Michel Devoret

Zierler:

This is David Zierler, oral historian for the American Institute of Physics. It is April 7th, 2021. I am delighted to be here with Professor Michel H. Devoret. Michel, it is great to see you. Thank you so much for joining me.

Devoret:

My pleasure. Thank you for organizing this interview.

Zierler:

Michel, to start, would you please tell me your title and institutional affiliation?

Devoret:

My official title is Frederick W. Beinecke Professor of Applied Physics at Yale University, where I have been teaching and doing research for the last two decades.

Zierler:

Michel, two questions associated with your title. First, who was or is Frederick W. Beinecke?

Devoret:

He was an engineer, entrepreneur and philanthropist. With his brother Edwin, he donated a rare book and manuscript library to Yale. My understanding is that Frederick’s contribution to Yale’s endowment is also funding a few professorships like mine.

Zierler:

Applied Physics, now the term "applied physics" means different things in different departments and also in different generations. What does applied physics mean at Yale, and how well does that encapsulate your area of research?

Devoret:

It's interesting that you mention there could be different meanings for the name of my department. The Applied Physics Department at Yale has indeed a rather complex history attached to it. First of all, I would like to say that prior to being at Yale, I have worked for 25 years at the Atomic Energy Commission center in Saclay, France, which is a government-sponsored research organization. The higher administration there had a tendency to insist that we physicists must concentrate on applied things. When I moved permanently to the US and became a professor at Yale in the Applied Physics Department, one of the many paradoxical twists in my new environment was that, all of a sudden, I felt completely free from the pressure of doing something applied. [laugh] The Applied Physics Department at Yale is historically the second physics department of that university. There is a Physics Department at Yale and an Applied Physics Department like in several other universities in the US, but the difference boils down pretty much to the emphasis on condensed matter physics in one department, and particle physics in the other. The story I was told was that Applied Physics at Yale was founded in the '60s by dissidents from the Physics Department who thought that they were too much under the domination of particle physicists and nuclear physicists. But in terms of how much fundamental physics we all work on, it's essentially the same in both departments.

Another nickname for the kind of physics we do in our department is low-energy physics, [laugh] to mark the contrast with high-energy physics, the other official name for particle and nuclear physics. I like this nickname because it has a sort of funny double meaning. It also expresses the fact that the physics we are doing doesn't involve large facilities. In high energy physics, the tools of choices are kilometer-wide accelerators that propel particles for collision experiments that break them into more basic constituents. We deal in our department with ordinary matter consisting of mundane particles like electrons and atoms, and our experiments tend to be more of the tabletop kind. But I think the phenomena that we probe are no less interesting than the phenomena of high-energy physics. You know, the big words for this aspect of our discussion are "emergent phenomena." You may not work on the very fundamental particles of nature like quarks and gluons, or their fundamental interactions, but fortunately, there is a complete separation between the still to be discovered extraordinary properties of ordinary matter at our common millimeter scale, and the properties of these particles, which operate at a much more reduced scale like a femtometer (10^-15m) and below. What we learn of the behavior of elementary particles does not change at all our understanding of, say, what happens when water boils.

Zierler:

Michel, this is a little bit outside of your field, but today is potentially a monumental day in physics with the reporting coming out of Fermilab and the muon experiment right now. Only because I'm talking to you today, it would wonderful just to have a snapshot moment for what it might mean across the board for physics if we find things beyond the standard model. What might you have to offer about that?

Devoret:

Well, you see, that's quite related to what I was talking about with emergence and the decoupling of physics laws operating at widely differing energy scales. So if somehow the standard model had to be corrected as a result of the kind of experiment we are learning about today, this would have essentially no incidence at all on phenomena playing out in our everyday experiments in the lab, like superconductivity. Better, let's talk about something even more basic like the competition existing between the vapor, liquid and solid phases of water. Well, our understanding of these phenomena would not change even if there's a major revision to the standard model in the wake of today’s announcement.

Zierler:

Michel, a question we're all dealing with, this year-plus in the pandemic, how has it affected your science for better or worse?

Devoret:

It has terribly affected the kind of science we have been doing because the social-distancing measures made it extremely difficult to train new students, particularly in an area like ours that depends on nanofabrication and very delicate lithography techniques involving a clean room. These techniques, and others that we use for measurements, absolutely require shoulder-to-shoulder supervision. Beyond our particular case, in every experiment in physics and probably in all highly technical enterprises, people need to very closely collaborate on the details of the apparatus they build. Many best practice moves are not readily transmissible at the explicit level, in a set of disincarnated instructions. Our experiments are more knowledge-intensive that they are work-intensive, and if you want to educate a new student, you have to be there with him or her in person for many hours. You cannot train people you do not know well at a distance, without first agreeing on a common language and a set of norms. So, the pandemic has complicated our life enormously at the scientific level. I would say that even if you want to do theory, which is most of the time blackboard work, it is quite difficult to remotely, despite the tools that are readily available over the internet. These internet tools can work only when you know the person on the screen so well that you can predict in advance their reactions. Then, you have a pretty good idea of what they are going to understand in what you're telling them, and what they won't understand. This way, you can somehow preempt the problems in this kind of remote conversation. Like in many universities, teaching at Yale went on entirely remotely in Spring 2020. So, with the students I had already worked with, we can have beneficial discussions. For the new students, I think it's very difficult. Most probably, it's also difficult for students to follow classes. They feel less engaged. That said, and despite all these difficulties, I should say that I feel very fortunate to have a job which can be done remotely, or at least which has a version of it that can be done remotely. Thus, I have been quite busy during this pandemic. There were a lot of meetings to attend. I even created a new course with another professor at Yale, which was entitled Cinema and Physics.

Zierler:

Oh! [laugh] That's fun.

Devoret:

Yes, it's a fun course. And we were very surprised, and a little bit stunned even, when we saw that the enrollment for the course was larger to that of "Differential Equations"….

Zierler:

[laugh] Right. Not surprising.

Devoret:

Yes, but problematic in the long run.

Zierler:

Michel, let's take it all the way back to the beginning. Let's go to France and let's start first with your parents. Tell me about them.

Devoret:

Ok. My mother was a schoolteacher at kindergarten level and my father was a biologist. He had been trained as a medical doctor and became a geneticist shortly after graduate school. He had taken only a few physics classes at the university, but he was very enthusiastic about physics, so that was probably an influence. Also, because he had temporarily worked as an occupational hazard physician in a physics lab, he had two good friends who were professional physicists. That gave me the impression that physicists were people that were in general fun to be around with, a feeling that I continue to have till this day.

One of the many things I owe to my parents is that they encouraged and supported the career that I chose. They were always willing to examine and discuss with me my career orientations. But as a college student, I didn't know right away that I would embark on a physics career, even though I had a fondness for this particular science and the type of people doing it. I vividly remember when came the crucial time to choose a graduate program. Artificial intelligence and neuroscience were very attractive to me intellectually, but I finally decided not to go into these fields. Nevertheless, I was, and remain, very interested in computers and electronics.

To explain the difficulties of the choice I was facing, I have to mention that physicists as a community were depressed at that moment -- or at least it is what they seemed to be when they were talking to prospective graduate students. It was the mid-'70s, and physicists, who had done great things during the 20 years following World War II, had the feeling their science was no longer what it used to be, and they were predicting doom to new students. They were saying that, well, pretty much all interesting subjects had been understood. Biology was the science of the future, they said. In my opinion, 45 years later, it is still the science of the future!

And I did actually try to study some biology. I took a serious course in biochemistry, but I never liked it very much in its everyday practice. Intellectually, biology as a whole is fascinating, but experiments in biology do not have the richness of physics experiments in terms of immediate sensory experience. Physics experiments are made of a huge variety of little parts like screws, springs, wires, transistors, resistors etc... They also have an incredible variety and sophistication in their instruments, like oscilloscopes or network analyzers. A physics lab, at least an experimental physics lab, is very rich in all kinds of basic supplies. In a biology lab, by contrast, it seemed to me, you had only a limited type of various transparent chemicals, test tubes and, as sophisticated equipment, centrifuges. It was a rather austere experimental environment, compared to those of physics labs I had visited. There were not that many things you had to do manually very skillfully, it seemed. Unlike physics which was well anchored in the childhood experience of mechanics and electricity, like moped repair, it was a rather abstract activity. The day-to-day practical, hands-on work seemed even less varied and fun, in my superficial opinion, than making a meal at home in your kitchen. [laugh]

Zierler:

[laugh]

Devoret:

To continue with the sensory experiences of physics, if you have to connect two pieces of copper wires with a soldering iron, there's that wonderful smell of the rosin which acts as a flux, to remove the oxide skin forming on the metal surface. There's a lot of manual labor joys that comes with working in a physics lab. Even now, I continue to feel very happy in the lab, even though unfortunately, I tend to work most of the time at a computer desk.

Zierler:

This is meaningful, making the culinary distinction, coming from a Frenchman.

Devoret:

Yes! By the way, I like to cook and these days, cooking has become my main experimental activity! To continue the story of the choice of my graduate school, I figured in the end that the important thing was not so much the subject, but to have fun during my PhD work, even though, as the mood seemed to indicate in the '70s, I was joining a science that was about to die. More importantly, by joining an established physics lab, it was clear that I was going to receive a genuinely excellent graduate education in science. There were a few famous physics labs I was having access to. On the other hand, I didn't see, either in biology or artificial intelligence, the kind of high-quality environment and mentoring I would receive in physics.

And you know what happened in the end? This sense of doom was totally artificial. Physics rebounded from its temporary lapse and there were a lot of exciting discoveries I witnessed. I like to tell this story because I find many students are often too much concerned about the future of the subfield they are joining. And now, since it is my turn to be the old guy who advises students, I have vowed I will never proclaim that physics is on the decline. I think it's just as an exciting science that it has been in the last 2,000 years. Just being old does not mean you are necessarily dying, you know.

Zierler:

[laugh] Michel, tell me about your parents' experience during World War II. Where were they?

Devoret:

On that topic, I have to mention that Pierre Goltman, a cousin of my mother and an Auschwitz survivor who wrote about his experience, once surprised everyone in a family lunch reunion when he said that I was owing my existence to Adolf Hitler. [laugh] Of course, it was provocative, but at the same time, that statement had some deep truth in it. World War II was, for many Frenchmen, a struggle. But it was in particular hard for my parents' families, because of their Jewish background, even though they were not at all religious. We would say now they were only from Jewish ancestry, but Hitler had generalized the branding of scapegoats and was an obsessive perfectionist. As French citizens, the situation of my parents was probably not as bad as if they had been native of the Netherlands, because the survival probability of Jews in the Netherlands during WWII was only 5 percent, whereas in France it turned out that it was more or less 75 percent, which says something about France relative to the Netherlands. As I think of it, while evoking the number of deaths in my family during WWII, both on my mother's side and my father's side, this grim statistics seems more or less to apply overall to my extended family. And you can imagine the collective traumatism of the survivors.

But coming back to the remark of our cousin Pierre, what linked my parents was that, even if they were from very different social backgrounds and upbringing, they had endured a lot of similar experiences during the war. They ended up doing voluntary work in the same summer camp for children. The organization was affiliated to the Communist party and that's how they met. As I have already explained, my father was trained as a physician, and he was running the infirmary in that summer camp while my mother was a counselor.

Zierler:

Michel, where were you born?

Devoret:

I was born in Paris.

Zierler:

What neighborhood?

Devoret:

Your focus on neighborhoods brings up interesting stories. Even though I was born in Paris city center, I had always lived in its suburbs until my mid-thirties. I started living in Paris proper only when I married my wife, who was from a small town in Burgundy, and who became very quickly a true Parisian, unlike me, when she attended the Ecole Normale Supérieure in Paris (a very elitist college where students enter as civil servants of the French Republic and are paid a salary to study). I was born in Paris because my mother and father, who were living in Montreuil – which was one of the so-called red cities around Paris – it is now gentrified – were interested in a particular Paris hospital where Dr. Lamaze was the head of the maternity ward and a pioneer of the “painless childbirth” movement, which is known in the US as the Lamaze method of childbirth. It was located Rue des Bluets in the 11th arrondissement. Can I tell you a funny story about this?

Zierler:

Absolutely. Absolutely.

Devoret:

That will give you a sort of idea of what was behind all these names and locations, some 60 years ago. The story happened when I was voting in a French election, a few years ago, in a poll station in Mamaroneck, New York. This little town is almost at the border between the states of Connecticut and New York, and it has a French high school that serves as voting place for expatriates in this part of the Northeast -- votes take place on Sundays in France. And in that place, there were of course French people acting as poll supervisors. A supervisor woman about my age was scrutinizing my passport. And she said, "Oh, you were born in the 11th arrondissement of Paris? I was born in that arrondissement, too!". I immediately replied "Then, you must have been born in the maternity ward of Clinique des Bluets". "Yes, absolutely. How did you guess? This is astounding!" she said. Well, you see, this Holmesian guess was not too difficult to make, because of a few clues like her apparent age and her New York State residence. There were not that many possibilities. There was a high probability she was belonging to the kind of social background that produces mothers giving birth in that very type of maternity ward popular in certain – shall we say progressive? – circles at that time. I like these interesting – it’s a small world – stories. But so, to come back to your question, while I was born in Paris, all my childhood and adolescence was spent in the suburbs of Paris. So, for a New Yorker, it would a little bit like being born, let's say, in the East Village and then living in New Jersey all the rest of your early life.

Zierler:

[laugh] Well, I'm a New Yorker, so I appreciate the focus. Michel, what kind of school did you go to growing up?

Devoret:

I went first in an elementary school in Montreuil until I was eight, and then our family moved to the town of Orsay, in another suburb south of Paris called "vallée de Chevreuse". It was the start of a blooming scientific ghetto at the time, with a lot of various government labs and a large science university. There, I continued my primary school education and then went to a high school called "lycée Blaise Pascal". I am proud to have attended only public schools during my childhood. But at that time, you see, there was nothing special about it because the public schools were the best place to get an education. Now probably this has changed a lot.

Zierler:

Michel, did your family observe any Jewish holidays or customs when you were growing up?

Devoret:

No, not at all. As I said, they were not religious, and on top of that, my family did not want to distinguish themselves in any way from an average French family. There was the trauma of World War II, during which French citizens of Jewish background had been singled out. Anything Jewish was not the sort of subject that my mother liked to talk about. Her older sister had been arrested by French cops. She was then sent to the Drancy concentration camp north of Paris managed by the French authorities and then on to Auschwitz. She was murdered there. My father, on the other hand, was never uncomfortable with any discussions and was more talkative about our Jewish ancestry. Maybe because his close family had been relatively spared. Let's say that this branch of the family had not been decimated in the same way as the branch on my mother’s side.

Zierler:

In the French system when you're thinking about undergraduate education, do you have to confirm a major right from the beginning?

Devoret:

In the French system, when I was growing up, yes, you had to opt very early for science versus humanities. This choice happened around what corresponds to eighth or ninth grade. I believe it is even more so today.

Zierler:

And so from then, that's when you settled on science and math?

Devoret:

Yes.

Zierler:

What about physics? When did you know it would be physics that you would focus on?

Devoret:

Physics as a specialization came rather late. To specialize in physics came during graduate school. But up to that point, I just had a general math, chemistry, biology, physics background. I have to say that my father had warned me about careers leading to scientific research. He said "Well, it's nice to want to become a scientist, but you first must learn a real job." It made research even more interesting, since it was not an ordinary job. And so I followed what a lot of kids in France do, those who manage to get good grades in scientific subjects. They go to this preparatory cycle of what is called Grandes Écoles. You have to understand that in France there are two parallel higher education systems. There's the university and there is this elite track, the name of which can be translated as “Great Schools of Engineering”. Maybe you've heard about some of them, for instance the Ecole Polytechnique, which was founded by Napoleon. You follow that elite parallel track just after high school and you get an extensive training in physics, chemistry, and, above all, math. Then, you take an entrance exam to these schools and end up with an engineer diploma. And basically, in France I would say that, in order to become a physicist, you have to be an engineer first. There are exceptions, but statistically, the majority of students follow this kind of track.

Zierler:

This is true even for people who are interested in theory?

Devoret:

Oh, yes. Yes, even more so perhaps, because in France, the more elitist, the more theoretical the training is.

Zierler:

Tell me about your undergraduate experience.

Devoret:

My undergraduate experience was mostly this preparation period, which is a bit a continuation of high school. You are still treated like a high school student, but you have to do math and physics all the time. To give you an example, we had something like twelve hours of math class per week and a lot of homework. Eight hours of physics. Maybe three or four of chemistry. There was English and French or so but only a few hours of each per week. It is a very, very intense program. I did not enjoy it, but I recognize that its value was all the tutoring we got in the form of continuously running oral exams. They made us learn to be responsible for the knowledge we had to absorb.

Zierler:

You graduated sort of at the later end of these things, but were student protests going on at all when you were an undergraduate, or was that over by that point?

Devoret:

Yes, there were a few remaining demonstrations in 1970 and 1971. During this period, I was at lycée Saint-Louis in Paris, just overlooking the boulevard Saint-Michel where a lot of the 1968 barricades had been set up. And so, we got to watch some interesting demonstrations, even some fights between the police and the students.

Zierler:

Michel, what laboratory work did you have as an undergraduate that may have been formative for your intellectual development?

Devoret:

During the kind of undergraduate education I received in France, you would never set a foot in a true research lab. It's absolutely not the way it's done. I had some practical experience because I had the good fortune in my high school to be involved in building a computer, of course a tiny computer. [laugh] I was involved as president in a sort of club of students interested in science, and we had undertaken the task of building a computer, which today would be no more powerful than a small pocket calculator. That was a fun part of my high school education. And we were very lucky because one of the teenager students in that club had a father that was an engineer at a famous electronics center close to our high school, and he gave us some surplus supplies from his lab. And it was a great, great introduction to research, to teamwork, and even to project management. Of course, we were reinventing things rather than inventing but, you know, you have to begin somewhere. And it's always very formative to reinvent something. That was really the beginning of my career as a scientist.

Maybe I should mention that I was also making rockets with other friends outside high school, and this kind of project were also highly educational. Highly dangerous, but highly educational. And when I think about this rocket making, the sad thing is that if I would do this now in the US as a school kid, I think I would immediately be arrested by the police.

Zierler:

[laugh]

Devoret:

We would be suspected to be terrorists.

Zierler:

Michel, between your own academic interests and perhaps some advice you got from your professors, what did you do next when you graduated? What did you want to pursue at that point?

Devoret:

So at that point in the track I was following, when you graduated, you would receive an electric engineer diploma. However, I had already set a foot in the world of scientific research. The last year of my engineering school coincided with my first year of graduate school. You see, in my senior year in engineering school, unlike most students who would choose to do an internship in a company, I had the opportunity to pursue a sort of master degree at a university. I took some graduate level classes and did some research at Orsay University.

Zierler:

Who was at Orsay? Who did you work with there?

Devoret:

I worked in the molecular physics lab at Orsay directed by Sydney Leach, under the direct supervision of John Eland. And this led me to a sort of master's degree that French universities called “thèse de 3^e cycle”. It was a sort of mini-PhD. It was done in two years. The first year was mostly consisting of classes and led to a DEA diploma (“Diplome d’Études Approfondies”), and then there was a second year entirely devoted to research. My project was to detect the fluorescence of molecular ions by a new method. That was my first stint with quantum mechanics. I chose this project because I was going to detect quantum particles. The project consisted in detecting electrons and UV photons in coincidence, and we were also detecting the resulting ions. It was a triple coincidence experiment. And it was really a very interesting project which fulfilled my dream, which was really to learn more about quantum mechanics. Because at this point, I had read enough physics to understand that there was this very mysterious theory and that I wanted to learn it.

Zierler:

When did you first meet Anatole Abragam?

Devoret:

After this project, it was clear that I was not going to get a graduate student fellowship in that molecular physics lab. They did not have the resources to fund me. So I started to look for other labs, and then I heard about the lab of Anatole Abragam. He was very famous and I thought it would be very difficult to join such a prestigious lab. But it turned out I was actually very welcome. It looks like students were a little bit afraid to work in this kind of government lab with a guard at the entrance. And it was not very close to the railway station, either, so I think all of this must have discouraged competitors. Also, I have already mentioned the sense of doom that students in physics were being communicated at that time by their elders. I think I was advantaged by being convinced that this was what I really wanted to do, and nothing else. It helps sometimes to have a very limited set of options. Again, I was very lucky, you see. I was attempting to do physics at a time when the field was rejuvenating itself, as a consequence of the oil crisis. This oil crisis gave me my chance, because all of a sudden, physicists were commissioned again, this time to find new sources of energy.

Zierler:

Right. There's urgency now.

Devoret:

There was urgency because, unlike today where we know we should not burn all the fossil fuel underground, there was a fear that the production of oil would peak because we would not find enough of it. Because there was more money poured into physics, there were more graduate student positions. And I benefitted from these new fellowships, and was able to start a PhD in the lab of Anatole Abragam, in the Atomic Energy Center of Saclay. The acronym of the time, CEA, now means “Commissariat aux Energies Alternatives”, a clever way to not appear downright pro-nuclear!

Zierler:

Michel, were there any civilian nuclear reactors up and running at that point in France?

Devoret:

Oh, yes. Saclay was actually running a relatively large civilian nuclear reactor, but for research only. I was not involved in it. I was involved in nuclear magnetic resonance, and while there's nuclear in the name, it is basically solid-state physics probed by RF signals.

Zierler:

Michel, as you say, Abragam was famous. What was he famous for at that point and what was he doing by the time you met him?

Devoret:

Well, there were several elements to the fame of Anatole Abragam. He was obviously leading one of the best labs in France in this new field at the time, nuclear magnetic resonance. But also there were other interesting elements in the career of Anatole Abragam. He had written a book which everyone in France in the subject was appropriately nicknaming "The Bible." It's a quite thick book, a complete review on nuclear magnetic resonance. And it was written in English, which was very unusual at that time. An academic authority in France would not write a book in English. He had directly written it in English, and it was translated in French later by one of his younger colleague, André Landesman, who was one of my direct managers at Saclay.

In addition, Anatole Abragam was also a professor at Collège de France, which is an institution which has a reputation of being very liberal in the choice of its subject matters. The rules at Collège de France are such that every chair is a new chair in the sense that it is created especially for an individual professor, and there is a tradition of choosing subjects which are new and in the pioneering stage. You're not supposed to continue something older that is in decline. And once the professor retires, his chair is essentially destroyed, and a new one is created in another subject. In principle, at least, a chair in math can become a chair in history or a chair in chemistry or physics. Anatole Abragam was holding the chair of nuclear magnetic resonance at the Collège de France, which as one of only four physics chairs, was really prestigious. He was also somebody who had a marvelous sense of humor. In the French labs, you eat at the cafeteria and, after that, you have coffee. And at coffee time, Anatole was telling always those fabulous jokes, which all carried a piece of deep wisdom. I don't have a fantastic memory like he had, but I remember very fondly some of them and tell them to my students.

Zierler:

Michel, what was the process leading to you focusing on NMR in solid hydrogen?

Devoret:

Well, you see, solid hydrogen, seemed tempting for a graduate student. You know, graduate students have this notion that they can understand and judge the research project that is proposed by their advisor [laugh], which is completely wrong. They should better look at whether the learning environment is good for them in their education as scientists. But, OK if I go into the mindset I had at that time, hydrogen is a very fundamental atom, you know. And so I thought that project was going to be also another chance for really learning basic quantum mechanics. You see, I think it was always clear to me right from the beginning that I was going into physics, for the aim to understand and possibly contribute to quantum mechanics.

Zierler:

Michel, why is hydrogen a fundamental atom, why is it such an excellent pathway to understanding quantum mechanics?

Devoret:

Hydrogen is the simplest possible atom. It consists of a single proton at the center and there's a single electron which orbits the proton. That’s it. Actually, the reason why I left the low-temperature properties of solid hydrogen, which I thought would be the simplest system to understand, was that it became clear to me during my thesis that in its solid form, hydrogen is already awfully complicated. [laugh] I learned I had to focus on something even simpler. But it was a great opportunity for me to learn about low-temperature techniques, building a dilution refrigerator, and so on. Moreover, this technique of NMR is an excellent preparation to quantum physics. You learn a lot of useful basic things in this sub-field, both theoretically and experimentally.

Zierler:

How long did you research this project before you knew you had enough to defend a dissertation?

Devoret:

Oh, in those days, the system was very generous in terms of time. From beginning to end my project took six years, but I must say that during one year, I was doing my military service, which I spent in another lab about two miles away from Abragam's lab. And for a while I was working on solid helium instead of solid hydrogen. Solid helium-3, actually, is another one of these solids that are supposed to be the most basic, most fundamental, the most simple. Actually, solid helium-3 is somewhat simpler than solid hydrogen. Helium-3 is an isotope of the much more common helium-4, with a nucleus composed of 1 neutron and 2 protons, instead of 2 neutrons and 2 protons. Solid hydrogen is a molecular solid, so when you are forming it, you're not packing spheres, you're packing ellipsoids. You see, I realized that to understand all the secrets of solid hydrogen, I would have to learn all the techniques that would be needed to explore it, and they included many other ones than just NMR. Other techniques like infra-red optics, neutrons and heat transport measurements were also crucial. This solid was too complicated. You could not really understand it from one particular angle.

Zierler:

Michel, was military service compulsory even for graduate students?

Devoret:

Yes, it was compulsory. But graduate students had it nice because we would spend it mostly in a lab, except for one month of real military training at the beginning. Usually, they would send us PhD students in tough regiments, so I did my one month in the French equivalent of the U.S. Marines; but the rest of the twelve months I spent in a research lab led by Jacques Friedel, under the direct supervision of Maurice Chapellier, who was funded by the military. It was also not a big change in terms of science or geographical location, since I was working only a few miles away from my regular lab. I was doing the same kind of condensed matter physics, but on a different solid. I was very lucky because this military service turned out to be extremely educational. I have to say “in retrospect”. At the time I was resenting the fact that I had to stop for one year my PhD research, but on the long term it was very beneficial, both personally and scientifically.

Zierler:

Michel, how parochial was your graduate education? In other words, did you feel connected with advances that were being made in NMR research elsewhere in Europe, and even in the United States?

Devoret:

Oh, yes. Actually, I was very much connected with a lab in the US at Duke University, the lab of Professor Horst Meyer. He had immigrated just after World War II. Originally born in Germany, he had been raised mostly in Switzerland, got a higher education in England and then moved to the US. And he was a fabulous experimentalist. You know, I was doing these experiments in France, but later in the US, particularly during my postdoc, I really realized that I was just a theoretician. [laugh] It is in the US, in close contact with real experimentalists, like Professor Horst Meyer and later, with Professor John Clarke, that I learned in what sense exactly Physics is an experimental science. Horst had, attached to the desk in its office, a sort of bumper sticker that said, "In God we trust. Everyone else must bring data." [laugh]

Zierler:

[laugh] That's great.

Devoret:

It was not unlike in Teddy Roosevelt's office who had "The buck stops here." I thought it was delightful. But we will have time, perhaps, to talk about my postdoc in Berkeley—

Zierler:

Of course.

Devoret:

— which was really, really the beginning of my true education in experimental physics.

Zierler:

Michel, before we leave France, is the thesis defense system similar to the United States? Is there an oral component and a written component?

Devoret:

Yes, there are close similarities, but at heart it's very different. Superficially, you write a dissertation, and you defend it in front of a committee. But I have found that some details in the rules of the game are extremely different. In the US, at least in most places, and Yale belongs to the norm, we professors examine the candidate in a closed-door meeting. The candidate defends their thesis is in public, but the questions are in a closed-door environment, probably so that the professors do not ridicule themselves in public.

Zierler:

[laugh]

Devoret:

In France, you see, the question period is fully public, so the defense is more an exam for the professors than for the student.

Zierler:

[laugh] Michel, how did the opportunity for the postdoc at Berkeley come about for you?

Devoret:

Oh, so this is also where I have to say I owe a lot to Anatole Abragam. At that point in your career, in those days in France, you had already a permanent position. It came typically during your PhD. I had thus a permanent position when I defended my dissertation. Although I was a Saclay employee, Anatole Abragam let me do a postdoc and complete my scientific education elsewhere. I think he understood the situation because he himself had received his higher education training in quantum mechanics at Oxford. You see, there is a crucial point that maybe we should talk about. The university in France, at the beginning of the '60s, did not teach quantum mechanics. This complete obscurantism had a lot to do with the fact that Louis de Broglie, one of pioneers of the early quantum theory, did not truly believe in quantum mechanics. He was under the impression, maybe in the wake of Albert Einstein, that this was a temporary theory. Since it would be revised soon, there would be no use of teaching it to students in its current form. In retrospect, of course, this looks very prejudiced and even totally crazy, given all the successes of quantum mechanics at that point. Fortunately, a rejuvenation of the teaching of quantum mechanics happened in 1962, which was awfully late, but brought up a very eager to learn, new generation of young physicists. It was people like Albert Messiah and Anatole Abragam who taught first quantum mechanics in France, at Saclay, which was a government lab independent of the university. Messiah got educated in the US and Abragam in England, and they both brought back a solid knowledge of quantum mechanics. One of their famous students was Claude Cohen-Tannoudji, who eventually won the Nobel Prize for laser cooling. And he himself took on the task of teaching quantum mechanics at the university. So, when I was learning quantum mechanics, the books written by Cohen-Tannoudji and his assistants were brand new, and they contained something that had not been taught in France before, a totally new subject. So maybe because Anatole Abragam valued physics education in the Anglo-Saxon world, I was authorized to take a leave of absence and become a postdoc at Berkeley.

Zierler:

So you had a job waiting for you if you chose to return to Saclay?

Devoret:

Yes. I have to mention that, at that stage, there was also another element of luck. At that time in France, you would not be an adult scientifically until you were around 40 or in some cases, even older. But Daniel Estève -- who was a close friend and a fellow student -- and I, we were very fortunate that our immediate thesis supervisor, Neil Sullivan, left Abragam’s lab. He was still a relatively junior scientist but moved to University of Florida, Gainesville and eventually became the Dean of the Physical Science there. So basically, his lab space was available to us, and we were able, when I returned from my postdoc, to have our own lab, Daniel Esteve and I, at a relatively early age. Later Cristian Urbina, another former student in Abragam’s lab, joined us. It was something that would never happen to anybody at that time. It still barely happens, because the age pyramid in French labs is more like a cylinder. But right now, the situation is getting a little bit better. Thanks to these European fellowships that allow very talented young researchers to start their own lab essentially after their postdoc. But, again, this is very, very new and was unknown to physicists of my generation. Therefore, when I returned from the two-year postdoctoral work at Berkeley, I was able to start our own laboratory. It was a little bit like what I would have done if I had taken an assistant professorship position in the US, but I had no teaching duties. Moreover, we, junior physicists at Saclay, could join forces and work on one collective project.

Zierler:

Michel, was your sense of what John Clarke was doing in quantum tunneling, was that happening anywhere else in the world, or this was really at the beginning?

Devoret:

That was completely at the beginning. The subject we attacked was not directly in the line of what John Clarke was doing at that time, which was more related to SQUIDS, superconducting devices that can measure extremely small magnetic fields. You see, John Martinis, I and John Clarke embarked in this project because of a conference given by Tony Leggett. As you perhaps know, at the beginning of the 80’s, Tony Leggett, was going around the world giving talks on the possibility that Josephson junctions could display truly macroscopic quantum phenomena, like macroscopic quantum tunneling. This is the tunneling of a macroscopic variable that is related to the voltage across a Josephson junction. This variable is called phase, but you should see it more like the electromagnetic equivalent of the angle a pendulum makes with the vertical. Maybe this is something that is not sufficiently recognized. Tony Leggett got the Nobel Prize in Physics for the explanation of the superfluidity of liquid Helium-3. But he also made this wonderful contribution of calculating, with his student Amir Caldeira, how would quantum tunneling of the Josephson junction phase be affected by its dissipation. They showed that it was observable, which was counterintuitive because usually macroscopic variables only obey classical physics. This is implicit in the Copenhagen interpretation. The experiment was very challenging, but the effect was, in theory at least, observable.

John Martinis was a graduate student at that time, I was a postdoc and we teamed up. I was coming with knowledge in dilution refrigeration, which no longer existed at that time in John Clarke's lab. And maybe there's more on this that I want later to tell you about. Anyway, we constructed this experiment from scratch, and one of my roles was the cryogenic part. On another level, I was bringing my NMR knowledge and what you could learn from electromagnetic irradiation techniques. John Martinis had some experience in building superconducting devices. So we were complementary. John Martinis was teaching me the fabrication of Josephson junction, and I was teaching him about NMR and dilution refrigerators.

Zierler:

Michel, how was your English before you arrived in Berkeley?

Devoret:

In retrospect, it was less than perfect, but still much better than what many French people of my age had managed to learn in school. You see, there had been an event in my adolescence that I haven’t told about yet, which was another one of these lucky, fortunate things that are important in one's life, and have enormous influence in deciding your future. When I was thirteen years old, my father, who was by that time an active biologist, spent a sabbatical at Yale University. He was working in the medical school, and at present I'm working on the other side of the freeway running through New Haven, on that other part of the Yale campus housing the school of arts and sciences. Because I joined at Yale the physicists and the engineers, I am in a professional environment totally different from what my father experienced, and maybe we will come to this again later. In any case, for my father’s family, which included me, this one-year sabbatical was a great opportunity to familiarize oneself with the American way of life and to learn real English, not the language we were learning in my high school in France, and which was called “Anglais” in our schedules. So I came to John Clarke's lab rather well equipped. I would not say I had a very solid knowledge of English, particularly written English, but I was familiar enough with the culture and language to get by.

Zierler:

What were your impressions of Berkeley when you first arrived?

Devoret:

Oh, it was marvelous! Before joining John Clarke's lab, I had toured a few interesting labs in the US, and in particular the lab of Professor Meyer at Duke University, which I have already mentioned. But when I walked across the sunny Berkeley campus, with its gorgeous Mediterranean trees and succulent plants, facing the Bay of San Francisco, I said to myself: “I must come here for my post-doc”. Another excellent impression I received was coming from John Clarke's students. I immediately sensed that this was a fantastic research group. And also, I had read John’s papers. I thought they were crystal clear. Intellectually and personally, there was no question that this was really the lab I had to join.

Zierler:

What was it like working with John Clarke?

Devoret:

I have to say that John remains—he is what you would call a role model in science. Being one of his advisees was wonderful and a great chance for my career. He was really closely advising everybody in his lab. I had not been through this kind of strict and intense training before. Basically, John Clarke was coming to see us every day in the lab and was asking, "What is new today?" [laugh] He was both jovial and respectful, besides being demanding. If we would find two new things in one day, we would store one in reserve for the next day in case it might be a lean day. Basically, we would store things in reserve to give John the impression that we were progressing steadily. I have excellent memories of this constant scrutiny, and this questioning of where we were heading.

A great habit of the group was also that every student and postdoc had to give a talk regularly, at least three times a year, in a group meeting. Thus, we were constantly stimulated into expressing ourselves scientifically, putting order in our thoughts and so on. [laugh] I realized during that post-doc that before that I was pretty much free-wheeling. On the other hand, to have the opportunity during six years to be free-wheeling in physics, there were advantages to that. Daniel and I had educated ourselves in various aspects of physics and had developed tools to think about problems. Thanks to the company of Daniel, who knew a lot and is a very deep thinker, I had had the time to absorb a good general physics background, which was also very useful in the long run. The French system had offered me the time to think about things and to mature scientifically, so I was ready to take advantage of the Berkeley lab at this point.

Zierler:

Michel, did you realize in real-time as these things were happening just how foundational this research was that you were involved with at Berkeley?

Devoret:

Yes, absolutely. I think we were pretty megalomaniac, both John Martinis and I, when doing our macroscopic quantum tunneling experiment. While we were of course incapable to predict where it would lead to, we were convinced that it was foundational. Ours was an ambitious experiment, containing elements that had not been attempted before. And John Clarke had been to that point involved in the physics of SQUIDs, which was both very fundamental – it touched on quantum noise -- and very practical. He was developing measurement tools, and his experience of the behavior of the superconducting phase, and the various ways it was perturbed, was really precious for our fundamental physics experiment. We wanted to do the best we could to bring proof that Leggett’s calculation was right, that macroscopic quantum tunneling (MQT) existed — It was obvious, the secret hope of Tony Leggett was that the phase, as a macroscopic variable, would not behave quantum mechanically, and for him, this would solve some of the mysteries of quantum theory. So our job was to make sure that if we would not see the predicted quantum effect, it would be for good reasons, not because the experiment did not work for some mundane flaw. Therefore, our job was to devise the right control experiments that would demonstrate that we were doing everything right.

By the way, Daniel Esteve came to visit us for a month at Berkeley and played a key role in the design of a key control experiment. So, as you know, the end result of this experiment was that quantum mechanics was beautifully explaining the tunneling behavior of the macroscopic phase difference. We did not find any deviation – within experimental uncertainties. So macroscopic variables were obeying quantum mechanics. This in itself was not a surprise to most people. What I think was new was that we were the first to make artificial atoms, macroscopic atoms to which you can connect by attaching wires, but which still behave as microscopic atoms. I think it was the beginning of superconducting qubits, but we did not call them this way yet.

Zierler:

[laugh] As you say, when you got to Berkeley, your notions of what a theorist was and an experimentalist, they changed.

Devoret:

Yes, yes, completely.

Zierler:

In what ways specifically with this experiment did those perceptions or notions of these divisions change for you intellectually?

Devoret:

Yes, so you see, the very important thing I understood with John Clarke was that a good physics experiment has to be, by and large, independent from theory. In the French system, you would do experiments to prove that the theory was correct. But you were not intellectually prepared to the idea of doing an experiment just to probe nature in a way that would maintain theory at arm’s length. You had to be open to surprises. You had to do a professional job in doing the meaningful control experiments, those that do the part of the devil’s advocate. The experiment had to stand alone. It was not in service of a unique theory. Its main components had to be independent of the theory you are interested in, even if it was directly motivated by a theoretical question. Grasping fully all these ideas was essential if you wanted to really understand the big differences between theory and experiment. And I have to say also that these superconducting circuits had some applied components, unlike the work I did for my PhD. After all, the SQUIDs were used as measurement devices. So there was this proximity with applications that I had never encountered before. We were making something that was concrete and that could be turned into – maybe not something as grandiose as a commercial product, but at least something that other scientists might want to use.

Zierler:

Michel, did you have a notion at this point that maybe you would want to make a long-term career in the United States and perhaps not go back to France permanently?

Devoret:

Yes. Well, there was some possibilities after this postdoc to take an assistant professorship position, but I have to say that the odds were a little bit — should I say not completely in my favor. After all, going back to France, I would team up with very talented people, have my own lab and my own funding, and also continue to hold my permanent position. So there was not a huge incentive in fighting the type of fights that young assistant professors have to do in the US. You know, the American system is rather cruel at the assistant professorship level. I've come to believe that there is an optimal country for each stage in your physics career. Certainly you want to be a postdoc in the US, but you don't necessarily want to be an assistant professor there.

Zierler:

What did you ultimately choose?

Devoret:

I chose to go back to France and set up this lab with Daniel Estève and Cristian Urbina, which became known as the Quantronics Lab at Saclay. And we did a certain number of experiments we are proud of. But that's another chapter in my career. [laugh]

Zierler:

Tell me about that. How was the transition for you?

Devoret:

Here, I have to mention that John Martinis, after his dissertation defense, came to join us for a postdoc. We embarked with him in a very, very ambitious experiment. The aim of the experiment was to measure the tunneling time of MQT. It's a quantum teaser, the notion of the time it takes for a particle to go under a barrier while tunneling. And we could talk about it for hours. The specialist, by the way, of this question now — we have ceased to work on this – is a younger scientist, Aephraim Steinberg, who is at Toronto, and who has done recently a very clever experiment on the measurement time of tunneling. His is an atomic physics experiment, whereas we measured this time for macroscopic quantum tunneling.

Zierler:

Michel, did you see this as a natural progression from what you were doing at Berkeley or a separate kind of research pursuit?

Devoret:

I think it was the best opportunity I had at that time to continue experiments on what I was calling "quantronics," which is a sort of contraction of quantum mechanical electronics. The idea that we would do fundamental quantum physics experiments but with electrical degrees of freedom, like currents and voltages, was something I found fascinating. You know, at that time, people doing fundamental quantum experiments like Alain Aspect were doing them on optical photons. As Aspect wannabees, basically, instead of doing experiments in quantum optics, we would do them with electrical degrees of freedom in superconducting circuits. This was very much in line with my former electrical engineer training.

Zierler:

Where were you publishing at this point? Given how important all of this research was, what was most important in terms of where you were publishing and what conferences you were attending?

Devoret:

Your question makes me realize that, in retrospect, we were in a state of bliss without knowing it. The high impact journal in those days was Physical Review Letters, so we were publishing in it, or hoping to publish in it. You were judged at that time by the number of articles you had published in Physical Review Letters. Unfortunately, now it tends to be in Nature and Science. Your questions open all kinds of subjects. [laugh]

Zierler:

That's what I do. [laugh] Michel, who coined the term—

Devoret:

Because we could talk about this also for—

Zierler:

So much, so much.

Devoret:

—a very long time, this evolution of the publication medium for physicists.

Zierler:

Who coined the term "quantronics," the Quantronics Group?

Devoret:

This is one of my main claims. I like to give names to things that crystalize an idea. I recently nicknamed one of the devices we invented in my group. I called it the "SNAIL" because I was envious of the SQUID and the SLUG, both John Clarke’s creatures, and we thought we needed another mollusk in the zoo of superconducting circuits—

Zierler:

[laugh]

Devoret:

—so we invented the SNAIL to complete a trio. [laugh]

Zierler:

What were some of the technical difficulties with creating the SNAIL?

Devoret:

Interestingly, the SNAIL came out of a very nice blackboard discussion which I had with my student Uri Vool. At some point, we realized that the SQUID, I mean the D.C. SQUID, which is a two-terminal device, is a reciprocal device, which is very strange at first sight. You see, even though the two sides might be made of different metals, and even though you might have asymmetry everywhere in the layout of this circuit, there's some fundamental symmetry at the level of the electrons which makes this two-terminal device completely symmetric. It behaves exactly in the same way when you exchange the terminals. In other words, it is not at all like a semiconductor diode, even though it might look like having the same kind of superficial asymmetry that the diode has. So, we created a new device that would break the symmetry. It was really a sort of Aha! moment to realize what exactly was needed to create a deep asymmetry. You can describe the SNAIL as a sort of inductive diode for superconducting circuits. We called it the SNAIL, which stands for Superconducting Nonlinear Asymmetric Inductive eLement.

Zierler:

Another new word for me. Tell me about the initial observation of Quantronium.

Devoret:

The quantronium was probably one of the highlights of the accomplishments of our group at CEA-Saclay. It came through a turn in my career that is very interesting. I have first to tell you that when I came back to Saclay, we continued for a while to do experiments on macroscopic tunneling, but there was something that was bothering us very much. We realized that the quality factor of the devices that we were using was only a hundred or so. The quality factor is basically the product of the frequency of the oscillation and its decay rate. So the dissipation was fairly large, at least when compared to natural atoms. It was not so large as to completely wash out all quantum phenomena, though. This is why we could observe tunneling, we could resolve energy levels and so on, but compared with atomic physics, compared to our competitor at that time, which was the Rydberg atom, which was studied in the group of Serge Haroche and Jean-Michel Raimond, our program looked like the road to failure. Although we had very advantageous coupling factors and our rigidly anchored atoms didn't fly in space, the quality factor of our artificial atoms was dismal compared with the Rydberg atom. So at some point we thought that maybe it was wise to stop. We did not have the right junction fabrication technology either.

We moved on and focused our interest on single electron effects, which had been predicted by Kostya Likharev and his team in Moscow. It was a good break because it allowed us to familiarize ourselves with aluminum tunnel junctions and e-beam nanolithography. Because, you see, these Berkeley experiments, they were done on Nb-Pb junctions that were proved later to be really not the right type of junctions you want to pursue for very long. These niobium-lead junctions had the nice property that they can be tested at liquid helium temperature because the superconducting transition temperature for niobium and lead are higher than liquid helium temperature, but these junctions don't behave very well in terms of their intrinsic dissipation. Actually, the foray we embarked on for almost 10 years, single-electron electronics, turned out quite useful later on for superconducting qubit work. This pushed us to familiarize ourselves with entirely new type of junctions which were the right objects for doing more ambitious work on quantronics, which was the Josephson junction as a qubit.

Zierler:

During this time, Michel, what struck you as some of the real potential breakthroughs in superconductivity, and what were the walls that you and your group were hitting up against in superconductivity at this time?

Devoret:

So, you know, these results of macroscopic quantum tunneling came at a point where one of the big popes of superconductivity, Pierre-Gilles de Gennes, a famous French physicist, a Nobel Prize winner and professor at the Collège de France, who had written a book on superconductivity, was declaring in various committee meetings that the field of superconductivity was finished. And that was mostly true, the kind of experiments that were done in the '60s didn't have much future. But a few years later, because there was all this interest in macroscopic quantum tunneling, and the momentous discovery of high-Tc superconductors, it became clear that superconductivity was not a finished field. What was opening was a completely new territory for superconductivity.

Zierler:

Michel, who were some of the international researchers who came to your lab at this point? Who was really excited about what you doing even from beyond France?

Devoret:

John Clarke did a sabbatical in our lab, but there were other important visitors, like Hermann Grabert from Essen, Germany. We also started a close collaboration with the group of Hans Mooij at Delft. As I started to tell, in 1987, there was a big stir in this community of people working on superconducting tunnel junctions, and that stir was produced by Konstantin Likharev, who had managed to attend an important conference in Belgium despite travel restrictions in what was still the USSR. Daniel Esteve and I were attending this conference and we heard this incredible review talk by Konstantin K. Likharev on the results of his team in Moscow. Everyone in the audience thought that the content was either complete rubbish or a revolution in our field. It turned out that it was really a revolution. By the way, unknown to us, the Soviet Union was about to crumble, as Kostya was predicting, but no one was believing him. This conference, and the visit that Kostya and Dima Averin did later at Saclay, provided us unique opportunities to observe what was going on beyond the iron curtain.

The Russian school of superconductivity was excellent. Of course, you know about Lev Landau. Among other qualities, Lev Landau attracted a lot of people who became great specialists of superconductivity. There were a lot of things they discovered independently of the Americans. So these single electron effects proposals that came in from the cold were a very interesting rejuvenation. They were quantum, but in a different way than Macroscopic Quantum Tunnneling. Up to that point, in terms of fundamental constants, we had occupied ourselves with the flux quantum and Planck’s action quantum in our macroscopic quantum mechanics experiment, but Kostya Likharev focused on the charge of single electrons and single Cooper pairs. He was promoting other new types of quantum effects that had some relationship to Macroscopic Quantum Tunneling, but emphasizing the charge degrees of freedom rather than the phase degrees of freedom. And Likharev was proposing to observe what he called Bloch oscillations. The Bloch oscillation is to charge what the so-called alternating current Josephson effect is to phase. The two effects are pretty much the electric dual of each other. To these days, the Bloch oscillations have not yet been observed completely convincingly in superconducting circuits. If somebody would manage to fully observe Bloch oscillations, this would be a way to have a fundamental standard of current mirroring exactly the way the Josephson effect provides a standard of voltage. Whereas the standard volt is based on the controlled passage of flux quanta one-by-one, the standard ampere based on Bloch oscillations would be based on the controlled passage of Cooper pairs one-by-one.

Zierler:

What were some of the advances generally in quantum electrodynamics at this time that were not happening in your group but may have been relevant for what you were working on?

Devoret:

At that time, outside France, there were two very powerful lab working on these single electron effects. One was the lab of Professor Mooij at Delft in the Netherlands and the other was the the lab of Professor Claeson at Chalmers in Sweden. And both labs were involved in rather advanced devices, quantum devices made with Al tunnel junctions. Chalmers became the home of some in the team of Kostya Likharev, and they embarked on experiments trying to demonstrate Bloch oscillations and the related phenomenon, which is single electron tunneling oscillation. The lab of Professor Mooij at Delft Technical University was focusing more on arrays of Josephson junctions, but they were also studying quantum effects, the quantum tunneling of vortices in those arrays of Josephson junctions. They were doing marvelous arrays of hundreds of Josephson junctions.

Zierler:

During these years were you teaching at all or this was strictly research work and supervising graduate students?

Devoret:

Yes, it was purely research work, only supervising graduate students or postdocs. I was teaching once in a while, but this was on purely voluntary basis and it was not a regular commitment.

Zierler:

Were you involved in metrology at all at this point or that came later?

Devoret:

Yes. That came about because of our experiment on the single electron pump. We were the first to propose to count electrons one-by-one with a device that we called the single-electron pump. Prior to that, there was the single-electron turnstile that we had developed in collaboration with Delft. But it was the electron pump which really arose the interest of metrologists, because it could produce a current even with zero voltage across it. It thus offered a pathway for a fundamental standard of current and the closure of the metrological triangle. You know that there is the Josephson effect, which transforms frequency into voltage. There is also the quantum Hall effect, which transforms a current into a voltage. The Josephson effect involves the flux quantum, the constant h/2e, Planck’s constant divided by the charge of Cooper pair. The quantum Hall effect involves the quantum of resistance h/e2 which involves Planck's constant divided by the square of the single-electron charge. And metrologists proposed to close the triangle formed by the 3 effects connecting the hertz, the volt and the ampere. In metrology, closing a triangle formed by three units is very important because it provides a rigorous simultaneous check of all the effects that transform a unit into another. It's a bit like in geodesic survey: if you can close a triangle of physical locations, you can verify all the distance measurements at the same time in a fool-proof manner. So closing the metrological triangle was very important to prove that all the constants appearing in all these effects are really fundamental constants and not some effective value involving complicated corrections. This remains a Holy Grail of electrical metrology, and its results are becoming better and better.

Unfortunately, the standard of current is still not exactly at the desired level to be on par with the Josephson effect and the quantum Hall effect. Let me remind you that the quantum Hall effect, we know its precision is at least 10^-10. For the Josephson effect, the precision is 10^-17. And, well, people hope to make the current standards at the 10^-8 level, but it is still work in the making. Since this verification of the metrological triangle is so important for metrology, I have spent a few years as a member of the Science Advisory Committee of the Metrology Bureau in France. Being part of a metrology institute is a very, very good education in physics. Metrologists have a view on science which is very sound and refreshing.

Zierler:

Michel, on the administrative side, what were the circumstances leading to you being named director of research of CEA at Saclay?

Devoret:

Let me clarify the title you are mentioning, which actually can be translated in English as “director of research at CEA-Saclay”. This title has to be understood as a research director located at the CEA-Saclay center. It did not mean that I was the director of the whole research effort done at Saclay. The title corresponds to a promotion I received which came with a few duties beyond pure research. I had to organize meetings and coordinate several departments, but in the CEA, such directors do not have real powers beyond their own research group. The people who have real power in that government institution are managers that are so busy with administration that they can no longer do research. Having lived during the last 20 years with road maps for quantum computing, I can tell you that this kind of top-down approach in organizing research is not always very fruitful. [laugh]

Zierler:

[laugh] Did you allow or was it possible to retain your research agenda when you were director, or you paused on that part?

Devoret:

No, fortunately this role of “research director” came with little administrative duties. It had actually been created for senior physicists who did not see themselves become full-time managers.

Zierler:

When did the move to Yale happen? When did you start thinking you might go somewhere else?

Devoret:

Well, when you reach, let's say, 45 or 50 in the French system, particularly at a government lab, you have two solutions. Either you are involved heavily in administration, and your activity becomes purely management, or you are put in a closet, figuratively speaking. So there's not much hope for change in your career, in the sense that whatever lab you have built, that’s it in terms of growth. So, by becoming a professor in the US, I had the good fortune to restart completely my career from scratch. And as you know, when you start building from zero, the derivative of your growth can be nothing else than positive! It’s much more satisfying to grow than being in steady state.

But I have to say that I was very happy at Saclay. I had wonderful colleagues, and our research was producing interesting results. It was just a bit disheartening that we had reached a sort of steady state. I was also worried that my leadership of our excellent team was not as creative as allowed by the means it had. I knew that in the next 20 years or less, because retirement at that time was very early, my activity would basically reach a plateau. As you can imagine, it was much more fun, and much more exciting, to start a new chapter of my career in the US. Another aspect is that I was in touch with many labs in the US. All my close competitors participated in US physics one way or the other, so that was another good motivation to move.

Zierler:

And how did Yale come together for you?

Devoret:

I told you about the sabbatical of my father, but I had become interested in Yale for reasons totally unrelated to his. I view my return to New Haven as a pure coincidence. Not my return to the US, because my wife and I wanted to give our children the same opportunity I had had in my youth, but the return to New Haven, yes, I think it is just chance. The beginning of this chapter of my career, I think, starts with the work of Yale Professor Dan Prober on particle detectors for radio-astronomy. Dan had been a graduate student of Professor Michel Tinkham at Harvard, and was among the best expert of superconductivity in the US. But Dan was also at heart a radio amateur. He had very strong knowledge of electrical engineering. He also had developed very interesting techniques in mesoscopic physics, borrowed from his radio-astronomy work. He was doing experiments with microwaves in a way we, at Berkeley or Saclay, had never attempted. We were sending microwaves to our circuits, but we were not listening to the microwaves coming back. We were extracting information from our sample in the audio range of electrical frequencies. Thus, in our so-called hybrid transport experiments, we were sending high frequencies in the GHz range, but we were receiving frequencies in the kHz range. And, you know, for an electrical engineer or communication engineer, the higher the carrier frequency of your signal, the better. A higher carrier frequency means you can extract more information from your system, for a given relative frequency excursion. As I mentioned, Dan Prober was also putting superconductivity in the service of radio-astronomy, realizing superconducting detectors for photons coming from celestial objects in the microwave range. He and the people at Chalmers were the only people in the world at that time that were doing something I thought was revolutionary in terms of measurement technique.

Zierler:

Michel, why was it so revolutionary exactly?

Devoret:

It was revolutionary in terms of information acquisition rate. So, you see, when you have an audio readout experiment, you are collecting signal that is modulated at 1 kilohertz or maybe 100 hertz, using a chain of amplifiers with a noise temperature of a few kelvin. But when you were using the amplifiers that the radio-astronomers had developed over the years, you could have the same noise temperature of a few kelvin, but with an enormous bandwidth—relatively speaking. This is because the carrier is now a few gigahertz, and with not too much engineering difficulties, you can have a bandwidth that is at least several hundreds of megahertz. So you basically are acquiring information at a rate that is a million time greater that with a readout in the audio range.

So imagine if you were in the position of an astronomer and you were doing experiments with simple binoculars, and now you have access to the Hubble telescope, what would you do? Would you keep watching the sky with your binoculars or would you switch? [laugh] That was the position I was in, so that was a no-brainer. I have also to say that at Yale there was Rob Schoelkopf, who had been a post-doc of Dan and had become a new professor in Applied Physics. Rob had also been trained as a radio astronomer in graduate school at Caltech. Prior to that he had worked at NASA for two years as a microwave engineer. Dan and Rob had done together experiments which personally I thought were very revolutionary. One of them measured quantum shot noise, which are quantum electrical fluctuations at microwave frequencies. As you can see, there was deep quantum mechanical physics associated with the application of radio-astronomy techniques to mesoscopic systems. This is what spurred me to do a one-year sabbatical visit in their lab, which took place in 1999-2000. And during that year, Yale proposed me to come permanently as a full professor.

At this point I should mention the important role of my close friend Douglas Stone, who was then the chair of the Applied Physics Department. My family and I went back to France, thought about the offer, which had to involve a position of my wife in the French department. In 2002, the family moved permanently to New Haven. Again, like during my post-doc, I was able to bring with me some useful techniques of my own lab like the fabrication of Al tunnel junction by e-beam lithography and e-beam evaporation. I was also bringing knowledge of the qubit we had developed at Saclay, the quantronium. In exchange, Rob and Dan were teaching me these fantastic measurement techniques based on microwave reflectometry. So from this point on, my research at Yale has been a happy marriage of the combination of the circuit ideas that we had invented at Saclay with the Yale measurement techniques based on the tools invented for radio astronomy.

Zierler:

When did you first meet Steve Girvin?

Devoret:

In this discussion with you, we are reviewing all the elements of luck that took place in my career and this new question leads to yet another one of them. As I was deciding to move to Yale, it happened that coincidentally, or maybe as a some sort of correlation effect, I don't know, both Leonid Glazman and Steve Girvin joined Yale. They are exceptionally talented theorists and great colleagues. I had actually already collaborated with Leonid Glazman, and he was a good friend. We had met at various conferences and spent time at Delft together, and to be able to tightly collaborate with him at Yale was an incredible opportunity. There is probably no one in the whole world who understands mesoscopic superconductivity as well as Leonid Glazman. Regarding Steve Girvin, his commitment to and understanding of quantum information physics is fabulous, too. He is also a formidable teacher and speaker, and can cast complicated concepts in very simple terms. We benefit from his deep wisdom like when he says “you should never attempt a calculation which you do not already know the answer of”. All my close Yale colleagues, and here I must include of course Rob Schoelkopf, with whom I am sharing a lab and who is one of the greatest experimentalists of his generation, created a wonderful scientific environment.

Zierler:

When did all of this come together to be something that was more than just your faculty appointments, that there was a quantum center to be built at Yale?

Devoret:

It's fortunate that physics departments now in the US realize that science is done by more than a single principal investigator group. This had been the model in the '60s, at least in this low-energy physics domain that I had been involved with. We are transitioning to another model now, necessitated by the evolution of experiments, which are becoming more and more complicated. You know, a closely knit community of individual talents is more than just the sum of its parts. There is incredible cross-fertilization and internal reviews and quality control that take place when you have a critical mass in an institute on a particular subject. When you have colleagues with whom you can talk about your experiments before they fully crystallize, you can bounce ideas back and forth, test them and get a lot of valuable input. And when the experiment seems finished, when you have obtained the results you were looking for, you have all these critical eyes that tell you that your control experiments are insufficient and so on.

So, yes, I think this idea of institutes where you have a coherent community of physicists is powerful. Niels Bohr seemed well aware of this. When I visited the Niels Bohr Institute, it looked like this community spirit was still living. That’s what I believe I've contributed to establish at Yale with my colleagues. We form a group of research teams that is sufficiently large to transcend what can be accomplished by single investigator groups. Departments are no longer as frightened as before to hire several professors working in the same field, and funding agents like to reward this kind of teamwork. Previously, there was a sort of exclusion principle, an “academic Fermi principle” if you like, where a professor would perceive a local colleague in his own field as dangerous competition, not beneficial stimulation. Maybe this isolationism was also driven by education, in order to have a palette of professors able to teach all corners of physics. But mentalities are changing. There are also now a larger number of tenure-track positions, and maybe the competition is not as high, so people have a strong incentive to collaborate.

Zierler:

Michel, when did you realize that nanofabrication was something that was very important for this research?

Devoret:

You are evoking a turning point in the day-to-day practicalities of our research on superconducting circuits. I have told you about our worries with the quality of our tunnel junctions. The tunnel junctions we used in our Berkeley experiments were made with optical lithography, pretty much with the type of techniques that were employed at that time in the semiconductor industries for computer chips. But in the meantime, Ted Fulton and Jerry Dolan at Bell Labs had developed a brand-new technique based on e-beam lithography which allowed physicists to make much, much smaller structures, tunnel junctions in particular. You know, there was a joke at that time saying that nanotechnology begins at dimensions lower than 999 nanometers. [laugh] Submicron would be a better term, because it is very difficult to control matter at the level of 1 nm, which is roughly the height of 10 atoms piled on top of one another. The new e-beam techniques were absolutely essential to create the junctions that were needed to go beyond these early experiments. And these techniques were actively developed at Bell Laboratories, at that time the temple of American solid-state physics and more generally, low-energy physics. So the challenge for labs in Europe was to import in an academic environment so-called nanolithography methods, which are based on the writing, with an electron beam, on a dedicated resist deposited on a substrate. This writing step, after development, creates a sort of stencil through which you evaporate high-purity aluminum. The overlap between two aluminum films obtained in this way forms a tunnel junction if you have allowed the bottom film to form the adequate oxide layer in an oxygen atmosphere. Executing this process correctly and with a high yield was a key aspect of our research. It is still a key component of our research. When I came to Yale, I became responsible for the establishment of in-house nanofabrication of tunnel junctions.

Zierler:

What have been some of the greatest technical challenges when you're achieving microwave measurements to do it in as ultra-low-noise as possible?

Devoret:

What I initially started at Yale was a whole research program on amplification at the quantum limit. The idea was making practical amplifiers whose performance are at the limit prescribed by quantum mechanics. Quantum mechanics imposes that when you amplify a signal, you are going to suffer from a minimum added noise. The best devices at that time were 20 or 30 times above the minimum limit. The radio astronomers who had devised those HEMT (High Electron Mobility Transistor) amplifiers were not caring too much about low noise. They were observing the sky, and you know that the temperature of the cosmic background is 3 kelvin. No matter the tricks they play with antennas, the sky is hot for radio-astronomers working in the band 2-16 GHz, so they are not especially interested in making ultra-low-noise amplifiers, whose noise temperatures would be well below the noise of the sky. But when you do a measurement of a superconducting qubit at 20 millikelvin, you want the least possible amount of noise to perturb your experiment. This is very important, particularly when you do correlation measurements. In this case, an increase of signal to noise ratio by a factor of two translates into a 16-fold shortening of the measurement time, a dramatic effect. I don't know if you've heard this phrase yet, but experimental physicists are usually ready to kill mother and father to improve their signal-to-noise ratio!

Zierler:

[laugh]

Devoret:

[laugh] So you now understand our motivation for making practical quantum limited amplifiers. Already, there had been pioneering work at Bell Labs by Bernard Yurke. He was actually the first to a make a device that operated probably around the level of the quantum limit. These pioneering experiments were wonderfully inspiring, but they did not lead to a practically useful amplifier. This was not the device you could use to make new measurements — it was more a proof of principle experiment. In this quest for practical quantum limited amplifiers, I was strongly influenced by my mentor, and post-doc advisor, Professor John Clarke. As I already mentioned, John was famous for doing both fundamental experiments on superconductivity and SQUIDs, and also developing SQUIDs as physics instruments for various projects like geophysics or brain research.

Zierler:

Michel, when did you realize that investigating single Cooper pair devices would be valuable for a broad range of things you were interested in?

Devoret:

At this point, I have to mention another important character in this story. It is Yasunobu Nakamura, a very, very talented Japanese physicist. Towards let's say the end of the '90s, it was pretty clear that our best hope for achieving the best quantum circuits would be in a regime that was completely opposite to the regime we had started with at Berkeley. The regime we were investigating at Berkeley when we were doing these experiments on macroscopic quantum tunneling was the regime of junctions that had a very low impedance. It means practically that the quantum fluctuations of the phase are very small, and, conversely, the quantum fluctuations of the charge are very large. But the problem with this kind of device is that they are very connected to the environment. They are very influenced by the dissipation in the environment, and your chance of observing pure quantum effects were slim in this regime.

On the other hand, the opposite regime where you had small fluctuations of the charge and you were therefore seeing the quantization of the charge of Cooper pairs was a much more advantageous regime to make fully quantum electrical circuits. We were naturally attracted to this regime of single Cooper pair tunneling. Single Cooper tunneling was, in our opinion, our best shot at creating a device that would be a superconducting qubit. And in fact, we did at Saclay the first indirect measurement of this tunneling of Cooper pairs, and the first author in this work is Vincent Bouchiat, who was a graduate student at the time. Thus was born the Cooper pair box, which is basically a circuit that demonstrated we were in this regime of single Cooper pair and, therefore, on track to actually make a superconducting qubit. In fact, the word "superconducting qubit" appears in the article we published on the Cooper pair box in Physica Scripta. But being perfectionists, Daniel Esteve and I were thinking at the time at various methods that would allow us to do measurement on the Cooper pair box that would reveal the quantum levels of the Cooper pair box without broadening them, a situation which is called quantum non-demolition.

The problem is the following: in quantum mechanics, if your measurement is too invasive, you kill quantum effects, or at least strongly suppress them. So we were very seduced by this idea of quantum non-demolition and making the best measurement possible. I have to say we fell in a trap here that we set for ourselves, because Yasunobu Nakamura invented a fantastic way to cut corners. He and his colleagues confirmed in a time domain experiment the superposition between these two states which we had indirectly measured in our experiment. Yasu used a readout phenomenon he had studied that was far from quantum non-demolition and actually strongly decohering the qubit, but for the low hanging fruit of demonstrating Rabi oscillations, it was OK. He made an excellent strategic choice and scooped us. After our pioneering 1998 Cooper pair box paper, there is thus this landmark 1999 paper of the Japanese group of observing the Rabi oscillations of the Cooper pair box. At that point on, superconducting qubits began to be taken very seriously. A couple of years later, with the quantronium, which satisfied our quantum non-demolition ambition, we observed both Rabi and Ramsey oscillations on a much longer-lived superconducting qubit and demonstrated full control of all its degree of freedom.

Zierler:

When did you first realize, along with your colleagues, that all of this might actually be foundational to the ultimate achievement of quantum computing?

Devoret:

You see, all along our focus was really to demonstrate the principle of quantum mechanics, but applied to an electrical circuit, like the one electrical engineer commonly consider. We were not particularly focused on quantum information, and I have to say that I remain more interested in exploring the foundations of quantum mechanics than the actual work of building a quantum computer, although this Holy Grail poses wonderfully interesting challenges.

Zierler:

You've never strayed from your initial interests as a college student, essentially?

Devoret:

Correct. But other people who are eloquent spokespersons for quantum computation, like Scott Aaronson, declared that the most interesting outcome of quantum computation research would be that the quantum computer cannot work for fundamental reasons. Then we would learn a lot about quantum mechanics. I do think that in the mind of many physicists, all these experiments on quantum computation actually test the foundations of quantum mechanics. For instance, is large scale entanglement truly possible? It is not clear that it is. Even when you achieve the best quantum error correction schemes, will Nature let us have this large-scale entanglement? We may just become more and more sensitive to subtle phenomena that no one has seen yet. But, yes, starting in, let's say, 2002, it was becoming clear that more than one platform for the implementation of quantum computers were needed and our rather confidential research on quantum superconducting circuits began to be taken very seriously at that point. But I have to say that realizing the best possible quantum mechanical operations with electrical circuits, this goal was pretty clear from day one to us. As you know, John Martinis always had also this vision and he has been completely consistent with this goal. Now, yes, we can of course discuss at what stage is really quantum computation these days. I have to say that in terms of the control of a single quantum object, the progress has been staggering. I think we are very far away from building a quantum computer, but in terms of controlling a quantum system to demonstrate quantum principles, the advances made in the last 20 years have been exhilarating. I have to tell you that I am amazed by the speed at which this field, which I helped develop, progresses. On one hand, it was clear to me right from the beginning what direction we were heading to, but, on the other hand, that we have arrived at the point we are today amazes me.

Zierler:

When did you get involved in looking at phenomena of fault-tolerant quantum memory? Is that more recent?

Devoret:

Yes. It is pretty clear that you cannot build a true quantum computer without heavy quantum error correction. And then there is also the problem of doing not only quantum error corrected memory, but quantum error corrected gates. There's an immense problem awaiting us, for which the question is: how do you do this quantum error correction in a way that is practical on a large scale? At present, we have only theoretical existence proofs that this is possible. Remember that in 1995 Peter Shor stunned the world in two ways. He invented his factorization quantum algorithm, but he also showed that quantum error correction was possible, which perhaps is even a greater achievement. Everyone thought any step in the direction of error correction was impossible. Because correcting means you have to do a measurement to find a fault, and since measurement is destructive in quantum mechanics, or at least furiously invasive, can error correction ever work? But Shor did find a fantastically clever solution to this conundrum, and founded this remarkable new field which deals with quantum error correction and fault-tolerant quantum operations. That you make an almost perfect quantum systems out of imperfect parts is really something amazing. However, we still have a long way to go in showing that the theoretical proposals can actually be implemented in practice with a not too large overhead.

Zierler:

And what about remote entanglement? When did you start to look at that in depth?

Devoret:

Remote entanglement is another very important component of a quantum computer. Quantum error correction is absolutely necessary because you want to prolong the decoherence time. The decoherence time that we have in all platforms is right now very mediocre. At present, you cannot do more than at best 1,000 operations before decoherence takes over. And 1,000 operations is just too little to do anything really far-reaching. So you have to somehow create new effective qubits of which you are going to prolong the decoherence time as much as possible. And in order to do that, you are going to devise new techniques to protect the qubits dynamically. In other words, you are going to prolong the lifetime of information in your system by correcting its errors that are revealed quantum-mechanically. In a classical computer, this is achieved by taking advantage of ordinary dissipation, but this dissipation is not allowed in the quantum computer. Its computing operations need to be very, very close to reversible operation, and that excludes stabilization against noise by using friction. But another important aspect of quantum computation is modularity. You don't want to make the machine in one go without testing the parts individually. That would be very foolish from an engineering perspective. Your chance of success would be very, very close to zero. So your quantum computer has to be modular in some way, and thus be made of parts for which the quantum behavior can be tested and verified independently of the whole. And then, at the last stage, you would just assemble all the parts together and the entire machine should work perfectly. It turns out that this requirement can be fulfilled only if you master the phenomenon of remote entanglement. It has some aspects that are almost magical, since correlations between different part appear to violate the constraint of relativity. Einstein talked about this effect as “spooky action at a distance”. When you master this phenomenon of remote entanglement, then you can create pairs of entangled qubits in physically disconnected circuits. In turn, these pairs allow you to make operations that combine information in separate location. Possessing the primitive of remote entanglement with high fidelity is the key to make a modular quantum computer.

Zierler:

And Michel, we must point out that its success is an interesting concept because it's still an open question what exactly quantum computing will be good for.

Devoret:

Yes. In fact, this is the terrible problem facing our field. [laugh] And if we compare quantum computing with flying, I don't think we are at the level of the Wright brothers yet.

Zierler:

[laugh]

Devoret:

We're probably only at the level of throwing a paper plane in the air.

Zierler:

[laugh]

Devoret:

You know, flight could be demonstrated to a large crowd. You would be flying with your machine above the crowd, and almost everyone would be enchanted. People had wanted to fly like birds for ages, so there was no problem “selling” the idea of aviation to the masses. The problem with the quantum computer is that we don't know yet of any really useful task which would be provably completely overwhelming for a classical computer.

Zierler:

Well, Michel, maybe we could ask the question like this: What are some of the obvious limitations that you see with classical computing, even as powerful as it is in its current form?

Devoret:

There's a lot of looming problems with classical computers now. For one thing, they consume way too much electricity. One figure I like to quote is that each time you do 10 Google searches you have consumed the electricity needed to boil a cup of tea. Right now, in the US, of order ten percent of the electrical consumption is used for computation or related information tasks. So if you envisage a society based on information, and let's say you want to do ten times more information processing than nowadays, it's totally out of the question that you can continue to use present-day technology. There's a huge energy bottleneck in classical computation today, and this has to change.

Zierler:

But is it so clear that quantum computing, if it's viable, will be less energy intensive?

Devoret:

Well, at least in its core functioning, yes. The quantum computer uses the minimum amount of energy in its operations. In terms of energy consumption, it is by far the most efficient use of energy, computation-wise. Of course, you may say that the circuits of the quantum computer will have to be cooled, and this requires energy. Yes, but even the classical servers of today have to be cooled. Cryogenics is becoming more and more a huge part of classical information processing. The cooling requirements should become less of a problem as the quantum machine is scaled up, because, at least in principle, you can benefit from economy of scale. And maybe this is what quantum computers would evolve into, specialized classical computers running at near the quantum limit of dissipation and therefore consuming much less electricity than today's ordinary semiconductor computers. You know, from a PR point of view, the good thing about quantum computers is that you can use this name and apply it to almost anything electrical that computes because, after all, electrons are quantum particles! The D-Wave company has played with this kind of games.

Zierler:

Yes.

Devoret:

There's not a big leap when rebranding quantum computation. [laugh]

Zierler:

Michel, when you came over to Yale, of course you took on teaching responsibilities. What have been some of your favorite courses to teach undergraduates at Yale?

Devoret:

At Yale, I have created two courses, and not surprisingly, these are the courses that are my favorite to teach. One is a course that I called Mesoscopic Physics, but which now mostly teaches the basis of quantum superconducting circuits. Pretty much, it's a combination of quantum optics and electrical engineering. We could also call it fundamental quantronics. There was also another subject that I thought was missing in the curriculum. I believe it is extremely useful for graduate students, at least in our condensed matter background. It is sometimes called out-of-equilibrium statistical physics. While this is a subject that has been developed in the 20th century, it's still not taught formally in most graduate schools around the world. I have called this course Noise, Information, Dissipation, and Amplification. It is a course whose main topic is what is called the fluctuation-dissipation theorem, which, you know, was formulated in the late '40s and early '50s. As far as I know, it's still not taught as a regular statistical physics basic class in American graduate schools, but it is an incredibly useful subject if you want to work on most quantum experiments.

Zierler:

Michel, just to bring our conversation up to the present, what are some of the key things that you've been working on in recent years?

Devoret:

Well, the subject that right now interests me the most is to find efficient ways to do quantum error correction. In particular, there is something that I find truly fascinating: autonomous quantum error correction. Autonomous quantum error correction would not need any external correction circuit. The quantum machine would correct itself by employing only what is called engineered dissipation. The system is created in such a way that it automatically ships out the entropy that noise produces in its dynamics. This flushing out of entropy requires of course that you put energy in the system, like in your home refrigerator. In the classical world, there's a good analog of this type of autonomous behavior: It is the escapement mechanism of clocks. As you know, in a grandfather’s clock, the pendulum, if left alone, would stop after 100 oscillations or so, due to friction. But the escapement mechanism resupplies energy into the pendulum, and it does so in an autonomous way. There's not an external mechanism correcting the effect of dissipation by measurement and feedback. Escapement mechanisms are really great fun to watch working because they are incredibly clever devices that manage to resupply energy in the pendulum without influencing the phase of the beat of the pendulum. This phase is actually connected to what time is.

In other words, you do not want to make the pendulum slow down or speed up as you resupply the energy. How you do that in a precise way is a captivating subject all in itself. But an even more advanced question is how you do that for the quantum equivalent of the classical pendulum. The quantum pendulum lives in a world with many more dimensions than the classical pendulum. How do you organize such autonomous error correction in this more complicated case is something that we intensely pursue at Yale. It turns out that a few days ago, a postdoc working with Steve Girvin at the Yale Quantum Institute, Baptiste Royer, found an idea in this direction which I thought was quite clever, and we are implementing it right now. So I'm sure there will be a lot of progress. You know, something that has never stopped in my career to surprise me, is how clever physicists are.

Zierler:

[laugh]

Devoret:

So while I'm unfortunately far from knowing what the quantum computer will be useful for, I have full confidence that physicists will be able to build one.

Zierler:

[laugh]

Devoret:

If there's a machine that is not forbidden by the laws of physics, then physicists will make one!

Zierler:

Michel, for the last part of our talk, now that we've worked all the way up to the present, I'd like to ask a few broadly retrospective questions about your career, and then we'll end looking to the future. We'll try to predict where things are going. So first, I don't want to try to burden you and to discuss all of the awards and honors that you've been recognized with over the course of your career, but I wonder if any one in particular sticks out in your memory as being most personally meaningful to you?

Devoret:

First of all, one career step which is personally very meaningful to me is to have been appointed as a professor at Collège de France. This was a great honor. There has been so many physicists that I admire that have been Collège professors, that being part, for a time, of this institution was very meaningful to me. As I said, Anatole Abragam, who was a very important mentor, was a professor at Collège de France. But there are other people that I deeply admire like Claude Cohen-Tannoudji, who was one of the discoverers of laser cooling, and many other great physicists like Pierre-Gilles de Gennes and Philippe Nozières, who also have done fantastic work. I'm also proud of belonging to the French Academy of Sciences. That was also a very moving recognition.

Concerning prizes, I won early on with Daniel Estève the Ampère Prize. It was particularly gratifying because Ampère is also a sort of scientific hero. You can actually see one of his experiments on a table at the Collège de France, in a special room where the professors can meditate and review their notes before they lecture. This brilliant French physicist was a great experimentalist. The Ampère Prize is, appropriately, funded by the French national company of electricity. It was great to receive this prize since we had worked on something at the forefront of electricity, namely counting electrons one by one, which of course is way too precise for the computation of your electricity bill [laugh].

Then there was the Bell Prize. That was also very meaningful to me because I greatly admire John Bell, who was the first to put the worries of Einstein concerning quantum mechanics in experimentally testable terms. For me, the prize was also marking the fact that I was actually taken seriously as a contributor to the foundations of quantum mechanics, which was always a hidden ambition from the very beginning. When I started my PhD, to be involved in experiments that were meaningful for the fundamental understanding of quantum mechanics was not necessarily a good career plan, as Alain Aspect has very eloquently explained. I should also mention the London Prize, which was managed by Horst Meyer. I mentioned him earlier in this discussion. Horst Meyer had actually studied at Oxford with Fritz London. Horst belonged to this group of scientists that came to the US from Europe, in the wake of the rise of Nazism. They contributed a lot to the establishment in American Universities of a culture I admire.

Zierler:

Michel, a very broad question. You've been involved in so much fundamental discovery over your career.

Devoret:

Oh, you're kind. [laugh]

Zierler:

What have you learned about the process of discovery that makes you understand when you're really onto something and when you've hit a wall and it's time to move on to the next project?

Devoret:

Well, I'm going to tell you what I'm telling graduate students working in my group. What I tell them is that I will not be able to give them the keys to win a Nobel Prize, but I will surely teach them how to avoid doing the bad science that will prevent them from ever winning a prize. [laugh] Maybe by sheer luck and by persevering along a sort of Brownian path, swerving around pitfalls, they will end up making discoveries. If you can avoid making all the numerous lethal mistakes frequent in scientific research, and if you keep working asking new questions, changing fields sufficiently often, and if you have some luck, you will eventually find something really groundbreaking.

What is very important in this quest and rarely mentioned is the type of masochistic attitude that you need to have in your research activity. It is easy to criticize the results of others and much more difficult to criticize the results coming from your own lab. I would even go as far as saying that being masochistic is not enough, you need also to be a bit schizophrenic. Because, while you should always remain passionate about what you do and a believer that the question you have asked has an interesting answer, you have also to be sharply critical about what you find. Do you see the big contradiction here, in the superposition of being both delusional and skeptical?

To repeat myself, you have to always question the validity of your experimental move choices, your understanding of your field, be lucid about all the mistakes you and your colleagues can make, all this while believing almost childishly that the answer you are looking for is extremely important and you are going to obtain it. It is crucial to have outstanding role models if you want to withstand the resulting psychological pressure. The quality of being very inquisitive is actually becoming softer as you age because, while it is easy to be cruel to your classmates, it is a little bit harder to do it with people who could be your grandchildren – at least in terms of scientific generations. [laugh] I realized this rather recently, you see, since the incoming graduate students at Yale are now of the age of my former postdocs’ children!

Zierler:

[laugh] It's an eyeopener.

Devoret:

Yes. [laugh] You know, the mellowing with age, which is a precious counterpart of diminishing capabilities in various aspects of life, is not always beneficial to science — you have to be careful about it because you could become too tolerant to mistakes. One should not become a "grand-daddy" advisor that is too soft when control experiments are discussed. It is the responsibility of someone with experience of failures, and having some authority, to do the devil's advocate part well.

Zierler:

Michel, your motivations in exploring the quantum world have always been from the basic research side, in other words, just discovering and understanding how things work. I wonder, though, what satisfaction you may take in terms of some of the more practical applications that have societal benefit and really whose foundation starts with what you've been involved with, you and your collaborators?

Devoret:

You are actually touching a nerve here. Here is an ambition that I inherited from my post-doctoral advisor John Clarke: in my opinion, the nirvana of experimental research is to be able to both touch fundamental concepts in physics and to discover things that have practical applications. As a matter of fact, one day, my colleague and friend John Martinis put this in extreme terms when he told me that only research work that had practical consequences was of any value. I think maybe this is a bit of an exaggeration, but to take the possibility of application as the gold standard for the value of scientific research has some merit. The utmost ambition would be to understand the full spectrum of physics, from theory side to the application side. In that sense, we have this kind of model in the history of physics. For instance, William Thomson, who became Lord Kelvin, was a contributor to fundamental thermodynamics, a very abstract subject, but he was also a sort of project manager for the first transatlantic telegraph cable. Thus, he was at the same time a fundamental physicist, an engineer and an entrepreneur. I have a lot of respect for this amazing combination of talents. We cannot always accomplish the marriage of these skills, and you have to be incredibly lucky to be in circumstances where you can possibly attain it, but it is fantastic if your research can have some practical application having an impact in society. Applications also play a positive role at the level of the quality of your research, even if it is very basic, because you are forced to confront uneasy questions like the reliability and repeatability of your experiment.

Zierler:

Michel, what are some of the enduring mysteries or ongoing research problems that continue to elude full understanding in your research field?

Devoret:

In my opinion, there is something in my own research field that remains deeply mysterious. Is quantum mechanics ultimately the bedrock of physics or a theory that is just preliminary and will be revised, like the Einsteinian revision of Newtonian mechanics? Is quantum mechanics some kind of emergent effective theory, with another more fundamental theory underneath? Elasticity, that supposes that media are continuous, is perfectly fine in some regime of stresses and strain but, sooner or later, you have to take into account the fact that matter is made of atoms and the usual theory of elasticity ceases to work at their level. Quantum electrodynamics, when you think of it, does not in itself fully encapsulates electromagnetism, because you don't have only electrons interacting through photons as fundamental particles linked to charge. Quantum electrodynamics, as fascinating and as successful as it is, is incomplete, and we know it should be included in a wider theory like the Standard Model. So what about quantum mechanics itself? Let’s take string theory, for instance, or more recent models of quantum gravity. String theorists take quantum mechanics for granted. They do not question the basic principles of quantum mechanics. But are they right in believing this? I would love to be involved in new experiments that would test quantum mechanics at a deeper level.

Zierler:

Is this to say, Michel, that the foundational disagreement between Albert Einstein and Niels Bohr is provisional perhaps?

Devoret:

My feeling is that if today we could talk with Albert Einstein and tell him about the new experiments that have been done recently in the domain of fundamental quantum measurements, we would probably convince him that his prejudice against quantum mechanics -- or more precisely his impression that it conflicted locality and realism -- is misplaced. I think we understand quantum measurements much better than during his time. However, there is still this nagging question of whether these principles of quantum mechanics are absolute or whether they have some limits. It is very hard to find a direction in which they could be false, but there are now some concrete proposals, and new generations of physicists will explore them. Although there are experiments that test the fundamental principles of quantum mechanics and agree with theoretical predictions at the level of many decimals, it's not impossible that there are some limitations in the postulates of quantum theory that will manifest themselves in new, specially designed experiments, for instance those testing large-scale entanglement. Daniel Esteve is also a proponent of this point of view, I think. Large scale quantum computers are prime tools in this endeavor, as Scott Aronson has pointed out.

Zierler:

Michel, we'll end looking to the future. So, as you well know, students today, they're profoundly interested in artificial intelligence and machine learning. Where do you see your research and the things that you're interested in contributing to these very exciting and also mysterious new areas?

Devoret:

As I explained previously, my first scientific love was artificial intelligence, but I thought then it was too new a science to safely join as a graduate student. Interestingly, in the last two years, one of my graduate students, Vlad Sivak has applied artificial intelligence techniques to quantum control and error correction. Here is the idea: when you want to control a quantum system, your experiment has to learn a certain number of things that you may not be able to directly measure or even conceptualize. So we are actually applying the new machine learning technique of so-called reinforcement learning in our experiment. I think this kind of techniques will play an important role in future of experimental physics.

What was particularly interesting in this work was that it allowed us to see what was possible to do with artificial intelligence and what was not possible. There are still some valuable tricks that physicists can invent after having studied for 25 years basic math and physics, and a machine will probably not know how to create these tricks by itself for some years yet. When you work on an artificial intelligence system, you develop a precious understanding of what these new amazing techniques can achieve and what they completely miss. Maybe, one day, we will be able to efficiently teach to an artificial intelligence creative thinking based on exploitation of symmetries for example, but right now it seems very difficult. Once we do it, once we communicate to a machine these physicist tricks or these intuitions, whatever you want to call them, then we are done with the monopoly of our thinking species.

A physicist colleague of mine, David Schuster, said something that struck me: one of the few major professions that artificial intelligence will suppress are poets and physicists. [laugh] While I think it's a provocative thought, it may be correct. What's more likely is that the artificial intelligence can greatly assist physicists in tedious tasks and help them do much better experiments. I imagine artificial intelligence is going to be more and more treated like a mathematical tool, a bit like calculus.

Zierler:

And, Michel, finally, for you, time is a limited resource. For all the things that you're involved in and for however long you want to be involved in them, what's most important to you scientifically and intellectually? What is the thing that you're most interested in seeing to completion as a capstone to all the things that you've been involved with over the course of your career?

Devoret:

Well, one immediate ambition I have is to make sure that the physics graduate students I advise currently graduate honorably. [laugh] But, also, right now, I am becoming more and more interested in something that I hope will be another chapter of my career: the popularization of science. I've been quite impressed by what people are able to do now on the internet. The videos that some physicists create are amazing. And I'm particularly interested in the popularization of quantum mechanics. Because quantum mechanics is very abstract, it is very difficult to write stories that simplify it without introducing wrong ideas. I believe that the presentation of this field through information concepts can make it more palatable to the layperson.

You know, there is this ambitious program connecting information theory and quantum mechanics summarized by John Wheeler in just four words: "From It to Bit". Basically, the idea would be that you would actually teach physics entirely starting from information, or in other words, that information is the basic concept in physics. As you can imagine, there is a lot of work to be done to reconstruct physics textbooks from this perspective, since it is difficult to define an information measure that would work in a fully general context.

In any case, there is a lot of room for new approaches in the popularization of quantum physics. Quantum mechanics remains a subject that is very opaque and impenetrable, even for seasoned physicists. While there are interesting science fiction films that allude to quantum phenomena, I would like to aim at a sort of intermediate level. Writing high-quality popularization of quantum mechanics at the sophomore undergrad level or at the senior high school level, would be a worthwhile ambition of the kind you mention. I am now looking for other people with whom I will collaborate on this project.

Zierler:

It's going to inspire, hopefully, a new generation of physicists.

Devoret:

Yes, we have to infect a new generation with the amazement at the wonders of quantum physics, without boring them with the rather obtuse debates of the past generations!

Zierler:

Michel, it's been a pure pleasure and absolutely captivating to listen to all of your insights and explanation over your career. Thank you so much for doing this.

Devoret:

Well, in turn, I would like to thank you for organizing and conducting this interview. You came up with questions that were really stimulating. I realize that it must have been some work on your part, and I'm thankful for this opportunity to discuss matters that are peripheral to science, but which are still important. When we are young, we tend to think that everything we want to leave scientifically is contained in our publications. As we age, and are questioned by younger people on our choices of career paths, we become more interested in the sort of record you are producing.

Zierler:

Well, it's my pleasure.

[End]