 So let us talk about the lithography, continue to what we did last time that is Wednesday. One of the thing which I was talking last time was something about lithography itself and I say first we will look into photolithography. The simple reason is even for 16 nanometer or 22 nanometer nodes, photolithography is still used. Some different forms but yes. One is expecting may be by 14 or 9 or 11 or 7 nanometers may be we will have extended UVs or extreme UVs. May be I will give a reference of extreme UV research from Intel and also from two other people. You can have a look at them. So that is fine. So let us look for mask. We were telling that we transfer the image and on the silicon, okay. So what is the mask? So mask is some kind of a glass plate. Even now it is glass plate. This glass plate is normally quartz glass plate and sometimes it has some impurities added so that its transparency improves. The quartz itself is very good transparent materials. Now on this there can be a pattern. There are two kinds of pattern possible. One in which this is a dark portion inside the glass which is transparent. Alternately I may have a dark portion and I may have clear region in that, okay. So window could be clear or window could be dark. So either of the way mask can be created. If the total area or larger area on a mask is dark, the mask is called dark field. Field is overall area. So if the overall area is larger area where the patterns are not there that is darker then we say it is dark field. On the contrary there is a small dark area. The rest is clear. The mask is called clear field. So we have two kinds of mask used. Clear field and dark field. Now this dark, the photo plates are this or the mask plates are created earlier on the even for photography on film, poster films. So what could we have? The transfer this kind of image, one method is that you create a directly from the optical light as other day showed on shine on glass plate which is coated with emulsion. Let us say this is your glass plate and this is your emulsion. Now this emulsion is normally photo emulsion which means it absorbs light. It absorbs light. Photo emulsion means it absorbs light photo emulsion. In the olden days for photography the material used was silver halides or silver chlorides or silver bromides and they were actually suspended in a gelatin solution which is called colloidal. It is not mixing it is called colloidals. So earlier the films used to be coated by colloidal solution of silver halides in gelatin. And that is an interesting thing that when such films were made earlier, some films showed very good resolutions in actual photography and they actually tried to find out what gelatin had that it allowed better photography. So it happens to be, it happened to know that this gelatin was purchased from Holland where of course probably may be knowing the dairy products are very popular from Holland. Those cows have eaten mustards. So they thought that if the mustard eating cows gives better gelatin. So they started adding what is called silver sulphide with some mustard decorations. Thinking then that may improve. We did not. So what kind of science one thinks you know you are very scientific, we do not believe in what they see and they start thinking over it in a different way. So mustard was added just to say it may better have better resolution. It never had. Mustard nothing to do with it. It incidentally happened those cows ate mustards. So this gelatin is essentially earlier used to be taken from animals. Now gelatin are chemicals which you can create in the lab. So most gelatin are now available and you can suspend silver halides particles in this, very fine particles. And when you coat on any glass substrate and you dry it, it sticks to it. So there is a sticking coefficient of gelatin with the glass plate. And because of that this is called photoemulsion plates. Now this plate, this photoemulsion is something in which I can create a pattern. So if I somehow restrict, let us say light in this area, if I do not allow light, I allow light everywhere but I see I have a darker portion where light cannot go. Then the property of this gelatin based emulsion is that it receives photons and it actually cyclized itself to a non-achable emulsions. Or there can be possibility nowadays, it can be achable emulsion. The rest place where light does not come, it remains hard and wherever light goes it becomes soft. Either it is possible, earlier it is to be that emulsions used to be hard and they used to keep, whenever you shine light they used to become soft. So it is positive kind of earlier emulsions where but both are possible. So if you shine light, something absorption will not take place in this region and the rest of the region, light will be absorbed. Now once the light is absorbed, then let us say it is so that initially it was hard and you shine light, so it depolymerizes itself or if it is a silver halide solution, silver halide and silver sulphide reacts with the light and actually silver is left out. Either cases could be done. In that case this will be clear portion and the rest is dark portion. It is only thing if it is a silver emulsion. Photon emulsions otherwise can be organics. In which case if this becomes light dependent and this doesn't, so you have a contrast. So if I remove this dark portion and develop in something where let us say the portion initially was hard which became soft, so that gets developed. The portion which was did not receive light remained hard. So if I develop this I may get a pattern which is just like this. This is cross section, so in a plan it will look something like this. So you can have patterns, you can actually expose the emulsions and wherever dark or clear areas are there, light will do contrast emulsion process and may remain black or white portions across the wafer, across the mass plate. Now the problem with mass plate was that firstly emulsion is some kind of a colloidal solution though dried but it is still sticky. It is not very sticky but it is still sticky. So if you put and we shall see next time on the wafer then the sticking part may also stick to wafer. So there is an issue which is called in contact printing. The emulsion touches the wafer itself. You want to avoid that and as many times you will do this emulsion lithography on different wafer after a sign that emulsion may go away partially or it may be spin holes may create there. So it will not a good mass. So we have to throw that mass and start fresh again. So what we did, I mean we means the newer generation did that instead of emulsions these are called soft mass. So we have a glass plate on which we actually deposited iron oxide, metal okay and again did the same lithography on this iron oxide. So let us say I see to it that this portion of iron oxide does not receive light. So that portion is not hable. This portion is hable and you remove this by etching. So you have iron oxide as the place where you want a window, closed window okay. This Fe2O3 has an advantage. It is called translucent. What does it mean? Partially transparent. So it has an advantage that in this is very dark. So for a given wavelength this may be similar for the photo resist below. But for seeing through Fe2O3 mass may be good and it is little harder than emulsion. So people thought it is better. But iron oxide it is own problems. Iron oxide gets oxygen from somewhere in the lab and it does not remain Fe2O3. It becomes Fe2O or FeO2 or Fe2O4 and because of that its property is not uniform. So people left this so called first version of hard mask. Since it was a compound it was harder than emulsion and could be probably better than soft mask actually. So next we came with chrome gold or chrome nickel as the metal instead Fe2O3. So I have a glass plate and I have chrome nickel or sometimes even gold also some people did. Some people do not want gold around in the lab so they do not want to have chrome mask which is gold plated. Now the same procedure use block something in that region. H out the rest of the pattern chrome nickel and you have chrome nickel dark areas all around and since these are metallic so even if you put n times on the wafer they normally do not get disturbed by their thicknesses neither defects come often and therefore most cases if we are using hard masks chrome nickel mask are normally used. So these are called masks. What is mask? Because wherever those darker portion will be there the light cannot pass through wherever clear portions are there light can pass through. So this is the simple contrast thinking we are doing. So this pattern or similar other kind of mask other patterns can be transferred on silicon through this mask system. Is that clear? So this was called mask stopping something to happen in certain restricted area. Why we are looking for this? Because let us I want to make a diode for example very simple and people believe that diodes are very trivial. Actually they are trivial in fact but to get a diode characteristics uniformly over a million diodes is not easy. It costs you health and it is very costly process. However all said and done. So in a lab normally if you get one diode and IV is okay you still shouting publish everywhere you know. But in 999 may not be working because there the leakage currents are very obviously should be less than certain given value breakdown has to be exact arm resistance has to be good capacitance should be low all features of circuit performance are required. So when you actually make a diode it will always work like a diode but not to any specification. So in industry no one believes that I can make a diode means I can diffuse something it is a diode as great as that. So why the diode? Let us say I start with n substrate and I somehow create a selective area p region okay and somehow I can make a contact to this p region and normally I want a planar diode which means I do not want to take a bottom contact. So I create an n plus region here which is same as contact to n. So this is you can say this is like anode cathode. So diode actually requires you can see from here now how many mass do you expect this will require? The first has to be for creation of p region, second for n plus contact region and third for actually creating the metal on that. So a simple diode may require 3 mask okay. However very interestingly you could have seen I did not make a p plus diffusion there. This is a technology gain in those days at least in most cases either you use titanium oxide or aluminum both are type 3 dopants. So as soon as I make a contact part of the aluminum goes down and actually creates a thin p plus region okay. So you do not need any additional contact to p while itself creates okay. So if you have a p-n junction diode the way I have shown then I require third mask but if I have p substrate I may only do n and n plus 2 mask and of course metal. So n plus wherever I will make it I will need additional mask is that clear to you whenever I will make n plus region in n I will require additional mask because that is only a contact to that region is that clear to you. So this 3 mask process is essentially what will make a diode. Similarly for a transistor, mass transistor you may require 4 mask minimum or even 6 masks in many cases. And if it is c mask minimum 6 or may be any number as I say 30 odd also. So this mask making is therefore crucial for us because it will actually allow us the areas how much the gaps are here how much is this area where this decide the resistances capacitances. So the circuit performance is essentially directly related to the device I make okay. So anything I do machine there is reflected immediately in electrical performance and since our ultimate aim is not to learn materials or may be some material science people are here they may be interested in that itself that is not a bad issue. But for us in a way we are interested to see my circuit functions or my chip functions and therefore anything which does not lead to betterment in my chip performance those technologies I will just avoid even if they may be fantastic okay. I will see that my performance of a circuit should not get deteriorated even a percent okay. And that is where the difference starts that we keep looking technologies which are cheaper, reliable and good and still function gives you a performance of electrical which you are looking for. So no technology is great or something. Technology is as good as the performance you are expecting. So I may reduce my technology content if you are giving me 5 rupees this is what you will get sir. So the game is in design or technology is the vendor who decides what I will use okay. He says this is the jai me itna hi paisa hai. So mera jai se bhi itna hi kush milega okay. This is bargain out okay and I want profit out of it. He also wants a profit out of it. So the real cost has to be even much lower for both sides okay. So whole technology details please remember though we keep thinking of great technologies here because we are trying to learn possibilities. But in real life not all possibilities are actually used okay. It is only because Intel wanted a processor to beat Apple processor tomorrow or AMD of course is losing track anyway. So they are working for better technology because if they do not do it the others will do it. The worry is not that my processor is bad. Uski kamis hum usse safet kyon. Jai problem okay. So that is where the technology progress starts because I have to beat the system so that I sell my more products compared to the other competitors. So this is an issue which you should realize that I keep giving you options but that does not mean every design house or every system house will actually use all kinds of technology. They may use the cheapest among it which gives the performance they are looking for. So one of the best method of now implementing a circuit is may be FPGA which is very cheap. So why go on a silicon when FPGA are available okay. But you need a RAM there. Now if this RAM access is bad or RAM goes bad then whole your FPGA circuit goes away. So you cannot be always reliable on that. But you say I do not want reliability in 3 months. Use FPGA okay. Huge very fast 100k to 1 mega gates FPGA are available running at very high speeds of typically now 800 mega hertz. So why go all of it on silicon. Go by FPGA, program it and sell okay. So do not come to see that so design. So whole course is always focused on this options okay. In reality the system person who delivers it has a choice and he decides how much to do okay. There is a standard cell technique, there are many methods which can money can be reduced on a chip okay. So look at the performance, look at the time frame he wants and then you decide where I will go. Technology is only to create something what options I can provide okay. So please learn technology seriously because these are the options you will have to know what is available in market. Of course Google gives lot of such information so you do not have to follow everything what I say. But all the same it is better if you know okay. Every time you go on this your so called these days mobile which is more like a tablet you keep actually looking for it and spoil your eyes it is not fair enough okay. So better if you have your mind keep your brain intact and remember as many smaller things as you can. So how do I do this? So I first start a vapor which is say N substrate and then I have the first mask which is shown here. Let us say I have I am going to use PPR okay. What does PPR means? Positive photo resist which is normally hard but when this is light it becomes soft. So since this region I have to open so first thing I do is oxidize this vapor. By the way lithography is always performed in oxide nitrides or metal films and never on silicon okay. Silicone is the region which you want to do process on. So that is not actually used for lithography. It is the top layer which allows you windows to create for silicon is what we do lithography. So most cases I will use silicon dioxide layer. This may be typically depends on the windows eye opening that thickness is adjusted. Now if this is oxide I coat this with a resist okay. If I coat it this resist then let us say this is the mask okay this area roughly. Now I want to see that this area is open and I am using PPR. Let us say resist use this PPR. So what should be this portion? What this should be dark or this should be clear? So if I put the mask on the top of resist this is my window and the rest is darker portion okay. So I put the mask plate on the resist and I see red and this where I have to put is called alignment okay is called alignment and we will go through that later. So if I put this and then shine light typically UV light on this depends on which UV we will use. This is the other part we will discuss. So what will happen? Light will go through clear regions okay and will not go through dark region. So the resist below this portions will what will remain hard because the material used is PPR which is hard but which resist will become soft the window part. So as soon as I have soft this I will develop. The next stage after this I will develop. So if I develop what will happen? I have an oxide I have a resist this is my oxide this is my oxide and the resist will go away except for this portions the resist will remain everywhere because light did not pass through those regions. This region the resist could be removed that is called development. So I removed the resist from the window. The resist is a very good material and that is why it was not developed that means it is unheachable hard in many of the chemicals I used most of the chemicals that is why it is very unheachable insoluble material okay. Now I will put this wafer into a solution which it is silicon dioxide which it is silicon dioxide. So what is the H n I use? Hydrophoric acid is the H n phi psi O 2. So what we see now that this resist will not allow the oxide below to be etched and only region where it was otherwise open the oxide goes away. So the pattern I will receive now is something like this and then I will strip resist. What do you mean by strip resist? I do not want to keep resist for next process. So it must be etched out but the normal etchants do not this the special etchants which remove resist is called strippers. So you put the wafer in strippers so that after the oxide is etched you remove the resist from the all regions where it was otherwise sticky okay. So I have a window now in silicon dioxide and I am now seeing an area there. So I start doing p diffusion either by solid state or by implant I am right now showing implant we will see you later and dry it also in oxygen. So what will happen? It will give a p region and we will have of course oxide region on that. If I dry it in oxygen after implant then I will create oxide on the top and impurities will go down because of the diffusion process and I will create a p region inside n region. So in a way diode is now available. Is that correct? P and n. Now the next process what should I do? This is the first mask has created p-n junction. Is that correct? The second mask I need which will open a window for either for n plus diffusion so which again I want a window to be there. I want to etch that area. So what should be the mask again? This is first mask. What should be the second mask should look like? Same kind dark field and a smaller window sorry clear head. Which is this n plus area you want? So I repeat this process again I deposit resist. Now remember oxide has already come so I do not have to go further anything. I have during my driving that is why I say every time I will do window only in oxide nitrides or sometimes in metals. These are the only things which I will etch. The rest silicon is always where I am actually going to do a process so that is not etched. Of course in a CMOS what is called a step isolation, yes we will do that but some other day. So if I do this you can see here as if this area will now get protected why? Because this is a dark area. Only the clear region is this small area somewhere here okay which will resist will go away because it will be developed softened. Then I will etch oxide. So I will create a window again for what this impurities now? n impurities which is either arsenic or phosphorus. So I can do either implant sub arsenic or phosphorus and I will create heavily doped n plus region below. But during the driving of this again what I will do? I will oxidize it because any driving cycle will be in oxygen and therefore as soon as I finish this oxide is grown on that region. Is that of course the thickness may vary that may now become like this thinner oxide at the n plus region thicker oxide rest portion but that is should be grown up for any other things to go through it okay so that much driving you have to do okay. So now I have a p region n plus region so what next I will do is you are drawn okay. So I do not need a contact so let us so this is the simplest lithography I will show you and then I will actually look into the details. So now I have a situation in which I have oxide lightly lower oxide for the p plus region then I have a slightly lower n plus region everywhere is oxide okay and here is n plus and here is p and substrate this is what I will get after 2 mask okay. Now I want to open windows okay before I put a metal I must open windows for making contact is where made to which silicon. So what should be the mask? Third mask should be now you can see now 2 mask has opened only made p and n the third mask so the first pattern which is seen here since there will be thickness change optically through microscope I will see that image is that clear because of the thickness variation the microscope will actually show me the pattern whichever I have printed okay because also the thickness are different and therefore the way optic shows that there will be a change in thickness means change in pattern with visibility. So I have this my first pattern I have this my second pattern second mask with the third mask which I want to see that I must open a window in the p at the same time I must open a window n plus so now again I am opening a window so this is a clear region mask sorry dark field mask where this third mask is only 2 small dots 2 small squares the only way is that this must get aligned in the earlier patterns which I have created it must go inside one this is only one diode on a refer the really hundreds of them or thousands of them or millions of them so everywhere similar patterns there will be n such squares will be there for purchase and each of them I must get those new windows getting into old ones is that clear this is called alignment is that clear this is called alignment so these windows must get aligned inside the old patterns okay which is visible through microscope I know this is one this is the other one so I see through it put the mask adjust my mask see that the new windows actually gets inside this safety is I will keep some margin all around so that I know that I can get easily inside I may make mistake over the total 3 inch wafer or 10 inch wafer so I have some margin of my error so this cannot be exactly at the same size it should be must follow so that there is always possibility even if this moves down I may have pattern here it goes left or slight it is still inside that okay so any error in printing as we say it should be taken care through your mask design okay this is what we do so once I print this by third times I come and do a lithography again put a resist put this mark shine light so I now say I have a window opened in the p region and in n region okay so now I have this I have oxide we want to edge anything in PPR is only edge if the light does not a light passes through that so if it is a clear area dark field means clear area of the window clear field dark windows dark can I am sorry I am awfully sorry but anyway what can I do then I will use an okay you are perfect so I use NPR which is opposite that it says that wherever light shines it becomes hard wherever it doesn't it remains soft so maybe the way I assume the pattern I mean using an NPR but you are right I should do you are very correct maybe I should do again this is the first window and there should be another window okay fair enough so I have the third mass which allowed me to open windows for contact to N and contact to P then I deposit metal why I show you the contours because the process of deposition as we shall see it picks up the contours okay so it will go everywhere uniformly but wherever it is dipping the metal will go through that okay it is a film so it goes through any undulations okay so if I now put the metalization typically it can be aluminium in olden days now it can be copper plus titanium plus vanadium plus tungsten many possible combinations and once I made this now I want the contact this is now connecting all diodes because metal is everywhere okay so I want selected I have to be made so I must retain metal for this region one and also for N plus so what should I do now another mask which kind of mass it should be no no no no no please remember now I want to retain metals in this region so now it is a clear filled mask with dark windows for PPRs okay the mask will be should cover more than the this window is that correct it should be more than this window even more than sometime this this is who is the dark which covers that all of it see only it covers then there is a separation between the two patterns okay there is no metal should come here so if I do this the final version I will get so how many masks I went through 4 masks so diode looks so trivial but requires 4 masks each mask cost a hell of course this dimensions are big enough so they are not that costly but even then okay so the way I now see it something like this this is metal same same procedure any film is independent of what the lithography is only by because we are exposing resist not the metal etching is going to be for metal the resist will sit on the metal film okay so you are actually etching resist so it is same okay so now what I get here this is my cathode this is my inode so in a 4 mask process I could make a diode okay for a NP trans this diode how many mass I would have required I would not have required P plus mask so I would have saved a mask out of that they will require only 3 masks because P does not require P plus it builds itself and requires always an N plus is that clear to you so there is an issue whether to do a PN junction device with NP or PN is decided by of course it is with the goes with the other technology part because diode is not singly made it is part of the circuit so whatever is the process you will have to keep doing according to that it can be either N plus P or a P plus N source case of karna hai ya P channel kaisa karna hai ya N channel kaisa karna hai so depends on what N you have or P you have you will decide which diode to make okay it is not so much in our hand but if you want independent diode you create another well and create another diode area for that another mask sets anything additionally from process mask it keep adding additional mask keep doing additional processing okay so a process of making a diode requires 4 times lithography to come and you will make a simple N plus P or P plus N junction diodes now you can see in a CMOS process you will have to do source drains will have to do contact to source drains you will have to delineate gate you will require contact to the gate and you will also require metals so at least 6 masks will be required to create either N channel will require more sometimes or depends on whether I want N plus poly or P plus poly or Y1 N plus and then P plus additional mask will be required for doing this okay. So process is is the lithography clear so how close I can bring this is the greatness of my work because if I am working on 10 nanometers or 22 nanometers the separation has to be such that they should be able to separate by that much amount okay any lithography do the way it will happen they will overlap how do you resolve the two lines that all that optics does all that the techniques we use trying to do number of ways we do it also there are possibility there are problems which we actually when I print something from the mask on the resist surface something else appears okay this is very interesting and from resist top surface to resist below till it touches the silicon or silicon dioxide the pattern is not exactly same okay so the upper portion is called airy patterns so this air patterns they do not same as wafer patterns so firstly from the mask I am creating air patterns and from air I am going to the resist patterns and there is an air going on everywhere so I start with something and I get something else okay and since I have very small window for adjustments my worries starts extremely high and the cost of therefore lithography increases just because you are asking smaller and smaller and smaller dimensions is that clear lithography is therefore the crux of integrate circuit design integrate circuit realizations and there should be understood well how fast how another way is how fast it can be done see if I take one wafer two days then I am broke anyway so I must do in 250 wafer an hour then only I am surviving so lithography technique cannot take each wafer require some three days to actually expose that I will not be able to sustain because the cost will be too high for me so processes we learn now which essentially optimizes many things at a time and errors trying to minimize as much as we can we are not saying we will be able to remove all errors but how minimum errors we can do through some kind of a systems is what optics does okay so today of course I am not looking into optics so much but I am now going to show you this is the basic lithography and mask I thought you should know because otherwise you will not be able to realize what is mask essentially okay so I just thought you should know mask so that when you learn this you also learn because we may not do diode we may do CMOS but CMOS requires so much okay so but lithography is similar everywhere actually iron oxide film used to be in those days 0.3 microns so typically energy required was 200 millijoules per this centimeter square that is the power density we are looking for less than 100 now okay for the case of translucent I mean this chrome this it is not so much the metal it is the resist or emulsion which is going to be put on that which you are going to H okay so that is still I line or G line in those days so the requirement is not more than 100 millijoules per centimeter square typically energy yes quads glass plate molly are not used because molly has a molly gates are used you are confusing with the mask yeah they are called super hard mask the problem with molly is it is not translucent you do not see an image below so there is something called laser imaging so you need additional mask system to do that so there it is more accurate much harder it remains for long time okay but molly are unless you are rich enough you will not do that okay so we are and today only I was saying the if you have extreme uv processes the cost of lithography is 55 percent of the net cost okay so we are across 33 percent we are now 55 percent cost only goes in lithography this is what I was saying from the mask on the top of the top of the resist this image is called aerial image and below this inside this is called latent image which is resist image so how much closer this latent image is to aerial and that is closer to mask is all that lithography is expertise is that clear so how much thickness I should have so that it goes very good lower down but if it is too low that can happen if too high thickness is high how much can happen so these are the choices one has to make our ultimate aim is whatever is in the mask identically comes it should come on silicon okay that is what we are looking one to one transfer okay if how much flavor we can get is our tricks of the trade okay so we have seen some day mass designs mass fabrication this we did last time we looked into light sources and then we also started looking into the wavelengths of the light many technologies are with F2 now which is 157 nanometers okay now we start looking into resist itself and first we will look into photoresist because it is a photon based lithography we can also looked into electron beam lithography but since electron beam lithography is not directly used in stepper as it is called direct stripping it is very costly process however masks are made using litho it electron beam lithography are very accurate okay the resist there maybe I will just tell you before I come is called PMMA what is it poly methyl methyl acrycelate okay so this PMMA is the standard resist for electron beams PMMA we will come to it I just thought I will tell you what it is okay so first look into photoresist photoresist are organic polymers and are generally viscous okay they can be thinned down by adding thinners or solvents typical resist thickness desired for lithography is with the order of 3000 Armstrong to a microns these days people are trying to reduce this to 1000 Armstrong but not successfully G there are two possible wavelengths of lights which have been tried one is called G line resist the other is I line resist G line is something 460 and the other is whatever that wavelength I said one is 436 and other is 365 okay so these are the two line resist earlier with so what are the problem with them all these resist contains the three things one is called resins the other is photosensitive compounds and third is thinning down of that is called solvents in which actually they dissolve okay these photo resist compounds are termed as if you add this some of the reasons have built in photo photosensitive material inside them that themselves are photosensitive whereas in most cases it is not so reasons are neutral so you add a compound which is called PAC okay we will see what is that PAC is essentially generally these are negative or positive photo resist depending on whether you are add you have to add a photo PACs or you do not have to add PACs somehow built in PACs the emulsion itself or resist itself is photosensitive okay some you have to add a photosensitive materials okay so generally there will be three resins are carbon based or carbon hydrogen based rings of benzene sometimes along with CH3 CH bonds okay so these are photo resist as I say what is different between other is a photo means they absorb light and change the and stop the chemical reaction whenever you shine light on resist they do chemical changes in themselves these are called photo resist okay now if you do a chemical reaction somewhere then you will be able to see that in a chemical reaction the a plus b equal to c plus d is essentially governed by thermodynamics okay so most cases we heat the we increase the temperature that means you are energizing it the process now this energy can also be given by photons and therefore photosensitive reactions also do similar as thermal reactions okay yeah I will come PACs photo activated some materials which we add okay photosensitive compounds or photo activated activating compounds PACs okay we will see that little quick few short time so the so far what we have seen is how to do a lithography and to do this I need a resist so I start looking into possibility of resist I can have so I said at least for the wavelength earlier we used G line and I line we have a typical photo resist which contains some reasons some photosensitivity of photo activating compounds and solvents solvent means it is a viscous material I want to reduce the viscosity I must add little extra solvent is that correct if I want to thin down that is reduce the viscosity that means I must add solvent to it thinners okay is it okay so let us look at G line I line O ones are generally PPRs and these are called DNQ resist DAISO naphtha quinons DAISO naphtho quinon is DNQ okay these resist are G line and I line PPRs okay are the ones who are close cyclic chains you can see from here typically resin chain is shown here those who are chemist oriented also should see if this is the orientation here this will be the orientation here okay so there is some chemistry involved but anyway we will not go into that detail okay these are Benzene rings and have OS CH2 CH3 bonds and these are called polymers or polystyles or anything related to resins the material which is base resin is essentially called Novolac okay Novolac and this is the typical equivalent chain of course this dotted means it is a long chain and at the end they combine itself okay so this for example may actually cyclic out okay a typical polymer had two methyl groups and one OH groups CH3 CH2 plus OH DNQs are hard resist okay and do not dissolve in normal solvent developer or that is developers they are hard itself and when they will become soft when they receive energy okay till that time they remains hard so this cyclic chain is unbroken without light so they remain bonded and unreachable in more solvents okay including the developers is that okay so these are hard resist because they are unreachable in most cases okay you can always remove by something but these strippings but otherwise they are normal agents they do not get attacked okay so is that okay so have you anyone drawn this this is by the way today now onward the lecture is mostly taken from Plummer's book so if you do not write also and if you wish to write I will be happy but if you do not everything what I am talking I read recently the after many days I read Plummer's book for this so now I can assure you that most of it what I am telling you is available in Plummer's book though some terms and some things which I tell I am not written by him because that is my experience but otherwise everything is what I am talking now is available in even more details than what I will talk because he may give 10 cases I may give you one okay so that is the difference okay so DNQ is the most commonly used this is it okay so when I expose these DNQs they convert itself into a what is called as carboxylic acids that means chain breaks and once CWH kind of formation takes place and that is called carboxylic acids and they are soluble in what is called as developers etchers for the resist okay. The basic developer is tetramethyl ammonium hydroxide or called Tima Tima I am spending so much time on lithography I keep telling you 55 percent if the amount goes so let us learn 55 percent nahi toh it is 10 percent toh karo so learn lithography because that is the there are still not good enough models in lithography there is a lot of problem in chemistry there is a lot of problem in optics there is lot of image processing going on there is lot of electromagnetic relation based designs are going on so anyone can touch this area for a novelty any day and come out with something and which will be great so please think it this is an area which is extremely fertile as of now okay everyone is trying some funny things hoping that he may be the one to get a novel price okay I mean that is the hope the second kind of resist this typical wavelength of the light which we used to work on for the resist exposure is from 2400 Armstrong to 2960 Armstrong's okay 290 nanometer 240 to 290 nanometers is typically the wavelength of UV lights which we used to use for or even higher sometimes but at least this much however now I want to do deep ultraviolet which includes some of these wavelengths as well okay so the new technique or new resist which has come is called DUV or deep ultraviolet resist what is deep word is that it has a different valence which is attacked better by the light okay for better resolution of pattern in the resist we need resist which are higher quantum efficiency what is quantum efficiency so better light absorption so we will model it how much energy is required to actually expose the thickness of resist okay r is equal to i0 e to the power minus alpha x so we will actually model and see how much absorption in this thickness will be so what kind of reaction like go deal model we will put a model how much thickness will give how much is the exposures okay so we will do that model tomorrow so for better resolution I say the mercury source normally have a lower intensity and hence provide lower quantum efficiency typically i resist or g line resist is quantum efficiency less than 0.3 to improve resolution we need to increase quantum efficiency and we add therefore some catalyst okay which will absorb more light is that correct we add some but what is the word catalyst means it should come back it should not it start the process but it should not participate in it or rather if it participate should be recovered okay so that is the catalyst the catalyst used in the case of duvs this is called duv okay maybe I should write popularly name as duv the duvs are essentially they also have polymer chain added with some what now we call PAG which is photo acid generator it is a catalyst is photo acid generator PAG okay this is a catalyst used along with polymers okay now if you look at the way sequence of processes and this is given in Plummer's book you have a polymer chain which includes this catalyst you expose so when you expose polymer chain and then you add some kind of acid here okay which is the PAG that is continuous process and this acid when it reacts with this insoluble photo resist it actually makes it soluble and when it soluble it reacts further with polymer and releases this acid is that point there I repeat acid reacts with or PAG reacts with light and polymer make it soluble and then it reacts with polymer further and releases the acid back so this acid is catalyst it did not remain with it it keeps increasing typical then this process continues you can say further you have acid now reacts with the rest of the part and it will now create both soluble and acid will again get released so that is why it is catalyst because acid you add and acid you take out so it is catalyst it did not participate in actual reactions but it added reactions okay typical temperature of this duvs where this resist could do exposures is 100 degree centigrade at and the accuracy we are asking is plus minus half degree centigrade you can think of it resist is thick enough the process is going down and down okay is that clear so till everything is done this acid will be released it will help to make some part of soluble again acid will come out reacting with polymers again come down keep doing at then it will remain as a part of the separate which is soluble anyway okay so yes so this this is essentially when we duv the wafer itself is the check itself with that high temperatures the check where the wafer are kept is actually under nitrogen ambient at high temperatures okay actually it is not shown properly what I am showing you is the thickness part so initially light shines part of the acid reacts and actually makes it soluble then acid is again released because it reacts with polymer that acid keeps going down till the whole resist is completely soluble so instead of looking like this you see as if this is the thickness okay I draw it because this is what the figure was given in plumbers I only copied it due regards okay but essentially saying that as you start it starts reacting reacting reacting like this is that okay so the word which we used in a lithography before today being this there are two things we are worried about one of course is a word called contrast I think we will talk about contrast little much more we can but briefly what we like to do is the following contrast is between black and white how much separation really you see is a gray world is gray it is neither white nor black but we want black and white so how much contrast I create so typically during the exposure the exposure starts with a this is called exposure dose dose means how I want the flux per unit area or intensity per unit area or intensity is essentially per unit area so if you have a exposure dose which is millijoule per centimeter square so somewhere let us say initially all of it resist no exposure and somewhere at D0 it starts reacting that much minimum energy it should receive so that it starts reactions and as you increase this the reaction is faster and faster and it reaches where the reaction is complete so this dose is called DF the start point is called D0 this is a PPR initially it was already hard and it becomes softened as it went through the light exposure in the case of negative photoresist it will require some D0 to start the reaction to become hardening now and it will keep hardening and somewhere at DF it will become fully hardened okay so the what is ideally you will expect D0 be equal to DF that is what we are expecting step but that we will not get how close we come to the step okay that fraction is called contrast number or gamma is that correct ideally I expect tag or tag that doesn't happen so how much closer we can get is all that expertise is all about so the resist choices additions search of catalyst okay all these gains okay is that okay everyone I repeat this is everything today has been taken from plumber's book so no worries okay people who normally don't want to see books at least go and now see the book I don't want the reason is yeah obviously if I had to keep what I expect is if I shine light or expose it whole resist should get exposed in one but it doesn't happen it goes down down down down so whenever any process goes slowly down it creates its own problems of non-equivalence what image area you are wanted what resist is going to give will change ideally if it transfers whatever in the top will come down the image on the air area image will be transferred immediately to the resist down but if it takes time it has possibility of light getting diffractions okay and that is where the image will get smashed okay so how much bad image will come will be how much df you are away from d0 ideally I want df to be equal to d0 okay as soon as I shine light everything should get exposed okay as often but it doesn't happen it goes through thickness as I showed you chain also it takes time before it actually that time taken that means dose is not what initially on the top is the you keep increasing the dose only than the lower portions get exposed actually okay so that essentially changes the patterns down okay okay the term which is used to define this accurately is not true this is some equivalent number they figured out it is 1 upon log and whenever at least to me log means to the power to the base 10 ln means to the base e so if I write log I am always talking of base 10 okay so log df by d0 inverse is called gamma typically df value of the order of 100 mj per centimeter square for most dnq's for g lines or i lines however for duv I want to increase gamma to the order of 5 to 10 with df values going down from 100 mj to 20 to 40 mj why this is necessary because lower intensity if I require to expose it then the diffraction of this will be smaller so I want to reduce intensities as much as possible but I don't know because if it doesn't expose I will give intensity I will give 100 mj okay but if it happens in lower I am happy okay so to and I get same contrast okay so if I get same contrast I will prefer this and that is where which resist will be most commonly used duv's dnq has a limit what the wavelengths are 465 and this 385 whatever it is since they are rolling cannot resolve the lines much better than their own wavelengths okay so 0.13 micron lines can be easily resolved through dnq's but below this dnq's will require much higher energy and therefore it will be difficult for using what we call final resolutions using dnq's okay so we will require duv's for this and otherwise what will further we will do is uv's extreme uv's are what we are now looking for extremes okay yes contrast is between the image which I will show you a figure you know there is an image in the air and there is image in the resist which is called latent image so how much latent image is different from airy image is essentially the contrast okay so some portion of that is not same okay I will come to figure then so there is an issue so how much do that what areas do not get exposed okay and therefore they will remain something that I want this but if it happens to be this this means this area additionally I either not exposed or exposed so I am worried on those terms okay there is another term which is coming from actually optics but as of now you just take it it is called minimum optical transfer function called cmtf which is given by df minus d0 upon df plus d0 which can also be written as 10 to the power 1 upon gamma minus 1 upon then this the cmtf for g line and i line is 0.4 whereas cmtf for duv is 0.1 and 0.2 so we want lower and more cmtf as much as possible we like to see how it is possible you can see if df is closer to d0 what is cmtf 0 so smaller the value of cmtf better is the resolutions you are going to get okay this is called minimal minimum optical transfer function maybe when I look for lens I will show you what is actually transfer functions this results in better resolution of image on the top of the resist to the bottom of the resist aerial to lower how much accuracy it goes down as I say I want this like this doesn't happen if it happens like this means this image firstly from the mass this image will not be same as mass image from the mass image to resist image further I have a problem okay and then finally it goes on the wafer some other thing may also happen on oxides during etching so you have etching problems you have resist problems and you have mass to this problem so some problems we will try to solve that whatever mass patterns I created are actually etched in the silicon that window we decided now based on what that in window you have decided in the diode the area okay now if your window is some area for a given current density you are designed that changes something else then the current density has changed for the diode is that correct if larger area appears then your current density is smaller so diode cannot give that much on currents okay so this transfer from the mass mass patterns are generated by whom the designers who think technology people should take care of everything now that procedure because those people just give me patterns I will have to worry that what they want am I transferring to the silicon that is all the lithography problems from gain mass coordinates I generate the patterns on silicon which are exactly same dimension or as close to that as much as this is what effort is all about is that that is why lithography is tough because of course I will be telling them that do something there okay that is the issue something can be done at the designs this itself okay some processing can be done during design which is called DSP methods so we can do some image processing and do some machine from the patterns itself what that can be let us say X pattern goes to Y okay and whatever lithography I do but I want X to X so I keep assume that I capture Y image okay and do lot of processing inverse processing see that at what time I can get X so that X modify X which is a bad looking X may actually transform into good looking Y okay this is called image processing is that clear I saw last image I say it is a source machine what X I should have which will lead to this Y which is good let us so I can always make games to realize what I really want but the cost accuracy of any DSP system processor speeds all these issues come there electronics can do everything in its own time there is no one time with us so we will have to worry here okay okay before I think there are few minutes or two things I may show you okay there are drawbacks wafer topography during processing has regions of hills and troughs you have seen the wafer during arching there are some areas are thicker some areas are thinner die also you see undulations resist are liquids even if they are viscous they are liquids so they actually follow these contours whatever shape the liquid goes through that shape but if that happens the thickness of resist from the mask okay is different from different regions is that clear so there is an error going on how much undulations you have liquid resist follow these contours this leads to unequal resist thickness because it will try to fill it up let us say this stuff is there so when the resist fails this thickness will be larger than the other areas now what is the problem you are shining same light okay the left side of resist which is on the thicker area exposed but this lower area where it went through it did not now the problem is I actually decide that wherever the thicker resist I did processing for that but then what will happen where it was thing if that problem then goes into the thinner areas then I have a problem is that clear one possibility that I only look for thinner resist then I see some areas I do not expose if I expose the thicker part then what will happen to that thinner part so I must worry about so there is a drawback in any system because resist goes through traps and hills and therefore thickness of resist is different at different points so it is called problem of over exposure or under exposures there is another problem particularly in the metals which instead of silicon dioxide if you have a metal which is you are reaching contacts there will be a metal layer over which there will be a resist and there will be light and one can see from here the way it happens even other materials can also do the same there is a refractive index here there is a refractive index here and there is a refractive index here depending on N1 N2 N3 the light which is incident on that may actually get reflected is that correct may get reflected the problem with reflection is not so bad just like that but essentially means photons come back photons went in they do not react they actually came out is that clear so this is called and if the phase of incident beam to the outgoing beam is 180 degree then it forms what is called as standing wave patterns then the resist cannot be exposed because standing wave will not allow etching to go through exposure to go through so this problem with metal surfaces or refractive index which is higher than the intrudes creates problem that there is a standing wave patterns what will it happen it will reduce the resolutions so what should we do easiest solution break the expose wafers as straight so we will see that how fast so somehow we change the refractive index and thickness of the resist and then we can probably get why refractive index is different in different temperatures because the solvent changes thick what is called solute percentage changes and that changes the refractive index so some breaking may help you to reduce if it is like what is called as total internal reflection so most of the beam may go out but I want most of the beam to get in okay so I must adjust both D as well as the density there or refractive index there so that most of the light is absorbed and not reflected is that clear to you this is essentially what is tried in baking system we will come back to it tomorrow the last but a figure but you can see of course this is the oldest 1965 mask liner you can see from here there is a check here below this on the wafer is kept there there is a check here where wafer is held by vacuum okay on the top there is a square plate which holds the mask so one side there is a sliding this you keep the mask and push inside okay this has x y z motion okay so this wafer to mass separation can be done or brought down touching itself it can align by x and y I want mass to this to align I will move x and y okay either I can move the wafer or I can move the mass generally mask is held constant and wafers are moved okay the z portion is has to be flat because you are separating and touching so everywhere it should touch okay so acfz is more important than even x and y okay there is a microscope which is not having a wavelength of light which is same as exposure light so you can actually see first the image so first separate the mass from the wafer closer to it and see the image on the mask pattern on the mass and the image on the wafer and then adjust x and y so that that image of the mass pattern on the mass gets inside the earlier patterns okay that is called alignment and this all is done under nitrogen environment very what should be the nitrogen flow if I keep my aligner I am doing it where from nitrogen should go come and go it will come from the top okay so where should it come it should go here and then it should go out okay why because it should take all the particles everything out so I am not adding my dust to that okay everything should come out so there is called laminar system so you have to maintain nitrogen laminar flows okay so it is not trivial it is very important how much accuracy pattern gets because if there is a particle sitting on the mask it will not get exposed that region or it will whatever light will not pass through that that region will go that chip will go okay so the issues are very cost business okay so maintaining a clean room and super clean room in the lithography is a must so we will come back tomorrow 9 30 and we will continue with lithography hopefully we will complete tomorrow so we have seen resist tomorrow we will see up