UK consortium starts work on cryogenic CMOS
Government agency Innovate UK has awarded a grant of £6.5m ($8.9m) to a seven-member consortium that plans to develop cryogenic memory and other IP cores to support quantum computing.
The aim of the project is to make it possible to put control circuitry much closer to the qubit elements of quantum computers, which generally need to operate at close to the temperature of liquid helium. Today, to allow conventional CMOS memories and circuits to operate and to avoid them heating the quantum circuits, the controls are placed some way away and linked by multiple cables. The amount of cabling required for all the qubits presents a significant barrier to scaling up qubit capacity as well as causing high processing latencies.
To support the more compact designs needed, CMOS circuits and processes need to be identified that can operate significantly below the typical limit of -40°C. The aim of this project is to model the change in behavior at cryogenic temperatures and use those models to inform the design of a portfolio of CryoCMOS IP. Those IP cores should then make it possible to design custom control chips able to operate inside a cryostat. That in turn would support both vertically integrated systems companies with quantum-computing projects as well as smaller startups.
The transistor-modeling work is is being done by SemiWise and the quantum-computing research group at the University of Glasgow, with the results fed into Synopsys’ TCAD tools. A combination of measurements and simulation data will be used by SemiWise to re-center a foundry PDK for cryogenic temperatures and to enable the cryogenic circuit design. SureCore has taken on the job of using the simulations to design memory macros, with the aim of both enabling cryogenic operation and storage while minimizing heat dissipation.
Quantum architectures
“As these quantum computers will be doing intense computations, there will be huge demand for memory so the savings in power and hence heat will be critical to the operational success of the cryo control chips,” said SureCore CEO Paul Wells.
Cryogenic-chamber expertise is provided by Oxford Instruments which manufactures cryogenic systems. The remaining consortium members, Universal Quantum and SEEQC, represent end-user needs and will determine what IP blocks the project will need to create for the CryoCMOS chips and will take part in the fabrication of test chips.
Matthew Hutchins, chief product officer and co-founder of SEEQC, said: “This project provides an exciting opportunity to incorporate CryoCMOS into our chip-based, single flux quantum architecture that can operate at the same temperatures as qubits and provides an advantageous interface for CryoCMOS.”
By working as a team, the project members expect to be able to achieve results in less than three years, but that this will be much shorter than if they tried working individually.
“This consortium enables us to gain vital knowledge in low-temperature electronics, streamline our ASIC development work and accelerate our development roadmap as we work towards building the world’s first million-qubit quantum computer,” said Adam Glibbery, IC design lead at Universal Quantum.
Universal Quantum’s design can operate at higher temperatures than many proposals for a quantum computer but still requires CMOS control circuits to run well below -40°C.
