Sign in with
Sign up | Sign in

Forget 3D, Purdue Scientists Have a 4D Transistor

By - Source: Purdue University | B 45 comments

Following trigate, or 3D transistors, there could be 4D transistors. But don't get excited just yet.

Researchers at Purdue University said that they succeeded in replacing silicon in transistors and create a path to much smaller chip structures. Instead of silicon, the team at Purdue used indium-gallium-arsenide that could become a critical material for the production of semiconductors below 10 nm. A prototype built at Purdue has been made in a 20 nm process.

According to Peide Ye, a professor of electrical and computer engineering at Purdue, three indium-gallium-arsenide wires were stacked on top of each other, while being progressively shorter to the top. Including the tapered cross section, the structures resembles the shape of a Christmas tree. So, why would we call this a 4D transistor? Here is the explanation in Peide's words:

"A one-story house can hold so many people, but more floors, more people, and it's the same thing with transistors," Ye said. "Stacking them results in more current and much faster operation for high-speed computing. This adds a whole new dimension, so I call them 4D." Curb your enthusiasm. No time travel yet.

However, indium-gallium-arsenide is, in fact, an interesting material to scale chip structures and, as Peide points out, silicon may run into physical limits in the 10 nm neighborhood. Whether that will be the case or not, we know that there are solutions that will keep Moore's Law alive for some time.

 

Contact Us for News Tips, Corrections and Feedback

Display 45 Comments.
This thread is closed for comments
Top Comments
  • 30 Hide
    sixdegree , December 10, 2012 6:38 AM
    I wonder when all this space-age tech ever gonna found their way into regular desktop. Also, stacking something on top of something else is adding a new dimension? I'll keep that in mind the next time i visit Burger King.
  • 19 Hide
    randomizer , December 10, 2012 6:56 AM
    Quote:
    This adds a whole new dimension, so I call them 4D.


    Does Purdue include a large marketing component in its engineering degrees? This sort of fluff normally comes from the PR squad, not the engineers.
  • 13 Hide
    diddo , December 10, 2012 7:08 AM
    OnihikageHow would the toxicity of an indium-gallium-arsenide chip and its production compare to those of silicon?
    I would not eat neither an i-g-a nor a Si based wafer...
Other Comments
  • 2 Hide
    tomfreak , December 10, 2012 6:06 AM
    Quote:
    Peide points out, silicon may run into physical limits in the 10 nm neighborhood
    with me skeptical about new changes, new material. I guess it is logical for me to get a high end 14nm Broadwell-E for my next rig. Since desktop life cycle is likely to last at least 5-6yrs, longer if I go highend. By that time the new material should have mature.
  • 30 Hide
    sixdegree , December 10, 2012 6:38 AM
    I wonder when all this space-age tech ever gonna found their way into regular desktop. Also, stacking something on top of something else is adding a new dimension? I'll keep that in mind the next time i visit Burger King.
  • 2 Hide
    army_ant7 , December 10, 2012 6:44 AM
    Quote:
    Quote:
    Peide points out, silicon may run into physical limits in the 10 nm neighborhood
    with me skeptical about new changes, new material. I guess it is logical for me to get a high end 14nm Broadwell-E for my next rig. Since desktop life cycle is likely to last at least 5-6yrs, longer if I go highend. By that time the new material should have mature.
    +1'ed you. Though it would still depend on what you'll be using that PC for. Hopefully, by that time, we'd have a lot more software that take advantage of multiple cores, 'coz if not, then you might not keep up as much as you may want to in terms of performance when newer architectures come out.

    I'm imagining a time when Intel will just have to resort to subsequent "Tock's" until smaller process transistors become available.

    Anyway, very interesting stuff! The reason behind calling them "4D" seems pretty cheesy to me... They should just call them "Stacked Transistor Technology" or something. Sounds cool to me... Actually, knowing how the tech works and what it's capable of are good enough to make it cool. :p 

    Its using indium-gallium-arsenide as the material is sort of a different thing on it's own compared to the way they made the transistors right?

  • 5 Hide
    Onihikage , December 10, 2012 6:53 AM
    How would the toxicity of an indium-gallium-arsenide chip and its production compare to those of silicon?
  • -1 Hide
    tomfreak , December 10, 2012 6:56 AM
    @ army_ant7,

    if they still researching now that means they still have no idea what and how is reliability of the material on a commercial product.(dont get me start asking if ever the new material CPU could last 5-6yrs or not).

    After we seen all the early SSD adopters are facing, u cant blame me skeptical on new technology. I'll be getting broadwell-E, the last of its kind base on a proven >40yrs old silicon technology that rig is likely to hold out for at least 7yrs b4 become outdated. :D 
  • 19 Hide
    randomizer , December 10, 2012 6:56 AM
    Quote:
    This adds a whole new dimension, so I call them 4D.


    Does Purdue include a large marketing component in its engineering degrees? This sort of fluff normally comes from the PR squad, not the engineers.
  • -7 Hide
    deksman , December 10, 2012 7:06 AM
    Synthetic diamonds could have been used for the very same purpose since 1996.
    Graphene since late 2008.
    And those 2 materials would be far better for this purpose.

    Synthetic diamonds also could have been used for production of displays in 1996.

    Oh but wait... nevermind the premise that we can create these superior synthetic materials in abundance... the 'market' works by using 'cheap' and 'cost efficient' materials (not what is technologically more efficient/better) and then they release the least efficient product first, followed by revisions in the upcoming years for the purpose of profits.

    I detest Capitalism... it doesn't promote innovation or gives us the best of what technology is capable of (in a sustainable capacity).

  • 13 Hide
    diddo , December 10, 2012 7:08 AM
    OnihikageHow would the toxicity of an indium-gallium-arsenide chip and its production compare to those of silicon?
    I would not eat neither an i-g-a nor a Si based wafer...
  • 11 Hide
    kensingtron , December 10, 2012 7:14 AM
    randomizerDoes Purdue include a large marketing component in its engineering degrees? This sort of fluff normally comes from the PR squad, not the engineers.


    Academics love talking them selves up; self marketing is something they excel at. 4d, bah, way to piss the physicists off.
  • 7 Hide
    godnodog , December 10, 2012 7:18 AM
    4d is clearly for marketing, as 4D objects can´t be created (at least by current technology), IF I am not misstaken.
  • -4 Hide
    Pyree , December 10, 2012 7:20 AM
    diddoI would not eat neither an i-g-a nor a Si based wafer...

    How about C H O C O La Te based wafer? :lol: 
  • 2 Hide
    photonboy , December 10, 2012 7:26 AM
    PROBLEM:

    When you increase the DENSITY of transistors you always run into heat dissipation problems. This sounds essentially no different from simply stacking chips.

    I remember for stacking chips they said "oh, we transfer the heat out the SIDE to solve that problem. Uh, really? Heat dissipation requires SQUARE AREA.

    PC GAMERS realized this issue with the new Intel CPU's (eg. i5-3570K). It could not achieve the same high-end frequency (i.e. 5GHz) as its predecessor. (This caused much confusion, as people said it "sucked" when in fact it used less power and offered more performance at the same frequency).

    I'm all for advances in technology but they have to pass the basic tests:
    - heat dissipation
    - manufacturing costs
  • 8 Hide
    Drazek , December 10, 2012 7:41 AM
    Quote:
    and it's the same thing with transistors," Ye said.


    Am I the only one who subconsciously slipped into a Pirate voice after reading that line?
  • 6 Hide
    greghome , December 10, 2012 8:21 AM
    For a second there, I thought this new transistor goes 88mp/h :p 
  • 0 Hide
    army_ant7 , December 10, 2012 8:41 AM
    Quote:
    @ army_ant7,
    if they still researching now that means they still have no idea what and how is reliability of the material on a commercial product.(dont get me start asking if ever the new material CPU could last 5-6yrs or not).

    After we seen all the early SSD adopters are facing, u cant blame me skeptical on new technology. I'll be getting broadwell-E, the last of its kind base on a proven >40yrs old silicon technology that rig is likely to hold out for at least 7yrs b4 become outdated. :D 
    A good point that applies to many other things, like games or software in general (some being very buggy at first). :) 
    What I was getting at is that hopefully, Broadwell-E (if ever it's made) would hold out in terms of performance for you. I doubt though that it would become "too slow" anytime soon after its theoretical release. :) 

    Quote:
    Synthetic diamonds could have been used for the very same purpose since 1996.
    Graphene since late 2008.
    And those 2 materials would be far better for this purpose.
    Synthetic diamonds also could have been used for production of displays in 1996.
    Oh but wait... nevermind the premise that we can create these superior synthetic materials in abundance... the 'market' works by using 'cheap' and 'cost efficient' materials (not what is technologically more efficient/better) and then they release the least efficient product first, followed by revisions in the upcoming years for the purpose of profits.
    I detest Capitalism... it doesn't promote innovation or gives us the best of what technology is capable of (in a sustainable capacity).
    Wait a moment. Have you thought why these materials aren't "cost efficient" in the first place? I have doubts Capitalism is solely if at all, to blame. My friend, ever heard of supply and demand? (Just tell me if you want me to explain further.) :) 

    Quote:
    PROBLEM:

    When you increase the DENSITY of transistors you always run into heat dissipation problems. This sounds essentially no different from simply stacking chips.

    I remember for stacking chips they said "oh, we transfer the heat out the SIDE to solve that problem. Uh, really? Heat dissipation requires SQUARE AREA.

    PC GAMERS realized this issue with the new Intel CPU's (eg. i5-3570K). It could not achieve the same high-end frequency (i.e. 5GHz) as its predecessor. (This caused much confusion, as people said it "sucked" when in fact it used less power and offered more performance at the same frequency).

    I'm all for advances in technology but they have to pass the basic tests:
    - heat dissipation
    - manufacturing costs
    Just wondering, is there a possibility that this new material (indium-gallium-arsenide) would release less heat (or maybe be more heat conductive that it allows heat to dissipate through neighboring transistors and then IHS or something) and thus allow this kind of stacking of transistors? :) 

  • 6 Hide
    vitreoushumor , December 10, 2012 9:24 AM
    Indium-gallium-arsenide is not new! It has been used for high power and high speed applications for 45 years or so. It is expensive and somewhat toxic. What is new here is the structure of the transistor.
  • 1 Hide
    wanderer11 , December 10, 2012 9:43 AM
    godnodog4d is clearly for marketing, as 4D objects can´t be created (at least by current technology), IF I am not misstaken.

    They mean 4d as in it goes in 4 directions. Not that there is a w, x, y, and z direction. Haven't you ever seen a 6-way intersection? You could call that 6d if you wanted.
  • -1 Hide
    tomfreak , December 10, 2012 9:45 AM
    vitreoushumorIndium-gallium-arsenide is not new! It has been used for high power and high speed applications for 45 years or so. It is expensive and somewhat toxic. What is new here is the structure of the transistor.
    if u see Intel 3D transistor problem with ivy bridge heat issue, u'll know that even these kind of problem is start to be tricky to Intel. I hope they iron out the issues @ broadwell. So it is unlikely I would take the risk to buy a new product under this new design b4 the mature, especially we are going beyond silicon limit 10nm
  • 3 Hide
    namecnassianer , December 10, 2012 9:58 AM
    Quote:
    For a second there, I thought this new transistor goes 88mp/h


    That second hasn't happened yet, future boy.
    - Dr. Emmett Brown
  • 3 Hide
    randomizer , December 10, 2012 10:08 AM
    wanderer11They mean 4d as in it goes in 4 directions. Not that there is a w, x, y, and z direction. Haven't you ever seen a 6-way intersection? You could call that 6d if you wanted.


    Dimensions and directions are different things, but we can overlook that in order to make it sound flashy.
Display more comments