Phase transition

Lore & Background

Phase transitions describe the physical process by which a medium changes from one state to another, most commonly between the four states of matter: solid, liquid, gas, and plasma. At the precise point of transition, such as the boiling point, the two phases involved possess identical free energies and are equally probable; below this point one phase is more stable, and above it the other dominates. While typical transitions occur when temperature or pressure changes, a system can sometimes be brought past a transition point without changing phase, resulting in a metastable state—such as superheating or supercooling—that is less stable than the equilibrium phase but not unstable. Beyond changes between states of matter, phase transitions also encompass structural changes within solids, where a substance alters its crystal structure without changing chemical composition (allotropy in elements, polymorphism in compounds). Examples include transformations from one crystal structure to another, from crystalline to amorphous solid, or between amorphous forms. Magnetic phase transitions occur when materials shift between different magnetic orderings, such as the ferromagnetic-to-paramagnetic transition at the Curie point, or between commensurate and incommensurate magnetic structures. In mixtures and solutions, transitions are more complex than in pure compounds; mixtures may exhibit congruent melting at a single temperature or a range of temperatures where solid and liquid coexist. Three-phase transformations include eutectic, eutectoid, and peritectic processes, where a single phase decomposes into two solid phases or a solid and a liquid upon heating or cooling.

Reader's Guide

Phase transitions are fundamental to understanding the behavior of matter under varying conditions. They encompass not only changes between solid, liquid, gas, and plasma but also structural transitions within solids (allotropy, polymorphism), magnetic ordering changes (e.g., ferromagnetic to paramagnetic at the Curie point), and complex transitions in mixtures (eutectic, peritectic, spinodal decomposition). The Ehrenfest classification distinguishes first-order transitions, where the first derivative of free energy is discontinuous (e.g., density change at boiling), from second-order transitions, where the first derivative is continuous but a second derivative is discontinuous (e.g., ferromagnetic transition). Phase transitions also occur in non-thermodynamic systems, such as quantum phase transitions and topological transitions, where parameters other than temperature drive the change. Their study is essential in materials science, condensed matter physics, and cosmology.

Did You Know?

• During a phase transition, certain properties of the medium change as a result of changes in external conditions such as temperature or pressure.
• Metastable states, like superheating and supercooling, do not appear on usual phase diagrams.
• The martensitic transformation in carbon steel is a model for displacive phase transformations.
• Isotope fractionation occurs during phase transitions, with heavier isotopes enriching in the liquid phase during condensation.

From Bernoulli's Insight to a New Discipline

In 1738, Swiss physicist Daniel Bernoulli published Hydrodynamica, a work that planted the seeds for what would eventually become statistical mechanics. His central argument—that gases are composed of vast numbers of molecules hurtling in every direction, that their collisions with surfaces produce the pressure we measure, and that what we call heat is nothing more than the kinetic energy of their motion—remains a foundational explanation to this day. Statistical mechanics grew out of classical thermodynamics, which had long described macroscopic quantities like temperature, pressure, and heat capacity. What statistical mechanics added was a microscopic explanation: these bulk properties emerge from parameters that fluctuate around average values and are described by probability distributions. The field's purpose is to clarify the properties of matter in aggregate by connecting them to the physical laws that govern atomic motion. Its reach extends far beyond physics, finding applications in biology, neuroscience, computer science, information theory, and even sociology.

Three Pioneers Who Built a Framework

The discipline is generally credited to three physicists whose contributions interlocked into a coherent whole. In 1859, Scottish physicist James Clerk Maxwell, inspired by reading Rudolf Clausius's paper on molecular diffusion, formulated the Maxwell distribution of molecular velocities—the first statistical law ever established in physics. This distribution specified the proportion of molecules occupying any given velocity range. Maxwell also provided the first mechanical argument showing that molecular collisions drive temperatures toward equality, establishing a natural tendency toward equilibrium. Five years later, a young student in Vienna named Ludwig Boltzmann encountered Maxwell's paper and devoted much of his career to extending the subject. His body of work, spanning roughly 2,000 pages in the proceedings of the Vienna Academy and other societies, covered the statistical interpretation of thermodynamics, the H-theorem, transport theory, thermal equilibrium, and the equation of state of gases. Boltzmann introduced the equilibrium statistical ensemble and was the first to investigate non-equilibrium statistical mechanics. The American mathematical physicist Josiah Willard Gibbs, who coined the term "statistical mechanics" in 1884, completed the picture with his 1902 book Elementary Principles in Statistical Mechanics, which formalized the field as a fully general approach applicable to all mechanical systems.

Bridging the Gap Between Law and Experience

Ordinary mechanics—whether classical or quantum—operates with two ingredients: a complete description of a system's state at a given moment (a phase point in classical mechanics, a pure state vector in quantum mechanics) and an equation of motion that carries that state forward in time (Hamilton's equations or the Schrödinger equation). In principle, knowing these two elements lets one calculate the state at any other time. Yet there is a profound disconnect between this formalism and everyday experience. When performing a chemical reaction or any process at the human scale, we neither need nor can track the exact simultaneous positions and velocities of every molecule. In quantum mechanics, doing so is not even theoretically possible. Statistical mechanics bridges this gap by introducing uncertainty about which precise state the system occupies. Instead of following a single trajectory, it employs the statistical ensemble: a large collection of virtual, independent copies of the system spread across various states. In classical settings, this ensemble is a probability distribution over phase points in a phase space with canonical coordinate axes. In quantum settings, it becomes a probability distribution over pure states.

Beyond Equilibrium: Modeling the Irreversible World

Classical thermodynamics is primarily concerned with systems at equilibrium, but the real world is full of processes driven by imbalances. Non-equilibrium statistical mechanics addresses exactly this: the microscopical modeling of the speed of irreversible processes. Examples include chemical reactions and the flows of particles and heat through a medium. The fluctuation-dissipation theorem stands as the fundamental result obtained when non-equilibrium statistical mechanics is applied to the simplest non-equilibrium scenario—a steady-state current flow in a system of many particles. This extension broadened the field's utility enormously. Gibbs's methods, though initially derived within classical mechanics, proved general enough to adapt seamlessly to quantum mechanics once it emerged, and they still form the foundation of the discipline today. The field's mathematical framework, applying statistical methods and probability theory to large assemblies of microscopic entities, has found applications in an unexpectedly wide variety of domains: biology, neuroscience, computer science, information theory, and sociology. Sometimes called statistical physics or statistical thermodynamics, it remains a unifying language for understanding how aggregate properties of matter arise from the motion of their constituent parts.