碳纳米管计算机 | Nature Podcast

来源：Nature自然科研

又到了每周一次的 Nature Podcast 时间了！欢迎收听本周由Benjamin Thompson和 Shamini Bundell 带来的一周科学故事，本期播客片段讨论碳纳米管的计算机。欢迎前往iTunes或你喜欢的其他播客平台下载完整版，随时随地收听一周科研新鲜事。

音频文本：

Host: Shamini Bundell

For decades, computers have been built using silicon. But if we want them to continue getting smaller and more powerful, soon silicon might not be up to the job. So, what could replace it? Lizzie Gibney is here to check out a material of the future.

Interviewer: Lizzie Gibney

At the heart of a computer is its microprocessor – a chip filled with tiny, silicon switches called transistors. These create the computer’s 1s and 0s, and the smaller and more transistors you have, the better the computer’s speed and power. But if we shrink silicon too far, then the chips start to heat up and they become inefficient. To keep improving computers, scientists think we’ll need to replace silicon, and a promising alternative is carbon. Well, carbon nanotubes, to be precise.

Interviewee: Max Shulaker

So, a carbon nanotube, or a CNT, is just a rolled-up sheet of graphene, and graphene is an atomically thin sheet of carbon atoms.

Interviewer: Lizzie Gibney

That’s Max Shulaker from the Department of Electrical Engineering and Computer Science at MIT. His team are interested in carbon nanotubes because they have some pretty unusual properties, like conducting electricity at an incredible pace. CNTs are also semiconductors, which means that this conduction can be turned on or off. Both these properties mean CNTs could be perfect for making transistors. I called Max up to find out what his team have been doing.

Interviewee: Max Shulaker

Here at MIT, we’re trying to use them to replace silicon, which is what all the devices inside of your computer chips are made from, and it’s projected that if we could build a computer chip using carbon nanotubes as the core of these devices that switch inside of your computer, then we can make a computer over an order of magnitude more energy efficient than the computers you’re using today in your house.

Interviewer: Lizzie Gibney

So, because carbon nanotubes are such speedy conductors, I remember that years ago everyone first started getting very excited about the idea that we could use them in computers or in processors that are at the heart of computers, but then nothing really came of that. What were the big issues that scientists and engineers found with actually trying to make processors out of carbon nanotubes?

Interviewee: Max Shulaker

Yeah, so, once CNTs were discovered there was a huge amount of excitement, but the challenge was how can you actually build a billion with a ‘b’, for instance, working transistors which are all perfect and uniform etc., and actually yield a working computer chip?

Interviewer: Lizzie Gibney

And what is it about carbon nanotubes that make it so difficult to do that?

Interviewee: Max Shulaker

Yeah, so, for carbon nanotubes, there are really three major intrinsic challenges with a material. There are what we like to refer to as material defects, manufacturing defects and then intrinsic variability. So, on the material defect side, when we grow our carbon nanotubes, they unfortunately don’t all grow perfectly uniform, and it turns out that if you don’t get the precise, right combination of diameter and chirality, then some of these carbon nanotubes, instead of being a semiconducting material, they will instead be metallic, which means it’s basically just a nanorod of metal and you can never turn it off. So, that’s the material defect. On the manufacturing defect side, when we have all these nanotubes, in order to get them on to the wafer after we grow them, we actually disperse them in a solution and then we pour that solution over the wafer, and when the wafer dries, it leaves behind the nanotubes on the wafer. The challenge is that the nanotubes, while we want them to be perfectly isolated from one another, in the solution, sometimes they’ll actually bundle together. That’s like a manufacturing defect. And then the third challenge with CNTs is variability. So, when we build a computing system, we need to be able to build billions of identical devices across an entire wafer, and we need to be able to tune each of these devices as well. And in the past, we’ve only been able to build one type of transistor using carbon nanotubes. So, in this work, we’ve figured out how we can build many different types and tune all these different transistors across our wafer and make them very uniform as well.

Interviewer: Lizzie Gibney

So, it sounds like they’re some pretty big hurdles to overcome. How did you manage to do that?

Interviewee: Max Shulaker

What it required was innovations in processing, so how do we actually build these chips, new types of processing techniques etc., as well as new circuit design techniques. So, for instance, for the manufacturing defects, these bundles of nanotubes that are particles that end up on our wafer, we found a way that we could selectively wash off just these big bundles, without washing off the good, single, isolated carbon nanotubes on our wafer. And then for the material defects, for the metallic carbon nanotubes, we came up with this design technique where the technique allows us to actually design circuits in such a way that they are immune to any of these remaining metallic CNTs that slips through the processing and end up in our circuit.

Interviewer: Lizzie Gibney

So, using all those techniques, you’ve made this microprocessor. What kind of tasks or calculations can it actually do?

Interviewee: Max Shulaker

So, it’s a 16-bit machine running 32-bit instructions, so we can add or multiply or divide or take the exponent or square root of these 16-bit numbers.

Interviewer: Lizzie Gibney

So, it’s not just some little toy machine – it could do some pretty hefty calculations.

Interviewee: Max Shulaker

Oh yeah, definitely. You can programme this computer just like you would programme pretty much any computer today. So, this is like a microcontroller than you can pick up at a hobbyist shop to programme and control a small robot, for instance.

Interviewer: Lizzie Gibney

Wow, and so I’m guessing it’s probably the biggest processor made of carbon nanotubes that’s out there at the moment?

Interviewee: Max Shulaker

This is definitely the largest and most complex digital system fabricated from carbon nanotubes, and even more broadly than that, it’s the most complex computing system fabricated from any beyond-silicon emerging nanotechnology.

Interviewer: Lizzie Gibney

And the idea with going beyond silicon then, as we started out by saying, is that you want your system to be more energy efficient and maybe faster. Is this system there yet? Is it actually more efficient than silicon?

Interviewee: Max Shulaker

So, right now, what we’ve been able to show is that this technology can work. So, the next step in the process is to focus on performance. We’re not quite all the way there yet, but the progress is really becoming more rapid, and I think we’re quickly approaching now a time and soon you’ll be able to see computer chips made from technologies beyond silicon which can outperform computer chips just made from silicon today.

Interviewer: Lizzie Gibney

So, how easy do you think it will actually be then to transfer this into making a commercial product?

Interviewee: Max Shulaker

We’re actually already partnering very closely with industry in order to transfer carbon nanotubes into their manufacturing facilities so we can actually get these CNT chips out into the real world.

Interviewer: Lizzie Gibney

So, how long do you think it will be before my mobile phone, say, has carbon nanotubes in it?

Interviewee: Max Shulaker

I think if you’d asked me that question a couple of years ago, the answer would be we’re not sure. I think now, given the progress that we’ve made in this work, that it won’t be too long. We’re talking under five years until you’ll begin to see chips that are still built with silicon but also have some carbon nanotube circuits built right on top of them, for instance. So, we’re not talking about science fiction anymore. It’s a matter of when and hopefully not just if.

Host: Shamini Bundell

That was Max Shulaker of the Massachusetts Institute of Technology in the US. You can find his paper over at nature.com, along with a News and Views article.ⓝ

Nature Podcast每周为您带来科学世界的全球新闻故事，覆盖众多科研领域，重点讲述Nature期刊上激动人心的研究故事。我们将话筒递给研究背后的科学家，呈现来自Nature记者和编辑的深度分析。在2017年，来自中国的收听和下载超过50万次，居全球第二。

↓↓iPhone用户长按二维码进入iTunes订阅

 ↓↓安卓用户长按二维码进入推荐平台acast订阅

点击“阅读原文”访问Nature官网收听完整版播客