A collaboration between Diraq and new venture Emergence Quantum has given rise to exquisite classical control that doesn’t compromise quantum performance.

Diraq is well on its way to building a quantum chip patterned with millions of silicon spin qubits, and one step along this path involves integrating classical hardware to control and measure the qubits.  A key advantage of silicon is its compatibility with the complementary metal-oxide semiconductor (CMOS) technology used widely in the microchip industry. But modern CMOS circuits generate heat and stray electrical fields that can degrade qubit performance. Reporting in Nature, Diraq and Emergence Quantum have shown that CMOS circuits held at cryogenic temperatures can control silicon spin qubits without affecting their performance.

Silicon spin qubits usually operate at temperatures that are even colder than the farthest reaches of deep space, whereas most CMOS circuits are designed to work at room temperature. Long cables can be used to connect the hot control system to each ultracold qubit, but scaling up to billions of qubits quickly renders this solution impossible. An alternative approach is to design CMOS technology that functions at ultracold temperatures — and to ensure that putting it right next to the quantum chip doesn’t compromise the qubits.

A new alliance

The two teams' solution is a cryo-CMOS control system that is integrated into the main quantum system as a distinct ‘chiplet’. It comes on the back of a long-standing research collaboration between UNSW Sydney, the University of Sydney and Diraq. And it heralds Diraq’s partnership with Emergence Quantum, a new company co-founded by Dr Thomas Ohki and Professor David Reilly at the University of Sydney.

The partnership has been cemented by Diraq’s recruitment of Dr Samuel Bartee, Reilly’s former student, and by Emergence joining the Diraq-led consortium for the Quantum Benchmarking Initiative, the United States’ Defense Advanced Research Projects Agency (DARPA) program.

Reilly’s team previously reported in Nature Electronics that this control chiplet could be integrated with quantum dots made from gallium arsenide. Now the teams have come together to show that cryo-CMOS is a viable strategy for minimizing disturbances to Diraq’s spin qubit system.

Low-power control

Bartee and co-authors report in Nature that the coherence time of silicon spin qubits — a key measure of qubit quality — is not measurably shortened when their cryo-CMOS chiplet is in control. This means that the chiplet can be integrated into Diraq's qubit system, doing away with the need for quantum–classical interconnects spanning 3 metres and 300 kelvin.

The best part is that the results were obtained within a power envelope of just 10 microwatts, the vast majority of which was expended on the digital systems. The analogue components dissipate only around 20 nanowatts per megahertz per qubit, which means that the system can be scaled up to millions of qubits without the need for a radical redesign of the cooling system.

Noise nuisance

Heat isn’t the only concern when integrating classical components into quantum chips. Qubit operations can also be affected by the electrical noise that arises from the constant switching of hundreds of thousands of transistors. Interactions between qubits are extremely sensitive to this noise, so logic operations involving two or more qubits are particularly at risk.  

The results in Nature show that interactions between qubits controlled with the cryo-CMOS chiplet resemble those between qubits manipulated with a standard room-temperature control system. The standard system is removed from the quantum system by means of long cables, so electrical noise is not an issue there. The fact that the cryo-CMOS system shows remarkably similar behaviour indicates that the effect of electrical noise from the chiplet is negligible.  

Looking forward

This is a win for integrated control of silicon spin qubits: the cryo-CMOS chiplet doesn’t reduce the coherence time of individual qubits, and it doesn’t degrade the performance of multi-qubit interactions. Nonetheless, the collaboration sees ways in which the design could be improved still further. Excess heat could be avoided by cooling the chiplet and qubits independently, instead of using the same equipment. The modular chiplet architecture should make this straightforward to implement. 

Watch the explainer video here.