sound behavior in these materials and offer a new perspective on their "disordered" nature.

What Problem Does This Paper Address?

Inelastic X-ray scattering (IXS) experiments have revealed a peculiar "high-frequency sound" phenomenon in structural glasses and supercooled liquids. This sound exhibits a linear dispersion relation at low momentum, but its width (linewidth) scales as $p^2$ (where $p$ is momentum), a behavior that deviates from standard Rayleigh scattering ($p^4$). While this $p^2$ broadening is observed across various glassy systems, its underlying physical origin remains unclear. Existing models, like the disordered-solid model, predict a $p^4$ scaling, and hydrodynamic descriptions fail to capture the observed temperature independence of the broadening constant. This paper aims to provide a theoretical explanation for this $p^2$ scaling by treating the disordered system as a problem in statistical mechanics of random matrices.

Methods

The core of the paper's methodology lies in applying random matrix theory to model the dynamics of topologically disordered systems. Specifically, the authors:

  • Model the system with Euclidean Random Matrices: They represent the dynamical structure factor as a problem of spectral properties of a specific type of random matrix, known as Euclidean matrices. These matrices have entries that are deterministic functions of the random positions of particles.
  • Focus on Vibrational Dynamics: The approach assumes that at high frequencies, the system's dynamics are dominated by vibrations around "quenched" configurations.
  • Develop a 1/p Expansion: They employ a perturbative expansion in the inverse particle density ($1/\rho$) to calculate the resolvent of the system, which is directly related to the dynamical structure factor. This expansion allows them to systematically analyze the contributions of different orders of interaction.
  • Introduce a Gaussian Model: For analytical tractability and comparison with numerical simulations, they adopt a specific Gaussian form for the interaction function $f(r)$.
  • Perform Numerical Simulations: The analytical predictions are rigorously tested against numerical simulations of systems with up to 32768 particles.

Datasets

Not discussed in the paper. The authors perform numerical simulations on model systems rather than analyzing experimental datasets.

Key Insights and Ideas

  • Random Matrix Theory as a Unifying Framework: The paper demonstrates that random matrix theory can effectively describe the complex dynamics of topologically disordered systems, offering a unified approach to understanding phenomena like high-frequency sound.
  • Origin of $p^2$ Broadening: The key insight is that when the distinction between the propagating medium and the scatterers becomes blurred (as in topologically disordered systems), the $p^2$ scaling for the spectral linewidth is recovered. This is a departure from models where scatterers are distinct entities.
  • Two-Loop Contributions Dominate Low Momentum Broadening: The analysis reveals that at low momenta, the two-loop contributions to the self-energy are dominant and lead to a broadening that scales as $p^{d-1}$ (where $d$ is the dimensionality). This is a crucial finding for explaining experimental observations.
  • "More Disordered" Characterization: The paper suggests that the $p^{d-1}$ broadening is a quantitative characterization of topologically disordered systems being "more disordered" than their ordered counterparts, as they spread energy over a wider range of frequencies.
  • Crossover Behavior: A crossover from the $p^{d-1}$ regime to a $p^{d+1}$ regime is predicted at higher densities, indicating a change in the dominant scattering mechanism.

Key Results

  • Analytical Explanation for $p^2$ Broadening: The paper provides an analytical derivation of the $p^2$ scaling of the spectral linewidth at low momentum, consistent with experimental findings in IXS.
  • Agreement with Numerical Simulations: The analytical results for both the dynamical structure factor and the density of states show excellent agreement with numerical simulations, validating the theoretical approach.
  • Prediction of a Crossover: A crossover momentum is predicted, above which the one-loop term dominates, leading to a different scaling behavior.
  • Qualitative Agreement with IXS Experiments: The authors suggest that the observed $p^2$ scaling in IXS experiments might correspond to a transient behavior in a specific momentum range.

Limitations and Future Work

  • Uncorrelated Particles: The current computation explicitly assumes uncorrelated particles, which is a significant simplification compared to the correlated nature of particles in real glasses and supercooled liquids.
  • Dependence on Interaction Choice: The coefficients ($A_1$, $A_2$) that determine the scaling behavior appear to be highly dependent on the choice of the interaction function $f(r)$, making it unclear how general the predictions are without further system-specific calculations.
  • Superposition Approximation: While particle correlations are partially addressed using the superposition approximation, its limitations are acknowledged.
  • Future Work: The authors suggest extending the computation to include particle correlations more rigorously, exploring different interaction potentials, and comparing predictions with mode-coupling theory. They also highlight the need for experimental verification across a wider range of exchanged momenta to test the proposed scaling picture.

Summary updated

Abstract

A computation of the dynamical structure factor of topologically disordered systems, where the disorder can be described in terms of euclidean random matrices, is presented. Among others, structural glasses and supercooled liquids belong to that class of systems. The computation describes their relevant spectral features in the region of the high frequency sound. The analytical results are tested with numerical simulations and are found to be in very good agreement with them. Our results may explain the findings of inelastic X-ray scattering experiments in various glassy systems.
BibTeX
@inproceedings{martnmayor2000dynamical,
  author = {V. Martı́n-Mayor and Marc Mézard and Giorgio Parisi and P. Verrocchio},
  title = {{The dynamical structure factor in topologically disordered systems}},
  booktitle = {HAL},
  year = {2000},
  doi = {10.48550/arxiv.cond-mat/0008472},
}

Loading related content...

Figure