WEBVTT 00:00:00.000 --> 00:00:01.967 align:middle line:90% [SQUEAKING] 00:00:01.967 --> 00:00:03.423 align:middle line:90% [RUSTLING] 00:00:03.423 --> 00:00:03.923 align:middle line:90% 00:00:03.923 --> 00:00:16.077 align:middle line:90% [CLICKING] 00:00:16.077 --> 00:00:16.660 align:middle line:90% JUEJUN HU: OK. 00:00:16.660 --> 00:00:18.000 align:middle line:90% So we'll get started. 00:00:18.000 --> 00:00:21.260 align:middle line:84% So as I said in the last lecture, 00:00:21.260 --> 00:00:23.940 align:middle line:84% we are going to be discussing about devices, 00:00:23.940 --> 00:00:27.240 align:middle line:84% microelectronic devices made out of amorphous silicon. 00:00:27.240 --> 00:00:27.740 align:middle line:90% OK. 00:00:27.740 --> 00:00:30.600 align:middle line:84% Specifically, we are going to focus on two classes of devices, 00:00:30.600 --> 00:00:34.280 align:middle line:90% display and solar cells. 00:00:34.280 --> 00:00:34.780 align:middle line:90% All right. 00:00:34.780 --> 00:00:36.080 align:middle line:90% Display. 00:00:36.080 --> 00:00:39.280 align:middle line:84% So for display, what we want to do essentially 00:00:39.280 --> 00:00:44.720 align:middle line:84% is that we want to be able to control an array of pixels 00:00:44.720 --> 00:00:47.200 align:middle line:84% so that we can actually control the spatial dependent 00:00:47.200 --> 00:00:48.360 align:middle line:90% light emission pattern. 00:00:48.360 --> 00:00:51.440 align:middle line:84% So the first question we want to address here 00:00:51.440 --> 00:00:53.360 align:middle line:84% is what is the architecture that we 00:00:53.360 --> 00:00:58.300 align:middle line:84% can use to actually realize such a pixel level address array. 00:00:58.300 --> 00:01:01.420 align:middle line:90% 00:01:01.420 --> 00:01:03.670 align:middle line:84% The conventional way of doing a display-- the earliest 00:01:03.670 --> 00:01:06.590 align:middle line:84% days of doing such a display architecture 00:01:06.590 --> 00:01:08.650 align:middle line:84% is to use an architecture called direct drive. 00:01:08.650 --> 00:01:12.230 align:middle line:84% So your direct drive, each pixel is essentially driven using 00:01:12.230 --> 00:01:16.790 align:middle line:84% a pair of electrodes so that if you have, say, 00:01:16.790 --> 00:01:20.470 align:middle line:84% n number of pixels, you need essentially 2n number 00:01:20.470 --> 00:01:23.550 align:middle line:84% of contacts so that you can realize individual addressing 00:01:23.550 --> 00:01:24.770 align:middle line:90% of each of the pixels. 00:01:24.770 --> 00:01:26.070 align:middle line:90% OK. 00:01:26.070 --> 00:01:28.910 align:middle line:84% One example of this kind of direct drive of a display 00:01:28.910 --> 00:01:31.910 align:middle line:84% is if you think about the early days of the liquid crystal 00:01:31.910 --> 00:01:33.110 align:middle line:90% display calculators. 00:01:33.110 --> 00:01:33.610 align:middle line:90% Right. 00:01:33.610 --> 00:01:36.430 align:middle line:84% They use, essentially, this kind of a seven bars 00:01:36.430 --> 00:01:38.190 align:middle line:90% to represent numbers. 00:01:38.190 --> 00:01:40.950 align:middle line:84% In that case, each of these bars is actually 00:01:40.950 --> 00:01:43.510 align:middle line:84% individually controlled by a pair of contacts. 00:01:43.510 --> 00:01:45.590 align:middle line:84% Now you can easily see what's the challenge 00:01:45.590 --> 00:01:47.230 align:middle line:90% with this kind of scheme here. 00:01:47.230 --> 00:01:49.110 align:middle line:90% It's really not scalable. 00:01:49.110 --> 00:01:52.130 align:middle line:84% If you are going to go with, hundreds, thousands, 00:01:52.130 --> 00:01:56.670 align:middle line:84% millions of pixels on a display, you obviously cannot afford 00:01:56.670 --> 00:01:58.800 align:middle line:84% having that large number of electrical contacts. 00:01:58.800 --> 00:01:59.500 align:middle line:90% OK. 00:01:59.500 --> 00:02:02.980 align:middle line:84% So that's why, to realize this kind of large scale 2D array 00:02:02.980 --> 00:02:08.300 align:middle line:84% of pixels, people use two sorts of architectures here, 00:02:08.300 --> 00:02:10.979 align:middle line:84% essentially called passive matrix and active matrix. 00:02:10.979 --> 00:02:11.700 align:middle line:90% OK. 00:02:11.700 --> 00:02:15.680 align:middle line:84% Both cases use a crossbar architecture. 00:02:15.680 --> 00:02:24.260 align:middle line:84% So this is actually quite similar to the obtain memory we 00:02:24.260 --> 00:02:26.000 align:middle line:84% talked about, the phase change memory, 00:02:26.000 --> 00:02:30.540 align:middle line:84% the 3D cross point memory we discussed about when we talked 00:02:30.540 --> 00:02:32.660 align:middle line:84% about phase change memory devices. 00:02:32.660 --> 00:02:33.220 align:middle line:90% OK. 00:02:33.220 --> 00:02:37.940 align:middle line:84% Essentially, you have electrical wires 00:02:37.940 --> 00:02:40.780 align:middle line:84% that goes in two orthogonal directions, horizontal 00:02:40.780 --> 00:02:42.860 align:middle line:90% and vertical directions. 00:02:42.860 --> 00:02:45.740 align:middle line:84% And then you can think of them as this wire 00:02:45.740 --> 00:02:47.260 align:middle line:90% goes in two different planes. 00:02:47.260 --> 00:02:49.220 align:middle line:90% OK. 00:02:49.220 --> 00:02:51.900 align:middle line:84% And then at the cross point between each horizontal and 00:02:51.900 --> 00:02:55.460 align:middle line:84% vertical wire, you have one pixel essentially sitting 00:02:55.460 --> 00:02:56.180 align:middle line:90% in between here. 00:02:56.180 --> 00:02:56.680 align:middle line:90% OK. 00:02:56.680 --> 00:03:00.400 align:middle line:84% So at each cross point, there's essentially one pixel. 00:03:00.400 --> 00:03:02.480 align:middle line:84% You can then light up each of the pixel 00:03:02.480 --> 00:03:07.880 align:middle line:84% by applying a bias across a pair of corresponding horizontal and 00:03:07.880 --> 00:03:09.120 align:middle line:90% vertical wires there. 00:03:09.120 --> 00:03:11.680 align:middle line:84% So this is the basic architecture 00:03:11.680 --> 00:03:15.160 align:middle line:84% for both passive and active matrix displays. 00:03:15.160 --> 00:03:17.560 align:middle line:84% Now the difference between passive and active matrix 00:03:17.560 --> 00:03:24.280 align:middle line:84% displays is that, for a passive matrix, 00:03:24.280 --> 00:03:28.452 align:middle line:84% there's no individual switch that's co-located 00:03:28.452 --> 00:03:29.660 align:middle line:90% with each of the pixels here. 00:03:29.660 --> 00:03:31.760 align:middle line:84% So for example, let's say if you want to light up 00:03:31.760 --> 00:03:35.040 align:middle line:84% say this pixel, what you want to do 00:03:35.040 --> 00:03:36.960 align:middle line:84% is you want to apply a bias essentially 00:03:36.960 --> 00:03:38.260 align:middle line:90% across these two wires here. 00:03:38.260 --> 00:03:39.960 align:middle line:90% It's going to light it up. 00:03:39.960 --> 00:03:40.600 align:middle line:90% OK. 00:03:40.600 --> 00:03:43.120 align:middle line:84% Now suppose if you want to light up another pixel, 00:03:43.120 --> 00:03:46.920 align:middle line:84% let's say maybe this one, if you want to light up this one, 00:03:46.920 --> 00:03:49.200 align:middle line:84% then what you have to do is you essentially apply bias 00:03:49.200 --> 00:03:50.240 align:middle line:90% across here and here. 00:03:50.240 --> 00:03:53.780 align:middle line:90% But now you see the problem. 00:03:53.780 --> 00:03:56.510 align:middle line:84% In addition to lighting up these two target pixels, 00:03:56.510 --> 00:04:01.750 align:middle line:84% you also light up this pixel and this pixel. 00:04:01.750 --> 00:04:04.830 align:middle line:84% It's also, essentially, you also apply bias across that two. 00:04:04.830 --> 00:04:07.110 align:middle line:84% So what this means is that if you 00:04:07.110 --> 00:04:11.590 align:middle line:84% don't have an individual control located within each pixel, 00:04:11.590 --> 00:04:14.110 align:middle line:84% then the best you can do with this kind of passive matrix 00:04:14.110 --> 00:04:19.470 align:middle line:84% display is to do a row by row scanning. 00:04:19.470 --> 00:04:22.050 align:middle line:84% So what you would do, you essentially, for example, 00:04:22.050 --> 00:04:25.310 align:middle line:84% apply bias across one of the horizontal lines. 00:04:25.310 --> 00:04:29.910 align:middle line:84% These are actually called scan lines. 00:04:29.910 --> 00:04:32.390 align:middle line:90% OK. 00:04:32.390 --> 00:04:34.390 align:middle line:84% Then depending on which pixel you 00:04:34.390 --> 00:04:38.150 align:middle line:84% want to light up in that row, you then correspondingly apply 00:04:38.150 --> 00:04:40.967 align:middle line:90% the bias on the vertical lines. 00:04:40.967 --> 00:04:42.550 align:middle line:84% The vertical lines, these are actually 00:04:42.550 --> 00:04:46.470 align:middle line:90% what are called the data lines. 00:04:46.470 --> 00:04:50.770 align:middle line:84% So now, for example, display these two pixels. 00:04:50.770 --> 00:04:51.530 align:middle line:90% They light up. 00:04:51.530 --> 00:04:52.450 align:middle line:90% Then you move on. 00:04:52.450 --> 00:04:54.580 align:middle line:84% You turn off this one, this scan line. 00:04:54.580 --> 00:04:57.020 align:middle line:84% You move on to the next scan line, apply bias, 00:04:57.020 --> 00:04:58.600 align:middle line:84% and then you display the next line. 00:04:58.600 --> 00:05:00.360 align:middle line:84% So you do this kind line by line scan. 00:05:00.360 --> 00:05:01.760 align:middle line:84% That's why it's called scan line. 00:05:01.760 --> 00:05:03.500 align:middle line:84% Data line, essentially, gives you the actual data 00:05:03.500 --> 00:05:04.840 align:middle line:90% or the actual display pattern. 00:05:04.840 --> 00:05:09.822 align:middle line:90% 00:05:09.822 --> 00:05:11.660 align:middle line:84% So the advantage of this kind of scheme 00:05:11.660 --> 00:05:13.500 align:middle line:84% here is that you can significantly 00:05:13.500 --> 00:05:16.260 align:middle line:84% simplify the electrical contact structure. 00:05:16.260 --> 00:05:19.540 align:middle line:84% If you have an m by n pixel array, 00:05:19.540 --> 00:05:23.648 align:middle line:84% you only need m plus n number of contacts. 00:05:23.648 --> 00:05:25.940 align:middle line:84% So this is vastly simplified compared to a direct drive 00:05:25.940 --> 00:05:27.060 align:middle line:90% display. 00:05:27.060 --> 00:05:29.740 align:middle line:84% Downside for this kind of passive matrix display, 00:05:29.740 --> 00:05:33.420 align:middle line:84% obviously, it it's slow because you can only 00:05:33.420 --> 00:05:35.280 align:middle line:90% display patterns line by line. 00:05:35.280 --> 00:05:42.060 align:middle line:90% 00:05:42.060 --> 00:05:45.420 align:middle line:84% And that's why, if you look at state of the art displays people 00:05:45.420 --> 00:05:50.180 align:middle line:84% use, most of the time, you'll be looking at an active matrix 00:05:50.180 --> 00:05:51.020 align:middle line:90% display. 00:05:51.020 --> 00:05:52.180 align:middle line:90% OK. 00:05:52.180 --> 00:05:55.730 align:middle line:84% So active matrix display is also a crossbar structure. 00:05:55.730 --> 00:05:59.790 align:middle line:84% But the main difference here is that, within each pixel, 00:05:59.790 --> 00:06:02.410 align:middle line:84% you actually integrate a transistor, a thin film 00:06:02.410 --> 00:06:03.810 align:middle line:90% transistor, TFT. 00:06:03.810 --> 00:06:05.930 align:middle line:90% OK. 00:06:05.930 --> 00:06:08.440 align:middle line:90% So TFT. 00:06:08.440 --> 00:06:12.530 align:middle line:84% A thin film transistor is basically an electrical switch. 00:06:12.530 --> 00:06:16.770 align:middle line:84% You can use it to essentially control a pixel being on or off. 00:06:16.770 --> 00:06:20.410 align:middle line:84% So therefore, you will be able to individually control 00:06:20.410 --> 00:06:22.810 align:middle line:84% all the pixels here without having 00:06:22.810 --> 00:06:24.850 align:middle line:84% to worry about the crosstalk, as you see here 00:06:24.850 --> 00:06:27.310 align:middle line:84% in the kind of passive matrix arrays. 00:06:27.310 --> 00:06:27.810 align:middle line:90% OK. 00:06:27.810 --> 00:06:32.050 align:middle line:90% 00:06:32.050 --> 00:06:34.250 align:middle line:90% Now the TFTs. 00:06:34.250 --> 00:06:38.010 align:middle line:84% This has to be made out of a semiconductor. 00:06:38.010 --> 00:06:39.450 align:middle line:84% And because most displays, they're 00:06:39.450 --> 00:06:41.450 align:middle line:84% actually made on glass substrates. 00:06:41.450 --> 00:06:44.558 align:middle line:84% So as we have mentioned in the previous lectures, 00:06:44.558 --> 00:06:46.850 align:middle line:84% you will not be able to grow single crystalline silicon 00:06:46.850 --> 00:06:48.490 align:middle line:90% on amorphous glass substrate. 00:06:48.490 --> 00:06:51.210 align:middle line:84% So that's why you have to look into amorphous 00:06:51.210 --> 00:06:52.418 align:middle line:90% or polycrystalline materials. 00:06:52.418 --> 00:06:54.627 align:middle line:84% And that's where actually amorphous silicon comes in. 00:06:54.627 --> 00:06:57.190 align:middle line:84% They are the material that's being used to make the thin film 00:06:57.190 --> 00:06:58.010 align:middle line:90% transistors. 00:06:58.010 --> 00:07:07.178 align:middle line:90% 00:07:07.178 --> 00:07:11.230 align:middle line:84% So that is the basic structure of the display arrays. 00:07:11.230 --> 00:07:13.630 align:middle line:84% Now, in this next slide here, I'm showing you, 00:07:13.630 --> 00:07:15.910 align:middle line:84% essentially, the specific architecture 00:07:15.910 --> 00:07:18.070 align:middle line:90% within each individual pixel. 00:07:18.070 --> 00:07:21.310 align:middle line:84% So for active matrix display, there 00:07:21.310 --> 00:07:23.150 align:middle line:84% are, in general, two main variants 00:07:23.150 --> 00:07:26.910 align:middle line:84% there, active matrix liquid crystal display or AMLCD, 00:07:26.910 --> 00:07:30.050 align:middle line:84% or Active Matrix Organic Light Emitting Diode, AMOLED. 00:07:30.050 --> 00:07:35.510 align:middle line:84% So if you look at the layer structure for liquid crystal 00:07:35.510 --> 00:07:38.590 align:middle line:84% display pixel, what you have backlight. 00:07:38.590 --> 00:07:40.850 align:middle line:84% Backlight essentially is a large area light source. 00:07:40.850 --> 00:07:42.230 align:middle line:90% It's constantly on. 00:07:42.230 --> 00:07:43.830 align:middle line:90% It's always emitting light. 00:07:43.830 --> 00:07:44.590 align:middle line:90% OK. 00:07:44.590 --> 00:07:45.610 align:middle line:90% White light, basically. 00:07:45.610 --> 00:07:45.750 align:middle line:90% OK. 00:07:45.750 --> 00:07:46.910 align:middle line:90% Broadband white light. 00:07:46.910 --> 00:07:50.480 align:middle line:84% It goes through a linear polarizer, 00:07:50.480 --> 00:07:53.700 align:middle line:84% which cuts immediately 50% of the light. 00:07:53.700 --> 00:07:55.620 align:middle line:84% It's only when you actually-- usually, 00:07:55.620 --> 00:07:57.375 align:middle line:84% the light source is not polarized. 00:07:57.375 --> 00:07:59.000 align:middle line:84% So when it goes through this polarizer, 00:07:59.000 --> 00:08:00.703 align:middle line:84% you lose 50% of linear polarized light. 00:08:00.703 --> 00:08:02.620 align:middle line:84% We'll explain why we actually need to do that. 00:08:02.620 --> 00:08:03.140 align:middle line:90% OK. 00:08:03.140 --> 00:08:05.060 align:middle line:84% Then it goes through a TFT backplane. 00:08:05.060 --> 00:08:08.020 align:middle line:84% So this is, essentially, on the glass substrate 00:08:08.020 --> 00:08:10.700 align:middle line:84% where you have the thin film transistors made 00:08:10.700 --> 00:08:12.437 align:middle line:84% of amorphous silicon or polysilicon, 00:08:12.437 --> 00:08:13.520 align:middle line:90% as we'll talk about later. 00:08:13.520 --> 00:08:16.340 align:middle line:84% And then the light continues to transmit through and then 00:08:16.340 --> 00:08:17.757 align:middle line:84% go through a liquid crystal layer. 00:08:17.757 --> 00:08:20.132 align:middle line:84% So you probably have already learned about liquid crystal 00:08:20.132 --> 00:08:20.632 align:middle line:90% materials. 00:08:20.632 --> 00:08:21.132 align:middle line:90% Right. 00:08:21.132 --> 00:08:23.300 align:middle line:84% This is the kind of material where they are actually 00:08:23.300 --> 00:08:24.960 align:middle line:90% intrinsically optically chiral. 00:08:24.960 --> 00:08:31.340 align:middle line:84% So when I say they are chiral, essentially 00:08:31.340 --> 00:08:33.299 align:middle line:84% means that they have this unique property 00:08:33.299 --> 00:08:36.480 align:middle line:84% that the polarization of light as a propagating liquid crystal, 00:08:36.480 --> 00:08:40.220 align:middle line:84% the polarization plane actually rotates as it propagates. 00:08:40.220 --> 00:08:42.179 align:middle line:84% And you can actually electrically control 00:08:42.179 --> 00:08:44.600 align:middle line:84% the chirality of the liquid crystal 00:08:44.600 --> 00:08:47.140 align:middle line:84% so that you can control the extent of this polarization 00:08:47.140 --> 00:08:48.460 align:middle line:90% plane rotation. 00:08:48.460 --> 00:08:49.020 align:middle line:90% OK. 00:08:49.020 --> 00:08:53.090 align:middle line:84% So for example, suppose that you don't apply any electrical bias. 00:08:53.090 --> 00:08:57.450 align:middle line:84% Then you have, essentially, say 0 degree by rotation. 00:08:57.450 --> 00:09:00.570 align:middle line:84% If you apply bias, you can have 90 degree rotation. 00:09:00.570 --> 00:09:04.650 align:middle line:84% So then if there is no electric bias, 00:09:04.650 --> 00:09:06.510 align:middle line:84% then the polarization plane will not change. 00:09:06.510 --> 00:09:08.890 align:middle line:84% It will just go through all the top structures 00:09:08.890 --> 00:09:11.370 align:middle line:84% and then directly go to the top polarizer. 00:09:11.370 --> 00:09:14.570 align:middle line:84% The top polarizer is set up in an orientation 00:09:14.570 --> 00:09:18.090 align:middle line:84% that's 90 degrees with respect to the bottom polarizer. 00:09:18.090 --> 00:09:21.450 align:middle line:84% So if you don't have any light polarization, 00:09:21.450 --> 00:09:23.810 align:middle line:84% then all the light coming from bottom polarizer 00:09:23.810 --> 00:09:25.950 align:middle line:84% is going to get blocked by the top right polarizer. 00:09:25.950 --> 00:09:26.910 align:middle line:90% So it's not going to go through. 00:09:26.910 --> 00:09:27.690 align:middle line:90% It's stuck. 00:09:27.690 --> 00:09:29.370 align:middle line:90% So the pixel is not lighting up. 00:09:29.370 --> 00:09:31.010 align:middle line:84% On the other hand, if you apply a bias, 00:09:31.010 --> 00:09:33.030 align:middle line:84% you rotate the liquid crystal molecules 00:09:33.030 --> 00:09:35.530 align:middle line:84% so that you actually turn the polarization plane of light 00:09:35.530 --> 00:09:36.610 align:middle line:90% by 90 degrees. 00:09:36.610 --> 00:09:39.090 align:middle line:84% Then that will align with polarization plane 00:09:39.090 --> 00:09:40.203 align:middle line:90% of the polarizer. 00:09:40.203 --> 00:09:41.870 align:middle line:84% Then you get light transmission through. 00:09:41.870 --> 00:09:44.010 align:middle line:84% So the pixel will light up in that case. 00:09:44.010 --> 00:09:45.970 align:middle line:84% So that's how you actually use, essentially, 00:09:45.970 --> 00:09:48.880 align:middle line:84% this kind of electric bias across the liquid crystal 00:09:48.880 --> 00:09:52.360 align:middle line:90% to control light transmission. 00:09:52.360 --> 00:09:54.160 align:middle line:84% You can think of liquid crystal essentially 00:09:54.160 --> 00:09:58.200 align:middle line:84% as a tunable filter that allows you to control light on or off. 00:09:58.200 --> 00:10:01.597 align:middle line:84% So then the next part is that you have color filter. 00:10:01.597 --> 00:10:03.680 align:middle line:84% For color displays, we need to essentially make up 00:10:03.680 --> 00:10:06.000 align:middle line:84% all the visible color using three primary colors-- 00:10:06.000 --> 00:10:07.400 align:middle line:90% red, green, blue. 00:10:07.400 --> 00:10:10.840 align:middle line:84% So the way to do it in liquid crystal display 00:10:10.840 --> 00:10:13.280 align:middle line:84% is to have three kind of color filters 00:10:13.280 --> 00:10:16.580 align:middle line:84% which, again, cuts off a lot of transmitted light 00:10:16.580 --> 00:10:19.740 align:middle line:84% because when you have a red light filter, 00:10:19.740 --> 00:10:22.360 align:middle line:84% you essentially throw away all the green and blue light. 00:10:22.360 --> 00:10:23.860 align:middle line:84% Same thing for the other two colors. 00:10:23.860 --> 00:10:28.225 align:middle line:84% So again, you're losing lots of light in this process here. 00:10:28.225 --> 00:10:30.100 align:middle line:84% Finally, then you go through the color glass. 00:10:30.100 --> 00:10:32.480 align:middle line:84% You go through a polarizer and then it comes out. 00:10:32.480 --> 00:10:33.060 align:middle line:90% Right. 00:10:33.060 --> 00:10:37.400 align:middle line:90% 00:10:37.400 --> 00:10:41.480 align:middle line:84% So what we're seeing here, AMLCD is basically a tunable light 00:10:41.480 --> 00:10:44.040 align:middle line:90% filter. 00:10:44.040 --> 00:10:49.930 align:middle line:84% It allows this voltage gated selective transmission of light. 00:10:49.930 --> 00:10:51.930 align:middle line:84% And that's how you actually see a color pattern. 00:10:51.930 --> 00:10:56.670 align:middle line:90% 00:10:56.670 --> 00:10:57.710 align:middle line:90% All right. 00:10:57.710 --> 00:10:59.230 align:middle line:84% Now what we have also seen here is 00:10:59.230 --> 00:11:04.350 align:middle line:84% that this kind of architecture actually wastes a lot of light 00:11:04.350 --> 00:11:06.603 align:middle line:84% from the polarizers, from the color filters. 00:11:06.603 --> 00:11:08.770 align:middle line:84% You really have to throw away a lot of useful light. 00:11:08.770 --> 00:11:15.350 align:middle line:84% And also the backlight source, this light is always on. 00:11:15.350 --> 00:11:15.850 align:middle line:90% Right. 00:11:15.850 --> 00:11:16.868 align:middle line:90% It always stays on. 00:11:16.868 --> 00:11:18.910 align:middle line:84% So that means that there is also a constant power 00:11:18.910 --> 00:11:20.610 align:middle line:84% supply you have to supply to the light source, 00:11:20.610 --> 00:11:23.193 align:middle line:84% regardless of whether or not a pixel is light up or it's dark. 00:11:23.193 --> 00:11:27.690 align:middle line:90% 00:11:27.690 --> 00:11:29.510 align:middle line:84% A more energy efficient alternative 00:11:29.510 --> 00:11:32.790 align:middle line:84% to this kind of AMLCD is so-called Active Matrix Organic 00:11:32.790 --> 00:11:34.330 align:middle line:90% Light Emitting Diode, AMOLED. 00:11:34.330 --> 00:11:42.070 align:middle line:84% This is also maybe a more advanced, more recent generation 00:11:42.070 --> 00:11:44.970 align:middle line:84% of displays that we're seeing also in high displays right now, 00:11:44.970 --> 00:11:47.130 align:middle line:84% especially in small consumer electronic devices 00:11:47.130 --> 00:11:48.490 align:middle line:90% like smartphones. 00:11:48.490 --> 00:11:51.250 align:middle line:84% For AMOLED, what you would do, you actually directly 00:11:51.250 --> 00:11:53.130 align:middle line:84% have this organic light emitting diodes. 00:11:53.130 --> 00:11:55.450 align:middle line:84% You can process them so that they directly 00:11:55.450 --> 00:11:56.590 align:middle line:90% emit the different colors. 00:11:56.590 --> 00:11:58.590 align:middle line:84% So for example, if you want to get three colors, 00:11:58.590 --> 00:12:01.890 align:middle line:84% you have one LED that emits a red, one emit a green, one emit 00:12:01.890 --> 00:12:02.930 align:middle line:90% blue. 00:12:02.930 --> 00:12:04.660 align:middle line:90% If you use them, you apply bias. 00:12:04.660 --> 00:12:05.910 align:middle line:90% You send current through them. 00:12:05.910 --> 00:12:06.810 align:middle line:90% They are going to light up. 00:12:06.810 --> 00:12:08.290 align:middle line:84% When you don't use them, you just simply 00:12:08.290 --> 00:12:09.830 align:middle line:84% don't send the current through them. 00:12:09.830 --> 00:12:11.330 align:middle line:90% They are going to be dark. 00:12:11.330 --> 00:12:13.330 align:middle line:90% OK. 00:12:13.330 --> 00:12:16.410 align:middle line:84% So you see that in this regard, AMOLED 00:12:16.410 --> 00:12:19.252 align:middle line:84% can be more energy efficient compared to AMLCD. 00:12:19.252 --> 00:12:21.710 align:middle line:84% Number one, you don't have that many light loss mechanisms. 00:12:21.710 --> 00:12:23.490 align:middle line:84% Number two, you don't have a backlight source 00:12:23.490 --> 00:12:24.270 align:middle line:90% that's always on. 00:12:24.270 --> 00:12:26.210 align:middle line:84% You only turn on LED when you actually need 00:12:26.210 --> 00:12:27.710 align:middle line:90% the display pixel to light up. 00:12:27.710 --> 00:12:30.562 align:middle line:90% 00:12:30.562 --> 00:12:32.490 align:middle line:84% Now you may also notice here there's 00:12:32.490 --> 00:12:34.510 align:middle line:90% still a polarizer on top. 00:12:34.510 --> 00:12:36.350 align:middle line:84% You may wonder why you need a polarizer. 00:12:36.350 --> 00:12:37.990 align:middle line:84% Again, this is actually a circular polarizer 00:12:37.990 --> 00:12:38.615 align:middle line:90% but same thing. 00:12:38.615 --> 00:12:41.370 align:middle line:84% If a circular polarizer, you throw away 50% of light. 00:12:41.370 --> 00:12:43.070 align:middle line:84% You have two circular polarizations. 00:12:43.070 --> 00:12:44.950 align:middle line:84% One is left-handed circular polarization. 00:12:44.950 --> 00:12:46.700 align:middle line:84% One is right-handed circular polarization. 00:12:46.700 --> 00:12:49.400 align:middle line:84% So if you have such a polarizer, again you 00:12:49.400 --> 00:12:50.880 align:middle line:90% throw away 50% of light. 00:12:50.880 --> 00:12:54.040 align:middle line:84% You may wonder why you need polarizer there. 00:12:54.040 --> 00:12:57.120 align:middle line:90% 00:12:57.120 --> 00:12:59.940 align:middle line:84% Now, the reason you need this polarizer here in this case 00:12:59.940 --> 00:13:02.720 align:middle line:84% is actually because it's customary 00:13:02.720 --> 00:13:06.880 align:middle line:84% that behind this TFT backplane, there's usually a metal mirror 00:13:06.880 --> 00:13:08.418 align:middle line:90% metal reflector. 00:13:08.418 --> 00:13:09.960 align:middle line:84% The reason you want a metal reflector 00:13:09.960 --> 00:13:14.240 align:middle line:84% there is that because in OLED, the light emission is isotropic. 00:13:14.240 --> 00:13:16.568 align:middle line:84% So 50% of light emitted is actually downwards. 00:13:16.568 --> 00:13:18.360 align:middle line:84% If you have a metal mirror, you essentially 00:13:18.360 --> 00:13:20.720 align:middle line:84% reclaim that 50% of emitted light. 00:13:20.720 --> 00:13:22.713 align:middle line:84% But now what happens-- suppose you 00:13:22.713 --> 00:13:24.880 align:middle line:84% don't have the polarizer-- is that if you are trying 00:13:24.880 --> 00:13:28.600 align:middle line:84% to use your display, especially in a bright ambient environment, 00:13:28.600 --> 00:13:30.980 align:middle line:84% then the ambient light is going to go through a screen. 00:13:30.980 --> 00:13:31.963 align:middle line:90% It's all transparent. 00:13:31.963 --> 00:13:33.880 align:middle line:84% It's going to go all the way to the reflector, 00:13:33.880 --> 00:13:36.400 align:middle line:84% [INAUDIBLE] reflector, and then comes back. 00:13:36.400 --> 00:13:39.760 align:middle line:84% So this gives rise to a very bright background 00:13:39.760 --> 00:13:41.240 align:middle line:90% due to the ambient light. 00:13:41.240 --> 00:13:41.740 align:middle line:90% OK. 00:13:41.740 --> 00:13:42.870 align:middle line:90% You don't have a dark screen. 00:13:42.870 --> 00:13:45.328 align:middle line:84% And this will severely decrease the contrast of the display 00:13:45.328 --> 00:13:47.510 align:middle line:90% pattern you actually see. 00:13:47.510 --> 00:13:50.170 align:middle line:84% So that's why they add this circular polarizer there. 00:13:50.170 --> 00:13:53.312 align:middle line:84% And the idea is that if you have any ambient light that comes in 00:13:53.312 --> 00:13:54.770 align:middle line:84% after it goes through a polarizer-- 00:13:54.770 --> 00:13:57.770 align:middle line:84% let's say polarizer is, say, left-hand circularly polarized. 00:13:57.770 --> 00:13:58.350 align:middle line:90% OK. 00:13:58.350 --> 00:14:01.310 align:middle line:84% So then the light is going to go through a metal mirror 00:14:01.310 --> 00:14:03.490 align:middle line:84% where it gets reflected by the metal mirror. 00:14:03.490 --> 00:14:06.350 align:middle line:84% Its handedness actually gets flipped. 00:14:06.350 --> 00:14:08.830 align:middle line:84% So from left hand polarized, it becomes a right hand 00:14:08.830 --> 00:14:10.190 align:middle line:90% circularly polarized. 00:14:10.190 --> 00:14:11.830 align:middle line:84% So when it comes up, this reflect 00:14:11.830 --> 00:14:15.230 align:middle line:84% light is going to get blocked by this circular polarizer 00:14:15.230 --> 00:14:18.430 align:middle line:84% so you don't see, essentially, this kind of reflected light. 00:14:18.430 --> 00:14:22.190 align:middle line:84% And that's why, if you look at the typical display screen, 00:14:22.190 --> 00:14:24.570 align:middle line:84% despite the presence of this reflector there, 00:14:24.570 --> 00:14:26.482 align:middle line:90% the screen typically looks dark. 00:14:26.482 --> 00:14:28.190 align:middle line:84% That's because a lot of the ambient light 00:14:28.190 --> 00:14:29.690 align:middle line:90% is actually getting blocked. 00:14:29.690 --> 00:14:30.950 align:middle line:84% The only reflected light you are seeing here 00:14:30.950 --> 00:14:33.408 align:middle line:84% is actually direct reflection on the top surface, the glass 00:14:33.408 --> 00:14:33.950 align:middle line:90% cover. 00:14:33.950 --> 00:14:35.390 align:middle line:84% The glass cover will reflect a little bit. 00:14:35.390 --> 00:14:36.848 align:middle line:84% That's the only reflection you see. 00:14:36.848 --> 00:14:40.750 align:middle line:84% That's why the screen appears mostly dark OK. 00:14:40.750 --> 00:14:46.460 align:middle line:84% OK, so in both cases, you always have this TFT backplane there. 00:14:46.460 --> 00:14:48.580 align:middle line:84% This is the backplane where you have an array 00:14:48.580 --> 00:14:50.660 align:middle line:90% of these thin-film transistors. 00:14:50.660 --> 00:14:53.993 align:middle line:84% Each transistor controls one pixel. 00:14:53.993 --> 00:14:55.660 align:middle line:84% This is how you control on/off switching 00:14:55.660 --> 00:14:58.140 align:middle line:90% of all the pixel arrays here. 00:14:58.140 --> 00:14:59.860 align:middle line:84% And as I said, these things are made 00:14:59.860 --> 00:15:02.700 align:middle line:84% of amorphous silicon, or low-temperature polysilicon, 00:15:02.700 --> 00:15:04.158 align:middle line:90% or LPPS. 00:15:04.158 --> 00:15:06.700 align:middle line:84% We're going to discuss about these materials in a little bit. 00:15:06.700 --> 00:15:07.820 align:middle line:84% Now before I go to that, I just want 00:15:07.820 --> 00:15:10.020 align:middle line:84% to quickly mention about the advantage and disadvantage 00:15:10.020 --> 00:15:11.728 align:middle line:84% of these two kinds of architectures here. 00:15:11.728 --> 00:15:15.980 align:middle line:84% So if you look at the AMLCD, you can 00:15:15.980 --> 00:15:17.980 align:middle line:84% see that it has a more complex layer structure. 00:15:17.980 --> 00:15:18.740 align:middle line:90% It's thicker. 00:15:18.740 --> 00:15:21.460 align:middle line:84% And as I said, it tends to be a less energy efficient compared 00:15:21.460 --> 00:15:24.320 align:middle line:90% to AMOLED. 00:15:24.320 --> 00:15:26.960 align:middle line:84% But AMLCD is a relatively mature technology. 00:15:26.960 --> 00:15:28.480 align:middle line:90% It's been very well-established. 00:15:28.480 --> 00:15:29.880 align:middle line:90% It's also relatively low cost. 00:15:29.880 --> 00:15:33.580 align:middle line:84% So that's why it's still used in a lot of more low-end display 00:15:33.580 --> 00:15:34.340 align:middle line:90% applications. 00:15:34.340 --> 00:15:36.040 align:middle line:84% For more high-end electronic devices, 00:15:36.040 --> 00:15:39.360 align:middle line:84% people now tend to use OLED instead of AMLCDs. 00:15:39.360 --> 00:15:43.120 align:middle line:90% 00:15:43.120 --> 00:15:43.660 align:middle line:90% OK. 00:15:43.660 --> 00:15:46.400 align:middle line:84% So now, let's take a look at-- this 00:15:46.400 --> 00:15:48.140 align:middle line:84% is actually what the display looks like. 00:15:48.140 --> 00:15:49.960 align:middle line:84% What I'm showing you here, this is actually 00:15:49.960 --> 00:15:52.440 align:middle line:84% a little optical micrograph of one 00:15:52.440 --> 00:15:56.920 align:middle line:84% of the older generations of iPhone displays, 00:15:56.920 --> 00:16:00.353 align:middle line:84% something like iPhone 1 or iPhone 2, where they use AMLCD. 00:16:00.353 --> 00:16:02.020 align:middle line:84% If you look under an optical microscope, 00:16:02.020 --> 00:16:03.500 align:middle line:90% you clearly see the pixels here. 00:16:03.500 --> 00:16:05.280 align:middle line:84% These are actually three-color pixels. 00:16:05.280 --> 00:16:08.560 align:middle line:84% You see some black shadow right here. 00:16:08.560 --> 00:16:12.280 align:middle line:84% These are actually the thin-film transistors. 00:16:12.280 --> 00:16:13.840 align:middle line:84% Now amorphous silicon is actually 00:16:13.840 --> 00:16:15.340 align:middle line:84% absorbing in the visible spectrum. 00:16:15.340 --> 00:16:17.220 align:middle line:84% That's why people can make solar cells out of them. 00:16:17.220 --> 00:16:18.470 align:middle line:90% So that's why they are opaque. 00:16:18.470 --> 00:16:19.420 align:middle line:90% And you can see that. 00:16:19.420 --> 00:16:22.880 align:middle line:84% It blocks some of the light there. 00:16:22.880 --> 00:16:26.120 align:middle line:84% So you will not be able to see the individual pixels 00:16:26.120 --> 00:16:29.360 align:middle line:84% in more recent generations of displays because they have 00:16:29.360 --> 00:16:30.943 align:middle line:90% what's called a retina display. 00:16:30.943 --> 00:16:33.360 align:middle line:84% "Retina display" essentially means that the human eye will 00:16:33.360 --> 00:16:35.180 align:middle line:84% not be able to resolve individual pixels. 00:16:35.180 --> 00:16:38.790 align:middle line:84% So you won't be able to see individual colors. 00:16:38.790 --> 00:16:40.870 align:middle line:84% However, there's actually a very simple trick 00:16:40.870 --> 00:16:43.990 align:middle line:84% that allows you to see the individual color pixels. 00:16:43.990 --> 00:16:45.450 align:middle line:84% So what I have here is a pipette. 00:16:45.450 --> 00:16:46.590 align:middle line:90% This is water. 00:16:46.590 --> 00:16:48.890 align:middle line:84% Let me try to put some small water droplets on top. 00:16:48.890 --> 00:17:01.830 align:middle line:90% 00:17:01.830 --> 00:17:03.950 align:middle line:90% It's not easy, but OK. 00:17:03.950 --> 00:17:05.369 align:middle line:84% I'm going to pass my phone around. 00:17:05.369 --> 00:17:07.270 align:middle line:84% You can try that using your phone as well. 00:17:07.270 --> 00:17:09.349 align:middle line:84% But if you look at the water droplets, 00:17:09.349 --> 00:17:12.990 align:middle line:84% because of small meniscus, they serve essentially like a lens. 00:17:12.990 --> 00:17:15.170 align:middle line:84% You can clearly see the magnified pixels. 00:17:15.170 --> 00:17:16.651 align:middle line:90% Three colors. 00:17:16.651 --> 00:17:17.609 align:middle line:90% Are you able to see it? 00:17:17.609 --> 00:17:23.630 align:middle line:90% 00:17:23.630 --> 00:17:24.670 align:middle line:90% Good. 00:17:24.670 --> 00:17:26.470 align:middle line:90% All right. 00:17:26.470 --> 00:17:29.390 align:middle line:84% But now if you look into what actually 00:17:29.390 --> 00:17:31.593 align:middle line:84% goes into this little transistor there, 00:17:31.593 --> 00:17:34.010 align:middle line:84% this is a cross-sectional structure of what it looks like. 00:17:34.010 --> 00:17:37.780 align:middle line:84% So you basically have a substrate. 00:17:37.780 --> 00:17:39.480 align:middle line:84% This will be some kind of display glass, 00:17:39.480 --> 00:17:42.980 align:middle line:84% typically made out of a sodium-free glasses that 00:17:42.980 --> 00:17:45.040 align:middle line:84% were actually prepared using a downdraw method. 00:17:45.040 --> 00:17:46.860 align:middle line:84% If you recall what the downdraw method is, 00:17:46.860 --> 00:17:48.360 align:middle line:84% this is a technique people developed 00:17:48.360 --> 00:17:51.900 align:middle line:84% so that you can actually prepare large flat glass sheets 00:17:51.900 --> 00:17:53.308 align:middle line:90% without surface contamination. 00:17:53.308 --> 00:17:55.100 align:middle line:84% The surface is not in contact with any kind 00:17:55.100 --> 00:17:56.020 align:middle line:90% of other materials. 00:17:56.020 --> 00:17:57.820 align:middle line:90% It's basically suspended in air. 00:17:57.820 --> 00:18:01.260 align:middle line:84% So those are what's designed for mainly stringent requirements 00:18:01.260 --> 00:18:02.780 align:middle line:90% of the display industry. 00:18:02.780 --> 00:18:09.560 align:middle line:84% On top, you have a gate, which basically is metal. 00:18:09.560 --> 00:18:12.560 align:middle line:84% Essentially some kind of an electrical conductor, 00:18:12.560 --> 00:18:14.300 align:middle line:90% some kind of electrical contact. 00:18:14.300 --> 00:18:18.140 align:middle line:84% Then you have a gate material, gate dielectric. 00:18:18.140 --> 00:18:21.540 align:middle line:84% The gate dielectric typically is silicon nitride 00:18:21.540 --> 00:18:24.180 align:middle line:84% that's deposited using PECVD in the same process 00:18:24.180 --> 00:18:25.400 align:middle line:90% as amorphous silicon. 00:18:25.400 --> 00:18:28.460 align:middle line:84% And then on top, you have your actual amorphous silicon 00:18:28.460 --> 00:18:29.940 align:middle line:90% material. 00:18:29.940 --> 00:18:32.040 align:middle line:84% And you have two electrodes here. 00:18:32.040 --> 00:18:33.860 align:middle line:90% These are the source and drains. 00:18:33.860 --> 00:18:37.130 align:middle line:84% So essentially, the current flow through a source and drain 00:18:37.130 --> 00:18:39.750 align:middle line:84% is where you want to switch on and off by this kind of thing 00:18:39.750 --> 00:18:41.090 align:middle line:90% going through this. 00:18:41.090 --> 00:18:43.970 align:middle line:84% So it's a field effect transistor, basically. 00:18:43.970 --> 00:18:45.610 align:middle line:84% So the way it operates is that you 00:18:45.610 --> 00:18:48.930 align:middle line:84% can control the electrical conductance or the magnitude 00:18:48.930 --> 00:18:50.450 align:middle line:84% of the current between the source 00:18:50.450 --> 00:18:54.570 align:middle line:84% and drain by controlling the voltage you apply on the gate. 00:18:54.570 --> 00:19:02.510 align:middle line:84% So the way it works is if you think about-- 00:19:02.510 --> 00:19:05.230 align:middle line:90% so this is the bottom gate. 00:19:05.230 --> 00:19:08.890 align:middle line:90% This is the gate. 00:19:08.890 --> 00:19:10.390 align:middle line:90% This is a dielectric material. 00:19:10.390 --> 00:19:12.250 align:middle line:84% You have your source and drain here. 00:19:12.250 --> 00:19:14.970 align:middle line:84% And then this is your semiconductor material. 00:19:14.970 --> 00:19:17.890 align:middle line:84% So let's say plus, this is minus. 00:19:17.890 --> 00:19:21.350 align:middle line:84% So if you apply a bias onto the gate, 00:19:21.350 --> 00:19:25.450 align:middle line:84% let's say if you apply, say, a positive bias, 00:19:25.450 --> 00:19:28.930 align:middle line:84% now this layer of gate dielectric, this 00:19:28.930 --> 00:19:32.050 align:middle line:84% is actually electrically insulating material. 00:19:32.050 --> 00:19:34.490 align:middle line:84% So what you have here, amorphous silicon is electrically 00:19:34.490 --> 00:19:36.490 align:middle line:84% conductive, gate is electrically conductive, 00:19:36.490 --> 00:19:39.190 align:middle line:84% separated by an insulating layer. 00:19:39.190 --> 00:19:43.430 align:middle line:84% What you have here is basically a parallel plate capacitor. 00:19:43.430 --> 00:19:46.070 align:middle line:84% So if you apply a positive bias here, 00:19:46.070 --> 00:19:49.190 align:middle line:84% you are going to accumulate a lot of positive charges 00:19:49.190 --> 00:19:50.807 align:middle line:90% on the gate. 00:19:50.807 --> 00:19:52.390 align:middle line:84% At the same time, it means that you're 00:19:52.390 --> 00:19:54.978 align:middle line:84% going to accumulate a lot of negative charges 00:19:54.978 --> 00:19:57.270 align:middle line:84% at the interface between amorphous silicon and the gate 00:19:57.270 --> 00:19:58.950 align:middle line:90% dielectric. 00:19:58.950 --> 00:20:02.750 align:middle line:84% So eventually, you will be able to form essentially 00:20:02.750 --> 00:20:07.790 align:middle line:84% what they call an n-channel, a thin layer where 00:20:07.790 --> 00:20:12.710 align:middle line:84% you have lots of electrons accumulated near the interface. 00:20:12.710 --> 00:20:15.377 align:middle line:84% So you're starting amorphous silicon, if it's undoped. 00:20:15.377 --> 00:20:17.210 align:middle line:84% Then, as we have mentioned, if it's undoped, 00:20:17.210 --> 00:20:18.877 align:middle line:84% it has very low electrical conductivity. 00:20:18.877 --> 00:20:21.570 align:middle line:84% Now, effectively, through this field effect, 00:20:21.570 --> 00:20:24.470 align:middle line:84% you are able to essentially dope electrostatically 00:20:24.470 --> 00:20:28.030 align:middle line:84% the surface layer with a lot of n-type electrons. 00:20:28.030 --> 00:20:30.530 align:middle line:84% So now this becomes a conductive channel, 00:20:30.530 --> 00:20:33.900 align:middle line:84% and you can now pass current through between the source 00:20:33.900 --> 00:20:35.620 align:middle line:90% and drain electrodes. 00:20:35.620 --> 00:20:38.120 align:middle line:84% So this device will now be in an on state. 00:20:38.120 --> 00:20:41.500 align:middle line:90% 00:20:41.500 --> 00:20:45.952 align:middle line:84% So simply put, by changing the bias on the gate, 00:20:45.952 --> 00:20:47.660 align:middle line:84% you can change the electrical conductance 00:20:47.660 --> 00:20:49.060 align:middle line:90% between the source and drain. 00:20:49.060 --> 00:20:51.580 align:middle line:84% And therefore, you can essentially switch this device 00:20:51.580 --> 00:20:55.600 align:middle line:84% on and off by simply changing the bias you apply on the gate. 00:20:55.600 --> 00:20:56.100 align:middle line:90% OK. 00:20:56.100 --> 00:21:01.620 align:middle line:90% 00:21:01.620 --> 00:21:02.120 align:middle line:90% All right. 00:21:02.120 --> 00:21:04.220 align:middle line:84% Now on the right-hand side plot here, I'm 00:21:04.220 --> 00:21:06.760 align:middle line:84% basically showing you what the I-V curve looks like. 00:21:06.760 --> 00:21:10.640 align:middle line:84% So vertical axis is the current, horizontal axis is voltage. 00:21:10.640 --> 00:21:12.900 align:middle line:84% Now we are referring to the voltage and the current 00:21:12.900 --> 00:21:14.680 align:middle line:90% between the source and drain. 00:21:14.680 --> 00:21:18.580 align:middle line:90% 00:21:18.580 --> 00:21:20.700 align:middle line:84% I have plotted a few different curves there 00:21:20.700 --> 00:21:23.460 align:middle line:84% that correspond to different gate biases, of different gate 00:21:23.460 --> 00:21:24.200 align:middle line:90% voltages. 00:21:24.200 --> 00:21:26.380 align:middle line:84% You can clearly see the curve actually goes 00:21:26.380 --> 00:21:28.020 align:middle line:90% into two different regimes. 00:21:28.020 --> 00:21:32.770 align:middle line:84% Initially, you have what's called a linear regime where 00:21:32.770 --> 00:21:36.770 align:middle line:84% the current essentially scales linearly with the applied bias. 00:21:36.770 --> 00:21:38.090 align:middle line:90% Now that's easy to understand. 00:21:38.090 --> 00:21:39.610 align:middle line:90% Initially, you have low bias. 00:21:39.610 --> 00:21:41.690 align:middle line:84% You have a conductive channel that's present. 00:21:41.690 --> 00:21:44.050 align:middle line:84% If you apply higher bias, you have more current 00:21:44.050 --> 00:21:46.910 align:middle line:84% flowing through the conductive channel, the n-channel. 00:21:46.910 --> 00:21:49.550 align:middle line:84% And therefore, you get just more current, basically. 00:21:49.550 --> 00:21:51.370 align:middle line:90% So higher bias, more current. 00:21:51.370 --> 00:21:54.330 align:middle line:84% So that's what happens in the linear stage. 00:21:54.330 --> 00:21:58.130 align:middle line:84% Now beyond the linear stage, you see the curves start to plateau. 00:21:58.130 --> 00:22:02.810 align:middle line:84% This is called the saturation regime. 00:22:02.810 --> 00:22:05.370 align:middle line:84% In the saturation regime, what happens 00:22:05.370 --> 00:22:10.890 align:middle line:84% is that you essentially apply a very large source-drain voltage. 00:22:10.890 --> 00:22:15.130 align:middle line:84% So what happens if you increase the source-drain voltage? 00:22:15.130 --> 00:22:16.830 align:middle line:84% So if you increase source-drain voltage, 00:22:16.830 --> 00:22:20.570 align:middle line:84% basically you put more charges here and here. 00:22:20.570 --> 00:22:21.900 align:middle line:90% This is positively charged. 00:22:21.900 --> 00:22:23.650 align:middle line:84% So it means the electrons in the n-channel 00:22:23.650 --> 00:22:25.890 align:middle line:84% is going to get attracted to this side. 00:22:25.890 --> 00:22:26.990 align:middle line:90% Same thing this side. 00:22:26.990 --> 00:22:28.930 align:middle line:90% This is negatively charged. 00:22:28.930 --> 00:22:31.640 align:middle line:84% So it means the electrons are going to get repelled away 00:22:31.640 --> 00:22:34.040 align:middle line:90% from this electrode. 00:22:34.040 --> 00:22:36.800 align:middle line:84% So as you increase the source-drain voltage, what's 00:22:36.800 --> 00:22:39.908 align:middle line:84% going to happen is that the shape of this n-channel 00:22:39.908 --> 00:22:40.700 align:middle line:90% is going to change. 00:22:40.700 --> 00:22:43.520 align:middle line:90% It's going to become asymmetric. 00:22:43.520 --> 00:22:45.603 align:middle line:84% You're going to have more electrons accumulating 00:22:45.603 --> 00:22:48.020 align:middle line:84% on this side and less electrons accumulating on that side. 00:22:48.020 --> 00:22:50.760 align:middle line:84% So this n-channel will look more like this. 00:22:50.760 --> 00:22:56.440 align:middle line:90% 00:22:56.440 --> 00:22:58.920 align:middle line:84% Eventually, if you continue to increase 00:22:58.920 --> 00:23:01.000 align:middle line:84% the bias between the source and drain, 00:23:01.000 --> 00:23:02.640 align:middle line:84% you're going to have a phenomenon 00:23:02.640 --> 00:23:04.660 align:middle line:90% of so-called pinch-off. 00:23:04.660 --> 00:23:08.720 align:middle line:90% 00:23:08.720 --> 00:23:12.160 align:middle line:84% Pinch-off means you have such a high bias such 00:23:12.160 --> 00:23:16.760 align:middle line:84% that the n-channel becomes disconnected 00:23:16.760 --> 00:23:20.720 align:middle line:90% from the negative contact here. 00:23:20.720 --> 00:23:23.080 align:middle line:84% Again, because of electrostatic repulsion. 00:23:23.080 --> 00:23:25.960 align:middle line:84% Because of the large negative bias. 00:23:25.960 --> 00:23:29.480 align:middle line:84% So it pushes all the electrons away from this electrode there. 00:23:29.480 --> 00:23:31.973 align:middle line:84% So then you only have the n-channel 00:23:31.973 --> 00:23:33.890 align:middle line:84% that's not connected to one of the electrodes. 00:23:33.890 --> 00:23:42.230 align:middle line:90% 00:23:42.230 --> 00:23:46.270 align:middle line:84% Now this is what leads to the current saturation. 00:23:46.270 --> 00:23:48.190 align:middle line:84% Once you have a pinch-off, if you continue 00:23:48.190 --> 00:23:50.137 align:middle line:84% to increase the bias, you're not going 00:23:50.137 --> 00:23:52.470 align:middle line:84% to increase the current going through between the source 00:23:52.470 --> 00:23:53.950 align:middle line:90% and drain. 00:23:53.950 --> 00:23:56.730 align:middle line:84% Because as you increase the bias, this region, 00:23:56.730 --> 00:23:58.510 align:middle line:84% the disconnection region, the separation 00:23:58.510 --> 00:24:01.590 align:middle line:84% is going to become bigger, which essentially restricts the more 00:24:01.590 --> 00:24:03.150 align:middle line:84% current you can pass through there. 00:24:03.150 --> 00:24:06.190 align:middle line:84% So that's why you see that in the saturation regime, as you 00:24:06.190 --> 00:24:08.030 align:middle line:84% increase the source and drain voltage, 00:24:08.030 --> 00:24:10.530 align:middle line:84% the current flowing through is actually not increasing. 00:24:10.530 --> 00:24:11.930 align:middle line:90% It pretty much plateaus. 00:24:11.930 --> 00:24:14.750 align:middle line:84% That's exactly because of this kind of pinch-off mechanism 00:24:14.750 --> 00:24:17.030 align:middle line:90% there. 00:24:17.030 --> 00:24:18.030 align:middle line:90% Yes? 00:24:18.030 --> 00:24:21.590 align:middle line:84% STUDENT: Why does it plateau and not drop to zero? 00:24:21.590 --> 00:24:24.150 align:middle line:84% JUEJUN HU: Well, basically because this is not 00:24:24.150 --> 00:24:27.740 align:middle line:84% going to-- when we say pinch-off, this is a schematic. 00:24:27.740 --> 00:24:29.490 align:middle line:84% You will actually have a decreased density 00:24:29.490 --> 00:24:30.670 align:middle line:90% of negative charges. 00:24:30.670 --> 00:24:32.575 align:middle line:90% It doesn't go to zero. 00:24:32.575 --> 00:24:34.450 align:middle line:84% There is still current that can actually pass 00:24:34.450 --> 00:24:36.450 align:middle line:90% through this small gap there. 00:24:36.450 --> 00:24:39.410 align:middle line:90% 00:24:39.410 --> 00:24:42.210 align:middle line:84% And also what you also notice here is that in this regime, 00:24:42.210 --> 00:24:44.090 align:middle line:84% this actually can be a very nice switch 00:24:44.090 --> 00:24:46.610 align:middle line:84% because you can actually control the current 00:24:46.610 --> 00:24:49.250 align:middle line:90% by changing the gate voltage. 00:24:49.250 --> 00:24:51.990 align:middle line:84% You can imagine that if you apply a larger gate voltage, 00:24:51.990 --> 00:24:53.830 align:middle line:84% you put more positive charges here, 00:24:53.830 --> 00:24:56.510 align:middle line:84% you can attract more negative charges on the other side. 00:24:56.510 --> 00:25:02.613 align:middle line:84% So the higher the gate voltage, the higher the current going 00:25:02.613 --> 00:25:04.530 align:middle line:84% through is going to be, the higher the current 00:25:04.530 --> 00:25:06.770 align:middle line:84% is going to be between the source and drain. 00:25:06.770 --> 00:25:08.290 align:middle line:84% Essentially a higher gate voltage 00:25:08.290 --> 00:25:13.050 align:middle line:84% can effectively impede or inhibit 00:25:13.050 --> 00:25:16.770 align:middle line:84% the formation of this kind of pinch-off phenomenon. 00:25:16.770 --> 00:25:20.370 align:middle line:84% So you can essentially operate these devices in this regime 00:25:20.370 --> 00:25:24.570 align:middle line:84% here as a very effective switch using the gate 00:25:24.570 --> 00:25:27.130 align:middle line:84% to control the current and voltage between the source 00:25:27.130 --> 00:25:27.780 align:middle line:90% and gate. 00:25:27.780 --> 00:25:36.340 align:middle line:90% 00:25:36.340 --> 00:25:39.240 align:middle line:84% So that's the basics about this kind of thin film transistor 00:25:39.240 --> 00:25:39.800 align:middle line:90% there. 00:25:39.800 --> 00:25:43.280 align:middle line:84% Now what I want to do, there are actually two main material 00:25:43.280 --> 00:25:44.333 align:middle line:90% systems that people-- 00:25:44.333 --> 00:25:46.000 align:middle line:84% there are two main materials that people 00:25:46.000 --> 00:25:48.840 align:middle line:84% use to construct these kind of thin film transistors, 00:25:48.840 --> 00:25:50.480 align:middle line:84% are amorphous silicon and what's called 00:25:50.480 --> 00:25:53.240 align:middle line:90% a low-temperature polysilicon. 00:25:53.240 --> 00:25:56.240 align:middle line:84% Before I go to that, I do want to highlight one difference, why 00:25:56.240 --> 00:25:59.222 align:middle line:84% people want to use one technology over the other. 00:25:59.222 --> 00:26:00.680 align:middle line:84% So there's one important difference 00:26:00.680 --> 00:26:04.480 align:middle line:84% between an AMLCD and AMOLED that we have not talked about yet. 00:26:04.480 --> 00:26:08.120 align:middle line:84% So if you think about a liquid crystal display, 00:26:08.120 --> 00:26:10.040 align:middle line:84% liquid crystals are an electrically-insulating 00:26:10.040 --> 00:26:11.000 align:middle line:90% material. 00:26:11.000 --> 00:26:13.360 align:middle line:84% The way you rotate a liquid crystal molecule 00:26:13.360 --> 00:26:16.340 align:middle line:84% is you apply a bias across the liquid crystal. 00:26:16.340 --> 00:26:19.140 align:middle line:84% The electric field essentially changes the polarization 00:26:19.140 --> 00:26:21.140 align:middle line:84% and the orientation of liquid crystal molecules. 00:26:21.140 --> 00:26:23.640 align:middle line:84% That's how you change your chirality. 00:26:23.640 --> 00:26:27.350 align:middle line:84% Now it's basically a capacitive device. 00:26:27.350 --> 00:26:30.790 align:middle line:84% So when I say "capacitive," it means that we only apply a bias. 00:26:30.790 --> 00:26:33.310 align:middle line:84% There's no current flowing through when it's actually 00:26:33.310 --> 00:26:36.550 align:middle line:90% in a steady state. 00:26:36.550 --> 00:26:41.230 align:middle line:84% Now this is different from the case of an AMOLED. 00:26:41.230 --> 00:26:43.417 align:middle line:84% For OLED, it's a light-emitting diode. 00:26:43.417 --> 00:26:45.250 align:middle line:84% Similar to any kind of light emitting diode, 00:26:45.250 --> 00:26:48.370 align:middle line:84% you need to have current flowing through a device 00:26:48.370 --> 00:26:50.870 align:middle line:84% in order to have electrons and holes will recombine and emit 00:26:50.870 --> 00:26:51.750 align:middle line:90% light. 00:26:51.750 --> 00:26:55.150 align:middle line:90% So it's a current-driven device. 00:26:55.150 --> 00:26:58.650 align:middle line:84% So LCD display, this is a voltage-driven device. 00:26:58.650 --> 00:27:01.710 align:middle line:90% That's a current-driven device. 00:27:01.710 --> 00:27:04.910 align:middle line:84% Now this difference is actually quite important 00:27:04.910 --> 00:27:09.150 align:middle line:84% because as I will talk about the Low Temperature 00:27:09.150 --> 00:27:12.430 align:middle line:84% Polysilicon, LTPS, you can afford much higher driving 00:27:12.430 --> 00:27:16.710 align:middle line:84% current compared to TFTs made of amorphous silicon. 00:27:16.710 --> 00:27:21.750 align:middle line:84% So that's why for AMLCD, most of the time, people 00:27:21.750 --> 00:27:24.900 align:middle line:84% are happy with amorphous silicon TFT. 00:27:24.900 --> 00:27:29.235 align:middle line:84% But for AMOLED, people tend to use LTPS rather than 00:27:29.235 --> 00:27:30.860 align:middle line:84% amorphous silicon because they actually 00:27:30.860 --> 00:27:33.193 align:middle line:84% need the current to flow through their devices, not just 00:27:33.193 --> 00:27:33.840 align:middle line:90% the bias. 00:27:33.840 --> 00:27:42.067 align:middle line:90% 00:27:42.067 --> 00:27:43.900 align:middle line:84% So now let's come back and talk a little bit 00:27:43.900 --> 00:27:46.580 align:middle line:84% about the difference between these two material systems 00:27:46.580 --> 00:27:47.620 align:middle line:90% there. 00:27:47.620 --> 00:27:50.940 align:middle line:84% So these two are the dominant technologies 00:27:50.940 --> 00:27:55.000 align:middle line:84% for essentially display thin-film transistors. 00:27:55.000 --> 00:27:57.420 align:middle line:84% There's another kind of a class of material called oxide 00:27:57.420 --> 00:28:05.020 align:middle line:84% TFTs, which mostly are made of a material called IGZO, Indium 00:28:05.020 --> 00:28:09.180 align:middle line:84% Gallium Zinc Oxide, which is basically 00:28:09.180 --> 00:28:14.390 align:middle line:84% a semiconductor primarily based on zinc oxide. 00:28:14.390 --> 00:28:16.140 align:middle line:84% Indium and gallium are added, essentially, 00:28:16.140 --> 00:28:17.360 align:middle line:90% as chemical modifiers. 00:28:17.360 --> 00:28:21.340 align:middle line:84% So you can think of it as like doped zinc oxide to some extent. 00:28:21.340 --> 00:28:23.600 align:middle line:84% It has many properties similar to amorphous silicon 00:28:23.600 --> 00:28:24.940 align:middle line:90% and polycrystalline silicon. 00:28:24.940 --> 00:28:27.000 align:middle line:84% But in terms of market share, this 00:28:27.000 --> 00:28:30.220 align:middle line:84% is still a lot smaller compared to LTPS or amorphous silicon. 00:28:30.220 --> 00:28:32.640 align:middle line:84% Those are the dominant technologies. 00:28:32.640 --> 00:28:36.840 align:middle line:84% So for amorphous silicon, LTPS is basically 00:28:36.840 --> 00:28:38.880 align:middle line:90% a polycrystalline silicon. 00:28:38.880 --> 00:28:43.900 align:middle line:84% Now to prepare LTPS, you have to first deposit amorphous silicon. 00:28:43.900 --> 00:28:44.900 align:middle line:90% That's the way to do it. 00:28:44.900 --> 00:28:46.780 align:middle line:84% You start by depositing amorphous silicon. 00:28:46.780 --> 00:28:51.320 align:middle line:84% You then crystallize it to get LTPS. 00:28:51.320 --> 00:28:54.000 align:middle line:84% You have a couple of different ways to do a crystallization. 00:28:54.000 --> 00:28:57.482 align:middle line:84% You can either just do a blanket annealing 00:28:57.482 --> 00:28:58.940 align:middle line:84% of the amorphous silicon thin film. 00:28:58.940 --> 00:29:00.690 align:middle line:84% You just put it into a furnace, heat it up 00:29:00.690 --> 00:29:01.900 align:middle line:90% to 600 degrees Celsius. 00:29:01.900 --> 00:29:03.080 align:middle line:90% It's going to crystallize. 00:29:03.080 --> 00:29:06.160 align:middle line:84% It's actually not usually preferred, that kind 00:29:06.160 --> 00:29:08.860 align:middle line:84% of technique, because-- for a couple of reasons. 00:29:08.860 --> 00:29:13.345 align:middle line:84% One is mostly because-- number one, 600 degrees Celsius, 00:29:13.345 --> 00:29:15.720 align:middle line:84% this temperature is higher than the softening temperature 00:29:15.720 --> 00:29:18.800 align:middle line:84% of most glass substrates, like sodalite. 00:29:18.800 --> 00:29:21.760 align:middle line:84% So it means that if you want to use this process here 00:29:21.760 --> 00:29:23.670 align:middle line:84% to prevent any kind of mechanical deformation 00:29:23.670 --> 00:29:25.878 align:middle line:84% of the glass substrate, you have to use a specialized 00:29:25.878 --> 00:29:27.765 align:middle line:84% glass with high softening point just 00:29:27.765 --> 00:29:29.390 align:middle line:84% to prevent that, which, of course, adds 00:29:29.390 --> 00:29:30.710 align:middle line:90% to the overall cost. 00:29:30.710 --> 00:29:34.510 align:middle line:84% And also, if you do this kind of a high-temperature annealing, 00:29:34.510 --> 00:29:37.670 align:middle line:84% there will be concerns about other kinds of-- different kind 00:29:37.670 --> 00:29:40.910 align:middle line:84% of components that's already existing on the substrate that 00:29:40.910 --> 00:29:43.110 align:middle line:84% may be getting damaged because of high temperature. 00:29:43.110 --> 00:29:45.870 align:middle line:84% So for example, if you're trying to do a flexible display 00:29:45.870 --> 00:29:47.550 align:middle line:84% with lots of polymer components, you 00:29:47.550 --> 00:29:48.590 align:middle line:84% will not be able to use this kind 00:29:48.590 --> 00:29:50.507 align:middle line:84% of high-temperature annealing across the board 00:29:50.507 --> 00:29:52.030 align:middle line:90% to crystallize the material. 00:29:52.030 --> 00:29:55.710 align:middle line:84% So another way to do this kind of crystallization 00:29:55.710 --> 00:29:57.630 align:middle line:90% is to use, essentially, lasers. 00:29:57.630 --> 00:30:00.110 align:middle line:90% Excimer Laser Annealing, ELA. 00:30:00.110 --> 00:30:03.450 align:middle line:84% So this typically uses a short wavelength excimer laser. 00:30:03.450 --> 00:30:07.030 align:middle line:84% These are lasers with wavelengths of typically 305 00:30:07.030 --> 00:30:10.190 align:middle line:90% nanometers, so in the UV regime. 00:30:10.190 --> 00:30:13.230 align:middle line:84% And then you essentially focus laser to a spot, 00:30:13.230 --> 00:30:17.630 align:middle line:84% heat it up to high temperature to induce local crystallization. 00:30:17.630 --> 00:30:20.230 align:middle line:84% The reason this kind of excimer laser annealing 00:30:20.230 --> 00:30:22.580 align:middle line:84% is preferred also is because you have 00:30:22.580 --> 00:30:26.140 align:middle line:84% a lot more control in terms of the kinetics of crystallization 00:30:26.140 --> 00:30:26.680 align:middle line:90% process. 00:30:26.680 --> 00:30:29.540 align:middle line:84% In fact, one popular way of doing 00:30:29.540 --> 00:30:32.220 align:middle line:84% this kind of laser annealing is a process 00:30:32.220 --> 00:30:41.280 align:middle line:84% called sequential laser solidification. 00:30:41.280 --> 00:30:47.380 align:middle line:90% 00:30:47.380 --> 00:30:50.460 align:middle line:90% This is also called SLS. 00:30:50.460 --> 00:30:53.580 align:middle line:84% Now, the way a sequential laser solidification works 00:30:53.580 --> 00:30:55.300 align:middle line:84% is you start out by essentially focusing 00:30:55.300 --> 00:30:58.120 align:middle line:90% laser spot into a small region. 00:30:58.120 --> 00:31:01.060 align:middle line:84% In many cases, people also make essentially a little shadow mask 00:31:01.060 --> 00:31:03.380 align:middle line:84% so that you can essentially tightly confine the laser 00:31:03.380 --> 00:31:05.580 align:middle line:90% spot only to the that region. 00:31:05.580 --> 00:31:07.020 align:middle line:84% Within the desired regime, you can 00:31:07.020 --> 00:31:09.900 align:middle line:90% start by inducing nucleation. 00:31:09.900 --> 00:31:12.260 align:middle line:84% And then once you have a nuclei, you then 00:31:12.260 --> 00:31:14.500 align:middle line:84% start to laterally translate the laser spot 00:31:14.500 --> 00:31:18.012 align:middle line:90% into different regimes. 00:31:18.012 --> 00:31:21.050 align:middle line:84% Now you can imagine that if you do a furnace annealing, what's 00:31:21.050 --> 00:31:22.850 align:middle line:84% going to happen is that nucleation is going 00:31:22.850 --> 00:31:24.130 align:middle line:90% to be completely statistical. 00:31:24.130 --> 00:31:25.450 align:middle line:90% It's going to be random. 00:31:25.450 --> 00:31:28.652 align:middle line:84% You have essentially nuclei popping up all over the place, 00:31:28.652 --> 00:31:31.110 align:middle line:84% across the entire volume of the amorphous silicon material. 00:31:31.110 --> 00:31:34.370 align:middle line:84% You have no control over the crystal nucleation or growth. 00:31:34.370 --> 00:31:36.430 align:middle line:84% In the case of sequential laser solidification, 00:31:36.430 --> 00:31:38.210 align:middle line:84% on the other hand, you basically start out 00:31:38.210 --> 00:31:40.130 align:middle line:90% by nucleating in one region. 00:31:40.130 --> 00:31:44.450 align:middle line:84% You then induce the growth by translating the laser laterally 00:31:44.450 --> 00:31:46.530 align:middle line:90% along one particular direction. 00:31:46.530 --> 00:31:49.470 align:middle line:84% So in the process, in this forming new nuclei, what's 00:31:49.470 --> 00:31:51.970 align:middle line:84% going to happen is that you are going to preferentially grow 00:31:51.970 --> 00:31:54.330 align:middle line:90% the existing nuclei. 00:31:54.330 --> 00:31:56.690 align:middle line:84% So then what you end up having is 00:31:56.690 --> 00:31:59.490 align:middle line:84% that you have this kind of directional lone crystalline 00:31:59.490 --> 00:32:02.810 align:middle line:84% grains that are almost single crystalline. 00:32:02.810 --> 00:32:04.970 align:middle line:84% You have very few crystalline grain boundaries 00:32:04.970 --> 00:32:08.050 align:middle line:84% because essentially, you are building upon existing crystal 00:32:08.050 --> 00:32:11.730 align:middle line:84% nuclei rather than trying to form new ones. 00:32:11.730 --> 00:32:15.070 align:middle line:84% And that's actually beneficial because the resulting material-- 00:32:15.070 --> 00:32:17.050 align:middle line:84% obviously, because of larger grains, 00:32:17.050 --> 00:32:18.310 align:middle line:90% you have a lot less defects. 00:32:18.310 --> 00:32:19.980 align:middle line:84% You have here-- for higher current, 00:32:19.980 --> 00:32:22.380 align:middle line:84% you have essentially less carrier recombination. 00:32:22.380 --> 00:32:23.968 align:middle line:90% You also have higher mobility. 00:32:23.968 --> 00:32:25.760 align:middle line:84% You have all the nice electronic properties 00:32:25.760 --> 00:32:28.520 align:middle line:84% because of this kind of a directional growth 00:32:28.520 --> 00:32:30.020 align:middle line:90% approach induced by lasers. 00:32:30.020 --> 00:32:36.360 align:middle line:90% 00:32:36.360 --> 00:32:37.280 align:middle line:90% Yes? 00:32:37.280 --> 00:32:39.080 align:middle line:90% STUDENT: And so-- 00:32:39.080 --> 00:32:42.400 align:middle line:84% I mean, it says that the processing temperature 00:32:42.400 --> 00:32:46.040 align:middle line:90% of this process is very high. 00:32:46.040 --> 00:32:49.332 align:middle line:90% So is it-- 00:32:49.332 --> 00:32:51.540 align:middle line:84% I mean, I guess-- but it's only in a very small spot. 00:32:51.540 --> 00:32:52.332 align:middle line:90% JUEJUN HU: Exactly. 00:32:52.332 --> 00:32:53.060 align:middle line:90% That's the idea. 00:32:53.060 --> 00:32:53.600 align:middle line:90% Yeah. 00:32:53.600 --> 00:32:55.700 align:middle line:84% In fact, in the next slide, I'm going to show you one example. 00:32:55.700 --> 00:32:57.158 align:middle line:84% You can even use this kind of laser 00:32:57.158 --> 00:32:58.750 align:middle line:84% annealing even for flexible displays 00:32:58.750 --> 00:33:00.500 align:middle line:84% where you have lots of polymers around it. 00:33:00.500 --> 00:33:02.480 align:middle line:84% As long as you design your structure so the laser 00:33:02.480 --> 00:33:03.760 align:middle line:84% spot-- the heat generation is thermally 00:33:03.760 --> 00:33:06.320 align:middle line:84% insulated from the surrounding environment, just a few tens 00:33:06.320 --> 00:33:07.820 align:middle line:84% of microns away from the laser spot, 00:33:07.820 --> 00:33:10.600 align:middle line:84% the temperature can go down to below 100 Celsius, 00:33:10.600 --> 00:33:12.213 align:middle line:90% which is well affordable. 00:33:12.213 --> 00:33:13.880 align:middle line:84% So that's another advantage of this kind 00:33:13.880 --> 00:33:16.080 align:middle line:84% of laser annealing approach is that it's localized. 00:33:16.080 --> 00:33:17.740 align:middle line:84% You don't have to worry about cooking everything 00:33:17.740 --> 00:33:18.640 align:middle line:90% on the substrate. 00:33:18.640 --> 00:33:23.320 align:middle line:90% 00:33:23.320 --> 00:33:25.100 align:middle line:84% Now, in terms of the properties, you 00:33:25.100 --> 00:33:28.060 align:middle line:84% can see that the biggest difference is the mobility. 00:33:28.060 --> 00:33:29.820 align:middle line:84% Amorphous silicon, as I mentioned, 00:33:29.820 --> 00:33:31.320 align:middle line:90% you get lots of trapped states. 00:33:31.320 --> 00:33:32.820 align:middle line:84% You also get lots of scattering just 00:33:32.820 --> 00:33:34.620 align:middle line:84% from disordered structure for carriers. 00:33:34.620 --> 00:33:37.180 align:middle line:84% So it tends to have quite low mobility. 00:33:37.180 --> 00:33:38.687 align:middle line:90% LTPS is actually better. 00:33:38.687 --> 00:33:41.020 align:middle line:84% It's still not comparable to single crystalline silicon. 00:33:41.020 --> 00:33:42.520 align:middle line:84% Single crystalline silicon mobility, 00:33:42.520 --> 00:33:44.060 align:middle line:90% we're talking about above 1,000. 00:33:44.060 --> 00:33:45.780 align:middle line:84% So it's still low [INAUDIBLE] silicon, 00:33:45.780 --> 00:33:47.080 align:middle line:90% but it's quite disatant. 00:33:47.080 --> 00:33:52.720 align:middle line:90% 00:33:52.720 --> 00:33:55.300 align:middle line:90% So that's why you can actually-- 00:33:55.300 --> 00:33:58.620 align:middle line:84% LTPS TFTs, they can be used as a current driver, 00:33:58.620 --> 00:34:00.620 align:middle line:84% while amorphous silicon TFTs, people typically 00:34:00.620 --> 00:34:03.660 align:middle line:84% only use them as voltage drivers. 00:34:03.660 --> 00:34:06.420 align:middle line:84% Another advantage of using LTPS is 00:34:06.420 --> 00:34:10.980 align:middle line:84% that you can directly integrate ICs, Integrated Circuits, 00:34:10.980 --> 00:34:12.620 align:middle line:90% on LTPS. 00:34:12.620 --> 00:34:15.320 align:middle line:84% For amorphous silicon, because of its very low mobility, 00:34:15.320 --> 00:34:18.290 align:middle line:84% people tend to make only TFTs out of them. 00:34:18.290 --> 00:34:20.770 align:middle line:84% They don't make ICs out of amorphous silicon. 00:34:20.770 --> 00:34:22.690 align:middle line:84% But you still need ICs to essentially program 00:34:22.690 --> 00:34:25.370 align:middle line:84% all the TFTs to get the desired display pattern. 00:34:25.370 --> 00:34:27.989 align:middle line:84% So the way they do it is they do a hybrid integration. 00:34:27.989 --> 00:34:30.190 align:middle line:84% They have ICs fabricated separately. 00:34:30.190 --> 00:34:32.969 align:middle line:84% They then attach ICs into the displays 00:34:32.969 --> 00:34:36.810 align:middle line:84% that's made with amorphous silicon TFT backplane. 00:34:36.810 --> 00:34:39.969 align:middle line:84% And in lots of cases, those ICs actually 00:34:39.969 --> 00:34:42.770 align:middle line:84% attach to the edge of the display panel. 00:34:42.770 --> 00:34:46.050 align:middle line:84% So if you look at amorphous silicon, some of the TFT display 00:34:46.050 --> 00:34:49.242 align:middle line:84% panels, the edge is usually dark, 00:34:49.242 --> 00:34:51.450 align:middle line:84% because the edge is here reserved and a lot of spaces 00:34:51.450 --> 00:34:53.650 align:middle line:90% are occupied by ICs. 00:34:53.650 --> 00:34:57.090 align:middle line:84% So there's essentially a little edge where it's not lighting up. 00:34:57.090 --> 00:35:00.210 align:middle line:84% The edge is a lot narrower, or can be a lot narrower 00:35:00.210 --> 00:35:02.490 align:middle line:84% in the case of LTPS, because ICs can 00:35:02.490 --> 00:35:05.290 align:middle line:84% be directly integrated within the polycrystalline material 00:35:05.290 --> 00:35:06.290 align:middle line:90% as well. 00:35:06.290 --> 00:35:09.850 align:middle line:84% So that's one additional advantage. 00:35:09.850 --> 00:35:13.590 align:middle line:84% And also LTPS, because of the much better properties, 00:35:13.590 --> 00:35:16.040 align:middle line:84% you can also get higher resolution of the pixels. 00:35:16.040 --> 00:35:19.760 align:middle line:84% So again, PPI, this is Pixel Per Inch, 00:35:19.760 --> 00:35:22.100 align:middle line:84% so you can get high resolution pixels with LTPS. 00:35:22.100 --> 00:35:25.480 align:middle line:84% The amorphous silicon is also a bit more difficult to reach. 00:35:25.480 --> 00:35:27.687 align:middle line:84% Now the biggest downside for LTPS, 00:35:27.687 --> 00:35:30.020 align:middle line:84% you can imagine that with all the additional processing, 00:35:30.020 --> 00:35:32.640 align:middle line:90% it gets more expensive. 00:35:32.640 --> 00:35:35.280 align:middle line:84% And that's why amorphous silicon is still 00:35:35.280 --> 00:35:37.400 align:middle line:84% the dominant technology, especially when you're 00:35:37.400 --> 00:35:39.100 align:middle line:90% looking at LCD displays. 00:35:39.100 --> 00:35:43.880 align:middle line:90% 00:35:43.880 --> 00:35:47.000 align:middle line:84% LTPS is really used more for high-end displays, 00:35:47.000 --> 00:35:49.420 align:middle line:90% like these kinds of devices. 00:35:49.420 --> 00:35:55.680 align:middle line:90% 00:35:55.680 --> 00:35:56.180 align:middle line:90% All right. 00:35:56.180 --> 00:35:57.820 align:middle line:84% But even if you want to make LTPS, 00:35:57.820 --> 00:35:59.100 align:middle line:84% you always start from amorphous silicon. 00:35:59.100 --> 00:36:00.767 align:middle line:84% You still have to start with essentially 00:36:00.767 --> 00:36:02.360 align:middle line:84% a PECVD deposited amorphous silicon 00:36:02.360 --> 00:36:03.943 align:middle line:84% and then go through the process there. 00:36:03.943 --> 00:36:05.920 align:middle line:84% So amorphous silicon is everywhere 00:36:05.920 --> 00:36:07.780 align:middle line:84% in this regard in display industry. 00:36:07.780 --> 00:36:10.740 align:middle line:90% 00:36:10.740 --> 00:36:12.420 align:middle line:84% As I promised you, in this slide, 00:36:12.420 --> 00:36:15.870 align:middle line:84% I'm going to show you one example where you can actually 00:36:15.870 --> 00:36:17.550 align:middle line:84% use this kind of laser annealing process 00:36:17.550 --> 00:36:20.805 align:middle line:84% to actually fabricate a flexible display in this case. 00:36:20.805 --> 00:36:22.930 align:middle line:84% So what I'm showing you on the left-hand side here, 00:36:22.930 --> 00:36:25.470 align:middle line:84% this is basically the process of fabricating a flexible display. 00:36:25.470 --> 00:36:27.637 align:middle line:84% You start out by essentially fabricating the display 00:36:27.637 --> 00:36:30.630 align:middle line:84% structures on two separate glass substrates. 00:36:30.630 --> 00:36:34.510 align:middle line:84% You then bond them together so that your display basically 00:36:34.510 --> 00:36:37.750 align:middle line:84% gets sandwiched between these two rigid glass substrates. 00:36:37.750 --> 00:36:40.790 align:middle line:84% You then delaminate the rigid glass substrates 00:36:40.790 --> 00:36:43.150 align:middle line:84% from the display and then transfer it 00:36:43.150 --> 00:36:45.670 align:middle line:84% onto your final flexible substrate. 00:36:45.670 --> 00:36:48.750 align:middle line:84% Now they use this kind of process here. 00:36:48.750 --> 00:36:50.990 align:middle line:84% You notice that the display devices are actually 00:36:50.990 --> 00:36:53.150 align:middle line:84% sandwiched right in the middle of the stack. 00:36:53.150 --> 00:36:55.773 align:middle line:84% This is mostly out of mechanical considerations. 00:36:55.773 --> 00:36:58.190 align:middle line:84% You may recall that when we were talking about essentially 00:36:58.190 --> 00:36:59.883 align:middle line:84% this kind of flexible glass substrates, 00:36:59.883 --> 00:37:02.050 align:middle line:84% we said that when you actually bend the glass plate, 00:37:02.050 --> 00:37:03.990 align:middle line:84% you have the so-called neutral plane. 00:37:03.990 --> 00:37:05.365 align:middle line:84% Right in the middle, you actually 00:37:05.365 --> 00:37:07.573 align:middle line:84% get zero strain and stress, despite the fact that you 00:37:07.573 --> 00:37:09.010 align:middle line:90% have a mechanical deformation. 00:37:09.010 --> 00:37:10.150 align:middle line:90% Same idea. 00:37:10.150 --> 00:37:14.490 align:middle line:84% By placing these devices aligned with the neutral plane, 00:37:14.490 --> 00:37:17.730 align:middle line:84% you can minimize any kind of mechanical strain exerted 00:37:17.730 --> 00:37:24.210 align:middle line:84% on your fragile devices despite the geometric deformation. 00:37:24.210 --> 00:37:25.510 align:middle line:90% So that's the idea. 00:37:25.510 --> 00:37:29.853 align:middle line:84% Now to get LTPS, people use laser annealing. 00:37:29.853 --> 00:37:30.770 align:middle line:90% And this is actually-- 00:37:30.770 --> 00:37:32.150 align:middle line:90% I quoted from this paper here. 00:37:32.150 --> 00:37:34.570 align:middle line:84% They did some simulations looking into-- 00:37:34.570 --> 00:37:37.570 align:middle line:84% you can locally heat up this amorphous silicon to a very high 00:37:37.570 --> 00:37:39.750 align:middle line:84% temperature, above 2,000 degrees Celsius. 00:37:39.750 --> 00:37:44.830 align:middle line:84% But just moving a few tens of microns away from the devices, 00:37:44.830 --> 00:37:46.490 align:middle line:84% you can actually quickly cool down 00:37:46.490 --> 00:37:48.810 align:middle line:84% to low temperature, below 100 degrees Celsius. 00:37:48.810 --> 00:37:52.410 align:middle line:84% So this kind of a highly localized heating characteristic 00:37:52.410 --> 00:37:57.690 align:middle line:84% is what enables laser annealing to process this kind of flexible 00:37:57.690 --> 00:38:00.290 align:middle line:84% displays with polymer components that cannot withstand high 00:38:00.290 --> 00:38:03.370 align:middle line:90% temperature. 00:38:03.370 --> 00:38:04.610 align:middle line:90% Questions? 00:38:04.610 --> 00:38:07.570 align:middle line:90% Make sense? 00:38:07.570 --> 00:38:11.240 align:middle line:84% STUDENT: Based on the image on the left, is the-- 00:38:11.240 --> 00:38:14.360 align:middle line:84% JUEJUN HU: Left is also zoomed in plot looking at locally where 00:38:14.360 --> 00:38:17.000 align:middle line:84% you have a focused laser spot hitting the amorphous silicone. 00:38:17.000 --> 00:38:17.500 align:middle line:90% STUDENT: OK. 00:38:17.500 --> 00:38:19.280 align:middle line:90% So it only goes-- 00:38:19.280 --> 00:38:21.887 align:middle line:84% the depth only goes to the [INAUDIBLE]. 00:38:21.887 --> 00:38:23.720 align:middle line:84% JUEJUN HU: Well, there's always a little bit 00:38:23.720 --> 00:38:25.120 align:middle line:90% of thermal diffusion. 00:38:25.120 --> 00:38:26.960 align:middle line:84% But the idea is that you design the layers 00:38:26.960 --> 00:38:29.440 align:middle line:84% and also you design the pulse so that you 00:38:29.440 --> 00:38:31.160 align:middle line:84% limit this kind of thermal diffusion 00:38:31.160 --> 00:38:33.200 align:middle line:90% to the surrounding regions. 00:38:33.200 --> 00:38:34.660 align:middle line:84% And this is the zoomed out region. 00:38:34.660 --> 00:38:36.420 align:middle line:84% This is actually where the focus spot is. 00:38:36.420 --> 00:38:38.295 align:middle line:84% But then, as soon as you move away from that, 00:38:38.295 --> 00:38:40.260 align:middle line:84% the temperature quickly kind of cools down. 00:38:40.260 --> 00:38:46.640 align:middle line:90% 00:38:46.640 --> 00:38:47.140 align:middle line:90% All right. 00:38:47.140 --> 00:38:48.932 align:middle line:84% Any more questions about displays before we 00:38:48.932 --> 00:38:52.040 align:middle line:90% move on to solar cells? 00:38:52.040 --> 00:38:52.720 align:middle line:90% All right. 00:38:52.720 --> 00:38:53.760 align:middle line:90% Good. 00:38:53.760 --> 00:38:56.460 align:middle line:84% So what we're seeing here is that in the display industry, 00:38:56.460 --> 00:38:58.880 align:middle line:84% as I said, amorphous silicon is still a dominant material. 00:38:58.880 --> 00:38:59.780 align:middle line:90% It's everywhere. 00:38:59.780 --> 00:39:01.600 align:middle line:84% People use it to make TFTs or use it 00:39:01.600 --> 00:39:04.040 align:middle line:84% to-- as a source material to create 00:39:04.040 --> 00:39:06.600 align:middle line:84% low-temperature polysilicon to make TFTs. 00:39:06.600 --> 00:39:08.227 align:middle line:84% So it's still a very useful material. 00:39:08.227 --> 00:39:10.060 align:middle line:84% Now we're going to switch gears a little bit 00:39:10.060 --> 00:39:11.510 align:middle line:90% and talk about solar cells. 00:39:11.510 --> 00:39:13.510 align:middle line:84% Solar cells is one industry where 00:39:13.510 --> 00:39:17.510 align:middle line:84% there used to be lots of interest in amorphous silicon. 00:39:17.510 --> 00:39:18.330 align:middle line:90% Not anymore. 00:39:18.330 --> 00:39:21.650 align:middle line:90% 00:39:21.650 --> 00:39:22.652 align:middle line:90% We're going to see why. 00:39:22.652 --> 00:39:24.110 align:middle line:84% So let's start by-- this slide just 00:39:24.110 --> 00:39:26.450 align:middle line:84% to give you a basic architecture of solar cells. 00:39:26.450 --> 00:39:29.263 align:middle line:84% You probably have seen it in 3033 or other classes. 00:39:29.263 --> 00:39:31.430 align:middle line:84% Solar cells are probably one of the most simple kind 00:39:31.430 --> 00:39:35.430 align:middle line:84% of semiconductor devices, basically a p-n junction. 00:39:35.430 --> 00:39:39.310 align:middle line:84% If light comes in, you are going to generate electron-hole pairs. 00:39:39.310 --> 00:39:41.630 align:middle line:84% Because of the presence of the building electric field 00:39:41.630 --> 00:39:43.950 align:middle line:84% in the p-n junction, these electrons and holes 00:39:43.950 --> 00:39:45.530 align:middle line:84% are going to get spatially separated. 00:39:45.530 --> 00:39:46.910 align:middle line:84% Electron goes one direction, hole 00:39:46.910 --> 00:39:48.210 align:middle line:90% goes the opposite direction. 00:39:48.210 --> 00:39:50.330 align:middle line:84% They are going to get collected by the electrodes. 00:39:50.330 --> 00:39:51.790 align:middle line:84% And then that's how you can have current flow 00:39:51.790 --> 00:39:53.750 align:middle line:84% in the external circuit to drive load. 00:39:53.750 --> 00:39:55.310 align:middle line:90% That's how solar cells work. 00:39:55.310 --> 00:39:57.150 align:middle line:84% If you look at I-V curve here, we 00:39:57.150 --> 00:40:00.972 align:middle line:84% are operating in this fourth quadrant here. 00:40:00.972 --> 00:40:02.430 align:middle line:84% If you have light, essentially this 00:40:02.430 --> 00:40:03.610 align:middle line:90% is basically a photocurrent. 00:40:03.610 --> 00:40:05.068 align:middle line:84% Photocurrent leads to, essentially, 00:40:05.068 --> 00:40:07.310 align:middle line:84% a lateral shift-- vertical shift of the curve here. 00:40:07.310 --> 00:40:09.160 align:middle line:84% This corresponds to a short circuit current. 00:40:09.160 --> 00:40:11.040 align:middle line:84% This corresponds to the open circuit voltage. 00:40:11.040 --> 00:40:16.700 align:middle line:84% I assume you learned a bit of that in 3033 or whatever. 00:40:16.700 --> 00:40:20.700 align:middle line:84% If not, go back to review the text in those classes. 00:40:20.700 --> 00:40:22.920 align:middle line:84% So it's basically just a p-n junction. 00:40:22.920 --> 00:40:23.820 align:middle line:90% Really simple. 00:40:23.820 --> 00:40:27.140 align:middle line:84% Now where does amorphous silicon fit in? 00:40:27.140 --> 00:40:33.560 align:middle line:84% So if you look back in about 15 years, in early 2000s, 00:40:33.560 --> 00:40:38.640 align:middle line:84% maybe early 2010s, there used to be lots of interest in thin film 00:40:38.640 --> 00:40:40.940 align:middle line:90% amorphous silicon solar cells. 00:40:40.940 --> 00:40:46.340 align:middle line:84% There were lots of startups trying to look into this. 00:40:46.340 --> 00:40:48.000 align:middle line:84% I think pretty much all of them failed. 00:40:48.000 --> 00:40:50.580 align:middle line:90% 00:40:50.580 --> 00:40:53.820 align:middle line:84% At the time, the premise was that you 00:40:53.820 --> 00:40:56.900 align:middle line:84% can deposit amorphous silicon over a large area 00:40:56.900 --> 00:40:59.220 align:middle line:84% with high-throughput CVD process. 00:40:59.220 --> 00:41:01.540 align:middle line:84% Remember the applied materials [INAUDIBLE] fab? 00:41:01.540 --> 00:41:03.620 align:middle line:84% You can deposit this on six-square meter glass 00:41:03.620 --> 00:41:04.660 align:middle line:90% substrates. 00:41:04.660 --> 00:41:08.660 align:middle line:84% So you can presumably we process these cells at low cost. 00:41:08.660 --> 00:41:12.240 align:middle line:90% 00:41:12.240 --> 00:41:14.520 align:middle line:84% An additional benefit of amorphous silicon compared 00:41:14.520 --> 00:41:16.728 align:middle line:84% to crystalline silicon is that you have a much higher 00:41:16.728 --> 00:41:18.280 align:middle line:90% absorption coefficient. 00:41:18.280 --> 00:41:23.640 align:middle line:84% Now this goes back to when we were talking about essentially 00:41:23.640 --> 00:41:24.390 align:middle line:90% amorphous silicon. 00:41:24.390 --> 00:41:26.723 align:middle line:84% Oh, sorry, crystalline silicon being an indirect bandgap 00:41:26.723 --> 00:41:27.460 align:middle line:90% semiconductor. 00:41:27.460 --> 00:41:29.840 align:middle line:84% So in order to have light absorption, 00:41:29.840 --> 00:41:31.500 align:middle line:84% you need a phonon-assisted process, 00:41:31.500 --> 00:41:33.520 align:middle line:90% which is pretty inefficient. 00:41:33.520 --> 00:41:37.920 align:middle line:84% So typically, in crystalline silicon solar cells, 00:41:37.920 --> 00:41:41.460 align:middle line:84% the typical wafer thickness is at least a few tens of microns, 00:41:41.460 --> 00:41:43.520 align:middle line:84% maybe 100 microns order of magnitude. 00:41:43.520 --> 00:41:47.760 align:middle line:84% So that's actually quite a lot of material. 00:41:47.760 --> 00:41:51.053 align:middle line:84% Versus in the case of amorphous silicon, because it's amorphous, 00:41:51.053 --> 00:41:52.720 align:middle line:84% you no longer have-- essentially, you're 00:41:52.720 --> 00:41:55.120 align:middle line:84% constrained coming from the momentum crystalline momentum 00:41:55.120 --> 00:41:55.940 align:middle line:90% conservation. 00:41:55.940 --> 00:41:58.180 align:middle line:84% It behaves more like direct bandgap semiconductor. 00:41:58.180 --> 00:42:00.472 align:middle line:84% So you can actually absorb pretty much all the sunlight 00:42:00.472 --> 00:42:02.600 align:middle line:90% within just a couple of microns. 00:42:02.600 --> 00:42:06.350 align:middle line:84% So you can save a lot of the semiconductor silicon material 00:42:06.350 --> 00:42:08.510 align:middle line:90% compared to crystalline silicon. 00:42:08.510 --> 00:42:14.053 align:middle line:84% So that's another factor that contributes to reduced costs. 00:42:14.053 --> 00:42:15.470 align:middle line:84% So that's basically the assumption 00:42:15.470 --> 00:42:18.070 align:middle line:84% people are making when they are exploring amorphous silicon 00:42:18.070 --> 00:42:19.510 align:middle line:84% as a solar cell material, that you 00:42:19.510 --> 00:42:22.990 align:middle line:84% can achieve lower cost compared to crystalline silicon. 00:42:22.990 --> 00:42:24.510 align:middle line:90% I put a question mark there. 00:42:24.510 --> 00:42:28.856 align:middle line:84% I'm not going to talk about whether or not that's correct. 00:42:28.856 --> 00:42:31.110 align:middle line:84% Now there are also some additional benefits 00:42:31.110 --> 00:42:32.090 align:middle line:90% of amorphous silicon. 00:42:32.090 --> 00:42:34.550 align:middle line:84% So for example, one very interesting concept people 00:42:34.550 --> 00:42:36.870 align:middle line:84% are exploring here is the so-called monolithic silicon 00:42:36.870 --> 00:42:37.770 align:middle line:90% tandem structures. 00:42:37.770 --> 00:42:39.990 align:middle line:84% So a tandem cell essentially means 00:42:39.990 --> 00:42:41.470 align:middle line:84% that you're stacking semiconductor 00:42:41.470 --> 00:42:42.810 align:middle line:90% of different bandgaps together. 00:42:42.810 --> 00:42:46.270 align:middle line:84% If you look from top to bottom, like sunlight coming 00:42:46.270 --> 00:42:49.030 align:middle line:84% from the top, the top cell is the one that has the largest 00:42:49.030 --> 00:42:51.110 align:middle line:84% bandgap, so it absorbs essentially the blue part 00:42:51.110 --> 00:42:53.450 align:middle line:84% of the spectrum and generates photons, 00:42:53.450 --> 00:42:55.670 align:middle line:84% generates current with very high voltage. 00:42:55.670 --> 00:42:57.530 align:middle line:84% The remaining red photons that continue 00:42:57.530 --> 00:43:00.030 align:middle line:84% to go through, because it's transparent to this wide bandgap 00:43:00.030 --> 00:43:03.070 align:middle line:84% semiconductor, it gets absorbed in a semiconductor with smaller 00:43:03.070 --> 00:43:06.220 align:middle line:84% band gap get underneath, which, again, generates 00:43:06.220 --> 00:43:09.173 align:middle line:84% some additional generally low voltage, but additional current. 00:43:09.173 --> 00:43:11.340 align:middle line:84% You can continue by stacking multiple of these cells 00:43:11.340 --> 00:43:13.132 align:middle line:84% with different bandgaps so that you collect 00:43:13.132 --> 00:43:15.020 align:middle line:84% different part of the spectrum of sunlight 00:43:15.020 --> 00:43:19.792 align:middle line:84% without losing any of the other energy there. 00:43:19.792 --> 00:43:21.580 align:middle line:84% Conventionally, when people are making 00:43:21.580 --> 00:43:22.997 align:middle line:84% this kind of tandem structure, you 00:43:22.997 --> 00:43:26.885 align:middle line:84% have to essentially grow semiconductors epitaxially using 00:43:26.885 --> 00:43:28.260 align:middle line:84% different compositions, typically 00:43:28.260 --> 00:43:32.180 align:middle line:84% in III-V semiconductor systems, which gets quite complicated. 00:43:32.180 --> 00:43:36.315 align:middle line:84% But in the case of silicon, recall that amorphous silicon 00:43:36.315 --> 00:43:37.940 align:middle line:84% actually has a larger band gap compared 00:43:37.940 --> 00:43:41.540 align:middle line:84% to crystalline silicon because of the presence of hydrogen. 00:43:41.540 --> 00:43:43.320 align:middle line:84% In hydrogenated amorphous silicon, 00:43:43.320 --> 00:43:45.403 align:middle line:84% you get lots of hydrogen, which increases the band 00:43:45.403 --> 00:43:47.180 align:middle line:90% gap to about 1.7 EV. 00:43:47.180 --> 00:43:48.920 align:middle line:84% So by controlling the hydrogen content, 00:43:48.920 --> 00:43:51.100 align:middle line:84% you can basically tune the bandgap 00:43:51.100 --> 00:43:52.900 align:middle line:90% of the amorphous silicone. 00:43:52.900 --> 00:43:55.780 align:middle line:84% So in principle, this gives you a method 00:43:55.780 --> 00:43:58.540 align:middle line:84% to realize a monolithic tandem structure 00:43:58.540 --> 00:44:00.420 align:middle line:84% by just putting amorphous silicon on top 00:44:00.420 --> 00:44:02.340 align:middle line:90% of crystalline silicon. 00:44:02.340 --> 00:44:04.310 align:middle line:84% And by changing the hydrogen content, 00:44:04.310 --> 00:44:06.330 align:middle line:84% you can actually get several different bandgaps 00:44:06.330 --> 00:44:08.970 align:middle line:84% so that you can have essentially just one second material, 00:44:08.970 --> 00:44:11.730 align:middle line:84% but you also get this kind of graded band gap, 00:44:11.730 --> 00:44:13.850 align:middle line:84% and therefore create a tandem structure. 00:44:13.850 --> 00:44:16.770 align:middle line:84% So this is quite a unique capability 00:44:16.770 --> 00:44:19.370 align:middle line:84% that other materials will be difficult to actually realize. 00:44:19.370 --> 00:44:24.430 align:middle line:90% 00:44:24.430 --> 00:44:28.810 align:middle line:84% Now let's come back to why amorphous silicone is not 00:44:28.810 --> 00:44:30.050 align:middle line:90% gaining popularity. 00:44:30.050 --> 00:44:32.850 align:middle line:84% One of the first big barriers there 00:44:32.850 --> 00:44:35.730 align:middle line:84% is the so-called Staebler-Wronski effect, 00:44:35.730 --> 00:44:39.250 align:middle line:84% also known as the Light-Induced Degradation, LID. 00:44:39.250 --> 00:44:42.150 align:middle line:84% So this effect was discovered, obviously, 00:44:42.150 --> 00:44:47.250 align:middle line:84% by Staebler and Wronski back in 1977, almost 50 years ago. 00:44:47.250 --> 00:44:49.730 align:middle line:84% What I'm quoting here, this figure 00:44:49.730 --> 00:44:53.050 align:middle line:84% at the bottom left corner, is the original graph 00:44:53.050 --> 00:44:54.970 align:middle line:84% from the paper there where they discovered 00:44:54.970 --> 00:44:57.210 align:middle line:84% this light-induced degradation effect. 00:44:57.210 --> 00:44:58.990 align:middle line:84% So what I'm plotting on this plot here? 00:44:58.990 --> 00:45:01.640 align:middle line:84% Vertical axis is basically electrical conductivity. 00:45:01.640 --> 00:45:05.160 align:middle line:84% Horizontal axis is the light intensity. 00:45:05.160 --> 00:45:09.560 align:middle line:84% So essentially, these curves are showing the photoconductivity 00:45:09.560 --> 00:45:10.452 align:middle line:90% in amorphous silicon. 00:45:10.452 --> 00:45:12.160 align:middle line:84% If you send light into amorphous silicon, 00:45:12.160 --> 00:45:13.920 align:middle line:84% you generate electron-hole pairs. 00:45:13.920 --> 00:45:14.900 align:middle line:90% You have more carriers. 00:45:14.900 --> 00:45:16.780 align:middle line:84% You have essentially a higher conductivity. 00:45:16.780 --> 00:45:19.440 align:middle line:90% That's photoconductivity. 00:45:19.440 --> 00:45:21.700 align:middle line:84% Now what they have noticed is that initially, 00:45:21.700 --> 00:45:23.780 align:middle line:84% if you just freshly prepare amorphous silicon, 00:45:23.780 --> 00:45:25.360 align:middle line:84% hydrogenated amorphous silicon, you 00:45:25.360 --> 00:45:27.720 align:middle line:90% get decent photoconductivity. 00:45:27.720 --> 00:45:31.840 align:middle line:84% But after extended exposure to light, 00:45:31.840 --> 00:45:35.460 align:middle line:84% the photoconductivity drops by orders of magnitude. 00:45:35.460 --> 00:45:39.040 align:middle line:90% 00:45:39.040 --> 00:45:41.860 align:middle line:84% So this is known as a light-induced degradation. 00:45:41.860 --> 00:45:46.520 align:middle line:84% Essentially what happens is that light radiation 00:45:46.520 --> 00:45:51.160 align:middle line:84% can generate defects in amorphous silicon, which 00:45:51.160 --> 00:45:53.920 align:middle line:84% leads to a degradation of the material quality, which 00:45:53.920 --> 00:45:56.365 align:middle line:84% is what compromises the photoconductivity. 00:45:56.365 --> 00:45:58.240 align:middle line:84% What happens in the case of photoconductivity 00:45:58.240 --> 00:45:59.615 align:middle line:84% is that these defects essentially 00:45:59.615 --> 00:46:01.380 align:middle line:90% serve as recombination centers. 00:46:01.380 --> 00:46:04.900 align:middle line:84% So your photogenerated electrons and holes, they will recombine. 00:46:04.900 --> 00:46:07.580 align:middle line:84% They get expedited by these defects. 00:46:07.580 --> 00:46:09.140 align:middle line:84% And therefore, you effectively have 00:46:09.140 --> 00:46:10.660 align:middle line:84% less amount of electrons and holes 00:46:10.660 --> 00:46:12.640 align:middle line:84% available in the defective material. 00:46:12.640 --> 00:46:15.420 align:middle line:84% And that's why the conductivity will drop. 00:46:15.420 --> 00:46:19.280 align:middle line:84% Now if you go to solar cell, the mechanism is very similar. 00:46:19.280 --> 00:46:23.040 align:middle line:84% Because these generated defects increase carrier recombination, 00:46:23.040 --> 00:46:24.540 align:middle line:84% it means that lots of these carriers 00:46:24.540 --> 00:46:26.460 align:middle line:84% will not reach your collection electrodes 00:46:26.460 --> 00:46:29.300 align:middle line:84% to contribute to external current because 00:46:29.300 --> 00:46:30.860 align:middle line:90% of the recombination. 00:46:30.860 --> 00:46:33.260 align:middle line:84% So then, effectively, the efficiency of solar cells 00:46:33.260 --> 00:46:34.160 align:middle line:90% will also drop. 00:46:34.160 --> 00:46:38.500 align:middle line:90% 00:46:38.500 --> 00:46:42.500 align:middle line:84% So this is actually quite a significant effect. 00:46:42.500 --> 00:46:45.482 align:middle line:84% In a typical amorphous silicon panel, 00:46:45.482 --> 00:46:46.940 align:middle line:84% if you look at efficiency when it's 00:46:46.940 --> 00:46:49.887 align:middle line:84% freshly prepared versus when it's actually saturated 00:46:49.887 --> 00:46:51.720 align:middle line:84% with this kind of light induced degradation, 00:46:51.720 --> 00:46:55.260 align:middle line:84% the efficiency can drop from anywhere from 10% to 15%, 00:46:55.260 --> 00:46:57.980 align:middle line:84% in some extreme cases even up to 30%. 00:46:57.980 --> 00:47:01.290 align:middle line:84% So it's quite a significant job of efficiency there. 00:47:01.290 --> 00:47:03.690 align:middle line:90% Now why's that the case? 00:47:03.690 --> 00:47:06.690 align:middle line:90% It goes back to a dangling bond. 00:47:06.690 --> 00:47:11.250 align:middle line:84% We mentioned that dangling bonds is the main defects that 00:47:11.250 --> 00:47:14.627 align:middle line:84% essentially lead to essentially these midgap states that 00:47:14.627 --> 00:47:16.710 align:middle line:84% compromise the electronic properties of a circuit. 00:47:16.710 --> 00:47:18.332 align:middle line:84% You need to pacify them with hydrogen 00:47:18.332 --> 00:47:20.290 align:middle line:84% so that you can actually dope amorphous silicon 00:47:20.290 --> 00:47:22.370 align:middle line:84% and therefore use it as a semiconductor material. 00:47:22.370 --> 00:47:25.410 align:middle line:84% Now what happens during-light induced degradation 00:47:25.410 --> 00:47:32.170 align:middle line:84% is that light actually releases dangling bonds. 00:47:32.170 --> 00:47:35.410 align:middle line:84% So how does light releases dangling bonds? 00:47:35.410 --> 00:47:38.490 align:middle line:84% There are a couple of different mechanisms people have proposed. 00:47:38.490 --> 00:47:43.710 align:middle line:84% One of them-- this is called a bond-switching model. 00:47:43.710 --> 00:47:47.850 align:middle line:90% 00:47:47.850 --> 00:47:50.530 align:middle line:84% So what bond-switching model says 00:47:50.530 --> 00:47:53.890 align:middle line:84% is that-- say this is your amorphous silicon. 00:47:53.890 --> 00:47:57.210 align:middle line:90% Silicon-silicon bonds. 00:47:57.210 --> 00:47:59.020 align:middle line:90% Now you can absorb this light. 00:47:59.020 --> 00:48:01.520 align:middle line:84% The light comes in, you generate electron-hole pairs. 00:48:01.520 --> 00:48:04.640 align:middle line:84% These electron-hole pairs, they're going to recombine. 00:48:04.640 --> 00:48:07.160 align:middle line:84% When they recombine, sometimes they 00:48:07.160 --> 00:48:11.400 align:middle line:84% can pass energy to this silicon-silicon bond. 00:48:11.400 --> 00:48:12.860 align:middle line:84% In the case of crystalline silicon, 00:48:12.860 --> 00:48:15.360 align:middle line:84% all the silicon-silicon bonds are identical from each other. 00:48:15.360 --> 00:48:16.560 align:middle line:90% They are all the same. 00:48:16.560 --> 00:48:18.920 align:middle line:84% However, in amorphous silicon, there 00:48:18.920 --> 00:48:21.360 align:middle line:84% are always some silicon-silicon bonds that are weaker 00:48:21.360 --> 00:48:24.180 align:middle line:84% compared to others, just because they are further apart, 00:48:24.180 --> 00:48:26.042 align:middle line:84% or maybe they are strained, or whatever. 00:48:26.042 --> 00:48:28.000 align:middle line:84% So there are some weaker silicon-silicon bonds. 00:48:28.000 --> 00:48:30.360 align:middle line:84% Now these weaker silicon-silicon bonds 00:48:30.360 --> 00:48:33.360 align:middle line:84% can be severed by this additional energy coming 00:48:33.360 --> 00:48:35.132 align:middle line:84% from electron-hole recombination. 00:48:35.132 --> 00:48:36.840 align:middle line:84% So now what we end up with is essentially 00:48:36.840 --> 00:48:40.900 align:middle line:90% these two unpaired electrons. 00:48:40.900 --> 00:48:44.520 align:middle line:90% 00:48:44.520 --> 00:48:46.720 align:middle line:84% Now over time, these two electrons 00:48:46.720 --> 00:48:49.920 align:middle line:84% can again recombine and form a chemical bond again. 00:48:49.920 --> 00:48:52.240 align:middle line:84% In that case, no defect is going to get generated. 00:48:52.240 --> 00:48:58.110 align:middle line:84% However, in amorphous silicon, you get lots of hydrogen. 00:48:58.110 --> 00:49:00.010 align:middle line:84% So the hydrogen can also be released by light 00:49:00.010 --> 00:49:01.510 align:middle line:84% because the light coming and cutting 00:49:01.510 --> 00:49:02.730 align:middle line:90% the silicon-hydrogen bond. 00:49:02.730 --> 00:49:05.310 align:middle line:84% So then this hydrogen sometimes-- 00:49:05.310 --> 00:49:07.270 align:middle line:84% opportunistically, they can essentially 00:49:07.270 --> 00:49:10.430 align:middle line:84% diffuse next to one silicon and form a chemical bond 00:49:10.430 --> 00:49:13.910 align:middle line:90% with one of the silicon atoms. 00:49:13.910 --> 00:49:17.510 align:middle line:84% Now you are left behind with one unpaired electron that can no 00:49:17.510 --> 00:49:19.910 align:middle line:90% longer restore a chemical bond. 00:49:19.910 --> 00:49:23.373 align:middle line:90% So this is a dangling bond. 00:49:23.373 --> 00:49:25.290 align:middle line:84% That's what's called the bond-switching model. 00:49:25.290 --> 00:49:27.498 align:middle line:84% Essentially you're switching the silicon-silicon bond 00:49:27.498 --> 00:49:32.390 align:middle line:84% to Si-H bond, leaving behind the dangling bond. 00:49:32.390 --> 00:49:38.710 align:middle line:84% The other model is called the hydrogen collision model. 00:49:38.710 --> 00:49:43.830 align:middle line:84% In the hydrogen collision model, you have Si-H bond. 00:49:43.830 --> 00:49:44.950 align:middle line:90% Same thing. 00:49:44.950 --> 00:49:47.350 align:middle line:84% The photon energy generating electron-hole pair 00:49:47.350 --> 00:49:49.147 align:middle line:84% can pass energy to silicon-hydrogen bonds. 00:49:49.147 --> 00:49:51.230 align:middle line:84% You can actually break this silicon-hydrogen bond. 00:49:51.230 --> 00:49:55.050 align:middle line:90% 00:49:55.050 --> 00:49:58.350 align:middle line:84% And this process can happen in lots of different locations 00:49:58.350 --> 00:49:59.970 align:middle line:90% in the actual material. 00:49:59.970 --> 00:50:01.750 align:middle line:84% You release this kind of hydrogen atom, 00:50:01.750 --> 00:50:04.450 align:middle line:84% which is highly mobile, becomes very small. 00:50:04.450 --> 00:50:09.090 align:middle line:84% So then, again, opportunistically this hydrogen 00:50:09.090 --> 00:50:12.050 align:middle line:84% can bump into each other and form this hydrogen complex, 00:50:12.050 --> 00:50:13.030 align:middle line:90% H2 complex. 00:50:13.030 --> 00:50:15.650 align:middle line:84% Now keep in mind this is not hydrogen gas. 00:50:15.650 --> 00:50:17.743 align:middle line:84% This is actually two atoms attached to each other. 00:50:17.743 --> 00:50:19.910 align:middle line:84% But still within the matrix of an armophous circuit. 00:50:19.910 --> 00:50:22.850 align:middle line:84% So this is not really H2 gas molecules. 00:50:22.850 --> 00:50:25.810 align:middle line:84% But in any case, this also leaves behind, essentially, 00:50:25.810 --> 00:50:28.450 align:middle line:84% an unpaired electron or silicon dangling bond. 00:50:28.450 --> 00:50:36.210 align:middle line:90% 00:50:36.210 --> 00:50:38.130 align:middle line:84% So bottom line, this always involves 00:50:38.130 --> 00:50:41.050 align:middle line:84% essentially a light-induced breakage of chemical bonds 00:50:41.050 --> 00:50:43.210 align:middle line:84% and therefore a generation of dangling bonds 00:50:43.210 --> 00:50:44.345 align:middle line:90% in an amorphous circuit. 00:50:44.345 --> 00:50:45.970 align:middle line:84% And these dangling bonds are the defect 00:50:45.970 --> 00:50:48.690 align:middle line:84% that leads to a degradation of the electronic properties 00:50:48.690 --> 00:50:51.690 align:middle line:90% of amorphous silicon. 00:50:51.690 --> 00:50:52.190 align:middle line:90% All right. 00:50:52.190 --> 00:50:55.200 align:middle line:84% Now you can also take a look at more specifically the kinetics 00:50:55.200 --> 00:50:57.100 align:middle line:84% of this kind of light-induced degradation. 00:50:57.100 --> 00:51:00.120 align:middle line:84% So on this plot here, I'm showing you 00:51:00.120 --> 00:51:02.040 align:middle line:84% the measured kinetics of the process, which 00:51:02.040 --> 00:51:04.160 align:middle line:90% this is a pretty crowded plot. 00:51:04.160 --> 00:51:06.380 align:middle line:84% Let's just look at-- forget about the inset. 00:51:06.380 --> 00:51:07.780 align:middle line:90% Let's look at these curves here. 00:51:07.780 --> 00:51:10.160 align:middle line:84% So the vertical axis is the number 00:51:10.160 --> 00:51:12.920 align:middle line:84% of dangling bond defects in log scale. 00:51:12.920 --> 00:51:15.680 align:middle line:84% Horizontal axis is the illumination time, 00:51:15.680 --> 00:51:17.120 align:middle line:90% also in log scale. 00:51:17.120 --> 00:51:19.600 align:middle line:84% You see that these curves-- essentially, they 00:51:19.600 --> 00:51:20.420 align:middle line:90% are all straight. 00:51:20.420 --> 00:51:21.250 align:middle line:90% They're all straight lines. 00:51:21.250 --> 00:51:23.208 align:middle line:84% They correspond to different light intensities. 00:51:23.208 --> 00:51:27.680 align:middle line:84% Now a straight line in this kind of double logarithm plot 00:51:27.680 --> 00:51:29.400 align:middle line:84% essentially means that you have a power 00:51:29.400 --> 00:51:32.320 align:middle line:84% dependence between the two parameters there. 00:51:32.320 --> 00:51:34.400 align:middle line:84% And the exponent in the power dependence 00:51:34.400 --> 00:51:37.160 align:middle line:84% is given by the slope of this curve here. 00:51:37.160 --> 00:51:39.800 align:middle line:84% So if you look at the slope, it actually 00:51:39.800 --> 00:51:42.760 align:middle line:90% corresponds to one third. 00:51:42.760 --> 00:51:45.120 align:middle line:84% So basically what it says is that the dangling bond 00:51:45.120 --> 00:51:50.840 align:middle line:84% generation scales with time to the one third power. 00:51:50.840 --> 00:51:53.790 align:middle line:84% Now if you look at the inset, vertical axis 00:51:53.790 --> 00:51:56.950 align:middle line:84% is still the number of dangling bond defects in log scale. 00:51:56.950 --> 00:52:00.910 align:middle line:84% Horizontal axis is the light density in log scale. 00:52:00.910 --> 00:52:04.110 align:middle line:84% You also have a straight line here. 00:52:04.110 --> 00:52:07.170 align:middle line:84% The straight line has a slope of about 0.6, 00:52:07.170 --> 00:52:10.230 align:middle line:84% according to this experiment here. 00:52:10.230 --> 00:52:13.583 align:middle line:84% Now the horizontal axis, light intensity, light intensity 00:52:13.583 --> 00:52:15.750 align:middle line:84% basically gives you the generation rate of carriers. 00:52:15.750 --> 00:52:18.750 align:middle line:84% Generation rate of this-- essentially the 00:52:18.750 --> 00:52:19.610 align:middle line:90% electron-hole pairs. 00:52:19.610 --> 00:52:23.150 align:middle line:84% So that's what we write as G. G, essentially, 00:52:23.150 --> 00:52:26.870 align:middle line:90% is our light intensity. 00:52:26.870 --> 00:52:29.390 align:middle line:84% If you do some kinetic modeling, if you go ahead and read 00:52:29.390 --> 00:52:30.870 align:middle line:84% the paper, what they did is they-- 00:52:30.870 --> 00:52:33.710 align:middle line:84% based on these different models here, 00:52:33.710 --> 00:52:36.790 align:middle line:84% you can perform a detailed balance study 00:52:36.790 --> 00:52:39.270 align:middle line:84% by writing different kinetic equations. 00:52:39.270 --> 00:52:42.350 align:middle line:84% And then from that, you can write the rate constants. 00:52:42.350 --> 00:52:45.790 align:middle line:84% Then that allows you to provide several equations that 00:52:45.790 --> 00:52:51.380 align:middle line:84% involves the parameters such as external light generation rate. 00:52:51.380 --> 00:52:53.575 align:middle line:84% Generating rate for this kind of-- for example, 00:52:53.575 --> 00:52:54.200 align:middle line:90% dangling bonds. 00:52:54.200 --> 00:52:57.020 align:middle line:84% So if you go ahead and solve those kind 00:52:57.020 --> 00:52:58.460 align:middle line:84% of coupled differential equations 00:52:58.460 --> 00:53:00.380 align:middle line:84% about the rate equations, you're going 00:53:00.380 --> 00:53:03.540 align:middle line:84% to end up with the conclusion that the defect generation 00:53:03.540 --> 00:53:06.500 align:middle line:84% should scale with generation rate to the two thirds 00:53:06.500 --> 00:53:09.540 align:middle line:84% power and time to the one third power. 00:53:09.540 --> 00:53:12.900 align:middle line:84% You notice that actually agrees pretty well with the experiment. 00:53:12.900 --> 00:53:14.820 align:middle line:90% With time, it's one third. 00:53:14.820 --> 00:53:18.420 align:middle line:90% With rate, it's 0.6. 00:53:18.420 --> 00:53:22.420 align:middle line:84% The model predicted 0.67, which is pretty good agreement. 00:53:22.420 --> 00:53:26.980 align:middle line:84% So you generate more defects with higher light flux 00:53:26.980 --> 00:53:29.880 align:middle line:84% and also with a extended exposure, which makes sense. 00:53:29.880 --> 00:53:33.824 align:middle line:90% 00:53:33.824 --> 00:53:37.100 align:middle line:84% Now next thing I also want to mention here 00:53:37.100 --> 00:53:42.660 align:middle line:84% is that you can also partially reverse 00:53:42.660 --> 00:53:46.740 align:middle line:84% this kind of light Staebler-Wronski effect 00:53:46.740 --> 00:53:47.977 align:middle line:90% by essentially annealing. 00:53:47.977 --> 00:53:49.560 align:middle line:84% So what I'm showing on this plot here, 00:53:49.560 --> 00:53:51.930 align:middle line:84% this is actually measured efficiency of a solar panel. 00:53:51.930 --> 00:53:55.690 align:middle line:84% If you just put it out in the sun, then what's going to happen 00:53:55.690 --> 00:53:58.450 align:middle line:84% is that efficiency is going to drop within the first few hours 00:53:58.450 --> 00:53:59.650 align:middle line:90% of light exposure. 00:53:59.650 --> 00:54:01.930 align:middle line:84% But if you knew it, you can essentially 00:54:01.930 --> 00:54:05.130 align:middle line:84% remove some of these dangling defects by annealing there. 00:54:05.130 --> 00:54:08.830 align:middle line:84% Essentially induce diffusion of the hydrogen atoms in the matrix 00:54:08.830 --> 00:54:11.090 align:middle line:84% so that they can combine with the dangling bonds 00:54:11.090 --> 00:54:13.370 align:middle line:84% during the annealing process, which annihilates some 00:54:13.370 --> 00:54:14.670 align:middle line:90% of the dangling bond defects. 00:54:14.670 --> 00:54:17.210 align:middle line:84% You see partial recovery of the efficiency. 00:54:17.210 --> 00:54:18.730 align:middle line:84% You notice it never really goes back 00:54:18.730 --> 00:54:20.330 align:middle line:90% to original efficiency there. 00:54:20.330 --> 00:54:22.945 align:middle line:84% But if you put it out in the sun again, you'll get the grades. 00:54:22.945 --> 00:54:25.070 align:middle line:84% You [INAUDIBLE] again, again it partially recovers, 00:54:25.070 --> 00:54:27.650 align:middle line:84% but then it will degrade furthermore. 00:54:27.650 --> 00:54:29.810 align:middle line:84% Until, at some point, eventually it's 00:54:29.810 --> 00:54:33.050 align:middle line:90% going to reach a steady state. 00:54:33.050 --> 00:54:35.930 align:middle line:84% This stabilized efficiency is what 00:54:35.930 --> 00:54:38.210 align:middle line:84% people should be reporting for the efficiency 00:54:38.210 --> 00:54:40.690 align:middle line:90% of amorphous silicon. 00:54:40.690 --> 00:54:42.950 align:middle line:84% Because for solar cells for solar panels, 00:54:42.950 --> 00:54:46.290 align:middle line:84% you typically call them for a lifetime of 25 to 30 years. 00:54:46.290 --> 00:54:48.710 align:middle line:84% So this first few hours really doesn't count. 00:54:48.710 --> 00:54:50.310 align:middle line:84% What really you should be reporting 00:54:50.310 --> 00:54:53.750 align:middle line:84% is the stabilized efficiency for amorphous silicon cells. 00:54:53.750 --> 00:54:57.090 align:middle line:84% So this is already one big challenge for amorphous silicon 00:54:57.090 --> 00:54:57.590 align:middle line:90% here. 00:54:57.590 --> 00:54:59.465 align:middle line:84% You see that there's a significant efficiency 00:54:59.465 --> 00:55:01.990 align:middle line:84% drop just coming from this light-induced degradation effect 00:55:01.990 --> 00:55:03.430 align:middle line:90% there. 00:55:03.430 --> 00:55:07.350 align:middle line:84% Now let's take a look at what efficiency looks like 00:55:07.350 --> 00:55:10.230 align:middle line:84% for state-of-the-art amorphous silicon cell after 00:55:10.230 --> 00:55:11.060 align:middle line:90% stabilization. 00:55:11.060 --> 00:55:11.810 align:middle line:90% Yeah, I apologize. 00:55:11.810 --> 00:55:12.977 align:middle line:90% The phones are really small. 00:55:12.977 --> 00:55:15.230 align:middle line:90% I know you can't read it. 00:55:15.230 --> 00:55:17.300 align:middle line:84% That's really not my fault. That's 00:55:17.300 --> 00:55:19.550 align:middle line:84% because people have so much interest in new solar cell 00:55:19.550 --> 00:55:20.330 align:middle line:90% materials and stuff. 00:55:20.330 --> 00:55:22.750 align:middle line:84% That's why it gets really crowded as you go to this regime 00:55:22.750 --> 00:55:24.450 align:middle line:90% here. 00:55:24.450 --> 00:55:27.510 align:middle line:84% But I don't think you can really see it. 00:55:27.510 --> 00:55:30.070 align:middle line:90% Let's see if I can zoom in. 00:55:30.070 --> 00:55:30.970 align:middle line:90% Oh, OK. 00:55:30.970 --> 00:55:31.510 align:middle line:90% I cannot. 00:55:31.510 --> 00:55:32.010 align:middle line:90% All right. 00:55:32.010 --> 00:55:33.590 align:middle line:90% Now I can read it on my screen. 00:55:33.590 --> 00:55:35.270 align:middle line:84% I have a very high pixel density. 00:55:35.270 --> 00:55:37.270 align:middle line:84% It's a retina display, so I can actually see it. 00:55:37.270 --> 00:55:40.290 align:middle line:84% I apologize this is not sufficient optical quality. 00:55:40.290 --> 00:55:42.810 align:middle line:84% But this is basically a summary of all the state-of-the-art. 00:55:42.810 --> 00:55:45.090 align:middle line:84% So the world record solar cell efficiency. 00:55:45.090 --> 00:55:47.060 align:middle line:84% Let's try to find amorphous silicon. 00:55:47.060 --> 00:55:48.540 align:middle line:84% Amorphous silicon should be here. 00:55:48.540 --> 00:55:49.820 align:middle line:90% [INAUDIBLE] You can see it. 00:55:49.820 --> 00:55:51.420 align:middle line:90% You can sort of read it. 00:55:51.420 --> 00:55:53.720 align:middle line:84% It's this big green open circle there. 00:55:53.720 --> 00:55:55.260 align:middle line:84% Now you notice it says, amorphous 00:55:55.260 --> 00:55:58.020 align:middle line:90% hydrogenated silicon stabilized. 00:55:58.020 --> 00:56:00.673 align:middle line:84% So it should only report stabilized efficiency. 00:56:00.673 --> 00:56:02.840 align:middle line:84% Now let's look at this big, green, open circle here. 00:56:02.840 --> 00:56:03.860 align:middle line:90% It's right here. 00:56:03.860 --> 00:56:05.460 align:middle line:90% 14%. 00:56:05.460 --> 00:56:06.660 align:middle line:90% Remember that number. 00:56:06.660 --> 00:56:08.420 align:middle line:84% Now what about crystalline silicon? 00:56:08.420 --> 00:56:10.420 align:middle line:84% For single-crystal silicon, you have 00:56:10.420 --> 00:56:12.300 align:middle line:84% two-- these are the square symbols here, 00:56:12.300 --> 00:56:14.220 align:middle line:90% concentrator, nonconcentrator. 00:56:14.220 --> 00:56:16.000 align:middle line:84% You should look at the nonconcentrator 00:56:16.000 --> 00:56:17.700 align:middle line:84% because concentrator actually increases 00:56:17.700 --> 00:56:21.583 align:middle line:84% the effective light flux, which increases the electron density. 00:56:21.583 --> 00:56:23.500 align:middle line:84% It actually increases the open circuit voltage 00:56:23.500 --> 00:56:27.060 align:middle line:84% by increasing the quasi-Fermi level separation. 00:56:27.060 --> 00:56:28.780 align:middle line:90% 3033. 00:56:28.780 --> 00:56:29.800 align:middle line:90% Go back and review that. 00:56:29.800 --> 00:56:32.580 align:middle line:84% But the idea is that you should look at nonconcentrator silicon 00:56:32.580 --> 00:56:34.060 align:middle line:90% solar cells for fair comparison. 00:56:34.060 --> 00:56:35.780 align:middle line:90% So that's a field square. 00:56:35.780 --> 00:56:40.460 align:middle line:84% Field square is right here, 26.1%. 00:56:40.460 --> 00:56:42.087 align:middle line:90% So what is in there? 00:56:42.087 --> 00:56:43.920 align:middle line:84% Crystalline silicon cells have an efficiency 00:56:43.920 --> 00:56:47.530 align:middle line:84% that's almost twice that of amorphous silicon. 00:56:47.530 --> 00:56:50.690 align:middle line:90% Now is that really a problem? 00:56:50.690 --> 00:56:51.450 align:middle line:90% We'll see. 00:56:51.450 --> 00:56:54.090 align:middle line:90% 00:56:54.090 --> 00:56:57.610 align:middle line:84% Now before I get to why amorphous silicon is losing 00:56:57.610 --> 00:57:00.850 align:middle line:84% ground, let's first take a look at this plot 00:57:00.850 --> 00:57:05.090 align:middle line:84% here that shows essentially the global PV production 00:57:05.090 --> 00:57:05.990 align:middle line:90% by technology. 00:57:05.990 --> 00:57:09.462 align:middle line:90% You see most of it-- 00:57:09.462 --> 00:57:11.670 align:middle line:84% so you have three different kinds of technology here. 00:57:11.670 --> 00:57:14.130 align:middle line:84% Monosilicon, this is basically single crystalline silicon. 00:57:14.130 --> 00:57:18.090 align:middle line:84% Multisilicon is essentially polycrystalline silicon, 00:57:18.090 --> 00:57:19.792 align:middle line:84% but they have a large grain sizes. 00:57:19.792 --> 00:57:21.250 align:middle line:84% So with large grain sizes, you have 00:57:21.250 --> 00:57:24.730 align:middle line:84% reduced the amount of crystalline boundary defects. 00:57:24.730 --> 00:57:27.690 align:middle line:84% So you have essentially better quality there right. 00:57:27.690 --> 00:57:29.290 align:middle line:90% Thin film. 00:57:29.290 --> 00:57:31.470 align:middle line:84% This is all type of thin-film technologies, 00:57:31.470 --> 00:57:33.850 align:middle line:90% including amorphous silicon. 00:57:33.850 --> 00:57:39.050 align:middle line:84% So now you can see, as I said, in early 2000s, 2010s, 00:57:39.050 --> 00:57:41.650 align:middle line:84% thin films indeed gained lots of traction. 00:57:41.650 --> 00:57:44.840 align:middle line:84% But then now today, it's pretty much 00:57:44.840 --> 00:57:48.400 align:middle line:90% just single-crystal silicon. 00:57:48.400 --> 00:57:49.980 align:middle line:90% Now let's also further zoom in. 00:57:49.980 --> 00:57:52.683 align:middle line:84% Look at this very small green bar here for thin films, 00:57:52.683 --> 00:57:54.600 align:middle line:84% because there are actually different competing 00:57:54.600 --> 00:57:57.640 align:middle line:90% thin-film technologies. 00:57:57.640 --> 00:58:00.100 align:middle line:84% There are three dominant thin-film technologies, 00:58:00.100 --> 00:58:03.000 align:middle line:84% cadmium telluride, Amorphous silicon, CIGS, 00:58:03.000 --> 00:58:04.780 align:middle line:90% copper indium gallium selenide. 00:58:04.780 --> 00:58:07.320 align:middle line:90% 00:58:07.320 --> 00:58:09.358 align:middle line:90% This is where we are. 00:58:09.358 --> 00:58:11.400 align:middle line:84% I actually tried to zoom in and look at the plot. 00:58:11.400 --> 00:58:13.560 align:middle line:90% It's basically one pixel. 00:58:13.560 --> 00:58:17.000 align:middle line:84% I don't know if it's even real for amorphous silicon. 00:58:17.000 --> 00:58:20.440 align:middle line:84% Cadmium telluride is practically the only thin film technology 00:58:20.440 --> 00:58:21.540 align:middle line:90% that actually survives. 00:58:21.540 --> 00:58:24.760 align:middle line:90% 00:58:24.760 --> 00:58:26.040 align:middle line:90% So sad? 00:58:26.040 --> 00:58:28.400 align:middle line:84% Now let's wonder why it is so sad. 00:58:28.400 --> 00:58:30.260 align:middle line:90% STUDENT: [INAUDIBLE]. 00:58:30.260 --> 00:58:33.400 align:middle line:90% I don't like to see anyone fail. 00:58:33.400 --> 00:58:40.200 align:middle line:84% JUEJUN HU: Well, you see that this is their high days 00:58:40.200 --> 00:58:42.900 align:middle line:84% when you actually get lots of amorphous silicon layers. 00:58:42.900 --> 00:58:46.780 align:middle line:84% But then it quickly drops out over the past 10 years. 00:58:46.780 --> 00:58:48.380 align:middle line:90% So why is that? 00:58:48.380 --> 00:58:50.860 align:middle line:84% Now keep in mind, why did people look at amorphous silicon 00:58:50.860 --> 00:58:52.140 align:middle line:90% in the first place? 00:58:52.140 --> 00:58:53.780 align:middle line:84% The argument was that they can have 00:58:53.780 --> 00:58:55.780 align:middle line:84% a lower cost compared to crystalline silicon 00:58:55.780 --> 00:58:57.220 align:middle line:90% for the solar panels. 00:58:57.220 --> 00:58:59.040 align:middle line:84% Because you can do large area deposition, 00:58:59.040 --> 00:59:00.280 align:middle line:90% high throughput fabrication. 00:59:00.280 --> 00:59:02.460 align:middle line:84% You also need less semiconductor material. 00:59:02.460 --> 00:59:03.893 align:middle line:90% Is that true? 00:59:03.893 --> 00:59:05.560 align:middle line:84% So what am I plotting in this plot here? 00:59:05.560 --> 00:59:09.005 align:middle line:84% This is actually the manufacturing cost. 00:59:09.005 --> 00:59:10.880 align:middle line:84% Sorry, not manufacturing cost, but the price. 00:59:10.880 --> 00:59:15.060 align:middle line:84% The module price, the sale price for modules. 00:59:15.060 --> 00:59:18.480 align:middle line:84% I don't have a number beyond 2014, 00:59:18.480 --> 00:59:20.113 align:middle line:90% but the trend is pretty similar. 00:59:20.113 --> 00:59:21.780 align:middle line:84% If you look at the curves here, this one 00:59:21.780 --> 00:59:25.040 align:middle line:84% is thin-film silicon, amorphous silicon. 00:59:25.040 --> 00:59:28.980 align:middle line:84% These are actually crystalline silicon by different suppliers. 00:59:28.980 --> 00:59:31.580 align:middle line:84% Now what you see here is that indeed 00:59:31.580 --> 00:59:33.980 align:middle line:84% amorphous silicon solar panels are 00:59:33.980 --> 00:59:36.920 align:middle line:84% cheaper than the crystalline counterpart. 00:59:36.920 --> 00:59:39.055 align:middle line:84% Now keep in mind this is already normalized 00:59:39.055 --> 00:59:40.180 align:middle line:90% with respect to efficiency. 00:59:40.180 --> 00:59:42.013 align:middle line:84% You may argue that they have low efficiency. 00:59:42.013 --> 00:59:44.570 align:middle line:84% But what we're showing here, this is dollars per Watt. 00:59:44.570 --> 00:59:48.610 align:middle line:84% So when you're doing this kind of dollar per Watt calculation, 00:59:48.610 --> 00:59:53.770 align:middle line:84% this is basically dollar per Watt calculation, what you have? 00:59:53.770 --> 01:00:02.810 align:middle line:84% The numerator is your efficiency in percent. 01:00:02.810 --> 01:00:06.350 align:middle line:84% Now on top, this will be your module cost. 01:00:06.350 --> 01:00:12.330 align:middle line:90% 01:00:12.330 --> 01:00:19.770 align:middle line:84% This would be the cost of your module. 01:00:19.770 --> 01:00:22.740 align:middle line:84% So you can argue that amorphous silicon has a lower cost. 01:00:22.740 --> 01:00:24.490 align:middle line:84% But it's really already taken into account 01:00:24.490 --> 01:00:27.730 align:middle line:90% in this expression here. 01:00:27.730 --> 01:00:31.730 align:middle line:84% So indeed, the assumption was at least partially correct, 01:00:31.730 --> 01:00:37.930 align:middle line:84% that the modules are cheaper for amorphous silicon. 01:00:37.930 --> 01:00:41.480 align:middle line:90% So then why did they fail? 01:00:41.480 --> 01:00:42.440 align:middle line:90% Yes. 01:00:42.440 --> 01:00:45.880 align:middle line:84% STUDENT: Is [INAUDIBLE] you have to cover-- 01:00:45.880 --> 01:00:47.000 align:middle line:90% JUEJUN HU: Very close. 01:00:47.000 --> 01:00:48.458 align:middle line:90% Very, very, very close. 01:00:48.458 --> 01:00:50.000 align:middle line:84% Well, there's more than that, but you 01:00:50.000 --> 01:00:51.480 align:middle line:84% are going in the right direction, 01:00:51.480 --> 01:00:53.720 align:middle line:84% because there are other additional costs. 01:00:53.720 --> 01:00:57.240 align:middle line:84% A solar system requires not just the module itself. 01:00:57.240 --> 01:00:58.960 align:middle line:84% So what I'm showing you here, this 01:00:58.960 --> 01:01:02.820 align:middle line:84% is basically the total price for the systems, 01:01:02.820 --> 01:01:05.440 align:middle line:84% not just the solar module, not the solar panels. 01:01:05.440 --> 01:01:07.960 align:middle line:84% You see there are actually two components. 01:01:07.960 --> 01:01:11.220 align:middle line:84% This orange bar corresponds to the actual solar modules, 01:01:11.220 --> 01:01:12.400 align:middle line:90% the panels. 01:01:12.400 --> 01:01:16.020 align:middle line:84% This blue bar is what's called a balance of systems, 01:01:16.020 --> 01:01:18.160 align:middle line:84% which is everything else outside the panel. 01:01:18.160 --> 01:01:21.520 align:middle line:84% You see that back in like early 2000s, 01:01:21.520 --> 01:01:24.340 align:middle line:84% the dominant price comes from the modules. 01:01:24.340 --> 01:01:25.660 align:middle line:90% Modules are the expensive ones. 01:01:25.660 --> 01:01:29.280 align:middle line:84% But over the years, modules get really cheap. 01:01:29.280 --> 01:01:31.880 align:middle line:84% But the problem is that this blue bar, the balance 01:01:31.880 --> 01:01:34.440 align:middle line:90% of the system, did not scale. 01:01:34.440 --> 01:01:36.757 align:middle line:84% So what are these balance of systems? 01:01:36.757 --> 01:01:39.340 align:middle line:84% Now for that, we have to look at essentially the whole system. 01:01:39.340 --> 01:01:41.750 align:middle line:84% So if you look at the whole solar system here, 01:01:41.750 --> 01:01:43.590 align:middle line:84% you essentially start out with cells. 01:01:43.590 --> 01:01:46.110 align:middle line:84% These are the individual wafers or individual substrates. 01:01:46.110 --> 01:01:48.808 align:middle line:84% You have to string them together in series or parallel 01:01:48.808 --> 01:01:50.350 align:middle line:84% to form a module so that you can have 01:01:50.350 --> 01:01:52.130 align:middle line:84% sufficient current or voltage output. 01:01:52.130 --> 01:01:54.410 align:middle line:84% The module can be further integrated. 01:01:54.410 --> 01:01:56.373 align:middle line:84% It can be tied together to form an array. 01:01:56.373 --> 01:01:57.790 align:middle line:84% And then in the end, you also need 01:01:57.790 --> 01:01:59.870 align:middle line:84% to put them together to form a system. 01:01:59.870 --> 01:02:02.310 align:middle line:84% To get the system, you get multiple different kinds 01:02:02.310 --> 01:02:03.650 align:middle line:90% of components there. 01:02:03.650 --> 01:02:05.790 align:middle line:84% You need to have this kind of charge controller. 01:02:05.790 --> 01:02:07.510 align:middle line:84% You need an operating point controller 01:02:07.510 --> 01:02:08.670 align:middle line:84% so that you ensure that you're always 01:02:08.670 --> 01:02:10.350 align:middle line:84% operating at the right voltage and right current 01:02:10.350 --> 01:02:11.610 align:middle line:90% to maximize efficiency. 01:02:11.610 --> 01:02:14.150 align:middle line:84% You also need to have an inverter, 01:02:14.150 --> 01:02:17.430 align:middle line:84% because the current coming from solar cells are DC. 01:02:17.430 --> 01:02:19.148 align:middle line:90% But all the grids use AC. 01:02:19.148 --> 01:02:21.190 align:middle line:84% So you need an inverter to essentially convert DC 01:02:21.190 --> 01:02:24.110 align:middle line:90% into AC and vice versa. 01:02:24.110 --> 01:02:25.430 align:middle line:90% DC to AC, basically. 01:02:25.430 --> 01:02:29.470 align:middle line:84% And then you also have all the additional hardwares. 01:02:29.470 --> 01:02:32.790 align:middle line:84% So these are all captured by the essentially so-called balance 01:02:32.790 --> 01:02:33.710 align:middle line:90% of systems. 01:02:33.710 --> 01:02:35.110 align:middle line:84% So in this plot here, I'm looking 01:02:35.110 --> 01:02:37.172 align:middle line:84% at the hardware components for balance of system. 01:02:37.172 --> 01:02:39.130 align:middle line:84% You can see there that lots of different things 01:02:39.130 --> 01:02:40.210 align:middle line:90% besides the modules. 01:02:40.210 --> 01:02:42.830 align:middle line:84% You have even mechanical mounting frames. 01:02:42.830 --> 01:02:45.310 align:middle line:84% You have all the cabling, you have junction boxes, 01:02:45.310 --> 01:02:47.270 align:middle line:84% you have all this power grid infrastructure. 01:02:47.270 --> 01:02:48.430 align:middle line:90% You have switch boxes. 01:02:48.430 --> 01:02:49.930 align:middle line:84% So there are lots of things you have 01:02:49.930 --> 01:02:52.970 align:middle line:84% to invest in order for solar cells 01:02:52.970 --> 01:02:55.890 align:middle line:84% to be useful to generate electricity. 01:02:55.890 --> 01:02:58.130 align:middle line:90% And this is only the hardware. 01:02:58.130 --> 01:03:00.930 align:middle line:84% There's also what's called a soft cost that 01:03:00.930 --> 01:03:02.145 align:middle line:90% goes into balance of systems. 01:03:02.145 --> 01:03:03.270 align:middle line:90% So what are the soft costs? 01:03:03.270 --> 01:03:05.650 align:middle line:84% As you said, you need to acquire land. 01:03:05.650 --> 01:03:09.730 align:middle line:84% You need to also essentially pay for people, staff time, 01:03:09.730 --> 01:03:11.830 align:middle line:84% personnel to actually install the solar cells. 01:03:11.830 --> 01:03:13.370 align:middle line:90% You need maintenance. 01:03:13.370 --> 01:03:15.598 align:middle line:84% Also, all the companies, they need to have profit. 01:03:15.598 --> 01:03:16.890 align:middle line:90% Otherwise, they cannot survive. 01:03:16.890 --> 01:03:20.170 align:middle line:90% They need to pay sales tax. 01:03:20.170 --> 01:03:22.970 align:middle line:84% So all of those things, they all go into the balance 01:03:22.970 --> 01:03:25.090 align:middle line:90% of systems as soft cost. 01:03:25.090 --> 01:03:27.770 align:middle line:90% Now you see what the problem is. 01:03:27.770 --> 01:03:30.610 align:middle line:84% Because if you look at the real dollar per one number, 01:03:30.610 --> 01:03:32.770 align:middle line:84% what you should really be gauging here 01:03:32.770 --> 01:03:34.810 align:middle line:90% is the system level cost. 01:03:34.810 --> 01:03:38.960 align:middle line:84% So you should also put in the numerator here the cost 01:03:38.960 --> 01:03:44.080 align:middle line:84% coming from a balance of systems. 01:03:44.080 --> 01:03:45.800 align:middle line:90% And now you see the problem. 01:03:45.800 --> 01:03:50.400 align:middle line:84% With low efficiency, the relative cost 01:03:50.400 --> 01:03:52.980 align:middle line:84% coming from balance of systems essentially increases. 01:03:52.980 --> 01:04:10.933 align:middle line:90% 01:04:10.933 --> 01:04:12.600 align:middle line:84% And I want to argue that that's actually 01:04:12.600 --> 01:04:15.100 align:middle line:84% a pretty universal trend, not just for rooftop solar system. 01:04:15.100 --> 01:04:17.058 align:middle line:84% In the previous slide, I showed you [INAUDIBLE] 01:04:17.058 --> 01:04:19.440 align:middle line:84% for rooftop solar systems, and the dominant cost 01:04:19.440 --> 01:04:21.200 align:middle line:90% comes from balance of systems. 01:04:21.200 --> 01:04:24.320 align:middle line:84% It's actually the same whether it's residential, or rooftop, 01:04:24.320 --> 01:04:26.720 align:middle line:84% or it's commercial solar cell systems, like for parking 01:04:26.720 --> 01:04:29.140 align:middle line:84% lots and stuff, or utility scale, 01:04:29.140 --> 01:04:33.680 align:middle line:84% which these are big power plants in the desert in Arizona 01:04:33.680 --> 01:04:34.420 align:middle line:90% or whatever. 01:04:34.420 --> 01:04:36.990 align:middle line:84% So in all cases, what I'm showing in this plot 01:04:36.990 --> 01:04:38.570 align:middle line:84% here-- again, this is pretty crowded. 01:04:38.570 --> 01:04:41.670 align:middle line:84% But if you look at the different cost breakup there, 01:04:41.670 --> 01:04:45.990 align:middle line:84% the yellow part is actually the actual module of the panel. 01:04:45.990 --> 01:04:47.750 align:middle line:84% You then have inverters, the balance 01:04:47.750 --> 01:04:50.090 align:middle line:84% of system hardware, other hardware, outside inverter. 01:04:50.090 --> 01:04:50.690 align:middle line:90% You get labor. 01:04:50.690 --> 01:04:52.090 align:middle line:84% You get other kind of soft costs. 01:04:52.090 --> 01:04:54.270 align:middle line:84% But all of these things essentially 01:04:54.270 --> 01:04:56.990 align:middle line:84% are the balanced of system in general. 01:04:56.990 --> 01:05:01.750 align:middle line:84% So you can clearly see that this module cost has decreased a lot. 01:05:01.750 --> 01:05:05.310 align:middle line:84% In all three cases, the module cost 01:05:05.310 --> 01:05:08.190 align:middle line:84% is really not the dominant component for the overall system 01:05:08.190 --> 01:05:09.948 align:middle line:90% cost. 01:05:09.948 --> 01:05:11.990 align:middle line:84% In fact, sometimes you go to the solar conference 01:05:11.990 --> 01:05:14.410 align:middle line:84% and you hear people saying that cells are for free. 01:05:14.410 --> 01:05:19.350 align:middle line:84% They even say that, because it's such a small fraction there. 01:05:19.350 --> 01:05:20.950 align:middle line:84% The real opportunity for cost saving 01:05:20.950 --> 01:05:22.790 align:middle line:90% comes from balance of systems. 01:05:22.790 --> 01:05:26.870 align:middle line:84% Now one way to more efficiently decrease 01:05:26.870 --> 01:05:30.745 align:middle line:84% the balance of system cost is by increasing efficiency, 01:05:30.745 --> 01:05:31.870 align:middle line:90% which makes a lot of sense. 01:05:31.870 --> 01:05:34.928 align:middle line:84% For the same solar panel, the installation cost 01:05:34.928 --> 01:05:35.720 align:middle line:90% is basically fixed. 01:05:35.720 --> 01:05:37.220 align:middle line:84% It scales the area of solar cells. 01:05:37.220 --> 01:05:40.060 align:middle line:84% The land acquisition scales the area of solar cells. 01:05:40.060 --> 01:05:42.000 align:middle line:84% So a lot of those costs are actually fixed. 01:05:42.000 --> 01:05:44.340 align:middle line:84% The scale is essentially the area. 01:05:44.340 --> 01:05:46.260 align:middle line:84% So if you have a high efficiency, 01:05:46.260 --> 01:05:49.780 align:middle line:84% you can essentially still amortize all those costs. 01:05:49.780 --> 01:05:52.660 align:middle line:84% And then therefore, you will be able to achieve a much lower 01:05:52.660 --> 01:05:55.700 align:middle line:90% overall dollar per watt number. 01:05:55.700 --> 01:05:58.720 align:middle line:84% So that's why efficiency is actually important. 01:05:58.720 --> 01:06:00.720 align:middle line:84% It's not just that you produce more electricity. 01:06:00.720 --> 01:06:03.220 align:middle line:84% It's that effectively, you are saving your cost there 01:06:03.220 --> 01:06:04.160 align:middle line:90% for your system. 01:06:04.160 --> 01:06:08.060 align:middle line:90% 01:06:08.060 --> 01:06:10.180 align:middle line:84% And that's also why people spend lots 01:06:10.180 --> 01:06:12.515 align:middle line:84% of-- essentially invest lots of money looking 01:06:12.515 --> 01:06:15.140 align:middle line:84% into new solar cell systems that potentially can give them even 01:06:15.140 --> 01:06:16.432 align:middle line:90% higher efficiency than silicon. 01:06:16.432 --> 01:06:17.900 align:middle line:84% Like recent years, a lot of people 01:06:17.900 --> 01:06:22.350 align:middle line:84% are looking into perovskite silicon tandem cells. 01:06:22.350 --> 01:06:24.100 align:middle line:84% Because if you put perovskite on silicon-- 01:06:24.100 --> 01:06:25.558 align:middle line:84% and if you go back to the plot, you 01:06:25.558 --> 01:06:29.100 align:middle line:84% can probably see that perovskite-silicon tandem. 01:06:29.100 --> 01:06:29.860 align:middle line:90% Let's see. 01:06:29.860 --> 01:06:31.440 align:middle line:90% Perovskite silicon tandem. 01:06:31.440 --> 01:06:34.210 align:middle line:90% Where is it? 01:06:34.210 --> 01:06:35.910 align:middle line:84% Where is the perovskite-silicon tandem? 01:06:35.910 --> 01:06:37.130 align:middle line:90% Right here. 01:06:37.130 --> 01:06:43.630 align:middle line:84% So this is a perovskite-silicon, this open yellow triangle here. 01:06:43.630 --> 01:06:46.170 align:middle line:90% 34.6. 01:06:46.170 --> 01:06:47.597 align:middle line:90% I mean, this is lab results. 01:06:47.597 --> 01:06:49.430 align:middle line:84% It's not like production, largescale models. 01:06:49.430 --> 01:06:51.263 align:middle line:84% But you can already see the potential there. 01:06:51.263 --> 01:06:53.853 align:middle line:84% You can increase more than 8% of efficiency 01:06:53.853 --> 01:06:55.770 align:middle line:84% compared to just silicon by putting perovskite 01:06:55.770 --> 01:06:57.917 align:middle line:90% on top of silicon. 01:06:57.917 --> 01:06:59.750 align:middle line:84% So that's why it's showing lots of interest. 01:06:59.750 --> 01:07:01.792 align:middle line:84% It's not just because of perovskite cells itself. 01:07:01.792 --> 01:07:03.875 align:middle line:84% It's actually because the potential of integrating 01:07:03.875 --> 01:07:05.370 align:middle line:90% perovskite with silicon. 01:07:05.370 --> 01:07:09.690 align:middle line:84% In this case, silicon basically comes for free, practically. 01:07:09.690 --> 01:07:11.450 align:middle line:84% So they are really looking for adding 01:07:11.450 --> 01:07:12.783 align:middle line:90% a few percent of efficiency. 01:07:12.783 --> 01:07:14.950 align:middle line:84% That's why, because of the cost-saving aspect there. 01:07:14.950 --> 01:07:17.690 align:middle line:84% And that's also partially why a lot of the other solar cell 01:07:17.690 --> 01:07:19.110 align:middle line:84% technologies with low efficiency, 01:07:19.110 --> 01:07:22.970 align:middle line:84% they never really take off, like organic PV, basically 01:07:22.970 --> 01:07:24.095 align:middle line:90% because of the same reason. 01:07:24.095 --> 01:07:24.595 align:middle line:90% Yeah. 01:07:24.595 --> 01:07:27.150 align:middle line:84% It's very difficult to match with other kind of technologies 01:07:27.150 --> 01:07:30.210 align:middle line:84% Even though you have the premise that some solar cell technology 01:07:30.210 --> 01:07:32.790 align:middle line:84% that can be very low cost, but if you don't have decent 01:07:32.790 --> 01:07:34.950 align:middle line:90% efficiency, you cannot win. 01:07:34.950 --> 01:07:37.458 align:middle line:90% 01:07:37.458 --> 01:07:39.750 align:middle line:84% Now does that mean that amorphous silicon is completely 01:07:39.750 --> 01:07:41.630 align:middle line:84% eliminated from the picture of solar cells, 01:07:41.630 --> 01:07:43.047 align:middle line:90% or from the solar cell industry? 01:07:43.047 --> 01:07:44.130 align:middle line:90% I mean, not quite exactly. 01:07:44.130 --> 01:07:46.930 align:middle line:84% They still have some unique applications there. 01:07:46.930 --> 01:07:48.970 align:middle line:84% So one example here, for example, 01:07:48.970 --> 01:07:50.310 align:middle line:84% what I'm showing you here, this is actually a silicon 01:07:50.310 --> 01:07:51.570 align:middle line:90% heterojunction solar cell. 01:07:51.570 --> 01:07:54.990 align:middle line:84% So normally in a typical silicon solar cell, 01:07:54.990 --> 01:07:57.790 align:middle line:84% you have a homojunction made of silicon, p-type silicon n type 01:07:57.790 --> 01:07:59.850 align:middle line:84% silicon, both made of crystalline silicon. 01:07:59.850 --> 01:08:04.830 align:middle line:84% In a heterojunction cell, one of the doped regions 01:08:04.830 --> 01:08:07.470 align:middle line:90% is actually amorphous silicon. 01:08:07.470 --> 01:08:09.150 align:middle line:84% Now the reason they like to use it 01:08:09.150 --> 01:08:12.590 align:middle line:84% is because amorphous silicon actually has a larger band gap. 01:08:12.590 --> 01:08:15.290 align:middle line:84% So if you put the junction, for example, on the surface here, 01:08:15.290 --> 01:08:16.707 align:middle line:84% where one of the junction material 01:08:16.707 --> 01:08:20.470 align:middle line:84% is amorphous silicon because the larger band gap essentially 01:08:20.470 --> 01:08:22.990 align:middle line:84% creates a barrier towards carrier 01:08:22.990 --> 01:08:25.229 align:middle line:90% diffusion to the surface. 01:08:25.229 --> 01:08:27.048 align:middle line:84% And this is actually useful because surface 01:08:27.048 --> 01:08:28.590 align:middle line:84% is where you get lots of-- typically, 01:08:28.590 --> 01:08:30.939 align:middle line:84% you get lots of defects because of, essentially, 01:08:30.939 --> 01:08:33.060 align:middle line:90% the unsatisfied chemical bonds. 01:08:33.060 --> 01:08:34.819 align:middle line:84% Those can cause carrier recombination. 01:08:34.819 --> 01:08:37.680 align:middle line:84% So by having a wide bandgap material on the surface, 01:08:37.680 --> 01:08:40.100 align:middle line:84% you can push away carriers from the surface 01:08:40.100 --> 01:08:45.020 align:middle line:84% and therefore mitigate this kind of surface recombination. 01:08:45.020 --> 01:08:47.279 align:middle line:84% On top of that, because amorphous silicon, 01:08:47.279 --> 01:08:49.939 align:middle line:84% you get lots of hydrogen due to the PECVD process, 01:08:49.939 --> 01:08:53.060 align:middle line:84% those hydrogens can also diffuse to the amorphous silicon-silicon 01:08:53.060 --> 01:08:53.939 align:middle line:90% interface. 01:08:53.939 --> 01:08:55.740 align:middle line:84% And then for passivate the surface 01:08:55.740 --> 01:08:57.279 align:middle line:84% with those kind of hydrogen atoms. 01:08:57.279 --> 01:08:58.862 align:middle line:84% So that's an additional benefit there. 01:08:58.862 --> 01:09:01.260 align:middle line:84% So amorphous silicon is still getting 01:09:01.260 --> 01:09:02.899 align:middle line:84% used in the solar industry, but only 01:09:02.899 --> 01:09:05.640 align:middle line:84% for these more specialized, niche applications, 01:09:05.640 --> 01:09:08.880 align:middle line:84% not using them as the solar absorber itself directly. 01:09:08.880 --> 01:09:13.500 align:middle line:90% 01:09:13.500 --> 01:09:14.040 align:middle line:90% All right. 01:09:14.040 --> 01:09:16.640 align:middle line:84% I think that's pretty much all we have for this lecture here. 01:09:16.640 --> 01:09:19.120 align:middle line:84% We have basically talked about amorphous silicon here. 01:09:19.120 --> 01:09:21.482 align:middle line:84% We mentioned about hydrogen dangling bonds. 01:09:21.482 --> 01:09:23.899 align:middle line:84% Dangling bonds are really important kind of defects there. 01:09:23.899 --> 01:09:25.660 align:middle line:84% In order to get amorphous silicon to be useful, 01:09:25.660 --> 01:09:28.187 align:middle line:84% you really need to passivate this kind of dangling bond, OK? 01:09:28.187 --> 01:09:29.770 align:middle line:84% If you don't passivate dangling bonds, 01:09:29.770 --> 01:09:32.609 align:middle line:84% one example is this Staebler-Wronski effect. 01:09:32.609 --> 01:09:34.430 align:middle line:84% If you release new dangling bond defects, 01:09:34.430 --> 01:09:37.050 align:middle line:84% you're going to get to lead to a degradation of the solar cell 01:09:37.050 --> 01:09:37.890 align:middle line:90% efficiency. 01:09:37.890 --> 01:09:43.810 align:middle line:84% We have mentioned that amorphous silicon-- now active matrix 01:09:43.810 --> 01:09:45.722 align:middle line:84% displays based on TFTs, this is probably 01:09:45.722 --> 01:09:47.930 align:middle line:84% the dominant industry where lots of amorphous silicon 01:09:47.930 --> 01:09:49.729 align:middle line:90% is still getting used. 01:09:49.729 --> 01:09:53.010 align:middle line:84% Solar cells, we have seen why it's losing favor 01:09:53.010 --> 01:09:54.250 align:middle line:90% against crystalline silicon. 01:09:54.250 --> 01:09:55.970 align:middle line:84% Again, because of the combination 01:09:55.970 --> 01:09:57.810 align:middle line:84% of light-induced degradation as well as 01:09:57.810 --> 01:10:03.030 align:middle line:84% the lower efficiency compared to crystalline silicon. 01:10:03.030 --> 01:10:04.327 align:middle line:90% Any more questions about-- 01:10:04.327 --> 01:10:04.910 align:middle line:90% STUDENT: Yeah. 01:10:04.910 --> 01:10:11.010 align:middle line:90% 01:10:11.010 --> 01:10:16.870 align:middle line:84% When you annea, and then you get back some of your efficiency, 01:10:16.870 --> 01:10:19.303 align:middle line:84% and you said that there was a stabilization point? 01:10:19.303 --> 01:10:19.970 align:middle line:90% JUEJUN HU: Yeah. 01:10:19.970 --> 01:10:23.550 align:middle line:84% STUDENT: After how many annealings? 01:10:23.550 --> 01:10:26.303 align:middle line:84% I feel like a limited number of annealing before you just-- 01:10:26.303 --> 01:10:26.970 align:middle line:90% JUEJUN HU: Yeah. 01:10:26.970 --> 01:10:30.220 align:middle line:84% When I say "stabilization," I don't mean, actually, annealing. 01:10:30.220 --> 01:10:33.820 align:middle line:84% Essentially, it's like if you don't do annealing, 01:10:33.820 --> 01:10:36.240 align:middle line:84% you just expose it to sunlight, eventually it's 01:10:36.240 --> 01:10:37.240 align:middle line:90% going to stabilize. 01:10:37.240 --> 01:10:38.880 align:middle line:90% That's what I mean. 01:10:38.880 --> 01:10:39.400 align:middle line:90% Yeah. 01:10:39.400 --> 01:10:42.033 align:middle line:84% Annealing, as I said, can help to partially recover 01:10:42.033 --> 01:10:44.200 align:middle line:84% some efficiency because annealing essentially allows 01:10:44.200 --> 01:10:45.720 align:middle line:90% the hydrogen to diffuse around. 01:10:45.720 --> 01:10:47.215 align:middle line:84% As they diffuse around, they find, 01:10:47.215 --> 01:10:48.840 align:middle line:84% essentially, some of the dangling bonds 01:10:48.840 --> 01:10:50.757 align:middle line:84% recombine with dangling bonds to passivate it. 01:10:50.757 --> 01:10:54.020 align:middle line:84% But people don't do it practically. 01:10:54.020 --> 01:10:55.660 align:middle line:84% They just essentially let it stabilize. 01:10:55.660 --> 01:10:56.160 align:middle line:90% STUDENT: OK. 01:10:56.160 --> 01:10:59.440 align:middle line:84% So these stabilizations are independent of the annealing. 01:10:59.440 --> 01:11:02.680 align:middle line:84% JUEJUN HU: Independent of the annealing process, yeah. 01:11:02.680 --> 01:11:05.240 align:middle line:84% STUDENT: Right at the start of this section on solar cells, 01:11:05.240 --> 01:11:09.680 align:middle line:84% you mentioned the layered structures 01:11:09.680 --> 01:11:12.687 align:middle line:84% of crystalline and amorphous silicon. 01:11:12.687 --> 01:11:13.520 align:middle line:90% JUEJUN HU: This one? 01:11:13.520 --> 01:11:16.117 align:middle line:90% In the last slide or what? 01:11:16.117 --> 01:11:16.700 align:middle line:90% STUDENT: Yeah. 01:11:16.700 --> 01:11:20.360 align:middle line:90% I mean, also in the front. 01:11:20.360 --> 01:11:23.400 align:middle line:84% It seemed similar, like, where the wider bandgap kind 01:11:23.400 --> 01:11:25.370 align:middle line:84% allows for the filtering effect where 01:11:25.370 --> 01:11:26.620 align:middle line:90% you can have a higher voltage. 01:11:26.620 --> 01:11:27.840 align:middle line:90% JUEJUN HU: Oh, yeah. 01:11:27.840 --> 01:11:28.480 align:middle line:90% The tandem? 01:11:28.480 --> 01:11:28.740 align:middle line:90% STUDENT: Yeah. 01:11:28.740 --> 01:11:29.900 align:middle line:84% JUEJUN HU: The tandem thing, yeah. 01:11:29.900 --> 01:11:30.483 align:middle line:90% STUDENT: Yeah. 01:11:30.483 --> 01:11:38.580 align:middle line:84% So why does that just not work, or why is that not in use? 01:11:38.580 --> 01:11:40.420 align:middle line:90% JUEJUN HU: It's complicated. 01:11:40.420 --> 01:11:44.740 align:middle line:84% In fact, in order to get the tandem cell to work properly, 01:11:44.740 --> 01:11:48.352 align:middle line:84% you need to essentially make separate junctions. 01:11:48.352 --> 01:11:50.060 align:middle line:84% You also need to have tunneling junctions 01:11:50.060 --> 01:11:52.480 align:middle line:84% to ensure that the current actually flows through, 01:11:52.480 --> 01:11:54.160 align:middle line:90% and then you build up the bias. 01:11:54.160 --> 01:11:57.890 align:middle line:84% You essentially need to have cells working theories. 01:11:57.890 --> 01:11:59.640 align:middle line:84% It's actually a quite complicated process. 01:11:59.640 --> 01:12:01.740 align:middle line:84% It's not nearly as simple as just 01:12:01.740 --> 01:12:03.500 align:middle line:90% deposit the different layers. 01:12:03.500 --> 01:12:04.940 align:middle line:90% You have gradient bandgap. 01:12:04.940 --> 01:12:06.360 align:middle line:90% It's going to work magically. 01:12:06.360 --> 01:12:08.760 align:middle line:84% You need to essentially make separate solar cells 01:12:08.760 --> 01:12:11.340 align:middle line:84% essentially, one solar cell on top of each other, each 01:12:11.340 --> 01:12:14.240 align:middle line:84% with complete structures of p-n junctions 01:12:14.240 --> 01:12:16.780 align:middle line:84% so that they are essentially strung together in series 01:12:16.780 --> 01:12:18.900 align:middle line:90% to add up the voltage. 01:12:18.900 --> 01:12:20.280 align:middle line:90% It's nontrivial in the end. 01:12:20.280 --> 01:12:21.660 align:middle line:90% So it's a nice concept-- 01:12:21.660 --> 01:12:24.440 align:middle line:84% STUDENT: Like, the efficiency gains are not actually worth it? 01:12:24.440 --> 01:12:25.107 align:middle line:90% JUEJUN HU: Yeah. 01:12:25.107 --> 01:12:28.210 align:middle line:84% It's a nice concept, but people do not really do it in the end. 01:12:28.210 --> 01:12:29.760 align:middle line:90% Yeah. 01:12:29.760 --> 01:13:25.000 align:middle line:90%