Constriction junction, do you function?

This research was conducted as part of the Co-design Center for Quantum Advantage (C²QA), a DOE National Quantum Information Science Research Center led by Brookhaven Lab, and it builds upon years of scientific collaboration focused on improving qubit performance for scalable quantum computers. Recently, scientists have been working to increase the amount of time qubits retain quantum information, a property known as coherence that is closely linked to the quality of a qubit's junction.

They have been particularly focused on superconducting qubits whose architecture includes two superconducting layers separated by an insulator. This part of the qubit is known as an SIS junction, for superconductor-insulator-superconductor. But reliable manufacturing of such sandwich-like junctions is not easy, especially at the precision needed for the large-scale production of quantum computers.

"Making SIS junctions is truly an art," said Charles Black, co-author of the paper that recently published in the Physical Review A and director of the Center for Functional Nanomaterials (CFN), a DOE Office of Science user facility at Brookhaven Lab.

Black and Mingzhao Liu, senior scientist at CFN and lead author on the paper, have been part of C²QA since its inception in 2020. And while they've been helping quantum scientists understand the materials science of qubits to improve their coherence, they've also grown curious about the scalability of this qubit-building art and its compatibility with the inevitable need for manufacturing large-scale quantum computers.

So, the scientists turned their attention to qubit architectures with superconducting junctions comprised of two layers connected by a thin superconducting wire, instead of a middle insulating layer. Known as a constriction junction, this architecture lays flat rather than stacking like a sandwich. And importantly, the process for fabricating constriction junctions is compatible with standard methods in semiconductor manufacturing facilities.

"In our work, we investigated the impact of this architectural change," said Black. "Our goal was to understand the performance tradeoffs of making the switch to constriction junctions."

Overcoming the increased current flow and linearity

The most prevalent superconducting qubit architecture works best when the junction connecting the two superconductors transmits only a little bit of current. Though the insulator in the SIS sandwich prevents nearly all current transmission, it is thin enough to allow a small amount via a mechanism known as quantum tunneling.

"The SIS architecture is ideal for today's superconducting qubits, even though it's tricky to manufacture," said Black. "But it's a little counterintuitive to replace the SIS with a constriction, which intrinsically conducts a lot of current."

Through their analysis, the researchers showed that it is possible to reduce the current traveling across a constriction junction to an appropriate level for a superconducting qubit. However, the method requires less traditional superconducting metals.

"The constriction wire would have to be impractically thin if we used aluminum, tantalum, or niobium," explained Liu. "Other superconductors that do not conduct as well would let us fabricate the constriction junction at practical dimensions."

However, constriction junctions behave differently from their SIS counterparts. So, the scientists also investigated the consequences of making this architectural change.

To work, superconducting qubits require some nonlinearity, which limits the qubit to operate between only two energy levels. Superconductors don't naturally exhibit nonlinear behavior -- it's the qubit junction that introduces this key property.

Superconducting constriction junctions are inherently more linear than tried-and-true SIS junctions, meaning they are less ideal for qubit architectures. However, the scientists found that the constriction junction nonlinearity can be tuned through the selection of a superconducting material and the appropriate design of the junction's size and shape.

"We're excited about this work because it points materials scientists towards specific targets based on the device requirements," explained Liu. For example, the scientists identified that for qubits operating between 5 and 10 gigahertz, which is typical for today's electronics, there need to be specific tradeoffs between the material's ability to carry electricity, determined by its resistance, and the junction's nonlinearity.

"Certain combinations of material properties just aren't workable for qubits operating at 5 gigahertz," said Black. But with materials that meet the criteria outlined by the Brookhaven scientists, qubits with constriction junctions can operate similarly to qubits with SIS junctions.

Liu and Black are currently working with their C²QA colleagues to explore materials that can meet the specifications outlined in their new paper. Superconducting transition metal silicides, in particular, have captured their attention because these materials are already used in semiconductor manufacturing.

"In this work, we showed that it is possible to mitigate the concerning characteristics of constriction junctions," said Liu. "So, now we can begin exploiting the benefit of the simpler qubit fabrication process."

This work embodies C²QA's foundational co-design principle, as Liu and Black explored a qubit architecture that could satisfy the demands of quantum computing and align with current electronics manufacturing capabilities.

"These types of interdisciplinary collaborations will continue bringing us closer to realizing scalable quantum computers," said Black. "It's almost hard to believe that humans have attained the quantum computers we have today. We're so excited to play a role in helping C²QA achieve its goals."