The material family in question carries the chemical formula A2T3X4, where A is an alkali metal such as rubidium or cesium, T is a transition metal such as palladium or nickel, and X is a chalcogen such as sulfur or selenium. Within each van der Waals layer, the transition metal atoms form a perfect kagome network, sandwiched between alkali atoms and chalcogen ligands. Although the bulk crystals of these compounds were chemically characterized decades ago, with early work on rubidium and potassium niobate and palladium chalcogenides establishing their kagome structure, their two-dimensional electronic behavior had never been directly measured. The new study demonstrates that these crystals can be exfoliated like graphene, encapsulated in inert layers, and probed atom by atom — and that the kagome band structure survives all the way down to a monolayer.

The central experimental tool was nanofocused angle-resolved photoemission spectroscopy, or nano-ARPES, a technique in which an extremely bright synchrotron beam is focused to a spot small enough to interrogate micron-scale flakes. Using this method at facilities including the Advanced Light Source in Berkeley, Diamond Light Source in the UK and the MAX IV Laboratory in Sweden, the researchers mapped the momentum-resolved electronic structure of A2Pd3S4 devices of varying thickness, from eight layers down to a single layer. The measurements revealed the characteristic kagome fingerprint: a nearly flat band sitting just below the Fermi level, flanked by linearly dispersing Dirac bands, together forming the textbook three-band spectrum of the kagome model.

What makes the result striking is the isolation of that flat band. In most real kagome materials, the idealized dispersion is smeared out by extra hopping pathways, multiple competing orbitals and interlayer coupling, so the flat band blends into a congested spectrum of other states. In A2T3X4, a strong crystal-field splitting arising from the particular local geometry of each transition metal atom — four chalcogen ligands arranged in a planar configuration with D2h symmetry — pushes all other d orbitals well away from the Fermi energy, leaving an isolated manifold of dxy-derived kagome bands. Density functional theory calculations, cross-checked against polarization-dependent ARPES measurements, confirmed this orbital hierarchy in every compound the team surveyed, spanning palladium and nickel variants with both sulfur and selenium.

The comparison with existing flat-band platforms is telling. Since the 2018 discovery of magic-angle twisted bilayer graphene, the dominant strategy for creating flat electronic bands has been geometric: stack two atomically thin crystals, rotate one by a precise twist angle, and let the resulting moiré interference pattern slow the electrons down. That approach has delivered spectacular physics, including correlated insulating states and signatures of fractional quantum anomalous Hall effect in twisted bilayer MoTe2, but it comes with inherent limitations. Moiré bands are confined to minuscule unit cells containing only a few electrons, which means the electron density in the flat band is extremely low, and the delicate superlattice potential produces particle-like quasiparticles with tiny energy scales that only reveal themselves at temperatures measured in fractions of a kelvin.

The monolayer kagome crystals change the calculus entirely. Because the flat band is an intrinsic property of the atomic lattice rather than an artifact of a superlattice, its electron density is set by the actual atomic spacing of the kagome network. According to the team’s analysis, the flat band in A2T3X4 hosts more than 220 times the electron density achievable in moiré-based artificial flat-band systems. Higher density translates directly into larger characteristic interaction energies, and larger interaction energies mean that the collective phenomena expected in flat bands — from ferromagnetism predicted by Mielke’s theorem for line-graph lattices to fractional Chern insulator states — should become accessible at far higher temperatures than the millikelvin regime that currently defines the field.

The theoretical analysis supporting this prospect is unusually thorough. The researchers evaluated the quantum geometry of the topmost Chern band in these materials, a mathematical quantity that determines how closely an engineered band resembles the lowest Landau level of a two-dimensional electron gas in a magnetic field — the ideal setting for fractional quantum Hall physics. They found that while individual local values fall short of the ideal, the Brillouin-zone-averaged quantum metric and Berry curvature for cesium and rubidium palladium sulfide remain in the same order of magnitude as those of twisted bilayer MoTe2. Exact diagonalization calculations at one-third filling of the topmost Chern band produced energy spectra exhibiting the hallmark threefold ground-state degeneracy and spectral flow of a Laughlin state, the canonical fractional quantum Hall phase, even when band mixing was included.

Equally important for practical experimentation is the tunability and robustness of the platform. In the pristine monolayer devices, the flat band sits slightly below the Fermi level, which would normally keep it fully occupied and inert. The team demonstrated, however, that electrostatic gating — applying a voltage through the graphite back-gate of the encapsulated device — can shift the band up to the Fermi level, realizing the partially filled flat band that theorists prize as the starting point for interaction-driven states. The fabrication itself required considerable finesse: the kagome flake is encapsulated between top graphene and bottom graphite using a polymer stamp, released by heating above the polymer’s glass transition temperature, cleaned in chloroform, and polished with an atomic force microscopy tip before nano-ARPES spatial maps verify that only the encapsulated regions retain their kagome electronic structure while exposed regions degrade.

The study also charts the family’s chemical versatility as a design axis. Across the surveyed compounds, the bandwidth of the flat band, the size of the spin-orbit coupling gap at the Dirac point and the degree of electronic correlation all vary systematically with composition, giving researchers a tunable toolkit rather than a single fixed material. Environmental stability tests showed that cesium palladium sulfide flakes survive ten days in an inert atmosphere and endure heating to 300 degrees Celsius without visible degradation, although bilayer samples show measurable air sensitivity and nickel-based variants prove less thermally robust — reminders that the two-dimensional limit always demands careful handling.

The path from this demonstration to a confirmed high-temperature correlated phase remains open, as the authors themselves frame the work as establishing a platform rather than reporting a broken-symmetry state. Nonetheless, the implications ripple across several active frontiers. Flat-band ferromagnetism, which has recently been observed in kagome metals at cryogenic temperatures, could gain hundreds of times more thermal headroom. Fractional Chern insulators, currently confined to the coldest corners of experimental physics, might one day operate in a dilution-refrigerator-free device. And because the kagome layers are genuine van der Waals crystals, they can be stacked with other two-dimensional materials into heterostructures combining flat-band physics with superconducting contacts, spintronic layers or optical cavities. What began as a decorative weaving pattern, abstracted by physicists into a lattice model and hunted for in three-dimensional crystals for decades, has now been isolated in its purest two-dimensional form — a single atomic sheet in which electrons arrive already arranged for the quantum collective behavior that researchers have been trying to coax into existence by ever more elaborate means.

Subject of Research: Monolayer van der Waals kagome materials hosting gate-tunable isolated flat bands for high-temperature flat-band physics

Article Title: Monolayer van der Waals kagome lattice for high-temperature flat-band physics

Article References: Bong, H., Won, C., Lee, S., Oh, S., Nguyen, P. X., Chaturvedi, R., Oh, S., Han, J., Malinowski, P., Mowers, C., Kim, J., Ryu, S. H., Lee, H., Choi, J., Yer, S., Jozwiak, C., Bostwick, A., Rotenberg, E., Polley, C. M., … Kang, M. (2026). Monolayer van der Waals kagome lattice for high-temperature flat-band physics. Nature Physics. https://doi.org/10.1038/s41567-026-03492-0

Image Credits: AI Generated

DOI: 10.1038/s41567-026-03492-0

Keywords: kagome lattice, flat band, two-dimensional materials, van der Waals, nano-ARPES, topological matter, Chern band, electrostatic gating, quantum geometry, fractional quantum Hall effect, correlated electrons, A2T3X4

The physics at the heart of the experiment concerns stacking order. When graphene sheets are stacked, each successive layer can sit in slightly different registries relative to the layer beneath it. In the most common configuration, known as Bernal stacking, alternate layers align in a pattern that gives the material an ordinary, semimetallic electronic structure. A second arrangement, rhombohedral stacking, shifts each layer by the same registry offset in sequence, producing a dramatically different electronic landscape. Rhombohedral multilayer graphene has emerged in recent years as one of the richest platforms in condensed matter physics, hosting superconductivity, orbital ferromagnetism, and exotic topological states that are entirely absent in Bernal-stacked samples of the same thickness. The catch is that rhombohedral stacking is metastable, and growing devices in which the desired stacking is stable and well controlled has been a persistent challenge.

The Cornell team, led by corresponding author Kenji Yasuda, took advantage of the fact that these two stacking arrangements can be converted into one another by sliding entire layers laterally across each other. Because the layers in graphene are held together only by weak van der Waals forces, the energy barrier between registries is small enough that an electric field, and the carrier density it induces, can tip the balance. Earlier work by other groups had shown that doping and electric fields can influence stacking transitions in few-layer graphene, and that interlayer sliding in van der Waals materials can produce ferroelectric-like switching of polarization. What distinguishes the new study is the demonstration that this structural sliding transition can be harnessed as a non-volatile toggle for a collective magnetic phase, converting a structural degree of freedom into magnetic functionality.

Experimentally, the researchers fabricated devices from marginally twisted double-bilayer graphene, effectively four carbon layers held between top and bottom gates in a dual-gated Hall bar geometry. In such a device, sweeping the carrier density n and the displacement field D gives experimenters two independent knobs: n controls how many electrons or holes populate the layers, while D controls how asymmetrically charge is distributed between the top and bottom of the stack. By measuring longitudinal resistance as the gates were swept forward and backward, the team observed clear bistable switching between a Bernal-dominant configuration and a mixed state containing both rhombohedral and Bernal domains. The hysteresis in the resistance indicates a first-order structural phase transition: rather than sliding smoothly, the layers snap between registries, retaining whichever state they were last driven into until a sufficiently different gate condition pushes them back.

The phase diagram the team mapped out is remarkably broad, with the bistable switching persisting across a wide swath of the carrier density and displacement field plane. Quantum oscillation measurements confirmed the structural interpretation. In regions of the device identified as Bernal-dominant, Landau fan diagrams aligned with the known spectrum of Bernal tetralayer graphene, while the mixed regions displayed blurred oscillations consistent with a nearly equal coexistence of rhombohedral and Bernal domains. Strikingly, the hysteresis in the longitudinal resistance persisted all the way up to room temperature, indicating that the structural transition itself is robust and does not rely on cryogenic conditions. The magnetic phase that emerges from it, however, lives at much lower temperatures, with a Curie temperature of around 8 kelvin in parts of the phase space and closer to 2 kelvin elsewhere.

That magnetic phase is the central discovery. When the device is in the rhombohedral-Bernal mixed state, low-temperature transport measurements reveal an anomalous Hall effect: a transverse voltage that appears even at zero external magnetic field and shows hysteresis under magnetic field sweeps, the classic fingerprint of ferromagnetism. When the device is switched back into the Bernal-dominant state, the anomalous Hall signal vanishes. The authors demonstrated that this switching between magnet and non-magnet is non-volatile and repeatable: by cycling the gate voltages through the same sequence, they could repeatedly write and erase the ferromagnetic phase at a fixed set of gate values, with the state of the device determined entirely by its sweep history. The magnitude of the anomalous Hall signal stayed roughly consistent across cycles, although the coercive field varied between sweeps, a variability the authors attribute to the small magnetic energy scale of the ferromagnetism in their sample.

One of the most intriguing technical observations concerns the origin of the Hall signal itself. When the researchers applied opposite displacement fields, the sign of the anomalous Hall response reversed. This behavior led the authors to suggest that the Hall signal may originate not from the bulk of the rhombohedral regions but from the domain boundaries where Bernal and rhombohedral regions meet. Such boundaries are known in the graphene literature to host interesting electronic behavior, including topological valley transport, and in this system they may act as the conduits through which the ferromagnetic order produces its anomalous Hall response. Understanding exactly how domain walls between stacking regimes carry or mediate magnetic order is one of the questions the work leaves open, and the authors note that a full accounting of the resistance changes tied to domain configuration will require simultaneous real-space imaging and transport measurements.

The broader significance of the result lies in its demonstration that structural phase transitions in two-dimensional materials can serve as an on-demand control mechanism for complex electronic phases. In conventional phase-change materials, structural switching is used for memory applications, but the phases involved are typically distinguished by resistivity rather than by magnetism. In two-dimensional magnets such as chromium triiodide, pressure and strain have been shown to tune magnetic order, but those materials require magnetic atoms to exist in the first place. The graphene device here achieves magnetic switching in an elementally pure carbon system, where the magnetism is an emergent consequence of correlated electron behavior in a particular stacking arrangement. This decouples the magnetic functionality from any specific magnetic element and ties it instead to geometry, which can be manipulated electrically with van der Waals sliding.

There are, of course, substantial gaps between a low-temperature laboratory device and a practical technology. The ferromagnetic phase survives only at a few kelvin, the coercive fields fluctuate from cycle to cycle, and the domain configurations that underlie the switching appear to vary between preparations, producing scan-to-scan scatter in the hysteresis. Yet the structural switching itself endures to room temperature, and the strategy of using gate-induced sliding as a first-order transition between competing stacking polytypes is applicable in principle to other van der Waals materials where stacking governs exotic phases. As research into rhombohedral graphene continues to reveal superconductivity, fractional Chern insulators, and chiral superconducting candidates, the ability to switch between stacking regimes without rebuilding the device offers a new dimension of control. For now, the Cornell experiment stands as a vivid proof of concept: atoms slid by an electric field can turn a carbon crystal into a magnet, and back again, on command.

Subject of Research: Gate-induced layer sliding and stacking-order control of non-volatile magnetic switching in tetralayer graphene

Article Title: Gate-induced layer sliding drives non-volatile switching of magnetism in tetralayer graphene

Article References: Gate-induced layer sliding drives non-volatile switching of magnetism in tetralayer graphene. (n.d.). https://doi.org/10.1038/s41565-026-02306-x

Image Credits: AI Generated

DOI: 10.1038/s41565-026-02306-x

Keywords: graphene, layer sliding, rhombohedral stacking, Bernal stacking, ferromagnetism, anomalous Hall effect, non-volatile switching, two-dimensional materials, van der Waals, gate voltage, phase transition, stacking domains

Excitons are quasiparticles formed when a photon excites an electron, leaving behind a positively charged hole; the electron and hole remain bound together by Coulomb attraction. In monolayer semiconductors such as transition metal dichalcogenides, excitons dominate the optical response and are central to proposals for exciton-based circuits, transistors and quantum light sources. When two such monolayers are stacked on top of one another with a slight twist or lattice mismatch, electrons and holes can spatially separate into different layers, forming so-called interlayer excitons. These interlayer excitons possess long lifetimes, inherit the spin-valley properties of their parent materials, and carry a permanent out-of-plane dipole moment—all qualities that make them attractive for future optoelectronic technologies.

Yet controlling the flow of interlayer excitons has proved challenging. Conventional approaches rely on external electric fields, strain gradients or patterned potentials to herd excitons across a device. Such methods are often slow, energy-intensive or limited by the mobility of the excitons themselves, which can become trapped by defects, moiré disorder or interactions with the surrounding environment. The new work demonstrates a fundamentally different mechanism. By carefully engineering the stacking geometry of a van der Waals heterobilayer, the researchers created a situation in which the local atomic registry between the two layers varies across the sample. The lattice, seeking to minimize its total energy, undergoes a spontaneous reconstruction: regions of high-symmetry stacking expand and contract, reshaping the local potential landscape experienced by the excitons.

The crucial insight is that this reconstruction is not a static curiosity. As the researchers show, the reconstructed lattice effectively creates internal electric fields and band-alignment variations that act on the interlayer excitons’ permanent dipole moment. The result is a built-in drift force that propels excitons across the material at remarkable speed—orders of magnitude faster than diffusion alone would allow. In ultrafast optical measurements, the team tracked exciton populations as they moved between different stacking domains, observing a coherent, directional flow that correlated precisely with the spatial pattern of lattice reconstruction. The excitons, in effect, ride a wave carved by the crystal itself.

The experimental strategy combined several state-of-the-art techniques. The heterobilayers were assembled from monolayers of transition metal dichalcogenides using deterministic transfer methods, allowing precise control over the relative orientation of the two lattices. Optical microscopy and spectroscopy revealed the characteristic photoluminescence signatures of interlayer excitons in different stacking configurations, while spatially and temporally resolved measurements mapped how exciton emission evolved across the reconstructed domains after pulsed laser excitation. Complementary theoretical modeling and atomistic simulations confirmed that the observed drift velocities and trajectories emerge naturally from the reconstructed band structure, without the need for any externally applied field.

From a technical standpoint, the mechanism can be understood through the interplay of moiré physics and exciton dipoles. In a twisted or lattice-mismatched bilayer, the superposition of the two atomic lattices generates a moiré superlattice with a period of several nanometers. At small twist angles, the system lowers its elastic energy by deforming into triangular domains of near-perfect high-symmetry stacking, separated by narrow domain walls where the stacking registry changes abruptly. This lattice reconstruction modifies the local band edges: the energy of the interlayer exciton, which depends on the relative alignment of the conduction and valence bands in the two layers, varies across the moiré cell. Because the interlayer exciton carries a permanent dipole perpendicular to the layers, any gradient in its potential energy produces a force. The reconstructed lattice thus acts as a nanoscale, self-assembled potential gradient, driving excitons along well-defined crystallographic directions.

The speed of the effect is what sets it apart. Exciton diffusion in conventional semiconductor structures typically proceeds at velocities limited by thermal randomization and scattering with phonons and impurities. In the reconstructed heterobilayer, the deterministic drift component dominates over diffusion, enabling exciton transport over micron-scale distances on picosecond timescales. Such speeds approach those seen in deliberately engineered exciton funnels and are far beyond what natural diffusion would achieve over comparable distances. Moreover, because the driving force is intrinsic to the crystal structure, the effect is robust against the disorder that often plagues external-field approaches, and it operates without any need for electrodes, gates or patterned substrates.

The implications ripple across several research frontiers. For excitonics—a proposed paradigm in which excitons rather than electrons or photons carry information—the demonstration of fast, field-free, structurally guided exciton drift provides a missing ingredient: a way to route excitonic signals through a device using nothing more than cleverly designed stacking geometry. Because interlayer excitons also couple strongly to light, such routing could be read out optically, enabling hybrid photonic-excitonic circuits on a chip. For moiré physics, the result underscores that lattice reconstruction is not merely a structural side effect but an active agent that can be exploited to shape excitonic and electronic behavior. And for fundamental science, the work offers a clean platform to study how topological defects in a crystal lattice interact with neutral quasiparticles carrying electric dipoles.

The discovery also raises tantalizing questions that the field will now pursue. Can the direction and magnitude of the reconstruction-driven drift be tuned dynamically, for example by applying strain, gating the layers, or modulating the twist angle? Could domain walls themselves—where the stacking registry flips—serve as reconfigurable exciton waveguides or beam splitters? Might similar mechanisms operate for other dipolar quasiparticles, including trions, polaritons or even exciton condensates in the strongly correlated regimes now accessible in moiré systems? The researchers suggest that the interplay between lattice dynamics and exciton transport could be engineered with a precision that was previously unimaginable, effectively turning crystallography into a design tool for quantum materials.

There are, of course, practical hurdles on the road to applications. Van der Waals heterostructures must be assembled with exceptional cleanliness to avoid contamination and bubbles that disrupt the delicate reconstruction patterns. The optical measurements used to observe exciton drift remain demanding, requiring ultrafast lasers and low-temperature cryogenics in many cases, although the underlying physics persists at elevated temperatures. Scaling from laboratory-scale flakes to wafer-scale devices will require advances in growth and transfer techniques. Nevertheless, the rapid progress in two-dimensional materials engineering over the past decade gives reason for optimism that these challenges will be met.

What makes the result resonate beyond the specialist community is its conceptual elegance. Rather than fighting against the intrinsic tendencies of a layered crystal, the researchers harnessed them. The lattice, under the gentle constraint of twist-angle engineering, reorganizes itself into a landscape of nanoscale potentials—and the excitons, obeying simple electrostatic forces, stream across that landscape at extraordinary speed. It is a vivid reminder that in the quantum materials of the twenty-first century, structure and function are inseparable: the very geometry of atoms in a stack of atomically thin sheets can become the engine that moves light-matter quasiparticles across a chip.

As laboratories worldwide continue to explore the rich physics of moiré materials, this demonstration of fast interlayer exciton drift driven by lattice reconstruction is likely to be remembered as a turning point—the moment when structural self-organization in van der Waals heterostructures was transformed from a phenomenon to be characterized into a resource to be exploited. Whether it leads to practical exciton circuits, new classes of optoelectronic devices, or deeper insights into correlated states of matter, one thing is clear: the humble crystal lattice has revealed itself to be a far more dynamic and powerful player in the story of two-dimensional materials than anyone had fully appreciated.