News Superconductor Breakthrough Replicated, Twice

[Admin]

It's been a rough few days in the condensed matter physics realm following claims of the world's first room-temperature superconductor being achieved. However, work to verify and replicate the results is catapulting forward, and two disparate teams have already shown promising results.

Superconductor Breakthrough Replicated, Twice : Read more

 

[InvalidError]

As with many scientific discoveries, discovering something seemingly revolutionary often ends up in disappointment when the discovery cannot be turned into anything actually usable.

LK-99 might be real, though you still need to grow the crystals to usable sizes and shapes. A superconductor will have limited uses if you cannot cost-effectively make wires, discs, cylinders and other basic yet very handy shapes from it.

 

[A Stoner]

Next up, room temperature fusion!

 

[thisisaname]

As with many scientific discoveries, discovering something seemingly revolutionary often ends up in disappointment when the discovery cannot be turned into anything actually usable.

LK-99 might be real, though you still need to grow the crystals to usable sizes and shapes. A superconductor will have limited uses if you cannot cost-effectively make wires, discs, cylinders and other basic yet very handy shapes from it.

Yes going from how rough the disc looked the material may be quite difficult to shape. Going to be interesting see how this progresses.

 

[InvalidError]

Yes going from how rough the disc looked the material may be quite difficult to shape. Going to be interesting see how this progresses.

They said themselves that the reason the disc wasn't levitating well was likely due to excess impurities. The disc looked so crumbly likely because it is just a clump of little crystals held together by leftovers, not a single somewhat cohesive chunk.

 

[bit_user]

 

[garylcamp]

There still seems some skepticism as an article in Science 2 reports a failure to duplicate. But things are moving fast so I expect more clarity soon. Note that, like Fusion or voice recognition, it may not be as easy as it appears at first. Need more info on making useful material to decide.

 

[bit_user]

Imagine if your 16-core mainstream CPU (which likely requires a competent watercooling solution to avoid incinerating itself) operated without power losses — no current leakage, no electricity waste in the form of heat. Superconductors mean almost perfectly efficient computing.

Even if there are room-temperature superconductors that we could use in CPUs, is this statement even credible? Doesn't a lot of the power dissipation of modern CPUs come from leakage and transistor-switching?

Switching Losses: Effects on Semiconductors - Technical Articles

We review switching losses involving diode recovery charge, transistor switching with clamped inductive load, device capacitance and leakage, package, and stray inductances, and the efficiency versus switching frequency curve.

www.allaboutcircuits.com

 

[bit_user]

As with many scientific discoveries, discovering something seemingly revolutionary often ends up in disappointment when the discovery cannot be turned into anything actually usable.

LK-99 might be real, though you still need to grow the crystals to usable sizes and shapes. A superconductor will have limited uses if you cannot cost-effectively make wires, discs, cylinders and other basic yet very handy shapes from it.

A bit like why Graphene has taken so long to find commercially viable applications?

But, unlike Graphene, there might be other materials in this same category that are easier to manufacture at scale.

 

[toffty]

LK99 is not a super conductor

View: https://m.youtube.com/watch?v=RjzL9cS3VW8&t=34

 

[InvalidError]

Even if there are room-temperature superconductors that we could use in CPUs, is this statement even credible? Doesn't a lot of the power dissipation of modern CPUs come from leakage and transistor-switching?

Leakage comes from the extreme proximity between traces, I don't think superconductors will do anything about that and leakage is becoming an increasingly significant loss factor as things get packed tighter together. CMOS logic works by charging and discharging gates, the amount of energy spent charging or discharging gate capacitors is Q=Cg*V^2/2, superconductors aren't going to change that either, nor the energy associated with charging and discharging the remainder of parasitic trace capacitances. The only thing superconductors might change is trace conduction losses assuming you can shape crystals in a way that lets you put them on wafers.

Carbon nanotubes were hyped as the solution to semiconductor copper losses 20 years ago. I don't remember the last time I read about any sign of progress on that. Getting tubes to grow to the necessary lengths and self-assemble on a wafer sounds like extremely tricky business.

A bit like why Graphene has taken so long to find commercially viable applications?

But, unlike Graphene, there might be other materials in this same category that are easier to manufacture at scale.

Graphene has plenty of applications when you only need micron-sized flakes... such as thermal pastes, high-performance greases and pencil leads

Likewise, if LK-99 is real but can only be made in micron-sized flakes, I it may still be useful in less glamorous applications such as magnetic filler in ferrite-like transformer and induction motor cores to eliminate most of the remaining eddy current losses there assuming it can also bear the associated current and magnetic flux densities before quenching.

No shortage of things that can go wrong besides the LK-99 experiment turning out to be a hoax.

 

[bwana]

America: we have a new name for twitter
Korea: we have a room temperature superconductor

 

[InvalidError]

America: we have a new name for twitter
Korea: we have a room temperature superconductor

Out of the handful of labs working on it, none have managed to produce a successful sample yet. One lab has at least managed to duplicate the atomic structure, another has run simulations that say it might work.
LK-99 isn't out of the competition with 𝕏 for being a waste of hot air yet.

 

[palladin9479]

Issue with this article, it was not replicated twice, both papers are the exact same thing but by two different sets of Korean authors. Originally all six worked on the project and for some reason split into two groups that each submitted their own papers. Also neither paper actually measured that electrical resistance was zero, only that there was a drop.

German quantum physicist Sabine Hossenfelder did a break down on it in her typical dry humor way.

Haha looks like toffty beat me to it.

 

[bit_user]

Graphene has plenty of applications when you only need micron-sized flakes... such as thermal pastes, high-performance greases and pencil leads

der8auer claimed such applications don't perform well, because the thermal-conductivity of graphene is orientation-dependent. I heard that in his sales pitch for KryoSheet, which supposedly stacks the graphene layers in such a way that maximizes thermal conductivity in the direction normal to the plane.

Thermal Grizzly High Performance Cooling Solutions - KryoSheet

Hochwertige Wärmeleitlösungen für Computerchips

www.thermal-grizzly.com

 

[InvalidError]

der8auer claimed such applications don't perform well, because the thermal-conductivity of graphene is orientation-dependent.

Yes, heat propagation is fastest across the sheet (that's the one doing the headline 5000W/mK) than through the stack. Randomly ordered graphene flakes should give you at least 11W/mK, which is better than the 7-9W/mK for aluminum/zinc oxide pastes and just as chemically stable, which is what you would want if you prefer thermal paste to be a "set and forget" kind of thing like me.

I'm still using silicone-ZnO paste from a 20+ years old tube in my PCs, don't need fancy paste to pass 125W max from my CPU's IHS to its HSF. That stuff just doesn't die.

 

[Ryan F. Mercer]

No, I will not expand the Tweet above to see the video. Let me know when it's on YouTube. Why journalists continue to pathetically debase themselves by remaining on the Twitter platform is beyond me.

 

[Rexer]

Imagine a super conductor to power all our AI replacements. They'll round us up to be crash test dummies in rat-rods. 'bot war games'. Heh, heh, heh.

 

[rluker5]

A bit like why Graphene has taken so long to find commercially viable applications?

But, unlike Graphene, there might be other materials in this same category that are easier to manufacture at scale.

Graphene does have some recreational applications. For example you could buy some pyrolytic graphite from Amazon or Ebay and it would float like this demonstration on a magnet. It's even a similar color and is a bit bubbly in texture on the flat, unchipped surface. Not that I'm saying that this stuff is that.

I even tried to use the isotropic heat transfer of some of that graphene to better equalize temps across the heatpipes in the NH-D15 cooler I'm using. But I didn't see any improvement in temps 🙁. At least it was more entertaining than lottery tickets and I have leftover bits to float on magnets.

 

[bit_user]

I'm still using silicone-ZnO paste from a 20+ years old tube in my PCs, don't need fancy paste to pass 125W max from my CPU's IHS to its HSF. That stuff just doesn't die.

I have a machine with a 130 W CPU that I built 10 years ago and the Arctic Cooling MX-4 that I used on it seems to work about as well as the day I built the thing. MX-4 has a reputation for good longevity.

I just bought a tube of MX-6, for my future builds. I hope it holds up as well.

 

[Hooda Thunkett]

As with many scientific discoveries, discovering something seemingly revolutionary often ends up in disappointment when the discovery cannot be turned into anything actually usable.

LK-99 might be real, though you still need to grow the crystals to usable sizes and shapes. A superconductor will have limited uses if you cannot cost-effectively make wires, discs, cylinders and other basic yet very handy shapes from it.

I'm pretty sure we could find uses for a material like this even if we can't form wires from it. I'm thinking we could conceivably use it for a maglev transport with only minimal refinement of the manufacturing process.
As to using it in CPUs/GPUs, I see that as being more far fetched. If we can use it or a related compound to make some kind of wires or bars, then it would probably be more useful in power transmission, motors, and other equipment than in chips. You would probably save much more energy that way.
Assuming you can make wires that is...

 

[tanon]

Leakage comes from the extreme proximity between traces, I don't think superconductors will do anything about that and leakage is becoming an increasingly significant loss factor as things get packed tighter together. CMOS logic works by charging and discharging gates, the amount of energy spent charging or discharging gate capacitors is Q=Cg*V^2/2, superconductors aren't going to change that either, nor the energy associated with charging and discharging the remainder of parasitic trace capacitances. The only thing superconductors might change is trace conduction losses assuming you can shape crystals in a way that lets you put them on wafers.

The main promising application of superconductors in microelectronics is replacement of FETs with Josephson Junctions, where they would indeed be revolutionary, allowing switching frequencies of 100s of GHz, with power dissipation in the milliWatts (or less).

It would probably be the largest leap in computing technology since the shift from vacuum tubes to transistors.

 

[bit_user]

The main promising application of superconductors in microelectronics is replacement of FETs with Josephson Junctions, where they would indeed be revolutionary, allowing switching frequencies of 100s of GHz, with power dissipation in the milliWatts (or less).

I immediately thought of Josephson Junctions, but then looked at the related Wikipedia page and didn't get enough out of it to say anything here.

I guess what I'm wondering is whether those advantages still apply. What's the switching time and loss of a transistor, on the latest process node? How well do Josephson Junctions scale down? Can they be made a comparable size? And what does that do to their performance and efficiency?

It would probably be the largest leap in computing technology since the shift from vacuum tubes to transistors.

It's exciting, but I see too many unknowns for me to get worked up over it. Maybe that's just my ignorance, but I wouldn't be surprised if nobody knew for sure how competitive they'd be at scale.

 

[InvalidError]

The main promising application of superconductors in microelectronics is replacement of FETs with Josephson Junctions, where they would indeed be revolutionary, allowing switching frequencies of 100s of GHz, with power dissipation in the milliWatts (or less).

Even if you make a practically loss-less transistor, you will still have a few watts lost to parasitic capacitance between traces and unless something much better than silicon regarding ohmic + tunnelling leakage is discovered, you will still be losing several watts to electrons passing through insulation. About half of modern chips' power consumption is leakage current, which is why modern CPUs' power management schemes have to resort to completely powering down as much unused circuitry as possible.

Also, if you are going to depend on quantum effects for transistors, you have to keep in mind that quantum computing requires cryogenic temperatures not only for the ability to use superconductors but also to reduce thermal noise for improved qubit stability and improved signal-noise ratio.

And how are you supposed to use single-electron transistors when thousands of electrons are jumping between traces every second due to leakage between any two traces running parallel to each other? I'd imagine thermal noise would be a problem here too.

These things may be great for quantum computing and metrology applications as solitary junctions doing a specific job but I doubt you could pack them close enough to make a chip that can compete with conventional CPUs at room temperature.

 

[tanon]

Also, if you are going to depend on quantum effects for transistors, you have to keep in mind that quantum computing requires cryogenic temperatures not only for the ability to use superconductors but also to reduce thermal noise for improved qubit stability and improved signal-noise ratio.

And how are you supposed to use single-electron transistors when thousands of electrons are jumping between traces every second due to leakage between any two traces running parallel to each other? I'd imagine thermal noise would be a problem here too.

These things may be great for quantum computing and metrology applications as solitary junctions doing a specific job but I doubt you could pack them close enough to make a chip that can compete with conventional CPUs at room temperature.

Don't get me wrong, I'm sure there's plenty of very significant (possibly even insurmountable) problems that will have to be worked out, before/if we ever see real, useful devices using JJs.

My point was simply that the potential advantages of a real STP superconductor aren't just limited to the obvious one, namely, the removal of losses due to resistance.

Interestingly, NIST demonstrated JJs with a feature size of <100nm around 3 months ago.