 Good morning and welcome to the next lecture in our course on chemical engineering principles of CVD processes. In the next few lectures I want to spend some time again talking about the various types of CVD processes and trying to differentiate between them. In particular one of the key variables in controlling a CVD process is the pressure. As I have mentioned earlier in the course the most CVD reactors are run either at atmospheric pressure or at lower pressures and according to the pressure at which the CVD processes run you can classify the CVD processes as atmospheric pressure CVD, low pressure CVD as well as plasma enhanced CVD which typically is done at pressures that are even lower. So when we talk about low pressure CVD we are typically talking about a tenth of an atmosphere whereas in terms of plasma enhanced CVD we are talking about pressures that are a hundredth of an atmosphere or even lower if you want to go to a ultra high density plasma CVD. So when do we use what kind of process? I mean what is the factor that differentiates between these processes? The simplest way to think about atmospheric pressure CVD is that it is a process that is used when you are looking for high rates of production and low cost of production. So it is typically used for components that do not have a high sale value essentially. So for example if you are trying to make a dielectric film like SiO2 then atmospheric pressure CVD is a very suitable process but if you are trying to make crystalline silicon probably is not. And also we use atmospheric pressure CVD when the rates of deposition have to be quite high by which I mean that the rate of thickness growth is of the order of about 1000 angstroms per minute. So when you are looking for such large rates of deposition and film growth you tend to use atmospheric pressure CVD. The primary advantage of an atmospheric pressure CVD process is that it can be designed as a continuous flow process. All the other types of CVD processes that we will talk about are essentially batch processes where you have to have a chamber which is depressurized, you know vacuum is applied, process runs and then you open up the chamber and take out the substrates. In the case of the atmospheric pressure process you can design it so that substrates are constantly being cycled through. Obviously that helps in terms of throughput. You can do a lot more product when you run anything in a continuous mode compared to a batch mode and also it helps in the simplicity of the equipment. You know any time you have to provide a vacuum in a process it is very difficult. It is expensive and it is prone to leaks unless your vacuum chamber is perfectly designed hermetically sealed there are always going to be molecular level leaks happening. The CVD process is especially sensitive even to very, very minute quantities of leakage because all it takes is a few molecules of a particular impurity to for it to react with the components inside the CVD reactor and thereby deposit on the film. So in the case of an AP CVD atmospheric pressure CVD reactor you can conceivably have an entirely open chamber so that the reactants are coming in this way and the byproduct gases are going out that way and when you talk about a continuous throughput process essentially you can have a belt drive mechanism where you are constantly taking the cassettes through on a conveyor belt which is constantly rolling right. So this is the this then becomes the load station and this becomes the unload station. You will load your substrates to be coated on the upstream side and then it will go through the CVD process and then you will download the substrates with film on the downstream side and then the belt itself will typically go through a cleaning solution to remove any impurities that might have deposited on it. Of course you would still need the heater arrangement so the substrates will be provided an energy source as they go through the CVD process. Now in this particular case the gases are typically introduced through a tube that is sitting on top of the film deposition location. So this could consist of a diluent gas plus the reactants and essentially this gas flow will be happening constantly continuously even as the substrates are moving through continuously. So in principle this is a very simple flow through process and has all the advantages. What are some of the disadvantages of this type of reactor? The first is it is open to the atmosphere right so there is really no way to prevent impurities from getting into the chamber. So your product the film has to be robust enough to withstand impurity. So if you are looking for 99.999% purity it is probably not going to happen. So you have to have a product that is a little more forgiving of impurities and other imperfections. The second problem with this is that this type of a system is typically run as a cold wall reactor where only the substrate is heated. However the fact that the pressure is relatively high you know atmospheric pressure means that even locations that are farther away from the substrate are susceptible to nucleation. So a frequent problem in atmospheric pressure CVD reactors is that you have a tendency for particles to nucleate in the gas phase and then arrive on the substrate in particular form which gives rise to again powdery or flaky deposits rather than a smooth and continuous film. So the way you try to address this is really by controlling how the reactants are fed in. You try to keep them as much separated as possible until they actually get close to the substrate. So you try to play around a little bit with a gas delivery mechanism to ensure that the reactions do not take place far away from the substrate. But essentially you pretty much have to live with it. In an atmospheric pressure CVD reactor the price you pay for the increased throughput and the lower cost and simplicity and so on is that you really cannot expect a perfect film. You get what you pay for essentially. So again that is the reason why this is primarily used for dielectric materials where the absolute purity or even the uniformity of the film is not a major issue. Now if you want to take this process and improve it in order to provide better controls, higher quality, more uniformity and so on the first thing you need to do is make it into a batch process. You have to provide an isolated and controlled environment for CVD to happen and that is done in a LP CVD process which is a low pressure CVD process where the atmospheric conditions are kept at sub-atmospheric pressure. So in this case you would have a reactor which is essentially a sealed reactor and the primary difference that you will see you will still have a substrate on which the CVD process is happening. The primary difference is you will have an arrangement to pull vacuum on the system so that you can achieve a certain low pressure before you start depositing materials. Now when we are doing low pressure CVD again there are two configurations in terms of the temperature distribution. There is the cold wall LP CVD and the hot wall LP CVD where CW LP CVD implies that the substrate is kept much hotter than the reactor walls and the pressure is kept low. And of course hot wall LP CVD simply implies that the temperature distribution is fairly uniform within the reactor. A cold wall reactor obviously has the advantage that nucleation is unlikely to happen in the gas phase and most of the condensation will happen at the substrate whereas a hot wall reactor has a disadvantage that nucleation can happen anywhere in the gas phase. You can have substantial deposition on the walls of the reactor which can then flake off and so on. The advantage of the hot wall reactor as we have seen before is that it makes the flow much more uniform. You do not have to deal with recirculation flows, natural convection and other effects that are associated with the steep temperature gradient in your reactor. And in fact a majority, a vast majority of LP CVD processes are run under hot wall conditions but why would you do that when you know that the reactor is now susceptible to homogeneous nucleation? Well there is one thing that helps. Can you think of why a hot wall LP CVD reactor would be better than a hot wall AP CVD reactor in terms of controlling homogeneous nucleation? It is a fact that the pressure is low. Chemical reactions are typically driven much faster under elevated pressure and elevated temperature conditions. So the fact that you are keeping the pressure low automatically implies that to some extent chemical reactions are suppressed. So by driving to a sufficiently low pressure you can minimize the amount of homogeneous nucleation that takes place in this reactor. A low pressure by the way also helps in terms of reducing macroscopic induced flow such as buoyant flow or natural convection flow. So that is another reason why the low pressure CVD process typically gives you much better uniformity of a film compared to the atmospheric pressure CVD reactor. Another way in which you can classify these reactors is that the low pressure CVD reactor is typically run under lower temperature conditions. What that means if you recall the figure that we drew relating deposition rate to temperature is that at lower temperatures the kinetics become the controlling factor and not the transport. So what that really means is that that gives you the potential of doing multiple substrates at a time. The drawback with an LP CVD process is that compared to the atmospheric pressure process it is slower because it has to be done in a batch mode but what we do instead typically this process if you look at it the coating is happening essentially one substrate at a time whereas in the case of a LP CVD process you do not do one substrate at a time you essentially take a cassette on which you mount multiple substrates and you coat all of them at the same time right. So this is like a basket in which you can have a number of wafers or substrates to be coated and it can happen simultaneously because the conditions are set so that the deposition process is kinetically controlled which means that as long as you have a tight distribution over the or control over the temperature distribution the uniformity should be fine. Now in this case there are actually two different configurations as I mentioned one is called horizontal flow reactor and the other is vertical flow reactor. This actually refers to how the wafers are positioned. The wafers can be positioned so that they are parallel to the flow of gases and that is called HF CVD reactor whereas the wafers can be positioned also so that instead of lying flat like this they are actually standing up straight like this and that is called a vertical flow LP CVD reactor. The advantage of the vertical flow reactor over the horizontal flow is something we have discussed in one of the earlier lectures that when you have a horizontal plate and you are flowing a gas through it the uniformity is very very difficult to achieve because of concentration depletion as well as boundary layer growth. Whereas in a vertical configuration where the wafers are actually you know standing up you can potentially direct the gases and in fact what you would do in this case is the gas inflow system will be designed so that there is a fresh flow of gas for every substrate so that the flow is not going across them like this but rather there are multiple flow inlets so that the each substrate that is vertically standing up sees a fresh feed of reactants and gas so it makes the uniformity much better. In fact with this configuration vertical flow hot wall LP CVD it is possible to achieve better than 2% uniformity in the CVD film that is if you measure the CVD film across its length and you look at the variation in the thickness of the film it can be controlled to be no more than 2% which by the way is very good in a CVD reactor. And so the this is again the preferred configuration the cassettes or the wafers are mounted in a vertical configuration in the cassette and the walls are kept at about the same temperature as the substrate and the pressure is taken down to subatmospheric levels. So it is obviously a very good process and by the way this process is mostly used to make polycrystalline silicon in fact this process is used to make many semiconductor devices so silicon, germanium, gallium, arsenic, S-I-G-E, G-A-A-S there are many semiconductors and semiconductor compounds that are made using LP CVD process. The characteristics of the film will be different in the case of LP CVD compared to AP CVD the fact that these LP CVD reactors are run at a lower temperature as we have discussed before essentially means that they would not be as crystalline in structure as the atmospheric pressure CVD reactors which are run at a higher pressure and higher temperature. So they are likely to be less dense likely to be more porous and also there is a possibility that they can actually absorb some of the product by product gases and other impurities in the system. So typically the LP CVD process after it is run you do a high temperature annealing to bring its properties physical properties of the film to be more like the film properties that you get in an atmospheric pressure CVD reactor. There are two important characteristics of the film that you try to measure and control. The first is the refractive index which is typically measured using ellipsometry and the specification for a poly silicon film is that the refractive index must be in the range of 1.8 to 2.2 nominally should be 2.0 for a good film. A high value of mu indicates that it is silicon rich and a lower volume of mu indicates that you have some impurities typically oxygen in your system. And so as you measure the refractive index you constantly tweak the process so that you keep trying to bring it to the nominal value of 2.0. So this is you know quantitative measure that is a fairly sensitive indicator of the composition of the film. The second metric that is used frequently for quality control is the H rate in 49% HF solution. Here the specification for a good CVD film is that this must be less than 1 nanometer per minute. So H rates that are in excess of this basically imply that you have to do something to improve the solidity or integrity of the CVD film and again the most common process for doing that is to expose it to higher temperatures and that will improve the H rate as it is as measured in 49% HF. In an LPCVD reactor if you actually look at how the rate of deposition changes as a function of pressure and temperature so if you plot film growth rate versus let us say temperature of the substrate and you look at this dependence as a function of pressure typically what you see is that you get a family of curves that look like that where the pressure is it increasing or decreasing in this? Would the pressure be increasing in this direction or decreasing in this direction? How many think increasing? So that is the change to decrease, some of you are not sure. See again remember that I said it is kinetically controlled. As soon as I say that it means higher pressure is good. If it is diffusion controlled the other way around will be true. Something is diffusion controlled, low pressure is better than high pressure but when a process is kinetically controlled higher pressure always helps. The reason being higher pressure just very simplistically brings molecules closer together so they are more likely to react compared to if your pressures were lower. So we have looked at two categories of CVD. One is the atmospheric pressure, the other is the low pressure CVD. Third kind of CVD process is what we call plasma enhanced CVD. Plasma enhanced CVD is typically run at even lower temperatures than LPCVD and at lower pressures as well. The pressures in fact are kept so low that you either introduce a source of ionization into your system or the pressures themselves at sufficiently low values can cause ionization to occur spontaneously as well. The advantage of a plasma enhanced CVD is when you have a temperature sensitive substrate and you cannot afford to heat it up much beyond room temperature clearly you cannot rely on thermal source of energy to provide the energization to the substrate. So as an alternative means of energization of the substrate things like laser CVD, photo electric CVD and plasma enhanced CVD are resorted to. When we do plasma enhanced CVD the surface temperatures can be as low as 300 Kelvin not too far above room temperature. The advantage of plasma enhanced CVD is that it is again from an energy viewpoint, from a substrate reactivity viewpoint it is much better than APCVD or LPCVD. The downside of plasma enhanced CVD is that it is the most complex of all CVD reactors. The design of the reactor has to be paid a lot of attention to in order to achieve the plasma conditions you are looking for and maintain them. The other disadvantage is see the way a plasma enhanced CVD reactor works is you are energizing the substrate using a plasma but at the same time you are also energizing the gas phase itself. So the molecules that are present in the gas phase are also getting some of this energy transferred to them. So the same kind of problem as in hot wall CVD can happen, your molecules gas phase can become sufficiently energized that they start reacting in the gas phase and homogeneously nucleating particles or aerosols and so on. In fact there is a process called ECR which stands for electron cyclotron resonance where what you really do is you provide sufficient plasma energy to take a radical or a species in the gas phase and turn it into a radical. For example if you are trying to make let us say silicon nitride by reacting SiO2 plus N2 not easy to happen, make that happen right or actually if I use SiH4 sorry not SiO2 while SiLine is a very reactive molecule N2 is not. So it is very difficult to use this process to make Si3N4 however if you can bombard this N2 with high energy plasma and turn it into N, N is now very reactive with SiH4 and it easily forms an SiN3, Si3N4 type of compound on the surface. So electron cyclotron resonance is a technique by which you essentially convert neutral species in the gas phase to an ionized radical form which then reacts readily with the precursor vapors that you are introducing into the C V D reactor to get you the film that you want. A variation of this is called the HDP C V D reactor which just stands for high density plasma which is an extension of the ECR to provide energization not only to the gas phase species and the substrate but in addition there is sufficient energy provided to start etching the surface. You know normally plasma enhanced C V D works in its simplest form by energizing the surface by providing an additional energy component to the surface. ECR C V D works by energizing the surface as well as the gas phase species. High density plasma essentially goes one step further it energizes the surface, it energizes the species in the gas phase and it actually etches the surface which is sometimes necessary if you are trying to simultaneously grow film on the surface as well as etch certain locations. So it is kind of dual purpose kind of a system where you are able to do both simultaneously. The plasma enhanced C V D films are even more amorphous or less crystalline compared to the L P C V D films and certainly the A P C V D films which means that some of these thermal annealing or other processes are even more necessary in order to provide densification of the film and certain functional properties that you are looking for. Now one of the advantages of a plasma enhanced C V D reactor is that it is a very tunable process. You can grow many different kinds of films in the same reactor with the same reactant species. For example suppose I have a case where I want to build a multi-layer deposit so that I want the film to have varying compositions across its thickness. How do you do that? A C V D reactor in principle offers you the ability to do that because all you have to do is tweak the deposition conditions as a function of time to change the composition of the film. However that is easier said than done. It is very very difficult to provide that kind of precision to the film formation process that you can have a graded composition that is tightly controlled. However in the case of plasma enhanced C V D that does become possible. One of the common applications is multi-layers of S I O 2 and S I 3 and 4. As we were discussing in the last class while they are both dielectric materials silicon dioxide and silicon nitride have somewhat different properties in terms of hardness and moisture permeability and so on. So there may be cases where you want to put down these dielectric layers of varying composition. So in the case of if you want to take let us say silane in a plasma enhanced reactor and you want to make silica film S I O 2 you have really three options as far as how you do that. The first is let us say that all the processes are P E C V D you can introduce oxygen into the system. So S I H 4 plus O 2 in a plasma enhanced C V D reactor will give you an S I O 2 film. But it is kind of an overkill. It is a what I would call a brute force technique because oxygen is an extremely reactive species and to try and do this in a plasma enhanced C V D reactor is not necessary. I mean if you really want to make S I O 2 by mixing S I H 4 and O 2 you might as well do it in an atmospheric pressure C V D reactor. The other possibility is to do this using C O 2 which actually works quite well but what will be the concern? Suppose you make S I O 2 film using C O 2 as your reactant gas. Do you have any issues with that? Do you see any issues? The carbon can be an impurity, right? Carbon can become part of the film and carbon is you know very difficult to get rid of once it gets absorbed. So the presence of carbon as an impurity is something that would worry some people depending on the process that you are running. So in fact the preferred process for making S I O 2 is by using N 2 O as the reactant gas. The other advantage of using N 2 O as the reacting species with silane is that by controlling the ratio of S I H 4 to N 2 O you can make either S I 3 N 4 or S I O 2. So simply by altering the ratio of this to this you can have virtually a continuous transition from a film that is essentially S I O 2 rich to one that is S I 3 N 4 rich or any combinations thereof. So the plasma enhanced C V D process enables you to do that mainly because it is less reactive environment compared to atmospheric pressure C V D and low pressure C V D. So you potentially have the ability to control the process reasonably well. Another process that is similar to this is what we call a UHV process or ultra high vacuum plasma enhanced C V D. So typically while as I was mentioning earlier L P C V D may be run at one tenth of an atmosphere and plasma enhanced C V D may be run at a hundredth of an atmosphere. Ultra high vacuum plasma enhanced C V D typically runs at one thousandth of an atmosphere or lower pressures. The advantage here once again is that it can be done to keep the temperatures very low and when you talk about this type of ultra high vacuum you can almost start manipulating at atomic level. Now the problem with the ultra high vacuum kind of plasma enhanced processes there is actually two issues with it. The first is that when you pull the vacuum that low the mean free path of the molecules becomes very large right. So diffusion becomes very fast and you cannot dependably get uniform film thickness. Essentially it becomes a process where the transport phenomena start to play a role. A high mean free path essentially means that the probability of an encounter with another molecule is much lower. As the probability of encounters becomes lower what that means is that you cannot repeatedly and reproducibly make the same film every time because it all depends on whether two vapor molecules are going to find each other or not. So that is one issue particularly on the substrate what we typically find is that uniformity starts to suffer as you lower the pressure. Of course the other parameter that you can control is the plasma intensity. If you increase the intensity of the plasma you will get a higher deposition rate but it also means that the uniformity will become poorer and in fact the deposit that you make is also less dense as you increase the power of your system. So there is a penalty to pay. You get thicker deposits but they are less uniform and they are not as dense by increasing the power of your plasma. The other problem that becomes intensified when you go to very low pressures and plasma enhancement is that again particles start forming in the gas phase. Under ultra high vacuum conditions or even under the high density plasma conditions you know the heterogeneous nucleation process which happens sporadically in a low pressure CVD reactor in a plasma enhanced reactor it occurs to such an extent that particle formation is highly likely. So how do you deal with that? I mean you do not want those particles to fall on your CVD film and damage its properties. So what people use what are known as particle traps? They are like scavenging type of devices where you use usually a thermal gradient so that you know if you have a reactor and this is your substrate on which you are trying to make the film but instead your particles are forming over here you provide a essentially a temperature gradient in a certain direction. For example the temperature gradient should be such that the particles in the gas phase are pulled away from the substrate and towards the exhaust. You do that by making use of a property called thermophoresis. Thermophoresis refers to the diffusion of particles down a temperature gradient. As you know diffusion is typically associated with a concentration gradient right that is called Fick diffusion in the case of vapors it is called Brownian diffusion in the case of particles but the movement of vapor molecules due to a temperature gradient by the way is known as Soray diffusion and the movement of particles from high temperature to low temperature is called thermophoresis. So what you do is you apply a temperature gradient so that the exhaust system is kept at a lower temperature compared to the bulk of the CVD reactor. So automatically any fine particles that are formed in the gas phase due to nucleation in the gas phase will be automatically sucked towards the exhaust by using this temperature gradient. The application of thermophoresis as a particle trap is widely used in these plasma enhanced CVD reactors and by the way thermophoresis or thermal diffusion can actually have an adverse effect on CVD film formation. We will talk about this later when we talk about transport processes and their effect on CVD phenomena because as you can imagine by the same token if you are keeping a substrate hotter than the surrounding gases then vapor molecules that are trying to deposit on the substrate can be essentially transported away from the substrate because of a temperature gradient. So even though the concentration gradient may be helping you in terms of pulling molecules towards the surface the thermal gradient particularly in a cold wall reactor can be quite a significant effect in driving diffusion away from the surface. And that is another reason why hot wall reactors are typically favored over cold wall reactors because thermal diffusion when it does set in can be an order of magnitude greater than the molecular diffusion or thick diffusion that we are familiar with. So it can significantly counteract the effect of the diffusional deposition mechanism particularly in the boundary layer region where convection is absent thermal diffusion can play a huge role. So you try to design the reactor you know in such a way that you do not have to deal with thermal effects and that is again done by having a fairly uniform temperature distribution inside the CVD reactor. So we have talked about atmospheric pressure CVD, low pressure CVD and plasma enhanced CVD today. In the next lecture we will talk a little bit more about metal CVD as well as metal organic CVD reactors which are two other important classifications of CVD reactors. So with the discussion of those two we would have covered the five most widely used CVD reactors in the world and after we do that we will start looking into some of the process parameters in a little more detail. Any questions on what we have talked about today? Generally atmospheric and low pressure CVD we have homogeneous nucleation because the pressure is higher. So how are particles forming in the CVD? That is because in PECVD even though the temperature is low and the pressure is low the plasma itself is providing the energy right. I mean two neutral species are unlikely to react but if you bombard them both with plasma you energize them, you turn them into ionized radicals they are very very very keen to interact and react with each other. So it is a fact that you know same reason why the deposition happens on the surface. You are energizing the surface using a plasma. Similarly the gas phase molecules are also energized because of the presence of the plasma and in fact I would say that the effect on the gas phase energization is more immediate and more direct than the effect on surface energization. Okay see you at the next lecture.