We are now ready to discuss yet another short channel effects, this one goes by the name of Drain-Induced Barrier Lowering or DIBL for short. Let us begin with the long channel situation, we've seen again and again. We have a device like this where the length is much Longer than the extent of the division regions around Source and Drain. Here is the conduction band edge, versus horizontal distance. In this picture, you can see the following. Let me first assume that I apply a gate vaulted sequel to flat. Then there is no depletion region, no inverse in charts. And the p region in the body is neutral throughout. And between the n and the p we have a junction. Which has a built in potential, which we have called. That corresponds to, a change, in, Ec, of. So, as the potential goes down from the N to the P by Phi B I, it's [INAUDIBLE] goes up by Q Phi B I. And I'm assuming a very small drain transvoltence, so practically drain by the end source, but the voltages are the same. So this is a symmetrical situation. Q phi b i is the barrier that the electrons have to cross as they go from the source towards the channel like this. And similar, if you have electrons in the drain that would have to cross a barrier on the right. But let's concentrate on the barrier on the left. If I now make the gate potential higher I am going to deplete the channel and eventually invert it. The surface potential will become larger and that is what makes possible for the inversion layer to appear. Now the surface potential becomes more positive, that is equivalent to saying that the conduction band edge is lower. So you go from curve one to curve two. Now the barrier is smaller, and some electrons in the source can cross into the channel, and this gives rise to inversion. Finally, let me take the drain potential and increase it. So if VDB becomes high, the corresponding near the channel could become low. So, this is lowered by an amount, qV(DS). So now we have the edge of the conduction band going down like this which would correspond to a potential going up. Now if you have an electron that is at the source. And because of random also finds its way a little to the right of this peak, one set is in the channel, it will go horizontally if there were no collisions. But because there are collisions it will lose its kinetic energy, go down. And the bottom of the conduction band represents the potential energy of the electron. So let's say it goes down here, then it starts accelerating because of the field. Eventually it finds, it encounters another collision. It loses, its kinetic energy goes down. So, it keeps traveling towards the right. And we have discussed how to describe this effect using the idea of drift current, which is a phenomenon that is explained by assume you have a large number of collisions in the path of the electrode. That's how we arrived at the notion of drift velocity and so on. Now let me repeat this discussion for a short channel device. For the short channel device things look like this. We have discussed again and again what happens because of this picture. Let's see what the energy E sub c looks like in this case. First of all, because as we said before the inverse on layer is there not only because of the field of the gate. But also because of the fields due to the source and the drain, this makes the surface potential larger for the same value of gate you had for the long channel case. The larger surface potential corresponds to a lower E sub c. So instead of E sub c being up to here that you had for the long channel device, it is now something less. So instead of this curve over here, for the same gate voltage you have a lower E sub c, that means that the barrier that face before they can enter from source. The channel is smaller so it is easier for the electrons to get into the channel which means that it is easier to invert the channel. If you now increase the gate voltage and you make it equal to the value you had when you obtained curve two up here, you get a new curve to over here. And the position of this would be smaller than the position of the corresponding long channel curve. So the barrier is even smaller here than it was for the long channel case. Because it is smaller, you can expect a larger electron density in the channel. In other words, again, this is a manifestation of the fact that in the short channel case, it is easier to invert the channel. Finally, let us increase V D b, the drain body voltage. So, that means that E sub c is lower on the drain by qV DS, and we have this curve. Now, because the field lines from the drain penetrate all the way from the source In other words, they terminate on the channel, but all the way to the source. Because the source, after all, is very, very close to the drain. So what happens at the source end of the channel is affected by the presence of the drain. The lowering of these curves, happens throughout here. So instead of the peak of the dark curve, being where it would have been for the long channel case, it is somewhat smaller. So, the barrier you see here, which is the distance between the peak of the solid line and this line over here, this barrier becomes smaller there because of the presence of the drain. So this barrier has been lowered because of the drain button, so this is called Drain-Induced Barrier Lowering or DIBL for short. Because now the barrier is lower, more electrons will find it easier to get into the channel. So that means for the same gate, voltage that you have for the long channel case, you're going to get more current. It's the same as saying that the threshold of the device became smaller. And in other fact that I should not forget to mention is that the DIBL effect increases even in saturation. If you keep increasing VDS, you keep increasing this effect. The barrier will be lowered more and more. It doesn't stop a prime, in other words. As I mentioned, DIBL is often described as a change in the effective threshold, so the effective threshold is V T + Delta V T because of DIBL. And delta V T DIBL is a negative number, so the effective threshold gets lower. In the book, you can find expressions for delta V T DIBL. And just as the threshold gets lowered, so is VM, which is the bottom of the moderate inversion region. Everything is moved towards smaller VGS values, whether it is the bottom of the strong inversion, the bottom of moderate inversion, the threshold, they all move towards smaller gate voltage values. So let's plot now the log of ID versus VDS for a given value of VDS equal to VDS1. This corresponds to the solid lines that you see here. I'm assuming for now a highly substring. And you'll see in a moment why. So if I decrease the length, I go from this curve to this curve. So it's as if the curve moved to the left. Defectively you see that decreasing the threshold I remind you that because this is a log of i versus VTS, you get a straight line and weak inversion. That's why the curve should look like that. Then you enter moderate inversion, then it gets strong inversion. So the more you decrease L, the more the curve move to the left. So for example, from this curve, now, you get eventually, to this curve over here. Now that we already knew from Chaff's Herring. But the new thing here is that if VDS becomes larger, then you get a further shift. Because of DIBL, the drain induced barrier lowering that I covered on the previous slide. So instead of, for example, having the solid line over here, you get another line which is broken. So here, which is slightly moved to the left even further because of DIBL. That is what makes the curve move. It's the DIBL effect. Here is the drain soscarden verses the VDS. This curves happen to be developed for weakened version. You can see that as you increase VDS table works on your effective threshold and reduces it more and more. And the current keeps going up. So, the main reason the current trend goes up in weak inversion is the DIBL effect. To some extent, you'll find this in other regions too, but here it is very clearly seen. Now something else that is related, called punchthrough. If I have a low substrate and everything else is the same as before. Now the depletion depths are larger, and the penetration of the drain field all the way in the tunnel, all the way towards the end of the tunnel near the source becomes stronger. Then the effect become stronger as well. So as you decrease the length, things quickly fall out of hand. Not only do the curves shift, but the slope deteriorates and eventually it's very hard to turn the device off. In fact it doesn't turn off. And when you go from one VDS value to another one, these curves really move. And become the broken line curve. So, of course, this device is useless for digital operation because you cannot turn it off. So, that happens if you have low substrate where the depletion region depth is significant. Of course you would not go to such short lengths if you had such a device, because if you did, then I DS versus V DS would be like this, and this would be a practically useless device. So, that would determine how small your L can be. Now, very roughly, let's assume that we have a gate voltage such that the substrate by itself would be neutral. You have a depletion region around the source, and a depletion region around the drain. When the combination of drain and source voltage is such that this happens, this is an indication, very approximate, that you're about to get large currents if you try to turn on the device and you will not be able to turn it off. In other words, you may be in this situation over here. And the current that you get, is no longer a or of the type you want to have. It's a current. A very strong current. This is called a surface punchthrough. It's a current that flows along the surface, and you cannot remove it by decreasing the gate voltage. Sometimes, because of iron implanted regions which make the substrate docking non-uniform, for example. You may have a region here which is highly bolted because you want to control the value of the first then the depletion region extent here might be small, but then it lowered into the substrate might be large again. In such cases, you may get the linkage current that follows a subsurface path like this. So this is called bulk punchthrough. Of course, needless to say, punchthrough is supposed to be totally avoided, in properly working devices. So, we have seen in this video, the drain induced lowering effect. How increasing the drain can affect the barrier next to the source and, therefore, the current. We saw that an effective threshold defined for these cases decreases as the drain voltage increases. And when things are taken to the extreme, you enter the region of punchthrough where divisors do not operate properly and we're not supposed to be there.