Anderson Localization: How Disorder Traps Waves

Waves versus Particles in a Disordered Medium

Picture a classical particle, say a marble, rolling through a landscape scattered with random bumps and obstacles. Each collision sends it off in a new direction, and over many collisions its motion becomes a random walk. Track its average position over time and you will find it slowly diffuses outward, spreading farther and farther from where it started, exactly the behavior described by the classical theory of electrical conduction in metals. Now replace that marble with a quantum wave, such as an electron's wavefunction. The wave does not simply bounce from obstacle to obstacle in the classical sense. Instead, it splits at every scattering site, sending out a spray of partial waves that travel along countless different paths through the disordered medium, then recombine and interfere with one another. This is the crucial difference: particles add up probabilities, but waves add up amplitudes, and amplitudes can interfere constructively or destructively. When the disorder is weak, the interference effects average out over the many possible paths, and the wave behaves almost like the classical particle, diffusing outward in a process called weak localization, itself a subtle precursor effect. But when disorder becomes strong enough, something qualitatively different can occur. The countless interfering paths no longer average out randomly; instead, they conspire so that the wave amplitude reinforces itself in one small region while cancelling almost completely everywhere else. This is not merely a slower diffusion. It is a fundamentally different phase of behavior, in which the wave stops spreading altogether and remains trapped indefinitely, no matter how long you wait. Understanding this distinction between classical random walks and quantum interference is the essential starting point for grasping why disorder can be so much more powerful at stopping waves than intuition from everyday physics would suggest.

The Interference Mechanism Behind Localization

The heart of Anderson localization lies in a subtle interference effect involving pairs of scattering paths that traverse the same sequence of random obstacles but in opposite directions, so-called time-reversed loops. Consider a wave that scatters off a series of impurities and eventually returns close to its starting point by two different routes: one path traveling clockwise around a loop of scatterers, and another traveling counterclockwise around that same loop. Because both paths have exactly the same length, they pick up exactly the same phase, and when they recombine, they interfere constructively, doubling the probability of finding the wave back near where it started compared to what a naive classical calculation would predict. This enhanced return probability is the seed of localization. As disorder increases, or as the system is examined at lower dimensionality, these constructive backscattering loops multiply and reinforce one another across many length scales, eventually overwhelming the wave's ability to propagate outward at all. Away from this favored trapping region, however, the vast multitude of other possible paths interfere destructively with each other, essentially cancelling out almost completely. The net result is a wavefunction whose amplitude is large in a compact pocket and decays exponentially with distance from that pocket, a mathematical signature quite different from the gentle power-law spreading of ordinary diffusion. Crucially, this is a purely wave-mechanical effect with no classical analogue. A classical marble has no memory of phase and cannot interfere with itself, so it simply cannot localize this way. It is precisely because electrons, light, and sound are described by wave equations that they are susceptible to this dramatic trapping when disorder crosses a critical threshold.

The Mobility Edge: A New Kind of Boundary

In an ordinary insulator, such as diamond or glass, electrical conductivity is blocked because there is an energy gap: electrons simply have no available quantum states to occupy between the filled valence band and the empty conduction band, so they cannot move even if given a small nudge. Anderson localization offers a completely different route to insulating behavior, one that requires no gap whatsoever. In a disordered system, quantum states at every energy still exist, but whether a given state is localized or extended depends on its energy relative to a critical value known as the mobility edge. States with energies on one side of the mobility edge are extended, meaning the wavefunction spreads throughout the entire sample and can carry current freely, behaving much like the delocalized electron waves in a clean conductor. States on the other side of the mobility edge are localized, trapped within a finite region no matter how large the sample is, and therefore unable to contribute to conduction at zero temperature. As disorder in a material increases, the mobility edge shifts, and the boundary between extended and localized states sweeps across the range of available energies. If the mobility edge moves past the Fermi energy, the energy level that determines which states are actually occupied by conduction electrons, the material undergoes the Anderson transition: a sharp change from metallic, conducting behavior to insulating behavior driven purely by disorder, with the underlying electronic band structure essentially unchanged. This is a genuinely new paradigm for the metal-insulator transition, one governed not by band structure or energy gaps but by the interplay between disorder strength, dimensionality, and interference.

Dimensionality and the Scaling Theory of Localization

One of the most striking predictions to emerge from the theoretical framework built around Anderson's original insight is that the dimensionality of a system profoundly changes how localization plays out. The scaling theory of localization, developed in the late 1970s, argued that in one and two spatial dimensions, arbitrarily weak disorder is enough to localize all electronic states eventually, given a large enough sample, meaning that in the strict mathematical limit, there is no true metallic phase at all in these lower-dimensional systems. In three dimensions, however, the situation is richer: weak disorder still permits an extended, conducting phase to survive, and a genuine mobility edge can separate localized states from extended ones, allowing for a real metal-insulator transition to occur at some finite critical disorder strength. This dimensional sensitivity connects Anderson localization to the broader family of phase transitions and critical phenomena studied throughout condensed matter physics, since the Anderson transition exhibits its own critical exponents and universal scaling behavior near the mobility edge. Experimentally, this dimensional dependence has been probed in thin films, wires, and bulk crystals, with two-dimensional and one-dimensional samples showing a persistent tendency toward localization even when disorder is quite modest, exactly as the scaling arguments predict. Temperature, interactions between electrons, and finite sample size all complicate the clean theoretical picture in real materials, but the core prediction, that reducing dimensionality makes waves dramatically more prone to localization, has held up as one of the most influential and well-tested ideas to come out of this field, shaping decades of subsequent research into disordered quantum systems.

Beyond Electrons: Light, Sound, and Cold Atoms

Although Anderson originally formulated his theory to explain the behavior of electrons in disordered crystals, the underlying mathematics applies to any wave phenomenon, and this universality has made localization one of the most widely tested ideas in modern physics. Researchers have observed localization of light in disordered photonic materials, including powders of titanium dioxide particles and specially engineered arrays of optical waveguides, where instead of spreading across a sample, laser light remains confined near its entry point once scattering disorder is strong enough. Similar experiments have demonstrated localization of sound waves and elastic vibrations in disordered granular materials and engineered acoustic structures, confirming that the phenomenon has nothing fundamentally to do with the electron's charge or spin, only with its wave nature. Perhaps the cleanest experimental demonstrations have come from ultracold atom experiments, where a Bose-Einstein condensate, a cloud of atoms cooled to nanokelvin temperatures and described by a single coherent matter wave, is released into a laser-generated speckle pattern that acts as a precisely controllable source of disorder. By tuning the strength of this artificial disorder, physicists have watched atomic matter waves transition from freely expanding clouds to sharply localized, exponentially decaying density profiles, providing textbook-quality confirmation of Anderson's predictions in a system where every parameter can be dialed in with extraordinary precision. These cross-disciplinary confirmations, spanning electrons, photons, phonons, and cold atoms, demonstrate that localization is not a quirk of solid-state electron physics but a deep and general consequence of wave interference in any sufficiently disordered medium.

Frequently asked questions

Try it live

Everything above runs in your browser — open Anderson Localization: How Disorder Traps Waves and change the parameters while it is running. Nothing is installed, nothing is uploaded, the whole model lives in one tab.