Research Topics
Our group uses electronic transport in graphene-based van der Waals heterostructures to study how electron-electron interactions shape quantum matter in low dimensional systems. Stacking crystalline layers free from lattice-matching constraints, controlling the twist angle between them, and placing gates and screening layers a few nanometres from the active region lets us set the dimensionality, the coupling and the interaction strength of a system inside a single device, instead of comparing samples that differ in disorder and geometry. We follow three connected directions.
One-dimensional electronic channels
When two graphene layers are stacked with a minute twist, the lattice reconstructs into domains of alternating Bernal (AB and BA) stacking, separated by domain walls. These twist domain walls host protected one-dimensional (1D) channels. Our first experiments showed that they are ballistic, single-mode conductors, with a conductance reaching 4e2/h and a width below 3 nm. The channels are gate-controllable and can be made chiral or non-chiral, depending on whether a magnetic or an electric displacement field opens the gap in the surrounding domains (Barrier et al., Nature 628, 741 (2024)).
In one dimension, electrons cannot be described as weakly perturbed free particles. The system becomes a Tomonaga-Luttinger liquid, whose properties are set by a single interaction parameter. Most experiments have measured this parameter in samples where material and geometry fix its value. We use domain-wall channels, together with the screening technique described below, to make the interaction strength a variable. We ask what a single 1D conductor does as interactions are tuned, and what happens when several channels are coupled into ladders, networks, or pairs that can only exchange momentum through Coulomb interactions. This is a clean setting in which to ask when and how one dimension turns into two.
Coulomb screening and interaction engineering
Strong Coulomb repulsion gives rise to collective many-body phenomena that are notoriously difficult to predict. A key methodological tool of our work is the use of van der Waals heterostructure-based screening to control Coulomb interactions in situ. A metallic layer placed within a few nanometres of a two-dimensional system reduces the strength of the interaction without altering doping, strain or chemistry. By thinning the spacer layer by layer, the interaction strength can be reduced progressively, and by decoupling the screening layer from the sample through a twist, it can be pushed an order of magnitude beyond conventional screening designs.
With this approach we showed that proximity screening greatly enhances the electronic quality of graphene (Domaretskiy et al., Nature 644, 646 (2025)), and that screening suppresses the superconducting phase of magic-angle twisted bilayer graphene, which points to a pairing mechanism driven by electron-electron interactions rather than phonons (Barrier et al., Phys. Rev. X 16, 031040 (2026)). Interaction-driven phenomena in a strongly interacting electron-hole Dirac plasma in monolayer graphene, which produces giant magnetoresistance (Xin et al., Nature 616, 270 (2023)), and correlated phases in rhombohedral and twisted graphene (Shi et al., Nature 584, 210 (2020); Xu et al., Nature Physics 17, 619 (2021)) are the earlier results on which this strategy builds. Together, they establish interaction tuning as a unifying experimental strategy for mapping the phase diagrams of correlated quantum materials.
Josephson junctions
Josephson junctions are a natural way to probe superconducting correlations in a low-dimensional channel, and they are also what led us to the 1D channels above. We were originally looking for ways to build junctions from quantum Hall edge states, which is difficult because the edges must come within the superconducting coherence length of the electrodes. Twist domain walls solved that problem. Junctions based on these channels carry a supercurrent that persists in high magnetic fields and is limited only by the quantum of conductance of strictly 1D channels. They are gate-controllable and show Fabry-Pérot resonances, much like superconducting quantum dots.
We build on this platform to use Josephson junctions as probes and as devices: to study how superconductivity enters a single 1D channel (be it a quantum Hall edge, an anomalous quantum Hall channel or an AB/BA domain wall), how it behaves when interactions in that channel are tuned by screening, and how it is shared between coupled channels.
Complementary probes
Alongside electrical transport, we use nanophotonic, photocurrent and non-linear transport techniques, mostly in collaboration with F. Koppens (ICFO). They give access to quantities that dc transport cannot measure. IR and THz photocurrent showed that magic-angle twisted bilayer graphene hosts spontaneous inversion symmetry-breaking correlated states (Krishna Kumar et al., Nature Materials 24, 1034 (2025)); optical spectroscopy mapped the formation of moiré minibands (Li et al., Nano Letters 24, 15956 (2024)); and negative differential conductance enabled single-photon detection in a bilayer graphene superlattice (Nowakowski et al., Science 389, 644 (2025)). We use these probes selectively, where they answer a question that transport alone cannot. They extend to non-linear transport, in which out-of-equilibrium physics can be probed and bandwidth limitations show up (Bertini et al., Nano Letters (2026)).
See the full list on the Publications page.