The Electronic Disguise

La₃Ni₂O₇ is a bilayer nickelate superconductor — it conducts electricity with zero resistance under high pressure. Like other nickelate superconductors, it undergoes an abrupt structural phase transition on cooling: the crystal lattice distorts, and the material’s properties change discontinuously. The standard interpretation is structural — atoms rearrange, bond lengths change, the lattice finds a lower-energy configuration. Density functional theory (DFT), the workhorse of computational materials science, models the transition as a structural event and produces plausible-looking results.

DFT is wrong about what drives the transition.

Oh, Baek, Kim, Roh, and others (arXiv:2603.12924, March 2026) show that the abrupt transition in La₃Ni₂O₇ is driven by orbital dimerization — the spontaneous formation of spin-singlet bonds between d-orbitals on neighboring nickel atoms across the bilayer. Two electrons, one from each layer, pair their spins and lock into a bonding orbital. The pairing is an electronic instability that happens to drag the lattice along with it. The structural distortion is a consequence, not a cause.

Standard DFT misses this because it treats electrons as independent particles in an average potential field. Orbital dimerization is a correlation effect — it requires two electrons on different atoms to form a coherent quantum mechanical bond, which the mean-field approximation cannot capture. The lattice distortion DFT predicts is real, but the mechanism DFT assigns to it is wrong. The calculation gets the right answer for the wrong reason.

The structural lesson: when a computational method produces results consistent with experiment, consistency is not validation. DFT predicts the lattice distortion correctly because the distortion is the observable consequence of dimerization, and DFT can model consequences even when it misses causes. The observable symptom (structural change) and the underlying disease (electronic instability) produce the same X-ray diffraction pattern. The disguise is perfect because the measurement doesn’t distinguish between structure-driven and electron-driven distortion.