4 ms·

China develops new method to mass-produce high-quality semiconductors

79 points by david927 1y ago

igravious 1y ago
“China Makes the Biggest Technology Breakthrough in Decades” https://www.youtube.com/watch?v=inQ9fKui5jw https://www.youtube.com/watch?v=inQ9fKui5jw
igravious 1y ago
“'Golden' chip production gets boost” https://www.chinadaily.com.cn/a/202507/21/WS687d99bca310ad07b5d90f29.html https://www.chinadaily.com.cn/a/202507/21/WS687d99bca310ad07...
clairemiller069 1y ago
[dead]
david927 1y ago
Quoting William Huo, InSe outperforms silicon in every category that matters: • 5–10x higher electron mobility • Atomic thickness • Tunable bandgap • Lower power leakage • Faster switching So how big of news is this?
- igravious 1y ago
  Literally world changing
- addaon 1y ago
  What's the story for an insulator? Silicon's pretty awesome, but in many ways silicon dioxide (grown in place) is the real magic.
  - adrian_b 1y ago
    Unfortunately, even if the thermally-grown silicon dioxide is what has enabled the appearance of the monolithic integrated circuits and of the MOSFET transistors, eventually it had to be replaced. There are already around two decades since silicon dioxide is no longer used as the gate insulator in high-performance transistors. The reason is that its dielectric constant is too low, so for very small transistors the gate would have to be too thin, so thin that it is impossible for it to not have holes and also impossible to prevent electrons from tunneling through it. Therefore silicon dioxide has been replaced by hafnium dioxide (a part of the hafnium may be substituted with zirconium or rare-earth elements), which has a much higher dielectric constant, and which is also chemically resistant enough to survive the following wafer processing steps. Because HfO2 cannot be grown in place, but it must be deposited in its entirety, it is much more difficult to ensure that the insulator-semiconductor interface is perfect, but the industry had to solve this problem decades ago, otherwise it would have been impossible to make the transistors smaller. For other semiconductors than silicon, the gate insulator always had to be deposited. Because this is hard, metal-insulator-semiconductor FETs not made of silicon or of silicon carbide had only very rarely been used in the past. The transistors that are not made of Si or SiC typically are either Shottky-gate FETs or heterojunction bipolar transistors.
- chasil 1y ago
  This does not appear to be an ingot, as silicon would be prior to being cut, but a film grown on some substrate, itself called a "wafer," so perhaps silicon? The film itself is one atomic layer in thickness? I don't know how you would make "wells" that form a FET, either for the source and drain, or for the larger complimentary wells of CMOS. I don't know how advanced the thinking is to do this, or an equivalent. https://www.science.org/doi/10.1126/science.adu3803 https://www.science.org/doi/10.1126/science.adu3803
  - adrian_b 1y ago
    Many semiconductor materials cannot be grown as ingots that are later cut into wafers, like it is done with silicon or germanium, but they are grown epitaxially as thin layers on wafers made of other semiconductor or insulator materials that have a compatible crystal structure. Most gallium nitride devices, like those that are used now in miniature chargers for laptops/smartphones, are made like this. Using this technique for indium selenide is a standard procedure, not something surprising. The "wells" are made by doping with various kinds of atoms, which is normally done by ion implantation, i.e. a ion beam inserts the desired impurities into the crystal, at the desired depth. In very thin devices, like in most modern CMOS technologies, the doped zones no longer look like "wells". For an N-channel MOSFET, you just have from source to drain 3 zones of alternating polarity, n-p-n. The middle zone is surrounded partially or even totally by the gate insulator.
- hristov 1y ago
  I vastly underperforms silicon in the most important category: cost.
  - gsf_emergency_2 1y ago
    Wait for Chinese to find a (mostly) aluminum based III-V/VI They won't tell anyone
  - 0cf8612b2e1e 1y ago
    Looks like indium is about $390/kg. That is not cheap, but does not seem significant given how much material goes into a high performance CPU.
    - adrian_b 1y ago
      The reserves of indium are extremely small. Any new use that would increase demand would also increase the price. Also that price is for commercially pure indium. After indium is purified enough to be usable in semiconductor devices the price increases many times, possibly much more than 10 times. Currently, the major consumers of indium are all the screens for monitors, laptops and smartphones, which use indium oxide as a transparent conductor, then the LEDs used in lighting and indicators. Also the power devices with gallium nitride contain indium, and their use is increasing. There exist no mines of indium. Indium can be obtained as a byproduct from the extraction of other metals, primarily from zinc mining, but its concentration in zinc minerals is very small. In order to produce more indium, one also has to produce more zinc, in an amount several orders of magnitude greater than the amount of produced indium. When the demand from indium will exceed that available from the current zinc production, further increases in demand will increase the indium price much steeper. Except for indium, among the other chemical elements only for the platinum-group metals there is a so great mismatch between the amount that would be required by potential applications and the amount that is available on Earth. Selenium and tellurium are also close of these from this point of view.
      - phkahler 1y ago
        Well if it's used as a thin film on screens, it wouldn't require much more production to make a few square centimeters more for chips to build a computer to use that screen. We also ramped production of those screens without some major supply chain problem requiring 10x zinc production. I don't see scarcity as a problem.
- ajross 1y ago
  There are vast lists of "better than silicon" semiconductor crystals. Most good power mosfets these days are gallium nitride, for example. Where they all fail is production process, which amounts to transistor size, basically. Sure, you can make a really great, very efficient, measurably improved very much macroscopic transistor. But no one knows how to put a hundred billion of them on a chip, so... basically who cares? Someday someone will figure it out, maybe. But announcing an exciting new chemistry says little to nothing.
  - adrian_b 1y ago
    The problem is not the transistor size, but the transistor cost. It is easy to make transistors as small as on silicon on most other semiconductors, but the cost of the final product would be many times greater. One important reason is that silicon is made into huge 12-in wafers, while most other semiconductors are made into small 2-inch to 4-inch wafers, like for silicon several decades ago. At each processing step on a silicon production line one machine processes 9 to 36 times more transistors than if another semiconductor were used. Large wafers also waste much less area when making big dies. In general, for silicon there exist very big processing machines with very high productivity, while for other materials there is little difference between lab equipment and what can be used for commercial production. An integrated circuit made on a non-silicon material also requires more processing steps and more expensive materials. In order to keep increasing the performance of CPUs and GPUs, the replacement of silicon with another semiconductor is unavoidable, perhaps in a decade from now. However, that will be done only after all other possibilities of improving silicon devices will be completely exhausted, in order to avoid the increase in production costs.
    - ajross 1y ago
      > It is easy to make transistors as small as on silicon on most other semiconductors Yikes, [citation needed] here. No, it absolutely is not. All the etch and litho chemistry is highly specific to the substrate and dopants. You can't just feed a germanium crystal through an etcher tool in a TSMC fab and get anything but a brick out the other side. I'm not aware of anyone anywhere doing low-nm lithography on anything but silicon (even in a demo context, or even announcing plans for the capacity), but I'm willing to be educated.
hristov 1y ago
I am not going to download an app to read an article.
- ta8645 1y ago
  FWIW, reader view in Firefox made it easy to read.
babelfish 1y ago
https://archive.is/UUTet https://archive.is/UUTet
AuthAuth 1y ago
Yep, we're cooked if this is true.
- thenthenthen 1y ago
  Before the west can blink thats for sure. Give it a year.
jvanderbot 1y ago
Is it because most HN readers are _not_ in China that we lump together all scientific results as "China does thing"? Is this also common for EU/USA? Do we say "UK developers new method for ..." or "Researchers at Cambridge"? I swear I'm not making any political statements, just wondering why we treat it as a homogeneous entity.
- ken47 1y ago
  Occam’s Razor applies.
- knome 1y ago
  >Do we say "UK developers new method for ..." or "Researchers at Cambridge" these would be completely normal headlines, in my experience.
- mitthrowaway2 1y ago
  If you think it should be phrased differently, you should tell it to CGTN, China's Beijing-based state-run media. HN merely accurately transcribed their headline, as per the site rules.
- nneonneo 1y ago
  CGTN is a Chinese state-run broadcaster; they themselves chose to use the phrasing “China does thing”. It may be in an effort to present a more unified face for the widely varied research happening in China, as well as showing a more easily-understood headline for Western audiences.
  - dingnuts 1y ago
    that's a polite way to describe propaganda
    - uni_baconcat 1y ago
      There’s another way to find this out. Just one tap to CGTN homepage. Try to read how they describe government news, domestic news, and international news.
abdullahkhalids 1y ago
Personally, I don't care about mass production of SOTA semiconductors. My personal computing resource needs haven't really increased in the last ten years, and I doubt they have for the median person. I want 10 million dollar factories that can make 10 year old semiconductor chips.
- jackyard86 1y ago
  You get the said 10 year old chip cheaper, with lower power consumption and smaller physical size, on newer and improved fabrication methods.
  - CamperBob2 1y ago
    Yeeeeaaaah... that's not how that works. Legacy processes are often difficult or impractical to transport between fabs at the same node, much less fabs at different ones.
    - Kirby64 1y ago
      GP is saying you get the same functional performance as older chips for less size, power, and cost. Nobody makes main processors in 22nm anymore, for instance. That’s basically what was used in production 10 years ago for processors.
      - CamperBob2 1y ago
        No, GP is saying 'you get the same chip' but on a different process node. You don't, pretty much. Exceptions exist, but not when it comes to mixed-signal processes.
        Kirby64 1y ago
        People do that too. Fab ports to smaller nodes are something that absolutely happens, especially as older legacy nodes close down. It happens all the time.
ta8645 1y ago
It's slightly worrying that they talk about its applicability in "smart terminals" rather than personal computers. I hope that's not them saying the quiet part out loud, about the future (or lack thereof) of personal computing.
adrian_b 1y ago
It should be noted that this is a technology for producing a semiconductor material with very high performance, but also with very high cost, a cost that is impossible to reduce, because both indium and selenium are among the least abundant elements on Earth (both having an abundance similar to silver, but being much more difficult to mine than silver, because they are very dispersed). This material will never replace cheap materials, like silicon or silicon carbide, or even gallium nitride, in the bulk of semiconductor devices, e.g. in CPUs and memories, or in power semiconductor devices. It will be reserved for a few high-speed devices, in special instruments that need high-speed signal processing or in radars or communication devices used in high-frequency bands (obviously these include military applications).
- sandspar 1y ago
  How is this method better for those uses than the old method?
  - adrian_b 1y ago
    According to TFA, they have succeed to produce 2-inch wafers of indium selenide with a small enough density of defects, which nobody did previously. The first roadblock on the use of new semiconductor materials is that in the beginning nobody succeeds to make crystals that are both big enough and free enough of defects. For size, usually achieving to make 2-inch wafers is the threshold for enabling commercial applications. The second problem is finding metallization systems that can achieve ohmic contacts and rectifying contacts on the semiconductor crystal and the third is finding impurities that allow to modify the polarity and the concentration of the charge carriers in wide enough ranges. These 2 problems are particularly difficult for wide-bandgap semiconductors, like gallium nitride, but they are unlikely to be difficult for indium selenide, which should behave similarly to zinc selenide or indium phosphide, for which there is much more experience.
- roarcher 1y ago
  Selenium gets washed down residential shower drains every day in the form of dandruff shampoo. Surely if we can afford that, there's enough to go around for the semiconductor industry?
jauntywundrkind 1y ago
We're definitely seeing an age where cost-is-no-limit, and the properties here look amazing. Still, I guess I'm more interested in what Nikon is doing. Simplifying the process—eliminating the need for masks—could significantly decrease initial costs. https://www.techpowerup.com/339060/nikon-introduces-600x600-mm-substrates-for-advanced-ai-silicon https://www.techpowerup.com/339060/nikon-introduces-600x600-...
karteum 1y ago
I suspect there is much less Indium and Selenium mining resources available on earth compared to silicium, so I wonder to what extent we can speak about scaling to "mass production"...

79 points by david927 1y ago

“China Makes the Biggest Technology Breakthrough in Decades” https://www.youtube.com/watch?v=inQ9fKui5jw https://www.youtube.com/watch?v=inQ9fKui5jw

igravious 1y ago

“'Golden' chip production gets boost” https://www.chinadaily.com.cn/a/202507/21/WS687d99bca310ad07b5d90f29.html https://www.chinadaily.com.cn/a/202507/21/WS687d99bca310ad07...

clairemiller069 1y ago

[dead]

david927 1y ago

Quoting William Huo, InSe outperforms silicon in every category that matters: • 5–10x higher electron mobility • Atomic thickness • Tunable bandgap • Lower power leakage • Faster switching So how big of news is this?

igravious 1y ago
Literally world changing
addaon 1y ago
What's the story for an insulator? Silicon's pretty awesome, but in many ways silicon dioxide (grown in place) is the real magic.
- adrian_b 1y ago
  Unfortunately, even if the thermally-grown silicon dioxide is what has enabled the appearance of the monolithic integrated circuits and of the MOSFET transistors, eventually it had to be replaced. There are already around two decades since silicon dioxide is no longer used as the gate insulator in high-performance transistors. The reason is that its dielectric constant is too low, so for very small transistors the gate would have to be too thin, so thin that it is impossible for it to not have holes and also impossible to prevent electrons from tunneling through it. Therefore silicon dioxide has been replaced by hafnium dioxide (a part of the hafnium may be substituted with zirconium or rare-earth elements), which has a much higher dielectric constant, and which is also chemically resistant enough to survive the following wafer processing steps. Because HfO2 cannot be grown in place, but it must be deposited in its entirety, it is much more difficult to ensure that the insulator-semiconductor interface is perfect, but the industry had to solve this problem decades ago, otherwise it would have been impossible to make the transistors smaller. For other semiconductors than silicon, the gate insulator always had to be deposited. Because this is hard, metal-insulator-semiconductor FETs not made of silicon or of silicon carbide had only very rarely been used in the past. The transistors that are not made of Si or SiC typically are either Shottky-gate FETs or heterojunction bipolar transistors.
chasil 1y ago
This does not appear to be an ingot, as silicon would be prior to being cut, but a film grown on some substrate, itself called a "wafer," so perhaps silicon? The film itself is one atomic layer in thickness? I don't know how you would make "wells" that form a FET, either for the source and drain, or for the larger complimentary wells of CMOS. I don't know how advanced the thinking is to do this, or an equivalent. https://www.science.org/doi/10.1126/science.adu3803 https://www.science.org/doi/10.1126/science.adu3803
- adrian_b 1y ago
  Many semiconductor materials cannot be grown as ingots that are later cut into wafers, like it is done with silicon or germanium, but they are grown epitaxially as thin layers on wafers made of other semiconductor or insulator materials that have a compatible crystal structure. Most gallium nitride devices, like those that are used now in miniature chargers for laptops/smartphones, are made like this. Using this technique for indium selenide is a standard procedure, not something surprising. The "wells" are made by doping with various kinds of atoms, which is normally done by ion implantation, i.e. a ion beam inserts the desired impurities into the crystal, at the desired depth. In very thin devices, like in most modern CMOS technologies, the doped zones no longer look like "wells". For an N-channel MOSFET, you just have from source to drain 3 zones of alternating polarity, n-p-n. The middle zone is surrounded partially or even totally by the gate insulator.
hristov 1y ago
I vastly underperforms silicon in the most important category: cost.
- gsf_emergency_2 1y ago
  Wait for Chinese to find a (mostly) aluminum based III-V/VI They won't tell anyone
- 0cf8612b2e1e 1y ago
  Looks like indium is about $390/kg. That is not cheap, but does not seem significant given how much material goes into a high performance CPU.
  - adrian_b 1y ago
    The reserves of indium are extremely small. Any new use that would increase demand would also increase the price. Also that price is for commercially pure indium. After indium is purified enough to be usable in semiconductor devices the price increases many times, possibly much more than 10 times. Currently, the major consumers of indium are all the screens for monitors, laptops and smartphones, which use indium oxide as a transparent conductor, then the LEDs used in lighting and indicators. Also the power devices with gallium nitride contain indium, and their use is increasing. There exist no mines of indium. Indium can be obtained as a byproduct from the extraction of other metals, primarily from zinc mining, but its concentration in zinc minerals is very small. In order to produce more indium, one also has to produce more zinc, in an amount several orders of magnitude greater than the amount of produced indium. When the demand from indium will exceed that available from the current zinc production, further increases in demand will increase the indium price much steeper. Except for indium, among the other chemical elements only for the platinum-group metals there is a so great mismatch between the amount that would be required by potential applications and the amount that is available on Earth. Selenium and tellurium are also close of these from this point of view.
    - phkahler 1y ago
      Well if it's used as a thin film on screens, it wouldn't require much more production to make a few square centimeters more for chips to build a computer to use that screen. We also ramped production of those screens without some major supply chain problem requiring 10x zinc production. I don't see scarcity as a problem.
ajross 1y ago
There are vast lists of "better than silicon" semiconductor crystals. Most good power mosfets these days are gallium nitride, for example. Where they all fail is production process, which amounts to transistor size, basically. Sure, you can make a really great, very efficient, measurably improved very much macroscopic transistor. But no one knows how to put a hundred billion of them on a chip, so... basically who cares? Someday someone will figure it out, maybe. But announcing an exciting new chemistry says little to nothing.
- adrian_b 1y ago
  The problem is not the transistor size, but the transistor cost. It is easy to make transistors as small as on silicon on most other semiconductors, but the cost of the final product would be many times greater. One important reason is that silicon is made into huge 12-in wafers, while most other semiconductors are made into small 2-inch to 4-inch wafers, like for silicon several decades ago. At each processing step on a silicon production line one machine processes 9 to 36 times more transistors than if another semiconductor were used. Large wafers also waste much less area when making big dies. In general, for silicon there exist very big processing machines with very high productivity, while for other materials there is little difference between lab equipment and what can be used for commercial production. An integrated circuit made on a non-silicon material also requires more processing steps and more expensive materials. In order to keep increasing the performance of CPUs and GPUs, the replacement of silicon with another semiconductor is unavoidable, perhaps in a decade from now. However, that will be done only after all other possibilities of improving silicon devices will be completely exhausted, in order to avoid the increase in production costs.
  - ajross 1y ago
    > It is easy to make transistors as small as on silicon on most other semiconductors Yikes, [citation needed] here. No, it absolutely is not. All the etch and litho chemistry is highly specific to the substrate and dopants. You can't just feed a germanium crystal through an etcher tool in a TSMC fab and get anything but a brick out the other side. I'm not aware of anyone anywhere doing low-nm lithography on anything but silicon (even in a demo context, or even announcing plans for the capacity), but I'm willing to be educated.

hristov 1y ago

I am not going to download an app to read an article.

ta8645 1y ago
FWIW, reader view in Firefox made it easy to read.

babelfish 1y ago

https://archive.is/UUTet https://archive.is/UUTet

AuthAuth 1y ago

Yep, we're cooked if this is true.

thenthenthen 1y ago
Before the west can blink thats for sure. Give it a year.

jvanderbot 1y ago

Is it because most HN readers are _not_ in China that we lump together all scientific results as "China does thing"? Is this also common for EU/USA? Do we say "UK developers new method for ..." or "Researchers at Cambridge"? I swear I'm not making any political statements, just wondering why we treat it as a homogeneous entity.

ken47 1y ago
Occam’s Razor applies.
knome 1y ago
>Do we say "UK developers new method for ..." or "Researchers at Cambridge" these would be completely normal headlines, in my experience.
mitthrowaway2 1y ago
If you think it should be phrased differently, you should tell it to CGTN, China's Beijing-based state-run media. HN merely accurately transcribed their headline, as per the site rules.
nneonneo 1y ago
CGTN is a Chinese state-run broadcaster; they themselves chose to use the phrasing “China does thing”. It may be in an effort to present a more unified face for the widely varied research happening in China, as well as showing a more easily-understood headline for Western audiences.
- dingnuts 1y ago
  that's a polite way to describe propaganda
  - uni_baconcat 1y ago
    There’s another way to find this out. Just one tap to CGTN homepage. Try to read how they describe government news, domestic news, and international news.

abdullahkhalids 1y ago

Personally, I don't care about mass production of SOTA semiconductors. My personal computing resource needs haven't really increased in the last ten years, and I doubt they have for the median person. I want 10 million dollar factories that can make 10 year old semiconductor chips.

jackyard86 1y ago
You get the said 10 year old chip cheaper, with lower power consumption and smaller physical size, on newer and improved fabrication methods.
- CamperBob2 1y ago
  Yeeeeaaaah... that's not how that works. Legacy processes are often difficult or impractical to transport between fabs at the same node, much less fabs at different ones.
  - Kirby64 1y ago
    GP is saying you get the same functional performance as older chips for less size, power, and cost. Nobody makes main processors in 22nm anymore, for instance. That’s basically what was used in production 10 years ago for processors.
    - CamperBob2 1y ago
      No, GP is saying 'you get the same chip' but on a different process node. You don't, pretty much. Exceptions exist, but not when it comes to mixed-signal processes.
      - Kirby64 1y ago
        People do that too. Fab ports to smaller nodes are something that absolutely happens, especially as older legacy nodes close down. It happens all the time.

ta8645 1y ago

It's slightly worrying that they talk about its applicability in "smart terminals" rather than personal computers. I hope that's not them saying the quiet part out loud, about the future (or lack thereof) of personal computing.

adrian_b 1y ago

It should be noted that this is a technology for producing a semiconductor material with very high performance, but also with very high cost, a cost that is impossible to reduce, because both indium and selenium are among the least abundant elements on Earth (both having an abundance similar to silver, but being much more difficult to mine than silver, because they are very dispersed). This material will never replace cheap materials, like silicon or silicon carbide, or even gallium nitride, in the bulk of semiconductor devices, e.g. in CPUs and memories, or in power semiconductor devices. It will be reserved for a few high-speed devices, in special instruments that need high-speed signal processing or in radars or communication devices used in high-frequency bands (obviously these include military applications).

sandspar 1y ago
How is this method better for those uses than the old method?
- adrian_b 1y ago
  According to TFA, they have succeed to produce 2-inch wafers of indium selenide with a small enough density of defects, which nobody did previously. The first roadblock on the use of new semiconductor materials is that in the beginning nobody succeeds to make crystals that are both big enough and free enough of defects. For size, usually achieving to make 2-inch wafers is the threshold for enabling commercial applications. The second problem is finding metallization systems that can achieve ohmic contacts and rectifying contacts on the semiconductor crystal and the third is finding impurities that allow to modify the polarity and the concentration of the charge carriers in wide enough ranges. These 2 problems are particularly difficult for wide-bandgap semiconductors, like gallium nitride, but they are unlikely to be difficult for indium selenide, which should behave similarly to zinc selenide or indium phosphide, for which there is much more experience.
roarcher 1y ago
Selenium gets washed down residential shower drains every day in the form of dandruff shampoo. Surely if we can afford that, there's enough to go around for the semiconductor industry?

jauntywundrkind 1y ago

We're definitely seeing an age where cost-is-no-limit, and the properties here look amazing. Still, I guess I'm more interested in what Nikon is doing. Simplifying the process—eliminating the need for masks—could significantly decrease initial costs. https://www.techpowerup.com/339060/nikon-introduces-600x600-mm-substrates-for-advanced-ai-silicon https://www.techpowerup.com/339060/nikon-introduces-600x600-...

karteum 1y ago

I suspect there is much less Indium and Selenium mining resources available on earth compared to silicium, so I wonder to what extent we can speak about scaling to "mass production"...