 We have seen already importance of technology in making realization of integrated circuits and now systems. The course really now starts, the way I have organized it is something like this. Since the starting material for all integrated circuit is silicon chips mostly we are working at and therefore we starting material is silicon. So we will see first how silicon is obtained, okay. Once we know how silicon can be achieved or best silicon or pure silicon is available in particular form of warm I want then I will say okay take this so called silicon wafers inside the clean room or inside the lab where actually fab is going to be done then we discuss therefore something about the conditions in the fab labs which are called clean rooms. We will then discuss something more about clean rooms then we will also see there that you know when the silicon comes packaged from elsewhere many times there it seems very clean, it is very polished surface and everything but it still has many other impurities particularly organic ones. So the first thing before any processing starts we have to clean the wafers and there is a standard procedure of cleaning and we will see what are those procedures and how silicon is ready for the actual processing. In most cases the first process is normally oxidation but maybe we will do something more about diffusion because some part of the diffusion characteristics are required in oxidation. So we will first start studying about the incorporation of impurities which is called diffusion of impurities in silicon followed by we will talk about oxidations variety of kinds and once we did oxidation we look for lithography, we look for metal other depositions, metal depositions, etchings all this will follow as if we are actually working inside the lab and we are trying to realize and then I will show you how many masks as we will say later are required to realize some kind of a CMOS chip or even a MOS chip typically you may require around as I said 400 steps, 4 to 450 steps and the masking may be as low as 16 mass process or can be as high as 34 or 36 mass process. So we will go into one by one, crystal per se is a solid material of 3 kinds, every solids are available in 3 kinds, one of course is the state is called amorphous, the second is called polycrystalline and third is called crystalline. The difference between the 3 is something like this, please remember all atoms in a solid are arranged in a particular manner and their arrangement itself decides the kind of material whether it is amorphous, polycrystalline or crystalline. I may start with the lowest one, crystalline first, crystalline materials are those whose atomic arrangement is such that they are periodic in nature, that is fixed, atomic arrangement repeats all sides, so this is called periodic structure for crystalline materials. So there is a periodicity, every lattice will work this word soon, every such smallest cell which is repeated is called lattice and there is a lattice periodicity which creates a solid. You can see from here, they are atoms in particular order are fixed and they are bonded exactly as they are nearest possible, bonding is possible. So this is called crystalline materials, so this is called orderly materials. If there is no order of atomic arrangement then we say such materials are amorphous in nature, okay, most things which you use many times other than metals are mostly amorphous in nature, most compounds are amorphous in nature, for example silicon dioxide which is the most important material for us in IC processing is generally available only in the amorphous style, however if I crystallize it, what is it called, yes, quads, in between also there is something but we will come to it later, okay. So quads, so quads are crystalline SiO2 whereas we are looking for right time amorphous material which is silicon dioxide as amorphous, this is used often, okay and in between this if there is a partial order that is the direction of lattice are different, there are number of such crystals in one area then we say it is polycrystalline, even this material is required in IC processing. So all three, please remember these though most of the time we talk about crystalline materials, the other two also play a lot of role in our fabrication, in particular if you look for solar cell these days we were working most of the time on crystalline silicon there to make a solar cell, the cost of making crystal is very high as we shall see today and to reduce the cost if the process which allows the films to be deposited which are amorphous in nature is much cheaper and solar cell people's major activity or major work is what they call as watts percent or sense per watt that is how many cent per watt is cost or how many watts per cent you can create is the feature, figure of merit for them, cheaper the one that is larger watts in smaller sense or larger sense, smaller sense in larger watts is essentially what is the feature of a good solar cell. Since the amorphous films are much cheaper to make and much cheaper in cost otherwise they may be the best candidate as far as the low cost solar cells are concerned. However if anything is good there will be something wrong, so they are very poor in efficiencies and therefore if you look for higher efficiency solar cell conversion of optical energy into electrical then the crystalline is the best and there is a micro crystalline also which is close to polycrystalline better than amorphous little worse than polycrystalline is called this and these also are these days work for reducing I mean increasing little cost but better efficiency. So if we look into the solar cell area later if time permitting this year I do not know then we will see that there also material choices are very crucial. Silicon ICs it persists starting material is always crystalline silicon and we like to see how that is created in actual way. As I just now said arrangement of atoms in solid is called arrangement of lattices one single arrangement which is the periodic in nature is called one lattice both semiconductors like silicon, germanium, 3-5 compound semiconductor like gallium arsenide, enium arsenide, enium phosphide, gallium phosphide these are compound semiconductors 3-5 materials as they are famously called they all show crystalline arrangement and normally they have what they call the kind of lattice which they show is cubic in nature. Of course in the cubic itself I will show you other lattices little later the simple cubic there is only one material which is popularly known there are maybe few I could only gather one, polonium which is more radioactive material is essentially has a simple cubic structure I will show you the lattice is little later. Then there is an arrangement called body centered cubic the materials which are known in this kind of lattices is tungsten, molybdenum, tantalum. Then the third possible arrangement which is face centered cubic copper, gold, silver, platinum and of course this zinc blend has one more possibility which is called diamond. Diamond essentially slight variation of zinc blend but it is classified in the same silicon lattice is diamond lattice whereas gallium arsenide is a zinc blend lattice. So what is the purpose of these lattices? Since these are the arrangement possible in atoms we are interested to know if particular kind of lattice exist and if I want to create a material of different kind for example for electrical conductivity we expect that the material should be either n type or p type that is it should have excess electrons or excess holes so that the current can actually transport. Now this idea that only semiconductors allow you these two bipolar transport is very important metals only show electron transport insulators do not show any transport if at all they show they will show electron transport but very difficult transport. So the materials which are used in everything in this interior circuit area essentially is either p type semiconductor or an n type semiconductor and therefore how the semiconductors can be doped that to make it p type n type is also part of this crystal truth. Initial waveform or initial substrate should be doped to a given value this given value and type of doping is very very crucial in actually making the device. So this fact has to be understood by you that why we are so much worried about this is the fact if you look at the electrical characteristics of mass transistor you can see the threshold voltage of a mass transistor is given by for n channel device v t n phi m s plus 2 phi f minus q ox by c ox I will name the term minus q bulk divided by c ox 2 phi f is 2 k t by q l n n e or n d by n i depends on what you have a type acceptor type or donor type q bulk essentially is q n a p x d x d is the depletion layer maybe max x d max is equal to twice k s epsilon not upon some phi f okay 2 phi f upon q n a or d so you can see from here if I want to control the threshold voltage of n channel device as shown here or p channel either of them then I must be able to have fixed value of acceptors or donor concentration in the substrate because if n a s or n d s are constant only then phi f is constant the bulk charges are constant and therefore to fix a v t I must have a process which will fix my initial concentration in the wafer okay. So one technique which will allow the substrate to have a known concentration is to date the crystal during its growth itself okay and the other we will go into other process called epitaxial growths but we will see this later so essentially I say do concentration of wafer is a very very important parameter many a times in deciding the electrical property of a transistors in the case of bipolar the base width and the base doping again decides the gain of a bipolar transistor so again the doping in the base in specific is very very crucial to design the beta of a transistor. So in some sense starting material resistivity or starting material donors or acceptor concentration is very crucial to us and therefore how do we dope the crystal okay so these are the issues which we like to solve in the coming time coming over okay so the lattices why I am worried about lattices because lattices has a minimum amount of volume in which so many atoms are residing now if I want to introduce an impurity there is no volume there okay so where this will go okay so there will be something where these so-called impurity atoms like phosphorus arsenic antimony blonde aluminum which are the dopants for p or n type n or p type can get in and how do they get in and they uniformly distribute everywhere because crystal will we have a large volumes or large areas in many cases and so we want to have uniform uniform doping everywhere that is our major worry how uniform are you really uniform so understanding the lattice has some advantage because then we know how much of the possible external atoms can come in without actually disturbing the periodicity of the lattice if crystal breaks or dissociates then it will go to polycrystalline or even amorphous the problem with amorphous semiconductor to crystalline is because in most electrical characteristics of any device if you see very carefully if it is a mass transistor the drain to source current is mu into Seahawks into voltages so there is a mobility term appearing if you look at bipolar transistors the diffusion coefficient of the carriers minority carriers in the base there the D is related to ability by Einstein relation say let us say DN so the mobility of the carriers essentially are going to decide the available current in the both bipolar and mass transistors and therefore we like to see the material should have a known mobility or at least controllable mobility okay and only then I can fix the current why I am interested so much in mobility control because if I am working at a circuit let us say what is that let us say a simple mass circuit I take typical mass transistor looks something like this and really giving you a cross section worth this is a n channel device okay this is gate oxide this is my gate this is my drain this crosses our metal connections sources this is bulk the current is actually electrons are moving under when I apply VDS apply VGS I create an inversion channel hopefully if VGS exceed threshold then source is grounded VDS is positive electrons move from source to drain because of the drift field there is an electric field in this direction and the electrons move towards drain okay now these electrons through the channel they are moving and I want this current to be known to me because I will decide if I have a capacitor as my load okay okay this is my resistor when I say I am want to a circuit to be faster I want charging to be faster and I want discharge to be faster obviously the current made available to you from the power supply and the one which is grounded down from the capacitor discharge both are important to include the speed of the circuit. So current control is our major aim so please do not think that this so-called metallurgy chemistry material science which I teach has no relevance because we are actually looking at my circuit which is what is driving me to do all this 2000 till maybe 90 most of the electrical engineers were also doing fabrication of last 20 years chemical engineers have taken away or taken our role though we are still the ones who probably manage them and tell them what we want and how should we should do but of late I figured out for example Intel's last chief was also a chemical engineer okay so it is not true that only electrical can do but the idea is that study of a circuit essentially forces me to study devices devices I want to may finally make from somewhere and I must know all technology to make it so that that so called circuit performance or a system performance is guaranteed is that clear I will like to make my technology independent of a designer as much as I try so that this designer should not actually ask me what did you do okay he said these are my parameters I gave you use them and design your chip but things are not becoming so simple we need to know what is happening in technology designers must know and what designers can do also technology must take care and therefore now understanding of technology for designers is as much relevant in 22 nanometer loaded below as much it was earlier in point five for one micron nodes technology people were told you do what you like tell us what is your limits designer will say I will use that don't worry okay now it is not so much true so the study of technology is still relevant now or rather it has become more relevant now which was earlier in 70s or 60s much more stronger separate disciplines okay so please remember that all of our thinking has something to do with this circuit and it's not just material which I am studying but as an electrical engineer I will like to know if I had to achieve this what should I make a choice for okay and that choice is something you must understand so many people get bored with this courses many times too much chemistry chemical and things goes on but they are part of the system you know and in a system you cannot say I will only use this part of the system system is a system okay okay so first thing let me first go to lattices how do you construct a crystal once you specify the lattice you can then hang a collection of atoms of each position in the lattice and that is how we can always create a crystal and this the lattice points are essentially like you know when you build a house or build a building there is scaffolding all around from which the building is created after after that so this is atomic structure which is bonding is essentially scaffolding and it has a translational symmetry that means you repeat and the crystal repeats generally there are only 14 ways in which atoms can be in solid can be arranged these are on shown here this is called triclinic when it's like a all three sides are not equal and their angles are also not equal this called triclinic then there is a monoclinic alpha is not 90 but the other two angles are fixed or beta gamma is 90 and alpha is not 90 with a slight variation in sizes and there is an atomic additional atoms at the center surface two surfaces then it is called monoclinic then there is a standard cubic structures which are slightly modified to look as orthorhombic none of the sides are equal but their angle can be 0 sorry 90 then none of the sides can be this but there can be an atom in the center we will come back this figure again and then can be this when you can have atoms at the surfaces and then there is another arrangement which is a not equal to b not equal to c but there are additional atoms connected and we will see this these structures are called orthorhombic these are available okay I just tell you this is the book which either either or some of you should jointly buy or Xerox it or don't Xerox is officially not allowed and officially I don't know just check up just check up if it is available in library I if there are fewer copies I may order some more otherwise those who are going to remain in the nano lab they keep saying that they want very high level research going on there they should at least buy this okay well then this may be a good bible for them okay so the book which I will suggest is of course this is second edition this is US book of this edition in addition maybe in the soft bound also I have no idea this was purchased for seven years ago so maybe third or fourth edition maybe they are not aware but I have this book I don't lend anyone please take from me I won't lend this book to anyone 250 dollars okay so don't come and just say so please either buy steel whatever you can do okay okay don't steal from me okay so this book is almost gives everything which I am talking not necessarily in the order not necessarily the way I am saying some things which I say is may by the way Dr. Professor Plummer who is the first author is the dean of engineering at Stanford University and he was the professor I mean he is still a professor in the e- department his friend is my friend or his colleague is my friend Krishna Saraswath he is also around same age as mine we both are from Billion Institute together he was junior to me and my son who was doing PhD at Sanford did a course with Plummer so I and also I made three times Jim's so I can tell you that the way he thinks he has been the strongest consultant technology in the Bay Area any problem finally comes they say pass on to Jim okay that he may solve it finally means some student of his will solve it okay there are many IITNs who did PhD with Plummer and Krishna Saraswath okay this is just to give an idea why I am saying you Plummer is one of the most famous person now here is something rhombohedral in which none of the angles are equal none of the sides are equal or sides may be equal but shape is slightly tilted then there will be tetragonal A and B A and C are not equal other three sides are like this then there can be an atom in the center which is essentially called tetragonal is called body centered so there are 14 possible ways in which atomic arrangement can be done and that is the basic thing we know about however as I said a lady there is the one which is the cubic lattice which I was just talking in this cubic lattice cubic means A B C are equal each angle for this is 90 degree so it is a cube the length of each side is A is called lattice age A is called lattice age each material has a different lattice age for example silicon has 5.43 am strong gallium arsenide is 5.68 am strong so depends on the material this lattice age will be different even for cubic or other other lattices then if you see this second one you can see from here there is a atom in the center of the lattice and each is then bonded to corners on every other possible this is called body centered and if there are atoms not at the body but every surface then there are six surfaces so each atom is there but very interesting thing you should remember since it is periodic in nature how many atoms this cell has a common cell one is correct why because it is a periodic there is a possibility of another lattice sitting left right top bottom so eight such corners can meet at a point so each corner will give you one eighth atom eight corners will give only one atom so ages of the lattice cubic lattice only contributes to one in body center there are two because four eight one from the corners and one in the center and the surface how many four because six means they share with the other ones so three plus one four however the one actual structure which we are interested in is silicon which is neither of them okay then why are we showing because from the face center we will create a structure which is silicon structure and it has number of atoms are eight okay so let us see how do we get to it. This is just to show you atomic arrangement of a crystal structure two different materials shown gallium arsenide for example okay this plumber slide that is why I said you all introduction this slide we had taken from him personally so all plumber slides due regards to him that he allowed me to have them only thing he said do not sell it to in a book I say fine I will not or here is an arrangement which is non periodic random and since you can see they are random there is no periodicity and therefore material is amorphous crystalline I forgot there the mobility of crystalline structure is the highest amorphous is the least now mobility why it is proportional to this atomic arrangement is the following whenever let us say electron is given by an atom like an entire material otherwise electron you apply electric field to a solid so electrons will try to go towards the positive polarity however there will be large number of electrons in the crystal okay and each will actually interact with this moving electrons okay this energy sharing is called relaxation or scattering so whenever electron starts moving it hits something and then it loses the path because once it this it may shift away there are many other space charts the number of scattering events happens and not all electrons in the amorphous are possible to reach to the positive terminal that means the mobility of average mobility of a carrier in amorphous material is very very low okay whereas in case of single crystal since they are all atomic arrangement if electron is moving between atoms they hardly scatter except the space charges it is of course therefore it is not infinite otherwise all of them would have gone some will not go even now and this scattering events reduces the possibility of all reaching there with a given velocity and this is essentially why crystalline material show highest mobility okay compared to polycrystalline and compared to amorphous of course is more like an insulator many times they may have a resistivity of the order of 10 to power 8 ohm centimeter semiconductors may have order of say 0.01 ohm centimeter to maybe as high as 10,000 ohm centimeter okay so this is amorphous here is the now there is a word what is it called packing you know this packing is a very interesting word which will be surprised to know but before we let me come to silicon first okay this is called a unit cell the smallest structure of atomic arrangement lattice as it is which allows you to repeat itself and create a crystal is called unit cell okay is it is called unit cell please remember this is unit cell but not a primitive cell now this word is very interesting this is a unit cell but not a primitive cell we like to see what is the primitive cell for silicon or for any other material this blue colored volume there are how many atoms 8 it is a cubic lattice so there are 8 atoms and you can see they can be repeated now all sides to create the large amount of large size of crystal there is also one possibility which is called hexagonal close packing some materials do show this for example diamond has similar structure hexagonal quartz has similar structure so hexagonal close pack structure is you can see on the top there are hexagon 6 edges and they are called hexagons now this why I draw this is to show you what is this word packing density the way packing density I said that let us say in the body center you can see from here the center one is the slightly enlarged version the actual lattice is shown in the right so we create a model which is called heart sphere model okay so each atom we blow it by size okay equal sizes if they are silicon they avoid all atoms are of equal size so we start increasing from the edges blowing the size of the atom whenever they will touch each other we say they are that is the distance between these two so you can see from here this sphere corner this is only shown half quarter of this but the upper and side part is not shown where the center one is fully shown okay so if I want to calculate the distance or the between the corner atoms and the so-called body centered atom how do I calculate this is the lattice edge they are at the center so it is diagonal from any atom okay so how will we calculate a by 4 square plus a by 4 square plus a by 4 square under root of this essentially is the distance between the two and since we are trying to make two atoms touch in this distance this body center and let us say this so what is the radius of each atom half of this that radius is called tetrahedral radius or maybe we will discuss so a by 4 to a by 2 and somewhere here so if I want this diagonal so if this is your body center and if I start increasing the die of this then I want to calculate this radius r isn't it this radius r since each is a by 2 a by 2 and since it is a diagonal so the tetrahedral radius is half of a by 4 sorry a by 2 square plus a by 2 square plus a by 2 square so essentially it is a square by 4 a square by 4 so root 3 a upon 8 is it okay so this is the tetrahedral radius of this so what is the volume of this sphere then 4 third pi r cube 4 third pi r cube is the volume is that okay in the body center how many atoms are there two atoms so what we do is number of atoms into volume divided by cell volume what is the cells cell volume a cube this is called packing density so we can say two times 4 third pi root 3 a by 8 this is not silicon a is only any any this this is not body centered is not silicon a by 2 a by 2 and a by 2 say diagonal okay calculation you can do that so the problem which I am saying is if I now calculate this by a cube you are right okay whatever I get a packing density is typically 78 percent is the packing density of a body centered that is if I do the same analysis for face centered how many atoms there four yeah sorry I think I said the I did not have the number so 52 percent for body centered face centered at 72 percent but if I do for silicon which is very interesting that comes out to be 34 percent and the number of atoms in silicon are 8 okay so which looks very funny that there are eight atoms in a lattice and the packing is only 34 percent which means the lattice is very loose impurities can easily come inside the silicon compared to any other material therefore platinum cannot be easily doped okay whereas silicon can be okay so the trick about calculation whether impurities can get into a material is decided by the impurity you are talking size of that impurity as well as how much available space for you to enter inside a lattice where it can still bond with the atoms and does not create any strain the most important word is it does not create any strain so without straining how many atoms can get in is essentially per CC if I calculate is called concentration so I am interested to know how much concentration I can have an impurities in a silicon crystal so that I can say okay this much atoms are impurity atoms are available and once I know if any RNDs are known to me then I can do all kinds of calculation for my device parameters and therefore my circuit performance so to essentially to know how much I am trying to understand what is happening actually then any times you may be right or I may be right anyway wafers do not no one may manufactures wafers in the world other than some few manufacturers okay maybe backer cam is one monocentro is one Japan is like is one because the process of making silicon wafer is very very costly typically silicon wafer plant from polysilicon to silicon is around 1 billion dollars typically and they cannot do more than 8 inch if you have more sizes you need even costly and you will make large number of wafers the users are only fixed okay so if every company has its own crystal growth system they will be just putting their use of their crystal blanks this 10% of the available capacity so that is a big loss in an industry anything which is idle including humans are thrown out okay so idle is the worry for everyone and therefore the capacity remaining idle also is not encouraged so just think of it why no one makes silicon as a material they get it buy it from somewhere at little literally higher cost because someone makes for you however they he can he or she company which are CEO she can he can supply you of any kind of doping any size of wafers with defect density as what will come as much as you specify okay this is something very very important okay so here is a face centered and here is okay at the before we go to the other packing density of our silicon I will come back there is something we need to know of the crystals the directions because if you look at if this weight is shown here this is the surface of the wafer and this direction is one zero zero of the crystal because the mobility of electrons in silicon is highest long one zero zero direction okay now as I said for a circuit I am looking for higher mobility and if I see one zero zero highest this so I must have a crystal which is called has a one zero zero plane and the direction is such that it is direction means if this is your plane this is your direction because you can create crystal like this so if this is your plane this is your direction so we want to create a known plane of the wafer which will use in the case of mass transistors as well as bipolar transistors normally are one one one oriented mass transistors are always or not not necessarily always now they are changing they are going for one one zero but most cases it was one zero zero for film fats one one zero plane may be better at least for some characteristics but for a normal mass transistors one zero zero planes are most preferred planes okay so we like to see why we are looking into this because I keep referring you back to circuit because that is where my ultimate performance is I am looking for circuit performance I am not interested in all this jumbo if that does not help me to improve what I am really looking for what are we looking for high speed circuit large density circuits okay low power dissipation since I have parameters in my mind if I look the technology back I will start looking if I do this where it hurts where can I improve here if I improve can I do that there someday we are trying to match best of technology to the best of requirement of designers or system users so please take it that this whole course was planned in 80s when I joined IIT Bombay was because I thought that most of the Indians including myself maybe we are very much shy of doing technology and also in 2014 now with so much of laptops and what iPads and pads and whatever it is the tendency is only to work on everything on cat tool everything can be seen okay so why do it okay unless you do it like for example other day I made a joke also you can do everything on cat tools computer tools design tools you can create almost a reality which is called virtual the only catch is can anyone inside a screen give me a cup of coffee which I can drink if that so that day I will say car is very good okay till that happens someone has to give me in hand something which is real that all other things remain virtual if you go to Japan 90% of their scientists or human professors actually work on technologies and while they are so of course they are not doing well recently but why they did so well because they are the best of technology why US does so well because they have the technology we do not have technology we always have cat tools infosys is our ideal I can fill up some data file I see I have to do something change the date I say why 2k and I say I know a lot of computer science gods forbid okay all said and done the course is therefore were written then that at least introduce our students to a hard job which is actually realization okay would say it now in 2010 or onwards or maybe 2008 onwards we had a good facility now to make devices circuits also no one makes circuits but at least devices at least study of materials technology is being worked out it is good to work on technology and see what is going on because I am revealing a mystery is the Nobel prize meaning were okay their directions of the crystals the way it can be shown in a cubic is the following you can see from here you can start from any point so ACB of the corners they say then you can either take X here Y here so this is your origin this is your origin then the X Y and the top one is Z each have a lattice age a okay so you can see if I have for each BCCFCC I have actually shown the same directions and then we define planes okay once you see XYZ you can at least make one lattice one cubic and you say XY sorry XY and Z but these are arbitrary you can call Z XY there is nothing very specific it is only a matter of my decision or people whose copy I am making their decision so once I decide this is my axis okay and this axis is easy because it is an orthogonal system so plane which is which has an like for example it is change in this case this is my origin so X is on this side Y on this side so what we define a plane is we find if there is an intercept on whichever axis okay that we will call one since this red colored one does not have any intercept on Y because for this plane Y remains constant Z remains constant but X is varying if I move like this X will be varying so I say it is 1 0 0 plane there is no intercept on Y and Z so 0 0 there is an intercept on X you can see here since there is a variation of this crystal will be pulled like this so this plane is moving therefore this has a variation in X so it has 1 0 0 plane if I want to make 1 1 then I must take a diagonal of that and put a plane across and then there is a Y and X both has intercepts whereas Z is constant so this is moving like this therefore it is 1 1 0 and if all three are varying then it is like a triangular shape plane 1 1 1 okay so as I said in most cases mass transistors use 1 1 0 plane wavers and as I said recently we are looking for 1 1 0 for particularly for FinFET or multi-FIN if it is now we are looking for maybe this may be a better idea okay so why I am looking for plane because plane has something to do with mobility and since my aim at the end of the day is to get highest mobility I will try to get of course this is not the only way I will improve mobility but this is one possible way at least starting should be good then worst is going to happen then I will again try to improve on that but before that worst happens start should be as good as possible now here is a silicon structure you can see now this we are not shown here is the silicon structure which is most important what is done in silicon is it is a phase centered lattice two phase centered lattice are interspersed pushed in okay 1 by 4 1 by 4 1 by 4 and 1 by 4 I repeat 2 FCC lattice you take it put aside get from one corner 1 by 4 1 by 4 and 1 by 4 so if you shift 1 by 4 1 by 4 the 1 age of lattice from there will enter the first FCC near the corner this for example this one all this red one these are the 4 which are 1 by 4 1 by 4 1 by 4 from the lattice age okay that is one of the atom so this 1 by 4 1 by so there are from 4 sides the other atoms will be there above one similar lattice will be above so the other combination will go above part of this will be done go in the top most portion this one has come from one lattice which has gone in this one has come from this lattice the other is coming from into the next lattice adjoining this so if you now see it is a by 4 a by 4 a by 4 inside okay so there are how many atoms now I said 4 from which are directly gone in 1 from the corner and 3 from the surface so there are 8 atoms in silicon lattice and if I multiply and now what is the radii radii now because it is half of please remember it is a by 4 is the distance and diagonal distance but actually radius will be half of that so you have this is a this is a by 4 only so a by 4 square plus a by 4 square plus a by 4 square under root of that then calculate number of atoms divided by a cube and you will get this number as 34% packing density but you see the lattice how much partly it looks okay now they are saying it is the most loosely packed lattice it has only packing density of 34% okay whereas FCC has 52 and BCC has even higher what does that mean that is a loose packed lattice so you cannot introduce impurities there where here limb seems to be a little funny but that is what the whole game is a silicon lattice allows because it has the lowest packing density more impurities to come inside and therefore it can dope itself to N kind or P kind or any other impurity can be introduced this is very very crucial in our decision to make any material useful for device okay before I go to the crystal growth there is another fault you know there are some interesting fault occurred during growth for example this lattice exchange this lattice was being grown and suddenly we figure out that one atom is missing for example between that the atom is missing missing atom is called vacancy missing atom is called vacancy please remember though it seems a fault as I keep saying it is a fault or it is a defect if there are no defects particularly vacancies there cannot be any doping at all is that correct so doping essentially is decided by available vacancies so even if during the growth of a material you create a defect which you really thought it is a defect it has actually helped us to incorporate impurities there is a possibility between the lattice there is a four portion there is additional space between the lattice in between atoms that is called interstitial between the space available to you there is a space where atom can go and sit so these atoms for example can be called interstitial so this is another defect which essentially is available further if you start moving from left to right suddenly you find there is an additional plane of atoms have appeared the rail line rail line is additional atomic plane has appeared the earlier ones let us say we have only seven atoms in height now there are eight there this additional atom you can see now the bonding is upper three are bonded lower four but the one where the additional plane has no bonding with the earlier left part of it and this is additional plane this essentially is called the area where it starts is called dislocations okay you are dislocated the material and the plane through which this dislocation is continuously seen is called stacking fault okay stacking fault can be one-dimensional three-dimensional okay and there is a possibility that this silicon atom may get some other atoms bonding there like oxygen in specific case it is called precipitation so the defects which during crystal growth one sees are the following one is the vacancy interstitial and then dislocation stacking fault precipitates we do not want precipitates we do not want stacking faults we do not want dislocations but we certainly want vacancies okay we certainly want vacancies so to create any device we need to dope and we need vacancies to be created so we will see how they can be generated in actual crystal growths okay this is a very interesting because these defects are material property during growth because you are going from high temperature to low temperature we will see this this will appear irrespective whether you want or you do not want you can make some correct game to minimize this number okay but cannot actually avoid it but certain techniques do around most of the dislocation to move towards the age okay outside the wafer or age of the wafer so the center of the wafer is very weird the age is very bad so we blind that out okay you say remove them out that is why there is someone asked other day circular wafer are always circular because age we want to age out remove the all dislocation from the ages okay so is that okay extra line of atom is stacking fault because of which it dislocates the material and there are possibilities of oxidations or carbon atoms may replace silicon which is called precipitates now these are essentially the problems in crystal growths and we will see now how crystals are built actually to get and how they are doped what kind of doping I am expecting uniform if I say ma the chip may have 200 chips wafer may have 200 chips each may have 10 million transistors for each of them I want this to be known exactly doping should be uniform because my designers have chosen NA as 10 to power 17 per CC and derived BT now if every part of the transistor is varying every chip to chip varying and a wafer to wafer start varying then there is no way we can say that the performance is guaranteed because VGS minus VT is the current VT is varying so is the current so is the speed so is the noise margin so the power so everything varies as soon as this is not attained and therefore control is the major crux in all technologies have you been through before we go to technology these are the three defects again shown here this is two-dimensional picture one can see sorry this atom is missing during growth so you say vacancy this atom has additionally said between the open space is called interstitial and this an atom was otherwise fine but moved from one place to the interstitial side and by doing so what it did it created a vacancy the pair of such interstitial and vacancy pair is called Frankel fault. Please remember this has not changed the lattice it has actually comments sat there in this case atom during growth itself was missing whereas here the lattice was fine however one atom due to during cooling somehow jumped okay so it created a pair of vacancy interstitials all of these will contribute to doping technologies as well as doping concentration availability and therefore these defects are very crucial to know about at a given temperature these numbers will change so as much doping will change as the available faults with you is that okay vacancy is missing atom interstitial is additional atom in the void and Frankel and atom moves from one to go to void area and leaves a vacancy down is called Frankel is that okay everyone so now we start with crystal growth silicon dioxide is popularly available in the earth as sand it is the second largest material on the crust of earth water is one second largest material on earth is sand so as far as your supply of silicon is concerned it is like a sun source there is an enough radiation for many many million years sand is always available okay you require very small amount of sand to create silicon but there is a enough number of sand enough amount of sand available so we start with a silo 2 and we want to create finally wafers so here is a quick path okay first thing we will that is the processing we will do first thing we will somehow make it from sand to a silicon which is called metallurgical silicon why this is called world has metallurgical grade silicon because in most steel technologies one of the bio product is silicon actually most steel technologies to harden the steel we add carbon and also silicon and so major metallurgical silicon is a bi-product which is actually not usable otherwise bias any industry is the actual available source so we need not actually need to have sand actually so much mg grade silicon is available from steel industry it can feed another 50 years for us and this will continue to go so much so we hardly have to do conversion from sand to mg grade but let us say if we I do not have silicon industry I have a steel industry then what so I have sand I can go to Mumbai one of the beaches collect sand so here is a sand to metallurgical silicon I treat with hydrochloric acid and we will see what and I convert it itself into a gas or liquid depends on kinds of treatment I do it is called chlorosilanes I may distill it by distillation I can improve the purity of particular silane I use will use show you this then I get the high purity trichlorosilane as shown here then I pass through a Siemens reactor which is called polycrystalline silicon this I reduce this silane trichlorosilane into silicon and it deposits by process called chemical vapor depositions then I actually put it into a some kind of a reactor which is called pure where I actually melt the poly pieces actually melt okay heating by 1400 12 and above once I start heating the silicon it melts then I put a single crystal on the top okay touching this liquid and start pulling that single crystal rod if you are done these days no one does experiment home so but just for the sake of it in earlier times we had in third fifth or fourth class an experiment to be done at home to go sugar crystals and sodium chloride crystals at home so school is to supply the so-called wire this is horse here you have to tie up with a good crystalline material which are make a solution dip it there and you say that will grow up next few days okay this was an experiment done in 50s by me and same technology is used in silicon okay nothing great once you have a lord of crystal which is highly pure then you actually saw it cut it into wafers polish it and these are polished wafers available for you okay which is what our technology is so today we start with these at least one or two of them will finish and we will show you that how do I go about from sand to the poly silicon sand is damn cheap very very cheap just to convert from here to here you may have a 1 billion dollar industry to put from here to here another billion dollars so to create a polish silicon wafers is a very huge costly business okay this polish silicon is very interesting material because this is most likely to be used by solar cell people because it is better than amorphous not as good as single crystal but cheaper okay compared to the actual wafers you get so the current technology for solar cells is essentially using polish silicon and of course their efficiency is not 24 percent or 28 percent they may have 12 percent to 14 percent efficiency amorphous as 3 percent 4 percent crystalline maybe 24 percent 28 percent even higher these days in some other materials silicon of course is not showing very big so we will see to first process we will go you can see from here this is a reactor in which this is a quartz reactor and with carbon liners inside okay these are carbon electrodes what is this furnace called it is called arc furnace okay when you apply electric voltage across the two electrodes depends on the in between gap it creates an arc arc temperature can be 3000 degree centigrade so you have a arc furnace okay you introduce quartz and carbon together from the top here due to this arcing between the two electrodes the temperature here is around 3000 degree centigrade it melts silica it melts silica 3032 is the actual melting temperature but start melting but we already introduced carbon along with it and here is a reaction which happens there so the final reaction is SiO2 plus 2 carbon is silicon mg grade plus 2 CO around 2000 degree centigrade somewhere on the top surface sorry lower surface is why lower surface the highest temperature is here actually where the arc starts okay so we get silicon grade plus carbon monoxide and that is taken out okay and you can see the liquid this actually comes out is collected into silicon metallic grade and it solidifies okay there are intermediate reaction for those who feel I might have suddenly wrote this so there are many reactions at around 1500 degree the SiO2 breaks into SiO bond and SiO plus carbon monoxide reacts with form SiO2 Si and then carbon dioxide then this SiO also reacts with carbon to form silicon carbide then silicon carbide and SiO2 themselves react to form silicon SiO CO and this is multiple reactions keep doing and around 2000 degree centigrade the final reaction SiO2 plus 2 C is Si mg plus CO is the final reaction there I mean some more in between but I thought too many will not you will feel bad how much chemistry so I thought okay this is enough so you can see using an arc reactor the first process was to get silicon out of silica so we got it now so once I get this material then since it is collected somewhere in the trough it when it solidifies you actually break into pieces as shown earlier okay this is silicon metallurgical grade this now I want to create from this I would like to convert into pure silicon metallurgical grade silicon has many impurities almost all kinds of including carbon is the highest but many other impurities transium is there nickel is there even the iron is there and their sulphates are there so many impurities so they are not a good silicon material I want to create highly pure why I am looking into highly pure silicon because I am going to dope this silicon wafer or silicon material by a given concentration of impurities typically silicon concentration is 5 into 10 to power 22 per cc for a given good silicon I want to dope it 5 into 10 to power 14 15 10 to power 16 per 60 17 per cc which means I am doping roughly one part per million or one few parts per million so if initially it says there are impurities and I introduce afterwards I do not know how many are I have okay so I want to make highly pure silicon to start with so that when I start introducing impurities the concentration which I want I want to achieve that because my circuit performance is heard on that so I am always worried that pure initially wafer should be highly pure as pure as possible and only dope with the impurities of my choice to a given concentration okay so this the first process to purify it is using a fluid based reactor essentially I did not bring the because a big figure when can see a tree you treat it with hydrochloric acid and it forms what is called as trichlorosilane and this temperature this is very low temperature process it is around 300 degree centigrade silicon react with the seal and you form chlorosilane now why I wrote only one of them depending on the temperature I choose I think some chemistry you may read there are two ways of reaction to happen one is called the thermodynamically controlled reactions the other is called kinetic based reactions in this case it is the thermodynamic based reactions which I am controlling so if I change my Gibbs free energy what is Gibbs free energy anyone thermodynamics I have to mean I could what is Gibbs free energy if you have done thermodynamics very first few lectures we introduced the term entropy which is order of which is the measure of disorder it is delta H which is called energy of formation or called enthalpy so H is enthalpy and S is entropy so H delta H minus T delta S is essentially your delta G change in Gibbs energy so at given temperature this both H and S T delta S can change and because of that delta G can change if it is positive reaction will be from left to right it is negative reaction will be from right to left that is what the thermodynamic reactions are okay. So by making a choice of temperature I can create varieties of chlorosilanes one of course is dichlorosilane Sih2Cl2 then I can create of course monochlorosilane also Sih3Cl I can create all chlorine atom replacing hydrogens these four atoms to silicon so I get silicon tetrachloride SihCl4 or everyone gets only hydrogen out of Sih and we get what is called as silane Sih4 so essentially we say silane if I replace one hydrogen atom it is monochlorosilane if I replace two hydrogen atoms this is dichlorosilane if I replace three hydrogen atoms I create trichlorosilane and if I replace all four hydrogen atom it is silicon tetrachloride silicon tetrachloride is a liquid all others are gaseous forms silane is very very very highly flammable material what does that mean this emulation has also changed in my time we used to call inflammable materials okay inflammable material which takes the fire fast or which fires itself fast but of late Oxford dictionary last 30 years said inflammable may actually contradict the flame therefore it is only called flammable English so the Sih4 is extremely flammable material as soon as you release in the air the 27 degree centigrade hydrogen will blast it out okay from silicon which is a silicon tetrachlorosilane we want to create polysilicone and please you know in between I forgot this trichlorosilane is done what is called as fractional distillations you have done some chemistry a condenser may liquid heat so every has a density so it collects a different this cooling and therefore you can have pure silicon trichlorosilane gas coming out at one end of the port so once that highly pure trichlorosilane gas is there we use another reactor which is essentially called Siemens reactor there are modification to Siemens reactor but process is similar okay so you can introduce trichlorosilane you can also introduce silane trichlorosilane anything but I have used the most famous present day technology trichlorosilane so what I do is that this is you know you can see here this black portion is a called silicon bridge internally there is an electrode which is a silicon core this is a electrode and this is silicon core this is polysilicon rod which is inside which we have a heating element now what we do is we connect this heating element to a power supply this is a inverter based system which gives pulses of power okay power electronics we like that then we have a reaction which is around 1000 degree centigrade silane trichlorosilane reacts with hydrogen which is also introduced to create silicon and that silicon deposits it is a and this is normally kept in a particular vapor pressure which is essentially better than vacuum so it is filled with nitrogen okay it is first evacuated and some nitrogen is filled and then only hydrogen reaction starts if you don't do this hydrogen may still blast it okay so there is some level up to which hydrogen should be allowed in these are all precautions which you actually take when in the lab as well so what happens this rod gets silicon depositions already available and keep this is called vapor phase depositions chemical vapor phase cvd as the word so the rod becomes thinner to raw wider okay and after sufficient amount of thickness which it can hold you stop the process take this out make them all this poly highly pure poly rods into small pieces which are called nuggets and these nuggets will then be used to create silane crystal silicon rod which is what we are shown that's on friday