Density-temperature scaling of the fragility in a model glass-former

S. Sengupta, Thomas B. Schrøder, Srikanth Sastry

TL;DR

This paper investigates how the "fragility" of a glass-forming liquid—a measure of how rapidly its dynamics slow down as it cools—depends on its density. By applying density-temperature (DT) scaling to the standard Kob-Andersen model, the authors demonstrate that kinetic fragility is remarkably insensitive to density changes, resolving previous literature contradictions through more robust, large-scale simulations.

What Problem Does This Paper Address?

In the study of glass-forming liquids, "fragility" is a central concept. It describes the "steepness" with which a liquid's viscosity or relaxation time increases as it approaches the glass transition temperature. A "strong" liquid (like silica) shows an Arrhenius-like slowdown, while a "fragile" liquid (like many polymers) slows down much more abruptly.

A major question in condensed matter physics is whether this fragility is a fixed material property or if it changes significantly with density. Recent theories, specifically "Isomorph Theory," suggest that for many liquids, dynamics depend on a specific combination of density ($\rho$) and temperature ($T$), known as Density-Temperature (DT) scaling (often expressed as $\rho^\gamma/T$). If this scaling were perfect, kinetic fragility should remain constant regardless of density. However, previous studies on the popular Kob-Andersen (KA) model reported a significant density dependence, creating a conflict between theory and simulation.

Methods

The authors performed extensive Molecular Dynamics (MD) simulations of the Kob-Andersen (KA) binary mixture, a standard model for glass-forming liquids.

Key methodological choices included:

  • System Size: Using $N=1000$ particles, significantly larger than the $N=256$ used in the earlier studies they aimed to re-evaluate.
  • Simulation Length: Running simulations for much longer durations (up to 100 times the $\alpha$-relaxation time) to ensure equilibrium at low temperatures.
  • Stickel Analysis: Instead of simple curve fitting, they used a linearized version of the Vogel-Fulcher-Tammann (VFT) equation to more accurately extract fragility values from noisy data.
  • Thermodynamic Analysis: They calculated the configurational entropy ($S_c$) to determine "thermodynamic fragility" and tested the validity of the Adam-Gibbs (AG) relation, which links dynamics to the available states in the energy landscape.

Datasets

The study utilizes the Kob-Andersen (KA) model, which is the "gold standard" benchmark in computational glass physics. It consists of an 80:20 mixture of two types of Lennard-Jones particles. The authors simulated five different number densities ranging from $\rho = 1.10$ to $1.35$.

Key Insights and Ideas

  • The "Isomorph" Connection: The authors confirm that the KA model behaves like a "strongly correlating" liquid. This means its dynamics can be collapsed onto a single master curve using the scaling variable $\rho^\gamma/T$, where $\gamma \approx 5.16$.
  • Fragility is (Mostly) Invariant: Contrary to previous reports of a 3-fold change, the authors found that kinetic fragility only changes by a factor of ~1.3 over the studied density range. This suggests that fragility is much more robust to density changes than previously thought.
  • The Low-Density Breakdown: The scaling works beautifully at high densities but deviates at low densities. The authors pinpoint the cause: at low densities and temperatures, the liquid enters "negative virial" states (negative pressure), where the attractive forces of the potential start to dominate, breaking the simple repulsive power-law scaling.
  • Resolution of Conflict: The paper argues that the strong density dependence reported in earlier work was likely a simulation artifact caused by small system sizes, shorter run times, and a narrower temperature range for fitting.

Key Results

  • Data Collapse: Both diffusivity and relaxation times showed impressive collapse onto a single curve when plotted against $\rho^\gamma/T$ (Fig. 1).
  • Scaling Exponent Consistency: The scaling exponent $\gamma$ calculated from pressure-energy correlations ($\approx 5.16$) was consistent with the exponent derived from the density dependence of the activation energy in the Adam-Gibbs relation.
  • Thermodynamic Fragility: While kinetic fragility is nearly constant, the thermodynamic fragility ($K_T$) follows a power-law dependence on density ($K_T \sim \rho^\gamma$), which is exactly what DT scaling predicts.

Limitations and Future Work

  • Model Specificity: While the KA model is a standard, it is a simple atomic mixture. Future work could investigate if these findings hold for more complex molecular glass-formers or polymers where intramolecular degrees of freedom might complicate the scaling.
  • The Low-Density Regime: The breakdown of scaling at low densities due to negative virials suggests that a more complex scaling law might be needed to bridge the gap between high-pressure liquids and low-density vapors.
  • Experimental Verification: While DT scaling has been observed in experiments, precisely measuring fragility across a wide range of densities (requiring massive pressure changes) remains a significant experimental challenge.

Summary updated

Abstract

Dynamical quantities such as the diffusion coefficient and relaxation times for some glass-formers may depend on density and temperature through a specific combination, rather than independently, allowing the representation of data over ranges of density and temperature as a function of a single scaling variable. Such a scaling, referred to as density - temperature (DT) scaling, is exact for liquids with inverse power law (IPL) interactions but has also been found to be approximately valid in many non-IPL liquids. We have analyzed the consequences of DT scaling on the density dependence of the fragility in a model glass-former. We find the density dependence of kinetic fragility to be weak, and show that it can be understood in terms of DT scaling and deviations of DT scaling at low densities. We also show that the Adam-Gibbs relation exhibits DT scaling and the scaling exponent computed from the density dependence of the activation free energy in the Adam-Gibbs relation, is consistent with the exponent values obtained by other means.
BibTeX
@article{sengupta2013density,
  author = {S. Sengupta and Thomas B. Schrøder and Srikanth Sastry},
  title = {{Density-temperature scaling of the fragility in a model glass-former}},
  journal = {The European Physical Journal E},
  year = {2013},
  doi = {10.1140/epje/i2013-13141-9},
}

Loading related content...

Figure