In the battle to build the world’s first useful quantum computers, one company has taken an entirely different approach to the other frontrunners. The conventional approach is to gradually increase the size and power of these devices and test as you go.
But PsiQuantum, a startup based in Palo Alto, California, is gambling on the opposite approach. The company is investing heavily in quantum technologies that are compatible with chip-making fabrication plants that already exist. By using these facilities, their goal is to mass-produce powerful silicon-based quantum computers from the very beginning.
This week, they reveal how well this approach is going and discuss the challenges that still lie ahead.
Photon Guns
Founded in 2016, PsiQuantum hit the headlines in 2021 when it raised $700 million to pursue its goal of building useful quantum computers within a decade. This week, it announced a similar injection from the Australian government bring its total funding to some $1.3 billion. That makes it one of the best funded startups in history.
The excitement is largely because of PsiQuantum’s unique approach. A key decision is its choice of quantum bits or qubits. Other companies are focusing on superconducting qubits, ion traps, neutral atoms, quantum dots and so on.
PsiQuantum has opted to use photons. The advantage is that photons do not easily interact with the environment, so their quantum nature is relatively stable. That’s important for computation.
Paradoxically, this reluctance to interact is also the main disadvantage of photons. It’s hard to make them interact with each other in a way that processes information.
But various groups have demonstrated optical quantum computing and PsiQuantum was founded by researchers in this area from Imperial College London and the University of Bristol.
Optical quantum computing works by creating photons or pairs of them, guiding them through channels carved into silicon where they can interact and then measuring their properties with highly specialized detectors.
PsiQuantum intends to do all this with silicon wafers. Their bold idea was that we already know how to make silicon chips on a vast scale by mass-production. Chip fabrication plants cost billions to build so there is a significant advantage in being able to use this current technology.
And by making bigger, more densely packed chips, optical quantum computers can scale relatively easily. Unlike other designs where scaling will be much harder.
So all the focus has been on how to make the manufacture optical quantum computing chips compatible with conventional fabrication plants.
That’s not as easy as it sounds. So this week’s paper outlining their advances has been eagerly awaited.
The team have reached numerous goals. “We modified an established silicon photonics manufacturing flow to include high-performance single photon detection and photon pair generation,” they say. “To our knowledge, this is the first realization of an integrated photonic technology platform capable of on-chip generation, manipulation, and detection of photonic qubits.”
But there are significant steps ahead. PsiQuantum still needs to develop a variety of “next generation technologies” to make large scale photonic quantum computation feasible. “It will be necessary to further reduce Silicon Nitride materials and component losses, improve filter performance, and increase detector efficiency to push overall photon loss and fidelity,” say the team.
For example, the on-chip photon detectors that are built into waveguides need to be able to count individual photons. The on-chip photon waveguides need to be lower loss. And perhaps the biggest challenge is in developing high speed optoelectronic switches that can rapidly reconfigure optical circuits.
Material Challenge
PsiQuantum is making these switches out of barium titanate (BTO), a material that must be incorporated into the fabrication process. “We have developed a proprietary process for the growth of high-quality BTO films using molecular beam epitaxy, compatible with foundry processes,” they say.
All that looks impressive, but the paper does not include a demonstration of quantum computing itself.
Perhaps it’s too early to expect that. To be fair, basic quantum computation with photons has long been possible with these kinds of systems at a small scale.
“The singular intent of our development is a useful fault-tolerant quantum computer,” they say. PsiQuantum has also said elsewhere that its goal is to achieve this by 2029.
Of course, it faces stiff competition from other manufacturers of quantum computers. It’ll be an exciting race and the (quantum) clock is ticking.
Ref: A manufacturable platform for photonic quantum computing : arxiv.org/abs/2404.17570