🔬 mesoscale physics

Nonlocal Josephson coupling and current redistribution in a large-area graphene multiterminal Josephson junction

This paper demonstrates that a large-area graphene multiterminal Josephson junction exhibits a pronounced negative differential longitudinal resistance arising from the interplay between nonlocal Josephson coupling and nonlocal normal-state conductance, offering a new platform for characterizing hybrid superconductor-semiconductor systems and exploring many-body ground states.

Original authors: Luojia Zhang, Kushagra Aggarwal, Jay D. Sau, Ethan G. Arnault, Kenji Watanabe, Takashi Taniguchi, Valla Fatemi

Published 2026-10-12

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Luojia Zhang, Kushagra Aggarwal, Jay D. Sau, Ethan G. Arnault, Kenji Watanabe, Takashi Taniguchi, Valla Fatemi

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the world of electricity, materials are usually sorted into two camps: conductors, which let electrons flow freely like water through a wide pipe, and insulators, which block that flow entirely. Between these two extremes lies a special state called superconductivity, where certain materials allow electricity to move without any resistance at all. When two superconductors are separated by a thin barrier, they can still talk to each other through a quantum effect, allowing a current to flow across the gap without a battery pushing it. This is the Josephson effect, a phenomenon that has become a cornerstone for building sensitive sensors and the most advanced quantum computers. For decades, scientists have studied these connections using simple setups with just two superconducting leads. However, by adding more leads and creating complex networks, researchers have predicted a whole new landscape of behavior where the current can be steered and shaped in ways that simple two-wire connections cannot achieve. The challenge has been building a device large enough to see these complex interactions clearly, without the signals getting lost in the tiny, crowded spaces of traditional microchips.

A team of researchers has now built such a device, creating a large-scale network that reveals a surprising and useful behavior in how electricity moves through it. They constructed a "Josephson Hall bar," a device made from a single layer of carbon atoms known as graphene, sandwiched between protective layers and connected to six different metal leads. Unlike previous experiments that were microscopic, this device is large enough to be seen with a standard microscope, stretching several micrometers across. The researchers sent an electric current through the device and watched how the voltage changed as they increased the flow. They discovered that when the current reached a specific point, the voltage across the middle of the device suddenly dropped as the current increased further. This counterintuitive behavior, where pushing harder makes the resistance go down, is called negative differential resistance. It is a phenomenon that usually requires complex circuitry to create, yet here it emerged naturally from the way the superconducting currents interacted across the large, open space of the graphene sheet.

The key to this discovery lies in the size of the device and the unique way the superconducting currents travel through it. In smaller, traditional devices, the supercurrent only flows between neighbors, like people passing a message down a short line. In this large graphene bar, the supercurrent can travel in a coordinated way between distant terminals that are not right next to each other. The researchers found that as the current increased, different parts of the network switched from a superconducting state to a normal, resistive state one by one, rather than all at once. The first part to switch was the connection in the middle of the bar. Once this middle section switched, the current was forced to find a new path. Because the device is so large, the current could redistribute itself through other routes that were still superconducting. This redistribution caused the voltage measured in the middle to drop, creating the negative resistance effect. The team confirmed this by building a computer model of the network that included both the long-range superconducting connections and the normal electrical paths, which successfully reproduced the strange voltage drop seen in the experiment.

To ensure this was a fundamental property of the network and not just a fluke of their specific setup, the researchers tested the device under different conditions. They changed the number of electrons in the graphene, varied the temperature, and even built several different versions of the device using different materials and layer thicknesses. In every case, the negative resistance effect appeared, showing it is a robust feature of this type of large-scale network. They also ruled out other possibilities, such as the effect being caused by heat or by the device acting like a simple, single junction. By applying a magnetic field, they observed that the supercurrents were flowing through multiple distinct paths with different sizes, further proving that the device was acting as a complex, interconnected web rather than a simple wire. The magnetic field patterns showed that some currents were flowing between nearby contacts, while others were taking long, looping paths across the entire device, a behavior that would be impossible in a smaller, more confined structure.

This work opens a new door for understanding how electricity behaves in complex quantum networks. The ability to generate a negative resistance effect simply by changing the current in a large, multi-terminal device suggests new ways to build electronic circuits that can amplify signals or mix frequencies without needing many separate components. Beyond its potential use in circuits, the large size of the device offers a unique platform for studying how superconductivity interacts with other exotic states of matter that can exist in graphene. Because the device is so large, scientists can place gates and sensors in the middle of the material, far away from the edges, to study the bulk properties of the electrons without interference from the contacts. This "Josephson Hall bar" provides a versatile stage where the collective dance of superconducting electrons can be observed and controlled, offering a clearer view of the quantum world than ever before.

Technical Summary: Nonlocal Josephson Coupling and Current Redistribution in a Large-Area Graphene Multiterminal Josephson Junction

Problem Statement
Multiterminal Josephson junctions (MTJJs) offer a versatile platform for exploring emergent quantum phenomena, such as nontrivial current–phase relations and topological band structure simulation, by controlling supercurrents within an $(n-1)$ -dimensional phase space. However, conventional MTJJs are typically highly compact, with electrode separations ranging from tens to hundreds of nanometers. In these devices, superconducting transport is dominated by local coupling between nearest-neighbor leads. This dominance obscures physics arising from nonlocal interactions, such as crossed Andreev reflection, elastic cotunneling, and phase dragging, which require substantial coupling between distant terminals. Furthermore, existing compact geometries often fail to support the long mean free paths necessary for ballistic transport in large-area devices, limiting the exploration of nonlocal superconducting networks.

Methodology
The authors realized a "Josephson Hall bar" device to address these limitations. The device consists of a six-terminal monolayer graphene Hall bar (3 µm wide, 6 µm source–drain separation) encapsulated in hexagonal boron nitride (hBN) and contacted via edge-contact geometry. This micron-scale geometry supports substantial supercurrents due to graphene's long mean free path, enabling near-ballistic transport across the device.

The study employed the following experimental and theoretical approaches:

Transport Measurements: Current–voltage ( $I$ – $V$ ) characteristics were measured in a dilution refrigerator (10 mK) using various four-probe and two-probe configurations. Measurements included differential resistance ( $dV/dI$ ) and DC voltage responses under varying carrier densities, temperatures (up to 0.6 K), and perpendicular magnetic fields.
Device Variations: To test robustness, the authors fabricated six different devices with variations in graphene thickness (monolayer/bilayer), superconducting contacts (Niobium/Aluminum), and geometric modifications (segmented channels or increased source-drain distances).
Theoretical Modeling: The experimental data were analyzed using a Resistively Shunted Junction (RSJ) network model. This model incorporated an experimentally extracted conductance matrix to define normal-state resistive channels and included nonlocal Josephson couplings ( $I_{c,ij}$ ) between all node pairs. The model solved coupled nonlinear differential equations to simulate the switching sequence and current redistribution.

Key Results

Observation of Negative Differential Longitudinal Resistance (NDLR): The primary experimental finding is a pronounced NDLR in the four-probe response (specifically $dV_{23}/dI_{14}$ ) near the switching currents. This feature appears as a "back-bending" in the transfer curve where the internal voltage $V_{23}$ decreases as the bias current $I_{14}$ increases. Crucially, this NDLR is absent in two-terminal measurements ( $dV_{14}/dI_{14}$ ), indicating it is an intrinsic property of the multiterminal network rather than a measurement artifact.
Mechanism of NDLR: The authors attribute the NDLR to a two-stage process involving two distinct forms of nonlocality:
- Nonlocal Josephson Coupling: Determines the sequence in which different superconducting branches switch to a resistive state. The network switches sequentially rather than simultaneously; the central 2–3 junction switches first, followed by the 1–2 and 3–4 junctions.
- Nonlocal Normal-State Conductance: Once specific regions switch to a resistive state, the normal-state conductance network redistributes the dissipative current. This redistribution reduces the current flowing through the 2–3 branch, causing a drop in the measured voltage $V_{23}$ despite the increasing total bias current.
Robustness and Generality: The NDLR effect was observed across all tested device variations, regardless of contact material or graphene thickness, suggesting it arises from a network-level mechanism rather than specific material parameters. Control experiments where the graphene channel was segmented or the source-drain distance was increased (preventing supercurrent flow between specific terminals) resulted in the disappearance of the NDLR, confirming its dependence on nonlocal coupling.
Magnetic Field Dependence: The device exhibits complex magnetic interference patterns. The critical current shows a superposition of two modes: a conventional sinc-like Fraunhofer pattern (Mode 1) associated with short-range coupling between adjacent contacts, and a Gaussian-like suppression (Mode 2) arising from ensemble averaging over long Andreev trajectories traversing the device interior. The NDLR vanishes at the Fraunhofer nodes where the supercurrent in the relevant regions is suppressed.

Significance and Claims
The paper claims that the Josephson Hall bar geometry provides a distinct platform for studying nonlocal superconducting transport, moving beyond the nearest-neighbor connectivity that limits conventional MTJJs. The authors highlight several specific contributions:

Characterization Tool: The emergent NDLR serves as a sensitive probe for characterizing superconductor–semiconductor hybrid systems, specifically revealing the interplay between nonlocal Josephson coupling and normal-state conductance redistribution.
Nonlinear Circuit Potential: The device exhibits a continuously tunable, nonlinear DC response with negative differential resistance values around $-0.2$ kΩ and low dissipated power ( $\sim 10$ pW), suggesting potential utility as a low-impedance nonlinear element for circuits, though the paper does not propose specific circuit implementations.
Platform for Many-Body Physics: The micron-scale normal region allows for the investigation of superconducting proximity effects in graphene systems hosting exotic ground states (e.g., spin–orbit engineered or rhombohedral graphene) without the strong boundary effects and contact-induced doping that plague smaller junctions. The geometry facilitates the integration of local gating schemes to define mesoscopic regions like quantum dots within the superconducting network.

The authors maintain a modest stance, noting that a complete quantitative model accounting for distinct energy-phase relations in SNS channels, normal scattering, and nonequilibrium quasiparticle dynamics is still required to fully describe the device physics. They emphasize that the observed phenomena are generic to superconductor-semiconductor platforms but require further theoretical development for precise benchmarking.

More like this