Professor Donald Fitzmaurice, University College Dublin
Part of the show The Science of Nanotechnology
Chris - Now we're going to find out about a very interesting application of nanotech from Donald Fitzmaurice, and that's powering the next generation of microprocessors potentially with DNA.
Donald - Indeed. Well maybe I'll just back up a bit and say why we want to do this kind of work. Every one of us has got used to going into our local computer store and buying a PC that's twice as powerful and half the cost compared to the last one we bought just a few years before. That's a huge achievement by the electronics industry where they're able to deliver improved performance at reduced cost year on year. That's governed by something called Moore's Law, where the number of transistors on a wafer doubles every 18 months.
Chris - In other words, the processing power. The number of transistors doing the computing power on the chip.
Donald - Right. And the reason that works is because if you want to do a computation on a computer, you have to move electrons around. If you want to shorten the amount of time you want to do a computation, then you have to bring everything closer together. The way you bring everything closer together is that you make it smaller and shrink it. Now that's been going on now for four decades since Gordon Moore first observed this. It's a huge achievement and it's really underpinned a revolution in the way we do everything in our lives nowadays in certain parts of the world. But for us to be able to continue increasing processing power at this rate, we're going to have to keep shrinking the bits that make up a computer. That presents us with two problems: one is that making stuff that small is really hard and very expensive, and secondly when you move electrons very close together, they start to know about each other.
Chris - Quantum effects.
Donald - Yes, you get quantum effects and also you get power-density effects. So the effect of current inside these things goes way up and the power-density inside a Pentium 4 processor is approaching something like the power density inside a small star.
Chris - Good grief! So quite literally, what you have is one tiny component here getting so close to the second component here that the two begin to interfere with each other and there's therefore a theoretical limit to how many things you can pack onto one computer chip.
Donald - Exactly. If one electron is in the wrong place and they're all repelling each other, you need a lot of energy to force them through. The amount of energy you're putting into one small space is so much that it's something like the same amount of energy you'd find in a star.
Chris - So does this mean that in the near future we're looking at a theoretical maximum processing power given current technology?
Donald - Well there's the so-called industry road map and this predicts the rate at which processing power will increase, or in other words, how fast the components of a computer will shrink. What they believe is that they can continue to halve this size of transistor every 18 months for about the next possibly ten years maximum. The industry will then hit what's called the red wall. They just don't know what's going to happen after that. So all the big semiconductor companies and lots of other types of companies are interested in developing different ways and different approaches to making electronic devices that will get us around these problems.
Chris - So what's the answer.
Donald - Well there's two things. One, you have to think of new ways of making them fundamentally. If you look around us you find that biology is very good at making lots of things that are every small and packing them very close together so that they all work without interfering with one another. The way biology does that is to start with atoms, assembles them into molecules, assembles those molecules into collections of molecules and into larger collections of molecules. You've got a hierarchy of function all integrated. Of course, you see this in every day of your life. You're having a shave in the morning and you cut your face. You come back 24 hours later and that cut is gone. Well millions of operations have taken place to remove the damaged cells, to clear up any infection, to build new cells and put them into exactly the right place. Now let's say that cut is on your finger. Not only does it repair the skin, but it puts exactly the same pattern of the fingerprint back. So it's extremely remarkable and there's a lot to be inspired by from nature.
Chris - So you're advocating borrowing from biology but translating from a cut on a face to a computer chip sounds challenging.
Donald - We're working closely with our colleagues in University College Dublin and a number of colleagues at the Centre for Research in Adaptive Nanostructures and Nanodevices at Trinity College Dublin and particularly now with the folks at Intel and various other companies. What we're trying to do is combine the best of both approaches. We're using conventional methods from the semiconductor industry to build really smart substrates that will help us organise molecules. Then we're taking lessons from nature and what we're doing is building molecules which organise on that substrate and assemble nanoparticles of metal and semiconductor and insulator into the right place and build the device as we want.
Chris - So literally you can use other molecules that know how to have a shape and a structure to move other particles into just the right place to do something useful.
Donald - Exactly.
Chris - So if you focus in on DNA for example, we know it forms a rather nice helix and if you have a certain sequence of DNA it forms other structures as well. I presume that's where you're going to go with this.
Donald - There are some very nice things about DNA as a potential molecule as a basis for this approach. One is that it has information intrinsically stored in it in the form of a sequence of base pairs. That's how biology stores information about it's own future and past. Secondly you can produce DNA in any sequence virtually at the touch of a button in a current laboratory. The whole process is automated. So it's not only an attractive molecule from an intrinsic point of view but it has the potential to be scaled industrially if you ever wanted to do that.
Chris - But is that actually do-able Donald? The idea of getting DNA to organise molecules in the right place to make an electrical circuit sounds like a great idea but also sounds pretty difficult.
Donald - It is.
Chris - Have you actually tried this yet?
Donald - We have tried it and indeed a number of other groups around the world have tried it and there is growing success in this approach. So much so that companies such as Intel and various other companies are actively exploring this approach and pushing a lot of resources into it.
Chris - So you've almost got a living chip then.
Donald - Exactly. Let me explain to you one or two of the things that have been done. I suppose the simplest experiment that has been done is you take a conventional electronic substrate which has somewhere where the electrons can come from and go to, and conventionally you might connect that up with a piece of metal. Now you can connect that up with a piece of DNA which is just the right length and swims through the solution and sits down on the surface because it recognises the place on the surface that it should sit. You then build nanopaticles of, say, gold or copper and they know how to recognise a certain part of the DNA and they sit down and lie between the two electrodes. Now you can pass electrons from one end to the other.
Chris - And it's much quicker because it's much smaller. But how many years of technology will this buy us?
Donald - There's two things: one is that in principle it should be possible to build very small wires but also, and this is more important, it's in principle possible to build lots of wires in parallel very inexpensively. There are two problems here. It's not only what you end up with, but it's how hard it is to make it. So this approach offers the possibility of building new things with new function that overcome the power-density and the interference problems, but it also offers a way of making things in a massively paralleled approach. I'll just give you an idea about that. Let's say you put molecule A and molecule B in a beaker. So you've got a beaker with molecule A and a beaker with molecule B, and you pour the two into another beaker. These come together to form molecule C. If you have just 100ml of each, you can be making something like 10 to the power of 27 new molecules in a few minutes.
Donald - Yes, so in principle, using this approach, you can make more transistors in a few minutes than have ever been made in the history of human kind before. Whereas ten years ago that was seen as very fanciful talk, and for good reason because the problems facing us were very significant, groups around the world have made enough progress to make this at least a possibility. I wouldn't for a second suggest we're there. There's a huge amount of work yet to be done and it's still a great challenge to turn this into a commercial scale technology, but it's no longer one that hasn't got a chance.