Quantitative verification of the energy-landscape barrier mechanism for super-Arrhenius relaxation below the threshold temperature in fragile glass-formers

Grigera, Cavagna, Giardina and Parisi [Phys. Rev. Lett. 88, 055502 (2002)] proposed a geometric scenario for the dynamic glass transition of fragile supercooled liquids: the average instability index of the stationary points of the potential energy, plotted against their energy density, is a well-defined function $k(u)$ that vanishes at a threshold energy $u_{\rm th}$; below the temperature $T_{\rm th}$ at which the equilibrium bare energy crosses $u_{\rm th}$, diffusion must proceed by activated barrier hopping; and the barriers $\Delta U(u)$ connecting adjacent minima grow as the minima deepen, with $\Delta U(u_{\rm th}) \approx 10\,k_B T_{\rm th}$ — predicting a sharp super-Arrhenius increase of the relaxation time below $T_{\rm th}$, consistent with the mode-coupling crossover. This prediction was not demonstrated in the source study, which also left open whether the small fraction of barriers much smaller than $k_B T_{\rm th}$ controls non-activated off-equilibrium dynamics, whether the landscape statistics persist quantitatively at larger system sizes, and whether the mechanism generalizes across interaction models and fragility classes. The problem remains open: the barrier-growth mechanism has been supported in Lennard-Jones mixtures, the sharp 'saddles disappear' reading has been revised into a mode-localization picture, and equilibrium relaxation can now be measured down to the experimental $T_g$ — but the central quantitative link between measured relaxation times below the crossover and measured barrier statistics within a single system remains missing, and 2022–2026 studies explicitly list the reliable determination of landscape activation energies and the characterization of activated dynamics in the deeply supercooled regime as important open problems.

Background

When a liquid is supercooled below its crystallization temperature without crystallizing, its structural relaxation time grows by many orders of magnitude before the laboratory glass transition is reached at the conventional temperature $T_g$ (where the viscosity reaches $10^{13}$ poise). In so-called strong liquids this growth is approximately Arrhenius, $\tau \propto \exp(E/k_B T)$ with a roughly temperature-independent activation energy; in fragile liquids (in Angell's classification) the growth is much faster than Arrhenius ('super-Arrhenius'), steepening over a narrow temperature window. Explaining the sharp fragile slowdown is a central open problem of liquid-state physics. A useful theoretical reference point is mode-coupling theory (MCT), a first-principles theory of liquid dynamics that predicts a dynamical arrest at a crossover temperature $T_c$; in real (finite-dimensional) systems this arrest is rounded into a sharp but smooth crossover.

A complementary description is the potential energy landscape (PEL): the potential energy as a function of all particle coordinates. Goldstein proposed in 1969 that flow in supercooled liquids occurs by activated transitions between local minima of this high-dimensional surface, and Stillinger and Weber formalized the decomposition of configurations into 'inherent structures' — the local minima reached by steepest descent from instantaneous configurations. Sastry, Debenedetti and Stillinger later showed that distinct dynamical regimes of a glass-former appear directly in the statistics of its inherent structures. Stationary points of the PEL with unstable directions ('saddles') are characterized by their instability index, the number of negative eigenvalues of the Hessian matrix at the point. In mean-field glass models such as the p-spin model, a threshold energy separates a low-energy minima-dominated region from a high-energy saddle-dominated one, and dynamics becomes trapped near the threshold. Angelani and co-workers and Broderix and co-workers found analogous behavior in Lennard-Jones liquids: the average index of sampled stationary points, plotted against their energy, extrapolates to zero at a threshold energy.

Grigera, Cavagna, Giardina and Parisi studied a 50:50 binary soft-sphere mixture ($V_{ij} \propto (\sigma_i + \sigma_j)^{12}/r^{12}$, unit density) with $N = 70$ particles ($N = 140$ for checks), which can be equilibrated below the mode-coupling crossover using the swap Monte Carlo algorithm of Grigera and Parisi, in which particles of different sizes are exchanged to accelerate relaxation. They found a well-defined average index density $k(u)$ vanishing at a threshold energy $u_{\rm th} = 1.75 \pm 0.01$, and located the temperature $T_{\rm th} = 0.242 \pm 0.012$ at which the 'bare' equilibrium energy $u_b(T) = u_{\rm eq}(T) - (3/2) k_B T$ (with the vibrational contribution removed) crosses $u_{\rm th}$; below $T_{\rm th}$ the system is confined to the minima-dominated region and must diffuse by barrier hopping. $T_{\rm th}$ is close to the independently estimated dynamical crossover $T_c \approx 0.24$ (standard value 0.226). By following the gradient along the unstable eigenvector of index-1 saddles they sampled connected minimum–saddle–minimum triples (1652 barriers) and found that the average barrier $\Delta U(u)$ grows with decreasing minimum energy $u$, with $\Delta U(u_{\rm th}) \approx 2.2 \approx 10\,k_B T_{\rm th}$. Hence, at the temperature where activation first becomes the only mechanism of diffusion, barriers already far exceed the thermal energy — proposed as the distinguishing feature of very fragile liquids — and, because typical barriers continue to grow on cooling while the thermal energy decreases, the relaxation time was predicted to increase super-Arrhenius below $T_{\rm th}$. A real-space analysis showed that activated events involve a few particles exchanging nearest-neighbor positions while many others move slightly to make way.

The source study demonstrated none of this dynamically: relaxation times below $T_{\rm th}$ were not measured (with the physical, non-swap dynamics the system falls out of equilibrium above $T_{\rm th}$), and the barrier statistics were obtained for small systems only. Three further issues were flagged: the distribution of $\Delta U$ at fixed $u$ is broad, with a small fraction of barriers much smaller than $k_B T_{\rm th}$, whose possible role in non-activated off-equilibrium dynamics was left as an open suggestion; the $N = 140$ system could not be equilibrated below $u_{\rm th}$; and the mechanism had been probed mainly in Lennard-Jones and soft-sphere models, leaving its generality across interaction models and fragility classes untested. Verifying the prediction requires measuring, in one and the same system, both equilibrium relaxation times below the crossover (accelerated equilibration combined with physical dynamics) and the statistics of connected barriers — a program that later literature has approached but has not completed.

Problem Statement

Within the geometric approach to the dynamic glass transition formulated by Grigera, Cavagna, Giardina and Parisi (2002), establish — or refute — by direct quantitative measurement whether the barrier statistics of the potential energy landscape controls the relaxation of fragile glass-formers below the landscape crossover temperature $T_{\rm th}$. Specifically:

1. (Core test) In a fragile glass-forming model whose stationary-point statistics are measured — the index–energy function $k(u)$ (the average instability index, i.e. the number of unstable directions of a stationary point, as a function of its energy density) with its zero $u_{\rm th}$, and the statistics of barrier heights $\Delta U(u)$ for connected minimum–saddle–minimum triples — measure equilibrium structural relaxation times $\tau_\alpha(T)$ systematically below $T_{\rm th}$, with equilibrium established by an accelerated protocol and relaxation measured with the physical dynamics, over a broad temperature range relative to $T_{\rm th}$; determine whether the observed temperature dependence, predicted to be super-Arrhenius, is quantitatively accounted for by the growth of the typical barriers $\Delta U(u)$ as the system occupies deeper minima, including the functional form of $\tau_\alpha(T)$ and its parameters, with the energetic-barrier attribution disentangled from entropic or pathway effects, that is, whether the measured barrier growth accounts for the observed temperature dependence or whether entropic or pathway effects are needed to explain it, with the conclusion supported by the measured barrier statistics.

2. (Rare fast barriers) Determine the dynamical role of the broad distribution of barrier heights at fixed minima energy — in particular whether the small fraction of barriers much smaller than $k_B T_{\rm th}$ dominates non-activated, off-equilibrium (aging) relaxation in some regimes, as suggested in the source study.

3. (Robustness and generality) Establish whether the reported numerical values and the detailed $\Delta U(u)$ dependence persist quantitatively in substantially larger systems approaching the thermodynamic limit, and whether the existence of a threshold energy and the ensuing barrier growth occur across different interaction models and different fragility classes, from fragile to strong glass-formers. The size question is a finite-size comparison of the measured $u_{\rm th}$ and $\Delta U(u)$ at more than one larger $N$, together with their drift with system size; the generality question is the same measurement carried to at least one model outside the source's soft-sphere mixture and to at least one representative of each fragility class named.

The three questions form one objective: question 1 is the core quantitative test of the geometric scenario, and questions 2 and 3 are the robustness and generality conditions under which that test is meaningful. The single scientific objective uniting them is to decide whether the landscape barrier mechanism quantitatively explains the dynamic glass transition of fragile liquids.

References

1. T. S. Grigera, A. Cavagna, I. Giardina, G. Parisi, 'Geometric Approach to the Dynamic Glass Transition', Physical Review Letters 88, 055502 (2002). DOI: 10.1103/PhysRevLett.88.055502, https://doi.org/10.1103/PhysRevLett.88.055502 (preprint arXiv:cond-mat/0107198, https://arxiv.org/abs/cond-mat/0107198)
2. T. S. Grigera, 'Geometrical properties of the potential energy of the soft-sphere binary mixture', The Journal of Chemical Physics 124, 064502 (2006). DOI: 10.1063/1.2151899, https://doi.org/10.1063/1.2151899 (preprint arXiv:cond-mat/0509301, https://arxiv.org/abs/cond-mat/0509301)
3. C. Cammarota, A. Cavagna, G. Gradenigo, T. S. Grigera, P. Verrocchio, 'Evidence for a spinodal limit of amorphous excitations in glassy systems', Journal of Statistical Mechanics: Theory and Experiment 2009, L12002 (2009). DOI: 10.1088/1742-5468/2009/12/L12002, https://doi.org/10.1088/1742-5468/2009/12/L12002 (preprint arXiv:0904.1522, https://arxiv.org/abs/0904.1522, titled 'Surface tension fluctuations and a new spinodal point in glass-forming liquids')
4. B. Doliwa, A. Heuer, 'Energy barriers and activated dynamics in a supercooled Lennard-Jones liquid', Physical Review E 67, 031506 (2003). DOI: 10.1103/PhysRevE.67.031506, https://doi.org/10.1103/PhysRevE.67.031506 (preprint arXiv:cond-mat/0209139, https://arxiv.org/abs/cond-mat/0209139)
5. B. Doliwa, A. Heuer, 'Hopping in a supercooled Lennard-Jones liquid: Metabasins, waiting time distribution, and diffusion', Physical Review E 67, 030501 (2003). DOI: 10.1103/PhysRevE.67.030501, https://doi.org/10.1103/PhysRevE.67.030501 (preprint arXiv:cond-mat/0205283, https://arxiv.org/abs/cond-mat/0205283)
6. D. Coslovich, G. Pastore, 'Understanding fragility in supercooled Lennard-Jones mixtures. II. Potential energy surface', The Journal of Chemical Physics 127, 124505 (2007). DOI: 10.1063/1.2773720, https://doi.org/10.1063/1.2773720 (preprint arXiv:0705.0629, https://arxiv.org/abs/0705.0629)
7. D. Coslovich, G. Pastore, 'Dynamics and energy landscape in a tetrahedral network glass-former: Direct comparison with models of fragile liquids', Journal of Physics: Condensed Matter 21, 285107 (2009). DOI: 10.1088/0953-8984/21/28/285107, https://doi.org/10.1088/0953-8984/21/28/285107 (preprint arXiv:0906.1684, https://arxiv.org/abs/0906.1684)
8. L. Berthier, J. P. Garrahan, 'Real space origin of temperature crossovers in supercooled liquids', Physical Review E 68, 041201 (2003). DOI: 10.1103/PhysRevE.68.041201, https://doi.org/10.1103/PhysRevE.68.041201 (preprint arXiv:cond-mat/0306469, https://arxiv.org/abs/cond-mat/0306469)
9. D. Coslovich, A. Ninarello, L. Berthier, 'A localization transition underlies the mode-coupling crossover of glasses', SciPost Physics 7, 077 (2019). DOI: 10.21468/SciPostPhys.7.6.077, https://doi.org/10.21468/SciPostPhys.7.6.077 (preprint arXiv:1811.03171, https://arxiv.org/abs/1811.03171)
10. M. Shimada, D. Coslovich, H. Mizuno, A. Ikeda, 'Spatial structure of unstable normal modes in a glass-forming liquid', SciPost Physics 10, 001 (2021). DOI: 10.21468/SciPostPhys.10.1.001, https://doi.org/10.21468/SciPostPhys.10.1.001 (preprint arXiv:2009.07972, https://arxiv.org/abs/2009.07972)
11. Y. Nishikawa, M. Ozawa, A. Ikeda, P. Chaudhuri, L. Berthier, 'Relaxation Dynamics in the Energy Landscape of Glass-Forming Liquids', Physical Review X 12, 021001 (2022). DOI: 10.1103/PhysRevX.12.021001, https://doi.org/10.1103/PhysRevX.12.021001 (preprint arXiv:2106.01755, https://arxiv.org/abs/2106.01755)
12. C. Scalliet, B. Guiselin, L. Berthier, 'Thirty Milliseconds in the Life of a Supercooled Liquid', Physical Review X 12, 041028 (2022). DOI: 10.1103/PhysRevX.12.041028, https://doi.org/10.1103/PhysRevX.12.041028 (preprint arXiv:2207.00491, https://arxiv.org/abs/2207.00491)
13. N. Oyama, Y. Hara, T. Kawasaki, K. Kim, 'Potential energy landscape picture of zero-temperature avalanche criticality governing dynamics in supercooled liquids', arXiv:2604.03580 (2026), https://arxiv.org/abs/2604.03580
14. P. Das, S. Sastry, 'Crossover in dynamics in the Kob-Andersen binary mixture glass-forming liquid', Journal of Non-Crystalline Solids: X 14, 100098 (2022). DOI: 10.1016/j.nocx.2022.100098, https://doi.org/10.1016/j.nocx.2022.100098
15. K. González-López, E. Lerner, 'An energy-landscape-based crossover temperature in glass-forming liquids', The Journal of Chemical Physics 153, 241101 (2020). DOI: 10.1063/5.0034719, https://doi.org/10.1063/5.0034719 (preprint arXiv:2010.07153, https://arxiv.org/abs/2010.07153)
16. D. Coslovich, A. Ikeda, 'Revisiting the single-saddle model for the β-relaxation of supercooled liquids', The Journal of Chemical Physics 156, 094503 (2022). DOI: 10.1063/5.0083173, https://doi.org/10.1063/5.0083173 (preprint arXiv:2202.08016, https://arxiv.org/abs/2202.08016)
17. L. Angelani, R. Di Leonardo, G. Ruocco, A. Scala, F. Sciortino, 'Saddles in the energy landscape probed by supercooled liquids', Physical Review Letters 85, 5356 (2000). DOI: 10.1103/PhysRevLett.85.5356, https://doi.org/10.1103/PhysRevLett.85.5356
18. K. Broderix, K. K. Bhattacharya, A. Cavagna, A. Zippelius, I. Giardina, 'Energy Landscape of a Lennard-Jones Liquid: Statistics of Stationary Points', Physical Review Letters 85, 5360 (2000). DOI: 10.1103/PhysRevLett.85.5360, https://doi.org/10.1103/PhysRevLett.85.5360
19. M. Goldstein, 'Viscous Liquids and the Glass Transition: A Potential Energy Barrier Picture', The Journal of Chemical Physics 51, 3728 (1969). DOI: 10.1063/1.1672587, https://doi.org/10.1063/1.1672587
20. F. H. Stillinger, T. A. Weber, 'Hidden structure in liquids', Physical Review A 25, 978 (1982). DOI: 10.1103/PhysRevA.25.978, https://doi.org/10.1103/PhysRevA.25.978
21. S. Sastry, P. G. Debenedetti, F. H. Stillinger, 'Signatures of distinct dynamical regimes in the energy landscape of a glass-forming liquid', Nature 393, 554 (1998). DOI: 10.1038/31189, https://doi.org/10.1038/31189
22. T. S. Grigera, G. Parisi, 'Fast Monte Carlo algorithm for supercooled soft spheres', Physical Review E 63, 045102 (2001). DOI: 10.1103/PhysRevE.63.045102, https://doi.org/10.1103/PhysRevE.63.045102
23. C. A. Angell, 'Perspective on the glass transition', Journal of Physics and Chemistry of Solids 49, 863 (1988). DOI: 10.1016/0022-3697(88)90002-9, https://doi.org/10.1016/0022-3697(88)90002-9