If quantum computing is going to become an every-day reality, we need better superconducting thin films, the hardware that enables storage and processing of quantum information. Too often, these thin films have impurities or other defects that make them useless for real quantum computer chips. Now, Yuki Sato and colleagues at the RIKEN Center for Emergent Matter Science (CEMS) in Japan have discovered a way to make a superconducting thin film from iron telluride, which is surprising because it is not normally superconducting. The fabrication process reduces distortion in the crystal structure, making it superconducting at very low temperatures, and thus suitable for use in quantum chips. This study was published in the scientific journal Nature Communications.

In quantum computing, information is often encoded in qubits that exist in special thin films. Because qubit states are determined by supercurrents of paired electrons, the thin films must be superconducting. Irregularities, impurities, and distortions in thin films can all destabilize the qubits, which affects the accuracy of quantum operations; we don’t want a quantum chip that sometimes tells us that two plus two equals four, and sometimes that it equals five. Iron telluride has few impurities and would be a good substance for quantum thin films, but it isn’t normally superconducting.

What is lattice matching?

Thin films are grown on top of another substance in a process called epitaxy. Each of these substances, called substrates, has a repeating atomic pattern that forms a grid-like lattice. When the thin film grows, it’s atomic structure aligns itself to this grid as best it can. So, thin films made from the same elements can have totally different crystal structures depending on the atomic grid they are grown on. Usually, when developing new thin films, the goal is to use a substrate that gives the best lattice match—atom-to-atom alignment —with the thin film.

The twist of this study is that Sato and the team used a substrate that did not align well and ended up creating a superior thin film.

The researchers used molecular beam epitaxy to spray two tiny beams of iron and telluride atoms onto a sheet of cadmium telluride. The iron and telluride then self-assembled into crystal layers as expected, but alignment to underlying atomic grid wasn’t very good. In fact, it was off by about 20%. Such a large misalignment usually means that the thin film will be defective. However, the opposite turned out to be true in this case.

What is different about the new thin film?

The new iron telluride thin film is superconducting, making it suitable for studying quantum phenomena and possibly for use in quantum computer chips. Analysis with a scanning transmission electron microscope revealed higher-order alignment to the atomic grid in cadmium telluride, which stabilized the crystal structure. This alignment was not atom-to-atom as is usually desired, but could be seen if groups of atoms were examined. Analysis with synchrotron x-ray diffraction at KEK in Tsukuba, Japan showed that this structural change reduced lattice distortion that is normally present in bulk iron telluride. It is this distortion that prevents superconductivity at the low temperatures needed for quantum computing. Without the distortion, the new iron telluride film is superconducting below 10° K (-263° C).

The researchers used the same process to grow iron telluride thin films on strontium titanate, which has a very high classical lattice match, only off by 1.8%. This thin film was not superconducting at all, reinforcing the idea that higher-order epitaxy was the key.

“Our findings indicate that intentionally creating higher-order epitaxial matching could be the future of thin-film research,” says Sato. “Although we used a substrate that should not allow good lattice matching, the film quality somehow improved. In research, seemingly contradictory results like this sometimes appear. Rather than dismissing such contradictions as trivial, we carefully searched for and identified the underlying mechanism. This approach to research thankfully led to this discovery.”