Transistors & The End of Moore's Law

The transistor is essentially an electrically actuated switch that allows or denies the passage of a current between two terminals, so these two points in this case would be something like this, they would be two contacts that we call source and drain, source and drain and Omega. made of silicon but modified to make it essential and metallic material, so what you are doing here is simply adding a very high concentration of impurities like phosphorus that have extra electrons, this basically has the consequence of giving many more electrons available and when the concentration is high enough, there are so many that this silicon effectively becomes like a metal and then these are like the two ends of the switch that can open or close properly, so the next thing you want to do is switch so that the switching action So you're going to have, let's say, some electrical contacts here and then you're going to want something that controls the switch, something that acts like you know that your hand, your finger, pushes the switch when it's turned on really hard, so this is done by adding an extra electrode to it. it will be made of metal or again very conductive silicon which is on top of this insulator and this is called gate now if you do nothing these two highly metallic silicon electrodes with the semiconductor silicon in the middle will not conduct any current it is like a switch. that's open, okay, there's no channel for current to pass through here, if instead you apply a very positive voltage here, let's say you have a one piece battery, let's say 2 volts, what will this do?

You will see that the positive is here and so on. you have a positive potential, what that will do is attract electrons and you have a metal with a positive potential. Electrons are negative charges, so they are attracted by the positive potential, so the electrons will start to accumulate under this insulating layer of let's say silicon oxide and when they do so at some point they form a conductive channel that connects the source and the drain, so now you have closed the switch, so you have gone from an open switch with the lights off to a closed switch with the lights on but without any mechanically moving parts all you have done is change the voltage on this electrode here in the old days people made structures like this with characteristic sizes on the order of microns now what people call the size of a transistor is this distance here is the distance between the source and the drain so this is the space minimum you need plus some space for these contacts to fit a transistor somewhere, so it's the most practical measure of how much space on a chip a transistor will take up, which tells you, given a chip of this size, how many

transistors

can you put?

In the old days, people started making these sized things, you know, several microns, you have fractions of a millimeter, essentially, Moore's Law tells you that this size is what Moore's Law does in its different incarnations. I think the most famous one is that the number of

transistors

per chip doubles every 18 months, which means that the size must be reduced accordingly to fit more transistors on the chip, and being an exponential law, it means that the size of the transistor is a lot shrinks potentially, so if you make a graph of the size versus the ear now we should look it up in a proper table, but I'm kind of making it up here, so let's say 1970 8090 2000 2010 and this would actually be a log of size, like this that you will. we have a line that looks like this and now we are in 2013 we are now at a 22 nanometer node which means this distance between the source and drain of this transistor is 22 nanometers now what to keep in mind is that 22 nanometers means there are about 50 silicon atoms here between the source and the drain, so we're really hitting atomic size, but not on some exotic lab device on every device you have on your computer and on your mobile phone.

How many transistors are legit now around a billion depending on what you buy? normal quad-core, say on a good computer with a quad-core processor, only 1-2 billion transistors, why is that number of transistors so important? Because essentially it's what gives you the computing power, that is, what determines how many operations are performed in parallel that your computer can do, that determines the size of a video that your iPhone can play, that determines how complicated it can be. do a calculation that your computer can do, whether to make proper calculations of things, to show you pictures, or to process words. or anything else you do with your computer, the size of this can go on forever and I keep falling, so essentially what determines whether the transistor works or not is whether you can stop the electrons effectively enough when these two electrodes meet come so close? each other, so essentially you have two electrodes, the two ends of the switch getting closer and closer and closer and you're still trying to find a way to block the current between them, at some point quantum mechanics becomes a problem. so in quantum mechanics we know that even if there is a potential bear year between two electrodes, electrons can still flow because of a quantum mechanical effect called quantum tunneling, essentially electrons can go through the wall, so to speak, but the The likelihood that they will do so depends on how high the barrier is, so the whole art of making this ultra-small transition transistor is essentially the problem of designing a potential barrier that is high enough yet still so thin that it stops the current, probably by the year 2025 you would get to the level where you literally have like three or four atoms and at that point it becomes very difficult to imagine that you can keep quantum mechanics out of the way, so in a sense the whole challenge for modern microelectronics o should call it nanoelectronics to keep Following Moore's Law and still having transistors that work just like the light switches you have in your room is to prevent quantum mechanics from starting to play a role because quantum mechanics would allow current to pass even though you are trying to stop it, so that's really the engineering problem is trying to keep quantum mechanics out of this, that's right, but your research is actually trying to put quantum mechanics into itself and what we're trying to do, in fact , it's not something that's a different way of applying Moore's Law that we're trying. build a completely different machine, don't you know, along the evolution path of today's computers it's just a completely different computing machine that actually uses the laws of quantum mechanics to give you exponentially more computing power, although in In reality you can only have a small number of bits or quantum bits as we call them computers: you have about 2 qubits to the 300 classic bits, which are as many particles as there are in the universe.