 finished the silicon wafers to get from company but how are they manufactured and their properties. Today we start with the next something about where we actually fabricate the chips and these are called IC fab labs and there are a lot of constraints one gets in fabrication of ICs. A few of the things which are listed here, semiconductor manufacturing is constrained by requirement of very, very, very clean rooms. The environment has to be very, very clean and I will give you some examples if they are not what can happen. Then in that we just clean it, as I said should be super clean and such an environment you have to create to make an IC fab. And today we will talk about the clean rooms and then whatever goes inside the clean room we will talk and then something about before we prepared our next processes, wafers need to be cleaned and there is a standard procedure of cleaning and finally we will also talk of time permitting about gate rings. The word gator as I say will come later but let us wait for that. So I repeat I am looking for very high class of clean area where as we shall see what are the properties of clean rooms, how do we manufacture certain things, some things we have to be away from the normal routine in the case of when you are in clean room. For example typically in a clean room no more than 2 to 3 people are allowed to be in at a given time. The reason of course is we will see later something related to particle motion due to Brownian system. So let us talk further if you are noted down. This is not exactly given in either any of the books. Yes partly it is given as discussed but not everything. I have designed first clean room of India way back in 83 to 86 or 85. So I was aware in those days what is the requirement of a clean room and even to get those items in India was very difficult. So anyway somehow we made our first clean room in 85 and of course I do not say it was first in India by me. Some sense the industry had, BL has some fab lab. So a semiconductor complex has started in Chandigarh but there were industries. No university boosted any IC lab till that time and we were the first to actually create IC fab area. So something which I have learnt over the years may be of interest. So what is problem in a normal room? Normal room has lot many contaminants and they may be classified as particles, surface contaminants and molecular contaminants. So there are three kinds of contaminants are seen in a area in any volume of space and as I say if we want to fab a chip then we need to somehow control all of them. There are some list of contaminations given. Particle contamination is major source are the people working in the IC lab. I just now said. So if you want to go inside the lab and you also want to work then that is that goes against the clean room requirement. So that is why many people do not want to work because then the clean room will be clean. But that is not the purpose. So limited people should be inside. Part of the people who, part of the things which come from people are skins, perspirations. We perspire a lot many times not everyone, many, almost everyone but larger or smaller and it contains lot of sodium. So one of the major contaminants which human body gives is sodium. Second are hair. Hair are the sea of particles. So if you even go to a normal good looking hotel these days people wear caps. Simple reason. The largest amount of contaminants come from here. Then there are whatever we wear, clothings, they have some small lints all along and they float when the air comes. So clothing lints these are the major source of contaminants from person. Even a wood. You cannot live without wood or particle boards or maybe polished boards or everything in making of a lab but in that case they are the source of contaminants of particles. Any machine which saws or do sand blasting or drilling can cause particle flow and these are the other contaminants. So first thing we will have to do in making a clean room is how to actually reduce this particle contamination and you will see that there are ways of doing it and that is what the word is clean room. So if first thing is clear that I want to reduce somehow particles which are floating around and as I said other day typically this room is slightly air conditioned, slightly closed. So it may be a 1 billion class of room which essentially is trying to say it is 1 billion 0.5 micron particle floating per FCFT, per cubic feet. We will come back the scores, numbers. So we are talking of a billion particle per cubic feet and the class of room which I am looking for is class 1 or better, m1 as it is called or even m2s. So we are looking for particle removals in a large numbers and we will see how do we do that. So first is particles, the second is surface contaminants, fingertips. We all have where we use fingers by design or otherwise we get some oil or grease on our fingertips. Then from the body itself there will be oil or from here there will be oil. These are actually stick to a surface of a wafer or around, it gets into over it, it is floating in the air but gets attached to silicon. Polish, polish is something due regards to an facial powder as I say, the fairer sex are not allowed to use any nail polish or any face powder or any facial during the work inside the clean rooms because they are the contaminants which sits on silicon surface. The third of course is molecular contaminants. There are lot many gases coming out inside the system, near the system. For example there is some kind of a evaporation system for electron beam or sputtering system. They have pumps which are firstly, the first pump is diffusion pump and below that is rotary pump. All uses some kind of oil and the heat and the oil vapors are available even inside a clean room. We try to reduce or balkanize certain areas so that those vapors do not come to other areas but there is vapour flow all along due to the system itself. Then there are wall paintings, paints on the walls, glues and epoxies, aromatics, alcohols all these things are molecular contaminants. They are available in large amount in many places and one has to take care that these are minimized during the clay creation of a clean room as well as during the fab of the chip. So I repeat, first is particle contaminant, the second is surface contaminant, third are molecular contaminant. The same number, same name I am now, little different name I gave you because though they are part of any other but there are two kinds of contaminants we see mostly in equipments and reagents. These are either inorganic or organic. Now there are heavy metals, we use stainless steel everywhere though it is best possible material which does not give enough particle contamination but it does give you some contamination which is oxides. Then there are pipes we use, furnace front ends are there, other metallic impurities present in reagents, all create some metals inside your solutions or inside your furnaces and these are sometimes as we shall see may be killing actually your device performance. From water, human perspiration here as I just now said they are inorganic materials like alkali ions which can occupy silicon surface and most likely alkali ions which are seen on inside a clean room even if it is clean is sodium and potassium, mostly sodium partly potassium. So one question is arising that is that really affecting us so much that you are worried about making a clean room which may cost a billion dollars, total area of say large fab not a small lab. Our may be around 50 crore to 100 crores our fab, new fab lab which is nano lab now it is called. Now the question arises why are we so worried about contaminants? So here are two foils, of course I do not know whether it is, I do not know because taken from a book or other paper so it is not very good but may be I will explain you. There is something called a mask which we have not yet discussed either a glass plate or some other form on which pattern is created. It is like a photo plate okay on which patterns are some is white region some are black regions. So what happens here is a mask plate which has this region black so if I have a film over which photoresist is quoted I said last time some photoresist when receive light actually either become hard or soft depending on PPR or NPR kind they are. However if there is a black portion the light will not pass through. So the expectation is that if there is a particle next door or next to that black portion particle is also opaque in most cases so light will also not pass through that. So the pattern which is getting printed on a resist now because hard and soft areas will be additional in this part will come there either it will be stout or retained whichever way resist is used. Similarly there is a possibility that the particle may sit on the dark area and that portion may actually be opaque. There are particles quartz particle for example they are translucent some light may pass through it. So the below there may be some portion cut out some portion additional portion may be blocked or cut out and this means whatever pattern I was transferring does not get transferred when I go from mask to a mask plate when I expose it on resist everything what is on the mask does not really go because of contaminant. It can also have a problem that the contaminant may sit on the resist itself and even then it can create the same problems okay. So contaminant particle size now question is how small this particle size will be in say 14 nanometer node or maybe 11 nanometer or maybe 0 nanometer later. The particle size is at least round 0.5 micron or maybe 0.05 smallest particle known and if your dimensions are of the same size so it may happen maybe here itself. I have two metal areas which I was having but and this gap is also of same order and a particle it is like this during resist. Now around metallides the two metal plates get shorted simply because metal also goes along with that okay. So you thought that two lines are separated by you but in fact they are shorted okay. This comes more at ages as we shall see in real life okay. So there is a problem with the actual reliability issue if number of particles are floating too high there is more chances of shorting or opening of the areas without you knowing unless you see finally wafer is not chips are not working okay. So this is one major worry which we say particle contaminations. For example the way impurities are incorporated in the silicon these days is by process called ion implantation high energy ion beams are bombarded on the wafer. The resist normally acts like a good mass so no resist implant can go through this. However this was a window where I was trying to implant certain impurities. Let us say this was petal semiconductor and I was doing a phosphorus implant or arsenic implant. However if there is a particle sitting in that so called window okay implant cannot then go through this it may actually scatter from there okay and may even give something called strontium as radioactive material sometimes if the metals are of that kind molybdenum for example. Now what happens that this particle has actually you have a total n region to be created and you found certainly two n regions and open between the two which you thought that oh I did everything correct but there is nothing working the mass transistor channel is broken down let us say somewhere there is a n type you had this implanted n type in between there is a so there is some kind of depletion channel things have appeared before even you started doing inversions okay. So there are issues which are very crucial in actual device performance. So these are particle contaminants which lead to failure of circuits or failure of device. There is another very particularly for people as I earlier said the technologies are normally specified three ways these days one is processor market or processor technology then the other is DRAM or memory technology and third of course is a flash okay. So most of the research is in these three areas flash memories of DRAM SRAM camps and the rest of course is in logic which is normal what we discussed. So if you look at a DRAM performance I hope some of you know or your next course in design should talk about a typical DRAM has one access transistor and a capacitor which stores the charge which stores one or zero if there is a charge on this capacitor we say one if there is no charge we say zero okay. So this is the word line X line and this is my bit line. So if I am accessing it then this turn on because I have turned on this word line so this capacitor shares charge with bit line capacitance and depending on this voltage on the bit line I can sense it was one or zero okay. Now the problem is when I am not accessing even if I am not accessing I find there are two ways charge is lost and that is why DRAM required refresh cycle every now and then whether you read it or you do not read it whether you excessive or do not excessive the charge is lost from the capacitor and because of that your one may not become one may not remain one if for long that bit is never exercised okay. Now this is a very serious problem and therefore there is a refresh has to go through which means during the refresh cycle you are not writing or reading anything okay that means there is a slow down of the total cycle of DRAM because every certain number of this you have to come back and refresh all of it again if it is a 256G memory you can think how much large time it will take to refresh all of bits of course there are faster way of rewriting but all said and done you need refresh on that and refresh cannot be selected refresh has to be through through and through okay. So there is a time required typically it may require around a 1 millisecond every millisecond you should have to come and refresh okay where your excess time may be in tens of nanoseconds or 30 nanoseconds. Now why this happens as I said there is a leakage in the capacitor and there is a leakage in the so called transistor okay because of the junction sitting in a MOS transistor they actually leak whether you like or you do not know there is a reverse bias current also the threshold which you thought is very correct threshold is not very correct there is a sub threshold current also flows in the device so there is a leakage due to junction there is sub threshold currents there is a leakage in the capacitor all and all finally charges lost okay and therefore you say you need re refresh every now replenish this charge you require to generate for example in a capacitance large amount of carriers should be generated back so that it comes back to the level you want okay. So there is a time associated with refresh cycle or even excess cycle also is called generation time which is a nuclear normally the recombination time in the case of transistors it is a recombination which generates this junction currents okay. So this called Shockley read Hall mechanisms which leads to junction leakages now these are major worries so you are worrying about the recombination time and equivalently the generation time larger the generation time that may charge is retained this much more because you are much more time to adjust the charge and therefore you will have much smaller other times or this now from where this SIH mechanisms or leakage mechanism occurs is essentially because of what we call as traps a SIH mechanism that says shockley read Hall mechanism is dominated due to presence of traps in silicon traps essentially gives a energy level in the band gap of any material particularly in silicon it may be close to half E g by 2 or E g by 2 closer to the E I larger is the trap current okay. Both hole electrons can have probability to reach E t faster and therefore recombine. So if you look at recombination times it is essentially given by sigma Vth Nt inverse of that where Nt is the number of I is the trap density of number per centimeter square and by semiconductor industry standards tau R tau g should be greater than 25 microsecond and if I know capture cross section of the most materials most systems which I what is capture cross section the area which a carrier actually sees before getting trapped is called capture cross section some other day in physics course or Professor Vasi will be more happy and much more serious about building the trap problems. So the capture cross section is typically 10 to power minus 15 and if I calculate back the minimum required tau g or tau r then each trap should be around 10 to power 12 per centimeter square okay ever CC. Now typical silicon doping is 10 to power 14 above and traps are around 10 to power 12 which is around 0.02 part per billion. Now traps are because of what? There are many materials like iron, copper, gold they are present there and there are other two sodium and collium they are essentially not so much in the band gap in the mid gap they are closer to the surface either mostly towards connection band and there are ions so these are the possible impurities and as I say when I showed you the list of contaminants I did show you all of them are contaminants for us and if they are present there empty may increase and if empty increases the tau g or tau r also goes down that means your refresh has to be not milliseconds but maybe 100 microsecond and in worst case maybe every microsecond before you access you read first write then why do you if you have to every time write and then read then as well directly you can read the data okay why my memory memory is kept because you do not want to write every now and then store something so the very purpose of memory will be lost if you require a refresh cycle every microsecond or every nanosecond 10 sub nanosecond access time is 30 nanosecond and you read write also every 30 nanosecond then why I read directly from wherever you are transferring data so the game is that this refresh cycle has to have larger time therefore track should be very very small numbers less than 10 to power 12 possible which means the processing has to see that these are not present in the silicon area anywhere around so that they just get in touch and sit there okay in any reason okay these may come even from the solutions or gases which will use so this they also should be pure enough that they do not introduce these mini impurities okay so these are very crucial problems which memory people are worried about and particularly with 4G and above or 256G DRAMs this is becoming a very serious issue of there are very interesting new technologies that come in DRAMs we do not right now want to use single transistor DRAMs we are going back to 3 transistor DRAMs for variety of reasons some other course some other day okay this third part as I say sodium potassium which I did not want to club there but I wrote are alkali ions and they particularly are part of silicon dioxide growths now if you see the gate oxide of a MOS transistor which is the major material which makes MOS transistor go it is essentially we say if gate oxide threshold if you write for either n or p type it is 5ms well function difference for me potential twice there from a potential Q NA and NA by NA by NA or ND by NA, KT by Q LN NA by NA or ND by NA plus or minus will sign come accordingly then there is a bulk charges because of doping in the silicon substrate concentration QB, QA, NA, XD, XD is the replacement layer so this is related to doping these are charges which are available fixed charges which are always available there is another term which one can add and maybe Professor Vasi is the he will be so very happy to see this number interface states they also contribute variable charge why I keep saying Professor Vasi because for 40 years he is with MOS surfaces recently 2 years ago he changed to photovoltaics but otherwise his research interest and his personal interest probably matched here he has been working on MOS surface systems for 35-40 years so I think he may be the number one person all over the world maybe I do not know many are shifted up I left it in 90s but I think he continued okay so QM is the mobile charge what is mobile sodium and potassium bang into into the oxide area or volume they actually are not fully bonded to SO2 lattice since they are not bonded or even if they are what is called electrovalent linkage the bond is very weak and sodium is released very fast little temperature and sodium will come out okay since sodium even at room temperature or at least 30-40 or 50 degree centigrade is extremely mobile it has large mobility so depending on in a mass structure this is your oxide this is your silicon this is your metal and let us say this is in a plus let us say I have a negative terminal here and positive to this minus VGS I am applying let us say so most of the sodium may come towards metal but if I put plus here and minus here which is plus VGS in which case sodium will be close to the silicon surface and when I calculated threshold whatever charges are available at the interface of oxide and silicon was taken care and that is the numbers these charges are actually very close to the surface since the metal then will be very this alkali ions would be very close to the surface they are charged species and because of that VT will actually be affected by QM by C ox term okay this is a calculation some codes can do that since the temperature varies VGS varies because you are turning on off device so it is not that fixed temperature or fixed voltages will appear so these sodium will be actually moving in the oxide if they move the QM value at a given surface point is time dependent which means threshold is time dependent which essentially means I was doing some logic suddenly I find the time of charging of that output load is reducing no reason I have done anything I applied same bias it did not work then I figured out oh it might have heated so you must have really pushed in or pushed out okay so this is called instability and there are two words which we use in most cases positive bias temperature instability and negative bias temperature instability particularly for flash ROMs this is becoming a major worry NBTI specifically and because in NB in the case of ROMs what is the way we do it we have a bit line which is charged to a voltage we have a E prom cell or E square prom cell and we are putting some gate voltage to actually ride and we assume that the threshold of this transistor is either enhanced or decreased and if it is enhanced then nothing no access to the output if it is smaller it has an access to the bit line this is a 1 0 the rate there okay now if the VT varies one doesn't know whether bit line gets connected or doesn't get connected even if it wants or it doesn't want so there is a issue in a flash ROM that particularly because flash voltage that are coming down this may actually dominate okay that's why contamination may actually dominate so we are extremely worried in case of flash to get this so-called temperature instability reduced and that is why contaminants like sodium potassium and other kinds should be minimized why I always give this actually I think even plumber talks about it because he is also electrical person so we all believe that we have reasons to think why we are doing something because electrically it will affect us at the end if it doesn't affect who cares even if it is dusty area who cares if my chips are working well without any problem I will not dare to do anything on that but I figure when nothing works then I will have to solve problem that's how we do okay okay so this is you must have understood that why we are so much worried okay there is a clean room concept site infrastructure facilities and building are has to be taken care designing a clean room you also need to think about power supplies and their distribution some are the standard power supplies can be 230 lines you need 60 volt line you need many equipment needs 60 volt so stabilized or non-stabilized both kinds of sources are required distribution is very important a priority we have to decide where power will be required so very which equipment will be there how much power will be required that kind of lines have to be put so power supply design is a major design in case of a clean room okay because once clean room is created nothing can be changed okay it's very expensive to change anything then there is also a system which needs communication from outside for one is telephones probably but even telephones box itself is a contaminant so we like to avoid it as much but we also want a system in which I can talk so there is a membrane kind of see through port which is sealed from both side but membrane vibrates so I talk and insert person can listen it okay so these are membrane blocks which essentially can allow you actual talking without any connections okay but of course telephones are still used blocked very way so whether it is okay then we have to worry about there is a large area where you think you are going to ICFAB but there will be some area which will be ultra clean which is called main clean room area also we have to worry about side areas which is feeding this main clean for example this gallery outside how much it will affect us is very strong dependent because if that is two particle content then the diffusion may actually come inside then we need lot of central facilities gas supplies line many other hundreds of things you have to supply in okay then we need process utilities you may have furnaces ALD machines so many machines lithography room so many facilities inside a clean room that's the purpose of making a chip if they are not there how do you make a chip so we need to them so we must take care how much thermal load they will create okay how much area contaminant they are going to give and which area they should be kept so that their dirty things do not go too far and of course one is always worried about environments because if I release gases in the air I may survive inside but someone outside this building or outside this IIT may also get affected may die as in one case some cases I have seen for example if it is released 30% of the people in first 50 meters will die okay before they know that's what's happening so that's very important in making designs this is just a good idea because this is what I took care when 30 years ago we of course over the years change things have changed but not much we also have to decide how much area each will have lithography should be around 25% diffusion LPC we do should be 20% implantation may be 10% thin film that is e-brim depositions etching and sputtering they should occupy 20% dry etching area may be 15% wet cleaning areas may be 10% so we actually distribute areas also where which areas what we are going to keep and we are going to see that these too much area because each has contaminant giving problems so you must mark your areas where you will keep here and how much is the area you will allocate as I said these numbers are mine and may vary from Intel lab to TI lab to other labs to even our own lab but this is typical numbers which wanna see a since I designed it I thought I will inform you that a person has to take care even if you are electrical person when you are making a lab for which you are going to work there don't believe that chemical engineers will do as good a job unless you are around so someone has to keep telling him this is what I want this is what I want okay so this is typical area as I said these are no meaning but it is just to say you that we divide certain areas for certain work firstly the sizes of the equipments no other is like for example diffusion furnaces actually release too much of gases okay don't exhaust them but there is a thermal currents which keep flowing there so if that area is too small then there is a possibility that it will go faster otherwise so I want to remove heat right there so I will keep little larger area so the power density thermal density reduces so each has some reason I mean as I said these are not specifically a curatorial but typically given by the sizes and the out gassing or out products they give one has to decide areas okay many times these are to be thrown out because this is the lab given by the institute this is the area now you put your brains how best you can achieve in this what okay but if given a space I will like to divide it properly okay before I wait I mean finish this there is a site given to them this is a center of excellence in nano whatever we have 50 percent of that center is an ISC buying load we shared the money we shared the work as well so this is taken from there I think Dr. Professor Anil and Professor Vijay Raghavan have actually created a video I have not seen it but hopefully it is there or I give you this is http www.nano.isc.ernet.in slash semiconductor clean rooms intropdf this is by Vijay Raghavan PM Vijay Raghavan www.nano.isc.ernet.in slash semiconductor clean rooms intropdf that PDA has this problem you cannot copy it so I could not other way I would have copied partly from this site itself but this is just some word print out and it is not coming good in colors okay but you can still see some way this is a corridor both side you have rooms which are glass paneled everywhere okay these are the furnaces other systems kept there this is the clean area for lithography so we do have areas which are unmarked for each activity is that clear yes so this site as I said please see it hopefully they have better maybe today I may also see it but when I asked Anil so he thinks that he told me he has put it of course this is not his slide this is from some other paper you can see from here the central area is the most clean area okay class 1 as the word written there class 1 so they they are central area which is the cleanest area and you can see outside that central area is called equipment area and which is class 1000 now this class word will define soon okay then there are process tools palms this everything the main major thing which you should look into is these three top areas okay this this this now these are essentially you can see filters kept there even here there are filters okay of course they are side filters but on the top there are filters so when the air comes they are filtered down and goes down actually these filters essentially decides the class of clean room you actually make okay so you need air filters what particles you can allow from the air to get in after you can't live without air so you have to survive so you say do block all air then contaminant will be zero so you will not be there so if you have to be there you need air okay so survival also needs you need survival so we do that okay for example this you can see the process tools and one person is standing on this so the front of the furnace or whichever this is only visible inside the main clean area where refer will be actually opened the back portion it goes into class 1000 area because that area may is not very much used for silicon opening okay so that is how we say so gradient at least should be 1000 to 1000 just outside where furnaces are sitting the front end of the furnace actually goes inside the clean room the rest is closed from our own only furnace and this is now this hip filters as we shall see soon so what are the contaminant control we need we need to have look for filters we need to have air conditioning because we need temperature humidity control typically in a clean room 22 degree centigrade is the temperature maintained 40 percent relative humidity and some areas may have 45 percent some may 40 liter of it should have as low possibly 40 or even lower if possible please remember temperature and humidity goes against each other larger the temperature humidity is much less relative humidity so if you increase the temperature to 30 degree obviously humidity may go 10 percent but then you cannot the particles will start floating because Brownian motion will be very high so you have to cool and you cannot cool too because again you will feel chill and you may not be able to work so 20 to 22 is the range up to which we normally survive and that is the temperature at which most of the clean rooms are maintained. There is a air pressure inside forward pressure is maintained which is 30 pascal I think some other day I will give you there are number of units for vacuum as well as pressure as I say foot per square inch is one but there are other star bars pascals many others five of them but we will give you some relationship and we will talk about pre-system air pressure so this is up to quarter quarter or something so if your air pressure should be larger than 30 pascals or 0.25 tower 760 tower is atmosphere okay so acoustic noise in another area it should be less than 60 dB this is what supreme court also says less than 60 dB so the same number should be inside clean room so no speeches no loudspeakers okay of course you can play what apps are not I do not know but that was not there in my time so I do not know how much vibration and noise pollution it gives okay typical vibration system should be very stable as I say we are looking for platforms where your wafer is going to sit and if it vibrates with some numbers then it can never align with anything okay so either the relative vibration should be same 0 or you should have vibrations as small as possible 3 microns per second is all that we are looking for okay there is a electrostatic discharge there is a because of the nature whenever we touch we provide some voltage electrostatic voltage typically it can be human body can be 1500 volts okay whereas we are expecting ESD is less than 50 volt we also do not want any magnetic fields it should be less than 1 milligos because many of their materials are affected by magnetic fields we need to want to have any hydrocarbons less than 100 ppp preferably 10 ppp and the many other contaminants whatever it is should be less than half ppp so this is kind of clean room requirements and you get in and this is the kind of main area these are the filter areas through which air is coming down air is coming from air conditioner and there are many filters micro filters and HEPA filters so air now this is an issue which is important because if I have 3 levels of filtering before the air is pushed in then the pressure which I am building from the system side from the pressure side has to be such that the air has either sufficient flow or not very large amount of flow if you have a large amount of flow then the particles will also come with that okay moon they move with that if you are too little this is coming it also it creates a problem that air is some globules of air is formed so that is not allowed it is called sucking systems so in between the major thing which we look into is essentially called if the air is coming from here it should flow exactly vertically down without any of the layers of air actually interacting with this is called laminar flow so there is some if you are not expressing this is the standard Rinal number we decide some kind of flow system it should be small enough so that they are not distributing themselves they do not interact so the air particles do not flow out they remain vertically down and also if this is my age the air comes and comes out and it doesn't go towards silicon side okay so these are some requirements we also look at it that the class thousand as we soon this has a lower cost there are no walls generally boards are put better than class and the environment and process tools are inside and there should be then the problem starts your other equipment is outside this clean area but you will have to carry your wafers inside the clean area so either there should be you should carry but then you are the dirtiest person around so you are carrying with all your dirt around so you must pass inside so there are called pass boxes which are air sealed as class one air sealing as they call you release the pressure the box instantly opens you put the box in close it then I will open it take the box out so this is all transfer boxes areas are also possible okay the HEPAS time for high efficiency particulate air filter they are typically of the sizes of 0.01 micron 0.05 and even 0.5 particle sizes are not 0.51 and 5 microns okay they can be large particles okay when I say it is a 0.05 HEPA I am actually talking this 9's you should now learn in the semiconductor technology or any technology we always talk about 9's 8 9's 8 9's means 99.6 9's 99.99999 is called 8 9's okay so you need 8 9 percent of 0.08 micron be removed when I say it is 0.05 HEPA these many of this kind should get out okay then there is another HEPA which is 0.01 which is 69 percentage of 0.04 particles moved okay so that is the HEPA part meaning clean room nowadays on the wall HEPAS because of much more air you want and much more laminar system you want to create certain you want a horizontal laminar you may off filter when the air comes from down upward and goes down you must understand there must be something a return air path from where air will go so there must be some kind of system where air is push pulled in and taken out okay these are called return air paths so air is coming down and you actually suck it from the bottom and that is path return every AC cycle should have 30 percent fresher every fresh this 30 percent fresher is pushed in 30 percent because otherwise heat becomes higher and higher the condenser may not work okay compressor may not work okay so these are called return paths which is very important a typical cross section I mean the plan area one can see from here these are racks green racks these are the actual benches on which systems are kept then there are shelf of glows shelf of this before you enter inside you must wear what is called as bungie this is made of terrain bungie suit which is completely covering yourself and as I say humans are the worst so first thing you cover yourself tie it up yourself use goggles sometimes in many companies you need triple goggles they call all hair should be completely invisible so you don't know who is inside actually when you see the person only by height roughly you know but otherwise you don't know who is he or she okay so there is an entrance where you go through a shower a small area you first enter there one gate is closed one gate is open you enter here there the gate goes and there is a huge hippo filter all around you and a large amount of air is pushed on you so all the part and then there is a grill below so if the grill below all the particles sit below the grills and you actually get completely dry washed okay so that is how you enter the lab okay same the same area which I shown you as the larger this okay this is something more important the the lower part of the labs are normally has small holes or drills as they called and the air actually is pushed down you actually walk around six inches grills so you walk tack tack tack on that okay there is as I say these are only these are and the areas around these are called service areas your cylinders every other things are kept backside that itself should be at least class 10,000 okay so this is mini environment for a flat floor okay these are the tools which you are working at there are also emergency exits many times there will be six in larger areas at least two in a normal if you see in our lab there is only one this is the class of rooms which I discussed of a long typically this class is specified by number of ways one is 209 east federal standards which is American standard and there is a British standard and there is a non-British non-American standard which is what most people agree this is called class of room typically this number which we are talking class 1000 10,000 is the number of 0.5 micron particles per cubic feet this is should be understood per cubic feet VLSR technology is the only area where we work on both FPS and MKS system together or CGS system together we specify something in micron and the area sometime in mills milli inches okay so class wise so CFT is volume by feet one feet one feet one feet and how many particles so class 10,000 0.5 micron particle has class 10,000 such particles class 1 has 0.5 particles is around one particle and there is a class M1 which is lower than that and can go even lower than 0.1 microns this is actually M1 0 is the latest one now which essentially is even better is used for below 16 nanometre nodes actually so you can see we are the people in right now as I say this may be class billion so we are talking of 0.5 micron particles of billion numbers per CFT floating here with air conditioners if you remove your condition will become trillions okay so also you must know when I move my hand from say one feet I move at least 10 lakhs or 1 million particles one feet I move hand 1 million particles I move so your motion in the lab also has to be restricted or number of particle itself should be small enough so they do not move okay all all tricks of the trade okay so typically as I say we are looking for class or subclass one clean rooms the reason why we are looking for this less than 0.05 now because these are the feature sizes we are working on and anything particle which is of same size and if same numbers then they nothing will come out so features has to be taken into considerations however all said and done as I keep saying nature sometimes works for you some unknown which you do not know actually helps you out so something happens good so company survives unknowingly okay this is fairly standard 299 299 in a electronic there are too many standards for examples military are standard 801 803 833 so many specifications they will give you space as its own standards clean rooms has its own standard Americans are different standard Germans are different standard communication actually communication only probably works on standards nothing else I do not know anything they do more than standards oh there should be another standard so another 20 people will meet and create third standard so every now and then we are finding standards okay good travel and good nice time missing so typically what we are saying M1 is has a 10 particles of 0.1 micron and higher of course will be smaller okay class one has 35 particles so I need M1 which should be less than 1.1 micron particle so I should actually work on higher than M1 in many cases so that of course 0.05 according to this will be even higher than 10 and we are expecting better than one inside that region so M1 also has been now there are M1 to M9 standards have appeared Intel is using M10 I do not know why that 0 appeared but that is their lab number clean room so I do not mean when I am clear the other particles there will be all size of particles even if you have a HEPA I myself may release machines may release so there are all sizes but their number should be correspondingly smaller if you go for a smaller or higher or smaller class of clean rooms and I already said what particles can actually trouble you in your circuit performance I am least interested if this particles come and go if it does not hurt my silicon thank you very much but it does and breathe so please remember all our theories or what we talk about has some advanced relationship with the performance of the chip and that is where we are actually bound to work for so all the courses in EE department in microelectronics probably were designed earlier to say how ICs are done things are improved too many these ways so I do not know it is the only thing spin transistor for example whether it will come or not but it is very interesting and its functionality is very very good it seems as if it will do everything what you want but whether it will do it at the end of the day and will go into chip is only God knows and I will certainly not because till 2050 it will not come as I say I am not going to be there may be another few years also but at least then last part of this is as I say many of the wafers when they come even if they are virgin wafers virgin wafers means coming from the company itself in a box which are very highly tight and actually there is a silica gel cap there so that no moisture is retained everything comes good as they claim they are mirror polished better than normal mirrors because the there you have a mercury oxide and other materials to shine here it is only silicon polish. However there are whenever you open a wafer in any clean area they are called clean benches there is a inside the clean room there is a benches on the top which is HEPA filters are there air is passed through this and the area below is laminar system so you open a wafer it is better than the actual clean room area and therefore much less particle contamination is expected on silicon so better than class one as I said. However silicon is extremely fond of oxygen and therefore as soon as you open it it will oxidize and that can be seen by its polish suddenly your face may not be as bright. There are also organics which are sitting whatever you touch it whichever way some grease carbon oils do comes on that. So even if we are getting fresh wafers but that is only first time but once you are you started using you are passing this wafer to number of chemicals number of gases so they are getting this every next step anyway. So some way for the before the next process step starts we must do something that the wafer is clean as earlier okay. So there is a cleaning procedure which was decided that we should have okay. Of course we also need pre-clean and post-clean things like for every process step diffusion oxidation CVD metal cylinder name any CMAS process spacer creation contact vias in CMAS for every process step you do you need pre-cleaning to do this process and post-cleaning after the process has been completed. So this pre and post is a part of the game all processes which you work to make a chip pre-clean followed by process followed by post-cleanage standard practice. So silicon cleaning is an essential part in a refabrication step and it follows every step in fact so it is very important. These process steps anyway this is the course we will do all these processes now the first process after our clean room system is diffusion or incorporation of impurities that is how device is made P and N unless we are together no junction. So first thing how to incorporate impurities is the next part of this course okay. As I say these are standards they need not have been written exactly in the same format as I wrote in the plummets book but I think everything must be available. This course I do write myself so I many times do not know from where it is or how it is but must be from some book some journal some idea of mine whichever I actually did 30 years I was in process line so I know what problems I faced. So as I say every process step should be followed by cleaning. Now most important thing is a intermediate circuit cannot be fabricated without a etching step after you are selectively doing certain thing that means certain areas you want to remove okay. So etching is most important step other than lithograph as we shall discuss and they actually form the major base for any IC probe okay. Now there are two kinds of etchings we used one are called solution based in which you actually dip the wafer into a solution okay and other is dry etching in which only gases are used to etch the things. Solutions liquids and dry in the dry etching also nowadays there is another etching which has started coming and that is called photon based okay you bombard with photons okay something new has appeared but anything you bombard photon is a practically no mass so one expect no damage but it is seen there is a damage okay. So we need to remove nitrites remove metals we want to remove even silicon to create trenches oxide is the major thing we keep etching every now and then okay. We also need silicon etching because to create wear down okay. So the procedure which most people believe is correct slightly modified by different companies there is a Siemens process there is a iMac process there is a RCA cleaning cycle which is most accepted by many companies with modifications which they never tell okay like Intel follows only 80% of RCA clean. So what is that 20% no one knows okay Piranha cleaning they have another process in which three steps are different from RCA why do they do it but their wafer processing is good actually. So each company has its own cleaning cycles and this is what a very standard cleaning which is most people believe it is good and also work partly at least 80% to 90% is here if you are written down I may start looking for RCA that is the major cleaning system which almost everyone uses. There is a small difference in some place steps which they back in 83 I have introduced which now they claim iMac has is doing it possibly. There are number of steps when you take the silicon first you boil the silicon or other heat maybe 120 to 150 you can also call it boil in H2SO4, H2O2 solution typically some people use 1 is to 1 some use 1 is to 4 that is the difference every person has his own standard procedure in mind and he keeps doing it okay. 1 is to 1 is mine maybe others I do not know it goes up to 1 is to 4 if you look the books or journals typically 10 minutes boil is sufficient. Please remember do not put water around when you are boiling H2O4, H2O4 actually splits and may hurt your face okay. So do not add water any time in boiling H2O4 okay. Why I know because once I did my whole hand got burned so I tell you now what is the purpose to remove organic materials grease and resist please remember H2O2 actually in the solution system immediately reduces to H2O plus nascent oxygen and nascent oxygen is extremely reactive okay. So what happens any material which by removal of their oxides are soluble this nascent oxygen will oxidize them into solvents soluble this into the liquid. So they will be removed from the surface okay same with carbons all carbon containing compounds get oxidized faster and our solubiles are at least separable from the silicon okay. The major reason why we also have H2O of course can remove many materials because of sulphates which are solubiles but we actually have H2O2 for one more reason because it creates silicon dioxide H2O2. So this oxygen reacts silicon faster and then makes SiO2 and this thin SiO2 layer actually contains impurities because that surface area is same so some of the impurities are well within this thin oxide. So if I next time remove this oxide even those impurities will be taken care those will not form sulphates not form oxides can still be retained in this top silicon dioxide layer which are naturally creating okay and they can be then removed out. Every process step of any kind must followed by DI water rinse we will discuss what is DI water and after DI water rinse we the third step is the HD wafer HF plus H2O HF is HN for SiO2 fluorine is a very strong oxidizing I mean reducing agent for SiO2. So it removes as a silicon fluoride silicon fluoride is soluble material okay. So SiO2 plus 4 HF is SiF4 plus H2O is the reaction okay. Typically this is called dilute HF 1 is to 10, 1 HF when it is in some people do even as low as 100 H2O which means dilutes the HF even lower. The reason why they dilute because thickness of silicon dioxide is very low during the last process. So this is sufficient for removing this much. But there are other people believe that HF also tends to actually create I mean why larger because it may give some pits. So they are people who are different times and temperatures. Typically 1 to 2 minute room temperature is sufficient. Again every process step must follow DI water rinse. So 3 to 5 minute you rinse to remove all HF traces okay. Then this is the most 2 steps which are coming now are the most important step. This is how RCA is most famous for. To remove organic metals and like hydroxides or oxides we put the wafers in NH4OH H2O to H2O bath. Typical ratios can be 1 is to 1 is to 5, 1 NH4OH, 1 H2O to 5 water. Or in some cases 0.05 NH4OH, 1 H2O to 5 H2O. This is called SC1 cleaner. Company names this as SC1. You boil the wafers for 10 minutes around 80 to 90 degree centigrade bath or by flames which are this heating plate. Nowadays we rarely use plates because plates themselves have particle cream in this. So we actually use baths okay. They remove as I say and since there is H2O2 it will again leave a trace of silicon dioxide when you remove the wafer from this solution. Again rinse it for 5 to 10 minutes because this NH4OH ions has to be removed much more from them. So we rinse it longer at room temperature. For example, IMEC not only adds in this first step either H2O4 and ozone or followed by H2O plus ozone. So they are ozone technology. Ozone again is similar like H2O2 because ozone breaks faster into O2 and O and it does the oxidation. So instead of H2O2, IMEC is the micro center at the Belgium where Anil has his PhD. So it is from his, he is using ozone. Ozone has an advantage that you can actually push through gas phase but then it bubbles and it actually does not allow proper rinsing of silicon. So there are issues. If you ask me I would have said that is not the best way but it does not matter. After this we want to remove remainder of metal ions or alkali ions as chlorides. So we have, we boiled this into SCL H2O2 plus H2O in a ratio of 1 is to 1 is to 6. Again boiled for 10 minutes and 80 to 90 degree centigrade. Chlorides are removed as alkali ions, metal ions as chlorides and again whenever you put H2O2 you will leave a thin layer of silicon dioxide. Now many people believe and that is the difference between our RCA clean and many rate RCA clean which started way back in 70s. They do not have this step 9. They do not etch the last oxide. They believe this 20 angstrom oxide is a good passivator. It is punchable by most impurities. So it is better to preserve it for the processing. We believe that 20 angstrom oxide may create interface states. Those interface states may actually reduce your mobility in a short channel device. So our belief was slightly different. So we normally used to age and many companies now I find do age this HF actually. So we give 1% HF treatment that is 1 to 100 in water, boil it and again rinse it very heavily now for 20 minutes for removing all kinds of fluorine traces, all kinds of water in the material on the surface should go away. Normally in a running water, spray water as we say so that most of the physically removed particles can be moved out. And many cases they can dry these wafers into N2O gas flow system or there is another which Intel uses isopropyl alcohol vapors IPA. See people do not think N2O is very pure compared to IPA. This is there because it can be multiple distilled IPA. So Intel still uses IPA compared to nitrogen flow. Please remember the process steps which I am talking are generic company to company. Siemens are slightly 30% different from this. Micron has even different. Global boundaries have even another small variation in their last steps. So each company has its own processing cycles and we keep doing before we leave just a minute. In IC processing most use reagent is water. Water is the major reagent we need. As I said other day we need roughly for a good foundation 54,000 gallons of water per day. So it is a company has to provide get this much water from a state government. 54,000 gallons per day is used by a company. So we decide the quality of water which is called ultra high pure water. Something nowadays most people are having you know aqua guard or water purity or whatever it is. It is essentially purifiers. Similar system is used. We have three quality markers for water. One is the resistivity. Other is particulate size in the water and third is bacterial content. Now typical tap water if you look at actually there is a statement made by some good doctors these days start actually drinking tap water. I do not believe me. I am just telling you. They say it is much more good. It is much better for human body but it produces what is called as cells which will actually protect you. So maybe. But essentially tap water are dissolved in organic compounds of sodium calcium, strontium also. Even some mercury traces are found in Mumbai. These are the organic compounds like living matter. Then there is a particulate like silica, sands particles, dust particles, paper pieces. Then there is a microbiological life like fungi, algae is there and there of course bacterias. Both can be viral or non-viral. Now the first thing and mostly we do it is to figure out resistivity and typically water resistivity we use is something quickly to one more slide and we will finish. H2O please. H2O goes H plus OH in any equation in equilibrium. H and OH ion concentration is equal and it is typically the order of 6 to 10 to power 13 per cc. Diffusivity of ion is essentially given, we have discussed this KT by Q mu where actually the mu or mobility of ions is given by Z QD by KT and this relation is not pure Einstein because there is a charge Z, ion charge Z is associated. So it is called Nernst-Einstein relation. So we calculate for mobility for mu H plus and mu OH by using this if I know the diffusivity. Diffusivity of OH and H1 are known from chemical people. This is 9.3 to the power minus 5 centimeter per second. This is 5.3 to this. So I get mobilities. If I know mobilities then I know the concentration. What do I calculate? Resistivity. I know the mobility. I know the number. I know the charge. So I know the resistivity and resistivity of water is therefore a measure of quality. We will just show you the numbers. H2O breaks into H plus OH equilibrium both are same concentrations. Diffusivity is given by Z QD by KT is mu is called Nernst-Einstein relation. So I calculate mobility for hydrogen ions, mobility for OH ions which is 3.59 and 2.04 centimeter square volt second. These numbers so if we can somehow do better then we will see what can be done better. Is it okay? So if I calculate resistivity which is Q mu N plus Q mu P like similar like H and OH. So I calculate mu's. I know concentrations. I substitute them here and I find that the pure water has 18.5 mega ohm centimeter resistivity which is extremely high. You need not put all this. You can write rho as this and directly write this. I just did maths. If I compare with tap water with deionized water then the resistivity is tap water is 200 kilo ohm, 200 ohm centimeter. You can see how bad it is. So many ions but as I say they may be useful some of them. 200 ohm centimeter is a tap water and di water is greater than 18 mega ohm centimeter. Electronics available inside the water is 10 to power 5 part per billion and it should be less than 10 part per billion in case of di water. Particulates 10 to power 5 numbers per cc in tap waters less than 10 numbers per cc in di water. Organics 10 to power 2 to 10 to power 5 numbers per cc in tap waters. Of course these are not necessarily bomb water. It can be worse or better in certain areas. Boreeville may have worse. We may have the best. Do you know why? The first water release comes from the cleaning this in Bandu to IIT without actually getting further contaminated by the pipelines. So we get probably the best water in whole Mumbai luckily. So these are the numbers which we get and maybe quickly next time we will show you some systems and they are not very important. And we will show you that once the process is now ready we can start actually the first process in corporation of impurity. Maybe I will show you di plant or something for 2 minutes and then start. See you then.