Title is a bit misleading - it's not pure germanium that superconducts here, it's germanium doped w/ Gallium atoms.
Superconducting germanium alloys have been known for decades, I used a Molybdenum/Germanium superconducting alloy in my PhD research 20 years ago, with much higher Tc.
The interesting aspect of this current experiment is the precise alignment of the Ga atoms into specific points of the Ge lattice, so preserving the crystalline structure order which leads to some interesting effects.
Leaves me wondering if this will allow for superconducting cryogenic transistors? If my hobby level understanding of how silicon doping works, this new superconducting germanium would be a p-type? I could imagine something like ion implantation could be able to establish n-type regions within the germanium while allowing bulk regions of the lattice to maintain superconducting properties.
Though admittedly, I'm not actually aware what parts of a semiconductor circuit are the biggest power dissipation sources, so I guess its entirely possible that most of the power is dissipated across the p-n junctions themselves.
Yes, this would be P type. Boron is usually the P type dopant of choice. I’m not sure what role they have in mind for this, but probably to replace polysilicon and metals as conductors. What you have to watch out for is that this will make diodes wherever it bumps up against n-type material. This is a problem for metals as well, because you can get accidental schottky junctions, and we usually solve it with degenerate doping under the contract. I’m not sure what a junction with this material would do though.
I want to note that in what has become the largest (by mass) application of semiconductors, silicon PV cells, boron has been replaced by gallium as the P type dopant of choice. Boron suffers from an annoying form of light-induced efficiency degradation that gallium avoids.
whole article is suspect in that it mentions cryogenic consumer products
or maybe this is a slip and next gen refrigerator advertising will be run by a self hosting AI
I don't know if they updated the article, but I don't see any reference to cryogenic consumer products.
They mention cryogenic electronics, which are used for high-sensitivity electronics in research labs and in medical tests (eg SQUIDs for magnetoencephalography).
> For decades, researchers have tried to create semiconductor materials that can also act as superconductors -- materials capable of carrying electric current without resistance. Semiconductors, which form the foundation of modern computer chips and solar cells, could operate far faster and more efficiently if they also possessed superconducting abilities.
Really? First I've heard of it. And it also doesn't make any sense, since defintionally a material can't be superconducting and semiconducting at the same time, any more than it could be conducting and insulating at the same time. Are they imagining some new kind of thermal-switching circuitry?
This reads to me like the researchers came up with an irrelevant novelty (which is, to be fair, a valid and important part of scientific progress; it still expands our understanding of the universe) and Science Daily asked an LLM to rationalize it as useful.
> And it also doesn't make any sense, since defintionally a material can't be superconducting and semiconducting at the same time
I'd say it gets interesting if one can get at least part of a die made out of superconductors. Getting power in into the die is a huge damn challenge, we're talking about hundreds of amps for modern CPUs and GPUs - if even a part of that could be shrunk that would be a huge gain.
The image on the article talks about making Josephson junctions with it, and the abstract talks about epitaxial superconductor-semiconductor devices.
It feels like the researchers were mainly interested in applicability to Josephson junctions, and the article mixed them up with semiconductor junctions.
It's not practical for your desktop computer, but a tank of nitrogen and some refrigeration hardware which fits in a single rack and you can run at 3.5K in a data center.
An MRI machine is a giant magnet with a closed loop helium cooler to keep the superconducting coils cold. A chiller is used to reject the heat outside.
Researchers have for the first time turned germanium—a widely used semiconductor—into a superconducting material by embedding gallium atoms in its crystal structure. This breakthrough could usher in a new era of quantum devices and ultra-efficient electronics.
It seems the breakthrough is that you could use familiar semiconductor manufacturing processes. However the temperature is still going to be a major issue. I don't want a computer that requires liquid helium cooling.
> I don't want a computer that requires liquid helium cooling.
True, but I /can/ see someone, such as Sandia National Labs, very much willing to install a liquid helium cooled computer if it provides a significant performance increase above their existing supercomputer installations.
Probably because they don’t behave well for normal lithography techniques? The high temp superconductors I know of are weird meta materials, and good luck getting them to exist in chip form at all.
Title is a bit misleading - it's not pure germanium that superconducts here, it's germanium doped w/ Gallium atoms.
Superconducting germanium alloys have been known for decades, I used a Molybdenum/Germanium superconducting alloy in my PhD research 20 years ago, with much higher Tc.
The interesting aspect of this current experiment is the precise alignment of the Ga atoms into specific points of the Ge lattice, so preserving the crystalline structure order which leads to some interesting effects.
Leaves me wondering if this will allow for superconducting cryogenic transistors? If my hobby level understanding of how silicon doping works, this new superconducting germanium would be a p-type? I could imagine something like ion implantation could be able to establish n-type regions within the germanium while allowing bulk regions of the lattice to maintain superconducting properties.
Though admittedly, I'm not actually aware what parts of a semiconductor circuit are the biggest power dissipation sources, so I guess its entirely possible that most of the power is dissipated across the p-n junctions themselves.
Yes, this would be P type. Boron is usually the P type dopant of choice. I’m not sure what role they have in mind for this, but probably to replace polysilicon and metals as conductors. What you have to watch out for is that this will make diodes wherever it bumps up against n-type material. This is a problem for metals as well, because you can get accidental schottky junctions, and we usually solve it with degenerate doping under the contract. I’m not sure what a junction with this material would do though.
> Boron is usually the P type dopant of choice.
I want to note that in what has become the largest (by mass) application of semiconductors, silicon PV cells, boron has been replaced by gallium as the P type dopant of choice. Boron suffers from an annoying form of light-induced efficiency degradation that gallium avoids.
Fair enough, my ion implant experience was DRAM / flash. I never worked on PV.
> which leads to some interesting effects.
Such as?
whole article is suspect in that it mentions cryogenic consumer products or maybe this is a slip and next gen refrigerator advertising will be run by a self hosting AI
I don't know if they updated the article, but I don't see any reference to cryogenic consumer products.
They mention cryogenic electronics, which are used for high-sensitivity electronics in research labs and in medical tests (eg SQUIDs for magnetoencephalography).
> For decades, researchers have tried to create semiconductor materials that can also act as superconductors -- materials capable of carrying electric current without resistance. Semiconductors, which form the foundation of modern computer chips and solar cells, could operate far faster and more efficiently if they also possessed superconducting abilities.
Really? First I've heard of it. And it also doesn't make any sense, since defintionally a material can't be superconducting and semiconducting at the same time, any more than it could be conducting and insulating at the same time. Are they imagining some new kind of thermal-switching circuitry?
This reads to me like the researchers came up with an irrelevant novelty (which is, to be fair, a valid and important part of scientific progress; it still expands our understanding of the universe) and Science Daily asked an LLM to rationalize it as useful.
> And it also doesn't make any sense, since defintionally a material can't be superconducting and semiconducting at the same time
I'd say it gets interesting if one can get at least part of a die made out of superconductors. Getting power in into the die is a huge damn challenge, we're talking about hundreds of amps for modern CPUs and GPUs - if even a part of that could be shrunk that would be a huge gain.
Not if you spend that energy, or more, to cool the device hundreds of degrees below room temperature using liquid helium.
The image on the article talks about making Josephson junctions with it, and the abstract talks about epitaxial superconductor-semiconductor devices.
It feels like the researchers were mainly interested in applicability to Josephson junctions, and the article mixed them up with semiconductor junctions.
.. at a temperature of 3.5K. So perhaps not super practical.
It's not practical for your desktop computer, but a tank of nitrogen and some refrigeration hardware which fits in a single rack and you can run at 3.5K in a data center.
Nitrogen freezes at 63K. That makes it a bad coolant for a continuously-running process at 3.5K.
3.5K is well below the point where nitrogen is liquid. The only option would be helium.
Just to point, but it would require actively cooled helium. You can't just drop it in liquid helium and expect boiling to cool your device.
Can you make a closed loop helium cooler? Also, that level of coldness seems like it would have negative interactions with other components.
> Can you make a closed loop helium cooler?
An MRI machine is a giant magnet with a closed loop helium cooler to keep the superconducting coils cold. A chiller is used to reject the heat outside.
Researchers have for the first time turned germanium—a widely used semiconductor—into a superconducting material by embedding gallium atoms in its crystal structure. This breakthrough could usher in a new era of quantum devices and ultra-efficient electronics.
> ...allows it to carry current with zero resistance at 3.5 Kelvin (about -453 degrees Fahrenheit)
Seems to me this is a problem.
It's an interesting result, but yeah, not a room temperature superconductor.
For that matter, we've had superconductors for decades that work at much higher temperatures than this one.
It seems the breakthrough is that you could use familiar semiconductor manufacturing processes. However the temperature is still going to be a major issue. I don't want a computer that requires liquid helium cooling.
> I don't want a computer that requires liquid helium cooling.
True, but I /can/ see someone, such as Sandia National Labs, very much willing to install a liquid helium cooled computer if it provides a significant performance increase above their existing supercomputer installations.
> you could use familiar semiconductor manufacturing processes.
Unclear to me why that's helpful. Materials that superconduct at a higher temperature than this one aren't hard to come by, or obscure:
> In 1913, lead was found to superconduct at 7 K,
Probably because they don’t behave well for normal lithography techniques? The high temp superconductors I know of are weird meta materials, and good luck getting them to exist in chip form at all.
Isn’t that very close to the practical limit for cooling in a lab?
Not that hard. A dilution fridge, used for instance for cooling quantum computers, can go much lower:
https://en.wikipedia.org/wiki/Dilution_refrigerator
Quantum devices are already cooled to that temperature (at least for some technologies), so it's not a problem in that use case.
Thanks!
Was gonna be lazy and say… temp or is doesn't matter.