WEBVTT 00:00:00.000 --> 00:00:02.770 align:middle line:90% 00:00:02.770 --> 00:00:04.270 align:middle line:84% JUDY HOYT: Go ahead and get started. 00:00:04.270 --> 00:00:06.270 align:middle line:84% We want to start with looking at the schedule, 00:00:06.270 --> 00:00:10.260 align:middle line:90% get ourselves reoriented. 00:00:10.260 --> 00:00:13.950 align:middle line:90% This is November 16. 00:00:13.950 --> 00:00:17.880 align:middle line:84% And we'll have our final lecture today on chapter 9. 00:00:17.880 --> 00:00:19.500 align:middle line:90% Chapter 9 has three lectures. 00:00:19.500 --> 00:00:24.360 align:middle line:84% We'll talk about modeling the topography of deposition. 00:00:24.360 --> 00:00:26.940 align:middle line:84% And you're working on-- hopefully everybody's working 00:00:26.940 --> 00:00:30.120 align:middle line:90% on homework number five. 00:00:30.120 --> 00:00:34.560 align:middle line:84% That homework is due next lecture on the 18th. 00:00:34.560 --> 00:00:37.410 align:middle line:84% And does anybody have any questions on the homework? 00:00:37.410 --> 00:00:39.570 align:middle line:84% Or if you do, make sure you see me after class 00:00:39.570 --> 00:00:42.320 align:middle line:90% or send me an email. 00:00:42.320 --> 00:00:44.870 align:middle line:84% In addition to the homework, everyone 00:00:44.870 --> 00:00:48.740 align:middle line:84% has now-- except one person on the clipboard-- has a topic. 00:00:48.740 --> 00:00:50.750 align:middle line:84% And I had my assistant type it up. 00:00:50.750 --> 00:00:51.840 align:middle line:90% I'll put it on the web. 00:00:51.840 --> 00:00:54.080 align:middle line:84% Everybody's signed up for a final report topic. 00:00:54.080 --> 00:00:57.830 align:middle line:84% And you can verify that she got it right. 00:00:57.830 --> 00:01:02.002 align:middle line:84% And then if you're doing an oral presentation, 00:01:02.002 --> 00:01:03.710 align:middle line:84% this week I'll start making up a schedule 00:01:03.710 --> 00:01:08.180 align:middle line:84% so you know which day you're going to speak. 00:01:08.180 --> 00:01:09.890 align:middle line:90% OK. 00:01:09.890 --> 00:01:13.970 align:middle line:84% So hopefully everybody recovered from the snowstorm 00:01:13.970 --> 00:01:15.170 align:middle line:90% we had over the weekend. 00:01:15.170 --> 00:01:19.130 align:middle line:84% And we'll go ahead and start this lecture 00:01:19.130 --> 00:01:22.200 align:middle line:90% with handout number 32. 00:01:22.200 --> 00:01:25.440 align:middle line:84% As I mentioned, this is the final lecture on chapter 9. 00:01:25.440 --> 00:01:29.190 align:middle line:84% So far we've by way of thin film deposition 00:01:29.190 --> 00:01:32.610 align:middle line:84% and epitaxy we've talked at least a little bit 00:01:32.610 --> 00:01:34.810 align:middle line:90% about silicon epitaxial growth. 00:01:34.810 --> 00:01:38.640 align:middle line:84% We discussed the low pressure chemical vapor deposition 00:01:38.640 --> 00:01:42.180 align:middle line:84% of the three most important films in front end processing. 00:01:42.180 --> 00:01:45.420 align:middle line:84% That's polysilicon, oxide, low temperature oxide, 00:01:45.420 --> 00:01:47.020 align:middle line:90% and silicon nitride. 00:01:47.020 --> 00:01:51.540 align:middle line:84% And we introduced some PVD techniques such as sputtering. 00:01:51.540 --> 00:01:55.080 align:middle line:84% We talked about systems for DC and RF sputtering, 00:01:55.080 --> 00:02:00.185 align:middle line:84% which is the most commonly used PVD technique in modern fabs. 00:02:00.185 --> 00:02:01.560 align:middle line:84% The nice thing about it is it can 00:02:01.560 --> 00:02:04.000 align:middle line:84% be adjusted to give reasonably good step coverage. 00:02:04.000 --> 00:02:05.490 align:middle line:84% This is in contrast to evaporation. 00:02:05.490 --> 00:02:06.990 align:middle line:84% And thermal evaporation is typically 00:02:06.990 --> 00:02:10.789 align:middle line:90% not used in CMOS processing. 00:02:10.789 --> 00:02:12.180 align:middle line:84% Today, what I want to talk about, 00:02:12.180 --> 00:02:15.150 align:middle line:84% though, is modeling of these processes. 00:02:15.150 --> 00:02:17.850 align:middle line:84% And you'll see how these models are a little bit different. 00:02:17.850 --> 00:02:20.120 align:middle line:84% They're much more quasi-empirical than the models 00:02:20.120 --> 00:02:22.380 align:middle line:90% we had for ion implantation. 00:02:22.380 --> 00:02:24.770 align:middle line:84% But there are methods to do modeling of these processes. 00:02:24.770 --> 00:02:29.590 align:middle line:84% And we'll give a couple of specific examples. 00:02:29.590 --> 00:02:33.910 align:middle line:84% OK, let's go on to slide number 2. 00:02:33.910 --> 00:02:37.490 align:middle line:84% Within the last 10 to 15 years ago or so, 00:02:37.490 --> 00:02:40.180 align:middle line:84% there have been a number of simulation tools 00:02:40.180 --> 00:02:42.910 align:middle line:84% have been developed for topography simulation. 00:02:42.910 --> 00:02:46.000 align:middle line:84% Now, I want to make clear, topography simulation 00:02:46.000 --> 00:02:47.580 align:middle line:90% is not what SUPREM-IV does. 00:02:47.580 --> 00:02:51.610 align:middle line:84% You've all been using SUPREM-IV to model ion implantation, 00:02:51.610 --> 00:02:54.730 align:middle line:84% oxidation, diffusion, a lot of the front end processing. 00:02:54.730 --> 00:02:57.250 align:middle line:84% SUPREM-IV doesn't have any topography simulation 00:02:57.250 --> 00:02:58.900 align:middle line:90% of any sort. 00:02:58.900 --> 00:03:00.760 align:middle line:84% If you tell it to make an etch, it 00:03:00.760 --> 00:03:02.980 align:middle line:84% assumes it's etched at a certain angle. 00:03:02.980 --> 00:03:04.270 align:middle line:90% You tell it the angle. 00:03:04.270 --> 00:03:07.330 align:middle line:84% It doesn't make any effort to calculate those angles. 00:03:07.330 --> 00:03:09.850 align:middle line:84% Or if you tell it that something is deposited, 00:03:09.850 --> 00:03:11.830 align:middle line:84% it assumes conformable deposition 00:03:11.830 --> 00:03:12.890 align:middle line:90% unless you say otherwise. 00:03:12.890 --> 00:03:15.760 align:middle line:84% So it's not a simulator for topography. 00:03:15.760 --> 00:03:17.020 align:middle line:90% There are other simulators. 00:03:17.020 --> 00:03:20.140 align:middle line:84% And one of the ones that also came out of Stanford 00:03:20.140 --> 00:03:20.980 align:middle line:90% is called SPEEDIE. 00:03:20.980 --> 00:03:23.020 align:middle line:84% And there are some examples in this lecture 00:03:23.020 --> 00:03:27.340 align:middle line:84% from the SPEEDIE simulator and from in your textbook. 00:03:27.340 --> 00:03:30.730 align:middle line:84% OK, so what is it that we need to think about if we want 00:03:30.730 --> 00:03:32.110 align:middle line:90% to do topography simulation? 00:03:32.110 --> 00:03:35.920 align:middle line:84% Let's say we are trying to model the topography of deposition 00:03:35.920 --> 00:03:36.730 align:middle line:90% of some film. 00:03:36.730 --> 00:03:38.860 align:middle line:84% This particular example on slide 2 00:03:38.860 --> 00:03:42.310 align:middle line:84% is shown assuming we're doing deposition of aluminum, 00:03:42.310 --> 00:03:45.050 align:middle line:90% say, in a sputtering chamber. 00:03:45.050 --> 00:03:49.030 align:middle line:84% And this is the surface that we're 00:03:49.030 --> 00:03:51.760 align:middle line:84% depositing the aluminum, for instance, onto. 00:03:51.760 --> 00:03:57.010 align:middle line:84% That surface has some starting initial topography. 00:03:57.010 --> 00:03:59.800 align:middle line:84% Well, this is a generalized picture 00:03:59.800 --> 00:04:01.630 align:middle line:84% of what's involved in the deposition. 00:04:01.630 --> 00:04:04.240 align:middle line:84% And you'll see that modeling topography ends up 00:04:04.240 --> 00:04:09.280 align:middle line:84% being a lot of bookkeeping of the different fluxes involved. 00:04:09.280 --> 00:04:12.760 align:middle line:84% In general, we are not going to consider gas phase boundary 00:04:12.760 --> 00:04:15.130 align:middle line:90% layer in these simple models. 00:04:15.130 --> 00:04:17.649 align:middle line:84% So this picture will not be accurate, 00:04:17.649 --> 00:04:19.390 align:middle line:84% for example, atmospheric pressure 00:04:19.390 --> 00:04:21.730 align:middle line:84% CVD where diffusion through a boundary layer 00:04:21.730 --> 00:04:22.880 align:middle line:90% is very important. 00:04:22.880 --> 00:04:25.690 align:middle line:84% We're not taking that into account. 00:04:25.690 --> 00:04:30.880 align:middle line:84% But it will be reasonably useful for modeling low pressure CVD 00:04:30.880 --> 00:04:34.060 align:middle line:84% or for modeling sputtering, which also takes place 00:04:34.060 --> 00:04:35.390 align:middle line:90% at reasonably low pressures. 00:04:35.390 --> 00:04:37.463 align:middle line:84% So these are the fluxes involved. 00:04:37.463 --> 00:04:39.880 align:middle line:84% For example, if we're depositing aluminum on this surface, 00:04:39.880 --> 00:04:44.570 align:middle line:84% we have a direct flux to the surface of neutral aluminum 00:04:44.570 --> 00:04:45.070 align:middle line:90% atoms. 00:04:45.070 --> 00:04:47.140 align:middle line:84% We also have a direct flux of ions 00:04:47.140 --> 00:04:49.240 align:middle line:90% that we need to keep track of. 00:04:49.240 --> 00:04:51.730 align:middle line:84% Aluminum can come down and it can actually-- 00:04:51.730 --> 00:04:52.900 align:middle line:90% we can have a desorb. 00:04:52.900 --> 00:04:54.700 align:middle line:84% It can desorb aluminum that had already 00:04:54.700 --> 00:04:56.260 align:middle line:90% been deposited at a point. 00:04:56.260 --> 00:04:58.640 align:middle line:84% And then we have a desorbed or emitted flux 00:04:58.640 --> 00:05:01.530 align:middle line:84% that can be redeposited elsewhere on the surface. 00:05:01.530 --> 00:05:04.280 align:middle line:84% So that's a flux we need to take account of. 00:05:04.280 --> 00:05:08.210 align:middle line:84% Argon ions, because again this is in the sputtering system, 00:05:08.210 --> 00:05:11.390 align:middle line:84% argon is nominally supposed to be directed towards the target, 00:05:11.390 --> 00:05:12.388 align:middle line:90% not towards the wafer. 00:05:12.388 --> 00:05:14.180 align:middle line:84% But it's possible that it would be directed 00:05:14.180 --> 00:05:16.010 align:middle line:84% towards the wafer, argon ion, some of them, 00:05:16.010 --> 00:05:18.385 align:middle line:84% and that they would sputter off aluminum that had already 00:05:18.385 --> 00:05:19.100 align:middle line:90% been deposited. 00:05:19.100 --> 00:05:21.240 align:middle line:84% And then that can redeposit elsewhere. 00:05:21.240 --> 00:05:22.880 align:middle line:84% And there's a surface diffusion flux. 00:05:22.880 --> 00:05:25.790 align:middle line:84% Aluminum can diffuse along the surface 00:05:25.790 --> 00:05:30.000 align:middle line:84% a certain distance and then deposit at a different point. 00:05:30.000 --> 00:05:33.320 align:middle line:84% So we end up just doing a lot of bookkeeping 00:05:33.320 --> 00:05:34.520 align:middle line:90% on all the different fluxes. 00:05:34.520 --> 00:05:37.010 align:middle line:84% And it'd be good if you think about this picture 00:05:37.010 --> 00:05:39.260 align:middle line:84% because we're going to use the exact same picture when 00:05:39.260 --> 00:05:40.177 align:middle line:90% we talk about etching. 00:05:40.177 --> 00:05:43.518 align:middle line:84% When you think about it in this terminology, 00:05:43.518 --> 00:05:45.060 align:middle line:84% etching is sort of just the opposite. 00:05:45.060 --> 00:05:48.410 align:middle line:84% Instead of depositing, we're talking about fluxes 00:05:48.410 --> 00:05:50.923 align:middle line:84% coming down and reacting and removing films. 00:05:50.923 --> 00:05:52.340 align:middle line:84% But it's not that much difference. 00:05:52.340 --> 00:05:56.700 align:middle line:84% It's mathematically the inverse of deposition. 00:05:56.700 --> 00:05:59.250 align:middle line:84% So in order to simulate that process that I just 00:05:59.250 --> 00:06:04.350 align:middle line:84% showed on slide 3 here, we need some kind 00:06:04.350 --> 00:06:07.170 align:middle line:84% of mathematical description of all of these fluxes. 00:06:07.170 --> 00:06:09.810 align:middle line:84% And then we need to add them all up 00:06:09.810 --> 00:06:12.580 align:middle line:84% and see what we get at any given point. 00:06:12.580 --> 00:06:14.610 align:middle line:84% So what we write is this equation shown 00:06:14.610 --> 00:06:18.690 align:middle line:84% at the top of slide 3 that we say the net flux at any point 00:06:18.690 --> 00:06:19.230 align:middle line:90% I. 00:06:19.230 --> 00:06:22.590 align:middle line:84% So point I, point I is just-- if I go back one, 00:06:22.590 --> 00:06:24.808 align:middle line:84% point I is just at some point on the surface. 00:06:24.808 --> 00:06:26.350 align:middle line:84% It could be at this point right here. 00:06:26.350 --> 00:06:27.933 align:middle line:84% It could be at that point right there. 00:06:27.933 --> 00:06:30.210 align:middle line:84% It's just at some point on the surface. 00:06:30.210 --> 00:06:33.180 align:middle line:84% The net flux is, well, it's the flux 00:06:33.180 --> 00:06:35.550 align:middle line:84% deposited at that point minus everything 00:06:35.550 --> 00:06:36.757 align:middle line:90% that was emitted, right? 00:06:36.757 --> 00:06:39.090 align:middle line:84% And that gives you the total atoms per square centimeter 00:06:39.090 --> 00:06:39.970 align:middle line:90% per unit time. 00:06:39.970 --> 00:06:44.070 align:middle line:84% So you add up the direct flux of neutrals, the direct flux 00:06:44.070 --> 00:06:48.570 align:middle line:84% of ions, any flux that's redeposited that's coming off 00:06:48.570 --> 00:06:51.128 align:middle line:84% from another area, any flux that's 00:06:51.128 --> 00:06:52.170 align:middle line:90% due to surface diffusion. 00:06:52.170 --> 00:06:53.880 align:middle line:90% All of those are positive terms. 00:06:53.880 --> 00:06:57.810 align:middle line:84% You subtract off any flux that's emitted from that point, 00:06:57.810 --> 00:07:01.170 align:middle line:84% any flux that diffuses out or that sputter from that point 00:07:01.170 --> 00:07:04.230 align:middle line:84% by virtue of argon ions coming down or something. 00:07:04.230 --> 00:07:07.195 align:middle line:84% So you can see, this is going to amount to basically 00:07:07.195 --> 00:07:08.070 align:middle line:90% a lot of bookkeeping. 00:07:08.070 --> 00:07:09.270 align:middle line:84% And this is something that computers 00:07:09.270 --> 00:07:10.260 align:middle line:90% are very good at doing. 00:07:10.260 --> 00:07:12.610 align:middle line:84% And you can and you can model those fluxes 00:07:12.610 --> 00:07:15.150 align:middle line:90% across the surface. 00:07:15.150 --> 00:07:18.920 align:middle line:84% So usually, what we need to do for a specific system-- 00:07:18.920 --> 00:07:21.580 align:middle line:84% let' say you're doing LPCVD or something-- 00:07:21.580 --> 00:07:24.080 align:middle line:84% we have to figure out which one of these fluxes is important 00:07:24.080 --> 00:07:26.840 align:middle line:84% and ignore all the others so we can simplify 00:07:26.840 --> 00:07:29.960 align:middle line:90% the models to a certain extent. 00:07:29.960 --> 00:07:32.840 align:middle line:84% And this picture gives you a rough idea 00:07:32.840 --> 00:07:36.770 align:middle line:90% of how we do the modeling. 00:07:36.770 --> 00:07:41.300 align:middle line:84% What we consider is we draw this dashed line here right 00:07:41.300 --> 00:07:42.990 align:middle line:90% above the surface of the wafer. 00:07:42.990 --> 00:07:46.940 align:middle line:84% And so it's a plane just above the surface. 00:07:46.940 --> 00:07:48.560 align:middle line:84% And the first thing we need to look at 00:07:48.560 --> 00:07:52.400 align:middle line:84% is what's the arrival angle distribution of atoms and ions 00:07:52.400 --> 00:07:54.480 align:middle line:84% that are coming down on that plane? 00:07:54.480 --> 00:07:56.270 align:middle line:84% And then later on, the next step will 00:07:56.270 --> 00:07:57.847 align:middle line:90% take into account topography. 00:07:57.847 --> 00:07:59.930 align:middle line:84% So first, you do it assuming you have a plane just 00:07:59.930 --> 00:08:02.240 align:middle line:90% above the surface. 00:08:02.240 --> 00:08:04.010 align:middle line:84% And now what we need to take into account 00:08:04.010 --> 00:08:06.020 align:middle line:84% are these direct flux of neutrals 00:08:06.020 --> 00:08:08.180 align:middle line:90% and the direct flux of ions. 00:08:08.180 --> 00:08:10.670 align:middle line:84% These we generally model with some kind 00:08:10.670 --> 00:08:14.480 align:middle line:84% of arrival angle distribution just above the wafer. 00:08:14.480 --> 00:08:16.158 align:middle line:84% And I've noted here what it says is 00:08:16.158 --> 00:08:17.450 align:middle line:90% it doesn't model the equipment. 00:08:17.450 --> 00:08:20.660 align:middle line:84% What that means is you have to put in the arrival angle 00:08:20.660 --> 00:08:21.320 align:middle line:90% distribution. 00:08:21.320 --> 00:08:23.765 align:middle line:84% You can't say to the simulator, oh, yes, I 00:08:23.765 --> 00:08:26.180 align:middle line:84% have an applied materials sputtering tool in the endura, 00:08:26.180 --> 00:08:28.610 align:middle line:84% and now tell me what it's going to look like. 00:08:28.610 --> 00:08:31.160 align:middle line:84% It doesn't have that level of knowledge 00:08:31.160 --> 00:08:32.210 align:middle line:90% of the individual tools. 00:08:32.210 --> 00:08:35.850 align:middle line:84% What you do is you put in flux, the direct flux, 00:08:35.850 --> 00:08:38.460 align:middle line:84% as a function of the angle theta and that's 00:08:38.460 --> 00:08:39.960 align:middle line:84% measured with respect to the normal. 00:08:39.960 --> 00:08:41.490 align:middle line:90% And you give it usually-- 00:08:41.490 --> 00:08:46.620 align:middle line:84% it's some normal flux times a cosine theta to the n power. 00:08:46.620 --> 00:08:50.730 align:middle line:84% So that's a variable that you need to put into the simulator. 00:08:50.730 --> 00:08:53.500 align:middle line:84% But there are some guidelines for different systems, 00:08:53.500 --> 00:08:55.260 align:middle line:84% different types of systems, what kind 00:08:55.260 --> 00:08:58.770 align:middle line:84% of arrival angle distribution for neutrals and ions 00:08:58.770 --> 00:08:59.460 align:middle line:90% you might use. 00:08:59.460 --> 00:09:02.520 align:middle line:90% 00:09:02.520 --> 00:09:04.970 align:middle line:84% So let's look at a couple different types 00:09:04.970 --> 00:09:08.270 align:middle line:90% of flux distributions. 00:09:08.270 --> 00:09:11.540 align:middle line:84% The first one here shown on the slide number 4 00:09:11.540 --> 00:09:13.820 align:middle line:84% is what we call isotropic flux arrival. 00:09:13.820 --> 00:09:19.520 align:middle line:84% As the name suggests, the flux is essentially isotropic. 00:09:19.520 --> 00:09:23.810 align:middle line:84% And in this we use an n equals 1 in the formula of cosine theta 00:09:23.810 --> 00:09:25.070 align:middle line:90% to the n. 00:09:25.070 --> 00:09:28.370 align:middle line:84% So what we get is that the direct flux 00:09:28.370 --> 00:09:30.390 align:middle line:90% f as a function of angle theta. 00:09:30.390 --> 00:09:32.600 align:middle line:84% So if theta is measured with respect to the normal, 00:09:32.600 --> 00:09:35.900 align:middle line:84% is whatever the flux is in the normal direction f0. 00:09:35.900 --> 00:09:41.660 align:middle line:84% So it's this flux number here times cosine of theta to the n. 00:09:41.660 --> 00:09:44.210 align:middle line:84% So what's happening is I'm moving from normal. 00:09:44.210 --> 00:09:46.250 align:middle line:90% I have a certain flux in. 00:09:46.250 --> 00:09:48.410 align:middle line:84% And as I go this direction, as I coming in 00:09:48.410 --> 00:09:51.520 align:middle line:84% at shallower and shallower angles, 00:09:51.520 --> 00:09:55.010 align:middle line:84% the flux goes down according to this cosine theta. 00:09:55.010 --> 00:09:58.820 align:middle line:84% And this little derivation that I've shown mathematically, 00:09:58.820 --> 00:10:01.370 align:middle line:84% the second half of slide 4 kind of explains 00:10:01.370 --> 00:10:03.140 align:middle line:90% where that comes from. 00:10:03.140 --> 00:10:05.660 align:middle line:84% What we're interested in when we're looking at deposition, 00:10:05.660 --> 00:10:10.520 align:middle line:84% we're interested in the normal component of the flux, OK? 00:10:10.520 --> 00:10:12.710 align:middle line:84% So let's say this is my wafer down here. 00:10:12.710 --> 00:10:17.150 align:middle line:84% This solid line here represents the wafer or the surface 00:10:17.150 --> 00:10:18.570 align:middle line:90% that I'm depositing on. 00:10:18.570 --> 00:10:21.230 align:middle line:84% And if we look at this little red region, 00:10:21.230 --> 00:10:23.390 align:middle line:84% that represents a certain normal flux coming. 00:10:23.390 --> 00:10:25.580 align:middle line:90% And that would be my f super 0. 00:10:25.580 --> 00:10:28.760 align:middle line:84% And this, I'm drawing this for the n equals 1 case. 00:10:28.760 --> 00:10:32.360 align:middle line:84% So I have a certain flux coming in normal in normal. 00:10:32.360 --> 00:10:36.110 align:middle line:84% Now, at some other angle theta I have that same flux. 00:10:36.110 --> 00:10:38.280 align:middle line:84% You notice I drew those same number of arrows. 00:10:38.280 --> 00:10:43.340 align:middle line:84% So it's coming in this direction normal to this plane here, 00:10:43.340 --> 00:10:44.330 align:middle line:90% normal to this plane. 00:10:44.330 --> 00:10:47.870 align:middle line:84% I have the same number of atoms per square centimeter per unit 00:10:47.870 --> 00:10:49.200 align:middle line:90% time. 00:10:49.200 --> 00:10:53.430 align:middle line:84% But I need to project that now onto this surface. 00:10:53.430 --> 00:10:58.580 align:middle line:84% So when I project this distance here of length l 00:10:58.580 --> 00:11:01.790 align:middle line:84% onto the surface, I have this sort of distance-- 00:11:01.790 --> 00:11:06.910 align:middle line:84% it's spread out over a lateral dimension h. 00:11:06.910 --> 00:11:09.340 align:middle line:84% Now, we have right triangles here with angle theta 00:11:09.340 --> 00:11:12.160 align:middle line:90% here and theta here. 00:11:12.160 --> 00:11:13.690 align:middle line:84% And you can write-- it's very simply 00:11:13.690 --> 00:11:15.640 align:middle line:84% that the cosine of the angle theta 00:11:15.640 --> 00:11:20.180 align:middle line:84% is just this distance l divided by the distance h, 00:11:20.180 --> 00:11:22.870 align:middle line:84% where h represents how far that unit flux is 00:11:22.870 --> 00:11:27.240 align:middle line:90% spread across the wafer. 00:11:27.240 --> 00:11:28.860 align:middle line:84% OK, so now it's just bookkeeping. 00:11:28.860 --> 00:11:33.300 align:middle line:84% If I want to know what is the flux here in this region, 00:11:33.300 --> 00:11:35.700 align:middle line:84% it's just a number per unit area per unit time 00:11:35.700 --> 00:11:40.170 align:middle line:84% that's going to be proportional to the number per unit 00:11:40.170 --> 00:11:43.350 align:middle line:84% time divided by h, right, because it's 00:11:43.350 --> 00:11:45.210 align:middle line:90% spread out over the distance h. 00:11:45.210 --> 00:11:51.480 align:middle line:84% But you can substitute in for h is cosine theta over l 00:11:51.480 --> 00:11:54.370 align:middle line:84% or l over cosine theta, Right so you can substitute that in. 00:11:54.370 --> 00:11:57.120 align:middle line:84% So you get the fact that for isotropic, 00:11:57.120 --> 00:12:00.600 align:middle line:84% so assuming you have just as much intensity in this red bar 00:12:00.600 --> 00:12:05.790 align:middle line:84% as I had in this direction theta in the black bar, 00:12:05.790 --> 00:12:08.940 align:middle line:84% assuming that, then I get this simple formula, which 00:12:08.940 --> 00:12:10.980 align:middle line:84% shows that the flux as a function of theta 00:12:10.980 --> 00:12:15.150 align:middle line:84% is just the normal flux times the cosine of theta. 00:12:15.150 --> 00:12:18.690 align:middle line:84% And that's assuming that the source is isotropic like this. 00:12:18.690 --> 00:12:20.460 align:middle line:84% If for some reason your source varies, 00:12:20.460 --> 00:12:21.990 align:middle line:84% and you have much greater intensity 00:12:21.990 --> 00:12:23.910 align:middle line:84% over here and over here or whatever, 00:12:23.910 --> 00:12:26.040 align:middle line:84% this cosine theta term will be modified. 00:12:26.040 --> 00:12:32.340 align:middle line:84% But again, this is assuming isotropic, an isotropic source. 00:12:32.340 --> 00:12:35.773 align:middle line:84% And the cosine theta just comes from the geometry factor. 00:12:35.773 --> 00:12:38.190 align:middle line:84% OK, you can sit down and stare at that for a little while. 00:12:38.190 --> 00:12:41.900 align:middle line:84% Hopefully, it'll make sense to you. 00:12:41.900 --> 00:12:44.000 align:middle line:90% Go to slide 5. 00:12:44.000 --> 00:12:46.700 align:middle line:84% So in contrast to that, on the left side 00:12:46.700 --> 00:12:49.130 align:middle line:84% I show what I just derived, which is the isotropic flux 00:12:49.130 --> 00:12:50.660 align:middle line:90% arrival distribution. 00:12:50.660 --> 00:12:53.000 align:middle line:84% On the right side, for some systems 00:12:53.000 --> 00:12:55.340 align:middle line:84% you may have very anisotropic flux. 00:12:55.340 --> 00:12:57.410 align:middle line:84% Most of the atoms or ions coming down 00:12:57.410 --> 00:12:59.660 align:middle line:84% may be directed mostly normal to the surface. 00:12:59.660 --> 00:13:02.210 align:middle line:84% There may be very few coming out at shallow angles 00:13:02.210 --> 00:13:04.980 align:middle line:84% depending on the particular machine that you're using. 00:13:04.980 --> 00:13:07.400 align:middle line:84% So here in the cosine theta to the n, 00:13:07.400 --> 00:13:08.618 align:middle line:90% we make n greater than 1. 00:13:08.618 --> 00:13:10.160 align:middle line:84% And you can see just based on the way 00:13:10.160 --> 00:13:12.140 align:middle line:84% these arrows are distributed most of them 00:13:12.140 --> 00:13:13.160 align:middle line:90% are coming in normal. 00:13:13.160 --> 00:13:16.340 align:middle line:84% The flux coming in at angles, higher angles theta, 00:13:16.340 --> 00:13:17.213 align:middle line:90% is much smaller. 00:13:17.213 --> 00:13:19.130 align:middle line:84% You can see that from the length of the arrow. 00:13:19.130 --> 00:13:20.810 align:middle line:84% So the length of these arrows here 00:13:20.810 --> 00:13:26.320 align:middle line:84% is supposed to represent the flux at any given angle. 00:13:26.320 --> 00:13:28.680 align:middle line:84% So generally, we have this arrival angle distribution 00:13:28.680 --> 00:13:30.390 align:middle line:90% at a plane just above the wafer. 00:13:30.390 --> 00:13:34.530 align:middle line:84% It has some cosine theta to the n type of dependence. 00:13:34.530 --> 00:13:36.300 align:middle line:84% And again, it's the normal component, 00:13:36.300 --> 00:13:39.060 align:middle line:84% normal to the surface, that strikes the surface that 00:13:39.060 --> 00:13:40.485 align:middle line:90% determines the deposition rate. 00:13:40.485 --> 00:13:44.850 align:middle line:90% 00:13:44.850 --> 00:13:50.000 align:middle line:84% So let's take a look at examples of distribution of flux arrival 00:13:50.000 --> 00:13:52.410 align:middle line:90% distribution functions. 00:13:52.410 --> 00:13:55.220 align:middle line:84% Here's that formula I just mentioned. 00:13:55.220 --> 00:13:59.360 align:middle line:84% And for example, if you have a high pressure system, 00:13:59.360 --> 00:14:00.980 align:middle line:84% reasonably high pressure, you're going 00:14:00.980 --> 00:14:04.010 align:middle line:84% to have a lot of gas phase collisions, which means you 00:14:04.010 --> 00:14:05.420 align:middle line:90% have a short mean free path. 00:14:05.420 --> 00:14:07.730 align:middle line:84% And you're going to tend to get isotropic arrival 00:14:07.730 --> 00:14:09.590 align:middle line:84% because those atoms are all bouncing around. 00:14:09.590 --> 00:14:11.450 align:middle line:84% And regardless of where they're coming from, 00:14:11.450 --> 00:14:12.992 align:middle line:84% it's just sort of the flux is equally 00:14:12.992 --> 00:14:15.290 align:middle line:90% distributed for all angles. 00:14:15.290 --> 00:14:17.870 align:middle line:84% If you're in a very low pressure system 00:14:17.870 --> 00:14:21.920 align:middle line:84% where you have very fewer gas phase collisions, a long 00:14:21.920 --> 00:14:24.628 align:middle line:84% the mean free path, and is going to be closer to 1, 00:14:24.628 --> 00:14:26.420 align:middle line:84% you're going to get more like line of sight 00:14:26.420 --> 00:14:28.745 align:middle line:84% from the source, whatever that source happens to be. 00:14:28.745 --> 00:14:29.870 align:middle line:90% It could be a point source. 00:14:29.870 --> 00:14:31.820 align:middle line:84% It could be a plane if it's sputtering. 00:14:31.820 --> 00:14:35.600 align:middle line:84% But you're going to have n greater than 1. 00:14:35.600 --> 00:14:36.860 align:middle line:90% That's all assuming neutrals. 00:14:36.860 --> 00:14:39.553 align:middle line:84% If you have ionic species and the wafer is biased, 00:14:39.553 --> 00:14:41.720 align:middle line:84% imagine the wafer has some bias and you have an ion, 00:14:41.720 --> 00:14:44.690 align:middle line:84% the ion can be attracted to it by the electric field, 00:14:44.690 --> 00:14:45.680 align:middle line:90% OK, that's set up. 00:14:45.680 --> 00:14:47.510 align:middle line:84% In that case, for ionized species, 00:14:47.510 --> 00:14:50.750 align:middle line:84% it tends to be very anisotropic because, again, there's 00:14:50.750 --> 00:14:53.420 align:middle line:84% a force driving it towards the wafer 00:14:53.420 --> 00:14:56.570 align:middle line:84% so it doesn't tend to bounce around. 00:14:56.570 --> 00:14:58.670 align:middle line:84% The species tend to go right to the wafer. 00:14:58.670 --> 00:15:02.640 align:middle line:84% So these are just some examples of flux distribution 00:15:02.640 --> 00:15:07.180 align:middle line:84% so you can see what it looks like for different values of n 00:15:07.180 --> 00:15:08.960 align:middle line:90% So this is the relative flux. 00:15:08.960 --> 00:15:12.070 align:middle line:84% So it's the flux at any given angle 00:15:12.070 --> 00:15:16.570 align:middle line:84% theta divided by the normal flux, divided by f0. 00:15:16.570 --> 00:15:17.930 align:middle line:90% And this is what it looks like. 00:15:17.930 --> 00:15:20.080 align:middle line:84% So it starts here at theta equals 0. 00:15:20.080 --> 00:15:22.270 align:middle line:90% So that's normal is one. 00:15:22.270 --> 00:15:24.580 align:middle line:84% And you can see for n equals 1 it drops down 00:15:24.580 --> 00:15:26.020 align:middle line:90% and the flux finally goes to 0. 00:15:26.020 --> 00:15:28.270 align:middle line:84% At 90 degrees, you're just skimming along the surface. 00:15:28.270 --> 00:15:31.060 align:middle line:84% Obviously, the normal component is 0. 00:15:31.060 --> 00:15:35.095 align:middle line:84% n equals 3, which is slight anisotropy. 00:15:35.095 --> 00:15:36.315 align:middle line:90% It looks something like this. 00:15:36.315 --> 00:15:38.440 align:middle line:84% It's starting to flux really dies down dramatically 00:15:38.440 --> 00:15:39.790 align:middle line:90% after 60 degrees. 00:15:39.790 --> 00:15:44.500 align:middle line:84% n equals 15, which could be appropriate for some ionized 00:15:44.500 --> 00:15:46.990 align:middle line:84% deposition systems, you can see really the flux 00:15:46.990 --> 00:15:49.340 align:middle line:84% goes to 0 by the time you get to about 30 degrees. 00:15:49.340 --> 00:15:54.680 align:middle line:84% So it's very anisotropic arrival at that plane. 00:15:54.680 --> 00:15:58.487 align:middle line:84% And again, in most models, these are inputs that-- 00:15:58.487 --> 00:15:59.695 align:middle line:90% parameters that you can vary. 00:15:59.695 --> 00:16:03.200 align:middle line:90% 00:16:03.200 --> 00:16:05.930 align:middle line:84% OK, so that gives us an idea of the direct fluxes 00:16:05.930 --> 00:16:07.880 align:middle line:90% and their angular dependence. 00:16:07.880 --> 00:16:11.300 align:middle line:84% The next thing we have to do is to take into account 00:16:11.300 --> 00:16:13.040 align:middle line:90% the surface topography. 00:16:13.040 --> 00:16:15.560 align:middle line:84% And just to give you an example of that, 00:16:15.560 --> 00:16:18.060 align:middle line:84% I'm showing here-- imagine your surface is not flat, 00:16:18.060 --> 00:16:21.050 align:middle line:84% but that you have a wafer and you are sputtering aluminum 00:16:21.050 --> 00:16:24.260 align:middle line:84% into it and you have a via or you have a hole, contact hole. 00:16:24.260 --> 00:16:27.500 align:middle line:84% So it's like this, and then you have this hole right here. 00:16:27.500 --> 00:16:32.480 align:middle line:84% Well if you have imagined each one of these little regions 00:16:32.480 --> 00:16:35.900 align:middle line:84% corresponds to an emitter with a certain flux, 00:16:35.900 --> 00:16:40.190 align:middle line:84% depending at any point i on this vertical surface, 00:16:40.190 --> 00:16:42.770 align:middle line:84% it's going to have a certain viewing angle. 00:16:42.770 --> 00:16:44.750 align:middle line:84% And if you're beyond that viewing angle, 00:16:44.750 --> 00:16:46.040 align:middle line:90% you're going to be shadowed. 00:16:46.040 --> 00:16:49.940 align:middle line:84% So this is a shadowing effect on the sidewall of the trench. 00:16:49.940 --> 00:16:53.210 align:middle line:84% It's not that uncommon depending on the depth and the aspect 00:16:53.210 --> 00:16:56.532 align:middle line:90% ratio of the trench. 00:16:56.532 --> 00:16:58.490 align:middle line:84% So we have to take into account the orientation 00:16:58.490 --> 00:17:01.310 align:middle line:84% of the surface, what its angle is, this 00:17:01.310 --> 00:17:03.920 align:middle line:90% viewing angle, and shadowing. 00:17:03.920 --> 00:17:09.530 align:middle line:84% And typically, in the near certain way for surface region 00:17:09.530 --> 00:17:13.760 align:middle line:84% we will neglect the gas phase collisions on that. 00:17:13.760 --> 00:17:16.160 align:middle line:84% So if you will, you can think of this-- 00:17:16.160 --> 00:17:18.450 align:middle line:90% imagine my reference plane. 00:17:18.450 --> 00:17:21.230 align:middle line:84% Remember that reference plane is sort of a plane just 00:17:21.230 --> 00:17:23.579 align:middle line:90% above the wafer surface. 00:17:23.579 --> 00:17:27.260 align:middle line:84% And each small area on this reference plane 00:17:27.260 --> 00:17:29.730 align:middle line:84% can be treated as an emission source. 00:17:29.730 --> 00:17:32.390 align:middle line:84% So if you have a emission coming down, 00:17:32.390 --> 00:17:35.990 align:middle line:84% you have all these little emission sources. 00:17:35.990 --> 00:17:40.250 align:middle line:84% And each source is emitting to the topography below it. 00:17:40.250 --> 00:17:44.330 align:middle line:84% And then the total direct flux at any point i 00:17:44.330 --> 00:17:46.070 align:middle line:84% can be calculated by just adding up 00:17:46.070 --> 00:17:48.920 align:middle line:84% the flux from each one of these little miniature emission 00:17:48.920 --> 00:17:52.370 align:middle line:84% sources, which each one of which has its certain angular 00:17:52.370 --> 00:17:55.100 align:middle line:84% distribution depending on the type of system 00:17:55.100 --> 00:17:58.330 align:middle line:90% that you're modeling. 00:17:58.330 --> 00:18:00.940 align:middle line:84% So you can now begin to see how in a computer model 00:18:00.940 --> 00:18:03.640 align:middle line:84% you can build equations and you can 00:18:03.640 --> 00:18:06.130 align:middle line:84% build structures that would be able to model 00:18:06.130 --> 00:18:09.360 align:middle line:90% this type of deposition. 00:18:09.360 --> 00:18:10.490 align:middle line:90% Let's go on to slide 8. 00:18:10.490 --> 00:18:12.900 align:middle line:84% And I'd wanted to give you a practical example. 00:18:12.900 --> 00:18:14.983 align:middle line:84% We're talking about all these mathematical things. 00:18:14.983 --> 00:18:19.610 align:middle line:84% But a practical example, which should be somewhat intuitive, 00:18:19.610 --> 00:18:22.310 align:middle line:84% here on slide 8 is the impact of having a finite target 00:18:22.310 --> 00:18:25.940 align:middle line:84% size compared to the wafer and if you're 00:18:25.940 --> 00:18:30.040 align:middle line:84% doing sputter deposition on asymmetry across a wafer. 00:18:30.040 --> 00:18:31.540 align:middle line:84% So if you look at this illustration, 00:18:31.540 --> 00:18:34.240 align:middle line:84% imagine I have a target that I'm doing sputtering. 00:18:34.240 --> 00:18:35.470 align:middle line:90% And this could be aluminum. 00:18:35.470 --> 00:18:36.350 align:middle line:90% And it's up here. 00:18:36.350 --> 00:18:38.110 align:middle line:90% It has this dimension. 00:18:38.110 --> 00:18:40.270 align:middle line:84% It's relatively small compared to, say, 00:18:40.270 --> 00:18:43.240 align:middle line:90% I'm doing a 6 or 8 inch wafer. 00:18:43.240 --> 00:18:44.800 align:middle line:90% So what's going to happen? 00:18:44.800 --> 00:18:46.570 align:middle line:84% And you have the wafer sitting there, 00:18:46.570 --> 00:18:49.150 align:middle line:84% and you're sputtering into a trench on the left side 00:18:49.150 --> 00:18:51.430 align:middle line:84% of the wafer, and you have a dye over here 00:18:51.430 --> 00:18:52.780 align:middle line:90% on the right side of the wafer. 00:18:52.780 --> 00:18:55.360 align:middle line:84% And let's say for now you're not rotating 00:18:55.360 --> 00:18:56.587 align:middle line:90% the wafers or anything. 00:18:56.587 --> 00:18:59.170 align:middle line:84% So you can imagine what's going to happen here on this trench. 00:18:59.170 --> 00:19:02.030 align:middle line:84% This white region indicates the region that you've sputtered, 00:19:02.030 --> 00:19:03.280 align:middle line:90% the material you've sputtered. 00:19:03.280 --> 00:19:06.400 align:middle line:84% You get an asymmetry on the sidewall 00:19:06.400 --> 00:19:08.560 align:middle line:84% because you've got here at this point in the wafer 00:19:08.560 --> 00:19:11.623 align:middle line:84% you have more flux coming from this direction. 00:19:11.623 --> 00:19:12.790 align:middle line:90% And so you're going to get-- 00:19:12.790 --> 00:19:14.320 align:middle line:84% because of shadowing you're going to not get 00:19:14.320 --> 00:19:15.700 align:middle line:90% much deposition on this wall. 00:19:15.700 --> 00:19:16.923 align:middle line:90% You'll get more on this wall. 00:19:16.923 --> 00:19:18.340 align:middle line:84% On the right side of the wafer, we 00:19:18.340 --> 00:19:21.670 align:middle line:84% have just the opposite type of shadowing. 00:19:21.670 --> 00:19:24.630 align:middle line:84% So you will tend to get asymmetric deposition 00:19:24.630 --> 00:19:26.820 align:middle line:84% on the left side of the wafer as opposed 00:19:26.820 --> 00:19:29.590 align:middle line:84% to the right side of the wafer with this particular geometry. 00:19:29.590 --> 00:19:31.740 align:middle line:84% So what do people do in a real system? 00:19:31.740 --> 00:19:33.480 align:middle line:84% Because obviously this is not desirable. 00:19:33.480 --> 00:19:34.417 align:middle line:90% We want to have it-- 00:19:34.417 --> 00:19:36.250 align:middle line:84% each side of the wafer should look the same. 00:19:36.250 --> 00:19:38.220 align:middle line:84% Well you can widen the target size. 00:19:38.220 --> 00:19:40.500 align:middle line:84% So I can take this target and make it much bigger, 00:19:40.500 --> 00:19:43.690 align:middle line:90% make it larger than the wafer. 00:19:43.690 --> 00:19:45.610 align:middle line:84% Or you can take the same size target 00:19:45.610 --> 00:19:48.700 align:middle line:84% and move it further from the wafer. 00:19:48.700 --> 00:19:52.300 align:middle line:84% So in either case, you're going to get an arrival angle 00:19:52.300 --> 00:19:56.890 align:middle line:84% distribution that is more symmetric across the wafer. 00:19:56.890 --> 00:19:59.900 align:middle line:84% Another thing people sometimes do is to rotate the wafers. 00:19:59.900 --> 00:20:01.450 align:middle line:84% So you can put the wafers on a plate 00:20:01.450 --> 00:20:03.820 align:middle line:84% and that rotates with respect to the source, which 00:20:03.820 --> 00:20:06.160 align:middle line:84% is essentially similar to widening, increasing 00:20:06.160 --> 00:20:08.080 align:middle line:84% the effective size of your target. 00:20:08.080 --> 00:20:10.717 align:middle line:84% This is why sputtering systems are very large because they 00:20:10.717 --> 00:20:11.800 align:middle line:90% tend to have huge targets. 00:20:11.800 --> 00:20:14.380 align:middle line:84% Imagine today people were processing 6 inch wafers-- 00:20:14.380 --> 00:20:15.070 align:middle line:90% not 6 inch. 00:20:15.070 --> 00:20:16.780 align:middle line:84% Here at MTL we're processing 6 inch, 00:20:16.780 --> 00:20:18.340 align:middle line:90% 12 inch wafers in industry. 00:20:18.340 --> 00:20:21.040 align:middle line:84% That target has to be huge basically. 00:20:21.040 --> 00:20:22.790 align:middle line:84% And then you would be rotating the wafers. 00:20:22.790 --> 00:20:24.902 align:middle line:84% And so this is why the equipment in the fabs 00:20:24.902 --> 00:20:26.110 align:middle line:90% is getting larger and larger. 00:20:26.110 --> 00:20:29.020 align:middle line:84% You're fundamentally limited by some of these processes 00:20:29.020 --> 00:20:31.190 align:middle line:90% to obtain uniformity. 00:20:31.190 --> 00:20:33.050 align:middle line:84% And on slide 9, in fact, there's just 00:20:33.050 --> 00:20:35.780 align:middle line:84% an example of how we would ameliorate 00:20:35.780 --> 00:20:38.332 align:middle line:84% this asymmetry from the left and the right side of the wafer. 00:20:38.332 --> 00:20:40.040 align:middle line:84% Here's an example where the target's been 00:20:40.040 --> 00:20:41.360 align:middle line:90% made as large as the wafer. 00:20:41.360 --> 00:20:43.700 align:middle line:84% In fact, most targets today are much bigger 00:20:43.700 --> 00:20:44.903 align:middle line:90% than the size of the wafer. 00:20:44.903 --> 00:20:46.820 align:middle line:84% And this is also why they cost a lot of money. 00:20:46.820 --> 00:20:48.080 align:middle line:84% If you have a target of platinum, 00:20:48.080 --> 00:20:49.455 align:middle line:84% if you're sputtering platinum, it 00:20:49.455 --> 00:20:53.060 align:middle line:84% has to be 2 feet by 10 inches or something like. 00:20:53.060 --> 00:20:54.140 align:middle line:90% That's a lot of platinum. 00:20:54.140 --> 00:20:55.920 align:middle line:84% That's a very large sheet of platinum. 00:20:55.920 --> 00:20:57.295 align:middle line:84% So it costs quite a bit of money. 00:20:57.295 --> 00:20:59.685 align:middle line:84% Or if you want, instead of making a target so big 00:20:59.685 --> 00:21:01.310 align:middle line:84% you can move it further away and you'll 00:21:01.310 --> 00:21:03.367 align:middle line:84% get a little better symmetry between the left 00:21:03.367 --> 00:21:04.700 align:middle line:90% and the right side of the wafer. 00:21:04.700 --> 00:21:07.250 align:middle line:84% But that makes the machine really big. 00:21:07.250 --> 00:21:09.590 align:middle line:84% And you can only have so much size in your clean room. 00:21:09.590 --> 00:21:11.340 align:middle line:90% So they kind of trade offs here. 00:21:11.340 --> 00:21:13.730 align:middle line:84% So in practice, people use larger targets 00:21:13.730 --> 00:21:17.150 align:middle line:84% at some reasonable distance, and scale the tool size, 00:21:17.150 --> 00:21:21.413 align:middle line:90% and rotate the wafers to help. 00:21:21.413 --> 00:21:22.580 align:middle line:90% There's a practical example. 00:21:22.580 --> 00:21:26.920 align:middle line:84% OK, so we talked about how to calculate the direct flux. 00:21:26.920 --> 00:21:29.420 align:middle line:84% Basically, you're just taking-- it's like a geometry problem 00:21:29.420 --> 00:21:30.080 align:middle line:90% essentially. 00:21:30.080 --> 00:21:32.840 align:middle line:84% And you have to have a cosine theta to the end dependence. 00:21:32.840 --> 00:21:34.730 align:middle line:90% How about the indirect flux? 00:21:34.730 --> 00:21:37.220 align:middle line:84% First of all, on slide 10, what do I mean by indirect? 00:21:37.220 --> 00:21:40.370 align:middle line:84% Well, I mean all those processes that occur on the surface 00:21:40.370 --> 00:21:43.640 align:middle line:84% or near the surface of the wafer during the deposition. 00:21:43.640 --> 00:21:47.400 align:middle line:84% So these indirect fluxes include surface diffusion. 00:21:47.400 --> 00:21:51.650 align:middle line:84% So I could have an aluminum atom that's that started here. 00:21:51.650 --> 00:21:53.210 align:middle line:84% But it can diffuse along the surface 00:21:53.210 --> 00:21:54.720 align:middle line:90% and end up somewhere else. 00:21:54.720 --> 00:21:57.650 align:middle line:84% So that represents a flux to that point. 00:21:57.650 --> 00:22:01.570 align:middle line:84% Emission, species come down, but they don't always stick. 00:22:01.570 --> 00:22:03.910 align:middle line:84% They don't always stick where they land. 00:22:03.910 --> 00:22:07.110 align:middle line:84% They might actually go there and then be re-emitted. 00:22:07.110 --> 00:22:10.320 align:middle line:84% So that's an indirect flux called re-emission. 00:22:10.320 --> 00:22:12.750 align:middle line:84% Sputtering, incoming atoms can knock off-- 00:22:12.750 --> 00:22:15.240 align:middle line:84% you could have a film of aluminum already on the wafer, 00:22:15.240 --> 00:22:17.010 align:middle line:84% but incoming atoms could knock them off. 00:22:17.010 --> 00:22:20.370 align:middle line:84% And redeposition, sputtered atoms 00:22:20.370 --> 00:22:22.090 align:middle line:90% can be redeposited elsewhere. 00:22:22.090 --> 00:22:23.110 align:middle line:90% Here's an example. 00:22:23.110 --> 00:22:24.330 align:middle line:90% Aluminum comes down. 00:22:24.330 --> 00:22:26.160 align:middle line:84% It hits an aluminum atom that was already 00:22:26.160 --> 00:22:27.450 align:middle line:90% deposited on the surface. 00:22:27.450 --> 00:22:33.155 align:middle line:84% It's then desorbed or emitted and redeposited down here. 00:22:33.155 --> 00:22:34.530 align:middle line:84% So there's a number of processes. 00:22:34.530 --> 00:22:36.655 align:middle line:84% We call these indirect because they're not directly 00:22:36.655 --> 00:22:38.458 align:middle line:90% coming from the source. 00:22:38.458 --> 00:22:40.500 align:middle line:84% They're coming from other regions on the surface, 00:22:40.500 --> 00:22:44.945 align:middle line:84% but that need to be taken into account in the topography. 00:22:44.945 --> 00:22:46.820 align:middle line:84% So let's go through each one of these surface 00:22:46.820 --> 00:22:50.450 align:middle line:84% processes, the diffusion, and the redeposition, and things, 00:22:50.450 --> 00:22:53.330 align:middle line:84% and talk about how we might model them. 00:22:53.330 --> 00:22:57.860 align:middle line:84% If we go on to slide 11, to surface diffusion, 00:22:57.860 --> 00:23:00.467 align:middle line:84% well, surface diffusion-- and I'll refer to your text. 00:23:00.467 --> 00:23:02.050 align:middle line:84% I'm not going to derive this equation. 00:23:02.050 --> 00:23:04.583 align:middle line:84% But it turns out diffusion along a surface 00:23:04.583 --> 00:23:06.250 align:middle line:84% is different from diffusion in the bulk. 00:23:06.250 --> 00:23:09.220 align:middle line:84% We spent a lot of time talking about diffusion in the bulk 00:23:09.220 --> 00:23:12.070 align:middle line:90% of single crystal silicon. 00:23:12.070 --> 00:23:15.700 align:middle line:84% Surface diffusion has slightly different driving forces. 00:23:15.700 --> 00:23:18.310 align:middle line:84% Not always just driven by a gradient, 00:23:18.310 --> 00:23:20.920 align:middle line:84% but it can actually be driven by the shape of the surface. 00:23:20.920 --> 00:23:22.390 align:middle line:84% It can actually be-- diffusion can 00:23:22.390 --> 00:23:25.138 align:middle line:84% be driven by the local curvature of the surface, 00:23:25.138 --> 00:23:26.680 align:middle line:84% if you're depositing a film, in order 00:23:26.680 --> 00:23:29.210 align:middle line:90% to minimize surface free energy. 00:23:29.210 --> 00:23:31.570 align:middle line:84% So we have a different type of equation 00:23:31.570 --> 00:23:34.600 align:middle line:84% from what you're used to seeing perhaps for diffusion. 00:23:34.600 --> 00:23:39.070 align:middle line:84% We see that if there's a flux into a point on the surface, 00:23:39.070 --> 00:23:42.430 align:middle line:84% minus the diffusion flux out, so that's the net diffusion flux 00:23:42.430 --> 00:23:46.630 align:middle line:84% to a point, it actually depends on a surface diffusion 00:23:46.630 --> 00:23:49.720 align:middle line:84% coefficient d sub s, diffusivity. 00:23:49.720 --> 00:23:53.667 align:middle line:84% But it also depends on this curvature, this derivative, 00:23:53.667 --> 00:23:55.750 align:middle line:84% which has to do with the curvature of the surface. 00:23:55.750 --> 00:23:57.178 align:middle line:90% And what I would do-- 00:23:57.178 --> 00:23:59.220 align:middle line:84% I don't want to go through it in detail in class. 00:23:59.220 --> 00:24:04.760 align:middle line:84% But if you look at equation 11.37 in the associated texts, 00:24:04.760 --> 00:24:07.190 align:middle line:84% in chapter 11, they talk a little bit more 00:24:07.190 --> 00:24:11.310 align:middle line:84% about surface diffusion and where this equation comes from. 00:24:11.310 --> 00:24:15.210 align:middle line:84% But the main point is because it increases with curvature 00:24:15.210 --> 00:24:19.140 align:middle line:84% during deposition, filling in of corners may be enhanced. 00:24:19.140 --> 00:24:23.340 align:middle line:84% And you can sometimes get more smoother, more conformal 00:24:23.340 --> 00:24:23.850 align:middle line:90% deposition. 00:24:23.850 --> 00:24:26.100 align:middle line:84% Of course, you have to be at a high enough temperature 00:24:26.100 --> 00:24:27.480 align:middle line:90% that these species can diffuse. 00:24:27.480 --> 00:24:30.000 align:middle line:84% And depending on the material and the temperature, 00:24:30.000 --> 00:24:31.710 align:middle line:90% it may not be diffusing. 00:24:31.710 --> 00:24:34.920 align:middle line:84% You may have to raise the wafer temperature slightly 00:24:34.920 --> 00:24:37.690 align:middle line:84% if you want to get more surface diffusion. 00:24:37.690 --> 00:24:39.310 align:middle line:90% So you can think of it this way. 00:24:39.310 --> 00:24:41.580 align:middle line:84% If I raise the temperature of the wafer, 00:24:41.580 --> 00:24:43.980 align:middle line:84% I can smooth out and get a little more conformal 00:24:43.980 --> 00:24:48.640 align:middle line:84% deposition by virtue of surface diffusion processes. 00:24:48.640 --> 00:24:50.410 align:middle line:84% So that's something that can be modeled 00:24:50.410 --> 00:24:52.930 align:middle line:90% in topography simulators. 00:24:52.930 --> 00:24:55.700 align:middle line:90% 00:24:55.700 --> 00:24:58.820 align:middle line:84% Page 12, a little bit simpler, this is very much 00:24:58.820 --> 00:25:00.350 align:middle line:90% a bookkeeping thing. 00:25:00.350 --> 00:25:05.240 align:middle line:84% There's something called surface sticking and emission 00:25:05.240 --> 00:25:07.650 align:middle line:90% and a sticking coefficient. 00:25:07.650 --> 00:25:09.830 align:middle line:84% So there's this flux at any given point 00:25:09.830 --> 00:25:14.030 align:middle line:84% i that's emitted that arises because not all molecules stick 00:25:14.030 --> 00:25:15.900 align:middle line:90% when they arrive at the surface. 00:25:15.900 --> 00:25:19.820 align:middle line:84% And so in fact, we write the emitted flux from any point 00:25:19.820 --> 00:25:23.510 align:middle line:84% as being equal to 1 minus s sub c. 00:25:23.510 --> 00:25:26.450 align:middle line:84% So the fraction emitted is 1 minus s sub c, 00:25:26.450 --> 00:25:28.490 align:middle line:84% where s sub c is called a sticking coefficient, 00:25:28.490 --> 00:25:31.180 align:middle line:90% times the flux going in. 00:25:31.180 --> 00:25:32.770 align:middle line:90% So what is s sub c. 00:25:32.770 --> 00:25:34.610 align:middle line:90% Well, it's a very simple thing. 00:25:34.610 --> 00:25:36.880 align:middle line:84% It's the fraction that stick basically. 00:25:36.880 --> 00:25:39.580 align:middle line:84% So sticking coefficient here, s sub c, 00:25:39.580 --> 00:25:41.740 align:middle line:84% it's the flux that react and stay 00:25:41.740 --> 00:25:45.230 align:middle line:84% put divided by the total incident flux. 00:25:45.230 --> 00:25:50.610 align:middle line:84% So to give yourself an idea, if you have a high s sub c-- 00:25:50.610 --> 00:25:52.630 align:middle line:84% s sub c is a number between 0 and 1 00:25:52.630 --> 00:25:54.040 align:middle line:90% really because it's a fraction. 00:25:54.040 --> 00:25:56.440 align:middle line:84% If you have a high s sub c, like s sub c equals 1, 00:25:56.440 --> 00:25:58.390 align:middle line:84% basically you have line of sight deposition. 00:25:58.390 --> 00:26:00.760 align:middle line:84% Everything that comes down just stops and sticks 00:26:00.760 --> 00:26:01.930 align:middle line:90% on the surface. 00:26:01.930 --> 00:26:04.030 align:middle line:84% If you have a type of deposition with 00:26:04.030 --> 00:26:07.760 align:middle line:84% a low sticking coefficient, what happens is atoms come down. 00:26:07.760 --> 00:26:10.450 align:middle line:84% They may touch the surface, but they don't stick. 00:26:10.450 --> 00:26:11.290 align:middle line:90% They don't react. 00:26:11.290 --> 00:26:14.590 align:middle line:84% They're kind of inert, in which case they'll be re-emitted. 00:26:14.590 --> 00:26:17.530 align:middle line:84% They touch somewhere else, and then they don't react. 00:26:17.530 --> 00:26:19.660 align:middle line:84% And then finally they might be emitted and stuck 00:26:19.660 --> 00:26:20.510 align:middle line:90% at this point. 00:26:20.510 --> 00:26:23.210 align:middle line:84% So when you have a low sticking coefficient, much, 00:26:23.210 --> 00:26:26.440 align:middle line:84% much less than 1-- and sticking coefficients can be 0.001-- 00:26:26.440 --> 00:26:29.260 align:middle line:84% it means that things bounce around a lot before they 00:26:29.260 --> 00:26:31.670 align:middle line:90% actually stick on the surface. 00:26:31.670 --> 00:26:35.060 align:middle line:84% So you're going to have more conformal type deposition. 00:26:35.060 --> 00:26:38.200 align:middle line:84% So this depends a lot on the deposition system, 00:26:38.200 --> 00:26:39.850 align:middle line:84% on the temperature, on the chemistry. 00:26:39.850 --> 00:26:41.230 align:middle line:90% Again, we don't model all that. 00:26:41.230 --> 00:26:44.110 align:middle line:84% We don't say, oh, I'm putting a wafer into an LPCVD system. 00:26:44.110 --> 00:26:48.220 align:middle line:84% The sticking coefficient must be 0.01 or 0.1. 00:26:48.220 --> 00:26:49.810 align:middle line:84% We don't know that for any given tool. 00:26:49.810 --> 00:26:51.700 align:middle line:84% You can put it in as a variable in the model 00:26:51.700 --> 00:26:54.020 align:middle line:84% and then you can compare that to what you get. 00:26:54.020 --> 00:26:56.028 align:middle line:84% You can compare the modeled deposition 00:26:56.028 --> 00:26:57.320 align:middle line:90% to the profile of what you get. 00:26:57.320 --> 00:26:59.903 align:middle line:84% And then you get an idea of what your sticking coefficient is. 00:26:59.903 --> 00:27:01.660 align:middle line:84% It's not a number you can calculate 00:27:01.660 --> 00:27:05.070 align:middle line:84% from a first principles point of view. 00:27:05.070 --> 00:27:08.430 align:middle line:84% But people have derived sticking coefficients 00:27:08.430 --> 00:27:10.170 align:middle line:90% for various types of deposition. 00:27:10.170 --> 00:27:12.990 align:middle line:90% 00:27:12.990 --> 00:27:15.960 align:middle line:84% OK, so slide 13, so this is the formula we said. 00:27:15.960 --> 00:27:20.100 align:middle line:84% The flux emitted is just 1 minus the sticking coefficient. 00:27:20.100 --> 00:27:22.830 align:middle line:84% It's those that don't stick times the flux incident 00:27:22.830 --> 00:27:24.430 align:middle line:90% at the point. 00:27:24.430 --> 00:27:25.720 align:middle line:90% We make some assumptions. 00:27:25.720 --> 00:27:27.920 align:middle line:84% Generally, we assume that ions stick. 00:27:27.920 --> 00:27:29.560 align:middle line:84% So if we have an ion coming down, 00:27:29.560 --> 00:27:31.240 align:middle line:84% we assume sticking coefficient is 1. 00:27:31.240 --> 00:27:34.720 align:middle line:84% Why is that? ions tend to be very reactive 00:27:34.720 --> 00:27:37.670 align:middle line:84% and they often come down with a certain amount of energy. 00:27:37.670 --> 00:27:40.210 align:middle line:84% So we usually say that they have sticking coefficient of 1. 00:27:40.210 --> 00:27:44.350 align:middle line:84% Neutrals can have any number one or less. 00:27:44.350 --> 00:27:46.300 align:middle line:84% And they're often assumed to be emitted 00:27:46.300 --> 00:27:49.520 align:middle line:84% with some kind of a cosine theta angle distribution. 00:27:49.520 --> 00:27:53.650 align:middle line:84% So that is they may have no memory of their arrival angle. 00:27:53.650 --> 00:27:56.350 align:middle line:84% So don't think of this, for example, in here, 00:27:56.350 --> 00:28:00.652 align:middle line:84% don't think of this as a little trough like a pool 00:28:00.652 --> 00:28:02.110 align:middle line:84% that we've taken all the water out, 00:28:02.110 --> 00:28:04.480 align:middle line:84% and you have a ping pong ball and it's bouncing around 00:28:04.480 --> 00:28:06.070 align:middle line:84% with billiard ball kind of collisions. 00:28:06.070 --> 00:28:07.820 align:middle line:84% It's not supposed to be thought like that. 00:28:07.820 --> 00:28:09.940 align:middle line:84% This atom is supposed to come down. 00:28:09.940 --> 00:28:12.220 align:middle line:84% And then atoms are coming down continuously, 00:28:12.220 --> 00:28:14.470 align:middle line:84% say at this point, and then it's readmitting 00:28:14.470 --> 00:28:17.050 align:middle line:84% at with some angular distribution, 00:28:17.050 --> 00:28:19.690 align:middle line:84% but it's not necessarily a billiard ball kind of collision 00:28:19.690 --> 00:28:22.800 align:middle line:90% type of thing happening. 00:28:22.800 --> 00:28:26.930 align:middle line:84% So you need to put in a re-emission flux distribution. 00:28:26.930 --> 00:28:29.530 align:middle line:90% 00:28:29.530 --> 00:28:35.430 align:middle line:84% So at any point i we get a flux of redeposited material 00:28:35.430 --> 00:28:39.840 align:middle line:84% because the emitted flux can land somewhere else 00:28:39.840 --> 00:28:40.990 align:middle line:90% on the surface. 00:28:40.990 --> 00:28:44.440 align:middle line:84% So we have this sort of, again, this ends up being bookkeeping. 00:28:44.440 --> 00:28:46.530 align:middle line:84% We have this something we express 00:28:46.530 --> 00:28:51.420 align:middle line:90% as f super ik redeposited. 00:28:51.420 --> 00:28:56.565 align:middle line:84% And what this is you take the flux, the direct flux, f sub k, 00:28:56.565 --> 00:28:58.440 align:middle line:84% that's coming into a point k on the surface-- 00:28:58.440 --> 00:29:00.540 align:middle line:90% let's say it's right here-- 00:29:00.540 --> 00:29:04.920 align:middle line:84% and we're looking at then point i. 00:29:04.920 --> 00:29:09.330 align:middle line:84% It's desorbed from point k and it's redeposited at point i. 00:29:09.330 --> 00:29:14.938 align:middle line:84% So this redeposited flux depends on some geometry factor, gik. 00:29:14.938 --> 00:29:16.980 align:middle line:84% So it has to do with the geometry between point i 00:29:16.980 --> 00:29:21.090 align:middle line:84% and point k times f sub k that's emitted, 00:29:21.090 --> 00:29:26.100 align:middle line:84% so times this emitted flux from this point k. 00:29:26.100 --> 00:29:28.020 align:middle line:84% Well, the emitted flux from point k 00:29:28.020 --> 00:29:29.460 align:middle line:84% just depends on the incoming flux 00:29:29.460 --> 00:29:33.630 align:middle line:84% to point k times the 1 minus the sticking coefficient. 00:29:33.630 --> 00:29:38.630 align:middle line:84% So the nice thing is here for point i can take any point k 00:29:38.630 --> 00:29:40.610 align:middle line:84% and figure out if I know the geometry 00:29:40.610 --> 00:29:42.350 align:middle line:90% factor between the two. 00:29:42.350 --> 00:29:46.640 align:middle line:84% How much is deposited at this point i 00:29:46.640 --> 00:29:52.100 align:middle line:84% resulting from having come from being emitted from point k? 00:29:52.100 --> 00:29:54.260 align:middle line:84% And then what you do is you take all 00:29:54.260 --> 00:29:57.020 align:middle line:84% of these at this point i you take redeposited fluxes 00:29:57.020 --> 00:30:00.350 align:middle line:84% from all points of the surface so that all points of a surface 00:30:00.350 --> 00:30:02.090 align:middle line:84% could be redepositing at this point. 00:30:02.090 --> 00:30:07.262 align:middle line:84% So you sum up for all i and k and you figure out 00:30:07.262 --> 00:30:07.970 align:middle line:90% what's deposited. 00:30:07.970 --> 00:30:11.510 align:middle line:84% So gik then accounts for the geometry between any point i 00:30:11.510 --> 00:30:12.050 align:middle line:90% and k. 00:30:12.050 --> 00:30:13.670 align:middle line:84% Because, obviously, if you're over here, 00:30:13.670 --> 00:30:15.170 align:middle line:84% let's say I'm over here, my point k. 00:30:15.170 --> 00:30:17.420 align:middle line:84% We're up at the surface, and it's re-emitting. 00:30:17.420 --> 00:30:20.750 align:middle line:84% Well, chances of it coming back down over to this point 00:30:20.750 --> 00:30:24.170 align:middle line:84% are dramatically lower than if it's re-emitting from the walls 00:30:24.170 --> 00:30:25.880 align:middle line:90% just based on geometry. 00:30:25.880 --> 00:30:30.380 align:middle line:90% 00:30:30.380 --> 00:30:35.660 align:middle line:84% OK, so let's take a look at some examples on slide 14. 00:30:35.660 --> 00:30:37.550 align:middle line:84% And hopefully this will become intuitive. 00:30:37.550 --> 00:30:39.380 align:middle line:84% But if you have a low sticking coefficient, 00:30:39.380 --> 00:30:41.780 align:middle line:84% s sub c much less than 1, you can end up 00:30:41.780 --> 00:30:44.720 align:middle line:84% with more conformal deposition or conformal coverage 00:30:44.720 --> 00:30:49.380 align:middle line:84% of the step because of the redeposition process. 00:30:49.380 --> 00:30:51.910 align:middle line:84% Redeposition and this low sticking coefficient 00:30:51.910 --> 00:30:53.910 align:middle line:84% is usually more important than surface diffusion 00:30:53.910 --> 00:30:55.500 align:middle line:84% because most of the time in these 00:30:55.500 --> 00:30:58.950 align:middle line:84% processes the wafer temperature is low enough that the surface 00:30:58.950 --> 00:31:03.090 align:middle line:84% diffusion is not all that high, especially if you're doing PVD. 00:31:03.090 --> 00:31:08.113 align:middle line:84% But let's just take a look here at the case A where I have 00:31:08.113 --> 00:31:09.280 align:middle line:90% a high sticking coefficient. 00:31:09.280 --> 00:31:10.697 align:middle line:84% So it's sticking coefficient is 1. 00:31:10.697 --> 00:31:12.330 align:middle line:84% Everything comes down and sticks. 00:31:12.330 --> 00:31:16.180 align:middle line:84% So what do I get if I'm depositing in this trench? 00:31:16.180 --> 00:31:18.240 align:middle line:84% Well, I get very non-conformal deposition. 00:31:18.240 --> 00:31:20.965 align:middle line:84% I get a film deposited at the bottom of the trench. 00:31:20.965 --> 00:31:22.590 align:middle line:84% There'd probably be a film up here too. 00:31:22.590 --> 00:31:23.673 align:middle line:90% It should have been shown. 00:31:23.673 --> 00:31:26.610 align:middle line:84% But we're only looking at what's happening in the trench. 00:31:26.610 --> 00:31:29.520 align:middle line:90% But nothing on the side walls. 00:31:29.520 --> 00:31:30.090 align:middle line:90% Why? 00:31:30.090 --> 00:31:35.550 align:middle line:84% Because I really don't have any opportunity for redeposition. 00:31:35.550 --> 00:31:39.520 align:middle line:84% Whereas, I can have that same angular arrival distribution. 00:31:39.520 --> 00:31:41.988 align:middle line:84% So with the exact same angle of arrival distribution, 00:31:41.988 --> 00:31:43.530 align:middle line:84% if I have a low sticking coefficient, 00:31:43.530 --> 00:31:46.660 align:middle line:84% things can move around a lot and hit the side walls. 00:31:46.660 --> 00:31:50.160 align:middle line:84% And then you can end up with more conformal deposition. 00:31:50.160 --> 00:31:52.290 align:middle line:84% And again, you would use this type of equation 00:31:52.290 --> 00:31:54.870 align:middle line:84% in the computer to be able to model that. 00:31:54.870 --> 00:31:56.770 align:middle line:84% You just vary the sticking coefficient. 00:31:56.770 --> 00:31:59.520 align:middle line:84% And you need to calculate the geometry factor 00:31:59.520 --> 00:32:01.570 align:middle line:90% as part of that. 00:32:01.570 --> 00:32:03.090 align:middle line:84% So low sticking coefficient is good 00:32:03.090 --> 00:32:04.298 align:middle line:90% if you want to coat a trench. 00:32:04.298 --> 00:32:05.928 align:middle line:84% Now, that may not be what you want. 00:32:05.928 --> 00:32:07.720 align:middle line:84% You may want to just deposit at the bottom. 00:32:07.720 --> 00:32:09.540 align:middle line:84% But again, it depends on the process 00:32:09.540 --> 00:32:13.300 align:middle line:90% that you're trying to do. 00:32:13.300 --> 00:32:15.580 align:middle line:84% OK, so that has to do with sticking coefficient. 00:32:15.580 --> 00:32:19.157 align:middle line:84% That's an important concept in basic surface physics. 00:32:19.157 --> 00:32:20.740 align:middle line:84% And we talked about surface diffusion. 00:32:20.740 --> 00:32:24.640 align:middle line:84% The third concept is sputtering, what happens on the surface. 00:32:24.640 --> 00:32:26.455 align:middle line:84% And that we talk about here on slide 15. 00:32:26.455 --> 00:32:29.540 align:middle line:90% 00:32:29.540 --> 00:32:31.840 align:middle line:84% There's a sputtered flux at any point 00:32:31.840 --> 00:32:35.380 align:middle line:84% i is caused by energetic usually incoming ions. 00:32:35.380 --> 00:32:36.550 align:middle line:90% Why do we care about ions? 00:32:36.550 --> 00:32:39.520 align:middle line:84% Well, the neutrals usually come down with low enough energy. 00:32:39.520 --> 00:32:41.600 align:middle line:84% They do a little bit of sputtering but not much. 00:32:41.600 --> 00:32:43.780 align:middle line:84% But an incoming ion can come in and it 00:32:43.780 --> 00:32:46.180 align:middle line:84% might have a couple hundred electron volts of energy. 00:32:46.180 --> 00:32:49.630 align:middle line:84% With enough energy, it can then hit something at the surface 00:32:49.630 --> 00:32:52.630 align:middle line:84% and cause it to be sputtered or emitted. 00:32:52.630 --> 00:32:55.000 align:middle line:84% So what we typically say is any point 00:32:55.000 --> 00:32:59.785 align:middle line:84% i the sputtered flux depends on the sputtering yield, y. 00:32:59.785 --> 00:33:04.270 align:middle line:84% y is just the number sputtered divided by the number incident. 00:33:04.270 --> 00:33:06.820 align:middle line:84% And this plot tells you how the sputtering yield varies 00:33:06.820 --> 00:33:10.840 align:middle line:84% with angle, depends on y times the flux 00:33:10.840 --> 00:33:15.370 align:middle line:84% of incident argon ions, flux of argon, 00:33:15.370 --> 00:33:18.860 align:middle line:90% and the direct flux of ions. 00:33:18.860 --> 00:33:22.580 align:middle line:84% So the total ionic flux is what we add up. 00:33:22.580 --> 00:33:24.580 align:middle line:84% So you might be curious to say, all right, well, 00:33:24.580 --> 00:33:26.663 align:middle line:84% how does the sputtering yield depend on the angle? 00:33:26.663 --> 00:33:28.690 align:middle line:84% Because that's going to tell you if you 00:33:28.690 --> 00:33:30.370 align:middle line:84% have a trench with a certain angle, 00:33:30.370 --> 00:33:32.820 align:middle line:84% and the atoms are coming in, and the ions are coming in, 00:33:32.820 --> 00:33:34.570 align:middle line:84% how is the sputtering yield going to vary. 00:33:34.570 --> 00:33:39.280 align:middle line:84% Well, this is a typical plot of yield versus angle. 00:33:39.280 --> 00:33:42.730 align:middle line:84% So again, a yield of 1 means it's 1 to 1. 00:33:42.730 --> 00:33:47.690 align:middle line:84% For every one that's incident on that point, one comes off. 00:33:47.690 --> 00:33:48.470 align:middle line:90% And you can see. 00:33:48.470 --> 00:33:50.970 align:middle line:84% And this is all measured with respect to the surface normal. 00:33:50.970 --> 00:33:53.660 align:middle line:84% So theta equals 0 means normal sputtering. 00:33:53.660 --> 00:33:56.890 align:middle line:90% So it's coming straight down. 00:33:56.890 --> 00:34:00.110 align:middle line:84% So at theta equals 0 for this particular example, 00:34:00.110 --> 00:34:01.300 align:middle line:90% the sputtering yield is one. 00:34:01.300 --> 00:34:02.290 align:middle line:90% And look what happens. 00:34:02.290 --> 00:34:04.870 align:middle line:84% As you increase theta-- so you're coming in a little more 00:34:04.870 --> 00:34:07.390 align:middle line:90% grazing, I'm increasing theta-- 00:34:07.390 --> 00:34:11.230 align:middle line:84% it peaks somewhere here around 50 or 60, 65 degrees. 00:34:11.230 --> 00:34:15.250 align:middle line:84% Gets up to about 2, 2 and 1/2, and then eventually 00:34:15.250 --> 00:34:16.943 align:middle line:84% at your very glancing incidence there's 00:34:16.943 --> 00:34:18.110 align:middle line:90% really no sputtering at all. 00:34:18.110 --> 00:34:20.780 align:middle line:90% It really drops dramatically. 00:34:20.780 --> 00:34:23.830 align:middle line:84% So this angular sensitive and you 00:34:23.830 --> 00:34:25.480 align:middle line:84% can use this to achieve more planar 00:34:25.480 --> 00:34:27.100 align:middle line:90% surfaces during a deposition. 00:34:27.100 --> 00:34:28.810 align:middle line:84% In fact, we'll show some examples 00:34:28.810 --> 00:34:33.070 align:middle line:84% where we use ionized species during deposition to give you 00:34:33.070 --> 00:34:34.870 align:middle line:90% a little better planarity. 00:34:34.870 --> 00:34:36.722 align:middle line:84% But so this angular dependence is something 00:34:36.722 --> 00:34:37.930 align:middle line:90% that we want to keep in mind. 00:34:37.930 --> 00:34:42.480 align:middle line:90% 00:34:42.480 --> 00:34:44.679 align:middle line:84% OK, and now on slide 16, so what happens? 00:34:44.679 --> 00:34:46.320 align:middle line:84% So let's say you do sputter that off, 00:34:46.320 --> 00:34:48.900 align:middle line:84% then you have redeposition of the sputtered molecules, 00:34:48.900 --> 00:34:50.250 align:middle line:90% how do you model that? 00:34:50.250 --> 00:34:52.123 align:middle line:84% The exact same way you modeled reemission 00:34:52.123 --> 00:34:53.790 align:middle line:84% when you were just coming off due to low 00:34:53.790 --> 00:34:55.260 align:middle line:90% sticking coefficient. 00:34:55.260 --> 00:34:58.230 align:middle line:84% You just say the redeposited flux between i 00:34:58.230 --> 00:35:01.980 align:middle line:84% at point i due to whatever is sputtering off from point k 00:35:01.980 --> 00:35:04.740 align:middle line:84% is just a geometry factor gik, which 00:35:04.740 --> 00:35:07.230 align:middle line:84% you which end up summing over the surface, 00:35:07.230 --> 00:35:10.140 align:middle line:84% times the sputter flux from point k. 00:35:10.140 --> 00:35:12.480 align:middle line:84% And that sputter flux from point k 00:35:12.480 --> 00:35:17.070 align:middle line:84% depends on the flux incident of ions times the sputtering yield 00:35:17.070 --> 00:35:21.140 align:middle line:84% y to see how the sputtering yield comes into it. 00:35:21.140 --> 00:35:23.120 align:middle line:84% And finally, if you want to substitute 00:35:23.120 --> 00:35:25.460 align:middle line:84% in the sticking coefficients, what 00:35:25.460 --> 00:35:27.560 align:middle line:84% you say is the redeposited at point 00:35:27.560 --> 00:35:32.840 align:middle line:84% i depends on the emitted from k, which depends on 1 00:35:32.840 --> 00:35:35.430 align:middle line:90% minus the sticking coefficient. 00:35:35.430 --> 00:35:39.200 align:middle line:84% So we have this redeposited due to emission 00:35:39.200 --> 00:35:41.510 align:middle line:84% is very, very similar to the redeposited 00:35:41.510 --> 00:35:45.020 align:middle line:90% due to the sputtering. 00:35:45.020 --> 00:35:47.960 align:middle line:84% And there's a final term in a lot of these. 00:35:47.960 --> 00:35:49.460 align:middle line:84% And this is the biggest fudge factor 00:35:49.460 --> 00:35:52.730 align:middle line:84% I would say in the whole model, which is that ions coming down 00:35:52.730 --> 00:35:55.460 align:middle line:84% on the surface can sometimes enhance the deposition 00:35:55.460 --> 00:35:59.330 align:middle line:84% rate in a way that's not completely obvious. 00:35:59.330 --> 00:36:01.020 align:middle line:90% The ions may come down. 00:36:01.020 --> 00:36:03.440 align:middle line:84% They may supply enough energy to drive 00:36:03.440 --> 00:36:06.360 align:middle line:84% a chemical reaction that might take place at the surface. 00:36:06.360 --> 00:36:09.390 align:middle line:90% So we have a deposition flux. 00:36:09.390 --> 00:36:11.390 align:middle line:84% So we can have a deposition rate that 00:36:11.390 --> 00:36:17.340 align:middle line:84% is called ion enhanced or ion induced chemical reaction. 00:36:17.340 --> 00:36:18.680 align:middle line:90% So it's kind of a fudge factor. 00:36:18.680 --> 00:36:22.100 align:middle line:84% We have an additional flux and a term with a k sub i, 00:36:22.100 --> 00:36:23.990 align:middle line:84% where k sub i is a fitting parameter, 00:36:23.990 --> 00:36:27.180 align:middle line:84% times the flux of ions coming in at any given point. 00:36:27.180 --> 00:36:31.400 align:middle line:84% So you can have this extra ion induced deposition rate. 00:36:31.400 --> 00:36:35.120 align:middle line:84% Obviously, that's only important in systems with ionized type 00:36:35.120 --> 00:36:38.130 align:middle line:90% of deposition. 00:36:38.130 --> 00:36:40.110 align:middle line:84% OK, so those are all the different processes 00:36:40.110 --> 00:36:41.580 align:middle line:84% at the surface and the direct flux. 00:36:41.580 --> 00:36:43.753 align:middle line:84% And now, starting on slide 17, I want 00:36:43.753 --> 00:36:46.170 align:middle line:84% to give a couple of examples because it seems like there's 00:36:46.170 --> 00:36:47.378 align:middle line:90% a lot of different equations. 00:36:47.378 --> 00:36:51.640 align:middle line:84% But it turns out a lot of them simplify for most systems. 00:36:51.640 --> 00:36:53.490 align:middle line:84% So let's do an example of low pressure CVD. 00:36:53.490 --> 00:37:00.000 align:middle line:84% Remember, LPCVD is probably the most common deposition process 00:37:00.000 --> 00:37:02.430 align:middle line:84% in a fab, typically takes place in a furnace that's 00:37:02.430 --> 00:37:06.690 align:middle line:84% pumped down to low pressure, say 10 to 100 millitorr. 00:37:06.690 --> 00:37:09.420 align:middle line:84% Now, the wafers are usually stacked in a batch. 00:37:09.420 --> 00:37:11.400 align:middle line:84% People do 25 to 50 wafers at a time. 00:37:11.400 --> 00:37:14.460 align:middle line:84% It's in a resistively heated furnace. 00:37:14.460 --> 00:37:15.663 align:middle line:90% There are no ions. 00:37:15.663 --> 00:37:16.830 align:middle line:90% This is not a plasma system. 00:37:16.830 --> 00:37:19.048 align:middle line:84% So we can get rid of all ionized processes. 00:37:19.048 --> 00:37:21.090 align:middle line:84% So we can ignore sputtering because, again, we're 00:37:21.090 --> 00:37:23.250 align:middle line:84% going to say that only in the presence of ions 00:37:23.250 --> 00:37:26.320 align:middle line:84% do you have enough energy to sputter something off. 00:37:26.320 --> 00:37:28.530 align:middle line:90% So we get rid of the sputtering. 00:37:28.530 --> 00:37:30.480 align:middle line:84% Usually, long range surface diffusion 00:37:30.480 --> 00:37:31.660 align:middle line:90% is not that important. 00:37:31.660 --> 00:37:33.670 align:middle line:84% So we can also ignore the surface diffusion. 00:37:33.670 --> 00:37:37.380 align:middle line:84% So if in my bookkeeping I have all these different fluxes, 00:37:37.380 --> 00:37:39.390 align:middle line:84% most of them are all not present. 00:37:39.390 --> 00:37:42.060 align:middle line:84% There's only two that dominate at any point i. 00:37:42.060 --> 00:37:43.690 align:middle line:90% It's the emitted flux. 00:37:43.690 --> 00:37:45.960 align:middle line:84% Yes, because we have a finite sticking coefficient, 00:37:45.960 --> 00:37:47.380 align:middle line:90% and the redeposited flux. 00:37:47.380 --> 00:37:49.300 align:middle line:84% Yes, we need to take that into account, 00:37:49.300 --> 00:37:51.100 align:middle line:90% and the neutral direct flux. 00:37:51.100 --> 00:37:52.830 align:middle line:84% So really in this example I really only 00:37:52.830 --> 00:37:54.622 align:middle line:84% going to have three fluxes I need to worry. 00:37:54.622 --> 00:37:57.330 align:middle line:84% About all the rest of these we don't 00:37:57.330 --> 00:38:01.380 align:middle line:84% have to take into account in the simulation. 00:38:01.380 --> 00:38:03.830 align:middle line:84% So then we can write down those fluxes 00:38:03.830 --> 00:38:07.980 align:middle line:84% pretty simply, and the net flux here on slide 18. 00:38:07.980 --> 00:38:12.350 align:middle line:84% So at any point i on the surface, f super i, 00:38:12.350 --> 00:38:15.650 align:middle line:84% the net flux is again the flux coming in 00:38:15.650 --> 00:38:16.970 align:middle line:90% minus the flux going out. 00:38:16.970 --> 00:38:19.880 align:middle line:84% So it's the flux of directs, neutral species, 00:38:19.880 --> 00:38:22.220 align:middle line:84% the redeposited flux that comes from other areas 00:38:22.220 --> 00:38:25.490 align:middle line:84% of the surface, minus the flux that goes out. 00:38:25.490 --> 00:38:27.620 align:middle line:84% Well, those that don't stick, that's 00:38:27.620 --> 00:38:32.540 align:middle line:84% 1 minus the sticking coefficient times the direct flux 00:38:32.540 --> 00:38:36.950 align:middle line:84% of neutrals and the redeposited direct flux. 00:38:36.950 --> 00:38:37.730 align:middle line:90% OK? 00:38:37.730 --> 00:38:40.340 align:middle line:84% So these are the ones that come in, 00:38:40.340 --> 00:38:41.840 align:middle line:84% and these are the ones that go out. 00:38:41.840 --> 00:38:44.940 align:middle line:84% And there's a negative sign in between the two. 00:38:44.940 --> 00:38:46.940 align:middle line:84% So you can rearrange this a little bit 00:38:46.940 --> 00:38:49.160 align:middle line:84% and you can write it then as shown here, 00:38:49.160 --> 00:38:51.920 align:middle line:84% the net flux at any point i is the sticking coefficient 00:38:51.920 --> 00:38:57.200 align:middle line:84% s sub c times the direct flux at that point of neutrals 00:38:57.200 --> 00:38:59.780 align:middle line:90% plus a term that relates-- 00:38:59.780 --> 00:39:03.650 align:middle line:84% a geometric term, this gik times 1 minus s sub c. 00:39:03.650 --> 00:39:06.800 align:middle line:84% That's those that don't stick times the flux 00:39:06.800 --> 00:39:11.210 align:middle line:84% into some other point on the surface k. 00:39:11.210 --> 00:39:13.740 align:middle line:84% And then we just need to sum this up 00:39:13.740 --> 00:39:17.460 align:middle line:84% over all points k on the surface. 00:39:17.460 --> 00:39:19.940 align:middle line:84% Now, what you can do to simplify this a little bit, 00:39:19.940 --> 00:39:26.060 align:middle line:84% you can define f sub d to be the deposition flux at each point 00:39:26.060 --> 00:39:29.600 align:middle line:84% to be equal to the direct flux of neutrals 00:39:29.600 --> 00:39:33.200 align:middle line:84% plus the redeposited flux if you want to do that. 00:39:33.200 --> 00:39:36.170 align:middle line:84% And then you can write the deposition rate more simply 00:39:36.170 --> 00:39:39.360 align:middle line:90% as the sticking coefficient. 00:39:39.360 --> 00:39:42.110 align:middle line:84% So this is the dep rate, then sticking coefficient 00:39:42.110 --> 00:39:46.520 align:middle line:84% times the flux at any given point i divided by n. 00:39:46.520 --> 00:39:48.770 align:middle line:84% So we can then calculate the deposition rate 00:39:48.770 --> 00:39:50.730 align:middle line:90% at any point i on the surface. 00:39:50.730 --> 00:39:54.110 align:middle line:84% And we use a cosine theta to the n distribution 00:39:54.110 --> 00:39:55.410 align:middle line:90% for the incoming molecules. 00:39:55.410 --> 00:39:59.385 align:middle line:84% So it's a relatively simple formula can be used for LPCVD. 00:39:59.385 --> 00:40:02.650 align:middle line:90% 00:40:02.650 --> 00:40:04.630 align:middle line:84% And in fact, on slide 19, I just wanted 00:40:04.630 --> 00:40:05.830 align:middle line:90% to go through an example. 00:40:05.830 --> 00:40:08.890 align:middle line:84% This took directly from your text. 00:40:08.890 --> 00:40:12.100 align:middle line:84% What we're asked to do is calculate the dep rate 00:40:12.100 --> 00:40:14.710 align:middle line:84% using this LPCVD expression that we just derived, 00:40:14.710 --> 00:40:18.430 align:middle line:84% sticking coefficient times a flux divided by density, 00:40:18.430 --> 00:40:20.090 align:middle line:90% for silicon dioxide. 00:40:20.090 --> 00:40:21.520 align:middle line:84% So this is low temperature oxide. 00:40:21.520 --> 00:40:22.730 align:middle line:90% And we have a flat surface. 00:40:22.730 --> 00:40:23.500 align:middle line:90% So this is nice. 00:40:23.500 --> 00:40:25.690 align:middle line:90% There's no topography. 00:40:25.690 --> 00:40:31.650 align:middle line:84% We are told that the sticking coefficient is 0.3 00:40:31.650 --> 00:40:34.290 align:middle line:84% and that the maximum unobstructed flux, so that's 00:40:34.290 --> 00:40:36.720 align:middle line:84% the maximum flux that's coming straight down, 00:40:36.720 --> 00:40:39.660 align:middle line:84% is equal to 3 times 10 to the 15th molecules 00:40:39.660 --> 00:40:41.250 align:middle line:84% per square centimeter per second. 00:40:41.250 --> 00:40:43.933 align:middle line:90% So that's a molecular flux. 00:40:43.933 --> 00:40:45.850 align:middle line:84% And we're given the density of the film that's 00:40:45.850 --> 00:40:50.930 align:middle line:84% deposited, 2.27 grams per cubic centimeter. 00:40:50.930 --> 00:40:54.120 align:middle line:84% So this is a relatively simple calculation. 00:40:54.120 --> 00:40:57.200 align:middle line:84% We want to get the deposition rate on a flat surface, 00:40:57.200 --> 00:40:59.460 align:middle line:84% therefore there is no shadowing effect. 00:40:59.460 --> 00:41:03.140 align:middle line:84% There's no redeposited flux because we're not coming off 00:41:03.140 --> 00:41:04.730 align:middle line:90% of a side wall of a trench. 00:41:04.730 --> 00:41:08.310 align:middle line:84% And each surface position is in a horizontal orientation. 00:41:08.310 --> 00:41:12.100 align:middle line:84% So we're only concerned about the direct flux. 00:41:12.100 --> 00:41:17.410 align:middle line:84% So f sub d, the deposited flux, is just equal everywhere 00:41:17.410 --> 00:41:19.570 align:middle line:84% to the maximum value, which is this 3 times 10 00:41:19.570 --> 00:41:21.925 align:middle line:84% to the 15th molecules per square centimeter per second. 00:41:21.925 --> 00:41:25.240 align:middle line:90% 00:41:25.240 --> 00:41:27.880 align:middle line:84% And you can look up this equation 9.53. 00:41:27.880 --> 00:41:30.370 align:middle line:84% In fact, that's exactly what I just 00:41:30.370 --> 00:41:34.590 align:middle line:84% was showing down here for the dep rate. 00:41:34.590 --> 00:41:36.000 align:middle line:90% So we have that equation. 00:41:36.000 --> 00:41:37.950 align:middle line:84% The rate is just the sticking coefficient 00:41:37.950 --> 00:41:39.720 align:middle line:90% times the flux divided by n. 00:41:39.720 --> 00:41:44.610 align:middle line:84% We need to do a little bit of conversion here 00:41:44.610 --> 00:41:47.700 align:middle line:84% to get ourselves from grams per cubic centimeter 00:41:47.700 --> 00:41:48.992 align:middle line:90% into molecules. 00:41:48.992 --> 00:41:50.700 align:middle line:84% The way we do that is we take the density 00:41:50.700 --> 00:41:53.538 align:middle line:84% of SiO2 being 2.27 grams per cubic centimeter. 00:41:53.538 --> 00:41:57.270 align:middle line:84% And we multiply that by Avogadro's number 00:41:57.270 --> 00:42:01.710 align:middle line:84% divided by the number of grams per mole of SiO2. 00:42:01.710 --> 00:42:03.270 align:middle line:84% So we need to do a little conversion 00:42:03.270 --> 00:42:06.390 align:middle line:84% to convert this density of grams per cubic centimeter 00:42:06.390 --> 00:42:09.570 align:middle line:84% into a number of molecules per cubic centimeter for SiO2. 00:42:09.570 --> 00:42:13.260 align:middle line:84% So you get 2 times 10 to the 22 molecules 00:42:13.260 --> 00:42:17.902 align:middle line:84% of SiO2, those units per cubic centimeter. 00:42:17.902 --> 00:42:19.610 align:middle line:84% And then you just use the simple formula. 00:42:19.610 --> 00:42:21.235 align:middle line:84% The rate is just a sticking coefficient 00:42:21.235 --> 00:42:23.305 align:middle line:84% times the flux, the maximum flux over n. 00:42:23.305 --> 00:42:24.930 align:middle line:84% You plug-in the numbers you were given. 00:42:24.930 --> 00:42:32.790 align:middle line:84% And you get about something like 0.023 microns per minute. 00:42:32.790 --> 00:42:36.930 align:middle line:84% But now, if you had a deposition on a non-flat surface, 00:42:36.930 --> 00:42:39.700 align:middle line:84% if you had a trench or something like that on the side wall, 00:42:39.700 --> 00:42:41.830 align:middle line:84% for example, the local dep rate would be less. 00:42:41.830 --> 00:42:45.960 align:middle line:84% So this gives you a maximum, a rough idea of the maximum. 00:42:45.960 --> 00:42:48.150 align:middle line:90% So it just gives you-- 00:42:48.150 --> 00:42:50.700 align:middle line:84% we'll talk later about SPEEDIE simulations, 00:42:50.700 --> 00:42:51.840 align:middle line:90% more complex topographies. 00:42:51.840 --> 00:42:54.573 align:middle line:84% But it gives you a rough idea of how you can model 00:42:54.573 --> 00:42:55.740 align:middle line:90% this simple type of process. 00:42:55.740 --> 00:42:59.140 align:middle line:90% 00:42:59.140 --> 00:43:02.200 align:middle line:90% OK, let's go on to slide 20. 00:43:02.200 --> 00:43:04.610 align:middle line:84% And we'll give some more examples. 00:43:04.610 --> 00:43:06.850 align:middle line:90% Now we're going to do PVD. 00:43:06.850 --> 00:43:09.410 align:middle line:84% So we've already talked about physical vapor deposition 00:43:09.410 --> 00:43:09.910 align:middle line:90% systems. 00:43:09.910 --> 00:43:14.410 align:middle line:84% We said there could be DC or RF sputtering or evaporation. 00:43:14.410 --> 00:43:18.190 align:middle line:84% Typically, ions do not play a significant role 00:43:18.190 --> 00:43:20.600 align:middle line:84% in these kind of PVD systems most of the time. 00:43:20.600 --> 00:43:25.150 align:middle line:84% So the modeling of simple PVD is very much analogous to LPCVD. 00:43:25.150 --> 00:43:28.390 align:middle line:84% The parameters are very different though. 00:43:28.390 --> 00:43:34.100 align:middle line:84% So here's an example of a simple DC sputtering system, 00:43:34.100 --> 00:43:36.410 align:middle line:84% which we talked about a couple of lectures ago. 00:43:36.410 --> 00:43:39.080 align:middle line:84% And these are the processes that go on. 00:43:39.080 --> 00:43:41.980 align:middle line:84% So just like LPCVD, we can say a rate 00:43:41.980 --> 00:43:44.620 align:middle line:84% at any point on the surface is a sticking coefficient times 00:43:44.620 --> 00:43:49.250 align:middle line:84% the flux, the total deposited flux divided by the density. 00:43:49.250 --> 00:43:53.410 align:middle line:84% However, the key difference is that the sticking coefficient 00:43:53.410 --> 00:43:56.170 align:middle line:84% and the arrival angle distribution, 00:43:56.170 --> 00:43:59.230 align:middle line:84% cosine theta to the n, will be very different between LPCVD 00:43:59.230 --> 00:44:00.550 align:middle line:90% and PVD systems. 00:44:00.550 --> 00:44:03.625 align:middle line:84% And that's where the real differences tend to come about. 00:44:03.625 --> 00:44:07.400 align:middle line:90% 00:44:07.400 --> 00:44:08.995 align:middle line:84% There is one other type of system 00:44:08.995 --> 00:44:11.120 align:middle line:84% for PVD, though, where you do have to pay attention 00:44:11.120 --> 00:44:13.370 align:middle line:84% to a little more complex processes than you would 00:44:13.370 --> 00:44:14.030 align:middle line:90% in LPCVD. 00:44:14.030 --> 00:44:17.210 align:middle line:84% And that's shown on slide 21 when you have ionized PVD. 00:44:17.210 --> 00:44:20.810 align:middle line:84% We did talk about the case where we have 00:44:20.810 --> 00:44:24.090 align:middle line:90% systems that are more complex. 00:44:24.090 --> 00:44:26.780 align:middle line:84% This because ions and neutrals both play a role. 00:44:26.780 --> 00:44:30.030 align:middle line:84% Here's an example of an ionized PVD system. 00:44:30.030 --> 00:44:32.542 align:middle line:84% So you have sputtering of aluminum 00:44:32.542 --> 00:44:33.500 align:middle line:90% coming off this target. 00:44:33.500 --> 00:44:35.500 align:middle line:84% But then you have an additional-- in addition to 00:44:35.500 --> 00:44:39.080 align:middle line:84% that, you have a coil that goes around 00:44:39.080 --> 00:44:41.810 align:middle line:84% that ionizes the aluminum as it comes off. 00:44:41.810 --> 00:44:45.470 align:middle line:84% Now you have a large density of ions that are being created. 00:44:45.470 --> 00:44:46.850 align:middle line:84% And then these can be accelerated 00:44:46.850 --> 00:44:49.130 align:middle line:90% towards the substrate. 00:44:49.130 --> 00:44:53.210 align:middle line:84% So this is not that uncommon for metal deposition. 00:44:53.210 --> 00:44:55.660 align:middle line:84% So you have aluminum or maybe titanium 00:44:55.660 --> 00:44:56.960 align:middle line:90% ions that'll be present. 00:44:56.960 --> 00:44:58.420 align:middle line:84% Therefore, when you have ions, you 00:44:58.420 --> 00:45:01.000 align:middle line:84% do have to take into account sputtering and other processes 00:45:01.000 --> 00:45:05.680 align:middle line:84% that will not be present in LPCVD or the simpler PVD 00:45:05.680 --> 00:45:06.310 align:middle line:90% systems. 00:45:06.310 --> 00:45:10.330 align:middle line:84% So ionized PVD, inductively coupled ionized PVD, 00:45:10.330 --> 00:45:14.170 align:middle line:84% is a little bit more complicated to model, just more 00:45:14.170 --> 00:45:18.330 align:middle line:90% bookkeeping in the equations. 00:45:18.330 --> 00:45:20.810 align:middle line:84% Here's an example, in fact, on slide 22, 00:45:20.810 --> 00:45:24.120 align:middle line:84% if we do ionize PVD, of all the different fluxes that really do 00:45:24.120 --> 00:45:25.370 align:middle line:90% need to be taken into account. 00:45:25.370 --> 00:45:28.310 align:middle line:84% And you see we have yes on all of them, 00:45:28.310 --> 00:45:31.580 align:middle line:84% the direct neutrals, the direct ions. 00:45:31.580 --> 00:45:34.083 align:middle line:84% The only one we're not taking into account in this case 00:45:34.083 --> 00:45:36.500 align:middle line:84% because we're assuming the wafer is low enough temperature 00:45:36.500 --> 00:45:38.780 align:middle line:84% that we can ignore surface diffusion. 00:45:38.780 --> 00:45:42.170 align:middle line:84% But all these fluxes would need to be taken into account. 00:45:42.170 --> 00:45:45.500 align:middle line:84% And if you do that, shown here on slide 22, 00:45:45.500 --> 00:45:48.080 align:middle line:84% this is the example of the possible terms 00:45:48.080 --> 00:45:50.400 align:middle line:84% that could be included in your computer model. 00:45:50.400 --> 00:45:53.530 align:middle line:84% So you would say that the rate at any point i 00:45:53.530 --> 00:45:55.290 align:middle line:84% depends on the sticking coefficient 00:45:55.290 --> 00:45:58.890 align:middle line:84% times the direct maximum deposited 00:45:58.890 --> 00:46:01.230 align:middle line:84% flux, where fd includes the direct 00:46:01.230 --> 00:46:03.780 align:middle line:84% and the redeposited neutral fluxes. 00:46:03.780 --> 00:46:08.723 align:middle line:84% Plus f sub i includes the direct redeposit and ion induced flux. 00:46:08.723 --> 00:46:09.640 align:middle line:90% So there's an f sub i. 00:46:09.640 --> 00:46:11.960 align:middle line:84% And again, f sub i doesn't have a sticking coefficient 00:46:11.960 --> 00:46:12.460 align:middle line:90% associated. 00:46:12.460 --> 00:46:14.680 align:middle line:84% We say for ions, we include all of them, 00:46:14.680 --> 00:46:17.800 align:middle line:84% we say they just come down and stick. 00:46:17.800 --> 00:46:20.500 align:middle line:84% And then you have a minus those that are sputtered off. 00:46:20.500 --> 00:46:24.620 align:middle line:84% You have a sputter term plus a redeposited term. 00:46:24.620 --> 00:46:28.120 align:middle line:84% So this frd models the redeposition due to sputtering. 00:46:28.120 --> 00:46:32.030 align:middle line:84% So you have extra terms here divided by the density n. 00:46:32.030 --> 00:46:34.450 align:middle line:84% And so you have a positive term and a negative term, 00:46:34.450 --> 00:46:37.810 align:middle line:84% some of which depending on the point i and the surface 00:46:37.810 --> 00:46:41.200 align:middle line:84% and the angles involved may cancel each other out. 00:46:41.200 --> 00:46:44.140 align:middle line:90% 00:46:44.140 --> 00:46:47.790 align:middle line:84% OK so those are some basic bookkeeping models. 00:46:47.790 --> 00:46:49.440 align:middle line:84% I want to actually do some examples 00:46:49.440 --> 00:46:51.810 align:middle line:84% of the values of these parameters 00:46:51.810 --> 00:46:53.730 align:middle line:84% that are useful for specific systems. 00:46:53.730 --> 00:46:55.920 align:middle line:84% And so far, we've talked about two parameters that 00:46:55.920 --> 00:46:59.520 align:middle line:84% are most important, the n value, which 00:46:59.520 --> 00:47:03.870 align:middle line:84% is the exponent in the cosine theta to the n arrival angle 00:47:03.870 --> 00:47:07.900 align:middle line:84% distribution, and the sticking coefficient s sub c. 00:47:07.900 --> 00:47:12.988 align:middle line:84% Those are the two most critical in modeling simple systems. 00:47:12.988 --> 00:47:15.280 align:middle line:84% And there are a couple of different model systems shown 00:47:15.280 --> 00:47:15.780 align:middle line:90% here. 00:47:15.780 --> 00:47:17.890 align:middle line:84% On the top we're talking about sputter deposition, 00:47:17.890 --> 00:47:21.950 align:middle line:84% either standard or ionized, and evaporation. 00:47:21.950 --> 00:47:23.500 align:middle line:90% So these are PVD systems. 00:47:23.500 --> 00:47:27.050 align:middle line:84% In the bottom, we have the CVD systems. 00:47:27.050 --> 00:47:28.600 align:middle line:90% So let's look at PVD. 00:47:28.600 --> 00:47:30.850 align:middle line:84% Well, physical vapor deposition, it 00:47:30.850 --> 00:47:33.470 align:middle line:84% tends to have much more vertical arrival angle distribution. 00:47:33.470 --> 00:47:35.530 align:middle line:90% So the n value is big, right? 00:47:35.530 --> 00:47:38.680 align:middle line:84% Almost everything in PVD is coming straight down. 00:47:38.680 --> 00:47:41.060 align:middle line:84% Certainly in evaporation you have very low pressure, 00:47:41.060 --> 00:47:42.220 align:middle line:90% so you have line of sight. 00:47:42.220 --> 00:47:46.690 align:middle line:84% Even in sputtering you tend to have close very anisotropic. 00:47:46.690 --> 00:47:49.285 align:middle line:84% So for sputter deposition, depending, 00:47:49.285 --> 00:47:52.690 align:middle line:84% you can have n equals up to five maybe. 00:47:52.690 --> 00:47:58.590 align:middle line:84% If you're ionized you can get up into the range of 50 or so. 00:47:58.590 --> 00:48:03.540 align:middle line:84% But in any case, n is usually greater than 1 for PVD. 00:48:03.540 --> 00:48:07.500 align:middle line:84% In contrast, CVD systems tend to have pretty much 00:48:07.500 --> 00:48:09.920 align:middle line:90% isotropic arrival angle. 00:48:09.920 --> 00:48:12.630 align:middle line:84% PVD, again, takes place you're sputtering 00:48:12.630 --> 00:48:16.410 align:middle line:84% in 10 to the minus 5 torr, or 1 times 10 00:48:16.410 --> 00:48:19.540 align:middle line:84% to the minus 3 torr, or something like that. 00:48:19.540 --> 00:48:23.340 align:middle line:84% CVD usually much higher, hundreds of millitorr, 00:48:23.340 --> 00:48:25.270 align:middle line:90% 0.1 torr or something like that. 00:48:25.270 --> 00:48:27.450 align:middle line:84% So you usually have higher pressure. 00:48:27.450 --> 00:48:29.130 align:middle line:84% You have a lot of gas phase collisions 00:48:29.130 --> 00:48:31.410 align:middle line:84% that randomize things and mostly neutrals. 00:48:31.410 --> 00:48:33.460 align:middle line:84% Usually, in CVD, you don't have many ions. 00:48:33.460 --> 00:48:35.310 align:middle line:90% So n is usually 1 in CVD. 00:48:35.310 --> 00:48:40.120 align:middle line:84% So the cosine theta to the n is usually fairly isotropic. 00:48:40.120 --> 00:48:41.440 align:middle line:90% What about PVD? 00:48:41.440 --> 00:48:44.800 align:middle line:84% Well, the sticking coefficient in PVD tends to be close to 1. 00:48:44.800 --> 00:48:47.140 align:middle line:84% There really isn't much surface chemistry involved. 00:48:47.140 --> 00:48:50.080 align:middle line:84% The aluminum or whatever it is that's being evaporated 00:48:50.080 --> 00:48:55.120 align:middle line:84% or sputtered usually comes down and pretty close to sticks. 00:48:55.120 --> 00:48:56.260 align:middle line:90% There isn't a whole lot. 00:48:56.260 --> 00:48:57.100 align:middle line:90% There can be. 00:48:57.100 --> 00:49:00.190 align:middle line:84% But most of the time the atoms arrive and stick. 00:49:00.190 --> 00:49:02.680 align:middle line:84% Very, very different, though, in the case of LPCVD-- 00:49:02.680 --> 00:49:04.960 align:middle line:84% look at LPCVD, the sticking coefficients, 00:49:04.960 --> 00:49:07.480 align:middle line:84% depending on what your chemistry using, if you're 00:49:07.480 --> 00:49:09.370 align:middle line:84% using a silicon source of silane, 00:49:09.370 --> 00:49:12.220 align:middle line:84% you might have a sticking coefficient of 0.5 or 0.3. 00:49:12.220 --> 00:49:15.400 align:middle line:84% For TEOS, we mentioned there's other ways of doing 00:49:15.400 --> 00:49:18.670 align:middle line:84% low temperature oxide, not just with silane, 00:49:18.670 --> 00:49:22.570 align:middle line:84% but there's this other silicon sources. 00:49:22.570 --> 00:49:26.780 align:middle line:84% Sticking coefficient can be much lower. 00:49:26.780 --> 00:49:28.730 align:middle line:84% LPCVD tungsten or polysilicon you 00:49:28.730 --> 00:49:31.500 align:middle line:84% have much lower sticking coefficients, 00:49:31.500 --> 00:49:34.910 align:middle line:90% so very different from PVD. 00:49:34.910 --> 00:49:37.767 align:middle line:84% That same chart is reproduced here again on slide 23. 00:49:37.767 --> 00:49:38.850 align:middle line:90% It's the exact same chart. 00:49:38.850 --> 00:49:41.183 align:middle line:84% I just wanted to reproduce it and talk a little bit more 00:49:41.183 --> 00:49:42.890 align:middle line:90% about the CVD thing. 00:49:42.890 --> 00:49:45.650 align:middle line:84% Because they have surface chemistry, 00:49:45.650 --> 00:49:48.470 align:middle line:84% sticking coefficient much, much less than 1, 00:49:48.470 --> 00:49:52.380 align:middle line:84% and things often evaporate before they react. 00:49:52.380 --> 00:49:53.410 align:middle line:90% What does this mean? 00:49:53.410 --> 00:49:55.510 align:middle line:84% Well, just by looking at these numbers, 00:49:55.510 --> 00:49:57.817 align:middle line:84% you know that CVD systems with their small sticking 00:49:57.817 --> 00:49:59.400 align:middle line:84% coefficients are going to tend to have 00:49:59.400 --> 00:50:01.350 align:middle line:90% more conformal deposition. 00:50:01.350 --> 00:50:03.750 align:middle line:84% If you want to cover a step and you 00:50:03.750 --> 00:50:05.310 align:middle line:84% want to cover it very conformally, 00:50:05.310 --> 00:50:05.920 align:middle line:90% what does that mean? 00:50:05.920 --> 00:50:08.250 align:middle line:84% That means the thickness along the flat part of the wafer 00:50:08.250 --> 00:50:09.660 align:middle line:84% is exactly equal to the thickness 00:50:09.660 --> 00:50:12.273 align:middle line:84% of the film along the trench wall is 00:50:12.273 --> 00:50:13.440 align:middle line:90% exactly equal in the bottom. 00:50:13.440 --> 00:50:14.940 align:middle line:90% It's just completely conformal. 00:50:14.940 --> 00:50:18.390 align:middle line:84% In that case, you probably want to use CVD. 00:50:18.390 --> 00:50:20.050 align:middle line:84% Now, that may not be what you want, 00:50:20.050 --> 00:50:22.590 align:middle line:84% but this is what we mean by conformal deposition, 00:50:22.590 --> 00:50:23.580 align:middle line:90% that equal thickness. 00:50:23.580 --> 00:50:29.220 align:middle line:84% And CVD tends to give you more conformality just because 00:50:29.220 --> 00:50:33.870 align:middle line:84% of this sticking coefficient argument. 00:50:33.870 --> 00:50:36.630 align:middle line:84% OK, now I want to show some actual examples of simulations 00:50:36.630 --> 00:50:38.430 align:middle line:84% to convince you of some of these things 00:50:38.430 --> 00:50:39.680 align:middle line:90% that we've been talking about. 00:50:39.680 --> 00:50:41.940 align:middle line:84% And we're going to vary the parameters n 00:50:41.940 --> 00:50:44.440 align:middle line:90% and s sub c in the model. 00:50:44.440 --> 00:50:47.670 align:middle line:84% So on slide 25, these are topography simulations 00:50:47.670 --> 00:50:49.500 align:middle line:84% that are taken from the simulator 00:50:49.500 --> 00:50:50.940 align:middle line:90% that I mentioned called SPEEDIE. 00:50:50.940 --> 00:50:52.710 align:middle line:84% Again, that's a topography simulator. 00:50:52.710 --> 00:50:55.950 align:middle line:84% You cannot simulate this type of topography in SUPREM-IV. 00:50:55.950 --> 00:50:59.110 align:middle line:90% SUPREM-IV doesn't do topography. 00:50:59.110 --> 00:51:05.910 align:middle line:84% So this is an example of you're doing a deposition using LPCVD 00:51:05.910 --> 00:51:08.310 align:middle line:84% of SiO2 and you originally have a trench that you have 00:51:08.310 --> 00:51:10.150 align:middle line:90% etched that looks like this. 00:51:10.150 --> 00:51:11.800 align:middle line:90% And each contour line-- 00:51:11.800 --> 00:51:13.410 align:middle line:90% it's a little bit hard to see. 00:51:13.410 --> 00:51:17.070 align:middle line:84% But each contour line represents the deposited topography 00:51:17.070 --> 00:51:18.490 align:middle line:90% at different time intervals. 00:51:18.490 --> 00:51:22.630 align:middle line:84% So each one might be after a 1 minute interval. 00:51:22.630 --> 00:51:26.910 align:middle line:84% So you get an idea of how the topography develops over time. 00:51:26.910 --> 00:51:29.430 align:middle line:84% In this particular example, we did a SPEEDIE simulation 00:51:29.430 --> 00:51:32.305 align:middle line:84% with a sticking coefficient of 1. 00:51:32.305 --> 00:51:33.930 align:middle line:84% Actually, the sticking coefficient of 1 00:51:33.930 --> 00:51:35.340 align:middle line:90% is probably more typical of PVD. 00:51:35.340 --> 00:51:38.880 align:middle line:84% So this really should be PVD in your mind. 00:51:38.880 --> 00:51:40.980 align:middle line:84% And all we did mathematically is vary 00:51:40.980 --> 00:51:44.230 align:middle line:84% the arrival angle distribution, the n value. 00:51:44.230 --> 00:51:47.010 align:middle line:84% So this is for n equals 1, that's isotropic. 00:51:47.010 --> 00:51:53.020 align:middle line:84% Here's for n equals 3, somewhat anisotropic, n equals 10. 00:51:53.020 --> 00:51:54.360 align:middle line:90% So let's look at this. 00:51:54.360 --> 00:51:56.910 align:middle line:84% Even for isotropic arrival distribution, 00:51:56.910 --> 00:52:00.360 align:middle line:84% because of shadowing effects and things, 00:52:00.360 --> 00:52:01.962 align:middle line:84% conformal coverage is not achieved 00:52:01.962 --> 00:52:03.420 align:middle line:84% with the sticking coefficient of 1. 00:52:03.420 --> 00:52:05.860 align:middle line:84% That is, what do I mean by conformal coverage is not 00:52:05.860 --> 00:52:06.360 align:middle line:90% achieved? 00:52:06.360 --> 00:52:07.980 align:middle line:84% Well, the thickness of the film here 00:52:07.980 --> 00:52:10.410 align:middle line:84% at the surface on the flat portions 00:52:10.410 --> 00:52:14.790 align:middle line:84% is not equal to the thickness along the side wall. 00:52:14.790 --> 00:52:18.060 align:middle line:84% The side wall is thinner even for n equals 1. 00:52:18.060 --> 00:52:20.890 align:middle line:90% 00:52:20.890 --> 00:52:25.720 align:middle line:84% And so what this implies that if you have a sticking coefficient 00:52:25.720 --> 00:52:30.940 align:middle line:84% s sub c equal to 1, the geometry and line of sight issues 00:52:30.940 --> 00:52:33.080 align:middle line:90% are going to be very important. 00:52:33.080 --> 00:52:35.380 align:middle line:84% So for example, for PVD, you really 00:52:35.380 --> 00:52:37.270 align:middle line:84% need to take into account the geometry 00:52:37.270 --> 00:52:40.510 align:middle line:90% of what you're depositing into. 00:52:40.510 --> 00:52:42.400 align:middle line:84% Look at this one for n equals 10. 00:52:42.400 --> 00:52:44.503 align:middle line:84% You almost get no deposition on the side wall. 00:52:44.503 --> 00:52:45.920 align:middle line:84% You get quite a bit on the bottom. 00:52:45.920 --> 00:52:48.940 align:middle line:84% And notice this is for an aspect ratio that's 00:52:48.940 --> 00:52:51.820 align:middle line:84% pretty gentle, I mean, pretty not that sophisticated. 00:52:51.820 --> 00:52:54.200 align:middle line:84% This aspect ratio of 0.3, what does that mean? 00:52:54.200 --> 00:52:58.330 align:middle line:84% That means the trench is 3 times wider than it is tall. 00:52:58.330 --> 00:53:01.270 align:middle line:84% So it's only 1 micron tall, but it's 3 microns wide. 00:53:01.270 --> 00:53:04.840 align:middle line:84% And you're already seeing these nonconformallity types 00:53:04.840 --> 00:53:07.200 align:middle line:90% of issues. 00:53:07.200 --> 00:53:11.450 align:middle line:84% So this is what you'd expect from PVD with these parameters. 00:53:11.450 --> 00:53:14.270 align:middle line:84% Just to remind you, on page 26 what we're talking about, 00:53:14.270 --> 00:53:16.700 align:middle line:84% the n equals 1 the flux as a function of angle 00:53:16.700 --> 00:53:18.050 align:middle line:90% looks like this. 00:53:18.050 --> 00:53:20.060 align:middle line:84% N equals 3 looks something-- it cuts out 00:53:20.060 --> 00:53:21.450 align:middle line:90% after about 60 degrees. 00:53:21.450 --> 00:53:24.900 align:middle line:84% And the n equals 15 cuts out after about 30 degrees. 00:53:24.900 --> 00:53:27.360 align:middle line:90% So this is close to n equals 15. 00:53:27.360 --> 00:53:30.420 align:middle line:84% So when I was getting more than about 30 degrees, 00:53:30.420 --> 00:53:33.380 align:middle line:84% so from a normal you can imagine going on to this surface, 00:53:33.380 --> 00:53:35.600 align:middle line:84% there is not a whole lot of flux. 00:53:35.600 --> 00:53:36.720 align:middle line:90% And that's the problem. 00:53:36.720 --> 00:53:38.405 align:middle line:84% That's why we're getting such low-- 00:53:38.405 --> 00:53:40.920 align:middle line:90% 00:53:40.920 --> 00:53:42.943 align:middle line:84% with the sticking coefficient of 1, 00:53:42.943 --> 00:53:44.360 align:middle line:84% you're getting such low deposition 00:53:44.360 --> 00:53:45.485 align:middle line:90% along those vertical walls. 00:53:45.485 --> 00:53:49.440 align:middle line:90% 00:53:49.440 --> 00:53:54.210 align:middle line:84% Let's make it a little more difficult even 00:53:54.210 --> 00:53:55.800 align:middle line:90% let's take an aspect ratio. 00:53:55.800 --> 00:53:57.690 align:middle line:84% We're going to take those same parameters, so 00:53:57.690 --> 00:54:01.020 align:middle line:84% the same parameters that we just saw. 00:54:01.020 --> 00:54:05.650 align:middle line:84% Sticking coefficient equals 1 here on slide 27, 00:54:05.650 --> 00:54:08.530 align:middle line:84% but now I've increased the aspect ratio from 0.3 00:54:08.530 --> 00:54:11.060 align:middle line:90% up to 1.3 or 1.25. 00:54:11.060 --> 00:54:15.913 align:middle line:84% So now the trench is 1.25 times tall 00:54:15.913 --> 00:54:17.080 align:middle line:90% compared to what it is wide. 00:54:17.080 --> 00:54:18.580 align:middle line:84% So it's much taller than it is wide, 00:54:18.580 --> 00:54:20.205 align:middle line:84% just the opposite of what I had before. 00:54:20.205 --> 00:54:23.550 align:middle line:90% 00:54:23.550 --> 00:54:27.430 align:middle line:84% Now, what you do see again is less deposition on the side 00:54:27.430 --> 00:54:27.930 align:middle line:90% walls. 00:54:27.930 --> 00:54:32.910 align:middle line:84% Here, this is A here on the left is for isotropic arrival, 00:54:32.910 --> 00:54:33.900 align:middle line:90% n equals 1. 00:54:33.900 --> 00:54:37.540 align:middle line:84% And b is for anisotropic, n equals 10. 00:54:37.540 --> 00:54:39.030 align:middle line:90% So let's look at the n equals 1. 00:54:39.030 --> 00:54:40.140 align:middle line:90% Well, what's happening? 00:54:40.140 --> 00:54:43.350 align:middle line:84% You see if for isotropic arrival angle, 00:54:43.350 --> 00:54:44.970 align:middle line:84% you're not getting much material down 00:54:44.970 --> 00:54:47.430 align:middle line:84% at the bottom of that narrow trench. 00:54:47.430 --> 00:54:50.400 align:middle line:84% Because, again, it's reasonably isotropic arrival, 00:54:50.400 --> 00:54:53.340 align:middle line:84% you're getting material coming deposited along here 00:54:53.340 --> 00:54:55.740 align:middle line:84% and you're sort of developing these lobes up 00:54:55.740 --> 00:54:58.320 align:middle line:84% at the top, which if you had a narrow enough trench, 00:54:58.320 --> 00:55:00.000 align:middle line:84% you can imagine these might cut off 00:55:00.000 --> 00:55:02.370 align:middle line:90% and you end up with a big void. 00:55:02.370 --> 00:55:05.220 align:middle line:84% If you're more anisotropic, you're coming down straight. 00:55:05.220 --> 00:55:07.470 align:middle line:84% So you do get quite a bit at the bottom of the trench. 00:55:07.470 --> 00:55:08.850 align:middle line:84% You get quite a bit on the flats. 00:55:08.850 --> 00:55:12.160 align:middle line:84% But you get nothing or very little on the side walls. 00:55:12.160 --> 00:55:14.410 align:middle line:84% So as you go to narrower and narrower trenches, 00:55:14.410 --> 00:55:16.120 align:middle line:84% you can imagine you might end up not 00:55:16.120 --> 00:55:17.140 align:middle line:90% being able to fill them at all. 00:55:17.140 --> 00:55:19.807 align:middle line:84% You might end up with voiding if you have a sticking coefficient 00:55:19.807 --> 00:55:23.770 align:middle line:84% of 1 is really what this boils down to for high aspect ratio 00:55:23.770 --> 00:55:26.570 align:middle line:90% trenches. 00:55:26.570 --> 00:55:30.090 align:middle line:84% So if you want to avoid that, let's 00:55:30.090 --> 00:55:32.580 align:middle line:84% say the purpose of your deposition 00:55:32.580 --> 00:55:33.930 align:middle line:90% is to fill this trench. 00:55:33.930 --> 00:55:36.360 align:middle line:84% You haven't done a very good job in this case. 00:55:36.360 --> 00:55:37.920 align:middle line:84% And you're not going to do a good job in this case at all. 00:55:37.920 --> 00:55:39.128 align:middle line:90% You're going to pinch it off. 00:55:39.128 --> 00:55:40.860 align:middle line:90% What do we need to do? 00:55:40.860 --> 00:55:42.860 align:middle line:84% Well, we need to lower the sticking coefficient. 00:55:42.860 --> 00:55:45.870 align:middle line:84% We need to have not such a line of sight distribution. 00:55:45.870 --> 00:55:48.080 align:middle line:84% We need for atoms to come down, come in here, 00:55:48.080 --> 00:55:50.570 align:middle line:84% and not stick, and move around so they can eventually 00:55:50.570 --> 00:55:53.150 align:middle line:90% cover these side walls. 00:55:53.150 --> 00:55:56.840 align:middle line:84% And that's exactly what is shown on slide 00:55:56.840 --> 00:55:59.270 align:middle line:90% 28 in the SPEEDIE simulations. 00:55:59.270 --> 00:56:02.690 align:middle line:84% Here's an example of now I've got 00:56:02.690 --> 00:56:06.110 align:middle line:84% that same trench, that same high aspect ratio, 1.25. 00:56:06.110 --> 00:56:08.630 align:middle line:84% But on the left here, A, I have the sticking coefficient 00:56:08.630 --> 00:56:09.810 align:middle line:90% of 0.1. 00:56:09.810 --> 00:56:12.020 align:middle line:90% So only 1 in 10 atoms sticks. 00:56:12.020 --> 00:56:15.323 align:middle line:84% The other 9, when they hit a point, they go somewhere else 00:56:15.323 --> 00:56:16.740 align:middle line:84% and they continue to bounce around 00:56:16.740 --> 00:56:18.240 align:middle line:90% until they finally stick. 00:56:18.240 --> 00:56:19.670 align:middle line:90% Here's 1 in 100 sticking. 00:56:19.670 --> 00:56:22.340 align:middle line:90% 00:56:22.340 --> 00:56:24.620 align:middle line:84% so, again, this is with n equals 1. 00:56:24.620 --> 00:56:26.973 align:middle line:84% So it's isotropic arrival, but with different values 00:56:26.973 --> 00:56:28.140 align:middle line:90% of the sticking coefficient. 00:56:28.140 --> 00:56:30.020 align:middle line:84% So sticking coefficient is the knob 00:56:30.020 --> 00:56:33.500 align:middle line:84% you want to turn if you want to fill a trench. 00:56:33.500 --> 00:56:36.300 align:middle line:84% Changing the n value helped, but not a whole lot. 00:56:36.300 --> 00:56:39.560 align:middle line:84% It was really lowering the sticking coefficient. 00:56:39.560 --> 00:56:41.240 align:middle line:84% You get much more conformal coverage. 00:56:41.240 --> 00:56:43.400 align:middle line:84% So you get on the sidewall just about what you had. 00:56:43.400 --> 00:56:46.070 align:middle line:90% 00:56:46.070 --> 00:56:52.010 align:middle line:84% So this is typical on A, this view of things, the parameter 00:56:52.010 --> 00:56:52.580 align:middle line:90% values. 00:56:52.580 --> 00:56:54.890 align:middle line:84% n equals 1 and sticking coefficient of 0.1 00:56:54.890 --> 00:57:00.290 align:middle line:84% is very typical of LTL, so low temperature oxide or CVD SiO2. 00:57:00.290 --> 00:57:02.930 align:middle line:84% b with a sticking coefficient of 1 and 100 00:57:02.930 --> 00:57:07.030 align:middle line:90% is more typical of tungsten CVD. 00:57:07.030 --> 00:57:09.490 align:middle line:84% So we need to reduce the sticking coefficient if you 00:57:09.490 --> 00:57:14.710 align:middle line:90% need more conformal deposition. 00:57:14.710 --> 00:57:16.300 align:middle line:90% How about the sidewall angle? 00:57:16.300 --> 00:57:22.120 align:middle line:84% So if I even go back here, I look at this, 00:57:22.120 --> 00:57:23.920 align:middle line:84% I didn't do a perfect job in this case 00:57:23.920 --> 00:57:24.878 align:middle line:90% of filling this trench. 00:57:24.878 --> 00:57:27.790 align:middle line:84% It look like it's going to pinch off and form a void before I 00:57:27.790 --> 00:57:29.218 align:middle line:90% get the whole thing filled. 00:57:29.218 --> 00:57:31.510 align:middle line:84% Even here it doesn't look all necessarily that perfect. 00:57:31.510 --> 00:57:33.730 align:middle line:90% So what can we do? 00:57:33.730 --> 00:57:36.260 align:middle line:84% You only have so much latitude, you're sticking coefficient. 00:57:36.260 --> 00:57:38.677 align:middle line:84% You can change the pressure and do things in your reactor. 00:57:38.677 --> 00:57:40.630 align:middle line:84% But you only have so much latitude. 00:57:40.630 --> 00:57:43.150 align:middle line:84% Well, what you do is you change the topography 00:57:43.150 --> 00:57:44.450 align:middle line:90% that you're starting with. 00:57:44.450 --> 00:57:48.700 align:middle line:84% So these examples are results of SPEEDIE simulations 00:57:48.700 --> 00:57:52.658 align:middle line:84% where you're doing LPCVD into a trench. 00:57:52.658 --> 00:57:54.200 align:middle line:84% And we've changed the sidewall angle. 00:57:54.200 --> 00:57:55.890 align:middle line:84% Here, the trench has a 90 degree angle, 00:57:55.890 --> 00:57:57.590 align:middle line:84% so the side walls go straight down. 00:57:57.590 --> 00:58:00.600 align:middle line:84% Here, it's opened up slightly by 5 degrees. 00:58:00.600 --> 00:58:02.610 align:middle line:90% So this is an 85 degree angle. 00:58:02.610 --> 00:58:06.380 align:middle line:84% And here you open up a little bit more, 80 degree angle. 00:58:06.380 --> 00:58:10.130 align:middle line:84% This has a sticking coefficient of 0.2 and an n of 1. 00:58:10.130 --> 00:58:14.150 align:middle line:84% So this is very typical of low temperature oxide deposition, 00:58:14.150 --> 00:58:16.580 align:middle line:90% for example, into a trench. 00:58:16.580 --> 00:58:21.097 align:middle line:84% So if you decrease the angle from 90 to 80 degrees, 00:58:21.097 --> 00:58:22.430 align:middle line:90% you greatly improve the filling. 00:58:22.430 --> 00:58:23.990 align:middle line:84% See at 90 degrees you're going to pinch off 00:58:23.990 --> 00:58:25.282 align:middle line:90% and you're going to get a void. 00:58:25.282 --> 00:58:26.720 align:middle line:84% At 80 degrees you're not going to. 00:58:26.720 --> 00:58:30.380 align:middle line:84% So it's very common in etching, when 00:58:30.380 --> 00:58:33.633 align:middle line:84% you go to etch, to try to etch a contact hole in a via, 00:58:33.633 --> 00:58:35.300 align:middle line:84% you never want to etch it straight down. 00:58:35.300 --> 00:58:39.680 align:middle line:84% You always want to have a slight angle, say make it 80 degrees 00:58:39.680 --> 00:58:43.940 align:middle line:84% to slope this so that you can get much better filling 00:58:43.940 --> 00:58:44.720 align:middle line:90% of the hole. 00:58:44.720 --> 00:58:47.330 align:middle line:90% Or if you're trying to make-- 00:58:47.330 --> 00:58:50.840 align:middle line:84% you want metal lines to go over a step on your wafer, 00:58:50.840 --> 00:58:53.500 align:middle line:84% you never want to make the step look like this. 00:58:53.500 --> 00:58:55.500 align:middle line:84% You usually want to have it more sloped. 00:58:55.500 --> 00:58:58.890 align:middle line:84% So you somehow slope the step when you etch it, 00:58:58.890 --> 00:59:01.120 align:middle line:84% or you put in side wall spacers, or do something. 00:59:01.120 --> 00:59:03.840 align:middle line:84% So the general idea, if you want good step coverage, 00:59:03.840 --> 00:59:07.380 align:middle line:90% don't make such abrupt steps. 00:59:07.380 --> 00:59:10.770 align:middle line:84% If you're doing vias, you can just decrease your angle 00:59:10.770 --> 00:59:14.820 align:middle line:84% from 90 to 80 and you get much better filling of the trench. 00:59:14.820 --> 00:59:18.780 align:middle line:90% 00:59:18.780 --> 00:59:22.890 align:middle line:84% OK, how about there are a little bit more complicated cases? 00:59:22.890 --> 00:59:26.370 align:middle line:84% I've just shown very simple cases, PVD without ions 00:59:26.370 --> 00:59:28.200 align:middle line:90% or LPCVD without ions. 00:59:28.200 --> 00:59:30.420 align:middle line:84% How about the case where you do have an ion flux? 00:59:30.420 --> 00:59:32.790 align:middle line:84% And there are a couple of cases like that. 00:59:32.790 --> 00:59:36.300 align:middle line:84% There is what's called HDP, high density plasma CVD. 00:59:36.300 --> 00:59:39.570 align:middle line:84% This has a very high ion flux, could 00:59:39.570 --> 00:59:41.070 align:middle line:84% be three or four orders of magnitude 00:59:41.070 --> 00:59:47.010 align:middle line:84% higher than what you have in ordinary CVD. 00:59:47.010 --> 00:59:49.360 align:middle line:90% So how does this work? 00:59:49.360 --> 00:59:52.620 align:middle line:84% Well, think we mentioned this at one point. 00:59:52.620 --> 00:59:56.550 align:middle line:84% This particular one has a microwave source. 00:59:56.550 --> 01:00:00.090 align:middle line:84% What we have is we have a plasma up here. 01:00:00.090 --> 01:00:03.960 align:middle line:84% And there's a microwave supply that's creating those ions. 01:00:03.960 --> 01:00:06.802 align:middle line:90% But then it's created remotely. 01:00:06.802 --> 01:00:08.760 align:middle line:84% So this ionized plasma has a very high density, 01:00:08.760 --> 01:00:11.710 align:middle line:84% but it's created remotely, not directly over the wafer. 01:00:11.710 --> 01:00:15.450 align:middle line:84% And then you have an RF bias and you extract those ions 01:00:15.450 --> 01:00:17.100 align:middle line:90% towards the wafer. 01:00:17.100 --> 01:00:19.540 align:middle line:84% So the nice thing about this is that you 01:00:19.540 --> 01:00:22.450 align:middle line:84% can get a very high density without having 01:00:22.450 --> 01:00:27.890 align:middle line:84% huge energies, huge voltages, to extract them to the wafer 01:00:27.890 --> 01:00:31.580 align:middle line:90% so you don't get so much damage. 01:00:31.580 --> 01:00:38.960 align:middle line:84% And this high density plasma can give you a very high CVD dep 01:00:38.960 --> 01:00:42.230 align:middle line:84% rate, even down to a room temperature 01:00:42.230 --> 01:00:44.940 align:middle line:90% with relatively low pressures. 01:00:44.940 --> 01:00:46.670 align:middle line:84% So the key is you have a separate RF 01:00:46.670 --> 01:00:48.770 align:middle line:90% bias applied to the substrate. 01:00:48.770 --> 01:00:50.600 align:middle line:84% That controls the angular dependence 01:00:50.600 --> 01:00:56.250 align:middle line:84% of what comes down and of ion sputtering during deposition. 01:00:56.250 --> 01:00:58.800 align:middle line:84% So we have a direct ion flux coming down 01:00:58.800 --> 01:01:00.150 align:middle line:90% like this on the surface. 01:01:00.150 --> 01:01:02.910 align:middle line:84% And we have ions that come down and sputter off. 01:01:02.910 --> 01:01:04.740 align:middle line:84% And now here's where our resputter flux 01:01:04.740 --> 01:01:07.800 align:middle line:84% becomes important for these high density plasma systems. 01:01:07.800 --> 01:01:09.660 align:middle line:84% And the thing that we need to realize 01:01:09.660 --> 01:01:12.030 align:middle line:84% is because of the angular dependence of sputtering, 01:01:12.030 --> 01:01:16.080 align:middle line:84% sputtering occurs preferentially on a slope surface rather than 01:01:16.080 --> 01:01:18.840 align:middle line:84% a surface that's either vertical or horizontal. 01:01:18.840 --> 01:01:21.600 align:middle line:84% Because, remember, that sputtering rate sort of 01:01:21.600 --> 01:01:23.230 align:middle line:90% peaked at around 60 degrees. 01:01:23.230 --> 01:01:25.020 align:middle line:84% So if you have a sloped surface, you 01:01:25.020 --> 01:01:26.400 align:middle line:90% get preferential sputtering. 01:01:26.400 --> 01:01:29.610 align:middle line:84% So you can actually use these high density plasma 01:01:29.610 --> 01:01:32.030 align:middle line:90% systems to do planarization. 01:01:32.030 --> 01:01:34.530 align:middle line:84% Now, all along I've been talking about conformal deposition. 01:01:34.530 --> 01:01:36.600 align:middle line:84% You may not want to get conformal deposition. 01:01:36.600 --> 01:01:38.010 align:middle line:90% You may want planarized. 01:01:38.010 --> 01:01:40.380 align:middle line:84% So what do I mean by a planarized deposition? 01:01:40.380 --> 01:01:42.090 align:middle line:90% Well, if you start out-- 01:01:42.090 --> 01:01:45.820 align:middle line:84% let's say your surface topography looks like this. 01:01:45.820 --> 01:01:47.280 align:middle line:84% You have a metal line or something 01:01:47.280 --> 01:01:49.110 align:middle line:84% and you want to cover it with oxide, 01:01:49.110 --> 01:01:52.830 align:middle line:84% but you want it to be planar when you're done. 01:01:52.830 --> 01:01:55.320 align:middle line:84% You'd like the surface to look like that after deposition, 01:01:55.320 --> 01:01:56.250 align:middle line:90% completely flat. 01:01:56.250 --> 01:01:58.740 align:middle line:84% That's just the opposite of conformal. 01:01:58.740 --> 01:02:02.570 align:middle line:84% If it were conformal, right, it would have looked like this. 01:02:02.570 --> 01:02:05.190 align:middle line:84% It would reproduce that big hump. 01:02:05.190 --> 01:02:08.250 align:middle line:84% But if you're trying to do planarization during deposition 01:02:08.250 --> 01:02:10.740 align:middle line:84% without using CMP, you'd love a process 01:02:10.740 --> 01:02:12.480 align:middle line:90% that would deposit like this. 01:02:12.480 --> 01:02:13.780 align:middle line:90% Well, how can you do that? 01:02:13.780 --> 01:02:16.350 align:middle line:84% Well, there are some processes that are self 01:02:16.350 --> 01:02:19.670 align:middle line:90% planarizing during deposition. 01:02:19.670 --> 01:02:21.920 align:middle line:84% And here's an here's an example of such a process. 01:02:21.920 --> 01:02:24.200 align:middle line:84% You need these ionized techniques, 01:02:24.200 --> 01:02:26.600 align:middle line:84% which in addition to depositing, are doing 01:02:26.600 --> 01:02:28.740 align:middle line:90% sputtering preferentially. 01:02:28.740 --> 01:02:33.320 align:middle line:84% So here's an example on slide 31 of two SPEEDIE simulations. 01:02:33.320 --> 01:02:37.850 align:middle line:84% On the left, I have low pressure CVD, regular low pressure CVD. 01:02:37.850 --> 01:02:39.590 align:middle line:84% So I have a line like this and I'm 01:02:39.590 --> 01:02:41.420 align:middle line:90% trying to cover it with oxide. 01:02:41.420 --> 01:02:43.415 align:middle line:84% And you see you started out with a bump 01:02:43.415 --> 01:02:44.540 align:middle line:90% and you end up with a bump. 01:02:44.540 --> 01:02:45.623 align:middle line:90% There is no planarization. 01:02:45.623 --> 01:02:47.970 align:middle line:90% It's perfectly conforming. 01:02:47.970 --> 01:02:53.020 align:middle line:84% In b, we've simulated this high density plasma CVD deposition, 01:02:53.020 --> 01:02:55.140 align:middle line:90% and it has a directed ion flux. 01:02:55.140 --> 01:02:59.620 align:middle line:84% And the key is it has this angle dependent sputtering, which 01:02:59.620 --> 01:03:02.560 align:middle line:84% you end up with a little bump but much more planarized 01:03:02.560 --> 01:03:04.030 align:middle line:90% topography after deposition. 01:03:04.030 --> 01:03:06.880 align:middle line:84% So if you had to do CMP to planarize, on this one 01:03:06.880 --> 01:03:08.650 align:middle line:90% you'd have to CMP quite a bit. 01:03:08.650 --> 01:03:11.920 align:middle line:84% On this one, you'd probably have to CMP only a little bit, just 01:03:11.920 --> 01:03:12.970 align:middle line:90% a small amount. 01:03:12.970 --> 01:03:15.160 align:middle line:84% And that's very desirable because CMP 01:03:15.160 --> 01:03:16.705 align:middle line:90% is a tricky process. 01:03:16.705 --> 01:03:19.550 align:middle line:84% It introduces non-uniformity and it's fairly expensive. 01:03:19.550 --> 01:03:22.330 align:middle line:84% So if you can do self planarizing deposition 01:03:22.330 --> 01:03:25.480 align:middle line:84% with these fancy ionized deposition systems, 01:03:25.480 --> 01:03:28.300 align:middle line:90% you're much better off. 01:03:28.300 --> 01:03:30.710 align:middle line:84% And I'll actually show some real examples of that. 01:03:30.710 --> 01:03:34.000 align:middle line:84% So besides depositing over a line, 01:03:34.000 --> 01:03:36.670 align:middle line:84% you can also use this high density plasma CVD. 01:03:36.670 --> 01:03:39.250 align:middle line:84% It's very good at filling open spaces 01:03:39.250 --> 01:03:41.720 align:middle line:90% and reducing void formation. 01:03:41.720 --> 01:03:44.740 align:middle line:84% So here's an example here, again, a SPEEDIE simulation 01:03:44.740 --> 01:03:45.650 align:middle line:90% on the left. 01:03:45.650 --> 01:03:48.290 align:middle line:84% I had a trench with a relatively high aspect ratio. 01:03:48.290 --> 01:03:51.970 align:middle line:90% This was that 1.25 aspect ratio. 01:03:51.970 --> 01:03:56.170 align:middle line:84% By LPCVD, you see it pinches off and you end up with a void. 01:03:56.170 --> 01:03:59.260 align:middle line:84% And in order to do this, we used a sticking coefficient 01:03:59.260 --> 01:04:02.920 align:middle line:84% of 0.2 and an isotropic arrival angle. 01:04:02.920 --> 01:04:04.570 align:middle line:90% On the one on the left-- 01:04:04.570 --> 01:04:06.790 align:middle line:84% on the right, I'm sorry-- is high density plasma 01:04:06.790 --> 01:04:09.070 align:middle line:90% deposition in the trench. 01:04:09.070 --> 01:04:12.310 align:middle line:90% And you get much better filling. 01:04:12.310 --> 01:04:14.690 align:middle line:90% You don't see the void. 01:04:14.690 --> 01:04:15.810 align:middle line:90% Well, why does it do this? 01:04:15.810 --> 01:04:18.020 align:middle line:84% Well, high density plasma gives you better filling 01:04:18.020 --> 01:04:20.103 align:middle line:84% because you combine this highly directed ion flux. 01:04:20.103 --> 01:04:22.687 align:middle line:84% So a lot of the ions are coming straight down, and that helps. 01:04:22.687 --> 01:04:24.170 align:middle line:90% They get into the trench. 01:04:24.170 --> 01:04:26.895 align:middle line:84% So they're towards the surface with this angle 01:04:26.895 --> 01:04:27.770 align:middle line:90% dependent sputtering. 01:04:27.770 --> 01:04:29.300 align:middle line:84% So the angular dependent sputtering 01:04:29.300 --> 01:04:31.610 align:middle line:84% tends to etch off any of these overhangs. 01:04:31.610 --> 01:04:33.530 align:middle line:84% As they form they get etched off by 01:04:33.530 --> 01:04:35.130 align:middle line:90% the sputtering preferentially. 01:04:35.130 --> 01:04:37.377 align:middle line:84% So overhangs that develop get sputtered away, 01:04:37.377 --> 01:04:39.710 align:middle line:84% and you end up being able to fill the trench pretty well 01:04:39.710 --> 01:04:41.542 align:middle line:90% without a void. 01:04:41.542 --> 01:04:43.250 align:middle line:84% Often, you do not want voids in the film. 01:04:43.250 --> 01:04:44.975 align:middle line:84% You don't want air spaces in your chips. 01:04:44.975 --> 01:04:47.225 align:middle line:84% So this is a way to do it is that high density plasma. 01:04:47.225 --> 01:04:50.770 align:middle line:90% 01:04:50.770 --> 01:04:54.630 align:middle line:84% Here's an example on slide 33 just to show you-- 01:04:54.630 --> 01:04:58.650 align:middle line:84% instead of just showing SPEEDIE simulations, actual data. 01:04:58.650 --> 01:05:02.610 align:middle line:84% These are scanning electron micrograph images 01:05:02.610 --> 01:05:05.040 align:middle line:90% of HDP oxide deposition. 01:05:05.040 --> 01:05:08.760 align:middle line:84% And on the left, this was a metal line you can see. 01:05:08.760 --> 01:05:11.860 align:middle line:84% And we we're trying to deposit oxide over that. 01:05:11.860 --> 01:05:13.380 align:middle line:84% And again, if it had been conformal, 01:05:13.380 --> 01:05:15.270 align:middle line:84% you'd have a big hump in your oxide. 01:05:15.270 --> 01:05:18.703 align:middle line:84% But with high density plasma CVD, which is not conformal, 01:05:18.703 --> 01:05:20.370 align:middle line:84% all you end up with pretty planar oxide. 01:05:20.370 --> 01:05:22.260 align:middle line:84% You still have a little bit of a bump there, 01:05:22.260 --> 01:05:24.870 align:middle line:84% a very small little bump, that you 01:05:24.870 --> 01:05:27.003 align:middle line:90% would have to planarize away. 01:05:27.003 --> 01:05:28.920 align:middle line:84% But the amount of CMP you'd have to do on that 01:05:28.920 --> 01:05:31.930 align:middle line:90% would be very minimal. 01:05:31.930 --> 01:05:35.410 align:middle line:84% And again, so that's the case of depositing over a line. 01:05:35.410 --> 01:05:37.390 align:middle line:84% If you're depositing oxide in a trench, 01:05:37.390 --> 01:05:39.400 align:middle line:84% so here's an example of a trench. 01:05:39.400 --> 01:05:42.040 align:middle line:84% You have two metal lines, one on the right, one on the left. 01:05:42.040 --> 01:05:44.290 align:middle line:84% They're separated by half a micron or so. 01:05:44.290 --> 01:05:46.390 align:middle line:84% You want to completely fill this without voids. 01:05:46.390 --> 01:05:50.200 align:middle line:84% You can see you've gotten excellent filling. 01:05:50.200 --> 01:05:53.230 align:middle line:84% You did get some of these strange, you know, 01:05:53.230 --> 01:05:57.590 align:middle line:84% topographies because of the ion effects. 01:05:57.590 --> 01:05:59.320 align:middle line:84% But that would have to be CMPed off. 01:05:59.320 --> 01:06:02.770 align:middle line:84% But the key is you were able to fill a very small space 01:06:02.770 --> 01:06:04.777 align:middle line:90% without inducing voids. 01:06:04.777 --> 01:06:06.610 align:middle line:84% And that would be a problem if you're trying 01:06:06.610 --> 01:06:09.460 align:middle line:90% to use, say, an LPCVD system. 01:06:09.460 --> 01:06:12.900 align:middle line:84% So you can imagine looking at these topographies, 01:06:12.900 --> 01:06:16.500 align:middle line:84% simulating, and then going back in SPEEDIE 01:06:16.500 --> 01:06:18.420 align:middle line:84% and varying your models to try to get 01:06:18.420 --> 01:06:22.890 align:middle line:84% something that looks like what you actually got from SCM. 01:06:22.890 --> 01:06:24.420 align:middle line:84% And that's what exactly what people 01:06:24.420 --> 01:06:25.650 align:middle line:90% do in topography simulation. 01:06:25.650 --> 01:06:27.150 align:middle line:84% Again, it's not first principles. 01:06:27.150 --> 01:06:30.030 align:middle line:84% You vary all these different parameters. 01:06:30.030 --> 01:06:33.040 align:middle line:84% The sputtering yield is a function of angle, the n value, 01:06:33.040 --> 01:06:35.040 align:middle line:84% which is the cosine theta to the n distribution, 01:06:35.040 --> 01:06:36.082 align:middle line:90% the sticking coefficient. 01:06:36.082 --> 01:06:38.910 align:middle line:84% You vary all those until you get something in your simulation 01:06:38.910 --> 01:06:42.780 align:middle line:84% that looks somewhat close to what you actually deposited. 01:06:42.780 --> 01:06:46.680 align:middle line:90% 01:06:46.680 --> 01:06:49.080 align:middle line:84% Slide 34 shows the case-- we haven't 01:06:49.080 --> 01:06:51.720 align:middle line:84% had much chance to talk about it, but this whole idea 01:06:51.720 --> 01:06:54.180 align:middle line:90% of surface diffusion. 01:06:54.180 --> 01:06:56.880 align:middle line:84% And this becomes important only when 01:06:56.880 --> 01:07:00.840 align:middle line:84% you're usually doing high temperature PVD, say 01:07:00.840 --> 01:07:01.590 align:middle line:90% for aluminum. 01:07:01.590 --> 01:07:05.250 align:middle line:84% That's a practical example that people sometimes do. 01:07:05.250 --> 01:07:08.070 align:middle line:84% Again, all of these issues tend to arise 01:07:08.070 --> 01:07:11.080 align:middle line:84% when you have high aspect ratios on the wafer. 01:07:11.080 --> 01:07:16.290 align:middle line:84% So you have a very high aspect ratio trench like this 01:07:16.290 --> 01:07:19.590 align:middle line:84% and you're trying to fill that with aluminum. 01:07:19.590 --> 01:07:25.200 align:middle line:84% And so on A on the left is a high aspect ratio trench 01:07:25.200 --> 01:07:26.310 align:middle line:90% using PVD. 01:07:26.310 --> 01:07:30.690 align:middle line:84% So we have a sticking coefficient of 1, n equals 4. 01:07:30.690 --> 01:07:32.880 align:middle line:84% So this could be a sputtered situation. 01:07:32.880 --> 01:07:34.560 align:middle line:90% And you have very poor filling. 01:07:34.560 --> 01:07:37.500 align:middle line:84% In fact, you don't have much at all aluminum on the side walls 01:07:37.500 --> 01:07:41.820 align:middle line:84% if you weren't able to fill the trench well at all. 01:07:41.820 --> 01:07:47.860 align:middle line:84% In B, this is again, now this is way of high temperature 01:07:47.860 --> 01:07:50.080 align:middle line:84% PVD where we've raised the surface of the wafer 01:07:50.080 --> 01:07:54.250 align:middle line:84% up to about 400 degrees, much better filling, not 01:07:54.250 --> 01:07:56.030 align:middle line:90% completely, but much better. 01:07:56.030 --> 01:07:59.080 align:middle line:84% And finally, if you do to a deposition temperature of 550, 01:07:59.080 --> 01:08:00.190 align:middle line:90% and you can't go to hot. 01:08:00.190 --> 01:08:02.117 align:middle line:90% Aluminum melts at six something. 01:08:02.117 --> 01:08:02.950 align:middle line:90% You'd be in trouble. 01:08:02.950 --> 01:08:05.405 align:middle line:84% But at 550, you get a lot of surface diffusion. 01:08:05.405 --> 01:08:07.780 align:middle line:84% And you get a lot of reflow of the aluminum, essentially, 01:08:07.780 --> 01:08:10.810 align:middle line:84% and you completely filled the trench, so much better 01:08:10.810 --> 01:08:12.410 align:middle line:90% much smoother topography. 01:08:12.410 --> 01:08:13.570 align:middle line:90% So that's an example. 01:08:13.570 --> 01:08:15.340 align:middle line:84% When you get to very high aspect ratios, 01:08:15.340 --> 01:08:18.710 align:middle line:84% you cannot rely just simply on sputtering. 01:08:18.710 --> 01:08:20.950 align:middle line:84% And so you need to add something to the system. 01:08:20.950 --> 01:08:23.920 align:middle line:84% You add the surface diffusion component 01:08:23.920 --> 01:08:25.330 align:middle line:84% by raising the wafer temperature. 01:08:25.330 --> 01:08:27.100 align:middle line:84% So then you need a special system 01:08:27.100 --> 01:08:28.779 align:middle line:84% where the sputtering system is designed 01:08:28.779 --> 01:08:32.740 align:middle line:84% so it can heat the wafer up to reasonably high temperatures 01:08:32.740 --> 01:08:34.855 align:middle line:84% at which the species will surface diffuse. 01:08:34.855 --> 01:08:39.720 align:middle line:90% 01:08:39.720 --> 01:08:43.080 align:middle line:84% OK, so on slide 35 I just want to summarize. 01:08:43.080 --> 01:08:47.790 align:middle line:84% As I mentioned, this is the last lecture on chapter 9. 01:08:47.790 --> 01:08:50.760 align:middle line:84% We talked about important issues for thin film deposition. 01:08:50.760 --> 01:08:52.660 align:middle line:84% We said there are two different properties, 01:08:52.660 --> 01:08:54.785 align:middle line:84% the physical properties and the chemical properties 01:08:54.785 --> 01:08:55.950 align:middle line:90% of the film. 01:08:55.950 --> 01:08:59.220 align:middle line:84% The coverage of topography, for example, 01:08:59.220 --> 01:09:01.859 align:middle line:84% this conformal deposition and step coverage we said 01:09:01.859 --> 01:09:03.395 align:middle line:90% were very important. 01:09:03.395 --> 01:09:04.770 align:middle line:84% There are two main techniques you 01:09:04.770 --> 01:09:07.140 align:middle line:84% need to keep in mind, chemical vapor and physical vapor 01:09:07.140 --> 01:09:09.890 align:middle line:90% deposition. 01:09:09.890 --> 01:09:12.439 align:middle line:84% If you have simple models, the dep rate 01:09:12.439 --> 01:09:15.410 align:middle line:84% is limited by either the surface reaction rate-- 01:09:15.410 --> 01:09:16.910 align:middle line:84% that's usually at low temperatures-- 01:09:16.910 --> 01:09:20.300 align:middle line:84% or the mass transfer through the gas phase boundary layer. 01:09:20.300 --> 01:09:21.800 align:middle line:84% That's usually at high temperatures. 01:09:21.800 --> 01:09:25.069 align:middle line:90% 01:09:25.069 --> 01:09:27.830 align:middle line:84% But you do have to take into account shadowing and surface 01:09:27.830 --> 01:09:32.029 align:middle line:84% features to model the actual coverage. 01:09:32.029 --> 01:09:34.310 align:middle line:84% For PVD, these processes, again, their names-- 01:09:34.310 --> 01:09:37.290 align:middle line:84% it's very physical rather than chemical. 01:09:37.290 --> 01:09:40.500 align:middle line:84% The arrival angle distribution at the wafer surface 01:09:40.500 --> 01:09:44.130 align:middle line:84% is important, that cosine theta to the n term. 01:09:44.130 --> 01:09:45.960 align:middle line:84% In a lot of these techniques, the species 01:09:45.960 --> 01:09:48.930 align:middle line:84% arrive more vertical to the wafer surface, 01:09:48.930 --> 01:09:52.319 align:middle line:84% and the sticking coefficient tends to be one for PVD. 01:09:52.319 --> 01:09:55.290 align:middle line:84% So PVD has a lot of shadowing issues 01:09:55.290 --> 01:09:59.090 align:middle line:90% that you need to keep in mind. 01:09:59.090 --> 01:10:03.890 align:middle line:84% We model this by having a sticking coefficient s sub c. 01:10:03.890 --> 01:10:08.120 align:middle line:84% And if you have a low sticking coefficient, 01:10:08.120 --> 01:10:11.870 align:middle line:84% say much less than 0.01 or something like that or 0.001, 01:10:11.870 --> 01:10:14.450 align:middle line:84% and you have an isotropic arrival distribution, 01:10:14.450 --> 01:10:17.580 align:middle line:84% you'll get much better filling of holes. 01:10:17.580 --> 01:10:19.000 align:middle line:90% But there are new techniques. 01:10:19.000 --> 01:10:21.420 align:middle line:84% But this only works to a certain extent this filling. 01:10:21.420 --> 01:10:26.550 align:middle line:84% So the latest techniques use these high density plasma 01:10:26.550 --> 01:10:29.220 align:middle line:84% and these ionized sputtering techniques, 01:10:29.220 --> 01:10:32.490 align:middle line:84% which have simultaneously sputtering going on 01:10:32.490 --> 01:10:34.290 align:middle line:90% at the same time as deposition. 01:10:34.290 --> 01:10:37.860 align:middle line:84% And this can be used to get good gap filling and planarization. 01:10:37.860 --> 01:10:42.070 align:middle line:84% So these are the most modern techniques people use today. 01:10:42.070 --> 01:10:43.510 align:middle line:84% There's a lot of things we didn't 01:10:43.510 --> 01:10:46.450 align:middle line:84% get to talk about in chapter 9 in this course. 01:10:46.450 --> 01:10:49.450 align:middle line:84% 6774 is about front end processing, 01:10:49.450 --> 01:10:52.835 align:middle line:84% but we give the lion's share of the time to oxidation diffusion 01:10:52.835 --> 01:10:53.710 align:middle line:90% and ion implantation. 01:10:53.710 --> 01:10:55.930 align:middle line:84% Most of the lectures are on that. 01:10:55.930 --> 01:10:58.810 align:middle line:84% If you really want to know a lot more, if your thesis 01:10:58.810 --> 01:11:01.540 align:middle line:84% research is going to be on thin films, for example, 01:11:01.540 --> 01:11:02.890 align:middle line:90% we didn't cover a lot of things. 01:11:02.890 --> 01:11:05.740 align:middle line:84% We didn't talk at all about the grain structure, the density, 01:11:05.740 --> 01:11:06.883 align:middle line:90% the stress in the film. 01:11:06.883 --> 01:11:08.800 align:middle line:84% None of these things did we have time to cover 01:11:08.800 --> 01:11:11.440 align:middle line:84% and how to model these properties. 01:11:11.440 --> 01:11:14.213 align:middle line:84% There are courses that are more thin film oriented courses, 01:11:14.213 --> 01:11:15.880 align:middle line:84% like in the material science department. 01:11:15.880 --> 01:11:18.400 align:middle line:84% And if you really want to know about thin films 01:11:18.400 --> 01:11:20.650 align:middle line:84% beyond what's in chapter 9, I would 01:11:20.650 --> 01:11:23.020 align:middle line:84% recommend you look towards some of those classes. 01:11:23.020 --> 01:11:24.940 align:middle line:84% But I just wanted to make it clear 01:11:24.940 --> 01:11:28.300 align:middle line:84% that this is sort of the minimum amount of information 01:11:28.300 --> 01:11:30.190 align:middle line:84% you need to know for front end processing 01:11:30.190 --> 01:11:33.850 align:middle line:84% as far as thin film dep and epi goes. 01:11:33.850 --> 01:11:36.550 align:middle line:84% OK, I think that's about all I have 01:11:36.550 --> 01:11:39.580 align:middle line:90% in this for today's lecture. 01:11:39.580 --> 01:11:42.250 align:middle line:84% As I mentioned at the beginning, if you came in late, hopefully 01:11:42.250 --> 01:11:46.240 align:middle line:84% you're working on homework number 5. 01:11:46.240 --> 01:11:47.200 align:middle line:90% It's due Thursday. 01:11:47.200 --> 01:11:49.510 align:middle line:90% And I did bring the clipboard. 01:11:49.510 --> 01:11:52.623 align:middle line:84% My assistant has typed up her version 01:11:52.623 --> 01:11:54.790 align:middle line:84% of what she thinks you're going to be either talking 01:11:54.790 --> 01:11:55.690 align:middle line:90% about or writing up. 01:11:55.690 --> 01:11:56.273 align:middle line:90% You can check. 01:11:56.273 --> 01:11:57.650 align:middle line:90% That I've approved all of them. 01:11:57.650 --> 01:11:59.108 align:middle line:84% There's only one person who doesn't 01:11:59.108 --> 01:12:00.233 align:middle line:90% have anything listed there. 01:12:00.233 --> 01:12:01.941 align:middle line:84% And if you're giving an oral report, what 01:12:01.941 --> 01:12:03.970 align:middle line:84% I'm going to do this week is start scheduling 01:12:03.970 --> 01:12:07.210 align:middle line:84% for the last week of classes so you know which day you 01:12:07.210 --> 01:12:09.420 align:middle line:90% have to give your oral report. 01:12:09.420 --> 01:12:10.980 align:middle line:84% OK, good, well, I'll look forward 01:12:10.980 --> 01:12:13.400 align:middle line:90% to seeing you on Thursday then. 01:12:13.400 --> 01:12:14.000 align:middle line:90%