Notice: We are in the process of migrating Oral History Interview metadata to this new version of our website.
During this migration, the following fields associated with interviews may be incomplete: Institutions, Additional Persons, and Subjects. Our Browse Subjects feature is also affected by this migration.
Please contact [email protected] with any feedback.
Credit: Tim Lee
This transcript may not be quoted, reproduced or redistributed in whole or in part by any means except with the written permission of the American Institute of Physics.
This transcript is based on a tape-recorded interview deposited at the Center for History of Physics of the American Institute of Physics. The AIP's interviews have generally been transcribed from tape, edited by the interviewer for clarity, and then further edited by the interviewee. If this interview is important to you, you should consult earlier versions of the transcript or listen to the original tape. For many interviews, the AIP retains substantial files with further information about the interviewee and the interview itself. Please contact us for information about accessing these materials.
Please bear in mind that: 1) This material is a transcript of the spoken word rather than a literary product; 2) An interview must be read with the awareness that different people's memories about an event will often differ, and that memories can change with time for many reasons including subsequent experiences, interactions with others, and one's feelings about an event. Disclaimer: This transcript was scanned from a typescript, introducing occasional spelling errors. The original typescript is available.
In footnotes or endnotes please cite AIP interviews like this:
Interview of Renata Wentzcovitch by David Zierler on September 17, 2020,
Niels Bohr Library & Archives, American Institute of Physics,
College Park, MD USA,
For multiple citations, "AIP" is the preferred abbreviation for the location.
Interview with Renata Wentzcovitch, professor of Applied Physics and Applied Mathematics and Earth and Environmental Sciences at Columbia University. Wentzcovitch recounts her childhood in Brazil, and she describes how her grandfather sparked her interest in science early on. She describes her education at the University of São Paulo’s Institute of Physics where she developed an interest in density functional theory. Wentzcovitch discusses her interest in pursuing a graduate degree in the United States, and her decision to attend UC Berkeley and study under the direction of Marvin Cohen. She describes her thesis research on pseudopotential plane-wave codes and super-hard materials such as boron nitride and diamonds. Wentzcovitch explains the impact of High Tc Superconductivity on both her career and the field generally, and she describes her postdoctoral research with joint appointments at Brookhaven and Stony Brook on evolving electronic wavefunctions via classical dynamics. She discusses her subsequent work with Volker Henie at Cambridge to study silicate perovskite, which in turn led to her first faculty appointment at the University of Minnesota. Wentzcovitch describes the importance of Minnesota’s Supercomputing Institute for her research, and she explains how her research focused more centrally on geophysics and the thermo-elasticity of minerals and their aggregates. She describes the founding of the Virtual Laboratory for Earth and Planetary Materials and explains her decision to join the faculty at Columbia and her involvement with VLab and the study of exchange-correlation functionals to address electronic interactions. At the end of the interview, Wentzcovitch discusses her current work on developing codes for thermodynamic computations and seismic tomography, and she conveys the value of pursuing international collaborations to fit her broad and diverse research agenda.
OK, this is David Zierler, Oral Historian for the American Institute of Physics. It is on September 17th, 2020. I'm so happy to be here with Professor Renata M. Wentzcovitch. Renata, thank you so much for being with me today.
Oh, thank you. It's an honor.
OK. So to start, would you please tell me your title and institutional affiliation.
I am a professor in two departments at Columbia University. Applied Physics and Applied Mathematics and Earth and Environmental Sciences. I have a 50% appointment in each department.
Now, did you have that joint appointment from the beginning, since you joined Columbia?
Yes. From the very beginning. That was the term of my appointment.
And what are some of the advantages in terms of your research in having that dual appointment?
Oh, it's history. I came across geophysics in a very spontaneous way. I will tell you more later, but it happened in my career's early stages after developing techniques to simulate materials and extreme conditions. Later on, when I started looking for good problems to address using those novel techniques, I came across planetary materials, Earth materials. I could simulate very complex materials at high pressures and temperatures. Their structure and chemistry are quite complicated, high-Tc cuprates level of structural and, to some extent, electronic complexity. First-principles methods could not address that level of structural intricacy under pressure before the development of those techniques.
So I came across an entirely new field with new possibilities. I couldn't turn my back, and I started investigating mineral phases and interacting with geophysicists. But it was just pursuing good and untouched problems that I came across this field, mineral physics problems and geophysics. So a great deal of my career has been devoted to studying minerals and materials at extreme conditions. I developed several techniques to study materials at high temperatures and pressures. And then, my research started focusing on geophysics-related problems.
Now, do you have two labs that are located within each department? Or do you have one lab that has a different focus relevant to each department?
I have two labs on two different campuses, but my research today focuses on Earth and planetary materials. Now we are going way beyond pressures and temperatures routinely achieved in laboratories. We are motivated by the discovery of exoplanets, figuring out their internal makeup, structure, and dynamics.
And you have graduate students from both departments as well.
Yes, graduate students from both departments.
So it sounds like it might be a 50/50 appointment, but more like 150 percent responsibility.
That's about right.
[Laughs] Well, Renata, let's take it all the way back to the beginning. Let's go back to Brazil and first tell me about your parents. Tell me a little bit about them and where they're from.
OK. I have sweet parents. I am getting a little emotional here because my mother is unwell, OK?
Both parents were educators.
I'm sorry. Is she with you in New York?
No. All my family is in Brazil.
It must be hard that you can't be there right now.
I just returned from there two weeks ago. So both parents were educators. They both had very rigid educations themselves. My father almost became a priest. He quit the seminary two years before finishing theology. He had studied philosophy and theology. And then, he decided to get a real university degree to become a professor in classics. He used to teach Latin and Greek and then Portuguese and Portuguese literature.
My mother was an elementary school teacher, and she was my teacher when in elementary school. And she evolved with us as we grew up. She went back to get a college degree and became a high school teacher and a school principal and then a school administrator, a regional school administrator before retiring. That is the type of career my parents had. They were both public servants in the state school system. I never paid for my education in Brazil. My father was a state-wide officer in the public education system in São Paulo state when he retired. So I grew up exposed to classical culture, educated to be studious, in a quiet and stable home. Teaching and learning was the rule in the house for kids and parents.
Now, your parents are both Brazilian born?
And the name Wentzcovitch? Is that your family name or that's a married name?
No. That was an addition to my name later in life when I was a university student in Brazil. Both my parents are Italian descendants. My father also has Spanish on his side, my mother also has French on her side, and my father has really traditional Portuguese ancestry. His surname is Martins, Spanish, and Matosinho, a well-known Portuguese name from Brazilian colonial times.
Where did you grow up? Where in Brazil?
I grew up in the suburbs of São Paulo in a town called Santo André. It is an industrial area surrounding São Paulo city. I was born in Campinas in the countryside, near Sumaré, where my Italian grandfather lived.
And what is the school system like in Brazil? Is there a public school private school divide? Do you have that there?
When I was growing up, public schools were the most common, even though my father taught in a private college at that time, a Jesuit college. But yes, I attended public elementary school, junior high, high school, and university. But by the time I was in high school, private colleges were becoming more popular. Today, private colleges, universities, and schools are equally popular, if not more.
Was the church a big part of your family and your upbringing?
Yes, sure. In the early stages of life, yes. My parents are practicing Catholics. My mother studied in a Catholic boarding school taught by nuns. But at the same time, we were exposed very early to the pros- and cons- of organized religion. We were always allowed to question. As a result, our attitude toward religion evolved and broadened.
Renata, when did you start to get interested in science and the natural world? Was that early on?
My Italian grandfather from Venice (by the way, I'm Italian also because I inherited citizenship from him) was a businessman. My family always went to Sumaré in the countryside of São Paulo to spend Summers. He was retired and spent a lot of time talking to me, his first grandchild. He wanted me to be a businesswoman and follow in his footsteps. One game we used to play was oral algebra. He used to give me simple oral tests to test my thinking speed. He used to ask me to do simple calculations mentally, but I had to respond very quickly. He gave me confidence that I was very good at math. That was one thing we used to do together.
But math for business. He thought you would apply those math skills for business.
Math for business. For example, how much is one-third of ninety, plus two times five, minus 20? I had to respond instantaneously. So he gave me a great deal of confidence. Later on, when I was in junior high, he gave me a book, The Universe, by Isaac Asimov. He asked me to read it and discuss each chapter with him. That was wonderful.
That is how I became interested in science. I could have become interested in the classics earlier, which I did later. Classical Greek and Latin cultures, but that was later. I could have gotten interested in that area earlier if I had spent that kind of quality time with my father, but he worked so hard. I barely saw him when I was growing up. My grandfather sparked my interest earlier, but later on, I became interested in my parents' topics of interest. My mother got a degree in psychology, so I developed an interest in psychology when I saw her studying this subject. But this was later after I was already in the physics track.
Renata, in some ways, it sounds like you grew up in a pretty traditional household and culture, and I'm curious—as a young girl and woman with an interest in science—were you generally encouraged that this would be a field to pursue? Or were you ever discouraged that this was not something that would be appropriate for you?
Oh, no. My parents encouraged me to be serious and work hard. So, whatever I chose, it was good. If it made me happy, it was perfect for them. No problem. Not at all. They just wanted me to be dedicated and accomplished. They saw I could become a physics teacher, which pleased them very much.
And in school, were you—
Actually, one person discouraged me from studying physics. When I was in high school, I had a doctor who said it would be very hard, and I should do architecture or something more applied and blended with arts. He was the only person who told me that physics was not a good idea.
And so already by high school, you knew you wanted to pursue physics.
Not really. I was probably in the second year of high school when I decided. Before high school, I thought I wanted to be an aerospace engineer. I was very impressed by the Apollo program, and I went to see an air show with my uncle. I was so impressed by fighter jets, an F-15 I think, when I saw it for the first time. So, I got interested in becoming an aerospace engineer. Because of that event, I decided to do two high schools at the same time. At the end of the first year of the traditional high school, I applied for a competitive technical high school in the neighboring town, São Bernardo, to speed up my studies to become an engineer. The specialization was "Tools Design", and I attended it in the evening. It was considered an introduction to mechanical engineering. The local auto-industry workers attended that particular technical school in the evenings. I lived in this region of São Paulo, the ABC region, where the auto industry flourished. Luiz Inácio Lula da Silva, the former president, was an auto industry union leader there. So in the evenings, my colleagues were auto-industry workers, and in the morning, the regular high school kids. I did that for two years.
Then, when I was in the second year, I read Einstein's autobiography. It changed my whole perspective. Besides, I had been exposed to technical subjects, which clarified that I liked more abstract and less technical topics. Both made me think physics was a more exciting subject. In the third year, I decided to become a physicist and prepare for the general university entrance examination, the Vestibular, as we call it in Brazil. My first option was the Physics Department at the University of São Paulo. I passed and started studying Physics from the very first year at the university.
But did you know—even with physics—that you wanted to stay on the applied side of things?
No, I didn't want to stay on the applied side. By then, I was interested in astrophysics; remember Asimov's book? So I was interested in astrophysics, and Einstein's biography raised my interest in cosmology. So that pulled me towards astrophysics, cosmology, Einstein, physics; these things were coming together, shaping up my preferences.
Was the department of physics at the University of São Paulo, was it a strong department?
It's not a department. It is an institute of physics. It has several departments. It is very strong, indeed.
An institute. What would be the equivalent of something like that in the United States, or is it really a unique kind of set up?
It's a school. It is almost equivalent to a college of engineering in the US, I would say. It has several departments, for example, materials and mechanics, experimental physics, theoretical physics, nuclear physics. Each department has its chair and faculty.
Given that you were interested in so many different physics branches and given how big the institute was, how did you go about defining the things you wanted to focus on?
I attended courses on various subjects and worked in different labs. In the third year, I got a fellowship from São Paulo State Foundation, FAPESP, to study and research astrophysics. In the second year, I was taking courses in astrophysics already. I approached a professor, Sueli Aldrovandi, in the Institute of Astronomy and Geophysics and said I wanted to specialize in astrophysics and start doing research, if possible. She asked me to read an introductory book, Astrophysics and Stellar Astronomy by Swihart, and discuss it with then an Assistant Professor, Ruth Grünwald. So that's how it started. I was in the second year. In the third year, I got a FAPESP fellowship.
Nevertheless, I continued taking traditional physics courses, including special relativity and nuclear physics. In the fourth year, I thought I wanted perhaps something more concrete. I wanted to do something more tangible. I also appreciated engineering, practical and concrete accomplishments, and I was missing that. I thought perhaps experimental optical spectroscopy would be nice. I was fascinated with quantum physics then. Besides, my astrophysics project consisted of analyzing the optical spectra of quasars, which we didn't understand much.
So, I decided to try to work in an optical spectroscopy lab in my fourth year. It was intriguing physics that could be applied to astrophysics as well, and it felt more accessible to me. I obtained another fellowship from the National Nuclear Energy Council, CNEN, to work with an experimental laser physicist. So, that was my fourth year, right before I started the MSc degree. But my interest kept shifting, mostly after I learned that it was possible to do quantum mechanical calculations of atomic and molecular spectra. The late seventies and early eighties were the early days of electronic structure calculations based on density functional theory. Those computations, which were also very technical, allowed me to understand much better experimental observations. It seemed terrific, and it was within my reach.
Was it a lab, or was it a course that turned you on to this field?
After one year with the fantastic CNEN fellowship, I started looking for an MSc degree advisor. That is when I found my mentor, José Roberto Leite, a computational condensed matter physicist. He had been a post-doctoral fellow at IBM's Yorktown Heights lab. He was doing semiconductor physics and developing electronic structure methods in the late seventies.
This is very advanced stuff.
Very advanced stuff at that time, indeed. Numerical, computational, it was very appealing because we were solving the Schrodinger equation numerically, and he was calculating spectra of molecules and doing very early calculations in solids. He and his former advisor, Luiz Gimarães Ferreira (affectionally Guima), had their method. They were very early Brazilian pioneers in the field, and José Roberto adopted me, I would say. His example and level of support ultimately shaped my interest and career.
Mostly because of him, and to work with him.
Yes, I worked with him and learned about electronic structure methods development. So you see, that was a significant shift in trajectory. There was a great deal of wondering before I entered the path I am on still today. I had the opportunity to try different topics. But José Roberto effectively opened doors for me. He opened my eyes and my mind to possibilities outside Brazil, and that was wonderful. He was a wonderful mentor, indeed. That is when I got interested in doing a PhD in the US.
Was his advice that going to the United States would be the best thing for you?
Who knows what best would have been? But I would have more options if I studied abroad.
Right. But I guess my question is there wasn't necessarily the sense that the best education would be had in the United States. It wasn't necessarily that assumption.
I suppose the best experience to become a successful physicist would involve studying abroad, have international experience and contacts, build relationships, learn other ways and approaches, and have examples from successful professionals in the field. So, yes. It was his suggestion and my endless interest in learning new things.
What schools did you apply to?
I applied to three schools. Caltech, UC Berkeley, and MIT. I was accepted at UC Berkeley.
Did you know about Marvin Cohen beforehand?
Of course! I knew about Marvin. José Roberto Leite and Cylon Gonçalves da Silva, a professor at the University of Campinas, organized the first Brazilian School in Semiconductor Physics in Campinas. Marvin Cohen and Leo Falicov, both from UC-Berkeley, were invited to give talks there. That's when I met Marvin. I spoke with Marvin and told him that I was applying to UC Berkeley. Once I learned about pseudopotentials, an exceedingly interesting and useful idea Marvin pioneered, I knew I wanted to work with him. That was it.
You clicked with him right away?
Yes. He seemed to be an inspiring physicist.
Given that you came in with a master's degree, how much course work did you have to do at Berkeley?
I wanted to go through the coursework, again, because it was different. Solid-state physics for me was an undergraduate course. In Berkeley, it was condensed matter physics.
So in Brazil, it would have been called solid-state still at that point. Condensed matter was not in use.
Yes. The textbooks were different. Condensed matter physics in Berkeley didn't really follow a textbook. We followed Marvin's and Steve Louie's lecture notes. We also used Charles Kittel's Quantum Theory of Solids. Marvin and Steve published a book with their lecture notes in 2016, Fundamentals of Condensed Matter Physics. So it was a very different course than the classic solid-state physics course. In Brazil, I took one semester only. I also wanted to take statistical physics, electromagnetism. I had taken an undergraduate version of electromagnetism. In Berkeley, I took the graduate version. I also took Group Theory with Leo Falicov, and I wanted to be sure I was well trained in all aspects relevant to my research. My MSc degree in Brazil gave me great confidence that I could do original research and was more than ready to do it.
Renata, on the cultural side, had you been to the United States before? Was this your first time?
No. It was my first time.
And how was your English?
Not good. [Laughs] Not good. I used to speak Spanish and French at that time. I could read and communicate in English, but I couldn't write well. One of the first things I did was to take a course in English writing, which was extremely helpful. I didn't complete the course because the workload was hefty. It was a course for journalists and professional writers. It was more than I asked for, but the first half of the semester was very helpful.
What were your impressions of Berkeley when you first got there?
Tradition, tradition, tradition. I could feel the weight of history shaping the present. I was surprised because Berkeley had the reputation of being at the forefront of progressive social movements, but it felt a bit conservative. Indeed, I came from an ultra-liberal environment. UC Berkeley has Nobel Prize winners, folks who made history in physics. Whenever the record is so long, strong, dominant, and contemporary at the same time, people want to keep things as they have been. So, it was maybe a bit rigid for me. But, I grew up in a traditional and disciplined household. So I knew the benefits of rigorous training, and I was enjoying the process. But I was accustomed to being more independent.
When I joined Marvin's group, I started doing things that he did not ask me to do. He then asked me to focus and work with him and not to diverge. But I thought I was ready to be independent—I had previous research experience. I was confident. I recall I set up a problem to determine the energy levels on a super-lattice starting built from two semiconductors with band offsets. A visiting professor in the department suggested the problem. I solved the problem analytically. Marvin was not researching this topic and was surprised. He said, “From now on, let's work together.” And that is how it went. But he later understood me and gave me more freedom. He suggested problems, outstanding problems, and let me run with them. I felt I had something to prove. So that's how my PhD went.
And how closely was your dissertation related to his research?
Oh, absolutely close to his work.
And what were some of the big questions that were propelling both his research and your dissertation?
It was at that point that I got involved in high-pressure research. Those were the early days of predictive ab initio calculations at high pressures. We were still testing its predictive power. Some of the best electronic structure methods and codes were available in his group—a pseudopotential-plane-wave code.
He, Steve Louie, other students, and I started exploring the predictive power of ab initio pseudopotential calculations on structural properties under pressure on materials such as first row compounds, e.g., diamond and BN. Besides, Marvin collaborated with a geophysicist, Raymond Jeanloz, also a high-pressure experimentalist who was very influential in my career. The study of materials under pressure was in its infancy then, even though Marvin was already making predictions of superconductivity in silicon under pressure.
I was one of the first students to study first row compounds such as carbon, boron, and nitrogen. My thesis consisted primarily of studying super-hard materials such as boron nitride and diamond, their layered version, graphite, and layered boron nitrides, and some novel carbon-boron-nitrogen alloys under pressure. I also studied some curious boron-based semiconductors, one of which seemed to have reversed iconicity. That was my PhD thesis. I investigated the electronic structure of these semiconductors and insulators under pressure, their structural properties, and the trends throughout this series of compounds under pressure. I also studied phase transition under pressure, e.g., graphite to diamond and similar ones in boron nitride. My thesis was all about materials behavior under pressure and phase transitions. That is when I learned the limitations of ab initio methods to address high-pressure materials problems.
That was my research with Marvin. Raymond Jeanloz used diamond anvil cells to study materials properties under pressure and synthesize novel materials. Geophysics and the need to understand material properties in the Earth's deep interior gave great impetus to high-pressure science and technology. It is not an accident I started contributing to this field later.
And Renata, it sounds like you had a pretty good appreciation more broadly of some of the bigger trends in the field in high pressure. Where did you see your dissertation fitting in with sort of the broader direction of the field, where it was going?
My MSc degree experience helped me finish my PhD quickly because I graduated in five years. At that point in Berkeley, theory students were graduating in six and a half years on average. It was clear at that point what was going to happen next. I was very open to new areas of research, as I had always been. My long-term plan was to do outstanding research, innovate, break barriers, and open trails. I aspired to make significant contributions. It was not clear at that point what that would be.
And does that include maybe or maybe not returning to Brazil? Did you know you wanted to sort of set up and make a life for yourself in the states?
At that point, I was living with my husband, Nathan Berkovits. My short-term goals were to get postdocs together, stay together, do well, and things would work out. Brazil has always been in my mind. But I disconnected during my PhD studies, especially after marrying Nathan. Graduate school and postdoc times were very intense, mainly because of the double careers issue. A lot of energy went into juggling professional and personal lives. Besides, my connection with physics in Brazil was restricted to my ex-advisor, Leite, and he passed before I was ready to return.
But there was more. I was also very attracted to opportunities in the US. High Tc superconductivity was discovered in the cuprates two years before I graduated. What a beautiful, strange phenomenon in strongly correlated oxides of which I knew very little. This fantastic problem attracted my interest. Ab initio calculations were not able to address these cuprate materials adequately, even in the normal state. They are strongly correlated oxides with very complex crystal structures and were very challenging to ab initio calculations then—research on strongly correlated materials used mainly analytical models at that time.
I was very attracted to this class of problems and wanted the chance to work on them. Nathan, a superstring theorist, and I started looking for coordinated postdoc positions. His field was popular in that particular year. He had several options. We decided to go to the U of Chicago where I would work with Kathy Levin, a theorist expert in strongly correlated systems. Unfortunately, before we arrived there, the superstring group announced they were leaving for Rutgers University, and I started looking immediately for another option near Rutgers. A fantastic opportunity came by, a collaboration between the Condensed Matter Theory Group at Brookhaven National Laboratory and Stony Brook's Physics Department. My experience at the U of Chicago was brief but very impactful in my research. When I came across another critical and beautiful type of problem, I did not hesitate to address the behavior of strongly correlated iron-bearing minerals under pressure. My research on this topic is well-known today.
The BNL/Stony Brook position was stimulating. The research topic was fascinating and hardcore computational materials physics. In 1985, Roberto Car, today at Princeton University, and Michelle Parinello, today at ETH, proposed a novel ab initio method to perform molecular dynamics MD. It was a form of fictitious dynamics where electronic wavefunctions evolved using classical dynamics. It was an efficient algorithm to perform dynamics, including electrons but mysterious and intriguing, which made it fascinating. It was an impactful and stimulating method. I was very eager to work with it too.
Why? What was so big about it?
In classical MD, ionic motion is treated classically. Until then, MD was carried out with interatomic potentials, not with electronic structure methods. They introduced a form of ab initio MD where the electronic wave function dynamic evolution was also calculated using classical Newtonian dynamics, and that was tricky. It was an algorithm inspired by classical physics, not real physics. That type of dynamics is referred to as fictitious or extended ensemble MD. It is essentially computational physics. These unfamiliar nuances and the possibilities it opened made the research in this area very attractive. Besides, I was eager to return to computational predictive work. After doing predictive studies in graduate school, the experience with analytical models and their parameter-dependent predictions was a little uncomfortable. I was ready to embrace and advance new methods in computational physics. It seemed very worthwhile.
My time at BNL and Stony Brook were one of the happiest times I had. I was back doing predictive calculations, and everything was in place. Nathan moved to Stony Brook as well. The electronic structure group in BNL developed computational methods for electronic structure, and I had the opportunity to do independent research in that position. I decided to scrutinize more closely the Car-Parrinello method and other aspects of classical fictitious MD.
What was the group that you joined at Brookhaven?
Jim Davenport was the group leader, and Phil Allen from Stony Brook was collaborating with that group. They had a grant from the Department of Energy, and I was hired to work on that project. It was then that the dots started connecting for me. Stony Brook's department of geosciences had gotten an NSF Science and Technology Center, the Center for High Pressure Research, CHiPR. Phil Allen from the Physics Department knew the CHiPR leadership well and thought my graduate research in high-pressure calculations fit well in CHiPR's mission.
In one of my first conversations with Phil at BNL, he suggested that I consider extending the Car-Parrinello method by combining it with another kind of dynamics, the Parrinello-Rahman constant pressure MD, another sort of fictitious MD. That was the original fictitious MD. Another name for this type of dynamics is variable-cell-shape MD. In this dynamics, the simulation cell walls vary dynamically under internal stresses or external pressure. Until then, constant pressure MD was performed using only interatomic potentials. There was a possibility of combining both to study materials at high temperatures and high pressures. It was a fantastic project which would allow me to do method development. It would be a more general dynamics that would enable us to simulate phase transitions as well. Soon after that meeting, Phil left for a sabbatical at Los Alamos, and I was left to work independently.
And this was probably also a vote of confidence in you that you can do this on your own.
A vote of confidence in me and the BNL group. He knew I would have support if I needed it. That was the kind of freedom I always aspired to have. Jim Davenport and BNL gave me a vote of confidence and allowed me to research as I wanted.
Was this your first time working in a national lab?
You were never at like, Lawrence Berkeley or SLAC, or anything like that?
I was at UC Berkeley, not at Lawrence Berkeley Lab. I worked in Birge Hall on campus. It was a privilege to work in such an open-minded group at BNL. I had a great deal of freedom to explore ideas. First, I tried to understand fictitious MD. The first was the constant pressure or variable volume MD by Andersen, later extended by Parrinello and Rahman to variable cell shape, VCS. It was clear then that the Car-Parrinello method was just another form of fictitious MD, now introducing electronic degrees of freedom. It was an efficient algorithm to do something conceptually straightforward but computationally very demanding, i.e., MD with interatomic forces calculated with ab initio methods. The challenge was to do with similar efficiency the real calculation with fewer approximations.
For me, the plan was clear: first, I would implement the VCS MD with interatomic potentials, then I would replace the potential with an ab initio code. While developing the first code, I noticed the fictitious constant pressure MD was quite unrealistic when used with a small number of atoms. There was a disturbing symmetry breaking. I was able to reformulate that dynamics and fix that problem. My version of constant pressure MD seemed more physical.
Next, I had to replace the interatomic potentials with an efficient MD code. I realized a colleague I had at UC-Berkely, José Martins, had developed a very efficient pseudopotential code with the same advantages as the Car-Parrinello code. He was an assistant professor at the U of Minnesota then. I called him and asked if he would like to collaborate in putting both codes together and develop a new kind of ab initio MD, a self-consistent one. He was very interested, and we started collaborating. It worked marvelously well.
We were able to bypass several challenges the Car-Parrinello method faced, e.g., implement the Mermin functional, a finite temperature version of density functional theory, DFT. This was important for the simulation of metallic systems with electrons in thermal equilibrium with ions. It was the best way of doing these simulations. Constant pressure or VCS MD also worked very well, and we could study complex systems under pressure and phase transitions. This form of ab initio MD today is called Born-Oppenheimer MDre.
Renata, was anybody outside of Brookhaven working on these issues as well?
Absolutely. John Johnopolous’ group at MIT was trying to accomplish the same thing. He had a postdoc, Mike Payne, from Cambridge University who carried the work to England. He is the principal author of the CASTEP code. In Austria, Georg Kresse and Juergen Hafner developed a similar code, which evolved into the VASP code. But we had a stable and productive code before anyone else. My colleague José Martins was also very successful in proposing a very efficient form of pseudopotential. The Troullier-Martins pseudopotential gave us another edge to these calculations. We started producing significant results right away. We had a compelling technique in our hands. I had a lot going on for me.
This was a good time for you at Brookhaven.
This was a wonderful time.
Did you ever think about making a career—
I think the stars aligned for me at that time.
Yeah. Did you ever think about making a career at Brookhaven? Was that a possibility?
Yes, I thought about it. I wanted to stabilize a little. No, BNL was not a possibility at that time. Also, I was married, and I was not contemplating separation. Superstring theory was no longer a popular field in 1991-92, and there was not much support for that field then. Nathan received another postdoc offer at Kings College in London, and we decided to relocate to England. Phil Allen had connections at the Cavendish Laboratory in Cambridge and I knew Mike Payne, who was also pioneering ab initio MD, as I mentioned before.
Therefore, I went to Cambridge, where I met Volker Heine, the Cavendish professor, and my mentor. So I arrived there with this powerful technique and quite confident. Volker looked at what I was doing and brought to my attention this material, a silicate perovskite, the primary phase of the Earth's mantle that could occupy about 50 percent of the Earth's volume, and nobody knew how it behaved under pressure. Some say it could transition to a cubic phase; others said no, it gets more and more distorted under pressure. So, he suggested that I investigated it. In three months, I had a PRL submitted.
Yeah. This happened fast.
It happened very fast because the technique was spotless, and I was doing the calculations. I had submitted another paper to PRL while in the US applying the VCS-MD to vanadium dioxide, VO2, a controversial oxide. It has a metal-insulator transition at about 70 centigrades. A fascinating material with practical applications. The metal-insulator transition had a crystallographic component too, and for the first time, we could address the structural complexity of the insulating phase. The debate was whether the insulating phase was a strongly correlated Mott insulator or the structural distortion opened a gap, a Peierls-type gap. I did the calculation on VO2 and submitted it to PRLs while still in the US, but that paper was rejected. Too much novelty at the same time probably brought distrust. It is more likely I had several influential competitors that were not happy to see that paper. It was an incredible piece of research at that time.
I arrived in Cambridge with my most important paper rejected. I worked very hard to submit another article quickly. And that was magnesium silicate perovskite, Mg-pv, a large gap insulator and the most abundant phase of the Earth’s mantle. The electronic structure of that material is straightforward. The article was quickly accepted. After the technique was proved to work on a complicated system with a simple electronic structure, I re-submitted the paper on VO2 to PRL, and it was accepted. So, the publication of the first ab initio variable cell shaped MD was delayed a bit by that rejection and my move to England. The VO2 paper is, in fact, my most cited paper to date.
At that point, it became clear I could address complex minerals of Earth's deep interior at extreme conditions of pressure and temperature, PT. I new field for ab initio calculations started right there. That type of work was not possible before. Volker Heine introduced me to a mineral physicist at UCL in London, David Price. He was very influential in my career by opening my eyes to another science domain to make breakthroughs. I found mineral physics and geophysical applications while looking for good problems to apply my techniques. Actually, I wanted to continue developing methods, but given all the moves and all the changes that were going on in my personal life, I was not in a position to do so. Therefore, as long as it was great science, I embraced it, and mineral physics was a remarkable new field for ab initio calculations.
I stopped developing methods and started applying the technique to Earth minerals motivated by geophysics problems. Indeed, Volker Heine has a lot to do with what I do today. He pointed me to an entirely new domain of ab initio applications. David Price welcomed me at UCL and gave me a lot of support to take this research off the ground. It was clear this was the most innovative science I could deliver. I could not turn my back.
Were you interested at all—given the fact that, theoretically at least—there are many commercial and industrial applications for this research, did you ever spend time thinking about those kinds of applications? Or was your sort of mode always basic science and discovery?
Yes. Basic science and discovery. As you see, I am into geophysics now, the doorway to planetary physics. I could see right away the connection to planetary physics overlapping strongly with astronomy, a very dear field of science to me. Regarding applications, I like challenges. Industrial applications are nice too if one is in the right environment and is exposed to good problems. That was not where I was. I was lucky to get exposed to mineral physics in England. From then on, I was interested in promoting the field. I knew the breakthrough I had done and what could follow. There weren't others in a similar position at that time.
Earth and planetary materials were very challenging, and mineral physics problems very complex from an atomistic perspective. Ab initio methods had not seriously addressed them because of those challenges. They have large primitive cells and complex crystal structures, often with low symmetry. They have iron and are strongly correlated oxides too. Besides, one needs to understand their properties at high temperatures, which posed a different kind of challenge. Minerals are fascinating, and one needs to understand their properties at high pressures and temperatures to understand planetary interiors. They offered motivation for several method developments. It was the perfect topic for me.
After two years in England, I moved back to the US. José Martins, my main collaborator during the development of ab initio MD, returned to Portugal, and an opening was announced at the University of Minnesota. I applied for the job and succeeded. It was very exciting to move to a place with a supercomputing institute. I could aspire to do method development again.
And this was your first tenure-line position?
Yes. It was in the Department of Chemical Engineering and Materials Science, where José Martins had had an appointment. I had visited that department twice while collaborating with him. I was known there and had another colleague who knew my work well there, Jim Chelikowsky. So it was a very appealing position.
I need personal relationships to connect, and I knew folks in Minnesota. I felt more comfortable there. I thought I had a connection with that department, a stellar department. Also, the Minnesota Supercomputing Institute was one of the largest academic computing centers in the US. It was a desirable position.
Renata, your home department was chemical engineering and material science, but did you have multiple appointments there as well?
Yes. In Minnesota, I had a zero-time appointment in the School of Physics and Astronomy from the very beginning. Then in Chemical Physics, in Earth Sciences, and in the Scientific Computing program. Most of my students were from Physics.
Yeah. Renata, can you talk a little bit about how your research agenda might have changed when you got to Minnesota? Were you looking to take on new projects? Or was this an opportunity—now that you have an academic appointment, you can continue to expand what you had been working on up to that point?
Yes, there were expectations, of course. For example, I submitted my first proposal to the National Science Foundation to continue developing methods and applying them to these Earth minerals. The proposal was rejected, and I was advised to seek funding for this research in the Earth Sciences program. I applied to that program, and I was funded right away, substantially funded.
Good advice you got.
Yes, but it opened a different path for my research, and I started investigating mineral physics problems right away. My application to the materials theory program was focused on method development, not on applications of methods. However, I teamed up with a geophysicist, Lars Stixrude, and developed close cooperation that allowed me to keep developing simulation methods. Among the properties of interest to geophysics that needed method development were thermodynamics and thermoelasticity. This research, as far as method development was concerned, was welcome in the chemical engineering department. I also collaborated with colleagues such as Jim Chelikowsky, a computational materials physicist, and with the Materials Research Science and Engineering Center, MRSEC, on engineered materials. But indeed, my research started transmuting more and more into geophysical applications.
After being able to tackle the structural complexity of minerals, I shifted focus to thermodynamic properties. MD was the popular method to study these properties at thousands of Kelvin, something I was doing very well. However, some early calculations pointed to the shortcomings of this approach, especially to investigate phase transitions, e.g., hysteresis as in experiments, and the quantum nature of atomic motion even at thousands of Kelvin at high pressures. I abandoned ab initio MD and decided to develop other methods.
I chose another method, discredited at that time, to study thermodynamic properties: the quasiharmonic approximation, or QHA. In this approach, atomic motion is described as a superposition of harmonic phonons, i.e., non-interacting quantized harmonic vibrations. Since the target temperatures were so high, i.e., up to 4,000 K, this approach was very discredited. At these temperatures, the atomic motion was expected to be classical. But nobody had tried it at these temperatures and at high pressures of interest, over one million atmospheres. It was a very successful approach.
So I want to ask on that point, it's so interesting to hear because you're involved in so many different areas, really such a diverse and broad research agenda, how does all of that sort of—I don't wanna say narrow because it doesn't narrow, but how does the focus become geophysics at this point from all of the different areas you were working on that could have gone in any number of different, exciting directions? How does geophysics sort of happen at this time for you?
It was like gravity. I got pulled into it. You contribute, receive positive feedback, develop an affinity, people invite you to collaborate on projects and proposals, etc. It is beautiful science that has evolved, especially with the introduction of materials theory and simulations. It is very challenging because it is not possible to access Earth's deep interior directly, and one needs to be imaginative and be a detective too. I have come so far from the point I started, and I am delighted because I feel I have helped to shape this field.
For example, the quasiharmonic approximation that I adopted, rather than MD, is the standard way to study the thermodynamics properties of materials today. And, of course, we have gone beyond harmonic phonons to study phonon quasi-particles. All I can say is that geophysics poses numerous materials physics, and chemistry challenges. It has been a playground for for the development of several methods for materials simulations.
So maybe part of it is that you had all of this experience and knowledge base in all of these different areas and they all went to inform what you wanted to learn more and explore more about with geophysics.
Exactly. Geophysics allowed me to do new things.
And another appealing aspect was that there were very few people thinking about these computational problems.
And what were those problems, Renata?
Essentially, everything I did was novel.
Can you talk a little bit more about what those problems were at that time as you identified them?
For example, in the beginning, I was trying to study the static compressive behavior of complex minerals at zero Kelvin. But of course, we needed to study the compressive behavior at high temperatures. Therefore, I developed techniques to study materials properties at high temperatures, to calculate thermodynamics properties in a continuum range of pressures and temperatures. Continuum is essential because pressure and temperature derivatives are important in geophysics. It also means computations are more intensive. It is not sufficient to study a mineral phase in one particular situation. Properties need to be investigated in a continuum of pressure, temperature, and often composition. Most mineral phases are alloys.
Once you understand the role of materials physics in geophysics, you see a range of problems that are not typically addressed in materials science or physics. The broad aim is to provide structural, thermoelastic, and equilibrium thermodynamics properties of minerals and mineral aggregates, a.k.a. rocks. The word equilibrium here is very important, with brutal implications for materials simulations. It is a highly interdisciplinary field. It is a humbling experience and one must learn to interact with scientists in other fields, for example, seismology, geodynamics, and geochemistry.
Most information we have about the Earth's interior comes from seismology. Seismic waves can be used to develop tomography images of the Earth's interior, mapping its internal structure. We see low and high seismic velocity regions and velocity discontinuities pointing to depth- or angle-dependent state changes. That is how we know the Earth's core is part liquid and part solid, and the mantle is layered, etc. But there are much more subtle features still not interpreted. They are related to the internal dynamics of the Earth, but there is more too.
It is a geophysicist's job to figure out the origin of these suble velocity structures. One needs to know the necessary materials properties, i.e., velocities vs. pressure, temperature, composition, etc. Thermoelasticity and acoustic velocities of minerals and their aggregates—I mean, rocks—are of first order importance in geophysics. I invested a great deal of time and effort in this topic. Once I had results before anyone else, it was very tempting to interpret these features myself. That meant I started transmuting into a hybrid materials-geo-physicist.
Structural features revealed by seismic tomography are also related to the internal dynamics of in Earth. Fluid dynamic simulations of mantle convection also use thermodynamics properties of minerals. Therefore, seismology, mineral/materials physics, and geodynamics are complementary and intertwin in a fundamental way in geophysics. That is how I started collaborating with geodynamicists and seismologists. I had to learn how to speak their very different languages, an experience familiar to a foreigner or an immigrant. But, you learn many things in the process. For example, that despite your lack of knowledge of their fields, your contribution is fundamental and very much appreciated. Everybody grows together, and real progress is made. It is a good sign that you have the right minds and souls working together, keeping their eyes on the prize. Common interests overlook the differences. It has been a tremendous and transformative experience.
So Renata, because this research is so multi-disciplinary and because you came to it from many multidisciplines of yourself, can you talk a little bit about within these larger collaborations, what did you bring to the table? What was your sort of area of expertise or talents as they fit with all of the people that you were working with?
I developed several techniques to address materials properties that are essential for understanding Earth's internal state. In particular, I was able to calculate thermodynamic and thermoelastic properties of minerals of the Earth's lower mantle, the region between 660 and 2,900 kilometer depth. Temperature and pressures in this region are still challenging to experiments with, up to 4,000 K and over one million atmospheres. Once these properties became available, we could start interpreting velocity patterns from this region, and geodynamicists could improve the realism of their simulations using these newly calculated properties.
Let me give you a concrete example. The average chemical composition of the lower mantle is still a matter of debate. There are three main phases in this region: bridgmanite, a magnesium silicate, ferropericlase, a magnesium oxide, and a small amount of a calcium silicate. Bridgmanite is the most abundant phase, and the main question has been the relative abundance of ferropericlase. There are arguments in favor of a low abundance and other arguments in favor of a normal abundance, meaning the same abundance as in the upper mantle, which is well known. Once, as I was able to compute the seismic velocities of these phases at lower mantle conditions, I could give an approximate answer to this question, which favored a composition more like the upper mantle. Some experimental data was available at low temperatures. High-temperature data were too sparse to allow for extrapolations to lower mantle conditions. It is still the case today.
Today we are even addressing this question but from a much more sophisticated perspective, for example, 3D variations in the lower mantle composition. All this by computing shear and compressional velocities of rocks versus pressure, temperature, and chemical composition. There are profound implications for mantle dynamics and Earth’s evolutionary history.
Renata, can you talk about the founding of the Virtual Laboratory for Earth and Planetary Materials? How did that come about?
It was around 2003-04. I had two significant papers coming out. One on the thermoelastic properties of bridgmanite and the composition of the lower mantle. The second one on the identification of the post-perovskite phase. It was clear that significant progress was being made. However, the calculations were extremely demanding. At that time, thermoelasticity calculations required an enormous number of calculations and, therefore, were very impractical. I did the calculations on bridgmanite during a sabbatical in Trieste. My collaborators and I were submitting and babysitting code executions for nearly three months to get results. We did it because it was a very important problem, but clearly, these calculations were very impractical.
The second important problem was on the identification of the post-perovskite phase, a polymorph of bridgmanite produced under pressure. Motohiko Murakami and Kei Hirose in Tokyo found a change in X-ray diffraction in bridgmanite under pressure, a phase change, at 2,500 Kelvin and 125 Gigapascals, conditions typical of the deep lower mantle. This mantle region is still quite mysterious, but there is a layer 250 km thinck above the core mantle boundary, CMB, whose origin was, and the details still are not understood. This phase transition observed in Hirose's Lab could be responsible for this lawer, a new mineral with new properties producing a layer with different velocities. Kei Hirose kindly sent me his X-ray diffraction before the phase was identified, and we started trying to determine the crystal structure of that phase.
There were three Japanese postdocs working in my group at that time, Koichiro Umemoto, and Jun and Taku Tsuchiya, all established in geophysics in Japan today. We all got very excited. It was around Thanksgiving, and I remember not sleeping thinking about how bridgmanite's perovskite structure would change under pressure. I had some insights on a possible transformation mechanism into a layered structure, and that was odd. The Tsuchiyas were actually doing the calculations, searching for the new phase using variable cell shape molecular dynamics, and found the layered structure of the post-perovkite phase that could explain the X-ray diffraction. Umemoto showed the possible transformation mechanism from perovskite to the layered post-perovskite structure.
We computed the thermodynamic phase boundary for the first time and quite reluctantly because we were still trying to assess the uncertainties of ab initio predictions. Indeed all pieces of evidence pointed out to a phase transformation at the right depth in the mantle. The D" layer, the ~300 km thick layer above the CMB was possibly a region where post-perrovskite was the dominant phase instead of bridgmanite. It was indeed a big splash. We submitted the paper to Science, and it was rejected with much skepticism. The community did not accept this finding.
Why not? What was the disconnect there, do you think?
The community was unfamiliar with quantum materials simulations. I had two important findings in one year using techniques foreign to the community. But there were more serious reasons for skepticism too. First, that region of the Earth does not consist only of bridgmanite in the pure form we were studying. Bridgnanite is a complex solid-solution whose phase transformation properties can be very different. But the evidence we were presenting was very compelling. One must admit there was, at least, a lot of coincidence. We had to test whether the changes in seismic velocities were similar to those observed across the D" discontinuity.
Again we need to undertake those tremendously time-consuming calculations of thermoelastic properties, but the magnitude of the problem justified the effort, and we did it. The results were even more impressive and genuinely unbelievable. It was irrefutable evidence that the D" discontinuity in some places was being produced by this phase change in bridgmanite. It was clear that to contribute seriously to geophysics. We needed to step up our game in materials simulations. We needed to do calculations in more realistic solid-solutions, and that was also very challenging.
Here you see the motivation behind the VLab development. Several kinds of complexities in the study of minerals and rocks require intensive high-throughput calculation. The number of calculations required in these studies was getting out of hand. We had to do better and automate the calculations. It was an ideal motivation for injecting computer science into materials simulations. We wanted to create a computational laboratory for high pressures mineral physics. The goal was to enable complex high-throughput calculations to be performed easily online through a portal accessible by non-experts. An ambitious goal and an excellent idea for technology development. We wanted our applications to focus on minerals, but the technology was applicable to all materials. The submission of our post-perovskite paper was delayed by at least one month because I was writing the Vlab proposal. It was a very interdisciplinary proposal and was very well-reviewed.
I published two influential papers while my grant was under review. My article on the perovskite phase transition was delayed because I was writing the Vlab proposal. Still, it was a timely proposal to write for a major grant. While experimental high-pressure labs were popular, high-pressure computational labs did not exist. This one was devoted entirely to understanding minerals of the Earth's interior, but it would help develop other fields as well. We need novel developments in condensed matter physics, e.g., predictive methods to compute thermal conductivity on complex mineral phases, better exchange-correlation functionals to improve the accuracy of phase relations predictions at high pressures and temperatures, etc. And we needed to inject computational science into these calculations to facilitate them. I was told it was the top reviewed proposal in the ITR (Information and Technology Research) program that year.
How was the field changing at this point, Renata? How were the questions changing? How was the technology changing, and in turn, what were some of the theories that might have been relevant for your work as well?
The field was changing quickly. We were in 2004 when the Vlab was created. Still, since the early nineties, I was seeing the changes that were going to happen, or better, how I was going to study materials at high pressures and temperatures differently. I knew that I was not going to be doing molecular dynamics. I was going to do quasiharmonic calculations. Ab initio phonon calculations became possible in the mid-nineties using density functional perturbation theory, a technique developed by Italian physicists Stehano Baroni, Stefano de Gironcoli, and Paolo Giannozzi. I needed those ab initio phonon dispersions as an input for my quasiharmonic calculations of thermoelastic properties.
This was an essential step to advancing the field. Therefore, we established a collaboration. That is how I got involved with the Quantum ESPRESSO project as well. I got their codes, and I also included my Born-Oppenheimer MD with variable cell shape in one of the Quantum ESPRESSO modules. It was an excellent collaboration. In 1999, I published my first paper on thermoelasticity in Science. It took a couple of years to publish a second paper on this topic, the thermoelasticity of bridgmanite, and two more years to publish the thermoelastic properties of post-perovskite. That is how difficult it was to do these calculations, and geophysicists needed results on hundreds of phases with variable compositions. We need to enable easier calculations.
Another significant challenge I started confronting in 2004 was the spin transition problem in iron in lower mantle phases. This phenomenon was observed in ferropericlase and bridgmanite in 2003 and 2004. These minerals form solid-solutions with FeO, a prototypical strongly correlated material familiar to condensed matter physicists. Essentially, the magnetic moment of iron collapses when these phases are compressed to approximately 50 Gigapascals, a pressure typical of the Earth's lower mantle. I decided to confront this problem because a new ab initio technique to investigate strongly correlated systems was being developed by some of my collaborators in Trieste, Stefano de Gironcoli, and Matteo Cococcioni. I started exploring this technique even before it was published.
Until then, this technique had a free parameter, the Hubbard U, but they introduced a technique based on linear response to obtain an ab initio U. At that point we did not know how successful the technique was and how important this spin transition phenomenon was going to be in geophysics, but it could have fundamental consequences. Now I think we cannot fully understand the Earth's mantle without understanding spin transitions in lower mantle phases for several reasons. The calculations were extraordinarily successful, but more than calculations were needed to understand this phenomenon.
To understand any phase change, one needs to do thermodynamic calculations and the thermodynamics of the spin transition is very peculiar. I had an opportunity to develop a theory of this phenomenon at high pressures and temperatures, which makes exotic predictions that have been confirmed. There is a subtle volume change associated with the change in the magnetic moment of iron which has enigmatic but dramatic consequences for seismic velocities. It is enigmatic because the phase change happens in a wide pressure range and is only detectable if one knows what to look for. It is dramatic because it affects seismic velocities throughout the deep lower mantle, and it is not possible to interpret seismic tomography images without considering these changes. It also affects the mantle dynamics. It is a purely quantum phenomenon manifesting in tomography images and affecting geophysical processes literally on a global scale.
You can see the iron spin crossover signal throughout the lower mantle in seismic tomography images if you know what to look for and look carefully.
And it sounds, Renata, like Minnesota is really one of the most exciting places to be to do all of this.
I would say the Minnesota Supercomputing Institute, MSI, enabled this science and stimulated me to push computations further and further. In that spirit, I developed methods for thermoelasticity.
Because computers allowed you to do that.
Right. And I did not have to write lengthy proposals to get computer allocation. I could get sufficient computer time for one year with half a page request as long as I was productive. It was a very favorable situation.
Let's talk about the switch to Columbia. How did that come about for you?
Well, I was getting more involved with geophysics, and I welcomed closer interactions with geophysicists too. I wanted to be able to influence the field more directly. Of course, the research was not easy to do in a chemical engineering department. I could attract postdocs very easily to work in this field, but students were more reluctant. Postdocs were more experienced and were looking for new areas of research with a future.
Also, I met my current husband in 2006, Tommy Vaughan, a professor at the University of Minnesota. He is an MRI scientist and was being recruited by Columbia University in 2015. For that reason, I started thinking about moving to New York. Once there was an opportunity for him to move, I began contacting Columbia people, and the opportunity opened for me. He moved in 2016 while I was still exploring possibilities here. I moved in 2017. So, again, I explored this opportunity because of the personal side of my life.
It was always like walking with two feet. Personal and professional aspects of life had to be balanced.
The two-body problem, as we say, right?
Yes. I always had that.
Always. It sounds like it.
Yeah. Did you take graduate students with you? Was that part of the consideration?
At that point, my students in Minnesota were almost graduating. One graduated a couple of months after I left. Another visited Columbia for one year and then graduated. One postdoc moved to Columbia with me.
And in what ways, Renata, to have the joint appointment, right—now you're really recognized in the field, you have connections all over—in what ways does this new kind of appointment take your research to the next level as you're envisioning what a move to Columbia might look like?
I have appointments in two departments now. When I arrived at Columbia, it felt like drinking from a fire hose.
Because everybody wanted some time with you. Yeah.
I also wanted to talk to everybody. I needed to establish a new group, and there were many possible new directions. My basis of operation is in the Applied Physics and Applied Mathematics Department, APAM. That is where I teach and recruit my students from. We established an MSc program focused on Materials Theory and Simulations. I have successfully recruited PhD students through this program. Students do some research under my guidance before they apply for the PhD program. They learn about what I do while working on their MSc degree. Some students who do well and want to continue to do a PhD in APAM. Others want to work with me in the Department of Earth and Environmental Sciences, DEES. I organize a seminar series in geophysics, “Cool Topics in Geophysics,” whose theme varies every semester or year. This year we are discussing the lower mantle. Next year I plan to discuss the deep Earth geological water. It is open for credit to students, to participants from other institutions, and to the participation of DEES faculty members, postdocs, and scientists.
In what ways does being at Columbia—between being in New York and having this multiple appointments—in what ways has that further broadened your research, generally?
It has not broadened. It has focused.
Yes. I am more focused on Earth science related problems now.
I am also recruiting students more easily now. Being in APAM, I have access to a pool of students that is well-trained to join me. I can also attract excellent postdocs, which I always did in Minnesota as well. So, I am doing my research and recruiting more efficiently. I am embracing more challenging geophysics and mineral physics problems. The remaining mineral physics problems are becoming more demanding.
Can you talk a little bit more, Renata, about Vlab?
Yes. Vlab was conceptually very ambitious because we wanted ab initio calculations to be performed more efficiently by non-experts. In other words, I wanted to go online, enter a simplified input for a complex sequence of calculations on a portal, submit it for execution in a supercomputing center, which today would be a cloud, and get the final results displayed in graphic form. We needed to develop a great deal of technology to achieve that. We needed to create workflows and codes for online calculations. At that time, I had an outstanding Brazilian postdoc, Cesar da Silva, who was a physicist and a computer scientist, a very creative mind. He led the technical developments. The most challenging part was to execute the codes online. Clouds were not common in 2004.
We developed the necessary Java codes to run these high-throughput ab initio codes online at the Minnesota Supercomputing Institute. Our medium-term goal was to create a master workflow for calculations of thermoelastic properties, which is vital in mineral physics. It used to take at least three months to perform these calculations by a team of three well-trained scientists. I could not ask a graduate student to do these calculations. We wanted students to be able to do these calculations in a couple of days only. We finally achieved that goal in 2008. Two of my students got their PhDs working together with the developers and doing these calculations. This computational infrastructure, i.e., software plus databases of results, was the virtual laboratory.
There were other components to this interdisciplinary project. The heart of ab initio calculations is still the development of exchange-correlation functionals to address electronic interactions. The predictive power of these calculations depends on the accuracy of the treatment of electronic interactions. Don Truhlar, a quantum chemist at the University of Minnesota and an expert on developing these functionals, was also part of this project. Phil Allen, a condensed matter physicist from Stony Brook University and an expert in charge and heat transport, another essential type of mineral physics problem, was also involved in the project. These properties are fundamental to understanding heat flow in the Earth.
There were several components to this virtual laboratory, but the laboratory's heart was computational technology. We achieved a lot in four years. Thermoelasticity calculations were submitted from a portal and were executed in heterogeneous environments, i.e., concurrently in multiple computers at the MSI. Of course, our focus was on mineral physics calculations. It is a much smaller community than the materials science community, and we did not have an audience, a consumer basis, or an organization to back us up. So, the idea of a computational community infrastructure did not gain broad community support.
Nevertheless, it was an ambitious and successful project because we achieved our goals. Several groups in the materials community embraced the idea of community infrastructure for ab initio calculations, e.g., the Materials Project in the US, the Nomad project in Europe, among others. These are large, organized community projects. Of course, computational technology was evolving quickly, and personnel developing software in my group were leaving. Software development requires more consistent and broad community support. My current students have been reinventing these codes since I came to Columbia. Many of the techniques I developed while in Minnesota and never published are being released now. It has been fun to upgrade not only codes but also the methods.
I think we accomplished much and planted the idea of computational infrastructure for public online materials calculations and databases of results. Since 2010 my developments have been supported by my single-PI grants, not by a group grant.
Renata, I am curious whether specifically with Vlab or some of your other collaborations—particularly with your work on planet modeling—in what ways is your research generally important for climate change science?
[Laughs] I wish it could be more important for climate science.
In other words, is that specifically a motivating interest for you?
No. I am interested in solid planets, especially the newly discovered exoplanets we know very little about, almost nothing. I study solids and solid planets. I don't see my research contributing to climate science, which is primarily fluid dynamics. Earth's atmosphere was produced by degassing Earth's interior, but that is as far as my research touches on this topic.
On the other hand, the first terrestrial exoplanets larger than Earth, super-Earths, were discovered between 2003-2006. We started right then, trying to understand their solid mantles. It was a new line of research that is one of the main motivations for my studies today. We had identified the post-perovskite phase in 2004. We just asked the question, what happens with post-perovskite if the pressures increases. In super-Earths, the CMB pressure can reach twenty times that pressure in the Earth. The question to answer is, what are the solid phases that form these planets?
We have made a great deal of progress in discovering several phases. Some of them have been confirmed experimentally in low-pressure analogs, i.e., phases with different chemical compositions that display similar pressure-induced behavior at much lower pressures. So this is another exciting aspect of my research. Discovery of new solid phases, crystal structure discovery. The post-perovskite phase was the first one that triggered a great deal of development in materials discovery. It was a pivotal point in materials discovery that brought greater credibility to computational mineral physics.
Mineral physicists and geoscientists were unaware of the possibilities and innovations ab initio calculations could inject into the field. There was a great deal of suspicion of ab initio results. If others cannot measure and confirm the properties you calculate, the results are viewed with suspicion. But once we predicted a new phase as crucial as the post-perovskite phase, the calculations grained much credibility. The community became more receptive and started hiring ab initio practitioners. Stony Brook University and UC Berkeley created positions in this field in the Earth Science departments. Of course, my colleague David Price at UCL London started a group in computational mineral physics very early, right after I left in 1994. He opened my eyes to this field, and I showed him what was possible to do. Our interaction was mutually influential. Geosciences at UCL has several ab initio practitioners in the faculty today. So what was the question? I think I drifted away.
We've been talking about so much. We started with climate change, and then we went on to exoplanets.
Yes. I just wanted to make another point about the computational discovery of the post-perovskite. Of course, it was an experimental discovery. We discovered the structure that explained x-ray diffraction, which Hirose's group in Tokyo also published independently. Our publication of an independent paper simultaneously did not go well with them at that time. We submitted our paper to Science with a lot more content than they had. I little competition developed at that time, and I think that is another reason our paper was rejected. Later we started collaborating again, and we still are. My three Japanese postdocs returned to Japan and are established professors or scientists working in this field today. This work was beneficial to their careers as well.
And Renata, just to bring the narrative up to the present, what are some of the projects you're working on now?
I am now publishing several codes and techniques I developed in the past. Workflows now are more practical, and most people can use them independently. Before, they were not available for independent use. Before, they were used for online computations through a portal only. We are publishing several codes for computations of thermodynamic and related properties and workflows for automated high-throughput calculations. At Columbia I have students who are proficient in scientific computing as well. I am also interacting with geochemists interested in the geological water cycle here at Columbia. There are challenging questions about water in the Earth's interior, the origin of oceans, and why there is plate tectonics and convection in the Earth's solid mantle and not in the other terrestrial solar planets today. We believe the incorporation of water in minerals in the Earth's interior might be the answer. It must decrease the viscosity of rocks. Understanding the behavior of minerals incorporating water in minor or significant form is a very challenging topic for materials simulations, which I started studying at the University of Minnesota.
I am working with seismologists much more closely now, especially trying to detect the effect of spin crossover in iron in seismic tomography images. It is an essential phenomenon to interpreting these images. It is another of those discoveries by computational mineral physics that is very foreign to the geoscience community. So we are developing a database of thermoelastic properties incorporating effects of the spin crossover in several minerals to interpret seismic tomography. I have fantastic seismologist colleagues at Columbia and at Princeton who are interested in this problem as well. As a team, we came together to look at these seismic tomography image details and interpret them using the velocities I calculate. So, my research is much more in geophysics now.
The discovery of new solid phases at extreme conditions of super-Earth-type exoplanets is going on very strongly now. We are discovering new solid phases and modeling the internal structure of super-Earths with up to 20 Earth masses in collaboration with geodynamicists. The nature of these solid phases and phase transitions determine the structure and dynamic state of these planets in the first order. I am opening a new scientific trail here. I am always looking for opportunities to do something new and significant.
Renata, I want to ask—before we get to the last part of our talk—I want to ask now in what ways are the fundamental questions you're asking now the same as those you've been asking all the way since your undergraduate education and in what ways are they different now?
Yes. I think you can see my early interest in astronomy and astrophysics, being expressed in my research today. It is true, but the circle has not closed. I am moving in that direction naturally. It is fascinating to research exoplanets, planetary systems being discovered, the relationship between these planets and their stars, etc. Of course, this work brings me closer to the subject that once was very stimulating to me, which motivated my studies. So the circle is not closed, but I am getting much closer to astronomy and astrophysics, my original interests.
Well, Renata, I want to ask you now a few sort of broadly retrospective questions about your career.
The first is supercomputing and computational power, of course, has been central to your research for a long time, and over the course of your career, computational power has increased exponentially, right?
So I wonder if you can talk about what impact supercomputing has had and what are you able to do today that maybe even five, 10, 15 years ago might have been inconceivable?
Being part of a supercomputing institute, I contacted several computational scientists, including a geodynamicist, Dave Yuen, with whom I talked a lot about pushing for more extensive computations. I developed a highly demanding numerical method to compute thermoelastic properties and seismic velocities because it was a computational challenge. Still, of course, it was also a fundamental materials problem in geophysics. I probably would have done something else if I did not have access to MSI’s computational resources. I had to explore to the fullest the possibilities around me.
This was in the early 2000s. Of course, computational power has increased a lot since then. The computations were very demanding. That was one reason we wanted to develop the Vlab, to have all calculations automated and executed concurrently to take full advantage of computational power. Since then, this type of calculation has evolved in the opposite direction. We developed a semi-analytical version of this fully numerical method that decreased the required computational power by at least two orders of magnitude. It is much easier and faster to do these calculations today.
For example, in ten years, we did thermoelasticity calculations on four fundamental phases of the mantle using the fully numerical method. It was an impractical calculation because it was too demanding. With the semi-analytical method, we can do in a few hours what in 2000 took several months. We made significant progress by developing a smarter method and scaling back the computations. Recently, we have been doing a different and very demanding type of computation. In 2014 we introduced an ab initio method to compute unharmonic phonons dispersions and unharmonic thermodynamic properties. The new method involves a combination of MD and phonon calculations. We calculate phonon quasiparticle properties by analyzing MD trajectories. MD can be computationally very intensive, and we need to perform hundreds of MD runs to calculate these properties.
And it seems like we're getting closer and closer to quantum computing. Is that something that's personally exciting for you if we get to that stage?
I am not the one who is going to develop quantum computing codes, I believe. The computers are not there yet, a tremendous amount of code development is needed, and I am not willing to shift my research focus. I could collaborate with someone interested in developing quantum computing codes, the quantum computing version of the techniques we use, but at the moment, I am not seeing that coming. And that will be branching out too much at the moment. I do not see the rewards there yet. I still have a lot to explore in GPU computing before quantum computing.
Renata, another thing I wanted to ask about is your emphasis on international collaborations, in Asia and in Europe, why is this important to you? What are some of the values in pursuing these broad scale collaborations outside of the United States?
They just fit my research goals and are a lot of fun. As I said, I am Italian also. Therefore collaborations in Italy were very natural and a great joy to explore another side of my identity. And the most advanced codes to calculate phonons in the nineties were being developed in Trieste. So this was very natural, almost a necessary event for me. From 1998 until 2006, I had a one-month summer job in Trieste every year. It was a long and productive collaboration, and it not over. I see it reigniting shortly.
The collaborations in Japan started because I received that X-ray diffraction of the post-perovskite phase from Kei Hirose. He contacted me through a mutual friend, Dave Yuen, who was visiting Tokyo. Besides, working with Japanese folks was something familiar to me. I grew up in São Paulo, which has the largest Japanese community outside Japan. I studied and played with Japanese kids in school and had Japanese colleagues at the university. They were second-generation Japanese. I had very positive experiences with Japanese friends in Brazil, translating into positive collaborations with Japanese scientists when the opportunity was presented. They made the overture.
Of course, the basis of our collaboration was our common scientific interests. I have a formal position as a principal investigator at the Earth-Life Science Institute at the Tokyo Institute of Technology. Kei Hirose was the founding director. It is a major international research center in Japan and unique in the world. My collaborations in China started with my first postdoc, Wenhui Duan from Tsinghua University and continued through his students. He is a member of the Chinese Academy of Sciences today.
And Renata, I can't help but ask, there's more that you've worked on than there isn't in physics, right? I mean, you've just been involved in so much. And so I want to ask you a very broad question. Again, going back to when you first started asking these questions at a high level—even in college—and that is over the course of your career—no matter all the different collaborations and labs and areas that you've worked on—what are the scientific concepts in physics that stay with you closely? That you bring with you to all of your projects and inform how you see the world, how you wanna set up an experiment, the kinds of people that you wanna work with. What are those physics concepts or theories or ways of looking at science that inform and infuse all of your work?
I am a solid-state physicist, a condensed matter physicist. I think about solids in a quantum way. On the other hand, there is a quantum-classical crossover, which I am very aware of and I need to address in my work. My condensed matter physics training enables me to distinguish when I need the quantum result or accept the classical one. In mineral physics applications, we frequently cross the quantum/classical borderline. My understanding of strong electronic correlation enabled me to bring something new, fresh, and essential to Earth sciences and materials physics. One needs to investigate spin transitions at high pressures and temperatures to grasp the phenomenon. It is quantum thermodynamics or statistical mechanics.
So the conceptual frame of mind is that of a solid-state physicist and will always be. Of course, I appreciate technology and engineering new methods, new codes, and the beauty of developing something concrete that others can use to produce useful results. I have an appreciation for the engineering aspect of this research. Right now, there is a lot of that going on in my group. It is a lot of fun too.
On the other hand, the great questions that drive my research and I address are problems in geophysics and planetary sciences. The combination of all these aspects of my research—I think—makes me inimitable.
Absolutely. I've never heard of somebody that has exactly your portfolio. It is unique.
Yes. It is unique because I contributed techniques that enabled new areas of research and solved significant simulation problems. Because I developed such methods, I was there first and able to solve significant mineral physics problems. But there are others now that followed on the same trail. I am happy to say that postdocs that have worked with me almost all are working in this area today. Once someone is exposed to this field, they do not go back. It is an interdisciplinary field. It has fundamental problems in condensed matter, chemistry, planetary sciences, and scientific computing, and always something new to test.
But the issues are quite complex, which translates into challenging problems in scientific computing. There is a lot of work to be done and professional opportunities if people do well. I have numerous postdocs that returned to China and work in this field today, except for the first one, Wenhui Duan. He returned to China and started sending his graduate students to work with me as postdocs and undergraduate students as graduate students. They have all returned to China as mineral physicists or material physicists researching mineral physics too. Six former postdocs and students are professors there. China is doing well in this field because they were all well-trained physicists first. There is a great deal of critical mass there.
Well, Renata, for my last question—it might be a dangerous one because there's so much that you can say, but I have to ask—
I will try to be brief. Did you get the answers you wanted from me?
My training in solid-state physics is essential to addressing mineral physics problems.
I also like to develop methods because I have an appreciation for engineering. I wanted to be an aerospace engineer, remember?
That's right. Take it back to the beginning. So looking to the future—because you could go anywhere you want, literally, there's so much that you could do—and yet resources are so tight. Time, money, availability, the people that you're able to work with—inevitably, those limiting factors will force you to make decisions on the things that you can work on and the things that you can't.
Limitations always inspire creativity too. I have had to deal with different kinds of challenges, like many people. But I always saw new doors that I could walkthrough.
So on that note, what are those doors, metaphorically speaking, that you want to walk through, both in terms of the kind of research that you feel is most fundamental and impactful and the kind of research that simply you personally find most exciting and intellectually stimulating?
I think research on planetary materials is exceedingly interesting and challenging. No question about that. I am walking a careful line to extend myself solidly into Earth Sciences. My group has expanded quickly since to Columbia. I brought interdisciplinary grants and I have the opportunity to expand even further. There is a lot or work to be done in this field and I am compelled to expand. I needed to retain talent to build an enduring group and to develop collaborations with other colleagues that bring a wealth of problems.
Of course, we need to have a cohesive unit in more thatn one way. First, we need a group of materials oriented scientists able to tackle different aspects mineral physics computations, as I once started building in the VLab. Second, we need to integrate further materials computations, seismology, and geodynamics into a coherehnt modeling field where all there aspects of a global problem are addressed simultaneously.
Hopefully we will realistically model complex planets from early stages, in the case of Earth, to the planet we know today. If we can predict how Earth evolved from a molten aggregate of primitive meteorites than hopefully we wil be able to predict the internal state of a wealth of other planets being discoverend today. Geophysics depends to a great extend on materials theory and simulations to understand how planets evolved from a complex ball of magma, in the case of terrestrial planets, to become chemically stratified and dynamic bodies we know. Sometimes they don't seem too different from far away. Just look at case of Venus and Earth. They are both terrestrial and somewhat similar in size. Nevertheless, they evolved into very different planets. Small differences in chemisty or initial state can settle planets in very different evolutionary paths.
Hopefully, one day our research will start from a ball of magma and predict the way a planet evolves. This is a complex geo-materials chemistry problem and, of course, a material simulation problem. But that's what I would like to understand better, planets, the reslting internal structure and dynamic state. I am sure I will come across challenging materials simulations problems along the way. There are complex materials problems to be addressed, but look where we started. I started at the point where we were trying to do electronic stratucture calculations in molecules, and look where we are today. So, why not think about modeling these very complex systems? To be predictive, this modeling will have to be based on sophisticated ab initio calculations.
Of course there is no limit to the modeling. There are very different kinds of planets, for example, the outer planets, Jupiter, Saturn, etc., that some colleagues started modeling. And why not stars? But I think stars may be more boring and there is a lot more than chemical reaction going on inside them. If I were starting today, plasma physics in astrophysics would be very attractive research topic for me.
Well, Renata, I'm excited just listening to all the things that you're going to contribute to the future. It's been so fun speaking with you today. I'm so happy—
You are very kind, David.
I'm so happy that we were able to connect and this will be a wonderful addition to our collection because your research agenda is really, truly unique and it's very special to hear you talk about it.
It is. Yes. Thank you, David.