Anderson localization is one of the most counterintuitive predictions in quantum mechanics. Throw a wave into a disordered medium and, if the disorder is strong enough, the wave stops. Not because it is absorbed — it is reflected back on itself by interference between scattered paths, and the destructive interference is so complete that the wave remains confined forever. In three dimensions, this produces a quantum phase transition: below a critical energy (the mobility edge), states are localized; above it, they are extended and transport is diffusive.
The theory is from 1958. Direct experimental observation in three dimensions has taken nearly seventy years.
Yu et al. achieve it using ultracold atoms released into a laser-speckle disorder potential. The key technical advance is energy resolution. Previous cold-atom experiments suffered from broad energy distributions — atoms with many different energies all expanding simultaneously, blurring the sharp boundary between localized and diffusive regimes. Yu et al. prepare matter waves with narrow energy distributions and track their expansion over long timescales. At energies below the mobility edge, the atoms stop spreading. At energies above it, they diffuse. The transition between these regimes is sharp.
The measurements agree with state-of-the-art numerical predictions across a wide range of disorder strengths. This resolves long-standing discrepancies between earlier experiments and theory — discrepancies that persisted for decades and led some to question whether the 3D Anderson transition could be cleanly observed in atomic systems at all.
Banon et al. provide the theoretical framework. Self-consistent localization theory, incorporating both the spectral structure and spatial properties of the prepared quantum states, reproduces the observed density profiles across all three regimes: localized, diffusive, and critical. The theory also clarifies the role of thermal atoms versus the Bose-condensed component — the energy distribution matters, and disentangling these contributions is necessary for quantitative agreement.
The result is satisfying in the way that a long-predicted phenomenon finally observed cleanly is satisfying. The physics was known. The mathematics was solved. What was missing was the experimental control to prepare a quantum state with narrow enough energy resolution to see the transition without it being washed out by averaging. The sixty-seven-year gap between prediction and observation is not a gap in understanding. It is a gap in experimental precision.