Apple Podcasts Stitcher Spotify Pocket Casts Google Podcasts


In this episode, Dr. Bryce Meredig and Prof. Mauro discuss:

  • Prof. Mauro’s early interest in computer programming, and how that shaped his career
  • Challenges and opportunities for applying data-driven modeling techniques, like machine learning, to materials development
  • Keys to a successful cross-functional materials research and development team
  • The importance of a data-driven culture and strategy for commercial success
  • How to bridge skill gaps within materials science through education, curriculum development, and collaborative research

“A company that has a data-driven culture is going to be much more effective at developing better products, faster, and at a lower cost. [This culture] will be essential for their long-term survival.” — Prof. John Mauro

Speaker Bio

Prof. John Mauro: After earning his PhD in glass science from Alfred University, Prof. John Mauro joined Corning Incorporated, where he eventually became the senior research manager of the Glass Research Department. He is the inventor or co-inventor of several new glass compositions for Corning, including Corning Gorilla® Glass products.

Dr. Mauro joined the faculty at Pennsylvania State University in 2017 and is currently a world-recognized expert in fundamental and applied glass science, statistical mechanics, computational and condensed matter physics, thermodynamics, and the topology of disordered networks.

He is the inventor of new models for supercooled liquid and glass viscosity, glass structure and topology, relaxation behavior, and thermal and mechanical properties. He is the author of more than 200 peer-reviewed publications and has given more than 200 presentations at international conferences and seminars. In addition, he is editor of the Journal of the American Ceramic Society.

Connect with Prof. John Mauro:
Faculty Website


Bryce Meredig: Welcome to DataLab, a Materials Informatics podcast, with Bryce Meredig, Chief Science Officer at Citrine Informatics.

Bryce Meredig: Okay. Welcome to DataLab. Our guest today is Professor John Mauro, who joined the faculty at Pennsylvania State University in 2017 after a long and distinguished career at Corning, Inc., where he was the inventor or co-inventor of several new glass compositions, including some Gorilla Glass products. Now John is a world-recognized expert in fundamental and applied glass science, statistical mechanics, computational and condensed matter physics, thermodynamics, and the topology of disordered networks. Professor Mauro is also the inventor of new models for supercooled liquid and gas viscosity, glass structure and topology, relaxation behavior and thermal-mechanical properties. He is the author of more than 200 peer reviewed publications and has given more than 200 presentations at international conferences and seminars. Finally, he is editor of the journal of the American Ceramics Society.

Bryce Meredig: John, thank you so much for joining us, and welcome to DataLab.

John Mauro: Thank you, Bryce. It’s a pleasure to be here.

Bryce Meredig: Well, I like to start the episodes with fun facts about our guests. We always have very interesting things to share about our guests. In your case, I learned that you started programming when you were just four years old. So I’d be interested to hear about your early engagement with computers and programming, and how that came about.

John Mauro: Yeah, thank you. Actually one of my earliest childhood memories, when I was four years old, was repeatedly asking my father, “How do you spell print?” Because I wanted to program, but I hadn’t yet figured out how to spell print. Which, of course, is an essential command when programming in BASIC. My first computer that my parents got for me was this old TI-994A. I don’t know if you’ve heard of that or seen it before. Actually, my family and I were just in England this year. We went to a science and technology museum, and they had one of these computers right there at the museum, which made me feel kind of old. Especially since there were dinosaurs there, too.

John Mauro: But anyway, this was a great computer at the time. It would hook up to your TV. It had 16 colors, it had a speech synthesizer. So it was really advanced for the early 1980s. And when you’d turn it on, it basically gave you a choice. You could press 1 to go into TI BASIC programming mode, or press 2 to insert a module and play some pre-programmed game. But my parents got me this book on an introduction to TI BASIC. And I was just totally into computers from the very start. I loved how I could type something for the computer to do, and it would execute exactly what I had written. So it was a great introduction to programming, and I’ve been passionate about computer programming ever since then.

John Mauro: So after my TI days, I later went to programming Apple II computers. Then it really took off in my Macintosh programming when I got to be in high school and in college. There I had programmed in several different games, some utility programs like a scientific calculator. Then I had released them as shareware, and they got distributed around the world. At that time, there were a lot of Mac computer magazines that came with CD-ROMs that were filled with these shareware programs, and I got a lot of users in the United States and Japan especially. But also quite a few users in Europe.

John Mauro: In fact … this is kind of funny, but my very first person who actually paid for my software was Steve Wozniak, the co-founder of Apple. I couldn’t believe it, as a high school student, and I got that email from steve@woz.org. He left a note about how much he liked my code and he actually bought three copies of it. So that was my very first sale. And of course, the words of encouragement from Steve Wozniak, who’s one of my greatest heroes, they just gave me more and more motivation to learn more and to write more programs.

Bryce Meredig: That’s a really neat story, especially since, I guess, many years later, Apple used your glass as well. So it was long-term business relationship.

John Mauro: Yeah. I guess I got a little bit on the software side as well as the hardware side.

Bryce Meredig: John, you studied computer science in your university as well, and I think nowadays we’re seeing more and more students pursue these dual backgrounds in physical sciences and engineering, as well as computer science. And you did that, of course, well before that became an established trend. What motivated you to continue studying these two fields simultaneously?

John Mauro: Well, I loved programming, but at the same time, I felt like I wanted to do more than just programming. I had also fallen in love with glass at a very young age. This was when I was, I think, six years old. My parents took me to the Corning Museum of Glass, which was about an hour away from where we lived. And I was just captivated by everything I saw in that museum, how glass could adapt so many different shapes and different colors, and how beautiful it could be.

John Mauro: So I was interested in glass, but I didn’t realize it was a valid career choice until I became a junior in high school. In my high school science class, I saw this video from Alfred University. It was by Professor Alex Clare, who is an outstanding scientist with a tremendous sense of humor, and just an incredibly creative person. She made this video called What is Ceramic Engineering that, at the time, was distributed on videotape. It was kind of in the style of Bill Nye the Science Guy, only it was all about ceramics and glasses. There was a part of that video where she talked about the potential use of glass in optical computing. So having glass interconnects to improve bandwidth and transmit information at the speed of light. And I felt like everything just came together for me. It was computers and it was glass, and I was interested in light and optics as well. It was like the synthesis of all this stuff that I was interested in. That’s what I decided I wanted to do.

John Mauro: So Alfred University was the only school that I applied to, and fortunately, they took me into their program. I ended up doing a double degree with a BA in computer science and BS in glass engineering science. It had been my goal from the start to try to combine the two. From there, that’s how I got into Corning in the Modeling and Simulation department, and was able to help establish the materials modeling capabilities at Corning.

Bryce Meredig: From your experience at Corning implementing these materials modeling capabilities, materials informatics capabilities, and applying them to glass development, in your experience, what were some of the challenges that you and your team overcame as you were applying these computational and data driven methods to glass in a large company?

John Mauro: I would say the biggest challenge is in communication and establishing a team where people from the materials science side and people from the data science side can talk to each other using a common language. Because the vast majority of people from a materials science point of view simply do not have that background in data science, and the reverse is true also, that the people who are trained in data science don’t have that background in materials science. So one of the biggest challenges is finding that common language where they can actually have conversations about the project and what they need to achieve, and be able to work effectively together as one unified team. If people don’t understand the other side, then they may be solving a problem that isn’t the right problem, or they might not necessarily trust each other and the different methods that they would use to approach the problem. So having that initial stage where the team can jell as one unit working together towards one goal, and keeping the lines of communication open, explaining things in a way that people from totally different backgrounds can understand you, I would say that’s the number one challenge I’ve encountered.

Bryce Meredig: Yeah, this language barrier issue between the material science community and the data science community is real challenge. Even, I would say, within material science between those with experimental backgrounds and those with computational backgrounds. Were there any specific strategies or approaches that you took or you found to be successful to help people overcome these natural barriers and collaborate together more effectively?

John Mauro: Well, I don’t think I have a universal solution to the problem, but I think that at the very beginning of the project, it’s really important to bring people together and discuss all the objectives as a team. To make sure everybody can understand what we’re driving towards and everybody’s in agreement on that. Also, to introduce the team members with each other to talk about their backgrounds and to understand everyone’s roles and responsibilities, and the important role that every one of them plays towards achieving the team’s objectives. If that doesn’t happen, if the team starts off in kind of a fragmented fashion, then that’s where I’ve seen the biggest difficulty. So I think starting off on the right foot is really essential for the long term success of a project.

Bryce Meredig: I think you’re right. Just un-siloing people, bringing them together. Avoiding this mode that I think happens too often in our community, where maybe the computationalists or the material informatics folks go away and make some predictions, and toss those predictions over the wall to their experimental colleagues. We’re missing a lot of opportunity for communication and collaboration when we do things that way.

John Mauro: Yeah. Exactly. And like you said, it’s the same type of thing even within the field of materials science. I can see that in the glass community. There are sub communities within the glass science community. There’s the more material science based glass community, which tends to be more focused on chemistry aspects and more experimentally driven, in terms of glass synthesis and characterization. And then there’s another group of people coming from the physics community, who are very interested in the physics of the glassy state, but they aren’t necessarily connected to people from the material science community. So as a consequence, oftentimes people in the glass physics community don’t understand the relevant glass chemistries or how their models can be translated into practical applications. At the same time, people coming from the more experimental material science community don’t necessarily know that there are these new advanced models available for them to try out and to use. So anything we can do to open the doors of communication between those communities, I think benefits everyone.

Bryce Meredig: And for organizations that are successful in doing this, and of course you and your team were extremely successful doing this at Corning, what are some of the opportunities, do you see? What is the benefit at the end of the road for these multi-disciplinary teams that are able to apply informatics successfully?

John Mauro: I think it’s going to be essential in order to be commercially successful, going into the ensuing decades. If you have two companies, and one of them has a very data driven culture where they place emphasis on the collection of high quality data, and making use of that high quality data, and that company is competing against a company with perhaps an older mentality, where maybe they don’t even bother to collect data or they don’t make use of these data driven approaches, It’s clear that the company that has the data driven culture is going to be so much more effective at developing better products, developing them faster, developing them with lower cost. It’s just going to be essential for companies to be competitive and to survive in the long term.

Bryce Meredig: I think a lot of people might argue that it’s intuitively reasonable that these data driven methods, capabilities like AI and machine learning, should be strong competitive differentiators for a materials company. But imagine a company’s on the fence, or is hesitant to adopt these new technologies to make the investments in people and software necessary to deploy these capabilities. What, if any, approaches have you seen be successful in terms of making the case? Obviously, a lot of companies are driven by, for example, metrics like return on investment. Are there things you would point to as evidence that these methods deliver on the promise?

John Mauro: I think the approach that we’re always competing against, the old school approach, is the method of developing materials by the so called cook and look approach, where scientists who have built decades worth of experience have some intuitive knowledge about where to begin with their exploration of new materials designs.

Bryce Meredig: And that intuitive knowledge is quite good.

John Mauro: Yeah, historically it’s been successful. But what will happen is, they’ll create one material or a small set of them. They’ll measure the properties, see what direction they need to go in, in order to achieve the target properties, and kind of chase the target until they reach the target property values. But what inevitably happens, and a lot of the scientists get so frustrated by this, is that the targets change. You think that you’re targeting a certain set of properties, but maybe the customers just didn’t understand what they really needed, or maybe they have some new knowledge that leads them to have some different targets in mind.

John Mauro: Inevitably, the targets will change. If you take this cook and look approach, you’re basically starting over again every time that happens. But if you take the time up front to build machine learning based models, based on all of the data that you have available and design experiments to maximize the amount of information that you can get for enhancing these models, then, no matter how the targets change, you’re always prepared. This is one of the biggest arguments that I would have, based on my own experiences. That, when you take the time to build these models, no matter how the targets evolve, you’re prepared to come back right away with what the optimized solution should be. That is just so much more efficient compared to the older method of just chasing individual targets, one at a time.

Bryce Meredig: This idea of increased responsiveness to a very rapidly changing customer landscape makes a lot of sense. Especially in the industries that you are serving, like consumer electronics, with the products you were developing at Corning, where the product cycles are so fast. Consumer preferences are changing and the bar is increasing so quickly, you don’t have any choice but to adapt quickly or get left behind.

John Mauro: Exactly. The product life cycles now are so short compared to what they used to be. It used to take years to development a new glass product. Now it has to be compressed onto the same cycle as smartphone manufacturers. So this is every one to two, to three years. You need to have something new with some substantial improvement compared to older generation products.

Bryce Meredig: I’d like to go to back to this idea of, what makes a team successful when they apply machine learning and data driven approaches to materials R&D. We mentioned already the importance of good communication, the importance of interdisciplinarity. Are there other factors that you would point to, based on your experience, that separate teams that are successful with materials informatics from those who aren’t?

John Mauro: I think you need to have a culture that values team successes over individual successes. I’ve seen it both ways, and if you’re in an environment that promotes individuals over the team, then there’s incentive to not share your data or not share your best ideas. But if you’re in an environment that values the team success, then we’re all in this together. We either succeed together, or we fail together. That’s the type of culture that promotes the sharing of information and ultimately leads to a more successful project. Honestly … When we talk about barriers to success, it’s really these cultural barriers and sort of soft skills that I see as the bigger differentiator compared to anything in a technical sense.

Bryce Meredig: It’s interesting because so often we want to focus on the technology and the progress of the technology. But it’s these human factors that can, in many cases, make the difference between success and failure.

John Mauro: Exactly. So, if companies want to prepare themselves for the future, they really need to emphasize these cultural aspects. Also, from a talent pipelining point of view, I now hire people who are able to be the bridge between the experimental sciences and theoretical and data driven sciences. Because the reality is, there aren’t that many people now who are fluent in both sides, and can serve as kind of a bridge between these two worlds.

Bryce Meredig: That’s a great segue into the question of skill gaps in training. You’re somebody who has a very unique perspective of this, having been on the hiring side for many years, on the industrial side for many years, and now you’re tasked with training the next generation of glass scientists. What are some of the skill gaps you see exist, and how can we best address them? Especially at the university level, where you are now.

John Mauro: I would say the biggest skill gaps that I see are bridging across these traditional boundaries. For example, in my group here at Penn State, my goal is for the group to be half theoretical and computational, and half experimental. Right now, it’s more like one-third and two-thirds. I want all of the students, regardless of where the greatest interests are, whether it’s on the theoretical side or the experimental side … I want all of them to at least gain exposure to the other side and be able to understand the terminology, to understand how to apply these techniques, to understand the strengths and weaknesses of different techniques. Even on the theoretical side, there’s so many different type of modeling approaches that are available.

John Mauro: From purely empirical models through models that incorporate detailed fundamental physics. The fact is, that in any practical problem that I have seen, it’s never just one approach that solves the problem. It’s not just an experimental approach or just a theoretical approach. It’s never only molecular dynamics that will solve the problem. It’s some combination of approaches that gives you insights into different aspects of the problems. It’s not until you put all of those aspects together that come from many different types of approaches. It’s only then that you get a deep comprehensive understanding of the problem. It’s based on that, that you can come up with the optimized solution for the problem at hand.

Bryce Meredig: What do you think we should be doing from a curriculum perspective? Are you approaching your courses through this lens as well?

John Mauro: I’m trying to do that. I’m still finding my way around on the teaching side, learning about all the different courses that are offered here. Let me take, for example, my kinetics class. I’m teaching the grad students in this required course, which is called Kinetics of Materials Processes. What I’m trying to do there, is emphasize all these different ways one can approach kinetics, because there are some students who are primarily theory or computationally driven students, others who are more into material synthesis, others that are into characterization. Some people are more focused on bulk properties, some people are more focused on surface properties. I’m trying to design the course to give a little bit of something for everyone. There are parts of the course that are really mathematically intensive. There are parts that are focusing on characterization techniques. We cover all types of materials, from metals and polymers to ceramics, glasses, composites, and so on.

John Mauro: I’m trying to have a little bit of something for everyone. The other aspect that I’m trying to incorporate into this, is on the side of communication skills. The number one thing that the grad students wish that we would spend more time on, is in developing writing skills. I’m incorporating a term paper into the course, and they’ll have the opportunity to turn it in early to get feedback on both the content as well as the writing style. I just have to make sure that I’m able to accommodate this for 60 students, but as long as they get the drafts in early enough, it should be okay.

John Mauro: The fact is, there’s all these different approaches to material science and to kinetics, specifically for this class. I want them to be able to understand these different approaches and see how different ways of thinking about things can inform them about different aspects of a particular problem. Hopefully it will be useful for them in their thesis research and then in their subsequent careers.

Bryce Meredig: If your students are anything like the PhD students in materials I’ve interacted with, they would really value the fact that you come from this industry background and are in a position to say, here are the skills that you’ll need to be successful in your career. Not just today, but over the next 10 or 20 years.

John Mauro: Right. I just finished teaching the undergrads my introduction to glass science class. I was really delighted by the enthusiasm and the passion that they had towards glass. They were particularly interested in my experience working in industry. In fact, just two weeks ago I devoted a whole class period to having a career day. To kind of give them some introduction about my thoughts about working in industry, but mostly to spend the time answering whatever questions they had. I just opened it up to ask them to ask me whatever questions that they had. They’re mostly seniors, so they’re in the position now to be thinking about what they want to do with their future careers. If they want to go on to graduate school or find a job directly in industry or the government labs. There were a lot of good questions, and it’s a great bunch of kids. I’ve been so impressed with the students here, both at the undergraduate level and the graduate level.

Bryce Meredig: Thinking about your own career trajectory, obviously the transition from industry to academia is relatively uncommon. Although, of course, it’s the background that my own advisor, Chris Wolverton, had going from Ford to Northwestern. Do you think that’s something that more future academics ought to consider? Spending time in industry to gain that background before going back to academia?

John Mauro: I think it is helpful, actually, because most of the students that we graduate are going to end up with jobs in industry. If you have professors with industrial experience, I think they can understand more about what industry is looking for in their graduates. I’m trying to focus on that. I think it is really helpful to have that industrial background. That being said, it is difficult for people from industry to get the track record that they need to move into an academic role, because …

Bryce Meredig: In terms of publications?

John Mauro: Exactly. Because when universities hire faculty, they’re looking for a really strong track record of research as evidence and the track record of publications. Most companies simply don’t emphasize that. Of course, in a company, the primary motivation is ultimately to make money by developing new commercial products or services. That has to be the primary focus for companies. The goals that a company has are different from the goals that a university has in mind. If somebody in industry has in mind that they may want to make the change to academia in the future, it’s really important for them to pay attention to not only what is delivering value to the company, but what is setting up their track record to make them suitable for hiring into a faculty position.

Bryce Meredig: I’d like to make sure that we have a chance to at least touch on your current research interests. How have your research interests and your portfolio evolved from when you were at Corning to now running your own group at Penn State?

John Mauro: Well, at Corning I started off in the modeling and simulation department and spent the first two-thirds of my career there. Then I moved into an experimental department called Glass Research and spent the last third of my career there. When I see these gaps between the experimental world and the modeling world, that’s one thing that I’m trying to address in my research group here.

John Mauro: In setting up my group to be half experimental and half modeling, I’m trying to address that gap but also trying to encompass the breadth of interests that I have in different aspects of glass science. I think that is largely a reflection of my previous career at Corning, because there I was very interested in computational modeling, analytical modeling, glass compositional designs, and if you looked across the various projects that I have here, I think that they reflect that breadth.

John Mauro: I think building a group that has that breadth is also good for attracting top students into the group, because some people are interested in designing new chemistries. Other people are more interested in characterization, other people are more interested in the modeling. No matter what it is, if they’re a really strong student, I can find a place for them in this group.

Bryce Meredig: Looking ahead to the next, let’s say, five years of glass science and informatics capabilities applied to glass, what are you excited about? What do you think is coming up for the community that will really change how we view things?

John Mauro: I think one of the biggest challenges is how to leverage all the data that’s already out there, and the literature. I think that there’s so much that we can benefit from that. The current state of affairs is not so great. There’s a professor from St. Petersburg, Oleg Mazurin, who is one of the first people to compile databases of glass compositions and properties. He did this decades ago, and initially these were published in paper books. Later, he developed the first database software called SciGlass. This has been an amazing tool, and really puts us in a good position from the point of view of data science. However, there have been some issues with keeping the database up to date, with having it accessible for purchase around the world. At this point, people aren’t even sure who owns the database. I think that one of the things I’m most excited about is also one of the biggest challenges for us in this community right now, and that is getting this world of data that’s out there into some organized format where people can really make use of it.

Bryce Meredig: John, I think that’s all the time we have for today. I just want to thank you again for taking the time to be our guest here on DataLab. It was a pleasure to have that conversation with you.

John Mauro: Yeah, likewise Bryce. Hope you have a great day.