 Okay, so here we go, today may be the last lecture in which we talk about diffusions and we will next time start with thermal oxidation. Let me quickly revisit something which we did earlier but little more carefully we look into some more problems in actual diffusion profiles, so we looked for theory again. What we see now that the flux density was given as d minus d dn by dx from the fixed first law but one can see from here at a given temperature of diffusion which may be around 900 degree plus, the intrinsic carrier concentration of silicon itself will be large enough, large hole electron pairs are any available, so Ni is higher. However at diffusion temperature if the ionized impurities are created larger than this Ni, they create a gradient actually and that essentially leads to space charge and because of that we say there is a electric field. Now this additional electric field actually provides mobility to ions, okay. So the diffusion species which are ionic actually find that there is not only a force due to the diffusion gradient but also there is a field added flux available which is mu times n into eta e which is the electric field. I hope you have done some devices somewhere, this is Einstein relation plus dkT by q plus we can always while built in electric field essentially is minus kT by q 1 upon n dn by dx. Look for theory of devices or any other book, this is a standard expression in our case n of course is the carriers available which you can say dn capital dn by n is the number of impurities and n are electrons, please remember n are electrons. So in real life if you have noted down the flux has now two terms, one due to gradient the other due to electric field and one can see from there if the electric field is in the direction no diffusion will actually occur. So impurities can go deeper during this higher gradients and because of that higher electric fields. These relations are standard, if you wish I can even derive them this is very simple derivations but those who want to see some other book in devices do see that, okay. So why are we doing this is some one example at least or maybe two examples I will give you why are we so keen about doing this because we figured out in some species at some temperature did go more than what we thought by normal diffusion equations. So why did they go? So we came back and say oh there must be some additional force on that which carried them ahead, okay. Or in some cases the profile suddenly drops down, so why what has happened suddenly? So some reasons which we saw in real profiles we wanted to explain and based for them we created some more physics and this is what it is. Have you noted down now? A major interest to us is these two equations which I wrote J is flux density d dn by dx minus of course even now I am assuming d is constant. In fact d is also function of n itself. So there is additional term may appear but that we will see little later. So if I expand this in real term it comes as 1 minus d 1 plus dn by dn into dn by dx and this term in the bracket 1 plus dn by dn, n is the electron concentration, n is the capital n is impurity concentration then it that factor 1 plus dn by dx is given a name H which is enhancement factor, which is enhancement factor. So d into H is essentially now effective d, d into H is now effective d and therefore we can even find what is H which I did. I hope that you know these two equations again as I said if you do not know these equation again the charge neutrality means p plus n d is equal to n a plus n, n a minus. So charge neutrality holds so this is the first equation we use and law of mass action is what? P n is equal to ni square. So using these two equations you can write this expression n by ni is equal to capital N by 2 ni plus bracketed N by 2 ni square plus 1 to the power half. In this case either nd or n is will be used because you will have only one kind of species, n will be either nd or n is because you are only diffusing a species. So if I solve this and use binomial expansion assuming that n is larger than ni that is the condition I set then this can be expanded and roughly this can be written as 1 plus n upon n square plus 4 ni square this is roughly good enough this may be around if you calculate this value this may be around 0.1 to 0.2 or sometimes even 0.8 if n is much ni is comparable to n it can be as high as 0.8. So depending on the value of n and ni at that temperature one gets the coefficient H which can be as I said can go up to maximum up to 2 but minimum at least is not 2 means 1, 1 plus 1 is 2 but typically around 1.2 to 1.4 is the value which you get total H is the factor which is 1.4. So what does this mean that the diffusivity will increase in the presence of large dopants at if the temperature is high ni is higher but dopants are even higher. So for higher doping cases additional diffusion appears simply because this field enhancement factor H there is another effect people see which is very important and we will see that this but if you have noted down as I said this is my expansion I solve it so you do not have to solve every step you just this is the expression and you may write this expression I write it everything because I do not copy so I had to solve myself so when I solve steps have to be written but you do not have to I already solved it. I just want to tell that how it is done because most people believe that books are final answer they give you know it is derivable from basics that is why I wrote two equations and everything is solvable okay. Okay so j is minus h dn by dx or d effective dn by dx if you have noted down here is an effect what it can give let us say initial profile was this one and there are impurities here and electrons here which are higher mobility remember electrons are higher mobility than ions since electrons are higher mobility than ions electrons move faster and that is why the electric field is there is a depletion layer created which increases the electric fields. If there is an ions electric field impurities try to follow the electric field and since they follow the new profile is something which is different this was initial profile and now this is additional diffusivity has been seen so it actually goes in okay but net concentration remaining same surface concentration slightly reduces to adjust the number of impurities which you are pushing. There is another effect which is not so obvious from here this is called isotropic diffusion which is also varying part to many solid state diffusion people if I am actually introducing impurities inside of a silicon substrate this is one effect this is another effect. So the impurities do not travel essentially the electric field to great extent pushes them down but there is a diffusion also sideways okay this is called lateral diffusions this is because the impurities are isotropic in there getting inside material there is except little direct duty I get it because electric field is with me but even then there is a site diffusion goes on. So the actual source of drain if I create and I am allowing some window and through which I am putting impurities the source drain will not be of the same size up through which the window I pushed in but there will be also lateral on the side okay and that sometimes varying because essentially it helps in some cases because channel length may reduce in fact good but in some cases it will actually improve it will allow more depletion there and more back charge will appear. So there are issues which will increase so there are issues in which one has to worry how much lateral diffusion is so if your electric field is stronger this lateral diffusions are smaller so larger the doping you are doing this is possible for the normal mode of diffusion this is not possible okay because if Ni is equal or less than larger than this at that temperature then it will not have an instant effect it will only go through laterally as well anywhere okay. This is not just three it will go all directions isotropic is the word if anything goes in one direction what is the word use anisotropic so we are actually looking for anisotropic situation but we never get it in normal diffusivities is that no no no there is in a way inside furnace there is no voltage this is called built in fields these are called because of the as I say electrons are higher and they move faster in compared to because of their mobility they are higher so as they move out the depletion layer is created and that depletion layer essentially creates electric field okay space charge Poisson's law okay so more impurities are than sucked in because the electron the electric field electrons going in means field is higher on the other side that is what I only higher doping this is seen at lower doping this is not so strong in effect that is why I say H is close to 1.1 to 2 depending on the doping actually you are going to do okay this is two effects which I seen the third also is another problem which we see in many cases we also know that these also function of impurity concentration that is what I said and it can be shown that the vacancies in presence of electrons and holes can be charged species the figure maybe I can draw for you this is your lattice let us say this is your vacancy and this is your impurity so this impurity finds a vacancy here but it does not really immediately then because it will create a vacancy here but normally what happens it exchanges something like this is called vacancy ions pair is called vacancy ion pair okay and actually this hole we can say ion pair actually jumps okay so when this goes ahead or this goes ahead it will create another vacancy and ion will move ahead of it okay. So essentially what happens this wherever vacancy is larger in earlier and there is a gradient so more and more impurities in a pair starts getting in so this is a similar it can also occur occur with interstitials but that diffusion is not very dominant for us why because they are not contributing electrically but there is in real life even that diffusion is interstitial vacancy pair is also formed okay so we are not looking that so carefully because they do not contribute to electricity or currents however in real life many things happen in materials which we do not look into it because they are not electrically active situations whereas in our case if the electron holes are available this normal vacancy is called neutral it is only there is no charge with it but if it forms a pair it picks up a hole okay then it is called V plus and if it can pick up more than one electron that is two vacancies or two ions close to one vacancy die vacancy as it is called then or maybe rth vacancy as it is called so the neutral vacancy plus r number of electrons r can be 1, 2, 3 but not more than 3 in different materials r e minus is we are this is the definition I give when it picks up a hole then it becomes positively charged vacancy if it picks up an electron it becomes negatively charged and it can pick up two electrons one electron two electrons and even three electrons so that is called charged vacancies now if there is a charge vacancy in a bank of a silicon what does that mean and if these vacancies have their energy state in the band gap then we have worries if they are outside we may not be by it will be the conduction band above who cares however when we did lot of other experiment Raman spectra and many others it has been found by many experiments that these energy particularly for impurities in silicon there are two my electron states vacancy state minus one minus v minus and v minus two which is energy minus one energy minus minus the upper one is only 0.11 ev from conduction band the other is 0.44 electron volt in the conduction band whereas in the case of holes there is only one possible mechanism in silicon v plus so that is only and that also varies in different materials for example indium if you have an indium species and with a vacancy it may have around 0.16 electron volts as the band is the energy here but in the case of go on or any other species it can be as low as 0.08 0.1 kind of situation 0.01 or 0.1 in many cases so depends on species and depends on material you are using this v plus plus v plus but please remember even v plus plus in some other material with some other impurities is possible in silicon only v plus is observed within the band gear whereas two electron states v minus and v minus minus or two minus as we call are also available. Now because of availability of this states yeah this acts like a trap that is what the vacancies and therefore the impurities are actually trapped on these vacancy charge vacancy sites okay. So they may form now pairs like for go on it may form an IV plus or IV 0 in case of phosphorus it may form IV 0 IV is impurity IV minus and IV minus minus these are pairs in arsenic same as phosphorus IV 0 IV minus IV minus minus. So what is the implication there are otherwise impurities are going and picking up vacancy site but there is additional mechanism of pure diffusion which is adding to the movement of ions and that means there will be additional diffusivity in the presence of vacancy ion pairs one of course is actual jumps which we have already done that is how we calculated D there is a jump frequency one point to the other this is the energy barrier it crosses reaches the next site. But there is additional mechanisms we see at least for higher doping cases in most cases then they form a pairs and these pairs may contribute to additional diffusivity. If you are noted down please remember this has been verified by number of experiments including infrared spectroscopy including many other analytical tools that there are charge vacancies is that okay everyone why I am doing all this I am show you one example why I thought of doing show you because of course these days many very few people are working on phosphorus diffusion in CMOS but mostly we are going for arsenic why we like to go arsenic no no no it is not light not mass part shallow junction one is this but the only reason arsenic can give me highest solid solubility is misfit factor is 0 same tetrahedral radius that of silicon 1.18 and strong so the largest number of impurities which I can push in for arsenic is only possible in silicon so 4 into 21 per CC I can get to you no other species can go to this concentration is that clear phosphorus has a this is around 1.1 or something so it will give 0.25 or something as misfit factor which means it cannot go to all the way it may go to 21 but it will not go 4 into 21 or 5 into 21 which is possible in arsenic. So for larger surface concentrations arsenic is ideal and the second of course is the diffusivity of arsenic as that figure other day showed as the least of diffusivity among all other species what is the advantage shallow junctions very good. So if you look at the of course you can solve this but I just wrote the final answer to show you the diffusivity then effectively is H time additional diffusivity because of pairs this P is the whole concentration with reference to Ni, N is to the Ni and remember these are two possible vacancy mechanisms one with positive D, negative 2 Ds and however this is general expression both are P or N. So for if you look for N alone this plus will be not there because there will not be any IV plus formation because say phosphorus is already having additional electron with us or arsenic and additional electron with it. So they form only D minus and D minus minus pairs so D effective for N is D0 plus D minus N by Ni plus this actually N by Ni is the ratio vacancies to the intrinsic vacancy at a given time some other day. Extrinsic case the number of vacancies available to the intrinsic case number of the ratio of that is essentially N by Ni. You just listen me what I say this is good enough D effective for P dopant will be D0 plus D plus only because borons are minus B minus ions okay. So this will only give one term. So now we see that for a given doping situations with that temperature we find D I know P and N so Ni and then I know how much is D effective for that species. Now all that we did for two reasons here is something an example you note down and then I will show you where these examples were used. There are two popular effects in particularly in the bipolar processes one of them is very important. Please remember I keep saying mass circuits are taken over bipolar by huge numbers but there is still 8 to 10 percent bipolar market so do not think that bipolar are zero okay. So never and firstly I like bipolar because it gives lot of physics of semiconductors junctions. You use anything in mass later but basic thinking is far better in bipolar then only you can think why mass did not do or as bad as bipolar have I did it better than bipolar. So in some sense do not think that bipolar throughout the only things silicon bipolar throughout okay that much I can say all high frequency HEMTs or whatever devices now are gallium arsenide or 3, 5 material based bipolar but they are very much there they are in optical devices everywhere and therefore do not think that BJTs are over okay their application area has certainly okay. So I want to show two cases quickly if I see this is my let us say I am making an NPN transistor vertical this is my base diffusion this is in substrate N and I make N plus diffusion for making emitter actually it is also here collector this is a BJT. Now what I thought that if I diffuse emitter here the base width actually is this much okay. So it is good if I diffuse emitter little this then I will get smaller base width so better gains of course it is not very true because it will also go in all directions but the minimum base width is available to you but it does not occur like this. In fact in real life exactly below emitter there is a dip in the base region it is called emitter dip effect. So what happens the base width did not change okay it remains same and therefore your gain did not get improved and did not get improved as what we all thought that it should but it did not okay. So this was an anomalous situation we thought it should do something but it did not. So we thought why why this is only a wherever N plus was coming why there was a additional diffusivity of boron down okay. Please remember boron is going down okay. So we figured out that emitter is heavily dope material N plus maybe phosphorus or something what kind of vacancies pair it can form phosphorus or unipharsening it can form minus V minus and V minus minus. So there is a diffusivity which is essentially available with this but as soon as it tries to go below the concentration itself reduces. Surface is highest but as emitter diffusion happens the concentration at this junction point is smaller it comes smaller and smaller. So now what time this pair formation was declared by us when N i is N is larger than N i but as N decreases this pairs do not remain paired and vacancies are released okay. They break P plus let us say sorry I V 0 I V minus let us say becomes I plus V minus. So vacancies are released as the vacancies are released vacancy does not mind it is and they are mostly initially minus but it may also happen that it may become I minus plus V 0. If there are neutral vacancies created there the boron now finds there is a additional place where they can go okay and therefore wherever this vacancy drift down which was the H factor going down so it finds that more and more vacancies are available just below that. So more and more boron goes down goes down through this. So first thing it happens that at high concentration the pairs occur but at lower it did not okay. So this was one reason we said that look this is one possibility that pair might have broken down. In the case of phosphorus there are many effect but a major effect I can show you let us say this is ideal profile I want for phosphorus N 0 X J I want to create somewhere here in B but what I have figured out in real life this is the profile whether Gaussian or something like this but in real life the profile is not like this actually. There is a kink at some point and suddenly more impurities get out the same reason one can ascribe that as the concentration of this impurities goes down below N IE the vacancy pair both V minus minus as well as V minus actually are broken down they come out and as more and more vacancies are available for you phosphorus starts following these vacancies and start getting more and more inside. So now the junction depth many times is orders of higher than what you thought in your process you thought it is a 0.1 micron junction it may become a micron junction okay. Now this effect has to be known a priori because otherwise the junction depth which is deciding the base width or any other parameter may actually go so early that your performance of the circuit will never be as what you wanted. Therefore this anomalous effect when first time we observed many years ago people realized that there must be something more to it and therefore they looked into physics once again and they say okay there is an additional diffusivity and breaking down of pairs which causes vacancy creations which allows excess diffusivity okay. This issues which I am talking about are real life issues of course some books might be I do not think plumber has given this diffusivity term but I think he did not explain much about the actual profiles which people observe okay maybe deep effect everyone talks okay. This was first worry because they were only making in 70s BJTs and every time we used to figure it out that the gain has not improved okay which we thought by so much diffusion will happen. So we have to replan ourselves how much we should do how much so all these issues came when BJT gains were not attainable and then we started looking for physics okay is it okay? So these issues are essentially what I must say are physics based and one can explain there are many anomalous effect there are 12 or so I am not going to each of them I am just trying to say this is the physics one can think of and can explain the real results. Why I am interested in these models because if I am doing this whole processing of the chip on computer then it computer cannot think hopefully so good of course there is a artificial intelligence word is used but it is AI is how much you teach so that is not a real artificial okay that is AI is only as much as you teach it is a neuron you teach neuron that much it remembers it is like putting in memory there okay. However in real life that cannot do the job of a humans so what we figured out that some results which we are getting we are not so ready but the actual process has shown something else so the computer should also know what could have gone through okay. So the models had to be substituted there okay that is why these models were created because a CAD tool we one needs to have exact models not exactly of course what do we do in spite of all this let us say I am not able to get the profile then I will add some 1 plus K times some factor okay and try to fit there first and then I start explaining K afterwards after I find it is fitting okay most due regards to all modeling people but that is what they will do they will add 1 plus K and as fit to it that if K does and then fit linearly K e to the point B they will put some number and then also get a physics okay oh this must be thermodynamic issue come back okay that is the but anyway at the end of the day computer has to give the values which we observe closest to what it should be having done modeling having done the physics having done the possible profiles you can get okay how do I calculate time temperatures for given depths and surface concentration now we actually want to see how it is done in a lab okay because unless we do in a lab there is no issue so essentially if you are looking for diffusion processes in a lab first start looking into what are the requirements okay what are the requirement for a good diffusion system these are some of my thoughts over the years if you have seen it that the fusing impurity should be always brought in contact with the pre-cleaned silicon surface only then impurities can go so they must touch somewhere but gaseous form solid form whatever form they must touch the surface of silicon so first system which you are going to create must have possibility of see that the source is touching silicon okay so that kind of system you must build something it should also remain in contact as long as your diffusion process is going to be done for given temperature time cycle the impurities must keep touching this silicon surface if you are depositing something you must remain uniform there so there must be constancy available it should not vary then you know suddenly some impurity is gone away suddenly they come this now uniform that should remain there so that is another issue the third requirement from a diffusion system should be such that it should be possible to vary surface concentrations NS up to its solid solubility limits it should allow me to reach the solid solubility limits at highest of temperatures 421 or kind of that value if I cannot reach that that means I am limited by the diffusion system which I do not want okay of course these are all obvious but whatever is obvious is stated okay the diffusion process be such that it does not damage the surface of the wafer because if surface damage occurs then there are more issues of what we call trans created and random diffusivity will start so it should not actually damage the surface of silicon okay this is very important the reason why I said that many gases or many materials we did not use because it started surface damages we please remember the wafer which we created first time was polished surface and polish is of the order of 0.01 micron that is the this at which our polish was and that is what I said it is better than normal mirror finish okay it is such a flat surface so anything pits go down that which you do not see by eye but may be nanos hundreds of nanometer pits it will create different diffusivities okay and therefore we do not want anything which goes there of course you can say I will remove that yeah there that exists that possibility that okay I will do some way that I will further polish it out after my process that is done this process is called CMP okay mechanical polish chemical mechanical polishing which is allowed now earlier it was not now it is allowed so it is not that this criteria is not so valid now we can say okay I have a CMP I will do that okay okay. So these are four there are few more which we will look into is that okay first is contact second is till diffusion it should remain all impurities should remain in contact it should be given it should be allowed to get me the best surface concentration I need and the process should be such that it does not damage the wafer these are four requirements after the diffusion process the impurities and other materials on the wafer surface should be easily removable that is most important for example other day I said the silicate glasses borosilicate or phosphosilicate glass I should be able to remove that and not by great polish I should just dip something and it should and get a mirror polish again. So this is an essential part of a system whichever gases sources you are using you must see that they finally give me etchability of the any surface which I want one of the major interesting companies at the last two the diffusion system should be able to give reproducible results if I do today may be another half an hour later some other diffusion it should be reproducible if the one result is not matching the other then I have a problem of reliability not only the problem is not only these are two slightly same but different reproducible essentially means for given process the whole number of wafers in a run should have same result and if I do number of such runs they also should produce same results okay. If there is a huge variation then my process is not standard my system is not standard because the yield is money I want almost 100% yield so if so as close I reach that much money IR okay. So these are manufacturing requirement that the process has to be 100% uniform throughout that any number of times I do and that is very important okay. There are two kinds of systems for diffusion we use one is called open tube systems the other is sealed tube which is rarely used but possibly we will see at least this one maybe I have a photograph of a furnace which will probably give some idea I do not know whether I have okay first let me have you written down? Okay. Typical furnace is shown here another photograph is better but this is just to show you this is typically what a furnace looks like okay it depends on how many tubes you want this is a 4 stack furnace 1 tube, 2 tube, 3 tube, 4, 4 stack furnace this part of the back side of the this essentially has all gas entry systems different kinds of gases are required at different flow rates and they have to be given to different tubes so this is a gas flow system which is just rear side of the entrance of the every tube for that. So this is a very huge system typically a furnace may be height around 9 feet to 12 feet depends on how many tubes you have and what size of tubes you are actually going to use the better figure is somewhere here this is a typical 4 stack furnace shown here okay and one can see these are the this exits you can see here a small table is I mean this rider is kept so that you can go and put something above so you hide you can guess what it is okay. The kind of clean room requirement traces is shown here bungy jumps, bungy suits absolutely not known who is inside okay that is the catch in that it should be so much covered that except for your eyes nothing should be visible and you have two specs on that one internal specs one upper specs so you will not be able to see who is in okay. This is how wafers are loaded as sure this is not very good wafers are put in a rack and loaded okay this is a quartz tube which is used in all diffusion furnace quartz is the highest purity glass and it is crystalline it is melting temperature is as high as 1800 and 12 degree centigrade so it is all used up to 1400 degree centigrade quartz tubes can be used okay now this is the cap which sits on this okay which sits on that okay this hole is inside so the way it is done is this furnace is in a semi clean area which may be class 1000 or class 100 area but it is project this tube outlet is projected in clean area so wafers are introduced from clean area clean benches and a clean environment so no dust sits on wafers so that is the kind of furnaces we used and as I say each furnace per stack as I say per tube these days please remember this size is roughly 3 inch wafers this is a 4 inch tube can have handle 3 inch wafers or preferably 2 inch wafers if you have a 12 inch wafers you need to have 16 inch tube okay. So what is the importance in that larger the tube means larger heating element has to be provided so much this larger amount of gases has to be flown volume is so high okay. However larger the size of the wafers more chips will come out of a single wafers in one run so that is why we still want to go higher and higher for the sake of profitability okay. I have started work in 60s with 1 inch wafers 2 inch wafers I have 3 inch wafers and then of course then I came here and everyone say why not make a 6 inch wafers furnace I say the cost of 6 inch wafers furnace is 12 times that of 3 inch wafers furnace. So I just wanted to show you that the furnaces as such you see are very very crucial and the kind of furnace system at the end maybe we will talk more about it or types of sources which we use in diffusion are 3 kinds solid source, liquid source and gaseous source. The solid sources which sits actually on the wafer is called primary source the impurity source which actually gets impurities into silicon is called its primary source. You may have any other source but must convert it into primary source okay so that diffusion will only occur through primary source okay. There is also fourth possibility which is called spinon sources good for cell steeper okay. All impurities used in silicon processing are n type as arsenic, phosphorus, antimony, p type, bowon, indium and I intentionally put here because aluminum, gallium are rarely used but they are available they can be implemented. Everyone thought indium has a very large energy gap energy from the valence band 0.16 so it will never be used because how many holes it will create since the energy is higher e to the power minus E s by kT so larger the energy smaller number of holes it will create bowon has 0.08 or 0.06 degrees so it will create almost a room temperature all can be ionized so all holes can be achieved whereas in indium unlikely to get larger concentrations so indium is not used. Aluminum has another problem what aluminum is also has a diffusivity in silicon but also is alive in silicon now if it does both then one does not know how much whether aluminum is going or not going and how much it is going it also gets too fast oxidized what is it aluminum oxide is called NINI what is the commercial anodized okay whatever aluminum frames you see they are all oxidized and polished and called anodized so they are aluminum only okay. So aluminum is not a very good impurity to diffuse because it may form an alloy quicker than it diffuses down but it does it is not that it cannot be it is a type 3 impurities it will create holes okay but there is a slightly higher energy of this so one does not use aluminum as any time an impurity source the aluminum has a strong advantage in source drain contacts because let us say I have a P type source and P type drain and I put aluminum contact so even if P is not heavily doped this aluminum top doping during the higher temperature processing will make a good alloy and good omics it is good for P channel devices particularly putting aluminum contact but aluminum is now almost out of all IC processes barring exceptions we are looking for higher conductivity materials aluminum is very good conductor everyone thought but copper is even better so why we did not use copper copper is very strong oxidizing itself it forms copper oxide which is insulating. So we want now protect oxidation but we run copper line okay earlier when we made first lab in 80s there was not a single copper tubing or copper material in whole lab because copper is a poison it gives almost a level in the silicon which is near mid gap so we wanted to avoid copper everywhere now it is the copper is the major interconnects okay 20 years earlier it started in 98 from IBM damson process the actually first time introduced copper interconnects what is the advantage the resistivity ratio of aluminum to copper is 3 times so obviously for the resistance of a line will be 3 times smaller compared to aluminum for same size which means the RC time constant will go down which means speeds can be improved the issue came circuit devices could give faster outputs but running from one point to other is slowing down so it says interconnect related problem so first attempt was made to improve interconnect itself okay so that is why aluminum so this is additional features you should know gallium has even deeper impurities so it hardly used okay there are some sensors now being used with gallium okay but you read some may be asked to say Ram Gopal antimony is n type dopant can be used but again it has a larger energy gap therefore rarely used and it's a much bigger it's other of course it's misfit factor is small so don't think it has a better compared to both phosphorus and arsenic of course is the best but as I say size is bigger and energy is too high to create large number of electrons okay there are also other impurities we introduced is good very important for us because faster devices particularly the bipolar devices speed can be improved by gold doping's okay these are called the combination centers okay I'll leave a word for you gold is amphoteric what does it mean yeah so it is an amphoteric impurity it gives both levels acceptor level is higher than the donor level if you look at phosphorus the impurity sources are solid sources the phosphorus available in the market even if it is purified is called red phosphorus it's a powder form and it's available as red phosphorus the red essentially word came because it gets little bit oxidized and it gives reddish color so people used to call that phosphorus as a red phosphorus but it's essentially only a phosphorus with little oxidation on that so you will have to reduce it to make full phosphorus the other source is what is called die basic or even die mono ammonia trichloro or die tri also is possible tri basic ammonium phosphate mono basic ammonium phosphate but mostly use this die basic NH42 HPO4 what will be mono one here three here okay so if you keep changing the phosphate with reference to NH4 it can be all three possibilities but most used source is die basic ammonium phosphate for phosphorus for the 542 forms P2O5 please remember P2O5 is the basic source for phosphorus is the primary source for phosphorus P2O5 is the primary source for phosphorus diffusion so if I oxidize it I shall create P2O5 the reaction further when during diffusion will be once I have P2O5 on the surface of silicon P2O5 will react with silicon and will form phosphorus and SiO2 okay so phosphorus will get in and oxide will be on the top okay this is process of diffusion phosphorus in oxide top okay it does it does but still thin enough and since you have a gas flow down it has a more concentration in the outside so the diffusivity is much stronger in than what is coming out internally there is no gradient it is opposing the gradient so fewer impurities will come out but larger will go in there flux both side in any thermodynamic system X is going to YY will come to X okay it depends on the given temperature followed reaction is favored or the reverse reaction is favored actually you must have seen I have not put equal anywhere error essentially says that at different temperatures reaction can be either way okay it can be from right to left or left to right okay this mixture which is mixture of SiO2 and P2O5 is called phosphorus silicate glass okay what is it called phosphorus silicate glass and actually that is the source of impurities in fact okay the liquid source which I may use is very only known source of phosphorus is phosphorus oxychloride which is POCl3 and it is a liquid at room temperature and the reaction is 4 POCl3 plus 302 is 2P2O5 plus 6Cl2 now we will see that this is not very favored because chlorine is a halogen which pits silicon okay silicon has a reaction with chlorine, silicate plus chlorine is silicon I mean silicon chloride 3 forms 4 forms so any of that may actually remove some part of silicon then there is a source which is gaseous which is called phosphine okay there was a gas called phosphine what is that gas oh so I have you are all born after that so I thought I may not be knowing hopal tragedy is very bad because of the phosphine gas which is methyl isocyanide now MIC so this the problem with all these gaseous sources which I am using in specific are extremely toxicity for example let us say if it is 1 part per million and 10 to power 6 one atom one molecule is the level at which phosphine was toxic to human body of course it was much higher than that in phosphate it was actually released so there was nothing stopping it but pH 3 is 10 ppb okay so if you can see even 10 ppb can be released and you have a problem the only advantage of pH 3 which I may say you since I have used it and maybe I have I do not know where I inhaled but I ran for my life was it has a fishy smell so when it leaks you can run you have all oxygen this inside just run outside as fast as you can okay break every door if possible okay so say it but do not think it all there are hardly any accidents in the labs many years ago there were few but now we take so much precautions that it is unlikely things will fail okay so it is called fail safe operations okay if things happen some walls will be closed immediately automatically we do not do it so much volume increases it will switch off everything okay the phosphine gas at 440 degrees centigrade breaks into 3 hydrogen plus 4 2p okay of course these equations are not every time balanced in case they are not balanced you balance is that's arrays only showing left to right that's it but I tried to balance but I am still telling you they are not balanced just have balanced them out so this 2p that is phosphorus which is coming out of phosphine at 440 degree so this reacts with oxygen and form P2O5 so your basic source you wanted it so this phosphorus then will react with oxygen and form phosphorus this and if you have silicon around it will also form silicon dioxide so phosphosilic glass which is the source of impurities 5 SiO2 plus phosphorus okay example shown here this is your P2O5 silicon sitting on silicon when I drive in or when I push or continue this these impurities get inside SiO2 is created okay this may also contain phosphorus but that I will etch so I will have fresh surface of N type in P as I said I have already written there pH 3 is highly toxic gas so what we used for safety is we diluted with nitrogen typical dilution ratio is 99.9 nitrogen 2.1 phosphine so if I have 100 liters of nitrogen I will add 1 cc of phosphine to that whenever I work in a lab I take lot of precautions that it does not leak through anywhere but luck is bad if your luck is bad we had a student who was working earlier I am talking of 80s everyone probably I do not know how much hydrogen is extremely flammable gas but we actually burn hydrogen that is the way the hydrogen is removed in fact but it should come out of a capillary you should actually fire it okay this is a typical diffusion system shown here the furnace internal part is shown here this is my quartz tube this is the cap which I showed you earlier just a minute this is the tube and this is the cap with exit tube okay there is a there is a heater here all around this is only shown cross section this heating element is resistive heaters actually the tube is surrounded by some kind of a former which is made out of a material called mullite what is it okay mullite is a plastic created resin which is like a bricks of white you know ceramics you must have seen white bricks so they are mostly mullites so actually there is a mullite rods are made hollow rods and tube gets inside and there is a slots in the mullite when making like a brick making and this then there is a heating wire which is bound around this mullite which covers this tube and you give a normal enough current 200 amps currents to heat this filament now the important is what material I should use for a good resistive this it should be very highly resistive and stand to higher temperatures so what could be no no no the wire should be what heating wires I want a tungsten is a good material firstly tungsten has a little fragile so when you form it it breaks many times so one does not use tungsten nickel chrome nitrome as the word so nitrome wires are they are more malleable and therefore they are bound around this now there are three coils one in the center area one in this two ends of the furnace there are also thermal sensors two here three here and two here now what is the purpose of these sensors because the problem is depending on the your run number of wafers the wafers I of course it is not shown here there is a quadrack and there are slots inside and wafers are sitting inside like this same height of course now since each wafer should see same temperature and same flows this is called center zone so the center zone temperature should remain unaffected by the flow as well as by the change in your currents so what they do is you have a end end temperature monitoring early temperature monitoring as well as center zone three monitors okay these are fed to a PID controller and we keep on monitoring when to switch on or switch off the power to the coils individual coils and therefore since PID controls are very fast we are able to control the temperature in the central zone 0.25 plus minus degree centigrade plus minus 0.25 degree centigrade so you have here furnace I may have 1100 plus minus quarter degree centigrade accuracy which is very good in any sense okay so these are called resistive heaters there are other heaters which are not resistive what are those called inductive heaters so we will see in other processes we may use that is called cold furnaces inductive heaters are normally called cold furnaces these are actually hot furnaces huge huge heat now to monitor this there are variety of thermal sensor RTD is one the thermistors are others but thermistor at that temperature you cannot this so we normally use thermocouples which is typically of the constant and copper if it is outside if it is inside it should be rhodium platinum inside a quartz tube there are new sensors are also appearing using mams okay they are also very accurate so we can monitor better temperature sensitivities and these are all newer techniques with new thing but PID controller is still as it was okay it has a micro resistor which is sitting along with PID and you set the temperature beforehand pass gas flows till you don't load the wafers okay and when the temperature settles down then only you push it there are digital displays to know where is what temperatures okay the gas flow control is something like this for example I want a O2 gas so there is a flow meter here with a wall I can pass oxygen I can pass nitrogen I can open both wall and pass nitrogen and oxygen together I can have a gas flow of a phosphine I can have phosphine with nitrogen both of them I can go through this if I don't want I have a liquid source so I have a bubble air I pass nitrogen inside or argon sometimes and this vapours of this liquid is picked up and passed inside okay if you have a solid source then the solid is actually solid source is kept right here small cube principle which is rarely used no input phosphorus oxide or red phosphorus near the vapours okay the importance of all this is that when the gases go in whichever form they go in they must form a laminar flow okay and that is very important two parameters which we control in this is called which numbers Reynolds numbers and Rayleigh's numbers so we actually monitored the Rayleigh number and our number to get laminar inside so that the wave every vapor when the gas goes it sees uniform flow like this this is very important in processes please remember there are two kinds of flow meters which we use what are these two anyone heard of them one is called what is called rotameters okay the rotameters are used in almost every industry whether it is a fluid chemical liquid flow or gaseous flow it is essentially a capillary which is graduated for particular gas densities you pass the gas and there is a some float inside it may be stainless steel or plastic depends on the gas density and this small capillary is around 8 inch height 10 inch height the gas flows and since the gas flow pressure is from below it pushes the float up okay and it is balanced by the actual atmospheric pressure from the down so depending on the gas flow you can calibrate where this float will so you can see a graduated scale you say okay 100 cc per 200 cc a liter whatever scales you want you can get these are called standard flow meters which are not very accurate because every flow meter is for different gases and their graduation is not very accurate but let us say I am passing 120 liters or say maybe or 50 liters per in a large tube it does not matter if it is 51 okay it does not matter if it is 49 but if I am passing 100 cc I am 100 cc there okay so the accuracy of flow is required in certain cases otherwise it is not to make a very accurate flow inside it should be independent of pressures so what are those called these are called mass flow meters they are very costly MFCs mass flow meters MFCs have advantage that you know it is like a tube here the gas goes there there is a small additional capillary tube goes up maybe I can show you how it is this is your main tube this is something like this another tube okay then depending on the gas you enter there are three elements here this is a heater element this is heated so what happens if the gas flow is larger and if it is already heated sensor it cools okay so there is a temperature initially because of the gas is T1 but since there is a heater here it actually increases the temperature here the difference between the temperature of two sensors is essentially monitor of the flow going in okay so we can set this temperature initially which is called set points and if the gas is larger then it shuts off this valve here will be shutting off if it is smaller it will allow more gas to come in and very accurate amount of flow can be adjusted through this this whole electronic system this is huge circuit here actually the flow mass flow meter is typically around 8 inch by 8 inch kind of system and it cost there is a magnetic plunger there are too many things inside to close or open and these are very costly typical flow meter 25,000 above okay whereas the rotameters are available of 150 rupees okay so accuracy to money okay many a time these do not work so our experience is that we open it and close clean it and do it again but 90% of the scientists do not do it by another one okay so this is a typical diffusion system and I just give you all this janta this which otherwise no one will know tell you or no one will write about it this is only we did many furnaces other cells so we know what we did okay so this is the better part of our technology all that we were making our own systems there was nothing available so we made our systems then the we have seen phosphorus let us look for boron this figure is nothing to do with phosphorus this is for all kinds of diffusing systems okay there are as I say aluminium and gallium has a better misfit factor than boron indium has 0.22 boron has 0.254 okay even then we only use boron as I already said aluminium is a bad it forms an alloy so we do not want to use then gallium is essentially a very high energy it has a very large activity energy so we do not want to create very few holes okay then indium also is 0.16 electron volt so we do not want to use it boron has 0.086 as the activation energy so it is used very extensively since this misfit factor is larger so what is the effect its solid solid will be relatively smaller so it is 400 upon 19 in some cases maybe 500 upon 19 per cc all that boron be so p plus doping is always less than n plus doping simply because the misfit factor of boron is larger than phosphorus or arsenic of course is the best but arsenic as this gentleman also said that is a slow diffuser so in a shallow junction they are good but if you are looking for deep junctions phosphorus is the only solution so first why others are not so I just gave you a why others are not used solar sources is B2O3 powder is available boron nitride wafers are available and a very marketable material boric acid which is used in preservation of grains okay so S3BO3 these are the three solid sources of course the one which we use in the market is not electronic grade they are only industrial grade and they have lot many impurities so never used in labs anyway okay we already said the kind of grades we are looking for is minimum 6 9 and above okay purities 6 9 is the minimum purity which we use preferably 9 9s okay 7 9.99 okay so these are the solid sources you I mean I already said twice why why not aluminum why not gallium why not any other only boron they are reactions the B2O3 plus 3 silicon is 4 boron plus 3 SiO2 and mixture of B2O3 plus SiO2 is called borosilicate glass popularly known as in market there are three kinds of glasses available in market borosilate essentially know it is exactly this borosilate glass it is called Pyrex okay very famously named Pyrex which is coming from borosil company okay so it is a Pyrex glass but it is not very pure and it is not very good which is worse glass than this soda glass the whole drinks all the time you I do not know I these days I do not see many people taking cold drinks but in our time I think it was very much popular so all cold drink bottles are essentially of soda they contain sodium and no glass which has sodium is allowed inside our lab because if that happens everything will be or if for us so sodium is avoided so soda glass is never allowed Pyrex at higher temperature may release impurities okay so we do not want high pure high Pyrex around glasses except when they are not used in heating areas they may use Pyrex systems but anytime when we are going we should use only quads plus this so that is the catch in word everywhere okay there is another as I say boric acid also can reduce to at 185 degree B2O3 plus water water flow steams out and you have a B2O3 source for this case two zone furnaces you first you put S3BO3 oxidize it little bit and when it pass it forms a B2O3 near the surface and that is picked up on silicon there is also a possibility of making a solar sail with cheaper sources which is called boron nitride sources there are wafers of boron nitride so you put silicon wafer in touch with boron nitride both sides silicon so both sides are so porous surfaces of silicon touches a boron nitride wafer in a slot and slot is so adjusted all three wafers are tied each other okay such number of wafers two silicon one boron nitride two silicon one boron nitride and that is heated to create boron nitride breaks with silicon into four boron and nitride instead of SiO2 now silicon nitride is the upper layer which is also a mask for imperatives of course not as good a mask as SiO2 but it is still a mask the last source is liquid which is boron tribromide BBR3 BBR3 reacts with oxygen to form B2O3 so what was our primary source of boron B2O3 is our primary source so whichever source you may have you must convert this into B2O3 at the surface of silicon then only the next reaction is possible now this is the problem boron is highly toxic as well as it is very highly reactive to silicon so it gives what is called as halogen pitting so when we are making 5 micron devices or depth of 3 micron junction depth that pitting was relevant but with 100 nanometer junctions the pitting may be of 100 nanometers so there are issues now which in 25 30 years we never bothered but now known as uses BBR3 earlier time almost everyone used BBR3 so this is their progression now mostly people will not even use either of them if you want still a liquid source then there is a source called trimethyl borate which is used now TMB is very popularly known TMB plus oxygen when it oxidizes it creates B2O3 carbon dioxide and water. TMB has a problem that it has a very high vapor pressure so it has to be refrigerated to keep vapor pressure down otherwise it will blast out so it is always kept refrigerated and even when you put it put in a water bath and then only use so the process is this is trimethyl borate try this is methyl group so 2 CH3OB plus 9 O2 at 900 degree become B2O3 6 O2 plus 9 H2O very popular source it also is very controllable source and therefore even now in many ICs some labs of course Intel is not using but TI of course TI is now selling of all its process but TI had TMB process many times people do not change to something else similarly if it is working just for the heck of it why should I change if it is not working for my requirement I will change so many companies have been doing something which they think is working so why change so why money but we know here better sources will be gaseous but it is a matter of your decision how much money you have the other possible gaseous sources are boron chloride or boron di boron as called B2O6 they are all gas sources BCL3 reacts with 302 to form B2O3 and 602 and again the same issue it may pit normally chlorine does not put as much unless there is a hydrogen around if it is a CL is around it will definitely quit but if your system has hydrogen free it will not give that much pitting it does reacts but it has a very marginal silicon reaction with H around a CL it will pit it why I am saying you because some of the gate oxide grows where earlier using chlorine as a improving species for interstates and we used to use chlorinated oxide and then we realize that it is actually creating more problem than this so we changed to other TCA's and other chlorine sources BCL3 plus 3 H2O2 boron plus 6 HCI and here is that issue if there is an HCl there is a pitting 4 HCl plus E this is what it means silicon chloride will form so some spots this may go if it is hydrogen free absolutely no issue quickly di boron is a highly toxic gas very carefully has to be handled and it oxidizes 300 degree to form B2O3 the gas flow is one parameter for the amount of oxygen plus this mixture of these two gases the mixture decides how much concentration you are going to get okay as I say they are and it should be laminar decided by Reynolds number and Rayleigh's numbers okay how do I test that the boron diffusion is okay this is not advised by for many but people like us have done that you see in the other end of the tube and if you see a two good lobes of heart you say you are perfectly laminar flow going in okay this is called boron lobes so you actually see otherwise it will be mashed up you will see a mixture of everything if you see a good lobes you say you are perfectly under this okay so I just told you how do we found out that it is okay this is how we figured out okay last source is that okay B2O6 by 3O2 at 300 degree becomes what is primary source I declared B2O3 so all processes I show you reactions I show to convert that whatever source you use into B2O3 at this elegant surface okay primary source so we are always converting all secondary sources to the primary source that is the way last slide for the day not last for the course the meaning more to come just you will finish and come back the last is arsenic impurity which may use very often these days for example the solid is arsenic oxide which is also called when arsenic oxide plus SiO2 is called arsenic silicate glass rarely used because AS2O3 is also little poisonous so it is hardly used or hardly sold of course sold I do not know these days anything is available whether yesterday there was something that cobra venom is people like so I am surprised I just fear so much cobra and they take venoms fine so ASH3 reduces around 320 25 degrees to arsenic and hydrogen arsenic reacts with oxygen to form AS2O3 and AS2O3 plus Si is SiO2 plus arsenic so that is how arsenic gets in and SiO2 is created on the top is that okay so next yes they are exhausted at least 30 feet up and then if there are toxic gases there is something called scrubbers okay so they pass through solution where they are converted into solids or solubles which are non-toxic