Potential energy landscape picture of zero-temperature avalanche criticality governing dynamics in supercooled liquids

Through molecular dynamics simulations, this study proposes that the complex slow dynamics and spatial heterogeneity of supercooled liquids can be unified under a zero-temperature avalanche criticality framework governed by the potential energy landscape, thereby explaining previously unexplained phenomena near the mode-coupling transition.

The Gist

Imagine a crowded dance floor where everyone is trying to move, but the music has slowed down to a crawl. This is what happens to a supercooled liquid (like honey that has been cooled below its freezing point but hasn't turned into a solid crystal yet). The molecules are stuck in a "glassy" state: they want to flow, but they are jammed together, moving only in tiny, sporadic bursts.

For decades, scientists have been arguing about why these molecules get stuck and how they eventually move. This paper proposes a new way to visualize the problem, using a concept called Avalanche Criticality.

Here is the story of the paper, broken down into simple analogies:

1. The Landscape of the Dance Floor (Potential Energy Landscape)

Imagine the dance floor isn't flat. It's a giant, bumpy mountain range with deep valleys and high peaks.

• The Valleys: These are comfortable spots where the molecules like to rest (low energy).
• The Peaks: These are the barriers they have to climb to move to a new spot.
• The Goal: The molecules want to find the deepest, most comfortable valley.

Usually, scientists thought the molecules just slowly "hopped" from one valley to another, like a hiker carefully stepping over rocks. But this paper suggests something more dramatic is happening.

2. The Snowball Effect (Avalanches)

The authors propose that movement doesn't happen one molecule at a time. Instead, it happens in avalanches.

Imagine a snowball sitting on a steep, snowy hill.

• The Trigger: One tiny snowflake falls (a thermal fluctuation).
• The Chain Reaction: That tiny flake nudges a small patch of snow, which triggers a small slide, which knocks over a larger patch, which causes a massive avalanche.
• The Result: A huge chunk of the hill moves at once, even though it started with a tiny nudge.

In the supercooled liquid, the "snow" is the molecules. When one molecule tries to move, it pushes its neighbors, who push theirs, creating a chain reaction. The paper shows that these "molecular avalanches" are the key to understanding how the liquid flows.

3. The "Zero-Temperature" Secret

The most surprising finding is about when these avalanches happen.

• The Theory: The authors suggest that the rules governing these avalanches are actually set at absolute zero (the coldest possible temperature).
• The Analogy: Think of the dance floor as a giant, frozen sculpture. Even though the dancers are moving (because it's warm), the shape of the sculpture (the rules of how they can move) was carved out when it was frozen solid. The "critical point" where the system changes its behavior is effectively at absolute zero, even though we are observing it at higher temperatures.

4. The "Jam" and the "Un-jam"

The paper investigates what happens as the liquid gets colder and colder, approaching a specific temperature called the Mode-Coupling Transition (MCT).

• The Traffic Jam: At high temperatures, the dance floor is crowded with many small avalanches happening at once. They bump into each other, like cars in a traffic jam. This limits how big any single avalanche can get.
• The Clearing: As the temperature drops, the "traffic" clears up. The avalanches become fewer but can grow much larger because they don't bump into each other.
• The Limit: However, the paper finds that even when the traffic clears, the avalanches stop growing after a certain size. It's as if the dance floor has a maximum size for a single group dance. Once the system hits this limit, the "avalanche rules" stop working, and something else takes over to control the movement.

5. The "Unstable" Dancers

To prove this, the scientists looked at the "unstable modes" of the system.

• The Metaphor: Imagine a wobbly tower of blocks. Some blocks are precariously balanced. If you nudge them, the whole tower might fall.
• The Finding: The researchers found that as the liquid cools, these "wobbly blocks" (unstable modes) stop being spread out across the whole dance floor. Instead, they become localized—they cluster together in small, tight groups.
• Why it matters: This clustering explains why the "traffic jam" (the avalanche) stops growing. The instability is no longer a global property of the whole system; it's trapped in small pockets.

The Big Picture Conclusion

This paper unifies several confusing observations about supercooled liquids:

1. Why they slow down: It's due to the growth of these molecular avalanches.
2. Why they stop slowing down: At a certain point, the avalanches hit a size limit (saturation), and the "avalanche rules" break down.
3. The Role of Structure: The shape of the energy landscape (the mountains and valleys) dictates these rules.

In short: The authors have shown that the slow, sluggish movement of supercooled liquids is governed by a "zero-temperature" rulebook of avalanches. However, just like a snowstorm that eventually stops growing because the hill runs out of snow, these avalanches hit a limit as the liquid gets very cold, suggesting that a different mechanism takes over to explain the final stages of the glass transition.

This helps scientists understand that the "glass transition" isn't just one simple event, but a complex dance between different physical mechanisms, all choreographed by the hidden shape of the energy landscape.

Technical Summary

1. Problem Statement

Supercooled liquids exhibit dramatic slowing of dynamics and the emergence of dynamical heterogeneity (DH) (spatially heterogeneous regions of mobile and immobile particles) as they are cooled toward the glass transition. Despite extensive study, the microscopic origin of DH remains debated.

• Competing Theories: Existing frameworks include static structural theories (e.g., Random First-Order Transition theory, frustration-limited domains) which rely on growing static length scales, and dynamic theories (e.g., Mode-Coupling Theory, kinetically constrained models).
• The Gap: Recent work by Tahaei et al. (Ref. [66]) proposed that DH in supercooled liquids is governed by zero-temperature avalanche criticality, similar to plastic deformation in amorphous solids under shear. However, this hypothesis lacked comprehensive validation in particle-based simulations, particularly regarding the connection to the Potential Energy Landscape (PEL) and the explanation of specific anomalies near the Mode-Coupling Theory (MCT) transition temperature ($T_{MCT}$), such as the saturation of dynamical susceptibility and the localization of unstable modes.

2. Methodology

The authors performed extensive Molecular Dynamics (MD) simulations of the three-dimensional Kob-Andersen Model (KAM), a prototypical binary Lennard-Jones mixture representing a fragile glass-former.

• Simulation Parameters:
  • System sizes ($N$): 200 to 1500 particles.
  • Temperature range ($T$): 0.41 to 1.0 (covering the supercooled regime down to near $T_{MCT} \approx 0.435$).
  • Ensemble: Canonical (NVT) using the Nosé-Hoover thermostat.
  • Crystallization Control: Samples were screened for crystallization, but the authors noted that excluding crystallized samples did not qualitatively alter the main results within the studied range.
• Key Observables:
  1. Structural Relaxation: Measured via the overlap function $Q(t)$.
  2. Dynamical Heterogeneity: Quantified by the dynamical susceptibility $\chi_4(t)$, specifically its peak value $\chi_4^*$ and peak time $\tau_4$.
  3. PEL Characterization:
     • Inherent Structures (IS): Obtained via energy minimization (FIRE algorithm). Analyzed their vibrational density of states (VDOS).
     • Saddle Points: Configurations where the gradient of potential energy is zero but the Hessian has negative eigenvalues. Analyzed the number of unstable modes ($N^\dagger_{saddle}$) and the spatial localization of their eigenvectors.
     • Energy Levels: Statistical analysis of the inherent structure energy per particle ($e_{IS}$).

3. Key Contributions and Results

A. Validation of Zero-Temperature Avalanche Criticality

The study confirms that the temperature ($T$) and system-size ($N$) dependence of dynamical observables can be unified under a zero-temperature avalanche criticality framework.

• Scaling Collapse: The peak dynamical susceptibility $\chi_4^*$ exhibits finite-size scaling collapse consistent with the ansatz:
  $$\chi_4^* \sim T^{-\gamma} \quad \text{and} \quad \xi \sim T^{-\nu}$$
  where $\xi$ is the critical correlation length.
• Critical Exponents: The authors determined critical exponents independently of $\chi_4^*$:
  • $\nu \approx 3.2$ (from the correlation length).
  • $\gamma \approx 6.0$ (derived from the scaling relation $\nu d_f = \gamma$ and the fraction of unstable modes).
  • $d_f \approx 1.9$ (fractal dimension of avalanches).
• Consistency: The successful scaling collapse validates that DH is driven by thermally activated avalanches with a critical point at $T=0$.

B. Connection to Potential Energy Landscape (PEL)

The authors provide a quantitative characterization of the PEL to explain the origin of this criticality:

1. Vibrational Density of States (VDOS):
   • For stable inherent structures, the low-frequency VDOS follows a non-Debye law $D(\omega) \sim \omega^{6.5}$.
   • The coefficient $A_4$ of this power law decreases monotonically below a threshold temperature $T_{ava} \approx 0.6$, indicating increased stability and a reduction in the density of quasi-localized vibrational modes (QLVs) that trigger avalanches. This defines the onset of the critical regime.
2. Localization of Unstable Modes:
   • The authors analyzed the spatial localization of unstable modes (negative eigenvalues) at saddle points.
   • Using a fractal scaling assumption ($N_p \sim L^{d_f}$), they identified a mobility edge ($\lambda_e$).
   • Key Finding: As $T$ approaches $T_{MCT}$, the mobility edge vanishes ($|\lambda_e| \to 0$), meaning all unstable modes become spatially localized. This coincides with the saturation of the dynamical susceptibility.
3. Inherent Structure Energy:
   • The mean energy $\langle e_{IS} \rangle$ decreases linearly with $1/T$ below $T_{ava}$.
   • The standard deviation $\Delta e_{IS}$ shows a qualitative change (turning upward) near $T_{ava}$ and saturates near $T_{MCT}$, suggesting the system reaches the bottom of a "metabasin."

C. Unified Explanation of MCT Anomalies

The paper proposes a unified PEL-based picture explaining two previously unexplained phenomena near $T_{MCT}$:

1. Saturation of Dynamical Susceptibility: The correlation length $\xi$ stops growing and saturates near $T_{MCT}$. The authors interpret this as an "unjammed" state of Soft Avalanche Quasiparticles (SAQs). At high $T$, avalanches interfere (jammed), limiting size. At low $T$, they are sparse, but a physical upper bound ($\xi_{max}$) exists, likely corresponding to the size of a metabasin.
2. Localization of Unstable Modes: The simultaneous localization of all unstable modes at $T_{MCT}$ is explained by the same mechanism: the unjamming of SAQs. When avalanches are sparse and non-interfering, the hybridization of modes (which creates extended states) ceases, leaving only localized modes.

4. Significance and Implications

• Unified Framework: The study successfully bridges the gap between dynamic facilitation/avalanche theories and static energy landscape theories. It demonstrates that the slow dynamics of supercooled liquids are governed by the topology of the PEL, specifically the statistics of saddle points and the hierarchical structure of metabasins.
• Resolution of MCT Paradox: It provides a physical mechanism for the breakdown of avalanche criticality near $T_{MCT}$. The criticality is not a property of the ideal glass transition at $T=0$ in the deeply supercooled regime; rather, a new mechanism (likely related to local activation energies or hopping) dominates below $T_{MCT}$.
• System Dependence: The authors note that this picture may not be universal. Comparisons with "swap" models (which lack attractive interactions) suggest that the temperature dependence of stability and the existence of an avalanche-critical regime may differ based on interaction potentials.
• Future Directions: The work highlights the need to determine the activation energy more accurately (addressing conflicting literature) and to apply these analyses to systems with suppressed crystallization to fully understand the deeply supercooled regime.

In conclusion, Oyama et al. provide strong numerical evidence that zero-temperature avalanche criticality governs the dynamics of supercooled liquids in the regime $T_{MCT} < T < T_{ava}$, driven by the hierarchical structure of the potential energy landscape, while identifying a distinct physical regime below $T_{MCT}$ where this criticality saturates.