 Here we go just for the case, in case someone have forgotten, a mixed signal design is very interesting because most of the design currently required are mixed signals. So you will see that there is no deposition, there is not something else I will start, I will talk about D2A converters, A2D converters, PLLs, all kinds of funny things are circuits, how do they interact? So this is a part of mixed signal. So we will see there something differently and at that time I will never say sputtering or any word of that. But today I will say something more about depolitions. We have done electron beam evaporation and normal resistive heatings evaporation. Today we will finish up by sputtering. Basically I already said sputtering uses, sputtering is a plasma ions which can be energised and they hit the target okay. The target is the material which you want to deposit and substrate is the way where you want to deposit. So these ions actually hit the target and they somehow provide sufficient energy to stationery atoms and when they come out there is something called momentum reversals. And those species of target they get accelerated and actually go and hit the substrate and they lose their energy and therefore stick there. So that is the word sputter is okay. In most cases sputtering is done with a neutral gas like argon but there is a possibility you can add some reactive gases like you can add oxygen okay. So if you are sputtering and you also oxygen plasma then you can oxidize something okay. So this is called reactive sputtering. But most cases people do neutral ions sputtering. I think that word will come soon and you will know why where implant starts and where sputtering ends okay. There also actually we do not have that cathode to blow there but there are also energetic ions. Here also there are energetic ions. The difference there and here is that there is the acceleration is very high. You know the column is so long and we accelerated to 300 keV. Here we will not like to go that kind of energies okay. So sputtering uses neutral gas plasma to sputter target material to be deposited essentially it displaces the atom and deposits on substrate. It normally requires moderate vacuum 1 to 100 millitars may be sometime 300 millitars. It is useful in deposition of metals, alloys, compounds and even insulators. So that is the strength of sputtering it can deposit anything actually. Mostly it is used for deposition of metals and alloys okay. Where do you think alloys are deposited? For example these days we may require something called molysilicide or platinum silicide or titanium silicide. So these can be sputtered films okay. Higher end metals like tungsten only can be sputtered because tungsten is a 3300 degree melting temperature. Electron beam uses targets which are of that temperature. So you cannot actually operate that material. So essentially the materials which are high temperature melting points they are normally sputtered okay. Please again once all in all again I have argon gas which is neutral or inactive. Argon has an atomic number of 18 and sufficiently large in size therefore. Please remember I keep saying when it is ionized it will have equal number of electrons and neutrals. So plasmas are always neutrals okay. Thus we have argon plasma available for sputtering. So instead of any other gas if I have a chamber in which I have argon which is under vacuum to some extent and I put DC bias against it gas can be ionized. And if the distance between 2 plates is smaller there will not be a positive glow there will not be any ferrata this. Only one glow plus small dark space which we call sheath and the cathode okay. This is what the figure I have shown you. This is the kind of plasmas you will get. Earlier I have shown you anode. So showed you in this case normally anodes are grounded and cathodes are given negative potentials. Since there are negative glow available here which has plasma and some electrons are accelerated out of it towards anode there is a small amount of voltage drop or dark space you can say even at the anode that is called positive sheath. Now or anode sheath. Now these 2 things are so in general in a sputtering system there is a small sheath at the cathode a larger sheath at the cathode and a smaller sheath at anode as well okay. Where from where you can try this? You have done an IV characteristics of a plasma system. If you have written down I will just show you the figure which we have shown n times but once again I will show you. Since anode is grounded the potential at the anode end should be 0 okay. Since it is negative potential at this end should be negative value okay which is the cathode voltage okay. I repeat I am creating a plasma organ by discharging it through a potential of minus Vc to ground and I expect some cathode drop here which I called sheath potential in the dark space then there is a glow and there is a small anode sheath at the anode side is that okay. So the typical figure which we have been keeping showing you all the time at the voltage drop across cathode to anode this is the minus cathode potential Vc and as the dark space you go towards plasma from cathode towards this the potential start rising towards 0 because in the glow there is no much potential but some potential is developed in the plasmas because electrons are leaving and ions are moving. So there is some moving charge which results in equivalently saying there is a some charge on this which gives some kind of potential that is called plasma potential Vp. At the anode end that is 0 because you are grounded it which has a voltage okay voltage in argon plasma is positive with reference to both cathode and anode. Now argon ions when accelerated towards cathode where will they cross? Please remember argon ions are in this and they are moving towards cathode and they are getting accelerated because of the electric field which we have applied. So these argons then will bombard cathode okay. So if I keep at the cathode of material which I want to deposit let us say tungsten or any other material. So these argon ions will go and hit that is called target. Cathode is normally a target at us area. So if ions hit the target and they give their energy to them and it so happens and which is what the word is a momentum reversal takes place. A stationary atom of target is removed from the target area and starts moving towards anode area which is our where we keep our substrates okay where we keep our substrates. Is it okay figure? This figure we have been drawing it end times okay. This is why I actually discussed plasma in detail because plasma is a as I keep saying 99% of the world is plasma and we do not talk about it okay that is very funny. It is a DC sputter. DC or RF also will behave similarly is not it at high frequency. It can same can be used in RF we will just show you the RF sputtering part. This is DC sputtering going on. DC has a problem which since he has said it the target has to have some metallic system. Otherwise you cannot put a bias there. In RF you do not have to be that to be a metallic part because RF will just move back and forth. So the advantage of RF is not just getting a fields but it also allows you targets of non-metallic materials okay that is the only difference okay. The argon ions accelerate through cathode sheath towards minus VC potential argon ions are sufficient in kinetic energy and now they bombard the cathode. In other case cathode is the target plate and anode is the substrate. Then the target atoms are displaced from cathode and due to process of what we call reversal of momentum they move towards substrate and deposits there. More energy energy momentum transfers can be discussed but as I say time not enough to really go through full theories how reversal occurs can be proved. Just take it the target atoms when released they are still energy enough to come out and go towards anode. We define the terms sputtering yield that is the most important term which we use it is essentially defined as number of atoms or molecules ejected from target for incident ions. So one incident ion of let us say argon how many or this comes out of the target is called sputter yield okay or sputtering yield. It has a I think my notation is s just check what plumber has I think he has not used s as the same name yes that is what I say argon ions hit the target give energy to them okay and the target atoms get excited and come out till they receive energy how can they come out yeah but the mean free parts are so that that is the whole game that mean free path for this is kept so high compared to this distance so the collisions are minimal I will not say there are no collisions so there is scattered there as well but that is a small scatter I will show you figure how it occurs for your sake here is a figure I knew he asked I knew what figure I had to show you so even if they this they are still in the field so they still both move not in same direction actually moment and real is possible because they are not going in vertical direction is it okay sputter yield is number of ejected atoms of target per incident ion the sputtering yield does depend upon the mass of ions okay incidenting on tar so argon is heavier so if you are lighter it is unlikely to actually do much damage actually it may not eject it actually also what energy they strike okay both energy as well as momentum is deciding how much kinetic energy it will have and how much it can transfer to stationary atoms this is typically what you are asking argon may hit may come down may hit another one but they again accelerate and there is some kind of a steady state achieved some angle where this and some target atoms will come down okay this is random process and can't be explained in just one this is only a figure just to show you this randomness allows you to spend momentum reversals the intention is this is probably given in plumber's book I do not realize but recollect this is just to show you even if there is a step somewhere since the atoms are coming all angles they will actually go what is called as conformal it will climb on this is that is the biggest advantage this process allows okay because some will come somewhere and hit if there is a step in the wafer it will climb on that right some atoms may come like this some may come like this some may this is random process it is uniform over the time because this process is random so it is average output as I say I am not doing lot of theory on it say you know there is another full book on sputtering so you cannot carry on that in a day what is the difficulty with DC sputtering is that you need cathode to be and cathode is not to be metallic plates because you need bias to be created there so they both are conducting plates okay particular target has to be have metal because otherwise it cannot receive the voltage okay this is just a figure and this is not drawn to the scale and this is just to show how random processes can occur and can create reversals okay here is something that is more relevant and which is DC sputtering are rarely used except maybe we used to use in 1985-88 now we also do not use I do not know whether we use it also but hopefully if we are using this using system which is often used in the VLSI technology is RS sputtering we are last time discussed the area ratios smaller area compared to this please recollect what I did say if cathode area is A1 cathode anode area is A2 we have two possibilities either areas are same or cathode is smaller than anode which is two possibilities we can create I applied to a cathode through a matching network what can why you need a matching network because matching the impedance so what kind of matching network it will be LC normally LC with variable resistors there okay which is not really a resistor but the device which provides you the resistor matching network system costs 6 lakhs in market so you can see it cannot be just wired LNC okay so just to tell you okay and in our lab if still existing old lab one of my student has designed and fabricated our matching network which we also spent 2.5 lakhs but at least cheaper than the 6 lakhs so we have a generator which essentially is going to be this then we have this is called susceptor or substrate holder which is normally heated okay around 300 degree centigrade and then on this substrate this you keep your wafers whichever number of wafers this this is a circular disc okay in which you can keep number of wafers larger the surface area your larger number of wafers you can keep this is grounded so is the chamber is grounded okay the chamber is also of metallic this so both chamber as well as substrate holder is grounded which acts like an anode there is a gas inlet there is a vacuum what do you do first we evacuate everything we actually evacuate to the pressure of how much it should be as low vacuum as as low pressure or as high vacuum possible to avoid oxygen or any other species which may interact okay so 10 to power minus 6 star is a minimum you should first go or if better pumps available even minus 9 you should go but generally larger the vacuum requirements costlier is the system but if your films are not that great you do not put that money okay. So between these two as we have discussed if the areas are equal the potential at both sheets will be some kind of zero at both ends and there will be a plasma potential BP in case the area of cathode is smaller than area of this it will have a more drops because it will be why it will be have more drop because the ion density will increase is that point area reaction means ion density will increase and larger the ion density more will be drop across J Sigma E okay so it happens that if I push smaller area I am actually increasing the potential at that end so it will be like earlier one DC kind and it will be something but then the net potential is applied average of RF plus DC which we are through matching so average is same as what it was area under the curves are same so more potential at the cathode but lesser plasma potentials the ratio of V1 and V2 potential here and here this is V1 V2 is essentially the area ratio and M is some factor which people believe it is around 1 to 2 many people take it to it is area squared but it is not true you have to figure out from different film measurements and find what is the ratio of A1 A2 which means the ratio of M1 M2 sorry V1 V2 why V1 V2 ratio I am interested in because how much is the plasma potential and how much is the cathode potential will decide the acceleration of ions how many available and how much acceleration I can give will decide by VC and VP so that adjustment I will have to do for a given species which I want to evaporate on this so what is the advantage of RF system compared to DC I am not saying which cathode has to be metal I can put an insulator there bad play I may put metal but I am not because it both side RF is moving plus and minus so it does not really matters too much and therefore even insulators can be sputtered in the film RF systems insulator has a problem like SiO2 when the ions when it will hit it it should be less than the binding and I mean it should not be larger than the binding energy of SiO2 otherwise silicon and oxygen will separate okay and this oxygen ions may go back to the silicon upper target and rather going to the below target so the catch is that how much energy in reactive sputtering you should have which will allow you to do silicon dioxide silicon nitride deposition so there is a every process has a different methods and so much pressure and so much even we do that just it okay. The advantage of all these sputtering is conformals sputtering what is conformal the shape of the surface it will pick up all the shape of the surface if it is step it will step it will go down it will go down this figure is probably available in every book whichever book you have seen or not seen okay. So basically two things I am saying sputtering helps you DC sputtering can sputter metals even high rare earth metals like titanium platinum high noble metals they can also be sputtered and other rare earth like molybdenum tungsten they also can be sputtered a operation cannot do that because there they are used as a source of basket where you heat the material so never use these for other sputtering is the only possibility what is the else possibility could be if I do not want to do sputtering the problem what do you expect this problem is sputtering since there are too many electrons and ions are going and their energies you are going to adjust there will be some damage on the wafer is going to come okay and this damage has to be annealed again so instead of trying to anneal the damage you can reduce the damage by another process which will now do is chemical vapor deposition so there we will not use physical bombardments okay okay. So this is advantage of RF system is conformal and even dielectric can be sputtered but before we go to CVD here is a model for simple model for sputtering this model is not the best model but you can one can use it to find the thickness of the film deposited what is our ultimate aim in the model I want after so much time how much is the film thickness I got because that is the decision given by the circuit people I want thickness of metal line to be so much because I am expecting R to be per unit length so much okay interconnect so I am told by them so I must provide them that much thickness okay is that clear why designers actually just say I want this and then how much we play for okay a very basic model as sputtering is provided we already said what are yield is the ejected atoms per incident ion and it is found that S is equal to alpha times non-proportional alpha times e to the power half minus e to the power th to the power half where is the incident ion energy and as a mass m1 and may be Z atomic number Z1 which what is for this incident means what in our case I have gone so I have gone as atomic number of 18 mass much larger typically 32 38 then eth is the threshold energy required for dislodging a target atom so unless e exceeds that target cannot atom cannot be displaced and also gives the momentum reverses and alpha is a proportionality coefficient and it has been found that this is a function of target atom atomic number the atomic number of incident ion gas atoms and also what we called as sublimation energy or it is also called binding energy substrate binding energy sbe alpha is a proportionality constant which is a function of target atom atomic number atomic number of incident gas atoms and also its sublimation energy which is also called the binding energy so we will get the formula for alpha so I can I will find I will be able to find alpha I know what e I am incidenting and if I can find eth then what should I do I must adjust e so that for a given species and given substrate gases one be able to get some yield of your choice what is yield to do with it how much ideal it will be one one e should give one or even more if possible even gain but that is very difficult okay so we are trying to see how much number of atoms I can get on the substrate per unit time per unit area flux as we call and for a given time and mass we are using we can find how much film thickness we have so that is the ultimate purpose of so I must first find what should be eth and once I know eth and if I know alpha I know for a given yield which is decided by the current requirements or the sputter requirement I will be able to adjust my e is that correct so how e is adjusted is adjusted for given as we decide the thickness now without going into kinetics I just wrote down final answers for it if if m2 is the atomic mass of the target and maximum energy transferred to it by incident energetic ion e energy is the ion and mass of that ion is m1 that is argon in our case this is not see it can be found by writing the conservation of energy momentum one can derive this expression e maximum which is transferred to stationary atom is 4 m1 m2 upon m1 plus m2 square into e e which is incident ion energy the maximum e max which you can achieve is called ed or what is called a displacement energy and that is also we can say in normal case this ed should be same as the threshold energy for removal of atoms so I must find ed from this I know m1 m2 I know incident energy so I know ed and if that ed I use that as a thermal energy and if e is larger than this only then we can say some sputtering will take place generally this energy transfers in the range of 10 to 35 e v please remember energy ed is this is the maximum transferred energy and not ed ed is this stationary atom energy which it can pick up if I hit with e max after exchange e max is the available energy there and ed is the one which is required to take the displacement of atom out of the target which essentially is threshold energy so ed is typically found to be 10 to 35 e v incident ion will be higher energy and normally ed will be smaller and only then sputtering can take place please remember unless heated and energy is received by stationary atom it cannot come out so it must get transferred enough that it ejects out okay so this as I say I just write down this right now because law of theory involved and I have no time to explain now I just thought I will give you the expression which this expression you should know this energy which I may give you what is the typical threshold energy is used and why I am doing it I want to finally get interested in what thickness of the film so I will try to relate that s some way is that okay alpha can be derived as 5.2 by u z t upon z t to the power 2 by 3 z x to the power 2 by 3 to the whole power 3 by 4 z x upon z x plus z t to the power 0.7 this is some kind of a function derived for alpha or incident on hitting the target stationary targets please remember in this analysis all internal energies are used in elect e v per mole so 1 kilo calories per mole is essentially 0.0434 e v per mole since all this formula has been derived using e v per mole so even if you are actual this is a kilo calories per mole you must first convert it into equivalent of 0.0434 e v mole ratio all incident ion energies and ETH are expressed in kilo electron volts okay the cathode voltage for example 100 volt is essentially saying has a ion energy of 0.1 k e v is that q into e so what is called k e v for a case of tungsten using the alpha value for tungsten and argon I have figured out at 100 volts s is typically 0.2 atoms per ions you can find out yourself I have given all values you can calculate s this is the value which I calculated roughly 0.19 something it came so I have just said 0.2 atoms per ion so one incident ion that is five incident ion will eject equivalent of one target atom so what is the rate of sputtering if s is the yield if ions number of ions per unit area per unit time is j which you are monitoring across the circuit which is why matching network you can actually monitor the current so it is the current density is j ion so what is the sputtering rate j into s divided by q divided by q okay because j has q term so just divide by q is that okay point clear to you this of course is not given anywhere this is my just calculation from the formulas which I had just try yourself put some values I used argon and tungsten okay tungsten has a rate of 74 atomic number is even I do not remember 32 or 31 something I have taken from a table so you can also that data I will give you so actually I had to go and look for these values so I had to ask you means I will have to first search that new atom the rate of sputtering is called RSP is s time j ion by q unit of RSP a number of atoms per unit area per unit time and RSP is related to growth rate which is what you are looking for is g time g is the growth rate row is the density atomic density evaluator number atomic weight m this we did somewhere I am plantation okay so it is similar expressions are obtained for this now the question asked by this if your incident on energy is much larger than threshold energy much larger than ions will get inside okay and not be able to this even if they display since the depth is higher they will sit inside so they will not get ejected is that clear to you if I have very large incident energy the ions will actually go deeper to lose the energy and the momentum reversal may not allow atoms to come from the surface or you can say there will be some but that yield will be very small and I am looking for sufficient yield of atoms to come back and therefore larger energies are used only in implantation smaller energies are used in sputtering so basic process is similar only thing is if you hit too far it will go inside and then the atoms coming out may not transfer the energy enough to come out of the substrate I mean target itself it may sit inside it may what we call damage the upper layer but nothing will come out that is what has happened in implantation the upper layer gets all damage atoms keep moving from the sides but nothing comes out no silicon deposition takes place that is exactly the difference between sputtering and implantations is that clear to you so do not get too different they are similar process and similar formulas have been used to derive both cases so this essentially we did one of the method of deposition is physical vapor why it was called physical because I am energizing and kinetic energy is provided by me and if they hit somewhere or they evaporate with another kinetic energy vapors and hit somewhere so physical vapor deposition are evaporations using filaments using electron beams are using sputtering before we go to the itching part later maybe next week last class or can you think the same sputter unit can be used as a hr what is being deposited is the target this so if you change the one upper to lower lower to upper it will become RF hr okay exactly what they do okay so this finishes PVD now we start with the next possible method of deposition which is called chemical vapor depositions again there are many many features of CVDs which this course should have talked about but will not be able to talk about so some gist only will now be provided okay I may be actually recording the rest part some other time and we will put it into my course but for your class of three hours three three and a half hour whatever it is I will not be able to do lot of justice to lot of thermodynamics which is involved in this okay I just mentioned what is the thermodynamics involved or there are number of ways CVDs are number of systems or number of methods available for CVD one is called atmospheric pressure CVD which is popularly known as APCVD then there is a low pressure CVD which is LP CVD then there are also of course they are not these and these are similar but not exact there is a atmospheric pressure low temperature CVDs particularly used for dope glasses then there is a plasma enhanced CVD then there is a very popular name atomic layer depositions okay essentially I will tell you ALDs are used for textured depositions and will some other time this word will be explained MBE molecular beam CVD actually it is not called CVDs MBE is deposition is going on molecular beam epitaxy as the word will discuss this word epitaxy here as well and we will see that this is the best possible deposition system extremely costly these days Samir has won for and it is only for one species if you are working on gallium arsenide you have that MBE machine is only dedicated to gallium arsenide unite right you are another machine and 10 crore is the cost as of now so it is a huge money unless you are really working on LEDs do not buy any such systems okay okay so these are possible CVD techniques we will not go into all of them but at least briefly to expose to all what films we need in VLSI we need dielectric films of SiO2 silicon nitride half new oxide and many such high cave materials okay like lanthanum oxide gallium oxide mini polysilicon films self aligned gate and small interconnects okay these are requirement for polyfins of course now pure polyfins are really used they are used as silicides with some material metal molybdenum or even titanium where do you think polyfins other than mass transistors are used LCDs LEDs what else there is a driver for the small displays for LCD display there is a circuit which is used out of thin film transistors all thin film transistors are made out of polysilicon because there are 25 things which polysilicon can do a poly plant 200 million tons 2 million tons a year may cost you 200 million dollars and unless that much 2 million or 2 tons of poly is used by us there is no point in putting a plant okay and therefore in the world there are 3 or 4 companies which manufacture silicon polysilicon and this one is Wacker Kami from Germany the other is monocenture from US do calling from this and then there is one Tagano or something in Japan thermal bath okay so further we also need atmospheric CVD for epitaxial films now the word epitaxy will come back again later epi stands for as it is okay tax means texture so if I have a silicon wafer and I want to deposit silicon crystalline wafer as it is because lower was crystalline wafer I want to deposit a layer of crystalline silicon then it is called epitaxy or epitaxial layer that process is also atmospheric pressure and maybe quickly we will talk about how they do that why much is your thermal budget and the quality FM desired decide the choice of process we have seen just now in the process which you have gone through we are used titanium we are used tantalum we are used tungsten we are used poly we are used nitrides we are used and of course high cave we did not show but that is also replacing a sigh to gate so everywhere we require one or the other CVD to go through and not all processes will be from one kind of CVD okay at one time I will use one kind of CVD for other time I will use other kind of CVD and in some cases I may use sputtering okay which I find is cheaper like the applied material has a sputter edge they are using etching using sputtering okay but they also can use it as a depositing system metal depositions okay so they are same system of course they are pores separated so that no contamination but the system is same have you been any time to that lab space try yourself if not force yourself through someone just for the heck of it before we start one of the process or the other or there is some comparison people do which I also did we are three kinds of CVD compared APCVD LPCVD and PCVD and there are certain advantages there are certain again we will when I come to the figures I will re give you some more data there okay but just for the heck of it I store the APCVD is a required simpler reactor it is a very fast depositing system and can be operated at low temperatures the disadvantage is poor step coverage there is a particle particle contamination possible and it has a very comparative lower throughput rate what is lower throughput rate number of refers per unit time how many come out is the throughput rate so it has much lower throughput rate of course we will show you one conveyor belt system where throughput is increased but not so great the way gases are amount of gas which I pass in APCVD is in liters tens of liters so the farm growth rate is very high if I do not do it then what I will show you a boundary layer cannot be formed that is laminar flow is not maintained so I want a laminar flow so I will maintain my gas pressures as well I mean atmospheric pressure water gas amount so that laminar is maintained I will show you this now so it is very fast depressions throughput oh sorry throughput and this is not correct throughput is a number of way first per unit time which you susceptor can hold so there are 16 way first so at best in one run 16 will come out okay that throughput the depression rate only decide how thick the films I can how fast I can deposit a given thickness so that is fast where throughput is from the system how many way first come out per minute per hour water is the rate so if you have if you have a susceptor which can hold 100 way first it is a throughput of 100 okay per say 30 minute of water but any flat system you know you have a 8 inch dia way first so how many way first you can really keep 6 8 on a big susceptor but if you start like this you can start 200 okay so normally at atmospheric CVD have a flat requirements and LP CVD all are vertical requirements and therefore they have a lower throughputs this they can be used in doped and undoped oxide these are called low temperature oxides okay and also of course I forgot major and EP but then it is not low temperature hence very high temperature 1100 degree LP CVD which is low pressure CVD it has an excellent purity very good uniformity very good coverage and very large throughputs the disadvantages it is high temperature process relatively and low deposition rates and it can do almost everything okay yeah that for poly silicon depends on what you are doing it can go anything amorphous can be gone at 400 and 300 poly will be at least 600 to 900 which is not atmospheric pressure you can do at 400 LTO low temperature oxides if the application of AP CVD is only deposition or dope glasses passivation as we call then it is 400 degree depositions okay but EP as I said if you are doing AP CVD EP taxial growths then it is 1100 that is a very high temperature because you want crystallinity okay is that okay poly has to be little larger temperature because you want some crystallinity to attain is that clear nothing else can create crystallinity except the thermal energy so you have to provide that you may do it either by furnace by RTO RTP or water process but energy will have to by lasers there are methods of doing halogen lamps all kinds of processing can be done but energy should be thermal for crystallization no other way atoms must settle okay there is a solution method it can go very thick films but they are call it quality quality is so poor that they cannot make any cell there if you make a solar cell out of such soldiers solution gel as they call it may are less than a 0.05% solar cell you may have a film so it depends on the application and what kind of purity you are looking that much thickness and that quality has to be attained so process is decided where is your application is that okay there are cheaper processes available now thin film depositions which we are trying for silicon is a poly silicon film just deposit keep doing on a ruler so large area solar cell with thin films were 3% 5% efficiency so it depends as I keep telling you all processes are available your application and your quality or requirement should decide that the mask as a high you just subject the same thickness since it is a gas it does not have to be hit by anything it just finds the slope wherever it climbs on that so thickness is roughly uniform even if you have steps if it is a large upper area upper scope that you have shown you know see we have to do a CMP just because the area I mean the surface is not planar okay so non-planar surface will create non-planar depositions there is nothing we can do the last is of course the PCVD which is low low temperature fast deposition system it has excellent shape coverage actually it is doing at pressures which is slightly better than LPCVD there is a problem of chemical and particle contaminations which is like APCVD you can do low temperature insulator depositions over metals this is many times required in interconnects where this is used there are 7 layers of interconnects I am creating and in between I need a low dielectric materials okay and these are normally deposited by plasma enhanced CVDs and also finally because you do not want to disturb the wafer ready for circuit so the passivating silicon nitride is normally done by PCVD or sometimes using low temperature atmospheric CVD because 300 degrees the uppermost temperature will be given to okay this can be done at 100 to 300 this can be done up to 354 yes both APCVD does not have because it goes gas goes like this I will show you the figures now see anything in which it is a of course the word is coming which is only temperature dependent then it will cover every area if it is mass dependent how much gas reaches then it depends on if you have to go in all mass cannot go in okay so there is an issue whenever there is a mass transfers there will be always less step coverage if there is a temperature reaction limited then everywhere everything can go here is what I am saying all that is it a typical deposition rate which is called v y of a CVD process at its temperature dependence is shown here sometimes of course instead of one this I may actually give you 1000 by T all said temperatures are measured in Kelvin's please remember unless specified if I give you 27 degrees centigrade make it 300 degrees Kelvin so if I plot 1000 by T versus so 1000 by T means T which side increasing T increasing on this side okay so if one it is found that at higher temperatures somewhere the process of deposition is normally controlled by mass transfers and relatively lower temperatures relatively not very low it is limited by what we called as reaction rate limited this names do you get somewhere from gradle model okay same deposition rate ln is law okay you forget about because it is a lock scale so it is a v y v y is the deposition rate okay deposition rate v y the y word was used because of excess people are using x y so it is just a matter of y is not necessary because they will show you some directions I also use the same method okay so is that clear so the point which I am trying to create is the following so before I go to the actual AP CVD and others typical CVD reactions can be shown in this figure for AP CVD in specific okay the for AP CVD system or maybe I will first show you AP CVD so that you will appreciate what I am talking this figure has been taken from Doe Corning's paper okay it is a copyright or this equipment is a copyright of Doe Corning why Doe Corning became very unfamous in India why Doe chemical became so bad because they took over union carbide which created all problems in Bhopal okay so here is a typical AP CVD system we will show you some more figures on that there is a heater here over which wafers are kept vertical then there is a nitrogen gas process gas hydrogen gas or nitrogen gas many gases they will exhaust and this is little higher throughput machine that is a standard machine what is this circle shown here please and say conveyor belt so it is in the chamber so wafers come they deposit go down keep going through this lower side so this is slightly higher input this is of course required in EP and I will come back to it later what is what are typical steps in AP CVD reactant reaches deposition region where substrate are horizontally kept is that clear the first thing is in AP CVD wafers are kept horizontal from the ambient gas stream the reactants diffuse through boundary layer and that is something word which we earlier we discussed in crystal growth there is a stagnant layer which is called boundary layer okay which is the region which is between the gas stream and the wafer there is a region where nothing is actually happening okay or we say velocity is balanced the stream velocity with the surface velocities are balanced okay is that okay first stage they come on the wafers are kept horizontal the gases come and moves over them okay and the way it is and that is figure we will draw little later there is a boundary layer between the gas stream and the wafer so any reaction has to take place and that is where that figure will be better off okay before I come back here is the continuous gas flow this is your substrate this layer of atoms is called boundary layer they do not move very much okay so new species have to actually diffuse through this gas stream molecules or atoms actually than to deposit is that clear there is a boundary layer of gas which is oh because of the pressure to adjust and stream velocity to adjust there is this layer above this which is constant typically this is a function of gas stream velocity but it is typically constant in its thickness generally so all the diffusion the what is the reaction I am talking for this silicon depositions or polysilicon silane is the gas I used which gas silane I oxidize it with oxygen to create SiO2 and gas H2 is exhausted okay this SiO2 also has another this if you want then SiH4 plus SiH2 gas this is called di-silane H2 gas may create H2 and polysilicon this is at low temperature that is at high temperature this is at high temperature this is at low temperature thermodynamics is avoided I just gave the final result for you so this is continuous gas flow there is a boundary layer reactant atoms goes through this and then starts settling there now when they settle there are many things whenever atom sits somewhere here it should ask sticking so what is the sticking related to one is called surface binding energy okay the other is it should be able to move ahead because the next atom has to come it is called volume energy these two when it is called nucleation sites so when first it should create nucleation sites so it must have enough surface binding energy so that atom can sit there and it should move so it should have some energy which is which should overcome the volume energy and then only they will move so there are few energies involved which thermodynamically found how much is available for each gas stream and each gas atoms okay now this process which I was discussing just now is the reactor the reactants are then adsorbed at the wafer this word absorption and adsorption is slightly different what is the difference adsorption is also at the surface where is it then no no it is called inside the surface it is still on the surface but inside the surface addition and subtraction is the word used in absorbed and adsorbed okay the reactants are then adsorbed at the wafer surface reactions occur here the chemical reactants take place reactions takes place deposition occurs conformally what is conformally step host step per v why because sites are where where it finds sites only there it can deposit so and also it will be simultaneously something will be emitted not all will be sticking there some will come out and some will also re-deposit okay so there is some kind of equilibrium will be attained in the system unreacted reactants and the bribe products are dissolved dissolved means they leave the surface okay they transport through the boundary layer and are exhausted out of the system we can have kinetics of such depositions similar to Goldie model I will just do this today and then maybe I have a boundary layer some gas stream is coming and reacting same thing only difference in grade in my there was a thickness of oxide through which oxidant was diffusing there is nothing called such in between okay only boundary layer which is also gas okay so that is the only difference is that correct there is no intermediate oxide layer so how many fluxes will be there too there was third flux there is not existing just note down and then I will show you the mass for the day and we will finish today for that all the reactants and byproducts all gases are not reacting so the some part will just go some byproducts like oxygen hydrogen if I use trichloride if I use silicon tetrachloride what will be the byproduct HCl so there will be always a byproduct of the system okay is that is that fine we assume normally that the boundary layer is fixed okay but it can be a function and can be solved also but right now I say delta X but then I say it is delta but as such delta X is not uniform okay is that okay to all of you is that okay so here is that simple kinetics which we do we have a gas stream we have a silicon substrate CG is the gas stream concentration and CS is the corresponding gas stream concentration gas concentration at the surface and this delta X or delta SH I am going to use constant later is the boundary layer thickness so anything which is coming from here has to go through the boundary and not all will reach because there will be diffusivity involved there okay so CG to CS and from here whatever flux available will react with silicon or deposit on silicon to create the new atomic layer okay so we have two fluxes to offer the diffusion flux of reactant to the wafer this process is mass transfer limit why it is called mass transfer available gas concentration is proportion to partial pressure so how much is available is decided by the partial pressure Henry's Lathropod of itself so here CG I have a F1 flux say then F2 is the flux which is reacting or depositing system okay which is surface reaction is that clear this SI creation there from SI H4 or SI CL4 at where at the site if you take away wafer the reaction can take also in the gas stream sometimes and we must avoid that it should always occur at surface okay. So we have two fluxes to matter we say F1 which is mass transfer limited HG is called mass transfer coefficient units of centimeter per second F1 is HG times CG minus CS and F2 is reaction rate constant KS times available concentration CS and in steady state F is equal to F1 equal to F2 same as grade will my what is the difference there there is no oxide layer where further third flux is occurring okay if F1 is equal to F2 these two are equal steady state both fluxes must match okay maths trivial maths so if I equate them and collect the terms I can write CS is equal to HG upon HG plus KS into CG or one upon one plus KS by HG into CS or in a this one plus KS by HG to the power minus 1 into CG. So I am now related CS and CG through what terms mass transfer coefficient and reaction rate constantly okay the growth rate as we did the Dx Dx ox by dt is similar here here thickness grown or deposition rate growth and deposition are slightly used in same fashion but they are different there is no growth that is no silicon is consumed here okay but assume that deposition and growth I am talking in same terms okay which is flux divided by number of available atoms which are incorporated on surface then VY I use F from here okay one of this case KS times CS is flux so KS times CS is flux divided by N okay now there is another term which we use in actually this there is a CT denote the concentration of all molecules in the gas phase in total chambers okay CG is the fraction of CT which is available for diffusion is that clear. So its ratio how much is available and how much is going down is called mole fraction ideal gas law. Please remember CT is in the whole system and CG is where actually reaction is going to take how much is coming in okay. So this is called mole fraction Y which is CG divided by CT why I am interested in this CT is proportional to what the actual pressure with which gas will be introduced how many liters per second and what pressure I know how much gas I am pushing it this is measurable quantity CT is measurable quantity okay is that point clear if I know CT and I know why how much is the reaction I can deny how much is the mole fraction I am having then I will be able to correlate the external gas flows with the growths okay okay is it okay is that okay. So why is CG by CT but we know by partial pressure law mass law this is proportioned to PG by PT PG is let us say if I am using trichlorosilane as the gas and the reaction which I used to show this is for epitaxial growth silicon tetrachloride plus 2 H2 at high temperatures reduces to reacts to give silicon plus SCL and whether this reaction will be favored or silicon may react back with SCL and give me silicon tetrachloride is decided by thermodynamics okay at delta G is positive or negative delta S is the change in entropy at temperature T and delta H is the entropy if this is this if positive this will be forward reaction if it is minus it is a reverse reactions okay. So you must adjust your T and the concentration such that forward reaction is favored okay so if I am now see why which is proportional to PG by PT then I say VY is KS HG upon KS plus HG CT by N into Y okay now there are 2 possibilities this is the growth rate of deposition rate what is this one of KS HG by KS plus HG means what is essentially I am talking is 1 upon KS plus 1 upon HG I am talking something like this is it so now because this lower portion will be much easier to understand if I say case is much smaller than HG case is much smaller than HG which means HG HG can cancel and you get VY is KS CT Y by N okay. So what is it limiting KS KS is related to what quantity temperature e to the power minus E by KT so if you can make the process in which KS is much smaller than H then the reaction will be governed the deposition will be governed by only temperature is that rest remaining constant only temperature but if I make a case in which HG is smaller than KS then the process will be VY HG CT Y by N HG is decided by what we called as mass transfer on what basis what is the equivalent it can be in a figure which I have shown you there is a boundary layer and reactant is diffusing through so what will be roughly HG value D by delta is essentially called mass transfer coefficient HG diffusivity divided by delta the boundary layer thickness okay. So what are the 2 equations are trying to say by adjusting whichever parameter you wish you can either make a process of deposition which is function only of temperatures or which is a function of the available mass mass transfer LPCVD uses the 1st case in which HG is much larger than KS APCVD uses the 2nd case in which HG is much smaller than KS and therefore HG its mass transfer limited means therefore has to be then kept flat because then only it can create a boundary layer and reactant can go in the case of LPCVD it is only function of temperature so how how efforts can be kept even like this as long as their temperature is known I know how much growth it will have deposition so now is that point clear what I said in LPCVD their throughputs are very high because they are only temperature dependent processes all APCVDs are mass transfer limited process so wafers have to be kept flat okay and then their throughput is less because susceptor cannot have hundreds of wafers of course that belt system does improve the throughput but still it is limited by the amount it will take how much belt you cannot have of course belt is right now 14 feet okay normally Intel has a process which has 14 feet belt which holds around 12 inch wafers around 12 all that said okay so 12 wafers see at a time built wafers whole chamber may entry hothay right but attack bar so there is a retention time you can see on the monitor what is happening it will say end wafers will come out never wafers will start coming so if you of course unless you join Intel Portland or Oregon Oregon elsewhere you will never be able to see the real life process try joining TSMC but in TSMC also if you go I do not know how many will go someday if you go you will be restricted to some 100 feet by 100 feet area you cannot go other groups okay so tomorrow we will start with tomorrow means Friday we will discuss more about the each process and do EP growths okay and then finish the this and start etching later.