 classes, we been looking at the various steps that are involved in IC fabrication. We saw that we could broadly divide them into four main types. One was layering, where layering means you are trying to grow a thin film on top of your wafer. Then we had doping, so doping is the addition of selected impurities to your wafer in order to create say a p-type region or an n-type region. Doping of course is important if you want to create devices like say a p-n junction or a transistor. The third one was lithography. So lithography is important because this is used in order to define the dimensions of the device and also to pattern the various devices that ultimately form part of your integrated circuit. And we also had the last one that was heat treatment which comes along with all the other three processes. So we could broadly divide the various steps in IC fabrication into these four categories and we saw each of them in detail. Today we are going to look at the evaluation of the IC fabrication process. So we are going to look at both process and device evaluation. So process evaluation is something very critical and if you think about it a typical IC fabrication process starts with a blank wafer at the end of the process. So the blank wafer goes through various fabrication processes. You can think of this in terms of an assembly line. At the end of this you get the finished integrated circuit. So usually this is a die and a given wafer can have many dies. For example a typical 12 inch wafer can have a few 100 dies. Each die represents one finished integrated circuit. So starting from the blank wafer to the finished circuit there are usually 500 odd processes. So these processes could be the layering or doping or lithography or heat treatment. But there are typically around 500 odd processes and the whole thing can take approximately one month. So even if any of this processes even if one of them goes wrong and has a defect that defect can ultimately destroy the finished integrated circuit. So even one defect and those kinds of defects which will ultimately destroy the electrical functionality of the die are called killer defects. So even one killer defect can essentially destroy the die. So because of this your device or a process evaluation is very important and it is usually done at each and every stage of the process to make sure that the finished that the wafers are getting exact treatment that they are intended to get. So process evaluation is done throughout and this is sort of obvious because if you have approximately 500 processes and we do not do an evaluation and the end product is bad. It is very hard to go back and pinpoint which of the processes is essentially bad. So process evaluation is done throughout. So after each and every step it is carried out. Typically process evaluation is done on different kinds of wafers. So the first one is the process wafer. Another name for this is your product wafer. So you can do evaluation directly on the product wafer. Sometimes within the product wafer certain dies called test dies are also fabricated and then evaluation is done on these test dies. So that the entire wafer is not evaluated but only certain dies or certain sections of the wafer are evaluated. So it can be done on test dies on product. Apart from that you can also have something called blank wafers which can also be used for monitoring. So these are the kind of wafers where evaluation is carried out during the process. When the process is done and you get the finished integrated circuit electrical testing is also done on the finished product. To understand this process evaluation a bit more we can divide the various processes that take place during IC fabrication into two main types. So if you look at an IC fabrication process it can either be a batch process or it can be a serial process. So this is one broad way of classifying the processes. A batch process is one where more than one wafer is processed at the same time. So you have more than one wafer. So it could be usually a bunch of wafers that are processed together. A typical example of a batch process is a furnace operation. For example we have looked at the layering process and one of the processes we looked at during layering was oxide growth. In the case of oxide growth the sample is heated in a furnace and usually either dry oxygen which is just pure oxygen or wet ox which is steam is used in order to form and grow an oxide layer on the silicon wafer. So this is an typical example of a batch process where multiple wafers are processed at the same time so that they see the same environment. So oxide growth is an example of a batch process. Not all of these wafers are product wafers. So along with the product wafers some blank wafers or test wafers are also used and these test wafers can then be evaluated. Because they receive the same treatment as that of the product wafers by measuring the test wafers we can get an idea of how the film is grown on the product wafers. So here blank wafers and these are usually flats. Flats means these are just bare wafers with nothing else on them. So blank wafers are used for monitoring. So a typical oxide furnace can have anywhere between 100 to 125 wafers. Out of these 4 or 5 of wafers are used for monitoring. The rest of the wafers some of those are product wafers and there are also usually baffles or again bare wafers that are put at the edges of the furnace in order to use for regulating the temperature and also for regulating the flow of the gas. So furnace operation is an example of a batch process. The other type of process is a serial process. In this the wafers are processed serially. So one at a time. So an example of a serial process could be polishing. We talked about chemical mechanical polishing which is part of planarization. So chemical mechanical polishing could be example of a serial process doping. If doping is carried out by ion implantation and not by thermal diffusion that could be an example of a serial process. So I will just say ion implantation or etching. Again this could be an example of dry etching. We also saw dry etching earlier. So that is an example of a serial process. So in this case the wafers are processed one at a time. So here again you could use blank wafers in order to monitor the process continuity to make sure that there is no variation from one wafer to the other. But actually each wafer is processed differently. So here evaluation is not only carried out on the blank wafers but also carried out on the process wafers to make sure that the process is running smoothly. And also blank wafers and blank wafers are used in order to make sure that the process repeatability is good. So far we have seen what process and device evaluation is and how it is carried out whether you have a batch process or whether you have a serial process. So let us look more closely at the evaluation process itself. We usually define process evaluation as part of metrology where metrology is nothing but the measurement of physical surface features. So if you look at fabrication some of the important parameters that need to be measured and again if you look at physical surface features it could be pattern widths especially in lithography. The depth so if you have films of different thicknesses so the film depth or film thickness the presence of any defects or contamination. So defect concentration the location of these defects are also important because if a defect is located on a metal line it could cause a short and a short would basically lead to the destruction of the IC defect making it a killer defect. So defect concentration, defect location or all important parameters that need to be measured. Another parameter is your pattern registry. So this refers to the fact again we saw in lithography we have multiple mass and all of these mass have to be aligned on top of each other so that the final device is formed. So this alignment is referred to as pattern registry. So these are some of the important parameters that need to be evaluated at almost each and every step of the IC device fabrication. What metrology should tell us is whether the wafers that pass through a given step are good enough in order to move on to the next step. So information we are looking for is whether the wafers are good enough and let me underline that move to the next step. So if you think of a fabrication process with more than 500 steps and I am at step 10 the metrology should tell me whether the wafers can move from step 10 to step 11 or whether they have some defects that need to be addressed and fixed before the wafers can move to the next step. So we are looking for an evaluation process that has to be fast. Fast is important because we saw that a typical IC fabrication can take up to a month and if you now add in process and device evaluation we want it to be fast or else it will again increase the overall time of fabrication. This is not good because we know time equals money in any factory so that if you increase the process time you also increase the cost of the product. So we want an evaluation process that is fast and we also want an evaluation product that is conclusive so that it should be able to tell you that okay your wafers are good and they are good to move on to the next step or they are bad and then they need to go and get some sort of defect treatment before they move to the next step or they should be scrapped or deleted from the line. So based upon process and device evaluation you usually define something called a process window. So a process window defines the acceptable range of parameters for that particular process. So to give an example let us go back to the batch process we saw earlier which is a furnace operation to grow an oxide. So if you think about oxide growth the important process parameter is the thickness of the oxide layer. Any furnace operation can also cause defects in your sample so a defect density is another parameter. So these two are the process parameters and then based upon the final device there is usually a range of thickness value and also a maximum amount of defect that is permissible within the wafer. So there is an acceptable range for both thickness and the defect level and because a furnace operation is a batch process when more than one wafer is processed at a time these can be measured on the blank wafers or flat wafers and whatever is measured on those is transported directly to your product wafers. So if the measurement on the blank wafers are good then it means that the measurement on the product wafers is also good and then these wafers can then move to the next step. So this is one of the advantages of having a batch process because the measurements are not carried out directly on the product but on the blank wafers. The drawback of course is if there is a problem with the process which causes either a thickness shift or a high level of density because it is a batch process you will essentially affect more than one wafer which means the wafers have to be scrapped when we say scrap those wafers are condemned and destroyed. You will not only destroy one product wafer but you will destroy multiples of product wafers. So if you think about a furnace operation with say 125 wafers typically 100 wafers of product so that if the furnace operation goes bad we would essentially scrap 100 product wafers which will again lead to increase cost. So that is the one drawback if you have a batch process. So far we have looked at defining a process window. We will now look at some of the process evaluation types then we will look at each of those in detail. So what are some of the process evaluations or measurements that are carried out? The first kind of measurement of course is electrical measurement or electrical testing. This is important because at the end we are manufacturing an electronic device. So the electrical characteristics of the device is very important. So electrical testing is carried out not only during the process but when the finished integrated circuit is formed electrical testing is also carried out to make sure that the device or the chip functions the way it is supposed to function. You also have measurements of physical parameters and we just saw an example of this. For example if you look at an oxide the thickness of the oxide would be an example of a physical parameter and the third type of measurement that is carried out is the defect measurement. All three are carried out for each and every process in the IC fabrication scheme. For example usually defect measurement is carried out for almost all processes. Electrical testing is carried out only at select locations when certain portions of the device are defined. For example if you do doping and you want to measure the dopant concentration you can use can get that by the electrical testing process. We will see that in a minute. Similarly physical parameter measurement is usually carried out when you are doing some sort of a layering where you are growing a layer or when you are trying to remove a layer by say etching or when you are doing lithography and you are trying to define a pattern. So combinations of these measurements are usually made at various steps of the processes. The combination again depends on the fact that the measurement has to be fast and it has to tell us whether the wafers are good enough to move to the next step. So let us first look at electrical measurements. So the simplest example of an electrical measurement is of course measurement of resistance. So resistance or more commonly resistivity measurement. One way to measure resistance is by using a 4 point probe. So this consists of 4 probes, 2 probes for measuring the voltage, 2 probes for sourcing the current. So that resistance is nothing but voltage over current. So to give an example of a 4 point measurement consider a wafer substrate and usually have a thin film or a layer on top and you want to measure the resistance or the resistivity of this layer. So there are 2 probes, one for measuring voltage and 2 probes for sourcing the current. So in this particular case the resistance of the film or the layer is approximately given by V over I. There is usually some constant term that is given in the front which depends upon the geometry of the 4 point probe arrangement but typical resistance is V over I and from the resistance you can also calculate the resistivity. If you know the length of the probes and the cross sectional area. So rho here is the resistivity. So 4 point probe is usually used for measuring the resistance of a film or layer. It is also called in plane resistance or sheet resistance and this process works if the resistivity of the film or the layer is lower than the resistivity of the substrate. So this means most of the current passes through the film and does not pass through the substrate. So 4 point probe is usually used for measuring say the resistivity of thin films that are deposited. For example you can have a metal film that is deposited on a silicon. Metal usually has a much lower resistivity on the resistance compared to the silicon. So that the current flows through the film. Similarly if you dope a specific layer of the substrate so that you have a dope layer with a lower resistance that again can be measured using a 4 point probe. Let us look at some of the other electrical measurements that are done. You can also use electrical measurements for measuring the concentration or depth profile of dopants. So concentration or depth profile this is usually done in conjunction with doping. For example if you have a P type material and you dope a certain region that is N type. We can measure the depth of the N type region and also the depth of the P in junction by looking at electrical resistivity. This is again usually done by creating a bevel on the surface so that this is usually a destructive technique. So we start off with a P type material. We have doped a certain region of this material N type up to a certain depth so that we now have your P in junction. We can find the depth of this P in junction by etching a bevel onto your wafer. This is again usually done by using lithography in order to expose a certain portion of the wafer which is then etched through and then measuring the resistance as a function of length along the bevel. So as you move along the bevel length you are also moving down in thickness so you can measure the resistance and relate that to the concentration. So I will directly plot the concentration. So near your surface you have an N type region as you go closer to the P in junction. We know the electrons and holes recombine forming a depletion region and the concentration dops beyond the P in junction. Once again the concentration increases because now you have your P type material. So by forming these bevels and again looking at resistance we can use this in order to calculate the depth profile of doping. Another technique for measuring doping profiles is called SIMS which is secondary ion mass spectrometer or mass spectroscopy. So this is again a technique for measuring the concentration of dopants. It is much more accurate than the bevel technique because you directly measure the concentration but it is again a destructive technique and it is lot more slower. So techniques like SIMS which is your secondary ion mass spectroscopy can be used as an offline technique for say qualifying a new tool, a new dopant tool that is come into the fab rather than using it as an inline technique for process or device evaluation. Usually techniques based on resistances are much more quicker than techniques based on SIMS. So these are some examples of techniques that are used that are based upon electrical measurement. You will now look at some techniques that are based on physical measurement that is film thickness, film widths and heights and so on. So we look at physical measurement methods and the most important of those is the layer thickness. Usually in the case of IC fabrication one of the processes is layering. So in layering we grow a thin layer of another material on top of a wafer and it is important to know the thickness of that layer. So one simple method which is used in order to measure the thickness of these thin films is the fact that these films are colored and by looking at the color it is possible to approximately calculate the thickness of the film. This idea is essentially called white light interferometry. So consider a wafer with a thin film on top. So this could be a film of an oxide layer or a nitride layer. So white light were to form, were to fall on the film. Some of the light will be reflected from the top surface of the film. Some of it will penetrate through and then get reflected from the wafer. These two lights that come out from different depths can essentially interfere and depending upon the thickness some of the wavelengths will undergo destructive interference. Some of the wavelengths will undergo constructive interference. So even though you have a white light that comes in your output essentially has some color. So by looking at the color of the film it is possible to approximately calculate the thickness. So this is the basic idea behind white light interferometry. It is mainly used for films like oxides and nitrides where there is no absorption in the visible region but the band gap is usually in the UV region and for measuring thin layers typically less than 500 nm. The color in the case of white light interferometry depends upon the film. So it depends upon the index of reflection. It depends upon the viewing angle. So whether you view the film from top or from the side will matter because it will change the distance or the path length of the light through the film and also the film thickness. To give some values consider an oxide overhead viewing. So your angle you are looking right down so that the angle is 90 degrees. So consider overhead viewing of an oxide layer. So depending upon thickness we will write the thickness in micrometers. You can have different values of color. So if your thickness is 75 nanometers 0.075 your color can be brown 0.075. So 1 it is violet 0.2 it is gold 0.3 it is blue and so on. So as your thickness increases the whole thing will again cycle. So white light interferometry is an approximate technique it is not very exact but it is an approximate technique for estimating the thickness of the film. Based upon white light interferometry there is another technique for measuring thickness which is called a spectrophotometer. Once again this is based on the same principle except that a monochromatic light is used. So here you have an example of a film of some thickness T. Light falls on this some angle theta. The light gets refracted where it gets reflected from the bottom surface and then once again undergoes refraction. This angle is theta prime. So if T is the thickness of the film your path difference for constructive interference is an integral multiple of the wavelength. So this is for constructive and for destructive interference it is an odd multiple. So by changing the wavelength of the light it is possible to produce fringes of light and dark bands and from knowing the fringes and from measuring the value of m it is possible to calculate the thickness. So thickness T nothing but delta m for 2 fringes divided by 2 square root of n square minus sin square theta n is the refractive index of your material 1 over lambda 2 minus lambda 1. So delta m is the number of fringes between 2 wavelengths lambda 1 and lambda 2 where you have constructive interference n is the refractive index and T is the thickness of the film. So this takes the idea of white light interferometry but by using a monochromatic light gives you a much more accurate measurement of thickness. There are some other ways of measuring thickness. One more technique is the ellipsometry technique. This uses polarized light instead of your normal unpolarized light. So that when light passes through a thin film and gets reflected it also gets rotated. So changing the polarization and this change in polarization is a function of both the refractive index n and the thickness T. So once again by measuring the change in polarization it is possible to calculate the thickness. So ellipsometry is especially useful if you have stacks of different films of different thicknesses so that by putting in the values of the stack you can calculate the individual thicknesses. So it is useful when you have multi layers. For example you can have an oxide layer on top of a nitride and so on. Another way of measuring thickness called profilometry. So in this case you measure the surface topography. So there is a stylus that moves across the surface. So that if you have a film that is deposited on the surface you will essentially have a step and by measuring the step height you can calculate the film thickness. An offshoot of the profilometry technique is the optical profilometer where instead of a stylus you now have a laser beam that scans the surface and this laser beam will get reflected of a step and based upon the reflection we can calculate the step height. So whether you have ellipsometer ellipsometry or whether you have profilometry especially optical profilometry or you have interferometry all of these are really fast techniques that can be used in order to measure film thickness. So one of these techniques are usually integrated along with the process evaluation in order to measure the film thickness. The last one we want to look at is the measurement of defects or contamination. So when we look at defects or contaminants on a surface some of the important parameters we like to know. One is the size and shape of the defects, the density of the defects. So how many they are on the surface? The location of the defects especially on product wafers because once again you can have defects at certain critical locations which can destroy the dye and also the chemistry or the nature of the defect. So the simplest way to look at defect density is some sort of a visual inspection. So this can be done with the naked eye but now we know that we have devices whose dimensions are of the order of nanometers. So typical defects are also of the order of tens of nanometers. So simple visual inspection with the eye is not possible. So usually some sort of a microscopy is used and because we have defects of the size of nanometers usually a scanning electron microscopy or SEM we will not go into the details of an SEM but an SEM also has an attachment for X-ray analysis. This is called EDACS, Energy Differsive X-ray Analysis and this can be used to get information on the chemistry of the defects. So by using an SEM you not only get information about the size, density and location but later we can also go and get specific chemical information about the defects. So a typical SEM arrangement consists of an electron source. There are usually lenses in order to focus the beam and scan it on the surface, electron lenses then you have the wafer. There is some sort of a detector which is used for imaging. The whole thing is evacuated. So the electron beam is scanned on to the wafer surface and wherever you have defects the images of the defects are collected and these are used to calculate the density and the concentration. Sometimes in the case of patterned wafers the SEM image of the patterned wafer is compared with the image of the ideal wafer so that we can look for defects during patterning. So SEM is usually used for defect monitoring in line because it is a fast technique but it does not provide any cross sectional information for doing cross sectional information TEM or transmission electron microscopy is used but that is an offline technique. So today we have looked at three main measurement techniques one for measuring electrical resistivity, the other for measuring physical parameters typically a thickness of a film but you can also use it for measuring the dimensions for example the width and the area of the films and finally defect measurements mainly using an SEM. All of these are used during the various steps of the process in order to evaluate the wafers and find out whether these wafers are good or bad for subsequent measurement. So the next class we are going to look at product yield or process yield where we see how good a particular IC fabrication process is.