 We have been looking in detail at the various processes or steps that are involved in your integrated circuit fabrication. Last class we looked at doping. So we saw there were two main kinds of doping. So we can dope by thermal diffusion which is normally used for doping larger areas. The other technique is ion implantation. This is especially used if you want to dope smaller regions because your substrate is held at room temperature. We also saw that ion implantation is compatible with conventional lithography techniques because of the fact that the implantation is done at near room temperature. Today we are going to look at lithography. Another term for this is called patterning. So this is one of the most important steps in IC fabrication. So the ability to pattern smaller and smaller regions is what drives the miniaturization of circuits. So when we started in the 1960s, typical circuits were few micrometers long. So pattern sizes were of the order of micrometers. But these days pattern sizes are of the order of nanometers. And this is possible because we now have newer technologies and newer methods to pattern smaller and smaller regions. So we can define lithography as a series of steps that establishes the shapes, dimensions and placement of the various components of the chip. So the important thing is it controls the shape of the various components. So whether you want to pattern a linear component or if you have some sort of a structure to it, the dimensions, so that goes with scaling. So originally in the 1960s, the typical dimensions were of the order of micrometers. Now the dimensions are of the order of nanometers. And placement because we have different components that are part of a given chip, all these components have to be aligned with respect to one another. So the placement is also important there. There are certain process goals which we can define for lithography in a similar way to what we defined for doping. So if you look at process goals, one is to create a pattern with specified dimensions. Usually the dimensions are specified by the circuit design. So and the pattern is also specified by circuit design. So this pattern has to be now created onto the wafer. Another thing is the correct placement of the pattern with respect to the base crystal orientation. So this is because typically we will have multiple layers or multiple patterns for a given chip. So that all of those multiple patterns have to be aligned to each other and they also have to be aligned to the base wafer. So both creating the pattern and placing of the pattern are important goals in the case of lithography. So usually in lithography we expose a certain portion of the wafer. We will see examples of this soon and this certain portion is either removed or maybe some material is added to it. Similarly we can also do doping of those certain portions. So lithography is always combined with some of the other steps that are part of the IC fabrication. So you could have lithography followed by layering, lithography followed by etching. So in layering you add material, in etching you remove material or lithography followed by doping where you add selective impurities which provide some sort of electrical functionality to that portion of the chip. As you mentioned earlier alignment is very critical in these steps especially because you have multiple patterns that are there as part of a single wafer. So let us look at a brief overview of the lithography process. So the first thing that you need for lithography is something called a reticle or a mask. We will see later how a reticle or a mask is made but this contains the hard copy of the design. So this is the copy that has to be transferred to the wafer. So usually if you have multiple layers each layer has its own reticle and in the case of IC fabrication process there will be as high as 40 reticles or 40 masks depending upon the complexity of the circuit that all need to be transferred on to the wafer. This is where alignment becomes so critical because you have all of these multiple masks and all these masks have to be aligned to each other and also with the wafer. So the pattern is first transferred from the reticle to a photoresist. So a photoresist is nothing but a light sensitive material which is deposited on top of your sample. We will again see examples of photoresist and how they are actually applied on to the sample. So the photoresist is applied on to the sample or the wafer and the pattern is first transferred to the photoresist. As I mentioned before a photoresist is a light sensitive material. So by using light selective portions of the photoresist are exposed and this changes its properties. So this process is called developing so you expose the photoresist through the mask. At this point the pattern is still sort of temporary. So there is some issue with how the pattern has been formed. It is possible to remove the photoresist from the wafer and then clean the wafer apply a new layer of photoresist and then start the process all over again. So this right now is a reversible process and it is only temporary because the pattern is only on the photoresist. The next step is to transfer the pattern from the resist to the wafer. So this is actually a permanent process. So this involve as I mentioned earlier either removing material from the wafer or adding material to the wafer or doping certain amount of impurities at those specific points. So this becomes a permanent process. So later we will look at various ways of measuring the wafer during the IC patterning process. So looking at the pattern when the photoresist is on there it is a critical step because of the pattern is misaligned or is not correct then it is easy to remove the photoresist and start this process over again. So you look at an overview of the process. So let us look briefly at photoresist. So the photoresist was adapted by the wafer fabrication industry somewhere in the late 1950s. So this is a concept that is somewhat taken from the world of photography. So where we know we always have a film that is exposed and developed and then printed. So photoresist can either be general photoresist or they can also be used for only specific applications. The case of a research lab usually general photoresist are used but in the case of manufacturing they may be specific photoresist available for specific applications. So these are usually tuned to a specific wavelength. So let us look at the various components of a photoresist. So the first component is your light sensitive polymer. So this is the material whose structure will change on exposure to light which is why you have the term photoresist. Then you have a solvent. So let me say structure changes on exposure. Then you have a solvent which is used to thin the resist. So a solvent is used to enable the photoresist to be applied onto the wafer. Usually a process called spin on is used for applying the photoresist. We will see that later how that works but a solvent is used to thin the resist. After applying the photoresist usually it is baked somewhere around 100 degree C for a few minutes to remove the solvent. So it is usually removed by a soft baked process after application. Another component is called a sensitizer which controls the chemical reaction during exposure. And you may also have some additives usually in the form of dyes. So different photoresist have different parts or different components and these could change depending upon the wavelength at which it is being exposed and the conditions in which spin on takes place and the conditions which the photoresist is supposed to be exposed to during the further processing. These are normally divided into two major categories called positive photoresist and negative photoresist. So we said that the photoresist is something that is sensitive to light. So it reacts upon exposure to light. Usually the light that is used can be UV or the visible range. You will also find that there are photoresists that are sensitive to say X-rays or electrons but we deal with resist that are react with light. These are essentially called optical resist. Usually the light changes the way the resist is dissolved in a particular solvent. Based upon that you can divide your photoresist into two types. One is called a positive resist and the other is called a negative resist. In the case of a positive resist exposure to light whether it is UV or visible makes the resist more soluble. So exposure makes it soluble and negative resist on the other hand exposure makes it insoluble. So the type of photoresist that you choose for a particular application depends upon the nature of the mask you have, the pattern that you want to transfer and also the base wafer. So let us look at a difference between the positive and the negative, it is a small example. So consider a wafer which say an oxide layer on top. So this could be a simple silicon wafer, an oxide layer. We now apply the photoresist and for the first time I am going to choose a positive resist. We then have a mask with a specific pattern. So the dark portion is the pattern and usually masks are made separately, we will look at mask later but the dark portion is your pattern and we now want to transfer this pattern on to your wafer. So we have a positive photoresist and we expose the photoresist through the mask. So we have a specific UV light or a visible light of some wavelength, we also have to choose the intensity of the light which again depends upon the type of photoresist you have. The exposure time will also change depending upon the type of photoresist but the photoresist is exposed through the mask and then we have two different regions. So wherever we have the pattern the resist is covered and wherever there is no pattern it is exposed. So we have one region of the resist that is covered and two regions that are exposed. After exposure if we now use a suitable solvent we said that in the case of a positive photoresist the exposed regions are more soluble. So we are left with a wafer with the oxide layer and then the layer of resist which is covered, the rest of the resist is removed. So this process is after developing, there is also the oxide layer in the remaining regions. So let me draw that. So we have exposed the resist and then developed it so that wherever it is exposed the resist is removed and wherever it is not exposed the resist is still there. If we now etch this wafer in order to remove the oxide layer the resist will protect the oxide layer that is directly below it while in the rest of the portions the oxide layer can be removed. This gives you a wafer with an oxide layer which is being protected by the resist above it. So this is after etching. We could now remove the remaining resist by again using some sort of solvent so that you are left with the wafer and the oxide layer. So the pattern of the oxide layer is the same pattern that is present in the mask. So a positive photoresist will transfer the pattern directly from the mask on to the wafer. So we can do the same thing with a negative photoresist. So once again I have my wafer, there is an oxide layer on top. We apply a resist but now we have a negative resist. We will use the same mask and again expose the wafer. So this is exposed, this is exposed and the center region is covered. In the case of a negative photoresist the exposure makes the region insoluble. So that these two regions are now hardened and are difficult to remove while the region that is covered can be easily removed. So after developing what we are left with is the oxide layer and two regions of resist. Once again we can etch the oxide layer away and whatever the region that is right below the photoresist is protected so that we can etch the oxide and create a hole. We can then remove the photoresist that is remaining so that you are left with a wafer with an oxide layer and a hole where the pattern was there in the mask. So in the case of a negative photoresist we are transferring a negative of the pattern onto the wafer. So depending upon the kind of application and the kind of mask you have can either have a positive photoresist or a negative photoresist. Usually in the case of IC fabrication you would end up having both kind of photoresist for different processes. This again depends upon what process you have and what mask material you have. So let us now look at how this mask is actually made. So when we looked at lithography we said that the first thing is to have a hard copy of the design which you want to transfer onto your wafer. So this hard copy is called your mask or reticle and this is the first thing that has to be made. So the pattern that needs to be transferred is usually developed by the circuit designers and this pattern is then broken into a series of mask. So mask is the hard copy of the pattern. So this pattern is usually made and decided by the circuit engineers. These are then converted to either one mask or a series of mask with the required dimensions. Usually a mask is made out of some sort of glass, borosilicate or quartz. So the base material is glass, it would be a borosilicate glass or quartz. A layer of chromium is putter deposited on top of this glass with chromium which is typically a 100 nanometers thick and then a layer of photoresist is applied. So in some ways a mask making is similar to the process we just saw when we are transferring the pattern onto the wafer. So these are usually supplied. So for example you are trying to make your own mask, it is possible for you to buy this blank mask which is made of glass having the chrome layer and the photoresist layer. Then the pattern is written onto the mask using a laser writer. So there is usually a digital copy of the design and the laser writer exposes specific areas of the photoresist on the glass by using depending upon the pattern. So the laser light can work at different wavelengths. And typical wavelengths used are 365 nanometers, 248 or 193. The first two are in the visible region, the last one is in the UV region. The wavelength defines the resolution of the features on the mask. So if you want to write finer and finer features then you go for a smaller wavelength. So the laser writer is used to write the pattern onto the mask and this is a process that can typically take hours. Again this depends upon how intricate the pattern is and what resolution is required. So after writing the pattern once again the photoresist is developed so that wherever you have the pattern being written those regions are removed and the chromium is exposed. This is then etched in order to remove the chromium and after some final photoresist removal you have the hard copy of the pattern. So after writing the pattern it is developed and then chromium is etched and the remaining photoresist is removed. So mask making is actually a very tedious process because it can take typically hours to make a single mask and depending upon the complexity of the pattern you may have to have more than one mask. But once a mask is made it acts as a hard copy so that it can be used to make multiple copies onto the wafers. So the next thing we look at is how we use the photoresist onto the wafer. So when we looked at process overview after making the mask the first thing we said was that the pattern had to be transferred onto the photoresist and this photoresist layer has to be applied onto the wafer. So the photoresist application is usually done by a process of spin on. So before application of the resist the wafer surface is cleaned in order to remove any defects or any other contamination typically it is cleaned in deionized water and then blow dried along with nitrogen. So the wafer has to be cleaned by a process that is usually called spin drying which involves washing the wafer with water and then blowing nitrogen to remove the excess water and then doing a dehydration bake in order to remove any of the excess moisture that is on top of the wafer. The wafer is then loaded onto a vacuum chuck so you have a vacuum chuck. This acts as the wafer holder and there are different chucks depending upon the dimensions of the wafer. This is the wafer. So when we apply the photoresist on the wafer the important thing is that it should form a uniform continuous layer and the thickness should not change from one point to another. This is because when later when we expose the photoresist through the mask if you have a very thin layer then your exposure will not be correct and you will not be able to transfer the pattern correctly. In order to build a thin layer the photoresist is first dropped onto your wafer. The wafer is then rotated so the wafer is rotated slowly so that the photoresist spreads onto the surface but it is still not a uniform layer. After some time usually few tens of seconds the wafer is spun at high rpm when you do this all the excess photoresist that is there on the surface is removed and you now have a uniform layer of photoresist on the surface. This photoresist is usually a few micrometers thick maybe one or two micrometers but it is uniformly present on the surface. In all of these cases the wafer is still held by vacuum using the vacuum chuck. So once the photoresist is spun on to the wafer this is now taken and then exposed using the mask. At this point the wafer is sensitive so it has to be protected from ambient light. So usually spin on and developing takes place under special lighting conditions so that the light is chosen in such a way that the wavelength does not match with the wavelength at which the photoresist become active due to the interaction with light. So that way the wafer are protected or the photoresist is protected before the exposure process. The next thing we look at is how the exposure process works. So we have the photoresist that is spun on to the wafer. The next step is to expose the wafer to the light through the mask. So this process is done by something called an alignment and exposure. Alignment because the mask has to be aligned to the wafer either to the wafer surface or to any existing pattern that is there on the wafer. So the first step is mask alignment and this is usually done by using something called an alignment marker or alignment marks. So these are used to align the different patterns they are not part of the circuit design but they are usually written in part of the design in order to help the alignment. So the first step is the alignment process this is usually done by using a machine called the stepper. Now alignment can be done both on the front side or on the back side of the wafer again this depends upon the complexity of the stepper that is there. After alignment the wafers are then exposed to the light in order to develop exposure. The exposure time and exposure light intensity are the variables here. So these can be calculated depending upon the type of photoresist and also upon the wavelength that is being used they are usually standard tables that are available that tell you how long to expose a particular photoresist and at what intensity this has to be done. So we align and then do the exposure after the exposure process is done the whole thing can take place in less than few minutes which is why lithography is such a fast process can go on making multiple patterns or multiple copies of your mask. So even though the original hard mask can take hours to make once that is done the pattern can be quickly and effectively transferred to different wafers. So after alignment and exposure the wafers are developed. So developing process removes the excess photoresist so that the pattern is now transferred to the photoresist. So this again as we saw earlier depends upon whether you have a positive resist or a negative resist. Now most steppers are essentially one is to one so that if you have a mask of a certain area then that area is directly copied onto the wafer. It is also possible to have steppers where you reduce the size of the features and copy them onto the wafer. So these are called reduction steppers so steppers are usually one is to one. So you have a direct transfer of the pattern you also have something called a reduction stepper where you can have a 5 to 10 time reduction in the size. So after we develop the pattern usually the resist is then baked in order to harden the photoresist. Then the wafer is taken for any of the other operations whether you want to do layering or etching or doping and after that is done the remaining photoresist is removed usually again using some hard acid as a sort of wet etching or by doing something called plasma etching using oxygen. That way the hard copy of the pattern is created onto the wafer. So this process can be repeated multiple times in order to transfer different patterns onto the wafer. And if you are doing this multiple times the important step is to make sure that the different patterns are aligned to each other. So one of the important things that lithography has to do is that with the reduction in feature sizes we need to be able to pattern in smaller and smaller regions. So the smallest feature size that is possible depends upon the wavelength and the numerical aperture of your system. So the feature size which is the smallest feature that can be patterned let me call this sigma is related to the wavelength and the numerical aperture. So k is a constant which is called the Rayleigh constant typically it has a value of 0.5, lambda is the wavelength, Na is called the numerical aperture. This again depends upon the lens system. To give you some typical numbers if lambda is 436 nanometers this is for a typical mercury source and the numerical aperture is 1 the feature size sigma is 218 nanometers. So this refers to the smallest size that can be patterned onto the wafer. But we know that typical wafers now have dimensions of the order of tens of nanometers. So one way to achieve that is to reduce the wavelength of the light that is being used. So to reduce sigma one option is to reduce the wavelength. So you can go from visible to extreme UV. So if you have visible which is 436 nanometers and you go to 135 nanometers then the feature size goes from 280 nanometers to around 68 nanometers. Another way to reduce the feature size is to increase the numerical aperture, increase Na. So usually water or oil is used as the medium instead of air. This is because water or oil has a higher refractive index. Na is related to the refractive index by the formula Na is mu sin alpha when mu is the refractive index and alpha is the semi angle of the lens. So by using oil or water you can have a higher refractive index and which will have a higher Na and a smaller sigma. So this process is called an immersion lithography system. For example if you use water, water has a mu of 1.44 so sigma is reduced by 70%. So if you have 68 nanometers to be the feature size by using water you can reduce the feature size to somewhere around 50 nanometers. There are other various tricks that are used. For example double patterning is a process that is used in order to get again feature sizes much smaller than what the lithography system can give you. So various tricks are used in order to pattern smaller and smaller regions. One option is to replace light with some sort of an X-ray source. So you have X-ray lithography. An electron beam can also be used. It is called e-beam lithography. E-beam lithography is a very slow process because it is a massless process. The pattern is directly returned on to the wafer. It can pattern really small features but because it is a direct process it is time consuming unlike the regular lithography process which takes place only a few minutes. So one of the challenges as we go to smaller and smaller devices is to always be able to pattern these onto substrates so that different strategies that are employed. That way lithography forms one of the most important steps in the IC patterning industry. It is that step that defines the critical dimensions of the devices that are possible.