Spin qubits · Silicon
Silicon spin qubits reached microsecond coherence both with and without isotope purification
Save this development or follow its topic to track what changes.
Short answer
Both isotopically purified and cleverly engineered natural-silicon dots reached microsecond-scale spin coherence, so purification is not the only lever.
What happened
Eng and colleagues built a voltage-controlled triple quantum dot in 28Si-enriched Si/SiGe and measured a dephasing time T2* of 2.31 μs, against about 10 ns in GaAs and 360–900 ns previously reported for natural-silicon dots; the remaining magnetic noise appeared to be electronic rather than nuclear. A year later Takeda and colleagues used a cobalt micromagnet in a natural-silicon double dot and reported T2* of 1.83 μs and an average single-qubit Clifford gate fidelity of 99.6%, with about 0.02% crosstalk between the two addressable spins.
Why it matters
Isotope purification is expensive and was seen as the route to long coherence in silicon. These results show device design can recover much of the gap, and that once nuclear noise is reduced, other noise sources (charge noise, paramagnetic defects, microwave heating) become the limit.
Evidence
- Study type
- Single-device low-temperature experiments on Si/SiGe quantum dots
- Sample
- One device in each paper
- Journal
- Science Advances · peer reviewed
- Replication
- Each result is a single device; broader device-to-device variability is not reported here
- Limitations
- Only single-qubit operations; noise sources are inferred rather than identified; the heating explanation for faster decay at high drive is not measured directly.
What this connects to
Sources
The 2 studies this explanation is built from, by the role each plays. Every source links to PaperFren’s explanation of it and to the original paper.
Primary study
- Can ordinary silicon host fast, accurate spin qubits?
A carefully designed tiny magnet let electron spins in ordinary (non-purified) silicon be flipped fast enough and cleanly enough to pass the error threshold for fault-tolerant quantum computing.
What it does not showLimitations
Only single-qubit operations are shown; a two-qubit gate, which is needed for real computation, is proposed but not demonstrated. The results come from one device, and detailed data mostly from one of the two dots because the charge sensor was more sensitive to it. The heating explanation for faster decay at high drive is inferred rather than measured directly, and nuclear-spin noise from natural silicon still limits dephasing.
PaperFren explanationStudy with cards and a quizOriginal paper (DOI)cc by
Supporting evidence
- Does purified silicon make an all-electric spin qubit quieter?
Using isotopically purified silicon, a three-dot spin qubit controlled only by voltages kept its phase hundreds of times longer than in gallium arsenide, leaving charge noise as the main limit.
What it does not showLimitations
It is a single device, and the authors say the experiments do little to identify the physical sources of either the magnetic or the charge noise; the paramagnetic source is only guessed to come from fabrication. Gate fidelities, two-qubit operations and randomized benchmarking were not measured, only argued to be feasible. The singlet-triplet splitting on one side of the device was too small to use, hinting at valley-splitting variability.
PaperFren explanationStudy with cards and a quizOriginal paper (DOI)cc by
Before
Nuclear spins in natural silicon were assumed to cap coherence well below a microsecond unless the material was isotopically purified.
Now
Purified silicon gave the longest dephasing time, but a natural-silicon device came close and demonstrated high single-qubit fidelity. Each result is from one device, neither shows a two-qubit gate, and the two use different qubit encodings, so they are not a like-for-like comparison.