SemiWiki.com - Podcast EP292: The Expanding Worldwide Focus of the 83rd Device Research Conference with Dr. Tania Roy

Episode Date: June 20, 2025

Dan is joined by Dr. Tania Roy, Associate Professor in Electrical and Computer Engineering at Duke University. She focuses on developing innovative hardware for artificial intelligence using advanced materials beyond silicon. Her research explores reliable gallium nitride (GaN) devices and new materials to move beyond … Read More

Hello, my name is Daniel Nenny, founder of SemiWiki, the open forum for semiconductor professionals. Welcome to the Semiconductor Insiders podcast series. My guest today is Dr. Tanya Roy, an associate professor in electrical and computer engineering at Duke University. She focuses on developing innovative hardware for artificial intelligence using advanced materials beyond silicon. Her research explores reliable gallium nitride devices and new materials to move beyond traditional silicon, aiming to power the next generation of chips. As the technical program chair for the 83rd Device Research Conference, or we call it

DRC 2025, she is shaping a dynamic program featuring breakthroughs in areas like quantum devices and chip integration. This conference is next week, set for June 22nd to the 25th at Duke. She also organizes other semiconductor conferences such as IEDM and SISC. Welcome to the podcast Tanya. Thank you Dan. So a question I like to ask first of our panelists is what first got you started in semiconductors? Do you have a story you can share? Sure it's been quite long since I've been interested in semiconductors. All started in college. When I have been an electrical engineer, so first thing we

learned to do is code. Then the question came up to me that what's underneath the code? A black box. Then we take other courses, they tell us, oh you make circuits like operational amplifiers. What's underneath that? That's also a black box. And then as I kept on taking courses, finally, I came up to this view where I can see electrons and holes moving when an electric field is being applied. And that satisfied me. I said, OK, I have got this view and this is the view I like the most.

I can't accept that electrons and holes are black boxes. So that's how it started. And my journey into semiconductors. Well, that's a good story. So DRC has been the launch pad for technologies like silicon MOSFETs and graphene transistors. What makes DRC 2025 at Duke University the place to be for the next big thing in semiconductors?

This is the 83rd year of device research conference and I will call it DRC from now on. DRC is a very academic conference environment is academic. We hold it in college campuses every year so that it is more accessible to students, faculties. Now for semiconductor research, an academic environment allows for new concepts to flourish. Devices that would be difficult to produce at an industrial scale or new materials which are still not introduced in commercial fabs, these kind of things will be showcased at DRC. And this year, particularly, we have a lot of interesting path-breaking works at the conference.

For example, what are the most interesting things that the device community is looking at right now? One is materials that can succeed silicon. And one of this is two-dimensional materials. This here at DRC, we are going to show the first CMOS microprocessor with two-dimensional materials. This work was done at Penn State. Then for another material is ferroelectrics for memory, ultrafast embedded memory applications.

We are going to show that sub nanosecond switching is possible and we can see the time switching dynamics at sub nanosecond scales for these material systems. And that's really important for for ultrafast memory applications. If we are going for back end of silicon technologies, what are the materials that can be put at the back end of silicon? One is amorphous oxide semiconductors. And this is actually something industry, it has its eyes on. It's very close to being industrialized. But they have their own issues, these materials. Some of these issues are their stability, their threshold voltage control. And we have

a whole set of talks based on these materials where we are showing great advancement in the fields. If you're thinking of wideband gap semiconductors for power applications, we are showing that these devices can be operated at really high temperatures, 600 degrees C, etc. Very good for space applications. DRC is a place where we can show new characterization techniques, device-based characterization techniques without destroying the devices, without opening up the devices and see things with a microscope. Just by electrical characterization techniques, we can see what is located where, what kind

of defects are present in these devices when they are operating. And neuromorphic computing, memory devices are still a very strong point of DRC and we have a lot of good papers on these devices. Dan. Great. As an organizer, what's got you personally excited about DRC 2025 and what is a standout memory from the past DRCs that shows, you know, it's magic?

That's a very good question. DRC has actually changed quite a lot since last year. We have doubled its size since last year. Sorry, since two years back. It used to be a cozy conference where a select few people would come and interact. But this year, actually since last year onwards, we've had a lot more device experts from Europe, Asia, and of course the US, attending and presenting their work. The

contributed papers for DRC have to be 100% original. You cannot show pre-published work at DRC. Everything is original. So all attendees get to see some very nice technical advancement in the field. Now what is DRC's magic, right? The most interesting part according to me is the RUMP sessions. They are a key part of DRC and this is an informal panel where people can speak their mind and they really do. When I was a student, the RUM sessions were the reason why I attended DRC. We could see in real time, faculties, industry experts sparring over their favorite technology, what would succeed the current technology. And it was a very open discussion. Everybody can give their opinions, it's not recorded. So people, and since the community is so close knit, everybody can speak their mind. This is also the best way to learn because it's not a dry

textbook. You can see the expert talking about their technology passionately in the most informal visual language and that impression lasts forever. We want to create the same magic this year as well. forever. We want to create the same magic this year as well. This year's topic is who defines important research directions for the field? Is it academia, industry or funding agencies? We have representatives from academia, government agencies, program managers, industry members who will be discussing this topic among each other and I'm looking forward to a very exciting conversation. Great. So DRC 2025 is tackling hot topics

like neuromorphic computing and Chiplik integration. Why are these sessions a big deal for today's chip designers? Dan, this is the future. If we are talking about AI today being all pervasive, we cannot keep depending on the cloud for AI. The hardware has to be brought closer to the sensor node. For example, we cannot keep serving food at the restaurant

table while the food is being cooked at a kitchen thousands of miles away. That's what is happening today for AI. The cloud is located in those data centers, which is thousands of miles away. That's what is happening today for AI. The cloud is located in these data centers which is thousands of miles away. As a semiconductor device community, it is our job to make new devices that will allow AI to run at the sensors even more advanced than what we understand by edge computing. So the question is why these hot topics, why are these sessions a big deal

for today's trip designers? Imagine your favorite sandwich, I would say. It has various layers. It has lettuce, it has tomato, cheese, onion, meat, whatever you want. And when you take a bite, you get all the nutrition and the delicious taste. Now, our job as device researchers today is to create that high tech sandwich where layers

of materials and devices can be stacked on top of each other in a 3D integrated pattern so that different functionalities can be embedded at each layer. You will get to see that happen at these sessions at DRC this year. OK. And who is the ideal conference attendee? Is this for engineers, students, or industry professionals? And what's the one thing that they will gain by showing up?

The broad answer would be engineers, device engineers, and also circuit engineers, circuit designers. Because the circuit designers will be able to see the latest devices that we are making and find ways of designing circuits out of them for best possible performance. Typically, as I said, DRC has been very academic. We have mostly students and faculties attending, but we want to change that. That's why we are doing this podcast, aren't we? We want industry to come and see the new devices that are being developed at the universities beyond

what they typically fund through the semiconductor research corporation, let's say, and see if these devices could be translated to commercialization. Today, there is a big divide between academia and industry as far as semiconductor devices are concerned. Of course, a university cannot match up to the scale of Intel or TSMC, but definitely it can produce new concepts, first demonstrations. There are many studies that can be done elegantly in academia, for example studying the reliability of a new technology or even the new microprocessor with 2D materials that I just talked about. I would urge industry professionals to come to DRC and see what is getting done at academia. So that this huge divide reduces

the way it was back in the 1960s or even the 1990s when academic research and industry research was quite close to each other. If we are talking about bringing semiconductor manufacturing to the US, we need to expand that base. GRC is quite an important conference. You will get to see almost all device experts and device motivated graduate students, some undergraduate students too. It's also great for recruitment for the semiconductor workforce

that is required. And this year, since last year essentially, we have introduced industry sessions where professionals from quite a few semiconductor companies like Micron, IBM, Intel, Teledyne, they will come and speak to the audience. That's a good start to have the integration happen. Also, we have invited several industry speakers to showcase their work. In our DRC technical program committee, we have industry members who get to review the papers and they have a say in which papers get selected. That also helps us because it tells us what type of work industry is looking for.

And as DRC is known for unveiling groundbreaking devices, what kind of innovations are you hoping to see this year? There's a lot of development in the field of in the various areas, various, which can, one was, as I mentioned, there is a lot of work being shown on two dimensional materials as the future, the microprocessors or many other types of devices with two dimensionaldimensional materials.

The work on amorphous oxide semiconductors for back end of line compatible technologies is also very good there. For wide bandgap semiconductors, we are showing a lot of prospects in power devices and for 5G, 6G applications. And this year's focus session is on heterogeneous integration. So there is a lot of good technologies for this heterogeneous integration, which is being shown here. You can see a lot of innovative devices for neuromorphic computing, in-memory computing, which are required for AI.

And can you paint a picture of the DRC vibe at Duke? Say researchers debating 2D materials over coffee or late night panel showdowns. Give us a sense of what it's like. Yeah. If you look at the schedule, you will see it is packed from early morning to late in the evening every day. The way we put up the schedule we have frequent breaks between sessions which is for a very important purpose. The conference venue is chosen such that there are seating areas around the conference rooms and we have unlimited coffee and tea. What we want to provide is an environment where the researchers

can talk to each other about the work that they are doing, learn from each other, and sit and ideate what the next step should be. This environment is very important for us. DRC started as a very close-knit community. It still is. All the device researchers that we know would be there. Another thing is that we don't have an online version of DRC.

Nothing is recorded, so people are free to speak their mind. Students and attendees can ask whatever question they want to ask the presenters. It is very typical of a DRC session to run into a debate on what is being presented. If people don't agree with the researchers, they are very open to criticize or give their suggestions. So every talk has the Q&A session after every talk is extremely lively. That is not something you will see in most of the device conferences. For example, at IEDM or VLSI, it's a lot more formal environment in DRC because people know each other very well and also there are new members. But still the environment is so friendly

and we have a strict no tie policy which makes people very comfortable when they are attending the conference and the discussions that happen at DRC is extremely lively and a lot can be learned through every talk Okay With big names like Nikki Lou speaking. How will DRC 2025 shape the future of the chip industry? Yeah, we have very good speakers this year. Nikki Lou is one of them Yeah, we have very good speakers this year. Diki Lu is one of them. All our planetary speakers are really big personalities,

Eli Yabdonovich, Suman Dutta, and the keynote speakers and the invited speakers. We are bringing a lot of big people from industry as well to express their, to showcase their work. And we want to show all the great work that is being done in the area. Now, I think that industry should come and see what breakthroughs are being shown at DRC.

And the interaction that we are expecting is that they will be able to tell academics like us how we can help them. And this is a very low cost investment for them because it's only just to register for the conference. They don't have to fund the research, but that discussion is very important. Faculties who are not funded through the industry, typically funded from federal agencies, they

can be helpful to the industry. Again, in tune with the round session topic. If we know what industry is looking for, we can channel our research in those directions and provide results. Also, I would urge industry to be open to accepting the new ideas that is generating at the conference. So I think that's how DRC 2025 will shape the future of the chip industry. There should be a very clean exchange of ideas.

Okay, and if you could hype up one session, what would your pick be and why? That is a little difficult for me because I curated the sessions meticulously. All the sessions are really very good. We have about 25% acceptance rate for oral presentations in this conference, which is really strict and very selective. The quality control is very high for DRC. That's why I'm saying it's very difficult to choose one particular session and say this is the best one. But still I would say if you are talking about research, then I would emphasize the sessions on the amorphous oxide semiconductors for back-end of line compatibility. Brilliant advancements are being shown on these devices at DRC this year, some of the best work so far in the field and definitely the short courses on Sunday is very good because it will help people be up to date with what's happening in the field.

Our topic, one of the topics, we have two short courses running in parallel. One topic is heterogeneous integration. The other is modeling of semiconductor devices. For the heterogeneous integrations short course, we have it covered from device level to the packaging levels. It gives a full picture of the current status and what is required in the area. Also, this short course is free for students this year because of generous grants from NSF and ERO, so we look forward to seeing a lot of people there. Okay, final question. What is the one step our listeners should take today to join DRC 2025 and be part of the device research revolution.

Just come to North Carolina, from June 22 to 2025, register for DRC and we can promise you, you'll have a very awesome four days. Thank you, Tanya. Appreciate your time and have a good conference. Maybe we can talk to you afterwards. Sure. Thank you so Tanya. Appreciate your time and have a good conference. Maybe we can talk to you afterwards. Sure. Thank you so much for the opportunity. That concludes our podcast. Thank you all for listening and have a great day.