Cavity-altered superconductivity

Nature News · Feb 25, 2026 · Collected from RSS

Full Article

AbstractIs it feasible to alter the ground-state properties of a material by engineering its electromagnetic environment? Inspired by theoretical predictions1,2,3,4,5,6,7,8,9,10,11,12, experimental realizations of such cavity-controlled properties without optical excitation are beginning to emerge13,14,15,16,17,18,19. Here we devised and implemented a new platform to realize cavity-altered materials. Single crystals of hyperbolic van der Waals (vdW) compounds provide a resonant electromagnetic environment with enhanced density of photonic states and prominent mode confinement20,21,22,23,24. We interfaced hexagonal boron nitride (hBN) with the molecular superconductor κ-(BEDT-TTF)2Cu[N(CN)2]Br (κ-ET). The frequencies of infrared hyperbolic modes (HMs) of hBN (refs. 25,26) match the infrared-active carbon–carbon (C=C) stretching molecular resonance of κ-ET implicated in superconductivity27. Nano-optical data supported by first-principles molecular Langevin dynamics simulations confirm the presence of resonant coupling between the hBN hyperbolic cavity modes and the C=C stretching mode in κ-ET. Meissner-effect measurements using magnetic force microscopy (MFM) demonstrate a strong suppression of superfluid density near the hBN/κ-ET interface. Non-resonant control heterostructures, including RuCl3/κ-ET and hBN/Bi2Sr2CaCu2O8+x (BSCCO), do not show the pronounced superfluid suppression. These observations suggest that hBN/κ-ET realizes a cavity-altered superconducting ground state. Our work highlights the potential of dark cavities devoid of external photons for engineering electronic ground-state properties of complex quantum materials. MainSome of the most exciting properties of solids arise from strong collective interactions among electrons, spins and the crystal lattice. The emergent effects born out of such strong interactions are abundant and drive the formation of varied electronic and magnetic phases. At the vanguard of present interest and debate is the question of whether the strong interaction of the quantum fluctuations of electromagnetic modes in photonic cavities or metastructures with elementary excitations in solids can also prompt phase transitions and lead to new quantum states of matter. A grand aspiration of cavity quantum materials research is to uncover fundamentally new routes for controlling properties of matter by judiciously tailoring the quantum electromagnetic environment1,2,3,4,5,6,7,8,9,10,11,12. Experiments with dark cavities revealed modified transport properties in the integer and fractional quantum Hall states of a 2D electron gas13,17,19, as well as cavity-assisted thermal control of the metal-to-insulator transition in charge-density-wave systems15. Pioneering theoretical works on cavity control of superconductivity explored dark-cavity modification of phonon-mediated pairing through an increase in electron–phonon coupling by means of phonon polaritons1,8,9, Amperean pairing directly mediated through dark-cavity photons28,29, as well as the driven-cavity extension of the Eliashberg effect30.Here we have chosen to focus on superconductivity as a purely electronic phase transition, devoid of lattice reconstructions and charge or spin orderings. We work with κ-ET, a widely studied layered organic salt that superconducts below the transition temperature Tc = 11.5 K (refs. 31,32) (Supplementary Information Sections 1 and 2). We make use of an electromagnetic environment structured by the dipole-active phonons of a thin vdW hyperbolic material, hBN. Hyperbolicity occurs when the permittivity has opposite signs along different axes25,26 and results in a highly enhanced photonic density of states33,34,35. Notably, HMs of hBN overlap with the frequency of the C=C stretching mode of κ-ET at 1,470 cm−1 (ref. 36) and the two modes hybridize at the hBN/κ-ET interface. Effectively, the hBN slab acts as an electromagnetic cavity (Supplementary Information Section 3) resonantly tuned to the C=C stretching mode of κ-ET that is implicated in superconductivity27. In this work, we examine how superconductivity is altered at the hBN/κ-ET interface, using advanced scanning-probe-based techniques capable of analysis of electrodynamics in buried interfaces.Meissner effect at resonant interfacesMFM examines the local superfluid density by means of the repulsive Meissner force experienced by a magnetized tip at the end of a cantilever oscillating above the surface of a superconductor37. The force results from the supercurrent screening the stray magnetic field of the tip. Notably, the Meissner force is not affected when non-magnetic materials are inserted between the tip and the superconductor and the data in Fig. 1a,b confirm that MFM detects superconductors encapsulated by insulating layers. We measure the shift in the resonant frequency of the cantilever, which is proportional to the force gradient ∂zFz (Methods). Analysis of ∂zFz allows the extraction of the superfluid density ρ0 (ref. 37) (Methods and Supplementary Information Section 5).Fig. 1: Electromagnetic environment alters the superfluid density at the interface between hBN and κ-ET.Schematic at the bottom depicts the Meissner force Fz experienced by the magnetic atomic force microscope tip above the surface of the κ-ET molecular superconductor. Exfoliated hBN and RuCl3 microcrystals sit on the surface of a bulk κ-ET crystal (optical images in Supplementary Information Section 4). a,b, MFM data shown in the form of the derivative of the Meissner force ∂zFz as a function of tip height on bare κ-ET (green), hBN/κ-ET (blue) and RuCl3/κ-ET (magenta), taken at temperature 2 K. The κ-ET surface is at z = 0 and the grey shading highlights the difference between the two curves. The vertical axes in panels a and b are identical. The superfluid density is greatly reduced near the hBN/κ-ET interface but not near the RuCl3/κ-ET interface. Inset, same as a except that the superconductor has been replaced by BSCCO. The bare BSCCO curve is plotted using a wider grey line for visual clarity. c, Model real part of the out-of-plane (OOP) permittivity for κ-ET (green) and the in-plane (IP) permittivities for hBN (blue) and RuCl3 (magenta), based on refs. 36,39,50. For clarity, the permittivities have been simplified to show only the relevant modes. The C=C stretching mode of κ-ET (labelled C=C) falls within the hyperbolic region of hBN between the TO and LO frequencies (grey shading). a.u., arbitrary units.Source DataFull size imageWe examine the impact of hBN on superconductivity in a representative heterostructure composed of a 60-nm-thick hBN microcrystal on the surface of a bulk κ-ET single crystal (about 200 µm thick). We report ∂zFz measured as a function of z, the separation between the tip and the κ-ET surface, in Fig. 1a. On a region of the κ-ET crystal unobscured by hBN (green), ∂zFz(z) is negative at T = 2 K, indicating a repulsive force. The magnitude of ∂zFz(z) grows with reduced tip–sample surface separation. The ∂zFz(z) signal collected above hBN (blue) shows a weakened ∂zFz, highlighted by the grey shading. As a control experiment, we placed a 55-nm-thick RuCl3 microcrystal on top of the same κ-ET sample. RuCl3 is an insulator with a static permittivity similar to that of hBN (refs. 38,39). However, optical phonons of RuCl3 all occur at much lower frequencies below 350 cm−1 (ref. 39) and therefore do not resonantly couple to the C=C stretching mode of κ-ET. Figure 1c schematically depicts the permittivities of hBN, κ-ET and RuCl3. Comparing measurements above RuCl3/κ-ET (Fig. 1b, magenta curve) and bare κ-ET (green curve, same as in Fig. 1a), we observe a much weaker (less than about 7%) effect than over hBN/κ-ET.The strong contrast between the RuCl3/κ-ET and hBN/κ-ET interfaces indicates that the suppression of superfluid density under hBN is probably prompted by resonant coupling across the hBN/κ-ET interface. We remark that the RuCl3 control experiment rules out a prominent role of charge transfer owing to work-function disparity40 in the suppression of the superfluid density in Fig. 1 (Supplementary Information Section 14) and similarly suggests that strain is unlikely to be important. We perform a second control measurement on a heterostructure of 8-nm hBN residing on a 28.5-nm-thick microcrystal of BSCCO with Tc = 55 K. Crystals of BSCCO host phonons below 650 cm−1 (ref. 41), far below the HMs in hBN. MFM data plotted in the inset of Fig. 1b for hBN/BSCCO (blue) are indistinguishable from the data for bare BSCCO (dark grey; Supplementary Information Section 6). Thus, the marked suppression of the superfluid density is specific to the resonant nature of the hBN/κ-ET interface.Suppression of superfluid densityWe now validate the association of reduced ∂zFz (Fig. 1a) with suppressed superfluid density by studying the ∂zFz temperature dependence. Over bare κ-ET, the large repulsive ∂zFz arises from the Meissner effect (Supplementary Information Section 5), decreases with increasing temperature (Fig. 2a) and vanishes near Tc. Having confirmed that Meissner repulsion is the main contributor to ∂zFz over bare κ-ET, we examine the effects at the heterostructure interfaces. We performed constant-height scans with the tip hovering above the κ-ET surface in a 16 × 16-µm2 region that includes both hBN and RuCl3 (Fig. 2b,c). To facilitate comparison between data collected at different temperatures, we focus on the differential signal ∆∂zFz(x, y) = ∂zFz(x, y) − ∂zFz|bare, in which the latter term stands for ∂zFz over bare κ-ET. At 2 K (Fig. 2b), ∆∂zFz ≈ 40 pN µm−1 over hBN, indicating suppressed superfluid density. Over RuCl3, ∆∂zFz is visibly much smaller. More quantitatively, in Fig. 2d, we show the histogram corresponding to the image of the ∆∂zFz signal. We witness distinct distributions representing bare κ-ET, RuCl3/κ-ET and hBN/κ-ET. The hBN/κ-ET peak is prominently separated from the other two, indicating the strong effect of hBN on the superflui