Semiconductors Meet the Quantum Future and Vice Versa

Taking a long range look at what's next, a panel discussed the intersection of semiconductors with the emerging quantum future during a Monday night event at the February 2020 SPIE Advanced Lithography conference. The consensus was that quantum devices will test the manufacturing capabilities of the semiconductor industry, but the form those challenges will take isn't yet completely settled.

In part, that's because quantum technology is still emerging. Consequently, the nature of requirements needed to make devices is also still developing.

For instance, panelist Mark Thompson of PsiQuantum noted that the company's proposed silicon photonic quantum computers would benefit from improved lithography. That would boost the performance of the computer's quantum bits, or qubits, and even incremental enhancements would have an outsize impact, he said.

"Small improvements in [qubit] performance can lead to huge improvements in system performance," Thompson said.

But, he said, the size of the device features, such as wave guides, are 500 nanometers, orders of magnitude larger than the size of state-of-the-art semiconductor chips. In the case of PsiQuantum's technology, what would help is fine feature control, with dimensional variations of a few nanometers.

In contrast, Google's quantum computer uses superconducting devices, with dimensions measured in microns. Hence, said panelist John Martinis of Google, better lithography will have virtually no impact. Material properties, on the other hand, are much more important, a consequence of what's needed to make the new computing technology work.

"We have to control the quantum system very precisely," Martinis said.

He noted that Google's approach has demonstrated its power, solving a problem that stumped a standard computer. Some problems that confound today's computers are easily handled with a quantum computer, a reason why there's a race to build a general purpose, error-correcting implementation of the technology.

Like Google, IBM has also built rudimentary versions of a quantum computer, some having more than 50 qubits. The company's Markus Brink, another panelist, pointed out that semiconductor and quantum technology already intersect. The support, control, interface and other circuitry in current quantum systems consist of today's electronics, which is semiconductor based.

"We rely on CMOS technology," Brinks said.

He added, though, that there is room for improvement. For instance, if quantum technology is superconducting, as is the case for IBM and Google, then there would be a benefit to having CMOS technology running at cryogenic temperatures. Operating at superconducting temperatures would decrease latency by reducing some shuttling of information between room temperature and cryogenic circuitry.

Others on the panel also saw a need for improvement, particularly in materials. These have to be almost perfect, said Robert Visser of Applied Materials. The company doesn't make quantum technology but does plan to support customers who do. To do that, Visser said, "We have to step up our game."

Today, the semiconductor industry routinely makes films that are only a few atoms thick and of extremely high quality. So, Visser said the material manufacturing capability will be there when it is needed for quantum applications.

Materials will indeed be a big challenge, agreed Rick Silver of NIST. But repeatable, near perfect lithography is also needed. For instance, he pointed out that some implementations of quantum technology use quantum dots, a semiconductor nanocrystal. As is the case with other quantum implementations, the variation acceptable in the macro world is excessive in the quantum realm. So, the quantum dots must be almost perfectly uniform, which implies the lithography that defines those dots needs to be both repeatable and precise.

Asked about the future of quantum technology, the panelists said this was an exciting time. Several also noted that at the moment quantum computing is aimed at solving problems that cannot be done cost effectively with current technology. That focus will limit quantum technology to small volumes. At the same time, it will mean that quantum computers could command a premium price, predicted Sasikanth Manipatruni of Kepler Computing.

"If a quantum computer can do something nothing else can, then it might be worth paying a lot for it," he said.

Finally, today there are a number of different ways to implement quantum technology. Currently, it's unclear which of the various approaches will ultimately be the one that prevails, much like no one in the 1950s could have predicted with certainty which semiconductor technology would dominate.

As an illustration of this fluid situation, in the Q&A that followed the panel discussion, a question came up about why qubits are used. It is possible, after all, to have systems with three or more states instead of just two. Irfan Siddiqi of University of California, Berkeley said his research group has worked with qutrits, which have three states instead of the two in qubits. It may be that an architecture with more than two states ends up winning out.

So, the future of quantum technology may not follow the binary model favored by semiconductors. "We should keep our minds open," Siddiqi said.

Hank Hogan is a science writer based in Reno, Nevada.