Abstract
An equilibrated model glass-forming liquid is studied by mapping successive configurations produced by molecular dynamics simulation onto a time series of inherent structures (local minima in the potential energy). Using this “inherent dynamics” approach we find direct numerical evidence for the long held view that below a crossover temperature, $T_x$, the liquid’s dynamics can be separated into (i) vibrations around inherent structures and (ii) transitions between inherent structures, i.e., the dynamics become “dominated” by the potential energy landscape. In agreement with previous proposals, we find that $T_x$ is within the vicinity of the mode-coupling critical temperature $T_c$. We further find that near $T_x$, transitions between inherent structures occur via cooperative, stringlike rearrangements of groups of particles moving distances substantially smaller than the average interparticle distance.
Introduction
The nature of the glass transition remains one of the most challenging problems in condensed matter physics. As a liquid is supercooled, its relaxation time increases by orders of magnitude with minimal structural changes. In 1969, Martin Goldstein proposed that at low temperatures, the dynamics of a glass-forming liquid are "landscape dominated," characterized by infrequent, thermally activated transitions between local minima (basins) of the potential energy landscape (PEL). At high temperatures, the system samples the landscape freely, but as it cools, it becomes increasingly trapped in these basins. This paper provides direct numerical evidence for this crossover and identifies the temperature $T_x$ where PEL-dominated dynamics emerge.
Methods
The study utilizes a binary Lennard-Jones (BLJ) mixture (80:20 ratio of 500 particles) to simulate a glass-forming liquid. The core methodology is the "Inherent Dynamics" approach. Molecular dynamics (MD) simulations generate a "true" trajectory $R(t)$. Each configuration along this trajectory is then "quenched" via local energy minimization to find its corresponding Inherent Structure (IS), $R^I(t)$, which represents the nearest local minimum in the PEL.
By comparing the "true" dynamics (including intra-basin vibrations) with the "inherent" dynamics (representing inter-basin jumps), the researchers can isolate the structural relaxation process.
Results and Discussion
The primary finding is the emergence of a quantitative match between true and inherent dynamics as temperature decreases. This is most clearly seen in the self-intermediate scattering function, $F_s(q,t)$.
At high temperatures ($T=1.06$), the inherent dynamics ($F_s^I$) do not match the true dynamics ($F_s$). However, as the temperature drops toward $T=0.59$, the inherent scattering function perfectly captures the long-time "alpha-relaxation" of the true system. This indicates that at low temperatures, structural relaxation is entirely governed by transitions between energy basins.
The crossover temperature $T_x \approx 0.59$ is found to be close to the mode-coupling theory (MCT) critical temperature $T_c$. This suggests that the breakdown of MCT, which typically occurs at $T_c$ due to its neglect of activated processes, marks the onset of landscape-dominated hopping. Furthermore, the transitions between basins are found to be cooperative and "string-like," involving small displacements (less than $0.2 \sigma$) of groups of approximately 10 particles.
Conclusion
This study confirms the validity of the potential energy landscape paradigm for supercooled liquids. By mapping MD trajectories to inherent structures, the authors demonstrated that structural relaxation at low temperatures proceeds via discrete, cooperative jumps between energy basins. While the current system size is relatively small, the identification of string-like transitions provides a microscopic mechanism for the observed macroscopic relaxation. Future work may explore the detailed statistical mechanics of these events in larger systems.
Summary updated