Semiconductor Wafer Containing Spin Qubits. Credit: Diraq

Although it may first appear that the difference between operating qubits at 15 millikelvin (0.015K) and 1 kelvin doesn’t sound like very much, but for quantum computing it actually makes a world of difference.

On one hand, the cooling power of the dilution refrigerators are more than an order of magnitude better at the higher temperature. The qubits and the associated control lines generate heat. Even though the heat generated from each qubit is small, it puts a limit on the number of qubits that can be cooled by the dilution refrigerator. For this and other reasons, GQI estimates that the maximum sized qubit module that can be operated at a 15 millikelvin temperature might be limited to 1500 qubits due to the 10 or 20 microwatts of cooling power delivered by the dilution refrigerator. At a 1 kelvin temperature, the dilution refrigerator might be have 1000 microwatts or more of cooling power which could cool individual modules containing 10’s or 100’s of thousands of qubits. On the other hand, raising the temperature increases the amount of thermal activity in the electrons which causes noise and decoherence which will degrade the accuracy of the quantum computation.

So the trick is to find a way to operate at the 1 kelvin temperature without degrading the qubit quality measures and this is what Diraq announced they have achieved. This advantage goes hand-in-hand with the other advantage of the spin qubit technology Diraq is using. Spin qubits require many orders of magnitude smaller die area than alternatives like superconducting technology. So a large number of qubits can be fit onto a small semiconductor die. The company plans on using this approach to develop a large scale quantum processor without needing to resort to a complicated multi-module approach that other companies are pursuing. Key metrics reported by the company include a single qubit gate fidelity of up to 99.85%, two qubit gate fidelity of up to 98.92%, and initialization and readout fidelity of up to 99.34%. These levels of are getting close to the realm where applying error correction algorithms start to become practical.

The company has published a technical paper in Nature magazine titled High-fidelity spin qubit operation and algorithmic initialization above 1K which can be accessed here. The company has already a press release announcing the development available here and has posted a blog on their website with additional technical detail here.

March 27, 2024