How Microchips Are Made

during the show is nothing in embarrasson nothing. Yeah, let's turn this episode off with a little oh listener mail. This listener mail comes from Jacob a k. A. Booger, and Jacob gave himself that name. I'm catching up to the present day podcasts and them on USB versus FireWire, and I still haven't heard you guys mentioned a m D yet. Please talk about them since I think they're better than Intel in more than a few ways. Thanks Jacob Booger, Well Booger, we thought we'd talked a little

bit about kind of microchips in general and how they're made. Um, we don't tend to to break things down to company by company. But of course Intel on a m D are both known for their microchip architecture. Yeah they uh. For the uninitiated, these two are pretty much the big guys in the computer segments like the personal computer. Intel an m D are wrestling back and forth with one another on a day to day basis. Now, there are lots and lots and lots and lots of other companies

who make microprocessors of different kinds. Right, Let's let's be fair. Yes, when you say wrestling, Intel's kind of the enormous sumo wrestler and a m D is like the luchador. Yes, but that doesn't necessarily mean that in this case the luchador is you know, a lesser. Oh no, no no, no, I'm just saying by market share, okay, yes, by market share, yes,

until has the the lion's share of the market. Um. But there there are many many other microprocessor manufacturers out there, um, you know, some of whom are limiting their their work to one or you know, one or two different products. You can find microprocessors and just about everything anything that that's electronic these days, because we've come to roll eye on them. Yeah, we also had another listener asked us about microchips, and I would like to apologize to that

listener because I could not find your email and or tweet. Um, I know that you sent it to me. And so we wanted and this listener wanted to know what microchips were, and you know, what they did and how they were made. And so we were going to kind of talk about a little bit of of what they do, but not a whole lot because we've talked about it before. But uh, a microchip. You know, when you hear that term, you might think, well, what the heck is that. Technically it's

an integrated circuit. Yes, that's what a microchip is. And so if you remember when we talked about the basics of electronics and electronic theory, a circuit is a pathway for electrons to flow through, Yes, and of course the flow of electrons we know better as electricity. At least that's the current definition. Oh my gosh, I totally forgot that we went down that road last time. Alright, So, yes, a circuit is a pathway for electricity. Integrated circuit is

a special kind of circuit. Now, let's let's I guess you want to are we gonna take the way back machine here, or you want to just talk about the past. Oh, we could just talk about the past, all right. So I don't think anybody gassed up the way back, Yeah, I think, well, right, yeah, I I noticed the last time I was pulling the cord, it wasn't really it wasn't really starting up the way you would expect it to. So anyway, back in the day, the day being a

long time ago. Uh, circuits were much larger than they are today. Um, they you could actually see each of the discrete elements pretty clearly because it was all on a much larger scale. And we're going all the way back to vacuum tubes here talking about So vacuum tubes acted the way transistors do. Now, so let's let's talk a little bit about what exactly they did. The whole purpose of it was to control the flow of electricity.

It would be to allow electricity to flow through that part of the circuit, and you could even use it to amplify the electricity if you needed to. Okay, you're just staring at me now. I wasn't planning on talking about vacuum tubes, so I didn't at any rate, Uh, The problem with vacuum tubes is they were really big and they produced a lot of heat. They do produce a lot of heat. I can tell you that right now. I have a a beaten up old amplifier that uses

vacuum tubes. And not only do they produce a lot of heat, they're really heavy and unwieldy because they take up a lot of space. Yes, so those are all problems, right, I mean, you've got it, produces a lot of heat, takes up a lot of space, and they're really heavy. So the early computers were these enormous machines that generated uh so much heat that it was hard to be in the same room as them. Um, of course, it's hard to be in the same room anyway, because they

pretty much took up an entire room. Well, but it this way. You're not going to see a vacuum tube powered iPod anytime soon because they take up a lot of space, right, So you can't put that in your pocket, Yeah, no,

you can't that. And and in fact, the development of the semiconductor was what led to you know, transistor radios, and that's when we started getting a lot of portable electronics because it was possible to to throw a portable radio in your pocket instead of you know, having to deal with the one that was plugged in you know, at home, on on the desk, right, the one that was almost the same size as the desk it was

on the Yeah. So, so vacuum tubes were limiting us quite a bit, uh because of their size, because they would also burn out, and if one burnt out, that meant that your whole system was no longer working and you would have to replace the burnt out tube. So we needed to find something that was smaller, generated less heat, was more reliable, and that ended up being the transistor. Now, uh, that was invented back in It would still be quite a while before we got transistors small enough to put

them on like a circuit board. The earliest transistors were actually pretty large. Um. And then even then, you're still talking about a bunch of discrete elements that you wired together to create a circuit. Yeah right, So yeah, I mean, if you've done this, you might have a physics classes you've done where you've created an electric circuit this way, where you know, you're using wires to connect various elements together and then you hook it up to a battery.

So if you haven't, you should because it's a lot of fun. It's a good learning experience. But that's an aside, right, so the sorry, yeah, I know, now I'm throwing off down like okay with notes notes, So let's talk about the complex circuits here. Um, there was an issue called the tyranny of numbers. The tyranny of numbers? Have you heard about this? Wasn't that a mystery novel? Uh? No? Not? Well, I mean I'm gonna say no, it could be true. I'm gonna write it if there wasn't one. Right, now,

here's the here's the definition. Advanced circuits contained so many components and connections they were virtually impossible to build. This problem was known as the tyranny of number burrs. So that, in other words, we get to a point where we can build circuits, right, but in order to build them at the complexity that we need, the tools we have are two crude, right, using it, building it by hand. It's just there's no way you can have the level of precision you need in order to build a circuit

that's small. What could possibly be the solution to the tyranny of numbers problem? Let's see I've mentioned it before already, the integrated circuit. Just say the integrated circuit. Okay, the integrated circuit. Very good. So in nineteen fifty eight, you've got Jack Kilby. Yes, and Kilby was working at Texas Instruments, one of those aforementioned companies that makes lots and lots of different kinds of microprocessors. That's correct. And Kilby at

the time was brand new to the company. You know, he'd only been working there for a little while and had not earned any vacation yet. So there was a point where pretty much everyone went on vacation except for Kilby, and he was left there pondering the tyranny of numbers problem,

and he came up with this idea. He thought, well, what if you were to build an entire circuit out of one piece of material, and you would essentially carve out the different elements that you would need out of that material, overlay it with some metal to be the connectors, and then you would have an entire circuit printed more or less on one single piece and you wouldn't have to build the individual elements. That was the idea behind

the integrated circuit. I apologize, so Essentially he found a way to make the whole thing fit in a much more compact space. Right, and by using this method of carving away from a single elements a single chip, really you could make the different individual um parts of the circuit much much smaller than you could before. Um. Which is good because it has led to lots and lots

of very very small devices. Well this is this is yeah, this is what eventually led up to Gordon Moore taking a look at the number of transistors uh that engineers were able to fit onto a single one inch diameter chip and say, look every at the time, I think it was every twelve months, they were doubling, although of course today we talked about it being every two years.

Um Gordon, So Moore's law has slowed down over time, but it's still you're still talking about doubling over a set time limit, which is pretty amazing because you know, we're in the we're in the billions now. So um so how well, first of all, we I guess we should talk about the material they use to build these chips. Its silicon. Yes, it's not just silicon, it's extremely pure silicon YEP. As a matter of fact, when I was doing research on this. I found a video by a

different part of our company. The Science Channel did a piece on silicon and it's it's um really fascinating what they do because they have to to melt down lumps of the material to to get you know, basically to reform it in the shape that they need for microprocessors. But it has to be extremely pure, so they have to clear out the chamber chamber with argonne gas to make sure that there's no air in there at all, melted down and then uh, they create essentially what looks

like a giant silicon pencil. Yeah, it's a big cylinder cylinder of silicon, and then they put the cylinder through Um, well, they use they use a very fine wire to cut wafers. Yes, it's water thin, Yes it is. As a matter of fact, it's it's just a few millimeters thick each each layer, right and and um, so there they cut these these slices off and each slice becomes its own the owner, its own silicon wafer that you used to imprint uh, circuitry onto and you could end up using one wafer

to produce dozens and dozens of chips. Yes, it's a matter of fact, every once in a while you'll see a picture in the news of some dignitary visiting a computer plant and they'll hold up a what looks like a giant disk and you're going, wait a minute, that's way too big to fit in my computer. Well, yes, this is. This is a cross section of that giant cylinder imprinted with in general and one the ones I've seen. When they show where they're holding it up there, you

can see that there's something etched on on the silicon. Well, those are all the different individual chips or what would become chips if they were you know, uncontaminated by whoever it is holding it up right at the at the end of the process, once everything's been printed, you you chop that wafer into the various, uh, actual individual chips. Shop I mean you actually use a very fine saw. It's a diamond saw as a matter of fact, that they used to cut the uh, the slice of silicon

up into lots and lots of little chips. But we've we've left out stuff all right, So well, the process is known as photolithography, and it's been in use for several years several by several, I mean many, it's actually kind of a it's really a neat neat idea. All right, so let's let's start with you've got your your silicon um also, well, well we'll go ahead and say where the stuff takes place, because we're talking about elements on these chips that are just a few nnometers in width.

So any kind of impurity, any mode of dust that got on this thing is going to ruin it. A mode of dust would be enormous compared to one of the transistors on these chips. Exactly. Yeah, so you can't have any sort of of dust in there. Well, think about your environment for a second. Uh, just the dust in the air alone in your environment right now, unless you're in a clean room, is gonna be way too much to ever try and produce. Use a microchip, ye,

so at least one that will function. Right, And we're humans. We give off lots of lots of dust. Dead skin cells have tons of dust every day if you look at the entire human population, literally tons of dust. So the these clean rooms can't have that. They don't have the luxury of being able to have that sort of dust flying around. So in order to prevent it, they have massive air conditioning systems that circulate the air in

these these clean rooms. These clean rooms can be enormous, by the way, like the size of a warehouse, but in most of them, the air conditioning system is so powerful that it can completely circulate all the air in that room under ten minutes. It's pretty impressive. And uh, this is where Intel's famous bunny suit campaign came from. Back in the sadly, not the kind of bunny suit

I was thinking. No, no, the bunny and the bunny suits are actually they they look like they're in some sort of high tech firefighting gear because their suits that they wear where they have hoods over their heads with a visor in them, and this basically keeps dust out of the air by you know if you see in the humans. But yeah, it keeps it inside the suit.

The humans then they can't shed inside the clean rooms. Yeah, and I remember in that video you were referencing, they actually talk about how the air in these rooms is cleaner than the air you'll find in hospitals. That doesn't surprise because of the way, the speed and the filters that they use. So we've got this this uh environment that is it's not dust free because you're never gonna get that unless you're in a complete vacuum. Um. But it's about as limited as you possibly can be and

still be on on earth. Uh. Now, let's talk about the actual process of photolithography. Yeah, but as we get into this, you're gonna see that we have improved on Dr Kilby's process. I'm assuming he is a doctor. I didn't see that part on the engineer at any rate. Yeah, on Kilby's process for creating microprocessor micro microchips. Because this is a very very sophisticated way of doing it. Right.

So you start with your your pure silicon uh wafer, Yes, that you have sliced off of the the whole uh cylinder. Then the next thing you need to do is you need to uh create a mask. Now, the mask is and this is not in all forms of lithography, but we'll get into that. The mask is essentially a pattern. Yes, right, it's the pattern. It's what the chip is supposed to

look like at the end. If you if you've ever worked in photography and uh with film, not with with the with the digital camera, and you are trying to make part of the picture a little darker when you're developing it, you would hold up something basically to block the light from getting to that part of the uh, to the photosensitive paper. Well that's basically what the mask is is blocking light from a very high energy ultra violet source which is going to shine onto the silicon wafer. Uh,

and the mask is blocking that. Now why would we do this, Well, it's because the silicon is not just silicon at that point. They have also overlaid on top of that a piece of uh, photosensitive film right right, So you've got this, You put the photos intoitive film on top of the wafer, and then you've got this uh,

this mask that ah that ultraviolet light shines through. When the ultraviolet light contacts the wafer, like, there's gonna be some parts that the mask blocks in, some parts that the mask lets through, right right, So the parts where the mask let's through the light, the light hits the wafer. And when you when you are finished with this first process and you wash the film uh away, the it takes anything that that the light has contacted gets essentially

carved away. Well, basically what happens is that the high energy ultraviolet light causes that film to break up, you know it, Uh, the film doesn't react to the light very well. So that's the point. They wash it with water and that washes away the broken away bits, but you still have that protective film on the places where the light didn't touch because of the mask. Well right, right, yeah, I worded it incorrectly. Well no, no, no, I was

just clarifying. And then um, they once they rinse that away, they bake it. Then they use a process called etching, in which they use chemicals to dissolve the remaining protective film. Uh to Basically, you know, now you have just the etched wafer of silicon, right, so are film there and and also and also the stuff that was no longer covered by the film which had been you know, the film got zapped away wherever the light touched it. Yes,

that gets etched away. Yes, so that's where you've actually carved away the the material. What's left standing is the stuff that was not touched by the light. That's just one round. You may have to do this dozens of times before you have actually carved out all the elements on your your your circuit. So well, there's another part of the process to the doping part, in which they

change the electrical properties of the silicon. Right, doping means that you are purposefully introducing impurities into the material in order to change it's it's um well, the way it what either conducts or insulates. I mean, that's that's what a semiconductor is. Under certain circumstances that can conduct electricity and others it does not, right, and where it can conduct some electric electricity, but a not at a rapid rate as rapid as it was maybe at just certain temperatures.

So again, this all depends upon the impurities that you introduce into it. So at the base you have to have the pure system so that it doesn't conduct electricity at all. Otherwise you do you have a chip that's not gonna work. Again, the ultimate goal here is to be able to direct the flow of electrons in a very specific way, and of course if your entire chip does that, If it doesn't, it's it's not specific at

that point, and so you'd have a broken chip. So um, they, as Jonathan pointed out a moment ago, they continue to do this layer by layer by layers, so that um, again you're taking up less space because you have layers of material on top of one another, which you know there it's no longer and no longer requires it to take up so much physical space because it's all basically

sandwiched on top of one another. But then there's the the metallization process, right, so you've carved away everything that doesn't need to be there. It's kind of like a sculptor. Yeah, it didn't carve everything away that doesn't look like a circuit. Very nice, it's essentially that's what how it works. I'm on board with that. But again it's on an incredibly tiny level. And the reason it can be so tiny, you know, we've talked about before that the nano scales

so small that you cannot view it through a light microscope. Yes, well that's because they're not using the way they can get at a small as. They're not using visible light. Again, they're using ultra violet light. Yes, and in fact, Intel uses um is using a process called extreme ultra violet lithography in their latest processors, which are the ones that have like the thirty two nnometer transistors, which is crazy. I figured they were building microchips sliding downhill on a

street luge and that's not extreme, not that extreme. Yeah, it's not gonna be in the X games. But so the motalization, this is where you've built all these elements by carving away the stuff that doesn't belong, but you still have to connect them together, right, you know, this

is normally where you would be building wires. Well, the way they build wires and uh in these transistor or in these uh these circuits, it's actually kind of similar to the process we just described, except what they do is they put a layer of metal UM on top of the wafer and then again you put the this uh, this UV sensitive photo resist on top of that metal,

and you have another mask. This mask is going to block out all the places that where hiring actually blocks out the places where you do want wiring, right, so you want the places you want protected. Yeah, so it'll block it'll block light. Wherever it blocks light, that's where a wire is going to be. Wherever light comes through, that's gonna carve away this metal. So again you go through another process where you run it through the system.

The light ends up hitting the certain the designated areas on the wafer UM and that ends up going to that's gonna end up breaking down the metal so that you are left with just the wires that you want you've carved away everything you don't want again, you may have to do this process several times because in order to get all the connections you need, Um, you may need to do several layers on wires up to I think five I think was the last I had read, right,

But um, it's again very similar to the lithography process of actually carving out the chip itself. Yep, and uh see there. This is all very cool and it has led to a number of you know, all kinds of different chips that are getting smaller and smaller and allowing Moore's law to continue more or less, except you know,

for the laws of physics, which makes things different difficults. Well, I mean it's already made things difficult because remember when I mentioned the extreme ultra violet lithography, Yes, I do ever mentioned I remember when you mentioned that, because it was just like three minutes ago or something, so here was here was a barrier they that Intel had to find a way. And I know that I'm probably ticking off our our listener who wanted to hear about a

m D and not Intel. But um, Intel had a problem when they were going to switch to EUV and it's it's just one of those fundamental things. The wavelength that they are using, the eu V wavelength is at thirteen point foreign denomenas. Okay, at that size, that is so small that practically every material absorbs it. So if you were to shine it on anything, it's just gonna

get absorbed. Doesn't necessarily work if you if you need it for lithography, they found that if they used the eu V in a vacuum and they used reflective surfaces instead of lenses to focus and direct this light, they could then use it in lithography. And there's also another form of lithography we can mention, which is the electron

beam lithography. Yes, now, in this case, if you were to use electron beam lithography, you can get very very very precise and build these incredibly tiny structures um and you don't need a mask. You can just put it through a computer program and the computer program will direct the beam properly so that you don't necessarily you don't

have to build the mask for the light to shine through. However, it's a much slower process than photolithography, so it's not really considered a viable alternative for the mass market right now. But you were talking about reaching the the physical limits of of size, which doesn't have anything to do with necessarily our manufacturing process. It has to do with very fundamental physical laws on the quantum level. Right, at a certain point those divider has been between the uh the

pathways that electricity uh travel down. Once they get to a certain point, the electrons are going to be able to pass through them, and it's going to cause a serious problem. Yeah, it's called electron tunneling, and uh it's a really to read. To read a description of electron tunneling is really really bizarre because imagine that you come up to a wall. All right, You're standing on one side of the wall, you lean against the wall, and then suddenly you're on the other side of the wall.

You didn't didn't necessarily pass through the wall. You just started on one side and you ended up on the other. That's kind of like electron tunneling. It doesn't actually move through the material. It's just it's so small. It the materials so thin, the electron access it acts like there's nothing there. It might as well not have been there in the first place. Right. So that's the problem is if you if you build these components UM too small,

then the electrons are no longer controllable. And again, just like in the situation where I said, if the entire chip were to conduct electricity, it would not work the same sort of thing. Yeah, you can't direct the electrons if they can pass through everything, you know, So uh, in that case, we will have to look at alternatives to the transistor. UM. And there are lots of engineers

working on this problem. Uh. You sometimes sometimes they find out by switching to a different material they can actually get a little smaller than they expected, which until found out when they switched from one kind of metal that they were using in their their chips to a different kind. But eventually we are going to hit that limit, and it's not going to be that long from now. Um. But that doesn't mean that we won't find a new way to work around the problem. I'm I'm sure it

won't be long before we don't know something anyway. So that's how microchips are made. We kind of talked about what they did, uh, And you know, there are several different companies that produced these. Like Pilette was saying, it's

not just for computers, it's for all sorts of electronics. Yeah, it's it's funny because, um, you know, for a company with this long and story to pass is Texas Instruments has, for example, Um, you know that you just don't they don't make chips for computers that end up on people's desktops at least not anymore. Um. But yeah, a m D UH is one of those companies that has had a hard time getting a toe hold in the consumer

desktop market. Um, partially due to you know, just the fact that they're they're you know, trying to get out the eight hundred town guerrilla or maybe eight hundred ton guerrilla in case of Intel, because Intel was one of the very very first UH chip manufacturers to make UH microprocessors for home and UH work computers. Yeah, they also made some very shrewd partnerships with various companies, which actually has has gained them a little trouble in the court

systems for anti competitive behaviors. Yes, that's true, but that's a totally different story. Maybe one what maybe what we should do is put down on a podcast list at some point the story of Intel versus a m D. That would be fun, that would be that would be a good way to put it. That would be less technical and more kind of political. But it's it's an interesting story. But I have something else we can talk about very quickly. It's a little listener mail. This listener

mail comes from Chris Now. Chris says, just by the way, it was directed not just to tech stuff, but the stuff you should know. You smell, and you started this war. Why would you ever be so unwise as to start a war with tech stuff. This is a war that you cannot win. This is a war you will not win. We expect your surrender to be handed to our leader, Jonathan Strickland in the form of a sticky note with the words you win written in black sharpie. We expect

your surrender soon. And then there's some French which I'm not gonna attempt to say because you pop off on say. But this is from Chris Now. Chris. It comes down to to this, Um, we're gonna We're gonn just shoot straight here. We like Josh and Chuck. They're nice guys. We even like stuff you should know. I listened to it.

I'm a fan. They do good stuff. Occasionally they pull us on um right, So so our rivalry is mostly funny and the only reason I'm saying this now is I don't want it to spiral out of control because just the other day I noticed that Chuck was not his normal chip herself. And it turns out he's worried

that people don't like him. I mean, earlier today I found him weeping in the corners, sucking his own thumb and someone with I know, well, I mean it's probably he probably wouldn't want me to share that, but his psyche is weak. People. I'm just saying they're not made of the stern stuff that Plat and I are made of. You know, we are like teflon. It just slides right off us. But Josh and Chuck they're like, well, they're like Teddy Bears, emotionally vulnerable teddy Bears. Yes, it's true.

Thanks for writing, Chris. If any of you would like to write us, our address is tech Stuff at how stuff works dot com. And we have a cool fan page on Facebook now, so if you're on Facebook, do a search for tech Stuff join our fan page because uh, we're trying to build up a nice, strong community, have more interaction with our fans. We always like that. And

we also have our own Twitter feed. Yep, we decided to UH to have a text stuff podcast twitter feed now, so instead of just following Jonathan and I individually, you can follow us together right. The handle, by the way is text Stuff h s W, So just go to Twitter dot com slash tech Stuff h s W and you'll find us there. You can also find UH both Facebook pages and Twitter pages for the other podcasts on

How Stuff Works. So if you've been putting off tuning into some of the others because you just haven't had a reason, now you got a reason. They're right there. People, Yeah, check them out and Chris and I will talk to

you again really soon. If you're a tech stuff and be sure to check us out on Twitter Tech Stuff hs WSR handle, and you can also find us on Facebook at Facebook dot com slash tech Stuff h s W. For more on this and thousands of other topics, visit how stuff Works dot com and be sure to check out the new tech stuff blog now on the House Stuff Works homepage. Brought to you by the reinvented two thousand twelve camera. It's ready, are you