In research, published today in Nature Electronics, the Quantum Motion team has demonstrated the fastest dispersive readout of a spin qubit in a silicon metal-oxide-semiconductor device. To achieve this result, the team developed a new technique for fast and scalable readout spin qubit readout called radiofrequency electron cascade. With this new approach, which is directly compatible with standard 300mm industrial silicon manufacturing, the team went on to demonstrate a 2 qubit gate between coupled spin qubits, positioning radiofrequency cascade as a compact, fast, and scalable readout solution for utility-scale quantum computers based on semiconductor spin qubits.
Moving from individual qubits to dense arrays requires moving away from bulky sensing components. The current standard, the radio-frequency single-electron transistor (SET), occupies significant chip area and limits qubit connectivity. Existing dispersive alternatives are more compact but have historically suffered from low sensitivity in planar metal-oxide-semiconductor (MOS) silicon devices, due to the relatively weak electrostatic coupling between the gates and the quantum dots.
The paper introduces a new technique called the radio-frequency (RF) electron cascade. Rather than measuring the small alternating-current signal generated by cyclic tunnelling between two charge states in a double quantum dot (DQD), the team introduced a third quantum dot coupled to a charge reservoir. This third dot acts as an amplifier: the strong capacitive coupling between the DQD and the third dot means that every charge transition in the DQD synchronously drives an electron in and out of the reservoir, substantially amplifying the detectable signal.
This cascaded charge signal provides more than 35 dB of power amplification compared to standard dispersive readout, while retaining a key advantage of these methods. The advantage is that the qubit state is preserved after readout.
The technique achieved a minimum integration time of 8 μs, the highest performance ever reported for dispersive readout in a planar MOS process, and an improvement of over two orders of magnitude compared to prior state-of-the-art results on the same platform.
The team also demonstrated singlet-triplet readout of two electron spin states and coherent exchange control in a double quantum dot device, the two-qubit interaction that forms the basis for entangling gates. Dephasing times of up to 500 ns were observed, with a gate quality factor exceeding 10.
“By applying a radio‑frequency tone, we found we can repeatedly trigger an electron cascade during dispersive spin qubit readout, effectively giving built-in signal amplification. In our planar silicon MOS quantum‑dot array, that lets us determine the qubit state in about 8 microseconds, substantially faster than prior in‑situ dispersive readout results in the same platform. Because it’s rf‑driven, it also points toward reading many qubits in parallel by assigning each one a different frequency.” says Jacob Chittock-Wood, Quantum Engineer and lead author.
While spin qubits have long shown potential, the challenge has always been moving from laboratory devices to the mass-producible chips that a large-scale quantum computer will require. By demonstrating these results in a 300mm natural silicon MOS process, the same infrastructure used to build the chips in your smartphone, the team have shown that the building blocks of a quantum processor are compatible with the world’s most advanced silicon manufacturing. Advanced semiconductor manufacturing offers a promising path to scaling up silicon-based quantum processors by improving yield, uniformity, and the long-term integration of qubits with classical control electronics.
The next phase of this work involves translating these results to isotopically enriched silicon to further extend coherence times and gate fidelities. Combining high-performance readout with industrial manufacturing, Quantum Motion continues to build the case that spin qubits are the most viable route toward building a large-scale, fault-tolerant quantum computer.
This work included contributions from: Jacob Chittock-Wood, Ross Leon, Michael Fogarty, Tara Murphy, Felix-Ekkehard von Horstig, Sofia Patomäki, Giovanni Oakes, James Williams, Nathan Johnson (UCL), Julien Jussot (IMEC), Stefan Kubicek (IMEC), Bogdan Govoreanu (IMEC), David Wise, John J. L. Morton and M. Fernando Gonzalez-Zalba.