Nanotechnology - The New Science of Small || 03 - From Micro to Nano Scaling in a Digital World
[Music] thank you let's talk about nanotechnology through the lens of the electronics revolution of the computer Revolution why does that sensible thing to do why does that make sense well we've talked from a really basic point of view about electrons flowing and controlling one another I'll make the case that electronics and computers are really about electrons controlling other electrons and of course our capacity to design integrated circuits chips that control program and take advantage of that I'll also show in this first of two lectures on electronics and computers that the March forward for electronics since its Advent in the 1950s and 60s has been taking us from kind of the macroscopic to the micrometer to the millionth of the meter and then in the second lecture I'll show that the March that Electronics is marching along is now well into the nanometer and so when you talk about Electronics today you're talking about nanotechnology there's nanotechnology inside your computer and in your cell phone uh and in your digital camera and then the final point is that the technologies that have enabled us to get down to this nanoscale they were the ones that took us first to the micro scale they were the basis for carving matter on the scale first of the micrometer and then the nanometer so these are platform Technologies for what we call top-down nanotechnology they are the basis for our ability to manipulate matter the way we need to for nanotechnology so first in this lecture we'll speak about going from the macro world around us to the scale of the micrometer and how that relates to building computers first though let's just look back at some old pictures that I think many of us have seen about computers like the eniac which should for the electronic numerical integrator and computer eniac was the first general purpose electronic computer does that mean general purpose well it meant that it could be reprogrammed so it wasn't something that you know was only a calculator only did a specific set of arithmetic operations you could use it to solve a whole bunch of different problems now it was actually designed for a fairly specific purpose it was injure it was in World War II and the United States Army needed to be able to rapidly calculate how to launch artillery uh this was for its ballistics research laboratory and so in 1943 in a secret agreement it started off with University of Pennsylvania's electrical engineering school on building this general purpose computer building this eniac if you adjust the dollar figures to today's values cost about six million dollars and I think impressively it kept going in continuous operation from 1947 to 1955 so while getting on for 10 years how many computers do you have that you've continuously used for eight or ten years without them going Obsolete and you upgrading to another one so I think that I think that's uh impressive that the eniac had such a long useful life now uh I mentioned this idea general purpose computer but the way the eniac was programmed wasn't the way we think of programming today it wasn't with software it was actually with specific plugs and electrical connections and wires connecting with each other and that idea of a reprogrammable via a stored program which is software idea actually was exploited a little bit after eniac and I think that's kind of interesting to think back that there was a day there was a time when we didn't think of kind of the world of computing is having Hardware on top of which software ran but that itself was a new idea so how do computers like eniac work how did the original ones work well the the founding feature that's comediniacin is common to what's in your computer and your cell phone today is as I was saying what we call Electronics which really is electrons controlling other electrons uh why is that a basis for computation or analysis or Computing or arithmetic well if you think of one simple component and I'll start with the vacuum tube but we'll move on to transistors start with a vacuum tube if you have a current that can flow between two points between two electrodes and if you can use a third either a current or a voltage but some kind of other electrode to steer that current around to turn it on and off well then you have the basis for digital Computing where the world is filled with ones and zeros and with ones and zeros we can accurately represent information over huge swaths of dynamic range and if you can control that current flow not just with zeros and ones not just to complete on and off but if you can control it to intermediate phases then you have the basis for what we call analog or kind of continuous Electronics where you can vary these different levels well that second phenomenon the analog Electronics if you can if I can change the control signal just a bit and I can get a large change in the current well then I have amplification and so the history of computing is really intimately tied to the original discoveries like lead to Forest work which allowed electronic amplification because then you can take a radio signal as was done then and you can amplify it so you can pick week signals just out of the air and then turn them into an audio signal in this case that you can listen to and so this field of the analog the field of the digital these are the sort of the two primordial bases of electronics but in a way they're sort of using the same building blocks these vacuum tubes initially which then evolved into the transistors which we'll we'll talk about in a sec uh they one of them uses them in this sort of continuously varying mode and the other is the on and the off okay so let's go to the vacuum tubes I think I think these these pictures of vacuum tubes are familiar when we were looking at eniac we saw all of these this is sort of panels with periodic Arrangements of things and those are the vacuum tubes it kind of looks a bit like a light bulb and it has something in common with the light bulb which is that it inside it is a vacuum now for this controlling of the flow of electrons why do we need a vacuum well the answer is that if we didn't have a vacuum if we had air then electrons trying to flow in the form of a current would basically run into the molecules that make up the air they Collide and so we'd lose our current we need to apply vastly More Voltage than otherwise in order to get a sufficient current to flow and so inside the vacuum tube we have the perfect environment for the free flow of electrons and so we just heat up in a vacuum tube one of the electrodes and enough energy is now available from thermal energy to result in the emission of an electron the phenomenon is called thermionic emission the therm being the temperature of the thermal uh and then with the aid of an electrical field these electrons can then really scream through this vacuum without any impediment because of the lack of anything else present now that's that's just a start that's not uh something that has electrons controlling electrons yet that's just establishing the current where the control comes in is with this third electrode which can control independently whether that current flows think of it as being kind of in the perpendicular Direction I will turn on or off the flow of that current using a third electrode conceivably without having to really invest much electrical current or very much electrical power at all that's the key to amplification is to have just a touch of a modulation on the control side have a big impact on the flow of current so we talked about how the vacuum was necessary it led to the free flow of electrons but you can easily picture and you already know the problems with a computer based on kind of a bunch of light bulbs I mean they're all going to be burning out and if they're thousands of light bulbs or thousands of vacuum tubes making up the computer at any given time the things can be down as a result of one of the vacuum tubes that's gone off and that's exactly what happened and that was that was exactly one of the problems so really the field of transistor Electronics of semiconductor Electronics was born of the desire to take advantage of the best properties of the vacuum tube which is its free flow of electrons and its enablement of control over that flow but get Beyond this reliability issue get beyond the size issue scale things down so this picture is showing one of the very first transistors and here the electronic current didn't flow through a vacuum it flowed within the semiconductor and that's the first function of the semiconductor let's let's think about a crystal of silicon it's this perfect array a perfect pure array of atoms uh they're completely ordered they're completely periodic and now think of of the electron we spoke about how it's actually a wave we don't usually think of it that way we we picture it sometimes as a pointer as a particle but it's actually a wave and So within a semiconductor we can have an idea of what's called commensurability and this means that the wave of the electron can follow perfectly the wave of silicon atoms you can have a silicon atom a silicon atom and another silicon atom and you can have commensurable flow of this electron wave it can be delocalized as we were speaking about with ways it can be ubiquitous it can be everywhere and so now without having to establish this perfect vacuum and make this large tube that has these reliability issues we're able through the Perfection through the order through the structure of semiconductors through their purity through their crystallinity we're able to achieve the same everywhere-ness in the electrons so that's this key element that we took the the learning from the vacuum tube field and that we managed to translate into the solid state so when we talk about solid-state Electronics we're talking about going from this evacuated phase the vacuum and we're talking about translating the good stuff from that into a solid state device that we can then also start to make many of and we can make them a lot smaller now before we talk about making them a lot smaller there's something crucial and it was a crucial Discovery in the field of electronics uh and it was it was only when we figured out how to manage the interfaces of these materials so the connection between say a piece of silicon or a piece of germanium is a lot of the early work was done in the connection between that and the outside world which of course we refer to as the surface or the interface here and not all the Silicon atoms are happy right we talked about how in diamond or in Silicon the carbon or the Silicon atoms respectively C4 nearest Neighbors well that's perfectly true right inside the bulk of the semiconductor but at the interface one of these silicon atoms could see nothingness above it and so from a chemical standpoint unless we take special steps to manage the interface silicon is not perfect right at its interface nor is any semiconductor now silicon it turns out has this very special property which is that when you take a interface of silicon with the rest of the world and you just let it oxidize you let it form sio2 which is glass and you do it at the right temperatures and you do it in the right humidity you can form this oxide we call it the native oxide because it's exactly what silicon forms when you introduce oxygen or water on top of it you can form this oxide that forms a very clean interface and so now these silicon atoms that are mostly satisfied in the bonds but they have this one Bond hanging up that Bond can be satisfied through the presence through the formation of this native oxide that's also really easy to make that's very convenient because oxidation we just need to cook things you know in the right environment and we grow this native oxide on the surface of the Silicon so what we've talked about here and what's in this picture is a single transistor of course that's not a basis for computing this discrete transistor we need to connect it together at least with some other transistors there's going to be wiring involved right there's going to be points of connection involved well okay let's connect together a couple of transistors here's what an undergrad electrical engineer still builds today it's called a breadboard and it's got a bunch of these individual transistors each one of these blue things black things is a transistor it's typically with three terminals and this breadboard allows for a relatively convenient connection you can put resistors in as well so you know every electrical engineer and I'm I'm an undergraduate electrical engineer I remember suffering through this in third year in our Electronics course and we build these circuits by actually jabbing these little discrete transistor elements into this board and then putting wires and resistors and connecting them all together how many of these can you put together until you go crazy 10 20 30 maybe 40 if you're lucky yeah most of the time you've got a wire in the wrong place and it doesn't work you've got to debug your circuit uh and so it's what we now would say with the insights of the integrated circuit Revolution it's what we now call not a scalable solution it's not something where uh if you come up with a good idea to make a circuit based on a manageable 30 transistors and then the next day you say wow if I could take two of those circuits and combine them or four or 16 of them combine them I could make a parallel computer that could do 16 times more computations well it's 16 times more work uh and so it's not conveniently or in a cost-effective way scalable technology and so that's where the integrated circuit comes in we use the word monolithic integration and the monolithic you know it makes sense it's one rock where we're making one piece of semiconductor now and people found out in the 50s and the 60s how they could carve many transistors into the same Rock and the rock is a crystalline piece of silicon it's a large piece of silicon it's a large substrate and what we needed was as well as the ability to make many of these transistors typically they're actually essentially all the same as each other we're just making many copies many repeats the same thing it's like you're taking a potato carving an image out of it and then just repeating it many times we then also have to connect them and so what we needed was a path to build many transistors and then systematically connect them and of course reliably connect them with each other and that wiring on a two-dimensional substrate that's fairly readily done once we first Define how we put these transistors down where we locate them relative to each other then we'll connect them with a bunch of wires so for example and and this and we do this all the time you can imagine having a mask that we call a shadow mask where you cut holes in it and you allow a metal you heat a metal up and you allow it to evaporate and it only penetrates through the holes and it doesn't penetrate where it's obscured and so you can form interconnect you can form layers of connectivity you can make wires on your two-dimensional chip and your wires can now connect together all of your transistors let's take a little bit uh take a little bit of a look here at how we do this I mean that that shadow mask technology uh it's a start and actually we do it you know in my own Lab at University when we're trying to do course Connections in many of them it's it's a great way to do things and you can go and carve out that mask yourself but it's not going to get you down even to the micrometer length scale so the technology that we use to get down to the micrometer and and somewhat Beyond is called photolithography and in photolithography instead of just taking that mask and sticking right up against your wafer uh you have a mask whose purpose is to be optically transparent in certain regions and to be opaque in the other regions and then you project light through a lens and you project it down onto the wafer onto the substrate the Silicon that you're going to go build your transistors on and you essentially just put an image of your mask onto that that wafer now what does that light have to do with anything well on that wafer at every stage is something we call the photoresist and it's really just like photographic film it's something where you expose this material and if it sees light then when you put it later in a developer solution like old-fashioned uh pre-digital photography it washes away but if it wasn't exposed to light it goes in the same developer but it doesn't develop it hasn't had this photocatalytic reaction occur it will stay intact and so now what you've done is you've formed a template on your silicon wafer that allows you to do quite a number of different possible things so one of the things you can do if you want to make wires is you evaporate that metal on here and then you subsequently wash away the remaining photoresist and where the metal may direct contact with semiconductor it sticks and where it was just sitting on this soluble removable photoresist you lift it off what if you want to build a transistor well the building blocks of transistor construction involve taking these silicon crystals and putting controlled levels of impurity very low concentrations of impurities in there so you can use this mask that you've made and you can introduce impurities onto the surface of your silicon and if they're in direct contact with the Silicon surface they'll diffuse in through that surface into the silicon and where you have a mask that blocks their contact their their interaction with the interface they don't and so you can selectively alter the properties of your silicon from on top so this lithography process that is done using photons today using light is the basis for being able to make incredibly scalable integrated circuits and one of the inevitable needs of the electronic sector is you you know we're we're hungry as consumers we're as soon as we get something we want something even better and so there's this desire to scale to make things more integrated to make integrated circuits that are faster that use less power that are more sophisticated more complex that can process a huge image a huge and complex image and can process it so fast that you can't even tell when you're interacting with your computer that billions of operations had to happen well how do we do that and and keep things keep size under control it involves making the transistor smaller and smaller with every generation uh and to make them smaller and smaller you can you can see where this is going if we're using photons which are waves which have extent uh then eventually we're going to sort of run out of steam there because as we said before we're not going to be able to confine that focal point of light onto something smaller than about the wavelength of of the light that we're working with and so the March forward of lithography is also the march to shorter and shorter wavelengths it's gone from being indivisible to in the ultraviolet to deeper into the ultraviolet and there's even work on x-ray lithography today and so here our understanding of waves and wavelengths and the ability to focus them uh is is key to understanding the technological March of photolithography as we move to shorter and shorter wavelengths to more and more energetic particles as we do so so how do we build these circuits in a way that's reliable I mean let's let's think about the number of transistors that we have to put down today it's in the billions in one of these integrated circuits well perhaps you've seen a picture of somebody working in a clean room as they're called obviously the purpose here is to minimize dust and particles because a little piece of dust that lands on one of our transistors in one region of this big wafer containing many integrated circuits and all of a sudden that integrated circuit is a chance of not working so in clean rooms the pictures you will have seen have people wearing what we call Bunny suits the technical term and they're they're covered head to toe you can't see who it is or if it's man or a woman they have little slippers on that are disposable slippers and and these allow you know all the little bits of dust little bacteria little viral particles people are very dirty they allow us to minimize the extent to which these particles get released into the room and potentially onto your wafer in fact the latest generations of clean room Technologies tried to just minimize the extent of there's people in there at all try to do everything as much by a robotic operation of moving these Wafers around where everything can be just kept in this completely pure and clean environment you know when I for my own PhD I used to go into a clean room at Cornell it was called Cornell Nano fabrication facility and uh and and there and and at every clean room you know if you wanted to write your notes or write in a lab book you'd have a special lab book or a special notebook a kind of paper that doesn't Slough off little bits of dust but that's allowed for use in a clean room because it's essentially dust free paper that's how sensitive we are to these kinds of issues now given this complexity given that we're talking about these rooms with Incredible environmental control with control over uh temperatures uh with control over humidity uh these are very costly things to build in fact today to build a clean room in which we can build the latest generations of integrated circuits costs three to four billion dollars so the number of these places where we build integrated circuits is actually very small now and what has emerged in the last couple of decades is what people call the foundry model so there are companies that make all their own branded integrated circuits but there are also places where anybody can come in and and you know for a price typically charged per wafer you can build using an established set of Technologies using these kinds of photo lithography you can choose your metal layers you can choose how you diffuse dopants into a semiconductor you can build using these available recipes and you can and send them your circuit designs and they can actually build your circuits for you so you don't end up spending the three or four billion dollars as a company or as a startup company in order to build your circuits but instead you just pay for the use of that facility people call this model The Fabulous model and by that they they don't mean that it's not fabulous they mean that you don't have to have your own Fab your own fabrication facility but instead you can leverage existing infrastructure so where are these integrated circuits I mean we know we know that they're inside our computers that's kind of the canonical example but these integrated circuits are in all sorts of places uh there so for example the digital camera that you use it has an integrated circuit where the Silicon is being used to absorb the light uh there are processors these days that are devoted entirely to the processing of Graphics in fact the processors these days have become so complex and so sophisticated that the amount of heat being generated energy being generated on them is so large uh that they're now starting to become segmented where instead of having a single processor there are what are called multiple cores where there are different regions of processors that talk to one another to some degree but most of their time they're off in parallel working on different computational problems and then kind of sharing the information um you know there's other sort of more surprising or more unusual things that you do with integrated circuits so your cell phone transmitter is now made with an integrated circuit that's more on the analog side of the world where uh just like in the first radios you're trying to transmit electromagnetic waves or receive and then amplify very sensitively electromagnetic waves uh this whole Trend uh this whole uh incredible race to the bottom as as Richard Feynman called it this race towards the smaller and smaller it was described in what has become known as Moore's Law and some people object to the use of the word law to describe Moore's Law uh the the the law here is that it's an empirical law it's an observation it's a description that Gordon Moore made Gordon one of the founders of Intel made uh in about 1965 and he described that from 1958 to 1965 that about every two years there had been a doubling of what you could put onto an integrated circuit of a given size so you could put twice as many transistors you could make a circuit a computer that was twice as sophisticated uh twice as complex every two years and typically you could do it for about the same price um this is a great this is a great recipe and this this is the uh it's that doubling every two years an exponential growth law that can be used to describe how the integrated circuit revolution has happened and we've gone from there was discrete elements where thinking about putting tens of transistors onto a breadboard is already difficult to being able to put billions of transistors onto a chip becomes possible and it's possible to do it for costs that are either measured in the few dollars or measured in the tens of dollars per chip it all is directly traceable to this scalability of integrated circuit Technologies so to summarize really what we've discussed in today's lecture it's kind of the precursor to the Nano Revolution but in fact we really through Moore's Law we we face right up against the nanometer now because we do have the capacity to use shorter and shorter wavelengths of light through photolithography to make our transistor smaller and smaller to pack so many of them on that we are invariably and inevitably at the cusp of the nanoscale in fact the integrated circuit you have in your computer today consists of transistors that are made using technologies that can access below 100 nanometers so they're well into the nanometer length scale and so the big question for the remaining half of this discussion of electronics and next lectures will be the following what happens when we start to build our transistors on the scale of the nanometer I argued in the discussion of the physics of the nanoscale that exciting things happened uh that this was an opportunity to engineer phenomena like Quantum mechanic phenomena on the other hand when you think about Gordon Moore's observation even if it's just an empirical law there's an impression of continuity there right there's the sense that we will be able to scale we'll be able to gradually transition to smaller and smaller geometries build greater and greater complexity and there's no brick wall there there isn't a change in the rules but that we're just extrapolating on the existing rules and so in the next lecture what we'll talk about is this tension on the one hand the fact that to extrapolate to smaller and smaller geometries to continue this tremendous Legacy of Moore's law that we're going to be doing so within a new physical regime the regime where we hit up against Quantum effects but at the same time there will be opportunity there will be these new physical effects things like electron tunneling things like really seeing the size of electronic waves not being able to ignore their waves anymore and there are people who are working at the nanometer scale in nanoelectronics to try to leverage those phenomena for new paradigms in Computing
2023-02-11 11:28
