We recently had the pleasure of talking with Michelle Simmons and John Martinis about John’s new role at Silicon Quantum Computing (SQC) and SQC’s overall approach for quantum computing.  John had only recently emerged a few days earlier from the mandatory traveler’s quarantine and when we talked to him he had only just started talking with the engineering team to fully understand the aspects of their technology.  He is on sabbatical leave from the University of California, Santa Barbara and will stay in Australia for the next six months. After that, they will decide what is needed and make appropriate plans at that time.

Perhaps it’s best to start with a description of the particular technology that Silicon Quantum Computing (SQC) is using. The technology is called Donor Spin Qubit technology and was originally conceived at the University of New South Wales. SQC was spun out as a commercial business from UNSW in August 2017 with $83 million AUS (about $60 million USD) in funding to pursue this technology.  

The technology consists of a (31P) phosphorus donor in embedded in an isotopically pure (28Si) silicon structure.  Normal silicon material contains a mixture of isotopes which is not an issue for standard semiconductor devices. But for quantum purposes 28Si is preferred because this isotope has zero spin and results in the best coherence times. SQC launched a project in December 2019 with Silex Systems Ltd. to develop a volume manufacturing process for this purified material.

Silicon Quantum Computing
Donor Spin Qubit Structure Showing a Phosphorus Atom Embedded in a Silicon Crystal

The Donor Spin Qubit technology has a lot of potential advantages. Donor qubits in silicon have been shown to have high fidelities (more than 99%) with long coherence times measured in seconds for the electron spin states. Recent results described in this paper in Nature magazine, show that these atom qubits also offer very fast two-qubit gates as low as 0.8 nanoseconds. For comparison, the delays in the Google Sycamore chip are 12-25 nanoseconds and other qubit technologies can have longer gate delays that could be as much as a microsecond or more. In addition, the die area of a spin qubit is significantly smaller than a superconducting qubit so you can have much smaller die sizes and pack many more qubits onto a 300 mm wafer.  This could provide significant cost and yield advantages in the future.

At Google, John spent considerable time looking at qubit noise issues and that will be one of the first areas he will investigate at SQC.  He is initially focused right now on charge noise and the overall goal will be to develop an extremely high quality qubit technology that they will be able to understand and model very accurately.  This will allow them to develop an initial 10 qubit test vehicle and then rapidly create scaled up follow-on designs that contain a larger number of high quality qubits. In addition, John’s experience allows him to think about how the whole system needs to fit together and provide guidance to the SQC team as they work to integrate the many disparate pieces needed to create a full system.

It was interesting talking to the John and Michelle together.  Although they have different backgrounds, one could see a lot of synergy between the two as they discussed their ideas and this will no doubt help the progress for spin qubit developments at SQC. We are looking forward to seeing many more interesting developments that will come out of this collaboration in the future.

October 2, 2020