WEBVTT 00:00:00.000 --> 00:00:00.640 align:middle line:90% 00:00:00.640 --> 00:00:03.223 align:middle line:84% JUDY HOYT: All right, I think we can go ahead and get started. 00:00:03.223 --> 00:00:05.202 align:middle line:90% I'm going to put this-- 00:00:05.202 --> 00:00:07.470 align:middle line:84% is anybody here-- I think everybody who's here 00:00:07.470 --> 00:00:09.008 align:middle line:90% has already signed up. 00:00:09.008 --> 00:00:10.800 align:middle line:84% I'm going to put the clipboard in the back. 00:00:10.800 --> 00:00:14.160 align:middle line:84% And a couple of people haven't signed up, but wanted to-- 00:00:14.160 --> 00:00:17.545 align:middle line:84% everyone who signed up, I've actually approved your topics. 00:00:17.545 --> 00:00:20.170 align:middle line:84% But there's a couple people who haven't filled in their topics. 00:00:20.170 --> 00:00:23.760 align:middle line:84% So obviously, I can't approve them until they're filled in. 00:00:23.760 --> 00:00:25.260 align:middle line:90% So I'll put this back here. 00:00:25.260 --> 00:00:27.330 align:middle line:84% There's two handouts for today, the lecture notes 00:00:27.330 --> 00:00:29.820 align:middle line:84% and the homework, your final homework. 00:00:29.820 --> 00:00:32.455 align:middle line:84% And I'm handing back some homework 00:00:32.455 --> 00:00:34.080 align:middle line:84% sets that people haven't picked up yet. 00:00:34.080 --> 00:00:36.740 align:middle line:90% They're in the back. 00:00:36.740 --> 00:00:40.370 align:middle line:84% OK, so let's just take a look at where we are schedule wise. 00:00:40.370 --> 00:00:46.430 align:middle line:84% So today is November 9, and we're on the second lecture here 00:00:46.430 --> 00:00:47.420 align:middle line:90% on chapter 9. 00:00:47.420 --> 00:00:50.690 align:middle line:84% So hopefully you're reading chapter 9 of your textbook. 00:00:50.690 --> 00:00:54.230 align:middle line:84% And homework number 5, I'm handing out today. 00:00:54.230 --> 00:00:56.040 align:middle line:90% And that's your final homework. 00:00:56.040 --> 00:00:59.300 align:middle line:84% So I think it's an interesting problem set. 00:00:59.300 --> 00:01:01.550 align:middle line:84% One thing I want to remind you is that there is a part 00:01:01.550 --> 00:01:04.910 align:middle line:84% B on the back, so don't miss part B 00:01:04.910 --> 00:01:06.860 align:middle line:84% on the back side of the homework. 00:01:06.860 --> 00:01:09.530 align:middle line:90% There's three problems. 00:01:09.530 --> 00:01:14.090 align:middle line:84% And particularly problem number 3, 00:01:14.090 --> 00:01:15.890 align:middle line:90% I think you'll find interesting. 00:01:15.890 --> 00:01:18.620 align:middle line:84% It's about transient enhanced diffusion. 00:01:18.620 --> 00:01:23.210 align:middle line:84% Part A, you'll be doing some hand calculations of TED 00:01:23.210 --> 00:01:29.120 align:middle line:84% and using some analytic formulas for ion implant and diffusion. 00:01:29.120 --> 00:01:33.980 align:middle line:84% And part B, which is on the back side, again-- don't miss that-- 00:01:33.980 --> 00:01:35.780 align:middle line:84% you'll do some SUPREM simulations 00:01:35.780 --> 00:01:38.970 align:middle line:84% and then compare it to the hand calculation of the TED. 00:01:38.970 --> 00:01:42.300 align:middle line:84% And the other thing for me to mention about part B, 00:01:42.300 --> 00:01:46.260 align:middle line:84% this is the first homework where you're not 00:01:46.260 --> 00:01:48.870 align:middle line:90% given the SUPREM code. 00:01:48.870 --> 00:01:50.670 align:middle line:84% SUPREM codes you've been running all along, 00:01:50.670 --> 00:01:53.003 align:middle line:84% have been given to you that have been written by the TA, 00:01:53.003 --> 00:01:54.570 align:middle line:90% and you just have to run them. 00:01:54.570 --> 00:01:56.820 align:middle line:84% 3B, you're actually asked to write the code. 00:01:56.820 --> 00:01:58.680 align:middle line:84% Now, what you can do, clearly, is 00:01:58.680 --> 00:02:00.150 align:middle line:84% you've been given a lot of codes. 00:02:00.150 --> 00:02:02.130 align:middle line:84% You can take the examples of what 00:02:02.130 --> 00:02:03.630 align:middle line:84% you've been given as your starting-- 00:02:03.630 --> 00:02:05.172 align:middle line:84% you don't have to start from scratch. 00:02:05.172 --> 00:02:08.710 align:middle line:84% You can take other files and just modify them accordingly. 00:02:08.710 --> 00:02:11.790 align:middle line:84% And it's a relatively simple process. 00:02:11.790 --> 00:02:13.290 align:middle line:84% But I just wanted to make you aware 00:02:13.290 --> 00:02:14.957 align:middle line:84% that there is a part B on the back side. 00:02:14.957 --> 00:02:17.670 align:middle line:84% And I think problem number 3 should 00:02:17.670 --> 00:02:20.230 align:middle line:90% be an interesting one for you. 00:02:20.230 --> 00:02:22.730 align:middle line:84% And as I mentioned, that's the last homework for the course. 00:02:22.730 --> 00:02:25.970 align:middle line:84% The reason I make just the last problem set, 00:02:25.970 --> 00:02:29.410 align:middle line:84% this one is due on November 18, because I want to give you 00:02:29.410 --> 00:02:31.390 align:middle line:84% enough time between that last problem 00:02:31.390 --> 00:02:33.280 align:middle line:84% set so you can get the problems out 00:02:33.280 --> 00:02:37.940 align:middle line:84% of the way, enough time to be working on your final project. 00:02:37.940 --> 00:02:42.410 align:middle line:84% So that's either the written report or the oral presentation. 00:02:42.410 --> 00:02:44.950 align:middle line:84% So definitely, if you haven't done so, 00:02:44.950 --> 00:02:48.610 align:middle line:84% get started on your project soon. 00:02:48.610 --> 00:02:51.710 align:middle line:84% All right, so that's the mechanics of where we are. 00:02:51.710 --> 00:02:54.670 align:middle line:90% And let's go on with today's-- 00:02:54.670 --> 00:02:58.400 align:middle line:84% I'll start with today's lecture handout. 00:02:58.400 --> 00:03:00.200 align:middle line:90% This is handout number 31. 00:03:00.200 --> 00:03:03.830 align:middle line:84% And this is the second lecture on chapter 9. 00:03:03.830 --> 00:03:07.880 align:middle line:84% Chapter 9 is about thin film deposition and epitaxy. 00:03:07.880 --> 00:03:10.340 align:middle line:84% Last time, we gave you an introduction. 00:03:10.340 --> 00:03:14.720 align:middle line:84% We developed a very simple model for atmospheric chemical vapor 00:03:14.720 --> 00:03:15.560 align:middle line:90% deposition. 00:03:15.560 --> 00:03:19.610 align:middle line:84% And remember, we had two rate-limiting regimes. 00:03:19.610 --> 00:03:23.630 align:middle line:84% We had a regime of growth where the growth rate was determined 00:03:23.630 --> 00:03:26.180 align:middle line:90% by the surface reaction rate. 00:03:26.180 --> 00:03:29.090 align:middle line:84% And it had a certain activation energy, dependent exponentially 00:03:29.090 --> 00:03:30.140 align:middle line:90% on temperature. 00:03:30.140 --> 00:03:32.960 align:middle line:84% And we had another regime where the growth rate was 00:03:32.960 --> 00:03:34.207 align:middle line:90% determined by mass transport. 00:03:34.207 --> 00:03:35.540 align:middle line:90% That was the rate-limiting step. 00:03:35.540 --> 00:03:38.290 align:middle line:90% 00:03:38.290 --> 00:03:40.900 align:middle line:84% We also talked about the fact that the mass transport 00:03:40.900 --> 00:03:42.610 align:middle line:84% regime gives you a very high growth rate. 00:03:42.610 --> 00:03:44.830 align:middle line:84% This tends to be at high temperatures, 00:03:44.830 --> 00:03:46.502 align:middle line:84% where the surface reaction is fast. 00:03:46.502 --> 00:03:47.710 align:middle line:90% Gives you a high growth rate. 00:03:47.710 --> 00:03:51.340 align:middle line:84% It's relatively independent of temperature in contrast 00:03:51.340 --> 00:03:53.890 align:middle line:84% to the surface reaction control regime. 00:03:53.890 --> 00:03:56.590 align:middle line:84% But it's also sensitive to reactor geometry. 00:03:56.590 --> 00:03:59.590 align:middle line:84% So the growth rate and uniformity 00:03:59.590 --> 00:04:05.100 align:middle line:84% depend quite a bit on how the gas flows through the reactor. 00:04:05.100 --> 00:04:09.150 align:middle line:84% We also introduced this LPCVD, which 00:04:09.150 --> 00:04:12.240 align:middle line:84% is an acronym that stands for low-pressure CVD. 00:04:12.240 --> 00:04:14.640 align:middle line:84% And we introduced the use of LPCVD 00:04:14.640 --> 00:04:18.670 align:middle line:84% in a hot-wall batch reactor, which is in just about every fab 00:04:18.670 --> 00:04:21.279 align:middle line:90% that you'll ever see. 00:04:21.279 --> 00:04:23.640 align:middle line:84% And it's used to deposit very commonly-used films, 00:04:23.640 --> 00:04:27.900 align:middle line:84% such as polysilicon for the gate or other applications, oxide, 00:04:27.900 --> 00:04:30.850 align:middle line:84% deposited oxide, and deposited silicon nitride. 00:04:30.850 --> 00:04:33.450 align:middle line:84% So in the next lecture or two, what I want to discuss 00:04:33.450 --> 00:04:36.450 align:middle line:84% are some specific examples of CVD. 00:04:36.450 --> 00:04:38.190 align:middle line:84% So I want to discuss specifically 00:04:38.190 --> 00:04:43.710 align:middle line:84% the aspects of the deposition of polycrystalline silicon, 00:04:43.710 --> 00:04:47.950 align:middle line:84% deposition of silicon dioxide, and silicon nitride. 00:04:47.950 --> 00:04:51.300 align:middle line:84% And also, what we'll talk about towards the end of today's 00:04:51.300 --> 00:04:55.020 align:middle line:84% lecture is introduction to the basics of physical vapor 00:04:55.020 --> 00:04:55.710 align:middle line:90% deposition. 00:04:55.710 --> 00:04:59.450 align:middle line:84% So we have chemical vapor deposition and physical. 00:04:59.450 --> 00:05:02.150 align:middle line:84% OK, let's go on to page number 2, 00:05:02.150 --> 00:05:03.930 align:middle line:84% or slide number 2 on your handout. 00:05:03.930 --> 00:05:07.370 align:middle line:84% This is just a chart I took right out 00:05:07.370 --> 00:05:11.030 align:middle line:84% of your textbook in the section in chapter 9 00:05:11.030 --> 00:05:13.490 align:middle line:84% on manufacturing methods for CVD. 00:05:13.490 --> 00:05:16.910 align:middle line:84% And what it gives you is just some common thin films 00:05:16.910 --> 00:05:18.080 align:middle line:90% that are deposited by CVD. 00:05:18.080 --> 00:05:21.080 align:middle line:84% I think these are the most common, episilicon, polysilicon, 00:05:21.080 --> 00:05:25.260 align:middle line:90% silicon nitride, and SiO2. 00:05:25.260 --> 00:05:27.250 align:middle line:84% It talks about the type of equipment used. 00:05:27.250 --> 00:05:29.910 align:middle line:84% For example, we talked last time about epitaxial silicon 00:05:29.910 --> 00:05:33.120 align:middle line:84% can be either atmospheric or low-pressure CVD. 00:05:33.120 --> 00:05:35.520 align:middle line:84% There's actually another type of epitaxial silicon 00:05:35.520 --> 00:05:37.980 align:middle line:84% that wasn't listed here because it's not commonly 00:05:37.980 --> 00:05:40.063 align:middle line:90% used in fabs that much. 00:05:40.063 --> 00:05:41.730 align:middle line:84% But it should be listed for completeness 00:05:41.730 --> 00:05:43.320 align:middle line:90% if you want to put it down. 00:05:43.320 --> 00:05:47.650 align:middle line:84% Ultra high-vacuum CVD, UHVCVD is another method 00:05:47.650 --> 00:05:52.710 align:middle line:84% of depositing or growing epitaxial silicon. 00:05:52.710 --> 00:05:54.250 align:middle line:84% And that wasn't listed in the text. 00:05:54.250 --> 00:05:57.610 align:middle line:84% Some typical reactions that we talked about, for instance, 00:05:57.610 --> 00:06:01.647 align:middle line:84% the use of silane to grow epitaxial silicon 00:06:01.647 --> 00:06:02.480 align:middle line:90% at high temperature. 00:06:02.480 --> 00:06:05.660 align:middle line:84% The silane decomposes into silicon and hydrogen and evolves 00:06:05.660 --> 00:06:09.860 align:middle line:84% hydrogen. Under comments here is listed some typical temperature 00:06:09.860 --> 00:06:11.720 align:middle line:90% range. 00:06:11.720 --> 00:06:14.060 align:middle line:84% Here, if you're an atmospheric pressure CVD, 00:06:14.060 --> 00:06:17.090 align:middle line:84% you typically don't grow too much below 1,000, 00:06:17.090 --> 00:06:19.280 align:middle line:90% maybe 900 degrees centigrade. 00:06:19.280 --> 00:06:21.240 align:middle line:84% Much more common these days, though, 00:06:21.240 --> 00:06:24.740 align:middle line:84% is reduced pressure deposition of epitaxial silicon. 00:06:24.740 --> 00:06:25.730 align:middle line:90% We talked about that. 00:06:25.730 --> 00:06:27.500 align:middle line:84% Here, the deposition temperatures 00:06:27.500 --> 00:06:29.300 align:middle line:84% are lowered significantly compared 00:06:29.300 --> 00:06:30.350 align:middle line:90% to atmospheric pressure. 00:06:30.350 --> 00:06:34.310 align:middle line:84% And people can deposit epitaxial silicon typically 00:06:34.310 --> 00:06:38.600 align:middle line:84% all the way down to 750, maybe 700 even 00:06:38.600 --> 00:06:40.790 align:middle line:90% in the reduced-pressure systems. 00:06:40.790 --> 00:06:44.600 align:middle line:84% Besides silane, we also talked about silicon tetrachloride 00:06:44.600 --> 00:06:48.830 align:middle line:84% being used, also, trichlorosilane, shown here, 00:06:48.830 --> 00:06:52.530 align:middle line:84% and dichlorosilane, probably one of the most common gases, 00:06:52.530 --> 00:06:57.170 align:middle line:84% so dichlorosilane for selective epi growth. 00:06:57.170 --> 00:06:59.570 align:middle line:84% Polysilicon, which we mentioned last time, 00:06:59.570 --> 00:07:04.430 align:middle line:84% it's put down in a batch furnace, typically, by LPCVD. 00:07:04.430 --> 00:07:07.490 align:middle line:84% The gas reactions and the precursors 00:07:07.490 --> 00:07:10.340 align:middle line:84% are the same, typically, as used as epitaxial silicon. 00:07:10.340 --> 00:07:13.310 align:middle line:84% I would say that silane is really 00:07:13.310 --> 00:07:17.480 align:middle line:84% the precursor that people use for polycrystalline silicon. 00:07:17.480 --> 00:07:19.910 align:middle line:84% Polysilicon is typically put down somewhere-- 00:07:19.910 --> 00:07:21.830 align:middle line:84% and it's pretty tight temperature range, 00:07:21.830 --> 00:07:26.300 align:middle line:90% 575 to 650 degrees centigrade. 00:07:26.300 --> 00:07:30.658 align:middle line:84% The morphology of the film and its grain structure-- 00:07:30.658 --> 00:07:32.450 align:middle line:84% and we'll talk about that in this lecture-- 00:07:32.450 --> 00:07:35.090 align:middle line:84% depends on the deposition conditions, the temperature 00:07:35.090 --> 00:07:36.840 align:middle line:90% and the doping quite a bit. 00:07:36.840 --> 00:07:39.170 align:middle line:84% So you adjust your temperature to achieve 00:07:39.170 --> 00:07:42.620 align:middle line:84% the type of grain structure that you need for your application. 00:07:42.620 --> 00:07:47.060 align:middle line:84% Silicon nitride can be put down by an LPCVD process as well 00:07:47.060 --> 00:07:50.150 align:middle line:84% as we talked last time about plasma-enhanced CVD. 00:07:50.150 --> 00:07:52.700 align:middle line:84% This is becoming more and more common. 00:07:52.700 --> 00:07:56.390 align:middle line:84% For a low-pressure CVD, there are two different types 00:07:56.390 --> 00:07:59.420 align:middle line:84% of reactions that are commonly used. 00:07:59.420 --> 00:08:04.772 align:middle line:84% Silane is reacted with ammonia to create a silicon nitride. 00:08:04.772 --> 00:08:06.230 align:middle line:84% The other one, the most common one, 00:08:06.230 --> 00:08:09.410 align:middle line:84% the one that's used here at MIT in LPCVD 00:08:09.410 --> 00:08:12.770 align:middle line:84% is to use dichlorosilane, react that with ammonia 00:08:12.770 --> 00:08:15.470 align:middle line:90% and obtain silicon nitride. 00:08:15.470 --> 00:08:17.950 align:middle line:84% This reaction typically goes or is typically 00:08:17.950 --> 00:08:19.930 align:middle line:84% deposited to get reasonable depth 00:08:19.930 --> 00:08:22.780 align:middle line:84% rates at around 800 degrees C. So 00:08:22.780 --> 00:08:25.120 align:middle line:84% LPCVD nitride is one of those things 00:08:25.120 --> 00:08:27.910 align:middle line:84% that it's at a temperature where we talked 00:08:27.910 --> 00:08:30.790 align:middle line:84% about in the last several lectures 00:08:30.790 --> 00:08:32.780 align:middle line:84% transit-enhanced diffusion can be an issue. 00:08:32.780 --> 00:08:36.740 align:middle line:84% So it's right in that intermediate temperature range. 00:08:36.740 --> 00:08:38.140 align:middle line:90% What else do people use? 00:08:38.140 --> 00:08:40.179 align:middle line:84% Well, an alternative people sometimes 00:08:40.179 --> 00:08:42.760 align:middle line:84% use is low temperatures plasma-enhanced 00:08:42.760 --> 00:08:45.770 align:middle line:90% CVD for a passivating layer. 00:08:45.770 --> 00:08:48.970 align:middle line:84% Doesn't have as good of properties 00:08:48.970 --> 00:08:51.370 align:middle line:84% or the same properties as LPCVD nitride. 00:08:51.370 --> 00:08:54.270 align:middle line:84% And I'll talk a little bit more about that. 00:08:54.270 --> 00:08:56.880 align:middle line:84% The last one here, probably one of the most important, silicon 00:08:56.880 --> 00:08:58.502 align:middle line:90% dioxide. 00:08:58.502 --> 00:08:59.960 align:middle line:84% Remember, earlier in the course, we 00:08:59.960 --> 00:09:01.533 align:middle line:84% talked about growing silicon dioxide. 00:09:01.533 --> 00:09:03.200 align:middle line:84% Well, now we're talking about depositing 00:09:03.200 --> 00:09:05.370 align:middle line:90% it, which is quite different. 00:09:05.370 --> 00:09:09.680 align:middle line:84% We react typically silane as a precursor with oxygen 00:09:09.680 --> 00:09:12.950 align:middle line:84% to form SiO2 so that you don't have to have 00:09:12.950 --> 00:09:14.420 align:middle line:90% silicon exposed on the wafer. 00:09:14.420 --> 00:09:17.580 align:middle line:84% You don't consume, necessarily, much silicon on the wafer. 00:09:17.580 --> 00:09:19.490 align:middle line:84% This is listed of between 200 and 800. 00:09:19.490 --> 00:09:21.500 align:middle line:90% That's a pretty wide range. 00:09:21.500 --> 00:09:24.090 align:middle line:84% Really, the most common is between 200 and 500. 00:09:24.090 --> 00:09:26.900 align:middle line:84% In fact, LTO is a very common acronym 00:09:26.900 --> 00:09:30.950 align:middle line:84% you'll find in the fab, LTO, as we call it. 00:09:30.950 --> 00:09:34.610 align:middle line:84% Typically, LTO has put down between 350 and 450. 00:09:34.610 --> 00:09:38.540 align:middle line:84% That's really a more reasonable range. 00:09:38.540 --> 00:09:41.220 align:middle line:84% In order to get reasonable properties, 00:09:41.220 --> 00:09:43.820 align:middle line:84% so to reduce the etch rate and things, 00:09:43.820 --> 00:09:46.460 align:middle line:84% people usually have to high-temperature anneal LTO. 00:09:46.460 --> 00:09:50.080 align:middle line:84% So they have to what they call densify it because it's not 00:09:50.080 --> 00:09:52.870 align:middle line:90% a stoichiometric SiO2. 00:09:52.870 --> 00:09:57.190 align:middle line:84% In modern fabs, you will also find a very common process 00:09:57.190 --> 00:09:58.660 align:middle line:90% called TEOS. 00:09:58.660 --> 00:10:03.220 align:middle line:84% People refer to LTO sometimes as TEOS. 00:10:03.220 --> 00:10:09.070 align:middle line:84% And TEOS refers to this particular organic compound. 00:10:09.070 --> 00:10:12.430 align:middle line:84% And you notice it's a compound that includes silicon along 00:10:12.430 --> 00:10:16.810 align:middle line:84% with an organic group, a C2H5 group. 00:10:16.810 --> 00:10:21.220 align:middle line:84% You react that with ozone at a very low temperature, say, 400 00:10:21.220 --> 00:10:27.070 align:middle line:84% or below, you can get very high DEP rates of SiO2. 00:10:27.070 --> 00:10:28.750 align:middle line:84% Now, the only thing about TEOS is 00:10:28.750 --> 00:10:30.280 align:middle line:84% that you need to be aware of, what's 00:10:30.280 --> 00:10:34.210 align:middle line:84% different is there's an organic element in the precursor, some 00:10:34.210 --> 00:10:36.610 align:middle line:84% of which can potentially be incorporated in the film. 00:10:36.610 --> 00:10:40.330 align:middle line:84% And so TEOS films may have more carbon incorporation 00:10:40.330 --> 00:10:41.350 align:middle line:90% or carbon contamination. 00:10:41.350 --> 00:10:43.690 align:middle line:84% And some of their properties will 00:10:43.690 --> 00:10:45.520 align:middle line:90% be a little different from LTO. 00:10:45.520 --> 00:10:50.880 align:middle line:84% But TEOS is very commonly used in fabs today. 00:10:50.880 --> 00:10:53.790 align:middle line:84% OK, so that's just a little background 00:10:53.790 --> 00:10:56.340 align:middle line:84% on the different types of materials. 00:10:56.340 --> 00:10:59.910 align:middle line:84% Let's remind ourselves about low-pressure, here on slide 3, 00:10:59.910 --> 00:11:03.420 align:middle line:90% about low-pressure CVD. 00:11:03.420 --> 00:11:07.090 align:middle line:84% Atmospheric pressure systems have some major drawbacks. 00:11:07.090 --> 00:11:10.290 align:middle line:84% For example, if you operate them at high temperature, 00:11:10.290 --> 00:11:13.180 align:middle line:84% the gas flow makes a really big difference. 00:11:13.180 --> 00:11:14.850 align:middle line:84% And they have to be designed just right. 00:11:14.850 --> 00:11:16.950 align:middle line:84% You typically need a horizontal configuration, 00:11:16.950 --> 00:11:18.300 align:middle line:90% as we showed last time. 00:11:18.300 --> 00:11:20.370 align:middle line:84% And you only can really do a few wafers at a time 00:11:20.370 --> 00:11:22.440 align:middle line:84% in order to get any kind of real uniformity. 00:11:22.440 --> 00:11:25.800 align:middle line:84% So atmospheric pressure systems are really not 00:11:25.800 --> 00:11:27.270 align:middle line:90% that commonly used anymore. 00:11:27.270 --> 00:11:29.460 align:middle line:84% If you try to operate them at low T, you say, 00:11:29.460 --> 00:11:32.010 align:middle line:84% OK, I'll go to low temperature where 00:11:32.010 --> 00:11:34.500 align:middle line:84% there will be surface reaction rate-limited, 00:11:34.500 --> 00:11:40.250 align:middle line:84% but the deposition rate goes down quite a bit. 00:11:40.250 --> 00:11:42.070 align:middle line:84% So there is a solution, though, which 00:11:42.070 --> 00:11:45.320 align:middle line:84% is to operate the reactor at lower pressures. 00:11:45.320 --> 00:11:50.140 align:middle line:84% So instead of 760 torr, you might operate at 1 to 10 torr. 00:11:50.140 --> 00:11:53.260 align:middle line:84% Now, if you go back to the mass transfer-limited regime, 00:11:53.260 --> 00:11:55.390 align:middle line:84% remember, there was a transport coefficient, 00:11:55.390 --> 00:12:01.120 align:middle line:84% H sub G. So this tells you this is the limiting step in getting 00:12:01.120 --> 00:12:04.390 align:middle line:84% from the mainstream of the gas to diffuse through the boundary 00:12:04.390 --> 00:12:06.550 align:middle line:84% layer to the surface where it can react. 00:12:06.550 --> 00:12:09.820 align:middle line:84% So this mass transport coefficient HG 00:12:09.820 --> 00:12:13.600 align:middle line:84% was equal to a diffusivity through the boundary layer 00:12:13.600 --> 00:12:15.590 align:middle line:84% divided by the boundary layer thickness. 00:12:15.590 --> 00:12:17.560 align:middle line:84% So delta here is the boundary layer thickness. 00:12:17.560 --> 00:12:21.310 align:middle line:84% But you notice the diffusivity in the gas phase 00:12:21.310 --> 00:12:23.470 align:middle line:84% is proportional to 1 over the pressure, 1 00:12:23.470 --> 00:12:24.470 align:middle line:90% over the total pressure. 00:12:24.470 --> 00:12:27.880 align:middle line:84% So this is the key leverage that you're 00:12:27.880 --> 00:12:30.410 align:middle line:84% going to use by reducing the pressure. 00:12:30.410 --> 00:12:34.300 align:middle line:84% So the diffusivity will go up by about 760 times. 00:12:34.300 --> 00:12:37.030 align:middle line:84% If I go from atmospheric pressure-- remember, 00:12:37.030 --> 00:12:41.380 align:middle line:84% atmospheric pressure is 760 torr-- if I go down to one torr, 00:12:41.380 --> 00:12:47.170 align:middle line:84% this is going to go up by 760, while the boundary layer 00:12:47.170 --> 00:12:49.810 align:middle line:84% itself is also going to increase but not quite as 00:12:49.810 --> 00:12:51.430 align:middle line:90% much, only seven times. 00:12:51.430 --> 00:12:54.400 align:middle line:84% So we can get about a factor of 100 00:12:54.400 --> 00:12:58.450 align:middle line:84% in HG if I drop the pressure from atmospheric down 00:12:58.450 --> 00:13:00.730 align:middle line:90% to one torr. 00:13:00.730 --> 00:13:04.180 align:middle line:84% And then in the one torr regime, the transport 00:13:04.180 --> 00:13:07.240 align:middle line:84% of reactants from the gas phase through the surface boundary 00:13:07.240 --> 00:13:09.080 align:middle line:84% layer is no longer rate-limiting. 00:13:09.080 --> 00:13:13.877 align:middle line:84% So it can really speed up that process. 00:13:13.877 --> 00:13:15.710 align:middle line:84% In fact, we can show this graph-- let's look 00:13:15.710 --> 00:13:19.040 align:middle line:84% at slide number 4, and you can see this graphically. 00:13:19.040 --> 00:13:21.290 align:middle line:84% This is a plot similar to what I showed last time. 00:13:21.290 --> 00:13:24.320 align:middle line:84% It's the growth rate on a log scale versus 1 over T. 00:13:24.320 --> 00:13:26.600 align:middle line:84% So we typically call this an Arrhenius plot. 00:13:26.600 --> 00:13:29.720 align:middle line:84% And for atmospheric pressure CVD, 00:13:29.720 --> 00:13:33.470 align:middle line:84% it's this lower line right here is the net growth rate. 00:13:33.470 --> 00:13:35.910 align:middle line:84% And at low temperatures here, your surface reaction 00:13:35.910 --> 00:13:36.410 align:middle line:90% controlled. 00:13:36.410 --> 00:13:37.260 align:middle line:90% But look what happens. 00:13:37.260 --> 00:13:38.090 align:middle line:84% You get to a certain temperature, 00:13:38.090 --> 00:13:39.620 align:middle line:90% and there's a knee in the curve. 00:13:39.620 --> 00:13:41.370 align:middle line:90% And it sort of flattens out. 00:13:41.370 --> 00:13:43.310 align:middle line:90% So at 760 torr, it flattens out. 00:13:43.310 --> 00:13:47.150 align:middle line:84% And you can only get a certain growth rate 00:13:47.150 --> 00:13:48.500 align:middle line:90% for atmospheric systems. 00:13:48.500 --> 00:13:50.870 align:middle line:84% Now, if I drop the pressure to one torr, 00:13:50.870 --> 00:13:53.480 align:middle line:84% again, I could get about a factor of 100 increase. 00:13:53.480 --> 00:13:54.740 align:middle line:90% Remember, this is a log scale. 00:13:54.740 --> 00:13:57.330 align:middle line:84% So going from here to here doesn't look like much. 00:13:57.330 --> 00:14:00.120 align:middle line:84% But this could be a factor of 10 to 100 00:14:00.120 --> 00:14:05.795 align:middle line:84% in growth rate at one torr because why is this happening? 00:14:05.795 --> 00:14:08.420 align:middle line:84% Because the gas phase diffusion rate through the boundary layer 00:14:08.420 --> 00:14:11.910 align:middle line:84% is increasing as I drop that pressure. 00:14:11.910 --> 00:14:16.770 align:middle line:84% So can really up the growth rate by going to low pressure. 00:14:16.770 --> 00:14:18.210 align:middle line:84% And so that's how the two curves-- 00:14:18.210 --> 00:14:21.290 align:middle line:84% notice the two curves converge, though, pretty closely. 00:14:21.290 --> 00:14:23.240 align:middle line:84% When you get down to low enough temperatures 00:14:23.240 --> 00:14:25.430 align:middle line:84% that you really are surface reaction controlled, 00:14:25.430 --> 00:14:29.090 align:middle line:84% you can drop the pressure by a factor of 1,000. 00:14:29.090 --> 00:14:30.710 align:middle line:84% You only get a very small increase. 00:14:30.710 --> 00:14:34.460 align:middle line:84% You'll get some increase or some change, rather, in the growth 00:14:34.460 --> 00:14:36.650 align:middle line:84% rate but not so much because the surface reaction 00:14:36.650 --> 00:14:40.500 align:middle line:84% rate is rate-limiting at that temperature. 00:14:40.500 --> 00:14:42.000 align:middle line:84% So that's one of the big motivations 00:14:42.000 --> 00:14:44.910 align:middle line:90% for going to low pressure. 00:14:44.910 --> 00:14:49.690 align:middle line:84% Let's go to slide number 5, and we talked about this last time, 00:14:49.690 --> 00:14:51.480 align:middle line:84% but I want to give more detail now 00:14:51.480 --> 00:14:53.370 align:middle line:84% on the deposition of polycrystalline silicon. 00:14:53.370 --> 00:14:56.370 align:middle line:84% We introduced this rudimentary reactor. 00:14:56.370 --> 00:14:58.290 align:middle line:84% This type of setup is found pretty much 00:14:58.290 --> 00:15:01.530 align:middle line:84% in every fab, maybe slightly different configuration. 00:15:01.530 --> 00:15:04.680 align:middle line:84% But it's essentially a hot-wall furnace. 00:15:04.680 --> 00:15:08.250 align:middle line:84% It's a hot-wall furnace and it's resistively heated. 00:15:08.250 --> 00:15:13.710 align:middle line:84% The wafers are usually put in quartz boats or quartzware. 00:15:13.710 --> 00:15:16.560 align:middle line:84% And they usually stand up together very close 00:15:16.560 --> 00:15:17.920 align:middle line:90% to each other. 00:15:17.920 --> 00:15:19.120 align:middle line:90% And it's a batch process. 00:15:19.120 --> 00:15:22.170 align:middle line:84% So you can get 25, 50, maybe 75 wafers 00:15:22.170 --> 00:15:24.600 align:middle line:90% in a batch, quite a few wafers. 00:15:24.600 --> 00:15:26.580 align:middle line:84% You have some gas sources, consisting 00:15:26.580 --> 00:15:29.520 align:middle line:84% of silane and nitrogen and maybe dopants 00:15:29.520 --> 00:15:32.340 align:middle line:84% if you want, that flow into this quartz tube 00:15:32.340 --> 00:15:34.530 align:middle line:90% into the hot zone in there. 00:15:34.530 --> 00:15:37.810 align:middle line:84% The pressure, the low pressure is maintained by a vacuum pump. 00:15:37.810 --> 00:15:40.830 align:middle line:84% So this is not an open tube furnace. 00:15:40.830 --> 00:15:44.670 align:middle line:84% The mouth of the furnace where the wafers come in and out 00:15:44.670 --> 00:15:50.040 align:middle line:84% actually is sealed with an O-ring so you can pull a vacuum. 00:15:50.040 --> 00:15:54.900 align:middle line:84% Typical depositions, usually, polysilicon 00:15:54.900 --> 00:15:57.750 align:middle line:84% is deposited on oxide, maybe, or on an oxide 00:15:57.750 --> 00:15:59.790 align:middle line:90% with silicon windows. 00:15:59.790 --> 00:16:01.900 align:middle line:90% It's a nonselective process. 00:16:01.900 --> 00:16:06.720 align:middle line:84% So it goes over the entire wafer on all structures. 00:16:06.720 --> 00:16:09.400 align:middle line:84% The diluent gas, so what is the usually flow? 00:16:09.400 --> 00:16:10.650 align:middle line:90% Sometimes people flow nothing. 00:16:10.650 --> 00:16:12.840 align:middle line:84% Sometimes it's just silane flowing 00:16:12.840 --> 00:16:14.640 align:middle line:90% in the tool, relatively low. 00:16:14.640 --> 00:16:17.730 align:middle line:84% Sometimes people will use a carrier gas, 00:16:17.730 --> 00:16:22.320 align:middle line:84% such as nitrogen. Notice that hydrogen is not used, typically. 00:16:22.320 --> 00:16:23.880 align:middle line:84% And there was a question, I believe, 00:16:23.880 --> 00:16:25.800 align:middle line:90% about this in the last lecture. 00:16:25.800 --> 00:16:28.200 align:middle line:84% Hydrogen is not used because it reduces the growth rate. 00:16:28.200 --> 00:16:31.980 align:middle line:84% Again, this is the reaction from silane going to solid silicon. 00:16:31.980 --> 00:16:34.795 align:middle line:84% And it involves hydrogen. If you put hydrogen into the reactor, 00:16:34.795 --> 00:16:36.420 align:middle line:84% it tends to push the reaction backwards 00:16:36.420 --> 00:16:39.690 align:middle line:90% and slow down the growth rate. 00:16:39.690 --> 00:16:44.130 align:middle line:84% Typical pressure for polydep is about 200 millitorr, 00:16:44.130 --> 00:16:46.950 align:middle line:90% maybe up to a torr. 00:16:46.950 --> 00:16:48.670 align:middle line:90% Torr is a little bit high. 00:16:48.670 --> 00:16:51.720 align:middle line:84% The low pressure here in the hundreds of millitorr range 00:16:51.720 --> 00:16:55.210 align:middle line:84% improves the film uniformity, the film thickness uniformity. 00:16:55.210 --> 00:17:01.250 align:middle line:84% And again, the temperature range is pretty tight, 575 to 650. 00:17:01.250 --> 00:17:03.210 align:middle line:84% And an important number to keep in your head, 00:17:03.210 --> 00:17:06.380 align:middle line:84% an important temperature for polysilicon is 580. 00:17:06.380 --> 00:17:11.069 align:middle line:84% That's right about the amorphous to crystalline phase transition. 00:17:11.069 --> 00:17:14.329 align:middle line:84% So if you go below 580, the film is pretty much amorphous, 00:17:14.329 --> 00:17:15.319 align:middle line:90% primarily. 00:17:15.319 --> 00:17:18.079 align:middle line:84% Although, it will contain some small polycrystalline-- 00:17:18.079 --> 00:17:21.710 align:middle line:84% or crystallites, which sometimes people call seeds. 00:17:21.710 --> 00:17:23.220 align:middle line:90% So 580 is the magic number. 00:17:23.220 --> 00:17:26.670 align:middle line:84% You go down below that, you get amorphous material as deposited. 00:17:26.670 --> 00:17:30.740 align:middle line:84% You go above that, and you get polycrystalline material. 00:17:30.740 --> 00:17:33.850 align:middle line:90% 00:17:33.850 --> 00:17:35.950 align:middle line:84% So if we go on to slide 6, in fact, 00:17:35.950 --> 00:17:39.660 align:middle line:84% there are some pictures of polycrystalline silicon. 00:17:39.660 --> 00:17:42.450 align:middle line:84% Although, this is a figure in your text in chapter 9, 00:17:42.450 --> 00:17:45.810 align:middle line:84% figure 9-32, it's actually originally taken 00:17:45.810 --> 00:17:48.263 align:middle line:84% from Ted Kamin's book on polycrystalline silicon. 00:17:48.263 --> 00:17:49.680 align:middle line:84% So I've given up here on the upper 00:17:49.680 --> 00:17:54.660 align:middle line:84% left the original reference of that work. 00:17:54.660 --> 00:17:57.750 align:middle line:84% And these are cross-section TEM micrographs 00:17:57.750 --> 00:18:01.590 align:middle line:84% under four different conditions of polysilicon put down 00:18:01.590 --> 00:18:04.480 align:middle line:90% by low pressure CVD. 00:18:04.480 --> 00:18:07.840 align:middle line:84% And all these films were deposited at 625. 00:18:07.840 --> 00:18:10.750 align:middle line:84% So here's just an example of a cross-section TEM 00:18:10.750 --> 00:18:11.677 align:middle line:90% of undoped film. 00:18:11.677 --> 00:18:12.760 align:middle line:90% It's a little hard to see. 00:18:12.760 --> 00:18:14.093 align:middle line:90% That didn't reproduce that well. 00:18:14.093 --> 00:18:17.470 align:middle line:84% But you see the very columnar grain structure. 00:18:17.470 --> 00:18:20.650 align:middle line:84% These dark bands correspond to regions 00:18:20.650 --> 00:18:22.990 align:middle line:84% where the electron beam is diffracted 00:18:22.990 --> 00:18:24.880 align:middle line:84% by the different crystalline orientations 00:18:24.880 --> 00:18:27.950 align:middle line:90% and by grain boundaries. 00:18:27.950 --> 00:18:31.720 align:middle line:84% So in an undoped film deposited at 625 in cross-section looks 00:18:31.720 --> 00:18:33.730 align:middle line:84% something like this, a very columnar grain 00:18:33.730 --> 00:18:37.810 align:middle line:84% structure with relatively small grain size. 00:18:37.810 --> 00:18:43.480 align:middle line:84% As deposited, if it's heavily phosphorus doped, again, 00:18:43.480 --> 00:18:45.610 align:middle line:84% at 625, so still a very low temperature 00:18:45.610 --> 00:18:48.580 align:middle line:84% but very heavily phosphorus doped, you see lots-- 00:18:48.580 --> 00:18:51.520 align:middle line:84% the grains are more equiaxed, larger grain size. 00:18:51.520 --> 00:18:54.948 align:middle line:84% So phosphorus is one of those dopants 00:18:54.948 --> 00:18:56.740 align:middle line:84% that tends to cause secondary grain growth. 00:18:56.740 --> 00:18:58.817 align:middle line:84% It causes the grains to grow quite large. 00:18:58.817 --> 00:19:00.650 align:middle line:84% So you get a very different film morphology. 00:19:00.650 --> 00:19:06.000 align:middle line:84% If you deposit it in situ doped, say, with phosphine. 00:19:06.000 --> 00:19:08.490 align:middle line:84% The third one down is the undoped film. 00:19:08.490 --> 00:19:11.410 align:middle line:84% So if you take film A here and heat treat it, 00:19:11.410 --> 00:19:14.520 align:middle line:84% so you anneal it afterwards at 1,000 degrees, 00:19:14.520 --> 00:19:16.800 align:middle line:84% you can see the grains are still columnar. 00:19:16.800 --> 00:19:19.882 align:middle line:84% But they've grown, so the average grain size looks larger. 00:19:19.882 --> 00:19:21.840 align:middle line:84% It's a little bit hard to tell an average grain 00:19:21.840 --> 00:19:22.757 align:middle line:90% size in cross-section. 00:19:22.757 --> 00:19:23.970 align:middle line:90% But you get that impression. 00:19:23.970 --> 00:19:27.600 align:middle line:84% But look what also has happened to the surface. 00:19:27.600 --> 00:19:30.330 align:middle line:84% As the grains grow, there's a certain amount 00:19:30.330 --> 00:19:32.830 align:middle line:90% of induced surface roughness. 00:19:32.830 --> 00:19:35.280 align:middle line:84% So the surface is much rougher than it 00:19:35.280 --> 00:19:37.830 align:middle line:90% was in the as deposited case. 00:19:37.830 --> 00:19:40.117 align:middle line:84% It's typical, of course, to anneal a polysilicon 00:19:40.117 --> 00:19:41.700 align:middle line:84% film during the source drain annealing 00:19:41.700 --> 00:19:43.033 align:middle line:90% or whatever to activate dopants. 00:19:43.033 --> 00:19:48.270 align:middle line:84% So this is more what it might look like in a gate material. 00:19:48.270 --> 00:19:51.810 align:middle line:84% And the last one is one that was in situ doped. 00:19:51.810 --> 00:19:55.200 align:middle line:84% So it's film B, but it was also annealed up 00:19:55.200 --> 00:19:58.150 align:middle line:84% at 1,000 degrees C. So look at the big difference. 00:19:58.150 --> 00:20:00.960 align:middle line:84% Both of these films deposited 625, 00:20:00.960 --> 00:20:03.750 align:middle line:84% and both annealed at 1,000 C. The real big difference is 00:20:03.750 --> 00:20:06.750 align:middle line:84% this film is phosphorus doped, very large grains. 00:20:06.750 --> 00:20:09.630 align:middle line:84% In fact, if you take a thin film course 00:20:09.630 --> 00:20:13.290 align:middle line:84% in material science, Carl Thompson's courses or whatever, 00:20:13.290 --> 00:20:17.080 align:middle line:84% you'll study the thermodynamics of this much more carefully. 00:20:17.080 --> 00:20:19.230 align:middle line:84% But basically, it turns out that the grains 00:20:19.230 --> 00:20:23.640 align:middle line:84% tend to grow laterally to about the same size 00:20:23.640 --> 00:20:25.080 align:middle line:90% as the thickness of the film. 00:20:25.080 --> 00:20:28.080 align:middle line:84% So the thickness of this film is about half micron, 00:20:28.080 --> 00:20:30.090 align:middle line:84% and the average grain size ends up being, 00:20:30.090 --> 00:20:33.600 align:middle line:84% in these large grain films, about half micron. 00:20:33.600 --> 00:20:36.630 align:middle line:84% And so it's phosphorus, the presence of the dopant 00:20:36.630 --> 00:20:39.090 align:middle line:84% really accelerates the grain growth. 00:20:39.090 --> 00:20:43.580 align:middle line:84% And you get a very different morphology film. 00:20:43.580 --> 00:20:46.440 align:middle line:90% OK, so let's go on to slide 7. 00:20:46.440 --> 00:20:47.690 align:middle line:90% Why do we care about all that? 00:20:47.690 --> 00:20:49.010 align:middle line:90% Why did we go through all that? 00:20:49.010 --> 00:20:52.140 align:middle line:84% Well, the grain structure, as it turns out, 00:20:52.140 --> 00:20:55.490 align:middle line:84% is critical in determining the properties of diffusion 00:20:55.490 --> 00:20:58.160 align:middle line:84% of dopants through polycrystalline silicon 00:20:58.160 --> 00:21:01.470 align:middle line:84% and also in determining the final resistivity of the film. 00:21:01.470 --> 00:21:03.470 align:middle line:84% Now, I don't know if anyone signed up for it, 00:21:03.470 --> 00:21:06.200 align:middle line:84% but on the clipboard, one of the topics 00:21:06.200 --> 00:21:10.130 align:middle line:84% that I had open to do a final report on 00:21:10.130 --> 00:21:13.175 align:middle line:84% is diffusion and/or dopant activation 00:21:13.175 --> 00:21:14.300 align:middle line:90% in polycrystalline silicon. 00:21:14.300 --> 00:21:16.580 align:middle line:84% I'm not going to talk about it in great detail in the course. 00:21:16.580 --> 00:21:18.470 align:middle line:84% Diffusion through polycrystalline silicon 00:21:18.470 --> 00:21:20.180 align:middle line:84% is quite different than what we've 00:21:20.180 --> 00:21:22.490 align:middle line:84% talked about diffusion in single crystal silicon 00:21:22.490 --> 00:21:25.100 align:middle line:84% because in polysilicon the dopants can diffuse along 00:21:25.100 --> 00:21:26.480 align:middle line:90% the grain boundaries. 00:21:26.480 --> 00:21:27.320 align:middle line:90% Look at this. 00:21:27.320 --> 00:21:31.220 align:middle line:84% This is an example up here labeled 00:21:31.220 --> 00:21:33.140 align:middle line:90% B called the columnar grains. 00:21:33.140 --> 00:21:35.540 align:middle line:84% The dopants, a lot of them, like arsenic, 00:21:35.540 --> 00:21:38.090 align:middle line:84% diffuse very rapidly down grain boundaries. 00:21:38.090 --> 00:21:41.370 align:middle line:84% They can also diffuse more slowly in the grains. 00:21:41.370 --> 00:21:44.020 align:middle line:84% So you can imagine that the diffusion of dopants 00:21:44.020 --> 00:21:47.405 align:middle line:84% in this cartoon labeled B here on the upper left 00:21:47.405 --> 00:21:49.530 align:middle line:84% is going to be quite different from the one labeled 00:21:49.530 --> 00:21:53.220 align:middle line:84% A. A you have more random grain structure. 00:21:53.220 --> 00:21:55.295 align:middle line:90% Looks more like a brick wall. 00:21:55.295 --> 00:21:56.670 align:middle line:84% So the diffusivity is going to be 00:21:56.670 --> 00:22:00.960 align:middle line:84% very different in this structure to get the dopants from the top 00:22:00.960 --> 00:22:03.565 align:middle line:90% down to the bottom. 00:22:03.565 --> 00:22:04.940 align:middle line:84% It's going to be quite different. 00:22:04.940 --> 00:22:06.180 align:middle line:90% Why do we care about that? 00:22:06.180 --> 00:22:10.793 align:middle line:84% Well, polysilicon is the gate material in CMOS, 00:22:10.793 --> 00:22:11.960 align:middle line:90% until we get to metal gates. 00:22:11.960 --> 00:22:13.918 align:middle line:84% Of course, metal gates are coming very quickly. 00:22:13.918 --> 00:22:16.370 align:middle line:84% But for the last 30 years or 25 years, 00:22:16.370 --> 00:22:18.710 align:middle line:84% polysilicon has been the gate material. 00:22:18.710 --> 00:22:22.370 align:middle line:84% Typically, how it's doped is it's put down in undoped state, 00:22:22.370 --> 00:22:24.680 align:middle line:84% and then it's ion plated in the near-surface region. 00:22:24.680 --> 00:22:26.990 align:middle line:84% And then you do a high-temperature anneal, 00:22:26.990 --> 00:22:29.540 align:middle line:84% typically during the source drain anneal or maybe 00:22:29.540 --> 00:22:31.130 align:middle line:84% a separate one, in which you hope 00:22:31.130 --> 00:22:32.870 align:middle line:84% that you-- try to get these dopants down 00:22:32.870 --> 00:22:34.250 align:middle line:84% from the surface all the way down 00:22:34.250 --> 00:22:39.530 align:middle line:84% to the interface with the oxide to form the gate. 00:22:39.530 --> 00:22:41.480 align:middle line:84% If don't get all the dopants down there, 00:22:41.480 --> 00:22:42.750 align:middle line:90% you're in deep trouble. 00:22:42.750 --> 00:22:45.140 align:middle line:84% You're in deep trouble because polysilicon then no longer 00:22:45.140 --> 00:22:47.360 align:middle line:84% acts metallic if it's not heavily doped. 00:22:47.360 --> 00:22:50.468 align:middle line:84% In fact, it will deplete under gate bias, 00:22:50.468 --> 00:22:52.760 align:middle line:84% and you'll end up with something called gate depletion. 00:22:52.760 --> 00:22:57.000 align:middle line:84% And it will not act as a good metallic-like electrode. 00:22:57.000 --> 00:22:59.150 align:middle line:84% So how you put the polysilicon down 00:22:59.150 --> 00:23:01.400 align:middle line:84% is very critical because it determines 00:23:01.400 --> 00:23:05.200 align:middle line:84% how the dope is diffused through it, how well they activate. 00:23:05.200 --> 00:23:09.360 align:middle line:84% So the temperature that you use is very important. 00:23:09.360 --> 00:23:12.340 align:middle line:84% And here's just an example of-- now, this is a planar view. 00:23:12.340 --> 00:23:15.417 align:middle line:84% So in this case, you're looking down on the top of the film. 00:23:15.417 --> 00:23:17.250 align:middle line:84% Again, it's transmission electron micrograph 00:23:17.250 --> 00:23:19.050 align:middle line:90% but plan view. 00:23:19.050 --> 00:23:22.110 align:middle line:84% This is about 400 nanometers thick. 00:23:22.110 --> 00:23:24.540 align:middle line:84% And these are all phosphorus-doped films. 00:23:24.540 --> 00:23:27.390 align:middle line:84% Again, I'm showing the enhancement in the growth rate 00:23:27.390 --> 00:23:28.980 align:middle line:90% of the grains with phosphorus. 00:23:28.980 --> 00:23:30.690 align:middle line:90% These are all annealed at 1,000. 00:23:30.690 --> 00:23:34.410 align:middle line:84% This one has an average doping of about 1 E20. 00:23:34.410 --> 00:23:37.710 align:middle line:84% Micrograph B, the sample has about 2 and 1/2 E20. 00:23:37.710 --> 00:23:40.830 align:middle line:84% And micrograph C is about 7 and 1/2 times 10 to the 20. 00:23:40.830 --> 00:23:42.240 align:middle line:90% You can see how large-- 00:23:42.240 --> 00:23:45.660 align:middle line:84% again, you have large grains on the order of half micron 00:23:45.660 --> 00:23:48.910 align:middle line:90% to a micron here in grain size. 00:23:48.910 --> 00:23:54.930 align:middle line:84% So phosphorus doping can really accelerate the grain growth. 00:23:54.930 --> 00:23:57.000 align:middle line:90% How about on slide 8? 00:23:57.000 --> 00:23:58.230 align:middle line:90% How about film resistivity? 00:23:58.230 --> 00:24:00.420 align:middle line:84% And besides the fact that-- so I mentioned to you 00:24:00.420 --> 00:24:01.878 align:middle line:84% we need to get the dopants that are 00:24:01.878 --> 00:24:05.310 align:middle line:84% implanted at the top of the poly down to the gate oxide 00:24:05.310 --> 00:24:06.060 align:middle line:90% interface. 00:24:06.060 --> 00:24:07.477 align:middle line:84% Otherwise, we're not going to have 00:24:07.477 --> 00:24:09.102 align:middle line:90% a good polycrystal electrode. 00:24:09.102 --> 00:24:10.560 align:middle line:84% We're going to have gate depletion, 00:24:10.560 --> 00:24:12.578 align:middle line:90% and your MOSFET won't work. 00:24:12.578 --> 00:24:13.870 align:middle line:90% But there's other requirements. 00:24:13.870 --> 00:24:16.380 align:middle line:84% We also care about the sheet resistance or the resistivity 00:24:16.380 --> 00:24:21.510 align:middle line:84% of the poly itself, particularly for some high-frequency 00:24:21.510 --> 00:24:22.440 align:middle line:90% applications. 00:24:22.440 --> 00:24:28.110 align:middle line:84% So this is a plot also took from Ted Kamin's book of resistive 00:24:28.110 --> 00:24:33.760 align:middle line:84% now in ohm's centimeter as a function of dopant concentration 00:24:33.760 --> 00:24:35.290 align:middle line:90% in polysilicon. 00:24:35.290 --> 00:24:40.390 align:middle line:84% These films were all ion implanted with these dopants 00:24:40.390 --> 00:24:43.330 align:middle line:84% and then annealed at 1,000 degrees C. 00:24:43.330 --> 00:24:44.890 align:middle line:84% So it gives you a rough idea of what 00:24:44.890 --> 00:24:48.340 align:middle line:84% happens if you were to do a reasonable job of ion planting 00:24:48.340 --> 00:24:49.000 align:middle line:90% and annealing. 00:24:49.000 --> 00:24:51.880 align:middle line:84% And you could see the boron and the arsenic and phosphorus, 00:24:51.880 --> 00:24:53.960 align:middle line:90% they all kind of come down here. 00:24:53.960 --> 00:24:56.723 align:middle line:84% And so as you increase the doping concentration, 00:24:56.723 --> 00:24:57.640 align:middle line:90% the resistivity drops. 00:24:57.640 --> 00:25:00.130 align:middle line:84% But all of a sudden, it sort of plateaus out. 00:25:00.130 --> 00:25:05.380 align:middle line:84% So above about 1 E20 or 2 E20, arsenic and boron-doped poly, 00:25:05.380 --> 00:25:10.510 align:middle line:84% they sort of flatten out here at a certain resistivity 00:25:10.510 --> 00:25:16.185 align:middle line:84% and a couple of hundred micron centimeter, 00:25:16.185 --> 00:25:17.060 align:middle line:90% or a couple thousand. 00:25:17.060 --> 00:25:17.560 align:middle line:90% I'm sorry. 00:25:17.560 --> 00:25:21.020 align:middle line:84% So this is around 2,000 micron centimeter, which is still 00:25:21.020 --> 00:25:22.250 align:middle line:90% pretty high resistivity. 00:25:22.250 --> 00:25:25.250 align:middle line:84% If you use phosphorous, you're a little better shape. 00:25:25.250 --> 00:25:28.010 align:middle line:84% You can continue getting the resistive down as you-- 00:25:28.010 --> 00:25:30.860 align:middle line:84% and it plateaus somewhere above 5 10 to the 20. 00:25:30.860 --> 00:25:34.950 align:middle line:84% It plateaus around 400 microohm centimeter. 00:25:34.950 --> 00:25:37.490 align:middle line:84% So if you need to get the lowest resistivity, 00:25:37.490 --> 00:25:40.220 align:middle line:84% phosphorus is a good dopant to use. 00:25:40.220 --> 00:25:42.210 align:middle line:90% And it activates pretty well. 00:25:42.210 --> 00:25:44.510 align:middle line:84% So the conduction mechanism, as you can imagine, 00:25:44.510 --> 00:25:46.700 align:middle line:84% if you go through Ted Kamen's book, 00:25:46.700 --> 00:25:50.150 align:middle line:84% the conduction of current in polycrystalline silicon 00:25:50.150 --> 00:25:52.550 align:middle line:84% is going to be very different from that in single crystal 00:25:52.550 --> 00:25:55.280 align:middle line:84% because one thing you're going through, if you go back, just 00:25:55.280 --> 00:25:57.470 align:middle line:84% go back one slide, you can imagine I've 00:25:57.470 --> 00:26:01.580 align:middle line:84% got a current passing through either this film or this film. 00:26:01.580 --> 00:26:03.200 align:middle line:84% There are all these grain boundaries 00:26:03.200 --> 00:26:06.350 align:middle line:84% that the carriers end up going through. 00:26:06.350 --> 00:26:09.380 align:middle line:84% The grain boundaries are regions of very imperfect 00:26:09.380 --> 00:26:12.380 align:middle line:90% crystalline quality. 00:26:12.380 --> 00:26:14.690 align:middle line:84% You can get scattering off of grain boundaries. 00:26:14.690 --> 00:26:18.270 align:middle line:84% The scattering rate can depend on the amount of dopant 00:26:18.270 --> 00:26:19.870 align:middle line:84% that segregated to the grain boundary. 00:26:19.870 --> 00:26:22.463 align:middle line:84% So there's a whole lot of information 00:26:22.463 --> 00:26:23.880 align:middle line:84% that you really need to understand 00:26:23.880 --> 00:26:26.160 align:middle line:84% if you want to understand the resistivity 00:26:26.160 --> 00:26:28.895 align:middle line:84% and diffusion of dopants in polycrystalline silicon. 00:26:28.895 --> 00:26:30.270 align:middle line:84% We don't have time in this class, 00:26:30.270 --> 00:26:32.010 align:middle line:84% but I just wanted to make you aware. 00:26:32.010 --> 00:26:34.290 align:middle line:84% And Ted Kamin's book is a good reference. 00:26:34.290 --> 00:26:39.860 align:middle line:84% He's got a couple of chapters that topic if you're interested. 00:26:39.860 --> 00:26:42.350 align:middle line:84% OK, so that's polycrystalline silicon, 00:26:42.350 --> 00:26:46.310 align:middle line:84% and again, it's commonly deposited pretty much every fab 00:26:46.310 --> 00:26:50.270 align:middle line:90% just by simple low-pressure CVD. 00:26:50.270 --> 00:26:54.590 align:middle line:84% On slide 9 now, I want to go on to plasma-enhanced CVD 00:26:54.590 --> 00:26:57.380 align:middle line:84% and just show you some examples of those reactors 00:26:57.380 --> 00:26:58.190 align:middle line:90% and how it works. 00:26:58.190 --> 00:27:00.620 align:middle line:84% And again, the acronym here is PECVD. 00:27:00.620 --> 00:27:04.245 align:middle line:84% The idea, why do I want to use a plasma? 00:27:04.245 --> 00:27:05.620 align:middle line:84% Why would I want to use a plasma? 00:27:05.620 --> 00:27:10.740 align:middle line:84% Well, we want to enter-- we want to give to the system 00:27:10.740 --> 00:27:13.230 align:middle line:84% some kind of nonthermal energy, which 00:27:13.230 --> 00:27:16.390 align:middle line:84% is the plasma energy in order to enhance the process. 00:27:16.390 --> 00:27:19.780 align:middle line:84% So it will go faster at lower temperatures. 00:27:19.780 --> 00:27:23.080 align:middle line:84% So we get enhanced DEP rate at low temperatures. 00:27:23.080 --> 00:27:26.790 align:middle line:84% And this is a typical reactor, if you want a canonical plasma-- 00:27:26.790 --> 00:27:29.700 align:middle line:90% a simple plasma reactor. 00:27:29.700 --> 00:27:31.510 align:middle line:84% Actually, it's got an RF input here. 00:27:31.510 --> 00:27:33.335 align:middle line:84% So there are two electrodes, one at the top 00:27:33.335 --> 00:27:34.710 align:middle line:90% and one at the bottom. 00:27:34.710 --> 00:27:38.010 align:middle line:84% The wafers sit on one of the electrodes. 00:27:38.010 --> 00:27:40.410 align:middle line:84% There is a plasma here in between, 00:27:40.410 --> 00:27:45.750 align:middle line:84% and this plasma consists of electrons, ionized molecules. 00:27:45.750 --> 00:27:49.110 align:middle line:84% They'll have neutral species, neutral molecules, 00:27:49.110 --> 00:27:54.450 align:middle line:84% neutral and ionized fragments of broken up molecules, 00:27:54.450 --> 00:27:56.350 align:middle line:84% excited molecules and free radicals. 00:27:56.350 --> 00:27:58.890 align:middle line:84% So there's a lot of things present in the plasma that 00:27:58.890 --> 00:28:01.430 align:middle line:84% are not there if I'm just flowing silane. 00:28:01.430 --> 00:28:03.420 align:middle line:84% If I'm flowing saline and the plasma is off, 00:28:03.420 --> 00:28:04.720 align:middle line:90% what do I have in the reactor? 00:28:04.720 --> 00:28:06.610 align:middle line:90% I have silane, OK? 00:28:06.610 --> 00:28:09.140 align:middle line:84% But if I'm flowing silane and the plasma is on, 00:28:09.140 --> 00:28:11.380 align:middle line:84% I can break up this molecule into 00:28:11.380 --> 00:28:14.530 align:middle line:84% all different subcomponents, some of which 00:28:14.530 --> 00:28:16.630 align:middle line:90% are highly reactive. 00:28:16.630 --> 00:28:18.730 align:middle line:84% And therefore, the deposition rate at the surface 00:28:18.730 --> 00:28:23.270 align:middle line:84% will be dramatically impacted by the presence of this plasma. 00:28:23.270 --> 00:28:26.950 align:middle line:84% So here's sort of a canonical plasma system, 00:28:26.950 --> 00:28:29.140 align:middle line:84% and there is a heater here that would typically-- 00:28:29.140 --> 00:28:31.540 align:middle line:84% because you need to do this at a finite temperature 00:28:31.540 --> 00:28:34.370 align:middle line:90% that would heat the wafers. 00:28:34.370 --> 00:28:36.215 align:middle line:84% OK, so free radicals I mentioned, 00:28:36.215 --> 00:28:40.390 align:middle line:84% if you go on to slide 10, they're really critical. 00:28:40.390 --> 00:28:44.950 align:middle line:84% Free radicals are electrically-neutral species 00:28:44.950 --> 00:28:47.680 align:middle line:90% that have incomplete bonding. 00:28:47.680 --> 00:28:50.957 align:middle line:84% They're not naturally found in nature in this free state. 00:28:50.957 --> 00:28:52.790 align:middle line:84% They're created because they're in a plasma, 00:28:52.790 --> 00:28:56.710 align:middle line:84% and you have energy stripping off atoms from these molecules. 00:28:56.710 --> 00:29:00.110 align:middle line:84% And therefore, they are extremely reactive. 00:29:00.110 --> 00:29:05.590 align:middle line:84% And so the deposition rate can be dramatically enhanced. 00:29:05.590 --> 00:29:09.490 align:middle line:84% So for example, here's an example of a free radical, SiO. 00:29:09.490 --> 00:29:11.800 align:middle line:84% Typically, to be electrically neutral 00:29:11.800 --> 00:29:16.960 align:middle line:84% this would be SiO2 or SiH3 instead of SiH4. 00:29:16.960 --> 00:29:21.130 align:middle line:84% So you have an extra unsatisfied bond here on the silicon. 00:29:21.130 --> 00:29:23.410 align:middle line:84% So it's going to be very reactive. 00:29:23.410 --> 00:29:25.870 align:middle line:84% Fluorine, in and of itself, typically, fluorine 00:29:25.870 --> 00:29:26.830 align:middle line:90% is present as F2. 00:29:26.830 --> 00:29:28.660 align:middle line:84% But you can break it up with a plasma 00:29:28.660 --> 00:29:32.380 align:middle line:84% to form a lot of free radical just free flowing, which 00:29:32.380 --> 00:29:34.940 align:middle line:90% is very reactive. 00:29:34.940 --> 00:29:37.840 align:middle line:84% So the net result from this breaking up-- 00:29:37.840 --> 00:29:41.620 align:middle line:84% this fragmentation, the free radicals and the ion bombardment 00:29:41.620 --> 00:29:43.360 align:middle line:84% is that the surface processes occur 00:29:43.360 --> 00:29:45.110 align:middle line:90% at much lower temperatures. 00:29:45.110 --> 00:29:49.570 align:middle line:84% So for dielectrics, you can do PECVD typically 00:29:49.570 --> 00:29:53.633 align:middle line:84% between 200 and 350 and get very high deposition rates. 00:29:53.633 --> 00:29:55.300 align:middle line:84% This is particularly important if you're 00:29:55.300 --> 00:29:56.930 align:middle line:90% doing a backend process. 00:29:56.930 --> 00:29:59.320 align:middle line:84% So backend meaning you have some metal on the wafer. 00:29:59.320 --> 00:30:01.900 align:middle line:84% You can't take it above 350, for example. 00:30:01.900 --> 00:30:05.615 align:middle line:84% Well, if you want to deposit LTO, you have to go to 400 00:30:05.615 --> 00:30:07.240 align:middle line:84% to get any kind of reasonable DEP rate. 00:30:07.240 --> 00:30:10.630 align:middle line:84% It's going to be difficult. So people can use, in that case, 00:30:10.630 --> 00:30:16.450 align:middle line:84% can use plasma-enhanced CVD to get the DEP rate up even if you 00:30:16.450 --> 00:30:17.900 align:middle line:90% can't use high temperatures. 00:30:17.900 --> 00:30:19.510 align:middle line:84% So that's a very common application 00:30:19.510 --> 00:30:23.060 align:middle line:90% of PECVDs in the back end. 00:30:23.060 --> 00:30:26.390 align:middle line:84% If we go to slide 11, actually, this is a picture of-- 00:30:26.390 --> 00:30:28.910 align:middle line:84% an example of a PECVD deposition tool. 00:30:28.910 --> 00:30:31.820 align:middle line:84% There's one here in the microsystems technology 00:30:31.820 --> 00:30:35.510 align:middle line:90% lab in building 39 here at MIT. 00:30:35.510 --> 00:30:37.250 align:middle line:84% It's made by a company called Novellus, 00:30:37.250 --> 00:30:38.990 align:middle line:90% and it's called the Concept One. 00:30:38.990 --> 00:30:43.898 align:middle line:84% And on the bottom of the slide, I put down the URL 00:30:43.898 --> 00:30:45.440 align:middle line:84% where you can go get more information 00:30:45.440 --> 00:30:49.800 align:middle line:84% about that particular piece of equipment and how it works. 00:30:49.800 --> 00:30:52.190 align:middle line:84% This is a photograph of it on the left, 00:30:52.190 --> 00:30:55.245 align:middle line:84% and it's set up right now processing 6-inch wafers. 00:30:55.245 --> 00:30:56.870 align:middle line:84% Look at the interesting thing about it. 00:30:56.870 --> 00:30:59.770 align:middle line:90% Look at the DEP rate for oxide. 00:30:59.770 --> 00:31:04.000 align:middle line:84% At low temperature, 350, 300, it can put oxide down 00:31:04.000 --> 00:31:05.380 align:middle line:90% at a micron per minute. 00:31:05.380 --> 00:31:07.772 align:middle line:84% That's an absolutely phenomenal growth 00:31:07.772 --> 00:31:08.980 align:middle line:90% rate when you think about it. 00:31:08.980 --> 00:31:10.771 align:middle line:90% Yeah, question. 00:31:10.771 --> 00:31:13.600 align:middle line:90% AUDIENCE: [INAUDIBLE]. 00:31:13.600 --> 00:31:16.947 align:middle line:84% JUDY HOYT: Well, if you need only a couple thousand 00:31:16.947 --> 00:31:18.530 align:middle line:84% angstroms, you don't want to use that. 00:31:18.530 --> 00:31:20.500 align:middle line:84% But you can slow down the DEP rate as well. 00:31:20.500 --> 00:31:24.700 align:middle line:84% But the kind of people who will be using it at this DEP rate 00:31:24.700 --> 00:31:27.050 align:middle line:84% might be somebody who wants to put down 10 microns. 00:31:27.050 --> 00:31:30.550 align:middle line:84% Maybe they're making a MEMS device, so they need 10 microns. 00:31:30.550 --> 00:31:32.227 align:middle line:84% To do that by any thermal technique, 00:31:32.227 --> 00:31:34.060 align:middle line:84% you'd be there running the reactor for days. 00:31:34.060 --> 00:31:35.668 align:middle line:84% And you'd be going through cylinder 00:31:35.668 --> 00:31:36.710 align:middle line:90% after cylinder of silane. 00:31:36.710 --> 00:31:39.310 align:middle line:84% And it would be totally ridiculous. 00:31:39.310 --> 00:31:42.830 align:middle line:84% So for people who need the thick films, this is the way to go. 00:31:42.830 --> 00:31:44.350 align:middle line:84% You can back off on the DEP rate. 00:31:44.350 --> 00:31:45.620 align:middle line:90% This is the highest DEP rate. 00:31:45.620 --> 00:31:49.600 align:middle line:84% You can go slower than that, and people do do that. 00:31:49.600 --> 00:31:50.920 align:middle line:90% It does more than oxide. 00:31:50.920 --> 00:31:56.290 align:middle line:84% It can deposit nitride, oxynitride, and other films. 00:31:56.290 --> 00:31:59.320 align:middle line:90% It's an RF plasma. 00:31:59.320 --> 00:32:00.680 align:middle line:90% It has an interesting thing. 00:32:00.680 --> 00:32:03.550 align:middle line:84% It has a multiple-station processing sequence. 00:32:03.550 --> 00:32:07.810 align:middle line:84% It cycles each wafer through different deposition chambers 00:32:07.810 --> 00:32:10.900 align:middle line:84% or stations to get an averaging effect so 00:32:10.900 --> 00:32:12.640 align:middle line:90% that you get better uniformity. 00:32:12.640 --> 00:32:14.990 align:middle line:84% In any given tool, you often find, 00:32:14.990 --> 00:32:18.140 align:middle line:84% if it has multiple stations, say, it has five slots, 00:32:18.140 --> 00:32:21.050 align:middle line:84% slot one and slot five never have exactly the same DEP rate, 00:32:21.050 --> 00:32:22.282 align:middle line:90% and it's always a problem. 00:32:22.282 --> 00:32:23.240 align:middle line:90% So you put five wafers. 00:32:23.240 --> 00:32:26.160 align:middle line:84% In wafer 1 and 5 come out differently. 00:32:26.160 --> 00:32:29.630 align:middle line:84% This is a very brute force obvious way to do it. 00:32:29.630 --> 00:32:32.040 align:middle line:84% But what they do is instead you take one wafer, 00:32:32.040 --> 00:32:34.830 align:middle line:84% and you cycle it through all five slots. 00:32:34.830 --> 00:32:37.510 align:middle line:84% And so one wafer sees a little bit of all five 00:32:37.510 --> 00:32:39.410 align:middle line:84% and kind of averages out the nonuniformity. 00:32:39.410 --> 00:32:43.830 align:middle line:84% Now, with a robotic control, you can do that kind of thing. 00:32:43.830 --> 00:32:47.030 align:middle line:84% So anyway, this is the Concept One, typically used 00:32:47.030 --> 00:32:48.260 align:middle line:90% more for backend processing. 00:32:48.260 --> 00:32:52.690 align:middle line:84% You don't see too many people using this in the frontend. 00:32:52.690 --> 00:32:53.950 align:middle line:90% And why is that? 00:32:53.950 --> 00:32:55.210 align:middle line:90% Why is that? 00:32:55.210 --> 00:32:59.380 align:middle line:84% Well, backend has to use it because typically you 00:32:59.380 --> 00:33:00.590 align:middle line:90% need very low temperatures. 00:33:00.590 --> 00:33:04.030 align:middle line:84% But if you go to slide 12, the characteristics 00:33:04.030 --> 00:33:06.820 align:middle line:84% of plasma-enhanced CVD-deposited films, 00:33:06.820 --> 00:33:08.980 align:middle line:84% like oxide or silicon nitride, they 00:33:08.980 --> 00:33:10.930 align:middle line:84% can be quite different from those 00:33:10.930 --> 00:33:14.050 align:middle line:84% that are deposited at higher temperatures by LPCVD. 00:33:14.050 --> 00:33:16.240 align:middle line:84% For example, very different stress, 00:33:16.240 --> 00:33:20.560 align:middle line:84% very different amounts of stress in PECVD and LPCVD films. 00:33:20.560 --> 00:33:24.030 align:middle line:84% Hydrogen content, in particular-- make a note-- 00:33:24.030 --> 00:33:29.110 align:middle line:84% PECVD films tend to be loaded with hydrogen, a lot. 00:33:29.110 --> 00:33:32.207 align:middle line:84% Could be like 10% hydrogen. They're 00:33:32.207 --> 00:33:33.790 align:middle line:84% just stuffed with hydrogen. And that's 00:33:33.790 --> 00:33:38.250 align:middle line:84% because a lot of these precursors, if you go back, 00:33:38.250 --> 00:33:42.420 align:middle line:84% look at some of the precursors you might be using, silane. 00:33:42.420 --> 00:33:45.510 align:middle line:84% Well, a lot of those hydrogens get ripped off during the plasma 00:33:45.510 --> 00:33:47.657 align:middle line:84% process, and you're growing so fast 00:33:47.657 --> 00:33:48.990 align:middle line:90% they can get buried in the film. 00:33:48.990 --> 00:33:53.310 align:middle line:84% So hydrogen incorporation, just by stuffing it in the film 00:33:53.310 --> 00:33:56.250 align:middle line:84% is kind of common in any low-temperature plasma process. 00:33:56.250 --> 00:33:57.750 align:middle line:84% Also, the temperature is low enough, 00:33:57.750 --> 00:33:59.710 align:middle line:84% hydrogen gets in the film and it just sticks. 00:33:59.710 --> 00:34:00.780 align:middle line:90% It doesn't evolve. 00:34:00.780 --> 00:34:02.675 align:middle line:84% At higher temperatures, like LPCVD, 00:34:02.675 --> 00:34:04.800 align:middle line:84% the sticking coefficient of hydrogen on the surface 00:34:04.800 --> 00:34:05.500 align:middle line:90% is very low. 00:34:05.500 --> 00:34:07.880 align:middle line:84% So you don't get so much hydrogen. Hydrogen would 00:34:07.880 --> 00:34:09.989 align:middle line:84% be less than a percent, maybe half a percent 00:34:09.989 --> 00:34:14.370 align:middle line:84% or less in an LPCVD film, oxide film or nitride film. 00:34:14.370 --> 00:34:18.030 align:middle line:84% Whereas, in a plasma-enhanced film it could be 10%, 00:34:18.030 --> 00:34:22.340 align:middle line:84% have a dramatic effect on the properties. 00:34:22.340 --> 00:34:24.260 align:middle line:84% And the stoichiometry of the film, 00:34:24.260 --> 00:34:26.179 align:middle line:90% it will not necessarily be-- 00:34:26.179 --> 00:34:28.100 align:middle line:84% LTO is not necessarily exactly SiO2. 00:34:28.100 --> 00:34:29.600 align:middle line:90% It could be SiOx. 00:34:29.600 --> 00:34:32.103 align:middle line:84% x is slightly different from 2, again, 00:34:32.103 --> 00:34:33.770 align:middle line:84% because you're not growing it thermally. 00:34:33.770 --> 00:34:36.820 align:middle line:90% You're depositing it. 00:34:36.820 --> 00:34:40.239 align:middle line:84% So these differences between PECVD and LPCVD 00:34:40.239 --> 00:34:42.550 align:middle line:84% can affect properties like the film adhesion, 00:34:42.550 --> 00:34:43.960 align:middle line:90% the presence of pinholes. 00:34:43.960 --> 00:34:47.320 align:middle line:84% Plasma-enhanced films tend to have a little more tendency 00:34:47.320 --> 00:34:50.050 align:middle line:84% to have pinholes because, again, as was mentioned, 00:34:50.050 --> 00:34:52.449 align:middle line:90% they're going down really fast. 00:34:52.449 --> 00:34:55.030 align:middle line:84% Surface roughness can be different. 00:34:55.030 --> 00:34:58.125 align:middle line:84% A very important thing is the dielectric constant. 00:34:58.125 --> 00:35:00.250 align:middle line:84% Again, because the presence of different impurities 00:35:00.250 --> 00:35:02.560 align:middle line:84% will be quite different for a PECVD. 00:35:02.560 --> 00:35:04.930 align:middle line:84% So the optical property-- the index of refraction 00:35:04.930 --> 00:35:07.600 align:middle line:84% can be dramatically different because 00:35:07.600 --> 00:35:12.700 align:middle line:84% of stoichiometric differences or because of hydrogen. Hydrogen 00:35:12.700 --> 00:35:16.600 align:middle line:84% can induce vibrational modes, which can absorb light. 00:35:16.600 --> 00:35:19.810 align:middle line:84% So dielectrics put down by PECVD can 00:35:19.810 --> 00:35:23.050 align:middle line:84% have much different light absorption characteristics. 00:35:23.050 --> 00:35:26.140 align:middle line:84% And finally, the important thing is etch rate 00:35:26.140 --> 00:35:28.660 align:middle line:84% the etch rate of a PECVD-deposited film 00:35:28.660 --> 00:35:31.485 align:middle line:84% will be quite different from that of an LPCVD film, which 00:35:31.485 --> 00:35:32.860 align:middle line:84% will be quite different from that 00:35:32.860 --> 00:35:34.110 align:middle line:90% of a thermally-deposited film. 00:35:34.110 --> 00:35:35.810 align:middle line:84% So depending on how you put it down, 00:35:35.810 --> 00:35:39.590 align:middle line:84% you need to know what the etch rate of oxide is in Hf 00:35:39.590 --> 00:35:43.600 align:middle line:84% or whatever you need to test for that. 00:35:43.600 --> 00:35:46.870 align:middle line:84% OK, so that's sort of what I wanted 00:35:46.870 --> 00:35:50.260 align:middle line:84% to say at this point about chemical vapor deposition. 00:35:50.260 --> 00:35:53.950 align:middle line:84% If we go on to slide 13, I want to move on 00:35:53.950 --> 00:35:58.510 align:middle line:84% to another topic, which is called physical vapor deposition 00:35:58.510 --> 00:36:00.840 align:middle line:90% that's discussed in your text. 00:36:00.840 --> 00:36:02.990 align:middle line:84% As the name implies, it's pretty obvious, 00:36:02.990 --> 00:36:05.310 align:middle line:90% PVD uses physical processes. 00:36:05.310 --> 00:36:12.090 align:middle line:84% It's not just chemistry to deposit films in the gas phase 00:36:12.090 --> 00:36:14.070 align:middle line:90% and to get them onto the wafer. 00:36:14.070 --> 00:36:17.640 align:middle line:84% The most common and the most classic physical process, 00:36:17.640 --> 00:36:19.830 align:middle line:84% and maybe some of you are familiar with this, 00:36:19.830 --> 00:36:21.790 align:middle line:90% is thermal evaporation. 00:36:21.790 --> 00:36:23.778 align:middle line:84% This is pretty much totally physical process. 00:36:23.778 --> 00:36:25.320 align:middle line:84% There's not a whole lot of chemistry. 00:36:25.320 --> 00:36:29.490 align:middle line:84% You're not flowing in gases or doing any chemistry. 00:36:29.490 --> 00:36:34.950 align:middle line:84% Here's an example of a bell jar, a large vacuum chamber. 00:36:34.950 --> 00:36:36.570 align:middle line:90% You heat the source material. 00:36:36.570 --> 00:36:40.260 align:middle line:84% There's a source material here on a heater that is then 00:36:40.260 --> 00:36:42.090 align:middle line:90% made molten, so it's liquefied. 00:36:42.090 --> 00:36:44.460 align:middle line:90% And it literally evaporates off. 00:36:44.460 --> 00:36:45.960 align:middle line:90% It has a finite vapor pressure. 00:36:45.960 --> 00:36:47.700 align:middle line:90% It evaporates into the vacuum. 00:36:47.700 --> 00:36:51.130 align:middle line:84% The vacuum pressure is low enough, less than 10 00:36:51.130 --> 00:36:54.060 align:middle line:84% to the minus 5 torr so that the mean-free path is quite long. 00:36:54.060 --> 00:36:57.630 align:middle line:84% So it literally just is like a line-of-sight spraying paint, 00:36:57.630 --> 00:37:02.590 align:middle line:84% if you will, where the paint is some atomic species 00:37:02.590 --> 00:37:05.940 align:middle line:84% onto the surface of these wafers. 00:37:05.940 --> 00:37:09.480 align:middle line:84% And so this is a very common physical process. 00:37:09.480 --> 00:37:15.520 align:middle line:84% Not a whole lot of chemistry involved in thermal evaporation. 00:37:15.520 --> 00:37:18.765 align:middle line:84% So what for thin-film evaporation, what 00:37:18.765 --> 00:37:19.890 align:middle line:90% are some of the properties? 00:37:19.890 --> 00:37:22.920 align:middle line:84% Well, it's mostly what we call line-of-sight deposition 00:37:22.920 --> 00:37:24.250 align:middle line:90% because the pressure is low. 00:37:24.250 --> 00:37:27.282 align:middle line:84% So there's not a whole lot of collisions. 00:37:27.282 --> 00:37:28.740 align:middle line:84% Let's say I'm evaporating aluminum. 00:37:28.740 --> 00:37:32.910 align:middle line:84% As this aluminum comes off, it has a certain flux 00:37:32.910 --> 00:37:34.320 align:middle line:90% in all directions. 00:37:34.320 --> 00:37:37.290 align:middle line:84% And pretty much, the aluminum atoms 00:37:37.290 --> 00:37:40.643 align:middle line:84% find their way straight to the wafer, depending, of course, 00:37:40.643 --> 00:37:41.310 align:middle line:90% on the pressure. 00:37:41.310 --> 00:37:42.768 align:middle line:84% If the pressure is low enough, they 00:37:42.768 --> 00:37:46.630 align:middle line:84% can get straight to the wafer with very few collisions. 00:37:46.630 --> 00:37:48.730 align:middle line:84% The DEP rate, the deposition rate 00:37:48.730 --> 00:37:51.280 align:middle line:84% is really pretty much determined by the emitted flux 00:37:51.280 --> 00:37:54.320 align:middle line:90% and the geometry of the target. 00:37:54.320 --> 00:37:56.770 align:middle line:84% So it's much less complicated in some ways 00:37:56.770 --> 00:38:00.010 align:middle line:84% to calculate or to model it than it is in CVD. 00:38:00.010 --> 00:38:02.320 align:middle line:84% In CVD, the DEP rate-- well, we had that model 00:38:02.320 --> 00:38:03.580 align:middle line:90% as a function of temperature. 00:38:03.580 --> 00:38:07.510 align:middle line:84% But it's very hard to get an absolute DEP rate model. 00:38:07.510 --> 00:38:09.250 align:middle line:84% You always have to do fitting in CVD. 00:38:09.250 --> 00:38:14.940 align:middle line:84% This physical deposition is quite a bit simpler. 00:38:14.940 --> 00:38:16.800 align:middle line:84% This evaporation source, what is it? 00:38:16.800 --> 00:38:22.230 align:middle line:84% Well, usually it's a little cup called a crucible that's heated, 00:38:22.230 --> 00:38:25.440 align:middle line:84% that holds the hot the molten liquid. 00:38:25.440 --> 00:38:26.940 align:middle line:84% And there's two ways of modeling it. 00:38:26.940 --> 00:38:29.418 align:middle line:84% Sometimes, if it's small enough, depending on the geometry, 00:38:29.418 --> 00:38:31.710 align:middle line:84% it can be modeled as a point source, just like you have 00:38:31.710 --> 00:38:35.100 align:middle line:84% a point in space that's emitting aluminum atoms 00:38:35.100 --> 00:38:38.190 align:middle line:84% or whatever you're trying to evaporate in all directions. 00:38:38.190 --> 00:38:41.243 align:middle line:84% Sometimes, the cup is large enough, 00:38:41.243 --> 00:38:42.660 align:middle line:84% then you have to actually model it 00:38:42.660 --> 00:38:46.980 align:middle line:84% as a finite source, a small-area surface source. 00:38:46.980 --> 00:38:49.440 align:middle line:84% This is typically more applicable to evaporation 00:38:49.440 --> 00:38:51.180 align:middle line:90% systems. 00:38:51.180 --> 00:38:55.170 align:middle line:84% And I should mention, the heater here 00:38:55.170 --> 00:38:57.888 align:middle line:84% is just shown as this little box underneath it. 00:38:57.888 --> 00:38:59.430 align:middle line:84% There are different types of heaters. 00:38:59.430 --> 00:39:01.800 align:middle line:84% One is a very, very simple resistor. 00:39:01.800 --> 00:39:04.890 align:middle line:84% It's just a piece of wire, of tungsten wire that's 00:39:04.890 --> 00:39:07.680 align:middle line:84% wrapped up into a filament like you would have in a light bulb. 00:39:07.680 --> 00:39:09.390 align:middle line:84% And you pass a current through this wire, 00:39:09.390 --> 00:39:11.220 align:middle line:90% and it gets really hot. 00:39:11.220 --> 00:39:14.050 align:middle line:84% And that will heat the aluminum or whatever it is you're 00:39:14.050 --> 00:39:16.470 align:middle line:90% trying to evaporate. 00:39:16.470 --> 00:39:18.870 align:middle line:84% You can only get so hot with resistive heaters, though. 00:39:18.870 --> 00:39:21.133 align:middle line:84% So there's another way of heating up 00:39:21.133 --> 00:39:22.800 align:middle line:84% the material in the crucible, and that's 00:39:22.800 --> 00:39:25.860 align:middle line:84% to take an electron beam and to shoot the electron beam 00:39:25.860 --> 00:39:31.230 align:middle line:84% and to deposit energy by the electron beam into this solid 00:39:31.230 --> 00:39:33.640 align:middle line:84% here and liquefy the aluminum or whatever. 00:39:33.640 --> 00:39:35.940 align:middle line:84% So that's called e-beam evaporation, or electron beam 00:39:35.940 --> 00:39:36.660 align:middle line:90% evaporation. 00:39:36.660 --> 00:39:38.920 align:middle line:84% Two different methods of creating-- either case, 00:39:38.920 --> 00:39:41.580 align:middle line:84% you're creating molten material that then flashes 00:39:41.580 --> 00:39:44.940 align:middle line:90% off and creates an atomic flux. 00:39:44.940 --> 00:39:48.540 align:middle line:84% And that flux hits the surface of the wafer and is deposited. 00:39:48.540 --> 00:39:52.050 align:middle line:90% 00:39:52.050 --> 00:39:55.260 align:middle line:84% So if we go on to slide 15, this is 00:39:55.260 --> 00:39:58.090 align:middle line:84% a picture I took directly out of your text. 00:39:58.090 --> 00:40:00.900 align:middle line:84% And if you sit down with it and spend some time, 00:40:00.900 --> 00:40:03.630 align:middle line:84% you'll realize this is primarily, really, 00:40:03.630 --> 00:40:05.580 align:middle line:90% just geometry. 00:40:05.580 --> 00:40:08.160 align:middle line:84% So there are two cases, case A on the left, 00:40:08.160 --> 00:40:09.060 align:middle line:90% case B on the right. 00:40:09.060 --> 00:40:12.390 align:middle line:84% Case A is for what we call a point source, 00:40:12.390 --> 00:40:15.090 align:middle line:84% and case B is for a small planar surface 00:40:15.090 --> 00:40:17.773 align:middle line:90% source of a finite extent. 00:40:17.773 --> 00:40:19.940 align:middle line:84% So if we just look for a minute at the point source, 00:40:19.940 --> 00:40:22.220 align:middle line:84% this is where the source is located. 00:40:22.220 --> 00:40:26.870 align:middle line:84% Imagine it's some distance away, h, some perpendicular distance. 00:40:26.870 --> 00:40:31.550 align:middle line:84% H, there is a wafer holder on which you can put wafers. 00:40:31.550 --> 00:40:34.910 align:middle line:84% And let's say you have your wafer out here 00:40:34.910 --> 00:40:37.820 align:middle line:84% on this wafer holder at a distance L 00:40:37.820 --> 00:40:40.170 align:middle line:90% from this point from the center. 00:40:40.170 --> 00:40:42.270 align:middle line:90% This makes a right triangle. 00:40:42.270 --> 00:40:48.920 align:middle line:84% So the line H, L, and R three together make a triangle. 00:40:48.920 --> 00:40:52.970 align:middle line:84% For any of these deposition processes what really counts 00:40:52.970 --> 00:40:55.890 align:middle line:90% is the perpendicular flux. 00:40:55.890 --> 00:40:59.450 align:middle line:84% So it'd be the flux perpendicular to the surface. 00:40:59.450 --> 00:41:03.380 align:middle line:84% So you need to usually project this flux onto the surface, 00:41:03.380 --> 00:41:06.890 align:middle line:84% and that's where you end up with a cosine theta 00:41:06.890 --> 00:41:09.890 align:middle line:84% kind of dependence of the deposition rate. 00:41:09.890 --> 00:41:12.290 align:middle line:84% So anyway, we have this point source. 00:41:12.290 --> 00:41:18.570 align:middle line:84% It has a certain flux that it's evaporating, that it's emitting. 00:41:18.570 --> 00:41:20.670 align:middle line:84% And that is the rate of evaporation, 00:41:20.670 --> 00:41:25.110 align:middle line:84% or evap, divided by the solid angle, omega times 00:41:25.110 --> 00:41:27.280 align:middle line:84% r squared, where r is the distance. 00:41:27.280 --> 00:41:30.030 align:middle line:84% So if I take my point source and I just move it further away, 00:41:30.030 --> 00:41:34.440 align:middle line:84% the flux at that surface goes down like 1 over r squared. 00:41:34.440 --> 00:41:36.420 align:middle line:84% So you can adjust the evaporation rate 00:41:36.420 --> 00:41:39.480 align:middle line:84% by taking the wafers and moving them further or closer 00:41:39.480 --> 00:41:41.760 align:middle line:90% to the point source. 00:41:41.760 --> 00:41:43.770 align:middle line:84% You can adjust the flux, I'm sorry, 00:41:43.770 --> 00:41:44.970 align:middle line:90% with the 1 over r squared. 00:41:44.970 --> 00:41:49.920 align:middle line:84% This omega is the solid angle over which the source emits. 00:41:49.920 --> 00:41:56.350 align:middle line:84% It's equal to 4 pi if it's emitting in all directions. 00:41:56.350 --> 00:41:58.860 align:middle line:84% And in this deposition rate term here on the left 00:41:58.860 --> 00:42:02.158 align:middle line:84% is this equation for the velocity, the deposition rate. 00:42:02.158 --> 00:42:04.200 align:middle line:84% n is the density of the material being deposited. 00:42:04.200 --> 00:42:07.620 align:middle line:84% So here's a very simple formula to calculate the deposition 00:42:07.620 --> 00:42:08.730 align:middle line:90% rate of your reactor. 00:42:08.730 --> 00:42:11.460 align:middle line:84% It's the evaporation rate, which you'll have to calculate-- 00:42:11.460 --> 00:42:14.160 align:middle line:84% it depends on the temperature and all that-- 00:42:14.160 --> 00:42:18.850 align:middle line:84% divided by the solid angle times the density over r squared. 00:42:18.850 --> 00:42:23.760 align:middle line:84% And there's a cosine theta, theta sub k here. 00:42:23.760 --> 00:42:28.180 align:middle line:84% So theta k here is this angle right here. 00:42:28.180 --> 00:42:30.900 align:middle line:84% So as I move out, if I take a wafer 00:42:30.900 --> 00:42:34.380 align:middle line:84% and move it from right here and I move it along this line, 00:42:34.380 --> 00:42:36.450 align:middle line:84% the deposition rate is going to go down according 00:42:36.450 --> 00:42:40.680 align:middle line:90% to this cosine function. 00:42:40.680 --> 00:42:44.420 align:middle line:84% A small planar surface source is a slightly different. 00:42:44.420 --> 00:42:47.240 align:middle line:84% There are two angles involved here. 00:42:47.240 --> 00:42:51.370 align:middle line:84% There is this angle here, theta i. 00:42:51.370 --> 00:42:54.320 align:middle line:84% That's the angle between the perpendicular of the planar 00:42:54.320 --> 00:42:54.820 align:middle line:90% source. 00:42:54.820 --> 00:42:56.620 align:middle line:84% So this planar source is like my hand. 00:42:56.620 --> 00:42:59.170 align:middle line:84% It has some angle that it's making, 00:42:59.170 --> 00:43:02.560 align:middle line:84% theta i to the point of evaporation. 00:43:02.560 --> 00:43:05.260 align:middle line:84% And there's also the cosine theta k. 00:43:05.260 --> 00:43:07.900 align:middle line:90% 00:43:07.900 --> 00:43:12.610 align:middle line:84% So the key point is this outward flux, FK, from a point source p 00:43:12.610 --> 00:43:15.790 align:middle line:84% is independent of theta i because theta i really 00:43:15.790 --> 00:43:18.050 align:middle line:90% equals theta k in this case. 00:43:18.050 --> 00:43:20.620 align:middle line:84% So it's independent of-- it's emitting-- 00:43:20.620 --> 00:43:22.690 align:middle line:84% a point source emits the same around a sphere 00:43:22.690 --> 00:43:24.522 align:middle line:90% in all directions. 00:43:24.522 --> 00:43:26.640 align:middle line:84% But the outward flux from a small area source, 00:43:26.640 --> 00:43:31.600 align:middle line:84% like from my hand, it depends on some variable. 00:43:31.600 --> 00:43:33.990 align:middle line:84% It depends in some way cosine theta 00:43:33.990 --> 00:43:36.300 align:middle line:84% i to the n, where n is going to be 00:43:36.300 --> 00:43:37.935 align:middle line:90% a variable you end up fitting. 00:43:37.935 --> 00:43:40.890 align:middle line:90% 00:43:40.890 --> 00:43:43.230 align:middle line:84% Let's give-- I'll maybe give you an example 00:43:43.230 --> 00:43:46.080 align:middle line:84% to make that a little bit clearer on slide 16. 00:43:46.080 --> 00:43:49.590 align:middle line:84% This is a plot of exactly that structure. 00:43:49.590 --> 00:43:52.290 align:middle line:84% Here, I have a substrate, and I'm 00:43:52.290 --> 00:43:58.590 align:middle line:84% looking at the evaporation, the evaporation rate or the flux-- 00:43:58.590 --> 00:44:03.420 align:middle line:84% yeah, the surface evaporation rate as a function of l over h. 00:44:03.420 --> 00:44:05.580 align:middle line:84% So again, l is the distance that you go out 00:44:05.580 --> 00:44:09.000 align:middle line:84% from being directly under the source, 00:44:09.000 --> 00:44:11.340 align:middle line:84% and h is the distance of the substrate holder 00:44:11.340 --> 00:44:12.600 align:middle line:90% from the source. 00:44:12.600 --> 00:44:16.570 align:middle line:84% And this r, l, and h makes this right triangle. 00:44:16.570 --> 00:44:18.298 align:middle line:84% So as a function of l over h, you 00:44:18.298 --> 00:44:20.340 align:middle line:84% see the DEP rate can be calculated just according 00:44:20.340 --> 00:44:22.300 align:middle line:90% to these two equations. 00:44:22.300 --> 00:44:27.480 align:middle line:84% So when l equals h here, you can calculate the relative DEP rate 00:44:27.480 --> 00:44:31.800 align:middle line:84% compared to the EP rate directly perpendicular from the source. 00:44:31.800 --> 00:44:33.480 align:middle line:90% It's down by about-- 00:44:33.480 --> 00:44:37.267 align:middle line:84% it's like 30% of what it would be the DEP rate 00:44:37.267 --> 00:44:38.100 align:middle line:90% right at this point. 00:44:38.100 --> 00:44:40.017 align:middle line:84% So you get an idea of what the uniformity will 00:44:40.017 --> 00:44:42.652 align:middle line:90% be along this surface. 00:44:42.652 --> 00:44:43.860 align:middle line:90% And there's different curves. 00:44:43.860 --> 00:44:46.560 align:middle line:84% The dashed line here is for the surface source, 00:44:46.560 --> 00:44:48.520 align:middle line:84% and the point source is the solid line. 00:44:48.520 --> 00:44:52.650 align:middle line:84% So they have slightly different dependence on those parameters. 00:44:52.650 --> 00:44:56.370 align:middle line:90% 00:44:56.370 --> 00:45:01.010 align:middle line:84% So if we go on to slide 17, as we mentioned, 00:45:01.010 --> 00:45:03.380 align:middle line:84% the outward flux from a point source, p, 00:45:03.380 --> 00:45:05.960 align:middle line:90% is independent of this theta i. 00:45:05.960 --> 00:45:07.790 align:middle line:84% Whereas, from a small area source, 00:45:07.790 --> 00:45:09.380 align:middle line:90% it goes like cosine theta i. 00:45:09.380 --> 00:45:12.067 align:middle line:84% And in fact, if you look at this-- 00:45:12.067 --> 00:45:14.150 align:middle line:84% I think it becomes more intuitive if you just look 00:45:14.150 --> 00:45:17.180 align:middle line:90% at this diagram on slide 17. 00:45:17.180 --> 00:45:20.120 align:middle line:84% Look at diagram A. I have a point source right 00:45:20.120 --> 00:45:20.930 align:middle line:90% at this point. 00:45:20.930 --> 00:45:23.990 align:middle line:84% And I'm assuming that source has uniform-- 00:45:23.990 --> 00:45:25.110 align:middle line:90% that it's isotropic. 00:45:25.110 --> 00:45:26.600 align:middle line:84% It's the same in all directions-- 00:45:26.600 --> 00:45:28.080 align:middle line:90% emission from a point source. 00:45:28.080 --> 00:45:31.550 align:middle line:84% So each line, the length of a line 00:45:31.550 --> 00:45:34.290 align:middle line:90% is meant to represent the flux. 00:45:34.290 --> 00:45:36.570 align:middle line:84% So the flux is the same in all directions. 00:45:36.570 --> 00:45:40.040 align:middle line:84% So this is a very spherically uniform isotropic source. 00:45:40.040 --> 00:45:43.790 align:middle line:84% That would be like a point source. 00:45:43.790 --> 00:45:48.460 align:middle line:84% If you look at number B here, this 00:45:48.460 --> 00:45:52.570 align:middle line:84% is what they call ideal cosine emission from a small planar 00:45:52.570 --> 00:45:53.350 align:middle line:90% surface source. 00:45:53.350 --> 00:45:55.720 align:middle line:84% So imagine right here at this point in space 00:45:55.720 --> 00:45:57.340 align:middle line:90% I have a planar surface source. 00:45:57.340 --> 00:46:01.380 align:middle line:84% The flux is largest straight above because it's 00:46:01.380 --> 00:46:02.370 align:middle line:90% the longest arrow. 00:46:02.370 --> 00:46:04.860 align:middle line:90% And the flux then goes down. 00:46:04.860 --> 00:46:07.410 align:middle line:84% As you go from here to here, it goes down 00:46:07.410 --> 00:46:10.470 align:middle line:84% like a cosine theta, where n equals 1. 00:46:10.470 --> 00:46:13.600 align:middle line:90% So the flux is going down. 00:46:13.600 --> 00:46:16.180 align:middle line:84% And they're also, in c, there are nonideal, 00:46:16.180 --> 00:46:19.510 align:middle line:84% more anisotropic emission from a different type of small planar 00:46:19.510 --> 00:46:20.080 align:middle line:90% source. 00:46:20.080 --> 00:46:21.880 align:middle line:84% This is cosine theta to the n, in general, 00:46:21.880 --> 00:46:24.633 align:middle line:90% where n is some number. 00:46:24.633 --> 00:46:26.050 align:middle line:84% And so you can see what this means 00:46:26.050 --> 00:46:28.092 align:middle line:84% is you're getting much higher flux going straight 00:46:28.092 --> 00:46:29.690 align:middle line:84% above the source than you are going 00:46:29.690 --> 00:46:31.690 align:middle line:84% as you make an angle with respect to the source. 00:46:31.690 --> 00:46:35.200 align:middle line:84% So it really depends on how the source is set up. 00:46:35.200 --> 00:46:37.878 align:middle line:90% Why do we care about this? 00:46:37.878 --> 00:46:40.170 align:middle line:84% Well, we care because, depending on the type of source, 00:46:40.170 --> 00:46:42.030 align:middle line:84% we have to figure out how to place 00:46:42.030 --> 00:46:44.460 align:middle line:84% the wafers in the evaporator so you get the most 00:46:44.460 --> 00:46:46.620 align:middle line:84% uniform thickness across any given wafer 00:46:46.620 --> 00:46:47.910 align:middle line:90% or from wafer to wafer. 00:46:47.910 --> 00:46:50.820 align:middle line:84% So there are two different places 00:46:50.820 --> 00:46:55.710 align:middle line:84% you can place in common-- in evaporators. 00:46:55.710 --> 00:46:58.770 align:middle line:84% So if you have a point source, something 00:46:58.770 --> 00:47:00.510 align:middle line:84% that behaves like an ideal point source, 00:47:00.510 --> 00:47:02.890 align:middle line:84% you typically put it right in the center of a sphere. 00:47:02.890 --> 00:47:06.150 align:middle line:84% So you would put the evaporation source of platinum or aluminum 00:47:06.150 --> 00:47:07.980 align:middle line:90% right here in the center. 00:47:07.980 --> 00:47:10.110 align:middle line:84% Then you get a uniform flux in all directions, 00:47:10.110 --> 00:47:15.480 align:middle line:84% and you put the wafers along the circumference facing that point 00:47:15.480 --> 00:47:17.820 align:middle line:90% source. 00:47:17.820 --> 00:47:20.940 align:middle line:90% So use a spherical wafer holder. 00:47:20.940 --> 00:47:22.500 align:middle line:84% If you have a small surface source, 00:47:22.500 --> 00:47:24.840 align:middle line:84% you usually put the source on the inside of the sphere, 00:47:24.840 --> 00:47:26.310 align:middle line:90% right here. 00:47:26.310 --> 00:47:30.330 align:middle line:84% And you put the wafers, again, on the surface of the sphere. 00:47:30.330 --> 00:47:32.803 align:middle line:84% So putting it here tends to compensate 00:47:32.803 --> 00:47:35.220 align:middle line:84% for the cosine theta at the end because look what happens. 00:47:35.220 --> 00:47:36.720 align:middle line:90% This is my source right here. 00:47:36.720 --> 00:47:39.490 align:middle line:84% I have the highest flux, according to this diagram, 00:47:39.490 --> 00:47:41.490 align:middle line:84% I have highest flux straight in front of me. 00:47:41.490 --> 00:47:43.350 align:middle line:90% That wafer is the furthest away. 00:47:43.350 --> 00:47:44.770 align:middle line:90% So I have a high flux. 00:47:44.770 --> 00:47:46.500 align:middle line:90% It's the furthest distance. 00:47:46.500 --> 00:47:48.793 align:middle line:84% That 1 over r squared dependence will 00:47:48.793 --> 00:47:50.460 align:middle line:84% tend to compensate for that higher flux. 00:47:50.460 --> 00:47:52.950 align:middle line:84% And so you get the same DEP rate on this wafer 00:47:52.950 --> 00:47:54.270 align:middle line:90% as you do on this wafer. 00:47:54.270 --> 00:47:56.370 align:middle line:84% This wafer has a lower flux, right? 00:47:56.370 --> 00:47:58.650 align:middle line:90% It's going down by cosine theta. 00:47:58.650 --> 00:48:00.390 align:middle line:84% But on the other hand, it's also closer. 00:48:00.390 --> 00:48:02.890 align:middle line:84% So the 1 over r squared dependence will cancel that out. 00:48:02.890 --> 00:48:04.440 align:middle line:84% So this is typically where you want 00:48:04.440 --> 00:48:11.010 align:middle line:84% to put it for a surface source in an evaporator. 00:48:11.010 --> 00:48:13.440 align:middle line:84% OK, a little bit more on evaporation, slide 18. 00:48:13.440 --> 00:48:14.790 align:middle line:90% This is an interesting plot. 00:48:14.790 --> 00:48:18.420 align:middle line:84% This shows the vapor pressure in torr 00:48:18.420 --> 00:48:22.890 align:middle line:84% as a function of temperature for a number of different elements. 00:48:22.890 --> 00:48:27.630 align:middle line:84% Remember, if I go back to my flux equation or my deposition 00:48:27.630 --> 00:48:30.210 align:middle line:84% rate, going back to slide 15 for a minute, 00:48:30.210 --> 00:48:33.780 align:middle line:84% the flux was directly proportional to the evaporation 00:48:33.780 --> 00:48:37.908 align:middle line:84% rate over something times r squared. 00:48:37.908 --> 00:48:40.200 align:middle line:84% So I need to know what the evaporation rate is in order 00:48:40.200 --> 00:48:44.490 align:middle line:84% to calculate the DEP rate on the wafer. 00:48:44.490 --> 00:48:47.250 align:middle line:84% So again, going back to slide 18, this tells you, 00:48:47.250 --> 00:48:48.750 align:middle line:84% here, I have a formula here that you 00:48:48.750 --> 00:48:51.120 align:middle line:84% can calculate the evaporation rate here, 00:48:51.120 --> 00:48:54.750 align:middle line:84% depending on this vapor pressure, p sub e 00:48:54.750 --> 00:48:57.460 align:middle line:90% at the evaporation temperature. 00:48:57.460 --> 00:49:00.810 align:middle line:84% So the evaporation rate is naturally 00:49:00.810 --> 00:49:05.370 align:middle line:84% proportional to the vapor pressure at the evaporation 00:49:05.370 --> 00:49:06.490 align:middle line:90% temperature. 00:49:06.490 --> 00:49:08.615 align:middle line:84% So what you can see is you just need to-- depending 00:49:08.615 --> 00:49:10.698 align:middle line:84% on the element, depending on whether you're trying 00:49:10.698 --> 00:49:11.670 align:middle line:90% to evaporate indium-- 00:49:11.670 --> 00:49:14.070 align:middle line:84% well, indium you can do that at pretty low temperatures. 00:49:14.070 --> 00:49:15.330 align:middle line:90% Look at its vapor pressure. 00:49:15.330 --> 00:49:19.140 align:middle line:84% It skyrockets above about 550 centigrade. 00:49:19.140 --> 00:49:21.600 align:middle line:84% If you're trying to evaporate some refractory metal, 00:49:21.600 --> 00:49:24.210 align:middle line:84% for instance, molybdenum or tungsten, 00:49:24.210 --> 00:49:26.280 align:middle line:84% good luck with a resistance heater. 00:49:26.280 --> 00:49:27.870 align:middle line:84% You'd have to use an electron beam. 00:49:27.870 --> 00:49:29.537 align:middle line:84% You have to go to very high temperatures 00:49:29.537 --> 00:49:33.100 align:middle line:84% to get any kind of reasonable vapor pressure here. 00:49:33.100 --> 00:49:36.240 align:middle line:84% But just about-- as you can design a heater, an electron 00:49:36.240 --> 00:49:38.535 align:middle line:84% beam or a resistance heater, then you can evaporate it. 00:49:38.535 --> 00:49:40.410 align:middle line:84% That's the name of the game with evaporation. 00:49:40.410 --> 00:49:43.090 align:middle line:84% You can evaporate just about any element. 00:49:43.090 --> 00:49:47.950 align:middle line:84% So in a thin films lab, you can do just about anything. 00:49:47.950 --> 00:49:49.600 align:middle line:84% The DEP rate of some elements is going 00:49:49.600 --> 00:49:53.980 align:middle line:84% to be very slow because the evaporation rate is low. 00:49:53.980 --> 00:49:56.470 align:middle line:90% Its vapor pressure may be low. 00:49:56.470 --> 00:49:58.040 align:middle line:84% One thing-- a good point, though, 00:49:58.040 --> 00:50:00.730 align:middle line:84% is it's very difficult to evaporate alloys or compounds. 00:50:00.730 --> 00:50:05.410 align:middle line:84% As you can see that, depending on the material between aluminum 00:50:05.410 --> 00:50:08.050 align:middle line:84% and nickel or chromium and nickel or whatever, 00:50:08.050 --> 00:50:09.820 align:middle line:84% at a given temperature, their evaporation, 00:50:09.820 --> 00:50:12.640 align:middle line:84% their vapor pressures vary by orders of magnitude, 00:50:12.640 --> 00:50:13.760 align:middle line:90% orders of magnitude. 00:50:13.760 --> 00:50:17.530 align:middle line:84% And so if you had put in your little crucible a little bit 00:50:17.530 --> 00:50:19.600 align:middle line:84% of one element, a little bit of another element, 00:50:19.600 --> 00:50:21.230 align:middle line:84% they're going to come off at very different rates. 00:50:21.230 --> 00:50:23.020 align:middle line:84% So trying to evaporate a compound is tricky. 00:50:23.020 --> 00:50:25.312 align:middle line:84% You probably need two crucibles with their temperatures 00:50:25.312 --> 00:50:27.040 align:middle line:84% controlled electronically very carefully. 00:50:27.040 --> 00:50:29.970 align:middle line:90% So that gets a little tricky. 00:50:29.970 --> 00:50:32.370 align:middle line:84% Probably the most important reason 00:50:32.370 --> 00:50:36.720 align:middle line:84% why evaporation is not used in fabrication, in semiconductor 00:50:36.720 --> 00:50:41.130 align:middle line:84% manufacturing anymore is this DEP coverage is very poor. 00:50:41.130 --> 00:50:42.810 align:middle line:90% It's pretty much line-of-sight. 00:50:42.810 --> 00:50:46.350 align:middle line:84% So if you're trying to fill a trench, for example, 00:50:46.350 --> 00:50:48.600 align:middle line:84% in A here, and let's say you uniformly 00:50:48.600 --> 00:50:52.350 align:middle line:84% want to deposit the material all along the trench, because 00:50:52.350 --> 00:50:55.157 align:middle line:84% of shadowing effects, as this aluminum or whatever comes down, 00:50:55.157 --> 00:50:57.240 align:middle line:84% it's pretty much just going to deposit right here. 00:50:57.240 --> 00:50:59.640 align:middle line:84% It's not going to hit the sidewalls. 00:50:59.640 --> 00:51:01.840 align:middle line:84% So if you need this conformal deposition, 00:51:01.840 --> 00:51:04.045 align:middle line:84% you're not going to be able to get it by-- 00:51:04.045 --> 00:51:05.670 align:middle line:84% you're going to need to use sputtering. 00:51:05.670 --> 00:51:07.980 align:middle line:84% You're not be able to get it by evaporation. 00:51:07.980 --> 00:51:10.230 align:middle line:84% There are other reasons it's rarely used, 00:51:10.230 --> 00:51:14.910 align:middle line:84% not just the line-of-sight, but the evaporator itself 00:51:14.910 --> 00:51:17.168 align:middle line:90% is very hot. 00:51:17.168 --> 00:51:19.210 align:middle line:84% When you have a little hot body, a hot metal body 00:51:19.210 --> 00:51:21.002 align:middle line:84% like that, any impurities on the metal body 00:51:21.002 --> 00:51:22.730 align:middle line:84% can end up being put onto your wafer. 00:51:22.730 --> 00:51:25.150 align:middle line:84% So it's a little bit tricky to control impurities. 00:51:25.150 --> 00:51:29.840 align:middle line:84% And the electron beam itself, when it strikes things, 00:51:29.840 --> 00:51:32.500 align:middle line:84% remember, if you have a high-energetic electron striking 00:51:32.500 --> 00:51:34.150 align:middle line:84% something, you can create x-rays. 00:51:34.150 --> 00:51:35.590 align:middle line:90% You can create photons. 00:51:35.590 --> 00:51:37.430 align:middle line:84% That's how x-rays are actually created. 00:51:37.430 --> 00:51:43.060 align:middle line:84% So there is an inadvertent creation of energetic material, 00:51:43.060 --> 00:51:46.750 align:middle line:84% energetic things, like x-rays, which can then hit the wafer 00:51:46.750 --> 00:51:51.135 align:middle line:84% and interact with the wafer, potentially causing oxide damage 00:51:51.135 --> 00:51:52.010 align:middle line:90% and things like that. 00:51:52.010 --> 00:51:57.040 align:middle line:84% So evaporation is used in MEMS or maybe in research these days. 00:51:57.040 --> 00:52:03.010 align:middle line:84% But you don't find it too much in CMOS manufacturing. 00:52:03.010 --> 00:52:06.790 align:middle line:84% What you do find in MOS manufacturing fabs today 00:52:06.790 --> 00:52:09.130 align:middle line:90% is sputter deposition. 00:52:09.130 --> 00:52:11.200 align:middle line:84% So we're going to spend more time talking 00:52:11.200 --> 00:52:13.330 align:middle line:90% about this in the PVD area. 00:52:13.330 --> 00:52:14.380 align:middle line:90% What is sputter DEP? 00:52:14.380 --> 00:52:17.740 align:middle line:90% Well, this uses a plasma. 00:52:17.740 --> 00:52:19.660 align:middle line:84% It's different from plasma-enhanced CVD. 00:52:19.660 --> 00:52:22.835 align:middle line:84% So it's not so much chemical process. 00:52:22.835 --> 00:52:24.460 align:middle line:84% You use a plasma, but this time, you're 00:52:24.460 --> 00:52:27.280 align:middle line:84% physically using the plasma to hit a target 00:52:27.280 --> 00:52:31.420 align:middle line:84% and sputter off or dislodge atoms, dislodge atoms. 00:52:31.420 --> 00:52:33.730 align:middle line:84% And they will then find their way onto the wafer 00:52:33.730 --> 00:52:36.030 align:middle line:90% to form the film. 00:52:36.030 --> 00:52:38.190 align:middle line:84% That's what sputtering is all about. 00:52:38.190 --> 00:52:40.415 align:middle line:84% Sputtering uses higher pressures than evaporation. 00:52:40.415 --> 00:52:41.790 align:middle line:84% Evaporation, remember, I said you 00:52:41.790 --> 00:52:43.457 align:middle line:84% need to have 10 to the minus 5 at least, 00:52:43.457 --> 00:52:46.080 align:middle line:90% probably 10 to the minus 6 torr. 00:52:46.080 --> 00:52:48.540 align:middle line:84% That way, that you can evaporate and just not 00:52:48.540 --> 00:52:51.720 align:middle line:84% hit anything as that atom goes to your wafer. 00:52:51.720 --> 00:52:53.220 align:middle line:84% Higher pressures here, we're talking 00:52:53.220 --> 00:52:55.680 align:middle line:90% about 1 to 100 millitorr. 00:52:55.680 --> 00:53:00.060 align:middle line:84% So that's 1 to 10 to the minus 3 torr, something in that range. 00:53:00.060 --> 00:53:02.850 align:middle line:84% Sputtering is typically better at depositing an alloy 00:53:02.850 --> 00:53:05.970 align:middle line:84% because the sputter rates vary a little bit with the material, 00:53:05.970 --> 00:53:08.800 align:middle line:84% with the element, but not so much-- 00:53:08.800 --> 00:53:12.510 align:middle line:84% this is an example on slide 19, a very elementary DC sputter 00:53:12.510 --> 00:53:14.490 align:middle line:90% deposition. 00:53:14.490 --> 00:53:17.550 align:middle line:84% I have some sputtering gas that I put in the inlet. 00:53:17.550 --> 00:53:19.710 align:middle line:90% And argon is very commonly used. 00:53:19.710 --> 00:53:21.300 align:middle line:90% And why do we use argon? 00:53:21.300 --> 00:53:22.740 align:middle line:90% It's inert, right? 00:53:22.740 --> 00:53:23.890 align:middle line:90% It's a noble gas. 00:53:23.890 --> 00:53:24.540 align:middle line:90% It's big. 00:53:24.540 --> 00:53:25.140 align:middle line:90% It's heavy. 00:53:25.140 --> 00:53:26.970 align:middle line:90% It's fat. 00:53:26.970 --> 00:53:30.300 align:middle line:84% Given enough kinetic energy, it can sputter off something 00:53:30.300 --> 00:53:35.460 align:middle line:84% off your target, your electrode, whatever the materials 00:53:35.460 --> 00:53:36.610 align:middle line:90% you're trying to get off. 00:53:36.610 --> 00:53:38.530 align:middle line:90% So argon is commonly used. 00:53:38.530 --> 00:53:42.400 align:middle line:84% You create this argon glow discharge of this plasma. 00:53:42.400 --> 00:53:45.367 align:middle line:90% The argon is accelerated. 00:53:45.367 --> 00:53:47.200 align:middle line:84% It's positive, so the argon accelerated here 00:53:47.200 --> 00:53:48.158 align:middle line:90% towards this electrode. 00:53:48.158 --> 00:53:50.260 align:middle line:84% It knocks off the aluminum or whatever, which 00:53:50.260 --> 00:53:52.250 align:middle line:90% finds its way onto the wafer. 00:53:52.250 --> 00:53:55.450 align:middle line:84% So this is a common DC sputtering system, which you 00:53:55.450 --> 00:53:57.355 align:middle line:90% would use to deposit metals. 00:53:57.355 --> 00:54:00.628 align:middle line:90% 00:54:00.628 --> 00:54:02.170 align:middle line:84% So let's look a little more carefully 00:54:02.170 --> 00:54:06.340 align:middle line:84% on slide 20 at what the plasma consists of. 00:54:06.340 --> 00:54:09.040 align:middle line:84% So this, again, now I've turned the plasma on its side. 00:54:09.040 --> 00:54:15.520 align:middle line:84% Here, the negative electrode has a voltage v sub c, the cathode, 00:54:15.520 --> 00:54:16.810 align:middle line:90% as it's called. 00:54:16.810 --> 00:54:19.070 align:middle line:90% The cathode is the target. 00:54:19.070 --> 00:54:21.850 align:middle line:84% So this material would be aluminum or platinum or whatever 00:54:21.850 --> 00:54:23.650 align:middle line:90% it is you're trying to sputter. 00:54:23.650 --> 00:54:26.980 align:middle line:84% On the other side, you have the anode, which in a DC system 00:54:26.980 --> 00:54:27.820 align:middle line:90% would be grounded. 00:54:27.820 --> 00:54:34.000 align:middle line:84% And you put your wafers on the grounded-- or on the anode. 00:54:34.000 --> 00:54:36.880 align:middle line:84% The plasma itself contains-- pretty much, 00:54:36.880 --> 00:54:38.320 align:middle line:84% it's electrically overall neutral. 00:54:38.320 --> 00:54:42.770 align:middle line:84% So it has equal numbers of positive argon ions. 00:54:42.770 --> 00:54:44.740 align:middle line:84% So the argon ends up being positively ionized, 00:54:44.740 --> 00:54:47.650 align:middle line:84% and electrons that are in the plasma, as well 00:54:47.650 --> 00:54:50.690 align:middle line:90% as some neutral argon atoms. 00:54:50.690 --> 00:54:52.660 align:middle line:84% And if you take plasma physics class, 00:54:52.660 --> 00:54:54.260 align:middle line:90% it'll explain in more detail. 00:54:54.260 --> 00:54:56.470 align:middle line:84% But for now, you can take this as a given 00:54:56.470 --> 00:54:59.350 align:middle line:84% that if you plot the voltage here 00:54:59.350 --> 00:55:03.100 align:middle line:84% in the plasma as a function of distance from the cathode, 00:55:03.100 --> 00:55:05.440 align:middle line:84% most of the voltage drop from here to here-- 00:55:05.440 --> 00:55:08.020 align:middle line:84% the plasma has a positive bias here-- most of it 00:55:08.020 --> 00:55:11.870 align:middle line:90% occurs over the cathode sheath. 00:55:11.870 --> 00:55:15.710 align:middle line:84% So most of it occurs right here right near the cathode. 00:55:15.710 --> 00:55:17.410 align:middle line:84% So that means that the argon then 00:55:17.410 --> 00:55:20.000 align:middle line:84% can be accelerated in this voltage drop going 00:55:20.000 --> 00:55:21.250 align:middle line:90% from the center of the plasma. 00:55:21.250 --> 00:55:23.653 align:middle line:84% Here, it'll be accelerated towards the cathode, 00:55:23.653 --> 00:55:25.570 align:middle line:84% towards the aluminum target or whatever it is. 00:55:25.570 --> 00:55:27.910 align:middle line:84% And then it'll gain enough kinetic energy 00:55:27.910 --> 00:55:33.170 align:middle line:84% that it can knock off the aluminum. 00:55:33.170 --> 00:55:37.070 align:middle line:84% So on slide 21, these positive argon ions 00:55:37.070 --> 00:55:40.340 align:middle line:84% are accelerated by this voltage, this voltage drop 00:55:40.340 --> 00:55:43.280 align:middle line:90% across this cathode sheath. 00:55:43.280 --> 00:55:46.580 align:middle line:84% And they hit the target, and they sputter off, say, aluminum. 00:55:46.580 --> 00:55:50.300 align:middle line:84% These aluminums now then travel through the plasma, 00:55:50.300 --> 00:55:56.980 align:middle line:84% and they deposit on the wafers that are sitting on the anode. 00:55:56.980 --> 00:56:01.290 align:middle line:84% So what rate limits you here is the sputtering rate, 00:56:01.290 --> 00:56:05.100 align:middle line:84% and it depends on the sputtering yield, which we call y 00:56:05.100 --> 00:56:07.050 align:middle line:84% as the number of atoms of aluminum 00:56:07.050 --> 00:56:10.110 align:middle line:84% or the number of atoms of a target that come off 00:56:10.110 --> 00:56:12.160 align:middle line:90% per incident argon ion. 00:56:12.160 --> 00:56:15.370 align:middle line:84% So that's the definition of the sputter yield. 00:56:15.370 --> 00:56:17.190 align:middle line:84% So if you have a very high sputter yield, 00:56:17.190 --> 00:56:20.400 align:middle line:84% for every argon ion that comes in you'll get an aluminum off. 00:56:20.400 --> 00:56:22.350 align:middle line:84% That would be a high sputter yield. 00:56:22.350 --> 00:56:25.770 align:middle line:90% 00:56:25.770 --> 00:56:28.350 align:middle line:84% Looking at slide 22, the sputter yield y 00:56:28.350 --> 00:56:30.428 align:middle line:90% is a function of the energy. 00:56:30.428 --> 00:56:32.970 align:middle line:84% So that'll be the dependent on the bias in the plasma, right, 00:56:32.970 --> 00:56:33.900 align:middle line:90% the voltage drop. 00:56:33.900 --> 00:56:37.260 align:middle line:84% And it'll be also dependent on the mass of the ion-- 00:56:37.260 --> 00:56:38.740 align:middle line:90% argon is very commonly used. 00:56:38.740 --> 00:56:41.580 align:middle line:84% It's very massive-- and on the target material. 00:56:41.580 --> 00:56:43.500 align:middle line:84% It'll also be a function of the incident 00:56:43.500 --> 00:56:47.240 align:middle line:90% angle of the argon coming in. 00:56:47.240 --> 00:56:48.830 align:middle line:84% Why the sputter yield doesn't vary 00:56:48.830 --> 00:56:52.340 align:middle line:84% that much, not nearly as much as the vapor pressure. 00:56:52.340 --> 00:56:55.340 align:middle line:84% Vapor pressure varies by 5 or 10 orders of magnitude 00:56:55.340 --> 00:56:56.540 align:middle line:90% among the elements. 00:56:56.540 --> 00:56:59.220 align:middle line:84% That's not the case with sputter yield. 00:56:59.220 --> 00:57:02.270 align:middle line:84% So sputtering is pretty good for depositing alloys. 00:57:02.270 --> 00:57:05.210 align:middle line:84% This is just a schematic on slide 22 of what's going on. 00:57:05.210 --> 00:57:07.100 align:middle line:84% Let's say I have an aluminum target here, 00:57:07.100 --> 00:57:09.710 align:middle line:84% and here's my wafer on the surface. 00:57:09.710 --> 00:57:12.560 align:middle line:84% And this is the cathode sheath of the dark space. 00:57:12.560 --> 00:57:14.420 align:middle line:84% The aluminum ions that are in the plasma 00:57:14.420 --> 00:57:16.485 align:middle line:84% are accelerated towards this negative electrode. 00:57:16.485 --> 00:57:18.860 align:middle line:84% They have enough energy that they can knock in a little-- 00:57:18.860 --> 00:57:20.360 align:middle line:84% I'm sorry, the argon accelerate it. 00:57:20.360 --> 00:57:21.650 align:middle line:90% They knock an aluminum off. 00:57:21.650 --> 00:57:23.300 align:middle line:84% It finds its way down to the wafer, 00:57:23.300 --> 00:57:26.120 align:middle line:84% where it may move around a little bit if the wafer is hot. 00:57:26.120 --> 00:57:29.135 align:middle line:84% But it pretty much will stick and deposit. 00:57:29.135 --> 00:57:31.760 align:middle line:90% 00:57:31.760 --> 00:57:33.830 align:middle line:84% There are other species, of course, around. 00:57:33.830 --> 00:57:37.040 align:middle line:84% There are electrons in the glow discharge. 00:57:37.040 --> 00:57:40.020 align:middle line:84% And depending on what's going on, 00:57:40.020 --> 00:57:42.140 align:middle line:84% there may be some other impurities as well, 00:57:42.140 --> 00:57:44.390 align:middle line:84% depending on the cleanliness of the vacuum. 00:57:44.390 --> 00:57:47.210 align:middle line:90% 00:57:47.210 --> 00:57:49.060 align:middle line:90% We'll go to slide 23. 00:57:49.060 --> 00:57:51.430 align:middle line:84% This just gives you a schematic, which 00:57:51.430 --> 00:57:55.240 align:middle line:84% is supposed to point out that a sputtering target itself 00:57:55.240 --> 00:57:56.530 align:middle line:90% is generally quite large. 00:57:56.530 --> 00:57:59.050 align:middle line:84% If you go to buy a target for aluminum, 00:57:59.050 --> 00:58:02.050 align:middle line:84% it can be big, a couple of feet on a side. 00:58:02.050 --> 00:58:05.120 align:middle line:84% It will be much larger than the wafer. 00:58:05.120 --> 00:58:08.140 align:middle line:84% And so because it's so large compared to the wafer, 00:58:08.140 --> 00:58:11.590 align:middle line:84% it gives you a wide range of arrival angles 00:58:11.590 --> 00:58:13.010 align:middle line:90% in contrast to a point source. 00:58:13.010 --> 00:58:17.440 align:middle line:84% So on the left, you can imagine if I had a point source, 00:58:17.440 --> 00:58:20.110 align:middle line:84% the arrival angle here is somewhat limited. 00:58:20.110 --> 00:58:21.640 align:middle line:84% But if I have a source that's really 00:58:21.640 --> 00:58:27.130 align:middle line:84% much bigger than the wafer, I get a range of arrival angles. 00:58:27.130 --> 00:58:31.030 align:middle line:84% And at any given point, I can get pretty good uniformity, 00:58:31.030 --> 00:58:33.490 align:middle line:84% reasonably low shadowing effects. 00:58:33.490 --> 00:58:35.863 align:middle line:84% In the evaporator, there were a lot 00:58:35.863 --> 00:58:37.780 align:middle line:84% of-- when we talked about thermal evaporation, 00:58:37.780 --> 00:58:40.760 align:middle line:84% if you have a step that you're evaporating into or a hole, 00:58:40.760 --> 00:58:42.010 align:middle line:90% you can get shadowing effects. 00:58:42.010 --> 00:58:44.770 align:middle line:84% Sputtering doesn't suffer nearly from those kind 00:58:44.770 --> 00:58:45.700 align:middle line:90% of shadowing effects. 00:58:45.700 --> 00:58:47.905 align:middle line:84% It's not really considered line of sight. 00:58:47.905 --> 00:58:51.460 align:middle line:90% 00:58:51.460 --> 00:58:56.830 align:middle line:90% OK, so let's look at slide 24. 00:58:56.830 --> 00:59:02.110 align:middle line:84% I just to copied this page from this text or this book called 00:59:02.110 --> 00:59:05.210 align:middle line:84% The User's Guide to Vacuum Technology by O'Hanlon. 00:59:05.210 --> 00:59:07.960 align:middle line:84% I just wanted to remind people of an important concept 00:59:07.960 --> 00:59:10.330 align:middle line:84% where you're talking about these vacuum processes, 00:59:10.330 --> 00:59:13.810 align:middle line:90% like evaporation and sputtering. 00:59:13.810 --> 00:59:18.040 align:middle line:84% The concept of a mean-free path, just 00:59:18.040 --> 00:59:20.440 align:middle line:84% to remind you what this is if you haven't seen it before. 00:59:20.440 --> 00:59:24.010 align:middle line:84% This mean-free path here, a lambda, 00:59:24.010 --> 00:59:26.230 align:middle line:84% of a particle from the kinetic gas theory, 00:59:26.230 --> 00:59:30.260 align:middle line:84% it depends on temperature, directly proportional 00:59:30.260 --> 00:59:31.530 align:middle line:90% to the temperature. 00:59:31.530 --> 00:59:34.740 align:middle line:84% And it's inversely proportional to the pressure. 00:59:34.740 --> 00:59:38.510 align:middle line:84% So for example, in just a typical sputtering system, 00:59:38.510 --> 00:59:42.260 align:middle line:84% the pressure, let's say, maybe is 5 millitorr. 00:59:42.260 --> 00:59:43.842 align:middle line:84% So this gives us a mean-free path 00:59:43.842 --> 00:59:45.050 align:middle line:90% on the order of a centimeter. 00:59:45.050 --> 00:59:49.370 align:middle line:84% So you can think of these argon atoms or ions moving around 00:59:49.370 --> 00:59:53.030 align:middle line:84% or the aluminum moving around in that five militorr 00:59:53.030 --> 00:59:57.110 align:middle line:84% system with about every centimeter it has a collision. 00:59:57.110 --> 00:59:59.960 align:middle line:84% So that's a lot less than the physical path length 00:59:59.960 --> 01:00:02.490 align:middle line:84% because a sputtering system might be this big. 01:00:02.490 --> 01:00:04.850 align:middle line:84% So you have your target up here, your wafers down here. 01:00:04.850 --> 01:00:06.060 align:middle line:84% It has to go a lot of centimeters. 01:00:06.060 --> 01:00:07.580 align:middle line:84% So that atom sort of jiggles around, 01:00:07.580 --> 01:00:10.250 align:middle line:84% and it has a lot of collisions as it finally gets 01:00:10.250 --> 01:00:11.930 align:middle line:90% to come to rest on the surface. 01:00:11.930 --> 01:00:14.295 align:middle line:84% Contrast to an evaporation, could be just as long 01:00:14.295 --> 01:00:15.170 align:middle line:90% or a longer distance. 01:00:15.170 --> 01:00:16.580 align:middle line:84% But evaporation is line-of-sight. 01:00:16.580 --> 01:00:17.080 align:middle line:90% Bingo. 01:00:17.080 --> 01:00:19.310 align:middle line:90% Goes right from here to here. 01:00:19.310 --> 01:00:22.470 align:middle line:84% I'm at 10 to the minus 6 torr there with no collisions. 01:00:22.470 --> 01:00:25.040 align:middle line:84% So this mean-free path is also what 01:00:25.040 --> 01:00:31.080 align:middle line:84% determines the conformality of the deposition in sputtering 01:00:31.080 --> 01:00:35.660 align:middle line:84% and how it's different from evaporation. 01:00:35.660 --> 01:00:41.690 align:middle line:84% Looking at slide number 25 here, contrasting two different cases. 01:00:41.690 --> 01:00:46.190 align:middle line:84% On the left, case A is for isotropic flux arrival. 01:00:46.190 --> 01:00:49.490 align:middle line:84% Now, I'm considering-- I have my surface of a wafer, 01:00:49.490 --> 01:00:53.498 align:middle line:84% and I've got a flux of things coming in from the sputtering. 01:00:53.498 --> 01:00:55.040 align:middle line:84% That's not from evaporation, but it's 01:00:55.040 --> 01:00:58.640 align:middle line:84% coming in at a certain angular distribution 01:00:58.640 --> 01:01:01.230 align:middle line:90% in the sputtering tool. 01:01:01.230 --> 01:01:06.020 align:middle line:84% And so if n equals 1, I'll get a cosine theta arrival angle 01:01:06.020 --> 01:01:13.190 align:middle line:84% distribution depending on the angle of this source 01:01:13.190 --> 01:01:16.600 align:middle line:90% with respect to the normal. 01:01:16.600 --> 01:01:20.290 align:middle line:84% If have an anisotropic flux, a rival flux, n will go up. 01:01:20.290 --> 01:01:23.570 align:middle line:84% n can be much greater than 1, maybe 10 or something like that. 01:01:23.570 --> 01:01:26.010 align:middle line:84% And if n is very large, it means most of the flux 01:01:26.010 --> 01:01:27.010 align:middle line:90% is coming straight down. 01:01:27.010 --> 01:01:30.750 align:middle line:84% Not much but it's coming from wide angles. 01:01:30.750 --> 01:01:32.420 align:middle line:84% So typically, for sputtering, we also 01:01:32.420 --> 01:01:35.690 align:middle line:90% use a cosine theta at the end. 01:01:35.690 --> 01:01:37.920 align:middle line:84% And it's the normal component of the flux. 01:01:37.920 --> 01:01:41.030 align:middle line:84% So it's the component in the normal direction 01:01:41.030 --> 01:01:45.610 align:middle line:84% that strikes the surface and determines the DEP rate. 01:01:45.610 --> 01:01:47.740 align:middle line:84% So the size and the type of source, 01:01:47.740 --> 01:01:51.310 align:middle line:84% the geometry of the overall sputtering system, the pressure 01:01:51.310 --> 01:01:53.950 align:middle line:84% and, therefore, the collisions and the gas phase, all of those 01:01:53.950 --> 01:01:56.620 align:middle line:84% are important in determining the arrival angle distribution, 01:01:56.620 --> 01:01:58.558 align:middle line:84% in determining what this flux looks 01:01:58.558 --> 01:01:59.725 align:middle line:90% like as a function of theta. 01:01:59.725 --> 01:02:03.120 align:middle line:90% 01:02:03.120 --> 01:02:05.490 align:middle line:84% So that says a little bit about DC sputtering. 01:02:05.490 --> 01:02:09.260 align:middle line:84% Now, you might say, well, that's fine 01:02:09.260 --> 01:02:12.500 align:middle line:84% if the target electrode is a conductor, like platinum 01:02:12.500 --> 01:02:13.580 align:middle line:90% or aluminum or whatever. 01:02:13.580 --> 01:02:16.530 align:middle line:84% What if I want to sputter a dielectric material? 01:02:16.530 --> 01:02:18.312 align:middle line:90% It doesn't conduct DC current. 01:02:18.312 --> 01:02:19.770 align:middle line:84% You're going to have some problems. 01:02:19.770 --> 01:02:23.360 align:middle line:84% Well, what people do is then you use an RF power source 01:02:23.360 --> 01:02:26.540 align:middle line:84% because you can transmit power through insulators 01:02:26.540 --> 01:02:29.570 align:middle line:90% in an AC sense by using RF. 01:02:29.570 --> 01:02:33.890 align:middle line:84% So it looks very much or somewhat similar to a DC system, 01:02:33.890 --> 01:02:37.460 align:middle line:84% except now you have an RF generator and a matching 01:02:37.460 --> 01:02:38.780 align:middle line:90% network. 01:02:38.780 --> 01:02:43.820 align:middle line:84% And you put your target or your electrode up here. 01:02:43.820 --> 01:02:45.710 align:middle line:84% And you notice, we'll talk about why this is, 01:02:45.710 --> 01:02:51.530 align:middle line:84% but it's a smaller area than the electrode that holds the wafers. 01:02:51.530 --> 01:02:55.610 align:middle line:84% Again, you still bring in argon, and it's accelerated 01:02:55.610 --> 01:02:56.690 align:middle line:90% towards this target. 01:02:56.690 --> 01:03:00.170 align:middle line:90% And you can sputter dielectrics. 01:03:00.170 --> 01:03:02.255 align:middle line:84% SiO2 can be sputtered in this way. 01:03:02.255 --> 01:03:05.500 align:middle line:90% 01:03:05.500 --> 01:03:10.530 align:middle line:84% Slide 27 shows an example of the steady state-- now, 01:03:10.530 --> 01:03:14.610 align:middle line:84% this is steady state voltage distribution in an RF sputtering 01:03:14.610 --> 01:03:15.580 align:middle line:90% system. 01:03:15.580 --> 01:03:18.660 align:middle line:84% And so it's a plot on the y-axis here 01:03:18.660 --> 01:03:21.130 align:middle line:84% is voltage as a function of distance. 01:03:21.130 --> 01:03:24.600 align:middle line:84% And here on the left is meant to be the target. 01:03:24.600 --> 01:03:27.460 align:middle line:84% This could be SiO2 if you're sputtering that. 01:03:27.460 --> 01:03:29.610 align:middle line:84% And on the right is the electrode 01:03:29.610 --> 01:03:31.925 align:middle line:90% where you put the wafer. 01:03:31.925 --> 01:03:33.550 align:middle line:84% And there are two different plots here. 01:03:33.550 --> 01:03:36.082 align:middle line:90% 01:03:36.082 --> 01:03:37.540 align:middle line:84% Well, first of all, because there's 01:03:37.540 --> 01:03:39.340 align:middle line:84% slower mobility of the ions, the ions 01:03:39.340 --> 01:03:41.020 align:middle line:84% are more massive than the electrons, 01:03:41.020 --> 01:03:44.320 align:middle line:84% the plasma always tends to bias positively with respect 01:03:44.320 --> 01:03:48.340 align:middle line:90% to the two electrodes. 01:03:48.340 --> 01:03:51.550 align:middle line:84% So the plasma tends to be positively biased. 01:03:51.550 --> 01:03:54.550 align:middle line:84% The question is, what does the voltage drop 01:03:54.550 --> 01:03:58.710 align:middle line:84% look like as a function of the area in these two electrodes? 01:03:58.710 --> 01:04:02.990 align:middle line:84% So it turns out, when the area of the electrodes is not equal, 01:04:02.990 --> 01:04:05.910 align:middle line:84% the field has to be higher at the smaller electrode. 01:04:05.910 --> 01:04:08.420 align:middle line:84% So this is my small-- let's say, if this is the smaller 01:04:08.420 --> 01:04:12.050 align:middle line:84% electrode, the field is higher because there's 01:04:12.050 --> 01:04:13.800 align:middle line:90% a higher current density. 01:04:13.800 --> 01:04:16.070 align:middle line:84% So to maintain overall current continuity, 01:04:16.070 --> 01:04:23.690 align:middle line:84% it turns out that this voltage drop here, v1, will be related 01:04:23.690 --> 01:04:27.990 align:middle line:90% to v2 inversely to the areas. 01:04:27.990 --> 01:04:31.100 align:middle line:84% So if I have equal area electrodes, 01:04:31.100 --> 01:04:32.580 align:middle line:90% I have the solid line. 01:04:32.580 --> 01:04:34.940 align:middle line:90% So there's sort of a drop. 01:04:34.940 --> 01:04:38.450 align:middle line:84% The voltage drop on the left, and the right is about the same. 01:04:38.450 --> 01:04:43.930 align:middle line:84% If I make the area-- this electrode where the wafer sits 01:04:43.930 --> 01:04:47.780 align:middle line:84% much larger, so by making the target electrode smaller, 01:04:47.780 --> 01:04:52.910 align:middle line:84% in other words, there's a larger voltage drop near the target. 01:04:52.910 --> 01:04:55.450 align:middle line:84% And so all the sputtering tends to occur near the target, 01:04:55.450 --> 01:04:57.800 align:middle line:84% not during your wafer, not to your wafer. 01:04:57.800 --> 01:05:03.280 align:middle line:84% So just by imbalancing the size of the electrode where 01:05:03.280 --> 01:05:04.750 align:middle line:84% you put the wafer and the electrode 01:05:04.750 --> 01:05:06.040 align:middle line:84% where you do the sputtering, you can 01:05:06.040 --> 01:05:07.750 align:middle line:84% cause most of the voltage drop to be 01:05:07.750 --> 01:05:10.330 align:middle line:84% on this side with a smaller electrode. 01:05:10.330 --> 01:05:14.200 align:middle line:84% And then that way you ensure that you get sputtering, 01:05:14.200 --> 01:05:16.490 align:middle line:84% not of your wafer, but sputtering of the target 01:05:16.490 --> 01:05:16.990 align:middle line:90% material. 01:05:16.990 --> 01:05:21.060 align:middle line:90% 01:05:21.060 --> 01:05:24.630 align:middle line:84% And going on to slide 28, the other thing you can do, 01:05:24.630 --> 01:05:27.900 align:middle line:84% you can have a separate RF bias, which 01:05:27.900 --> 01:05:34.930 align:middle line:84% allows you to control the voltage drop on this side. 01:05:34.930 --> 01:05:36.910 align:middle line:90% So you can do a cleaning step. 01:05:36.910 --> 01:05:39.030 align:middle line:84% So at the very beginning of sputtering, 01:05:39.030 --> 01:05:40.740 align:middle line:84% you often want to sputter a little bit 01:05:40.740 --> 01:05:42.030 align:middle line:90% off the wafer surface. 01:05:42.030 --> 01:05:45.180 align:middle line:84% Let's say you have a little bit of deposition, 01:05:45.180 --> 01:05:49.660 align:middle line:84% or you can also do a bias sputter deposition. 01:05:49.660 --> 01:05:57.540 align:middle line:84% So you can actually have the wafer be at some finite bias, 01:05:57.540 --> 01:06:00.060 align:middle line:84% and this ends up giving you more conformal deposition 01:06:00.060 --> 01:06:04.680 align:middle line:84% because you can get a more highly directional sputtering 01:06:04.680 --> 01:06:05.950 align:middle line:90% flux. 01:06:05.950 --> 01:06:09.450 align:middle line:84% So there are a couple different methods in using sputtering 01:06:09.450 --> 01:06:14.100 align:middle line:84% where you can control the arrival angle with bias 01:06:14.100 --> 01:06:15.180 align:middle line:90% sputter deposition. 01:06:15.180 --> 01:06:19.140 align:middle line:90% 01:06:19.140 --> 01:06:22.800 align:middle line:84% OK, so the final technique that I 01:06:22.800 --> 01:06:25.500 align:middle line:84% want to talk about today for PVD is 01:06:25.500 --> 01:06:28.950 align:middle line:84% this ionized sputter deposition or high-density plasma 01:06:28.950 --> 01:06:30.280 align:middle line:90% sputtering. 01:06:30.280 --> 01:06:32.230 align:middle line:84% So far, let's just make it clear, 01:06:32.230 --> 01:06:35.560 align:middle line:84% what we've been talking about is the argon has been ionized. 01:06:35.560 --> 01:06:37.840 align:middle line:84% It's been accelerated towards the target. 01:06:37.840 --> 01:06:40.930 align:middle line:84% But the atoms coming off are pretty much neutral, 01:06:40.930 --> 01:06:44.690 align:middle line:84% the aluminum coming off or the platinum or whatever. 01:06:44.690 --> 01:06:46.600 align:middle line:84% But in some systems, you can arrange 01:06:46.600 --> 01:06:50.050 align:middle line:84% it, that the depositing atoms themselves are ionized, 01:06:50.050 --> 01:06:53.540 align:middle line:84% so not just the sputtering material. 01:06:53.540 --> 01:06:56.200 align:middle line:84% And so what you do is you place-- 01:06:56.200 --> 01:07:02.020 align:middle line:84% in addition to the target and the substrate 01:07:02.020 --> 01:07:06.580 align:middle line:84% that are biased with some kind of RF, you can put an RF coil. 01:07:06.580 --> 01:07:08.170 align:middle line:90% So now we have two RF sources. 01:07:08.170 --> 01:07:12.280 align:middle line:84% And you can put an RF coil around the plasma, 01:07:12.280 --> 01:07:13.822 align:middle line:84% and that's going to induce collisions 01:07:13.822 --> 01:07:16.363 align:middle line:84% in the plasma, which is going to tend to ionize the aluminum. 01:07:16.363 --> 01:07:17.800 align:middle line:84% So the aluminum comes off neutral, 01:07:17.800 --> 01:07:21.880 align:middle line:84% but you can ionize it with this inductively-coupled antenna, 01:07:21.880 --> 01:07:22.960 align:middle line:90% essentially? 01:07:22.960 --> 01:07:25.910 align:middle line:84% So now I have a different situation. 01:07:25.910 --> 01:07:30.100 align:middle line:84% I'm using the argon ions to be accelerated towards the target, 01:07:30.100 --> 01:07:31.150 align:middle line:90% get the aluminum off. 01:07:31.150 --> 01:07:34.300 align:middle line:84% And then I ionize the aluminum as it comes down. 01:07:34.300 --> 01:07:35.510 align:middle line:90% Why would I want to do that? 01:07:35.510 --> 01:07:38.150 align:middle line:84% Well, if I have now ions coming down, 01:07:38.150 --> 01:07:43.180 align:middle line:84% and if I bias the substrate, I can get a much narrower arrival 01:07:43.180 --> 01:07:45.983 align:middle line:84% angle distribution because they're being accelerated 01:07:45.983 --> 01:07:46.900 align:middle line:90% towards the substrate. 01:07:46.900 --> 01:07:49.120 align:middle line:84% Before, they were just sort of coming down and sort 01:07:49.120 --> 01:07:50.620 align:middle line:90% of collisions, mean-free path. 01:07:50.620 --> 01:07:52.750 align:middle line:84% And the aluminum was just hitting the surface. 01:07:52.750 --> 01:07:55.630 align:middle line:84% But now you can actually accelerate them 01:07:55.630 --> 01:07:56.560 align:middle line:90% towards the substrate. 01:07:56.560 --> 01:08:00.070 align:middle line:84% Maybe you can use this to fill a deep contact hole. 01:08:00.070 --> 01:08:01.750 align:middle line:90% Here's an example. 01:08:01.750 --> 01:08:03.550 align:middle line:84% On slide 29, you have a deep hole. 01:08:03.550 --> 01:08:06.760 align:middle line:84% And look at the-- schematically, on the right, in B, 01:08:06.760 --> 01:08:09.460 align:middle line:84% the arrival angle distribution is very narrow. 01:08:09.460 --> 01:08:11.350 align:middle line:84% They're pretty much all coming straight down, 01:08:11.350 --> 01:08:13.017 align:middle line:84% compared to here where they're coming in 01:08:13.017 --> 01:08:14.130 align:middle line:90% at different directions. 01:08:14.130 --> 01:08:18.140 align:middle line:84% So you can then adjust in this high-density plasma 01:08:18.140 --> 01:08:22.370 align:middle line:84% or inductively-coupled-- people sometimes call this ICP 01:08:22.370 --> 01:08:24.109 align:middle line:90% sputtering-- 01:08:24.109 --> 01:08:27.109 align:middle line:84% you have another knob that you can turn, literally. 01:08:27.109 --> 01:08:30.529 align:middle line:90% 01:08:30.529 --> 01:08:33.287 align:middle line:84% So what are some common PVD films, 01:08:33.287 --> 01:08:34.370 align:middle line:90% and how are they put down? 01:08:34.370 --> 01:08:37.430 align:middle line:84% That's listed on this chart took from your text, 01:08:37.430 --> 01:08:39.439 align:middle line:90% here on slide 30. 01:08:39.439 --> 01:08:44.210 align:middle line:84% Aluminum put down in a number of different ways, 01:08:44.210 --> 01:08:47.060 align:middle line:84% but sputter deposition is really the most common. 01:08:47.060 --> 01:08:49.330 align:middle line:84% The standard deposition of aluminum 01:08:49.330 --> 01:08:51.830 align:middle line:84% is to hold the wafers near to room temperature, 01:08:51.830 --> 01:08:54.830 align:middle line:84% pretty close to room temperature. 01:08:54.830 --> 01:08:57.710 align:middle line:84% There are some special sputter deposition systems 01:08:57.710 --> 01:09:00.300 align:middle line:84% where we'll put it down in about 400 to 500, 01:09:00.300 --> 01:09:01.670 align:middle line:90% so in other words, hot aluminum. 01:09:01.670 --> 01:09:04.430 align:middle line:84% And this is if you need to reflow the aluminum for better 01:09:04.430 --> 01:09:07.479 align:middle line:90% step coverage. 01:09:07.479 --> 01:09:10.390 align:middle line:84% You don't see aluminum deposited by chemical vapor deposition. 01:09:10.390 --> 01:09:13.450 align:middle line:90% That's extremely rare. 01:09:13.450 --> 01:09:15.430 align:middle line:90% It's usually sputtered. 01:09:15.430 --> 01:09:19.720 align:middle line:84% Another very common material you'll find in fabs 01:09:19.720 --> 01:09:21.010 align:middle line:90% is Ti or Ti-tungsten. 01:09:21.010 --> 01:09:23.620 align:middle line:90% This is also sputter deposited. 01:09:23.620 --> 01:09:28.100 align:middle line:84% Again, CVD can be very difficult. 01:09:28.100 --> 01:09:30.399 align:middle line:90% So people will use sputtering. 01:09:30.399 --> 01:09:33.890 align:middle line:84% Tungsten, tungsten is used to fill plugs. 01:09:33.890 --> 01:09:36.670 align:middle line:84% It's often put down by low-pressure CVD 01:09:36.670 --> 01:09:44.229 align:middle line:84% using tungsten WF 6, tungsten hexafluoride. 01:09:44.229 --> 01:09:48.170 align:middle line:84% Ti-silicide can be put down in a couple of different ways. 01:09:48.170 --> 01:09:49.300 align:middle line:90% It's usually sputtered. 01:09:49.300 --> 01:09:51.700 align:middle line:84% The titanium is usually sputtered initially, 01:09:51.700 --> 01:09:52.899 align:middle line:90% and then it's reacted. 01:09:52.899 --> 01:09:55.450 align:middle line:84% And we'll talk a little bit more in I 01:09:55.450 --> 01:09:57.400 align:middle line:84% think one of the last couple lectures 01:09:57.400 --> 01:09:59.770 align:middle line:90% on how silicides are formed. 01:09:59.770 --> 01:10:04.050 align:middle line:84% Ti-nitride is also sputtered but usually in a nitrogen plasma. 01:10:04.050 --> 01:10:06.540 align:middle line:84% Instead of using an inert gas like argon, 01:10:06.540 --> 01:10:09.180 align:middle line:84% they use nitrogen, which then can, not only 01:10:09.180 --> 01:10:11.190 align:middle line:84% knock the titanium off, but then react with it 01:10:11.190 --> 01:10:14.010 align:middle line:90% and then go down as Ti-nitride. 01:10:14.010 --> 01:10:18.580 align:middle line:84% So those are some common films you'll find in the fabs. 01:10:18.580 --> 01:10:24.570 align:middle line:84% So let me just summarize that the LPCVD uses low pressures. 01:10:24.570 --> 01:10:25.777 align:middle line:90% And why do we do that? 01:10:25.777 --> 01:10:27.360 align:middle line:84% We want to increase the diffusion rate 01:10:27.360 --> 01:10:30.750 align:middle line:84% through the boundary layer so you can get more uniform depth 01:10:30.750 --> 01:10:35.460 align:middle line:84% across the wafer, even when the wafers are very closely packed. 01:10:35.460 --> 01:10:37.320 align:middle line:84% So the nice thing about LPCVD tubes 01:10:37.320 --> 01:10:40.140 align:middle line:84% is you don't have to worry about how the gas flows. 01:10:40.140 --> 01:10:43.500 align:middle line:84% You just have to get a very uniform temperature. 01:10:43.500 --> 01:10:47.490 align:middle line:84% Polysilicon is the most commonly deposited material by LPCVD. 01:10:47.490 --> 01:10:49.718 align:middle line:90% It's the gate in CMOS. 01:10:49.718 --> 01:10:51.510 align:middle line:84% We've barely scratched the surface of poly. 01:10:51.510 --> 01:10:53.070 align:middle line:84% I think it's a reasonably good topic 01:10:53.070 --> 01:10:55.710 align:middle line:84% to do if you want to do on one of your reports. 01:10:55.710 --> 01:10:58.470 align:middle line:84% You need to optimize the temperature and the pressure 01:10:58.470 --> 01:11:02.590 align:middle line:84% and the exact conditions to get the right grain structure so you 01:11:02.590 --> 01:11:04.960 align:middle line:84% can get diffusion of dopants properly 01:11:04.960 --> 01:11:07.030 align:middle line:84% and good electrical activation in the grains. 01:11:07.030 --> 01:11:09.970 align:middle line:90% 01:11:09.970 --> 01:11:12.090 align:middle line:84% This plasma enhancement can be used 01:11:12.090 --> 01:11:14.860 align:middle line:84% to enhance the DEP rate at low temperatures, 01:11:14.860 --> 01:11:16.620 align:middle line:90% especially for dielectrics. 01:11:16.620 --> 01:11:18.360 align:middle line:84% We talked about the Novellus Concept 01:11:18.360 --> 01:11:19.480 align:middle line:90% One, which is an example. 01:11:19.480 --> 01:11:22.530 align:middle line:84% There are many, many examples of PECVD systems 01:11:22.530 --> 01:11:24.630 align:middle line:90% that are used in fabs. 01:11:24.630 --> 01:11:27.780 align:middle line:84% Typically, use plasmas when you have a limited thermal budget 01:11:27.780 --> 01:11:30.570 align:middle line:84% or if you happen to be doing MEMS or you have something where 01:11:30.570 --> 01:11:33.120 align:middle line:84% you really need thick layers and doing it thermally just 01:11:33.120 --> 01:11:35.880 align:middle line:90% is not practical. 01:11:35.880 --> 01:11:37.830 align:middle line:84% The most commonly used PVD technique 01:11:37.830 --> 01:11:45.040 align:middle line:84% is sputtering, either DC for metals or RF for dielectrics. 01:11:45.040 --> 01:11:47.350 align:middle line:84% In contrast, to DEP evaporation, you 01:11:47.350 --> 01:11:51.820 align:middle line:84% can get very good step coverage and filling of high aspect ratio 01:11:51.820 --> 01:11:52.780 align:middle line:90% holes. 01:11:52.780 --> 01:11:55.132 align:middle line:84% So far, we've talked very qualitatively. 01:11:55.132 --> 01:11:56.590 align:middle line:84% What I want to go through next time 01:11:56.590 --> 01:11:59.740 align:middle line:84% is some actual models that people have developed, 01:11:59.740 --> 01:12:02.710 align:middle line:84% methodologies for modeling these processes, 01:12:02.710 --> 01:12:06.820 align:middle line:84% both CVD and particularly sputtering of PVD. 01:12:06.820 --> 01:12:07.790 align:middle line:90% So that's all I have. 01:12:07.790 --> 01:12:10.450 align:middle line:84% If you came in late, remind you, the homework number five 01:12:10.450 --> 01:12:11.020 align:middle line:90% went out. 01:12:11.020 --> 01:12:14.620 align:middle line:84% And there is a problem on the back, so don't miss part 3B. 01:12:14.620 --> 01:12:17.293 align:middle line:90% 01:12:17.293 --> 01:12:18.710 align:middle line:84% And if you haven't signed up, make 01:12:18.710 --> 01:12:21.000 align:middle line:84% sure-- today is the last day to sign up for a topic. 01:12:21.000 --> 01:12:24.250 align:middle line:84% So I've got the clipboard there in the back. 01:12:24.250 --> 01:12:27.000 align:middle line:90%