1 00:00:03,303 --> 00:00:05,739 This is the Discovery Files podcast from the U.S. 2 00:00:05,739 --> 00:00:08,742 National Science Foundation. 3 00:00:10,110 --> 00:00:13,680 Semiconductors are the backbone of most modern electronic devices, 4 00:00:13,680 --> 00:00:17,250 from smartphones and home appliances to MRI scanners and satellites. 5 00:00:17,517 --> 00:00:20,086 Breakthroughs in semiconductors and microelectronics 6 00:00:20,086 --> 00:00:23,757 will be key to overcoming limits in critical areas, including artificial 7 00:00:23,757 --> 00:00:27,594 intelligence, quantum computing, manufacturing and communications. 8 00:00:27,861 --> 00:00:29,396 We're joined by Zetian Mi, 9 00:00:29,396 --> 00:00:31,898 professor of electrical engineering and computer science 10 00:00:31,898 --> 00:00:33,500 at the University of Michigan. 11 00:00:33,500 --> 00:00:36,036 His group is working on a new class of semiconductors 12 00:00:36,036 --> 00:00:39,205 with great potential for next generation microelectronic devices. 13 00:00:39,305 --> 00:00:41,541 Professor Mi, thanks so much for joining me today. 14 00:00:41,541 --> 00:00:42,542 Thank you very much Nate. 15 00:00:42,542 --> 00:00:46,813 It is my great honor and pleasure to have this opportunity to talk to you. 16 00:00:46,846 --> 00:00:50,483 So I need to ask you about wurtzite ferroelectric nitrites. 17 00:00:50,683 --> 00:00:52,786 What is this? 18 00:00:52,786 --> 00:00:56,556 So my group is working on three nitride semiconductors. 19 00:00:56,589 --> 00:01:00,527 This is material that's commonly used in our daily lives. 20 00:01:00,827 --> 00:01:04,497 For example, we use this material for LED lighting. 21 00:01:04,531 --> 00:01:09,002 They are also in our cell phones and literally everywhere. 22 00:01:09,002 --> 00:01:13,273 And the second most produced semiconductors on this planet 23 00:01:13,339 --> 00:01:15,642 next only to silicon. 24 00:01:15,642 --> 00:01:19,446 So this material has been around, but for decades. 25 00:01:19,712 --> 00:01:23,349 This material is known to be piezoelectric. 26 00:01:23,583 --> 00:01:27,020 Let me briefly explain what piezo electric is. 27 00:01:27,187 --> 00:01:32,358 So this material has fixed the polarization inside of the material. 28 00:01:33,026 --> 00:01:37,964 And, when you apply electric field, you can generate a string and vice versa. 29 00:01:37,997 --> 00:01:40,567 So they can also be used as sensors. 30 00:01:41,801 --> 00:01:43,803 So the ferroelectric 31 00:01:43,803 --> 00:01:47,240 is a subset of the piezoelectric family. 32 00:01:47,407 --> 00:01:52,345 By saying ferroelectric we mean the electrical polarization 33 00:01:52,345 --> 00:01:55,682 in the material can be reversed back and forth. 34 00:01:56,015 --> 00:01:59,018 When we apply an external electric field. 35 00:01:59,586 --> 00:02:02,589 Ferroelectric is not new, first discovered 36 00:02:02,822 --> 00:02:06,960 around 100 years ago, 1920 by a graduate student 37 00:02:06,993 --> 00:02:09,963 when he was doing PhD at University of Minnesota. 38 00:02:10,263 --> 00:02:14,100 But for nearly 100 years, the ferroelectric materials 39 00:02:14,100 --> 00:02:15,935 are mostly oxide based. 40 00:02:15,935 --> 00:02:20,440 Oxide based feroelectric they have been used almost everywhere. 41 00:02:20,440 --> 00:02:24,344 For example, in sensors, in industry settings, in medical devices, 42 00:02:24,344 --> 00:02:25,678 in ultrasonic. 43 00:02:25,678 --> 00:02:30,550 However, the oxide ferroelectric also has some fundamental limitations. 44 00:02:30,550 --> 00:02:34,654 For examble, they are not stable in harsh environments. 45 00:02:34,754 --> 00:02:39,492 It’s very difficult to make them compatible with a mainstream 46 00:02:39,692 --> 00:02:44,430 semiconductor processing, and therefore limit some of their applications. 47 00:02:45,098 --> 00:02:49,602 And here we have one of the most produced semiconductor material. 48 00:02:49,602 --> 00:02:53,439 And now we can turn it into ferroelectric. 49 00:02:53,573 --> 00:02:57,810 You can imagine the enormous opportunities this can open. 50 00:02:57,810 --> 00:03:00,446 And that's what we are really excited about. 51 00:03:00,446 --> 00:03:04,918 So how do you store information inside an electric field? 52 00:03:05,084 --> 00:03:08,087 So electric field in the semiconductor 53 00:03:08,521 --> 00:03:11,791 can have positive or negative directions. 54 00:03:12,125 --> 00:03:14,928 Which means we can switch the direction back 55 00:03:14,928 --> 00:03:17,931 and forth for ferroelectric material 56 00:03:18,131 --> 00:03:21,467 and then give us the opportunity to store the information. 57 00:03:21,467 --> 00:03:25,605 For example we can define a positive direction 58 00:03:25,605 --> 00:03:31,344 polarization as the information one, a negative as zero right. 59 00:03:31,377 --> 00:03:33,313 Then we can store the information. 60 00:03:33,313 --> 00:03:36,683 We can electrically reverse the polarization, 61 00:03:36,783 --> 00:03:41,321 which means we can write or reset the information in the material. 62 00:03:41,721 --> 00:03:45,458 Or we can read the material by applying a electrical signal. 63 00:03:45,458 --> 00:03:48,461 We can read the information stored inside. 64 00:03:48,528 --> 00:03:53,800 The most beautiful part of this is the information is stored there 65 00:03:54,000 --> 00:03:57,103 without applying any external bias. 66 00:03:57,136 --> 00:04:00,139 So once you set the polarization, there, 67 00:04:00,173 --> 00:04:04,410 it will stay there without supplying any external power. 68 00:04:04,410 --> 00:04:08,948 So this is a so-called nonvolatile memory device. 69 00:04:09,482 --> 00:04:12,852 And so imagine if we can build 70 00:04:13,219 --> 00:04:17,423 atomic scale memory cells with a very little power consumption, 71 00:04:17,624 --> 00:04:22,528 and we can integrate billions or trillions of them on a very small size chip. 72 00:04:23,029 --> 00:04:25,965 And that's going to really revolutionize 73 00:04:25,965 --> 00:04:29,535 computing communication in the years to come. 74 00:04:29,902 --> 00:04:33,339 So what is the benefit of lower power consumption with semiconductors? 75 00:04:33,673 --> 00:04:40,079 Computing communication is consuming enormous amount power as we speak. 76 00:04:40,079 --> 00:04:43,516 And with the AI this is going to grow exponentially. 77 00:04:43,549 --> 00:04:48,988 So how to reduce the power consumption at the device level will be very critical. 78 00:04:49,222 --> 00:04:53,793 Now let's first talk about this, this device power consumption. 79 00:04:53,926 --> 00:04:57,997 There are many factors that will affect the device power consumption. 80 00:04:58,164 --> 00:05:00,700 For example, for memory device. Right. 81 00:05:00,700 --> 00:05:04,470 How much power we need to reset or right to the signal, 82 00:05:04,637 --> 00:05:09,676 how much power we need to use to read the signal, or how much power 83 00:05:09,809 --> 00:05:14,080 we need to use to store the information to maintain the information. 84 00:05:14,347 --> 00:05:18,651 So what's really great about this material is nonvolatile. 85 00:05:18,751 --> 00:05:24,023 We do not need to have extra power to maintain the information. 86 00:05:24,991 --> 00:05:28,628 Now come back to the power consumption due to the writing 87 00:05:28,628 --> 00:05:31,998 or reading for a memory device or, other devices. 88 00:05:32,265 --> 00:05:36,669 It directly related to the efficiency and also relate to the device size. 89 00:05:37,136 --> 00:05:41,107 And for the wurtzite ferroelectric nitrides. 90 00:05:41,107 --> 00:05:45,078 Nowadays we can grow them atomic layer by atomic layer 91 00:05:45,378 --> 00:05:47,513 so we can make them very, very small. 92 00:05:47,513 --> 00:05:51,484 And therefore the power consumption can be drastically reduced 93 00:05:51,551 --> 00:05:55,388 together with very high integration density billions, 94 00:05:55,388 --> 00:05:59,726 billions of devices can be integrated on a very small size chip. 95 00:06:00,259 --> 00:06:03,663 What is the unique polarization capability in this material? 96 00:06:03,863 --> 00:06:05,298 Yes. I'm glad you asked. 97 00:06:05,298 --> 00:06:08,534 So for the wurtzite ferroelectric? 98 00:06:08,601 --> 00:06:15,241 There have been interest and studies by many research groups but for a long time. 99 00:06:15,241 --> 00:06:20,413 There is a fundamental question is for this wurtzite ferroelectric. 100 00:06:20,513 --> 00:06:24,851 Once we switch the polarization then we have all the this 101 00:06:25,051 --> 00:06:29,522 polarization within the same material, just like a bar magnet right. 102 00:06:29,522 --> 00:06:32,191 The north and south pole. 103 00:06:32,191 --> 00:06:37,830 When you try to put them together within one material, how can they be stabilized? 104 00:06:38,398 --> 00:06:41,401 And specifically in the wurtzite ferroelectric? 105 00:06:41,701 --> 00:06:44,203 When we switch the polarization 106 00:06:44,203 --> 00:06:48,341 only part of the material, the polarization is switched. 107 00:06:48,741 --> 00:06:52,011 So there will be domains of different polarization. 108 00:06:52,011 --> 00:06:57,683 At the interface, we may have two positive electric field facing each other. 109 00:06:58,284 --> 00:06:59,852 How can this be stabilized. 110 00:06:59,852 --> 00:07:02,855 This has remained a fundamental question. 111 00:07:02,855 --> 00:07:05,725 So that's one of the studies that we recently published. 112 00:07:05,725 --> 00:07:07,326 We showed that. 113 00:07:07,326 --> 00:07:10,263 And and this dummy interface, 114 00:07:10,263 --> 00:07:13,266 we have a new atomic scene structure 115 00:07:14,100 --> 00:07:17,270 and this atomic structure different 116 00:07:17,270 --> 00:07:22,041 from the host material and has unbonded electrons, 117 00:07:22,041 --> 00:07:26,412 which can stabilize the polarization discontinuity. 118 00:07:26,813 --> 00:07:29,882 Not only that, this electron density 119 00:07:30,450 --> 00:07:34,020 is much higher than that in the traditional gallium 120 00:07:34,020 --> 00:07:38,324 nitride transistors, that this is like 100 times higher. 121 00:07:38,624 --> 00:07:42,562 So provides future opportunities to design 122 00:07:42,962 --> 00:07:46,032 nanoscale device with better performance. 123 00:07:46,666 --> 00:07:49,368 How is this discovery going to really impact 124 00:07:49,368 --> 00:07:53,372 the next generation of semiconductors that people see in their devices. 125 00:07:53,739 --> 00:07:56,943 It’s going to impact in several different ways. 126 00:07:57,677 --> 00:08:03,916 Gallium nitride devices are already in many business sectors, right? 127 00:08:03,916 --> 00:08:06,052 Almost in our daily lives. 128 00:08:06,052 --> 00:08:09,121 And now we have this ferroelectricity 129 00:08:09,555 --> 00:08:12,225 and it can enhance the performance 130 00:08:12,225 --> 00:08:15,394 functionality of existing devices. 131 00:08:15,394 --> 00:08:20,032 For example, it can make our cell phone to have stronger, signal 132 00:08:20,233 --> 00:08:23,236 less noise to operate more efficiently, 133 00:08:23,870 --> 00:08:29,175 and secondly, it also broadens the scope of the traditional oxide 134 00:08:29,175 --> 00:08:32,311 ferroelectric beyond what can possibly be done. 135 00:08:32,812 --> 00:08:36,415 Oxide ferroelectric is a wonderful material. 136 00:08:36,449 --> 00:08:42,455 However, this material has some challenges when operating in harsh environment. 137 00:08:42,455 --> 00:08:45,658 For example in aerospace, 138 00:08:45,658 --> 00:08:49,562 or spacecraft, in electrical vehicles where you often 139 00:08:49,562 --> 00:08:53,132 have very high temperature or other extreme conditions. 140 00:08:53,666 --> 00:08:56,302 These wurtzite ferroelectric nitrides 141 00:08:56,302 --> 00:09:00,139 are known to be very stable at very high temperature. 142 00:09:00,139 --> 00:09:03,175 Up to 1000 degree Celsius. 143 00:09:03,209 --> 00:09:06,445 That can have immediately important applications. 144 00:09:07,113 --> 00:09:11,350 And also this wurtzite ferroelectric nitrides can be grown 145 00:09:11,717 --> 00:09:14,854 or synthesize atomic layer by atomic layer. 146 00:09:15,321 --> 00:09:21,160 And you can really scale these to very small size and with excellent performance. 147 00:09:21,460 --> 00:09:25,398 So that will help to make electronic devices 148 00:09:25,398 --> 00:09:28,668 with better performance with high integration density. 149 00:09:28,935 --> 00:09:33,039 And also these material can be integrated with the mainstream 150 00:09:33,272 --> 00:09:38,010 electronics silicon and gallium nitride which will all enhance 151 00:09:38,010 --> 00:09:41,647 the functionality of our future microelectronics. 152 00:09:42,148 --> 00:09:46,352 The exact scope and impact remains to be seen, but, 153 00:09:46,552 --> 00:09:51,223 many colleagues in academia, in the industry, we are very excited 154 00:09:51,223 --> 00:09:55,428 with the enormous potential of this new class of semiconductors. 155 00:09:56,162 --> 00:10:00,032 Are there challenges getting buy in from industry partners 156 00:10:00,032 --> 00:10:03,803 when you develop a new material or a new strategy 157 00:10:04,136 --> 00:10:08,741 to introduce these materials into devices that they are currently producing? 158 00:10:08,941 --> 00:10:13,346 Yes. As you correctly mentioned, for any new material, 159 00:10:13,346 --> 00:10:17,717 there are important considerations how they are going to be adopted 160 00:10:18,150 --> 00:10:23,656 by the industry, how this is going to be integrated with their production lines. 161 00:10:23,956 --> 00:10:26,959 Some materials may have wonderful properties, 162 00:10:27,026 --> 00:10:30,162 but if they cannot be adopted by industry, 163 00:10:30,296 --> 00:10:33,899 they will only remain as a subject of research interest. 164 00:10:34,500 --> 00:10:37,403 Very fortunately, for this material family. 165 00:10:37,403 --> 00:10:41,340 It is based on an existing material, gellium nitride 166 00:10:41,340 --> 00:10:45,277 is the second most produced semiconductor in industry, 167 00:10:45,711 --> 00:10:50,483 and we only add 1 or 2 elements into this material. 168 00:10:50,650 --> 00:10:52,952 And then we can transform this material 169 00:10:52,952 --> 00:10:57,023 to how wonderful properties not existing before. 170 00:10:57,289 --> 00:11:00,960 As a matter of fact, this material has already been 171 00:11:01,060 --> 00:11:04,096 adopted in our next generation cell phones. 172 00:11:04,430 --> 00:11:07,433 And we'll see more of those in the future. 173 00:11:07,466 --> 00:11:12,171 I want to ask you about the importance of critical minerals in semiconductors. 174 00:11:12,471 --> 00:11:16,542 Indeed, critical minerals are very important for semiconductors. 175 00:11:16,976 --> 00:11:22,515 For example, gallium, indium those are being used in the industry, 176 00:11:22,515 --> 00:11:25,951 but those are also considered critical materials. 177 00:11:26,318 --> 00:11:27,787 And the part of the research 178 00:11:27,787 --> 00:11:32,558 we are doing, among many other researchers is to develop new semiconductors, 179 00:11:32,958 --> 00:11:36,062 then potentially can use more abundant elements. 180 00:11:36,362 --> 00:11:39,298 And have more enhanced functionality. 181 00:11:39,298 --> 00:11:42,768 How hard will it be to adopt other materials 182 00:11:42,768 --> 00:11:48,040 that are more common or easier to use to replace the current critical minerals? 183 00:11:48,340 --> 00:11:52,411 As you also mentioned earlier, for any new semiconductor material, 184 00:11:52,745 --> 00:11:56,082 it often takes time for the industry to adopt. 185 00:11:56,515 --> 00:11:59,852 There's always a question about compatibility 186 00:12:00,019 --> 00:12:02,421 with semiconductor processing. 187 00:12:02,421 --> 00:12:05,424 The existing ones, among other factors. 188 00:12:05,624 --> 00:12:08,060 So it's not an easy 189 00:12:08,060 --> 00:12:12,064 answer to your question, but, I can give one example. 190 00:12:12,064 --> 00:12:14,700 For example, in the material we are working on, 191 00:12:14,700 --> 00:12:18,003 we are adding new elements into gallium nitride. 192 00:12:18,304 --> 00:12:21,707 So potentially we can have enhanced functionality 193 00:12:21,841 --> 00:12:28,013 by using less gallium or less indium, and therefore have less demand 194 00:12:28,013 --> 00:12:33,452 on the critical minerals without affecting the processing too much. 195 00:12:33,753 --> 00:12:37,857 How has NSF support impacted your work so far? 196 00:12:38,657 --> 00:12:42,128 I'm very grateful for the support from NSF. 197 00:12:42,495 --> 00:12:45,131 NSF not only provides support 198 00:12:45,131 --> 00:12:48,200 for graduate students, materials, supplies, 199 00:12:48,534 --> 00:12:53,572 but it provides the immense flexibility for us to explore 200 00:12:53,706 --> 00:12:58,410 the unknown domain in semiconductor and related research. 201 00:12:58,444 --> 00:13:02,982 It also help us to build a network to collaborate and connect 202 00:13:02,982 --> 00:13:04,483 with other researchers, 203 00:13:04,483 --> 00:13:08,654 not only at the University of Michigan, but also in other institutions. 204 00:13:09,121 --> 00:13:12,558 So this is really a very important resource 205 00:13:13,058 --> 00:13:16,862 for us and for university research to continue 206 00:13:16,862 --> 00:13:20,366 to make breakthrough advances in many sectors. 207 00:13:20,733 --> 00:13:24,870 For my last question today, I want to ask you about the future of semiconductors. 208 00:13:24,870 --> 00:13:25,938 Where do you see it 209 00:13:25,938 --> 00:13:29,408 developing or where do you see your work going in the next few years? 210 00:13:29,775 --> 00:13:34,046 So it's a very exciting time for semiconductor research, 211 00:13:34,079 --> 00:13:38,651 not only because it is important for our existing microelectronics, 212 00:13:39,151 --> 00:13:43,656 for example, to make our computers, cell phones to be more efficient, consume 213 00:13:43,656 --> 00:13:48,561 less power, to make our communication to be faster, to be more secure. 214 00:13:48,894 --> 00:13:53,299 But also it's important to open up future opportunities, 215 00:13:53,499 --> 00:13:58,671 for example, for medical, for healthcare, for future quantum technologies. 216 00:13:59,071 --> 00:14:02,308 These are some of the most exciting time for semiconductor 217 00:14:02,308 --> 00:14:05,311 research from a historical point of view. 218 00:14:05,578 --> 00:14:08,781 And for my own research at the University of Michigan, 219 00:14:09,048 --> 00:14:13,252 we have very talented colleagues working in materials, 220 00:14:13,619 --> 00:14:17,489 in device systems, circuits and applications. 221 00:14:17,523 --> 00:14:21,827 So I'm very fortunate to be a faculty member at the University of Michigan. 222 00:14:22,061 --> 00:14:26,799 So basically, for my own research, I focus on the development 223 00:14:26,799 --> 00:14:30,903 of next generation wide bandgap semiconductors. 224 00:14:31,237 --> 00:14:34,373 And then semiconductors can potentially be used 225 00:14:34,740 --> 00:14:38,010 to help make more efficient electronic 226 00:14:38,377 --> 00:14:42,648 optoelectronic devices, and also have some implication 227 00:14:42,848 --> 00:14:47,019 to make future quantum technologies closer to our life. 228 00:14:47,253 --> 00:14:48,787 Special thanks to Zetian Mi. 229 00:14:48,787 --> 00:14:50,422 For the Discovery Files, I'm Nate Pottker. 230 00:14:50,422 --> 00:14:51,490 You can watch video versions 231 00:14:51,490 --> 00:14:54,793 of these conversations on our YouTube channel by searching @NSFscience. 232 00:14:54,960 --> 00:14:58,030 Please subscribe wherever you get podcasts and if you like our program, 233 00:14:58,030 --> 00:15:00,232 share it with a friend and consider leaving a review. 234 00:15:02,034 --> 00:15:02,868 Discover how the U.S. 235 00:15:02,868 --> 00:15:05,604 National Science Foundation is advancing research at NSF.gov.