Heat Can Switch Off a Topological Insulator. Above 500 Kelvin, Bismuth Selenide Stops Being One.
Physicists in Shanghai calculated that spin-orbit coupling — long treated as a fixed property of a material's atoms — weakens as temperature rises, closing the band inversion that makes topological states possible.
Spin-orbit coupling is usually treated as a fact about a material rather than a condition it happens to be in. Heavy atoms produce strong coupling, light atoms produce weak coupling, and the strength is set by what the crystal is made of. A group at the Shanghai Institute of Ceramics, part of the Chinese Academy of Sciences, has calculated that this is not quite true — and that temperature can turn the effect down far enough to destroy a material's topological character altogether.
The team, led by Professor Sun Yiyang with Ph.D. candidate Lu Lingyan as first author, worked on bismuth selenide, Bi2Se3, the textbook three-dimensional topological insulator. A topological insulator does not conduct through its interior but carries current along its surfaces in states that are protected by the material's band structure, which is what makes them attractive for spintronics and for certain quantum computing architectures. That protection depends on band inversion: the ordering of two electronic bands being flipped relative to a normal insulator, an inversion driven by spin-orbit coupling.
Across a temperature range from 0 to 900 kelvin — absolute zero to about 627 degrees Celsius — the calculations show the spin-orbit-induced correction to the band gap shrinking by as much as 0.19 electron volts. That is enough to undo the inversion. Above roughly 500 kelvin, the structures come out topologically trivial. Surface spectral function calculations confirm it directly: the gapless Dirac cones that are the signature of the protected surface states disappear.
The result was reproduced in Bi2Te3 and Sb2Te3, the two closest relatives of bismuth selenide and both workhorses of the field, which suggests the effect is general to this family rather than a quirk of one compound.
Sun's group frames the finding as a control knob rather than a warning. "Temperature could serve as a new 'knob' for controlling the topological properties of materials," the team writes — a way of switching a topological phase on and off in a single sample without changing its chemistry, applying strain or building a heterostructure. Most proposed devices in this space assume the topology is a permanent feature of the material and design around it.
The corollary is less comfortable. If spin-orbit coupling is temperature-dependent, then predictions of topological behavior computed at zero temperature — which is how nearly all of them are computed — describe a material that does not exist in a laboratory, let alone in a device running warm. How much of a correction that implies for the wider catalogue of predicted topological materials is not yet known.
The work appears in the journal Newton, published September 2, 2026, DOI 10.1016/j.newton.2026.100621.
Originally reported by Phys.org.