WEBVTT 00:00:00.000 --> 00:00:00.780 align:middle line:90% 00:00:00.780 --> 00:00:03.840 align:middle line:84% JUDY HOYT: OK, so there are two handouts for today 00:00:03.840 --> 00:00:05.040 align:middle line:90% that are in the back there. 00:00:05.040 --> 00:00:09.240 align:middle line:84% There is the lecture notes and handout 8, 00:00:09.240 --> 00:00:12.510 align:middle line:84% and also, problem set number 2, or homework number 2 00:00:12.510 --> 00:00:14.100 align:middle line:90% is going out today. 00:00:14.100 --> 00:00:19.120 align:middle line:84% Your homework 1s are graded, and I'll have them back next time. 00:00:19.120 --> 00:00:23.740 align:middle line:84% So this handout is a fairly long discussion. 00:00:23.740 --> 00:00:28.990 align:middle line:84% There's two major areas I want to cover in today's lecture. 00:00:28.990 --> 00:00:30.480 align:middle line:90% I want to finish up chapter 4. 00:00:30.480 --> 00:00:31.980 align:middle line:84% There was a section in chapter 4 we 00:00:31.980 --> 00:00:34.480 align:middle line:84% didn't get to talk about yet, which is about characterizing. 00:00:34.480 --> 00:00:36.660 align:middle line:84% We talked all about impurities and how they 00:00:36.660 --> 00:00:38.160 align:middle line:90% can be issues and problems. 00:00:38.160 --> 00:00:41.190 align:middle line:84% I'm going to talk about how we characterize those impurities. 00:00:41.190 --> 00:00:43.530 align:middle line:84% And then we're going to skip chapter 5. 00:00:43.530 --> 00:00:45.160 align:middle line:84% Remember, chapter 5 is on lithography. 00:00:45.160 --> 00:00:47.340 align:middle line:84% And we have a whole one-semester course here 00:00:47.340 --> 00:00:50.040 align:middle line:84% at MIT on lithography, Hank Smith's class. 00:00:50.040 --> 00:00:51.378 align:middle line:90% I'm going to skip that. 00:00:51.378 --> 00:00:52.920 align:middle line:84% You're welcome to read it, of course. 00:00:52.920 --> 00:00:56.250 align:middle line:84% But if you don't, you should go ahead and start reading chapter 00:00:56.250 --> 00:00:58.800 align:middle line:90% 6 on thermal oxidation. 00:00:58.800 --> 00:01:01.200 align:middle line:90% I'm going to start that. 00:01:01.200 --> 00:01:03.510 align:middle line:84% In fact, the next three lectures are going 00:01:03.510 --> 00:01:05.220 align:middle line:90% to be on thermal oxidation. 00:01:05.220 --> 00:01:06.393 align:middle line:90% And I've listed here-- 00:01:06.393 --> 00:01:07.560 align:middle line:90% I won't go through them all. 00:01:07.560 --> 00:01:09.180 align:middle line:90% You can read them later-- 00:01:09.180 --> 00:01:11.640 align:middle line:84% some of the major topics from chapter 6 00:01:11.640 --> 00:01:13.510 align:middle line:90% that we're going to discuss. 00:01:13.510 --> 00:01:16.740 align:middle line:84% So let's go on to talk about-- finish up the characterization 00:01:16.740 --> 00:01:19.930 align:middle line:90% on slide number 2. 00:01:19.930 --> 00:01:22.810 align:middle line:84% This slide is supposed to give you 00:01:22.810 --> 00:01:25.720 align:middle line:84% an idea of the different surface analysis 00:01:25.720 --> 00:01:28.390 align:middle line:84% techniques and their detection sensitivity, 00:01:28.390 --> 00:01:32.890 align:middle line:84% how low of a level of an impurity can they detect? 00:01:32.890 --> 00:01:34.750 align:middle line:84% And I apologize right off the bat 00:01:34.750 --> 00:01:38.710 align:middle line:84% because not all of these acronyms are defined. 00:01:38.710 --> 00:01:41.170 align:middle line:84% And it's hard when you haven't seen these 00:01:41.170 --> 00:01:42.910 align:middle line:84% before to see all these ridiculous 00:01:42.910 --> 00:01:45.340 align:middle line:84% looking acronyms, Auger, [? OAES. ?] 00:01:45.340 --> 00:01:48.310 align:middle line:84% We're going to talk about Auger in this lecture. 00:01:48.310 --> 00:01:51.640 align:middle line:84% Raman, FTIR, we'll talk about a few of these. 00:01:51.640 --> 00:01:53.890 align:middle line:84% These are techniques you can see just 00:01:53.890 --> 00:01:56.630 align:middle line:84% based on where they lie on this vertical bar. 00:01:56.630 --> 00:01:59.810 align:middle line:84% The vertical bar is atoms per cubic centimeter. 00:01:59.810 --> 00:02:03.040 align:middle line:84% So here, 100% would be 5 [INAUDIBLE] 22. 00:02:03.040 --> 00:02:06.610 align:middle line:84% That's 100% of silicon lattice atomic density. 00:02:06.610 --> 00:02:09.810 align:middle line:84% If you want to measure an impurity on the level of 1 00:02:09.810 --> 00:02:13.570 align:middle line:84% to 0.1 atomic percent, that level, you would use Auger, 00:02:13.570 --> 00:02:16.157 align:middle line:84% or [? ESCA, or ?] one of these types of techniques. 00:02:16.157 --> 00:02:18.490 align:middle line:84% Most of the time that's not going to be nearly adequate. 00:02:18.490 --> 00:02:19.910 align:middle line:84% It depends on what you're making. 00:02:19.910 --> 00:02:24.450 align:middle line:84% But we talked about needing to know impurity levels much, 00:02:24.450 --> 00:02:28.760 align:middle line:84% much lower than that in the 10 to the 10th or so atoms 00:02:28.760 --> 00:02:29.570 align:middle line:90% per square. 00:02:29.570 --> 00:02:33.410 align:middle line:84% This is if you integrate over some depth, or 10 00:02:33.410 --> 00:02:36.000 align:middle line:84% to the 15th per cubic centimeter on the left here. 00:02:36.000 --> 00:02:37.550 align:middle line:84% So the types of techniques people 00:02:37.550 --> 00:02:41.690 align:middle line:84% use for that are listed here in this lower shaded bar. 00:02:41.690 --> 00:02:44.930 align:middle line:84% X-ray fluorescence, which we'll talk about and surface SIMS, 00:02:44.930 --> 00:02:47.150 align:middle line:84% these are the two most common techniques 00:02:47.150 --> 00:02:48.470 align:middle line:90% for measuring impurities. 00:02:48.470 --> 00:02:52.310 align:middle line:84% And you can see they measure them down to the something 00:02:52.310 --> 00:02:55.690 align:middle line:84% like tens of parts per billion range. 00:02:55.690 --> 00:02:59.590 align:middle line:84% So those are some of the options available to us. 00:02:59.590 --> 00:03:03.690 align:middle line:84% Next slide, on slide 3, gives you a rough estimate 00:03:03.690 --> 00:03:05.700 align:middle line:84% of the depth of analysis, the depth 00:03:05.700 --> 00:03:07.717 align:middle line:84% from the surface of some of these techniques. 00:03:07.717 --> 00:03:09.300 align:middle line:84% And I've circled the ones that will be 00:03:09.300 --> 00:03:11.020 align:middle line:90% of interest for this lecture. 00:03:11.020 --> 00:03:16.200 align:middle line:84% So X-ray fluorescence, which is the acronym is TXRF, 00:03:16.200 --> 00:03:21.030 align:middle line:84% it's samples typically the near-surface region 00:03:21.030 --> 00:03:22.170 align:middle line:90% depending on how it's done. 00:03:22.170 --> 00:03:25.200 align:middle line:84% But it's in the near surface region, say, the top 100 00:03:25.200 --> 00:03:27.012 align:middle line:90% angstroms or so, roughly. 00:03:27.012 --> 00:03:28.470 align:middle line:84% Maybe it's a little less than that. 00:03:28.470 --> 00:03:30.000 align:middle line:84% There's a special technique called 00:03:30.000 --> 00:03:32.940 align:middle line:84% surface SIMS, secondary ion mass spectrometry. 00:03:32.940 --> 00:03:34.740 align:middle line:90% We'll talk about that. 00:03:34.740 --> 00:03:36.750 align:middle line:84% That's designed to, again, sample 00:03:36.750 --> 00:03:40.620 align:middle line:84% just the top 100 angstroms, just the surface region. 00:03:40.620 --> 00:03:45.450 align:middle line:84% And then there's a standard SIMS or depth-profiling SIMS. 00:03:45.450 --> 00:03:48.330 align:middle line:84% You can profile pretty much as deep as you need to. 00:03:48.330 --> 00:03:53.520 align:middle line:84% Typically, you profile here in the 1,000 to 10,000 angstroms 00:03:53.520 --> 00:03:56.560 align:middle line:90% or 0.1 micron to 1 micron. 00:03:56.560 --> 00:03:58.840 align:middle line:84% And this is a depth profiling type of SIMS. 00:03:58.840 --> 00:04:01.510 align:middle line:84% Whereas, this measures surface SIMS, as the name suggests, 00:04:01.510 --> 00:04:04.910 align:middle line:84% measures just what's in the very near-surface region. 00:04:04.910 --> 00:04:07.990 align:middle line:84% So those are some those three are three important techniques. 00:04:07.990 --> 00:04:11.050 align:middle line:90% 00:04:11.050 --> 00:04:14.920 align:middle line:84% Slide 4 is meant to be a generic view-- 00:04:14.920 --> 00:04:16.630 align:middle line:84% and I took this out of your text-- 00:04:16.630 --> 00:04:20.230 align:middle line:84% of how surface contaminants are measured in silicon wafers. 00:04:20.230 --> 00:04:21.730 align:middle line:84% A lot of the time, what you're doing 00:04:21.730 --> 00:04:23.320 align:middle line:84% is you're taking a silicon wafer, which 00:04:23.320 --> 00:04:24.862 align:middle line:84% is shown schematically on the bottom, 00:04:24.862 --> 00:04:27.550 align:middle line:84% with some surface layer, however deep it 00:04:27.550 --> 00:04:31.840 align:middle line:84% might be, with some contamination level in it. 00:04:31.840 --> 00:04:36.100 align:middle line:84% And we interact that surface with some kind of beam. 00:04:36.100 --> 00:04:39.070 align:middle line:84% It could be an incident electron beam, as shown on the left. 00:04:39.070 --> 00:04:41.260 align:middle line:90% And we look at what comes off. 00:04:41.260 --> 00:04:44.920 align:middle line:84% It could be an incident X-ray beam or an incident ion beam, 00:04:44.920 --> 00:04:46.600 align:middle line:84% as in Rutherford backscattering or SIMS. 00:04:46.600 --> 00:04:49.090 align:middle line:84% So a lot of these-- these are beam techniques, 00:04:49.090 --> 00:04:49.870 align:middle line:90% where we interact. 00:04:49.870 --> 00:04:51.590 align:middle line:84% We shoot a beam into the surface. 00:04:51.590 --> 00:04:53.380 align:middle line:84% It has some energetic interaction. 00:04:53.380 --> 00:04:54.800 align:middle line:90% Something comes off. 00:04:54.800 --> 00:04:58.000 align:middle line:84% Could be an X-ray, could be an electron, could be an ion. 00:04:58.000 --> 00:04:59.950 align:middle line:90% And we analyze what comes off. 00:04:59.950 --> 00:05:04.110 align:middle line:84% That's the basic idea, very schematically. 00:05:04.110 --> 00:05:07.620 align:middle line:84% Let's go on to slide 5, and slide 5 00:05:07.620 --> 00:05:12.030 align:middle line:84% is about what happens when we bombard the surface of silicon 00:05:12.030 --> 00:05:13.110 align:middle line:90% with electrons. 00:05:13.110 --> 00:05:15.750 align:middle line:84% And what it shows very schematically 00:05:15.750 --> 00:05:18.480 align:middle line:84% is a plot on the vertical axis of the number of electrons 00:05:18.480 --> 00:05:21.580 align:middle line:84% coming off as a function of their energy. 00:05:21.580 --> 00:05:24.120 align:middle line:84% And you can see various humps here. 00:05:24.120 --> 00:05:27.120 align:middle line:84% At very low energies, there are a large number 00:05:27.120 --> 00:05:28.140 align:middle line:90% of electrons coming off. 00:05:28.140 --> 00:05:29.860 align:middle line:84% These are called secondary electrons. 00:05:29.860 --> 00:05:32.910 align:middle line:90% Maybe they're in the 5 eV range. 00:05:32.910 --> 00:05:35.250 align:middle line:84% These are the electrons that are coming off 00:05:35.250 --> 00:05:37.980 align:middle line:84% that are forming images if you look at a scanning electron 00:05:37.980 --> 00:05:38.760 align:middle line:90% micrograph. 00:05:38.760 --> 00:05:41.400 align:middle line:84% I've shown a couple of scanning SEM micrographs. 00:05:41.400 --> 00:05:43.505 align:middle line:84% In fact, last time we had that gettering picture 00:05:43.505 --> 00:05:44.880 align:middle line:84% with all those little dots in it. 00:05:44.880 --> 00:05:46.500 align:middle line:84% And you can see throughout the wafer, 00:05:46.500 --> 00:05:50.100 align:middle line:84% see all those dots which were the oxygen precipitates. 00:05:50.100 --> 00:05:51.930 align:middle line:84% That was a scanning electron micrograph. 00:05:51.930 --> 00:05:55.620 align:middle line:84% So that was looking at these secondary electrons. 00:05:55.620 --> 00:05:59.430 align:middle line:84% At the very highest energies are backscattered electrons. 00:05:59.430 --> 00:06:01.680 align:middle line:84% Basically, the incoming ones just bounce off the atoms 00:06:01.680 --> 00:06:05.590 align:middle line:84% and get reflected back with some energy similar to what 00:06:05.590 --> 00:06:08.050 align:middle line:90% you shot the electron in. 00:06:08.050 --> 00:06:09.452 align:middle line:90% They don't tell you much-- 00:06:09.452 --> 00:06:11.410 align:middle line:84% neither the lowest energies or the highest ones 00:06:11.410 --> 00:06:13.810 align:middle line:84% tell you much about what you bounced off of. 00:06:13.810 --> 00:06:16.000 align:middle line:84% But there is a region of intermediate energies 00:06:16.000 --> 00:06:18.730 align:middle line:84% here called the Auger electrons, which is shown in here-- 00:06:18.730 --> 00:06:21.100 align:middle line:84% there aren't as many of them-- that 00:06:21.100 --> 00:06:24.820 align:middle line:84% give very specific information about the species that they hit 00:06:24.820 --> 00:06:28.048 align:middle line:84% or that they interacted with because they interact 00:06:28.048 --> 00:06:30.340 align:middle line:84% with the atoms in the substrate with the core electrons 00:06:30.340 --> 00:06:33.020 align:middle line:84% of those atoms that are more tightly bound to the nucleus. 00:06:33.020 --> 00:06:35.200 align:middle line:84% So here, at these intermediate energies 00:06:35.200 --> 00:06:36.700 align:middle line:84% is where we can get some information 00:06:36.700 --> 00:06:39.070 align:middle line:84% about what it was that the electron interacted with, 00:06:39.070 --> 00:06:41.950 align:middle line:90% what element. 00:06:41.950 --> 00:06:45.810 align:middle line:90% So we go to slide 6. 00:06:45.810 --> 00:06:48.240 align:middle line:84% Again, this is a very schematic illustration 00:06:48.240 --> 00:06:52.315 align:middle line:84% of the energy levels involved in this type of process. 00:06:52.315 --> 00:06:53.940 align:middle line:84% So let's just go through the schematic. 00:06:53.940 --> 00:06:57.270 align:middle line:84% What's shown up here in the upper center dot 00:06:57.270 --> 00:06:59.700 align:middle line:84% is a primary electron, which is this black dot. 00:06:59.700 --> 00:07:04.590 align:middle line:84% It's coming in, and it's interacting-- in energy space, 00:07:04.590 --> 00:07:08.430 align:middle line:84% it's interacting with an atom, let's say, 00:07:08.430 --> 00:07:11.370 align:middle line:84% down here that has a certain energy level. 00:07:11.370 --> 00:07:15.030 align:middle line:84% And what happens is basically it knocks-- 00:07:15.030 --> 00:07:18.630 align:middle line:84% these are the energy levels, this Ek, El1, El2, 00:07:18.630 --> 00:07:22.110 align:middle line:84% these are meant to represent the core energy 00:07:22.110 --> 00:07:25.650 align:middle line:84% levels of the electrons that are bound closer to the nucleus 00:07:25.650 --> 00:07:26.310 align:middle line:90% in the core. 00:07:26.310 --> 00:07:28.260 align:middle line:84% We've been talking all about valence electrons 00:07:28.260 --> 00:07:30.950 align:middle line:84% up here in the valence band and electrons in the conduction 00:07:30.950 --> 00:07:31.450 align:middle line:90% band. 00:07:31.450 --> 00:07:33.420 align:middle line:84% It doesn't interact with those, but it 00:07:33.420 --> 00:07:36.150 align:middle line:84% goes into the atomic core and interacts 00:07:36.150 --> 00:07:38.700 align:middle line:84% with the electrons that are there. 00:07:38.700 --> 00:07:42.910 align:middle line:84% OK, and basically, it kicks out one of these core level 00:07:42.910 --> 00:07:45.730 align:middle line:84% electrons from this Ek energy, where this 00:07:45.730 --> 00:07:47.110 align:middle line:90% becomes a secondary electron. 00:07:47.110 --> 00:07:49.910 align:middle line:84% Now you have an open energy level. 00:07:49.910 --> 00:07:52.990 align:middle line:84% And one of the higher level electrons, this one at El, 00:07:52.990 --> 00:07:54.910 align:middle line:90% can drop down into that level. 00:07:54.910 --> 00:07:58.060 align:middle line:84% And the spacing of this level between the L1 and the K level 00:07:58.060 --> 00:08:00.950 align:middle line:84% is very much a characteristic spacing of the atom. 00:08:00.950 --> 00:08:03.280 align:middle line:90% Depends on the element, OK? 00:08:03.280 --> 00:08:07.180 align:middle line:84% So you can imagine that this energy here, this transition 00:08:07.180 --> 00:08:09.430 align:middle line:84% energy is very specific to the atom. 00:08:09.430 --> 00:08:12.670 align:middle line:84% That transition energy is then, in this particular case, 00:08:12.670 --> 00:08:17.740 align:middle line:84% given off as kinetic energy to an ejected electron, which 00:08:17.740 --> 00:08:19.710 align:middle line:90% is called an Auger electron. 00:08:19.710 --> 00:08:21.780 align:middle line:84% That ejected electron then has an energy 00:08:21.780 --> 00:08:24.030 align:middle line:84% that's characteristic of this level spacing. 00:08:24.030 --> 00:08:27.030 align:middle line:84% So by measuring its the energy spectrum 00:08:27.030 --> 00:08:29.640 align:middle line:84% of those outgoing Auger electrons, 00:08:29.640 --> 00:08:31.410 align:middle line:84% it tells me something about the atoms 00:08:31.410 --> 00:08:34.278 align:middle line:84% that were in the substrate, what those spacings were. 00:08:34.278 --> 00:08:36.570 align:middle line:84% And as long as you know you have a fingerprint for what 00:08:36.570 --> 00:08:39.150 align:middle line:84% those spacings should be, you can tell what atom 00:08:39.150 --> 00:08:40.590 align:middle line:90% the electron interacted with. 00:08:40.590 --> 00:08:44.670 align:middle line:84% So that's the Auger electron spectroscopy. 00:08:44.670 --> 00:08:46.140 align:middle line:84% There's another way to do it where, 00:08:46.140 --> 00:08:50.370 align:middle line:84% instead of releasing an electron, an Auger electron, 00:08:50.370 --> 00:08:52.320 align:middle line:84% you may actually release an X-ray. 00:08:52.320 --> 00:08:55.800 align:middle line:84% This energy spacing between El1 and Ek 00:08:55.800 --> 00:08:58.800 align:middle line:84% can be given off to eject an X-ray. 00:08:58.800 --> 00:09:01.380 align:middle line:84% So you may also look for heavier elements 00:09:01.380 --> 00:09:03.455 align:middle line:90% at the X-rays coming off. 00:09:03.455 --> 00:09:05.830 align:middle line:84% And that will give you some species-specific information. 00:09:05.830 --> 00:09:07.830 align:middle line:84% So you're get the qualitative idea here. 00:09:07.830 --> 00:09:09.480 align:middle line:84% I'm putting in some energetic particle. 00:09:09.480 --> 00:09:13.930 align:middle line:84% I watch transitions that happen in the core levels. 00:09:13.930 --> 00:09:15.840 align:middle line:84% And then I look at what comes off, 00:09:15.840 --> 00:09:19.260 align:middle line:84% either another electron or an X-ray. 00:09:19.260 --> 00:09:24.430 align:middle line:84% So that was an incident in that last slide, slide 6. 00:09:24.430 --> 00:09:28.560 align:middle line:84% The incident particle was an electron. 00:09:28.560 --> 00:09:31.050 align:middle line:84% You don't have to use electrons, as it turns out. 00:09:31.050 --> 00:09:33.450 align:middle line:84% You just need something that can bring some energy 00:09:33.450 --> 00:09:36.360 align:middle line:90% into the atomic core. 00:09:36.360 --> 00:09:39.690 align:middle line:84% So here on slide 7, you can imagine the incident energy 00:09:39.690 --> 00:09:41.158 align:middle line:90% might be provided by an X-ray. 00:09:41.158 --> 00:09:42.450 align:middle line:90% Doesn't have to be an electron. 00:09:42.450 --> 00:09:46.460 align:middle line:84% So you may bombard the surface with X-rays instead. 00:09:46.460 --> 00:09:49.880 align:middle line:84% And there are two different categories here, 00:09:49.880 --> 00:09:53.270 align:middle line:84% XPS, X-ray photoelectron spectroscopy, 00:09:53.270 --> 00:09:56.570 align:middle line:84% where you bring an X-ray in that provides 00:09:56.570 --> 00:09:58.520 align:middle line:84% the energy and an electron comes out, 00:09:58.520 --> 00:10:03.950 align:middle line:84% or X-ray fluorescence, where you bring an X-ray in, 00:10:03.950 --> 00:10:06.560 align:middle line:84% and another X-ray is emitted with 00:10:06.560 --> 00:10:10.480 align:middle line:90% a characteristic wavelength. 00:10:10.480 --> 00:10:13.510 align:middle line:84% And what's happening here in this XRF, 00:10:13.510 --> 00:10:16.240 align:middle line:84% this X-ray fluorescence, is pictured schematically here 00:10:16.240 --> 00:10:18.850 align:middle line:84% on the left for an element that's 00:10:18.850 --> 00:10:25.150 align:middle line:84% relatively light, has fewer shells in its electrons, 00:10:25.150 --> 00:10:26.740 align:middle line:84% in its core levels, and one that's 00:10:26.740 --> 00:10:30.620 align:middle line:84% relatively heavy on the right, which has lots of core levels. 00:10:30.620 --> 00:10:35.050 align:middle line:84% So you can imagine an incident X-ray coming in. 00:10:35.050 --> 00:10:38.510 align:middle line:90% It's knocking an electron out. 00:10:38.510 --> 00:10:41.620 align:middle line:84% And you get a transition energy between two core levels. 00:10:41.620 --> 00:10:44.590 align:middle line:84% And what comes off is an electron with-- 00:10:44.590 --> 00:10:46.960 align:middle line:84% an X-ray with some energy N nu with 00:10:46.960 --> 00:10:48.590 align:middle line:90% a characteristic wavelength. 00:10:48.590 --> 00:10:50.590 align:middle line:84% The problem with doing this with a light element 00:10:50.590 --> 00:10:52.820 align:middle line:84% is there are too many electron shells. 00:10:52.820 --> 00:10:57.030 align:middle line:84% So this process is not that favored, 00:10:57.030 --> 00:10:59.440 align:middle line:90% and it becomes difficult to do. 00:10:59.440 --> 00:11:02.080 align:middle line:84% For heavier elements, it's much more practical, 00:11:02.080 --> 00:11:05.110 align:middle line:84% heavier elements being the transition metals or things 00:11:05.110 --> 00:11:06.820 align:middle line:90% that are heavy, like silicon. 00:11:06.820 --> 00:11:08.320 align:middle line:84% There are lots of these core levels. 00:11:08.320 --> 00:11:09.590 align:middle line:90% You bring an X-ray in. 00:11:09.590 --> 00:11:12.200 align:middle line:84% It has a high probability of interacting, producing 00:11:12.200 --> 00:11:15.030 align:middle line:84% an X-ray coming out that has a characteristic spectrum. 00:11:15.030 --> 00:11:18.770 align:middle line:84% So you see, if you plot the X-ray density 00:11:18.770 --> 00:11:21.560 align:middle line:84% or intensity coming out versus energy, you see these peaks. 00:11:21.560 --> 00:11:24.560 align:middle line:84% And this spectrum becomes a fingerprint 00:11:24.560 --> 00:11:31.700 align:middle line:84% of what elements the X-ray has interacted with. 00:11:31.700 --> 00:11:36.790 align:middle line:84% So if we go on to slide 8, this is a very common technique 00:11:36.790 --> 00:11:40.840 align:middle line:84% that's used in semiconductor analysis. 00:11:40.840 --> 00:11:42.610 align:middle line:84% I'm showing here a typical, what's called, 00:11:42.610 --> 00:11:46.030 align:middle line:84% total X-ray fluorescence, TXRF, the spectrum. 00:11:46.030 --> 00:11:49.510 align:middle line:84% And the plot shows on the y-axis the intensity 00:11:49.510 --> 00:11:54.670 align:middle line:84% of the X-rays coming off in counts per second, detected 00:11:54.670 --> 00:11:59.120 align:middle line:84% as a function of the X-ray energy, so that H nu 00:11:59.120 --> 00:12:01.490 align:middle line:84% or whatever you want to call it, that's coming off. 00:12:01.490 --> 00:12:03.140 align:middle line:84% And again, these energies are going 00:12:03.140 --> 00:12:06.150 align:middle line:84% to be characteristic of the core levels of the atom. 00:12:06.150 --> 00:12:10.015 align:middle line:84% So next to each peak, the person who did the spectrum actually 00:12:10.015 --> 00:12:11.390 align:middle line:84% is familiar enough that they were 00:12:11.390 --> 00:12:12.680 align:middle line:90% able to identify the peaks. 00:12:12.680 --> 00:12:16.280 align:middle line:84% For instance, this peak right here, little above 2 keV, 00:12:16.280 --> 00:12:17.750 align:middle line:90% corresponds to chlorine. 00:12:17.750 --> 00:12:20.217 align:middle line:84% And you can integrate the area under that peak 00:12:20.217 --> 00:12:21.800 align:middle line:84% and get a rough estimate of the amount 00:12:21.800 --> 00:12:24.800 align:middle line:84% of chlorine atoms per square centimeter 00:12:24.800 --> 00:12:27.573 align:middle line:90% that exists on that surface. 00:12:27.573 --> 00:12:29.240 align:middle line:84% Silicon, of course, is going to be large 00:12:29.240 --> 00:12:31.250 align:middle line:90% because it's a silicon surface. 00:12:31.250 --> 00:12:33.020 align:middle line:84% There's a lot of chromium on this sample-- 00:12:33.020 --> 00:12:34.478 align:middle line:84% look at the peak labeled chromium-- 00:12:34.478 --> 00:12:35.690 align:middle line:90% and a fair amount of iron. 00:12:35.690 --> 00:12:38.270 align:middle line:84% So the nice thing, it gives you a broad spectrum. 00:12:38.270 --> 00:12:40.670 align:middle line:84% You can just shoot the X-ray onto the surface 00:12:40.670 --> 00:12:42.170 align:middle line:90% in about a 1-centimeter area. 00:12:42.170 --> 00:12:45.830 align:middle line:84% And you can immediately survey in the transition metals what's 00:12:45.830 --> 00:12:47.040 align:middle line:90% on that surface. 00:12:47.040 --> 00:12:50.820 align:middle line:84% It's very useful and commonly-used technique. 00:12:50.820 --> 00:12:53.180 align:middle line:90% So let's go on to slide 9. 00:12:53.180 --> 00:12:56.330 align:middle line:84% So this TXRF, what are some of the good things about it? 00:12:56.330 --> 00:12:59.030 align:middle line:84% Well, I just showed you, it's a survey technique 00:12:59.030 --> 00:13:01.910 align:middle line:84% to get a broad idea of what different elements are 00:13:01.910 --> 00:13:02.880 align:middle line:90% on the wafer. 00:13:02.880 --> 00:13:04.850 align:middle line:90% And it can be automated. 00:13:04.850 --> 00:13:06.290 align:middle line:90% It's reasonably quantitative. 00:13:06.290 --> 00:13:09.810 align:middle line:84% We can integrate those peaks and compare them to standards. 00:13:09.810 --> 00:13:11.210 align:middle line:90% It doesn't destroy the wafer. 00:13:11.210 --> 00:13:14.900 align:middle line:84% You notice the X-ray just comes in, interacts, and goes out. 00:13:14.900 --> 00:13:16.610 align:middle line:84% It's not like I have to sputter off, 00:13:16.610 --> 00:13:18.350 align:middle line:84% or I don't even have to break the wafer. 00:13:18.350 --> 00:13:19.725 align:middle line:84% You can do it on an entire wafer. 00:13:19.725 --> 00:13:22.490 align:middle line:84% They have machines for 12-inch wafers you can put in there. 00:13:22.490 --> 00:13:23.990 align:middle line:84% And they'll tell you what impurities 00:13:23.990 --> 00:13:26.190 align:middle line:90% are on that wafer surface. 00:13:26.190 --> 00:13:27.380 align:middle line:90% So that's really nice. 00:13:27.380 --> 00:13:31.130 align:middle line:84% A weakness, it doesn't detect low-z elements 00:13:31.130 --> 00:13:34.860 align:middle line:84% because the physics of-- there aren't enough core level 00:13:34.860 --> 00:13:35.360 align:middle line:90% shells. 00:13:35.360 --> 00:13:39.140 align:middle line:84% So aluminum, potassium, and sodium, both of which 00:13:39.140 --> 00:13:42.570 align:middle line:84% are very nasty, and silicon can't be detected. 00:13:42.570 --> 00:13:44.030 align:middle line:90% So that's a major limitation. 00:13:44.030 --> 00:13:46.580 align:middle line:84% It doesn't tell you about how it's distributed in depth. 00:13:46.580 --> 00:13:49.580 align:middle line:84% It doesn't tell you anything about the depth distribution. 00:13:49.580 --> 00:13:51.180 align:middle line:90% And it's pretty big spot. 00:13:51.180 --> 00:13:53.220 align:middle line:84% About 10 millimeters is the smallest 00:13:53.220 --> 00:13:55.080 align:middle line:90% you can focus the beam down. 00:13:55.080 --> 00:13:57.660 align:middle line:84% So you're looking at large spots. 00:13:57.660 --> 00:13:58.212 align:middle line:90% That's OK. 00:13:58.212 --> 00:13:59.670 align:middle line:84% But if you want to look, let's say, 00:13:59.670 --> 00:14:02.003 align:middle line:84% inside an individual device and see what impurities were 00:14:02.003 --> 00:14:04.260 align:middle line:84% in a bad device, if you had a fail device, 00:14:04.260 --> 00:14:06.485 align:middle line:84% TXRF is not the way to go because the device isn't 00:14:06.485 --> 00:14:08.860 align:middle line:84% much bigger than-- it's much smaller than 10 millimeters. 00:14:08.860 --> 00:14:12.210 align:middle line:84% That's the whole die on that wafer being passed around. 00:14:12.210 --> 00:14:14.280 align:middle line:84% And the surface has to be specular and polished. 00:14:14.280 --> 00:14:17.520 align:middle line:84% If the surface is patterned-- like on this wafer here, 00:14:17.520 --> 00:14:20.020 align:middle line:84% you see some patterns, it's not perfectly specular-- you're 00:14:20.020 --> 00:14:22.020 align:middle line:84% going to have a problem because you're not going 00:14:22.020 --> 00:14:23.320 align:middle line:90% to get the kind of reflectance. 00:14:23.320 --> 00:14:25.860 align:middle line:84% So it's typically done in an area that's 00:14:25.860 --> 00:14:28.650 align:middle line:90% relatively well polished. 00:14:28.650 --> 00:14:33.030 align:middle line:84% OK, there is an alternative technique, which actually 00:14:33.030 --> 00:14:34.200 align:middle line:90% is quite complementary. 00:14:34.200 --> 00:14:36.990 align:middle line:84% Where TXRF has weaknesses, surface SIMS 00:14:36.990 --> 00:14:38.430 align:middle line:90% has some strengths. 00:14:38.430 --> 00:14:40.620 align:middle line:84% So people often use both of them. 00:14:40.620 --> 00:14:43.600 align:middle line:84% This is an example of an ion beam technique. 00:14:43.600 --> 00:14:46.380 align:middle line:84% So instead of bringing in X-rays, I'm bringing in ions. 00:14:46.380 --> 00:14:49.990 align:middle line:84% And the ions that come in are called the primary ions 00:14:49.990 --> 00:14:52.740 align:middle line:90% or the primary particles. 00:14:52.740 --> 00:14:57.030 align:middle line:84% The two typical ones to use are either cesium ions or oxygen 00:14:57.030 --> 00:14:58.530 align:middle line:90% ions. 00:14:58.530 --> 00:15:01.850 align:middle line:84% So you have a primary beam of particles. 00:15:01.850 --> 00:15:03.260 align:middle line:90% Energy varies. 00:15:03.260 --> 00:15:04.970 align:middle line:90% This says 10 keV. 00:15:04.970 --> 00:15:08.060 align:middle line:84% In some cases, you can bring this energy all the way down 00:15:08.060 --> 00:15:10.640 align:middle line:90% to 1 keV. 00:15:10.640 --> 00:15:13.040 align:middle line:84% And there's an advantage in bringing the primary energy 00:15:13.040 --> 00:15:15.328 align:middle line:90% down lower. 00:15:15.328 --> 00:15:17.120 align:middle line:84% What you see is this primary beam comes in, 00:15:17.120 --> 00:15:18.180 align:middle line:90% and what does it do? 00:15:18.180 --> 00:15:21.620 align:middle line:84% Well, it interacts with the solid, the surface. 00:15:21.620 --> 00:15:24.710 align:middle line:84% And of course, the atoms, the ions penetrate in, 00:15:24.710 --> 00:15:27.840 align:middle line:84% and they get implanted some depth. 00:15:27.840 --> 00:15:29.360 align:middle line:90% They penetrate in some depth. 00:15:29.360 --> 00:15:32.130 align:middle line:84% That depth that they penetrate depends on their energy. 00:15:32.130 --> 00:15:38.110 align:middle line:84% So if I'm at 10 keV, I may go down to 100 angstroms. 00:15:38.110 --> 00:15:42.108 align:middle line:84% Now, the ions that come off, the atoms that are sputtered off, 00:15:42.108 --> 00:15:43.150 align:middle line:90% some of them are ionized. 00:15:43.150 --> 00:15:44.950 align:middle line:84% And that's what we analyze in the machine. 00:15:44.950 --> 00:15:46.810 align:middle line:84% And we measure the atoms that come off. 00:15:46.810 --> 00:15:48.730 align:middle line:84% They only come off the top 10 angstroms 00:15:48.730 --> 00:15:51.478 align:middle line:84% or so because that's the escape depth of which they 00:15:51.478 --> 00:15:52.270 align:middle line:90% have enough energy. 00:15:52.270 --> 00:15:55.850 align:middle line:84% These other particles that come off can get out. 00:15:55.850 --> 00:15:56.530 align:middle line:90% So that's good. 00:15:56.530 --> 00:15:58.090 align:middle line:84% We can sample the top 10 angstroms. 00:15:58.090 --> 00:16:01.060 align:middle line:84% But notice, I disturbed-- in doing the measurement, 00:16:01.060 --> 00:16:02.890 align:middle line:84% like Heisenberg, in doing the measurement, 00:16:02.890 --> 00:16:03.970 align:middle line:90% I disturbed the crystal. 00:16:03.970 --> 00:16:06.550 align:middle line:84% I shot a bunch of ions in here, and now they're 00:16:06.550 --> 00:16:08.290 align:middle line:90% down here at 100 angstroms deep. 00:16:08.290 --> 00:16:10.870 align:middle line:84% And they knocked things around in the substrate. 00:16:10.870 --> 00:16:13.210 align:middle line:84% So whatever atoms were down there, they disturbed them. 00:16:13.210 --> 00:16:14.168 align:middle line:90% They moved them around. 00:16:14.168 --> 00:16:16.840 align:middle line:84% They might have knocked them in deeper. 00:16:16.840 --> 00:16:19.720 align:middle line:84% So if I can lower this primary particle energy down 00:16:19.720 --> 00:16:23.110 align:middle line:84% to one keV, they won't be knocked in so deep. 00:16:23.110 --> 00:16:25.220 align:middle line:84% And they won't disturb as much of the crystal. 00:16:25.220 --> 00:16:27.700 align:middle line:84% So you get better depth resolution in general 00:16:27.700 --> 00:16:31.280 align:middle line:84% as you bring the primary ion beam energy down. 00:16:31.280 --> 00:16:33.290 align:middle line:84% Basic idea is you have a primary ion beam. 00:16:33.290 --> 00:16:33.800 align:middle line:90% It comes in. 00:16:33.800 --> 00:16:36.860 align:middle line:84% It sputters off a certain number of monolayers 00:16:36.860 --> 00:16:37.940 align:middle line:90% in a certain period. 00:16:37.940 --> 00:16:40.280 align:middle line:84% And you measure the ions that come off the substrate. 00:16:40.280 --> 00:16:44.360 align:middle line:84% And you count how many ions are of silicon, how many coming off 00:16:44.360 --> 00:16:47.110 align:middle line:84% are copper, how many are gold, whatever. 00:16:47.110 --> 00:16:48.110 align:middle line:90% And you get an estimate. 00:16:48.110 --> 00:16:50.240 align:middle line:84% You can get a spectrum of what's coming off. 00:16:50.240 --> 00:16:51.920 align:middle line:84% You can't count too many because it has 00:16:51.920 --> 00:16:53.128 align:middle line:90% to go through a spectrometer. 00:16:53.128 --> 00:16:56.120 align:middle line:84% And the spectrometer can only count so many at one time. 00:16:56.120 --> 00:16:59.300 align:middle line:84% But that's the basic idea of secondary ion mass 00:16:59.300 --> 00:17:01.970 align:middle line:90% spectrometry. 00:17:01.970 --> 00:17:04.210 align:middle line:84% If we go on to slide 11, I actually 00:17:04.210 --> 00:17:06.520 align:middle line:90% took this off the website. 00:17:06.520 --> 00:17:08.079 align:middle line:84% One of the commercial places that 00:17:08.079 --> 00:17:12.040 align:middle line:84% does a lot of this analysis commercially 00:17:12.040 --> 00:17:13.972 align:middle line:84% is called Charles Evans and Associates. 00:17:13.972 --> 00:17:15.430 align:middle line:84% And if you go on to their website-- 00:17:15.430 --> 00:17:16.513 align:middle line:90% you just search on Google. 00:17:16.513 --> 00:17:17.780 align:middle line:90% You can get to their website. 00:17:17.780 --> 00:17:19.197 align:middle line:84% And they have a lot of information 00:17:19.197 --> 00:17:20.200 align:middle line:90% on these techniques. 00:17:20.200 --> 00:17:22.180 align:middle line:84% I won't go into slide 11 in any great detail. 00:17:22.180 --> 00:17:23.410 align:middle line:90% I'll let you read it later. 00:17:23.410 --> 00:17:26.829 align:middle line:84% I just put it up there to show that companies 00:17:26.829 --> 00:17:29.470 align:middle line:84% have done a lot of work in order to quantitatively 00:17:29.470 --> 00:17:31.870 align:middle line:84% be able to measure in the top, say, 00:17:31.870 --> 00:17:34.930 align:middle line:84% 10 to 20 nanometers or 100 to 200 angstroms, 00:17:34.930 --> 00:17:39.342 align:middle line:84% quantitatively measure the contaminants 00:17:39.342 --> 00:17:40.300 align:middle line:90% in that surface region. 00:17:40.300 --> 00:17:41.890 align:middle line:84% It wasn't always possible to do that. 00:17:41.890 --> 00:17:44.927 align:middle line:84% But they have nice techniques to do that now. 00:17:44.927 --> 00:17:47.260 align:middle line:84% And if you're interested, you can do more detailed study 00:17:47.260 --> 00:17:49.430 align:middle line:90% on the web. 00:17:49.430 --> 00:17:52.310 align:middle line:84% Let's go on to slide 12, which lists 00:17:52.310 --> 00:17:54.830 align:middle line:84% some of the key features of surface SIMS. 00:17:54.830 --> 00:17:57.860 align:middle line:84% Well, it's been approved by ASTM for measuring 00:17:57.860 --> 00:18:00.560 align:middle line:84% some of the elements and exactly some of the elements 00:18:00.560 --> 00:18:02.370 align:middle line:90% that the TXRF can't measure. 00:18:02.370 --> 00:18:03.290 align:middle line:90% So that's nice. 00:18:03.290 --> 00:18:05.660 align:middle line:84% It can quantitatively measure contamination 00:18:05.660 --> 00:18:09.780 align:middle line:84% for sodium, aluminum, potassium, iron, 00:18:09.780 --> 00:18:11.073 align:middle line:90% on into [INAUDIBLE] silicon. 00:18:11.073 --> 00:18:12.740 align:middle line:84% And it can measure lots of the elements. 00:18:12.740 --> 00:18:14.407 align:middle line:84% But it's actually been approved as being 00:18:14.407 --> 00:18:16.340 align:middle line:90% a standardized technique. 00:18:16.340 --> 00:18:17.730 align:middle line:90% It's quite accurate. 00:18:17.730 --> 00:18:20.300 align:middle line:84% It can detect many elements and isotopes. 00:18:20.300 --> 00:18:22.220 align:middle line:84% But detection limits is quite low. 00:18:22.220 --> 00:18:23.015 align:middle line:90% Look at this. 00:18:23.015 --> 00:18:24.890 align:middle line:84% Look at these numbers, 10 to the eighth to 10 00:18:24.890 --> 00:18:27.860 align:middle line:84% to the ninth atoms per square centimeter for most metals. 00:18:27.860 --> 00:18:28.920 align:middle line:90% Now, why do we care? 00:18:28.920 --> 00:18:30.290 align:middle line:90% Well, what was on the ITRS? 00:18:30.290 --> 00:18:33.260 align:middle line:84% When you read your ITRS in your homework, 00:18:33.260 --> 00:18:35.300 align:middle line:84% the number of atoms per square centimeter 00:18:35.300 --> 00:18:39.350 align:middle line:84% that people were interested in was what order of magnitude? 00:18:39.350 --> 00:18:40.877 align:middle line:90% 10 on the 10th, exactly. 00:18:40.877 --> 00:18:42.710 align:middle line:84% So you better be able to measure below that. 00:18:42.710 --> 00:18:44.600 align:middle line:84% Otherwise, you can't even tell what you have. 00:18:44.600 --> 00:18:47.137 align:middle line:84% So it's right in the range where we need 00:18:47.137 --> 00:18:48.345 align:middle line:90% to be measuring for the ITRS. 00:18:48.345 --> 00:18:51.570 align:middle line:90% 00:18:51.570 --> 00:18:54.940 align:middle line:84% It also can give you information about the profile in depth. 00:18:54.940 --> 00:18:55.940 align:middle line:90% And that's kind of nice. 00:18:55.940 --> 00:18:58.490 align:middle line:84% If you go back, if we go back a couple slides 00:18:58.490 --> 00:19:02.420 align:middle line:84% to slide 10, what happens is you collect these ions 00:19:02.420 --> 00:19:03.620 align:middle line:90% as they come off. 00:19:03.620 --> 00:19:05.840 align:middle line:84% And you measure them as a function of time. 00:19:05.840 --> 00:19:07.400 align:middle line:84% In the first couple of seconds, you 00:19:07.400 --> 00:19:09.110 align:middle line:84% sputter off the first few monolayers, 00:19:09.110 --> 00:19:10.970 align:middle line:84% the second few seconds you sputter deeper. 00:19:10.970 --> 00:19:13.700 align:middle line:84% So you're constantly sputtering the surface 00:19:13.700 --> 00:19:15.350 align:middle line:84% of at a constant rate, and you're 00:19:15.350 --> 00:19:18.230 align:middle line:84% measuring what comes off as a function of time. 00:19:18.230 --> 00:19:21.260 align:middle line:84% So you know by collecting this as a function of time 00:19:21.260 --> 00:19:23.970 align:middle line:84% pretty much exactly what depth those ions came from. 00:19:23.970 --> 00:19:26.360 align:middle line:84% So if you want to sputter a little deeper, 00:19:26.360 --> 00:19:30.220 align:middle line:84% you can get some information about the profile. 00:19:30.220 --> 00:19:32.530 align:middle line:84% And finally, going back to slide 12, 00:19:32.530 --> 00:19:35.260 align:middle line:84% it has a relatively small detection area, 00:19:35.260 --> 00:19:36.790 align:middle line:90% say, 50 by 50 microns. 00:19:36.790 --> 00:19:39.430 align:middle line:84% Still not that small compared to the size of a device. 00:19:39.430 --> 00:19:41.930 align:middle line:84% But it's better than 10 millimeters. 00:19:41.930 --> 00:19:43.645 align:middle line:90% So you can get in-- 00:19:43.645 --> 00:19:46.930 align:middle line:84% you can put special structures on the mask 00:19:46.930 --> 00:19:50.532 align:middle line:84% with this kind of size area that are dedicated for device 00:19:50.532 --> 00:19:52.240 align:middle line:84% characterization if you're doing research 00:19:52.240 --> 00:19:55.750 align:middle line:84% and you want to be able to measure on the chip, measure 00:19:55.750 --> 00:20:01.280 align:middle line:84% the characterization, the contamination. 00:20:01.280 --> 00:20:03.110 align:middle line:90% Let's go on to slide 13. 00:20:03.110 --> 00:20:05.090 align:middle line:84% What I'm showing here, again, I took this off 00:20:05.090 --> 00:20:07.340 align:middle line:84% of the Charles Evans and Associates website. 00:20:07.340 --> 00:20:08.937 align:middle line:90% And it's a little bit dated. 00:20:08.937 --> 00:20:10.520 align:middle line:84% It's a few years old, but it gives you 00:20:10.520 --> 00:20:14.990 align:middle line:84% a rough idea of comparing total X-ray fluorescence, TXRF, 00:20:14.990 --> 00:20:18.440 align:middle line:84% in these columns on the left, to surface SIMS. 00:20:18.440 --> 00:20:21.260 align:middle line:84% In terms of its detection limit, so the detection limit 00:20:21.260 --> 00:20:24.440 align:middle line:84% is the lowest concentration, which 00:20:24.440 --> 00:20:25.837 align:middle line:90% either technique can measure. 00:20:25.837 --> 00:20:27.170 align:middle line:90% And so you can get a rough idea. 00:20:27.170 --> 00:20:31.910 align:middle line:84% Let's say just for comparison, let's look at potassium here. 00:20:31.910 --> 00:20:34.070 align:middle line:84% TXRF doesn't measure it very well. 00:20:34.070 --> 00:20:36.620 align:middle line:84% The lowest it can measure here is about 20 times 00:20:36.620 --> 00:20:40.475 align:middle line:84% 10 to the 10th, or 210 to the 11th, not all that sensitive. 00:20:40.475 --> 00:20:42.350 align:middle line:84% At least a few years ago, that was the limit. 00:20:42.350 --> 00:20:43.433 align:middle line:90% Maybe they've improved it. 00:20:43.433 --> 00:20:46.250 align:middle line:84% But look at the detection limit using surface SIMS. 00:20:46.250 --> 00:20:50.540 align:middle line:84% It's 0.01 times 10 to the 10th or 10 to the eighth per square, 00:20:50.540 --> 00:20:51.818 align:middle line:90% much more sensitive. 00:20:51.818 --> 00:20:54.110 align:middle line:84% So it gives you an idea of what the detection limit is. 00:20:54.110 --> 00:20:57.770 align:middle line:84% So if I know that I'm concerned about potassium, 00:20:57.770 --> 00:20:59.990 align:middle line:84% I would shoot for SIMS, not TSRF. 00:20:59.990 --> 00:21:02.280 align:middle line:84% So it's a nice quantitative comparison 00:21:02.280 --> 00:21:04.320 align:middle line:90% for the different elements. 00:21:04.320 --> 00:21:08.340 align:middle line:90% 00:21:08.340 --> 00:21:12.710 align:middle line:84% So I went through that relatively quickly, I admit. 00:21:12.710 --> 00:21:14.540 align:middle line:84% But you can read in your textbook 00:21:14.540 --> 00:21:16.130 align:middle line:84% on the characterization techniques. 00:21:16.130 --> 00:21:17.930 align:middle line:90% In chapter 4, they're covered. 00:21:17.930 --> 00:21:19.640 align:middle line:84% And I would also encourage you, if you're 00:21:19.640 --> 00:21:22.350 align:middle line:84% doing research in this area and you need to know more, 00:21:22.350 --> 00:21:23.810 align:middle line:90% go to the Charles Evans website. 00:21:23.810 --> 00:21:26.060 align:middle line:90% They are a commercial vendor. 00:21:26.060 --> 00:21:28.160 align:middle line:84% They do sell characterizations for a business. 00:21:28.160 --> 00:21:31.220 align:middle line:84% But they're the world's experts as a result on that. 00:21:31.220 --> 00:21:34.950 align:middle line:84% So it's a nice place to get information. 00:21:34.950 --> 00:21:37.910 align:middle line:84% So I want to move on now starting with slide 14 00:21:37.910 --> 00:21:41.750 align:middle line:84% and talk about oxides and the IC industry. 00:21:41.750 --> 00:21:44.330 align:middle line:84% And I think we hinted at this in the first lecture or two. 00:21:44.330 --> 00:21:46.790 align:middle line:84% But really, it's silicon-silicon-dioxide 00:21:46.790 --> 00:21:47.330 align:middle line:90% interface. 00:21:47.330 --> 00:21:48.950 align:middle line:90% And it's perfect properties. 00:21:48.950 --> 00:21:52.160 align:middle line:84% That is the number one reason why silicon dominates 00:21:52.160 --> 00:21:54.710 align:middle line:84% integrated circuits today compared to germanium, 00:21:54.710 --> 00:21:56.210 align:middle line:84% compared to gallium arsenide or any 00:21:56.210 --> 00:21:58.100 align:middle line:90% of the other semiconductors. 00:21:58.100 --> 00:22:00.680 align:middle line:84% And these are some properties here listed 00:22:00.680 --> 00:22:02.960 align:middle line:84% in this bullet list of silicon dioxide 00:22:02.960 --> 00:22:05.090 align:middle line:90% that are quite desirable. 00:22:05.090 --> 00:22:09.140 align:middle line:84% It's easy to selectively etch silicon dioxide 00:22:09.140 --> 00:22:11.420 align:middle line:90% and to pattern it. 00:22:11.420 --> 00:22:14.480 align:middle line:84% It also masks the diffusion of a lot of common impurities, which 00:22:14.480 --> 00:22:16.070 align:middle line:90% is a really great thing. 00:22:16.070 --> 00:22:18.440 align:middle line:84% You can put down SiO2 and pattern it, 00:22:18.440 --> 00:22:22.940 align:middle line:84% and some impurities will not diffuse through it. 00:22:22.940 --> 00:22:24.460 align:middle line:84% It has great electrical properties. 00:22:24.460 --> 00:22:27.860 align:middle line:84% It's a very good insulator, at least when it's thick enough. 00:22:27.860 --> 00:22:29.860 align:middle line:90% It has a high breakdown field. 00:22:29.860 --> 00:22:33.240 align:middle line:84% It tends to passivate junctions very well. 00:22:33.240 --> 00:22:36.183 align:middle line:84% It has a very stable and reproducible interface. 00:22:36.183 --> 00:22:37.600 align:middle line:84% For any of you who have ever tried 00:22:37.600 --> 00:22:41.110 align:middle line:84% to make oxides of this quality on other semiconductors, 00:22:41.110 --> 00:22:42.970 align:middle line:84% like germanium or gallium arsenide, 00:22:42.970 --> 00:22:46.060 align:middle line:84% you immediately appreciate the properties of silicon dioxide 00:22:46.060 --> 00:22:48.280 align:middle line:84% because on those other semiconductors 00:22:48.280 --> 00:22:53.200 align:middle line:84% it's 10 PhD thesis to try to get a decent passivation 00:22:53.200 --> 00:22:54.970 align:middle line:90% of those semiconductors. 00:22:54.970 --> 00:22:56.350 align:middle line:90% So silicon is really-- 00:22:56.350 --> 00:22:59.770 align:middle line:84% we were very, very lucky coming upon silicon-silicon-dioxide 00:22:59.770 --> 00:23:00.490 align:middle line:90% interface. 00:23:00.490 --> 00:23:03.400 align:middle line:90% It's really unique. 00:23:03.400 --> 00:23:07.930 align:middle line:84% On slide 15, this is a little bit dated, 00:23:07.930 --> 00:23:11.470 align:middle line:84% but this just shows some of the uses of silicon dioxide in IC 00:23:11.470 --> 00:23:12.910 align:middle line:90% technology. 00:23:12.910 --> 00:23:14.230 align:middle line:90% And there are two columns here. 00:23:14.230 --> 00:23:17.320 align:middle line:84% On the left, I'm showing thermally grown oxides, 00:23:17.320 --> 00:23:20.140 align:middle line:84% and we'll talk about that process 00:23:20.140 --> 00:23:21.850 align:middle line:90% in the next few lectures. 00:23:21.850 --> 00:23:24.490 align:middle line:84% On the right are deposited oxide layers. 00:23:24.490 --> 00:23:27.070 align:middle line:84% These are put on the wafer, rather 00:23:27.070 --> 00:23:28.840 align:middle line:84% than growing and consuming the wafer, 00:23:28.840 --> 00:23:31.510 align:middle line:84% by a process called chemical vapor deposition. 00:23:31.510 --> 00:23:34.450 align:middle line:84% And we'll talk about these a little bit later in the class, 00:23:34.450 --> 00:23:37.760 align:middle line:90% in the course later on. 00:23:37.760 --> 00:23:40.520 align:middle line:84% And these deposited oxides are different. 00:23:40.520 --> 00:23:42.070 align:middle line:84% They're usually not used for layers 00:23:42.070 --> 00:23:43.780 align:middle line:90% below about 10 nanometers. 00:23:43.780 --> 00:23:45.100 align:middle line:90% It's harder to control. 00:23:45.100 --> 00:23:47.590 align:middle line:84% And the properties of the silicon silicon dioxide 00:23:47.590 --> 00:23:50.800 align:middle line:84% interface for a deposited as opposed to a grown oxide 00:23:50.800 --> 00:23:53.280 align:middle line:90% are not nearly as good. 00:23:53.280 --> 00:23:54.280 align:middle line:90% But you still need them. 00:23:54.280 --> 00:23:56.500 align:middle line:84% You need them in the back end between metal layers, 00:23:56.500 --> 00:23:57.887 align:middle line:90% and you need them to do masking. 00:23:57.887 --> 00:23:58.970 align:middle line:90% So they're very important. 00:23:58.970 --> 00:24:02.020 align:middle line:84% But we're going to focus now on thermal oxides. 00:24:02.020 --> 00:24:04.600 align:middle line:84% And this just shows thickness and the type 00:24:04.600 --> 00:24:05.650 align:middle line:90% of oxides that are used. 00:24:05.650 --> 00:24:08.170 align:middle line:84% The thickest oxides are used in the field. 00:24:08.170 --> 00:24:10.790 align:middle line:84% The field surrounds the active device region. 00:24:10.790 --> 00:24:15.640 align:middle line:84% So it isolates individual devices. 00:24:15.640 --> 00:24:17.830 align:middle line:84% Thinner oxides, like 100 angstroms, 00:24:17.830 --> 00:24:19.660 align:middle line:84% are used for gates or pad oxides. 00:24:19.660 --> 00:24:23.180 align:middle line:84% In fact, I put a little arrow here and showed it going down. 00:24:23.180 --> 00:24:25.900 align:middle line:84% In fact, gate oxides today in the highest performance devices 00:24:25.900 --> 00:24:29.290 align:middle line:84% are quite a lot thinner than when this textbook was written. 00:24:29.290 --> 00:24:31.370 align:middle line:84% So they're actually more like 20 angstroms. 00:24:31.370 --> 00:24:34.990 align:middle line:84% So they're actually down in this range of tunneling oxides. 00:24:34.990 --> 00:24:37.000 align:middle line:84% And then finally, the very thinnest oxides 00:24:37.000 --> 00:24:39.640 align:middle line:84% are those chemical oxides that result from RCA clean, 00:24:39.640 --> 00:24:43.150 align:middle line:84% as we talked about last time, that are grown chemically 00:24:43.150 --> 00:24:47.220 align:middle line:90% rather than in a hot furnace. 00:24:47.220 --> 00:24:48.450 align:middle line:90% Let's go on to slide 16. 00:24:48.450 --> 00:24:51.720 align:middle line:84% This is just-- the quality didn't come out all that well. 00:24:51.720 --> 00:24:54.870 align:middle line:84% But this is a high-resolution cross-section transmission 00:24:54.870 --> 00:24:56.430 align:middle line:90% electron micrograph. 00:24:56.430 --> 00:25:00.900 align:middle line:84% And what we're looking at here on the top is polysilicon gate. 00:25:00.900 --> 00:25:04.710 align:middle line:84% This very thin layer right here, labeled gate oxide, 00:25:04.710 --> 00:25:07.800 align:middle line:84% is 1.2 nanometers thick, or 12 angstroms. 00:25:07.800 --> 00:25:12.000 align:middle line:84% That's the gate oxide that's been thermally grown. 00:25:12.000 --> 00:25:14.160 align:middle line:84% And at the bottom is the silicon. 00:25:14.160 --> 00:25:17.370 align:middle line:84% And in fact, it's the formation of these gate insulators 00:25:17.370 --> 00:25:20.280 align:middle line:84% that really is probably the most critical application 00:25:20.280 --> 00:25:23.520 align:middle line:84% of the process of silicon thermal oxidation in ICs today. 00:25:23.520 --> 00:25:25.290 align:middle line:84% We'll talk about thermal oxidation 00:25:25.290 --> 00:25:29.222 align:middle line:84% to grow thicker oxides, field oxide, et cetera, which is, 00:25:29.222 --> 00:25:30.180 align:middle line:90% of course, of interest. 00:25:30.180 --> 00:25:32.490 align:middle line:84% But in terms of the most critical application 00:25:32.490 --> 00:25:34.350 align:middle line:84% of thermal oxidation, it's really 00:25:34.350 --> 00:25:38.800 align:middle line:84% to try to form the gate insulator. 00:25:38.800 --> 00:25:42.540 align:middle line:84% However, remember we said that silicon dioxide is 00:25:42.540 --> 00:25:47.560 align:middle line:84% a perfect insulator if we go on to slide 17. 00:25:47.560 --> 00:25:49.030 align:middle line:84% There's some basic physics, though, 00:25:49.030 --> 00:25:52.630 align:middle line:84% that we can't get around, which has nothing 00:25:52.630 --> 00:25:53.780 align:middle line:90% to do with silicon dioxide. 00:25:53.780 --> 00:25:56.230 align:middle line:84% It's just the fact that there's something called 00:25:56.230 --> 00:25:58.270 align:middle line:90% quantum mechanical tunneling. 00:25:58.270 --> 00:26:01.270 align:middle line:84% If you have two regions, one on the left and one on the right, 00:26:01.270 --> 00:26:02.968 align:middle line:84% in the middle there's an insulator. 00:26:02.968 --> 00:26:04.510 align:middle line:84% There's a barrier, an energy barrier. 00:26:04.510 --> 00:26:05.927 align:middle line:84% If you've taken quantum mechanics, 00:26:05.927 --> 00:26:08.795 align:middle line:84% you know that when that energy barrier becomes thin enough, 00:26:08.795 --> 00:26:11.170 align:middle line:84% depending on its height, but when it becomes thin enough, 00:26:11.170 --> 00:26:13.510 align:middle line:84% there's a finite probability that electrons will find 00:26:13.510 --> 00:26:14.650 align:middle line:90% their way through that barrier. 00:26:14.650 --> 00:26:15.460 align:middle line:90% They will tunnel through. 00:26:15.460 --> 00:26:16.300 align:middle line:90% They don't go over. 00:26:16.300 --> 00:26:18.220 align:middle line:90% They go through. 00:26:18.220 --> 00:26:20.860 align:middle line:84% And this isn't a problem when you're thicker. 00:26:20.860 --> 00:26:23.470 align:middle line:84% But when you get down to this regime, for example, 00:26:23.470 --> 00:26:26.620 align:middle line:84% below about 30 angstroms, you can have direct tunneling 00:26:26.620 --> 00:26:30.130 align:middle line:84% right through from the gate electrode into the channel. 00:26:30.130 --> 00:26:31.930 align:middle line:84% And this constitutes a gate current, 00:26:31.930 --> 00:26:34.030 align:middle line:84% which, again, the whole idea of CMOS 00:26:34.030 --> 00:26:37.510 align:middle line:84% is that there's no current through the gate insulator. 00:26:37.510 --> 00:26:38.930 align:middle line:90% That's not true anymore. 00:26:38.930 --> 00:26:41.200 align:middle line:84% There is current flowing through these gate insulators 00:26:41.200 --> 00:26:43.040 align:middle line:90% because they're so thin. 00:26:43.040 --> 00:26:46.690 align:middle line:84% And this is just a diagram I took from an older paper now. 00:26:46.690 --> 00:26:49.890 align:middle line:84% But on the vertical axis is a log scale, 00:26:49.890 --> 00:26:51.320 align:middle line:90% shows the gate current density. 00:26:51.320 --> 00:26:53.737 align:middle line:84% So that's how much current is going through that insulator 00:26:53.737 --> 00:26:55.180 align:middle line:90% as a function of gate voltage. 00:26:55.180 --> 00:26:57.520 align:middle line:84% As I increase the gate voltage, of course, that goes up. 00:26:57.520 --> 00:26:59.840 align:middle line:84% And the different parameters shown here 00:26:59.840 --> 00:27:03.240 align:middle line:84% are the thickness of the SiO2 layer. 00:27:03.240 --> 00:27:06.050 align:middle line:84% So here in the bottom, it's showing for 36 angstroms, 00:27:06.050 --> 00:27:06.710 align:middle line:90% you see. 00:27:06.710 --> 00:27:09.170 align:middle line:84% And you can see it goes up exponentially. 00:27:09.170 --> 00:27:12.020 align:middle line:84% Below about 30 angstroms, we're going up exponentially. 00:27:12.020 --> 00:27:15.250 align:middle line:84% Here, I'm only going from 29 to 25 to 20 angstroms. 00:27:15.250 --> 00:27:19.160 align:middle line:84% And we're going up five orders of magnitude in just five 00:27:19.160 --> 00:27:21.280 align:middle line:84% angstroms in the amount of current 00:27:21.280 --> 00:27:22.530 align:middle line:90% that flows through that oxide. 00:27:22.530 --> 00:27:25.130 align:middle line:90% So it's an exponential process. 00:27:25.130 --> 00:27:27.740 align:middle line:84% It's very, very highly dependent on [INAUDIBLE].. 00:27:27.740 --> 00:27:31.498 align:middle line:84% As we go thinner, as we go here to, say, to 15 angstroms, 00:27:31.498 --> 00:27:34.040 align:middle line:84% really getting quite a bit of current tunneling through that. 00:27:34.040 --> 00:27:38.090 align:middle line:84% And that's a major issue in modern technology. 00:27:38.090 --> 00:27:40.670 align:middle line:84% So as a result, if we go on to slide 18, 00:27:40.670 --> 00:27:41.720 align:middle line:90% what are people doing? 00:27:41.720 --> 00:27:44.370 align:middle line:84% Well, you have a couple of options. 00:27:44.370 --> 00:27:46.760 align:middle line:90% One is don't scale the oxide. 00:27:46.760 --> 00:27:47.840 align:middle line:90% Don't make it so thin. 00:27:47.840 --> 00:27:48.750 align:middle line:90% Stop scaling it. 00:27:48.750 --> 00:27:51.072 align:middle line:84% Well, it turns out that's an issue because we need 00:27:51.072 --> 00:27:52.280 align:middle line:90% to get higher current drives. 00:27:52.280 --> 00:27:55.020 align:middle line:84% And one way we do that is to make the oxide thinner. 00:27:55.020 --> 00:27:56.480 align:middle line:84% So if we want higher performance, 00:27:56.480 --> 00:27:58.010 align:middle line:84% we want to make the oxide thinner. 00:27:58.010 --> 00:28:00.908 align:middle line:84% On the other hand, the gate is no longer a perfect insulator, 00:28:00.908 --> 00:28:01.700 align:middle line:90% the gate insulator. 00:28:01.700 --> 00:28:04.370 align:middle line:90% So there's another idea. 00:28:04.370 --> 00:28:06.710 align:middle line:84% To achieve higher current drive, well, we 00:28:06.710 --> 00:28:08.570 align:middle line:84% need to increase the capacitance. 00:28:08.570 --> 00:28:11.930 align:middle line:84% Usually, we just decrease t-ox in this formula 00:28:11.930 --> 00:28:13.262 align:middle line:90% shown on the slide. 00:28:13.262 --> 00:28:14.720 align:middle line:84% Well, the other thing you can do is 00:28:14.720 --> 00:28:17.830 align:middle line:84% increase the dielectric constant, k-ox 00:28:17.830 --> 00:28:20.270 align:middle line:90% of the insulator. 00:28:20.270 --> 00:28:22.725 align:middle line:84% So that's the name of the game people 00:28:22.725 --> 00:28:24.350 align:middle line:84% have been looking at and doing research 00:28:24.350 --> 00:28:26.060 align:middle line:90% for the last five or 10 years. 00:28:26.060 --> 00:28:28.820 align:middle line:84% What other insulators can we make 00:28:28.820 --> 00:28:31.980 align:middle line:84% besides SiO2 that have all the great properties, 00:28:31.980 --> 00:28:35.510 align:middle line:84% but they have a somewhat higher dielectric constant, k 00:28:35.510 --> 00:28:38.340 align:middle line:84% nu, which is shown in this formula here. 00:28:38.340 --> 00:28:41.300 align:middle line:84% So I try to increase the dielectric constant by-- 00:28:41.300 --> 00:28:44.060 align:middle line:84% one way is by adding a little nitrogen to the insulator. 00:28:44.060 --> 00:28:45.860 align:middle line:84% And people use oxide nitrides, or you might 00:28:45.860 --> 00:28:48.260 align:middle line:90% go to a whole new material. 00:28:48.260 --> 00:28:53.920 align:middle line:84% So what people refer to is for the same gate capacitance, 00:28:53.920 --> 00:28:56.980 align:middle line:84% what we typically do is we define an equivalent gate oxide 00:28:56.980 --> 00:29:00.370 align:middle line:84% thickness called t-ox equivalent according 00:29:00.370 --> 00:29:02.240 align:middle line:90% to the formula on the bottom. 00:29:02.240 --> 00:29:04.690 align:middle line:84% So what is t-ox equivalent? t-ox equivalent 00:29:04.690 --> 00:29:07.270 align:middle line:84% is equal to just k, the dielectric constant of silicon 00:29:07.270 --> 00:29:10.870 align:middle line:84% dioxide, which is at 3.9, divided 00:29:10.870 --> 00:29:14.440 align:middle line:84% by the dielectric constant of my new gate insulator, my high k. 00:29:14.440 --> 00:29:22.190 align:middle line:84% So let's say you went from 4 here to 8 times the thickness. 00:29:22.190 --> 00:29:26.290 align:middle line:84% So what that means is, basically, you can actually 00:29:26.290 --> 00:29:29.770 align:middle line:84% increase the thickness, go to a slightly higher thickness 00:29:29.770 --> 00:29:32.450 align:middle line:84% of the high k and get the same capacitance. 00:29:32.450 --> 00:29:35.780 align:middle line:84% So I'm upping k here, k nu, so I can increase thickness, 00:29:35.780 --> 00:29:37.780 align:middle line:84% which means your quantum mechanical tunneling is 00:29:37.780 --> 00:29:38.728 align:middle line:90% going to go down. 00:29:38.728 --> 00:29:41.020 align:middle line:84% You'll still get the high capacitance and the high gate 00:29:41.020 --> 00:29:41.520 align:middle line:90% drive. 00:29:41.520 --> 00:29:43.840 align:middle line:84% So it sounds like a perfect, from a mathematical point 00:29:43.840 --> 00:29:45.970 align:middle line:84% of view, physics will look at this, say, perfect. 00:29:45.970 --> 00:29:49.440 align:middle line:84% Increase k, increase t, everybody's happy. 00:29:49.440 --> 00:29:50.990 align:middle line:84% The problem is it's hard to come up 00:29:50.990 --> 00:29:53.870 align:middle line:84% with a material that has a higher dielectric constant that 00:29:53.870 --> 00:29:57.140 align:middle line:84% has all the perfect and ideal properties of the silicon 00:29:57.140 --> 00:29:58.410 align:middle line:90% silicon dioxide interface. 00:29:58.410 --> 00:30:00.590 align:middle line:84% In fact, such a material has not been discovered yet 00:30:00.590 --> 00:30:02.930 align:middle line:84% despite millions and millions of dollars that's 00:30:02.930 --> 00:30:06.350 align:middle line:84% been spent in trying to comb the periodic chart for such 00:30:06.350 --> 00:30:08.480 align:middle line:90% a material. 00:30:08.480 --> 00:30:12.770 align:middle line:84% So if we go on to slide 19, let's 00:30:12.770 --> 00:30:15.860 align:middle line:84% look at what some of the future projections 00:30:15.860 --> 00:30:18.830 align:middle line:84% are for the scaling of the gate insulator. 00:30:18.830 --> 00:30:21.290 align:middle line:84% And I should say gate insulator in this slide 00:30:21.290 --> 00:30:23.390 align:middle line:84% because we're not sure-- it won't be SiO2. 00:30:23.390 --> 00:30:26.150 align:middle line:84% It'll be an oxynitride with a higher 00:30:26.150 --> 00:30:30.650 align:middle line:84% k or maybe a new material that is yet to be figured out. 00:30:30.650 --> 00:30:33.320 align:middle line:84% And what I'm showing here, each column 00:30:33.320 --> 00:30:35.540 align:middle line:84% corresponds to a different year, just like we usually 00:30:35.540 --> 00:30:37.190 align:middle line:90% do on the ITRS. 00:30:37.190 --> 00:30:41.480 align:middle line:84% And what I wanted to focus on is a couple of interesting rows. 00:30:41.480 --> 00:30:45.240 align:middle line:84% This row right here, if I highlight this row, 00:30:45.240 --> 00:30:47.690 align:middle line:84% that says equivalent physical oxide thickness 00:30:47.690 --> 00:30:49.800 align:middle line:90% for microprocessor units. 00:30:49.800 --> 00:30:51.570 align:middle line:84% So this would be for high performance. 00:30:51.570 --> 00:30:53.300 align:middle line:90% So this is t-ox equivalent. 00:30:53.300 --> 00:30:56.420 align:middle line:84% In 2004, for high-performance devices, 00:30:56.420 --> 00:30:59.910 align:middle line:84% it's supposed to be about 1.2 nanometers, or 12 angstroms. 00:30:59.910 --> 00:31:01.370 align:middle line:84% So the high-performance technology 00:31:01.370 --> 00:31:03.690 align:middle line:84% is supposed to be about 90 nanometer technology, 12 00:31:03.690 --> 00:31:04.190 align:middle line:90% angstroms. 00:31:04.190 --> 00:31:05.180 align:middle line:90% That's how thin we are. 00:31:05.180 --> 00:31:06.500 align:middle line:90% And look at the current. 00:31:06.500 --> 00:31:09.810 align:middle line:84% The gate dielectric leakage, the maximum current tolerable 00:31:09.810 --> 00:31:14.340 align:middle line:84% in such a device is about 170 nanoamps per micron. 00:31:14.340 --> 00:31:17.520 align:middle line:90% So it's pretty high current. 00:31:17.520 --> 00:31:19.440 align:middle line:84% So that's for the highest performance device. 00:31:19.440 --> 00:31:21.240 align:middle line:84% And if we go out in that same row 00:31:21.240 --> 00:31:23.310 align:middle line:84% and we work to higher years, look what happens. 00:31:23.310 --> 00:31:25.500 align:middle line:90% By 2006, we end up in yellow. 00:31:25.500 --> 00:31:27.570 align:middle line:84% Yellow means people have an idea how to do it 00:31:27.570 --> 00:31:29.820 align:middle line:84% but maybe not totally manufacturable. 00:31:29.820 --> 00:31:30.990 align:middle line:90% But there's some ideas. 00:31:30.990 --> 00:31:34.900 align:middle line:84% And if we go beyond that, 2007, it's all red. 00:31:34.900 --> 00:31:37.600 align:middle line:84% So we want to achieve equivalent oxide thicknesses 00:31:37.600 --> 00:31:41.410 align:middle line:90% of 0.9, 0.8, and 0.8 by 2008. 00:31:41.410 --> 00:31:43.120 align:middle line:84% People just don't know-- don't really 00:31:43.120 --> 00:31:45.020 align:middle line:84% have the right material at this point. 00:31:45.020 --> 00:31:47.930 align:middle line:84% So there's a lot of research in this area. 00:31:47.930 --> 00:31:51.370 align:middle line:84% So what people do instead, we haven't figured that out yet, 00:31:51.370 --> 00:31:53.013 align:middle line:84% is they make different types to reduce 00:31:53.013 --> 00:31:55.180 align:middle line:84% the total amount of power that's burned on the chip. 00:31:55.180 --> 00:31:56.920 align:middle line:84% They make different types of devices. 00:31:56.920 --> 00:31:59.980 align:middle line:84% You have a very high-performance device that's quite fast. 00:31:59.980 --> 00:32:04.600 align:middle line:84% But it has reasonably thin physical gate oxide thickness. 00:32:04.600 --> 00:32:08.230 align:middle line:84% But then you can also, on the same chip, 00:32:08.230 --> 00:32:10.690 align:middle line:84% going down here in this region that I've circled down here 00:32:10.690 --> 00:32:13.870 align:middle line:84% where it says equivalent physical oxide thickness 00:32:13.870 --> 00:32:15.920 align:middle line:90% for low operating power. 00:32:15.920 --> 00:32:18.800 align:middle line:84% So this is a device burns less power. 00:32:18.800 --> 00:32:20.890 align:middle line:90% Notice, the oxide is thicker. 00:32:20.890 --> 00:32:23.980 align:middle line:84% It's 1.5 nanometers instead of 1.2. 00:32:23.980 --> 00:32:29.170 align:middle line:84% And the maximum tolerable gate leakage is on the order of 1, 00:32:29.170 --> 00:32:32.768 align:middle line:84% 1 nanoamp per micron instead of 170. 00:32:32.768 --> 00:32:35.060 align:middle line:84% So this device is supposed to be lower operating power. 00:32:35.060 --> 00:32:36.602 align:middle line:84% You make it a little thicker, and you 00:32:36.602 --> 00:32:38.917 align:middle line:90% change the VT to achieve that. 00:32:38.917 --> 00:32:40.500 align:middle line:84% Finally, if you go down to this column 00:32:40.500 --> 00:32:42.480 align:middle line:84% for the low standby power device, 00:32:42.480 --> 00:32:44.160 align:middle line:90% it has even thicker oxide. 00:32:44.160 --> 00:32:48.960 align:middle line:84% So on the same chip you might have a 2 nanometer, or 2.1-- 00:32:48.960 --> 00:32:50.310 align:middle line:90% 21 angstrom oxide. 00:32:50.310 --> 00:32:52.290 align:middle line:84% And the leakage current much lower 00:32:52.290 --> 00:32:55.680 align:middle line:84% through the gate for this device, about three picoamps 00:32:55.680 --> 00:32:56.580 align:middle line:90% per micron. 00:32:56.580 --> 00:32:59.880 align:middle line:84% So three orders of magnitude lower than for the other device 00:32:59.880 --> 00:33:02.130 align:middle line:84% and five orders of magnitude lower than for this. 00:33:02.130 --> 00:33:05.692 align:middle line:84% So quite a bit different on the same chip. 00:33:05.692 --> 00:33:07.650 align:middle line:84% This is the solution people are coming up with. 00:33:07.650 --> 00:33:11.880 align:middle line:84% Only the fastest devices have to have the very thinnest oxides. 00:33:11.880 --> 00:33:13.560 align:middle line:84% Nevertheless, in a couple of years, 00:33:13.560 --> 00:33:15.143 align:middle line:84% we will be hitting the red brick wall. 00:33:15.143 --> 00:33:18.120 align:middle line:84% So there's a lot of work to be done on alternative gate 00:33:18.120 --> 00:33:20.580 align:middle line:90% insulators. 00:33:20.580 --> 00:33:21.930 align:middle line:90% Let's go to the next slide. 00:33:21.930 --> 00:33:25.376 align:middle line:84% I've also taken this from the 2003 ITRS. 00:33:25.376 --> 00:33:30.330 align:middle line:84% It's a little bit complicated and not the most pretty plot 00:33:30.330 --> 00:33:31.080 align:middle line:90% in the world. 00:33:31.080 --> 00:33:34.110 align:middle line:84% But what it does show is a couple of interesting things. 00:33:34.110 --> 00:33:37.372 align:middle line:84% It's a plot of the leakage current density, JG, 00:33:37.372 --> 00:33:38.080 align:middle line:90% through the gate. 00:33:38.080 --> 00:33:40.205 align:middle line:84% So this is the current going through the gate oxide 00:33:40.205 --> 00:33:44.020 align:middle line:84% as a function of year on the left axis. 00:33:44.020 --> 00:33:49.622 align:middle line:84% And you can see that this red line represents the JG limit. 00:33:49.622 --> 00:33:51.330 align:middle line:84% This is what device designers would like. 00:33:51.330 --> 00:33:52.747 align:middle line:84% No higher than this amount of gate 00:33:52.747 --> 00:33:54.180 align:middle line:84% current is what they can tolerate 00:33:54.180 --> 00:33:56.280 align:middle line:90% for a high-performance device. 00:33:56.280 --> 00:33:59.610 align:middle line:84% This black curve is actually what people think, 00:33:59.610 --> 00:34:01.800 align:middle line:90% a dielectric called oxynitride. 00:34:01.800 --> 00:34:05.567 align:middle line:84% So this is not SiO2, but it's SiO2, SiO, 00:34:05.567 --> 00:34:07.650 align:middle line:84% where a certain amount of nitrogen has been added. 00:34:07.650 --> 00:34:10.440 align:middle line:84% And you can see the gate current on the left 00:34:10.440 --> 00:34:11.460 align:middle line:90% that is being projected. 00:34:11.460 --> 00:34:13.168 align:middle line:84% It looks something like this black curve. 00:34:13.168 --> 00:34:14.909 align:middle line:90% It's increasing exponentially. 00:34:14.909 --> 00:34:18.730 align:middle line:84% And somewhere around 2006, 2007, these curves cross. 00:34:18.730 --> 00:34:22.940 align:middle line:84% So beyond that oxynitride, you cannot stay on the same scaling 00:34:22.940 --> 00:34:23.440 align:middle line:90% curve. 00:34:23.440 --> 00:34:25.659 align:middle line:84% It won't satisfy the gate leakage current. 00:34:25.659 --> 00:34:27.639 align:middle line:84% It will leak too much through the gate. 00:34:27.639 --> 00:34:30.497 align:middle line:84% So beyond 2007, it's not clear if people can use oxynitride. 00:34:30.497 --> 00:34:32.330 align:middle line:84% So they have to come up with something else. 00:34:32.330 --> 00:34:34.600 align:middle line:84% This brown curve up here that's marked 00:34:34.600 --> 00:34:37.060 align:middle line:84% EOT, the equivalent oxide thickness, 00:34:37.060 --> 00:34:40.027 align:middle line:84% it's referenced to the right-hand vertical axis, which 00:34:40.027 --> 00:34:42.110 align:middle line:84% shows in angstroms the equivalent oxide thickness. 00:34:42.110 --> 00:34:44.750 align:middle line:84% So this is just exactly the data I showed you going down. 00:34:44.750 --> 00:34:47.739 align:middle line:84% So starting in 2003, we have an EOT 00:34:47.739 --> 00:34:49.870 align:middle line:90% of 13 angstroms on the right. 00:34:49.870 --> 00:34:54.489 align:middle line:84% And it's going down and down so that by 2007 or so here, 00:34:54.489 --> 00:34:57.740 align:middle line:84% the EOT is nominally 9 angstroms. 00:34:57.740 --> 00:35:00.400 align:middle line:84% So this is the way people would like to scale. 00:35:00.400 --> 00:35:02.680 align:middle line:84% But clearly, in the next few years, 00:35:02.680 --> 00:35:04.070 align:middle line:90% oxynitride is not going to work. 00:35:04.070 --> 00:35:06.610 align:middle line:84% You're going to need some other material, like hafnium 00:35:06.610 --> 00:35:11.900 align:middle line:84% oxide or one of the silicates that people are studying. 00:35:11.900 --> 00:35:13.520 align:middle line:90% And so one more-- 00:35:13.520 --> 00:35:15.140 align:middle line:84% go to the next slide, on slide 21, 00:35:15.140 --> 00:35:17.030 align:middle line:84% one more thing I just wanted to point out 00:35:17.030 --> 00:35:19.730 align:middle line:84% about alternative gate dielectrics is-- 00:35:19.730 --> 00:35:23.210 align:middle line:84% this I took from IEDM, International Electron Devices 00:35:23.210 --> 00:35:26.880 align:middle line:84% Meeting publication from about two years ago. 00:35:26.880 --> 00:35:28.460 align:middle line:90% There's been a lot of data. 00:35:28.460 --> 00:35:30.080 align:middle line:84% I don't mean that people have searched 00:35:30.080 --> 00:35:32.570 align:middle line:84% the entire periodic chart, but there's 00:35:32.570 --> 00:35:35.270 align:middle line:84% a lot of work that's been done by different people 00:35:35.270 --> 00:35:37.200 align:middle line:90% to study different materials. 00:35:37.200 --> 00:35:40.100 align:middle line:84% And this is just a way of summarizing them. 00:35:40.100 --> 00:35:42.500 align:middle line:84% One parameter people look at is the current density 00:35:42.500 --> 00:35:45.710 align:middle line:84% that leaks through the gate as a function of the EOT, equivalent 00:35:45.710 --> 00:35:46.920 align:middle line:90% oxide thickness. 00:35:46.920 --> 00:35:48.420 align:middle line:84% And I won't go through it in detail. 00:35:48.420 --> 00:35:50.950 align:middle line:84% But just the point is that a number of different materials 00:35:50.950 --> 00:35:51.950 align:middle line:90% are under consideration. 00:35:51.950 --> 00:35:53.570 align:middle line:84% I think right now the most popular 00:35:53.570 --> 00:35:59.360 align:middle line:84% are hafnium oxides and hafnium silicates, HfSiON. 00:35:59.360 --> 00:36:01.070 align:middle line:84% But this is an example of a topic, 00:36:01.070 --> 00:36:04.130 align:middle line:84% if you're interested in doing some of the library 00:36:04.130 --> 00:36:06.620 align:middle line:84% research for your final report, of a topic you 00:36:06.620 --> 00:36:10.040 align:middle line:84% might want to consider looking at is high k dielectrics. 00:36:10.040 --> 00:36:13.170 align:middle line:84% What some of their properties are and how they're formed 00:36:13.170 --> 00:36:16.870 align:middle line:84% is an interesting thing to study. 00:36:16.870 --> 00:36:18.540 align:middle line:84% So given that background information, 00:36:18.540 --> 00:36:22.450 align:middle line:84% I want to go on to page 22, slide 22 here. 00:36:22.450 --> 00:36:25.500 align:middle line:84% And even though I've just finished telling you 00:36:25.500 --> 00:36:29.970 align:middle line:84% that in the future pure SiO2 probably won't be used, 00:36:29.970 --> 00:36:32.737 align:middle line:90% it's still the most important-- 00:36:32.737 --> 00:36:34.320 align:middle line:84% there's nothing that's ever beaten it, 00:36:34.320 --> 00:36:35.340 align:middle line:90% so it's the standard. 00:36:35.340 --> 00:36:38.340 align:middle line:84% So we're going to study the process of thermal oxidation 00:36:38.340 --> 00:36:39.660 align:middle line:90% in this course. 00:36:39.660 --> 00:36:42.660 align:middle line:84% And this shows schematically, very schematically 00:36:42.660 --> 00:36:43.830 align:middle line:90% the basic process. 00:36:43.830 --> 00:36:47.070 align:middle line:84% We have an oxide that's growing at a high temperature 00:36:47.070 --> 00:36:48.330 align:middle line:90% on a silicon wafer. 00:36:48.330 --> 00:36:52.260 align:middle line:84% And what happens is an oxidant, such as oxygen or water, 00:36:52.260 --> 00:36:55.950 align:middle line:84% diffuses through that oxide, has to react at the surface. 00:36:55.950 --> 00:36:59.410 align:middle line:84% And new oxide is formed right at this interface right here. 00:36:59.410 --> 00:37:02.760 align:middle line:84% So we have a chemical reaction taking place at the interface. 00:37:02.760 --> 00:37:06.000 align:middle line:84% These are examples of the chemical reactions, silicon 00:37:06.000 --> 00:37:09.630 align:middle line:84% plus oxygen going to SiO2, or silicon plus moisture 00:37:09.630 --> 00:37:10.890 align:middle line:90% going to SiO2. 00:37:10.890 --> 00:37:14.250 align:middle line:84% These are the reactions that take place at that interface. 00:37:14.250 --> 00:37:16.630 align:middle line:84% And we'll spend, not this lecture, but probably 00:37:16.630 --> 00:37:20.800 align:middle line:84% next lecture talking in detail about mathematical models 00:37:20.800 --> 00:37:23.970 align:middle line:90% for that process. 00:37:23.970 --> 00:37:26.490 align:middle line:84% But rather, before we get into all the mathematical detail 00:37:26.490 --> 00:37:29.850 align:middle line:84% models and the Deal-Grove model, there 00:37:29.850 --> 00:37:31.950 align:middle line:84% are a few basic properties about the process 00:37:31.950 --> 00:37:36.450 align:middle line:84% we should think about, that, in fact, when the silicon 00:37:36.450 --> 00:37:39.810 align:middle line:84% surface is oxidized, there is a volume expansion associated 00:37:39.810 --> 00:37:40.480 align:middle line:90% with that. 00:37:40.480 --> 00:37:44.610 align:middle line:84% And this upper schematic picture shown here in the upper right 00:37:44.610 --> 00:37:48.990 align:middle line:84% is something I took from chapter 6, figure 6-39. 00:37:48.990 --> 00:37:50.970 align:middle line:84% What it represents is a cross-section 00:37:50.970 --> 00:37:54.330 align:middle line:84% in depth of the crystal, where these open circles here 00:37:54.330 --> 00:37:57.480 align:middle line:84% are meant to represent the silicon atoms in a lattice. 00:37:57.480 --> 00:38:00.990 align:middle line:84% This interface right here is meant to represent the silicon 00:38:00.990 --> 00:38:02.340 align:middle line:90% silicon dioxide interface. 00:38:02.340 --> 00:38:05.550 align:middle line:84% And this upper region here that has both black atoms, which 00:38:05.550 --> 00:38:08.640 align:middle line:84% are oxygen, and the white open circles, which are 00:38:08.640 --> 00:38:10.680 align:middle line:90% silicon in it, that's the SiO2. 00:38:10.680 --> 00:38:13.300 align:middle line:90% So here's the interface. 00:38:13.300 --> 00:38:15.990 align:middle line:84% So what it shows schematically is that, in order 00:38:15.990 --> 00:38:18.090 align:middle line:84% to oxidize this surface, what has to happen 00:38:18.090 --> 00:38:21.295 align:middle line:84% is the silicon bonds have to be broken because the silicon is 00:38:21.295 --> 00:38:22.170 align:middle line:90% bonded to each other. 00:38:22.170 --> 00:38:23.640 align:middle line:90% They have to be broken. 00:38:23.640 --> 00:38:25.500 align:middle line:84% Oxygen, these little black atoms, 00:38:25.500 --> 00:38:27.930 align:middle line:84% have to be inserted in between the silicon 00:38:27.930 --> 00:38:31.770 align:middle line:84% to form SiO bonds, such as shown here. 00:38:31.770 --> 00:38:34.662 align:middle line:84% And so the volume, that whole area 00:38:34.662 --> 00:38:36.870 align:middle line:84% has to expand to a certain extent because of the room 00:38:36.870 --> 00:38:38.770 align:middle line:90% taken up by the oxygen atoms. 00:38:38.770 --> 00:38:41.870 align:middle line:84% So there is a volume expansion associated with it. 00:38:41.870 --> 00:38:45.830 align:middle line:84% And the lower cartoons show schematically what 00:38:45.830 --> 00:38:47.558 align:middle line:90% that expansion looks like. 00:38:47.558 --> 00:38:49.850 align:middle line:84% So let's say I take a unit volume of a substrate that's 00:38:49.850 --> 00:38:54.560 align:middle line:84% a cube, that's 1 by 1 by 1 cube in all three dimensions, 00:38:54.560 --> 00:38:56.240 align:middle line:84% shown schematically here on the left. 00:38:56.240 --> 00:38:59.390 align:middle line:84% And just take that square that cube 00:38:59.390 --> 00:39:01.498 align:middle line:84% and expand it 30% in all three directions. 00:39:01.498 --> 00:39:02.540 align:middle line:90% So this is a bigger cube. 00:39:02.540 --> 00:39:05.520 align:middle line:84% This blue cube in the center is bigger by 30%. 00:39:05.520 --> 00:39:07.340 align:middle line:84% So this would be if it were unconstrained. 00:39:07.340 --> 00:39:09.950 align:middle line:84% There were no constraints, just in free space. 00:39:09.950 --> 00:39:12.710 align:middle line:84% But actually, the substrate restricts the expansion 00:39:12.710 --> 00:39:14.030 align:middle line:90% to one dimension, right? 00:39:14.030 --> 00:39:17.450 align:middle line:84% I can't just take the cube and expand it in the xy plane 00:39:17.450 --> 00:39:19.950 align:middle line:84% because I've got the substrate holding on to it. 00:39:19.950 --> 00:39:22.520 align:middle line:84% So in fact, the expansion only takes place 00:39:22.520 --> 00:39:24.840 align:middle line:90% in the vertical direction. 00:39:24.840 --> 00:39:28.100 align:middle line:90% So for every one unit consumed-- 00:39:28.100 --> 00:39:30.000 align:middle line:84% silicon consumed in the vertical direction, 00:39:30.000 --> 00:39:38.750 align:middle line:84% we have basically a height here of 2.2 units when it actually 00:39:38.750 --> 00:39:40.370 align:middle line:90% grows the SiO2. 00:39:40.370 --> 00:39:45.610 align:middle line:84% So the substrate restricts this to one dimension. 00:39:45.610 --> 00:39:46.720 align:middle line:90% So what does it look like? 00:39:46.720 --> 00:39:48.310 align:middle line:84% What does an oxide look like grown-- 00:39:48.310 --> 00:39:50.980 align:middle line:84% especially an oxide grown in a complicated two-dimensional 00:39:50.980 --> 00:39:51.480 align:middle line:90% fashion? 00:39:51.480 --> 00:39:53.685 align:middle line:84% We're going to spend a lecture talking about this. 00:39:53.685 --> 00:39:55.500 align:middle line:90% But this is a pretty picture. 00:39:55.500 --> 00:39:57.420 align:middle line:84% It's a scanning electron micrograph 00:39:57.420 --> 00:39:58.920 align:middle line:90% of something called LOCOS. 00:39:58.920 --> 00:40:00.960 align:middle line:84% I think the first lecture we talk 00:40:00.960 --> 00:40:03.330 align:middle line:84% about local oxidation of silicon, 00:40:03.330 --> 00:40:06.930 align:middle line:84% where we would put a nitride mask down on the chip 00:40:06.930 --> 00:40:08.610 align:middle line:84% where we don't want the oxide to grow. 00:40:08.610 --> 00:40:10.950 align:middle line:84% And everywhere outside the nitride mask 00:40:10.950 --> 00:40:13.450 align:middle line:84% is where the field oxidation takes place. 00:40:13.450 --> 00:40:16.230 align:middle line:84% So this is an actual example of field oxidation 00:40:16.230 --> 00:40:19.740 align:middle line:84% that's taking place and what it looks like. 00:40:19.740 --> 00:40:24.330 align:middle line:84% This rectangular bar marked the location of the silicon nitride 00:40:24.330 --> 00:40:26.280 align:middle line:84% mask, that's the region on the chip where 00:40:26.280 --> 00:40:30.900 align:middle line:84% the nitride was there protecting the surface against oxidation. 00:40:30.900 --> 00:40:34.900 align:middle line:84% The gray region down here is the silicon substrate. 00:40:34.900 --> 00:40:39.420 align:middle line:84% This region marked SiO2, you can see the volume expansion. 00:40:39.420 --> 00:40:42.960 align:middle line:84% It's expanded compared to-- the original silicon surface is 00:40:42.960 --> 00:40:46.770 align:middle line:84% shown by this white dashed line because we know oxidation 00:40:46.770 --> 00:40:50.640 align:middle line:84% involves the volume expansion by about 2.2x. 00:40:50.640 --> 00:40:54.780 align:middle line:84% So this occupies about twice, 2.2 times, 00:40:54.780 --> 00:40:56.430 align:middle line:84% the volume of the silicon from here 00:40:56.430 --> 00:40:57.905 align:middle line:84% to here that was actually consumed. 00:40:57.905 --> 00:40:59.280 align:middle line:84% So you're consuming the substrate 00:40:59.280 --> 00:41:00.750 align:middle line:90% when you're doing this. 00:41:00.750 --> 00:41:03.265 align:middle line:84% Interestingly, though, in two dimensions 00:41:03.265 --> 00:41:04.890 align:middle line:84% and these three-dimensional structures, 00:41:04.890 --> 00:41:07.573 align:middle line:84% stress plays an important dominant role. 00:41:07.573 --> 00:41:09.490 align:middle line:84% We're going to talk about that in this course. 00:41:09.490 --> 00:41:11.320 align:middle line:84% So look at this little bird's beak. 00:41:11.320 --> 00:41:13.140 align:middle line:84% They call this a bird's beak that's 00:41:13.140 --> 00:41:15.930 align:middle line:84% squeezing out like this because it looks like a bird's beak. 00:41:15.930 --> 00:41:18.300 align:middle line:84% And sometimes they call this the bird's head. 00:41:18.300 --> 00:41:20.160 align:middle line:84% You have to have a good imagination in order 00:41:20.160 --> 00:41:21.060 align:middle line:90% to understand that. 00:41:21.060 --> 00:41:22.770 align:middle line:84% But the bird's beak shape is actually 00:41:22.770 --> 00:41:24.840 align:middle line:84% due to lateral oxidation, some of the oxidant 00:41:24.840 --> 00:41:27.810 align:middle line:84% diffusing in here under the nitride mask. 00:41:27.810 --> 00:41:30.210 align:middle line:84% And also, its shape is controlled by the nitride mask 00:41:30.210 --> 00:41:31.163 align:middle line:90% pushing down on it. 00:41:31.163 --> 00:41:32.580 align:middle line:84% So the oxide is trying to grow it. 00:41:32.580 --> 00:41:33.830 align:middle line:90% Has to push up on the nitride. 00:41:33.830 --> 00:41:35.100 align:middle line:90% That takes some force. 00:41:35.100 --> 00:41:38.310 align:middle line:84% So this whole shape and the amount of lateral encroachment 00:41:38.310 --> 00:41:40.350 align:middle line:84% is very much dependent on the stress effects. 00:41:40.350 --> 00:41:45.500 align:middle line:84% And we'll talk about that in the next couple of lectures. 00:41:45.500 --> 00:41:49.825 align:middle line:84% So complex shapes can be formed by local oxidation. 00:41:49.825 --> 00:41:52.990 align:middle line:90% We'll go on to slide 25. 00:41:52.990 --> 00:41:55.990 align:middle line:84% This is a very high-resolution transmission electron 00:41:55.990 --> 00:41:56.590 align:middle line:90% micrograph. 00:41:56.590 --> 00:41:58.257 align:middle line:84% And again, on your handouts, it probably 00:41:58.257 --> 00:41:59.440 align:middle line:90% doesn't look very good. 00:41:59.440 --> 00:42:01.990 align:middle line:84% But all those little dots, you see them in rows, 00:42:01.990 --> 00:42:05.170 align:middle line:84% correspond to pairs of atoms in the silicon lattice. 00:42:05.170 --> 00:42:06.667 align:middle line:90% So these are planes here. 00:42:06.667 --> 00:42:08.500 align:middle line:84% And they're all very regular because silicon 00:42:08.500 --> 00:42:10.375 align:middle line:84% is a crystal material, right, single crystal, 00:42:10.375 --> 00:42:11.380 align:middle line:90% perfect structure. 00:42:11.380 --> 00:42:13.480 align:middle line:90% SiO2, however, is amorphous. 00:42:13.480 --> 00:42:14.782 align:middle line:90% It has no long-range order. 00:42:14.782 --> 00:42:16.990 align:middle line:84% It has some short-range order but no long-range order 00:42:16.990 --> 00:42:18.920 align:middle line:84% even though it's growing on a single crystal. 00:42:18.920 --> 00:42:22.090 align:middle line:84% So this is an amorphous material even though it's growing 00:42:22.090 --> 00:42:24.680 align:middle line:90% on a single crystal substrate. 00:42:24.680 --> 00:42:27.250 align:middle line:90% It's a glass. 00:42:27.250 --> 00:42:32.110 align:middle line:84% If we go on to slide 26, there are no crystalline forms 00:42:32.110 --> 00:42:36.730 align:middle line:84% of SiO2 that match the lattice size of silicon. 00:42:36.730 --> 00:42:38.620 align:middle line:84% There is a little bit of short-range order 00:42:38.620 --> 00:42:41.320 align:middle line:90% in thermally-grown SiO2. 00:42:41.320 --> 00:42:45.110 align:middle line:84% And in fact, if you look at this schematic on the left, 00:42:45.110 --> 00:42:49.990 align:middle line:84% this shows that there's a tetrahedral type of structure 00:42:49.990 --> 00:42:50.600 align:middle line:90% on the left. 00:42:50.600 --> 00:42:54.250 align:middle line:84% You see a silicon atom in the center and four oxygen atoms 00:42:54.250 --> 00:42:59.200 align:middle line:84% bonded to it nearby forming some sort of tetrahedron. 00:42:59.200 --> 00:43:01.990 align:middle line:84% These tetrahedra, though, are not actually bound 00:43:01.990 --> 00:43:03.160 align:middle line:90% in any real structure. 00:43:03.160 --> 00:43:06.130 align:middle line:84% They may form a ring structure a little bit 00:43:06.130 --> 00:43:07.300 align:middle line:90% by sharing oxygen atoms. 00:43:07.300 --> 00:43:08.842 align:middle line:84% But you notice, this lattice does not 00:43:08.842 --> 00:43:13.850 align:middle line:84% repeat itself throughout space to form some single crystal. 00:43:13.850 --> 00:43:15.850 align:middle line:84% So there is short-range order in the tetrahedra. 00:43:15.850 --> 00:43:19.390 align:middle line:84% But it is an open network, and it 00:43:19.390 --> 00:43:22.860 align:middle line:90% doesn't have long-range order. 00:43:22.860 --> 00:43:26.120 align:middle line:84% So let's go on to slide 27 and just 00:43:26.120 --> 00:43:28.220 align:middle line:84% talk for a few minutes about stress and oxide. 00:43:28.220 --> 00:43:30.830 align:middle line:84% We're going to spend a lot more time talking about it 00:43:30.830 --> 00:43:33.260 align:middle line:84% in the next lecture but just to make the point 00:43:33.260 --> 00:43:38.150 align:middle line:84% that when oxide layers grow on silicon, 00:43:38.150 --> 00:43:40.310 align:middle line:84% they're under compressive stress. 00:43:40.310 --> 00:43:42.630 align:middle line:84% And it's for a couple of different reasons. 00:43:42.630 --> 00:43:45.710 align:middle line:84% But basically, because the growing oxide 00:43:45.710 --> 00:43:48.140 align:middle line:84% is constrained by the interface that it's growing on 00:43:48.140 --> 00:43:50.390 align:middle line:84% it's constrained by the substrate 00:43:50.390 --> 00:43:51.960 align:middle line:90% this interface right down here. 00:43:51.960 --> 00:43:56.840 align:middle line:84% So it can only expand upward And this gives you 00:43:56.840 --> 00:44:00.200 align:middle line:84% an idea, 5 10 the ninth dynes per square centimeter 00:44:00.200 --> 00:44:01.940 align:middle line:84% for the amount of compressive stress 00:44:01.940 --> 00:44:06.330 align:middle line:84% that this oxide is feeling when it's grown on the wafer. 00:44:06.330 --> 00:44:08.270 align:middle line:84% Now, at high enough temperatures, 00:44:08.270 --> 00:44:11.270 align:middle line:84% say, if you put it in the furnace to oxidize above 1,000 00:44:11.270 --> 00:44:14.180 align:middle line:84% or anneal it, the oxide can relieve some of the stress. 00:44:14.180 --> 00:44:18.310 align:middle line:84% The oxide can actually flow a little bit by a viscous flow. 00:44:18.310 --> 00:44:20.810 align:middle line:84% But that's only at a high enough temperature where the glass 00:44:20.810 --> 00:44:22.430 align:middle line:90% starts to soften a little bit. 00:44:22.430 --> 00:44:26.380 align:middle line:84% At lower temperatures, it's just too viscous in order for it 00:44:26.380 --> 00:44:26.880 align:middle line:90% to flow. 00:44:26.880 --> 00:44:29.460 align:middle line:90% So the stress stays built in. 00:44:29.460 --> 00:44:31.940 align:middle line:84% So there's these intrinsic stresses just because 00:44:31.940 --> 00:44:33.830 align:middle line:90% of the volume expansion. 00:44:33.830 --> 00:44:36.530 align:middle line:84% There's also differences in the thermal expansion coefficients 00:44:36.530 --> 00:44:37.610 align:middle line:90% of SiO2 and silicon. 00:44:37.610 --> 00:44:40.880 align:middle line:84% And that leads to a certain amount of intrinsic stress. 00:44:40.880 --> 00:44:43.950 align:middle line:84% Both of these effects are going to put the silicon in tension. 00:44:43.950 --> 00:44:46.010 align:middle line:84% So the silicon wafer at the surface 00:44:46.010 --> 00:44:48.740 align:middle line:84% is sort of being pulled by the oxide, which 00:44:48.740 --> 00:44:50.130 align:middle line:90% is being compressed. 00:44:50.130 --> 00:44:53.870 align:middle line:84% So if you were to grow an oxide on both sides of a wafer 00:44:53.870 --> 00:44:56.030 align:middle line:84% and strip it off the back, the wafer 00:44:56.030 --> 00:44:59.353 align:middle line:84% would actually be slightly bowed, very so slightly, 00:44:59.353 --> 00:45:00.770 align:middle line:84% nothing that you could see by eye. 00:45:00.770 --> 00:45:03.200 align:middle line:84% But you can actually measure it in a special laser 00:45:03.200 --> 00:45:05.520 align:middle line:84% apparatus called a wafer curvature measurement. 00:45:05.520 --> 00:45:09.590 align:middle line:84% So there are ways of actually detecting the amount of stress 00:45:09.590 --> 00:45:11.780 align:middle line:84% just by measuring the curvature of the wafer when 00:45:11.780 --> 00:45:13.340 align:middle line:84% you remove the oxide from one side 00:45:13.340 --> 00:45:15.342 align:middle line:90% and leave it on the other side. 00:45:15.342 --> 00:45:17.300 align:middle line:84% That's how people determine some of the stress. 00:45:17.300 --> 00:45:20.020 align:middle line:90% 00:45:20.020 --> 00:45:21.130 align:middle line:90% Let's go on to slide 28. 00:45:21.130 --> 00:45:23.750 align:middle line:84% This is a very schematic, and I apologize for this cartoon. 00:45:23.750 --> 00:45:25.750 align:middle line:84% If you've ever actually been in a fab clean room 00:45:25.750 --> 00:45:28.030 align:middle line:84% and you know what an oxidation furnace looks like, 00:45:28.030 --> 00:45:29.620 align:middle line:84% it doesn't look exactly like this. 00:45:29.620 --> 00:45:31.840 align:middle line:90% But it's an artist's conception. 00:45:31.840 --> 00:45:36.340 align:middle line:84% The equipment is actually relatively simple in concept. 00:45:36.340 --> 00:45:39.040 align:middle line:84% It's essentially a quartz tube, which is shown here 00:45:39.040 --> 00:45:40.640 align:middle line:90% by this black line. 00:45:40.640 --> 00:45:41.560 align:middle line:90% So it's a cylinder. 00:45:41.560 --> 00:45:44.230 align:middle line:84% It's a long quartz tube into which 00:45:44.230 --> 00:45:46.930 align:middle line:84% one can push a boat of a quartz carrier 00:45:46.930 --> 00:45:49.360 align:middle line:90% with wafers standing up in it. 00:45:49.360 --> 00:45:51.990 align:middle line:84% So all these wafers are put vertically 00:45:51.990 --> 00:45:53.490 align:middle line:84% in the boat in little slots, and you 00:45:53.490 --> 00:45:54.510 align:middle line:84% have a whole bunch of them sitting 00:45:54.510 --> 00:45:56.550 align:middle line:84% right next to each other, within a few millimeter of each other. 00:45:56.550 --> 00:45:58.008 align:middle line:84% You have the whole stack of wafers. 00:45:58.008 --> 00:46:02.310 align:middle line:84% So you can oxidize 25 or 50 or 100 wafers in a furnace 00:46:02.310 --> 00:46:04.380 align:middle line:90% at a time. 00:46:04.380 --> 00:46:06.030 align:middle line:84% You essentially just flow in oxidants, 00:46:06.030 --> 00:46:09.390 align:middle line:84% like oxygen and hydrogen, to create water vapor 00:46:09.390 --> 00:46:10.050 align:middle line:90% and moisture. 00:46:10.050 --> 00:46:12.240 align:middle line:84% You heat the whole thing by resistive [? heater. ?] 00:46:12.240 --> 00:46:15.390 align:middle line:84% So the quartz gets very hot, and the wafers get very hot. 00:46:15.390 --> 00:46:17.700 align:middle line:84% And then the oxidant goes out the back side. 00:46:17.700 --> 00:46:19.500 align:middle line:90% So it's relatively simple. 00:46:19.500 --> 00:46:22.680 align:middle line:84% In practice, though, there are some modern furnaces 00:46:22.680 --> 00:46:26.170 align:middle line:84% or oxidation systems that are a little bit different. 00:46:26.170 --> 00:46:28.260 align:middle line:84% They also have vertical furnaces these days, 00:46:28.260 --> 00:46:33.480 align:middle line:84% where the wafers sit vertically like this to help avoid 00:46:33.480 --> 00:46:35.490 align:middle line:90% warpage and things like that. 00:46:35.490 --> 00:46:37.890 align:middle line:84% There are rapid thermal oxidation systems 00:46:37.890 --> 00:46:40.800 align:middle line:90% and fast-ramped furnaces. 00:46:40.800 --> 00:46:45.140 align:middle line:84% In fact, I think on the next slide, slide 29, 00:46:45.140 --> 00:46:49.190 align:middle line:84% there is an example nothing like an oxidation furnace, 00:46:49.190 --> 00:46:54.320 align:middle line:84% but a system in which people do grow SiO2 and oxynitrides 00:46:54.320 --> 00:46:56.660 align:middle line:84% by a process called rapid thermal processing 00:46:56.660 --> 00:46:58.400 align:middle line:90% or rapid thermal oxidation. 00:46:58.400 --> 00:46:59.847 align:middle line:90% This is a photograph. 00:46:59.847 --> 00:47:01.680 align:middle line:84% I took it off the Applied Materials website. 00:47:01.680 --> 00:47:04.010 align:middle line:84% It's a little bit fuzzy because I blew it up. 00:47:04.010 --> 00:47:06.350 align:middle line:84% But this is a particular system called 00:47:06.350 --> 00:47:08.390 align:middle line:84% the Centura rapid thermal processing system. 00:47:08.390 --> 00:47:10.200 align:middle line:90% And it's a single wafer system. 00:47:10.200 --> 00:47:11.870 align:middle line:84% And you can see it's like a clamshell. 00:47:11.870 --> 00:47:14.510 align:middle line:84% The clamshell has been opened up. 00:47:14.510 --> 00:47:16.430 align:middle line:90% A wafer goes in, sits here. 00:47:16.430 --> 00:47:19.770 align:middle line:84% In fact, you can see where a wafer might sit. 00:47:19.770 --> 00:47:22.992 align:middle line:84% The clamshell would be down, and the infrared lamps-- 00:47:22.992 --> 00:47:25.200 align:middle line:84% you can see them glowing-- the infrared lamps heat up 00:47:25.200 --> 00:47:30.560 align:middle line:84% very rapidly, heat up the wafer within a few seconds. 00:47:30.560 --> 00:47:33.060 align:middle line:84% It's in a little quartz chamber where there's flowing oxygen 00:47:33.060 --> 00:47:34.020 align:middle line:90% or whatever. 00:47:34.020 --> 00:47:34.650 align:middle line:90% Heats it up. 00:47:34.650 --> 00:47:37.630 align:middle line:84% The reaction takes place and then cools it down. 00:47:37.630 --> 00:47:41.430 align:middle line:84% So this is a so-called single wafer rapid thermal processing, 00:47:41.430 --> 00:47:42.750 align:middle line:90% a method of making oxide. 00:47:42.750 --> 00:47:44.640 align:middle line:84% So you don't have to necessarily do 00:47:44.640 --> 00:47:47.230 align:middle line:84% gate oxides in an old fashioned furnace anymore. 00:47:47.230 --> 00:47:51.690 align:middle line:84% You can also do it in these new pieces of equipment. 00:47:51.690 --> 00:47:55.320 align:middle line:84% So I want to go on to slide 30 now. 00:47:55.320 --> 00:47:58.800 align:middle line:84% And I want to talk a little bit now 00:47:58.800 --> 00:48:01.290 align:middle line:84% about a little more theoretical things, which 00:48:01.290 --> 00:48:04.933 align:middle line:84% is to talk about the silicon silicon dioxide interface. 00:48:04.933 --> 00:48:06.600 align:middle line:84% This is important from the point of view 00:48:06.600 --> 00:48:09.990 align:middle line:84% of understanding electrically the quality of what 00:48:09.990 --> 00:48:11.610 align:middle line:90% you've produced. 00:48:11.610 --> 00:48:14.220 align:middle line:84% And it does feed into how we actually 00:48:14.220 --> 00:48:18.510 align:middle line:84% do the processes to grow thermal oxides. 00:48:18.510 --> 00:48:20.787 align:middle line:84% So this is a cross section here in pink. 00:48:20.787 --> 00:48:22.620 align:middle line:84% On the bottom is supposed to be the silicon, 00:48:22.620 --> 00:48:25.680 align:middle line:84% and the white region is the SiO2. 00:48:25.680 --> 00:48:28.680 align:middle line:84% And back in the 1980s, Bruce Thiel 00:48:28.680 --> 00:48:32.850 align:middle line:84% suggested that this picture of electrical defects that 00:48:32.850 --> 00:48:35.250 align:middle line:84% might exist, the evidence of which 00:48:35.250 --> 00:48:38.260 align:middle line:84% had been observed experimentally. 00:48:38.260 --> 00:48:41.670 align:middle line:84% And so there are different types of charges 00:48:41.670 --> 00:48:45.720 align:middle line:84% here that we're going to talk about, Qit, Qf, Qot, and Qm. 00:48:45.720 --> 00:48:48.240 align:middle line:84% On the next slide, I'll define those. 00:48:48.240 --> 00:48:50.710 align:middle line:84% One point to make, though, is that to first order 00:48:50.710 --> 00:48:52.152 align:middle line:90% the interface is perfect. 00:48:52.152 --> 00:48:54.610 align:middle line:84% We're going to focus on defects because we're all neurotic, 00:48:54.610 --> 00:48:56.260 align:middle line:90% and we're interested in defects. 00:48:56.260 --> 00:48:58.810 align:middle line:84% But just to stand back for a moment and think, 00:48:58.810 --> 00:49:04.230 align:middle line:84% it's perfect in that only one in about 10 to the fifth atoms 00:49:04.230 --> 00:49:05.077 align:middle line:90% has a defect. 00:49:05.077 --> 00:49:06.660 align:middle line:84% So if you were to go onto that surface 00:49:06.660 --> 00:49:08.760 align:middle line:84% and walk around at the interface and just 00:49:08.760 --> 00:49:11.070 align:middle line:84% count the number of atoms that where 00:49:11.070 --> 00:49:13.080 align:middle line:84% there was an unsatisfied or broken bond, 00:49:13.080 --> 00:49:15.880 align:middle line:84% you would count 1 for every 10 to the fifth. 00:49:15.880 --> 00:49:18.090 align:middle line:90% So that's pretty darn perfect. 00:49:18.090 --> 00:49:19.048 align:middle line:90% It's still not perfect. 00:49:19.048 --> 00:49:20.882 align:middle line:84% It's not as perfect as we'd like to make it. 00:49:20.882 --> 00:49:21.960 align:middle line:90% We always want better. 00:49:21.960 --> 00:49:23.610 align:middle line:84% But it's pretty darn good compared 00:49:23.610 --> 00:49:26.740 align:middle line:84% to any other semiconductor insulator system. 00:49:26.740 --> 00:49:29.910 align:middle line:84% So as a result, the defect densities 00:49:29.910 --> 00:49:32.460 align:middle line:84% at the interface we measure are in this range 00:49:32.460 --> 00:49:35.640 align:middle line:84% per square centimeter, 10 to the ninth to 10 00:49:35.640 --> 00:49:39.690 align:middle line:84% to the 11th defects on a per square centimeter basis. 00:49:39.690 --> 00:49:41.850 align:middle line:84% Remember, the surface atom density is about 10 00:49:41.850 --> 00:49:42.420 align:middle line:90% to the 15th. 00:49:42.420 --> 00:49:45.480 align:middle line:84% So again, it's about 1 in 10 to the 5, something like that. 00:49:45.480 --> 00:49:48.380 align:middle line:90% 00:49:48.380 --> 00:49:52.180 align:middle line:84% So let's go on to slide 31 and talk 00:49:52.180 --> 00:49:55.190 align:middle line:90% about these different charges. 00:49:55.190 --> 00:49:58.070 align:middle line:84% So first, there's basically these four types 00:49:58.070 --> 00:50:01.490 align:middle line:84% of charges associated with the insulator itself 00:50:01.490 --> 00:50:04.340 align:middle line:84% or the semiconductor insulator interface. 00:50:04.340 --> 00:50:08.000 align:middle line:84% The first one you may have heard of called Qf, or Q sub f, 00:50:08.000 --> 00:50:09.860 align:middle line:90% it's the fixed oxide charge. 00:50:09.860 --> 00:50:14.570 align:middle line:84% It is represented here by these little positive plus signs. 00:50:14.570 --> 00:50:17.090 align:middle line:84% It's a sheet of positive charge, and it's 00:50:17.090 --> 00:50:20.580 align:middle line:84% supposed to be within about 2 nanometers of the interface. 00:50:20.580 --> 00:50:22.262 align:middle line:84% So you might ask me, well, what happens 00:50:22.262 --> 00:50:24.470 align:middle line:84% when you make an oxide that's less than 2 nanometers? 00:50:24.470 --> 00:50:27.080 align:middle line:90% Well, that's a good question. 00:50:27.080 --> 00:50:28.740 align:middle line:84% What happens to the fixed charge? 00:50:28.740 --> 00:50:30.590 align:middle line:84% So it's not it's not exactly clear. 00:50:30.590 --> 00:50:35.480 align:middle line:84% But the fixed charge, it has a very unique characteristic 00:50:35.480 --> 00:50:37.670 align:middle line:90% to it, and it's always positive. 00:50:37.670 --> 00:50:40.190 align:middle line:84% And we'll talk about why people think that is. 00:50:40.190 --> 00:50:42.800 align:middle line:84% And you can reduce its concentration 00:50:42.800 --> 00:50:46.100 align:middle line:84% by heat treating the oxide after it's grown, heat treating it 00:50:46.100 --> 00:50:48.930 align:middle line:90% at a certain temperature. 00:50:48.930 --> 00:50:52.550 align:middle line:84% So we put into our recipes after we grow an oxide very often 00:50:52.550 --> 00:50:57.270 align:middle line:84% an anneal to reduce the fixed charge to as low as it can go. 00:50:57.270 --> 00:50:57.870 align:middle line:90% That's Qf. 00:50:57.870 --> 00:51:00.730 align:middle line:90% It's always positive. 00:51:00.730 --> 00:51:01.730 align:middle line:90% And it's a fixed number. 00:51:01.730 --> 00:51:03.397 align:middle line:84% It's fixed in the sense of, after you've 00:51:03.397 --> 00:51:05.570 align:middle line:84% done the process, you have the chip, 00:51:05.570 --> 00:51:09.420 align:middle line:84% depending on how you bias the chip, it doesn't change. 00:51:09.420 --> 00:51:12.230 align:middle line:90% That's not true of Qit. 00:51:12.230 --> 00:51:15.860 align:middle line:84% Interface trap charge, Qit, can be either, 00:51:15.860 --> 00:51:18.380 align:middle line:84% as pictured here by the little x, x meaning we 00:51:18.380 --> 00:51:20.780 align:middle line:84% don't know its sign, it can be either plus, 00:51:20.780 --> 00:51:22.650 align:middle line:90% it can be neutral or negative. 00:51:22.650 --> 00:51:25.970 align:middle line:84% And it may change during normal device operation 00:51:25.970 --> 00:51:29.870 align:middle line:84% as you change the bias because electrons or holes can 00:51:29.870 --> 00:51:31.860 align:middle line:90% be captured. 00:51:31.860 --> 00:51:36.770 align:middle line:84% So it has a behavior similar to bulk deep levels, the way 00:51:36.770 --> 00:51:39.720 align:middle line:84% iron and nickel can trap electrons and holes, 00:51:39.720 --> 00:51:42.170 align:middle line:84% the way that's discussed in chapters 1 and 4. 00:51:42.170 --> 00:51:45.270 align:middle line:90% So Qit is not fixed. 00:51:45.270 --> 00:51:47.780 align:middle line:84% It changes with time depending on how we bias. 00:51:47.780 --> 00:51:52.170 align:middle line:84% And we'll talk about how it changes with bias. 00:51:52.170 --> 00:51:55.360 align:middle line:90% Third type, QM, mobile charge-- 00:51:55.360 --> 00:51:59.520 align:middle line:84% well, as the name suggests-- these ions are mobile. 00:51:59.520 --> 00:52:02.550 align:middle line:84% That is, they can move in the oxide from the top surface 00:52:02.550 --> 00:52:04.680 align:middle line:84% to the bottom surface, the bottom interface. 00:52:04.680 --> 00:52:07.770 align:middle line:84% They can move up and down upon application of electric field. 00:52:07.770 --> 00:52:08.980 align:middle line:90% They're ionized. 00:52:08.980 --> 00:52:13.025 align:middle line:84% So if you heat the oxide, say, to 100, 200, 300 degrees 00:52:13.025 --> 00:52:14.400 align:middle line:84% and you put electric field on it, 00:52:14.400 --> 00:52:15.900 align:middle line:84% you can actually cause these sodium 00:52:15.900 --> 00:52:19.860 align:middle line:84% and potassium to move from here to here and back. 00:52:19.860 --> 00:52:22.800 align:middle line:84% Mobile ions are less important in modern manufacturing 00:52:22.800 --> 00:52:24.540 align:middle line:84% because clean rooms are so clean. 00:52:24.540 --> 00:52:27.150 align:middle line:84% People are never allowed to touch anything. 00:52:27.150 --> 00:52:29.720 align:middle line:84% Unless you're in a research lab where people are sloppy 00:52:29.720 --> 00:52:31.470 align:middle line:84% or something like that, you don't really-- 00:52:31.470 --> 00:52:33.150 align:middle line:84% you might worry about mobile ions. 00:52:33.150 --> 00:52:35.900 align:middle line:84% We do worry about them, of course, because here at MIT 00:52:35.900 --> 00:52:36.900 align:middle line:90% we're in a research fab. 00:52:36.900 --> 00:52:38.020 align:middle line:90% We have a lot of students. 00:52:38.020 --> 00:52:41.295 align:middle line:84% But in a real fab, people, in manufacturing sense, 00:52:41.295 --> 00:52:43.170 align:middle line:84% they're very, very careful about mobile ions. 00:52:43.170 --> 00:52:45.900 align:middle line:84% So usually, the mobile ion constant charge 00:52:45.900 --> 00:52:50.160 align:middle line:84% is relatively small unless you've made a mistake. 00:52:50.160 --> 00:52:53.520 align:middle line:84% Fourth type, Qot, oxide trap charge, this 00:52:53.520 --> 00:52:55.980 align:middle line:90% is represented right here. 00:52:55.980 --> 00:52:57.285 align:middle line:90% It's in the bulk of the oxide. 00:52:57.285 --> 00:52:58.410 align:middle line:90% Notice, it's drawn up here. 00:52:58.410 --> 00:53:01.110 align:middle line:90% It's not at the interface. 00:53:01.110 --> 00:53:04.080 align:middle line:84% It may be created by other processes 00:53:04.080 --> 00:53:07.020 align:middle line:84% that the oxide was subjected to after it was grown. 00:53:07.020 --> 00:53:09.810 align:middle line:84% The oxide might have been subjected to plasma etching. 00:53:09.810 --> 00:53:12.480 align:middle line:84% The plasma could put traps in because the plasma has 00:53:12.480 --> 00:53:15.900 align:middle line:84% energetic ions going into this thing. 00:53:15.900 --> 00:53:18.630 align:middle line:84% Or it may capture electrons that are 00:53:18.630 --> 00:53:21.750 align:middle line:84% holes that are injected into the oxide during device operation. 00:53:21.750 --> 00:53:24.480 align:middle line:84% So the energetic electrons may be coming along the surface 00:53:24.480 --> 00:53:26.910 align:middle line:84% and, boop, get popped up by some scattering event 00:53:26.910 --> 00:53:28.290 align:middle line:90% and create oxide trap charge. 00:53:28.290 --> 00:53:30.060 align:middle line:90% So it's in the bulk. 00:53:30.060 --> 00:53:32.790 align:middle line:84% And it may depend, not only on how the oxide was grown, 00:53:32.790 --> 00:53:34.930 align:middle line:84% but what happened to the oxide afterwards. 00:53:34.930 --> 00:53:38.280 align:middle line:84% So those are the four characteristic types, 00:53:38.280 --> 00:53:41.790 align:middle line:84% and we can see their signature by doing certain types 00:53:41.790 --> 00:53:43.110 align:middle line:90% of electrical measurements. 00:53:43.110 --> 00:53:45.680 align:middle line:90% 00:53:45.680 --> 00:53:48.992 align:middle line:84% So let's take a look at slide 32. 00:53:48.992 --> 00:53:51.200 align:middle line:84% There are a lot of different measurement techniques-- 00:53:51.200 --> 00:53:53.480 align:middle line:84% and your book talks about them-- to understand 00:53:53.480 --> 00:53:54.740 align:middle line:90% the properties of oxide. 00:53:54.740 --> 00:53:56.495 align:middle line:84% You can do physical measurements. 00:53:56.495 --> 00:53:58.370 align:middle line:84% A very simple physical measurement any of you 00:53:58.370 --> 00:54:02.000 align:middle line:84% can do in the lab, take an oxide and etch it 00:54:02.000 --> 00:54:05.150 align:middle line:84% with hydrofluoric acid, and measure its etch rate. 00:54:05.150 --> 00:54:07.340 align:middle line:84% And it turns out, the etch rate in Hf, 00:54:07.340 --> 00:54:10.010 align:middle line:84% so how quickly it gets removed by Hf 00:54:10.010 --> 00:54:12.890 align:middle line:84% is a function of the density and the index of refraction 00:54:12.890 --> 00:54:14.280 align:middle line:90% of the density oxide. 00:54:14.280 --> 00:54:16.160 align:middle line:84% So it gives you an idea of the density. 00:54:16.160 --> 00:54:18.230 align:middle line:84% You can do scanning electron micrograph 00:54:18.230 --> 00:54:23.930 align:middle line:84% or AFM, atomic force, to see how rough the surface is. 00:54:23.930 --> 00:54:26.930 align:middle line:84% A very common measurement is called ellipsometry. 00:54:26.930 --> 00:54:29.630 align:middle line:84% This is a technique where you put a laser 00:54:29.630 --> 00:54:33.200 align:middle line:84% beam into the oxide, and you look at its reflection. 00:54:33.200 --> 00:54:35.360 align:middle line:84% And you measure-- actually, it's a polarized beam. 00:54:35.360 --> 00:54:37.820 align:middle line:84% You measure the shift in the polarization. 00:54:37.820 --> 00:54:40.400 align:middle line:84% And it's a very nice method of measuring accurately 00:54:40.400 --> 00:54:43.280 align:middle line:84% the thickness of the oxide and its index 00:54:43.280 --> 00:54:45.430 align:middle line:84% of refraction, which tells you about the density. 00:54:45.430 --> 00:54:48.960 align:middle line:84% So almost every fab, and even the research labs, 00:54:48.960 --> 00:54:52.110 align:middle line:84% have ellipsometers, where you can take your wafer. 00:54:52.110 --> 00:54:54.487 align:middle line:84% You'd be able to take this wafer here from Intel, 00:54:54.487 --> 00:54:55.695 align:middle line:90% stick it on the ellipsometer. 00:54:55.695 --> 00:54:59.310 align:middle line:84% It would come back and say 5,000 angstroms or something, 00:54:59.310 --> 00:55:01.050 align:middle line:90% or a 532 angstroms. 00:55:01.050 --> 00:55:04.128 align:middle line:84% It can tell you the oxide thickness. 00:55:04.128 --> 00:55:05.920 align:middle line:84% What we're going to talk about a little bit 00:55:05.920 --> 00:55:08.290 align:middle line:90% here is electrical measurements. 00:55:08.290 --> 00:55:10.390 align:middle line:84% And some of you may be familiar with these, 00:55:10.390 --> 00:55:13.480 align:middle line:84% but there's a measurement called a capacitance voltage 00:55:13.480 --> 00:55:14.080 align:middle line:90% technique. 00:55:14.080 --> 00:55:16.780 align:middle line:84% It's probably the most powerful measurement technique 00:55:16.780 --> 00:55:19.060 align:middle line:84% that you can subject an oxide to. 00:55:19.060 --> 00:55:22.210 align:middle line:84% If you look in the text, section 6.4.3, 00:55:22.210 --> 00:55:23.890 align:middle line:84% there's two or three pages on it. 00:55:23.890 --> 00:55:26.410 align:middle line:84% I suggest you want to read through that, especially 00:55:26.410 --> 00:55:28.450 align:middle line:84% if you're not familiar with CV because it 00:55:28.450 --> 00:55:33.530 align:middle line:84% has a lot more detail than what we can do in the lecture. 00:55:33.530 --> 00:55:37.180 align:middle line:84% So let's just go through some of these CV measurements. 00:55:37.180 --> 00:55:39.502 align:middle line:84% If you've had basic courses in electronics, 00:55:39.502 --> 00:55:40.960 align:middle line:84% you'll be quite familiar with this. 00:55:40.960 --> 00:55:42.190 align:middle line:84% And it'll be boring and a review. 00:55:42.190 --> 00:55:44.800 align:middle line:84% If you haven't, it'll probably seem a little bit mysterious. 00:55:44.800 --> 00:55:46.770 align:middle line:90% But bear with us. 00:55:46.770 --> 00:55:49.270 align:middle line:84% And again, if you go back and you read through that section, 00:55:49.270 --> 00:55:50.350 align:middle line:90% it hopefully will help. 00:55:50.350 --> 00:55:52.000 align:middle line:84% The main point it will make, though, hopefully 00:55:52.000 --> 00:55:54.417 align:middle line:84% is that, by doing a simple measurement, which doesn't cost 00:55:54.417 --> 00:55:57.730 align:middle line:84% that much time or money, you can learn a tremendous amount 00:55:57.730 --> 00:55:59.980 align:middle line:84% about how that oxide was grown and what 00:55:59.980 --> 00:56:02.270 align:middle line:90% happened to it in its life. 00:56:02.270 --> 00:56:04.960 align:middle line:84% So I'm going to talk about a technique called high frequency 00:56:04.960 --> 00:56:07.130 align:middle line:90% capacitance voltage. 00:56:07.130 --> 00:56:08.720 align:middle line:84% And in order to do this measurement 00:56:08.720 --> 00:56:10.910 align:middle line:84% you do need to make a device, a very simple one. 00:56:10.910 --> 00:56:12.860 align:middle line:84% In fact, it's the simplest device probably 00:56:12.860 --> 00:56:15.120 align:middle line:84% that you can ever make on silicon. 00:56:15.120 --> 00:56:18.720 align:middle line:84% What you have is you take a silicon wafer. 00:56:18.720 --> 00:56:20.060 align:middle line:90% In this case, it's n-type. 00:56:20.060 --> 00:56:21.170 align:middle line:90% It's doped with donors. 00:56:21.170 --> 00:56:22.910 align:middle line:90% And you grow an oxide. 00:56:22.910 --> 00:56:24.590 align:middle line:84% You take it into the clean room, put it 00:56:24.590 --> 00:56:28.010 align:middle line:84% in at 900 degrees in an oxygen ambient for half an hour. 00:56:28.010 --> 00:56:29.990 align:middle line:84% You grow a certain oxide thickness. 00:56:29.990 --> 00:56:33.770 align:middle line:84% You put metal down, just evaporate aluminum and pattern 00:56:33.770 --> 00:56:34.820 align:middle line:90% into little dots. 00:56:34.820 --> 00:56:37.550 align:middle line:84% And then you make a contact to the back of the wafer, 00:56:37.550 --> 00:56:39.620 align:middle line:84% and you put a probe down the front of the wafer. 00:56:39.620 --> 00:56:41.150 align:middle line:84% So relatively simple measurement. 00:56:41.150 --> 00:56:45.860 align:middle line:84% We do this here at MIT in a class called 6.152J. 00:56:45.860 --> 00:56:46.490 align:middle line:90% Very simple. 00:56:46.490 --> 00:56:49.820 align:middle line:84% You make an MOS capacitor in the lab. 00:56:49.820 --> 00:56:51.830 align:middle line:84% This is physically what it looks like. 00:56:51.830 --> 00:56:54.350 align:middle line:84% It's got silicon, oxide, and metal. 00:56:54.350 --> 00:56:56.240 align:middle line:84% This little symbol in the middle is 00:56:56.240 --> 00:56:57.980 align:middle line:84% meant to represent from the circuit 00:56:57.980 --> 00:57:00.290 align:middle line:84% engine or the electrical engineers or physics point 00:57:00.290 --> 00:57:01.800 align:middle line:90% of view what it looks like. 00:57:01.800 --> 00:57:02.570 align:middle line:90% It's a capacitor. 00:57:02.570 --> 00:57:06.920 align:middle line:84% It's two electrodes with contacts on either side. 00:57:06.920 --> 00:57:09.500 align:middle line:84% And what I'm showing in this plot here 00:57:09.500 --> 00:57:13.880 align:middle line:84% is the capacitance on the y-axis as measured as a function 00:57:13.880 --> 00:57:14.760 align:middle line:90% of the gate voltage. 00:57:14.760 --> 00:57:17.090 align:middle line:84% So the amount of voltage put on the gate 00:57:17.090 --> 00:57:19.370 align:middle line:84% is referred to as that little aluminum dot 00:57:19.370 --> 00:57:21.570 align:middle line:90% on the surface of the wafer. 00:57:21.570 --> 00:57:25.280 align:middle line:84% And we've biased this particular wafer in such a way 00:57:25.280 --> 00:57:28.940 align:middle line:84% that it's got a positive bias on the aluminum dot. 00:57:28.940 --> 00:57:31.310 align:middle line:84% In that case, electrons from the substrate 00:57:31.310 --> 00:57:32.870 align:middle line:84% will be attracted to the surface. 00:57:32.870 --> 00:57:35.480 align:middle line:84% And we do what we call-- we accumulate 00:57:35.480 --> 00:57:37.610 align:middle line:90% the substrate, the surface. 00:57:37.610 --> 00:57:40.940 align:middle line:84% The primary majority carrier in the silicon 00:57:40.940 --> 00:57:42.170 align:middle line:90% is electrons anyway. 00:57:42.170 --> 00:57:45.000 align:middle line:84% But I get more of them right at that silicon-silicon-dioxide 00:57:45.000 --> 00:57:45.500 align:middle line:90% interface. 00:57:45.500 --> 00:57:46.890 align:middle line:90% I accumulate them. 00:57:46.890 --> 00:57:50.588 align:middle line:84% And so if I were to measure the capacitance of this thing, 00:57:50.588 --> 00:57:52.880 align:middle line:84% it looks just like a parallel plate from basic physics. 00:57:52.880 --> 00:57:54.080 align:middle line:90% I have aluminum. 00:57:54.080 --> 00:57:55.190 align:middle line:90% I have a dielectric. 00:57:55.190 --> 00:57:56.870 align:middle line:84% I have another sheet of electrons. 00:57:56.870 --> 00:57:58.310 align:middle line:90% Acts sort of metallic. 00:57:58.310 --> 00:57:59.630 align:middle line:90% That's called a capacitor. 00:57:59.630 --> 00:58:03.343 align:middle line:84% The capacitance you measure is the dielectric capacitance 00:58:03.343 --> 00:58:04.760 align:middle line:84% related to the dielectric constant 00:58:04.760 --> 00:58:06.770 align:middle line:90% and the thickness, c-ox. 00:58:06.770 --> 00:58:09.990 align:middle line:84% So as long as I bias it was sufficiently positive voltage, 00:58:09.990 --> 00:58:11.070 align:middle line:90% I get a constant. 00:58:11.070 --> 00:58:13.600 align:middle line:84% The capacitance is independent of the gate bias. 00:58:13.600 --> 00:58:15.740 align:middle line:90% So that's accumulation. 00:58:15.740 --> 00:58:17.510 align:middle line:84% Now we start taking the gate bias 00:58:17.510 --> 00:58:19.380 align:middle line:90% and start making it negative. 00:58:19.380 --> 00:58:24.510 align:middle line:84% So this vg now goes to 0 and then to a negative number. 00:58:24.510 --> 00:58:27.020 align:middle line:84% Well, when that happens, we get something actually calling 00:58:27.020 --> 00:58:27.650 align:middle line:90% depletion. 00:58:27.650 --> 00:58:29.640 align:middle line:84% And depletion, as the name suggests, 00:58:29.640 --> 00:58:32.630 align:middle line:84% means that the surface region is now, instead of accumulated, 00:58:32.630 --> 00:58:34.610 align:middle line:90% it's depleted of free electrons. 00:58:34.610 --> 00:58:36.650 align:middle line:84% The free carriers are now pushed away 00:58:36.650 --> 00:58:38.840 align:middle line:84% from the surface, shown sort of schematically 00:58:38.840 --> 00:58:40.850 align:middle line:84% by this arrow, the electrons going away. 00:58:40.850 --> 00:58:43.640 align:middle line:84% And we form a region here of a certain thickness, 00:58:43.640 --> 00:58:46.820 align:middle line:84% x sub d, where there are no electrons. 00:58:46.820 --> 00:58:48.470 align:middle line:90% There are no majority carriers. 00:58:48.470 --> 00:58:50.540 align:middle line:90% It's depleted of free carriers. 00:58:50.540 --> 00:58:53.090 align:middle line:84% And this x sub d region is called the depletion region. 00:58:53.090 --> 00:58:54.950 align:middle line:90% It has a certain thickness. 00:58:54.950 --> 00:58:57.470 align:middle line:84% And in fact, the thickness of that depleted region, 00:58:57.470 --> 00:59:00.080 align:middle line:84% the thickness of xd depends on the bias. 00:59:00.080 --> 00:59:04.010 align:middle line:84% As I make it more negative, xd grows. 00:59:04.010 --> 00:59:06.155 align:middle line:84% In fact, what we have electrically, 00:59:06.155 --> 00:59:08.030 align:middle line:84% if you want to look at the circuit schematic, 00:59:08.030 --> 00:59:09.470 align:middle line:84% we have two capacitors in series. 00:59:09.470 --> 00:59:12.240 align:middle line:84% We have an oxide capacitor from here to here. 00:59:12.240 --> 00:59:14.900 align:middle line:84% And we have a depletion capacitance 00:59:14.900 --> 00:59:17.270 align:middle line:84% in the semiconductor, which is a variable. 00:59:17.270 --> 00:59:18.810 align:middle line:90% And it depends on the voltage. 00:59:18.810 --> 00:59:21.765 align:middle line:84% So in fact, if you measure the capacitance, 00:59:21.765 --> 00:59:22.640 align:middle line:90% it starts to go down. 00:59:22.640 --> 00:59:27.680 align:middle line:84% And you're sweeping out a curve as you make it more negative. 00:59:27.680 --> 00:59:29.530 align:middle line:84% So if you go on to slide 34 and you 00:59:29.530 --> 00:59:31.120 align:middle line:84% continue on making it more negative, 00:59:31.120 --> 00:59:35.100 align:middle line:84% you finally hit a point called v threshold. 00:59:35.100 --> 00:59:39.300 align:middle line:84% And right at that point, you're at something called inversion. 00:59:39.300 --> 00:59:40.800 align:middle line:84% And in fact, what's happened, you've 00:59:40.800 --> 00:59:42.930 align:middle line:84% put so much negative charge on the gate 00:59:42.930 --> 00:59:46.530 align:middle line:84% that you actually start getting positive holes attracted 00:59:46.530 --> 00:59:48.810 align:middle line:84% to that interface, to that top interface. 00:59:48.810 --> 00:59:51.480 align:middle line:84% So you've now formed an inversion layer. 00:59:51.480 --> 00:59:54.300 align:middle line:84% If you're a circuit design person, right at VT, 00:59:54.300 --> 00:59:57.510 align:middle line:84% that's where the device turns on for a MOSFET. 00:59:57.510 --> 01:00:00.240 align:middle line:84% That's where you've just formed a conducting channel. 01:00:00.240 --> 01:00:02.790 align:middle line:84% This bright red region here is the holes 01:00:02.790 --> 01:00:04.600 align:middle line:90% in the inversion layer. 01:00:04.600 --> 01:00:06.810 align:middle line:84% So in an n type semiconductor, in inversion, 01:00:06.810 --> 01:00:12.270 align:middle line:84% you have a p type in a layer of holes at the interface. 01:00:12.270 --> 01:00:15.510 align:middle line:84% This xd, which I said was expanding, it stops expanding. 01:00:15.510 --> 01:00:17.160 align:middle line:90% It reaches its maximum. 01:00:17.160 --> 01:00:19.590 align:middle line:84% And all of a sudden, you get a constant capacitance. 01:00:19.590 --> 01:00:21.090 align:middle line:84% You notice the capacitance curve now 01:00:21.090 --> 01:00:24.040 align:middle line:84% is flattened out, which is just a series 01:00:24.040 --> 01:00:29.440 align:middle line:84% capacitance of this c-ox and this c depletion region, 01:00:29.440 --> 01:00:31.920 align:middle line:84% where the depletion region is maxed out. 01:00:31.920 --> 01:00:34.100 align:middle line:84% So you trace out a very simple looking 01:00:34.100 --> 01:00:38.810 align:middle line:84% curve, which we'll see how the properties of this oxide 01:00:38.810 --> 01:00:39.830 align:middle line:90% affect this curve. 01:00:39.830 --> 01:00:42.770 align:middle line:84% But the basic idea is that all regions, no matter where you're 01:00:42.770 --> 01:00:45.620 align:middle line:84% operating this device, the amount of charge that's 01:00:45.620 --> 01:00:47.810 align:middle line:84% on the gate here, on the aluminum, 01:00:47.810 --> 01:00:50.840 align:middle line:84% has to be balanced by the charges in the substrate. 01:00:50.840 --> 01:00:53.300 align:middle line:84% And the charges in the substrate will be either 01:00:53.300 --> 01:00:56.180 align:middle line:84% the depletion charge consisting of the charge associated 01:00:56.180 --> 01:00:57.980 align:middle line:84% with the donors in the depletion region, 01:00:57.980 --> 01:01:00.800 align:middle line:84% or the inversion charge, which is also positive, which 01:01:00.800 --> 01:01:02.580 align:middle line:90% are these holes in this region. 01:01:02.580 --> 01:01:05.522 align:middle line:84% So that's kind of a basic charge neutrality overall 01:01:05.522 --> 01:01:06.230 align:middle line:90% that has to hold. 01:01:06.230 --> 01:01:09.100 align:middle line:90% 01:01:09.100 --> 01:01:12.660 align:middle line:84% So let's go on to slide 36 briefly. 01:01:12.660 --> 01:01:14.848 align:middle line:84% This is a little more detailed solid state physics 01:01:14.848 --> 01:01:16.140 align:middle line:90% than most you need to get into. 01:01:16.140 --> 01:01:18.432 align:middle line:84% If you're familiar with it, again, it'll be a reminder. 01:01:18.432 --> 01:01:22.170 align:middle line:84% If not, you have to take it as a truth. 01:01:22.170 --> 01:01:25.980 align:middle line:84% The question is, once I've formed this inversion layer 01:01:25.980 --> 01:01:28.860 align:middle line:84% and I have this Q sub I and I add 01:01:28.860 --> 01:01:31.260 align:middle line:84% additional charge to the gate, it's 01:01:31.260 --> 01:01:33.300 align:middle line:84% balanced by more inversion charge. 01:01:33.300 --> 01:01:36.880 align:middle line:84% You might say, well, why doesn't xd just keep growing? 01:01:36.880 --> 01:01:38.380 align:middle line:90% People always wonder about that. 01:01:38.380 --> 01:01:40.540 align:middle line:84% Why do we always get more inversion charge instead? 01:01:40.540 --> 01:01:43.180 align:middle line:84% Well, it turns out it's just a lot easier at that point 01:01:43.180 --> 01:01:45.550 align:middle line:84% to create inversion charge, to create 01:01:45.550 --> 01:01:46.940 align:middle line:90% these holes at the interface. 01:01:46.940 --> 01:01:49.630 align:middle line:84% In fact, if you're taking a solid state physics class, 01:01:49.630 --> 01:01:52.670 align:middle line:84% you know that that hole density looks something like this. 01:01:52.670 --> 01:01:57.010 align:middle line:84% In fact, it has an exponential dependence on that voltage, 01:01:57.010 --> 01:01:58.990 align:middle line:84% on how we move that potential up and down. 01:01:58.990 --> 01:02:00.593 align:middle line:84% So exponential means we only have 01:02:00.593 --> 01:02:02.260 align:middle line:84% to move the potential just a little bit, 01:02:02.260 --> 01:02:04.930 align:middle line:84% and we can get exponentially more carriers. 01:02:04.930 --> 01:02:07.330 align:middle line:84% So it's a lot easier to create this inversion 01:02:07.330 --> 01:02:10.690 align:middle line:84% charge at this point than it is to create more depletion 01:02:10.690 --> 01:02:11.230 align:middle line:90% charge. 01:02:11.230 --> 01:02:13.462 align:middle line:84% So that's kind of a reason why that happens. 01:02:13.462 --> 01:02:14.920 align:middle line:84% If you're familiar with it, that'll 01:02:14.920 --> 01:02:16.495 align:middle line:90% just remind you how that works. 01:02:16.495 --> 01:02:19.120 align:middle line:90% 01:02:19.120 --> 01:02:20.555 align:middle line:90% So let's go on to slide 36. 01:02:20.555 --> 01:02:21.930 align:middle line:84% And now, I've traced out-- again, 01:02:21.930 --> 01:02:24.450 align:middle line:84% this is a very schematic sort of view. 01:02:24.450 --> 01:02:26.850 align:middle line:84% But I've traced out a plot of capacitance 01:02:26.850 --> 01:02:28.110 align:middle line:90% on the vertical axis. 01:02:28.110 --> 01:02:31.410 align:middle line:84% And the x-axis is DC gate voltage. 01:02:31.410 --> 01:02:33.600 align:middle line:84% And we just traced out for you this black curve, 01:02:33.600 --> 01:02:36.180 align:middle line:84% where we started at c-ox, a constant number. 01:02:36.180 --> 01:02:39.510 align:middle line:84% As I sweep down through 0, I go in depletion. 01:02:39.510 --> 01:02:43.710 align:middle line:84% And it goes down, and then the capacitance reaches a minimum. 01:02:43.710 --> 01:02:47.070 align:middle line:84% And that's what an ideal high-frequency capacitance 01:02:47.070 --> 01:02:50.430 align:middle line:90% curve would look like. 01:02:50.430 --> 01:02:56.085 align:middle line:84% Now, basically, so what happens when 01:02:56.085 --> 01:02:57.210 align:middle line:90% we go to a lower frequency? 01:02:57.210 --> 01:02:58.830 align:middle line:84% Well, I have another curve here where 01:02:58.830 --> 01:03:01.260 align:middle line:84% we swept through this region, through depletion. 01:03:01.260 --> 01:03:02.760 align:middle line:84% And then all of a sudden, the curve, 01:03:02.760 --> 01:03:06.690 align:middle line:84% right after threshold in inversion at low frequencies, 01:03:06.690 --> 01:03:09.750 align:middle line:84% the capacitance goes back up to c-ox. 01:03:09.750 --> 01:03:11.730 align:middle line:84% Well, it's because if I have a signal that's 01:03:11.730 --> 01:03:14.790 align:middle line:84% going at low enough frequencies, the inversion layer carriers 01:03:14.790 --> 01:03:16.480 align:middle line:90% can actually follow that signal. 01:03:16.480 --> 01:03:19.720 align:middle line:84% So instead of just getting this minimum capacitance 01:03:19.720 --> 01:03:22.640 align:middle line:84% of these two in capacitors in series, 01:03:22.640 --> 01:03:27.040 align:middle line:84% it actually goes back up to look like a normal parallel plate 01:03:27.040 --> 01:03:28.820 align:middle line:90% capacitor with capacitance c-ox. 01:03:28.820 --> 01:03:30.320 align:middle line:84% In fact, if we go to the next slide. 01:03:30.320 --> 01:03:34.750 align:middle line:84% I think it explains it a little bit more physically on slide 01:03:34.750 --> 01:03:37.400 align:middle line:90% 37, the different regions. 01:03:37.400 --> 01:03:39.280 align:middle line:84% So in accumulation, remember, we said 01:03:39.280 --> 01:03:41.290 align:middle line:84% we just had a regular parallel plate capacitor, 01:03:41.290 --> 01:03:42.340 align:middle line:90% capacitance c-ox. 01:03:42.340 --> 01:03:45.893 align:middle line:84% I wiggle charge here on the metal gate, 01:03:45.893 --> 01:03:47.560 align:middle line:84% and you wiggle a little bit equal amount 01:03:47.560 --> 01:03:49.720 align:middle line:90% of charge in the substrate. 01:03:49.720 --> 01:03:51.382 align:middle line:90% In depletion, what's happening? 01:03:51.382 --> 01:03:52.840 align:middle line:84% You're wiggling charge on the gate. 01:03:52.840 --> 01:03:56.280 align:middle line:84% And instead of accumulating, you're actually depleting. 01:03:56.280 --> 01:03:58.030 align:middle line:84% And you're wiggling charge at the back end 01:03:58.030 --> 01:04:00.580 align:middle line:84% of the space charge region, which is sweeping out. 01:04:00.580 --> 01:04:02.660 align:middle line:84% And that's why the capacitance goes down. 01:04:02.660 --> 01:04:06.490 align:middle line:84% Now, if I'm at low frequency, and I'm inverted, 01:04:06.490 --> 01:04:09.820 align:middle line:84% I've created this inversion layer here of positive charge. 01:04:09.820 --> 01:04:13.990 align:middle line:84% If I wiggle charge on the gate, it's low enough frequency 01:04:13.990 --> 01:04:17.410 align:middle line:84% that I can actually wiggle charge in the inversion layer. 01:04:17.410 --> 01:04:20.075 align:middle line:84% However, if I'm at a bias where I'm in inversion 01:04:20.075 --> 01:04:21.700 align:middle line:84% but I'm doing it a very high frequency, 01:04:21.700 --> 01:04:25.990 align:middle line:84% say, megahertz or 100 kilohertz, the inversion layer 01:04:25.990 --> 01:04:27.190 align:middle line:90% can't respond that fast. 01:04:27.190 --> 01:04:29.200 align:middle line:84% But the depletion region actually can. 01:04:29.200 --> 01:04:33.980 align:middle line:84% So you actually get the series capacitance 01:04:33.980 --> 01:04:36.300 align:middle line:84% of both the inversion layer and the depletion layer. 01:04:36.300 --> 01:04:38.660 align:middle line:84% So this is all reasonably well understood if you 01:04:38.660 --> 01:04:40.490 align:middle line:84% want to read in [? Peres ?] book, 01:04:40.490 --> 01:04:43.010 align:middle line:84% it has a little more detailed explanation 01:04:43.010 --> 01:04:45.500 align:middle line:84% of these different regimes or regions of the curve. 01:04:45.500 --> 01:04:48.180 align:middle line:90% 01:04:48.180 --> 01:04:51.470 align:middle line:84% So there's another region that tells us 01:04:51.470 --> 01:04:53.450 align:middle line:84% something of the CV curve, which is 01:04:53.450 --> 01:04:54.950 align:middle line:84% going to end up telling us something 01:04:54.950 --> 01:04:56.640 align:middle line:90% about the quality of the oxide. 01:04:56.640 --> 01:04:58.280 align:middle line:90% And it's called deep depletion. 01:04:58.280 --> 01:05:00.992 align:middle line:84% And I've marked that region here on a high-frequency curve, 01:05:00.992 --> 01:05:02.450 align:middle line:84% where it actually, instead of being 01:05:02.450 --> 01:05:06.390 align:middle line:84% flat going straight across, it actually starts to go down. 01:05:06.390 --> 01:05:08.930 align:middle line:84% And what we've done in deep depletion-- 01:05:08.930 --> 01:05:12.140 align:middle line:84% remember, in this regime we have two capacitors 01:05:12.140 --> 01:05:14.780 align:middle line:84% in series, the oxide and the depletion region-- 01:05:14.780 --> 01:05:19.180 align:middle line:84% is that we sweep the DC voltage very fast so that the inversion 01:05:19.180 --> 01:05:20.770 align:middle line:90% layer carries can't follow it. 01:05:20.770 --> 01:05:23.950 align:middle line:84% So xd ends up expanding a little bit. 01:05:23.950 --> 01:05:26.683 align:middle line:84% And in fact, if you see deep depletion, 01:05:26.683 --> 01:05:29.350 align:middle line:84% it's a sign that you have a very high minority carrier lifetime. 01:05:29.350 --> 01:05:31.300 align:middle line:90% You have very few traps. 01:05:31.300 --> 01:05:33.710 align:middle line:84% You don't have much iron and things like that. 01:05:33.710 --> 01:05:35.830 align:middle line:84% So people will often take capacitors 01:05:35.830 --> 01:05:38.320 align:middle line:84% and try to intentionally sweep the CV quickly, 01:05:38.320 --> 01:05:41.230 align:middle line:84% see if they can get this lowering, this deep depletion. 01:05:41.230 --> 01:05:43.900 align:middle line:84% The faster you sweep it, the further down it goes. 01:05:43.900 --> 01:05:45.572 align:middle line:84% If you don't see that lowering, you 01:05:45.572 --> 01:05:47.530 align:middle line:84% don't have very good minority carrier lifetime. 01:05:47.530 --> 01:05:49.780 align:middle line:84% You probably have a lot of traps or things 01:05:49.780 --> 01:05:51.410 align:middle line:90% somewhere at that interface. 01:05:51.410 --> 01:05:54.070 align:middle line:84% So this is a way of getting a rough idea of what 01:05:54.070 --> 01:05:56.500 align:middle line:84% kind of quality interface have you produced? 01:05:56.500 --> 01:05:58.615 align:middle line:84% Can you deep deplete the capacitor or not? 01:05:58.615 --> 01:06:02.540 align:middle line:90% 01:06:02.540 --> 01:06:06.430 align:middle line:84% So basically, from looking at these capacitance voltage 01:06:06.430 --> 01:06:09.430 align:middle line:84% measurements, we can extract a lot 01:06:09.430 --> 01:06:11.360 align:middle line:90% of quantitative information. 01:06:11.360 --> 01:06:13.080 align:middle line:84% In fact, we can get the oxide thickness. 01:06:13.080 --> 01:06:14.360 align:middle line:90% Remember, we had c-ox. 01:06:14.360 --> 01:06:18.430 align:middle line:84% C-ox just depends on the dielectric constant 01:06:18.430 --> 01:06:19.540 align:middle line:90% and the oxide thickness. 01:06:19.540 --> 01:06:21.580 align:middle line:84% So once I get that number, I know the area, 01:06:21.580 --> 01:06:23.380 align:middle line:84% I can calculate the oxide thickness. 01:06:23.380 --> 01:06:28.150 align:middle line:84% I can get the substrate doping from the depletion capacitance. 01:06:28.150 --> 01:06:31.060 align:middle line:84% And you can get all these different interface charges, 01:06:31.060 --> 01:06:35.650 align:middle line:84% Qf, Qit, the mobile, and the oxide trap charge, all of those 01:06:35.650 --> 01:06:38.860 align:middle line:84% just by doing CV measurements in their different configurations 01:06:38.860 --> 01:06:39.670 align:middle line:90% and frequencies. 01:06:39.670 --> 01:06:41.440 align:middle line:84% We can get this information, which 01:06:41.440 --> 01:06:45.400 align:middle line:84% tells you about the electrical quality of the device. 01:06:45.400 --> 01:06:48.490 align:middle line:90% 01:06:48.490 --> 01:06:51.600 align:middle line:84% So in fact, on slide 40, what I'm showing 01:06:51.600 --> 01:06:56.280 align:middle line:84% are some quote unquote "realistic" capacitance voltage 01:06:56.280 --> 01:06:58.590 align:middle line:84% curves, much more like you would measure in the lab. 01:06:58.590 --> 01:07:01.348 align:middle line:84% What we've shown so far on a [? cv ?] plot, 01:07:01.348 --> 01:07:02.140 align:middle line:90% it looks like this. 01:07:02.140 --> 01:07:04.290 align:middle line:90% So we have this curve here. 01:07:04.290 --> 01:07:06.210 align:middle line:84% Here in accumulation we're at c-ox. 01:07:06.210 --> 01:07:09.740 align:middle line:84% It comes down and sweeps out like this. 01:07:09.740 --> 01:07:11.820 align:middle line:84% So that's the ideal high-frequency curve. 01:07:11.820 --> 01:07:14.320 align:middle line:84% In fact, that's not what you would measure in a real device. 01:07:14.320 --> 01:07:17.720 align:middle line:84% You would measure that curve but shifted over to the left. 01:07:17.720 --> 01:07:19.930 align:middle line:84% So if you look at this curve right here, 01:07:19.930 --> 01:07:22.510 align:middle line:90% you see it's shifted. 01:07:22.510 --> 01:07:25.090 align:middle line:84% And in fact, it's the high-frequency curve 01:07:25.090 --> 01:07:27.070 align:middle line:90% but shifted by two terms. 01:07:27.070 --> 01:07:28.900 align:middle line:90% And the terms look like this. 01:07:28.900 --> 01:07:33.550 align:middle line:84% There's one term that goes like the coulombic charge, little q, 01:07:33.550 --> 01:07:38.150 align:middle line:84% times Qf, the fixed charge, divided by c-ox. 01:07:38.150 --> 01:07:40.070 align:middle line:90% So this is the Qf term. 01:07:40.070 --> 01:07:42.230 align:middle line:84% This is just-- remember, Qf is positive. 01:07:42.230 --> 01:07:44.270 align:middle line:84% So it just literally takes the threshold voltage 01:07:44.270 --> 01:07:46.580 align:middle line:90% and shifts it by Qf over c-ox. 01:07:46.580 --> 01:07:49.280 align:middle line:84% So you see this shift because of this positive fixed charge. 01:07:49.280 --> 01:07:52.160 align:middle line:84% The bigger the fixed charge, the bigger that shift. 01:07:52.160 --> 01:07:54.590 align:middle line:84% And in fact, it's the shift from the ideal curve 01:07:54.590 --> 01:07:55.400 align:middle line:90% that people use. 01:07:55.400 --> 01:07:59.120 align:middle line:84% That's the way people use to measure Qf. 01:07:59.120 --> 01:08:02.360 align:middle line:84% Phi [? M S ?] is the metal semiconductor work function 01:08:02.360 --> 01:08:05.360 align:middle line:84% difference, which is just sort of a property of the metal 01:08:05.360 --> 01:08:09.750 align:middle line:84% that you use, whether you use aluminum or what you use. 01:08:09.750 --> 01:08:11.750 align:middle line:84% So there's also a shift associated with that. 01:08:11.750 --> 01:08:14.000 align:middle line:84% And that's a fixed number depending on the polysilicon 01:08:14.000 --> 01:08:16.200 align:middle line:90% gate that you used. 01:08:16.200 --> 01:08:17.750 align:middle line:84% So with this positive fixed charge 01:08:17.750 --> 01:08:20.960 align:middle line:84% shifts the CV to the left, and you get this curve. 01:08:20.960 --> 01:08:22.170 align:middle line:90% That's a good sign. 01:08:22.170 --> 01:08:23.810 align:middle line:84% Now, most of you would actually, if you 01:08:23.810 --> 01:08:26.090 align:middle line:84% were to measure your devices, you'd get a distorted curve. 01:08:26.090 --> 01:08:27.740 align:middle line:84% Instead of looking just like with this, 01:08:27.740 --> 01:08:29.667 align:middle line:84% with this abrupt drop off, you'd see 01:08:29.667 --> 01:08:32.000 align:middle line:84% it looks a little bit distorted from what you calculated 01:08:32.000 --> 01:08:34.100 align:middle line:90% in the ideal case. 01:08:34.100 --> 01:08:36.890 align:middle line:84% And in fact, this distortion is due to the fact 01:08:36.890 --> 01:08:39.380 align:middle line:84% that there is variable interface trap charge. 01:08:39.380 --> 01:08:41.960 align:middle line:90% Remember, we talked about Qit? 01:08:41.960 --> 01:08:44.660 align:middle line:84% And we said the amount of charge and Qit 01:08:44.660 --> 01:08:47.250 align:middle line:90% depends on the bias voltage? 01:08:47.250 --> 01:08:50.120 align:middle line:84% So as I'm sweeping from here from right to left, 01:08:50.120 --> 01:08:53.330 align:middle line:84% I'm uncovering or causing extra-- 01:08:53.330 --> 01:08:57.020 align:middle line:84% my ability to see, to charge an uncharge Qit. 01:08:57.020 --> 01:08:59.670 align:middle line:84% And that ends up shifting the curve by variable amounts. 01:08:59.670 --> 01:09:01.109 align:middle line:90% So it gets distorted. 01:09:01.109 --> 01:09:03.350 align:middle line:84% So the amount of distortion is a way 01:09:03.350 --> 01:09:04.970 align:middle line:84% of telling how much Qit you have. 01:09:04.970 --> 01:09:07.920 align:middle line:90% 01:09:07.920 --> 01:09:10.109 align:middle line:84% And in fact, if you look on slide 41, 01:09:10.109 --> 01:09:12.569 align:middle line:90% this is pictured for you. 01:09:12.569 --> 01:09:14.670 align:middle line:84% If you've had solid state physics class 01:09:14.670 --> 01:09:17.939 align:middle line:84% and you know about energy band diagrams and states, 01:09:17.939 --> 01:09:20.729 align:middle line:84% then this is a way of thinking about that. 01:09:20.729 --> 01:09:22.649 align:middle line:84% If you haven't, again, it's the type of thing 01:09:22.649 --> 01:09:27.390 align:middle line:84% you'll have to take with a grain of salt or study up on it. 01:09:27.390 --> 01:09:31.460 align:middle line:84% So what we have pictorially here up in this upper band diagram 01:09:31.460 --> 01:09:34.880 align:middle line:84% is the conduction band here shown of the silicon. 01:09:34.880 --> 01:09:36.172 align:middle line:90% And here's the valence band. 01:09:36.172 --> 01:09:37.880 align:middle line:84% So here's the band gap between those two. 01:09:37.880 --> 01:09:39.649 align:middle line:84% This is the semiconductor of a silicon, 01:09:39.649 --> 01:09:41.149 align:middle line:90% and on the left is the oxide. 01:09:41.149 --> 01:09:42.890 align:middle line:84% And notice, right at the interface, 01:09:42.890 --> 01:09:44.540 align:middle line:84% there's a whole bunch of little levels. 01:09:44.540 --> 01:09:47.750 align:middle line:84% Each little bar here represents an energy level 01:09:47.750 --> 01:09:49.340 align:middle line:90% that electron can occupy. 01:09:49.340 --> 01:09:51.350 align:middle line:84% And you notice that there are energy levels 01:09:51.350 --> 01:09:52.880 align:middle line:90% within the entire band gap. 01:09:52.880 --> 01:09:56.850 align:middle line:84% There are certain density of states throughout the band gap. 01:09:56.850 --> 01:09:58.890 align:middle line:84% And we're going to assume they're donor type, 01:09:58.890 --> 01:10:00.810 align:middle line:90% and they're distributed somehow. 01:10:00.810 --> 01:10:02.460 align:middle line:84% Their donor type, meaning that they're 01:10:02.460 --> 01:10:04.180 align:middle line:90% positive when they're empty. 01:10:04.180 --> 01:10:05.860 align:middle line:84% So if there's no electron in there, 01:10:05.860 --> 01:10:07.680 align:middle line:90% this has a positive charge. 01:10:07.680 --> 01:10:12.320 align:middle line:84% If it captures an electron, it's neutral, OK? 01:10:12.320 --> 01:10:14.030 align:middle line:84% So now we look at this energy band 01:10:14.030 --> 01:10:17.690 align:middle line:84% diagram in the center set of curves for different bias 01:10:17.690 --> 01:10:18.720 align:middle line:90% conditions. 01:10:18.720 --> 01:10:20.810 align:middle line:84% So we can look at it under inversion. 01:10:20.810 --> 01:10:25.880 align:middle line:84% So when the surface is inverted, it turns out that Qit is large, 01:10:25.880 --> 01:10:26.900 align:middle line:90% and it's positive. 01:10:26.900 --> 01:10:28.080 align:middle line:90% It's mostly empty. 01:10:28.080 --> 01:10:29.180 align:middle line:90% We have the bands bent. 01:10:29.180 --> 01:10:32.370 align:middle line:84% None of these have any electrons in them. 01:10:32.370 --> 01:10:36.110 align:middle line:84% So I have now a lot of positive charge at that interface 01:10:36.110 --> 01:10:38.840 align:middle line:90% under that bias condition, OK? 01:10:38.840 --> 01:10:43.010 align:middle line:84% So here I am at inversion, region C on this bottom curve. 01:10:43.010 --> 01:10:45.260 align:middle line:84% And lo and behold, I have a lot of positive charge. 01:10:45.260 --> 01:10:48.140 align:middle line:84% That means I am shifted from the ideal case quite 01:10:48.140 --> 01:10:49.400 align:middle line:90% a bit to the left. 01:10:49.400 --> 01:10:52.760 align:middle line:84% And you see why in region C of this realistic curve 01:10:52.760 --> 01:10:54.930 align:middle line:90% I'm shifted quite far over. 01:10:54.930 --> 01:10:56.700 align:middle line:90% So I'm distorted. 01:10:56.700 --> 01:10:58.650 align:middle line:84% So now let's go to region B of that curve. 01:10:58.650 --> 01:11:01.920 align:middle line:84% So that's here under depletion in the center. 01:11:01.920 --> 01:11:03.910 align:middle line:84% When I'm depleted, the bias voltage, again, 01:11:03.910 --> 01:11:05.573 align:middle line:90% is a little bit closer to 0. 01:11:05.573 --> 01:11:06.990 align:middle line:84% I don't have as much band bending. 01:11:06.990 --> 01:11:09.630 align:middle line:84% And in fact, some of these density of interface traps 01:11:09.630 --> 01:11:12.430 align:middle line:84% is now filled with electrons, maybe about half of them. 01:11:12.430 --> 01:11:16.360 align:middle line:84% So the amount of Qit that's uncovered at that interface 01:11:16.360 --> 01:11:18.270 align:middle line:84% now in this bias condition is less. 01:11:18.270 --> 01:11:20.250 align:middle line:84% So I have less interface trap charge. 01:11:20.250 --> 01:11:23.160 align:middle line:84% So indeed, if you look at the ideal curve 01:11:23.160 --> 01:11:25.380 align:middle line:84% and you look at the distorted curve, at point B, 01:11:25.380 --> 01:11:28.830 align:middle line:84% there's less of a shift than there was at point C 01:11:28.830 --> 01:11:31.680 align:middle line:84% because there's less charge at that interface, less 01:11:31.680 --> 01:11:32.970 align:middle line:90% positive charge. 01:11:32.970 --> 01:11:37.000 align:middle line:84% And now, let's finally go to accumulation. 01:11:37.000 --> 01:11:39.160 align:middle line:84% I've bent things-- the voltage now 01:11:39.160 --> 01:11:42.560 align:middle line:84% is such that all of these traps are occupied by electrons. 01:11:42.560 --> 01:11:44.890 align:middle line:90% So Qit has its minimum value. 01:11:44.890 --> 01:11:46.300 align:middle line:90% They're all occupied. 01:11:46.300 --> 01:11:48.440 align:middle line:90% So Qit is low. 01:11:48.440 --> 01:11:51.920 align:middle line:84% And so in region A, where we're in accumulation, in fact, 01:11:51.920 --> 01:11:53.990 align:middle line:84% the shift from the ideal is very, very small. 01:11:53.990 --> 01:11:55.220 align:middle line:90% It's a small shift. 01:11:55.220 --> 01:11:59.080 align:middle line:84% So as I sweep here, from inversion, I get the curve. 01:11:59.080 --> 01:12:01.930 align:middle line:84% A lot of shift to depletion, not so much shift, 01:12:01.930 --> 01:12:04.000 align:middle line:90% accumulation very little shift. 01:12:04.000 --> 01:12:05.890 align:middle line:84% So the distortion of the curve has 01:12:05.890 --> 01:12:07.930 align:middle line:84% to do with these traps that depend 01:12:07.930 --> 01:12:14.510 align:middle line:84% on the bias level and the fact that it varies with bias. 01:12:14.510 --> 01:12:16.150 align:middle line:90% So how do people measure Qit? 01:12:16.150 --> 01:12:19.558 align:middle line:84% Well, there's one way of doing it. 01:12:19.558 --> 01:12:21.100 align:middle line:84% There are a couple of different ones, 01:12:21.100 --> 01:12:23.260 align:middle line:84% but what people do is they measure 01:12:23.260 --> 01:12:25.870 align:middle line:84% a curve under high frequency, and they 01:12:25.870 --> 01:12:27.730 align:middle line:84% choose the frequency such that the traps, 01:12:27.730 --> 01:12:29.930 align:middle line:90% the Qit's can't respond. 01:12:29.930 --> 01:12:34.800 align:middle line:84% So they get kind of an ideal like curve, shown here. 01:12:34.800 --> 01:12:36.360 align:middle line:84% Or you can calculate the ideal curve. 01:12:36.360 --> 01:12:39.420 align:middle line:84% If you're a more theoretical person, you can calculate it. 01:12:39.420 --> 01:12:41.670 align:middle line:84% And then they do a low-frequency curve, 01:12:41.670 --> 01:12:44.640 align:middle line:84% and they choose low enough so that the traps can respond. 01:12:44.640 --> 01:12:47.280 align:middle line:84% And in fact, they see a difference. 01:12:47.280 --> 01:12:49.950 align:middle line:84% In a certain bias regime, you can see a difference 01:12:49.950 --> 01:12:52.173 align:middle line:90% in the capacitance. 01:12:52.173 --> 01:12:53.590 align:middle line:84% And that difference in capacitance 01:12:53.590 --> 01:12:57.400 align:middle line:84% can be related to the density of the traps that 01:12:57.400 --> 01:13:00.820 align:middle line:84% are in a certain region of the bandgap 01:13:00.820 --> 01:13:03.100 align:middle line:84% because where you are in the bandgap corresponds 01:13:03.100 --> 01:13:04.400 align:middle line:90% to where you are in bias. 01:13:04.400 --> 01:13:08.410 align:middle line:84% So you could give someone this data who is familiar with it, 01:13:08.410 --> 01:13:10.330 align:middle line:84% and they could plot then for you the density 01:13:10.330 --> 01:13:15.010 align:middle line:84% of traps in the bandgap at that interface between silicon 01:13:15.010 --> 01:13:17.710 align:middle line:84% silicon dioxide as a function of the energy in the bandgap. 01:13:17.710 --> 01:13:20.260 align:middle line:84% And you can see it has this familiar u-shaped profile. 01:13:20.260 --> 01:13:22.450 align:middle line:84% In the center, it has some number. 01:13:22.450 --> 01:13:25.780 align:middle line:84% Mid-gap it 's typically 5 10 to the 10th traps 01:13:25.780 --> 01:13:26.860 align:middle line:90% per square centimeter. 01:13:26.860 --> 01:13:31.550 align:middle line:84% As you go closer to the edges, it goes up dramatically. 01:13:31.550 --> 01:13:35.230 align:middle line:84% But this is a way by CV of measuring Dit or Qit. 01:13:35.230 --> 01:13:38.020 align:middle line:90% 01:13:38.020 --> 01:13:40.520 align:middle line:84% And in fact, we want to keep it as low as possible 01:13:40.520 --> 01:13:42.860 align:middle line:84% so our threshold voltage is moving all over the place 01:13:42.860 --> 01:13:44.527 align:middle line:84% and we're getting ideal characteristics. 01:13:44.527 --> 01:13:47.560 align:middle line:84% And what we do is, after we do an oxidation, 01:13:47.560 --> 01:13:51.650 align:middle line:84% the last high temperature step is typically some kind 01:13:51.650 --> 01:13:52.820 align:middle line:90% of forming gas anneal. 01:13:52.820 --> 01:13:55.340 align:middle line:84% And after annealing, this Dit goes down 01:13:55.340 --> 01:13:57.180 align:middle line:84% by about two orders of magnitude. 01:13:57.180 --> 01:14:01.490 align:middle line:84% So you notice the last step in a MOSFET or an MOS flow, 01:14:01.490 --> 01:14:03.890 align:middle line:84% if you've done it, is to do an annealing 01:14:03.890 --> 01:14:05.930 align:middle line:84% and form a gas, which has hydrogen in it, 01:14:05.930 --> 01:14:07.640 align:middle line:90% at about 450 or 500. 01:14:07.640 --> 01:14:10.640 align:middle line:84% And that takes this Dit from up here all the way down to here 01:14:10.640 --> 01:14:16.690 align:middle line:84% to a level that's tolerable for the device to operate. 01:14:16.690 --> 01:14:18.810 align:middle line:84% So I know I went through a lot of that CV stuff 01:14:18.810 --> 01:14:20.962 align:middle line:84% fairly rapidly, and I apologize for that. 01:14:20.962 --> 01:14:22.920 align:middle line:84% But I wanted to give you more than the details, 01:14:22.920 --> 01:14:24.610 align:middle line:90% give you a flavor for it. 01:14:24.610 --> 01:14:27.570 align:middle line:84% If you've had courses in electrical or solid state 01:14:27.570 --> 01:14:29.130 align:middle line:84% physics, you'll recognize some of it. 01:14:29.130 --> 01:14:30.600 align:middle line:84% If you haven't, you can go through 01:14:30.600 --> 01:14:33.140 align:middle line:90% in the text in more detail. 01:14:33.140 --> 01:14:35.820 align:middle line:84% But just to summarize on the oxidation, 01:14:35.820 --> 01:14:39.030 align:middle line:84% it's really probably one of the most critical processes 01:14:39.030 --> 01:14:40.650 align:middle line:90% for CMOS. 01:14:40.650 --> 01:14:42.840 align:middle line:84% Thermal oxides that are thermally grown 01:14:42.840 --> 01:14:46.260 align:middle line:84% are used for a lot of different things, tunnel insulators, 01:14:46.260 --> 01:14:48.570 align:middle line:90% masking oxide, field oxides. 01:14:48.570 --> 01:14:50.155 align:middle line:84% If nothing else, you should remember 01:14:50.155 --> 01:14:52.530 align:middle line:84% that the electrical properties of silicon silicon dioxide 01:14:52.530 --> 01:14:55.612 align:middle line:84% are the best, and they're superior to all other cases. 01:14:55.612 --> 01:14:57.570 align:middle line:84% And that's why it's taking so many years for us 01:14:57.570 --> 01:15:00.570 align:middle line:84% to find another insulator, another high k. 01:15:00.570 --> 01:15:04.470 align:middle line:84% It really is tough to beat, or it's tough even to equal it. 01:15:04.470 --> 01:15:05.700 align:middle line:90% There are charges, though. 01:15:05.700 --> 01:15:07.350 align:middle line:90% It's not perfect, remember? 01:15:07.350 --> 01:15:08.580 align:middle line:90% Nobody's perfect. 01:15:08.580 --> 01:15:11.700 align:middle line:84% 1 and then 10 to the fifth is a problem at the interface. 01:15:11.700 --> 01:15:13.770 align:middle line:84% But there are charges that exist and things 01:15:13.770 --> 01:15:17.470 align:middle line:84% like fixed charge, which is positive, Dit or Qit, 01:15:17.470 --> 01:15:18.460 align:middle line:90% mobile charge-- 01:15:18.460 --> 01:15:21.100 align:middle line:84% you don't have mobile charge if you're really clean-- 01:15:21.100 --> 01:15:22.570 align:middle line:90% and Qot. 01:15:22.570 --> 01:15:24.400 align:middle line:84% These charges do exist, but they can 01:15:24.400 --> 01:15:27.550 align:middle line:84% be characterized by relatively simple techniques, like MOSCV. 01:15:27.550 --> 01:15:30.760 align:middle line:84% So we'll go on next time and talk 01:15:30.760 --> 01:15:33.310 align:middle line:84% about the physics or the kinetics 01:15:33.310 --> 01:15:35.560 align:middle line:84% of how this process of thermal oxidation takes place. 01:15:35.560 --> 01:15:39.520 align:middle line:84% But please go ahead and start reading chapter 6. 01:15:39.520 --> 01:15:43.513 align:middle line:84% Also, your handout, homework number 2 was handed out. 01:15:43.513 --> 01:15:45.430 align:middle line:84% So you can go ahead and start working on that. 01:15:45.430 --> 01:15:47.360 align:middle line:90% That's due next Thursday. 01:15:47.360 --> 01:15:47.860 align:middle line:90% Thanks. 01:15:47.860 --> 01:15:54.740 align:middle line:90% 01:15:54.740 --> 01:15:56.290 align:middle line:90% it. 01:15:56.290 --> 01:16:00.000 align:middle line:90%