Dates first: this study was published August 17 in PRX, with a same-day APS Viewpoint written by Harvard's Kaxiras; a September 4 phys.org report brought the discussion to a wider audience. The paper is three weeks old—but the question it answers is eight years old.
In March 2018, Yuan Cao and Jarillo-Herrero at MIT twisted two graphene layers by 1.1 degrees, flat bands appeared, and superconductivity emerged at 1.7 K—a Nature cover, and "magic-angle graphene" got its name. For eight years one fundamental question hung unanswered: why do the electrons pair? Textbook superconductors use lattice vibrations (phonons) as the glue—"conventional." Magic-angle graphene, with its flat bands, strong correlations, and extremely compressed phase diagram, looked like the "unconventional" cuprate route. The two camps fought for eight years, neither convincing the other.
Eight years of experiments, all one nanometer short
The verdict-style idea sounds simple: place a tunable metal next to the superconducting layer to screen out the Coulomb interaction between electrons. If pairing relies on phonons, screening Coulomb repulsion should make pairing easier and Tc should rise; if pairing relies on electron-electron interactions, screening removes the glue and Tc should collapse.
All previous attempts failed on distance. Coulomb screening decays exponentially, acting only over one or two nanometers. In 2020, two experiments with graphite gates had 6-10 nm of hBN in between—superconductivity was unmoved. In 2021, one group squeezed the gap to 3 nm, and Tc rose 2-3%: the textbook signature of conventional superconductivity. The physics hid in the last nanometer, and no insulating substrate could fit there.
Manchester's solution: no substrate, just a twist
The Barrier team (National Graphene Institute, University of Manchester, led by Geim—the 2010 graphene Nobel laureate) built a "twisted four-layer" device: below, a 1.15-degree magic-angle bilayer (the superconductor); above, a 0.46-degree bilayer (the screening layer). Nothing separates them—direct atomic stacking, with a ~10-degree relative twist scattering the two moiré momentum sets apart: electrons do not tunnel between them, yet Coulomb interactions pass face-to-face. hBN only encapsulates the outside for shaping.
Then they switched it on. Doping the screening layer with carriers turns it metallic, and Coulomb screening kicks in. The result speaks for itself: the superconducting dome continuously shrinks, critical current goes to zero, and beyond roughly 2×10¹² cm⁻² doping, superconductivity vanishes completely even at 35 mK—a Tc change exceeding an order of magnitude. Correlated insulating states die in sync. The sample is fine: normal-state transport carries on, and the supplement rules out the three "fake death" explanations—parallel shunt, displacement field, and proximity effects.
How to read the verdict
The paper's conclusion is measured: "the screening behavior rules out conventional (BCS-like) Cooper pairing mechanisms and supports an electronic pairing mechanism." The Viewpoint's Kaxiras adds a layer of honesty: phonons may still play a stabilizing role, just "unlikely to play a dominant role in pairing." As for which electronic mechanism—plasmon models can reproduce this collapse (an old lead from Adam's group, theoretical collaborators on this paper), but the paper doesn't commit. This paper's victory belongs to the method of elimination—very clean elimination.
Looking back at those two "failed" predecessor experiments is especially telling: the 2-3% Tc rise at 3 nm spacing looked exactly like conventional superconductivity—at the time, read as evidence that magic-angle graphene was conventional. Now it's clear: that was just an artifact of screening decaying to nearly zero at 3 nm. The physics didn't change; the ruler wasn't short enough.
Two human touches. Kaxiras, who wrote the Viewpoint explaining the work, was a co-author of the 2018 Nature cover paper—the person who helped open the question eight years ago wrote the guide to the verdict. In the same period, a Nature Physics paper by Gao et al. reached the same conclusion via a different route, mutual corroboration. And a note of lab honesty: of this batch of devices, two achieved the magic angle, and only one superconducted. The sample yield in this field has been this moving for eight years.
Geim's own interest, told to reporters, lies elsewhere: he says he only cares about room-temperature superconductivity, "preferably at room temperature or above," then adds "Rome wasn't built in a day." This twisted four-layer device is far from that dream, but it leaves behind a handy tool: a screening knob that can be twisted right up against a correlated material. Whose unconventional superconductor gets turned next? The paper doesn't say. The knob is in hand.
References: PRX paper | Kaxiras Viewpoint | phys.org report | arXiv full text