When a complex system like a spin glass is cooled, it does not settle into a single stable state. Instead, it enters a state of perpetual "aging," where its internal configuration slowly evolves toward lower energy states over timescales that can last for years. For decades, physicists observed two baffling phenomena during this process: rejuvenation (where a small temperature drop makes the system act as if it has just started aging from scratch) and memory (where returning to the original temperature restores the system’s previous aged state).
Recent research has identified the microscopic driver of these behaviors: Temperature Chaos (TC). This is a phenomenon where even an infinitesimal change in temperature radically reorganizes the entire equilibrium landscape of the system.
The Fragility of the Energy Landscape
In simple materials, lowering the temperature merely narrows the range of possible states. In glassy systems, however, the "energy landscape"—the map of all possible configurations and their energies—is rugged and frustrated. Florent Krzakala and others have shown through "state-following" formalisms that as temperature shifts, the relative stability of different states reshuffles unpredictably.
A state that was a deep, stable minimum at 10 Kelvin might become a high-energy peak at 9 Kelvin. This is not a gradual shift; L. A. Fernández, V. Martı́n-Mayor, and Giorgio Parisi demonstrated that these transitions resemble "level crossings" in quantum mechanics—rare, sudden events where a previously irrelevant configuration abruptly becomes the new global minimum.
Why Chaos is Hard to See
If temperature chaos is so fundamental, why was it historically difficult to simulate? Olivier Martin proposed that TC is driven primarily by entropy fluctuations rather than energy alone. These fluctuations scale with the system size ($N$). In small-scale Monte Carlo simulations, the "chaos" is often too faint to detect. It only becomes dominant when the system is large enough ($N \gtrsim 1000$) for these entropic shifts to overwhelm the local energy barriers.
Furthermore, Federico Ricci‐Tersenghi noted that the final state of a glass depends heavily on its "cooling history." If a system exhibits temperature chaos, it cannot simply "track" a metastable state as it cools; the landscape shifts so drastically that the system is constantly forced to find new paths.
Rejuvenation and the Multi-Scale Paradox
The most striking evidence for temperature chaos is the coexistence of rejuvenation and memory. When a system is cooled from $T_1$ to $T_2$, it appears to "rejuvenate"—the aging process resets because the configuration it reached at $T_1$ is no longer a minimum at $T_2$. It must start exploring the new landscape from scratch.
However, if the system is warmed back to $T_1$, it "remembers" its previous state. The Janus Collaboration resolved this paradox by showing that these effects are governed by multiple length scales. Using the Janus II supercomputer, Enrico Calore and colleagues demonstrated that:
- Small scales reorganize quickly to match the new temperature (causing rejuvenation).
- Large scales remain "frozen" or dynamically inactive during the temperature dip.
- When the temperature returns to $T_1$, these large-scale structures act as a template, allowing the system to recover its previous configuration (memory).
Bridging Theory and Experiment
For years, a gap existed: temperature chaos was a theoretical property of systems at equilibrium, but real glasses are always off-equilibrium. Work by Enrico Calore and I. Paga finally proved that temperature chaos exists in off-equilibrium dynamics.
By defining new coefficients to quantify memory, researchers have linked these macroscopic "memory" effects directly to the growth of microscopic correlation lengths. This confirms that the bizarre thermal history of a spin glass is not just a surface-level curiosity, but a direct window into the chaotic, ever-shifting nature of frustrated energy landscapes.