Forgot your password?
typodupeerror
Australia Science

Researchers Create 4nm Transistor With Seven Atoms 120

Posted by CmdrTaco
from the xanthir-has-almost-70k-hp dept.
EmagGeek writes "University researchers have created a transistor by replacing just seven atoms of silicon with phosphorous. The seven-atom transistor has hopeful implications for the future of quantum cryptography, nuclear and weather modeling, and other applications. 'The significance of this achievement is that we are not just moving atoms around or looking at them through a microscope,' says Professor Michelle Simmons, a co-author of a paper on the subject that is being published by Nature Nanotechnology. The paper is entitled 'Spectroscopy of Few-Electron Single-Crystal Silicon Quantum Dots'."
This discussion has been archived. No new comments can be posted.

Researchers Create 4nm Transistor With Seven Atoms

Comments Filter:
  • 7 Atoms? (Score:3, Funny)

    by Anonymous Coward on Wednesday May 26, 2010 @10:32AM (#32348420)

    Should've used a VIA C7 instead.

  • by Anonymous Coward on Wednesday May 26, 2010 @10:34AM (#32348454)

    It sounds like they did this by moving single atoms at a time, and not through any kind of lithography, or mass-producible process. So while neat, like the single atom transistor story from a while back, it doesn't look like they really have a way to produce billions of these at a time. We may have to wait a long time before we see anything like this in our home PCs.

    • by $RANDOMLUSER (804576) on Wednesday May 26, 2010 @11:01AM (#32348784)
      You probably don't need a quantum computer to wait on I/O in your home PC.
      • Re: (Score:2, Insightful)

        by Anonymous Coward

        You probably don't need a quantum computer to wait on I/O in your home PC.

        You also probably don't need more than 640K.

        • Re: (Score:2, Informative)

          by Anonymous Coward

          This isn't insightful. The analogy drawn is invalid on many levels.

          The argument isn't that a quantum computer isn't necessary. It's pointing out the fact that computing is often I/O limited - how fast a computer can move data around to be processed. He's saying that advances need to come in other areas before things like this are significant.

          And, as someone with a background in these things - you don't make a good transistor with 7 phosphorus atoms. There has to be more to it. The fact that they create

      • You probably don't need a quantum computer to wait on I/O in your home PC.

        That depends. Will it speed up the Gaussian Blur filter in Photoshop?

    • Re: (Score:3, Interesting)

      by dpilot (134227)

      It'll take a really wicked manufacturing process to ever make, too. 7 atoms? What if you get only 6? What if you get 8? What if one is slightly off position? We've already been at sub-100nM processes for years now, and things are already too "grainy" for real comfort.

      Oh yeah, what's the difference between "on" current and "off" current?

      • Re: (Score:3, Interesting)

        by wiredlogic (135348)

        Even if you could mass produce this structure it will always be impractical because natural diffusion processes will cause the atoms to migrate out of position.

        • by Bakkster (1529253)

          Even in a crystaline structure? Forgive me, IANAMS.

          • by bain itic (1593851) on Wednesday May 26, 2010 @12:52PM (#32350220)

            Even in a crystaline structure? Forgive me, IANAMS.

            Yes, even in a crystalline structure. Diffusion in solids at the macroscopic scale seems slow compared to say, cream in your coffee, but at the atomic scale... They did this at the surface, which makes it even worse. I can't imagine this lasting any useful amount of time without some SEVERE cooling measures. I'm not sure if even liquid nitrogen could save it. IAAMS.

            • by Bakkster (1529253)

              Any idea how close (or far) we are from being able to produce stable room-temperature quantum dots and wells? Perpetually 20 years away?

              • I'm not involved directly in any research using quantum dots/wells, so I don't feel I qualified to make that kind of prediction. However, the article shows the quantum dot the transistor was contained in, which looks like its an order of magnitude larger (70-80 atoms). It's one or two atoms out of 70 move, there isn't going to be nearly the change in properties compared to one or two in the seven that act as the gate. Computer components need to be able to last for years at or near room temperature, and
      • Re: (Score:3, Insightful)

        by Thanshin (1188877)

        It'll take a really wicked manufacturing process to ever make, too. 7 atoms? What if you get only 6? What if you get 8? What if one is slightly off position?

        Building a car with 4 wheels? What if you only get 3? What if you get 5? What if one is slightly off position?

        An automated process doesn't care about size. What they did, can be replicated. Thus, it can be automated, unless there's a creative process involved that implies the use of a human mind, which I strongly doubt.

        If the automation is too slow, it can be multiplied. If multiplying is still not enough, the process itself of creating and assembling multiple automatons can be multiplied.

        Price vs usefulnes

        • Re: (Score:1, Informative)

          It'll take a really wicked manufacturing process to ever make, too. 7 atoms? What if you get only 6? What if you get 8? What if one is slightly off position?

          Building a car with 4 wheels? What if you only get 3? What if you get 5? What if one is slightly off position?

          An automated process doesn't care about size. What they did, can be replicated. Thus, it can be automated, unless there's a creative process involved that implies the use of a human mind, which I strongly doubt.

          If the automation is too slow, it can be multiplied. If multiplying is still not enough, the process itself of creating and assembling multiple automatons can be multiplied.

          Price vs usefulness of the final product may well be a problem, but size isn't. It was until it was solved, which is precisely the point of the news.

          In macroscopic terms the world is simple. The finer the resolution the more complex the world gets. In nanoscopic terms the world is complicated.

          Our current technology allows us to automate macroscopic processes with high precision. Nanotechnology however is one leading edge technology, and as such the precision certainly isn't there to make a fair comparison to automated macroscopic processes.

          Think of a doctor performing surgery: a large benign tumor in section of fat could be easily removed, while a min

          • Re: (Score:3, Insightful)

            by Thanshin (1188877)

            In macroscopic terms the world is simple. The finer the resolution the more complex the world gets. In nanoscopic terms the world is complicated.

            Making chips is considerably harder than making bricks; and yet we do make both.

            Our current technology allows us to automate macroscopic processes with high precision. Nanotechnology however is one leading edge technology, and as such the precision certainly isn't there to make a fair comparison to automated macroscopic processes.

            Making chips was once leading edge technology, not comparable to making bricks; and yet we made both.

            Think of a doctor performing surgery: a large benign tumor in section of fat could be easily removed, while a miniscule brain tumor would probably be one of the most difficult to remove.

            Removing a minuscule brain tumor is much harder than amputating a leg; and yet we do both.

            That's precisely the point of science and technology. Some guy spends years doing something that was previously impossible. Some other guys try little variants on the same action. And then a guy develops a process of doing the exact same thi

            • Re: (Score:1, Informative)

              by Anonymous Coward

              The problem is that even at current sizes, we experience a large amount of process variation, which is basically deviation of the actual device sizes from the ones you specified due to it being so damn hard to make something that small:

              http://en.wikipedia.org/wiki/Process_variation_%28semiconductor%29

              Process variation is becoming one of the biggest problems as chips shrink, because the variation in transistor sizes means that every circuit has to be designed with some amount of safety buffer, which increase

              • by Thanshin (1188877)

                Exactly. That, indeed, is a problem specific to the mechanization of the process.

                Additionally to improvements in fabrication techniques and design alterations (which I don't think will be possible in this case) there's also the often used option of discarding the bad results, which, as always, turns into a pure production cost problem.

            • Re: (Score:2, Funny)

              In macroscopic terms the world is simple. The finer the resolution the more complex the world gets. In nanoscopic terms the world is complicated.

              Making chips is considerably harder than making bricks; and yet we do make both.

              Our current technology allows us to automate macroscopic processes with high precision. Nanotechnology however is one leading edge technology, and as such the precision certainly isn't there to make a fair comparison to automated macroscopic processes.

              Making chips was once leading edge technology, not comparable to making bricks; and yet we made both.

              Think of a doctor performing surgery: a large benign tumor in section of fat could be easily removed, while a miniscule brain tumor would probably be one of the most difficult to remove.

              Removing a minuscule brain tumor is much harder than amputating a leg; and yet we do both.

              That's precisely the point of science and technology. Some guy spends years doing something that was previously impossible. Some other guys try little variants on the same action. And then a guy develops a process of doing the exact same thing but better, faster and cheaper.

              Once the action passes through the imposibility barrier, the steps from "breakthough" to "mundane" are well known. We've spent several thousand years walking those steps on each new discovery.

              So then, just so I'm clear, leg amputation is just as difficult as brain surgery; bricks are just as hard to make as silicon wafers.

              Thanks for clearing all that up. Now that you've enlightened me on now the world works I will fly home after work this evening using nothing but my arms. Because I can walk with my legs.

              • Re: (Score:1, Redundant)

                by Thanshin (1188877)

                So then, just so I'm clear, leg amputation is just as difficult as brain surgery; bricks are just as hard to make as silicon wafers.

                The point is precisely that being harder doesn't stop us from doing things.

                This conversation started with someone pointing the extra difficulties of a new, just proven, process. My point is that those difficulties, that obviously make the problem a hard one, were exactly what was proven resoluble. The news are that those problems were surpassed. We're now on the mechanizing the solution phase.

                The point is that the initially mentioned difficulties are the "already solved" ones. Not that it's an easy process

                • by dave420 (699308)
                  What are you smoking? Reliably being able to produce something so intricate and tiny, on any useful scale, is fantastically more complicated than producing current silicon chips. The process isn't proven - not even close.
                  • by WNight (23683)

                    The process is proven possible, not that there's a "proven process". That'd be more akin to a well-tested, bug free process.

            • by damnfuct (861910)
              There's a big difference between "something that can be done" and "something that can be done for a profit"; the latter having considerably more motivation (and desirability) behind it. Note that a good case-in-point for your argument is the first transistor made by Bell Labs in 1947. This transistor was large and crude by today's standards, but the fact that it *could* be done made it desirable to pursue the technology.
            • by geekoid (135745)

              All your example are macroscopic examples. You did nothing to counter the poster point. What you have is faith couples with bias confirmation.

              How many breakthrough never got to mundane?

        • by geekoid (135745)

          "What they did, can be replicated. Thus, it can be automate"

          Just because you can do something many times, doesn't mean it can be automated.

          Why you think it is 'solved' is beyond me. Will this turn into a practical application of optical drives? I hope so, but there are a lot of steps and process to be worked on before we can say it's possible to mass produce.

          Where in the paper does it say it's been solved?

      • In very simple terms, transistors work like a switch. When a voltage is applied to the base (one terminal), they allow current to flow between the collector and emitter (other two terminals). "On" current is when the transistor is allowing the current to flow. "Off" is...off. No current flows. If I'm wrong, someone correct me. I'm helping someone on a project with some transistors, and need to know if I'm messing it up.
        • Re: (Score:3, Informative)

          by Bakkster (1529253)

          "Off" is almost never zero current. There's usually just a tiny amount of 'leakage' current, although some quantum designs (such as this one seems to be) can have exactly no current while off.

          Basically, while all our computers and data are binary, they operate in an analog environment. We just treat any value greater than (for example) analog 0.8 as a digital 1, and anything less than analog 0.2 as a digital 0. The problem has been as we shrink the gate size and thickness and reduce supply voltage in or

        • Re: (Score:3, Informative)

          by tool462 (677306)

          You've got the theory basically correct, but in the real world the "off" current is just less current, not zero current. To get a good signal to noise ratio, you want Ion / Ioff to be as big as possible. In older processes (or thick oxide devices) you can get really good ratios. You could have an Ion of 10mA and an Ioff of 10nA, for example, for a ratio of 1e6. For newer process nodes on thin oxide devices, that ratio may get as low as 1e3 or worse. In that range, the device still works well for digita

          • by dpilot (134227)

            But see, that's exactly why I made my remark. I've been watching the characteristics of these "transistors" the process guys give us degrade with each generation. Active current is kept under fairly decent control, but standby current is rising fast. For a silly joke, I like to think of 2 lines on a chart, active and standby currents, and wonder when standby will surpass active.

            BTW, in spite of your saying the device isn't suitable for digital circuitry - analog circuitry is a heck of a lot tougher. By

            • by tool462 (677306)

              I agree completely and thought it was a good question in your post. I was just trying to give a simple explanation of why you would be interested in Ion vs. Ioff for guppysap13.

              Wiring up the device is a whole other issue with this 7-atom gate. The small process nodes already have the total device area dominated by the source/drain just to be able to wire it up. Going to effectively zero gate area will only get you maybe a 10% improvement in total area usage without some major changes in how the circuitry

              • by dpilot (134227)

                I missed a post in there - it didn't show the one you were responding to, and I thought you were responding to me. My mistake.

        • by geekoid (135745)

          depend is it's PNP or NPN.

          And that is extremely simple terms. You need to read up on Gain. Yes it can be used to turn it on or off(with some leakage) but it also controls the output in proportion to the input.

          When getting into radio circuitry, it becomes very interesting.

      • We've already been at sub-100nM processes for years now, and things are already too "grainy" for real comfort.

        Yeah, this is the interesting part of the post, I think. We have already 10x this resolution on sale at NewEgg for $79. 10x ago was the "Pentium with MMX" processor, say 12 years or so ago.

        So, to keep on track we'll have to be placing individual atoms in a consumer product on sale for a half day's wages in 2022.

        Hmmm, I'd have never thought to bet against Moore's Law before.

    • While I would agree that this is a proof of concept rather than mass production, I don't think we need billions of quantum computers. Does every home PC need to be a quantum computer anyway? The value of a single quantum computer may offset the cost of building it one atom at a time.
    • by jnnnnn (1079877)

      They did it using an atomic force microscope (very sharp needle) to make room for the phosphorus atoms on the silicon surface (removing the hydrogen termination in certain places). Phosphene gas PH_3 then places phosphorus atoms in the vacated holes, and finally silicon is grown over the top using a low-temperature CVD process. It's a beautiful technique that took them several years to get right.

  • by Errol backfiring (1280012) on Wednesday May 26, 2010 @10:36AM (#32348476) Journal

    Just wait until you get an error message that says:

    * * * ATOM NOT PRESENT ERROR * * *

  • by flahwho (1243110) on Wednesday May 26, 2010 @10:37AM (#32348496)
    I once created a transistor with seven raisins. It didn't last long and I think Kelloggs stole the patent!

    --I forgot my sig.
    • I used them as 'nads to glue onto my Nazi Party action figures: Himmler, Goebbels, Hitler and Göring.

    • "'Spectroscopy of Few-Electron Single-Crystal Silicon Quantum Dots"
      It's a shame that the good Professor could not have applied the same logic to the title of his paper.
  • Interesting (Score:4, Interesting)

    by daveime (1253762) on Wednesday May 26, 2010 @11:01AM (#32348788)

    I thought that phosphorus was one of those elements that is never present in atomic form, it's so reactive it immediately oxidizes to form phosphorus compounds.

    Does this mean the 7 atom transistor has to remain in a vacuum ?

    • by irreverant (1544263) on Wednesday May 26, 2010 @11:08AM (#32348866)
      We move forward while we move back, if it needs to be in a vacuum then it would use a vacuum tube, while it's good for music it's bad for computers since we moved forward from these to transistors. I'm thoroughly confused now.
      • by mcgrew (92797) *

        if it needs to be in a vacuum then it would use a vacuum tube

        It would have to be in a tube to be a vaccuum tube, but it would still be a transistor. The way a vaccuum tube works is electricity heats a filiment (cathode), analogous to a transistor's emitter, which throws out electrons and photons. There is a mesh, analogous to a transistor's gate, that the current to be amplified is fed to which controls controls how much energy reaches the tube's anode. The anode is analogous to a transistor's collector.

        Eve

        • by ajrs (186276)

          if it needs to be in a vacuum then it would use a vacuum tube

          It would have to be in a tube to be a vaccuum tube, but it would still be a transistor. The way a vaccuum tube works is electricity heats a filiment (cathode), analogous to a transistor's emitter, which throws out electrons and photons. There is a mesh, analogous to a transistor's gate, that the current to be amplified is fed to which controls controls how much energy reaches the tube's anode. The anode is analogous to a transistor's collector.

          Even if this were inside a vaccume tube, it would still be a transistor, while an old-fashioned amplifier tube is not a transistor.

          So, were talking about a series of tubes then?

    • If only there was some way we could enclose the whole active device in some form of vacuum assembly made from, say, a stable material like glass? Maybe we could also create a temperature controlled environment for the unit to ensure consistent operation by including perhaps a heater?
    • Re:Interesting (Score:5, Interesting)

      by jo_ham (604554) <joham999&gmail,com> on Wednesday May 26, 2010 @11:30AM (#32349162)

      You can just keep it in an inert atmosphere or cover it in an inert insulator.

      You can store phosphorus under oil relatively easily. No need for a vacuum per se.

      • Or you could just store it in a silicon crystal. That aught to do the trick...

        • by jo_ham (604554)

          Silicon is a semiconductor itself - it may not be suitable if you pack millions of these little 7 atom transistors together, but who knows.

    • by MattskEE (925706)

      Current transistors use phosphorous, and it is not a problem. The phosphorous bonds to silicon atoms. But in the long term reactions will still happen. So all current chip dies have "die passivation", where the die is covered with something like silicon dioxide (glass), silicon nitride (ceramic), or maybe other things. It's a very effective hermetic seal.

  • That may or may not be a transistor. Probably both.
  • That's good and all. Unless someone comes up with a 6 atom transistor. Then you're in trouble, huh?

    • by dohzer (867770)

      Step into my semiconductor fabrication laboratory, 'cos you're f**king fired!

    • Re: (Score:2, Funny)

      by LHorstman (572584)
      Would someone tell me how this happened? We were the fucking vanguard of quantum transistors in this country. The University of New South Wales' Centre for Quantum Computer Technology (CQCT) Mach7 was the quantum transistor to own. Then the other guy came out with a seven-atom transistor. Were we scared? Hell, no. Because we hit back with a little thing called the Mach7Turbo. That's seven atoms and an aloe strip. For moisture. But you know what happened next? Shut up, I'm telling you what happened--the bast
  • Seven atom transistors, quantum cryptography, nuclear and weather modeling applications and journals called "Nature Nanotechnology"...

    Ladies and gentlemen, if this isn't the future then what is?

  • How does semiconductor doping works in these cases?

    In "big" transistors, silicon is doped with other elements, to a very high ratio silicon/other element (can't remember, but I recall something aroung 10E6 ratio)

    So when you have only seven atoms, what happens?? (ok, I guess this is already a problem with current tech, but still)

    • by vlm (69642)

      In "big" transistors, silicon is doped with other elements, to a very high ratio silicon/other element (can't remember, but I recall something aroung 10E6 ratio)

      So when you have only seven atoms, what happens?? (ok, I guess this is already a problem with current tech, but still)

      I'm guessing the doping ratio is somewhat higher, like 33% for the emitter and collector, and 66% for the base?

      That does seem a wee bit high. I'd like to see a 4 atom diode, while you're at it.

      • Re: (Score:3, Informative)

        by Bakkster (1529253)

        Doping really isn't relevent here, since we're not talking CMOS or FET transistors. While it's still a transistor operationally, the structure is completely different, so there is no p- or n-type material, per-se.

        What this is, is a quantum dot [wikipedia.org] which acts as a single electron transistor [wikipedia.org]. It's as different from a CMOS transistor as CMOS is from a vaccuum tube. So, asking for a doping ratio of a quantum dot transistor is like asking for the grid spacing of a CMOS, or the oxide thickness of a JFET: it doesn

    • by imgod2u (812837)

      The headline is (again) inaccurate. It's not a 7-atom transistor, it's a transistor with 7 phosphorus atoms (dopant) inserted into the silicon crystal structure by placing the atoms using a tunneling microscope.

  • Applications (Score:3, Insightful)

    by dohzer (867770) on Wednesday May 26, 2010 @11:19AM (#32349020) Homepage

    The seven-atom transistor has very hopeful implications for the future of quantum cryptography, nuclear and weather modeling, and other applications.

    Why not just say that it will lead to faster computers?

    • Why not just say that it will lead to faster computers?

      There's no grant money for faster computers.

    • Why not just say that it will lead to faster computers?

      In the 80's and 90's, whenever a new computer chip came out, they magazines always said they were aimed at CAD and :mumble: users. I think that just meant that they'd be expensive as hell upon release.

    • You think quantum computer=faster computer? OMG.

    • by geekoid (135745)

      Because maybe it doesn't?

      Faster does not equal more powerful, and more powerful doesn't equal faster.

      While there has been a marketing correlation in the lay world, it isn't a scientifically accurate one.

  • I'm always leery of new technology. I'll wait until the 7-atom transistor 2.0 is announced before I spring for one. They should have the bugs worked out by then.
  • ... Imagine what you could do with a Beowulf cluster of these!
  • The paper linked in the article is regarding advances in optical storage.

    Correct paper [nature.com]

  • the big question is: Can it run Linux?
  • The fact that a transistor with only 7 atoms is 4nm in size makes me think Moore's law is about to end. 4nm is only one order of magnitude less than what Intel is using in production now.
    • by imgod2u (812837)

      It's not just 7 atoms. It's 7 phosphorus atoms inserted into a silicon crystal that's 4nm. A silicon atom is roughly ~0.4nm in diameter.

    • by tippe (1136385)

      In general [wikipedia.org], it's taken the semiconductor industry 10-20 years to shrink a process by an order of magnitude (e.g. 1995=350nm, 2010=32nm). 11nm isn't really expected until 2022 (that's mass production, not just tech demos which are typically several years ahead). Presumably the 4nm process, which is perhaps still a couple of process sizes smaller than 11nm, would happen some time after that. There are still a lot of hurdles between producing a technology demo in a lab setting and the point where you've dev

    • by geekoid (135745)

      Look at th cost vs transistors per sqr mm. over the last 20 years. It has ended. It ended about 5 years ago.

  • Every time I try to solder one of these newfangled transistors to the breadboard, the drop of solder overflows, from the atom I am trying to connect over onto one of the other atoms, and the two atoms converge in a glob of molten solder causing a short circuit. I can never get the glob of solder off of the other atom after that.

Nobody said computers were going to be polite.

Working...