By David Reilly
‍

Some years ago, many of us realized that scaling up quantum computers would likely require developing integrated cryogenic control electronics, but in that era, as a bunch of academic physicists, this looked like a daunting challenge. I was keen to learn more of what was possible and better understand what the limits might be, even if that meant being regarded by the physicists as a nuts-and-bolts engineer. The real engineers of course always regarded us as physicists.

‍

With support from the ARC, US-Gov, and Microsoft Research, our group at the University of Sydney was able to grow our physics team to now include talented ASIC engineers with PhDs. In Kushal Das, I found a person filled with creative ideas and together we had some real fun thinking about ways of controlling qubits with tiny amounts of power.

‍

We took inspiration from different technologies: the electronics of the 1970s where the number of transistors were limited and memory was expensive, provided examples of clever functionality without the brute force resources. Then there are pixel refresh circuits in display screens and finite state machines that operate vending machines, the circuits that powered synthesizers and drum machines (like the LinnDrum or Wendel) achieved unbelievable timing precision, looping, branching, and waveform synthesis with tiny circuits and miniscule power. Necessity as the mother of invention and all that.

‍

Making voltage pulses from room temperature instruments meant milliwatts of power just to charge the cable. We need to get rid of the low-impedance cables! Wait! If we just move small amounts of charge between capacitors we can make big voltages!

‍

Over a decade, what started as perhaps naïve ambition began yielding real fruit.

‍

Showing that we could design chips that worked at these temperatures and generated the signals needed for control opened the next chapter in scepticism, “Yeah, fine, it works, but it will kill qubit fidelity”.

‍

Spin qubits provided an ideal test-vehicle. Two-qubit entangling gates based on the exchange interaction can be exponentially sensitive to voltage noise on the gate electrode. If — and it was a big if — our integrated CMOS control technology could be used to perform 2-qubit gates without degrading fidelity, this would be an important demonstration.

‍

Enter Diraq and longtime friend and colleague Andrew Dzurak with the best spin qubits around. We set out to put cryo-CMOS control to the test with MOS-based silicon qubits. The meticulous work of Sam Bartee, Will Gilbert, and Kun Zuo drove the experiments that benchmarked performance under cryo-CMOS control.

‍

And the result? Even with all the digital and analog blocks powered up and functioning, the qubits hardly notice the 100,000 transistors switching in the control chip right next door! The all-important 2-qubit gate performs much like room temperature control!

This is obviously good news for the future of spin qubits and opens the prospect of scaling this platform with integrated control. But, why does it work?

Truth is, we are still trying to under why this approach works so well. Minimizing the power dissipation leverages low leakage of transistors at cryogenic temperatures. The FDSOI platform allows us to carefully set the back gate to optimize threshold voltage with temperature. But what about the noise and crosstalk? Likely, there are a few aspects that conspire to reduce the impact of proximal cryo-control on qubit fidelity. Firstly, the thermal noise of the control circuits is suppressed by 2-orders of magnitude or more. Further, the capacitive circuit breaks the galvanic connection between hot noisy control and the sensitive qubits – effectively ac coupling. This limits 1/f noise coupling to the qubits. Lastly, heterogenous integration, the idea of having control proximal rather than totally monolithically integrated with qubits opens the prospect of independent thermal management and cooling for each chip.

‍

So, in the end, integrated milli-kelvin control might just work. Certainly, we will now bring to bear the know-how used in this experiment to design many new and scaled-up cryo-control systems.

‍

At Emergence Quantum we want to see this know-how having impact across a diverse range of technologies, from different qubit platforms to scaled-up sensing systems, and new energy-efficient compute paradigms that will power the data centres of the future.

‍

Interested? Talk to us!