Researchers integrate superinductor into commercial silicon chip

Researchers working with CIC nanoGUNE, Quantum Motion, and UK university partners have demonstrated a superconducting superinductor fabricated within a commercial silicon integrated circuit, using titanium nitride structures already present in the semiconductor manufacturing process.

The work, published in Nature Electronics on 4 November 2026, establishes a route for integrating high-inductance superconducting circuitry into advanced CMOS devices without requiring the complete circuit to be manufactured through a separate specialist superconducting process.

The researchers used GlobalFoundries’ 22 nm fully depleted silicon-on-insulator technology to produce the superinductor and integrate it with a silicon quantum dot. The resulting arrangement operates as a radiofrequency single-electron transistor, a highly sensitive charge sensor used in semiconductor quantum computing systems.

The research involves scientists from nanoGUNE’s Quantum Hardware group, Quantum Motion, the London Centre for Nanotechnology, Imperial College London, and the University of Cambridge. An earlier version of the work was publicly available as a preprint in July 2025, while the new peer-reviewed journal article reports the completed experimental study.

Superinductors are circuit elements engineered to provide exceptionally high inductance and impedance. Unlike conventional integrated inductors, which generally rely on the magnetic field generated around a conducting structure, superconducting superinductors can exploit the kinetic properties of the material carrying the electrical current.

In a superconductor, electrical current is transported by paired charge carriers known as Cooper pairs. Their inertia contributes to an effect called kinetic inductance, which can become substantial in thin films with suitable superconducting properties.

This creates an opportunity to obtain high inductance within a much smaller physical area than would be practical using a conventional patterned inductor. High-impedance resonant circuits are particularly relevant to quantum electronics because they can modify the interaction between a sensing element and the electrical measurement system.

Many superconducting circuit implementations use specialised Josephson junction arrays or superconducting nanowires. These structures can provide the necessary electrical properties, but their fabrication and integration may introduce additional process requirements beyond those of conventional semiconductor manufacturing.

The new work instead exploits titanium nitride already associated with the commercial CMOS process. The researchers investigated thin-film structures available in the 22 nm technology and measured kinetic inductance of approximately one nanohenry per square.

Kinetic inductance expressed per square provides a way to describe the electrical behaviour of a thin film independently of the absolute dimensions of an individual patterned conductor. The effective inductance of the finished circuit still depends on its geometry, material characteristics, and operating conditions.

The team used the material to create an integrated nonlinear superinductor with a reported footprint of approximately 18 square micrometres. The journal paper describes an improvement in inductance density of four orders of magnitude compared with conventional spiral inductors used in CMOS circuitry.

That comparison concerns the density of the inductive element rather than demonstrating that an entire quantum processor or associated control system can be reduced by the same factor. Packaging, readout wiring, supporting electronics, and other circuit elements continue to influence the dimensions of a complete device.

The researchers integrated the superinductor into a resonator and coupled it to a silicon quantum dot fabricated within the same integrated circuit. The combination forms a radiofrequency single-electron transistor, in which changes in the electrical state of the quantum dot influence the response of the resonant measurement circuit.

A resonator responds strongly near its characteristic frequency, allowing small electrical changes to alter the phase or amplitude of a reflected radiofrequency signal. Measuring that response provides a means of detecting charge changes associated with the quantum dot.

The sensitivity of such a measurement depends on electrical noise, resonator behaviour, signal strength, and the losses introduced by the surrounding circuit. Connections between separate devices can add unwanted capacitance and other parasitic effects, reducing the response available for measurement.

Integrating the resonator and sensing element within the same chip reduces some of these parasitic contributions. The high impedance of the superconducting element further increases the electrical response associated with changes in the quantum dot.

The researchers also identify nonlinear kinetic inductance as an important part of the measured behaviour. In a nonlinear resonator, the response changes with the applied signal, creating operating conditions that can enhance the sensitivity of the charge measurement.

The published study reports a power signal-to-noise ratio of 4,500 over an integration time of four nanoseconds. Its analysis indicates an improvement exceeding two orders of magnitude compared with previously reported systems under the comparisons used in the research.

These measurements relate to the demonstrated charge-sensing device and its experimental operating conditions. They should not be interpreted as evidence that the team has already achieved high-fidelity spin-qubit readout in four nanoseconds or demonstrated a complete multi-qubit computing system.

Practical spin-qubit readout requires the state of a quantum system to be converted into a measurable electrical response while maintaining sufficient fidelity and controlling interactions with neighbouring devices. Performance also depends on the readout protocol, device characteristics, and the measurement chain.

The research team’s next objective is to integrate the resonator with multi-gate transistor structures and demonstrate high-speed, high-fidelity spin readout at timescales substantially below one microsecond.

That development would require additional system-level experiments beyond the reported charge-sensor characterisation. Device variation, readout fidelity, power dissipation, and operation within larger arrays will become increasingly important as the architecture is extended.

The compatibility with an established semiconductor manufacturing process also creates possibilities outside quantum computing. High-kinetic-inductance structures may be useful in cryogenic sensing, resonator arrays, parametric circuits, and instrumentation requiring compact superconducting elements.

Radio astronomy detectors and experimental quantum simulators are among the possible applications identified by the researchers. Each would require additional engineering and validation, and the current study does not establish commercial readiness for those systems.

The result demonstrates that specialised superconducting circuit behaviour can be obtained from material structures present in commercial advanced CMOS technology. Future work will determine how the integration method performs when extended from a single charge sensor to larger and more complex cryogenic electronic systems.