6 ms·

LK-99 research continues, paper says superconductivity could be possible

210 points by nsoonhui 3y ago

bluehat974 3y ago
Original paper link?
- Mayzie 3y ago
  It's in the first paragraph of the article: https://doi.org/10.48550/arXiv.2309.07928 https://doi.org/10.48550/arXiv.2309.07928
justinclift 3y ago
This sounds (to my layman self) like one of those situations where the IBM team that assembles nifty constructs atom by atom could be of assistance. https://research.ibm.com/projects/atom-and-molecule-manipulation https://research.ibm.com/projects/atom-and-molecule-manipula...
- chii 3y ago
  It'd take all of forever to construct any real world material atom by atom!
  - justinclift 3y ago
    Even just the bare minimal size for testing this concept conclusively? Ok then, perhaps using ULV lithography (aka microprocessor design) could do it. Pity those machines take billions of $$$ to work with, so it's unlikely someone would try. Hmmm, unless there is a research institute doing something like those lines.
  - godelski 3y ago
    Theoretically or just current technology? Big difference.
    - Retric 3y ago
      Theoretically, if you’re adding one atom at a time making a meaningfully thick object requires trillions of layers. Add 1 atom a second across a full layer ~= several years per centimeter of thickness. Lithography is about as close as we can do, but that takes multiple stages to do each individual layer otherwise you can’t get patterns just undifferentiated layers. If you don’t need atomic precision for arbitrary patterns there are many options. Crystals can grow much faster, but you have minimal control over how they grow etc.
      - yreg 3y ago
        If you have 12 machines and then slap it together, it would take ~= several months per centimeter of thickness.
      - godelski 3y ago
        > Add 1 atom a second across a full layer Who said that's the max theoretical speed? And that it can't be parallel. You're sure putting a lot of constraints on a theoretical argument.
        Retric 3y ago
        That was already assuming an unachievable level of parallel behavior and operations per second. Something like 10^15th print heads and feed lines each slapping down 100’s of thousands of atoms per second, is still orders of magnitude to slow to be practical for human scale objects. Further, that’s the tip of the iceberg in terms of difficulty here. It’s ignoring chemical interactions, thermal issues, sensors, control mechanisms etc etc.
        godelski 3y ago
        I don't see why this is a theoretical limit. Clearly the existence of crystal growth and similar things show that there is of a much faster speed. Especially with repeating patterns those can be built in parallel and combined. Crystals can grow really fast. If it took so long for atoms to combine and make things then we wouldn't have anything actually made. You can definitely call molecular binding a "trick" when building materials from atomic levels. This is a parallelization method. And we can certainly do trillions of operations a second with atomic level combinations. Just take pure sodium and pure chloride and put them together. Or oxygen and pure iron. These reactions happen down at the "atomic level" and we can do these things at quite a macro level. So I'm just not understanding your argument here.
  - Matticus_Rex 3y ago
    With today's tech, yes. With some of the nanotechnology we think is very possible, this is one of the things we could potentially see massive gains in.
A_D_E_P_T 3y ago
Let's assume they're right. Even so, they emphasize that the structure of LK99 is so finicky, so sensitive to a misplaced atom here or there, so prone to confusing performance in the real world due to inhomogeneity, that it's practically impossible to "get it right" with current materials synthesis technologies. It's something that literally has to be grown from the bottom-up, atom by atom --- in bulk, that's essentially a UFO-tier tech. And that's if their models and assumptions are correct. If it works. The way I see it, the best this material can do for us is advance our theoretical understanding of superconductivity. There's no grand theory of how molecular/crystal structure relates to superconductivity. But maybe we've gotten a little bit closer.
- TowerTall 3y ago
  > It's something that literally has to be grown from the bottom-up, atom by atom --- in bulk, that's essentially a UFO-tier tech. The UFO-tears were also at our level at some point. How far out in the future are we talking before humans can do this? Is there anything fundamental that would prevents us from doing it? Also, it's the replicator from Star Trek, isn't it?
  - highduc 3y ago
    > It's something that literally has to be grown from the bottom-up, atom by atom --- in bulk, that's essentially a UFO-tier tech. isn't that how your CPU is made?
    - A_D_E_P_T 3y ago
      No, your CPU is made from the top-down, via lithographic techniques. The silicon wafers are grown from the bottom-up, but silicon is as chemically simple as it gets -- a pure element! LK99 would be tougher to grow (correctly, and assuming that there is a specific superconducting configuration from among several semiconducting configurations,) by at least several orders of magnitude. Growing pristine crystals of such a complex and entropic material is something that has never been attempted.
danbruc 3y ago
Was this not already pointed out in one of the theoretical papers during the initial replication attempt including the judgment that the interesting crystal structure is unlikely to be producable because it is not energetically favored?
brucethemoose2 3y ago
I still think it will be a finicky and difficult to produce "1D" topological superconductor, that conducts in only one direction. So not the bulk superconductor we need, indeed, but that could be interesting for microelectronics.
- amelius 3y ago
  What if you built a 2D or 3D mesh out of that 1D superconductor?
  - willis936 3y ago
    Would you not need 4 or 5 spatial dimensions (something unavailable to us atm)?
    - candiodari 3y ago
      No, because you're misunderstanding what electricity is and how it propagates. Electricity is disturbances in the electromagnetic field. A difference between the electromagnetic field on the surface of the conductor, super or not, and the empty space around it. What conductors do is to propagate that field, perfectly in the case of superconductors, across their surface. It is not necessary for surfaces of pieces of the superconductor to actually touch to propagate that field. There can even be stuff in between, if it doesn't conduct it won't heat up and usually won't disturb the superconductivity. (what it does is it slightly lowers the point, in current, where superconductivity breaks down) So you can just stack them close to each other in our 3 traditional dimensions, because the superconductive region doesn't have to be continuous, just close enough. That will work fine, that's how it works in current superconductors.
      - willis936 3y ago
        What you're describing is not a 1D superconducting plane in a 3D molecular structure.
  - brucethemoose2 3y ago
v3ss0n 3y ago
we are so back?
- lamontcg 3y ago
  we are drunk and desperately texting that we should get back together.
- gus_massa 3y ago
  It's a theoretical paper. We don't have a good theory about high temperature superconductivity, so they are calculating related properties like "flat bands". [It's not my are, so I'm not sure how good is the relation.] My guess is that they started this during peak LK-99, and it got published just now, too late for the news wave. I was very optimistic during the first week, but the evidence so far is not convincing. Unless you work in superconductors, you can probably just ignore this until some experimental evidence appears.
jamseason 3y ago
This article contains various unfounded statements. The manuscript that it refers to does not contain any calculations concerning superconducivity of the material, so it cannot say anything about it aside from vague guesses. There are no manuscripts I know on arXiv that contain calculations that suggest the material "should" be superconducting. This paper and several other discuss flat bands, but as noted in https://arxiv.org/abs/2308.05143 https://arxiv.org/abs/2308.05143 and elsewhere flat bands are no reason to expect superconductivity, especially the kinds that this material supports. So there is no theoretical support for this material to be in any way good for superconductivity. There is also no good experimental support. The crystal structure suggested in the original papers has been reproduced, but is found to be an insulator. Reproduction attempts of superconducting features have universally failed, and alternative explanations for the original measurements (which are fairly sloppy) exist. So there's not much reason to believe there ever was anything interesting there.
ck2 3y ago
I sometimes check the wikipedia page to see what's been done https://en.m.wikipedia.org/wiki/LK-99#Replication_attempts https://en.m.wikipedia.org/wiki/LK-99#Replication_attempts Even if it doesn't pan out, maybe it will still advance superconductor development otherwise via government and private investment.
- Accujack 3y ago
  What everyone following the stuff online seems to forget is that the replication attempts and samples so far are based on a leaked, apparently incomplete version of the LK99 paper. Supposedly, a University in Taiwan is working to validate the full version of the paper and the material, and the authors of the paper say that once confirmation is complete (and is successful) they will release the full version of the paper... sometime in 2024. We'll see what happens, I guess.
yborg 3y ago
Pons and Fleischmann also announced they were still searching for the real killer after their initial claims couldn't be replicated. As far as I know, they still are.
- lamontcg 3y ago
  Nice mash-up.
- badinsie 3y ago
  [dead]
gaze 3y ago
I bet a friend a nice bottle of wine that LK-99 was bullshit. Looks like I’m never getting that bottle but neither is he. It is, obviously, bullshit and always has been.

210 points by nsoonhui 3y ago

bluehat974 3y ago

Original paper link?

Mayzie 3y ago
It's in the first paragraph of the article: https://doi.org/10.48550/arXiv.2309.07928 https://doi.org/10.48550/arXiv.2309.07928

justinclift 3y ago

This sounds (to my layman self) like one of those situations where the IBM team that assembles nifty constructs atom by atom could be of assistance. https://research.ibm.com/projects/atom-and-molecule-manipulation https://research.ibm.com/projects/atom-and-molecule-manipula...

chii 3y ago
It'd take all of forever to construct any real world material atom by atom!
- justinclift 3y ago
  Even just the bare minimal size for testing this concept conclusively? Ok then, perhaps using ULV lithography (aka microprocessor design) could do it. Pity those machines take billions of $$$ to work with, so it's unlikely someone would try. Hmmm, unless there is a research institute doing something like those lines.
- godelski 3y ago
  Theoretically or just current technology? Big difference.
  - Retric 3y ago
    Theoretically, if you’re adding one atom at a time making a meaningfully thick object requires trillions of layers. Add 1 atom a second across a full layer ~= several years per centimeter of thickness. Lithography is about as close as we can do, but that takes multiple stages to do each individual layer otherwise you can’t get patterns just undifferentiated layers. If you don’t need atomic precision for arbitrary patterns there are many options. Crystals can grow much faster, but you have minimal control over how they grow etc.
    - yreg 3y ago
      If you have 12 machines and then slap it together, it would take ~= several months per centimeter of thickness.
    - godelski 3y ago
      > Add 1 atom a second across a full layer Who said that's the max theoretical speed? And that it can't be parallel. You're sure putting a lot of constraints on a theoretical argument.
      - Retric 3y ago
        That was already assuming an unachievable level of parallel behavior and operations per second. Something like 10^15th print heads and feed lines each slapping down 100’s of thousands of atoms per second, is still orders of magnitude to slow to be practical for human scale objects. Further, that’s the tip of the iceberg in terms of difficulty here. It’s ignoring chemical interactions, thermal issues, sensors, control mechanisms etc etc.
        godelski 3y ago
        I don't see why this is a theoretical limit. Clearly the existence of crystal growth and similar things show that there is of a much faster speed. Especially with repeating patterns those can be built in parallel and combined. Crystals can grow really fast. If it took so long for atoms to combine and make things then we wouldn't have anything actually made. You can definitely call molecular binding a "trick" when building materials from atomic levels. This is a parallelization method. And we can certainly do trillions of operations a second with atomic level combinations. Just take pure sodium and pure chloride and put them together. Or oxygen and pure iron. These reactions happen down at the "atomic level" and we can do these things at quite a macro level. So I'm just not understanding your argument here.
- Matticus_Rex 3y ago
  With today's tech, yes. With some of the nanotechnology we think is very possible, this is one of the things we could potentially see massive gains in.

A_D_E_P_T 3y ago

Let's assume they're right. Even so, they emphasize that the structure of LK99 is so finicky, so sensitive to a misplaced atom here or there, so prone to confusing performance in the real world due to inhomogeneity, that it's practically impossible to "get it right" with current materials synthesis technologies. It's something that literally has to be grown from the bottom-up, atom by atom --- in bulk, that's essentially a UFO-tier tech. And that's if their models and assumptions are correct. If it works. The way I see it, the best this material can do for us is advance our theoretical understanding of superconductivity. There's no grand theory of how molecular/crystal structure relates to superconductivity. But maybe we've gotten a little bit closer.

TowerTall 3y ago
> It's something that literally has to be grown from the bottom-up, atom by atom --- in bulk, that's essentially a UFO-tier tech. The UFO-tears were also at our level at some point. How far out in the future are we talking before humans can do this? Is there anything fundamental that would prevents us from doing it? Also, it's the replicator from Star Trek, isn't it?
- highduc 3y ago
  > It's something that literally has to be grown from the bottom-up, atom by atom --- in bulk, that's essentially a UFO-tier tech. isn't that how your CPU is made?
  - A_D_E_P_T 3y ago
    No, your CPU is made from the top-down, via lithographic techniques. The silicon wafers are grown from the bottom-up, but silicon is as chemically simple as it gets -- a pure element! LK99 would be tougher to grow (correctly, and assuming that there is a specific superconducting configuration from among several semiconducting configurations,) by at least several orders of magnitude. Growing pristine crystals of such a complex and entropic material is something that has never been attempted.

danbruc 3y ago

Was this not already pointed out in one of the theoretical papers during the initial replication attempt including the judgment that the interesting crystal structure is unlikely to be producable because it is not energetically favored?

brucethemoose2 3y ago

I still think it will be a finicky and difficult to produce "1D" topological superconductor, that conducts in only one direction. So not the bulk superconductor we need, indeed, but that could be interesting for microelectronics.

amelius 3y ago
What if you built a 2D or 3D mesh out of that 1D superconductor?
- willis936 3y ago
  Would you not need 4 or 5 spatial dimensions (something unavailable to us atm)?
  - candiodari 3y ago
    No, because you're misunderstanding what electricity is and how it propagates. Electricity is disturbances in the electromagnetic field. A difference between the electromagnetic field on the surface of the conductor, super or not, and the empty space around it. What conductors do is to propagate that field, perfectly in the case of superconductors, across their surface. It is not necessary for surfaces of pieces of the superconductor to actually touch to propagate that field. There can even be stuff in between, if it doesn't conduct it won't heat up and usually won't disturb the superconductivity. (what it does is it slightly lowers the point, in current, where superconductivity breaks down) So you can just stack them close to each other in our 3 traditional dimensions, because the superconductive region doesn't have to be continuous, just close enough. That will work fine, that's how it works in current superconductors.
    - willis936 3y ago
      What you're describing is not a 1D superconducting plane in a 3D molecular structure.
- brucethemoose2 3y ago

v3ss0n 3y ago

we are so back?

lamontcg 3y ago
we are drunk and desperately texting that we should get back together.
gus_massa 3y ago
It's a theoretical paper. We don't have a good theory about high temperature superconductivity, so they are calculating related properties like "flat bands". [It's not my are, so I'm not sure how good is the relation.] My guess is that they started this during peak LK-99, and it got published just now, too late for the news wave. I was very optimistic during the first week, but the evidence so far is not convincing. Unless you work in superconductors, you can probably just ignore this until some experimental evidence appears.

jamseason 3y ago

This article contains various unfounded statements. The manuscript that it refers to does not contain any calculations concerning superconducivity of the material, so it cannot say anything about it aside from vague guesses. There are no manuscripts I know on arXiv that contain calculations that suggest the material "should" be superconducting. This paper and several other discuss flat bands, but as noted in https://arxiv.org/abs/2308.05143 https://arxiv.org/abs/2308.05143 and elsewhere flat bands are no reason to expect superconductivity, especially the kinds that this material supports. So there is no theoretical support for this material to be in any way good for superconductivity. There is also no good experimental support. The crystal structure suggested in the original papers has been reproduced, but is found to be an insulator. Reproduction attempts of superconducting features have universally failed, and alternative explanations for the original measurements (which are fairly sloppy) exist. So there's not much reason to believe there ever was anything interesting there.

ck2 3y ago

I sometimes check the wikipedia page to see what's been done https://en.m.wikipedia.org/wiki/LK-99#Replication_attempts https://en.m.wikipedia.org/wiki/LK-99#Replication_attempts Even if it doesn't pan out, maybe it will still advance superconductor development otherwise via government and private investment.

Accujack 3y ago
What everyone following the stuff online seems to forget is that the replication attempts and samples so far are based on a leaked, apparently incomplete version of the LK99 paper. Supposedly, a University in Taiwan is working to validate the full version of the paper and the material, and the authors of the paper say that once confirmation is complete (and is successful) they will release the full version of the paper... sometime in 2024. We'll see what happens, I guess.

yborg 3y ago

Pons and Fleischmann also announced they were still searching for the real killer after their initial claims couldn't be replicated. As far as I know, they still are.

lamontcg 3y ago
Nice mash-up.
badinsie 3y ago
[dead]

gaze 3y ago

I bet a friend a nice bottle of wine that LK-99 was bullshit. Looks like I’m never getting that bottle but neither is he. It is, obviously, bullshit and always has been.