Paul van Gerven
4 May

TU Delft researchers have demonstrated the superconducting equivalent of a diode, which they think could take computers into the terahertz domain.

A research team at Delft University of Technology has demonstrated one-way superconductivity without magnetic fields. This experimental result, which was long thought to be impossible, could clear a path to speeding up conventional computing by a factor of 300-400 while saving energy. “If the 20th century was the century of semiconductors, the 21st can become the century of the superconductor,” commented Mazhar Ali, associate professor at TU Delft and lead author of the study published in Nature.

It’s impossible to build a computer without having control over the direction in which currents flow. That’s why modern electronics are based on semiconductors; these allow for one-way electrical conduction. The textbook example of this is the pn junction, in which a semiconductor with an excess of holes is connected to one with extra electrons. Due to the interaction of the charges, a potential builds up, making it harder for an electron to travel in one direction than the other. Pn junctions are the building blocks of several electronic devices, including diodes, transistors and LEDs. The diode, which conducts primarily in one direction, consists of a single pn junction.

“Superconductors never had an analog of this one-directional idea without a magnetic field, since they’re more related to metals, which always conduct in both directions and don’t have any built-in potential. Similarly, Josephson Junctions, which are sandwiches of two superconductors with non-superconducting, classical barrier materials in between the superconductors, also haven’t had any particular symmetry-breaking mechanism that resulted in a difference between ‘forward’ and ‘backward,’” Ali explains.

TU Delft Josephson Junction
Credit: Ella Marushenko/TU Delft

Real

Ali’s team managed to sidestep this limitation of Josephson Junctions (JJs) by replacing the classical barrier with a so-called quantum material, which is capable of modulating the coupling between the two superconductors. To build the Josephson diode, Nb3Br8 was introduced, a graphene-like material predicted to host a net electric dipole to create a similar effect as in pn junctions.

 advertorial 

Hands-on GaN Doherty amplifier design

During the Benelux RF Conference at Van der Valk Nijmegen on 1 June, Martijn Brethouwer (Bruco IC) will take you through the paces of a Doherty amplifier development process with all its pitfalls and hurdles. Take a look at the complete program. We only have a limited number of tickets left, so sign up in time.

“We were able to peel off just a couple atomic layers of this Nb3Br8 and make a very, very thin sandwich – just a few atomic layers thick – which was needed for making a Josephson diode for the first time. This wasn’t possible with normal 3D materials,” Ali says.

To make sure the discovery was real, the Delft researchers made many batches of the device and had them measured all over the world by colleagues. In all instances, the results came back the same: current flows without resistance in one direction and with normal resistance in the other. This effect is present in the absence of a magnetic field.

Scaling

That last characteristic is a boon for technological applications since magnetic fields at the nanometer scale are very difficult to control and limit. Potential uses include any technology device that currently relies on semiconductors, including computing. Superconductor computers could take computers into the terahertz domain, Ali claims. “For server farms or supercomputers, it would be smart to implement this,” he contends. Given the need for cryogenic cooling, superconductor consumer PCs probably aren’t practical.

Next, Ali and his team want to decrease the amount of cooling by raising the operating temperature. “We used a very simple superconductor that limited the operating temperature. Now we want to work with the known so-called high-temperature superconductors and see whether we can operate Josephson diodes at temperatures above 77 K since this will allow for liquid nitrogen cooling.” Another priority is the scaling of production. “While it’s great that we proved this works in nanodevices, we only made a handful. The next step will be to investigate how to scale production to millions of Josephson diodes on a chip.”