In the race to the quantum computer, different physical media are competing to encode quantum information. What do they have in common? They form systems with two energy levels - the quantum bits or qubits - which must be able to be initialized, read and manipulated.
Electron or holes spins are promising candidates because they can now be isolated in silicon using technology compatible with industrial microelectronics processes.
Electron spins can be manipulated while locally applying a magnetic field that oscillates at a microwave frequency. However, this manipulation is rather slow and it dissipates locally a lot of heat that reduces the performance of qubits. An alternative is to insert micro-magnets that couple the spins to an electric field at the microwave frequency, but this increases the size of the qubits and makes their large-scale integration more complex.
In 2016, IRIG and CEA-Leti researchers demonstrated that this situation is quite different with holes. Their spin can be controlled "naturally" via an electric field, thanks to "spin-orbit coupling". These qubits therefore materialize in the form of transistors cooled to a very low temperatures, for which coherent spin manipulation only requires sending a microwave signal to the transistor gate.
Recently, researchers at the Delft University of Technology (Netherlands) developed a 4-qubit processor based on hole spins in germanium, a feat hailed by the entire spin qubit community. However, in their experiment (as in the first CEA experiments), the electrical control of the qubits exposes them to the surrounding electrical noise, limiting their coherence times to much lower values than those of electron qubits.
By finely controlling a single hole spin in silicon, Irig researchers could demonstrate that there is a sweet spot where the hole qubit becomes almost insensitive to electrical noise, while remaining tunable. This ideal configuration, which corresponds to a particular orientation of the static external magnetic field, is in agreement with theoretical models and should be realizable in other materials such as germanium.
The coherence times obtained in this sweet spot are close to 100 µs, surpassing the previous reported values by more than an order of magnitude. They are now very close to the values for electron qubits electrically controlled with micro-magnets and obtained in isotopically purified silicon. This option reduces the perturbations caused by the nuclear spins of silicon 29Si which is present in natural silicon at level up to 5% and optimizes the coherence time of electron spins.