 So we start, we will just go back few slides. We were talking yesterday start with some contrast and then went over but the few things which I forgot to mention, we did say that PPR or NPR the fraction of resist after exposure, it may start at initial dose of T0 and may end up in DF when all of PPR is exposed or start from D0 to start exposing and finish at DF and we define a term gamma which is called the contrast factor and we said it is 1 upon log of this. However one fact which I forgot to tell, it seems as if DF and D0 are constant for a resist. It is not really and that is why I thought I should rewrite. I wrote this line afterwards after I saw in the book. Gamma is really not a constant even for a resist. It varies with process parameters like chemistry involved, baking times, baking temperatures, before and after exposures, underlying layers and wear lengths. So it is not that for a given resist DF by DF, D0 ratio is fixed. It can vary with so many of them and we are expecting gamma to be higher enough such that the required DF values are not very high and still it is able to expose all the resist. This is what yesterday I, these lines I wrote yesterday and I thought I should show you. There is also a interesting feature which is related to contrast which we will see now. I keep repeating my problem is essentially the sufficient intensity of light should get inside and get inside as early as possible so that full resist is exposed and the amount of days required to do this should be smaller. That is what I am really looking for in lithography. How good I do it is what my expertise. So this is of course as I said I read yesterday the book again and I realized that these are issues which you should know that gamma is not really a constant. Though to a great extent for a given all these parameters gamma will be a constant. Is that clear? So if your time lithography is fixed, so okay gamma will be fixed but otherwise gamma is a function. If I change the pre-wet temperature or post-wet as we shall see, this gamma may vary. So there is a possibility of improving gamma by this is what I wanted to stress. So it is not just that I chose one kind of resistance I am through. I cannot then help. So I have a possibility of playing with gamma values as well. Is that point clear to you? So why suddenly I said 5 to 10. So you may ask so if this is fixed what is 5 to 10? Yes I can do some machine and I can get little higher gammas if I need so that I have better resolution or better contrast. This is something which last I forgot. This of course we did but what we did not show I think that slide was double it just got mixed up there so that slide was there. The slideshow was showing that there is a mask, good mask. I put it on a resist and depends on the exposure dose and position. Position is along x axis from here to here. I figure it out that depending on the thickness of resist all other parameters this upper portion you know you can see this of course the other portion is not shown here. This is with the second metal line or second mask area. So let us say if this is your corresponding to this, this is your aerial image pattern for a given exposure and this is your dfd0, this is d0, this is df. And if you have another area which is under mask and it is so that the aerial image has a lower intensity and this can happen because of the thickness of resist not being constant. In that case one can see it may be little flattening up and this hatched areas are essentially called partially exposed resist areas. You are going from d0 to df or df to d0 or this. In one case you can see the partially exposed areas are larger. Ideally it should be 0. I want sharp this and that may not happen. I may have some gray areas as shown here and this happens reality why because on the wafer the thickness of resist is not uniform throughout the wafer. So there is a possibility that you may actually get gamma differently at different positions because dfd0 to shaded areas may be different points. And this means some contrast will be lost even if you have everything good but there is a possibility the mask may not be the culprit, the culprit may be the resist thickness. But these are all issues because at the end of the day I cannot blame anyone. If it does not work someone else will blame me. So obviously I must prior you know what could I could be done so that this is minimized. Is that clear? Please remember every time I am looking for a smaller dose however at the same time I want full exposure to the resist thickness. Is that clear? So of course there are tricks and trade of how to get rid of even if I make mistake I can slightly correct it and that can be shown little later. We already looked into the drawbacks of yesterday this was shown and also we showed the actual alignment system. These are some of it. So now we finally show you the actual lithography sequence. Yesterday we did show you a diode we made but here is exact process which is given in a book given in many journal papers and it is available everywhere wherever you wish and maybe if I see this lady wearing an INUP so must have heard in INUP end times. The lithography process I now create some kind of sequence of processes which leads to a good lithography. The first thing in the process of doing lithography is cleaning the wafer. It is most important. There must be some oxide or some other material which you want to in which you want to create windows but the surface of whatever you have has to be very good clean. So you may actually go through RCA clean and dry it very heavily. This drying process is also very very important because this is called surface preparation prior to coating of the resist. If there is a moisture resist may actually crumble or sometime does not stick. So because of that the drying process is very crucial. Normally run in a little higher temperature ovens and oven is running with a huge amount of nitrogen flow inside. Typical flow in a normal ovens is around few liters per minute but we need at least 30 liters per minute flow inside the oven chambers so that there is no oxidation of wafers and there is no dust going on it. Once the wafer sometime even the furnace itself we may put and take it out but if there is a metal do not put it into furnace. If it is oxide nitride or anything yes you can push after for drying itself in the heated furnace at the age and nitrogen is anyway passing in that. So you can put it there for a while and then take it for masking okay. Then we apply what is called as primer. Primer is the sticking material. I hope many of you have heard about this some home many of I do not know any of you have seen how the pens are put on the walls or any furniture. They always coat with a primer okay. The primer is essentially is the creator of a bond between the resist and the layer which you want to like SiO2 or nitride. So if I put some primer layer it will allow resist to stick and it will also allow SiO2 to stick with the primer. So of course a monolayer is put so it is also applied by putting into a chuck and rotating it okay or spinning it. The primer generally used is hexamethyl disilane HDMS very famously known HDMS and layer typically reacts the kind of thing which I am looking for this is SiO2 this is photo resist and there is a primer in between which bonds okay. Now this primer application is only for resist adhesion. So otherwise resist will rule really you have not seen it but resist film is coated and after the while it actually rules okay. So everything what you did is lost half way okay. So we normally wish to see that it sticks well okay. Then of course we after the primer we coat the wafer with resist and normally there is a different dispensation earlier we used to put through a series of drops now there are a formal diffusing dispensers which actually put sufficient small drops around all the wafers and there is a chuck which is rotating bulb and the dispenser is on the top so fixed amount of resist is dispensed on the wafer and then we give it spin which may be of the order of 3000 to 8000 rotations per minute okay. Now no no no when you go to remove the resist primer will go actually then RC I know you will have every time you will go for a cleaning so that everything will get cleaned up in RC everything there is HCL and H2O so forth everything will be coming out. Every process the wafer should be fresh the last process you finish you go to cleaning then you will actually remove everything when you strip your resist you will strip everything so of course this depending on the viscosity of the resist used and the thickness you are trying to build this thickness is decided by feature sizes and thickness is also decided by viscosity and as well as the kind of source you are going to use expose and kind of lithography three of them will show you. So this layer could be please remember why resist layer is important because resist when even if it is hardened somewhere by after exposure it should not get etched out when the other part which is non resist soft part is being etched that resist should not actually itself come out okay so it should stop etchants actually attacking the surface below so you need to have sometime thicker resist so that it does not attack layers below but to a thick two thicker resist actually rolls it does not remain enough sticking coefficient for it so it rolls. So one has to worry how much thickness for a given kind of films by experience okay typical thickness could be 3000 Armstrong to a 10000 Armstrong then the most important step in the case of resist after before the exposure is given is pre-bake this is very very important because of many things are happening one is of course that since the resist has a solvent and we want a film which has solvent free only resist part so I want to remove that solvent so I heat the wafers or either give what called pre-baking and pre in the sense pre to the exposure okay. So I actually bake it around 90 degree to 100 degree for 10 to 30 minutes depending on the resist and depending on the thicknesses you have deposited this time may vary. The first advantage is it removes solvents and it also has an advantage of releasing stress any thermal this is called anneals and it releases stress in the material okay so it is stress release so the film is more uniform it does not crumble so it needs to have some pre-bake. Now I will just think of it I just put a question mark what if I do it at 120 or what if I do it at 60 degree what if I do at higher times and what if I do at lower times possible four combinations why do I choose some combination which suits me okay think of it one example I can give you if I actually bake at higher temperatures the upper surface will get hardened faster okay because it will dry faster but lower one will not so there will be some kind of a gel below the hard surface which will not get exposed okay so one example I give you now think of it similar things four possible combinations higher temperature higher times higher temperature lower times lower time lower temperature higher and lower time think of it why only some selective combinations work because there is a possibility of resist not getting properly exposed okay so I one I itself the other three you look for it okay think of it there is nothing great and why I am leaving it maybe I can ask then okay so of course these are all experiences you do I mean and you really are not happening well so think over it why it has happened then you create some solution for yourself and then you find out works okay so this is not something which is very popularly given in books but these are something experience we know how much I do so for example this temperatures are also not very sank through sank for all resist okay some may do at 85 but some other time some may do at 95 some other time so there is a combination but there is a maximum min bars which you have to follow in either case okay the next step of course is the alignment which is the most important by the pre-baked wafer is the then put on a chuck which is the mask aligner chuck yesterday I showed you and over which the mass plate is actually sitting okay and there is a gap between the mass plate and the wafer there is a gap between this may be 50 microns or even lower but they are not touching is that point clear they are not touching what is the why they are not touching because when I move one of them refer related to each other they should not scratch each other so there is a gap between them so that I can move wafer with reference to mask or mask with reference for alignment of patterns now many a times there are alignment patterns we create and these are created by designers they have they have been told to do this which I will show you quickly just a minute I will come back to that what we do is there are alignment marks the first alignment mark may be something like this this is blown up version not necessarily that big this is put on four corners sometimes in the edges of this this is one pattern we create sometimes the pattern could be like this sometimes the pattern could be pattern could be like this these are called alignment marks okay so these are for what is called as global aligning since each mask corners will have these some kind of alignment marks and in the last month they will be similar this so if I am for the second mask it must be either if you want to cover it it should be either this or it may be for depends on what being h 10 what being retail the other pattern may be inside or outside depends on whether you are covering something or you are etching something so the next alignment mark on the second mass will be correspondingly same position either slightly lower or slightly higher depending on what patterning you are doing and these alignment marks are created by designers during the layouts that they provide you everywhere this because layout I am not creating layout must come from design side so these are the patterns which they create and these are major corners or something sometimes in the side edges as well so when I align the first time I align is this alignment marks which is called course alignment I don't see patterns I only see these two corners and four corners or six corners six areas I align so major wafer may get aligned but the problem is if you have a wafer let us say like this sorry your mass may be like this this area still may not get aligned you may have on some chips you saw the alignment went and some area may not be so you have one has to guarantee that full wafer corners so not only on chip you align but also align on edges of the wafer those chips so that roughly you know your mask and your wafer which has earlier some prints is actually now just below that this can be done by X square column in this stage or it can also be done optically by mirrors but that we see later now this alignment once done then individual wafer some X and some Y's are actually chosen and their finer alignments are done from corner to corner okay we believe that if two diagonals are matching then most of the time wafer will match okay the diagonals match so the wafer must have got aligned now the problem starts in most cases the kind of lithography I do is called contact printing I have after this alignment is done which may be 25 micron mask away I bring it down and allow the mask to touch to the wafer the emulsion portion is below that means there is no gap between the mask and the initial pattern between the resist okay now there is a problem there itself because in any XYZ system it is mechanical there is always backlashes okay or there is what is called inertial motions so if I have if I have a wafer and I am putting this on the top of it I thought it is I will go like this but actually I may go like this I may go like this okay so there is a error and in general this error is many a time systematic errors unless you every time change the direction of X and Y this you know roughly which side it goes okay so apparently your initial misalignment itself is to be such that when I will stick it will get aligned if it does not you come back okay in the align keep doing after a while you realize how to align okay it takes months sometime years to learn exact alignments for a smaller patterns every time you do you scratch something you go out of it some chips get aligned some chips do not but that is only experience and you know how to align large number of chips in one wafer and number of such wafers and therefore these are tricks which as I say I cannot just say do it try it yourself and learn it okay so this is called alignment typically the ultraviolet light source energy or source intensity is 150 millijoules per centimetres for DNQs and 20 to 40 millijoules per centimetre square for DUVs so this then we expose once we align we expose so wherever dark portions are there on the mask light will not pass through wherever clear portions are light will pass through depending on the resist either it will get hardened clear areas or it will become softened if it is PPR it is softens if it is NPR it is hardened okay so this is important step this is essentially what the major step is in all the business how to put patterns and this how many times I may have to do in a diode four times and for a C mask it may be as I say 16 to 24 mask or even higher iso C mask process may require 32 mask and current Intel processor is requiring 38 mask so it is some kind of FinFET based extreme controls side walls so many things we are trying there so it requires 36 mask or 38 mask if you have wireless chip you may have further 4 mask whether the lengths you have to adjust for the RF lines so you have even 40 process for a wireless receiver chip or using as even standard Intel process so these are all you must understand every masking state will create some error is that clear so the next masking should take care of that error itself otherwise error may start building and finally it may not get aligned anywhere is that clear so there is a trick of so please remember this the way we do it is something like this to avoid some errors first time I create some patterns next time these 2 may be aligned and I may create the new pattern on the next mask and to this when the 4th mask comes I will align first with them and then align with the earlier ones so I keep creating a new pattern and all once I keep aligning and that is how I can minimize my errors if possible and that is the trick which and so therefore you must remember every mask where to create the new pattern and where to keep the old patterns match so this is all this called test area designs and sorry frame designs and frame designs are very crucial at the end in chips designer may look into only circuit layout but if this is not there things may never work for him so some experience of knowing technology problems is essential for designers of course which you are anyway experts the game in internet circuit is something like this that's what all of us wish to do it but sometimes you have to do novelty as well okay after the exposure is done the resist has to be please remember you are shining light and your assumption was that the light did not pass through the dark regions yeah it may not pass through dark regions that's true okay but if this is your window let's say you thought light did not go I mean did not pass through this but light can pass through edges this is diffractions okay and this is major worry in our case okay so the actual pattern to pattern which you are doing are not necessarily same this is isotropic it's a diffraction area and we'll see that that is how the different alignment system does how to minimize diffraction problems okay so please remember that masking is not a very trivial issue it's a very important issue and the proof that your masking was good at the end chip works and if it doesn't work there is some way your masking has a problem or maybe patterns created by circuit people may be a problem now the purpose of this post exposure bake is typically this is also around 90 degree to 100 degree sometimes from 20 degrees depending on the kind of resist you have if it is duv then it is this yesterday say the acid generator photo acid generator it reacts with polymer because you are released you are just bonded it reacted and polymer but it has to come back so this catalytic reaction can now take place during this post bake system so as it is released again it also allows you to some kind of reducing the standing wave patterns because the reflections are because the density of resist becomes higher so reflections are minimal relatively smaller and therefore one says that post bake reduces standing waves and post bake also reduces release acids this is very important step if this is not properly baked when you develop resist from everywhere resist will go away okay so this post bake for 30 minutes 100 to 120 degree some may require one some resist require 150 degree hard bakes okay so it depends different resist people then after resist post bake is given we actually put the wafers either by spray development there is a gun which sprays the developer or in earlier time we used to actually do the wafers itself in the bath so you have a number of racks wafers on the rack and there is a bath you just dip there okay either way but now it is mostly spray developments it takes roughly 30 to 60 seconds at room temperature for resist to develop what do in the resist to develop the soft portions of resist is solved dissolved okay and hard portion is not attacked so here is again if I do too long what can happen I will just show you I will come back to it this is my resist area which is hard okay this is my soft so this was getting aged or developed but part of the developer can get inside and actually lift the hardened resist okay so it is too long development can also actually pull the actually hardened resist areas and therefore the 30 second is not that we figure out it is enough that the hard portion remains there itself otherwise from below something like this the developer goes through below and it just takes away so in the plain wafer will come there is no resist anywhere okay so this tricks of time and temperature is very crucial and thickness how much you have put that decides the time and temperatures normally room temperature is good enough and therefore the lifts process is smaller if it is higher temperature it will lift definitely this technique has been used in what we call shadow lithography but some other day okay lift something going wrong I used it for my advantage okay then I rinse because there is a particles of resist sitting in the clear areas so I want to remove all because that is the clear area everything has to be etched later so I want to remove any resist particles sitting on the portion from where resist has been taken away or developed out so this has to be rinsed and also there is many a times it is not necessarily only in water they are fixers like in the case of photography we used to put fixing in hypo what is hypo have you heard of the word hypo you are right but why it was called hypo because initially when it was people thought it is sodium hypo sulphide so it was given a name hypo but later on it figured out it is the thiosulphide the it is more acidic actually so then they figured out it's not hypo okay but that name stuck okay so history hi time millita so anyway you this and then after the PR resist particles are removed you clean it and then you develop the areas spin right completely then actually you post baking because I just now said there is a possibility of ages slightly getting lifted okay so you want to stick them back so that actually etching of oxide and nitride during that time HF may go in actually HF is a very strong HN and it will just gets in okay as soon as it sees SiO2 below it will just walk in okay so there the age has to be very strict the age is very important than the rest if you see here the resist may actually enter faster here okay where there is oxide below here you only want this oxide to go but actually it will lift this so all that pattern which you are trying will go away okay after the resist because you are in a solution which is solving resist so it may also attack the age part easily where age has the maximum what we call stress areas so the resist can penetrate in the particles there so there we must harden it again so that resist is not that strong HN there because a hardened resist but HF or any other HN for the other surface may actually enter because this it not only removes this SiO2 it will also remove sideways also so you are creating windows in nothing because there is no oxide anywhere okay so you find there is no pattern okay then stripping will be a problem nothing great otherwise happens so it is typically the bakes are given for 10 to 30 minutes 100 to 150 degree centigrade here also the earlier version also I put a question mark if you rinse it for too long or if you rinse for very short time what happens okay these are all issues you answer okay so after photo resist is hardened from the edges once again then you H put this wafers into etching solution for the layer which you want to H like SiO2 nitride metal whatever you are which is your patterning that region should be now that HN should be put in the material so for example nitride essentially goes normally not in the normal HF but you have to need sodium not sodium ammonium fluoride mixed with HF probably can etch with some one drop of nitric acid probably can etch nitrides so there are different etchants for different materials at different concentrations so one has to look at it which is relatively strong but not very strong that is the way so once the etching of area is desired which can be oxide nitride poly metal whatever it is and that area is then etched out and once you put everything in this all it has become messy so you must rinse and clean the wafer once again is that madam clear every time you go you rinse the wafer completely again go through RCA clean okay but before going to RCA right now you only rinse in water or xylene many times you this and then remove the resist how to remove the resist we already say some TMHA TMH at higher temperature itself can act like a stripper the best stripper is nitric acid what is the nitric acid we do there is something called fuming nitric acid what is fuming nitric acid anyone anyone heard of this chemistry fuming nitric acid fuming so if you take if you have seen the nitric acid it constantly brown vapors keep coming so why all others are not called fuming and some this so if there is a copious fumes coming they can only occur if the concentration of HN nitric acid is 48 percent which is maximum possible so if it is one N solution then the fumes coming from nitric which is the most concentrated nitric acid normally we 10 percent we use nitric acid thin I mean this diluted by 10 percent 10 times water but here we use 100 percent nitric acid which is actually 48 percent which is called maximum possible nitric acid concentration you actually keep your wafers on the wafers itself that it keeps fumes so wafers are held in the fumes resist is attacked very fast however many people don't like acids so they have some organics which are called sopes snoopies okay they also alkaline materials and they also can strip the resist okay so some acetone can be used for a PPR also okay so once so earlier during the etching of SiO2 nitride or any other I did not want resist to be removed I want that to stick because area wise I was etching but once I am over I want resist to go away because resist is a carbon which I don't want to be around okay so I must remove all of it as much as possible and then give full RCA clean to the wafer except if it is metal don't go through acids okay because metal will also go okay but otherwise for oxides and nitride go through full RCA clean once again and wafers are ready for next process next diffusion next implant next whatever you are doing you do that and come back for the next lithography keep this keeps doing every time you do a process selective lithography do process keep doing till all of the masks are consumed and the circuit is completely available on silicon that's what lithography is all about okay so we have now seen the procedure of doing actual lithography yesterday I showed you diode also I can show you for mask as well and why I am doing it then I am at least the plumbers 16 mask standard CMOS process I want to explain you later so there at time you shouldn't ask I will only say lithography done okay do you want to say how so that is why I keep saying this you learn because there we will only use all the processes which we learn we know how they are done what is the physics behind what is the chemistry behind and once I know what is happening then I will say okay do this do this do this use this mask only I show you mask use this mask I get this pattern if I h and this I get this 16 times I go through such lithography and I realize standard CMOS okay that is what the 16 mask process is so if I do Finfeld it may be 26 24 26 mask if you only double gate it may require some 21 or 22 mask so different kind of structures if surround Finfeld's if you are working it may be 30 odd mask so there are structures which are different and require different maskings now we come to so far we looked into resist and we looked into mass design now we look into the exposure systems there are three exposure possibilities one of course is very pop just now I showed told you all the time was contact printing essentially contact printing means the this is my optical system which focuses the source which is a point source shown here ultraviolet source and this comes in this such an angle such that this becomes parallel the distance is adjusted so that if this is kept at the focal point the beams comes out in parallel so I put it into beams are coming parallel to the direction and they pass through mask patterns the light passes through mask patterns and then attacks resist what is the difference between other things this here the mask and wafers are touching contact okay this is contact printing this is like a photography in photography so called negatives are actually connect test with a photo paper and actually exposed so whether black it becomes why whether why it becomes black that's the photograph okay so this is essentially contact printing the word has been taken from photography contact prints you go ask a sketch are contact printing exactly like this so the emulsion side is on the side of this resist so it touches and that is one problem I last time said that if keep every now and then if you touch the mask to a resist to a different wafers because the mask is costly so you keep doing after sometime the resist I mean the resist met attached to the mask or mask emulsion may go and therefore it cannot be used after few amount numbers of lithographies which cost goes to you then why this is so important yeah it is it is still used once a while since this does not have any great attachments actually have a optical system which is sometime multiple convex with convex or convex lens for focusing properly or concave convex lens or plano convex there are different kinds of lenses which focuses the beams or by focus beam comes to the paddle okay so adjust focal lengths this is relatively cheaper 300 to 4,400 thousand dollars which people think cheap okay total salary whenever I listen 171.76 lakh crore loss I just count the number I can't count is that okay so what is contact print the mask is touching the wafer or resist okay then you shine the light it will give very good result we will show you later because whatever age you are saying light possibly can go there easily okay so it may be one of the best lithography possible but it has its own problems is it okay as mask is contact with resist it is called you know you can see since this age is it is a very sharp image can be transferred however the it is a very high resolution resist process actually sharp mostly sharp however as I keep touching the mask will goes bad as well as therefore may go bad okay resist may be not straight all the time every area in some area resist thickness and mass thickness may not be touching so it may have another problem create so contact printing is better only when the patterns are large then it is 100% transfers okay no errors smaller patterns do not try contacts we will show you what we will do then see second possibility is separate the mask from wafer or resist okay by some small amount this is called gap distance it may be 50 microns it can be 25 microns it can be even 100 microns in some proximity aligners okay but 100 is normally not favored typically it is 25 to 50 micron separation between the mass plate and the resist well so this is gap between resist and the mass plate again it only can do large features can be done through this kind of this I will show you what numbers one get actual values okay it has a relatively poor resolutions okay and the cost is slightly more it is 1000 kilo okay 400 say of 1000 okay that is 1 million 1 million dollars okay in our lab if you have a MJB some old SWS kind of companies aligner in those days is to cost 4 lakhs of rupees now such mass aligners are not even available okay but that time my patterns were 5 microns so it was perfect everything went well okay both places some resist may stick to the mask some so since its densities are different the reflect index are different so the next time when you use it will create a problem secondly if I remove the resist for some part the exposure thickness will be varying so it may spoil the image the present one and it will spoil the image for the second one okay is that clear so avoid it if possible touching be avoided the problem is if it exactly sits like this it does not happen so much both are dried and everything but as I told you what will happen it slide on that that may be 2 microns slide but it will remove that much area from there at times okay and that is where the problem starts but one can see if it is a large area okay so this is called proximity aligners and these are very of course very few places it is still used like you are using HEMTs or you are you are the number of transistors are 10 20 we are looking for an IC which has million or billion transistors such processes will never be used so I know there are certain people who are still working on different things and buying a huge cost of projection printer which will show which may be around 10 million to 20 millions is worthless because you are not even doing 10 papers in next one year so putting so much money is best okay so buy cheaper one play some games look for results okay and then you are publishing so a one bit okay but in industry it is numbers so there you need to have a proximity aligners okay so the last and not the one which is mostly used uh actually the proximity printers can do as much as one microns one to five microns okay but below one micron certainly is not possible accurately I mean you put it something will happen and there are games of retrieving as well maybe some other day when you know something called necessity is mother of inventions in my time many things were not available but we still did it okay like when I first time did my phd when during mine there was no matrix inversion program in any of the libraries of computers which we have we are 16 20 first IBM computers so I wrote my matrix inversion myself okay I learned a lot when how to write a program for such a large arrays the problem came that the memory of those systems used to be 32k usable memory that 32k soon cake it now but I like them mega giga or a k so no arrays can be put there so you think a scan karo us go release karo the first I use karnai first I will be low so you keep doing that job and times but don't store okay up here be on a key and so I can so it is not that we couldn't solve a matrix the system didn't allow us so yeah when you think that can't be done probably your ingenuity plays so don't think that earlier time we could never do anything we still could do better as good in those days possible the cost money was very crucial for us a proximity aligners has a typically image reduction system which is 4 to 5 x initial pattern which can be reduced to 1x it has two optical areas one is essentially the collimator and then there is a mask and then through another optical collimating system it is focused on the silicon wafer okay so mask is far away from the wafer actually mask is far away from the wafer it's a very fantastic it seems to be otherwise but it is very high resolution process it creates very very low defect densities because it doesn't touch anything anywhere okay the only problem it gives is its number of wafer per hour are pure compared to counterparts are fastest thug thug thug a process line but if you see a conveyor belt there are exposure systems the wafer come thug wafer next next thug it keeps moving okay this is not possible in approximate others projection aligners actually they do some four or five at a time but that's that okay so this is throughput could be as low as 25 to 50 wafer another whereas a counter printers can give you 2000 wafer in a day okay so that's the cost but at the cost is resolutions this machine is typically 10 millions and this price is of 2002 now with all inflation and everything the why the cost will not be lower because not that many foundries are buying them very few people are buying it so even more cost because more and more is the one founded closer is the next person has to pay more money because he has invested so he will charge the second vendor even more okay so that's the game so actually prices never go down in case of any of this semiconductor systems because more and more companies are folding so those who are sticking has to pay more I invested okay so this is a interesting projection printing now all is fine if the dimensions of the mask patterns which you are looking a smaller a larger up to 0.13 microns all everything looks so fine one can say something which is now I will come back to and which is most important part for our lithography some optics because everywhere we are showing you optics so what is the optics playing part in this whole game yeah this figure is available in this these are the features high relation low defect density low throughput higher cost okay now these issues are very crucial for us if the dimensions of the object and apertures are large we can always use what is called as geometric optics or the ray nature of light is ray travels okay however if the aperture is small and the light then at the edges of the apertures light beam diffracts and this diffraction mechanism can be explained by saying that light beam has wave nature okay it is not the photon beam but it is it has a wave kind of structure what is this law called de Broglie's law okay every matter can be thought of as waves okay so what is the why de Broglie's was also famous firstly he was count that means he was a one of the king's family in France in those days actually there were seven de Broglie's three one the noble price okay of course because they were counts so no one could refuse them okay of course they did great so this de Broglie's PhD thesis that the waves are matter is matter can be explained as waves actually has only one line in his PhD thesis and he was awarded doctorate okay so we need we had to leave geometric optics and we actually go to wave optics just for those of course this is also given in the book just for that all the point sources actually inmates the EM waves okay and this is the direction of propagation so it is electric vector and this vector is normal to the propagation EH and then third Z is the propagation so the EM waves theory same wave theory so if if the propagation is this these overlapping waves around are called wavelets or final wave rates okay these travel and actually hit the object okay and depending if they are also small objects there if they also form similar wavelets around those object points and this is my image and since if they are very close to each other there is a overlapping wavelet patterns so you actually see a full pattern so whole of the object is seen as one unit if this point are very close to each other and there is overlapping eigenfunctions okay this is very theory so if that happens you see this image however if this object is not the kind shown here but has a slit which is called which has an aperture so this is small aperture then when the wavelets travel here as in this this center ones of course may go straight because they are unhindered but the edges do get some kind of edge effects okay and that is essentially word use is diffraction okay. So now the wavelets travel upwards this wavelet travels downwards so the image has right now because of the aperture this object is only now restricted to this and is not same as what you thought okay so diffraction pattern actually modifies your object image of the object and that happens because of the wavelets trans going from aperture to the image plane what is image plane in our case in this resist there the light is actually coming so that is the resist plane okay what is object the mask the mask which you have dark and white portion is the plane these are slits for us 2 separated dark portions by clear portion acts like a apertures so light is passing through a mask through their apertures if they are clear why will they close because as your notes going to 10 nanometers down you are looking for smaller and smaller gaps therefore smaller apertures and smaller apertures will lead to diffractions okay and this is provable by random loss okay yes oh yeah the typical source if it is distributed if it is a point source I may show you something but then I will I will okay I will show you a point source system this is only a source which is like a UV light I have a lamp which shines everywhere okay so everywhere I shine but in real life I may focus that okay and here is what you are asking that was only to show you the diffract why it diffracts okay that is the way theory okay this is our system this is our projection printing a projection mask liner you are a point source and the first thing it passes through is collimator okay which is essentially a convex lens the distance is so adjusted focal length so that it becomes parallel beams here is your mask which is having a slots which is apertures is that okay black portion and clear portion again black portion so open space in between clear areas these are like a disc circular disc okay now this aperture as I just now said whenever the beams come here the center ones may straight go but the edge ones may diffract now the problem is I may refocus them for the image okay the problem start something like this depending on the actual apertures part of the beam may actually be outside the next focusing system so this essentially is the last information information was coming from the source to this but it has gone outside the range of the next focusing system so this much information typically one can say it is a high frequency information is lost and only what is the information available which is impinging on the lens okay so now this of course I should have shown this also so once you see a lens system it focuses down again okay now this lens has a what is we call as diameter how I how the lenses are made lenses are made out of spheres of any material glass or anything and then they are cut into some shapes okay so curves are created out of that so because you know it is spherical so it gives you some kind of convex or concave shapes okay so whatever is the this ball you use and you cut that diameter of that sphere is essentially diameter of the lens so this is the diameter of the lens okay now this is the distance f focal points and you see because of the diffraction all of them are not focusing at the midpoint okay and there is a information which is slightly in a spaced area and of course you can now see it roughly reflects whatever is your aperture it comes down by adjusting f and d probably you may get almost similar pattern seen aperture pattern seen at the image plane is that clear what is the game we are looking for I want an airy image should be as same as mask image so I can create by choice of proper lens system which will create this size is roughly same as this size and that is what and the way masks are this is mask so mask is not touching to the wafer or silica resist and still able to project the image which you actually wants to okay so this is essentially called projection alignment system okay now typically that is lost what is lost is cannot be gained okay yeah there is a there is a lack of some information but as I said normally the you can see from here this intensity pattern at higher frequency is much lower intensity so it is not that you really lose all of it 90 percent is anyway covered through lens system if you really want increase the dye and faster lens goes by the diameter you have okay so it now what all ends like okay but then there is a money involved it has been found that the image intensity II at this plane is essentially forms the solution of wave equation is in the basal form basal function form and this is how the basal function looks these are called first minimats this is the maxima these are called first minimats here and here this is the intensity pattern along x axis so one can see this first minima or this is called central maximum by release theory one can see that this essentially is 1.22 lambda f by D this is derived from basal function through release formula okay so if you see the radius that is the maximum available intensity only at the maximum at this point then this radius is essentially this minima to minima how sharp it is what is the what are we looking for or intensity sharply availability maximum intensity so if this is smaller it is even better okay so we find the radius of center maximum is 1.22 lambda f by D where D is the diameter of the lens f is the focal length lambda is the wavelength of the light used and okay now if you really want the image to be point image very sharp point image to be created that means this should be 0 you know smaller the radius means sharper the image so if I want a very sharp image what should I do lambda be 0 focal length be 0 and the diameter of lens is infinite yeah it is the ideal system which never can be done so obviously the radius cannot be point so point image it will always be slightly larger than that because lambda is finite focal length is finite and the data is also finite beam which actually I can create okay this is image available to you energy available to you for exposures so I essentially decided by how much is f how much is D and what is the wavelength of the light used and it gives a basal functions okay I will read it further if not I will give a paper where more details are available now this ideals points our image cannot be created so how best is what I have shown you keep finite everything and image will be slightly diffracted all that I am now seeing if this can be almost equal to the aperture I have solved my problems okay what are the advantages there are also two diffraction patterns or diffraction laws one is called final laws if the image plane is very close to the lens system then it is follows what is called final laws and if it is a far field you are far away your image plane is at infinite or very far then the image transfers are essentially due to the effect called front of our images okay optics panayat okay see our egg then poor optics okay so since in projection alignment the distance of image to this is far little larger than proximity is the minimum contact is 0 okay here the image plane is away so we say it is front of our case it has a resolution performance as following things we look for any mask printing how good is the resolution then the word which we want to explain quickly is depth of focus field of view MTF alignment accuracy and throughput these are the features we are looking for lithography expertise okay of course this image now I will show you what is each term I am talking about the word depth of focus is something like this why we are interested in the word depth of focus anyone no it is not true I mean you are saying it is not wrong but the problem is essentially this the resist on a wafer does not have uniform thickness is that correct so some image is away from the resist some images closer to resist but at the image plane both should come same place okay let us look for something more about quickly we have an image we have a figure which I am going to show you there are two problems one is of course the depth of focus other is two point sources and how do they transform there or through a lens system I repeat this is given in Lambert's book not exactly in the order not exactly the way I explain or not exactly the way I write but in general whatever I am teaching for last two three days is somewhere the other or maybe identical in some cases because I read it since I read it some of my words may be coming from them okay okay so this just to give you D is the diameter of the lens and we are interested to see what is word we are going to see now is the word numerical aperture a lens system two point sources at A and B passing through a lens system and they are they are so this is the focal length so we what we see is let us say you start with B and it appears at B dash here so it goes through a angle alpha then it is it focuses down to B the maximum angle up to which the point source can be accepted by the lens is called numerical aperture is that clear the maximum angle or maximum this hypotenuse value to this value ratio distance to this height this is sin theta okay how much is this sin alpha essentially gives you numerical aperture 60 degree 45 degree is what possibly is easier to attain the alpha angles and this alpha is also so actually it is half alpha in fact I should I have shown you alpha but many books show actually this as total alpha okay this and this because this is D but I had I had done slightly different as he also he uses D by 2 if you use D then the alpha should be used as to alpha but since he uses D by 2 so it is only this as defined as alpha so some books if you see old ones or some paper they show you this alpha this is exactly what is really aperture how much is the total beam ages to ages lens can see if that angle is called numerical aperture but the definition we there we say okay use instead of D use D by 2 so it is similar and therefore need not worry the numbers may come same okay so this the if you have 2 such point sources different reflected this it has some small angle they both will pass through an angle alpha even this angle is also alpha okay I should have shown you the center line also okay so this also is alpha so what we are now saying that to resolve to what is our object there are 2 point that is 2 areas of the mass I want to separate how quick 2 lines I can separate that is my resolution okay that is my resolution so these are the 2 such point sources and I say I want to see how whether they resolve at the other end now for each of them what is the pattern intensity pattern will be Bessel function so this is one Bessel function for a and this is another Bessel function for B what really says and that is most important it is called criteria the resolution between these 2 maxima which is the R value which is gives will be such that the central maxima are separated by R if the first minima of the 1 matches with maximum of the second if that happens you have the resolutions if anything otherwise they will come closer and then you will not be able to resolve so the one of the minima's should match with the maximum minima offers maxima should match with the minima should match with the maximum is that clear this is release criteria okay we also should realize that this right now we assume beams are in air air has a refracted index of 1 let us say I put this system in water water has a refracted index of 1.33 and let us put some immersion oil is which is what the new technology is it has a 1.4 to 1.5 refracted indexes okay 1.4 is most likely but there are new oils have come which has 1.48 or 1.5 kind of refracted index so the numerical aperture is defined as the refracted index multiplied by sin alpha n sin alpha is defined as in physics what we are saying the maximum amount of angle through which the lens can receive from any source is this numerical aperture is that clear that is aperture okay so is that clear definition and this R we want to find R now why all this match because I want to relate numerical aperture with the D as well as R R is my ultimate I want to separate two lines how closer I can come what in lithography expertise can I separate 10 nanometer lines can I separate 50 nanometer lines so I want to see what is the reservation I get from this so-called projection alignment system and no better system we know so this is the system will use okay now if you look at the geometry maybe is okay this is D by 2 this is sin alpha so D by 2 by F you look at this side this is F this is D by 2 this is alpha so D by 2 by F is sin alpha is that okay perpendicular divided by base is sin so D by 2 divided by F is sin alpha so D is 2 F sin alpha however in this assumption was the air was the where there was I mean where let's you are we are going in the air if you now substitute R what was R we figured out from the air airy pattern 1.22 lambda F by D that's the function we got is that okay if we got this function replace D here by 2 F sin lambda and if you include the refractive index in this then it will be instead of F it will be NF sin alpha so 0.61 lambda over N sin alpha but N sin alpha is numerical aperture so it is 0.61 lambda by numerical aperture that is the resolution is that clear so how to improve the resolution or resolution means R should be smaller I want to resolve as close as possible okay so what should I increase NA so I must somehow increase the NA so I must increase refractive indexes so that is why immersion word has come I can change the refractive index okay if I change the alpha which is possible but then the lens dia will correspondingly increase so I can't increase alpha too much okay the other possibilities lambda be smaller that's what we are trying we are actually reducing the wavelength of the light from G line to I line to 193 okay 465 to 248 to 193 and we are 5 fluoride is calcium fluoride is 157 okay so one of the why we are reducing is it now clear why wavelengths are going down because I want to improve R so if I improve R improve me smaller smaller the two two points can be separated I must have lambda smaller so I am trying very hard to get very low lambdas I will improve my NA by just increasing the refractive index and that is why I will do what is called as immersion lithography this 0.61 is true for certain values so we normally name it as K1 it has also depending on the machine used and typical values around 0.6 to 0.8 now another term simultaneously appears okay the word and now what is the second parameter first I said how much you can separate the second was what I was saying Maragiri depth of focus because you know my resist is not uniform thickness everywhere so I want to see a depth of focus is that okay this is also given in plumbers book okay this is what I already have said now if I increase any and my worries are and now I am proving that the depth of focus will go down okay let's see what happens depth of focus is you have a lens system and this is the delta is called depth of focus two planes up to which image should be possible okay why it should be like this because I want to see if the thickness of resist is here or here image should be same point okay so I call that as depth of focus for the lens okay so if I this angle is theta let's say I am sorry this is the external line not this don't use this okay so this goes to the this is delta this is the focal length so we see tan theta and this is d by 2 d by 2 by f plus lambda is tan theta is that okay this is geometry d by 2 by f is f plus delta is tan theta and if theta is smaller that is f is long enough then f is much larger than delta which it will be depth of focus is much smaller compared to focal lengths okay same method microscope believe actually goes on same principle depth of focus you adjusted okay so theta tan theta is theta therefore theta is d upon 2 f if f is much larger than delta then theta is roughly d upon 2 f okay is that point clear I did tan theta is this upon perpendicular the point upon divided by base base is f plus delta sorry f plus delta d by 2 upon f plus delta is d by 2 I divide over this is tan theta is theta is smaller tan theta is theta essentially how it is written tan theta is equal to sin theta is theta is smaller and sin theta is theta is theta is smaller okay you can keep series of expansions so I get now what is the difference between delta minus delta cos theta what is delta cos theta remember this is delta so I am looking for this okay so delta minus delta cos theta by optics law must be proportional to lambda by 4 okay please remember what I am talking this is my delta and this is delta yeah this is delta divided by minus delta cos theta sorry sorry I am sorry I am only saying the same so the separation between these two delta cos theta minus delta is essentially always lambda by 4 by optics okay so lambda 4 is delta minus 1 cos theta and if I expand everything again this you can this probably is not given in the way I had done but may be given so I can prove that I can write sin square theta is 1 minus cos square theta cos theta is under root of 1 minus sin square theta by namely expansion if sin theta is smaller so it is half sin sin square theta after all this delta is lambda upon 2 theta square okay simple maths or delta is lambda by 2 and theta square is 1 upon d square by f square theta just now I found so theta is d by f so d square by f square so this is half lambda but d by f is also numerical apertures d by f so it is numerical aperture square and this half is normally the machine dependent many times so it is given a name k2 which is 0.5 or close to that so if I want larger depth of focus what should I do from this expression I want even if there is a variation in resistiveness I want focusing be done is that clear to you so I want depth of focus be larger what should I do from the expression lambda be larger and nab smaller what was the condition for resolution lambda be smaller and nab larger and now we are asking just the opposite okay so which means some tradeoff is essential either you will get greater resolutions or greater depth of focus okay both cannot be simultaneously achieved is that clear so you cannot say I will have a variation in resist of 0.5 microns and we will also have a resolution of 0.10 nanometer that is not possible in any case is that clear to you so either you can adjust higher depth of focus systems if your lithography is so bad you look for delta larger then you will have a worse resolutions is that clear or vice versa okay this is what our ultimate yeah but that is pixel based if it is ccd based otherwise okay last few just a minute I know you are tired the problem which there are few more problem which you can read in this one of them is called mtf maybe quickly few two slides maybe is enough for you typical intensity pattern for the two areas maybe like this but in real life when it goes through a mask in this projection printing actually you get something like this okay this is called I minimum and this is called I maximum and we define a term modulation transfer function as I max minus I mean divided by I max minus I mean and if I plot mtf versus dimensions this is 1.0 what is it trying to show you that if your dimensions are larger this is I max to I mean this is a one can see what I am plotting for a larger dimensions separation that is larger dimension of the patterns mtf can always be achieved 1 I minimum can go as low as 0 okay image can go all the way down I mean light can go all the way down however if it is very close and there is a diffraction going on this I minimum may not be 0 but I mean you may be higher since the thickness of resist is higher I max may not be this so it may be something like this in which case mtf will become smaller okay which means that means the actual intensities are not passing as I reduce dimensions is that clear so as I scale down the technology one of the problem I see is that the actual available intensity for exposure starts reducing that clear because of this mtf factors okay the last but not the least some figures I may show you the rest you can read from the book yesterday I did say there is something called optical proximity corrections let us say this is my actual image okay this is my actual image and I do optical proximity corrections and I get this image but when I print if I use this without OPC I get this mtf image and if I use OPC corrected mask then I get little better than this what is OPC it is a software program which actually do image processing okay it is a software program which does image processing it sees that given that this if I transfer the image on a resist what do I get if I modify that initial image itself mask image then I may get little different outputs this is called optical proximity corrections these are possible okay is that clear so by lithography prior to exposures by prior to mask you have to actually do some kind of proximity corrections so you create the actual mass which the mass which you really want from the cloud designers you actually modify okay this only can be done if you roughly know from image processing how does it transfer in real life and then start correcting so that better image is then transferred on the resist this is called OPC techniques okay yes it is preparation of mask if you have a laser you can remove some part of the resist it is called scrolls but that is not every it is very costly system last sheet on this there is something very interesting thing happened since it said relates on a smaller dimensions this is your mask and this is the electric field or amplitude of the light as per the mask okay this electric field at the mask and electric field at the wafers and wafers will be early patterns so this is some kind of this patterns I get and I know the intensity is proportion to electric field square so actually I get this so what is the resolution between these two the merging has happened okay so what I do is when I reduce dimension the second part of this I have a some material I put which gives me 180 degree out of phase this is called grating optics we actually coat some material on the other side which gives you 180 degree out of phase so the electric field at the mask is something like this electric field means electric field vector for EM ways do not think I have bias it okay so the E for this and this is this but since it is a square two separated intensity patterns can be get this is called phase shifting lithography okay so if a smaller dimension comes do phase shifting do OPC do projection alignments adjust the numerical apertures you do immersion lithography and you will be able to get far better patterns on paper okay is that clear it is photo you see here in the book how to improve numerical aperture that is what I said this angle was the limiting factor so I put a liquid somewhere here okay this is called immersion this okay and generally best best liquid is what water so most immersion technology is used water as the liquids okay okay so this is one technique of improving numerical apertures before we quit this is the last item of this show this is what we are looking for now this is called extreme ultraviolet lithography technique you have a carbon dioxide laser through a lot of collecting system and focusing system we pass the beam this is a plasma chamber so we actually convert this gas laser beam and up put into plasma now this plasma beams actually are absorbed even here but you want larger intensity so I do not want any absorption but 30% will still absorb here then I have mirrors I do not use lenses lenses will absorb much more so I use number of mirrors even mirror absorb some energies and you deflect the beam and put it on the wafer okay this deflecting system using mirrors which are controlled by motors is very costly and very accuracy you want a order of nanometers okay moments so very difficult the lambda which you get from this is 13.5 nanometers please remember the numbers if I use extreme lithography the lambda which I can get is 193 158 and what number now I am talking 13 nanometers so even 5 nanometer technology is doable if I can achieve extreme UV system okay it is very expensive right now it is in billion dollar system going on and no one has a full success okay however this is what the ultimate will be in the lithography you will have a deep extreme UV system very costly system may have a smaller throughputs but may give you features of as low as less than 5 nanometers because it is its wavelength is just 13 nanometers this is something what ultimately every company is dreaming okay so that zero dimension FETs could be created okay at the end of the day thank you for today for patients okay so next Wednesday we will start with implants