0 00:00:00,000 --> 00:00:30,000 Dear viewer, these subtitles were generated by a machine via the service Trint and therefore are (very) buggy. If you are capable, please help us to create good quality subtitles: https://c3subtitles.de/talk/857 Thanks! 1 00:00:14,600 --> 00:00:15,600 Welcome, Ari. 2 00:00:24,930 --> 00:00:27,059 Oh, actually, one thing 3 00:00:27,060 --> 00:00:29,459 I sort of don't have my notes here 4 00:00:29,460 --> 00:00:30,480 they are my backpack. 5 00:00:34,710 --> 00:00:35,710 Thank you very much. 6 00:00:43,930 --> 00:00:47,199 Great. Now, I hope we are actually ready. 7 00:00:47,200 --> 00:00:49,269 So let's talk about 8 00:00:49,270 --> 00:00:51,279 it's about the making of a chip and I 9 00:00:51,280 --> 00:00:52,950 started on see the slides here? 10 00:00:55,460 --> 00:00:56,569 Oh, there we go. 11 00:00:56,570 --> 00:00:57,570 Finally. 12 00:01:04,060 --> 00:01:05,018 Great to see you all. 13 00:01:05,019 --> 00:01:07,689 So we are talking about fabrication 14 00:01:07,690 --> 00:01:09,999 of a microchip and integrated circuit, 15 00:01:10,000 --> 00:01:12,069 not exactly cutting edge 16 00:01:12,070 --> 00:01:14,349 technology, but is something 17 00:01:14,350 --> 00:01:17,199 that could build us 80 to 86 18 00:01:17,200 --> 00:01:19,199 now. Thirty four years old. 19 00:01:19,200 --> 00:01:20,289 Thirty five years old. 20 00:01:20,290 --> 00:01:22,119 Eight years older than I am. 21 00:01:22,120 --> 00:01:24,429 But it has something like. 22 00:01:24,430 --> 00:01:25,430 Oh. 23 00:01:26,440 --> 00:01:28,449 It has something like hundred thirty five 24 00:01:28,450 --> 00:01:30,849 thousand transistors, there should be 25 00:01:30,850 --> 00:01:33,039 a Dilshad over here and I 26 00:01:33,040 --> 00:01:35,349 hope the rest of my presentation 27 00:01:35,350 --> 00:01:37,599 is not the same 28 00:01:37,600 --> 00:01:39,069 because on my laptop it worked. 29 00:01:40,070 --> 00:01:41,070 Um, 30 00:01:42,700 --> 00:01:44,979 anyway, um, so 31 00:01:44,980 --> 00:01:48,279 on on this day we have 32 00:01:48,280 --> 00:01:50,799 so there's a die in here. 33 00:01:50,800 --> 00:01:52,359 There are one hundred thirty five 34 00:01:52,360 --> 00:01:54,789 thousand transistors and 35 00:01:54,790 --> 00:01:56,259 the minimum feature size. 36 00:01:56,260 --> 00:01:58,419 So the minimum size 37 00:01:58,420 --> 00:02:00,789 that you have there 38 00:02:00,790 --> 00:02:02,999 in inside there is one, 39 00:02:03,000 --> 00:02:04,149 not five micrometers. 40 00:02:04,150 --> 00:02:06,170 So that's already something. 41 00:02:08,050 --> 00:02:10,629 So just let me quickly 42 00:02:10,630 --> 00:02:12,819 give an overview of our field effect 43 00:02:12,820 --> 00:02:14,319 transistor works. 44 00:02:14,320 --> 00:02:16,899 So down here you have, 45 00:02:16,900 --> 00:02:19,149 uh, the silicon 46 00:02:19,150 --> 00:02:20,150 substrate. 47 00:02:20,930 --> 00:02:23,089 And on itself, silicon is not very 48 00:02:23,090 --> 00:02:25,279 conductive, but 49 00:02:25,280 --> 00:02:27,259 you have two ways to make it conductive, 50 00:02:27,260 --> 00:02:29,779 and one is to implant 51 00:02:29,780 --> 00:02:32,209 charges into the silicon 52 00:02:32,210 --> 00:02:35,639 and this creates positive. 53 00:02:35,640 --> 00:02:37,709 In the case of the substrate or 54 00:02:37,710 --> 00:02:39,779 negative charges, 55 00:02:39,780 --> 00:02:41,999 and they can move around and transport 56 00:02:42,000 --> 00:02:44,069 car in that way, and so you 57 00:02:44,070 --> 00:02:46,529 see here that their source 58 00:02:46,530 --> 00:02:48,639 region and the drain region, and they 59 00:02:48,640 --> 00:02:50,909 are conducted by themselves, but 60 00:02:50,910 --> 00:02:54,299 they are separated from the substrate 61 00:02:54,300 --> 00:02:56,429 p region by the depletion 62 00:02:56,430 --> 00:02:58,829 region where there are no free charges. 63 00:02:58,830 --> 00:03:02,069 And so there is no connection anywhere. 64 00:03:02,070 --> 00:03:04,409 But if you now turn on the gate, 65 00:03:04,410 --> 00:03:05,490 apply a voltage. 66 00:03:07,610 --> 00:03:10,069 Um, this voltage on the gate 67 00:03:10,070 --> 00:03:12,769 here, it attracts charges 68 00:03:12,770 --> 00:03:15,139 down here in the substrate 69 00:03:15,140 --> 00:03:17,329 and these, uh, these 70 00:03:17,330 --> 00:03:19,819 charges are the same or similar 71 00:03:19,820 --> 00:03:22,339 ones then in the, 72 00:03:22,340 --> 00:03:25,189 uh, surrounding soils and drain areas. 73 00:03:25,190 --> 00:03:27,409 And so there is a channel that 74 00:03:27,410 --> 00:03:29,629 can conduct current and you have turned 75 00:03:29,630 --> 00:03:30,630 on the transistor. 76 00:03:32,230 --> 00:03:34,989 And so from this, we see already 77 00:03:34,990 --> 00:03:37,149 the things that we need to put 78 00:03:37,150 --> 00:03:39,699 together to form transistors 79 00:03:39,700 --> 00:03:41,529 and in the end to form a chip. 80 00:03:41,530 --> 00:03:43,809 So we need to have a way to 81 00:03:43,810 --> 00:03:46,209 oxidize silicon. 82 00:03:46,210 --> 00:03:48,549 We need to place polysilicon. 83 00:03:48,550 --> 00:03:50,859 That's the material of which the gate is 84 00:03:50,860 --> 00:03:52,149 made of. 85 00:03:52,150 --> 00:03:54,429 We need to do to implant 86 00:03:54,430 --> 00:03:56,859 other charges into 87 00:03:56,860 --> 00:03:59,649 the, uh, silicon substrate. 88 00:03:59,650 --> 00:04:01,989 We need to put insulators because 89 00:04:01,990 --> 00:04:04,359 later on, we'll, um, place 90 00:04:04,360 --> 00:04:06,579 connections that consist of metal. 91 00:04:06,580 --> 00:04:07,989 And we don't want to short circuit 92 00:04:07,990 --> 00:04:09,429 everything. 93 00:04:09,430 --> 00:04:11,919 And lastly, and that's 94 00:04:11,920 --> 00:04:14,109 maybe half of it, we 95 00:04:14,110 --> 00:04:17,109 need a way to pattern to form 96 00:04:17,110 --> 00:04:19,599 the, um, materials 97 00:04:19,600 --> 00:04:22,509 that we created there because. 98 00:04:22,510 --> 00:04:25,149 Well, the the usual, 99 00:04:25,150 --> 00:04:27,249 um, waste pattern stuff that 100 00:04:27,250 --> 00:04:29,489 we know in the macroscopic world, 101 00:04:29,490 --> 00:04:31,779 things like printing or, 102 00:04:31,780 --> 00:04:33,849 uh, milling or molding, they don't 103 00:04:33,850 --> 00:04:36,189 scale down to these hundred 104 00:04:36,190 --> 00:04:38,289 thirty thousand features and not 105 00:04:38,290 --> 00:04:41,019 down to this one micrometer size. 106 00:04:41,020 --> 00:04:43,339 But we have a way to pattern light. 107 00:04:43,340 --> 00:04:44,979 We know that pretty well. 108 00:04:44,980 --> 00:04:47,409 And we also know materials which undergo 109 00:04:47,410 --> 00:04:49,899 chemical reactions and 110 00:04:49,900 --> 00:04:52,599 we use both of that to 111 00:04:52,600 --> 00:04:54,849 selectively take away 112 00:04:54,850 --> 00:04:56,859 or add um. 113 00:04:58,060 --> 00:04:59,160 Layers of material, 114 00:05:00,400 --> 00:05:02,949 so enough with the theory will 115 00:05:02,950 --> 00:05:05,079 start to go to the cleanroom, 116 00:05:05,080 --> 00:05:07,419 first of all, we'll, uh, 117 00:05:07,420 --> 00:05:10,269 well put on Cleanroom Garment and 118 00:05:10,270 --> 00:05:12,189 clean shoes and also gloves. 119 00:05:12,190 --> 00:05:14,259 And this is, uh, in order not 120 00:05:14,260 --> 00:05:15,609 to contaminate the room. 121 00:05:15,610 --> 00:05:18,459 So we have all these particles 122 00:05:18,460 --> 00:05:20,679 on the clothing that's 123 00:05:20,680 --> 00:05:23,019 dead skin cells, 124 00:05:23,020 --> 00:05:24,459 and we don't want to follow them 125 00:05:24,460 --> 00:05:25,599 everywhere. 126 00:05:25,600 --> 00:05:27,939 We also take the silicon wafer over here 127 00:05:27,940 --> 00:05:29,920 and that's where we build our chip on. 128 00:05:31,340 --> 00:05:33,769 So first of all, for the lithography, 129 00:05:33,770 --> 00:05:35,899 um, that's sort of the answer to 130 00:05:35,900 --> 00:05:38,059 the question, how we can produce 131 00:05:38,060 --> 00:05:40,279 a pattern on this film of silicon, which 132 00:05:40,280 --> 00:05:42,739 is in yellow silicon dioxide. 133 00:05:42,740 --> 00:05:43,879 Sorry. 134 00:05:43,880 --> 00:05:44,880 Um. 135 00:05:45,780 --> 00:05:47,879 And not into 136 00:05:47,880 --> 00:05:50,219 this silicon substrate, 137 00:05:50,220 --> 00:05:52,499 which is green gray here. 138 00:05:52,500 --> 00:05:54,569 So first of all, 139 00:05:54,570 --> 00:05:57,389 you, um, code 140 00:05:57,390 --> 00:05:59,699 the substrate with a layer 141 00:05:59,700 --> 00:06:00,810 of photoresistor. 142 00:06:02,390 --> 00:06:04,459 And then you put on a 143 00:06:04,460 --> 00:06:07,249 mask, this, uh, selectively 144 00:06:07,250 --> 00:06:09,319 blocks light in some areas 145 00:06:09,320 --> 00:06:11,479 and other word households which light 146 00:06:11,480 --> 00:06:13,909 can go through, then you 147 00:06:13,910 --> 00:06:17,179 shine light here through this mask. 148 00:06:17,180 --> 00:06:19,729 And this gives you areas 149 00:06:19,730 --> 00:06:21,949 of photoresistor which 150 00:06:21,950 --> 00:06:24,259 undergo some reaction, which makes 151 00:06:24,260 --> 00:06:26,539 them soluble in 152 00:06:26,540 --> 00:06:28,759 some developed solution, 153 00:06:28,760 --> 00:06:30,109 which is exactly what you do. 154 00:06:30,110 --> 00:06:32,719 You place it in developing a solution 155 00:06:32,720 --> 00:06:35,209 which takes away developed, um, 156 00:06:35,210 --> 00:06:37,879 exposed resist and 157 00:06:37,880 --> 00:06:39,979 leaves the silicon dioxide 158 00:06:39,980 --> 00:06:42,229 in some areas and otherwise 159 00:06:42,230 --> 00:06:43,230 coated. 160 00:06:44,250 --> 00:06:46,589 And then you can at last 161 00:06:46,590 --> 00:06:48,659 make your actual step of 162 00:06:48,660 --> 00:06:51,449 well, taken away the silicon dioxide 163 00:06:51,450 --> 00:06:53,729 and afterwards, um, 164 00:06:53,730 --> 00:06:56,069 dissolve away the resist, 165 00:06:56,070 --> 00:06:59,279 and you have your hole here in your, 166 00:06:59,280 --> 00:07:01,379 uh, in your wafer in the 167 00:07:01,380 --> 00:07:02,459 silicon dioxide layer. 168 00:07:02,460 --> 00:07:03,460 In the wafer. 169 00:07:04,170 --> 00:07:05,370 Great. So. 170 00:07:06,630 --> 00:07:08,699 Uh, let's look at the equipment and 171 00:07:08,700 --> 00:07:11,069 the steps on detail which contain, 172 00:07:11,070 --> 00:07:13,649 uh, which make up for the lithography. 173 00:07:13,650 --> 00:07:15,899 So, um, 174 00:07:15,900 --> 00:07:18,059 first thing is you code your 175 00:07:18,060 --> 00:07:20,219 wafer and that, uh, 176 00:07:20,220 --> 00:07:23,339 works by placing the wafer on, 177 00:07:23,340 --> 00:07:25,499 uh, into the 178 00:07:25,500 --> 00:07:27,269 spin coger. 179 00:07:27,270 --> 00:07:30,179 Um, plays liquid, photoresistor 180 00:07:30,180 --> 00:07:32,429 on top of the, um, wafer 181 00:07:33,630 --> 00:07:35,869 and, uh, spinning 182 00:07:35,870 --> 00:07:36,870 it quickly. 183 00:07:37,780 --> 00:07:39,849 What happens here is that the 184 00:07:39,850 --> 00:07:42,099 photoresistor that sits just the top 185 00:07:42,100 --> 00:07:44,919 of the surface of the way for 186 00:07:44,920 --> 00:07:47,379 this will stick to the 187 00:07:47,380 --> 00:07:49,689 wafer, but the 188 00:07:49,690 --> 00:07:51,789 excess wafer which sits away from 189 00:07:51,790 --> 00:07:54,249 the surface, this is accelerated 190 00:07:54,250 --> 00:07:55,899 and spun off the disk. 191 00:07:55,900 --> 00:07:58,149 You can see here nicely droplets 192 00:07:58,150 --> 00:08:00,399 that's been just off the disk. 193 00:08:00,400 --> 00:08:02,859 And this leaves you with a very even 194 00:08:02,860 --> 00:08:04,779 layer of photoresistor. 195 00:08:07,220 --> 00:08:09,619 Next thing you need is the mosque, and 196 00:08:09,620 --> 00:08:12,049 here you see an example mosque, 197 00:08:12,050 --> 00:08:13,969 which has the patterns that you want to 198 00:08:13,970 --> 00:08:16,579 transfer onto your device. 199 00:08:16,580 --> 00:08:18,679 And this one has five times four 200 00:08:18,680 --> 00:08:19,699 devices on them. 201 00:08:19,700 --> 00:08:21,949 And in each of them, you can sort 202 00:08:21,950 --> 00:08:24,109 of see the geometry 203 00:08:24,110 --> 00:08:26,119 that you want to transfer. 204 00:08:26,120 --> 00:08:28,489 You take both the mosque and 205 00:08:28,490 --> 00:08:31,669 the coated wafer 206 00:08:31,670 --> 00:08:33,709 to this mosque, Alina. 207 00:08:33,710 --> 00:08:35,989 And, um, the important 208 00:08:35,990 --> 00:08:38,579 part is here, the center. 209 00:08:38,580 --> 00:08:40,739 Yeah, so in 210 00:08:40,740 --> 00:08:42,808 this thing here, you put, uh, the 211 00:08:42,809 --> 00:08:45,089 mask and this is the check where you put 212 00:08:45,090 --> 00:08:46,090 the wafer. 213 00:08:48,000 --> 00:08:50,439 And here you see a microscope, 214 00:08:50,440 --> 00:08:52,799 so what you do here is 215 00:08:52,800 --> 00:08:54,899 you, um, move you 216 00:08:54,900 --> 00:08:56,999 can move the chalk, um, 217 00:08:57,000 --> 00:08:58,000 and you move it. 218 00:08:59,550 --> 00:09:01,409 Uh, so that it's bylined. 219 00:09:02,690 --> 00:09:04,760 With the the mosque on top. 220 00:09:06,550 --> 00:09:08,859 Um, especially 221 00:09:08,860 --> 00:09:11,469 if you already have patents on 222 00:09:11,470 --> 00:09:13,839 your wafer, you need to align 223 00:09:13,840 --> 00:09:15,939 them to the, uh, patents on the 224 00:09:15,940 --> 00:09:18,369 mask when you're fine with the alignment 225 00:09:18,370 --> 00:09:20,499 and you have to be really careful there 226 00:09:20,500 --> 00:09:22,719 because, I mean, you have details which 227 00:09:22,720 --> 00:09:24,069 are one micrometer. 228 00:09:24,070 --> 00:09:26,649 So you need to have 229 00:09:26,650 --> 00:09:28,809 that order of accuracy 230 00:09:28,810 --> 00:09:31,299 in placing the mask to the, 231 00:09:31,300 --> 00:09:33,369 um, wafer. 232 00:09:34,510 --> 00:09:36,759 When you're fine with that, you switch 233 00:09:36,760 --> 00:09:38,859 the microscope to your light 234 00:09:38,860 --> 00:09:41,259 source, and that's a light source 235 00:09:41,260 --> 00:09:43,660 that, um, is. 236 00:09:44,710 --> 00:09:46,779 Well, it emits ultraviolet light, it's 237 00:09:46,780 --> 00:09:48,130 contained in this boxier. 238 00:09:51,570 --> 00:09:53,639 Uh, it it shines ultraviolet 239 00:09:53,640 --> 00:09:56,729 light through this mirrors here on 240 00:09:56,730 --> 00:09:58,799 your device, partially 241 00:09:58,800 --> 00:10:01,289 blocked by the mask, and so 242 00:10:01,290 --> 00:10:03,419 you end up with your resist. 243 00:10:04,470 --> 00:10:06,030 Um, Expo's. 244 00:10:08,360 --> 00:10:09,710 To kick it up a notch 245 00:10:11,240 --> 00:10:14,119 nowadays, you don't use these, um, 246 00:10:14,120 --> 00:10:15,979 mask liners anymore, but usually the 247 00:10:15,980 --> 00:10:18,049 STEPA scanners, their 248 00:10:18,050 --> 00:10:19,969 boxes like this and you don't really see 249 00:10:19,970 --> 00:10:21,679 what's going on there. 250 00:10:21,680 --> 00:10:24,019 So what happens 251 00:10:24,020 --> 00:10:26,149 inside is that the mask 252 00:10:26,150 --> 00:10:27,769 is not placed on top of the wafer 253 00:10:27,770 --> 00:10:30,319 anymore, but it is separated 254 00:10:30,320 --> 00:10:31,320 by a lens. 255 00:10:32,460 --> 00:10:34,859 And so there 256 00:10:34,860 --> 00:10:37,189 you go, step by step, 257 00:10:37,190 --> 00:10:39,629 uh, expose all the 258 00:10:39,630 --> 00:10:41,520 chips that you have on the wafer. 259 00:10:44,480 --> 00:10:46,999 And this gives you several 260 00:10:47,000 --> 00:10:49,759 upsides, you don't need 261 00:10:49,760 --> 00:10:52,399 several copies of your 262 00:10:52,400 --> 00:10:53,840 pattern on the mask. 263 00:10:55,150 --> 00:10:57,279 And it also allows you to 264 00:10:57,280 --> 00:10:59,559 scale down the image here, 265 00:10:59,560 --> 00:11:01,419 which sort of, um. 266 00:11:02,800 --> 00:11:05,889 Takes away some of the will 267 00:11:05,890 --> 00:11:07,989 necessary to have this 268 00:11:07,990 --> 00:11:08,990 very small. 269 00:11:11,460 --> 00:11:13,650 Um, structures well-defined. 270 00:11:16,750 --> 00:11:18,879 But on the other side, you 271 00:11:18,880 --> 00:11:21,459 you need to now move 272 00:11:21,460 --> 00:11:23,500 the mosque and. 273 00:11:24,610 --> 00:11:26,619 Also, the way for her, because you have 274 00:11:26,620 --> 00:11:28,929 this, um, this 275 00:11:28,930 --> 00:11:30,009 line of light. 276 00:11:31,200 --> 00:11:32,969 And you need to do this with absolute 277 00:11:32,970 --> 00:11:34,530 accuracy, so. 278 00:11:35,980 --> 00:11:38,109 I mean, nowadays, the features 279 00:11:38,110 --> 00:11:40,809 are just 20 nanometers anymore, and, 280 00:11:40,810 --> 00:11:43,269 uh, so you need to have 281 00:11:43,270 --> 00:11:45,699 the accuracy of these both movements 282 00:11:45,700 --> 00:11:47,320 here. Um. 283 00:11:48,580 --> 00:11:50,709 Well, just has to be perfectly 284 00:11:50,710 --> 00:11:51,710 aligned. 285 00:11:55,500 --> 00:11:57,599 So let's actually 286 00:11:57,600 --> 00:11:59,999 develop to resist, and that's 287 00:12:00,000 --> 00:12:03,029 now easy, you just put it inside. 288 00:12:03,030 --> 00:12:04,030 Uh. 289 00:12:05,050 --> 00:12:06,549 OK, next thing that doesn't work. 290 00:12:06,550 --> 00:12:08,649 Sorry about that, so you just put 291 00:12:08,650 --> 00:12:10,750 it inside a developing solution. 292 00:12:12,290 --> 00:12:14,389 And this will 293 00:12:14,390 --> 00:12:16,669 dissolve away the exposed parts 294 00:12:16,670 --> 00:12:19,159 of the resistance, and it might 295 00:12:19,160 --> 00:12:21,629 leave you with this thing here. 296 00:12:21,630 --> 00:12:23,779 Um, so the 297 00:12:23,780 --> 00:12:25,999 darker areas here, that's 298 00:12:26,000 --> 00:12:28,039 areas where you don't have resist 299 00:12:28,040 --> 00:12:29,389 anymore. 300 00:12:29,390 --> 00:12:31,669 The lighter areas are 301 00:12:31,670 --> 00:12:34,099 silicon covered with resist. 302 00:12:34,100 --> 00:12:36,049 And you also see these crosses here. 303 00:12:36,050 --> 00:12:38,149 They're used for alignment of 304 00:12:38,150 --> 00:12:40,249 especially when you have, uh, 305 00:12:40,250 --> 00:12:42,319 details already on there, you can use 306 00:12:42,320 --> 00:12:44,899 them to align the mask to 307 00:12:44,900 --> 00:12:46,249 your details. 308 00:12:46,250 --> 00:12:48,469 You also see here that there is 309 00:12:48,470 --> 00:12:50,659 quite a bit of, uh, uh, 310 00:12:50,660 --> 00:12:53,209 contamination particle here, something 311 00:12:53,210 --> 00:12:55,279 like five micrometers in size. 312 00:12:55,280 --> 00:12:57,799 And if you imagine this would be here, 313 00:12:57,800 --> 00:13:00,409 uh, abstracting this connection here, 314 00:13:00,410 --> 00:13:02,269 then the device would be unusable. 315 00:13:02,270 --> 00:13:04,489 So that's why we need to have 316 00:13:04,490 --> 00:13:06,949 this level of clean, clean 317 00:13:06,950 --> 00:13:09,129 uncleanliness inside 318 00:13:09,130 --> 00:13:11,479 the, uh, clean room 319 00:13:11,480 --> 00:13:13,639 in order to have working devices. 320 00:13:14,810 --> 00:13:16,939 So all in all, you have the 321 00:13:16,940 --> 00:13:19,009 resources taken away here 322 00:13:19,010 --> 00:13:20,720 by the development step. 323 00:13:22,660 --> 00:13:25,329 So it's gone to the actual, 324 00:13:25,330 --> 00:13:28,749 um, material, 325 00:13:28,750 --> 00:13:29,750 um, 326 00:13:31,120 --> 00:13:33,219 actual, um, take 327 00:13:33,220 --> 00:13:35,649 away of material and 328 00:13:35,650 --> 00:13:37,749 you could use the WCH 329 00:13:37,750 --> 00:13:40,179 etching process that you probably know. 330 00:13:40,180 --> 00:13:42,309 No, just some liquid acid and just 331 00:13:42,310 --> 00:13:44,529 put your, uh, wafer 332 00:13:44,530 --> 00:13:45,789 inside of this. 333 00:13:45,790 --> 00:13:48,159 But rather we use this plus matching 334 00:13:48,160 --> 00:13:50,469 machine. So you place your wafer 335 00:13:50,470 --> 00:13:51,470 inside of this. 336 00:13:52,470 --> 00:13:55,199 And, uh, Flo 337 00:13:55,200 --> 00:13:57,839 flowering inside of the chamber 338 00:13:57,840 --> 00:13:59,159 and then ignite a plasma. 339 00:13:59,160 --> 00:14:01,199 So the situation is now this. 340 00:14:01,200 --> 00:14:03,719 You have Florian Ion's here 341 00:14:03,720 --> 00:14:05,849 and they are accelerated to the wafer 342 00:14:05,850 --> 00:14:06,850 below. 343 00:14:07,840 --> 00:14:10,119 And etch in the exposed 344 00:14:10,120 --> 00:14:12,070 side here, so. 345 00:14:13,400 --> 00:14:15,739 Um, well, the oxide 346 00:14:15,740 --> 00:14:17,389 gets away where it's not protected by 347 00:14:17,390 --> 00:14:20,239 resist if you had used, 348 00:14:20,240 --> 00:14:22,669 um, wet etching, so 349 00:14:22,670 --> 00:14:25,159 just, uh, placed your wafer 350 00:14:25,160 --> 00:14:27,529 inside liquid, um, 351 00:14:27,530 --> 00:14:29,869 then you'd have the problem that, 352 00:14:29,870 --> 00:14:32,119 uh, liquid etchings isotropic and 353 00:14:32,120 --> 00:14:34,489 it would further go 354 00:14:34,490 --> 00:14:36,559 and EDG, the oxide below 355 00:14:36,560 --> 00:14:37,679 the resist. 356 00:14:37,680 --> 00:14:39,679 And with this, uh, plasma etching, you 357 00:14:39,680 --> 00:14:42,139 have the ions that just 358 00:14:42,140 --> 00:14:44,359 go down here and have this really 359 00:14:44,360 --> 00:14:47,119 fine motion and they are the resistors 360 00:14:47,120 --> 00:14:49,279 capable of really, 361 00:14:49,280 --> 00:14:51,559 well, covering and protecting the oxide. 362 00:14:51,560 --> 00:14:53,839 So this gives 363 00:14:53,840 --> 00:14:56,509 you, in fact, uh, better resolution 364 00:14:56,510 --> 00:14:57,950 and that's why you're using it. 365 00:14:59,770 --> 00:15:01,870 So need to, um. 366 00:15:03,350 --> 00:15:05,779 Dope or silicone, now 367 00:15:05,780 --> 00:15:07,969 this happens inside the 368 00:15:07,970 --> 00:15:10,129 planter. And again, you can't really see 369 00:15:10,130 --> 00:15:13,009 anything. So we are trying to schematic. 370 00:15:13,010 --> 00:15:15,349 So you start with this iron source 371 00:15:15,350 --> 00:15:16,279 here. 372 00:15:16,280 --> 00:15:18,559 Um, this emits some ions which are 373 00:15:18,560 --> 00:15:20,779 accelerated, fly 374 00:15:20,780 --> 00:15:23,089 away into this magnetic field 375 00:15:23,090 --> 00:15:25,429 over here. And this magnetic 376 00:15:25,430 --> 00:15:28,759 field sends them on a curved path 377 00:15:28,760 --> 00:15:31,009 and different types of ions. 378 00:15:31,010 --> 00:15:33,349 They follow differently 379 00:15:33,350 --> 00:15:34,909 curved paths. 380 00:15:34,910 --> 00:15:37,099 So you have this sluttier, 381 00:15:37,100 --> 00:15:38,959 which selects only the ions that you 382 00:15:38,960 --> 00:15:41,629 want. They fly onto your substrate 383 00:15:41,630 --> 00:15:42,799 and create. 384 00:15:44,330 --> 00:15:46,399 Uh, and 385 00:15:46,400 --> 00:15:48,979 in this case, they create, uh, negatively 386 00:15:48,980 --> 00:15:51,289 charged wall inside your positively 387 00:15:51,290 --> 00:15:53,840 charged, um, substrate. 388 00:15:57,130 --> 00:15:59,409 You now continue to 389 00:15:59,410 --> 00:16:02,199 grow a thin oxide 390 00:16:02,200 --> 00:16:04,719 that separates your gates from 391 00:16:04,720 --> 00:16:06,789 the substrate, and that's quite 392 00:16:06,790 --> 00:16:08,529 easy, you just placed them into a 393 00:16:08,530 --> 00:16:10,689 furnace, um, just 394 00:16:10,690 --> 00:16:12,939 the wafers flow oxygen 395 00:16:12,940 --> 00:16:14,289 to it through it. 396 00:16:16,450 --> 00:16:18,759 And then, uh, the silicon 397 00:16:18,760 --> 00:16:21,429 oxidizers and Sigatoka site, 398 00:16:21,430 --> 00:16:24,009 so you can now vary in the thickness 399 00:16:24,010 --> 00:16:26,619 by varying the time and, 400 00:16:26,620 --> 00:16:28,809 uh, oxygen pressure and 401 00:16:28,810 --> 00:16:29,810 also the temperature. 402 00:16:32,270 --> 00:16:34,699 So next part of your gate, the actual 403 00:16:34,700 --> 00:16:37,279 gate material, um, 404 00:16:37,280 --> 00:16:38,899 this is deposited in one of these 405 00:16:38,900 --> 00:16:40,879 chemical vapor deposition machines. 406 00:16:41,930 --> 00:16:43,820 Um, so. 407 00:16:45,390 --> 00:16:47,609 You see here that it's another 408 00:16:47,610 --> 00:16:50,069 chamber, um, you follow 409 00:16:50,070 --> 00:16:53,149 silicon containing gas through it. 410 00:16:53,150 --> 00:16:55,279 Um, you have the 411 00:16:55,280 --> 00:16:57,469 ability to heat the chamber with 412 00:16:57,470 --> 00:17:00,049 these, uh, all these 413 00:17:00,050 --> 00:17:01,050 heating elements. 414 00:17:02,460 --> 00:17:05,229 And then whenever. 415 00:17:05,230 --> 00:17:07,719 Uh, molecule of this gas hits 416 00:17:07,720 --> 00:17:09,909 some surface, it gets converted into 417 00:17:09,910 --> 00:17:10,959 atomic silicon. 418 00:17:12,490 --> 00:17:14,559 And in case of the wait for this 419 00:17:14,560 --> 00:17:16,779 now cover, you have now a covering 420 00:17:16,780 --> 00:17:18,969 of silicon of everywhere 421 00:17:18,970 --> 00:17:19,970 on your wafer. 422 00:17:23,300 --> 00:17:25,578 So you need now to, 423 00:17:25,579 --> 00:17:27,709 um, form the gates, just 424 00:17:27,710 --> 00:17:29,869 etch everything away, you already know 425 00:17:29,870 --> 00:17:32,089 this, uh, plasma etching 426 00:17:32,090 --> 00:17:33,679 device. 427 00:17:33,680 --> 00:17:36,519 So you take away the, uh, 428 00:17:36,520 --> 00:17:38,989 the gate, uh, oxide and 429 00:17:38,990 --> 00:17:41,839 the, um, 430 00:17:41,840 --> 00:17:43,909 gate polysilicon everywhere that 431 00:17:43,910 --> 00:17:45,229 you don't want it. 432 00:17:45,230 --> 00:17:47,899 And this for this, you also 433 00:17:47,900 --> 00:17:50,479 have for lithography step, 434 00:17:50,480 --> 00:17:52,099 which I'm not going to show here. 435 00:17:53,880 --> 00:17:55,949 Next thing is, you, um, 436 00:17:57,210 --> 00:17:59,519 you construct the source and the drain 437 00:17:59,520 --> 00:18:01,609 regions, and for this, you 438 00:18:01,610 --> 00:18:03,839 already seen the ion implantation 439 00:18:03,840 --> 00:18:04,840 machine. 440 00:18:06,430 --> 00:18:09,129 And reform here positively 441 00:18:09,130 --> 00:18:11,529 charges inside the 442 00:18:11,530 --> 00:18:14,109 substract, you might wonder 443 00:18:14,110 --> 00:18:16,239 why you need, 444 00:18:16,240 --> 00:18:17,240 um. 445 00:18:18,360 --> 00:18:20,729 That's why we have deposited 446 00:18:20,730 --> 00:18:22,979 this, uh, well, of negative 447 00:18:22,980 --> 00:18:25,049 charges first and then 448 00:18:25,050 --> 00:18:27,209 continue to, uh, 449 00:18:27,210 --> 00:18:29,279 positive charges, and that's 450 00:18:29,280 --> 00:18:31,259 because actually there are two types of 451 00:18:31,260 --> 00:18:33,509 transistors in, uh, 452 00:18:33,510 --> 00:18:35,489 inside the chips. 453 00:18:39,530 --> 00:18:41,689 These are complementary, so they 454 00:18:41,690 --> 00:18:43,759 use either 455 00:18:43,760 --> 00:18:46,309 negative charges or positive charges 456 00:18:46,310 --> 00:18:48,470 to, um, conduct current. 457 00:18:49,480 --> 00:18:52,119 And that actually gives you the, 458 00:18:52,120 --> 00:18:54,339 um, the possibility 459 00:18:54,340 --> 00:18:56,649 to actually switch things, otherwise 460 00:18:56,650 --> 00:18:58,839 you'd have just turned on and you 461 00:18:58,840 --> 00:19:01,029 need something to turn it off. 462 00:19:01,030 --> 00:19:03,400 And that's the complimentary transistor. 463 00:19:05,820 --> 00:19:08,339 You also have here, 464 00:19:08,340 --> 00:19:10,859 um, the situation that you don't 465 00:19:10,860 --> 00:19:12,959 need to talk graphy 466 00:19:12,960 --> 00:19:15,269 to, uh, define the, 467 00:19:15,270 --> 00:19:18,089 um, the regions which are 468 00:19:18,090 --> 00:19:20,729 doped, but you just use the, 469 00:19:20,730 --> 00:19:22,859 um, oxygen which is already there 470 00:19:22,860 --> 00:19:24,179 and also the gates. 471 00:19:24,180 --> 00:19:26,459 And so everywhere where you have, 472 00:19:26,460 --> 00:19:28,529 uh, naked substrate showing, 473 00:19:28,530 --> 00:19:30,989 you change the, um, 474 00:19:32,010 --> 00:19:34,409 the doping by implanting 475 00:19:34,410 --> 00:19:35,679 ions. 476 00:19:35,680 --> 00:19:37,949 That's called a cell line process. 477 00:19:37,950 --> 00:19:40,469 So you don't need to align 478 00:19:40,470 --> 00:19:42,659 any, um, photo 479 00:19:42,660 --> 00:19:44,879 lithography step to your existing 480 00:19:44,880 --> 00:19:46,769 pattern. So this. 481 00:19:47,950 --> 00:19:50,349 Um, this gives you much 482 00:19:50,350 --> 00:19:52,419 higher accuracy because you don't have 483 00:19:52,420 --> 00:19:54,669 any alignment errors 484 00:19:54,670 --> 00:19:56,829 and, well, this 485 00:19:56,830 --> 00:19:58,959 basically allows you to 486 00:19:58,960 --> 00:20:01,479 scale down your devices, 487 00:20:01,480 --> 00:20:03,189 which is obviously what you're going to 488 00:20:03,190 --> 00:20:04,190 do. 489 00:20:06,330 --> 00:20:08,129 Now we are nearing the end of the 490 00:20:08,130 --> 00:20:10,740 fabrication, um, 491 00:20:11,850 --> 00:20:13,319 we. 492 00:20:13,320 --> 00:20:15,449 Have up until now, we 493 00:20:15,450 --> 00:20:17,429 have working transistors, but they are 494 00:20:17,430 --> 00:20:19,649 not connected really anywhere, 495 00:20:19,650 --> 00:20:22,029 and while they are not very useful, 496 00:20:22,030 --> 00:20:22,949 they are not connected. 497 00:20:22,950 --> 00:20:25,829 So we want to place metal. 498 00:20:25,830 --> 00:20:28,409 But until now, the, 499 00:20:28,410 --> 00:20:29,410 um. 500 00:20:30,320 --> 00:20:32,179 They are conductive part exposed 501 00:20:32,180 --> 00:20:34,369 everywhere. So first of 502 00:20:34,370 --> 00:20:36,439 all, we are depositing some 503 00:20:36,440 --> 00:20:37,440 insulator. 504 00:20:38,230 --> 00:20:40,269 That's again, done in this chemical 505 00:20:40,270 --> 00:20:42,909 weapons chemical vapor deposition 506 00:20:42,910 --> 00:20:45,309 chamber, but now not with silicone, 507 00:20:45,310 --> 00:20:47,589 but with silicon dioxide forming gas. 508 00:20:49,770 --> 00:20:52,019 And you see what happens here 509 00:20:52,020 --> 00:20:54,289 from, uh. 510 00:20:54,290 --> 00:20:56,869 Form a layer of insulator 511 00:20:56,870 --> 00:20:58,279 and then afterwards. 512 00:20:58,280 --> 00:21:00,829 Now you use a step of lithography 513 00:21:00,830 --> 00:21:03,109 to edge, 514 00:21:03,110 --> 00:21:05,409 um, holes inside 515 00:21:05,410 --> 00:21:08,059 the insulator to actually, 516 00:21:08,060 --> 00:21:10,469 uh, connect your gait 517 00:21:10,470 --> 00:21:11,599 sources and drains. 518 00:21:15,240 --> 00:21:16,259 Almost ready. 519 00:21:16,260 --> 00:21:17,519 You know, um, 520 00:21:18,570 --> 00:21:21,629 deposited the actual, 521 00:21:21,630 --> 00:21:23,969 uh, connection there, and 522 00:21:23,970 --> 00:21:25,739 this is the last piece of equipment that 523 00:21:25,740 --> 00:21:26,999 I'm going to show. 524 00:21:27,000 --> 00:21:29,069 So this is now, 525 00:21:29,070 --> 00:21:31,049 again, vapor deposition device. 526 00:21:31,050 --> 00:21:32,339 But this time it's the physical 527 00:21:32,340 --> 00:21:35,789 deposition, physical vapor deposition. 528 00:21:35,790 --> 00:21:38,249 Um, you load 529 00:21:38,250 --> 00:21:40,169 this load log here, your wafer 530 00:21:40,170 --> 00:21:42,209 transferred into the process chamber 531 00:21:42,210 --> 00:21:43,210 here. 532 00:21:43,970 --> 00:21:46,189 And turn on the deposition. 533 00:21:47,580 --> 00:21:49,439 It's a sputtering machine, which means 534 00:21:49,440 --> 00:21:51,719 that you have argan 535 00:21:51,720 --> 00:21:54,119 ions which bombard some 536 00:21:54,120 --> 00:21:55,410 material source. 537 00:21:56,600 --> 00:21:58,879 And you can see here 538 00:21:58,880 --> 00:22:00,819 that, uh. 539 00:22:00,820 --> 00:22:02,589 Due to this iron bombardment, there is 540 00:22:02,590 --> 00:22:04,959 the plasma blowing up and 541 00:22:04,960 --> 00:22:07,029 this ions, they sputter 542 00:22:07,030 --> 00:22:09,159 out atoms in all 543 00:22:09,160 --> 00:22:11,229 directions and this 544 00:22:11,230 --> 00:22:13,809 creates a covering of, uh, metal 545 00:22:13,810 --> 00:22:14,810 on our wafer. 546 00:22:19,010 --> 00:22:21,829 We, uh, so 547 00:22:21,830 --> 00:22:24,379 we almost have a finished ship, 548 00:22:24,380 --> 00:22:25,380 um. 549 00:22:26,550 --> 00:22:29,249 We have basically, 550 00:22:29,250 --> 00:22:31,739 hopefully a working chip, but 551 00:22:31,740 --> 00:22:34,269 it's still we have several copies, 552 00:22:34,270 --> 00:22:37,199 um, sitting on a wafer, 553 00:22:37,200 --> 00:22:40,229 so we need to dice the wafer. 554 00:22:40,230 --> 00:22:42,749 That means cutting it into, uh, 555 00:22:42,750 --> 00:22:44,190 the single chips that we have. 556 00:22:45,450 --> 00:22:48,029 And then we, 557 00:22:48,030 --> 00:22:50,219 uh, put it into some package, 558 00:22:50,220 --> 00:22:51,220 um. 559 00:22:52,770 --> 00:22:54,899 Connected to the outside world, that's 560 00:22:54,900 --> 00:22:57,119 by placing these wires 561 00:22:57,120 --> 00:22:58,829 that you hardly see here 562 00:22:59,960 --> 00:23:02,279 and then, 563 00:23:02,280 --> 00:23:04,769 um, close the package and 564 00:23:04,770 --> 00:23:07,169 well, then you have your chip 565 00:23:07,170 --> 00:23:09,339 just sort of have to test 566 00:23:09,340 --> 00:23:10,859 it, whether it actually works. 567 00:23:10,860 --> 00:23:12,640 But, hey, you're finished. 568 00:23:13,890 --> 00:23:16,199 So, um, what 569 00:23:16,200 --> 00:23:19,409 would be different in 570 00:23:19,410 --> 00:23:21,509 actual fabs? 571 00:23:21,510 --> 00:23:24,839 So I mostly 572 00:23:24,840 --> 00:23:27,269 showed your research equipment, which 573 00:23:27,270 --> 00:23:29,489 basically works the same way. 574 00:23:29,490 --> 00:23:31,739 But in industrial 575 00:23:31,740 --> 00:23:33,899 fabs, you, uh, 576 00:23:33,900 --> 00:23:36,059 basically have robots doing all 577 00:23:36,060 --> 00:23:38,249 the work because, I mean, it's repetitive 578 00:23:38,250 --> 00:23:40,379 work. And humans in 579 00:23:40,380 --> 00:23:42,479 this process are well, 580 00:23:42,480 --> 00:23:45,059 they they contaminate the clean room. 581 00:23:45,060 --> 00:23:47,469 So you take the, 582 00:23:47,470 --> 00:23:49,679 uh, the humans out of the clean room and 583 00:23:49,680 --> 00:23:51,449 have robots are doing all the work and 584 00:23:51,450 --> 00:23:53,219 then you're better off. 585 00:23:53,220 --> 00:23:55,949 So, uh, this 586 00:23:55,950 --> 00:23:57,749 sort of leads to that. 587 00:23:57,750 --> 00:23:59,339 You don't see everything. 588 00:23:59,340 --> 00:24:01,649 You just see some robot 589 00:24:01,650 --> 00:24:04,079 carrying your wafer into the next, 590 00:24:04,080 --> 00:24:05,660 uh, processed chamber. 591 00:24:07,110 --> 00:24:07,709 All right. 592 00:24:07,710 --> 00:24:10,049 And, um, 593 00:24:10,050 --> 00:24:11,669 I'm now already finished. 594 00:24:11,670 --> 00:24:14,399 Somehow this was much faster 595 00:24:14,400 --> 00:24:16,739 than, uh, 596 00:24:16,740 --> 00:24:18,479 in my testing. 597 00:24:18,480 --> 00:24:20,639 So we can already have 598 00:24:20,640 --> 00:24:21,640 Kunhardt. 599 00:24:30,890 --> 00:24:32,959 OK, you all know the Q&A 600 00:24:32,960 --> 00:24:35,239 game, there's four microphones, 601 00:24:35,240 --> 00:24:38,029 please cue up behind them, 602 00:24:38,030 --> 00:24:40,039 and if you're on the stream, our Signal 603 00:24:40,040 --> 00:24:42,499 Angel will relay your questions 604 00:24:42,500 --> 00:24:45,109 as our human interface device. 605 00:24:45,110 --> 00:24:48,139 Um, so let's start with Mike to please. 606 00:24:48,140 --> 00:24:49,819 Hi. Thanks for the great talk. 607 00:24:49,820 --> 00:24:52,069 Um, my question is, can 608 00:24:52,070 --> 00:24:53,479 you do multiple layers? 609 00:24:53,480 --> 00:24:55,909 I thought there are chips with multiple 610 00:24:55,910 --> 00:24:58,009 layers and I wonder how this 611 00:24:58,010 --> 00:24:59,010 is done. 612 00:24:59,570 --> 00:25:01,699 Yeah. So there are sort 613 00:25:01,700 --> 00:25:03,439 of two answers to that. 614 00:25:03,440 --> 00:25:05,509 So in new chips then 615 00:25:05,510 --> 00:25:07,759 this one, you have multiple layers. 616 00:25:07,760 --> 00:25:10,429 And this is done by basically, 617 00:25:10,430 --> 00:25:12,529 uh, repeating these 618 00:25:12,530 --> 00:25:14,539 two steps over and over again. 619 00:25:14,540 --> 00:25:16,789 So the you 620 00:25:16,790 --> 00:25:18,919 build an insulator and 621 00:25:18,920 --> 00:25:22,099 then on top of that place, the, 622 00:25:22,100 --> 00:25:23,749 uh, well. 623 00:25:23,750 --> 00:25:26,000 Edge contacts and place the connections. 624 00:25:27,160 --> 00:25:29,409 And nowadays you have on 625 00:25:29,410 --> 00:25:31,659 the order of 10 Conexion 626 00:25:31,660 --> 00:25:32,660 layers. 627 00:25:33,540 --> 00:25:36,419 And I don't know if your question 628 00:25:36,420 --> 00:25:39,869 goes on to three dimensional 629 00:25:39,870 --> 00:25:43,109 flash ships or things like that, 630 00:25:43,110 --> 00:25:45,419 but it's only the connections that 631 00:25:45,420 --> 00:25:46,499 it's multiple layers. 632 00:25:46,500 --> 00:25:48,689 It's not only the connections in CPUs 633 00:25:48,690 --> 00:25:50,489 and stuff, it's only the connections. 634 00:25:50,490 --> 00:25:51,490 Oh, OK. You. 635 00:25:52,950 --> 00:25:54,689 Mike, No one, please. 636 00:25:54,690 --> 00:25:55,889 Yes, good evening. 637 00:25:55,890 --> 00:25:58,169 What kind of techniques could 638 00:25:58,170 --> 00:26:00,809 you list that would enable 639 00:26:00,810 --> 00:26:02,879 a user to check the 640 00:26:02,880 --> 00:26:05,339 conformity of what was 641 00:26:05,340 --> 00:26:07,619 printed on 642 00:26:07,620 --> 00:26:10,079 a dice with the original 643 00:26:11,250 --> 00:26:12,959 X-ray microscope? 644 00:26:12,960 --> 00:26:16,139 Uh, electronic microscope. 645 00:26:16,140 --> 00:26:18,029 Yeah. So what kind of technology do you 646 00:26:18,030 --> 00:26:19,799 know? And we did. 647 00:26:19,800 --> 00:26:21,899 And my second question, would it be 648 00:26:21,900 --> 00:26:24,359 possible, according to you, to invent 649 00:26:24,360 --> 00:26:26,789 or develop a very Low-Cost 650 00:26:26,790 --> 00:26:28,889 technology that would allow all 651 00:26:28,890 --> 00:26:31,019 of us to check the integrity 652 00:26:31,020 --> 00:26:33,269 of the integrated circuits we would have 653 00:26:33,270 --> 00:26:35,579 in our devices like 654 00:26:35,580 --> 00:26:37,949 we do Ashar 655 00:26:37,950 --> 00:26:40,049 256 on a 656 00:26:40,050 --> 00:26:41,050 piece of cotton? 657 00:26:43,940 --> 00:26:46,219 All right, so first part of the question 658 00:26:46,220 --> 00:26:48,289 is how can we 659 00:26:48,290 --> 00:26:49,789 analyze the functioning? 660 00:26:49,790 --> 00:26:52,339 And this is really a really hard 661 00:26:52,340 --> 00:26:54,559 task to do so 662 00:26:54,560 --> 00:26:57,019 nowadays you're talking about billions 663 00:26:57,020 --> 00:26:59,119 of transistors and they don't 664 00:26:59,120 --> 00:27:00,259 exist on their own. 665 00:27:00,260 --> 00:27:02,839 They live. So they they only function 666 00:27:02,840 --> 00:27:05,569 through the connections in between. 667 00:27:05,570 --> 00:27:07,310 And so you. 668 00:27:08,320 --> 00:27:10,389 Can check whether the 669 00:27:10,390 --> 00:27:12,759 fabrication produces 670 00:27:12,760 --> 00:27:15,039 the transistors that you want and 671 00:27:15,040 --> 00:27:17,649 how this works is you 672 00:27:17,650 --> 00:27:19,719 while you first dissolve 673 00:27:19,720 --> 00:27:22,359 the package and then 674 00:27:22,360 --> 00:27:23,749 take. 675 00:27:23,750 --> 00:27:25,849 Um, successive, 676 00:27:25,850 --> 00:27:28,069 uh, micro photographs, 677 00:27:28,070 --> 00:27:30,379 um, microphone, uh, not 678 00:27:30,380 --> 00:27:33,139 microscopic images of 679 00:27:33,140 --> 00:27:35,929 the layers, so you start with the topmost 680 00:27:35,930 --> 00:27:38,389 metal layer and then selectively 681 00:27:38,390 --> 00:27:40,289 etch away every layer. 682 00:27:40,290 --> 00:27:42,679 And, uh, well, you basically 683 00:27:42,680 --> 00:27:44,809 can reconstruct the layers that your 684 00:27:44,810 --> 00:27:46,469 device is made of even. 685 00:27:46,470 --> 00:27:48,649 Is it possible to ensure 686 00:27:48,650 --> 00:27:51,049 that some parts were correctly 687 00:27:51,050 --> 00:27:52,050 dubbed? 688 00:27:52,910 --> 00:27:55,279 I'm not sure 689 00:27:55,280 --> 00:27:56,809 that's possible. 690 00:27:56,810 --> 00:27:58,999 I mean, the doping level, you 691 00:27:59,000 --> 00:28:01,069 can actually see that the, 692 00:28:01,070 --> 00:28:03,289 um, the semiconductor just 693 00:28:03,290 --> 00:28:05,569 looks differently, OK, but 694 00:28:05,570 --> 00:28:08,719 I mean, it's still 695 00:28:08,720 --> 00:28:11,269 possible to, uh, 696 00:28:11,270 --> 00:28:13,549 to just change the functioning 697 00:28:13,550 --> 00:28:16,219 a bit and have completely 698 00:28:16,220 --> 00:28:18,319 different, uh, functioning 699 00:28:18,320 --> 00:28:19,579 of your device. 700 00:28:19,580 --> 00:28:21,409 This is something that other people 701 00:28:21,410 --> 00:28:23,479 actually have demonstrated, that you 702 00:28:23,480 --> 00:28:25,789 can sort of change a couple 703 00:28:25,790 --> 00:28:27,889 of transistors and 704 00:28:27,890 --> 00:28:30,230 that creates a backdoor in your CPU 705 00:28:31,400 --> 00:28:32,579 and attack. 706 00:28:32,580 --> 00:28:34,159 Exactly. 707 00:28:34,160 --> 00:28:36,109 And my second question. 708 00:28:36,110 --> 00:28:37,159 Yes, ma'am. 709 00:28:37,160 --> 00:28:38,299 Do you think it's a question? 710 00:28:38,300 --> 00:28:40,399 Yes. Do you think it would be possible to 711 00:28:40,400 --> 00:28:42,469 invent a Low-Cost technology to check? 712 00:28:42,470 --> 00:28:44,059 Uh, no, I don't think so. 713 00:28:44,060 --> 00:28:45,779 I think that's out of reach. 714 00:28:45,780 --> 00:28:47,749 OK, thank you, Mike. 715 00:28:47,750 --> 00:28:49,939 Number four, please ask one question, 716 00:28:49,940 --> 00:28:52,249 please. Thanks for the task. 717 00:28:52,250 --> 00:28:54,589 I'm just wondering myself how the photo 718 00:28:54,590 --> 00:28:57,319 mask is going to be printed, 719 00:28:57,320 --> 00:28:58,429 how this is working. 720 00:28:59,870 --> 00:29:02,029 So you also, uh, 721 00:29:02,030 --> 00:29:04,189 use some lithography, but this 722 00:29:04,190 --> 00:29:06,469 time you don't use, uh, rays 723 00:29:06,470 --> 00:29:08,659 of light, rays of electrons 724 00:29:08,660 --> 00:29:10,099 and electrons. 725 00:29:10,100 --> 00:29:12,769 You can very accurately, 726 00:29:12,770 --> 00:29:15,109 um, shine somewhere. 727 00:29:15,110 --> 00:29:17,629 You can make an array 728 00:29:17,630 --> 00:29:19,819 of electrons go into 729 00:29:19,820 --> 00:29:21,049 some exact location. 730 00:29:21,050 --> 00:29:23,269 And that actually works with much 731 00:29:23,270 --> 00:29:26,389 more precision than optical lithography. 732 00:29:26,390 --> 00:29:28,669 You could somehow use this 733 00:29:28,670 --> 00:29:30,889 to, um, create the 734 00:29:30,890 --> 00:29:32,509 CPU themselves. 735 00:29:32,510 --> 00:29:34,939 But this doesn't really scale because 736 00:29:34,940 --> 00:29:37,009 you have just well, 737 00:29:37,010 --> 00:29:39,199 it basically is very similar 738 00:29:39,200 --> 00:29:42,649 to this old cathode ray tubes, 739 00:29:42,650 --> 00:29:44,839 the big monitors that we have before, 740 00:29:44,840 --> 00:29:47,479 the array of electrons going someplace. 741 00:29:47,480 --> 00:29:49,579 But it's just one very tiny dot 742 00:29:49,580 --> 00:29:51,349 of maybe 10 nanometers. 743 00:29:51,350 --> 00:29:53,839 And you need to scan all 744 00:29:53,840 --> 00:29:56,149 of your surface with this electron 745 00:29:56,150 --> 00:29:58,219 beam and then you have 746 00:29:58,220 --> 00:30:00,259 resist, which are sensitive to this 747 00:30:00,260 --> 00:30:02,419 electron beam, change 748 00:30:02,420 --> 00:30:04,699 their chemistry there and you can 749 00:30:04,700 --> 00:30:07,309 dissolve the partially away. 750 00:30:07,310 --> 00:30:08,749 Thank you very much. 751 00:30:08,750 --> 00:30:11,159 Thank you. Number three, please. 752 00:30:11,160 --> 00:30:12,199 Uh, hello. 753 00:30:12,200 --> 00:30:14,549 Uh, why wait for 754 00:30:14,550 --> 00:30:16,669 around one dies 755 00:30:16,670 --> 00:30:17,670 square. 756 00:30:18,500 --> 00:30:20,299 Um, that's. 757 00:30:21,550 --> 00:30:23,599 Because, well, the, um, 758 00:30:25,000 --> 00:30:27,609 the the wafers, they are cut 759 00:30:27,610 --> 00:30:30,249 from the cylinder of, 760 00:30:30,250 --> 00:30:33,159 uh, silicon, and that's one crystal, 761 00:30:33,160 --> 00:30:35,259 um, and it's just the 762 00:30:35,260 --> 00:30:37,839 most simple way of producing 763 00:30:37,840 --> 00:30:40,149 single crystals in this big 764 00:30:40,150 --> 00:30:41,589 size. 765 00:30:41,590 --> 00:30:44,619 Um, I hope I can 766 00:30:44,620 --> 00:30:46,699 open some this 767 00:30:46,700 --> 00:30:48,490 square crystal complicated. 768 00:30:49,630 --> 00:30:52,179 Um, I 769 00:30:52,180 --> 00:30:54,639 have not seen growing of a square 770 00:30:54,640 --> 00:30:56,799 crystal actually. 771 00:30:56,800 --> 00:30:58,089 I don't know. 772 00:30:58,090 --> 00:31:00,219 But I mean, 773 00:31:00,220 --> 00:31:02,349 it's for it's 774 00:31:02,350 --> 00:31:04,179 definitely more simple to have it in the 775 00:31:04,180 --> 00:31:06,099 surround, uh, size. 776 00:31:08,690 --> 00:31:10,189 Mike, number two, please. 777 00:31:10,190 --> 00:31:13,429 Sorry, so you said the 778 00:31:13,430 --> 00:31:15,619 lifeforce used for look and talk 779 00:31:15,620 --> 00:31:17,989 graffitists UV, but 780 00:31:17,990 --> 00:31:20,269 in the modern spoof, the 781 00:31:20,270 --> 00:31:22,729 size of the dye is so small, 782 00:31:22,730 --> 00:31:24,829 what lights off to what was used 783 00:31:24,830 --> 00:31:25,909 there? 784 00:31:25,910 --> 00:31:29,329 It's actually still ultraviolet 785 00:31:29,330 --> 00:31:32,149 light. And that's really a wonder 786 00:31:32,150 --> 00:31:34,579 how you can make patterns 787 00:31:34,580 --> 00:31:37,009 so small. So they use 788 00:31:37,010 --> 00:31:40,129 one hundred ninety three nanometers light 789 00:31:40,130 --> 00:31:42,529 to produce features 790 00:31:42,530 --> 00:31:44,929 the size of 20 nanometers. 791 00:31:44,930 --> 00:31:47,419 And usually there's a 792 00:31:47,420 --> 00:31:49,459 sort of rule of thumb 793 00:31:50,510 --> 00:31:52,819 that says that you can only have 794 00:31:52,820 --> 00:31:55,499 features half of the wavelength, so 795 00:31:55,500 --> 00:31:58,129 a little bit over 90 nanometers. 796 00:31:58,130 --> 00:32:00,380 And you use some techniques which 797 00:32:01,460 --> 00:32:03,769 basically you 798 00:32:03,770 --> 00:32:06,229 can overlap two 799 00:32:06,230 --> 00:32:08,389 beams sequentially, the image 800 00:32:08,390 --> 00:32:11,449 of two beams and uh, then, 801 00:32:11,450 --> 00:32:13,819 well, multiple times expose 802 00:32:13,820 --> 00:32:14,959 your resist. 803 00:32:14,960 --> 00:32:17,539 And this, if done correctly, 804 00:32:17,540 --> 00:32:20,239 can create smaller features. 805 00:32:20,240 --> 00:32:21,240 Thank you. 806 00:32:21,800 --> 00:32:23,170 Mike, number one, please. 807 00:32:25,330 --> 00:32:27,429 So in the plasma etching, how do 808 00:32:27,430 --> 00:32:29,049 they modulate how much material is 809 00:32:29,050 --> 00:32:30,999 actually removed and is the photoresistor 810 00:32:31,000 --> 00:32:32,640 destroyed at the same in the same step? 811 00:32:34,330 --> 00:32:36,829 So first part, 812 00:32:36,830 --> 00:32:40,059 um, well, you can control 813 00:32:40,060 --> 00:32:42,519 and measure the currents 814 00:32:42,520 --> 00:32:44,049 of plasma. 815 00:32:44,050 --> 00:32:46,569 Let me check where the slide is. 816 00:32:46,570 --> 00:32:47,570 This one here. 817 00:32:48,970 --> 00:32:50,740 So, uh. 818 00:32:55,580 --> 00:32:57,829 So basically, you can I mean, 819 00:32:57,830 --> 00:32:59,899 these are also charges, 820 00:32:59,900 --> 00:33:02,149 right, and you have 821 00:33:02,150 --> 00:33:04,349 plates here which are which 822 00:33:04,350 --> 00:33:05,959 have a potential difference and you can 823 00:33:05,960 --> 00:33:07,189 measure the current through this. 824 00:33:07,190 --> 00:33:09,859 And this will be proportional 825 00:33:09,860 --> 00:33:12,289 to the, um, amount 826 00:33:12,290 --> 00:33:14,359 of ions that, um. 827 00:33:15,760 --> 00:33:17,109 That act on your device. 828 00:33:18,590 --> 00:33:20,239 Uh, what was the second part, 829 00:33:21,410 --> 00:33:22,940 is this destroyed in the same step? 830 00:33:24,260 --> 00:33:26,539 No, it's not 831 00:33:26,540 --> 00:33:28,549 necessarily, um, destroyed. 832 00:33:28,550 --> 00:33:30,529 So you have some attack on the 833 00:33:30,530 --> 00:33:32,959 photoresistor, but 834 00:33:32,960 --> 00:33:35,329 if you have, uh, the photo is taken 835 00:33:35,330 --> 00:33:36,589 off, you don't care. 836 00:33:36,590 --> 00:33:39,679 I mean, uh, this is, uh, 837 00:33:39,680 --> 00:33:41,209 sacrificial layer. 838 00:33:41,210 --> 00:33:43,519 It gets to solve the way afterwards 839 00:33:43,520 --> 00:33:44,520 anyway. 840 00:33:45,810 --> 00:33:46,810 OK, thank you. 841 00:33:48,190 --> 00:33:50,049 Mike, number three, please. 842 00:33:50,050 --> 00:33:52,239 Thank you. Hi, I would like to ask 843 00:33:52,240 --> 00:33:54,309 you about vertical stacking and just 844 00:33:54,310 --> 00:33:56,329 mentioned and the technology of through 845 00:33:56,330 --> 00:33:57,219 Silicon Valley. 846 00:33:57,220 --> 00:33:59,319 How do you make the holes 847 00:33:59,320 --> 00:34:00,320 in the chip? 848 00:34:03,370 --> 00:34:05,619 I'd say also with etching, 849 00:34:05,620 --> 00:34:07,809 but actually I don't know yet. 850 00:34:07,810 --> 00:34:09,339 Sorry. 851 00:34:09,340 --> 00:34:10,340 OK. 852 00:34:11,500 --> 00:34:13,388 And back to one, please. 853 00:34:13,389 --> 00:34:15,579 Uh, yeah, regarding just before 854 00:34:15,580 --> 00:34:17,499 the small feature size, what I've heard 855 00:34:17,500 --> 00:34:19,299 before is to actually do also photo 856 00:34:19,300 --> 00:34:21,488 lithography, not in air, but in other 857 00:34:21,489 --> 00:34:23,319 mediums. Is this actually true or not 858 00:34:23,320 --> 00:34:25,509 sort of like underwater or 859 00:34:25,510 --> 00:34:27,129 within a liquid where the wavelength 860 00:34:27,130 --> 00:34:28,569 changes? Exactly. 861 00:34:28,570 --> 00:34:30,339 You can use, uh, water. 862 00:34:30,340 --> 00:34:32,379 And this says, uh, dispersion. 863 00:34:32,380 --> 00:34:34,599 Well, it changes 864 00:34:34,600 --> 00:34:36,908 the, uh, the the wavelength of 865 00:34:36,909 --> 00:34:38,979 your ultraviolet light by a 866 00:34:38,980 --> 00:34:41,079 factor of one four four. 867 00:34:41,080 --> 00:34:43,419 And so now you have 868 00:34:43,420 --> 00:34:45,009 one dot four, four times smaller 869 00:34:45,010 --> 00:34:47,559 wavelength and you can image 870 00:34:47,560 --> 00:34:49,329 things that are one dot four, four times 871 00:34:49,330 --> 00:34:50,888 smaller. That's correct. 872 00:34:50,889 --> 00:34:53,049 Yeah. But it doesn't it's 873 00:34:53,050 --> 00:34:55,238 not the whole answer to 874 00:34:55,239 --> 00:34:57,609 how can you produce patterns that are 20 875 00:34:57,610 --> 00:34:59,619 nanometers with one hundred ninety three 876 00:34:59,620 --> 00:35:00,620 nanometers wide. 877 00:35:03,890 --> 00:35:05,979 Number three, please, I 878 00:35:05,980 --> 00:35:08,529 have a question regarding the bonding 879 00:35:08,530 --> 00:35:10,599 and a modern c.p.u, you 880 00:35:10,600 --> 00:35:12,969 have up to the thousand contacts, 881 00:35:12,970 --> 00:35:15,759 not to mention the letters, 882 00:35:15,760 --> 00:35:17,979 um, it's 883 00:35:17,980 --> 00:35:18,909 the bonding done. 884 00:35:18,910 --> 00:35:21,819 And to that in a two dimensional 885 00:35:21,820 --> 00:35:22,869 way as well. 886 00:35:22,870 --> 00:35:25,179 Or is it squeezed around 887 00:35:25,180 --> 00:35:27,579 the edges as 888 00:35:27,580 --> 00:35:29,139 though it's done in a two dimensional way 889 00:35:29,140 --> 00:35:30,039 nowadays? 890 00:35:30,040 --> 00:35:32,379 So basically, it looks 891 00:35:32,380 --> 00:35:34,569 somehow like your, 892 00:35:34,570 --> 00:35:37,029 uh, bogert or a package. 893 00:35:37,030 --> 00:35:39,309 But, uh, I mean, this migratory package 894 00:35:39,310 --> 00:35:41,319 is also some kind of substrate. 895 00:35:41,320 --> 00:35:43,499 And on the other side of the 896 00:35:43,500 --> 00:35:45,969 spogli, you have a similar array. 897 00:35:45,970 --> 00:35:48,099 And this time it's not. 898 00:35:48,100 --> 00:35:50,379 Well, it's similar to solar, but it's 899 00:35:50,380 --> 00:35:52,089 actually bonding Balts. 900 00:35:52,090 --> 00:35:54,519 And you use 901 00:35:54,520 --> 00:35:56,829 it while you put the wafer 902 00:35:56,830 --> 00:35:58,479 upside down on this. 903 00:35:58,480 --> 00:36:00,999 And, uh, bonders, 904 00:36:01,000 --> 00:36:03,219 I actually, 905 00:36:03,220 --> 00:36:05,379 uh, technique somehow like solar. 906 00:36:05,380 --> 00:36:07,509 You heat it up and then the the 907 00:36:07,510 --> 00:36:08,739 balls will attach. 908 00:36:11,490 --> 00:36:13,050 Do we have any more questions? 909 00:36:14,770 --> 00:36:16,690 No, then please think of speaker. 910 00:36:25,920 --> 00:36:27,839 We have a bit of time left now. 911 00:36:27,840 --> 00:36:29,639 Yeah, I mean, if you want to continue, 912 00:36:29,640 --> 00:36:32,459 actually, I have a question myself 913 00:36:32,460 --> 00:36:34,649 and this question was, 914 00:36:34,650 --> 00:36:37,439 is it somehow possible to 915 00:36:37,440 --> 00:36:40,409 recreate something like this process 916 00:36:40,410 --> 00:36:42,570 in a sort of Do-It-Yourself way? 917 00:36:43,590 --> 00:36:45,809 And I think it 918 00:36:45,810 --> 00:36:47,969 actually might be possible, 919 00:36:47,970 --> 00:36:49,509 so. 920 00:36:49,510 --> 00:36:51,669 I think well, I mean, I'm 921 00:36:51,670 --> 00:36:54,039 a physicist, so I think the 922 00:36:54,040 --> 00:36:56,109 the most well, 923 00:36:56,110 --> 00:36:58,329 significant step to this is 924 00:36:58,330 --> 00:37:00,959 this, um, just for 925 00:37:00,960 --> 00:37:04,419 the topography and 926 00:37:04,420 --> 00:37:06,549 well, turns out 927 00:37:06,550 --> 00:37:09,009 actually they sell, 928 00:37:09,010 --> 00:37:11,349 um, devices that 929 00:37:11,350 --> 00:37:13,509 create some kind of pattern in 930 00:37:13,510 --> 00:37:15,669 this microscopic dimensions, 931 00:37:15,670 --> 00:37:18,429 and that's called a Blu ray burner. 932 00:37:18,430 --> 00:37:19,430 So. 933 00:37:20,240 --> 00:37:23,209 Maybe we could use Rabenau 934 00:37:23,210 --> 00:37:25,369 to create a pattern not 935 00:37:25,370 --> 00:37:28,129 in a desk, but in. 936 00:37:28,130 --> 00:37:30,230 Well, uh, photoresistor. 937 00:37:31,710 --> 00:37:33,899 And that would be sort of the first 938 00:37:33,900 --> 00:37:36,239 step that is necessary to 939 00:37:36,240 --> 00:37:38,339 create, um, 940 00:37:38,340 --> 00:37:40,589 integrated circuits of our own, which 941 00:37:40,590 --> 00:37:43,169 are, you know, there 942 00:37:43,170 --> 00:37:45,519 know, we can trust these things. 943 00:37:45,520 --> 00:37:47,849 So because we have 944 00:37:47,850 --> 00:37:50,149 develop them all of the way. 945 00:37:52,920 --> 00:37:54,839 If someone wants to comment on microphone 946 00:37:54,840 --> 00:37:55,979 one, please. 947 00:37:55,980 --> 00:37:58,169 So that raises the question, what is 948 00:37:58,170 --> 00:38:00,929 the size of a I guess, 949 00:38:00,930 --> 00:38:03,299 the maximum size of a usable AC? 950 00:38:03,300 --> 00:38:05,129 If one were to try to carry this at home, 951 00:38:05,130 --> 00:38:06,899 how small would it have to be to actually 952 00:38:06,900 --> 00:38:07,900 be usable? 953 00:38:09,030 --> 00:38:11,219 Well, I mean, you saw this 82, 954 00:38:11,220 --> 00:38:13,709 86, and this is, uh, 955 00:38:13,710 --> 00:38:15,749 patterns of the size of one, not five 956 00:38:15,750 --> 00:38:17,099 microns. 957 00:38:17,100 --> 00:38:19,289 And I mean, Rurales 958 00:38:19,290 --> 00:38:21,239 have much smaller in the order of 400 959 00:38:21,240 --> 00:38:24,179 nanometers, small feature size. 960 00:38:24,180 --> 00:38:26,369 So, well, 961 00:38:26,370 --> 00:38:28,469 it sort of points in the direction that 962 00:38:28,470 --> 00:38:30,839 you could create patterns 963 00:38:30,840 --> 00:38:33,149 this size. And I mean, this would 964 00:38:33,150 --> 00:38:35,819 create a functional if not very fast. 965 00:38:35,820 --> 00:38:36,900 Uh, Chip. No. 966 00:38:39,450 --> 00:38:40,450 He's nodding. 967 00:38:43,600 --> 00:38:45,709 Mike, number four, please. 968 00:38:45,710 --> 00:38:48,129 Um, if you want to create 969 00:38:48,130 --> 00:38:50,619 a chip at your homebrew, 970 00:38:50,620 --> 00:38:52,779 uh, equipment, how will 971 00:38:52,780 --> 00:38:54,279 you do the doping? 972 00:38:54,280 --> 00:38:56,349 Because I think you need some 973 00:38:56,350 --> 00:38:58,539 kind of way cume for it and all 974 00:38:58,540 --> 00:39:00,339 that stuff. And it's not that easy to 975 00:39:00,340 --> 00:39:01,340 handle. 976 00:39:02,400 --> 00:39:04,439 That's not the only part when you need a 977 00:39:04,440 --> 00:39:07,229 vacuum, by the way, but, um, 978 00:39:07,230 --> 00:39:09,539 well, you can just buy 979 00:39:09,540 --> 00:39:11,789 vacuum pumps and 980 00:39:11,790 --> 00:39:14,459 for the implantation thing, you actually 981 00:39:14,460 --> 00:39:16,349 and that's how it was done in the old 982 00:39:16,350 --> 00:39:18,689 days, you can just flow 983 00:39:18,690 --> 00:39:21,749 gas over the, uh, the wafer. 984 00:39:21,750 --> 00:39:24,299 And this will partially diffuse 985 00:39:24,300 --> 00:39:26,369 into the silicon under the right 986 00:39:26,370 --> 00:39:29,039 conditions. I mean, you have to, uh, 987 00:39:29,040 --> 00:39:30,210 heat up the way for. 988 00:39:31,450 --> 00:39:33,759 And then, uh, gas 989 00:39:33,760 --> 00:39:36,279 atoms will diffuse into, uh, 990 00:39:36,280 --> 00:39:37,280 the silicon. 991 00:39:38,320 --> 00:39:40,539 So that's much easier than this 992 00:39:40,540 --> 00:39:42,639 iron implantation device, but it's also 993 00:39:42,640 --> 00:39:44,859 not as accurate, you 994 00:39:44,860 --> 00:39:47,239 can control the doping to that accuracy. 995 00:39:50,590 --> 00:39:51,789 Do we have any comments from the 996 00:39:51,790 --> 00:39:53,319 Internet? By the way, I haven't seen the 997 00:39:53,320 --> 00:39:55,329 signal angel in ages, OK. 998 00:39:55,330 --> 00:39:56,830 Uh, my phone number two, please. 999 00:39:58,360 --> 00:40:00,819 Concerning the self-made stuff, 1000 00:40:00,820 --> 00:40:02,979 what are the tolerances for the 1001 00:40:02,980 --> 00:40:05,079 thickness of the substrate of 1002 00:40:05,080 --> 00:40:06,189 the removing and so on? 1003 00:40:06,190 --> 00:40:08,439 Because if we get it within 1004 00:40:08,440 --> 00:40:10,509 like 10 1005 00:40:10,510 --> 00:40:12,609 percent or so, then probably we 1006 00:40:12,610 --> 00:40:14,499 can do it. But I don't know if I can do 1007 00:40:14,500 --> 00:40:16,719 it by hand with the accuracy needed. 1008 00:40:18,170 --> 00:40:20,329 Well, actually, that's what you need 1009 00:40:20,330 --> 00:40:22,999 for a photo lithography for, 1010 00:40:23,000 --> 00:40:25,429 uh, you don't rely 1011 00:40:25,430 --> 00:40:28,189 on the etching going 1012 00:40:28,190 --> 00:40:31,069 something like 20 nanometers 1013 00:40:31,070 --> 00:40:34,129 plus minus zero point one percent, 1014 00:40:34,130 --> 00:40:36,499 but you rely on the next material 1015 00:40:36,500 --> 00:40:39,049 being different and the etching 1016 00:40:39,050 --> 00:40:41,569 stopping at the layer between 1017 00:40:41,570 --> 00:40:43,969 the well, the the the, 1018 00:40:43,970 --> 00:40:45,739 uh, material change. 1019 00:40:47,210 --> 00:40:49,459 And so that would be actually easy, 1020 00:40:49,460 --> 00:40:50,660 I hope. Thanks. 1021 00:40:52,610 --> 00:40:54,060 At Magnon before your killing. 1022 00:40:55,100 --> 00:40:57,019 OK, go ahead. 1023 00:40:57,020 --> 00:40:59,300 So you're talking about 1024 00:41:00,470 --> 00:41:02,659 you making your own chips, 1025 00:41:02,660 --> 00:41:04,549 but I don't really think it would be very 1026 00:41:04,550 --> 00:41:06,979 economical or feasible 1027 00:41:06,980 --> 00:41:09,079 and reasonable way to make it 1028 00:41:09,080 --> 00:41:10,080 at home. 1029 00:41:10,940 --> 00:41:12,499 There are a lot of fabrications out 1030 00:41:12,500 --> 00:41:14,629 there. We can just send your design and 1031 00:41:14,630 --> 00:41:15,589 make it there. 1032 00:41:15,590 --> 00:41:17,449 Wouldn't it be more interesting to really 1033 00:41:17,450 --> 00:41:19,759 create a platform to make that more 1034 00:41:19,760 --> 00:41:22,159 affordable so we can just 1035 00:41:22,160 --> 00:41:24,739 order our own chips made custom 1036 00:41:24,740 --> 00:41:25,740 at the fab? 1037 00:41:27,010 --> 00:41:29,109 Actually, there are people who 1038 00:41:29,110 --> 00:41:31,239 actually more companies that go 1039 00:41:31,240 --> 00:41:33,339 into that direction, you 1040 00:41:33,340 --> 00:41:35,979 may have heard of one of these risk free, 1041 00:41:35,980 --> 00:41:38,489 um, processors out there, 1042 00:41:38,490 --> 00:41:40,659 um, I think they're called 1043 00:41:40,660 --> 00:41:41,889 sci fi or something. 1044 00:41:41,890 --> 00:41:44,169 That and actually one 1045 00:41:44,170 --> 00:41:45,999 of the things that they want to do is 1046 00:41:46,000 --> 00:41:49,479 offer, well, fabrication. 1047 00:41:49,480 --> 00:41:50,480 Um. 1048 00:41:51,740 --> 00:41:54,379 As sort of as a service, 1049 00:41:54,380 --> 00:41:56,989 I mean, you can if you're a company 1050 00:41:56,990 --> 00:41:58,849 and come up with your own design, you can 1051 00:41:58,850 --> 00:42:01,399 already, uh, do the manufacturing 1052 00:42:01,400 --> 00:42:03,659 as a service. That's one of the 1053 00:42:03,660 --> 00:42:05,959 well, basically all of the big fabs 1054 00:42:05,960 --> 00:42:06,960 offer. 1055 00:42:07,980 --> 00:42:10,349 But if you're just some 1056 00:42:10,350 --> 00:42:13,169 people, then, uh, 1057 00:42:13,170 --> 00:42:15,299 it gets easier, but it's still, 1058 00:42:15,300 --> 00:42:16,489 well, expensive. 1059 00:42:20,340 --> 00:42:22,559 We get Mike one, please. 1060 00:42:22,560 --> 00:42:25,679 Yes, I also wanted to comment on 1061 00:42:25,680 --> 00:42:28,049 making these chips at home and yes, 1062 00:42:28,050 --> 00:42:30,779 I agree it would probably be, 1063 00:42:30,780 --> 00:42:32,939 well, at least very hard to do this 1064 00:42:32,940 --> 00:42:35,069 because you would need to miniaturize 1065 00:42:35,070 --> 00:42:36,809 the equipment and stuff and it would need 1066 00:42:36,810 --> 00:42:39,029 to be cheap and 1067 00:42:39,030 --> 00:42:40,030 well, 1068 00:42:41,100 --> 00:42:43,469 if you could actually make 1069 00:42:43,470 --> 00:42:45,269 chips that small at home, you'd have a 1070 00:42:45,270 --> 00:42:47,549 lot of problems, I think, with. 1071 00:42:47,550 --> 00:42:49,349 Well, as you said, you need dust, free 1072 00:42:49,350 --> 00:42:51,179 environments and stuff. 1073 00:42:51,180 --> 00:42:53,279 And I 1074 00:42:53,280 --> 00:42:55,559 don't know, it might be possible, but 1075 00:42:55,560 --> 00:42:58,049 I think it's probably extremely expensive 1076 00:42:58,050 --> 00:43:00,239 or it would take a really, really long 1077 00:43:00,240 --> 00:43:02,519 time to mature the process 1078 00:43:02,520 --> 00:43:04,619 that much. And I don't think 1079 00:43:04,620 --> 00:43:06,239 it would have that many people in the 1080 00:43:06,240 --> 00:43:07,829 world interested in making their own 1081 00:43:07,830 --> 00:43:08,849 chips. 1082 00:43:08,850 --> 00:43:10,949 And if you don't have that 1083 00:43:10,950 --> 00:43:13,049 kind of economic factor driving 1084 00:43:13,050 --> 00:43:16,109 it, I don't think it would ever work. 1085 00:43:16,110 --> 00:43:18,329 And regarding the other 1086 00:43:18,330 --> 00:43:20,789 guys saying, well, you can just 1087 00:43:20,790 --> 00:43:22,949 send your own plans to the fab 1088 00:43:22,950 --> 00:43:23,950 lab. 1089 00:43:24,630 --> 00:43:27,209 Well, if you're talking about modifying 1090 00:43:27,210 --> 00:43:29,279 these chips or preventing the 1091 00:43:29,280 --> 00:43:30,629 modification of these chips and 1092 00:43:30,630 --> 00:43:32,849 preventing someone from installing back 1093 00:43:32,850 --> 00:43:34,559 doors, you can't really send it away and 1094 00:43:34,560 --> 00:43:36,329 be sure it's not modified. 1095 00:43:36,330 --> 00:43:38,519 I mean, you need some way to 1096 00:43:38,520 --> 00:43:40,619 actually validate that your chip is 1097 00:43:40,620 --> 00:43:41,579 not modified at all. 1098 00:43:41,580 --> 00:43:43,559 And as you said, you only need to change 1099 00:43:43,560 --> 00:43:45,839 a few transistors on a 1100 00:43:45,840 --> 00:43:48,329 scale of, well, 20 nanometers 1101 00:43:48,330 --> 00:43:50,549 and you have a back door and 1102 00:43:50,550 --> 00:43:52,709 you wouldn't be able to prevent that if 1103 00:43:52,710 --> 00:43:54,449 you just sent your design away. 1104 00:43:55,730 --> 00:43:56,730 Yeah, that's it. 1105 00:43:59,560 --> 00:44:01,779 I'm not really on the same 1106 00:44:01,780 --> 00:44:03,969 side of you, but we'll 1107 00:44:03,970 --> 00:44:05,800 just leave it as a comment, I guess. 1108 00:44:07,210 --> 00:44:08,780 Oh, yeah, well, I don't know if 1109 00:44:10,040 --> 00:44:11,469 you can answer, of course. 1110 00:44:13,490 --> 00:44:15,589 I mean, it just also as 1111 00:44:15,590 --> 00:44:17,749 a comment, I think, I 1112 00:44:17,750 --> 00:44:18,750 mean, I. 1113 00:44:20,310 --> 00:44:22,529 For for the fabrication, 1114 00:44:22,530 --> 00:44:25,109 I think for me as a person, 1115 00:44:25,110 --> 00:44:27,449 it would be extremely rewarding 1116 00:44:27,450 --> 00:44:29,579 to get to that 1117 00:44:29,580 --> 00:44:31,679 point, to have my own chip 1118 00:44:31,680 --> 00:44:34,349 made it sort of the the final frontier 1119 00:44:34,350 --> 00:44:35,350 of making. No. 1120 00:44:36,860 --> 00:44:38,659 Yes, yes, I don't doubt that. 1121 00:44:38,660 --> 00:44:40,999 I was just I don't know, I 1122 00:44:41,000 --> 00:44:42,439 don't I don't say it's impossible and I 1123 00:44:42,440 --> 00:44:43,759 don't say you shouldn't do it. 1124 00:44:43,760 --> 00:44:45,109 If you want to do it, do it. 1125 00:44:45,110 --> 00:44:46,249 You can do anything. 1126 00:44:46,250 --> 00:44:48,439 But, uh, well, there's 1127 00:44:48,440 --> 00:44:50,599 a lot of really high obstacles 1128 00:44:50,600 --> 00:44:51,709 to get there, I think. 1129 00:44:51,710 --> 00:44:52,949 Definitely, yes. 1130 00:44:52,950 --> 00:44:54,289 Yes. 1131 00:44:54,290 --> 00:44:55,699 And concern the other thing. 1132 00:44:55,700 --> 00:44:58,159 Well, it's you 1133 00:44:58,160 --> 00:45:00,559 it's much simpler, actually, 1134 00:45:00,560 --> 00:45:03,169 to etch a chip and then 1135 00:45:03,170 --> 00:45:05,389 see it to micrographs and 1136 00:45:05,390 --> 00:45:07,519 well, see whether it 1137 00:45:07,520 --> 00:45:08,989 sort of checks out. 1138 00:45:08,990 --> 00:45:10,010 So it's. 1139 00:45:11,330 --> 00:45:13,729 I think it's 1140 00:45:13,730 --> 00:45:15,829 quite hard still to 1141 00:45:15,830 --> 00:45:17,809 put a back door like that. 1142 00:45:17,810 --> 00:45:20,209 It's definitely much 1143 00:45:20,210 --> 00:45:22,639 better. You're in a much better position 1144 00:45:22,640 --> 00:45:24,739 than when you just buy some intel 1145 00:45:24,740 --> 00:45:25,999 chip. 1146 00:45:26,000 --> 00:45:27,409 Oh, yeah. 1147 00:45:27,410 --> 00:45:29,149 I mean, yeah, you would have to take the 1148 00:45:29,150 --> 00:45:31,309 intel chip apart. Of course, 1149 00:45:31,310 --> 00:45:33,379 I well, I got the impression it was 1150 00:45:33,380 --> 00:45:36,439 hard from what you said earlier, because 1151 00:45:36,440 --> 00:45:38,599 I understood that you needed 1152 00:45:38,600 --> 00:45:41,269 to take every layer apart to actually 1153 00:45:41,270 --> 00:45:43,159 verify it and not you can just use a 1154 00:45:43,160 --> 00:45:44,689 microphone microscope. 1155 00:45:44,690 --> 00:45:46,759 But if well, if you can then 1156 00:45:46,760 --> 00:45:49,309 of course, as well it's easier. 1157 00:45:49,310 --> 00:45:52,019 I mean, it's sort of both as true. 1158 00:45:52,020 --> 00:45:53,539 Uh, it's just. 1159 00:45:57,840 --> 00:46:01,139 You're not in the best position by 1160 00:46:01,140 --> 00:46:03,539 matching up a chip and then sort of 1161 00:46:03,540 --> 00:46:05,639 hoping that, uh, what 1162 00:46:05,640 --> 00:46:07,889 you want, what you're checking 1163 00:46:07,890 --> 00:46:10,079 resolves to, um. 1164 00:46:11,620 --> 00:46:13,959 Well, when your check turns 1165 00:46:13,960 --> 00:46:16,059 out to that, you're 1166 00:46:16,060 --> 00:46:18,219 actually fine, but it's also 1167 00:46:18,220 --> 00:46:20,499 well, the the other side is 1168 00:46:20,500 --> 00:46:22,929 not really a good point either. 1169 00:46:22,930 --> 00:46:24,129 Yeah, I just got an idea. 1170 00:46:24,130 --> 00:46:26,439 You could maybe implement some kind of 1171 00:46:26,440 --> 00:46:28,689 check in your own chip design to check 1172 00:46:28,690 --> 00:46:30,309 if it's been tampered with. 1173 00:46:30,310 --> 00:46:32,609 You know, some kind of check, 1174 00:46:32,610 --> 00:46:34,719 some what 1175 00:46:34,720 --> 00:46:36,789 the other guy who was a bit rude 1176 00:46:36,790 --> 00:46:39,219 said earlier, but, 1177 00:46:39,220 --> 00:46:41,589 well, not like on the hardware 1178 00:46:41,590 --> 00:46:43,959 level by implementing 1179 00:46:43,960 --> 00:46:46,389 some gates that 1180 00:46:46,390 --> 00:46:48,849 return a different value once your design 1181 00:46:48,850 --> 00:46:51,580 has been altered by a few transistors. 1182 00:46:53,520 --> 00:46:55,619 Yeah, but you need to prove that this 1183 00:46:55,620 --> 00:46:56,939 actually works out. 1184 00:46:56,940 --> 00:46:59,009 Yes, of course, it's also very hard. 1185 00:46:59,010 --> 00:47:00,779 I think there might be people who are 1186 00:47:00,780 --> 00:47:02,939 doing this, but I don't know 1187 00:47:02,940 --> 00:47:05,249 whether this has actual positive 1188 00:47:05,250 --> 00:47:06,719 results yet. 1189 00:47:06,720 --> 00:47:09,269 Yes, it's it's pretty interesting. 1190 00:47:09,270 --> 00:47:10,769 I'm going to go ahead and cut you off 1191 00:47:10,770 --> 00:47:12,119 now. Sorry, Mike. 1192 00:47:12,120 --> 00:47:13,349 No fall, please. 1193 00:47:13,350 --> 00:47:14,350 Am I 1194 00:47:15,870 --> 00:47:17,219 having at home? I'm not going to be 1195 00:47:17,220 --> 00:47:19,229 possible because of the vibrations. 1196 00:47:19,230 --> 00:47:21,029 You have been in the lab where you 1197 00:47:21,030 --> 00:47:22,589 manufacture them themselves and you 1198 00:47:22,590 --> 00:47:24,209 remember it's probably a vibration, 1199 00:47:24,210 --> 00:47:27,239 isolated or dampened basement. 1200 00:47:27,240 --> 00:47:29,279 And only your neighbors using the 1201 00:47:29,280 --> 00:47:31,589 staircase is already vibrating your 1202 00:47:31,590 --> 00:47:34,319 equipment so much that at modern 1203 00:47:34,320 --> 00:47:36,239 nanometer size, it's just not going to be 1204 00:47:36,240 --> 00:47:37,240 feasible. 1205 00:47:38,020 --> 00:47:40,029 For nanometer size, your true yeah, 1206 00:47:40,030 --> 00:47:42,279 right, but, uh, the the cleanroom 1207 00:47:42,280 --> 00:47:44,949 I was at was actually not 1208 00:47:44,950 --> 00:47:47,379 vibration free vibration dampened. 1209 00:47:47,380 --> 00:47:49,659 So I mean, we, 1210 00:47:49,660 --> 00:47:51,839 uh, colleagues of mine produced, 1211 00:47:51,840 --> 00:47:54,099 uh, pictures of 1212 00:47:54,100 --> 00:47:56,949 10 nanometers or so, and they are fine. 1213 00:47:56,950 --> 00:47:59,739 So this is probably not really 1214 00:47:59,740 --> 00:48:00,969 repeatable. 1215 00:48:00,970 --> 00:48:03,159 But, um, I mean, I'm 1216 00:48:03,160 --> 00:48:05,799 not really wanting to produce 1217 00:48:05,800 --> 00:48:08,109 20 nanometers, but I would be fine 1218 00:48:08,110 --> 00:48:09,579 with the micrometer size. 1219 00:48:09,580 --> 00:48:12,129 And I think at that resolution, 1220 00:48:12,130 --> 00:48:14,339 it's not that much of a problem. 1221 00:48:16,790 --> 00:48:17,989 My number two, please. 1222 00:48:19,350 --> 00:48:21,649 I'm asking, is it possible to buy 1223 00:48:21,650 --> 00:48:23,839 used equipment so something which 1224 00:48:23,840 --> 00:48:26,089 was used sometime some years ago 1225 00:48:26,090 --> 00:48:29,059 and some Fairpoint, which is now 1226 00:48:29,060 --> 00:48:32,149 thrown away because we have better stuff 1227 00:48:32,150 --> 00:48:33,319 for research equipment? 1228 00:48:33,320 --> 00:48:34,790 That's definitely the case. 1229 00:48:36,010 --> 00:48:38,079 But I mean, you're still looking, I 1230 00:48:38,080 --> 00:48:40,269 think, at a couple 1231 00:48:40,270 --> 00:48:41,709 of ten thousand bucks. 1232 00:48:43,800 --> 00:48:44,800 For us, it. 1233 00:48:46,100 --> 00:48:47,100 My sister. 1234 00:48:48,120 --> 00:48:50,549 I don't know, maybe other universities 1235 00:48:50,550 --> 00:48:52,229 which are not that well off. 1236 00:48:56,390 --> 00:48:57,739 OK, to again. 1237 00:48:59,480 --> 00:49:01,609 In case we could talk about 1238 00:49:01,610 --> 00:49:03,829 equipment like 1239 00:49:03,830 --> 00:49:06,139 rent equipment, would you be into 1240 00:49:06,140 --> 00:49:08,539 helping developing a 1241 00:49:08,540 --> 00:49:10,909 process for making 1242 00:49:10,910 --> 00:49:13,399 trips with this Blu ray? 1243 00:49:13,400 --> 00:49:16,469 Uh, let's talk about this afterwards. 1244 00:49:16,470 --> 00:49:18,529 So the thing that you 1245 00:49:18,530 --> 00:49:20,929 have to know is that we know each other 1246 00:49:20,930 --> 00:49:22,609 and, uh, well, there is a previous 1247 00:49:22,610 --> 00:49:24,089 discussion going on. 1248 00:49:24,090 --> 00:49:25,090 Uh. 1249 00:49:27,340 --> 00:49:28,570 Do you have anything to add? 1250 00:49:30,670 --> 00:49:32,349 Apparently not. 1251 00:49:32,350 --> 00:49:35,079 Oh, actually, someone had my number three 1252 00:49:35,080 --> 00:49:37,539 is I so 1253 00:49:37,540 --> 00:49:39,699 silicone is not the only process 1254 00:49:39,700 --> 00:49:41,949 you can use to make chips can speculate 1255 00:49:41,950 --> 00:49:44,589 maybe if there is a possible 1256 00:49:44,590 --> 00:49:47,229 future other than silicon, 1257 00:49:47,230 --> 00:49:48,230 either in silicon. 1258 00:49:50,020 --> 00:49:52,419 So for us, 1259 00:49:52,420 --> 00:49:54,969 produce chips, basically. 1260 00:49:54,970 --> 00:49:57,819 No, for processers, I think 1261 00:49:57,820 --> 00:49:59,919 Silicon will be there 1262 00:49:59,920 --> 00:50:02,139 for a while. I mean, you have 1263 00:50:02,140 --> 00:50:04,479 processes which enabled you 1264 00:50:04,480 --> 00:50:06,909 to communicate with light 1265 00:50:06,910 --> 00:50:09,339 to the outside world and 1266 00:50:09,340 --> 00:50:11,409 that sort of where 1267 00:50:11,410 --> 00:50:13,899 today's processor are, well, 1268 00:50:13,900 --> 00:50:15,819 reaching a boundary with electrical 1269 00:50:15,820 --> 00:50:16,959 communication. 1270 00:50:16,960 --> 00:50:19,029 So once this is 1271 00:50:19,030 --> 00:50:21,369 out there, um, 1272 00:50:21,370 --> 00:50:23,619 silicon based devices will be fine for 1273 00:50:23,620 --> 00:50:25,749 a while. I mean, sort of the 1274 00:50:25,750 --> 00:50:26,949 the next thing. 1275 00:50:28,010 --> 00:50:30,529 That drastically would change, 1276 00:50:30,530 --> 00:50:32,179 uh, once you go. 1277 00:50:33,650 --> 00:50:35,779 Um, for a while, 1278 00:50:35,780 --> 00:50:38,509 not switching based on charges, 1279 00:50:38,510 --> 00:50:41,779 but on photons while 1280 00:50:41,780 --> 00:50:43,639 everything changes, but I don't really 1281 00:50:43,640 --> 00:50:45,499 see that happening. 1282 00:50:45,500 --> 00:50:47,089 That's where your question was going, 1283 00:50:47,090 --> 00:50:48,309 right? 1284 00:50:48,310 --> 00:50:51,189 I mean, just as another 1285 00:50:51,190 --> 00:50:53,559 semiconductor material, so, 1286 00:50:53,560 --> 00:50:55,959 oh, the other semiconductor material, so 1287 00:50:55,960 --> 00:50:57,489 yeah, there are other semiconductor 1288 00:50:57,490 --> 00:50:59,619 materials, but 1289 00:50:59,620 --> 00:51:01,809 I don't think for 1290 00:51:01,810 --> 00:51:04,599 processing for for basically, 1291 00:51:04,600 --> 00:51:06,699 um, well, 1292 00:51:06,700 --> 00:51:08,859 data stuff, uh, there is something else 1293 00:51:08,860 --> 00:51:09,309 coming up. 1294 00:51:09,310 --> 00:51:11,509 I mean, for power devices, the 1295 00:51:11,510 --> 00:51:13,959 silicon carbide and 1296 00:51:13,960 --> 00:51:16,239 obviously for light emitting 1297 00:51:16,240 --> 00:51:19,149 diodes and stuff, you use, uh, other, 1298 00:51:19,150 --> 00:51:20,409 um, materials. 1299 00:51:24,340 --> 00:51:25,939 OK, someone someone's walking into a 1300 00:51:25,940 --> 00:51:26,940 microphone for 1301 00:51:28,420 --> 00:51:30,579 hour. Do you have anything to to 1302 00:51:30,580 --> 00:51:31,580 say? 1303 00:51:31,960 --> 00:51:32,859 No, I'm fine. 1304 00:51:32,860 --> 00:51:34,630 OK, then let's thank Ari. 1305 00:51:40,990 --> 00:51:42,759 Thank you for this great discussion we 1306 00:51:42,760 --> 00:51:43,760 had. 1307 00:51:48,330 --> 00:51:50,309 Also, thanks for the guy with the laptop.