Sabine
patreon
This new semiconductor could revolutionize computing
🕑 Added 2024-01-09 15:45:00 +0000 UTC[This is a transcript with links to references.]
Researchers at the Georgia Institute for Technology have found a new semiconductor that’s a really good candidate for making computers faster and smaller than ever. Let’s have a look.
In 1965, Gordon Moore, one of the brains behind Intel, noticed that the number of transistors on a microchip was doubling roughly every two years. It’s become known as “Moore’s Law” and it has roughly remained valid until a year or two ago when Nvidia declared Moore’s law dead.
The problem is that by now the size of transistors, that are the elementary logic components of computers, are mere nanometres in size and it’s becoming increasingly difficult to squeeze more of them into smaller space.
You see the issue is that transistors are getting close to the size of single atoms. At that point quantum physics becomes important and that makes things much more complicated. Doesn’t make it impossible to shrink things down further, but you’d need to come up with something entirely new.
Sure enough, researchers are working on entirely new stuff such as spintronics or quantum dots etc. But these are still quite far in the future and from a manufacturing perspective they’re somewhat unappealing because they’d require many changes.
The path that most chip makers are pursuing for the moment is to stack transistors on top of each other. At present transistors are etched into flat silicon surfaces next to each other. But if you could stack them then you could get more computing power into small space. And that’d continue Moore’s law, if by other means than previously.
If you stack transistors, however, you get a new problem, which is that they just get too hot.One way you can try to address this issue is with some sort of cooling technology like tiny channels with liquids or such. But again, the issue is that from a manufacturing perspective this is unappealing because you can’t continue producing your microchips as previously.
The most convenient thing you could do to continue Moore’s law is to find a material for transistors that’s better at giving off heat than silicon but that can largely be used with the existing production technologies. And we know such a material, it’s graphene.
You have probably seen images of graphene. It’s a single layer of carbon atoms, set up in a honey/kəʊm/ pattern. The great thing about graphene is that it gives off heat enormously well. So it’s rather unsurprising that the idea of graphene transistors has been around ever since the material was discovered in 2004.
But while graphene is good at giving up heat and is very good at conducting electricity, it’s unfortunately not a semiconductor. You see, a semiconductor is a material to which you can apply a current to make it switch it from a non-conducting to a conducting state. For this you need what’s called a “band gap” in the electron bands. If you apply a sufficiently high current, that will propel electrons to the upper band and the semi-conductor will conduct. That’s why you can use them as logical elements. Because you can switch them on and off.
But graphene doesn’t have a good band gap. Graphene is considered a quasi-metal because it isn’t strictly a metal, it conducts electricity much like one. And this means it can’t fulfil the function of a transistor to act as a switch, at least not the way it is.
Scientists have come up with several ways to try and fix that. For example, they’ve used graphene rolled up to tubes called carbon nanotubes. These can be switched between conducting and nonconducting by twisting them, and that can be used to make transistors from them. But the issue is again that producing these nanotubes and using them to build microchips is too cumbersome for mass production.
The authors of the new paper now did something entirely different. They grew a layer of graphene on wafers of silicon carbide. Silicon carbide itself is a semiconductor and it’s one with a pretty big band gap. Basically they figured out that they can combine the silicon carbide with the graphene to get a stable material that both has a band gap and still gives off heat 10 times better than silicon. They have also very thoroughly documented exactly how they produced the material and have tested that it’s strong and stable enough to be used for transistors.
This is pretty exciting, honestly. After 20 years of graphene talk that didn’t amount to much, it could finally go somewhere. But it’s a long way from the laboratory to the shopping mall, and many nice ideas have died on that path. So I’m afraid we’ll have to wait a little longer until we can breathe transistors together with the microplastics.
Comments
Making smaller transistors, diodes and the rest of the microcircuit semiconductor flora and fauna has two problems that cannot be solved with ingenuity and the use of better materials alone: (1) Quantum tunneling, that mixes up the currents and reduces the capacity to calculate correctly with really small transistors, if packed together very closely in huge numbers. (2) Greater sensitivity to the impact of radiation, in particular high-energy particles in cosmic rays. The second of these two reasons has already caused some serious problems, when computing with integrated circuits consisting of many tiny transistors in critical applications, or less seriously so, but more commented about, when such so-called Single Event Upsets (SEUs) happen during computer games: Actually noticeable, but not really serious when happening on smartphones, but some really bad cases in other devices are listed here as well: https://www.sciencealert.com/rogue-cosmic-rays-from-outer-space-are-causing-havoc-with-our-smartphones Games: https://www.nintendolife.com/news/2021/09/random_a_rogue_space_particle_could_be_responsible_for_this_unsolved_super_mario_64_speedrun_glitch But there are other components in integrated circuits that are not transistors or other kinds of semiconducting switches, for example resistors, inductors and capacitors; minuterizing them a well is a whole other story: https://hackaday.com/2021/11/08/smaller-is-sometimes-better-why-electronic-components-are-so-tiny/
After reading this, I imagined people of the distant future unknowingly dealing with sand that has imprinted the transistors we were using on our devices...
