 So far we have seen the theory of implantation. We have also seen the profiles which ions can get into silicon or for that matter any material. This sheets are taken from website from Sherifa University of Iran. I hope some of you may be knowing, someone is here from that university. Those who are doing their projects probably many of them may be knowing there is someone who is here in our faculty who is from Iran on same university. So I have just once thought what is our university. So I have found this slide. So I just thought it is a good photographs. Okay. We are trying to do something what is an ion implanter. The basic ion implanter is something like this. We have a source of impurities which is normal in the gaseous form. For example in the case of boron, we have boron fluoride BF3. In the case of arsenic it is arsine. In the case of phosphorus it is phosphine. Okay. Or in the case of antimony or any other you have to first create its gaseous source. Okay. Now this gas is then such as BF3 or arsenic arsine you know it is accelerated at high potential through in a chamber which is shown here and as this gas inlet starts in there is a large electric field which actually breaks this ions into BF3 breaks into fluorine as well as boron ions. Now if we get arsenic boron or antimony or phosphorus ion off either of these species along with hydrogen. Okay. So first thing is I must separate arsine or anything from or some part may not be even ionized. So there will be some gases which are anionized. There will be some hydrogen and some species which you want to implant and there will be some impurities as well in the gas stream. Now since I want only one impurity to come out okay that is I want arsenic to be implanted. So the chamber may have all kinds of ions but I must take out only one of them which I want to implant. Okay. So this here that is called mass analyzer. We will see what is it mass analyzer. So I must somehow get out of this whole magnetic system which we will show you discuss only those species which I want okay and that when it comes out then I accelerate them by large electric fields then there is a deflating system scan system which is also electromagnetic system and the beam then moves along accelerated ions or beam ion and there is a wafer holder here on which they impinge okay. So few things you have to understand that the first thing I must get some kind of a source of impurities in a gaseous form. So let us say I have solid form okay fine then there may be a crucible inside you actually heat okay and see to it that you get vapors or gaseous form. These are then introduced in this chamber called plasma chamber and in this large electric field breaks the gas and it releases ions. These ions then pass through a magnet which is electromagnet as shown here and somehow bend only those ions which I want okay. Rest will be hit elsewhere and only in the slit only those ions which I am looking for probably will pass through. They will be passed through a large electric field so they get accelerated ions get energy from the electric field and they essentially get half mv square which is kinetic energy which essentially is equal to qv okay energy of the electric due to electric fields. So if we go through this they will accelerated large energies this is what it requires energies so you say I want 10 kv, 100 kv, 300 kv it is this acceleration system which is typically a periodic system which accelerate these ions to energy of your choice okay may be any amount you can of course larger the energy this electric fields here will be very high and any transforming system which puts may not be able to stand that high voltages. So therefore no more than 300, 400 kv implanters can be made easily not impossibly then of course as I say there is a standard deflecting system you have done through this and all CRT systems you have magnetic fields and electric fields which can scan the beam X and Y okay. And this beam passes and there is a place where the wafers are kept is called Faraday Cup okay it was given an honor to Faraday okay. So why it was given because this tube kind of system is essentially a normal cathode to anode kind of electron going or ions going from one place to the other and that has a Faraday discharge system so the place where ions are picked up on the wafer is essentially called Faraday Cup okay so and of course there is a positioner where we can actually change the position of wafers in the sense the beam itself can be scanned or some mechanical motion also can be given to the wafer holders okay we will show you the checks which we have. So this is typically what an ion implanted does okay so few things one must understand that how only one kind of particle one kind of ions only come out and how they are accelerated and how they impinge on the target wafers which may be more than tens of 20s or 100s okay there must be some wafer feed mechanism through which why wafer feed become please remember gas can only be ionized if there is a low pressure okay gases can be only ionized if there is a lower pressure so typically pressure may be the order of 10 to power minus 3 torr highest torr related to atmospheric pressure 760 torr is 1 atmospheric pressure 1335 Pascal's is 1 torr so there are 6 units of pressures okay so look for I may give any even I may give a change also okay do not worry. So typically this is 10 to power minus 3 vacuums are normally expressed in terms of the pressures which is torr cell is torrs and one can see from here that needs some kind of evacuation system called vacuum system so this chamber has to be and it has to be evacuated by pumps so there are 2 kinds of pump rotary pump followed by diffusion pump if you want much lower vacuums then you will have to put some kind of turbo pumps or some kind of turbo pumps but as of now diffusion pump can go up to 10 to power minus 6 torrs so that is sufficient for an implanted okay. If you go to the lab all rotary pumps are kept outside but diffusion pumps are sitting just below the vacuum system okay so a rotary pump is required to start a diffusion pump at least it should give some torr lower torr then only diffusion pump starts acting okay some other day when I show you actually evaporation I may show you what is the diffusion pump. So this is a machine the only relevant parts have been shown here it is the same thing what I said this is an ion source there is an analyzer there is an accelerator there is a scan horizontal vertical scanner and this is the way for target. So this is the brief what an implanted is trying to do you start with the ion source pass through a magnet which is electro magnet which is analyzer and then it you get a particular ions you actually accelerate and focus using this vertical and horizontal scanners and a pointed beam is actually hitting the target. These are all electro magnetic lenses which is actually actually can focus any ion beam so this is typically an ion implanted is a machine works maybe if I have a figure I will just quickly show you you can see this is a 20 by 10 20 by 15 kind of a room which can take care of one implanted larger the energy implanted you want and larger the ion mass you are you will have larger magnets so larger area larger acceleration tube length has to be larger so everything become larger as you large ion numbers you want large energies and you also want large mass ions to be separated larger the everything will equal also require larger space a typical implanted room is of this size actually this is the panel and this is that check you can see from here on that check right now for example they are shown some 12 or 14 wafers you can see from here so what is the throughput at a time I can at best put 12 wafers or 14 wafers in a any other diffusion furnace 200 is the minimum lot so I can push diffusion for 200 here 12 14 16 of course nowadays they are checks which are 48 but 48 means such big 8 inch wafers think of it 48 so how much implanted will be large enough it should be able to scan such a large lengths and widths so huge money okay huge power this is what an implanted looks like the next as I said we look into the each part the first is the plasma source typically I just now said a plasma chamber you have an gas feed from here and you have a pump system which evacuates the chamber okay and one can apply electric field be external voltage here between the plates and this this and as you actually start ionizing the gas there is a slit here which also should have some electric field so that ions can which kind of field it should be a positive ions have to come out negative charge plate must come so that they are accelerated outside okay so there is another chamber out which has the negatively biased so ions come out now as I say the pressure inside this chamber is around 10 to power minus 4 to 10 to power minus 2 Torr and ion source is characterized by ion current density okay this d distance is essentially between the extractor and the source this is called extractor so from extractor to the slit the distance is called d so the ion current density is 5.5 10 to power minus 8 external voltage whatever extractor voltage I am extractor means which extracts the ion out okay that extractor so this extractor voltage to the power 3 by 2 d square the gap to the m to the power half and its units of course are ampere per centimeter square please remember the area is not very large here slits are very thin small so the currents are relatively larger okay so if you have a larger currents the cost of implanter increases typically it can be say 100 1 milli amp 1 amp current implanter will be cheaper than 10 amps and 100 amps okay because larger the ion current density you are looking out larger will be the cost of the equipment which will require okay so typically I can decide the decision of ion density is decided by what extractor voltage is I am allowed what is the distance between slit out and the external extractor chamber and of course the mass of the ions which you are taking out which ion like arsenic is 72 or 73 phosphorus 31 so depends on which species or boron is 11 so which species is taking out that will decide the ion current density d is the distance between slit to the extractor chamber this is the from slit to the extractor is the d okay so the first part is ionize the gas and extract out ions the next part is as I say this plasma chamber have all kinds of possible impurity gases along with the so you thought that you have got all ultra pure gas introduced but the species which you are introducing are seen or this may be ultra pure but ultra pure is only say 69 or 89 purities that means there is a residual impurities the chamber itself is made of steel itself d gases so it also releases impurities and there are many other source of impurities inside and they may be small amount part per billion or less than part per billion but they are there but I do not want any one of them to come out of the slit that is what I am really looking but they will right now extractor will pull all ions whichever ion is there it will pull out okay now these ions I want to somehow now see that only species of my choice comes out okay so I pass through a magnetic system let us look at a electromagnet maybe first sheet again okay this is a cylindrical electromagnet and since it is a magnet is taken in a ring form it has a radius of curvature it is a radius of curvature is known the magnet size and this radius of curvature is actually known to me okay now this means if an ion is entering this magnetic field it will experience magnetic field for magnetic force what is that force called Lorentz force and since it is passing it has also accelerated and it has a mass it has a kinetic energy of half MV square which was also equal to QV external voltage extractor voltage that is the energy which you give them okay when they pass through a this slit through a magnet they will be actually passing through and they will have a motion which is which receives a force called centrifugal force so it actually moves in a direction which is decided by MV square by R V is the velocity okay which must be balanced by the Lorentz force which is velocity cross B or VXBZ if X is the direction of motion across orthogonal to that is the magnetic field so V is QV cross B MV square by R by solving all this I can get a term R is the radius of curvature B into B is the magnetic flux available with this it is expressed in Gauss what does Gauss means how much lines per square inch 6.4 magnetic lines my memory force lines per square inch is the density which is equal to one Gauss value okay so if I calculate B into R is the radius of curvature where MV square by R is R is the radius of curvature through which they are bending okay so if B into R is 2 QV external by M divided by QM QQ will cancel so it actually can be written as 4.55 under root M times P extractor voltage. Now this B into R is essentially called magnetic rigidity and that is the feature of any permanent magnet this has nothing to do with this but there is another small BR anyone has done second years well I hope so there is a BR of a magnet which is small R not capital R what is it that is not just now I said BR is the magnetic flux density or so many lines per square per square inch or per centimeter square that is if I plot a BH curve okay may be little bit of what do you do this will be a star this is so the maximum value of this is BR what is this value called SC what is SC X is a magnetization current density at which the magnetization becomes 0 okay SC these are the numbers which are given for a given magnets okay and the water energy store is B cross H inside this BH curve okay some other time since I designed a permanent magnet motor way back I have studied lot many magnetic systems okay so once of course this BR and that BR should not be confused this is R is the radius of curvature so if R is fixed for the magnet is that clear if R is fixed for the magnet I can now say that this magnetic intensity B or magnetic flux density B is proportional to root of mass is that clear everything else is if I keep extractor fixed this is constant R is constant so B is proportional to root of M M means atomic mass or atomic weight of that species. So if I want this any species to come through the area of curvature of R which is fixed by me all that I have to do is to adjust B which means the curve and how do I adjust B B is proportional to what how do I if I have a core which is normally iron or steel and I put a wire around what is the light follows and pierce law okay if any of the number of turns I is the current flowing in this in the coil and I am you by L L is the length of the core so N i mu mu is the probability N i mu by L is the magnetic flux which it can receive this is the Ampere's law very old Ampere's law so this simple Ampere's law tells us how much is the magnet so what is that other things are constant so what B will change I since R the everything is fixed for a electromagnet all that I changes the current in the electromagnetic coil and that decides my B if that decides my B that decides my M is that correct for a D and B there is only one M is possible come through R okay so if I want arsenic so I figure out if so much Ampere per this current I pass through this coil so many Ampere's this arsenic arsenic will come I change the current go on may come I change the current first one is may come is that clear so all that I do is fixed currents are available known to us at for each PC which I want to take out is that clear however if you want to change further if you change the extracted voltage which you are not allowed but there are systems in which these days they allow extractor voltage to change then the BR changes and then the different currents you have to plant with current now I should have this reason is there are 2 reasons it is called pre-acceleration and post-acceleration V extractor will accelerate ions then there is accelerating voltage also further ahead of it so how much pre-acceleration should be done and how much should be post there may be additional post acceleration pre-acceleration system so now it is decided that this voltage itself is variable you can actually vary it okay but then your currents will also proportionately different for different M because B will be varying for that is that clear so B has to be changed corresponding to M required and that is how called mass analysis is that correct this system is called mass analyzer so any species may come but through the radio curvature which has a slit inside the magnet only one species will come out because you have fixed a current through which only it will bend through R okay this normal angle is 45 degree around like this but it can be 60 degree then you will have to recalculate how much you have where do you want to take it out but there is no harm of any angles generally it is 45 degree which you move your ions come and then through 45 degree they become 90 45 45 so normally at 90 degree it moves out okay now how do I know that which gas is coming also so I can actually take a spectra through a mass I do not know how much your chemistry there is a infrared spectra can be obtained and I can see for spectra for each of the gases spectra this so I know if I have a gas I want this specific species please remember these are isotopes of BF3's there will be fluorine there will be BF2 there will be so these which species also is to be decided through what is the maximum mass it is going to give us that you will have to evaluate a priori okay so this spectra for each gas I get it first I get the spectra for it for phosphine system boron boron system as well as for of course these are provided by the people who is apply your gas you do not have to analyze this analytical graphs are always provided to know which species has what amount of gas or what amount of impurities or other other spectra it has okay now this the next of course of as I said you after I know which species analyzed I also know I pass through magnet I got one species out and then I apply some kind of a ion acceleration through electric fields electric field coils are shown here you have a this is called resolving aperture then you have to maintain certain amount of pressure inside okay and you accelerate the ions to it what is the acceleration kinetic energy is Q times V x extractor plus V which you now apply is that how much energy it will pick up Q into V extractor plus whatever V accelerating voltage now you will apply so net energy now is Q V x plus V V up you are further apply okay so that is the net energy and that energy is what is provided to the ions is that clear I repeat you already accelerated partly extractor through extractor so extractor plus additional voltage which you apply Q into that is the energy which ions will actually receive so for example they are given example if external voltage is 30 K extractor is 30 KV then I apply accelerated voltage are 70 KV and the ion will have energy of 100 KV so this that is why I say nowadays this also is adjustable okay in some implanters we extractor is already already you have extracted out plus whatever accelerating voltage you will put Q V x plus V accelerated is the net energy this ion will receive okay so you have energetic ions okay which are coming out is that okay gas is ionized accelerated by extractor voltage pass through magnet re accelerated and your ions are specific ions with a given energy which you want to actually want to implant on okay is it okay then you have a XY scanner this is a vertical scan and this is a horizontal scan right now it is both electrostatic but it could be magnetic as well and then there is a some kind of a chuck where this wafers are holding okay maybe this system is also decided with the current of ions which you are actually picking up is that current larger currents will require larger deflections because so many ion what does that mean number of ions per centimeter square will be very high current density is higher so now you require much higher voltages to deflect them out okay smaller currents smaller displacements okay smaller number of ions so this is something a deflecting system which will have to design for a given amperage which you want okay what is the current to do something is more important because why we are constantly talking of current I have first day talked about but let us say again why are we worried about current in the implantors okay what does current means it essentially measures the ions per unit time moving is it that is the current so larger the current means larger number of ions are flowing through this okay so the number of ions which will impinge silicon will be larger larger current means number of ions will be impinging in a larger numbers okay and that is what we want to say how much what word we said days how many impurities I want to push in it per unit area is decided by current which I will pass through okay okay okay so here is the last part of the implanter or the ion beam is coming there is some kind of a cylindrical system which is called the Faraday cup or Faraday chamber please remember in earlier version the wafer holder how do you measure the current as the ions strike the wafer since they are charged and if the wafer are grounded then the charge flows through and current is monitored is that correct so this is the emitter which I shown here this is how the current is monitored but as the ions keeps coming this current will be also time function is that correct it is a time function and what I need at a after a given time I want to see how many impurities have gone in per centimeter square okay that is called dose so it is a 1 upon QA 0 to T dash is the time for which implantation was performed I T DT so what is this circuit will be I should be an integrator an integrator is required to actually find the dose is that clear integral I DT is essentially used a integrator there so this integrator along with this is a constant we know QA which is scaled down directly so direct measurement on emitter you can see how many how much is the dose QA is fixed because you know area where the beam will strike how much area and you also know this integrator will actually integrate the I T DT is what exchange essentially I charge per unit time is that correct so this is the net charge received is that correct this is the net charge received okay per unit area is that correct I want to give only charge okay monitored is current okay so this is how actually we involved so normally how do I know can I keep monitoring what is the dose no there will be some kind of comparator you fix your dose value for a given voltage equivalent and you keep monitoring this not you might it will you fix that value and when you turn on when the comparator will switch off the iron source whenever that dose is same as this this is a very simple technique which allows you to actually fix the dose okay so you set the dose thought I am a planter and after when the dose is same as what you said the machine will switch off so is that point clear so what is an iron implanter essentially I can put number of and those has been in our case what is the name we gave for those NS in our profile NS is the dose we were talking about in the Gaussian profile and what is the value we said integral NX DX minus infinity to plus infinity is the amount of impurity per unit area gone in so that is the dose is that clear so any profile should be integrated from minus infinity to plus infinity because a Boschian goes to minus infinity to plus infinity so this do I did you get the point how do I fix the dose I actually said one of the comparator value of fixed day whatever I want then the integrator gives output correspondingly and it keeps comparing with the set value as soon as the set value matches that comparator change the state and switch off the we creates a pulse with which of the answers okay so we automatically dose stops I mean implantation stops so this is what I on implanter is all about and what is the advantage of an implantation we said impurities can be put below surface is that clear gosh any other diffusion start from the surface there is no other way I can do that is that clear so the first thing I got an advantage that I can go below the surface anywhere I can fix the dose at any position some device I said I want lower concentration on a higher concentration in the lower side and lower concentration higher side that is also possible called hyper abrupt because I can put impurities higher dose below low I can change I can take any arbitrary profile you give me this is the profile I want I can adjust number of multiple implants and I will give any kind of profiles okay I can adjust any amount of days by adjusting the time okay till high some damage so I will anneal that damage and I also expect during that impurities will also get into substantial sites so essentially iron implanter has replaced solid state diffusion as a source of impurity but after impurities are running still follows diffusion law please remember DT product is not neglect DT is same so why we talk so much about solid state diffusion because we were looking for diffusion as a theory the source we use that time was constant source or limited source but we actually looked into how impurities get into silicon or diffuse into silicon implantation only does one job to put a fixed amount of impurities present in square at a place where you want that is the only difference between the two however energy smaller and lighter can also decide or heavier can decide how much is the depth you want okay that is another advantage I can decide just below so much all impurities at the surface slightly below much below these are the advantages with iron iron implanter allows at the solid state diffusion furnaces with the 4 star 4 tubes may cost around 30 40 lakhs rupees not even dollars and iron implanter may cost 30 million dollars so that is the kind of money which you have to understand so if I do in my lab I may not buy implanter I may still work with solid state diffusion okay so this so far we have discussed now almost all this there are two or three more processes which I have to do first before I start other thing but I thought that should wait and I should first show you how the ICs are made because that is the purpose of this course is to show how the IC factory is so we will first do this and wherever the word deposition comes I can do it I will say yeah I can do it okay and then we will look back we have to see physical vapor depositions we have to see chemical vapor depositions and we have to look into etching okay can you think etching and depositions are same after if I etch something this material will go somewhere if it deposits somewhere else so that is the deposition this is etching okay identical same thing I will use it okay let us start today and maybe tomorrow I hope I will be able to finish I want to see the process flow of an IC making particularly using CMOS okay so we start with IC processing and as I said there are two process mode deposition these are lab but we assume we know and continue with this okay so this is the crux of the course this is what we are trying to learn and to do this whatever processes we have to do we will do that okay so we have learned so far 341 maybe 2 are still remaining okay which is the major step in the case of indigo circuit lithography if you cannot print correctly all your game is over okay that is the major crux why I am keep telling in exam also lithography I will just talk about basic process given in Plummer's book and of course at the end maybe after the other two process time permitting I will also show you how finfits are actually fabbed okay why finfits because most mass transistors are now replaced by at least 3D if not 3D at least normal finfits with at least few fins okay choice of substrate first thing we have to do is substrate choice mass circuits are normally 100 oriented wafers bipolar normally use 111 wafers okay why 90% bipolar transistor action is vertical is that correct base emitter base collector okay mass transistor is always lateral so look for mobility is which direction it is maximum and therefore utilize it for your advantage also of course there are many other reasons of bipolar using 111s I do not have time otherwise I will show you bipolar process is even more interesting and little more difficult also okay but since it is not having enough market I think we will go for mass typical doping of the substrate is around 5 to 50s and ohm centimeter it is better to have a larger I mean wafers of higher this or rather it should be more intrinsic lower the sheet resistance it is better okay if the wafers are intrinsic sometimes it is better but intrinsic is too bad and too difficult and too costly so we do not buy why intrinsic wafers are costly what is that intrinsic means no impurities so you have to purify a wafer for almost all impurities out that takes money so MZ crystals are costlier than CZ crystals by 2 orders 100 times because it removes impurities okay I keep telling you money because you must realize why certain companies do only this much and certain do not they do not do it because their volumes are such that they cannot afford the wafers are normally P type boron rubbed lightly duped essentially and for typical 0.25 micron process which what plumber is working about the concentration is around 10 to power 15 per cc then there is a EDP count what is it it is essentially called electronic defect count which is expressed a number per centimeter square the wafers one expects are less one defect per centimeter square is what is expected no one will get this but if you have and pay for it what is it to do with the why we should have a low EDP counts the number of chips on a wafer will be proportional to how many defects are on a crystal okay so those many having a defect will not be working so the either paisa bachauge hudder loss may jauge so you have to think how much money the fourth is the wafer size and the thickness of the wafer please remember wafer sizes are decided by companies throughput requirements 8 inch wafers 10 inch way for 12 inch wafers and now people are looking for 16 inch wafers okay one wafer of 16 inch I have worked from half inch wafers 1 inch wafers 2 inch wafers 3 inch wafers 4 inch wafers and then I did not work okay so so best I have used is 4 inch wafers but nowadays they are talking of 16 inch wafers why they are looking for higher size more chips out of one same processing because gas a bit under dollar 29 okay so throughput may be better of course reality is also issue the first thing we are we actually go through is called active area for the transistors okay now let me actually before I come to this just minute was let me get rid of this implant sheets okay what we are really trying to do is the something following like this of course this is rudimentary I am not showing fully okay just a minute I will come back read I will just try to show you what is what about then I want some oxide right now I am not showing you how I will do this okay this is of course then there is a metallization also here is to be metallization here here here and here and of course from the gate this is a CMOS process which I want to create okay that is my job I want to have a N channel device I also want to have a P channel device and of course I may connect one of these two to make a common complementary part in that and I must one thing important thing is each area of N channel should be separated from the P channel because the substrate has to be opposite so this is done in the area which is called P well this is done in the AP devices are made in the area which is N well okay. I also want each transistor be separated from the other one okay so this is called isolation so they must get isolated from each other okay so the process which I need to know is isolation this region where actually transistor appears is called active area is that here wherever transistor occurs that area is called active area so the first mask is the one to create where transistors are going to come and where the other part get isolated from this areas is that here I want one transistor here one transistor here in between something should block so that they do not connect to each other okay. So this is what essentially first mask will do it will allow me to have individual transistor area isolated okay and that is called active area mask is that okay so standard CMOS will go through and as the it is only 2 metal process they are done if you have a 7 metal process another 5 mask so 16 plus 5 is only if same process with only additional metals if you add any other extra things it will keep increasing the number of mask so we will actually look into 16 mask process and then we will say active area that the regions where mask transistors will be created okay all transistor need to be separated and process which allows this is called isolation the first mask is used to have used to delineate active areas I am in a silicon wafer I want these are the transistor here please remember I am only showing you cross sections but then actually here will be the wafer will be something like this so one transistor will be here one will be here one will be here one will be here so it is the plan where number of transistors are will be actually there but where I will see it only on this side okay cross section okay so the first mask as I said is to delineate active area delineate me separate okay I said and the mask is therefore many times called active area mask is called active area mask or some people call it isolation mask isolation is provided by thick oxide the easiest way to separate transistors are through thick oxide why oxide current cannot pass through insulators that is the hope there are two process of creating this oxide of course process is same but one is called low cost process which is old one process slightly modified version is the second one the first process which creates which is called what is low cost means local oxidation of silicon which can create two kinds of figures very interesting figures if you see on the cross section one is called birds beak the other is called birds crest okay no it looks like that that is why the names but of course nowadays slightly different isolations have been tried in a lower down technologies and they actually created trench isolations shallow trench isolations which is called STI so almost every new technology will have STI isolations not just birds beak or birds crest okay not much different not much different but there is little extra processing has been done in the second one why why this was tried because this was not able to isolate because this oxide thickness could not be very thick so between two transistors there was still some leakage paths so I said okay I said not deep oxide dull that either says okay so we start with the first one I will do first this process is it okay two possible ways of doing it of course process is still local oxidation one earlier version we used to do what is called creating locus and which used to create birds crest or birds beak both have some problems the second one nowadays we actually only do STI's here is the flow we start looking in the way I start only one figure I you first draw whichever I am discussing then draw the second one the first is you start with silicon okay and grow thin oxide on that please draw the figure because this is the only way you can learn please draw the figure you have a silicon substrate which is thermally oxidized Si plus and dry oxidation Si plus O2 is Si O2 typically 400 Armstrong's of oxide was used for 0.25 micron process this thicknesses will be only true for 0.25 0.35 processes 90 nanometer down their values are different 45 it is even very different 12 22 it is even very different so do not use same values for all nodes okay but this anyway the process will remain say not much different in the next device also okay so the first oxide which I grow is called pad oxide okay something padding is to be done so I said okay I call it pad over which the second figure now you draw I have a silicon I have a silicon dioxide and I deposit silicon nitride I deposit silicon nitride what do I deposit silicon nitride the reaction is 3 Si H4 Si H4 is silo 3 Si H4 plus 4 NH3 using a low pressure CVD system at around 600 to 800 degree centigrade may be 800 I can deposit silicon nitride plus hydrogen will be released in the ambient okay typical nitride thickness for please remember these numbers are taken from Plummer's book and that technology is 0.25 to 0.35 so otherwise do not go by these numbers are valid numbers for 22 nanometers when I show you a pin 5 at least for 45 these number I tell you what are they are using now so the second step is first is pad oxide on the top I what is the process now it is not growth this was silicon converted to Si A2 now I am depositing silicon nitride by reaction this this process will show low pressure CVD chemical vapor depositions so this is what we are not done deposition etching we are not done but we will do that there after this depression is done I hope this with photo resist hope is by either by dispensing by tube or by dispenser called diffuser and then spinning the wafer on the chuck okay this is deposition plus spin and I get a thin layer of photo resist which can be different thinking if I want thicker what is the way I can change the thickness of resist spinning speed larger the spinning speed thinner will be the resist layer okay but too small speed also has a problem liquid does not dry out too thick resist actually the lower part does not dry okay then it curate bubbles there are catch words technology wise so the third process step is deposition spinning layer of resist okay typical resist requirement may be 0.6 micron to 1 micron 0.6 micron to 1 micron the fourth step is the first lithography fourth step is the first mask has been used if I am using a PPR this resist is positive for what does that property of a positive photo resist it is hard initially when exposed it becomes soft means a table developed it can be developed so since it is a PPR and I want to retain certain areas their light should not go so the mask has two windows for this two areas this is the mask two windows clear field with two dark windows the light will not pass through the dark areas and therefore below the resist will remain hard and wherever the light will go through the clear regions the resist is developed out is that okay PPR properties it becomes softened when it receives photons okay essentially what does it do you have a cross-linked resin which actually breaks and therefore it is achievable NPR is opposite it said uncross-linked sorry it is a straight chain it cross-links when receives energy okay so unachievable okay so the fourth step we now got after using lithography which I have not shown you how do I do this I put this mask on the top of this I discuss this shine light when the light will not pass through dark areas these areas will become soft that can be developed and only these areas resist will remain so this is called patterning first pattern has appeared this is your pattern this has appeared please remember this is cross-section the Y Z is inside okay this is only XY I shown Z is inside maybe you should show something like this okay so the after the fourth step we actually have delineation of so what is this area is about these areas of the area where transistor is going to come is that correct where transistor is going to come how do you decide the active area one is channel length what else source the width is not blood pressure source the lengths drain lengths some mass masking areas ages okay so at least this three the minimum feature size and the length channel length the sum of the three is essentially decides the active area is that clear active area channel length source and drain this total area is essentially where transistor is going to come okay so soon must come and channel must come gate must come so all three together should actually create a transistor so that is the area which you have delineated so below that only transistor will come and the rest places transistor should not be allowed to happen that means one transistor here one transistor here should get separated by process what we call just now as isolations the same figure is slightly better modified in this case I forgot after the PPI is developed even the nitride is etched out so here is the fifth one the nitride is also removed from this areas where the areas outside active region I have only silicon dioxide thin layer over which the active region has nitride as well as resist as well as resist how does nitride nitride is agent is hydromeric acid which is same as for SiO2 yeah I mean that is the problem you are right the reason why H rate is adjusted for that you know there is for SiO2 you need something called buffers in nitrides you do not need buffers buffer means do not put ammonium fluorides for oxide you need some ammonium fluoride okay thin downs so anyway etching selective etching so I have now after this I get oxide and nitride okay only resist has been stripped out what is why resist is called stripper because otherwise it was unachievable you said no and even now I am removing so I say it is stripped okay normally what are the stripper either organics like xylene acetones or some dark color resins like you know is asphalt you use in the road making carbon jelly that also removes resist but that is bad so we do not use it okay after this I start oxidation that is the local oxidation word came now thin oxide nitride active areas I started acting like a I start put this wafer into oxidation furnace typically it is normally this process of because this thicker oxide is required what is this process called why I am saying so any oxidation cycle normally is weight cycle is dry weight and dry initial oxide is good if you are dry oxidation no other species then you want to go you want do not want to wear some 20 hours so you want a lower time so you do weight oxidation the final sin there will be some steam left out so you do is dry oxidation which oxygen rest of the oxygen picks up the vapors so always weight oxidation cycles are dry weight dry though this oxide thickness is due to dry may be very very small may be few and strongs compared to the weight which you are growing as soon as I start doing oxidation you can see this figure oxygen does not pass through nitride nitride is a mass for oxygen okay so any oxidation cannot take place below nitride regions is that clear below nitride region oxidation cannot happen but oxygen will allow oxygen to get in oxide is allow oxygen to grow deal grow model the problem now is you are holding this wafer holding this surface of nitride at one level this level okay you are oxidizing but I already told you that 0.45 micron of silicon is consumed to create 1 micron of Si u2 so the volume roughly doubles is that correct so if you consume 0.45 micron down 0.55 must come up because you want 1 micron if totally you create so 0.45 will go below and 0.55 will go up why because nitride is holding the surface layer so half will go up half will go down so that is the shape is that clear so part of the oxide is above part of the oxide is below okay now if you look at and then you remove strip the nitrides now okay how do I remove as I say by using this method is it figure okay this is 5 maybe 6 7 maybe TKB is that okay how this oxidation has taken place why it is called local oxidation is taking place only in these regions but not in the active area so localized oxidation has been performed therefore low cost local oxidation of silicon as someone asked me if I would have aged oxide from here also it would not have mattered there is some advantage of retaining thin oxide but it need not be retained because anyway I am going to do local oxidation is that okay so it is not a compulsory that if nitride during aging oxide word is okay but normally I preferentially so that some thin oxide is maintained okay because the new growth will be then better than fresh silicon okay so I will do this preferential but if not nothing serious happens is that okay everyone wrong figures because these are the figures which will ask you in the exam as well so if I look at little magnified version of this I find these and then there is a thin oxide here okay I removed nitride from there and I see this area so if I actually see this small area at the edges I see this right side figure that is essentially called birds beak birds beak now why this birds beak has appeared no birds beak appeared because when the oxidation must be performed for this what is below there thin oxide some oxygen entered that laterally below that nitride area okay so it created a thin line inside the transistor area it gives what is the problem with that the active area is reduced okay your gate length will not reduce where source drain areas will reduce so what is the problem if source drain area goes down resistance increases so whole your speed goes down so it is not real I am worried about 6 gigahertz I may lose 6 to 4 by now okay so this birds beak so we said okay why did you put a pad I put a pad because nitride on silicon has a very bad combination it has a different thermal coefficient of expansion so it does not sit very well okay it is mechanical strength is different from oxide so I said okay between nitride and silicon I will put a material which matches both okay so I put a thin layer of SiO2 that is why I called pad it is like a buffer layer there so above is nitride below is silicon in between thin oxide of pad was used but that created because there was a oxygen here it entered from sideways as well and created some oxidation inside also okay so your active area becomes smaller relatively okay of course these are all exaggerated things not so bad or something but just for later so we say okay if that is all that you were looking for then I remove the pad okay so I started with nitride without oxide masked it everything and I have a wafer now which is active area on the nitride okay active area only on nitride sitting on silicon but since their thermal coefficient of expansion is different so when I start oxidation this ages starts lifting okay this ages because their thermal coefficient is different the stress here is excessively high okay what we say sticking coefficient goes down okay some other dig more chemical so the nitride layer from the ages actually lifts off okay if this lifts off and when I am now doing oxidation from here you can see it will go up and come down from that age will go up and then so this portion is like a crest of the bird it is called birds crest why are we so worried about these shapes because this if area is not uniform on a one plane the next process may be have a problem of accuracy all around okay is that okay and then I remove the nitride so what I got now is birds crest as isolating areas is that okay is that why crest I repeat the lift off nitride allows oxidation to proceed like this okay and because of that you see a huge top oxide and that is called birds crest so this is non-play much non-planar than this but what is the problem here it encourages the active area here it does not okay but it creates crest non-planar surfaces so we are still not done first implant as well what we have done so far is only to check get the active area where transistors are going to come even then it is not over okay the other possible oxidation isolation oxide isolation is done through what is called shallow trench isolation called STI the first 4 or 5 steps are identical as we did earlier however I have now oxide nitride resist and isolated this is same as what was done earlier for low cost before low cost starts oxidation start this is the step which is common for both now I please remember this is my initial surface where this oxide is is that okay this is my initial surface okay silicon I start etching silicon itself okay so if I start etching silicon silicon will be seen here will be seen here so silicon will start etching down now the question is normal etchings which we use which are the liquid etching system is isotropic in nature what does that mean so shape will not be vertical but some angle it will show 57 degree normal silicon V it will create a V here I do a process of etching which is exactly vertical which is called iron etching okay dry etching as the word wet etching is always isotropic and isotropic etchings are done by dry etching so we will do dry etching so this area of silicon has been removed this area of silicon has been removed and this area of silicon has been taken away okay so now I have oxide night like this and you have trenches side wise is that clear these islands are now sitting in a in the trenches just about trenches this depth of this trench can be a micron or even lower these days okay because the device is within 1000 Armstrong's now so I even a micron is more than sufficient so once how do I silicon silicon has no direct etching etchant either you do a dry etching is fine but if I do weight etching what is the etchant for silicon could be first you know HFHS silicon dioxide so I must convert silicon into silicon dioxide nitric acid does that 533 is the ratio 5 water 3 nitric acid 3 HF if I mix our 531 depends on company plus water dilution if you want 10% of this these calls silicon agents or call RH now this etching will do silicon oxidizer silicon and HF removes the oxides okay but in the case of this we will use only flora and fluorine ions to H through okay once this is done trenches are created is that okay figures this deep trench these are shallow but shown deep compared to this okay they are called shallow trenches because it is less than a micron wafer is 10000 these days millimeters and I am talking of few microns so it is very shallow why 1 millimeter thick wafer will be required larger the size larger will be thickness okay after etching of trenches in silicon one removes resist by stripping okay and you know than oxide nitride oxide nitride after this then I oxidize this wafer in a dry ambient dry oxidation is performed okay so thin oxide layer fills up the trenches thin oxide please remember this is thin you can also say but nothing will grow above silicon nitride because silicon nitride does not oxidizes okay so all the trench edges now are filled up with thin oxide okay typically this thickness may be 100 Armstrong to 200 Armstrong's okay or 10 nanometer to 20 nanometers if you want nanometers after this initial dry oxide is done then you may go for weight oxidation and that weight cycle will be how much again dry weight dry so the first rise this the next stage weight dry cycle ahead but that will be very thin I mean wait till it stay that time because this is a trench which is to be filled up now so I do not use any growth techniques now so what do I do I dump that is I deposit silicon dioxide from the top is that correct this is a oxide deposition process not growth process is that clear so I deposit silicon dioxide from the top such that and says a gaseous system so gases goes down everywhere nitride does not allow oxygen to go through so it fills up all the trenches with oxides it follows the contours however on the surface I did not show you properly you will see this because the thickness proportionality will be seen on the surface okay dump coming in a jahapar jahadaya here to get hold on each a chala you got it this is deeper this is not deeper so it will actually show some one portion higher than slightly lower the reason is this this thickness which is growing over this is not same as this deposition rate is same everywhere but since you are going through the time taken to grow this should be same as time taken to deposit this the deposition process is ionic it is not just LPCVD it is RF depositions it will just go down just go down even if it does but silicon nitride is very thick okay so no no no this way deposition laterally is very difficult because you are depositing now growth if it is a growth yes it will enter but there is no growth here you are just pushing gas down okay so it goes like this okay so this is a non-planar surface is seen okay and this non-planar surface I can make planarized by using a process which is called chemical mechanical CMP is polish what is the word I use chemical mechanical polish CMP which is the most important process state these days in almost all CMOS processes we keep planarizing the wafer every now and then what is chemical mechanical polish you take the wafer put it on a chemical on a chuck which has some mechanical slurry like aluminum oxide and you put some chemical on that and you keep rotating in 8 direction 8 forming an 8 so it HS partly and scrubs partly scrubs partly HS partly so you get chemical mechanical polish and as fine the powder you take final is the polish okay polish means what is polish word surface is planarized is very best polish means non-completely uniform plane so how much I will remove I will remove just above nitride kind just above I can keep watching when it comes you have drawn the figure the last sheet for the day please remember etching the CMP wafer is it okay is it figure okay so here is the final this STI has seen I removed now all that non-planar everything has been taken care by single plane because polish will come on single plane there is a nitride there is oxide this and there is an STI okay this is called recessing because it now fits to the surface it is called recess anything which is fitting with the surface so this is recessed oxide this is trench isolations so where are the active regions now these are the active transistor areas they are separated by trenched silicon dioxide the wave pval everything will be much below only in smaller areas so nothing this transistor can interact with this transistor any way okay is that okay so STI is a very standard process nowadays CMP is a major processing step nowadays and therefore this is very very crucial in making any chips these days is it okay so so far out of 16 mask how many masks we have done one so imagine there are 15 more masks to go through if the minimum C must be shown okay so something I will hurry up something I will show I thought that this is first time I am showing you so how we actually keep doing things so all process steps I shown you everything in between there also connects I do this and I will get this because you know you have seen once okay is that okay so far I have shown you every process step in between how it will happen but next time I say here to here I can come you look