In a paper published in Nature Sensors, the Quantum Motion team has demonstrated the first single-electron box (SEB) integrated within a planar silicon metal-oxide-semiconductor (MOS) process, achieving state-of-the-art readout fidelities with a significantly smaller chip footprint than conventional sensors. The work also introduces a Hidden Markov Model approach to spin-state classification that provides a more accurate picture of qubit readout than standard methods.
Readout of a spin qubit (determining whether its spin is 0 or 1) is one of the most basic operations in a quantum processor, and one of the hardest to scale. The current standard sensor, the single-electron transistor (SET), provides high fidelity but occupies substantial chip area, limiting how densely qubits can be connected to one another. More compact dispersive alternatives exist, but until now they had not been demonstrated in planar MOS devices MOS fabrication processes are the industry standard for Very Large-Scale Integration.
The team integrated an SEB, a single quantum dot coupled to a charge reservoir via a tuneable barrier gate, within a bilinear unit cell of planar MOS quantum dots fabricated on an industrial-grade 300 mm wafer. The sensor sits in a parallel row to the qubit dots rather than co-linearly, meaning it does not restrict qubit connectivity as the array scales.
Using the electrical tunability provided by multiple overlapping gates, the team independently optimised the SEB tunnel rate and the coupling to the double quantum dot (DQD) to maximise signal contrast. This yielded readout fidelities of 99.92% which is state-of-the-art for any dispersive charge sensing implementation. .
The team also developed a Hidden Markov Model (HMM) of the two-electron spin dynamics. Standard readout methods assume only two spin states are relevant, which can systematically overestimate fidelity when a third state has not fully decayed before the measurement window closes. The HMM accounts for all three accessible states providing a more accurate fidelity estimate and, in certain measurement regimes, improving actual classification performance.
Constance Lainé, lead author and Quantum Engineer at Quantum Motion, said: “ The SEB’s compact footprint opens a path toward denser, more connected qubit arrays. Unlike previous MOS sensor demonstrations, the SEB is placed non-collinearly with the qubit array, making it compatible with two-dimensional architectures.”
The work also shows preliminary evidence of reproducibility across multiple devices fabricated on the same process. This is an important indicator for the yield and uniformity required at scale.
Fernando González-Zalba, Principal Quantum Engineer at Quantum Motion, said: “The HMM framework developed in the work is not limited to spin qubits. It provides a general tool for distinguishing between multiple overlapping charge signals, with potential applications across spin qubits in semiconductor, Majorana, and electron-on-helium qubit platforms. Furthermore, the techniques to improve the SEB have broader implications for fields such nanoscale thermometry, low-power cryogenic electronics and quantum-limited parametric amplification.”
This work included contributions from Constance Lainé, Giovanni Oakes, Virginia Ciriano-Tejel, Jacob Chittock-Wood, Lorenzo Peri, Michael Fogarty, Sofia Patomäki, Stefan Kubicek (IMEC), David Wise, Ross Leon, M. Fernando Gonzalez-Zalba, and John J. L. Morton.
Read the full paper here: https://www.nature.com/articles/s44460-026-00084-6