 You can follow along with this presentation using printed slides from the Nanohub. Visit www.nanohub.org and download the PDF file containing the slides for this presentation. Print them out and turn each page when you hear the following sound. Enjoy the show. So today we'll be talking about reliability of MOSFET. This is a very important topic. Most of the time for last 10 lectures or so we have been talking about how MOSFET operates, MOS capacitor, MOS transistor, how it operates. But most of the time what people do not realize that what makes a technology successful is not how much it performs or how well it performs initially in the laboratories, but how well it survives the harsh condition that it is supposed to operate over a prolonged period of time many times for 10 years or so and that is exactly what reliability of MOSFET, the physics of it, tries to describe. So I'll begin by explaining why this is a very important problem. Turns out one of the most important problem in MOS technology these days is the reliability of MOSFET. I'll explain why. We'll talk about three types of reliability problems. The most important one is something called a negative bias temperature instability. Now this you should distinguish between this bias temperature instability I talked about about the sodium ion moving from within the oxide left and right right coming from human hand and how it was solved in 1960s to solve and make MOSFET possible practical MOSFET possible. This is different. This is also discovered in 1960s. It went away for about 20-30 years and then in mid 90s it resurfaced and nowadays this is one of the most important reliability challenge for MOSFET. I'll talk about gate dielectric breakdown as I said that this is a very thin oxide that you have about a nanometer or so and when you apply a voltage one volt one volt may sound small but if you divide one volt divided by one nanometer you have more than 10 megavolts per centimeter electric field in there. If you look up in any book then it will say at 10 megavolts this should break down completely be destroyed yet our transistors do not get destroyed and I'll try to explain very briefly why and how to think about that problem. Finally this issue about radiation induced damage particles coming from sun and outer space and as well as various environments. This is very important every time you take a transatlantic trip there is a probability almost close to one that one of the bits of your flash memory drive will be flipped so that if you take a power point or any file your music file it's almost certain that a few of the bits might be flipped and the reason it doesn't hurt you because there are error correction code that you have to put in every storage device now because of this bit flip problem. I'll explain a little bit about that also before concluding. Now the point I wanted to make and the beginning that in this course you are learning about how MOSFET operates right and how to analyze them how to think about surface potential inversion charge and all but of course what you realize that these MOSFETs are under incredibly harsh conditions they operate then by the time you throw away a computer every almost all the transistors will have switched close to about 10 to the power 14 number of times 10 to the power 14 is a very large number it's like a hundred I mean a 10 to the power 14 is about a hundred million oh no not million this okay you get the 10 to the power 14 it's a large number I forget how to how else to put it and when you turn things on and off this this large number of times this is a hard this reader that you all of you have in your computers and you can see on the periphery there are all input output MOSFET transistors that essentially connect the output world the outside wall to the inside world of the microprocessor and you can see that they have been lighting up this is a liquid crystal image of the temperature and they have been lighting up that means those places are incredibly hot and what will happen that when you use it under such a harsh harsh conditions over a prolonged period of time that the transistors are going to degrade and essentially eventually they are going to fail and this is certainly not acceptable because once you have this type of degradation let's say the gate oxide you remember the MOSFET right the MOSFET the two vertical columns are source rain contact the fingers the black fingers coming from the two side are source and rain the very thin region in between the thin region is oxide and the black little black strip on the top is a gate contact so get metal so you can see that those regions are heavily dope show us as black in this picture and if you blow up that region many times you will see the gate oxide has been completely blown because of the harsh operating conditions that the transistor find itself in incredibly this the physics of this is actually one can calculate when things will break after a certain period of operation that's that's the essence of the reliability physics so in order to understand that that this reliability issues of reliability you will have to distinguish between two types of bonds one bond that I have already discussed is the passivation of silicon with hydrogen at the silicon silicon dioxide interface this is that forming as a nail remember that silicon dioxide cannot really satisfy all the bonds of silicon therefore you need to tie up the extra bonds with hydrogen hydrogen has one proton this has one dangling bond so they make a covalent bond and everybody is happy now the thing is that something that comes easy also goes easy that means that something that you did in the last process step with a low temperature this hydrogen stitching up the substrate with hydrogen bonds as soon as you begin to operate the transistor some of the hydrogen begins to leave with the electric field and I'll explain why and as they begin to leave what will happen that gradually the number of interface states will continue to rise again remember you bought it down to 10 to 10 and gradually after the hydrogen begins to leave you will gradually the number will gradually go up and if it reaches on the order of 10 to 11 per centimeter square then you are already in trouble because then your threshold voltages shifted a lot and the current is no longer stable the other type of bond that could be broken is silicon oxygen bonds and that's the second bond that can also be broken now you'll notice that in this list I didn't put silicon bond breaking why not and there are lots of silicon silicon bond in the crystalline phase why not because crystalline materials are deposited at very high temperature they are their bonding is very strong and they are in the crystalline phase everybody is wiped they are supposed to be and therefore it is not easy to break it at room temperature as you are just operating the transistor right in your computer you're just using it in room temperature of course then therefore it's not easy to break but the amorphous material those are randomly structured bonds easy to break similarly the silicon hydrogen bonds easy to break and that allows you to categorize the reliability problems into two broad groups one is this broken silicon hydrogen bonds and the negative bias temperature instability and hot carrier degradation these are related to silicon hydrogen bond dissociation primarily this surface association and if you have silicon oxygen broken bonds right in that case those the various mechanisms that are associated with it will call be will be called gate dielectric breakdown electrostatic discharge radiation induced gate rapture this type of thing these are various things I'll give you some example by the way do you see this hot carrier degradation it should be hcd but what it is called hci so this is hot carrier induced degradation and so this the shortened version doesn't really go with the full name and similarly the gate dielectric breakdown there is this extra t this is because it is time dependent dielectric breakdown tddb and that's something these things we'll talk about so I'll give you some examples I'll talk about negative bias temperature instability I'll tell you a little bit about the physics I'll tell a little bit about the gate dielectric breakdown and radiation induced radiation induced damage okay so that will give you some flavor of how to think about this problem now silicon hydrogen bond as I said that those are at the interface between silicon silicon dioxide and silicon oxygen within the bulk of the silicon dioxide right now you remember how thin these things are right six atoms five or six atoms vertically and over a wafer 12 inch wafer so I sometimes I may have given this example before that if you covered the whole united states with six inch of snow not a inch variation across the whole whole country this is the level of control that you have to have in order to make a integrated circuit possible simply amazing that anyone can do this and now we do it routinely take it in your pocket on this flash drives and anything and think nothing about it and that's the that's the amazing part of this technology let's talk about negative bias temperature instability first of all the word negative so it happens with negative voltage and it only happens in p MOSFET doesn't happen primarily in n MOSFET what I have shown here on the left is an inverter now I'm sure you many of you may know what inverter is but for the time being we'll assume that this is a p MOS on the top series connected to an n MOS on the bottom and together the input are connected together and connected to an input now if we apply a zero volt on the input that one then what happens then the n MOS is off do you realize that because it has a positive threshold voltage and at zero voltage the n MOS is off p MOS is on right p MOS is on and as a result the VDD the supply voltage VDD gets connected to the output VDD output and becomes equal to VDD and that's why it's called an inverter because your input is zero and output is one now if you think about what is happening to the p MOS then you will realize that the source drain is connected to VDD and the gate is connected to zero volt now if I assume that source drain is connected to zero volt then equivalently I would say that the gate is connected to a negative voltage and that way you understand why the transistor is on because the threshold is higher than the threshold voltage of the transistor the negative voltage this negative voltage what it does that it creates defects at the silicon silicon dioxide interface between this yellow and the white region and what it does then that if you look at the IV characteristics have you seen this IV characteristics before this is that square law or vg minus vt to the power alpha which could be between one and two now the black line is before stress meaning when you just bought your computer from your store just bought it home I haven't just yet opened it that's your black line but what's going to happen and something that you don't expect is that as you are writing your microsoft word various documents and other things gradually the current the output current will gradually come down and that it will become the rate curve over a period of time it will become the rate curve and this change in the current this change in the current in the drain current you can take it and divide it by the original current so that's the percentage change right that's the percentage change and if you look at the percentage degradation as a function of time so you see 10 to the power one second what is this red line 10 to the power nine seconds around 10 to the power nine seconds so it is close to this 10 year lifetime so what is 10 year 10 year is 3 times 10 to the 8 I think that's the 10 year so many seconds in 10 years so many times your computer will be guaranteed for 10 years so that's your warranty period or guarantee whatever is a manufacturing period and what you will see that the black curve will gradually come down and the rate curve if you look at the difference and plot it then the degradation will continue to increase do you understand that as this curve is going up there on the right hand side that means the rate curve is going down right that's the difference divided by the original value that's the degradation value and what is amazing that this curve is generally you say if you plot it on a log log curve it becomes in straight line I'll explain the physics in a second but you realize that if it changes crosses 10% exceeds 10% degradation so the rate curve is weighed down by 10% then the circuit you have designed assuming a given level of current that circuit will not work anymore because if the transistor the preceding stage of the transistor will not be able to supply the current in succeeding stages and so your computer is going to die you can always blame it on the software and you can say that that's why your computer died but many times the hardware may be the problem as well now so therefore we have to understand how it degrades and how much how long will it take because you realize that if it degrades that much and all on a sudden a million transistor is returned then the company we whoever made it they will go out of business so this physics explaining or estimating how many will come back as a field return of a computer is of primary importance when you design a MOSFET now as I said that if you look at the delta vt the threshold voltage shift as a function of stress time at various temperature 25 90 and 150 degrees C then you will see that on a log log plot this has a straight line I'll explain that in a second but let me first explain what am I talking about about delta vt delta vt is essentially when the black curve became equal to the came down to the rate curve because it's because this threshold voltage changed now why did this threshold voltage change because the as the hydrogen bonds hydrogens are diffusing away you have more defects those more defects are catching more electrons that causes a qf do you remember the qf and qit so it increases the qf and a qit as a result then your threshold voltage changes and therefore your current changes so this tells you how much current change there would be now for nbti this threshold voltage shift is given by this very strange formula you have a constant and exponential dependence on temperature you see if the temperature is high then the degradation is more right so you can see that that's that's exactly what is seen that 150 degree the triangle down that is significantly higher than the 25 degree c that's the triangle down now why why does 150 degree c I mean nobody even in the equator no there's no 150 degree c region where you want to live at least why do I care about this higher temperature the reason is yes not in the equator the environment may still be room temperature but inside the computer remember as they are turning on and off each transistor are getting incredibly hot this is dumping energy so the transistors around them can have significantly higher temperature than the your room temperature right that's why your computer often gets hot your room may be cold but your computer gets hot so this high temperature is important the amazing thing is that many times this is described by a power law power law means that something raised to the power a constant time raised to the power in constant power law to the power 0.25 and amazingly this should happen over many orders of magnitude it could happen over six seven orders of magnitude in time and this quarter power law will keep going keep going keep going and so this is a very robust and unusual type of degradation and if you apply larger electric field larger electric field than your a will depend on the electric field so and so therefore as you're turning transistors on and off your degradation will change with respect to the voltage waveform now before I explain why this is a t to the power 0.25 remember this is a hydrogen we are talking about silicon hydrogen bond distillation we are not talking about electron walls anymore we're just talking about how bonds break so we want to explain to you how this happens but before I do that let me explain something about diffusion distance right now this is something you have seen many times and so I really don't have to explain too much if you put a drop of ink in water then you know that the ink spreads and it spreads why because that that the molecules in that drop of ink they have a random motion they go in various directions and that's what I'm trying to show the red point is a sort of a drop of ink sort of with many molecules and they could have a random work the white jumble regions are sort of the molecules moving around now how far do they spread over a given period of time you know this right square root of dt that is how far it goes and how does it work that if you put a drop of ink like this then in the beginning it's very tightly focused over a very small per region and then really you are look at it a little bit later it has spread a little bit more and you look at it a little bit later it's spread a little bit more and I don't want to go through the map but essentially how far it will go on the average this will be square root of dt and that's the x is proportional to it will be maybe a factor of 2 or 4 up front but I'm not including that yet by the way so this is the diffusion coefficient of the molecule whatever molecules it's not electron holes this is whatever molecule we are talking about if it's the ink then it's the ink molecule diffusion coefficient of that okay had it been ballistic then of course with time it would have gone linearly proportional to time there wouldn't have any square root because of this back and forth scattering back and forth it cannot go as fast as it would have gone as far as it could have gone in ballistic case well remember this right how far it goes square root of dt and that is something I'm going to use in a second okay now on the very left I have rotated the MOSFET 90 degrees so first of all you see the silicon substrate on the left on the top curve the yellow region the is the silicon dioxide and the right side I have written as poly it could be a metal also the semi the metal gate now what happens if you see that I have also drawn some silicon hydrogen there are of course lots of silicon oxygen bonds there as well right do you remember only a fraction of them are silicon hydrogen but I have just drawn the silicon hydrogen one without the silicon oxygen one and in so with some arrows I have shown that some of the hydrogen is going away going away why because I have applied the electric field this is a strong electric field and therefore the bonds are being broken and then they are diffusing away now if I wanted to know how many hydrogens are diffusing away remember then I will get the number of interface traps then I will get the qf and then I will get the delta vt so that's my that's my game plan so how many bonds are being broken I can write a simple equation you see no rocket science here very simple dn it dt is the number n it is the number of bonds that got broken dn it dt means number of bonds that go broken per unit time is equal to kf kf is some dissociation rate at which the bonds are being broken you know depends on the electric field n naught is the number of original silicon hydrogen bonds I had minus n it so n naught minus n it is number of unbroken that I all I still have okay now of course once they are broken the hydrogen is broken they'll be walking around some of them will be going away diffusing to the right and going away some of them will come back and see that there's a dangling silicon bond sitting there and it will come back and anneal anneal that bond because it it is around and there is a dangling bond so it can just do just the original forming as anneal it can again do the passivation so that's the second term on the right of the first equation now you will notice that I have written nh0 h4 hydrogen nh means number of hydrogen 0 means number of hydrogen which are at the interface because the hydrogen which have diffused far out they cannot tie up the any silicon bonds right so it has to be whatever number which are on the surface now most of the time the rate is very very small very small and the number of broken bonds is also relatively small so compared to so you can set dn it dt equal to approximately equal to 0 right approximately equal to 0 and you can neglect in the first term the red n it n0 minus n it you know n it is small let's say n0 is on the order of do you remember what this number was five times 10 to the 12 I showed you in the last class or the class before and n it I'm thinking about on the order of 10 to the power 11 so it's less than one tenth of it so I can drop that one also that gives you the second equation do you see that kf n0 I have taken the kr on the other side divided it throughout and so I have a relation between nh and n it two unknowns these two are unknowns one equation okay now this is the crucial second step and see whether you can understand it there's a number of bonds that got broken right hydrogen got got away now those will be diffusing and you can see in the second plot I have shown how far they will diffuse at a given time square root of dt that's how far the hydrogen has gone right okay now the area under the curve is equal to the number of hydrogen which has been released now where will the hydrogen come from this is coming from because bond goes broken so the number of bonds that got broken is equal to the number of hydrogen that was that was released in this oxide region so therefore I can write this following statement and see whether you agree with this statement n it number of bonds that got broken is equal to the area under this curve half you see triangle right up nh0 is the ordinate and for a time t it got dht is how far it diffused the hydrogen how far it diffused so now I have two equations and two unknowns right and that immediately tells me the number of broken bonds number of broken bonds goes as so that was the experiment by the way and look at that t to the power one fourth so it is telling you by very simple this three line of algebra no rocket science here and it's immediately telling you the number of bonds that will be broken as a function of time meaning that let's say this is one month after you bought your computer five month after you bought your computer so you put five month entire time t and that will tell you how many bonds got broken you put it in your threshold voltage calculation that tells you how much the threshold voltage got shifted and how much current reduction happened as a result right so therefore this is what people do before they sell a computer they do the extensive characterization and they find out that how it's going to degrade that is only how they will make a computer and sell it realizing that you'll not bring it back before a certain period of time now remember this nbti if you go to interviews and other places people will be very impressed if you know how this works and what happens the physics of it because this is one of the most important reliability problem today let me give you some example this was about silicon hydrogen bond dissociation right many times you will have also hot carrier degradation that when you turn the transistor on very large drain bias the electrons in the drain is very hot right because the long voltage drop and when this hot energetic electron comes in and strikes next to the silicon silicon dioxide surface then silicon hydrogen bonds are also released so that's hot carrier degradation another very important reliability problem but it was more important in 1980s no longer as important now i'll maybe i'll explain that statement a little later now let's give me let me give you some examples about silicon oxygen bonds and in silicon oxygen bonds and the dissociation of it so starting with gate dielectric breakdown now in the gate dielectric breakdown the first thing before even gate dielectric breakdown happens first thing you realize that many times the oxygen in silicon dioxide is not where it should be and so there is a trap level because the oxygen is not where it needs to be as a result when the electron goes in a state of going straight from source to drain many times it get trapped in that trap level within the oxide within the silicon dioxide and as a result if you have a series of electrons trapped then you see in the bottom figure i have that red arrow that red arrow is the amount of trap charge right and if you have over a period of time huge amount of trap charges as the electrons are going by a small fraction of them are getting trapped into this into this vacant sites and as a result over a time what will happen the arrow magnitude of the arrow will keep rising and that will change your threshold voltage you can see the q ox at x1 t that's a time dependent thing so that will change your threshold and as a result you'll see v car will shift and vt will shift and there'll be a problem in terms of drive current so that's one problem for thick oxide it's a big problem but these days they make such good quality oxide such fantastic oxide that number of these defects are miniscule these days very small however there is another trouble the trouble is that although you initially didn't have any defects but remember you're applying an incredible voltage on this i often say that if you if you apply one volt one one nanometer that's like a 10 000 times larger than the high voltage power line that you see going next to your homes right if you see a bar sits between those two they'll be fried instantly be fried uh if you have in sits in a power line and this is about 10 000 times more voltage than that feel electric field and that so you can realize that over a period of time there'll be bonds will be broken as a result and defects will be formed and you also realize so that when a certain amount of defects have been formed this defects which are shown here in little red boxes those defects might line up and if they line up then you have a short circuit between the substrate and the gate and the electrons will simply flow out your MOSFET is gone no longer a MOSFET maybe a very bad bipolar transistor you have now with the base gone back and so what you see in the gate current as a function of time again this is a used time as a function of time the gate current remains essentially unchanged for a long period of time and then all of a sudden this red defects line up all of a sudden the current shoots up and there'll be such a violent uh the current shall rise that most of the time the transistor will be transistor will be destroyed what you see here on the right hand plot is a time to break down that's what's TBD it's not to be determined it's time to break down uh and you can see that this is 10 to the power so that means you can see the 10 to the power 8 10 to the power 8 is that 10 year thing right on the on the right hand side and so most of the time experiments will be made uh first let's look at the red one vg1 and on the y-axis I have f capital f f means fraction of the oxides broken so that means let's say you start with 100 transistors the first one breaks so the f is 1 divided by 100 is a cumulative failure then the fifth transistor breaks breaks so f is 5 divided by 100 so you y-axis is you note when the transistor failed broke and in the x-axis evaluate that function it's called a waible distribution so evaluate the function and put a dot there and you keep putting the dots and that creates this straight line uh in the in the red straight line similarly if you reduce the voltage a little bit then you can get the uh the blue one why because you know if you reduce the voltage defects will form a little slower and so it will take a little bit more time for the bonds to break and as a result just shift it to the right a little bit to the right by the way where I draw a dotted line that's sort of where 50 point is 50 percent which is a very strong log log plot so therefore 50 percent will happen about there and also the green is a little lower voltage so you do it at three voltages and you see they're all parallel so if you want to say what will happen at one volt then you will simply extrapolate to one volt and see what will happen now this point where the bottom horizontal line is that is one part in 10 to the power minus 12 that is this point and that is because every company wants that if you sell 10 000 i c 10 000 i c no more than one should come back now every computer may have a billion transistor 10 000 right so this number would be on the order of 10 to the power 12 10 to the power 13 so you have to make sure that this time where this magenta car intercepts the black car that is actually more than 10 years because you don't want anyone bringing their computer back to you for for money back more than one in 10 000 persons and so therefore you cannot accept more than one in this number of transistors to fail because as soon as one transistor fails the whole ic is gone and so that person is going to bring back your computer bring back his computer and so we want to explain very quickly what the physics is of this in order to understand that there are these two things that i'll quickly explain just qualitatively one is that when you apply a a positive bias this is a MOSFET bottom part is the substrate what i call cathode oxide you see and the metal is metal or the polysilicon is on the top side you apply a positive bias so these days what will happen that electron will tunnel from the substrate to the substrate to the gate and they will come back they will have impact ionization there and the holes will come back in the reverse side and in the process of coming back they will break bonds and this is called an anode hole injection because holes are coming back from the anode region now you don't really have to follow the details because i'll not be asking you to make calculations just get the flavor of how people do such a things now so therefore the time to break down simply depends on the number of electrons that are flowing through right because that determines on how many how much impact ionization will occur so that's j sub e on the denominator and alpha or a is the number of electron hole pairs that was generated per injected electron and then tp is the number of transmission probability of the holes for coming back so if you have more holes then your lifetime will be shorter right that's all i'm saying and nbd is the number of defects that you need to create a short circuit between source and rate so that's that's the calculation one does and this is the trap generation efficiency k is a trap generation efficiency because every hole cannot generate a trap so a fraction of them will be able to be able to do that so from this type of calculation people say that they determine that when will the oxide will break but remember this one is saying every oxide should break at one time if you're given gate voltage means j is a constant a is a constant tp is a constant but that simply saying that at everybody should break at the same time point now let me just quickly explain one thing that what type of dependency do i expect so if the voltage is very high then do you see that the top expression for j e is given by this following order type expression this is a tunneling current expression a exponential of b divided by e alpha the number of electron hole pair that is generated per injected electron is one the order of one and two tp as high energy is more or less a constant so time to break down goes as inversely with the electric field multiply this three out and take the inverse and then you'll get that it will go inversely with the electric field that's one if the voltage is low then the tunneling current has some complicated dependence on the electric field the impact ionization has this particular expression for the voltage dependence and therefore at low voltages that lie time to break down goes with v the voltage v the point is that there is certain electric field dependence of what the lifetime is going to be now let me explain this final one about about tdb because not all of them not all of them are breaking at the same time so if you look at the rate points that they fail at different times so how would you determine how they are failing in order to do that let's start on the top side now probability this is just high school algebra or so let's start on the top side probability of a field column with the rate rate field column what is the probability if the probability of individual one is p and if there is m rows then p to the power m or sorry q to the power m there is the probability of a field column because all of them have to be present now the probability that you will have exactly one breakdown spot on the left hand side is called p1 now you see this vertical column the rate column could occur anywhere anywhere within all the columns right so you take nc1 that is you say choose n one out of n vertical columns then p to the power one because one cell is taken and one minus p n minus one what is that that the remaining n minus one are not shorted right that's simple binomial binomial expression so if you simply take that one for p1 and do it for large n you can show that the p1 and just try to do it at home and see whether you can express the p1 as equal to chi exponential of minus chi with the chi given by the following value just when you have large n and then the f1 f1 is essentially the probability that cumulative probability it will be equal to one minus p0 p0 is the probability of not failing if p0 is the probability of not failing then f1 says the probability of failure at a given time so you insert that value for p1 in the other one and that will give you this y as a function of logarithm of time with a certain slope so people can in fact the point i'm trying to make you can do this at home but the point i'm trying to make here is that both the voltage dependence as well as the slope of this curve people can easily calculate on the back of the envelope calculation for to predict oxide reliability okay now very quickly about radiation damage now what happens for radiation and this is for a flash memory charges are stored in the floating gate called fg now what happens is that when a radiation strikes one of the charge very large charge particle comes in and strikes is so energetic that it creates a shower of electrons and holes it creates a showers of electrons and holes there i'm not sure whether you can see it from here there's a track starting from the top and as it is losing energy the energy is huge it's 58 mega electron volts and as it is coming down it is generating a shower of electrons and this shower of electrons causes huge amount of damage in the oxide and that again traps charges causes racial voltage shift and again causes reliability problem so every time you take a flight across the country this is happening to your computer that particles are striking and because planes move about 40 000 feet right 40 000 feet 36 000 feet is sort of 90 percent of the atmosphere is within at 40 000 feet so you are sort of exposed to a huge amount of radiation that's coming in not huge in terms of meaning that that will going to kill you but huge in terms of it will give going to flip some cells on your flash drive and depends on the energy see that was 8 mev electron and this is a chlorine atom and coming in and you can see a huge huge number of electron hole pierce like explosion has happened projectile has come in and explosion has happened creating a large amount of defects in your in your oxide whichever gets stuck that's dead again what happens very simply and this is a very old problem many satellites have essentially been disabled because of this problem is because when a strike happens you have a huge amount of electron holes generated the picture on the left right is shown taken from your book and what happens with the electrons very quickly because of the electric field very quickly it leaves but the holes are not as fast they are more heavy and so it takes a certain time and then they can get generally get trapped and become like an interface trap so this is that qf the fixed charge of the interface and what will happen that although the charge has radiation has gone away but the blue arrow is indicating that the trapped holes will be there for a long period of time just sitting there shifting your threshold voltage and as a result that transistor you cannot turn on anymore if you cannot turn on that transistor then for a given given supply voltage then of course your computer is not working so uh so this effect in thin oxides of course this is this picture is taken from 1970s but these days oxides are not this thick so many times the trap charge effect is not as important but what happens because the radiation that deposits so much energy that many times it ruptures the gate gate is very thin you have five six atoms particle comes in it breaks all the bonds and shorts the substrate to the gate so that's a slightly different problem that is the radiation induced gate rapture looks like all biblical references here and uh so let me summarize so that the reliability is really a serious concern for mosfet and in fact this is one of the first things that you realize going from uh university to a company that you really didn't really understand the details of the reliability physics unless you begin to understand the reliability problems now there are many different degradation modes and people have developed very nice models of predicting those individual phenomena and they can in fact predict very well that how your transistor is going to degrade and how it's going to behave in 10 years time now at present as I said nbti is the most important one negative biostempre instability this is not the sodium problem right this is associated with silicon hydrogen bond dissociation and this is the most difficult reliability problem and followed by fii heart carry and reduced degradation which I did not discuss get the electric breakdown which I showed you how this is done and radiation effects radiation effects because most of the interplanetary missions are nearly impossible satellites and other things you really have to work very hard to make those work under those harsh conditions so then that's it okay thank you