Researchers at the University of Hong Kong have demonstrated a single silicon carbide MOSFET behaving like a biological neuron at 10 millikelvin, about one thousandth of a degree above absolute zero. The result, published in Nature Communications on June 12, 2026, rests on an "S-shape" negative differential resistance effect that appears in SiC devices only when cooled below 2 kelvin.

Why does the quantum stack need cryogenic control circuits?

The team framed the work as a remedy for a thermal bottleneck that has constrained quantum machines since the field began. Qubits must sit at millikelvin temperatures, yet the silicon control electronics that read and drive them generate waste heat and are therefore parked at higher stages of the dilution refrigerator. That separation forces long, signal-degrading cable runs between qubit and controller.

Each cable adds resistance and thermal conductance to the stack, degrading qubit readout fidelity and forcing engineers to keep the largest machines at a few thousand qubits. Moving the controller to the same cold stage would shorten those paths and lift the wiring cap.

"Our work introduces a hardware platform that can be integrated alongside quantum processors," said Professor Yuhao Zhang of HKU's Department of Electrical and Computer Engineering and the Centre for Advanced Semiconductors and Integrated Circuits. He added that circuits built this way are "thousands of times more energy-efficient than conventional electronics, significantly reducing the thermal load on cryogenic systems."

Takeaway: putting the controller in the same cold stage as the qubit is the prize; less heat, fewer cables, larger machines become feasible.

How does a standard SiC transistor act like a neuron?

The mechanism is electron-donor impact ionization, in which electrons accelerated by the device's own electric field knock bound carriers free inside the silicon carbide lattice. That avalanche produces an S-shaped current-voltage curve, and the curve lets a single transistor emit brief voltage spikes rather than smooth analog signals. The spikes reproduce the energy-efficient firing pattern of biological neurons.

Unlike other negative differential resistance devices that rely on internal heating, the silicon carbide mechanism traces directly to the material's atomic structure. As a result, the team reported, the behavior stays stable across different manufacturing batches, a critical requirement for any device intended for volume deployment.

Lead PhD student Xin Yang emphasized the manufacturing angle. "This is a robust and scalable approach," Yang said. "Because SiC is already used globally in electric vehicles and power grids, we can leverage existing industrial foundries to manufacture these cryogenic chips on 300-millimeter wafers." SiC is a mature, high-volume material already shipping in commercial EVs and grid infrastructure, not a laboratory curiosity.

Takeaway: the innovation is the operating mode, not the device itself, which keeps the path to high-volume production open.

What changes when the neurons can be cascaded?

The HKU group also showed that the artificial neurons can be linked into larger networks that still operate at cryogenic temperatures. That capability matters for two quantum workloads: quantum error correction and real-time quantum control, both of which demand dense, low-power computation beside the qubits themselves.

The same cold tolerance could carry the circuits beyond Earth. Because the devices function in extreme cold without active heating, the team suggests future systems could operate on the lunar surface or in the outer solar system, where ambient temperatures already approach the chip's working range.

Takeaway: a cold-tolerant, energy-efficient neuromorphic layer may eventually sit beside both qubits and spacecraft instruments.

With SiC already flowing through automotive and grid foundries on 300-millimeter lines, the research points to a commercial route that does not require exotic new fabs. Integration with live quantum hardware remains the next milestone on the team's roadmap.