Silicon-28 spin qubits and what still limits coherence

Short answer: Isotopically enriched silicon-28 removes most nuclear spins that disturb electron spin qubits, but the three papers here show that other limits remain. A 2019 Nature Communications paper found that residual silicon-29 nuclei limit coherence at low magnetic field (T2* of 1.4 microseconds at 150 millitesla), a 2024 npj Quantum Information paper reported fast singlet-triplet oscillation with a quality factor above 580, and a paper listed on exa.ai reported a CNOT gate in 270 nanoseconds in a quadruple dot. These are small-device laboratory results.

2019: low-field operation and residual nuclei

The Nature Communications abstract says single-electron spin qubits usually use magnetic fields of about 1 tesla or more for spin-dependent-tunnelling readout, which requires demanding microwave engineering and limits large-scale prospects. Singlet-triplet readout works in much lower fields.

The authors demonstrate low-field operation of metal-oxide-silicon quantum dot qubits by combining coherent single-spin control with high-fidelity, single-shot, Pauli-spin-blockade readout. They find the qubits decohere faster at low field, with T2 Rabi of 18.6 microseconds and T2* of 1.4 microseconds at 150 millitesla. Coherence is limited by spin flips of residual silicon-29 nuclei in the isotopically enriched silicon-28 host, which happen more often at lower fields. They conclude that new trade-offs are needed for frequency stabilization, and that isotopic enrichment of device substrates is important for a scalable silicon processor.

2024: a micromagnet and fast oscillations

The npj Quantum Information abstract concerns baseband control of encoded spin qubits with a micromagnet in isotopically purified silicon. The authors demonstrate fast singlet-triplet qubit oscillation, about 100 MHz, in a gate-defined double quantum dot in silicon-28 and silicon-germanium with an on-chip micromagnet, with an oscillation quality factor above 580. They analyze T2 as a function of detuning and external field. In weak fields, coherence is limited by frequency-independent noise faster than the typical data acquisition time of about 100 ms, which holds T2 below 1 microsecond in the ergodic limit. They also present evidence of coherent coupling of the qubit with the spin states of a nearby dot, suggesting a charge-based two-qubit gate may be possible in a (1,1) charge configuration. That is a suggestion.

A quadruple dot

An exa.ai library listing reproduces the abstract of a paper by Sigillito, Loy, Zajac, Gullans, Edge and Petta, with the journal and year not shown in the text read here, so its venue is unconfirmed. It reports coherent spin control in a quadruple quantum dot fabricated with isotopically enriched silicon-28. The authors tune the charge configuration down to the single-electron regime and demonstrate tunable interdot tunnel couplings as large as 20 GHz, which enables exchange-based two-qubit gates. They achieve site-selective single-spin rotations using electric dipole spin resonance in a magnetic field gradient, and execute a resonant CNOT gate between two adjacent spins in 270 nanoseconds.

Limits

Enrichment helps but does not remove every limit. The 2019 result shows low-field coherence is limited by residual nuclei, and the 2024 result shows noise limiting T2* below 1 microsecond in weak fields. None of the abstracts reports a fidelity for the CNOT gate.

FAQ

Why use silicon-28?

It has no nuclear spin, so it reduces nuclear noise on electron spins, as the 2019 abstract's reference to enriched host material implies.

What still limits coherence?

Residual silicon-29 nuclei in 2019, and fast noise in weak fields in 2024.

What gate speed was reported?

A resonant CNOT in 270 nanoseconds, in the exa.ai listing.

Is the exa.ai paper verified?

Its venue and year could not be confirmed.

Is this a large silicon processor?

No. These are devices with a few dots.