WEBVTT 00:00:00.000 --> 00:00:02.570 align:middle line:84% JUDY HOYT: In a polycrystalline state, 00:00:02.570 --> 00:00:05.300 align:middle line:84% or it can be grown epitaxially, which 00:00:05.300 --> 00:00:07.970 align:middle line:84% epitaxially-- "epitaxial" is a Greek word, which 00:00:07.970 --> 00:00:11.750 align:middle line:84% means, having the same crystal structure as the substrate. 00:00:11.750 --> 00:00:14.000 align:middle line:84% So silicon itself can be deposited. 00:00:14.000 --> 00:00:17.220 align:middle line:84% And we'll talk about that in the next couple of lectures. 00:00:17.220 --> 00:00:19.580 align:middle line:84% So we're going to talk about the methods, the equipment, 00:00:19.580 --> 00:00:23.210 align:middle line:84% and some basic concepts used to deposit or grow 00:00:23.210 --> 00:00:27.090 align:middle line:90% the-- these thin films. 00:00:27.090 --> 00:00:29.720 align:middle line:84% So let's just go back to a picture 00:00:29.720 --> 00:00:31.740 align:middle line:84% that I think I showed earlier in the course. 00:00:31.740 --> 00:00:38.120 align:middle line:84% But it's a cross-section view of, say, devices in a CMOS chip. 00:00:38.120 --> 00:00:40.400 align:middle line:84% And we've got our silicon substrate down here 00:00:40.400 --> 00:00:41.630 align:middle line:90% at the bottom. 00:00:41.630 --> 00:00:44.540 align:middle line:84% We've talked a lot about how to create n-wells and p-wells 00:00:44.540 --> 00:00:47.120 align:middle line:84% by ion planting, and diffusing, and the source drains, 00:00:47.120 --> 00:00:49.153 align:middle line:90% and all that. 00:00:49.153 --> 00:00:50.570 align:middle line:84% But there are a lot of other films 00:00:50.570 --> 00:00:52.080 align:middle line:90% here besides silicon substrate. 00:00:52.080 --> 00:00:55.220 align:middle line:84% So everything that's in a color, other than white here, 00:00:55.220 --> 00:00:56.870 align:middle line:90% is some other film. 00:00:56.870 --> 00:00:59.690 align:middle line:84% For example, here, we might have copper 00:00:59.690 --> 00:01:03.500 align:middle line:84% that's used in the interconnect wiring, or aluminum, 00:01:03.500 --> 00:01:05.209 align:middle line:90% or tungsten. 00:01:05.209 --> 00:01:08.690 align:middle line:84% The-- shown here in this bright magenta. 00:01:08.690 --> 00:01:11.000 align:middle line:84% These hatched regions that are blue 00:01:11.000 --> 00:01:13.040 align:middle line:84% could be a low temperature oxide. 00:01:13.040 --> 00:01:15.560 align:middle line:84% Now, it's a deposited oxide, very 00:01:15.560 --> 00:01:17.330 align:middle line:90% different from a grown oxide. 00:01:17.330 --> 00:01:20.060 align:middle line:84% So we're going to talk about how we deposit oxides 00:01:20.060 --> 00:01:22.700 align:middle line:84% by a process called chemical vapor deposition. 00:01:22.700 --> 00:01:24.890 align:middle line:90% Or it might not be oxide. 00:01:24.890 --> 00:01:28.200 align:middle line:84% It could be generally referred to as an interlayer 00:01:28.200 --> 00:01:32.160 align:middle line:84% dielectric in between some of the metal layers, 00:01:32.160 --> 00:01:34.380 align:middle line:84% and that has a low dielectric constant. 00:01:34.380 --> 00:01:37.500 align:middle line:84% And we'll talk about that in a little minute, why we want that. 00:01:37.500 --> 00:01:40.710 align:middle line:84% Other films that are absolutely critical for the MOSFET itself. 00:01:40.710 --> 00:01:41.860 align:middle line:90% We haven't talked about it. 00:01:41.860 --> 00:01:43.710 align:middle line:84% We've been talking about polysilicon gates. 00:01:43.710 --> 00:01:46.648 align:middle line:84% So far, I just draw them in in PowerPoint 00:01:46.648 --> 00:01:47.940 align:middle line:90% just by slapping them in there. 00:01:47.940 --> 00:01:49.650 align:middle line:84% But they don't get slapped in PowerPoint. 00:01:49.650 --> 00:01:50.902 align:middle line:90% They have to be deposited. 00:01:50.902 --> 00:01:52.860 align:middle line:84% So we'll talk a little bit about the deposition 00:01:52.860 --> 00:01:55.020 align:middle line:84% of polycrystalline silicon, which is critical, 00:01:55.020 --> 00:01:56.823 align:middle line:84% the deposition of the materials that 00:01:56.823 --> 00:01:58.740 align:middle line:84% are going to form the spacers, which is either 00:01:58.740 --> 00:02:01.920 align:middle line:84% this bright green silicon nitride, 00:02:01.920 --> 00:02:04.740 align:middle line:84% again, deposited, or it could be low-temperature oxide. 00:02:04.740 --> 00:02:06.630 align:middle line:84% So there are a lot of different materials 00:02:06.630 --> 00:02:08.672 align:middle line:84% that we want to talk about, and how they go down. 00:02:08.672 --> 00:02:11.000 align:middle line:90% 00:02:11.000 --> 00:02:13.880 align:middle line:84% Besides how they go down, there are some requirements, or goals, 00:02:13.880 --> 00:02:17.630 align:middle line:84% if we go onto slide number three, for thin-film deposition. 00:02:17.630 --> 00:02:20.130 align:middle line:84% We need to be able to control the material composition. 00:02:20.130 --> 00:02:27.830 align:middle line:84% So either a stoichiometric SiO2, for example, or a silicon 3 N4 00:02:27.830 --> 00:02:30.020 align:middle line:84% for silicon nitride, or it could be 00:02:30.020 --> 00:02:33.110 align:middle line:84% a oxynitride, a mixture of silicon, oxygen, and nitrogen. 00:02:33.110 --> 00:02:36.170 align:middle line:84% But we need to be able to have a process that reliably controls 00:02:36.170 --> 00:02:38.300 align:middle line:90% the atomic composition. 00:02:38.300 --> 00:02:40.853 align:middle line:84% We would like to have low contamination in general to-- 00:02:40.853 --> 00:02:42.770 align:middle line:84% in order to control the electrical and optical 00:02:42.770 --> 00:02:45.230 align:middle line:90% properties of this material. 00:02:45.230 --> 00:02:47.510 align:middle line:84% And good electrical, and mechanical, 00:02:47.510 --> 00:02:51.160 align:middle line:84% and optical properties, generally. 00:02:51.160 --> 00:02:54.310 align:middle line:84% An obvious-- somewhat obvious thing, but very important, 00:02:54.310 --> 00:02:57.100 align:middle line:84% is we need to control the thickness uniformity across 00:02:57.100 --> 00:03:00.550 align:middle line:84% the wafer and from wafer to wafer so that when we go to etch 00:03:00.550 --> 00:03:04.813 align:middle line:84% it, or whatever, we know what the unit-- we know how-- 00:03:04.813 --> 00:03:06.730 align:middle line:84% what the uniformity is, what the thickness is, 00:03:06.730 --> 00:03:09.100 align:middle line:90% at any given point on the wafer. 00:03:09.100 --> 00:03:11.585 align:middle line:84% There's something called step coverage, which I'll give you 00:03:11.585 --> 00:03:13.210 align:middle line:84% an example in the next couple of slides 00:03:13.210 --> 00:03:15.250 align:middle line:84% of what we mean by step coverage. 00:03:15.250 --> 00:03:20.250 align:middle line:84% We often want conformal deposition. 00:03:20.250 --> 00:03:21.960 align:middle line:84% And we'll give illustrations of that. 00:03:21.960 --> 00:03:24.550 align:middle line:84% We want to be able to fill spaces in between lines. 00:03:24.550 --> 00:03:26.610 align:middle line:84% We don't want to have voids, usually. 00:03:26.610 --> 00:03:30.480 align:middle line:84% And there is a need to planarize films. 00:03:30.480 --> 00:03:33.240 align:middle line:84% That is, create-- we have a topography that 00:03:33.240 --> 00:03:37.110 align:middle line:84% often has a lot of ups-- hills and valleys. 00:03:37.110 --> 00:03:39.300 align:middle line:84% We often need to planarize it to give 00:03:39.300 --> 00:03:41.160 align:middle line:90% a smooth, flat, top surface. 00:03:41.160 --> 00:03:44.940 align:middle line:84% It turns out that certain deposition techniques themselves 00:03:44.940 --> 00:03:46.980 align:middle line:90% tend to be self planarizing. 00:03:46.980 --> 00:03:48.850 align:middle line:84% And we'll talk a little bit about that. 00:03:48.850 --> 00:03:51.270 align:middle line:84% So these are some requirements that people 00:03:51.270 --> 00:03:53.370 align:middle line:84% worry about when they're talking about coming up 00:03:53.370 --> 00:03:57.830 align:middle line:84% with a deposition procedure, or process. 00:03:57.830 --> 00:04:01.130 align:middle line:84% Let's give some examples of these issues, some of the ones 00:04:01.130 --> 00:04:04.730 align:middle line:84% that may be not so obvious here on slide number four. 00:04:04.730 --> 00:04:06.500 align:middle line:90% What is step coverage? 00:04:06.500 --> 00:04:10.470 align:middle line:84% OK, well, step coverage is when we have a step that exists. 00:04:10.470 --> 00:04:13.340 align:middle line:84% So let's say, I already have this step here. 00:04:13.340 --> 00:04:18.079 align:middle line:84% You see this-- there's maybe a material here, this metal line, 00:04:18.079 --> 00:04:20.060 align:middle line:84% and then we have a step going over it. 00:04:20.060 --> 00:04:21.140 align:middle line:90% It could be an oxide. 00:04:21.140 --> 00:04:23.090 align:middle line:84% And now, I want to put a metal down. 00:04:23.090 --> 00:04:25.130 align:middle line:84% So I'm going to either evaporate it, 00:04:25.130 --> 00:04:27.570 align:middle line:84% or I'm going to sputter it, or somehow put the metal down. 00:04:27.570 --> 00:04:30.540 align:middle line:84% But you can see, here, I'm getting very uniform step 00:04:30.540 --> 00:04:31.040 align:middle line:90% coverage. 00:04:31.040 --> 00:04:32.810 align:middle line:84% Regardless of when I go over the step, 00:04:32.810 --> 00:04:37.280 align:middle line:84% I get the same thickness and conformality of the film. 00:04:37.280 --> 00:04:37.910 align:middle line:90% That's good. 00:04:37.910 --> 00:04:39.240 align:middle line:90% That's uniform step coverage. 00:04:39.240 --> 00:04:42.117 align:middle line:84% This is what's called poor step coverage. 00:04:42.117 --> 00:04:42.950 align:middle line:90% What does that mean? 00:04:42.950 --> 00:04:44.590 align:middle line:84% Well, right at the point of the step, 00:04:44.590 --> 00:04:47.350 align:middle line:84% you can see that the film thickness is, 00:04:47.350 --> 00:04:51.280 align:middle line:84% with respect to the surface here, is actually less. 00:04:51.280 --> 00:04:53.132 align:middle line:90% It's thinned down here. 00:04:53.132 --> 00:04:54.340 align:middle line:90% So that's poor step coverage. 00:04:54.340 --> 00:04:56.350 align:middle line:90% It's not conformal. 00:04:56.350 --> 00:04:58.570 align:middle line:84% And so the problem with this, then, 00:04:58.570 --> 00:05:01.300 align:middle line:84% is if your step coverage degrades at some point, 00:05:01.300 --> 00:05:03.700 align:middle line:84% this can be a weak spot in the metal line. 00:05:03.700 --> 00:05:06.620 align:middle line:90% And you could get an open. 00:05:06.620 --> 00:05:09.050 align:middle line:84% So step coverage is an important property 00:05:09.050 --> 00:05:10.590 align:middle line:90% of the type of deposition. 00:05:10.590 --> 00:05:13.760 align:middle line:84% So this would be considered poor step coverage. 00:05:13.760 --> 00:05:15.560 align:middle line:90% How about a filling issue? 00:05:15.560 --> 00:05:18.690 align:middle line:84% OK, well, what we mean by this is 00:05:18.690 --> 00:05:22.280 align:middle line:84% this is an example of metal that has been-- that shows 00:05:22.280 --> 00:05:24.060 align:middle line:90% good filling into a via. 00:05:24.060 --> 00:05:28.790 align:middle line:84% So here, I have a flat surface, I have a deposited oxide, 00:05:28.790 --> 00:05:30.110 align:middle line:90% and I've etched a hole. 00:05:30.110 --> 00:05:33.080 align:middle line:84% And I want to deposit the metal onto the surface 00:05:33.080 --> 00:05:35.280 align:middle line:90% and fill the hole. 00:05:35.280 --> 00:05:38.510 align:middle line:84% So that's good filling of the via. 00:05:38.510 --> 00:05:40.280 align:middle line:84% Here's an example where we were trying-- 00:05:40.280 --> 00:05:43.130 align:middle line:84% we had metal to start with, these two metal lines. 00:05:43.130 --> 00:05:46.010 align:middle line:84% They had a certain height, and width, and spacing. 00:05:46.010 --> 00:05:48.230 align:middle line:84% And we were trying to deposit oxide over them. 00:05:48.230 --> 00:05:50.870 align:middle line:84% And you notice, we did not fill the space completely 00:05:50.870 --> 00:05:52.280 align:middle line:90% in between the two metal lines. 00:05:52.280 --> 00:05:54.480 align:middle line:90% We end up with a void. 00:05:54.480 --> 00:05:56.940 align:middle line:84% So that would be kind of a poor fill. 00:05:56.940 --> 00:05:58.830 align:middle line:90% We have voids. 00:05:58.830 --> 00:06:02.280 align:middle line:84% So we not only have to fill vias, which where we've etched, 00:06:02.280 --> 00:06:04.740 align:middle line:84% this is a via, where I've etched into a film, 00:06:04.740 --> 00:06:08.880 align:middle line:84% but we may have etched features that we're trying to cover up. 00:06:08.880 --> 00:06:11.250 align:middle line:84% And we want to fill the space in between them uniformly. 00:06:11.250 --> 00:06:14.040 align:middle line:90% And then here's an example of-- 00:06:14.040 --> 00:06:17.100 align:middle line:84% here, the third example on the right 00:06:17.100 --> 00:06:19.770 align:middle line:84% is showing, trying to etch, or rather 00:06:19.770 --> 00:06:24.360 align:middle line:84% fill a via where we have poor bottom filling. 00:06:24.360 --> 00:06:27.120 align:middle line:84% Here, we get a nice metal film on top. 00:06:27.120 --> 00:06:31.140 align:middle line:84% The metal is not very well uniformly deposited 00:06:31.140 --> 00:06:31.740 align:middle line:90% on the edges. 00:06:31.740 --> 00:06:33.760 align:middle line:84% And the bottom doesn't fill very well at all. 00:06:33.760 --> 00:06:37.170 align:middle line:84% So there's an example of poor bottom filling. 00:06:37.170 --> 00:06:39.682 align:middle line:84% So this particular methodology, we'll 00:06:39.682 --> 00:06:40.890 align:middle line:90% talk about how this happened. 00:06:40.890 --> 00:06:43.455 align:middle line:84% This probably happened in some kind of evaporation 00:06:43.455 --> 00:06:46.710 align:middle line:84% or some type of line of sight type of deposition. 00:06:46.710 --> 00:06:48.450 align:middle line:84% So depending on how we do the deposition, 00:06:48.450 --> 00:06:52.200 align:middle line:84% we can get different quality of filling. 00:06:52.200 --> 00:06:53.400 align:middle line:90% Those were cartoons. 00:06:53.400 --> 00:06:57.570 align:middle line:84% Here on slide number five, these are some actual micrographs. 00:06:57.570 --> 00:06:59.310 align:middle line:84% These are scanning electron micrographs 00:06:59.310 --> 00:07:01.530 align:middle line:90% taking in an SEM microscope. 00:07:01.530 --> 00:07:04.740 align:middle line:84% I took these directly out of your text. 00:07:04.740 --> 00:07:08.100 align:middle line:84% The one on the left labeled A here 00:07:08.100 --> 00:07:12.220 align:middle line:84% is an example of poor metal step coverage. 00:07:12.220 --> 00:07:17.820 align:middle line:84% This is a sandwich layer, where Thai tungsten was put down, 00:07:17.820 --> 00:07:20.340 align:middle line:84% aluminum, then aluminum, and then Thai tungsten. 00:07:20.340 --> 00:07:23.610 align:middle line:84% So it's a tri-layer stack deposited by sputter deposition 00:07:23.610 --> 00:07:24.970 align:middle line:90% over an oxide step. 00:07:24.970 --> 00:07:30.900 align:middle line:84% So this is the oxide step here right at this point. 00:07:30.900 --> 00:07:32.975 align:middle line:84% And you can see the tri-layer stack here, 00:07:32.975 --> 00:07:35.100 align:middle line:84% it's kind of hard to identify the different layers, 00:07:35.100 --> 00:07:38.310 align:middle line:84% but it has this sort of-- this sort of thickness over here, 00:07:38.310 --> 00:07:39.730 align:middle line:84% and a certain thickness over here. 00:07:39.730 --> 00:07:41.510 align:middle line:90% But it did not cover the step. 00:07:41.510 --> 00:07:44.407 align:middle line:84% And the PowerPoint, unfortunately, this 00:07:44.407 --> 00:07:45.490 align:middle line:90% didn't scan all that well. 00:07:45.490 --> 00:07:47.490 align:middle line:84% But you can see that the layer, the metal layer, 00:07:47.490 --> 00:07:51.350 align:middle line:84% is thinned down where it goes over the step. 00:07:51.350 --> 00:07:54.530 align:middle line:84% On the right-hand side, these are a series of metal lines. 00:07:54.530 --> 00:07:57.290 align:middle line:84% Each one of these is a metal line going into the board. 00:07:57.290 --> 00:08:02.510 align:middle line:84% And oxide was deposited over these metal lines 00:08:02.510 --> 00:08:04.130 align:middle line:90% by chemical vapor deposition. 00:08:04.130 --> 00:08:06.410 align:middle line:84% And there were very narrow spaces between the lines. 00:08:06.410 --> 00:08:09.230 align:middle line:84% The spacing is less than a micron, maybe half micron. 00:08:09.230 --> 00:08:12.680 align:middle line:84% And you can see in the deposition of the oxide, when 00:08:12.680 --> 00:08:15.050 align:middle line:84% it gets-- the spacing gets to be quite narrow, as 00:08:15.050 --> 00:08:16.820 align:middle line:84% in the case between these two metal lines. 00:08:16.820 --> 00:08:18.050 align:middle line:90% You actually get voids. 00:08:18.050 --> 00:08:20.810 align:middle line:84% So the oxide did not fill in here. 00:08:20.810 --> 00:08:23.660 align:middle line:84% When the spacing was wider, as in this case, 00:08:23.660 --> 00:08:25.770 align:middle line:90% you didn't get the voiding. 00:08:25.770 --> 00:08:28.140 align:middle line:84% So when the net-- we get below a certain spacing. 00:08:28.140 --> 00:08:29.940 align:middle line:84% And this is a typical test structure 00:08:29.940 --> 00:08:32.700 align:middle line:84% one could use to figure out, what's 00:08:32.700 --> 00:08:35.400 align:middle line:84% my critical spacing, below which I start to get voids. 00:08:35.400 --> 00:08:37.409 align:middle line:84% And so again, this is a property-- 00:08:37.409 --> 00:08:40.890 align:middle line:84% this voiding will be a strong function of the method by which 00:08:40.890 --> 00:08:43.480 align:middle line:90% the oxide is deposited. 00:08:43.480 --> 00:08:45.850 align:middle line:84% OK, so those are some practical examples 00:08:45.850 --> 00:08:50.020 align:middle line:90% of issues with film deposition. 00:08:50.020 --> 00:08:51.400 align:middle line:90% Let's go on to slide number six. 00:08:51.400 --> 00:08:53.470 align:middle line:84% There is an important concept when 00:08:53.470 --> 00:08:55.540 align:middle line:84% people are doing thin film, describing 00:08:55.540 --> 00:08:58.280 align:middle line:90% thin films, deposition. 00:08:58.280 --> 00:09:00.400 align:middle line:90% And that is called aspect ratio. 00:09:00.400 --> 00:09:02.710 align:middle line:84% We abbreviate that in this class, AR. 00:09:02.710 --> 00:09:05.860 align:middle line:84% And the aspect ratio is just simply defined 00:09:05.860 --> 00:09:07.820 align:middle line:90% as the height of a feature. 00:09:07.820 --> 00:09:10.510 align:middle line:84% So if we have a metal line, it says, height h divided 00:09:10.510 --> 00:09:13.240 align:middle line:90% by the width of the feature, w. 00:09:13.240 --> 00:09:14.830 align:middle line:84% So a meta line, it looks like this 00:09:14.830 --> 00:09:16.930 align:middle line:90% is the aspect ratio, h over w. 00:09:16.930 --> 00:09:20.170 align:middle line:84% For a contact hole, they also speak of an aspect ratio. 00:09:20.170 --> 00:09:22.360 align:middle line:84% It's the height, or the depth, of the hole 00:09:22.360 --> 00:09:26.070 align:middle line:90% divided by its width. 00:09:26.070 --> 00:09:28.590 align:middle line:84% Aspect ratio tells you something about the topography. 00:09:28.590 --> 00:09:31.290 align:middle line:84% In general, a high aspect ratio structure 00:09:31.290 --> 00:09:35.130 align:middle line:84% is more difficult to fabricate than a low aspect ratio one. 00:09:35.130 --> 00:09:37.900 align:middle line:84% For example, it's hard to make a very, 00:09:37.900 --> 00:09:39.840 align:middle line:84% very deep-- imagine this being very deep 00:09:39.840 --> 00:09:41.460 align:middle line:90% and narrow contact hole. 00:09:41.460 --> 00:09:44.850 align:middle line:84% It has a high aspect ratio, say 2 to 1. 00:09:44.850 --> 00:09:47.012 align:middle line:84% And it's harder to etch it for one thing, 00:09:47.012 --> 00:09:49.470 align:middle line:84% and it's going to be harder to fill it, as you can imagine. 00:09:49.470 --> 00:09:51.160 align:middle line:84% As you go deeper and deeper down, 00:09:51.160 --> 00:09:53.950 align:middle line:84% it's going to be harder to access the bottom of the hole. 00:09:53.950 --> 00:09:57.390 align:middle line:84% So in one of the requirements people talk about is, 00:09:57.390 --> 00:10:02.430 align:middle line:84% what kind of aspect ratio can a given deposition technique fill? 00:10:02.430 --> 00:10:05.350 align:middle line:84% And how does it vary with aspect ratio? 00:10:05.350 --> 00:10:07.365 align:middle line:84% So that's just a simple definition, h over w. 00:10:07.365 --> 00:10:10.700 align:middle line:90% 00:10:10.700 --> 00:10:14.141 align:middle line:84% Slide number seven, I took from the-- 00:10:14.141 --> 00:10:18.470 align:middle line:84% your hand-- from the 2003 ITRS, the International Technology 00:10:18.470 --> 00:10:19.560 align:middle line:90% Roadmap on Semiconductors. 00:10:19.560 --> 00:10:21.310 align:middle line:84% Remember, in the beginning of this course, 00:10:21.310 --> 00:10:23.942 align:middle line:84% you read as a homework set, you read a couple of chapters. 00:10:23.942 --> 00:10:25.400 align:middle line:84% Well, I took this from the chapter. 00:10:25.400 --> 00:10:27.200 align:middle line:84% They have a whole chapter on interconnect. 00:10:27.200 --> 00:10:28.970 align:middle line:90% So this is table 81. 00:10:28.970 --> 00:10:31.490 align:middle line:84% And these are interconnect-- predicted 00:10:31.490 --> 00:10:34.710 align:middle line:84% interconnect technology requirements in the near term, 00:10:34.710 --> 00:10:35.210 align:middle line:90% so to speak. 00:10:35.210 --> 00:10:38.930 align:middle line:84% Near term being going up to the year 2009 00:10:38.930 --> 00:10:41.000 align:middle line:90% for microprocessor units. 00:10:41.000 --> 00:10:44.967 align:middle line:84% If you're talking about DRAM or other types of devices, 00:10:44.967 --> 00:10:46.800 align:middle line:84% the requirements are a little bit different. 00:10:46.800 --> 00:10:49.040 align:middle line:84% But this is specifically for microprocessors. 00:10:49.040 --> 00:10:51.540 align:middle line:84% And we've seen these type of charts before. 00:10:51.540 --> 00:10:52.790 align:middle line:90% Let's just go through it here. 00:10:52.790 --> 00:10:54.200 align:middle line:90% The columns are the years. 00:10:54.200 --> 00:10:57.200 align:middle line:90% So 2004, where we are right now. 00:10:57.200 --> 00:11:00.080 align:middle line:84% Remember, this was written a little over a year ago or so. 00:11:00.080 --> 00:11:03.470 align:middle line:84% So this is a prediction going up to 2009. 00:11:03.470 --> 00:11:08.300 align:middle line:84% And each row tells us some kind of a characteristic. 00:11:08.300 --> 00:11:09.892 align:middle line:84% In this case, for interconnect, I 00:11:09.892 --> 00:11:12.600 align:middle line:84% have pointed out with the arrows a couple of key characteristics. 00:11:12.600 --> 00:11:15.450 align:middle line:84% Let's look at the number of metal levels. 00:11:15.450 --> 00:11:18.600 align:middle line:84% In 2004, that was expected on average 00:11:18.600 --> 00:11:21.210 align:middle line:84% for a microprocessor is 10 levels 00:11:21.210 --> 00:11:23.730 align:middle line:90% of circuit wiring of metal. 00:11:23.730 --> 00:11:25.290 align:middle line:90% So that's quite a bit. 00:11:25.290 --> 00:11:27.720 align:middle line:84% And you notice that number is increasing over time 00:11:27.720 --> 00:11:31.230 align:middle line:90% as we go through 2007 going-- 00:11:31.230 --> 00:11:37.290 align:middle line:84% five and seven is 11 metal layers and then going up to 12. 00:11:37.290 --> 00:11:39.190 align:middle line:84% And it's going to continue on beyond that. 00:11:39.190 --> 00:11:41.730 align:middle line:84% So we're getting higher and higher numbers of metal levels. 00:11:41.730 --> 00:11:45.690 align:middle line:84% That's one, which means we're going to have more deposition 00:11:45.690 --> 00:11:48.300 align:middle line:90% steps basically. 00:11:48.300 --> 00:11:52.230 align:middle line:84% Another characteristic that is quoted here is metal one. 00:11:52.230 --> 00:11:55.380 align:middle line:84% The metal layers are numbered, by the way, from 1 up to n. 00:11:55.380 --> 00:11:57.840 align:middle line:84% So in 2004, if you had 10 layers, 00:11:57.840 --> 00:11:59.670 align:middle line:90% they'd be numbered 1 to 10. 00:11:59.670 --> 00:12:02.370 align:middle line:84% So the metal one is the first metal layer. 00:12:02.370 --> 00:12:07.290 align:middle line:84% And you're being told here what the aspect ratio is. 00:12:07.290 --> 00:12:10.260 align:middle line:84% They use an A slash R in the ITRS roadmap. 00:12:10.260 --> 00:12:12.570 align:middle line:84% We just define the aspect ratio as being 00:12:12.570 --> 00:12:15.930 align:middle line:84% the height to the width for the metal line for copper. 00:12:15.930 --> 00:12:19.470 align:middle line:84% And its aspect ratio is somewhere around 1.7. 00:12:19.470 --> 00:12:23.092 align:middle line:84% And you see the aspect ratio is increasing over time. 00:12:23.092 --> 00:12:24.550 align:middle line:84% And in fact, I didn't show it here, 00:12:24.550 --> 00:12:27.420 align:middle line:84% but if you go to the year 2010 and beyond, 00:12:27.420 --> 00:12:30.120 align:middle line:84% it goes up to 2, a little over 2. 00:12:30.120 --> 00:12:36.240 align:middle line:84% So the aspect ratio, we need to make that an increasing number. 00:12:36.240 --> 00:12:39.420 align:middle line:84% And then look at the number-- look at the interlevel metal 00:12:39.420 --> 00:12:42.780 align:middle line:84% insulator dielectric constant, the very bottom row 00:12:42.780 --> 00:12:43.708 align:middle line:90% of this chart. 00:12:43.708 --> 00:12:45.000 align:middle line:90% I actually truncated the chart. 00:12:45.000 --> 00:12:46.530 align:middle line:84% There was a whole bunch of rows in the middle. 00:12:46.530 --> 00:12:48.540 align:middle line:84% And you can see right here where this little discontinuity, 00:12:48.540 --> 00:12:49.260 align:middle line:90% I took it out. 00:12:49.260 --> 00:12:50.677 align:middle line:84% If you want to see the full chart, 00:12:50.677 --> 00:12:52.460 align:middle line:90% you can go to the website. 00:12:52.460 --> 00:12:54.830 align:middle line:84% So look at the dielectric constant. 00:12:54.830 --> 00:12:57.830 align:middle line:84% The bulk dielectric constant of the-- of the insulator 00:12:57.830 --> 00:13:00.230 align:middle line:84% is going to between each layer of metals 00:13:00.230 --> 00:13:01.550 align:middle line:90% is actually going down. 00:13:01.550 --> 00:13:04.010 align:middle line:90% Here it is supposedly in 2003. 00:13:04.010 --> 00:13:05.480 align:middle line:90% It was around three. 00:13:05.480 --> 00:13:07.730 align:middle line:84% And you notice, we get into the yellow region, which 00:13:07.730 --> 00:13:09.950 align:middle line:84% means solutions-- people have some idea of what 00:13:09.950 --> 00:13:12.800 align:middle line:84% the solutions are, but they're not ready for manufacturing. 00:13:12.800 --> 00:13:16.820 align:middle line:84% People want to go to less than 2.7 dielectric constant. 00:13:16.820 --> 00:13:19.340 align:middle line:84% And in the year 2007, they're looking 00:13:19.340 --> 00:13:22.250 align:middle line:84% at-- would like dielectric constants less than 2.4. 00:13:22.250 --> 00:13:24.270 align:middle line:84% And that's part of the red region, 00:13:24.270 --> 00:13:27.010 align:middle line:84% which means nobody knows exactly how to do that. 00:13:27.010 --> 00:13:29.273 align:middle line:84% So just the opposite of the gate insulator. 00:13:29.273 --> 00:13:30.690 align:middle line:84% Remember, in the gate insulator, i 00:13:30.690 --> 00:13:33.090 align:middle line:84% was trying to get higher capacitance between the gate 00:13:33.090 --> 00:13:33.880 align:middle line:90% and the substrate. 00:13:33.880 --> 00:13:35.440 align:middle line:84% So I can get better charge control. 00:13:35.440 --> 00:13:38.130 align:middle line:84% So in the gate insulator that dielectric constant, 00:13:38.130 --> 00:13:39.570 align:middle line:90% we want to go up. 00:13:39.570 --> 00:13:42.750 align:middle line:84% For in the interconnect, we want the dielectric constant 00:13:42.750 --> 00:13:43.322 align:middle line:90% to go down. 00:13:43.322 --> 00:13:44.280 align:middle line:90% We want it to be lower. 00:13:44.280 --> 00:13:47.250 align:middle line:84% And these are the so-called low k dielectrics 00:13:47.250 --> 00:13:49.720 align:middle line:90% that are used in the back end. 00:13:49.720 --> 00:13:51.630 align:middle line:84% And the reason we want that is just simply 00:13:51.630 --> 00:13:54.240 align:middle line:84% if I have a big stack here, if I'm going back 00:13:54.240 --> 00:13:58.770 align:middle line:84% a couple of viewgraphs here to, say, page number two, 00:13:58.770 --> 00:14:00.150 align:middle line:84% I'm only showing a couple layers. 00:14:00.150 --> 00:14:03.570 align:middle line:84% But between this metal layer and this metal layer, 00:14:03.570 --> 00:14:05.820 align:middle line:84% I'd like to have minimum capacitance 00:14:05.820 --> 00:14:09.840 align:middle line:84% between the different lines because the capacitance leads 00:14:09.840 --> 00:14:13.050 align:middle line:90% to RC time delay in the circuit. 00:14:13.050 --> 00:14:16.380 align:middle line:84% And so-- and in fact, the RC delays, the interconnect 00:14:16.380 --> 00:14:18.523 align:middle line:84% delays are very rapidly approaching the delays 00:14:18.523 --> 00:14:19.690 align:middle line:90% associated with the devices. 00:14:19.690 --> 00:14:22.200 align:middle line:84% So making the devices faster doesn't do a whole lot of good 00:14:22.200 --> 00:14:24.340 align:middle line:90% if your interconnect limited. 00:14:24.340 --> 00:14:26.190 align:middle line:84% So there's a trend to get new dielectrics 00:14:26.190 --> 00:14:28.980 align:middle line:84% in here, all these shading here in blue 00:14:28.980 --> 00:14:30.990 align:middle line:84% to come up with new materials that have lower 00:14:30.990 --> 00:14:32.375 align:middle line:90% and lower dielectric constants. 00:14:32.375 --> 00:14:33.750 align:middle line:84% Just the opposite of what we want 00:14:33.750 --> 00:14:35.240 align:middle line:84% to do with the gate insulator here. 00:14:35.240 --> 00:14:37.710 align:middle line:84% We want to make a higher and higher one. 00:14:37.710 --> 00:14:40.760 align:middle line:84% So we're going to find new materials coming all the time. 00:14:40.760 --> 00:14:42.738 align:middle line:84% So let me skip back to that on slide seven. 00:14:42.738 --> 00:14:44.530 align:middle line:84% And again, if you want to see more of this, 00:14:44.530 --> 00:14:45.760 align:middle line:90% you just go to the ITRS. 00:14:45.760 --> 00:14:47.020 align:middle line:90% And you've been there before. 00:14:47.020 --> 00:14:50.495 align:middle line:84% And you just pick the chapter on interconnect. 00:14:50.495 --> 00:14:51.870 align:middle line:84% So besides interconnect though, I 00:14:51.870 --> 00:14:54.210 align:middle line:84% don't want to give you the idea that thin films are only 00:14:54.210 --> 00:14:55.200 align:middle line:90% used in the back end. 00:14:55.200 --> 00:14:56.730 align:middle line:84% They're actually used in front end. 00:14:56.730 --> 00:14:58.570 align:middle line:84% And that's what I'm going to emphasize in this course. 00:14:58.570 --> 00:14:59.400 align:middle line:90% So I chopped off-- 00:14:59.400 --> 00:15:01.650 align:middle line:84% I took the diagram I just showed you and I chopped off 00:15:01.650 --> 00:15:06.030 align:middle line:84% the back end to mean what we're talking about sort of front end. 00:15:06.030 --> 00:15:08.430 align:middle line:84% And there are maybe three different types 00:15:08.430 --> 00:15:11.880 align:middle line:84% of critical thin films that we're going to emphasize, 00:15:11.880 --> 00:15:13.960 align:middle line:84% because this is a course on front-end processing. 00:15:13.960 --> 00:15:16.260 align:middle line:84% So we're going to emphasize the deposition of silicon, 00:15:16.260 --> 00:15:18.270 align:middle line:84% either single crystal by epitaxy, 00:15:18.270 --> 00:15:20.820 align:middle line:90% or polycrystalline by LPCVD. 00:15:20.820 --> 00:15:22.860 align:middle line:84% That's-- and here's the poly gate. 00:15:22.860 --> 00:15:26.220 align:middle line:84% Silicon dioxide, which is also called 00:15:26.220 --> 00:15:29.760 align:middle line:84% LPCVD oxide, Low-Pressure Chemical Vapor Deposition, 00:15:29.760 --> 00:15:32.260 align:middle line:84% or low-temperature oxide, and silicon nitride, 00:15:32.260 --> 00:15:33.630 align:middle line:90% which I've shown here in green. 00:15:33.630 --> 00:15:36.270 align:middle line:84% The spacers are often made of nitride these days, 00:15:36.270 --> 00:15:39.075 align:middle line:84% or combinations of nitride and LTL. 00:15:39.075 --> 00:15:40.950 align:middle line:84% So that's what we'll emphasize in this class. 00:15:40.950 --> 00:15:42.600 align:middle line:84% All the rest above this, which I've 00:15:42.600 --> 00:15:46.260 align:middle line:84% erased, if you take 6.773, that hasn't been offered recently 00:15:46.260 --> 00:15:49.140 align:middle line:84% because Professor Reif is department head of EECS. 00:15:49.140 --> 00:15:50.500 align:middle line:90% So he's been kind of busy. 00:15:50.500 --> 00:15:54.130 align:middle line:84% But that course deals with all the types of thin films 00:15:54.130 --> 00:15:55.620 align:middle line:90% that are kind of above here. 00:15:55.620 --> 00:15:58.120 align:middle line:84% I'll talk a little bit about deposition of metals and things 00:15:58.120 --> 00:15:58.630 align:middle line:90% like that. 00:15:58.630 --> 00:16:03.130 align:middle line:84% But we won't emphasize it as much as that class does. 00:16:03.130 --> 00:16:04.870 align:middle line:84% OK, so let's go on to slide number nine. 00:16:04.870 --> 00:16:08.020 align:middle line:84% And we'll just talk a little bit, historically, 00:16:08.020 --> 00:16:10.300 align:middle line:84% about how some of these films are deposited. 00:16:10.300 --> 00:16:13.480 align:middle line:84% And the films we're going to emphasize initially today 00:16:13.480 --> 00:16:15.790 align:middle line:84% will be a deposition of silicon, either 00:16:15.790 --> 00:16:17.890 align:middle line:90% polycrystalline or epitaxially. 00:16:17.890 --> 00:16:21.670 align:middle line:84% So there's really two main types of methods 00:16:21.670 --> 00:16:23.020 align:middle line:90% that are used in CMOS. 00:16:23.020 --> 00:16:24.820 align:middle line:90% And you'll hear about them. 00:16:24.820 --> 00:16:27.190 align:middle line:84% And one thing I want to apologize 00:16:27.190 --> 00:16:29.223 align:middle line:84% for right off the bat is in thin films, 00:16:29.223 --> 00:16:30.640 align:middle line:84% one of the things I really dislike 00:16:30.640 --> 00:16:33.970 align:middle line:84% is the number of acronyms is really kind of ridiculous. 00:16:33.970 --> 00:16:37.150 align:middle line:84% And you'll see that as we go through, just this first bullet 00:16:37.150 --> 00:16:38.770 align:middle line:90% here, number one. 00:16:38.770 --> 00:16:41.800 align:middle line:84% So the first type of deposition method, and very common, 00:16:41.800 --> 00:16:44.210 align:middle line:84% is called chemical vapor deposition. 00:16:44.210 --> 00:16:47.500 align:middle line:84% And it is just like the name sounds. 00:16:47.500 --> 00:16:50.260 align:middle line:84% You have a vapor of different gases, 00:16:50.260 --> 00:16:54.100 align:middle line:84% and you deposit from the vapor phase a film on the wafer. 00:16:54.100 --> 00:16:55.600 align:middle line:90% And there are different types. 00:16:55.600 --> 00:16:58.330 align:middle line:84% People call atmospheric pressure, CVD, 00:16:58.330 --> 00:17:00.280 align:middle line:90% and that's abbreviated APCVD. 00:17:00.280 --> 00:17:05.390 align:middle line:84% Low-pressure CVD, which we'll talk about, abbreviated LPCVD. 00:17:05.390 --> 00:17:06.859 align:middle line:90% Now, we get a little crazy. 00:17:06.859 --> 00:17:09.230 align:middle line:84% There's also a plasma-enhanced types of CVD. 00:17:09.230 --> 00:17:13.790 align:middle line:84% So PECVD, where we use not only thermal energy, but we 00:17:13.790 --> 00:17:17.300 align:middle line:84% use plasma energy to break up the constituents 00:17:17.300 --> 00:17:19.190 align:middle line:90% and cause the deposition. 00:17:19.190 --> 00:17:22.710 align:middle line:84% And there's also high-density plasma CVD, or HDPCVD. 00:17:22.710 --> 00:17:25.099 align:middle line:84% So it goes on and on, the number of acronyms that 00:17:25.099 --> 00:17:28.900 align:middle line:90% are all associated with CVD. 00:17:28.900 --> 00:17:33.290 align:middle line:84% PVD, not-- maybe not so many in acronyms, but this stands for-- 00:17:33.290 --> 00:17:36.110 align:middle line:84% so the second major method is called physical vapor 00:17:36.110 --> 00:17:37.460 align:middle line:90% deposition. 00:17:37.460 --> 00:17:40.080 align:middle line:84% This is just as distinguished from chemical. 00:17:40.080 --> 00:17:42.170 align:middle line:84% It's primarily a physical process. 00:17:42.170 --> 00:17:44.500 align:middle line:84% So there's not as much chemistry involved. 00:17:44.500 --> 00:17:46.960 align:middle line:84% And the physical processes that people use, 00:17:46.960 --> 00:17:50.470 align:middle line:84% you may be familiar with, is thermal evaporation, which we'll 00:17:50.470 --> 00:17:52.300 align:middle line:90% talk about, and sputtering. 00:17:52.300 --> 00:17:54.595 align:middle line:84% These are primarily physical, have 00:17:54.595 --> 00:17:58.790 align:middle line:84% less chemical sort of-- less chemistry associated with them. 00:17:58.790 --> 00:18:01.420 align:middle line:84% So let's talk about CVD here first. 00:18:01.420 --> 00:18:06.130 align:middle line:84% This is a very old-fashioned sort of classic diagram 00:18:06.130 --> 00:18:09.550 align:middle line:84% of an example of an atmospheric cold wall, 00:18:09.550 --> 00:18:11.650 align:middle line:84% or cool wall system, that might have 00:18:11.650 --> 00:18:14.740 align:middle line:84% been used for a number of years to grow epitaxial silicon. 00:18:14.740 --> 00:18:17.680 align:middle line:84% And I put it here in quotations, "old fashioned" 00:18:17.680 --> 00:18:22.023 align:middle line:84% because it's pretty far from what modern epitaxial reactors 00:18:22.023 --> 00:18:22.690 align:middle line:90% look like today. 00:18:22.690 --> 00:18:24.520 align:middle line:84% And I have some examples of modern reactors 00:18:24.520 --> 00:18:25.490 align:middle line:90% in the next few slides. 00:18:25.490 --> 00:18:26.750 align:middle line:90% But just to give you an idea. 00:18:26.750 --> 00:18:27.820 align:middle line:90% The basic idea was this. 00:18:27.820 --> 00:18:30.470 align:middle line:84% You had a series of different gas lines. 00:18:30.470 --> 00:18:33.730 align:middle line:84% So here is, for example, hydrogen 00:18:33.730 --> 00:18:35.570 align:middle line:90% with diborane or phosphine. 00:18:35.570 --> 00:18:40.090 align:middle line:84% These could be used for dopants for boron and phosphorus doping. 00:18:40.090 --> 00:18:44.020 align:middle line:84% You had carrier gases like argon and hydrogen. Hydrogen chloride 00:18:44.020 --> 00:18:49.060 align:middle line:84% could be used to etch a-- etch deposits off the courts. 00:18:49.060 --> 00:18:51.100 align:middle line:84% And here's a liquid source for silicon, 00:18:51.100 --> 00:18:54.190 align:middle line:90% silicon tetrachloride, SiCl4. 00:18:54.190 --> 00:18:55.930 align:middle line:84% Maybe hydrogen was bubbled through that 00:18:55.930 --> 00:18:59.690 align:middle line:84% and transported that vapor into the chamber. 00:18:59.690 --> 00:19:01.600 align:middle line:84% The silicon wafers typically used 00:19:01.600 --> 00:19:03.850 align:middle line:84% to sit on a big block of graphite. 00:19:03.850 --> 00:19:07.085 align:middle line:84% And the graphite was heated not by a nichrome heater, 00:19:07.085 --> 00:19:08.710 align:middle line:84% but it was heated by induction heating. 00:19:08.710 --> 00:19:10.960 align:middle line:90% So there were RF coils. 00:19:10.960 --> 00:19:13.780 align:middle line:84% And this is why it's called cold wall, because these RF coils, 00:19:13.780 --> 00:19:17.435 align:middle line:84% the RF energy is not absorbed by the quartz, 00:19:17.435 --> 00:19:18.310 align:middle line:90% just by the graphite. 00:19:18.310 --> 00:19:19.930 align:middle line:84% So just the graphite gets hot here. 00:19:19.930 --> 00:19:22.660 align:middle line:84% And the hot graphite then heats the wafers. 00:19:22.660 --> 00:19:24.700 align:middle line:90% The chemical vapor passes over. 00:19:24.700 --> 00:19:28.120 align:middle line:84% And there's decomposition on the surfaces we'll talk about. 00:19:28.120 --> 00:19:31.217 align:middle line:84% And you grow the epitaxial silicon. 00:19:31.217 --> 00:19:33.550 align:middle line:84% That's a-- that's a relatively old-fashioned atmospheric 00:19:33.550 --> 00:19:36.310 align:middle line:90% reactor. 00:19:36.310 --> 00:19:37.690 align:middle line:90% Let's go on to slide number 10. 00:19:37.690 --> 00:19:41.300 align:middle line:84% Now, this is an example of something that you will find. 00:19:41.300 --> 00:19:42.733 align:middle line:84% That old epi reactor, you probably 00:19:42.733 --> 00:19:44.150 align:middle line:84% won't find in too many fabs today. 00:19:44.150 --> 00:19:45.500 align:middle line:90% Maybe a few of the bipolar-- 00:19:45.500 --> 00:19:46.460 align:middle line:90% old bipolar fabs. 00:19:46.460 --> 00:19:48.650 align:middle line:84% But this is a reactor that you will 00:19:48.650 --> 00:19:51.600 align:middle line:84% find in most fabs, something very much like it. 00:19:51.600 --> 00:19:55.340 align:middle line:84% It's called low-pressure hot wall system, or LPCVD. 00:19:55.340 --> 00:19:58.910 align:middle line:84% And it's used pretty much every day now for the deposition 00:19:58.910 --> 00:20:01.550 align:middle line:84% of polycrystalline silicon that is used to form the gate, 00:20:01.550 --> 00:20:04.730 align:middle line:90% or for amorphous silicon. 00:20:04.730 --> 00:20:08.030 align:middle line:84% And-- or even the silicon dioxide. 00:20:08.030 --> 00:20:09.200 align:middle line:90% This type of setup. 00:20:09.200 --> 00:20:12.830 align:middle line:84% And what it is is a regular furnace 00:20:12.830 --> 00:20:14.240 align:middle line:90% that is resistively heated. 00:20:14.240 --> 00:20:15.990 align:middle line:90% So it's a hot wall system. 00:20:15.990 --> 00:20:18.140 align:middle line:84% So the entire quartz tube comes up 00:20:18.140 --> 00:20:19.865 align:middle line:84% to the temperature of deposition. 00:20:19.865 --> 00:20:23.910 align:middle line:84% Could be 600-- anywhere from 400 to 700, something in that range. 00:20:23.910 --> 00:20:25.760 align:middle line:90% So everything is hot. 00:20:25.760 --> 00:20:27.260 align:middle line:84% And you notice what's very different 00:20:27.260 --> 00:20:28.670 align:middle line:90% from the previous diagram. 00:20:28.670 --> 00:20:31.010 align:middle line:84% I have some kind of a quartz holder here. 00:20:31.010 --> 00:20:33.230 align:middle line:90% And the wafers are standing up. 00:20:33.230 --> 00:20:34.590 align:middle line:90% They're not sitting flat. 00:20:34.590 --> 00:20:36.090 align:middle line:84% And there's a lot of wafers in here. 00:20:36.090 --> 00:20:37.445 align:middle line:90% It could be 25. 00:20:37.445 --> 00:20:38.720 align:middle line:90% Could be a hundred. 00:20:38.720 --> 00:20:43.235 align:middle line:84% But they're stacked very, very close to each other. 00:20:43.235 --> 00:20:45.290 align:middle line:84% And so you can do batch processing. 00:20:45.290 --> 00:20:47.273 align:middle line:90% Lots of wafers at once. 00:20:47.273 --> 00:20:49.940 align:middle line:84% And the whole reason you can do this, and we'll talk about that, 00:20:49.940 --> 00:20:53.120 align:middle line:84% is because it turns out the mean-free path is quite 00:20:53.120 --> 00:20:54.380 align:middle line:90% long when the pressure is low. 00:20:54.380 --> 00:20:56.210 align:middle line:84% So the gas can get-- effectively, 00:20:56.210 --> 00:20:58.980 align:middle line:84% the reactants can be transported in between the wafers 00:20:58.980 --> 00:21:00.240 align:middle line:90% very efficiently. 00:21:00.240 --> 00:21:03.540 align:middle line:84% So there's no need to spread them out over a large susceptor, 00:21:03.540 --> 00:21:06.470 align:middle line:84% as we had to do in the case of an atmospheric process 00:21:06.470 --> 00:21:07.970 align:middle line:84% like this, where we can only get two 00:21:07.970 --> 00:21:11.010 align:middle line:90% or three wafers in at a time. 00:21:11.010 --> 00:21:15.440 align:middle line:84% In this low-pressure case, you can put in quite a few wafers. 00:21:15.440 --> 00:21:18.300 align:middle line:84% The pressure is kept low in this case by a vacuum pump. 00:21:18.300 --> 00:21:20.670 align:middle line:84% So the quartz tube has an opening 00:21:20.670 --> 00:21:23.310 align:middle line:84% in the front, which is sealed by an O ring seal 00:21:23.310 --> 00:21:24.660 align:middle line:90% and a stainless steel plate. 00:21:24.660 --> 00:21:27.090 align:middle line:84% So that closes the quartz, and you 00:21:27.090 --> 00:21:30.240 align:middle line:84% can suck on it with this pump, and pull a vacuum. 00:21:30.240 --> 00:21:32.047 align:middle line:84% And you have a couple of source gases 00:21:32.047 --> 00:21:33.630 align:middle line:84% that-- with mass flow controllers that 00:21:33.630 --> 00:21:35.940 align:middle line:84% control the flow of whatever it is you're using. 00:21:35.940 --> 00:21:38.080 align:middle line:84% Could be silane, could be dichloro, or whatever. 00:21:38.080 --> 00:21:38.580 align:middle line:90% Yeah. 00:21:38.580 --> 00:21:42.490 align:middle line:84% AUDIENCE: Why is it that when the mean-free path is long, 00:21:42.490 --> 00:21:44.590 align:middle line:84% do you get the atoms to go [INAUDIBLE]?? 00:21:44.590 --> 00:21:46.340 align:middle line:84% JUDY HOYT: We're going to talk about that. 00:21:46.340 --> 00:21:50.590 align:middle line:84% But basically, if the mean-free path is sufficiently long, 00:21:50.590 --> 00:21:53.230 align:middle line:90% what it means is before-- 00:21:53.230 --> 00:21:55.240 align:middle line:84% the atom can go, or the molecule can go, 00:21:55.240 --> 00:21:58.480 align:middle line:84% a long distance before it hits a wafer and would stick. 00:21:58.480 --> 00:22:00.092 align:middle line:84% The mean-free path is very short. 00:22:00.092 --> 00:22:02.050 align:middle line:84% And I would have put it in a reactor like this, 00:22:02.050 --> 00:22:05.295 align:middle line:84% you'd get deposition all on the tips of the wafers right 00:22:05.295 --> 00:22:05.920 align:middle line:90% on these edges. 00:22:05.920 --> 00:22:07.540 align:middle line:84% But it wouldn't get down in between 00:22:07.540 --> 00:22:10.720 align:middle line:84% because the mean-free path tells you how often is it 00:22:10.720 --> 00:22:12.280 align:middle line:90% that you have a collision. 00:22:12.280 --> 00:22:15.575 align:middle line:84% If that mean-free path is very short, 00:22:15.575 --> 00:22:17.200 align:middle line:84% basically, you're going to-- the chance 00:22:17.200 --> 00:22:19.428 align:middle line:84% of you having a collision with a wafer is quite high 00:22:19.428 --> 00:22:20.470 align:middle line:90% and you'll tend to react. 00:22:20.470 --> 00:22:22.652 align:middle line:84% AUDIENCE: Is it a collision of particles? 00:22:22.652 --> 00:22:24.360 align:middle line:84% JUDY HOYT: It's a collision of particles. 00:22:24.360 --> 00:22:25.050 align:middle line:90% Right. 00:22:25.050 --> 00:22:28.650 align:middle line:84% But it also relates to the interaction 00:22:28.650 --> 00:22:32.760 align:middle line:84% of those particles with whatever you have in the reactor 00:22:32.760 --> 00:22:33.990 align:middle line:90% with the geometry. 00:22:33.990 --> 00:22:37.230 align:middle line:84% AUDIENCE: It means [INAUDIBLE] to have 00:22:37.230 --> 00:22:39.910 align:middle line:90% more shielding [INAUDIBLE]? 00:22:39.910 --> 00:22:41.245 align:middle line:90% JUDY HOYT: Well, it's not-- 00:22:41.245 --> 00:22:43.000 align:middle line:90% this is a chemical process now. 00:22:43.000 --> 00:22:45.080 align:middle line:84% Again, when we talk-- when we kind of go through this process, 00:22:45.080 --> 00:22:46.372 align:middle line:90% it may become a little obvious. 00:22:46.372 --> 00:22:47.320 align:middle line:90% It's not evaporation. 00:22:47.320 --> 00:22:50.150 align:middle line:84% You still have the gas flowing and getting in between. 00:22:50.150 --> 00:22:52.870 align:middle line:84% So it's still a gas phase type of process. 00:22:52.870 --> 00:22:56.620 align:middle line:90% We're not evaporating onto it. 00:22:56.620 --> 00:23:00.835 align:middle line:84% So the gas can still flow here in between each of these wafers. 00:23:00.835 --> 00:23:03.190 align:middle line:84% And the pressure is low enough that the reactants 00:23:03.190 --> 00:23:04.270 align:middle line:90% can get in between. 00:23:04.270 --> 00:23:06.820 align:middle line:84% And you can get nice uniform deposition. 00:23:06.820 --> 00:23:11.888 align:middle line:84% That's not the case if you do at atmospheric pressure 00:23:11.888 --> 00:23:13.930 align:middle line:84% where you need to have the wafers well separated, 00:23:13.930 --> 00:23:15.680 align:middle line:84% because you develop boundary-layer effects 00:23:15.680 --> 00:23:18.380 align:middle line:90% and things like that. 00:23:18.380 --> 00:23:20.900 align:middle line:84% OK, hopefully, that'll become a little more obvious 00:23:20.900 --> 00:23:25.220 align:middle line:84% as we go on for the rest of this lecture and the next. 00:23:25.220 --> 00:23:27.350 align:middle line:84% OK, what else about this reactor you should know? 00:23:27.350 --> 00:23:28.670 align:middle line:90% Remember, it's a hot wall. 00:23:28.670 --> 00:23:30.890 align:middle line:84% That's-- the problem with that is you get deposition 00:23:30.890 --> 00:23:33.530 align:middle line:84% of whatever you're depositing on the wafers also deposits 00:23:33.530 --> 00:23:35.390 align:middle line:90% on the walls. 00:23:35.390 --> 00:23:37.550 align:middle line:84% That can be a part-- give you particulate problems. 00:23:37.550 --> 00:23:39.680 align:middle line:84% Because you can eventually develop so much silicon, 00:23:39.680 --> 00:23:42.020 align:middle line:84% develop deposit on the walls, it's-- 00:23:42.020 --> 00:23:43.635 align:middle line:90% it'll start to flake. 00:23:43.635 --> 00:23:44.760 align:middle line:90% And then you get particles. 00:23:44.760 --> 00:23:50.060 align:middle line:84% So LPCVD is notorious for having particulate issues. 00:23:50.060 --> 00:23:52.820 align:middle line:84% So that's kind of a basic characteristics 00:23:52.820 --> 00:23:55.390 align:middle line:90% of LPCVD reactor. 00:23:55.390 --> 00:23:57.700 align:middle line:84% Let's give an example of two types of deposition. 00:23:57.700 --> 00:24:00.780 align:middle line:90% Here's on slide number 11. 00:24:00.780 --> 00:24:05.130 align:middle line:84% The first example was epitaxial silicon, single-crystal growth. 00:24:05.130 --> 00:24:08.190 align:middle line:84% And I showed you a picture of a cold wall atmospheric pressure 00:24:08.190 --> 00:24:09.180 align:middle line:90% system. 00:24:09.180 --> 00:24:12.150 align:middle line:84% This particular equation, equation number one, 00:24:12.150 --> 00:24:14.310 align:middle line:84% is, again, a little bit old fashioned 00:24:14.310 --> 00:24:17.715 align:middle line:84% using silicon tet, silicon tetrachloride. 00:24:17.715 --> 00:24:20.010 align:middle line:84% And it reacts at high temperature, 00:24:20.010 --> 00:24:25.140 align:middle line:84% and can be decomposed into silicon solid plus HCl 00:24:25.140 --> 00:24:27.840 align:middle line:90% gas, which has evolved. 00:24:27.840 --> 00:24:30.870 align:middle line:84% You can also more commonly today, rather than silicon tet, 00:24:30.870 --> 00:24:33.540 align:middle line:84% people typically use silane, SiH4. 00:24:33.540 --> 00:24:35.880 align:middle line:84% Again, in the gas phase at high temperature, 00:24:35.880 --> 00:24:37.680 align:middle line:84% it can react to form solid silicon 00:24:37.680 --> 00:24:41.580 align:middle line:84% and give and evolve off hydrogen. 00:24:41.580 --> 00:24:45.510 align:middle line:84% So this might be for epitaxial growth. 00:24:45.510 --> 00:24:47.910 align:middle line:84% The second example here, equation number three 00:24:47.910 --> 00:24:50.290 align:middle line:84% shows the deposition of amorphous silicon dioxide. 00:24:50.290 --> 00:24:53.160 align:middle line:84% So this is that famous LTO that we've been talking about, 00:24:53.160 --> 00:24:54.990 align:middle line:90% low-temperature oxide. 00:24:54.990 --> 00:24:57.810 align:middle line:84% That is deposited very commonly in a hot-walled low-pressure 00:24:57.810 --> 00:25:01.220 align:middle line:84% system, just like what I just showed on page 10. 00:25:01.220 --> 00:25:07.920 align:middle line:84% This is exactly a low-pressure CVD oxide, or LTO furnace. 00:25:07.920 --> 00:25:11.970 align:middle line:84% And the way the LTO is made is typically with silane, 00:25:11.970 --> 00:25:15.840 align:middle line:84% a gaseous silane combined with oxygen, usually 00:25:15.840 --> 00:25:18.750 align:middle line:84% in the temperature range of 400 to 500, something like that. 00:25:18.750 --> 00:25:20.560 align:middle line:90% Decomposes on the wafer surface. 00:25:20.560 --> 00:25:21.690 align:middle line:90% Both of these decompose. 00:25:21.690 --> 00:25:24.960 align:middle line:84% And they react to form SiO2 in the solid phase, 00:25:24.960 --> 00:25:26.850 align:middle line:84% and evolved hydrogen. And notice, 00:25:26.850 --> 00:25:29.310 align:middle line:84% because you are putting in silane, 00:25:29.310 --> 00:25:32.210 align:middle line:84% you don't have to consume any silicon on the wafer. 00:25:32.210 --> 00:25:33.210 align:middle line:90% So that's the advantage. 00:25:33.210 --> 00:25:37.080 align:middle line:84% If you need to put down a film, and there's no silicon exposed, 00:25:37.080 --> 00:25:38.730 align:middle line:90% then you use this low-- 00:25:38.730 --> 00:25:41.610 align:middle line:90% this LPCVD process to get-- 00:25:41.610 --> 00:25:44.450 align:middle line:90% put down an oxide. 00:25:44.450 --> 00:25:47.530 align:middle line:84% OK, there's a couple of examples. 00:25:47.530 --> 00:25:50.103 align:middle line:84% So let's look at this process now on slide number 12 00:25:50.103 --> 00:25:51.270 align:middle line:90% a little bit more carefully. 00:25:51.270 --> 00:25:54.630 align:middle line:84% And here, maybe we'll see how the pressure 00:25:54.630 --> 00:25:57.390 align:middle line:90% comes to play in all this. 00:25:57.390 --> 00:26:01.170 align:middle line:84% Slide number 12 talks about atmospheric pressure, APCVD. 00:26:01.170 --> 00:26:04.440 align:middle line:84% And what it is is, schematically, this 00:26:04.440 --> 00:26:06.270 align:middle line:90% is the top wall of your reactor. 00:26:06.270 --> 00:26:07.020 align:middle line:90% Could be quartz. 00:26:07.020 --> 00:26:08.970 align:middle line:90% Here's the bottom wall. 00:26:08.970 --> 00:26:10.720 align:middle line:84% Here's a graphite susceptor, which is hot. 00:26:10.720 --> 00:26:12.520 align:middle line:84% And there's a wafer sitting on it. 00:26:12.520 --> 00:26:14.590 align:middle line:84% And there's a gas stream up here above. 00:26:14.590 --> 00:26:17.290 align:middle line:84% Some distance is the gas is flowing 00:26:17.290 --> 00:26:19.225 align:middle line:90% at a reasonably high velocity. 00:26:19.225 --> 00:26:22.690 align:middle line:84% And steps one through seven here are the different steps 00:26:22.690 --> 00:26:25.690 align:middle line:84% that people have identified as being involved in chemical vapor 00:26:25.690 --> 00:26:26.230 align:middle line:90% deposition. 00:26:26.230 --> 00:26:29.120 align:middle line:84% So notice how it's not evaporation. 00:26:29.120 --> 00:26:31.510 align:middle line:84% So the first step, we take the reactants 00:26:31.510 --> 00:26:34.210 align:middle line:84% that are-- that have to be transported from the main gas 00:26:34.210 --> 00:26:36.548 align:middle line:84% stream into the deposition region. 00:26:36.548 --> 00:26:38.840 align:middle line:84% So they're coming in the reactor over here on the left. 00:26:38.840 --> 00:26:40.382 align:middle line:84% They have to get over near the wafer. 00:26:40.382 --> 00:26:41.410 align:middle line:90% OK, that's obvious. 00:26:41.410 --> 00:26:43.870 align:middle line:84% It's not usually rate limiting in any way. 00:26:43.870 --> 00:26:46.450 align:middle line:84% Step number two is the transport from the main gas 00:26:46.450 --> 00:26:49.350 align:middle line:84% stream through what's called a boundary layer to the wafer 00:26:49.350 --> 00:26:49.850 align:middle line:90% surface. 00:26:49.850 --> 00:26:52.150 align:middle line:84% Now, a boundary layer exists whenever 00:26:52.150 --> 00:26:54.520 align:middle line:84% you have a static surface and a gas 00:26:54.520 --> 00:26:56.080 align:middle line:90% or some fluid flowing above it. 00:26:56.080 --> 00:26:57.640 align:middle line:84% And the boundary layer is a layer 00:26:57.640 --> 00:27:00.010 align:middle line:84% in which the velocity of the particles 00:27:00.010 --> 00:27:02.170 align:middle line:84% actually goes from the velocities in the center 00:27:02.170 --> 00:27:03.550 align:middle line:90% of the tube down to 0. 00:27:03.550 --> 00:27:06.820 align:middle line:84% Because right at the wafer's surface, the velocity is 0. 00:27:06.820 --> 00:27:10.790 align:middle line:84% So you have to somehow get through the stagnant layer. 00:27:10.790 --> 00:27:12.620 align:middle line:84% And so transport through the boundary layer 00:27:12.620 --> 00:27:14.390 align:middle line:84% is one of the critical processes. 00:27:14.390 --> 00:27:18.520 align:middle line:84% And it will be rate limiting in certain types of reactors. 00:27:18.520 --> 00:27:20.797 align:middle line:84% Then there are three surface processes here. 00:27:20.797 --> 00:27:22.630 align:middle line:84% You could lump them all together if you want 00:27:22.630 --> 00:27:24.490 align:middle line:84% and just call them surface processes. 00:27:24.490 --> 00:27:26.680 align:middle line:84% But they are called out individually here. 00:27:26.680 --> 00:27:29.260 align:middle line:84% Number three is the adsorption of the reactant 00:27:29.260 --> 00:27:31.600 align:middle line:90% onto the wafer surface. 00:27:31.600 --> 00:27:34.960 align:middle line:84% Four is the surface reaction itself, including a chemical 00:27:34.960 --> 00:27:38.110 align:middle line:84% decomposition, like what I just showed you, a reaction, if we 00:27:38.110 --> 00:27:39.730 align:middle line:90% go back to slide 11. 00:27:39.730 --> 00:27:41.360 align:middle line:84% This is what I mean by that step. 00:27:41.360 --> 00:27:44.500 align:middle line:84% So it's the actual decomposition of the gas on the surface 00:27:44.500 --> 00:27:46.060 align:middle line:90% into to-- 00:27:46.060 --> 00:27:49.610 align:middle line:84% to crack the gas, or decompose it. 00:27:49.610 --> 00:27:53.685 align:middle line:84% Could also be a surface migration involved 00:27:53.685 --> 00:27:58.770 align:middle line:84% in the surface in step four, or attachment to kinks and ledges. 00:27:58.770 --> 00:28:01.110 align:middle line:84% So there are a number of surface processes that happen. 00:28:01.110 --> 00:28:04.800 align:middle line:84% And then step number five is desorption of the byproducts. 00:28:04.800 --> 00:28:06.900 align:middle line:84% And then we have to transport the byproducts 00:28:06.900 --> 00:28:08.430 align:middle line:90% through the boundary layer. 00:28:08.430 --> 00:28:11.250 align:middle line:84% And then finally, transport of that out of the reactor, 00:28:11.250 --> 00:28:13.860 align:middle line:90% out of the deposition zone. 00:28:13.860 --> 00:28:16.260 align:middle line:84% And I've highlighted here in red steps two 00:28:16.260 --> 00:28:19.740 align:middle line:84% through five because they're most important in determining 00:28:19.740 --> 00:28:21.310 align:middle line:90% the growth rate. 00:28:21.310 --> 00:28:24.870 align:middle line:84% In fact, the two rate-limiting steps tend to be step two, 00:28:24.870 --> 00:28:27.630 align:middle line:84% which is transport the reactants through the boundary layer, 00:28:27.630 --> 00:28:28.500 align:middle line:90% through the-- 00:28:28.500 --> 00:28:31.980 align:middle line:84% which is usually typically a diffusion type of process, 00:28:31.980 --> 00:28:34.710 align:middle line:84% and a reaction-- the second rate-limiting step 00:28:34.710 --> 00:28:36.210 align:middle line:84% is the surface processes themselves, 00:28:36.210 --> 00:28:37.502 align:middle line:90% the chemical reactions usually. 00:28:37.502 --> 00:28:40.240 align:middle line:90% 00:28:40.240 --> 00:28:43.900 align:middle line:84% OK, so let's go on to slide 13 and talk about a way 00:28:43.900 --> 00:28:48.220 align:middle line:84% to quantitatively come up with a very simple quantitative model. 00:28:48.220 --> 00:28:50.292 align:middle line:84% Now, I've turned everything by 90 degrees on you 00:28:50.292 --> 00:28:52.000 align:middle line:84% because that's what was done in the text. 00:28:52.000 --> 00:28:53.290 align:middle line:90% I took this from your text. 00:28:53.290 --> 00:28:56.030 align:middle line:84% So now, be a little more careful here. 00:28:56.030 --> 00:28:57.950 align:middle line:84% The silicon wafer is now sitting vertically. 00:28:57.950 --> 00:28:59.650 align:middle line:84% It doesn't really matter if it's horizontal or vertical. 00:28:59.650 --> 00:29:01.330 align:middle line:84% You just have to tilt your head, if you 00:29:01.330 --> 00:29:02.990 align:middle line:90% want, from the last picture. 00:29:02.990 --> 00:29:06.220 align:middle line:84% And the gas here, so the gas is flowing vertically. 00:29:06.220 --> 00:29:09.130 align:middle line:84% So the gas is going up here the way the laser pointer is. 00:29:09.130 --> 00:29:09.940 align:middle line:90% It's flowing by. 00:29:09.940 --> 00:29:13.420 align:middle line:84% And it has some concentration here in the main gas stream. 00:29:13.420 --> 00:29:15.520 align:middle line:84% And this region that's labeled from here 00:29:15.520 --> 00:29:17.980 align:middle line:84% to here by these two arrows from the silicon 00:29:17.980 --> 00:29:21.880 align:middle line:84% surface to this point here is called the boundary layer. 00:29:21.880 --> 00:29:23.490 align:middle line:84% So that's the stagnant layer where 00:29:23.490 --> 00:29:25.920 align:middle line:84% the gas velocity in the stream direction 00:29:25.920 --> 00:29:26.910 align:middle line:90% is actually going down. 00:29:26.910 --> 00:29:27.952 align:middle line:90% Eventually, it reaches 0. 00:29:27.952 --> 00:29:30.720 align:middle line:84% So within the boundary layer, we typically 00:29:30.720 --> 00:29:33.420 align:middle line:84% have a gradient of the concentration of the silane, 00:29:33.420 --> 00:29:37.275 align:middle line:84% or whatever it is, the concentration of the reactant. 00:29:37.275 --> 00:29:41.550 align:middle line:84% And the gas stream is called c sub g here. 00:29:41.550 --> 00:29:43.350 align:middle line:84% When you reach the surface of the silicon, 00:29:43.350 --> 00:29:45.480 align:middle line:84% at the very surface, there's a concentration 00:29:45.480 --> 00:29:48.135 align:middle line:84% of that species called c sub s at the surface. 00:29:48.135 --> 00:29:50.830 align:middle line:90% 00:29:50.830 --> 00:29:53.380 align:middle line:84% And these two-- these two f's here, 00:29:53.380 --> 00:29:56.510 align:middle line:84% this flux f1 is a diffusion flux. 00:29:56.510 --> 00:29:59.980 align:middle line:84% So that's the flux of the reactant species to the wafer. 00:29:59.980 --> 00:30:03.010 align:middle line:84% And it's mass transfer-- it's called a mass transport, 00:30:03.010 --> 00:30:04.660 align:middle line:90% or mass transfer flux. 00:30:04.660 --> 00:30:08.740 align:middle line:84% And on the previous page, it represented step number two. 00:30:08.740 --> 00:30:11.490 align:middle line:90% So that's this flux, f1. 00:30:11.490 --> 00:30:14.940 align:middle line:84% Flux f2 is the actual reactant consumed 00:30:14.940 --> 00:30:16.390 align:middle line:90% by the surface reaction. 00:30:16.390 --> 00:30:19.530 align:middle line:84% So that's the surface reaction rate, or surface reaction flux. 00:30:19.530 --> 00:30:21.810 align:middle line:84% And that's what I meant by steps three to five 00:30:21.810 --> 00:30:22.870 align:middle line:90% on the previous page. 00:30:22.870 --> 00:30:28.853 align:middle line:84% So f2, refer to these surface reactions. 00:30:28.853 --> 00:30:30.020 align:middle line:90% So we have these two fluxes. 00:30:30.020 --> 00:30:32.240 align:middle line:84% We got to get through the boundary layer by diffusion, 00:30:32.240 --> 00:30:33.782 align:middle line:84% and we've got to react to the surface 00:30:33.782 --> 00:30:35.820 align:middle line:90% by some chemical reaction. 00:30:35.820 --> 00:30:38.240 align:middle line:84% And so where-- we're just going to write down 00:30:38.240 --> 00:30:41.320 align:middle line:84% simple equations for flux one, transport through a boundary 00:30:41.320 --> 00:30:41.820 align:middle line:90% layer. 00:30:41.820 --> 00:30:45.290 align:middle line:84% We're going to say that flux, the number of particles going 00:30:45.290 --> 00:30:47.762 align:middle line:84% through per square centimeter per unit time 00:30:47.762 --> 00:30:49.970 align:middle line:84% is going to be just proportional to the concentration 00:30:49.970 --> 00:30:52.435 align:middle line:90% gradient, cg minus cs. 00:30:52.435 --> 00:30:53.810 align:middle line:84% And the proportionality constant, 00:30:53.810 --> 00:30:58.220 align:middle line:84% we're going to call hg, which is the mass transfer coefficient, 00:30:58.220 --> 00:31:00.660 align:middle line:84% and typically have units of length per unit time, 00:31:00.660 --> 00:31:01.980 align:middle line:90% or centimeters per second. 00:31:01.980 --> 00:31:03.200 align:middle line:90% So that's the first flux. 00:31:03.200 --> 00:31:06.640 align:middle line:84% That's just transport through the boundary layer. 00:31:06.640 --> 00:31:09.340 align:middle line:84% Flux number two is the reaction at the surface, 00:31:09.340 --> 00:31:10.550 align:middle line:90% and is a chemical reaction. 00:31:10.550 --> 00:31:13.270 align:middle line:84% So we typically write that with its chemical constant, 00:31:13.270 --> 00:31:18.200 align:middle line:84% k sub s, some surface reaction rate in centimeters per second, 00:31:18.200 --> 00:31:20.450 align:middle line:84% times the concentration of the species at the surface. 00:31:20.450 --> 00:31:22.910 align:middle line:84% So again, this is a little cartoon of what we just saw. 00:31:22.910 --> 00:31:25.120 align:middle line:84% So in steady state, we're going to let this-- 00:31:25.120 --> 00:31:28.510 align:middle line:84% the fluxes through the boundary layer and at the surface, 00:31:28.510 --> 00:31:29.260 align:middle line:90% be equal. 00:31:29.260 --> 00:31:33.160 align:middle line:84% f1 equals f2 so we just equate equations four and five, which 00:31:33.160 --> 00:31:34.630 align:middle line:90% we just had. 00:31:34.630 --> 00:31:35.570 align:middle line:90% That's pretty simple. 00:31:35.570 --> 00:31:38.410 align:middle line:84% We can solve, then, for c sub s, the concentration 00:31:38.410 --> 00:31:41.680 align:middle line:84% at the surface, in terms of c sub g, the concentration 00:31:41.680 --> 00:31:42.730 align:middle line:90% in the gas. 00:31:42.730 --> 00:31:49.200 align:middle line:84% And we get its cg divided by 1 plus k sub s over hg. 00:31:49.200 --> 00:31:54.210 align:middle line:84% Now, we just need to define the growth rate of the film. 00:31:54.210 --> 00:31:56.820 align:middle line:84% So the growth rate of the film is this flux, 00:31:56.820 --> 00:31:58.560 align:middle line:90% f, which we now know. 00:31:58.560 --> 00:32:01.230 align:middle line:84% We can write f as k sub s times c sub s. 00:32:01.230 --> 00:32:04.050 align:middle line:84% And I just solve for c sub s in equation seven. 00:32:04.050 --> 00:32:06.540 align:middle line:84% So the velocity of the growth rate 00:32:06.540 --> 00:32:08.070 align:middle line:84% of the film is a velocity v. It's 00:32:08.070 --> 00:32:11.460 align:middle line:84% just the flux divided by n, where n is the number of atoms 00:32:11.460 --> 00:32:13.560 align:middle line:90% per cubic centimeter. 00:32:13.560 --> 00:32:15.450 align:middle line:84% Just you can just do this dimensionally. 00:32:15.450 --> 00:32:17.340 align:middle line:84% The velocity-- the flux has units 00:32:17.340 --> 00:32:21.420 align:middle line:84% of number of atoms per square centimeter per time. 00:32:21.420 --> 00:32:23.740 align:middle line:84% And this is number of atoms per cubic centimeter. 00:32:23.740 --> 00:32:25.410 align:middle line:84% So I just-- I end up with a velocity 00:32:25.410 --> 00:32:28.600 align:middle line:84% of centimeters per second, or centimeters per unit time. 00:32:28.600 --> 00:32:30.960 align:middle line:84% So we just divide this flux by the density. 00:32:30.960 --> 00:32:35.600 align:middle line:84% In silicon, for example, the density is 510 to the 22nd. 00:32:35.600 --> 00:32:39.290 align:middle line:84% Multiplying this out, what we get is we get this quantity, 00:32:39.290 --> 00:32:42.260 align:middle line:84% k sub s hg divided by k sub s plus hg. 00:32:42.260 --> 00:32:47.390 align:middle line:84% That times the concentration in the gas phase of the reactants 00:32:47.390 --> 00:32:48.665 align:middle line:90% divided by the density. 00:32:48.665 --> 00:32:50.840 align:middle line:84% This kind of makes sense if you look at it. 00:32:50.840 --> 00:32:55.080 align:middle line:84% Or you can look at it here in terms of this quantity y, 00:32:55.080 --> 00:32:58.050 align:middle line:84% where y is the mole fraction of the incorporating species. 00:32:58.050 --> 00:33:00.740 align:middle line:84% So it's the partial pressure of silane, for example, 00:33:00.740 --> 00:33:04.820 align:middle line:84% divided by the total pressure in the-- of the gas. 00:33:04.820 --> 00:33:07.100 align:middle line:84% And it kind of makes sense, as you might imagine, 00:33:07.100 --> 00:33:09.230 align:middle line:84% the velocity of growth, or the growth rate, 00:33:09.230 --> 00:33:14.630 align:middle line:84% depends on the concentration of the species in the gas phase. 00:33:14.630 --> 00:33:17.470 align:middle line:90% That's not too surprising. 00:33:17.470 --> 00:33:19.430 align:middle line:84% And therefore, it depends on the mole fraction, 00:33:19.430 --> 00:33:23.120 align:middle line:84% the partial pressure of the silane divided 00:33:23.120 --> 00:33:24.720 align:middle line:90% by the total pressure. 00:33:24.720 --> 00:33:26.370 align:middle line:90% So that's a very simple model. 00:33:26.370 --> 00:33:28.070 align:middle line:84% And the important thing is now to look 00:33:28.070 --> 00:33:32.030 align:middle line:84% at the dependency on k sub s and hg 00:33:32.030 --> 00:33:35.310 align:middle line:84% and what their temperature dependences are. 00:33:35.310 --> 00:33:38.300 align:middle line:84% So let's go on to slide 15 and look at that. 00:33:38.300 --> 00:33:39.920 align:middle line:90% So this is the-- 00:33:39.920 --> 00:33:42.500 align:middle line:84% equation eight, I just repeated exactly that same equation 00:33:42.500 --> 00:33:46.520 align:middle line:84% we just saw for the deposition rate. 00:33:46.520 --> 00:33:49.490 align:middle line:84% And you notice what it-- because of the way these 00:33:49.490 --> 00:33:51.920 align:middle line:84% add here, the deposition rate is going 00:33:51.920 --> 00:33:55.550 align:middle line:84% to-- determined by the smaller of the two of k sub s or hg. 00:33:55.550 --> 00:33:58.620 align:middle line:84% So let's say k sub s is very small. 00:33:58.620 --> 00:34:02.138 align:middle line:84% Then I can ignore it here and forget about it 00:34:02.138 --> 00:34:02.930 align:middle line:90% in the denominator. 00:34:02.930 --> 00:34:04.400 align:middle line:90% Then the hg's go out. 00:34:04.400 --> 00:34:06.950 align:middle line:84% And then in that case, if k sub s is very small, 00:34:06.950 --> 00:34:10.218 align:middle line:90% the velocity just depends on-- 00:34:10.218 --> 00:34:11.389 align:middle line:90% but I'm doing this case. 00:34:11.389 --> 00:34:16.540 align:middle line:84% Then it just depends on hg times this concentration. 00:34:16.540 --> 00:34:19.440 align:middle line:84% Let's see. k sub s is much, much less than hg. 00:34:19.440 --> 00:34:20.739 align:middle line:90% Then I can ignore it here. 00:34:20.739 --> 00:34:22.179 align:middle line:90% The hg's go out. 00:34:22.179 --> 00:34:23.770 align:middle line:90% Oh, and I end up-- 00:34:23.770 --> 00:34:24.969 align:middle line:90% it's the rate-limiting step. 00:34:24.969 --> 00:34:28.164 align:middle line:84% And you end up being proportional to k sub s. 00:34:28.164 --> 00:34:32.070 align:middle line:84% So in that case, we have what's called the surface reaction 00:34:32.070 --> 00:34:32.570 align:middle line:90% case. 00:34:32.570 --> 00:34:34.659 align:middle line:84% So this would be a case where the mass transfer 00:34:34.659 --> 00:34:37.570 align:middle line:84% through the boundary layer is very fast. 00:34:37.570 --> 00:34:40.270 align:middle line:84% You want to think about that, or you could call it in a case 00:34:40.270 --> 00:34:45.050 align:middle line:84% where the surface reaction is very, very slow. 00:34:45.050 --> 00:34:47.110 align:middle line:84% And so the growth rate in this case 00:34:47.110 --> 00:34:49.730 align:middle line:84% is directly proportional to the reaction rate. 00:34:49.730 --> 00:34:52.210 align:middle line:84% The important thing about k sub s is to notice 00:34:52.210 --> 00:34:55.550 align:middle line:84% is that it's exponentially dependent on temperature. 00:34:55.550 --> 00:34:58.840 align:middle line:84% So in this regime, we expect to see the growth 00:34:58.840 --> 00:35:01.780 align:middle line:84% rate exponentially proportional, or exponentially dependent, 00:35:01.780 --> 00:35:03.550 align:middle line:90% on temperature. 00:35:03.550 --> 00:35:05.920 align:middle line:84% On the other hand, if k sub s is very-- 00:35:05.920 --> 00:35:09.430 align:middle line:84% if the reaction goes very fast compared to the mass transport, 00:35:09.430 --> 00:35:12.620 align:middle line:84% or the mass transport here is slow, hg is small, 00:35:12.620 --> 00:35:15.700 align:middle line:84% then we have the mass transfer, or gas phase, diffusion control 00:35:15.700 --> 00:35:16.370 align:middle line:90% case. 00:35:16.370 --> 00:35:20.260 align:middle line:84% So here, the velocity depends on hg, which it turns out, 00:35:20.260 --> 00:35:23.710 align:middle line:84% diffusion through the gas is not a very tempered-- unlike 00:35:23.710 --> 00:35:26.200 align:middle line:84% reaction at the surface, is not very temperature sensitive. 00:35:26.200 --> 00:35:30.910 align:middle line:84% So we expect in this regime to have a very weak temperature 00:35:30.910 --> 00:35:32.580 align:middle line:90% dependence. 00:35:32.580 --> 00:35:36.050 align:middle line:84% So just, we can immediately see these two different regimes 00:35:36.050 --> 00:35:39.110 align:middle line:84% happening depending on the relative ratio of k sub 00:35:39.110 --> 00:35:40.580 align:middle line:84% s to hg, and therefore, depending 00:35:40.580 --> 00:35:41.900 align:middle line:90% on the relative temperature. 00:35:41.900 --> 00:35:44.600 align:middle line:90% 00:35:44.600 --> 00:35:49.870 align:middle line:84% So if we go to slide 16, what people find experimentally 00:35:49.870 --> 00:35:52.540 align:middle line:84% what these rate constants look like. 00:35:52.540 --> 00:35:56.470 align:middle line:84% k sub s actually goes like a constant, some number, 00:35:56.470 --> 00:35:59.350 align:middle line:84% k naught, whatever it happens to be, times e to the minus 00:35:59.350 --> 00:36:00.580 align:middle line:90% e over kt. 00:36:00.580 --> 00:36:03.700 align:middle line:84% So the surface reaction rate has a certain temperature 00:36:03.700 --> 00:36:05.200 align:middle line:84% dependence, it's exponential, and it 00:36:05.200 --> 00:36:09.190 align:middle line:90% has an activation energy of ea. 00:36:09.190 --> 00:36:13.540 align:middle line:84% So if I were to plot that on the growth velocity, or the growth 00:36:13.540 --> 00:36:17.260 align:middle line:84% rate, on a log scale, versus 1 over t, what-- 00:36:17.260 --> 00:36:19.000 align:middle line:84% the k sub s term is just going to be 00:36:19.000 --> 00:36:21.580 align:middle line:84% a straight line because it's just, 00:36:21.580 --> 00:36:23.440 align:middle line:84% I'm taking the log of an exponential. 00:36:23.440 --> 00:36:25.150 align:middle line:84% And therefore, I just get a straight line 00:36:25.150 --> 00:36:28.320 align:middle line:84% directly proportional to 1 over t. 00:36:28.320 --> 00:36:30.870 align:middle line:84% So down in this region here, the slope 00:36:30.870 --> 00:36:33.730 align:middle line:84% is directly proportional to the activation energy. 00:36:33.730 --> 00:36:35.940 align:middle line:90% Now, what happens is, as I go-- 00:36:35.940 --> 00:36:37.578 align:middle line:84% as I change the temperature here, 00:36:37.578 --> 00:36:39.370 align:middle line:84% and I go to higher and higher temperature-- 00:36:39.370 --> 00:36:41.578 align:middle line:84% so higher temperature means going to the left, right? 00:36:41.578 --> 00:36:43.095 align:middle line:90% Going to lower 1 over t. 00:36:43.095 --> 00:36:46.950 align:middle line:84% At some point, what we see is we come 00:36:46.950 --> 00:36:50.340 align:middle line:84% rate limited by the surface reaction rate 00:36:50.340 --> 00:36:52.440 align:middle line:90% by the mass transport rather. 00:36:52.440 --> 00:36:55.200 align:middle line:84% The surface reaction is happening so fast at this point 00:36:55.200 --> 00:36:57.600 align:middle line:90% that its rate overtakes that. 00:36:57.600 --> 00:37:00.750 align:middle line:84% And you-- the growth rate becomes limited by transport 00:37:00.750 --> 00:37:02.115 align:middle line:90% through the boundary layer. 00:37:02.115 --> 00:37:03.990 align:middle line:84% And that transport through the boundary layer 00:37:03.990 --> 00:37:05.073 align:middle line:90% is pretty much a constant. 00:37:05.073 --> 00:37:07.260 align:middle line:84% It doesn't depend much on temperature. 00:37:07.260 --> 00:37:09.900 align:middle line:84% It depends on pressure and the design of the reactor. 00:37:09.900 --> 00:37:12.165 align:middle line:84% So in this regime, we expect this 00:37:12.165 --> 00:37:15.420 align:middle line:84% to be relatively temperature insensitive. 00:37:15.420 --> 00:37:18.000 align:middle line:84% And of course, the growth rate is-- the net growth rate 00:37:18.000 --> 00:37:20.340 align:middle line:84% is neither one of these dashed lines. 00:37:20.340 --> 00:37:23.280 align:middle line:84% It's the solid line, which is the combination of the two, 00:37:23.280 --> 00:37:25.950 align:middle line:84% where we combine them just like we did here 00:37:25.950 --> 00:37:28.290 align:middle line:84% according to this equation, equation eight. 00:37:28.290 --> 00:37:29.910 align:middle line:84% Tells you exactly how to add them up. 00:37:29.910 --> 00:37:32.460 align:middle line:90% 00:37:32.460 --> 00:37:35.335 align:middle line:84% For example, if you are growing single crystal silicon, 00:37:35.335 --> 00:37:36.710 align:middle line:84% typically, the activation energy, 00:37:36.710 --> 00:37:38.502 align:middle line:84% just to give you a rough idea, is typically 00:37:38.502 --> 00:37:41.410 align:middle line:90% in the range of 1.6 to 2 EV. 00:37:41.410 --> 00:37:45.280 align:middle line:84% So that's what the slope down here will be. 00:37:45.280 --> 00:37:48.190 align:middle line:84% And again, hg is relatively constant as a function 00:37:48.190 --> 00:37:51.460 align:middle line:90% of temperature. 00:37:51.460 --> 00:37:55.880 align:middle line:84% So as an example, let's go to slide number 17, where-- 00:37:55.880 --> 00:37:58.650 align:middle line:84% which illustrates some data I took out of your textbook 00:37:58.650 --> 00:38:01.590 align:middle line:90% on silicon epitaxial growth. 00:38:01.590 --> 00:38:03.443 align:middle line:84% And this is all at atmospheric pressure. 00:38:03.443 --> 00:38:04.860 align:middle line:84% So the data is a little bit older. 00:38:04.860 --> 00:38:09.090 align:middle line:84% But what you see here is growth rate. 00:38:09.090 --> 00:38:11.070 align:middle line:84% And the units are microns per minute. 00:38:11.070 --> 00:38:15.210 align:middle line:84% And again, this is a logarithmic scale versus 1 over t, 00:38:15.210 --> 00:38:16.380 align:middle line:90% or actually 1,000 over t. 00:38:16.380 --> 00:38:19.530 align:middle line:84% So this is a typical Arrhenius-type plot. 00:38:19.530 --> 00:38:21.270 align:middle line:84% If you want to read the temperature, 00:38:21.270 --> 00:38:24.120 align:middle line:84% you can conveniently read the temperature right off here, 00:38:24.120 --> 00:38:27.150 align:middle line:84% going from say 600 all the way up to 1,200. 00:38:27.150 --> 00:38:30.100 align:middle line:84% And there are a couple different curves here. 00:38:30.100 --> 00:38:33.150 align:middle line:84% So here's the curve for silicon tetrachloride. 00:38:33.150 --> 00:38:36.580 align:middle line:84% Silicon tet doesn't react very well at low temperatures. 00:38:36.580 --> 00:38:38.760 align:middle line:84% So its growth rate is relatively slow 00:38:38.760 --> 00:38:42.540 align:middle line:84% compared to trichlorosilane, dichlorosilane, 00:38:42.540 --> 00:38:44.640 align:middle line:90% which is SiH2Cl2-- 00:38:44.640 --> 00:38:47.790 align:middle line:84% that's dichlorosilane, very commonly used today-- 00:38:47.790 --> 00:38:49.020 align:middle line:90% or silane. 00:38:49.020 --> 00:38:51.708 align:middle line:84% Silane's the most reactive of all of these. 00:38:51.708 --> 00:38:54.250 align:middle line:84% And so at any given temperature, it has a higher growth rate. 00:38:54.250 --> 00:38:55.588 align:middle line:90% You can see that. 00:38:55.588 --> 00:38:57.130 align:middle line:84% But look at the shapes of the curves. 00:38:57.130 --> 00:38:59.410 align:middle line:84% It's very similar to that previous model. 00:38:59.410 --> 00:39:00.220 align:middle line:90% Not exact. 00:39:00.220 --> 00:39:02.740 align:middle line:84% But you see for any given reactant, let's say silane, 00:39:02.740 --> 00:39:06.940 align:middle line:84% we have a region where the growth rate on Arrhenius plot 00:39:06.940 --> 00:39:08.210 align:middle line:90% is a straight line. 00:39:08.210 --> 00:39:09.940 align:middle line:90% So it's exponentially activated. 00:39:09.940 --> 00:39:11.290 align:middle line:90% This is so its mass-- 00:39:11.290 --> 00:39:13.870 align:middle line:84% it's limited here by a surface reaction. 00:39:13.870 --> 00:39:15.527 align:middle line:84% Very sensitive to the temperature. 00:39:15.527 --> 00:39:17.860 align:middle line:84% And then you finally get up to a high enough temperature 00:39:17.860 --> 00:39:19.040 align:middle line:90% and it starts to roll off. 00:39:19.040 --> 00:39:20.560 align:middle line:90% It never is completely constant. 00:39:20.560 --> 00:39:23.185 align:middle line:84% But it reaches a point where the temperature dependence is very 00:39:23.185 --> 00:39:26.720 align:middle line:90% small at high temperatures. 00:39:26.720 --> 00:39:27.980 align:middle line:90% And so and this-- 00:39:27.980 --> 00:39:29.480 align:middle line:84% we call this high temperature regime 00:39:29.480 --> 00:39:31.915 align:middle line:90% the mass transport limited. 00:39:31.915 --> 00:39:33.290 align:middle line:84% So you're talking about diffusion 00:39:33.290 --> 00:39:34.415 align:middle line:90% through the boundary layer. 00:39:34.415 --> 00:39:36.560 align:middle line:84% And down here, where it's clearly exponential, 00:39:36.560 --> 00:39:38.390 align:middle line:84% the surface reaction rate limited. 00:39:38.390 --> 00:39:41.510 align:middle line:90% 00:39:41.510 --> 00:39:44.920 align:middle line:84% So in the old fashioned silicone epi, 00:39:44.920 --> 00:39:47.890 align:middle line:84% people used to do deposition at very high temperatures 00:39:47.890 --> 00:39:49.670 align:middle line:90% to get high crystal quality. 00:39:49.670 --> 00:39:52.600 align:middle line:84% So in the old days, people typically grew at 1,00 or 1,150. 00:39:52.600 --> 00:39:54.640 align:middle line:84% So they were almost always growing 00:39:54.640 --> 00:39:59.360 align:middle line:84% with these reactants in the mass transport limited regime. 00:39:59.360 --> 00:40:01.460 align:middle line:84% So they were almost always hg controlled. 00:40:01.460 --> 00:40:03.280 align:middle line:84% So the old-fashioned reactors had 00:40:03.280 --> 00:40:05.320 align:middle line:84% to have the horizontal reactor configuration 00:40:05.320 --> 00:40:08.140 align:middle line:84% because mass transport through the boundary layer 00:40:08.140 --> 00:40:10.600 align:middle line:84% depends on exactly how the reactor is designed. 00:40:10.600 --> 00:40:13.660 align:middle line:84% Modern epitaxy often operates, however, 00:40:13.660 --> 00:40:16.960 align:middle line:84% people growing silicon germanium and other materials 00:40:16.960 --> 00:40:19.390 align:middle line:84% need to be grown at lower temperatures, like 800 or so, 00:40:19.390 --> 00:40:21.880 align:middle line:84% or 700, often operates in this regime now 00:40:21.880 --> 00:40:24.260 align:middle line:84% these days, where you're in the exponential regime. 00:40:24.260 --> 00:40:26.770 align:middle line:84% So we're controlling the temperature becomes extremely 00:40:26.770 --> 00:40:28.060 align:middle line:90% important because that-- 00:40:28.060 --> 00:40:31.890 align:middle line:84% in order to get good growth rate control. 00:40:31.890 --> 00:40:33.450 align:middle line:90% Yeah. 00:40:33.450 --> 00:40:35.430 align:middle line:90% AUDIENCE: [INAUDIBLE] 00:40:35.430 --> 00:40:37.120 align:middle line:90% 00:40:37.120 --> 00:40:39.270 align:middle line:84% JUDY HOYT: The question on slide 17 00:40:39.270 --> 00:40:42.480 align:middle line:84% is about, what happened to the silane in the nitrogen ambient? 00:40:42.480 --> 00:40:43.560 align:middle line:90% You see this dashed line. 00:40:43.560 --> 00:40:46.590 align:middle line:84% You notice the growth rate for silicon using a nitrogen 00:40:46.590 --> 00:40:48.090 align:middle line:84% ambient instead of a hydrogen. These 00:40:48.090 --> 00:40:50.160 align:middle line:84% are all grown in hydrogen, which is typically 00:40:50.160 --> 00:40:51.390 align:middle line:90% used for higher purity. 00:40:51.390 --> 00:40:55.120 align:middle line:84% The growth rate pops way up at the same temperature. 00:40:55.120 --> 00:40:57.210 align:middle line:84% And the reason for that is if you go back and look 00:40:57.210 --> 00:41:00.390 align:middle line:84% at a couple of slides, let's see if I can find 00:41:00.390 --> 00:41:02.850 align:middle line:90% the chemical decomposition. 00:41:02.850 --> 00:41:06.960 align:middle line:84% On slide number 11, if you look at equation number two 00:41:06.960 --> 00:41:11.580 align:middle line:84% for the chemical decomposition of silane, silane and the gas 00:41:11.580 --> 00:41:15.240 align:middle line:84% forming silicon, solid silicon, it evolves. 00:41:15.240 --> 00:41:17.760 align:middle line:84% The reaction, when you grow epitaxial silicon with silane, 00:41:17.760 --> 00:41:20.435 align:middle line:84% it evolves hydrogen. So hydrogen has to come off. 00:41:20.435 --> 00:41:21.810 align:middle line:84% Now, this is a chemical reaction. 00:41:21.810 --> 00:41:24.240 align:middle line:84% So you know you can push it to the left. 00:41:24.240 --> 00:41:26.430 align:middle line:84% You can tend to slow down this reaction 00:41:26.430 --> 00:41:29.230 align:middle line:84% if I put a lot of hydrogen in the ambient. 00:41:29.230 --> 00:41:30.990 align:middle line:84% So using the hydrogen carrier gas 00:41:30.990 --> 00:41:33.940 align:middle line:84% tends to push this to the left, and tends to cause this surface 00:41:33.940 --> 00:41:35.980 align:middle line:90% reaction to be less probable. 00:41:35.980 --> 00:41:38.405 align:middle line:90% And the reaction rate goes down. 00:41:38.405 --> 00:41:40.780 align:middle line:84% But if you, instead of using hydrogen as the carrier gas, 00:41:40.780 --> 00:41:44.080 align:middle line:84% if you use an inert, more inert gas like nitrogen, 00:41:44.080 --> 00:41:46.520 align:middle line:84% you won't have that pushing-to-the-left effect. 00:41:46.520 --> 00:41:48.650 align:middle line:90% And so the growth rate goes up. 00:41:48.650 --> 00:41:51.320 align:middle line:90% AUDIENCE: [INAUDIBLE] 00:41:51.320 --> 00:41:54.740 align:middle line:90% 00:41:54.740 --> 00:41:56.060 align:middle line:90% JUDY HOYT: Well, let's see. 00:41:56.060 --> 00:41:58.910 align:middle line:84% In the transport limited regime, you're still going to have-- 00:41:58.910 --> 00:42:01.070 align:middle line:90% I mean, you still have-- 00:42:01.070 --> 00:42:02.570 align:middle line:90% well, there is a little bit. 00:42:02.570 --> 00:42:04.310 align:middle line:90% Yeah. 00:42:04.310 --> 00:42:06.380 align:middle line:84% Well, in the transport limited regime, 00:42:06.380 --> 00:42:09.020 align:middle line:84% it looks like we never really get there with this dashed data. 00:42:09.020 --> 00:42:11.760 align:middle line:84% They sort of petered out a little bit. 00:42:11.760 --> 00:42:14.090 align:middle line:84% So it's almost getting to the transport limit regime. 00:42:14.090 --> 00:42:16.973 align:middle line:84% But the problem is the surface reaction rate is so fast now, 00:42:16.973 --> 00:42:18.890 align:middle line:84% it's hard to see the transport limited regime. 00:42:18.890 --> 00:42:20.690 align:middle line:84% But it looks like they're converging. 00:42:20.690 --> 00:42:23.000 align:middle line:84% It looks like the two are converging. 00:42:23.000 --> 00:42:25.820 align:middle line:84% Because at that point, the reaction rate really 00:42:25.820 --> 00:42:27.080 align:middle line:90% isn't what's controlling it. 00:42:27.080 --> 00:42:29.040 align:middle line:84% It's the transport through the boundary layer. 00:42:29.040 --> 00:42:30.740 align:middle line:84% So the question is, well, what's the difference 00:42:30.740 --> 00:42:32.540 align:middle line:84% in transport between a boundary layer 00:42:32.540 --> 00:42:34.043 align:middle line:84% and a nitrogen versus a hydrogen? 00:42:34.043 --> 00:42:35.960 align:middle line:84% There's probably some small differences there. 00:42:35.960 --> 00:42:38.590 align:middle line:90% 00:42:38.590 --> 00:42:42.290 align:middle line:84% OK, so let's go on to slide number 18. 00:42:42.290 --> 00:42:45.040 align:middle line:84% We were just saying how transport through the boundary 00:42:45.040 --> 00:42:47.890 align:middle line:84% layer is important when you're in the high temperature 00:42:47.890 --> 00:42:49.960 align:middle line:84% regime, or the mass transport limited regime. 00:42:49.960 --> 00:42:53.200 align:middle line:84% And therefore, it turns out that the reactor geometry 00:42:53.200 --> 00:42:55.330 align:middle line:84% in the high temperature regime is very important, 00:42:55.330 --> 00:42:56.920 align:middle line:90% how you build the reactor. 00:42:56.920 --> 00:42:59.260 align:middle line:84% And I've taken this on slide 18 here. 00:42:59.260 --> 00:43:03.370 align:middle line:84% I've taken this picture from your textbook, figure 9.9. 00:43:03.370 --> 00:43:07.810 align:middle line:84% And what it's showing is the velocities in the boundary 00:43:07.810 --> 00:43:09.400 align:middle line:84% layer, the velocities are represented 00:43:09.400 --> 00:43:14.620 align:middle line:84% by these little arrows along the susceptor 00:43:14.620 --> 00:43:16.880 align:middle line:90% and going from left to right. 00:43:16.880 --> 00:43:19.090 align:middle line:84% So the gas is coming in at the left, 00:43:19.090 --> 00:43:21.640 align:middle line:84% flowing over the wafers and the susceptor, 00:43:21.640 --> 00:43:24.400 align:middle line:90% and going out the right. 00:43:24.400 --> 00:43:28.390 align:middle line:84% And according to some gas flow laws, Newton's second law, 00:43:28.390 --> 00:43:31.430 align:middle line:84% the boundary layer, the thickness of the stagnant layer, 00:43:31.430 --> 00:43:34.510 align:middle line:84% which is represented by the height of this layer 00:43:34.510 --> 00:43:36.370 align:middle line:84% above the wafer surface, this delta sub 00:43:36.370 --> 00:43:40.520 align:middle line:84% s, you see this height, or the thickness of the stagnant layer, 00:43:40.520 --> 00:43:42.720 align:middle line:90% is actually increasing. 00:43:42.720 --> 00:43:45.612 align:middle line:84% So and this boundary, this dark region, 00:43:45.612 --> 00:43:47.570 align:middle line:84% represents the thickness of the boundary layer. 00:43:47.570 --> 00:43:50.210 align:middle line:84% And you notice beyond the boundary layer, above it, 00:43:50.210 --> 00:43:53.030 align:middle line:84% the gas velocity has reached the same as the velocity 00:43:53.030 --> 00:43:54.632 align:middle line:84% as it is in the center of the tube. 00:43:54.632 --> 00:43:56.090 align:middle line:84% It's only within the boundary layer 00:43:56.090 --> 00:43:57.423 align:middle line:90% that the velocity is decreasing. 00:43:57.423 --> 00:44:00.210 align:middle line:84% So these arrows, their lengths are decreasing. 00:44:00.210 --> 00:44:05.150 align:middle line:84% So as you flow gas through a pipe against the surface, 00:44:05.150 --> 00:44:07.760 align:middle line:84% the boundary layer thickness is actually increasing. 00:44:07.760 --> 00:44:12.080 align:middle line:84% Now, this process of diffusion through the boundary layer, 00:44:12.080 --> 00:44:15.950 align:middle line:84% this quantity, the hg, the mass transport coefficient, 00:44:15.950 --> 00:44:18.230 align:middle line:84% it has to do with diffusion through that delta 00:44:18.230 --> 00:44:21.130 align:middle line:84% sub s, that thickness delta sub s. 00:44:21.130 --> 00:44:24.290 align:middle line:84% So people write in a crude way that hg 00:44:24.290 --> 00:44:27.050 align:middle line:84% is equal to some diffusion coefficient in the gas 00:44:27.050 --> 00:44:30.056 align:middle line:84% phase divided by the thickness delta s. 00:44:30.056 --> 00:44:32.530 align:middle line:84% But see, delta is not constant as the gas 00:44:32.530 --> 00:44:34.310 align:middle line:90% flows along the surface. 00:44:34.310 --> 00:44:38.620 align:middle line:84% So people try to manipulate things 00:44:38.620 --> 00:44:42.110 align:middle line:84% to try to make delta s more constant across the reactor 00:44:42.110 --> 00:44:42.610 align:middle line:90% surface. 00:44:42.610 --> 00:44:45.040 align:middle line:84% Otherwise, what will happen is the growth rate here 00:44:45.040 --> 00:44:46.990 align:middle line:84% will be much higher on the first wafer 00:44:46.990 --> 00:44:51.480 align:middle line:84% than it would be on the last wafer in the reactor. 00:44:51.480 --> 00:44:56.040 align:middle line:84% So people try to compensate for that. 00:44:56.040 --> 00:44:57.960 align:middle line:84% A certain type of reactor geometry 00:44:57.960 --> 00:45:00.810 align:middle line:84% is usually required, therefore, to get uniform deposition 00:45:00.810 --> 00:45:02.820 align:middle line:84% across a number of different silicon wafers 00:45:02.820 --> 00:45:06.780 align:middle line:84% if you're working in this mass transport limited regime. 00:45:06.780 --> 00:45:10.420 align:middle line:84% A typical trick is to tilt the susceptor. 00:45:10.420 --> 00:45:12.940 align:middle line:84% So take the susceptor and tilt it. 00:45:12.940 --> 00:45:17.380 align:middle line:84% And how-- this increases the gas velocity, which keeps delta sub 00:45:17.380 --> 00:45:18.200 align:middle line:90% s constant. 00:45:18.200 --> 00:45:21.050 align:middle line:84% So as I'm going, the gas velocity at this point-- 00:45:21.050 --> 00:45:23.440 align:middle line:84% so imagine I'm flowing the same volume of gas 00:45:23.440 --> 00:45:25.090 align:middle line:84% a certain number of liters per minute 00:45:25.090 --> 00:45:27.670 align:middle line:84% through this cross section from here to here. 00:45:27.670 --> 00:45:29.320 align:middle line:90% The gas velocity has some value. 00:45:29.320 --> 00:45:32.008 align:middle line:84% Now, if I go here, the cross section has decreased. 00:45:32.008 --> 00:45:34.300 align:middle line:84% So the velocity-- so again, I'm getting the same volume 00:45:34.300 --> 00:45:35.600 align:middle line:90% number of liters per minute. 00:45:35.600 --> 00:45:38.680 align:middle line:84% So the velocity of the gas in a smaller cross section tube 00:45:38.680 --> 00:45:39.950 align:middle line:90% must therefore go up. 00:45:39.950 --> 00:45:42.800 align:middle line:84% So v is increasing. now from left to right, 00:45:42.800 --> 00:45:46.210 align:middle line:84% which was not the case when the susceptor is flat. 00:45:46.210 --> 00:45:48.940 align:middle line:84% When the susceptor is flat, the velocity here is called u. 00:45:48.940 --> 00:45:52.460 align:middle line:84% And the center is the same from left to right. 00:45:52.460 --> 00:45:56.170 align:middle line:84% And so what we're doing is we're-- to compensate for this-- 00:45:56.170 --> 00:45:58.960 align:middle line:84% to make the boundary layer thickness perfectly flat, 00:45:58.960 --> 00:46:02.032 align:middle line:84% we tilt the susceptor to up that velocity 00:46:02.032 --> 00:46:03.115 align:middle line:90% in the center of the tube. 00:46:03.115 --> 00:46:06.830 align:middle line:90% 00:46:06.830 --> 00:46:09.173 align:middle line:84% So but there's another-- there's another issue, 00:46:09.173 --> 00:46:10.340 align:middle line:90% though, with non-uniformity. 00:46:10.340 --> 00:46:12.923 align:middle line:84% Not just so that-- and you can take care of the boundary layer 00:46:12.923 --> 00:46:15.110 align:middle line:84% effect by designing your reactor appropriately. 00:46:15.110 --> 00:46:17.180 align:middle line:84% The source gases themselves can become 00:46:17.180 --> 00:46:19.100 align:middle line:84% depleted along the length of the susceptor. 00:46:19.100 --> 00:46:22.040 align:middle line:84% So you have-- maybe have less silane back here than you 00:46:22.040 --> 00:46:24.980 align:middle line:84% have here because a lot of it may have reacted. 00:46:24.980 --> 00:46:27.740 align:middle line:84% What people sometimes do, then, is they sometimes introduce 00:46:27.740 --> 00:46:29.910 align:middle line:84% a temperature gradient along the flow direction. 00:46:29.910 --> 00:46:32.180 align:middle line:84% So you might make the temperature here just 00:46:32.180 --> 00:46:34.190 align:middle line:84% a little bit hotter than it is here 00:46:34.190 --> 00:46:36.140 align:middle line:84% and try to compensate for that, and try 00:46:36.140 --> 00:46:38.660 align:middle line:84% to get uniform thickness on the first wafer, 00:46:38.660 --> 00:46:42.875 align:middle line:84% and the fifth wafer in such a type of reactor. 00:46:42.875 --> 00:46:45.800 align:middle line:90% 00:46:45.800 --> 00:46:49.000 align:middle line:84% So if we go on to slide 20, now, there 00:46:49.000 --> 00:46:51.670 align:middle line:90% are some implications of this-- 00:46:51.670 --> 00:46:54.070 align:middle line:84% what we know about how CVD works now 00:46:54.070 --> 00:46:57.370 align:middle line:84% based on our simple model of surface reaction rate and mass 00:46:57.370 --> 00:47:00.700 align:middle line:84% transport late on the reactor design. 00:47:00.700 --> 00:47:02.270 align:middle line:84% If we look at a horizontal reactor, 00:47:02.270 --> 00:47:03.700 align:middle line:84% and this is what I was just-- this 00:47:03.700 --> 00:47:05.950 align:middle line:84% is supposed to represent the tilt susceptor. 00:47:05.950 --> 00:47:09.580 align:middle line:84% If the deposition occurs in the range of, say, one torr to 760 00:47:09.580 --> 00:47:13.520 align:middle line:84% torr, relatively high pressures and relatively high 00:47:13.520 --> 00:47:16.430 align:middle line:84% temperatures, then the mass transport to the wafer surface 00:47:16.430 --> 00:47:18.535 align:middle line:90% is the most important thing. 00:47:18.535 --> 00:47:20.660 align:middle line:84% It's more important than the surface reaction rate. 00:47:20.660 --> 00:47:23.240 align:middle line:84% And that places some severe restrictions 00:47:23.240 --> 00:47:27.050 align:middle line:84% on the geometry of the reactor, and on the gas flows, 00:47:27.050 --> 00:47:28.970 align:middle line:90% and how the wafers are stacked. 00:47:28.970 --> 00:47:31.700 align:middle line:84% So typically in that case, atmospheric pressure, 00:47:31.700 --> 00:47:35.090 align:middle line:84% or relatively high pressure, epi, you'll 00:47:35.090 --> 00:47:38.570 align:middle line:84% see it in reactors that only have a few wafers in them. 00:47:38.570 --> 00:47:40.958 align:middle line:84% And the geometry of the reactor and how you stack them 00:47:40.958 --> 00:47:41.750 align:middle line:90% is really critical. 00:47:41.750 --> 00:47:44.310 align:middle line:90% 00:47:44.310 --> 00:47:46.250 align:middle line:84% So when the deposition occurs primarily 00:47:46.250 --> 00:47:48.800 align:middle line:84% at low temperatures, and low pressures-- so we're 00:47:48.800 --> 00:47:52.310 align:middle line:84% in the millitorr regime, or hundreds of millitorr, 00:47:52.310 --> 00:47:54.020 align:middle line:84% we're surface reaction rate controlled. 00:47:54.020 --> 00:47:56.270 align:middle line:84% So it's very sensitive to the temperature. 00:47:56.270 --> 00:47:58.340 align:middle line:84% But it's not so sensitive to transport 00:47:58.340 --> 00:48:01.250 align:middle line:90% through the gas boundary layer. 00:48:01.250 --> 00:48:03.080 align:middle line:84% And the mean-free path is long enough 00:48:03.080 --> 00:48:05.540 align:middle line:84% that the mean-free path is just don't come into effect. 00:48:05.540 --> 00:48:08.090 align:middle line:84% In this case, for this type of process, 00:48:08.090 --> 00:48:12.325 align:middle line:84% then, where we're working in the hundreds millitorr range, 00:48:12.325 --> 00:48:13.700 align:middle line:84% and the temperature is low enough 00:48:13.700 --> 00:48:15.408 align:middle line:84% that we're surface reaction rate limited, 00:48:15.408 --> 00:48:18.230 align:middle line:84% we can stack the wafers together relatively closely. 00:48:18.230 --> 00:48:22.040 align:middle line:84% So you see in an LPCVD process, which LPCVD usually 00:48:22.040 --> 00:48:24.660 align:middle line:84% works at around a few hundred millitorr, 00:48:24.660 --> 00:48:29.280 align:middle line:84% you can stack a lot of wafers close to each other. 00:48:29.280 --> 00:48:32.100 align:middle line:84% You still have to worry about gas depletion in LPCVD. 00:48:32.100 --> 00:48:35.400 align:middle line:90% So what people do is when-- 00:48:35.400 --> 00:48:37.260 align:middle line:84% assuming the gases are coming in here 00:48:37.260 --> 00:48:38.910 align:middle line:84% and going out, say, they're going out-- 00:48:38.910 --> 00:48:41.040 align:middle line:84% going from left to right, people will again 00:48:41.040 --> 00:48:43.530 align:middle line:84% tilt the profile of the temperature in the furnace. 00:48:43.530 --> 00:48:46.980 align:middle line:84% So typical LPCVD system will have three zones, a front zone, 00:48:46.980 --> 00:48:48.450 align:middle line:90% a middle zone, and a back zone. 00:48:48.450 --> 00:48:50.940 align:middle line:84% And the back zone is usually kept a little bit hotter 00:48:50.940 --> 00:48:53.425 align:middle line:84% than the middle or the front zone. 00:48:53.425 --> 00:48:55.800 align:middle line:84% Because again, the silane, or whatever you're depositing, 00:48:55.800 --> 00:48:56.980 align:middle line:90% tends to get depleted. 00:48:56.980 --> 00:48:59.630 align:middle line:84% So you make up for that by increasing the surface reaction 00:48:59.630 --> 00:49:01.630 align:middle line:84% rate a little bit by increasing the temperature. 00:49:01.630 --> 00:49:04.230 align:middle line:84% So you hope you can get the same thickness on the first wafer 00:49:04.230 --> 00:49:07.230 align:middle line:84% as you got on the 50th wafer, or whatever. 00:49:07.230 --> 00:49:11.970 align:middle line:84% And that needs to be done somewhat empirically. 00:49:11.970 --> 00:49:15.560 align:middle line:84% So on slide 21, I'm showing some classical geometries 00:49:15.560 --> 00:49:17.660 align:middle line:90% for epitaxial reactors. 00:49:17.660 --> 00:49:19.430 align:middle line:90% I took this from Simon Z's book. 00:49:19.430 --> 00:49:23.420 align:middle line:84% He has a VLSI technology book that he edited. 00:49:23.420 --> 00:49:25.400 align:middle line:84% There's a whole chapter there on epitaxy 00:49:25.400 --> 00:49:27.810 align:middle line:84% if you want to hear more-- see more details on it. 00:49:27.810 --> 00:49:29.720 align:middle line:84% These are some schematics of some 00:49:29.720 --> 00:49:32.360 align:middle line:84% of the common old-fashioned epi reactors. 00:49:32.360 --> 00:49:35.480 align:middle line:84% There was something called a radiant barrel. 00:49:35.480 --> 00:49:37.430 align:middle line:84% The barrel reactor was shaped like a barrel. 00:49:37.430 --> 00:49:39.890 align:middle line:84% The gaseous-- gas would come in at the top 00:49:39.890 --> 00:49:41.630 align:middle line:90% and exit out the bottom. 00:49:41.630 --> 00:49:47.420 align:middle line:84% And there would be sort of cylindrically shaped, 00:49:47.420 --> 00:49:50.888 align:middle line:84% but with flat surfaces, sort of susceptor. 00:49:50.888 --> 00:49:52.430 align:middle line:84% And you see each one of these circles 00:49:52.430 --> 00:49:54.847 align:middle line:84% is supposed to represent a wafer sitting on the susceptor. 00:49:54.847 --> 00:49:57.770 align:middle line:84% So you could get a fair number of wafers in it at one time. 00:49:57.770 --> 00:50:00.320 align:middle line:84% And notice, the-- they were trying 00:50:00.320 --> 00:50:03.290 align:middle line:84% to make to do the equivalent of susceptor tilting 00:50:03.290 --> 00:50:07.070 align:middle line:84% by decreasing the cross-section through which the gas flows 00:50:07.070 --> 00:50:09.380 align:middle line:84% from the top, which the wafer at the top 00:50:09.380 --> 00:50:12.200 align:middle line:84% to the bottom, where the gas exits. 00:50:12.200 --> 00:50:14.120 align:middle line:84% There are also vertical reactor geometries 00:50:14.120 --> 00:50:16.162 align:middle line:84% where the gas would come in and go out like this. 00:50:16.162 --> 00:50:18.120 align:middle line:84% And the horizontal, we've already talked about. 00:50:18.120 --> 00:50:21.110 align:middle line:84% So all of these were in the cases where-- 00:50:21.110 --> 00:50:26.210 align:middle line:84% are typically multi-wafer tools where you grow anywhere from 00:50:26.210 --> 00:50:29.660 align:middle line:84% five to 20 wafers in a typical epi-- 00:50:29.660 --> 00:50:32.130 align:middle line:90% in an epi batch. 00:50:32.130 --> 00:50:34.620 align:middle line:84% It's the more classical, or maybe old-fashioned, 00:50:34.620 --> 00:50:37.370 align:middle line:90% type of reactor. 00:50:37.370 --> 00:50:40.640 align:middle line:84% Slide 22 is a more modern epitaxial system 00:50:40.640 --> 00:50:44.900 align:middle line:84% that you will see in fabs today, particularly fabs 00:50:44.900 --> 00:50:48.890 align:middle line:84% that are growing silicon germanium for bipolar 00:50:48.890 --> 00:50:51.410 align:middle line:84% transistors, or even silicon germanium now used 00:50:51.410 --> 00:50:54.710 align:middle line:84% in CMOS in source drain regions of CMOS intel 00:50:54.710 --> 00:50:57.080 align:middle line:90% has a commercial process. 00:50:57.080 --> 00:51:00.080 align:middle line:84% In both of those cases, bipolar or CMOS, 00:51:00.080 --> 00:51:03.380 align:middle line:84% people have moved to-- away from the old fashioned more 00:51:03.380 --> 00:51:07.250 align:middle line:84% to the more modern single-wafer epitaxial, silicon, 00:51:07.250 --> 00:51:11.750 align:middle line:84% in order to get the kind of uniformity they would need over 00:51:11.750 --> 00:51:13.010 align:middle line:90% a single wafer. 00:51:13.010 --> 00:51:14.660 align:middle line:84% As the wafer diameter has increased, 00:51:14.660 --> 00:51:16.850 align:middle line:84% people are now at 12-inch wafers. 00:51:16.850 --> 00:51:19.910 align:middle line:84% The reactor would become gigantic if you had 12-- 00:51:19.910 --> 00:51:23.100 align:middle line:90% 12-inch wafers along a system. 00:51:23.100 --> 00:51:25.760 align:middle line:84% It's much more efficient to do a single wafer 00:51:25.760 --> 00:51:28.040 align:middle line:84% in a chamber at a time, and get the throughput high 00:51:28.040 --> 00:51:30.120 align:middle line:90% enough that you can do that. 00:51:30.120 --> 00:51:32.960 align:middle line:84% So this is an example of a system that's manufactured 00:51:32.960 --> 00:51:34.400 align:middle line:90% by Applied Materials. 00:51:34.400 --> 00:51:37.220 align:middle line:84% We actually have one here at MIT. 00:51:37.220 --> 00:51:39.530 align:middle line:84% This is sort of an overall view of the system. 00:51:39.530 --> 00:51:42.140 align:middle line:84% What it has is two different load locks. 00:51:42.140 --> 00:51:44.660 align:middle line:84% So the load locks are designed so you can open them up 00:51:44.660 --> 00:51:48.020 align:middle line:84% to air without putting air into the reactor-- 00:51:48.020 --> 00:51:49.020 align:middle line:90% into the reactor itself. 00:51:49.020 --> 00:51:50.660 align:middle line:84% So you can keep the reactor very clean. 00:51:50.660 --> 00:51:53.750 align:middle line:84% So you can open this door, put the wafer cassette in, 00:51:53.750 --> 00:51:56.357 align:middle line:84% and close it, and pump it down, get most of the air out. 00:51:56.357 --> 00:51:58.190 align:middle line:84% So you don't just open up the whole reactor, 00:51:58.190 --> 00:52:00.428 align:middle line:84% and get all the air in, and contaminate it. 00:52:00.428 --> 00:52:02.470 align:middle line:84% Because there's a couple of different load locks. 00:52:02.470 --> 00:52:04.810 align:middle line:84% There's a transfer-- a chamber in the center 00:52:04.810 --> 00:52:07.820 align:middle line:84% called the transfer chamber, which has a robot in it. 00:52:07.820 --> 00:52:09.430 align:middle line:84% And here's a picture of the robot. 00:52:09.430 --> 00:52:10.810 align:middle line:90% It has two little arms. 00:52:10.810 --> 00:52:12.010 align:middle line:90% Looks like a frog arm. 00:52:12.010 --> 00:52:15.730 align:middle line:84% The robot will come out, grab a wafer from the cassette, 00:52:15.730 --> 00:52:18.543 align:middle line:84% put it into the transfer chamber, 00:52:18.543 --> 00:52:19.960 align:middle line:84% and close this door, again, so you 00:52:19.960 --> 00:52:21.443 align:middle line:90% don't allow any air to get in. 00:52:21.443 --> 00:52:23.110 align:middle line:84% And then it'll take the wafer, according 00:52:23.110 --> 00:52:24.527 align:middle line:84% to how it's programmed, and put it 00:52:24.527 --> 00:52:26.838 align:middle line:84% in one of these three chambers depending on what is-- 00:52:26.838 --> 00:52:27.880 align:middle line:90% how you've programmed it. 00:52:27.880 --> 00:52:30.970 align:middle line:84% And there-- you can have up to three chambers here growing 00:52:30.970 --> 00:52:33.950 align:middle line:84% different materials, silicon or silicon germanium. 00:52:33.950 --> 00:52:36.790 align:middle line:84% So they increase the throughput by having three chambers 00:52:36.790 --> 00:52:39.370 align:middle line:84% instead of having three wafers all along the susceptor 00:52:39.370 --> 00:52:40.960 align:middle line:90% in a single chamber. 00:52:40.960 --> 00:52:43.390 align:middle line:84% Typical growth pressure in this type of equipment 00:52:43.390 --> 00:52:45.670 align:middle line:90% is in the 1 to 100 torr range. 00:52:45.670 --> 00:52:50.400 align:middle line:84% So both-- it turns out both surface reactions and mass 00:52:50.400 --> 00:52:52.020 align:middle line:90% transport are important. 00:52:52.020 --> 00:52:54.570 align:middle line:84% Growth temperatures can be anywhere from 400, 00:52:54.570 --> 00:52:57.480 align:middle line:84% very low if you're growing some special material. 00:52:57.480 --> 00:53:01.860 align:middle line:84% Typically, silicon germanium is grown between 600 and 800. 00:53:01.860 --> 00:53:05.020 align:middle line:84% And high-temperature silicon can be grown up to 1,100 00:53:05.020 --> 00:53:08.490 align:middle line:90% in this type of system. 00:53:08.490 --> 00:53:10.350 align:middle line:84% Just taking a little more zoomed-in look 00:53:10.350 --> 00:53:13.560 align:middle line:84% at this more modern reactor on the slide 23, 00:53:13.560 --> 00:53:15.053 align:middle line:90% this is that same picture. 00:53:15.053 --> 00:53:16.470 align:middle line:84% The only difference is here, we're 00:53:16.470 --> 00:53:19.560 align:middle line:84% showing in a single chamber right here, one of the chambers. 00:53:19.560 --> 00:53:21.810 align:middle line:84% And this green object is meant to be 00:53:21.810 --> 00:53:26.190 align:middle line:84% a wafer that is sitting on a susceptor that's spinning. 00:53:26.190 --> 00:53:29.890 align:middle line:84% So the wafer is spun at high velocity, 00:53:29.890 --> 00:53:32.010 align:middle line:84% say 30 revolutions per minute, to try 00:53:32.010 --> 00:53:35.160 align:middle line:84% to get good uniformity across the wafer. 00:53:35.160 --> 00:53:38.340 align:middle line:84% And the gas comes in here through a series of injectors 00:53:38.340 --> 00:53:41.970 align:middle line:84% as a plane flows across the entire wafer. 00:53:41.970 --> 00:53:43.770 align:middle line:90% So the inject is over here. 00:53:43.770 --> 00:53:47.580 align:middle line:84% And it flows laterally across, and the exit, or the exhaust, 00:53:47.580 --> 00:53:48.600 align:middle line:90% would be over here. 00:53:48.600 --> 00:53:50.760 align:middle line:90% And the wafer is spinning. 00:53:50.760 --> 00:53:53.400 align:middle line:84% So the chamber is designed with both in mind. 00:53:53.400 --> 00:53:57.660 align:middle line:84% Both the control of the boundary layer is important 00:53:57.660 --> 00:54:00.300 align:middle line:84% as well as the control of the temperature. 00:54:00.300 --> 00:54:02.700 align:middle line:84% There's a lamp bank that sits on top and bottom that 00:54:02.700 --> 00:54:05.940 align:middle line:84% tries to uniformly control the temperature from center to edge 00:54:05.940 --> 00:54:10.180 align:middle line:84% to get good uniformity of the epi. 00:54:10.180 --> 00:54:15.100 align:middle line:84% So that's a more modern single wafer piece of equipment. 00:54:15.100 --> 00:54:18.480 align:middle line:84% So let's just go back one more time to think about slide 00:54:18.480 --> 00:54:22.020 align:middle line:84% 24, the basic example of silicon epi. 00:54:22.020 --> 00:54:25.740 align:middle line:84% To show the slide before of temperature effects, 00:54:25.740 --> 00:54:28.637 align:middle line:84% depending on the relative speed of the surface reaction rate, 00:54:28.637 --> 00:54:30.720 align:middle line:84% and the mass transport through the boundary layer, 00:54:30.720 --> 00:54:34.920 align:middle line:84% you'll be in one of these two regimes probably. 00:54:34.920 --> 00:54:37.360 align:middle line:84% This curve on the lower right is interesting. 00:54:37.360 --> 00:54:43.140 align:middle line:84% It shows the reactant flow, or the partial pressure, effects. 00:54:43.140 --> 00:54:45.680 align:middle line:84% So this is-- before, I was showing growth rate versus 1 00:54:45.680 --> 00:54:46.550 align:middle line:90% over t. 00:54:46.550 --> 00:54:48.350 align:middle line:84% So these are at fixed temperature. 00:54:48.350 --> 00:54:50.310 align:middle line:84% Each curve is for a fixed temperature. 00:54:50.310 --> 00:54:52.730 align:middle line:84% And I'm showing growth rate versus the percent 00:54:52.730 --> 00:54:55.140 align:middle line:90% of the reactant. 00:54:55.140 --> 00:54:57.350 align:middle line:84% So this is the percent of dichlorosilane 00:54:57.350 --> 00:54:58.640 align:middle line:90% that's in the gas stream. 00:54:58.640 --> 00:55:02.310 align:middle line:84% Or you could think of it as the dichlorosilane flow. 00:55:02.310 --> 00:55:04.670 align:middle line:84% So interesting, if you go here at lower-- 00:55:04.670 --> 00:55:06.420 align:middle line:84% with dichlor-- in this particular reactor, 00:55:06.420 --> 00:55:08.003 align:middle line:84% again, this will be reactor dependent. 00:55:08.003 --> 00:55:09.930 align:middle line:84% At 840 degrees, the growth rate is 00:55:09.930 --> 00:55:13.110 align:middle line:84% independent of the flow of the dichlorosilane. 00:55:13.110 --> 00:55:14.220 align:middle line:90% So it doesn't matter. 00:55:14.220 --> 00:55:17.120 align:middle line:84% You can increase the partial pressure by a factor of 2 or 3. 00:55:17.120 --> 00:55:18.870 align:middle line:84% You're not upping the growth rate anymore. 00:55:18.870 --> 00:55:21.000 align:middle line:90% You have enough there. 00:55:21.000 --> 00:55:25.410 align:middle line:84% And you're really limited by the set of the partial pressure 00:55:25.410 --> 00:55:28.470 align:middle line:84% limited by the temperature, the rate at which the reaction can 00:55:28.470 --> 00:55:30.070 align:middle line:90% occur. 00:55:30.070 --> 00:55:31.650 align:middle line:84% So at low enough temperatures, it 00:55:31.650 --> 00:55:33.395 align:middle line:84% doesn't do you any good to flow more gas. 00:55:33.395 --> 00:55:34.770 align:middle line:84% As you get to higher temperatures 00:55:34.770 --> 00:55:36.990 align:middle line:84% here, say at a thousand, you can see 00:55:36.990 --> 00:55:39.120 align:middle line:84% that the growth rate is pretty much almost linearly 00:55:39.120 --> 00:55:41.850 align:middle line:84% proportionate up to a certain flow. 00:55:41.850 --> 00:55:43.950 align:middle line:84% It's proportional to the flow, or the-- 00:55:43.950 --> 00:55:50.580 align:middle line:84% because basically at this point, the surface reaction 00:55:50.580 --> 00:55:53.820 align:middle line:84% is going faster and faster as you give it 00:55:53.820 --> 00:55:57.640 align:middle line:84% more reactants than at some point, it saturates out. 00:55:57.640 --> 00:56:01.080 align:middle line:84% So depending on what temperature you're in, increasing the flow 00:56:01.080 --> 00:56:03.280 align:middle line:84% may or may not increase your growth rate. 00:56:03.280 --> 00:56:06.950 align:middle line:90% 00:56:06.950 --> 00:56:09.610 align:middle line:84% So we go to slide 25, there's another process 00:56:09.610 --> 00:56:11.140 align:middle line:90% that we want to talk about. 00:56:11.140 --> 00:56:15.100 align:middle line:84% We've talked a little bit about design of the reactor for growth 00:56:15.100 --> 00:56:18.520 align:middle line:84% rate uniformity and things like that, and epitaxial growth. 00:56:18.520 --> 00:56:20.440 align:middle line:84% But the uniformity of the layer thickness 00:56:20.440 --> 00:56:22.910 align:middle line:90% is really only one aspect. 00:56:22.910 --> 00:56:26.920 align:middle line:84% And these days, it's pretty well nailed. 00:56:26.920 --> 00:56:30.490 align:middle line:84% But another aspect that's just as important 00:56:30.490 --> 00:56:33.580 align:middle line:84% is called autodoping during epitaxial growth. 00:56:33.580 --> 00:56:35.815 align:middle line:84% And it's sort of unique to epitaxial growth. 00:56:35.815 --> 00:56:38.440 align:middle line:84% We don't usually think about it as being an issue in deposition 00:56:38.440 --> 00:56:39.940 align:middle line:90% of other types of films. 00:56:39.940 --> 00:56:42.740 align:middle line:84% But autodoping, as the name implies, 00:56:42.740 --> 00:56:45.310 align:middle line:84% is automatic, or unintentionally-- 00:56:45.310 --> 00:56:48.400 align:middle line:84% unintentional introduction of dopants into the epitaxial 00:56:48.400 --> 00:56:50.390 align:middle line:90% growing layer. 00:56:50.390 --> 00:56:52.960 align:middle line:84% And there's a couple of different regions 00:56:52.960 --> 00:56:53.800 align:middle line:90% it can come from. 00:56:53.800 --> 00:56:57.430 align:middle line:84% First, before we look at the reactor at the top, let's just-- 00:56:57.430 --> 00:57:00.910 align:middle line:84% look at the-- this diagram at the bottom of slide 25. 00:57:00.910 --> 00:57:03.160 align:middle line:84% And it's a plot of the concentration 00:57:03.160 --> 00:57:06.850 align:middle line:84% on a log scale of a dopant as a function of your distance 00:57:06.850 --> 00:57:08.400 align:middle line:90% into the epi layer. 00:57:08.400 --> 00:57:10.990 align:middle line:84% So if you-- it's a little bit backwards, 00:57:10.990 --> 00:57:14.530 align:middle line:84% but going from 0 to here is your distance into the epi. 00:57:14.530 --> 00:57:16.780 align:middle line:84% So the epi is getting thicker from left to right. 00:57:16.780 --> 00:57:20.810 align:middle line:84% And everything to the left of 0 is the substrate. 00:57:20.810 --> 00:57:22.150 align:middle line:90% So here, I'm in the substrate. 00:57:22.150 --> 00:57:26.055 align:middle line:84% So what we're doing in autodoping, you can imagine, 00:57:26.055 --> 00:57:28.180 align:middle line:84% I might have a substrate that's heavily doped, say, 00:57:28.180 --> 00:57:30.590 align:middle line:90% for a silicon bipolar process. 00:57:30.590 --> 00:57:32.950 align:middle line:84% So I have a substrate that has some concentration 00:57:32.950 --> 00:57:35.140 align:middle line:90% at the surface. 00:57:35.140 --> 00:57:38.410 align:middle line:84% And some of that dopant is going to make its way into the epi 00:57:38.410 --> 00:57:41.530 align:middle line:90% layer as the epi layer grows. 00:57:41.530 --> 00:57:43.760 align:middle line:90% And one process of-- 00:57:43.760 --> 00:57:45.760 align:middle line:84% say I'm trying to grow a lightly-doped epi layer 00:57:45.760 --> 00:57:46.720 align:middle line:90% on a heavily-doped substrate. 00:57:46.720 --> 00:57:49.220 align:middle line:84% One process by which it will get in there, you already know. 00:57:49.220 --> 00:57:50.262 align:middle line:90% And that's diffusion. 00:57:50.262 --> 00:57:52.720 align:middle line:84% And in fact, we know it's going to be a complementary error 00:57:52.720 --> 00:57:55.150 align:middle line:84% function type of profile if it's simply 00:57:55.150 --> 00:57:58.030 align:middle line:84% up diffusion from the substrate into the epi layer. 00:57:58.030 --> 00:57:59.600 align:middle line:90% So that's relatively simple. 00:57:59.600 --> 00:58:00.880 align:middle line:90% We know how to solve that. 00:58:00.880 --> 00:58:03.010 align:middle line:84% And we can limit the temperature and all that. 00:58:03.010 --> 00:58:08.410 align:middle line:84% And that will help us limit how far this autodoping goes 00:58:08.410 --> 00:58:09.800 align:middle line:90% into the epi layer. 00:58:09.800 --> 00:58:11.960 align:middle line:84% But there's another process here, 00:58:11.960 --> 00:58:14.920 align:middle line:84% which I'm showing by this sort of exponential dependence, 00:58:14.920 --> 00:58:17.530 align:middle line:84% exponentially-decreasing function. 00:58:17.530 --> 00:58:18.910 align:middle line:90% It has a certain decay length. 00:58:18.910 --> 00:58:23.090 align:middle line:84% And that's called front side, or vertical, autodoping. 00:58:23.090 --> 00:58:26.360 align:middle line:84% And vertical autodoping refers to the fact, not so much 00:58:26.360 --> 00:58:30.320 align:middle line:84% up diffusion, but just the fact that as I 00:58:30.320 --> 00:58:34.460 align:middle line:84% grow a single layer, if there is a phosphorus atom there 00:58:34.460 --> 00:58:37.440 align:middle line:84% on the surface, it tends to want to be up on the surface. 00:58:37.440 --> 00:58:40.160 align:middle line:84% So it tends to want to exchange sites instead of being 00:58:40.160 --> 00:58:41.570 align:middle line:90% buried with the silicon. 00:58:41.570 --> 00:58:44.520 align:middle line:84% And the phosphorus will ride up the surface. 00:58:44.520 --> 00:58:45.780 align:middle line:90% Same thing for arsenic. 00:58:45.780 --> 00:58:49.050 align:middle line:84% These dopants tend to like to be on a free surface. 00:58:49.050 --> 00:58:50.630 align:middle line:84% So as you're growing the crystal, 00:58:50.630 --> 00:58:53.750 align:middle line:84% they will ride themselves not by diffusion, but just 00:58:53.750 --> 00:58:56.870 align:middle line:84% by surface exchange at the surface. 00:58:56.870 --> 00:59:00.232 align:middle line:84% They'll ride themselves up the epi layer. 00:59:00.232 --> 00:59:01.690 align:middle line:84% And it's called vertical autodoping 00:59:01.690 --> 00:59:03.970 align:middle line:84% because it happens directly from the buried layer. 00:59:03.970 --> 00:59:06.220 align:middle line:84% It just goes vertically straight up. 00:59:06.220 --> 00:59:08.380 align:middle line:84% So that surface riding effect is really 00:59:08.380 --> 00:59:10.450 align:middle line:84% a function of the nature of the dopant 00:59:10.450 --> 00:59:12.700 align:middle line:84% and the temperature to a certain extent. 00:59:12.700 --> 00:59:15.430 align:middle line:84% But the nature of the dopant, it turns out the n-type dopants, 00:59:15.430 --> 00:59:17.800 align:middle line:84% arsenic, phosphorus in particular, 00:59:17.800 --> 00:59:19.690 align:middle line:90% like to be on the free surface. 00:59:19.690 --> 00:59:21.610 align:middle line:84% They're very happy being on a free surface. 00:59:21.610 --> 00:59:23.020 align:middle line:90% So they float to the surface. 00:59:23.020 --> 00:59:26.000 align:middle line:84% And they make it very hard to make a lightly-doped epi layer 00:59:26.000 --> 00:59:27.250 align:middle line:90% because they tend to autodope. 00:59:27.250 --> 00:59:29.507 align:middle line:90% They tend to have a long tail. 00:59:29.507 --> 00:59:31.090 align:middle line:84% Boron on, the other hand, doesn't mind 00:59:31.090 --> 00:59:32.990 align:middle line:90% being buried in the crystal. 00:59:32.990 --> 00:59:34.840 align:middle line:84% So boron doesn't have much autodoping. 00:59:34.840 --> 00:59:37.810 align:middle line:84% It does not surface ride very much. 00:59:37.810 --> 00:59:40.820 align:middle line:84% So this tail is not a diffusion-limited process. 00:59:40.820 --> 00:59:43.540 align:middle line:84% It has to do with the affinity of the dopant 00:59:43.540 --> 00:59:46.090 align:middle line:84% for the free surface during the growth process. 00:59:46.090 --> 00:59:48.160 align:middle line:84% And then there's this final background 00:59:48.160 --> 00:59:50.920 align:middle line:84% that you'll see in an epi layer, which 00:59:50.920 --> 00:59:53.260 align:middle line:84% is I'm showing here in green, called-- sometimes 00:59:53.260 --> 00:59:55.450 align:middle line:84% called backside, or lateral, autodoping. 00:59:55.450 --> 00:59:58.630 align:middle line:84% And what that refers to is you can get-- 00:59:58.630 --> 01:00:01.083 align:middle line:84% say if the back of the wafer is heavily doped, when 01:00:01.083 --> 01:00:02.750 align:middle line:84% you heat this thing up, you can actually 01:00:02.750 --> 01:00:06.470 align:middle line:84% get dopants coming off the back and finding their way 01:00:06.470 --> 01:00:08.150 align:middle line:90% onto the front of the wafer. 01:00:08.150 --> 01:00:10.400 align:middle line:84% They could be coming off the susceptor. 01:00:10.400 --> 01:00:14.720 align:middle line:84% Let's say you just grew a doped epi layer. 01:00:14.720 --> 01:00:17.330 align:middle line:84% The run before this, you're growing doped epi. 01:00:17.330 --> 01:00:18.680 align:middle line:90% And you put in some phosphorus. 01:00:18.680 --> 01:00:21.493 align:middle line:84% So that phosphorus now can still hang around in the susceptor. 01:00:21.493 --> 01:00:23.660 align:middle line:84% And then it can be transported when the susceptor is 01:00:23.660 --> 01:00:25.040 align:middle line:90% heated onto the wafer. 01:00:25.040 --> 01:00:27.140 align:middle line:84% So it could be from the back of the wafer, 01:00:27.140 --> 01:00:28.670 align:middle line:84% it could be-- from the susceptor, 01:00:28.670 --> 01:00:31.190 align:middle line:84% it could be from another wafer that has 01:00:31.190 --> 01:00:32.660 align:middle line:90% a heavily-doped region on it. 01:00:32.660 --> 01:00:35.300 align:middle line:84% So it's lateral because it's coming from somewhere else. 01:00:35.300 --> 01:00:38.030 align:middle line:84% And this is very often a function of what you just grew, 01:00:38.030 --> 01:00:40.430 align:middle line:84% maybe a memory effect that was grown in the reactor, 01:00:40.430 --> 01:00:43.435 align:middle line:84% and how it's designed, how well the susceptor is sealed, 01:00:43.435 --> 01:00:44.310 align:middle line:90% and things like that. 01:00:44.310 --> 01:00:46.610 align:middle line:84% So in the epitaxial growth process, 01:00:46.610 --> 01:00:49.430 align:middle line:84% we don't worry just about growth rate and its uniformity. 01:00:49.430 --> 01:00:51.890 align:middle line:84% We have to think carefully about temperature, and pressure, 01:00:51.890 --> 01:00:53.810 align:middle line:84% and the nature of the reactor design 01:00:53.810 --> 01:00:55.955 align:middle line:84% to avoid this type of autodoping effect. 01:00:55.955 --> 01:01:00.030 align:middle line:90% 01:01:00.030 --> 01:01:00.645 align:middle line:90% Slide 26. 01:01:00.645 --> 01:01:02.520 align:middle line:84% Let me-- let's say a little bit about epitaxy 01:01:02.520 --> 01:01:04.980 align:middle line:84% has evolved in CMOS and bipolar technology 01:01:04.980 --> 01:01:07.920 align:middle line:84% just so you have a view of where it comes from. 01:01:07.920 --> 01:01:12.120 align:middle line:84% Originally, in the old days, epi was originally primarily used 01:01:12.120 --> 01:01:14.130 align:middle line:84% just to provide a lightly-doped layer 01:01:14.130 --> 01:01:17.310 align:middle line:84% on top of a heavily-doped substrate for either bipolar 01:01:17.310 --> 01:01:19.650 align:middle line:90% or CMOS device processing. 01:01:19.650 --> 01:01:24.478 align:middle line:84% So epitaxy in the old days, and there is no chalk, because-- 01:01:24.478 --> 01:01:25.770 align:middle line:90% we had colored chalk last time. 01:01:25.770 --> 01:01:28.890 align:middle line:84% So it was simply a way if you had a heavily-doped wafer, 01:01:28.890 --> 01:01:31.942 align:middle line:84% and you needed a lightly-doped layer on top, maybe 10 microns, 01:01:31.942 --> 01:01:33.150 align:middle line:90% it was a way to produce that. 01:01:33.150 --> 01:01:34.567 align:middle line:84% There really wasn't any other way. 01:01:34.567 --> 01:01:39.750 align:middle line:84% And CMOS and bipolar foundries, or companies, didn't do epi. 01:01:39.750 --> 01:01:42.570 align:middle line:84% They bought epi wafers from one or two places 01:01:42.570 --> 01:01:45.210 align:middle line:84% in the world, half a dozen places where people grew it. 01:01:45.210 --> 01:01:48.150 align:middle line:84% So you would just buy an epi wafer. 01:01:48.150 --> 01:01:50.670 align:middle line:84% There's-- that's still a very large market, 01:01:50.670 --> 01:01:53.340 align:middle line:84% a very large component of epitaxial technology. 01:01:53.340 --> 01:01:55.455 align:middle line:84% It's simply to produce wafers that 01:01:55.455 --> 01:01:58.080 align:middle line:84% are heavily-doped substrate, and you need a lightly-doped layer 01:01:58.080 --> 01:01:59.270 align:middle line:90% on top. 01:01:59.270 --> 01:02:02.510 align:middle line:84% But actually, the last 10 or 15 years, the technique 01:02:02.510 --> 01:02:03.740 align:middle line:90% really has evolved. 01:02:03.740 --> 01:02:05.840 align:middle line:84% And now, it's used to do much more than that. 01:02:05.840 --> 01:02:09.050 align:middle line:84% People grow very complex doping profiles. 01:02:09.050 --> 01:02:11.330 align:middle line:84% They grow layers of different materials. 01:02:11.330 --> 01:02:14.930 align:middle line:84% Silicon germanium is grown on silicon, for example. 01:02:14.930 --> 01:02:16.690 align:middle line:90% It could be grown selectively. 01:02:16.690 --> 01:02:19.000 align:middle line:84% That means in only certain regions of the wafer 01:02:19.000 --> 01:02:23.110 align:middle line:84% where there are openings in the oxide, could be-- selective 01:02:23.110 --> 01:02:27.070 align:middle line:84% epi can be used for the source drains on top 01:02:27.070 --> 01:02:28.967 align:middle line:84% of the source drains to elevate them. 01:02:28.967 --> 01:02:30.550 align:middle line:84% It could be used in the active regions 01:02:30.550 --> 01:02:32.300 align:middle line:84% of the heterojunction bipolar transistors. 01:02:32.300 --> 01:02:35.380 align:middle line:84% So epi has gone from being just a type of wafer 01:02:35.380 --> 01:02:40.510 align:middle line:84% you buy to the point where different-- where manufacturers 01:02:40.510 --> 01:02:42.070 align:middle line:84% of silicon chips now, a lot of them 01:02:42.070 --> 01:02:44.260 align:middle line:90% have epi reactors in their fabs. 01:02:44.260 --> 01:02:46.270 align:middle line:90% And they use them. 01:02:46.270 --> 01:02:46.960 align:middle line:90% Thanks. 01:02:46.960 --> 01:02:53.574 align:middle line:84% And they use those reactors to actively grow structures. 01:02:53.574 --> 01:02:57.130 align:middle line:84% So as these-- as this evolution took place, 01:02:57.130 --> 01:02:59.310 align:middle line:90% several things had to happen. 01:02:59.310 --> 01:03:01.810 align:middle line:84% And that's why these new single wafer reactors have come out 01:03:01.810 --> 01:03:02.920 align:middle line:90% to a certain extent. 01:03:02.920 --> 01:03:04.630 align:middle line:84% This-- the cleanliness of the system 01:03:04.630 --> 01:03:06.890 align:middle line:84% and the purity of the gases had to be improved. 01:03:06.890 --> 01:03:09.100 align:middle line:84% And that's because the epitaxial growth temperatures 01:03:09.100 --> 01:03:10.270 align:middle line:90% are going down. 01:03:10.270 --> 01:03:12.760 align:middle line:84% And there's a tendency for impurities, like oxygen, 01:03:12.760 --> 01:03:15.590 align:middle line:84% to be incorporated much more readily at low temperatures. 01:03:15.590 --> 01:03:18.130 align:middle line:84% So in order to deal with that, as we 01:03:18.130 --> 01:03:20.408 align:middle line:84% lower the temperature of growth, we 01:03:20.408 --> 01:03:21.700 align:middle line:90% need to have a cleaner reactor. 01:03:21.700 --> 01:03:23.830 align:middle line:84% Otherwise, you're going to put a lot of oxygen 01:03:23.830 --> 01:03:25.315 align:middle line:90% in your epitaxial layer. 01:03:25.315 --> 01:03:27.820 align:middle line:90% 01:03:27.820 --> 01:03:30.130 align:middle line:84% Again, and pre-clean temperatures have also lowered. 01:03:30.130 --> 01:03:32.590 align:middle line:84% So the reduction in the growth temperature and reduction 01:03:32.590 --> 01:03:34.480 align:middle line:84% in the pre-clean temperature means epi now 01:03:34.480 --> 01:03:35.740 align:middle line:90% has a lower thermal budget. 01:03:35.740 --> 01:03:38.440 align:middle line:84% It used to be 1,150 degrees for an hour. 01:03:38.440 --> 01:03:40.630 align:middle line:84% Now, people grow silicon germanium layers 01:03:40.630 --> 01:03:43.257 align:middle line:90% at 700 for 10 minutes. 01:03:43.257 --> 01:03:44.090 align:middle line:90% Something like that. 01:03:44.090 --> 01:03:46.625 align:middle line:84% So it can be a relatively low thermal budget process now. 01:03:46.625 --> 01:03:47.500 align:middle line:90% And that's important. 01:03:47.500 --> 01:03:49.292 align:middle line:84% Because that means you can have epitaxial-- 01:03:49.292 --> 01:03:52.250 align:middle line:84% you can have other dope structures on the wafer, 01:03:52.250 --> 01:03:54.740 align:middle line:84% and not have to worry too much about TED, 01:03:54.740 --> 01:03:56.207 align:middle line:90% or other things going on. 01:03:56.207 --> 01:03:58.040 align:middle line:84% But in order to have low-temperature growth, 01:03:58.040 --> 01:04:00.380 align:middle line:84% you really have to have a clean reactor. 01:04:00.380 --> 01:04:01.760 align:middle line:84% So the new paradigms I've already 01:04:01.760 --> 01:04:04.400 align:middle line:84% showed you are single-wafer tools. 01:04:04.400 --> 01:04:06.900 align:middle line:84% There are rapid thermal epitaxy systems. 01:04:06.900 --> 01:04:08.480 align:middle line:84% So just like rapid thermal annealers, 01:04:08.480 --> 01:04:10.260 align:middle line:90% but you can grow epi in them. 01:04:10.260 --> 01:04:13.040 align:middle line:84% And there's also something that is not used in production very 01:04:13.040 --> 01:04:15.320 align:middle line:90% often, but is used in research. 01:04:15.320 --> 01:04:18.110 align:middle line:84% It's called molecular beam epitaxy. 01:04:18.110 --> 01:04:20.660 align:middle line:84% It is not a manufacturing process generally 01:04:20.660 --> 01:04:22.220 align:middle line:84% because the throughput is horrible. 01:04:22.220 --> 01:04:24.620 align:middle line:84% But it's a good-- it's very good for research. 01:04:24.620 --> 01:04:27.650 align:middle line:84% It's good for growing new types of materials for the first time. 01:04:27.650 --> 01:04:29.990 align:middle line:84% And then people figure out how to grow them by CVD. 01:04:29.990 --> 01:04:32.540 align:middle line:90% And they go from there. 01:04:32.540 --> 01:04:35.320 align:middle line:84% So that evolution that I just explained in words 01:04:35.320 --> 01:04:39.520 align:middle line:84% is shown here pictorially on slide 27 in the 1970s, 01:04:39.520 --> 01:04:42.113 align:middle line:84% probably before you guys were born, maybe. 01:04:42.113 --> 01:04:43.280 align:middle line:90% Or you were just being born. 01:04:43.280 --> 01:04:43.810 align:middle line:90% I don't know. 01:04:43.810 --> 01:04:44.530 align:middle line:90% Something like that. 01:04:44.530 --> 01:04:45.130 align:middle line:90% No, no. 01:04:45.130 --> 01:04:47.200 align:middle line:90% That's before your-- yeah, yeah. 01:04:47.200 --> 01:04:49.160 align:middle line:90% That's a long time ago. 01:04:49.160 --> 01:04:51.700 align:middle line:84% This is what epi was, the concentration versus depth 01:04:51.700 --> 01:04:53.200 align:middle line:90% in a bipolar transistor. 01:04:53.200 --> 01:04:56.197 align:middle line:90% The epi was simply formed-- 01:04:56.197 --> 01:04:57.280 align:middle line:90% the whole-- the collector. 01:04:57.280 --> 01:04:59.560 align:middle line:84% You just grew a silicon epi layer, 01:04:59.560 --> 01:05:02.840 align:middle line:84% you would buy a wafer with 5 microns of epi on it. 01:05:02.840 --> 01:05:04.450 align:middle line:84% And on top of that, you would form-- 01:05:04.450 --> 01:05:07.198 align:middle line:84% you would diffuse in or ion implant your base, 01:05:07.198 --> 01:05:08.740 align:middle line:84% and then you diffuse in your emitter. 01:05:08.740 --> 01:05:12.500 align:middle line:84% And you make an NPN bipolar transistor. 01:05:12.500 --> 01:05:14.900 align:middle line:84% And the epi layer was pretty thick, 5 microns. 01:05:14.900 --> 01:05:18.500 align:middle line:84% In the 1980s, again, it was silicon, epi, 01:05:18.500 --> 01:05:21.080 align:middle line:84% and the epi layer was shrunk by a factor of 10. 01:05:21.080 --> 01:05:24.440 align:middle line:84% Bipolar transistors, the total thickness was about half micron. 01:05:24.440 --> 01:05:27.110 align:middle line:84% So in those 10 years, the epi layer thickness 01:05:27.110 --> 01:05:28.670 align:middle line:90% had to drop by a factor of 10. 01:05:28.670 --> 01:05:31.855 align:middle line:84% People had to learn how to do autodoping reduction. 01:05:31.855 --> 01:05:33.980 align:middle line:84% They had to change the pressure and the temperature 01:05:33.980 --> 01:05:36.140 align:middle line:84% to try to reduce this out diffusion 01:05:36.140 --> 01:05:37.730 align:middle line:84% and the autodoping of this n plus 01:05:37.730 --> 01:05:40.490 align:middle line:84% buried layer into the active region. 01:05:40.490 --> 01:05:43.430 align:middle line:84% And finally, the 1990s, what people started to do, 01:05:43.430 --> 01:05:45.170 align:middle line:84% instead of just growing the collector, 01:05:45.170 --> 01:05:47.870 align:middle line:84% people started to grow not only the collector by epi, 01:05:47.870 --> 01:05:49.068 align:middle line:90% but the base. 01:05:49.068 --> 01:05:51.110 align:middle line:84% Instead of using an ion-implanted, look at this-- 01:05:51.110 --> 01:05:55.500 align:middle line:84% you can see this very broad profile with a retrograde. 01:05:55.500 --> 01:05:58.342 align:middle line:84% That-- not necessarily what the device designer wanted. 01:05:58.342 --> 01:06:00.550 align:middle line:84% That's just what they got when they ion implanted it. 01:06:00.550 --> 01:06:04.850 align:middle line:84% In fact, you can grow a very thin, narrow profile 01:06:04.850 --> 01:06:08.910 align:middle line:84% with a very abrupt base doping by epitaxial growth. 01:06:08.910 --> 01:06:12.710 align:middle line:84% And so epi transformed from just growing the substrate to being 01:06:12.710 --> 01:06:14.120 align:middle line:90% able to grow the base layer. 01:06:14.120 --> 01:06:17.660 align:middle line:84% Either a silicon base or people then inserted silicon germanium 01:06:17.660 --> 01:06:18.170 align:middle line:90% into it. 01:06:18.170 --> 01:06:22.130 align:middle line:84% And IBM and others commercialized the silicon-- 01:06:22.130 --> 01:06:24.270 align:middle line:84% silicon germanium, silicon heterojunction bipolar 01:06:24.270 --> 01:06:24.770 align:middle line:90% transistor. 01:06:24.770 --> 01:06:26.690 align:middle line:84% And that was all based on the development 01:06:26.690 --> 01:06:30.260 align:middle line:84% of silicon and silicon germanium epitaxial-- 01:06:30.260 --> 01:06:32.000 align:middle line:90% epitaxial growth technology. 01:06:32.000 --> 01:06:34.560 align:middle line:84% So it's really come from tens of microns. 01:06:34.560 --> 01:06:37.220 align:middle line:84% Now do people grow typical base today 01:06:37.220 --> 01:06:40.977 align:middle line:84% in a heterojunction bipolar transistor is 400 angstroms. 01:06:40.977 --> 01:06:43.310 align:middle line:84% So people are growing 400 angstroms in silicon germanium 01:06:43.310 --> 01:06:45.320 align:middle line:84% commercially in production, and thousands 01:06:45.320 --> 01:06:48.920 align:middle line:84% of transistors on circuits all the time, and selling them. 01:06:48.920 --> 01:06:51.690 align:middle line:90% 01:06:51.690 --> 01:06:53.570 align:middle line:84% So what are the most common gases 01:06:53.570 --> 01:06:58.320 align:middle line:84% here on page 28 today for epi, silane is quite common. 01:06:58.320 --> 01:07:00.860 align:middle line:84% And we saw the decomposition reaction before. 01:07:00.860 --> 01:07:04.580 align:middle line:84% Dichlorosilane is also popular, has a lower growth rate, 01:07:04.580 --> 01:07:07.010 align:middle line:84% but it gives you a better selectivity 01:07:07.010 --> 01:07:10.010 align:middle line:84% if you're trying to deposit the epi in windows. 01:07:10.010 --> 01:07:14.780 align:middle line:84% It's not too unusual to have a wafer where you-- 01:07:14.780 --> 01:07:19.550 align:middle line:84% in cross-section where you have oxide like this 01:07:19.550 --> 01:07:22.670 align:middle line:90% and you've patterned the oxide. 01:07:22.670 --> 01:07:25.820 align:middle line:84% And you can grow, it turns out, by a process 01:07:25.820 --> 01:07:26.870 align:middle line:90% called selective epi. 01:07:26.870 --> 01:07:30.200 align:middle line:84% You can grow it in such a way that the epi only 01:07:30.200 --> 01:07:33.840 align:middle line:84% grows in the silicon, where the silicon is exposed. 01:07:33.840 --> 01:07:36.980 align:middle line:84% So you get epi here and you get epi growth here. 01:07:36.980 --> 01:07:40.200 align:middle line:84% But you get nothing on top of the oxide. 01:07:40.200 --> 01:07:43.880 align:middle line:84% So it's automatically sort of self-aligned. 01:07:43.880 --> 01:07:45.870 align:middle line:84% And this is called selective epi growth. 01:07:45.870 --> 01:07:48.830 align:middle line:84% And to do that, you need dichlorosilane, 01:07:48.830 --> 01:07:50.630 align:middle line:90% and this is the reaction-- 01:07:50.630 --> 01:07:56.250 align:middle line:84% the silicon plus H2 plus Cl2 decomposes into another species. 01:07:56.250 --> 01:07:58.530 align:middle line:84% And that eventually decomposes on the surface to grow. 01:07:58.530 --> 01:08:00.060 align:middle line:90% And it forms HCl. 01:08:00.060 --> 01:08:03.090 align:middle line:84% And the HCl is important, it turns out, 01:08:03.090 --> 01:08:06.270 align:middle line:84% in etching the silicon off of the oxide, 01:08:06.270 --> 01:08:08.190 align:middle line:84% sort of keep it so you only deposit 01:08:08.190 --> 01:08:12.340 align:middle line:84% in the holes in the selective epi growth. 01:08:12.340 --> 01:08:15.230 align:middle line:84% So those are two very commonly-used gases. 01:08:15.230 --> 01:08:15.730 align:middle line:90% All right. 01:08:15.730 --> 01:08:17.390 align:middle line:90% Let me just summarize. 01:08:17.390 --> 01:08:21.319 align:middle line:84% So far on thin film dep and epi, we have two main types. 01:08:21.319 --> 01:08:23.319 align:middle line:84% We said there's chemical vapor deposition, which 01:08:23.319 --> 01:08:25.760 align:middle line:84% we talked about today, and physical vapor deposition, 01:08:25.760 --> 01:08:29.170 align:middle line:84% which we'll talk about in the next couple lectures. 01:08:29.170 --> 01:08:32.260 align:middle line:84% Traditional epi growth uses atmospheric pressure. 01:08:32.260 --> 01:08:36.880 align:middle line:84% More modern systems use low pressure methods, typically 01:08:36.880 --> 01:08:40.000 align:middle line:84% in the 1 to 100 torr, these days. 01:08:40.000 --> 01:08:43.660 align:middle line:84% We have a very simple model we developed for atmospheric CVD. 01:08:43.660 --> 01:08:46.210 align:middle line:84% And it led to rate-limiting regimes. 01:08:46.210 --> 01:08:49.810 align:middle line:84% Surface reaction rate, where the growth rate is exponentially 01:08:49.810 --> 01:08:51.370 align:middle line:84% dependent on temperature, activation 01:08:51.370 --> 01:08:54.250 align:middle line:90% energy of about 1.6 to 2 EV. 01:08:54.250 --> 01:08:56.290 align:middle line:84% And then at higher temperatures, you 01:08:56.290 --> 01:08:57.729 align:middle line:90% become mass transport limited. 01:08:57.729 --> 01:09:00.080 align:middle line:84% And you have very little dependence. 01:09:00.080 --> 01:09:03.130 align:middle line:84% Very slight dependence on temperature. 01:09:03.130 --> 01:09:06.040 align:middle line:84% The mass transport limit regime is good in some ways. 01:09:06.040 --> 01:09:07.840 align:middle line:90% It gives a high growth rate. 01:09:07.840 --> 01:09:10.090 align:middle line:84% But the bad thing about it, it's very sensitive to how 01:09:10.090 --> 01:09:11.270 align:middle line:90% you design your reactor. 01:09:11.270 --> 01:09:13.240 align:middle line:90% So it's a little bit tricky. 01:09:13.240 --> 01:09:16.720 align:middle line:84% Reactor geometry has to be well designed. 01:09:16.720 --> 01:09:20.920 align:middle line:84% LPCVD in the 100-millitorr range, in a hot well batch 01:09:20.920 --> 01:09:23.859 align:middle line:84% reactor is used every day in fabs, 01:09:23.859 --> 01:09:26.260 align:middle line:84% particularly to deposit things such as a polysilicon 01:09:26.260 --> 01:09:29.087 align:middle line:84% gate, low-temperature oxide, and silicon nitride. 01:09:29.087 --> 01:09:31.420 align:middle line:84% And we're going to talk about these three types of films 01:09:31.420 --> 01:09:35.000 align:middle line:84% and this type of process next time. 01:09:35.000 --> 01:09:35.500 align:middle line:90% All right. 01:09:35.500 --> 01:09:38.838 align:middle line:84% That's all I have for today's lecture. 01:09:38.838 --> 01:09:41.380 align:middle line:84% Just in case you came in late, the clipboard is going around. 01:09:41.380 --> 01:09:46.210 align:middle line:84% Make sure today, you put down on that clipboard your topic. 01:09:46.210 --> 01:09:48.700 align:middle line:84% I really want to have them so I can approve them. 01:09:48.700 --> 01:09:50.899 align:middle line:90% Tuesday is the deadline on that. 01:09:50.899 --> 01:09:52.899 align:middle line:84% Also, if you didn't pick up any prior homeworks, 01:09:52.899 --> 01:09:55.990 align:middle line:90% they're in the back. 01:09:55.990 --> 01:09:58.440 align:middle line:90% [AI Auditory Hallucination] 01:09:58.440 --> 01:10:00.000 align:middle line:90%