When a liquid is cooled rapidly enough to avoid crystallization, it becomes "supercooled," and its viscosity increases by many orders of magnitude over a narrow temperature range. To understand why this happens, researchers look past the chaotic motion of individual molecules and instead analyze the Potential Energy Landscape (PEL)—a high-dimensional "topography" where the coordinates represent the positions of all particles in the system and the height represents the system's total potential energy.
Stripping Thermal Noise with Inherent Structures
In a hot liquid, particles have enough kinetic energy to zip across this landscape, ignoring the small bumps and dips in the terrain. However, as the liquid cools, these features begin to dictate the system's behavior. The challenge for researchers is that constant thermal vibrations "blur" the underlying topography, making it difficult to see the stable configurations that actually govern the glass transition.
To solve this, Francesco Sciortino and P. Tartaglia popularized a technique called inherent structure mapping. As detailed in The phase space structure of supercooled liquids, this involves taking a "snapshot" of a liquid and computationally "quenching" it—removing all kinetic energy to let the system slide down the nearest slope to a local potential energy minimum.
These minima are the inherent structures. By mapping a continuous trajectory of a liquid onto a discrete sequence of these minima, researchers like Thomas B. H. Schroeder, Jeppe C. Dyre, and Sharon C. Glotzer showed that we can separate fast, irrelevant vibrations from the slow, structural "hops" that characterize the approach to a glass (Crossover to potential energy landscape dominated dynamics in a model glass-forming liquid).
The Onset of Landscape Dominated Dynamics
The PEL approach reveals a specific "onset temperature" ($T_s$) where the physics of a liquid fundamentally changes. Above $T_s$, the liquid is a "normal" fluid; its particles move freely, and the underlying energy minima have little influence on its flow.
As the system cools below $T_s$, it becomes "landscape-influenced." As described in Onset Temperature of Slow Dynamics in Glass Forming Liquids, the system begins to spend most of its time vibrating within a single "basin" of the landscape, only occasionally jumping over a saddle point into a neighboring basin. This transition marks the beginning of the dramatic slowdown associated with glass formation.
Recent work by Marco Baity-Jesi, Giulio Biroli, and David R. Reichman suggests that this navigation isn't just about jumping over high walls. Instead, the system engages in entropic selection, searching for rare, nearly barrierless pathways to move between "metabasins"—clusters of similar inherent structures (Revisiting the concept of activation in supercooled liquids).
Linking Geometry to Thermodynamics
The power of the inherent structure formalism is that it provides a bridge between the microscopic geometry of the landscape and macroscopic thermodynamic properties like configurational entropy.
Configurational entropy is essentially a measure of how many different "valleys" (inherent structures) are available for the system to explore. In Fragility, Configurational Entropy and the Potential Energy Landscape of Glass Forming Liquids, researchers demonstrated that as a liquid cools, the number of accessible basins drops sharply. This reduction in "navigational choices" is what causes the viscosity to skyrocket.
This framework also explains fragility—the rate at which a liquid's dynamics slow down. "Fragile" liquids have landscapes where the energy of the basins varies wildly, while "strong" liquids (like silica) have more uniform landscapes. For example, Francis W. Starr and colleagues used PEL analysis to show that water's unique properties in the supercooled regime stem from its landscape forcing it into a highly ordered, tetrahedral network (Supercooled Water: Dynamics, Structure and Thermodynamics).
Universality Across Disordered Systems
The utility of the potential energy landscape extends beyond liquids to other disordered systems, most notably spin glasses. By applying inherent structure mapping to vector spin glasses, Giorgio Parisi demonstrated how the complexity of the landscape depends on the number of components in the system (Infinite-temperature Inherent Structures in m-component Spin Glasses).
Whether the system is a collection of molecules in a supercooled liquid or a lattice of magnetic spins, the core insight remains the same: the transition from a fluid or disordered state to a "frozen" one is a journey through a rugged, high-dimensional landscape, where the geometry of the valleys determines the ultimate fate of the system.