Physicists discover "impossible" one-way superconductor that could lead to dramatically...

Daniel Sims

Posts: 1,372   +43
Staff
Why it matters: Modern computers are based on semiconductors, partially because they can direct a current in one direction. Doing that in a superconductor has been thought impossible for over a century, but researchers in the Netherlands have found a way. The discovery could make computers hundreds of times faster than they are currently.

This week, researchers at the Delft University of Technology published a paper on how they achieved one-directional superconductivity. This could let computers replace semiconductors with superconductors, which can carry a current indefinitely with no energy loss, potentially increasing computer speeds by orders of magnitude.

According to Associate Professor Mazhar Ali, superconductor-based computers could reach speeds up to a terahertz. Superconducting might not be usable for consumer computers in the near term, but Ali thinks server farms and supercomputers could implement it.


Ali (middle), along with fellow researchers Dr. Yaojia Wang (left) and Dr. Heng Wu (right)

Normally, currents run through superconductors without any resistance, which makes stopping or directing their flow impossible. Ali says his group was able to do it by sandwiching a quantum material between two semiconductors.

Currently however, the research team has only tested the method at extremely low temperatures. Thus far, any superconductor-based system using this process would be extremely sensitive to heat. Ali's team plans to see if the method can work at temperatures above 77 Kelvin (about -321 Fahrenheit), at which point computers might be able to use these superconductors with the help of liquid-nitrogen cooling. The next step will be to figure out how to produce enough superconductors for a chip.

The (im)possibility of applying superconducting

In the 20th century and beyond, no one could tackle the barrier of making superconducting electrons go in just one-direction, which is a fundamental property needed for computing and other modern electronics (consider for example diodes that go one way as well). In normal conduction the electrons fly around as separate particles; in superconductors they move in pairs of twos, without any loss of electrical energy. In the 70s, scientists at IBM tried out the idea of superconducting computing but had to stop their efforts: in their papers on the subject, IBM mentions that without non-reciprocal superconductivity, a computer running on superconductors is impossible.

Permalink to story.

 
Clickbait bullshit.

"Thus far, any superconductor-based system using this process would be extremely sensitive to heat. Ali's team plans to see if the method can work at temperatures above 77 Kelvin (about -321 Fahrenheit), at which point computers might be able to use these superconductors with the help of liquid-nitrogen cooling."
 
As usual with articles about alleged breakthroughs, there's enough catches that it's not going to make as meaningful of a difference as the headline implies.

  • Not useable in the near term for consumers, maybe useable eventually in server farms and supercomputers.
  • Only tested at extremely low temperatures.
  • If they get it to work at above -321 F, they could keep it cooled... with liquid nitrogen.
  • They haven't figured out how to make enough superconductors for a chip.

The research paper may still be worth reporting on, but perhaps with a less sensational headline.
 
With so high frequencies in the area of terahertz even the tiniest improvement in IPC of a CPU it will be multiplied by a lot.

For example if two CPUs which both run at 1 GHz and they have 1% difference in performance (lets say one run a game at 100fps and the other at 101fps) then that difference in 1 THz it would be x1000 so they will have difference in performance of 1000 fps instead of 1 fps and both they will have already improved by a total factor of x1000 (the 100fps will be 100k fps) in comparison with the 1GHz.

Until now we are waiting 5 years for just a 20% improvement. So even if they can make a CPU with superconductors but "only" at 100 nm if it can run at 1 Thz the difference it will be huge with one ordinary with the same transistors but at a lot lower nm.
 
Last edited:
Will never find it's way into desktop computer in the next 20 years minimum. High temperature superconductivity was discovered in like 1987 and all the hype about superconducting wires to revolutionise the world coming never eventuated. The fact this is a regular superconductor meaning ultra-low temperatures renders it useless for most applications.
 
IBM may have composed a "proof" that it was impossible to use superconductors in computers, but shortly after they gave up on their Josephson Junction project, some engineers from Russia found a way to make superconductors useful by using pulsed currents instead of steady ones.
This was called RSFQ - Rapid Single Flux Quantum computing.
 
Wouldn't get too excited till this has been vetted by some skilled industry RnD and alternative materials are considered. Its sort of a nothing burger in a way, this isn't a new discovery they came upon apropos of nothing.
 
Wouldn't get too excited till this has been vetted by some skilled industry RnD and alternative materials are considered. Its sort of a nothing burger in a way,
Well that, and the fact that uber cold temperatures have been know to cause materials to super-conduct for decades
 
You can do a lot of things with ultra cold temps, duh! Talk to me when you can super conduct at room temperatures.
 
Back