Robert Schoelkopf

Zierler:

Okay. This is David Zierler, Oral Historian for the American Institute of Physics. It is September 30th, 2020. I am so happy to be here with Professor Robert Schoelkopf. Rob, thanks so much for joining me today.

Schoelkopf:

Hi, David, it's great to be here.

Zierler:

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

Schoelkopf:

Right, so I am the Sterling Professor of Applied Physics and Physics at Yale University. I am also the director of the Yale Quantum Institute. Yeah.

Zierler:

Now, the Sterling name, you see Sterling a lot at Yale. Is there any particular connection with physics or with your chair in particular?

Schoelkopf:

So, Sterling is our sort of special category of endowed professorship. So, there are Sterling Professorships in all the various fields. You know, I am not enough of a Yale historian to give the accurate description of Sterling, but he was a major-- his family was a major, supporter of Yale. The library is named after them and so on. And so, there are a handful of Sterling professors in a variety of different disciplines. I think I'm the first Sterling Professor of Applied Physics, actually.

Zierler:

When did you create the Yale Quantum Institute? What year was that?

Schoelkopf:

So, we've been very active in quantum research at Yale for quite a while. The institute's an idea that's also been around for a while. I think we launched it officially about five or six years ago, and it occupies a sort of small amount of space on the fourth floor of 17 Hill House on campus, which is sort of where we hold seminars and interactions between people researching quantum on campus and so on.

Zierler:

And is it a separate program? Is it housed administratively within the Department of Physics?

Schoelkopf:

Yeah, Yale Institutes are sort of our mechanism for trying to promote interdisciplinary research and initiatives. So, it's not the same as a department. It's not necessarily under any particular department. But for instance, the institute can work with different departments to kind of collaborate and hire people in. So, you know, in this burgeoning field of quantum computation. For instance, how does one find, attract, and appropriately house in a university someone who is working at the interface of physics and computer science that doesn't maybe fully fit in a traditional computer science department and is too much of a computer scientist to be considered a traditional physicist—and so the institutes are one way to kind of bring people together from different departments to kind of support and recruit those types of people. And so that's one of the reasons they have the institute. And then in general, you know, what it's also doing is trying to create a community and more interchange between people who are already on campus in perhaps different buildings and different departments, but are interested in some of the same kind of intellectual topics.

Zierler:

Rob, of course, right now life in applied physics and experimentation is a lot more complicated than your colleagues in theory. Can you talk a little bit about how you're doing your best to keep up the labs and the research and all of the physicality that goes into applied physics and experimentation during COVID?

Schoelkopf:

Yeah, it's certainly a challenge. There are many aspects there. I mean, to some extent we are in a good position where we may have some experiments that are these days fairly automated. And even pre-COVID, people would sometimes log in to check on how their data collection is doing from home. I remember sitting in the meeting at the APS meeting next to one of my students who was plugging away on his phone. I was like, “What are you doing?” He's like, “I'm taking data.” [laugh] So we were off at a conference, and he was still taking data back home. The tricky part is, well, how do you debug it? Or what do you do when you need to change something out? And so we've now, for several months, been able to go in in sort of limited density and have people work in the lab and do things like that. So we're regaining some of the productivity on these things that weren't kind of in full flight. But it's been challenging. I think, you know, it's hard to schedule these things, and it's always of course much more fun and much more productive to work side-by-side at the bench with someone, and you say, “I still can't fit this in,” and they point out that you're using the wrong size screwdriver or something, right? So just basically talking out loud as you're experiencing the problems, you often go, “Oh, okay, right. I know what I'm doing wrong.” And so it's a little hard to do that over FaceTime or Zoom or what have you. I think the other concern that I have, which will grow as this pandemic drags on, is what do you do about training new people?

Zierler:

Yeah.

Schoelkopf:

Right? And so that's kind of, still experimental physics is a little bit of an apprenticeship system, right? You say, “Well, come on in, and tag along.” And, you know, “Look, here I'm doing this and that, and you may ask questions about why I'm doing it that way. There's a reason we do it that way.” And that's pretty hard to kind of codify in a PowerPoint or, again, dictate to someone and have them really get it just over an interaction. So, I'm a little worried that there will be a few years of students that are a bit delayed in their learning of the practical side of science.

Zierler:

Has that influenced the number of students, graduate students, you've taken on this year?

Schoelkopf:

No, I don't think so. I mean we also had already admitted our class for this year, you know, pretty much before all this happened. And I think everyone's having to adapt, and we will move on. Yale's been pretty good, Connecticut's pretty good. We have essentially no cases here, but people are still being very, very cautious, which is good.

Zierler:

Well, that's probably why there are no cases. [laugh] The two are definitely connected. Well Rob, let's take it all the way back to the beginning. I'd like to hear about your parents first. Tell me a little bit about them and where they're from.

Schoelkopf:

[laugh] Yeah so, my dad was born in New York. My mom in York, Pennsylvania. Pennsylvania Dutch country, as they would say, of course, but that's also a German heritage. And they-- originally, I was born in Manhattan, and we lived in Manhattan early on, and then later on we moved to the suburbs, to Chappaqua, New York. My parents were art dealers.

Zierler:

SoHo?

Schoelkopf:

No, they, originally the gallery was on Madison Avenue. 69th and Madison. And then later on, they moved down to 57th Street, which was a big gallery row as well.

Zierler:

They were partners? It was a shared business, your parents?

Schoelkopf:

It was a business that my dad started, and over the years my mom got more and more involved, as my brother and I got a little older. She sort of took over a lot of managing the business. My dad was, you know, very highly regarded and well-known for his eye and for his connoisseurship. Maybe not as much for his personal skills and his business thing, and my mom was like, you know, way off-scale on interpersonal things. It only took a ten-block cab ride to know the names of all the siblings of the cab driver.

Zierler:

Right.

Schoelkopf:

Every time I rode a cab with her, right? So, she was very good as the sort of front of the office person, as it were.

Zierler:

Rob, it sounds like you probably got a front-row seat to upper crust New York as a kid.

Schoelkopf:

Uh, yeah, to some extent. There were always interesting things going on and sometimes I got to go with my dad to visit artists in their studios, or-- I don't remember how old I was. I remember one time, because my dad represented the estate of Gaston Lachaise, who was a pretty well-known sculptor, and most of his works are bronzes. He took me to the foundry, I think in Brooklyn or Queens or somewhere, where they cast these life size or larger than life size bronze things. And it was a little bit like a steel mill, so that was a pretty cool thing to do. And then, you know, right, lots of interesting discussions around the table. None of them about science.

Zierler:

Did you go to public school or private school in New York, and then in Chappaqua?

Schoelkopf:

Moved to Chappaqua when I started fourth grade and went to public schools in Chappaqua. First couple years, grades one through three, was at the Trinity School, upper east side in Manhattan.

Zierler:

And when did you get interested in science? Were there any family members beyond your parents who were into science or were influential on your thinking?

Schoelkopf:

It's one of these funny things. There isn't really much science in my family. And I can't really say why I'm interested in science, but I remembered being interested in science since I was five years old or something. And as I was growing up, I was sure I wanted to be a microbiologist. I think because I got one year a microscope for a birthday present or something. And then another year I wanted to be a geologist. And for a while I really loved Jacques Cousteau and wanted to be a marine biologist. Still sad that I didn't do that, and my field work isn't in like Aruba or something. Sort of messed up there. And went through all these things until sort of finally found my way to physics. And you know, the thing I say sometimes kind of half-jokingly is, it seemed to me growing up that my dad knew everything about languages, art, literature, history. He spoke like four languages. He was kind of clueless with his hands and he didn't know anything about science and like there was my lane to shine.

Zierler:

Were you good in science in school? Were you a stand-out student in math and science in middle school and high school?

Schoelkopf:

Yeah. Maybe not such a stand-out in math. Always had a little trouble with math as a physicist, actually. But yeah, I sort of-- for example in high school, by the time I got to my junior year, I'd already taken every science course offered in the school. Right? As well as the welding shop, wood shop, and all of those. And I think I took welding shop like three times, which was not typical for one of the students also in AP classes. And then, you know, I was very lucky that at the high school there were some great teachers that allowed me to do kind of more advanced things or independent study projects. I had a chemistry teacher, Roslind Zook, who mentored me and was really wonderful. We also had at the high school, surprisingly, a little observatory. There was a small dome out on the football field, or on the corner above the football field, with a ten-inch reflecting telescope. And I spent quite a bit of time and late nights in that telescope taking star images. I built a little spectrograph that I could use to identify stellar types. You know, just simple sort of slit thing, you could disperse and you could easily count the number of lines and figure out which ones were the giant stars and which ones were the other ones. So that was a lot of fun and was a really good environment. It was a great public school in Chappaqua.

Zierler:

Rob, were you thinking about physics programs specifically when you were thinking about what colleges to apply to?

Schoelkopf:

Yeah. I guess in this serial love of various science disciplines, kind of came to physics last, and I'm not quite sure why it's the one that stuck the most. I guess I always liked gadgets and things. But also physics had a little bit, this mystique. You know, it's the one you can't get to until you've done all the other stuff and you have enough math and calculus and all of that. And it's often viewed as the hardest one, right? You know, so you still get this frustrating thing where you talk to people and you say, “Oh, yes, I'm a physicist.” And they go, “Oh. Oh that was really hard. I hated that in high school,” or... But I think by the time I was sort of finishing high school, I really wanted to do physics, and so I was already sort of thinking I would be a physics major when I applied to colleges.

Zierler:

Now, in terms of working with your hands and not being so strong in math, was applied physics, experimentation, was that sort of your pathway even from the beginning?

Schoelkopf:

Yeah. I mean, I liked to, you know, build stuff, even as a young kid. So, I built a tree house, which slept three people and had a roof and a trap door and was like 30 feet up in the air and you had to get there with a grappling hook, and-- I built a forced air wood fired furnace out of some old bricks and a fan I stole from somewhere, and like made castings in lead. Which was probably not very safe or very smart. [laugh] Built go karts and things like that. And so yeah, I sort of always wanted to be an experimental physicist, I think. And I really wanted to have a lab. I just knew enough even at an early age to like-- it'll be really cool to have a lab and do your own thing.

Zierler:

What schools did you apply to as an undergraduate?

Schoelkopf:

Oh, quite a few. Wow, we're really going to go through everything, aren't we? Gosh, Princeton, Yale, Chicago... Stanford, Caltech for sure... Didn't apply to Harvard, though. Not sure why. Maybe Brown? I don't remember. And I had applied early to Princeton, but was deferred. And eventually got into several schools and went to Princeton for maybe not a very substantial reason, but just like Princeton had this reputation as being so great in physics because you know, Einstein was there.

Zierler:

Right. [laugh] That works.

Schoelkopf:

Yeah, yeah.

Zierler:

And so, you declared a major right away? In physics?

Schoelkopf:

Yeah.

Zierler:

What kind of courses did you find you were best at? What was most enjoyable for you as an undergraduate?

Schoelkopf:

Well, let's see. I really liked, there was an electricity and magnetism course, not surprisingly, that I really liked a lot. You know, one thing, I think one of the reasons I wanted to go to Princeton, one thing that's really great there, is there's a fair amount of undergraduate research. I really wanted to go to Caltech, because like I heard that was a really great place, and also allowed you to get involved in research early on. I sort of was a little impatient. Like, I don't want to do classes, I want to like invent stuff and create stuff and learn new things. Even discover things. My dad said, “No, you have the opportunity to go and get a liberal arts education. You can go to Caltech for grad school. If you decide to go to Caltech for undergrad, I'm not paying for it, but you can go anywhere else.” So indeed, I went to Princeton as an undergrad and then grad school Caltech. But at Princeton, they have these junior papers, which are sort of pretty substantial, independent research thing, even though you're only a junior. And then a senior thesis. You know, everyone at Princeton does a senior thesis, which is a big effort. It's sort of, they only have three courses other than the thesis during your senior year. And that's a really great thing, and good opportunity, and I liked those very, very much, and did some in quantum mechanics and some in cryogenics and sort of foreshadowing of a lot of things to come.

Zierler:

Rob, did you become close with any professors at Princeton in the physics department?

Schoelkopf:

Yeah. I think two of the professors that supervised my independent research projects there. One was a guy named Steve Boughn, B-O-U-G-H-N, and then there was also Jeff Kuhn, who have both moved on, but they're not at Princeton, but they're still around, I think. And they were very good. They were-- was Steve part? I think they were both part of what at Princeton was called the Gravity Group. This was a sort of loose affiliation of several professors, including for instance Dave Wilkinson. One of the pioneers of cosmic microwave background types of things. And so, these were some pretty neat labs where people did things like build a Mazer amplifier that would be very sensitive that would go on a balloon with horns to actually detect and look for signs of the cosmic microwave background and things like that. One professor I had who was really great—I think I had Joe Taylor for freshman physics, actually, who was well-known as the discoverer of the binary neutron star system—pulsar system. And later on he won the Nobel Prize. And then another person who I interacted with a little bit, but left a big impression on me, was Bob Dicke. And he was already super well-known. I mean I guess in a way the Wilkinson and the Gravity Group was also like sort of under his umbrella, or you know following in his footsteps. And so, he was already, I think, emeritus and stuff, but was still open enough that a junior undergraduate could come in and discuss, like, “Hey, I don't understand what the vacuum fluctuations of the electromagnetic field in a microwave cavity are about.” And I remember several discussions like that that we had. And he was a reader of my junior paper and my senior thesis, I think, as well. Even though he wasn't a direct mentor.

Zierler:

One of the possible advantages as an undergraduate of going to a place like Princeton as opposed to a Caltech or an MIT is that you can really avail yourself of a more liberal arts education. Did you do a good job of taking advantage of that? Or did you try to hang out as much as possible within the world of physics?

Schoelkopf:

Oh, no, I mean you're of course exactly right, and I think that's what my dad was on about in his steering of things. No, there were a lot of interesting classes and things that one was forced to take because of the distribution requirements, but when would you ever get a chance to do that again? To read great literature and hear it discussed, or... I remember, there was a great course, which was about architecture of gothic cathedrals and things like that which had a little physics in it. Or at least, I found the physics in the evolution of the flying buttresses to help support the things they were building that they didn't quite understand the force laws for. But yeah, no, I think that was a nice experience. And I would do it again that way, I think. There's also maybe a thing that one gets from a liberal arts education which I think is under-appreciated in its importance in science, which is being able to write clearly and well and to be able to speak and present clearly and well. And you know, I think I see this to this day. As a professor running a research group, and these days with the size of our group, like when we go to the March meeting and everyone goes and give their ten-minute talks, there are like 20 talks I have to research. And I'm very thankful for those people who have had this broader education, and already know how to construct a sort of ten-minute talk with a thesis and a problem statement and a conclusion. And that's something that's pretty important in science. Doesn't matter if you do it in the lab, if you don't write it down in the lab book, it didn't happen.

Zierler:

Right. [laugh]

Schoelkopf:

And if nobody reads the paper, sort of also doesn't matter. And the worst thing you could do is have somebody read the paper and say, “Yeah, so it was about what again?” And so really kind of saying clearly why you've reached the conclusions you've reached and what it's all about, that's pretty important. And I don't know exactly how it is that a liberal arts education gets you there, but it seems like it correlates. Maybe it's the people who go that way, or already have that skillset a little bit in their makeup, in their personality. But maybe also it helps to practice.

Zierler:

Rob, between your senior thesis and the professors that you sort of naturally gravitated toward, when you were thinking about graduate school, how well-defined were your interests in the kinds of physics you wanted to pursue in graduate school, and did that influence the kinds of programs you were thinking about applying to?

Schoelkopf:

Right. Yeah, well I was a little bit—So, I was a little unsure of where I wanted to go. And actually sort of near the tail-end of my undergraduate, I was kind of debating whether I wanted to go into more of an engineering field or into physics.

Zierler:

And engineering so much that even pursuing a career in industry was something you considered?

Schoelkopf:

Yeah, it was something I considered. Or you know I said, well, maybe I should try some other things before I go straight into graduate school. So, I also had like a sort of lousy senior year with a terrible bout of mononucleosis and spent way too much time in the infirmary and stuff, and so I was a little bit sort of like out of-- I didn't do the whole GRE and applying to all those things. So I ended up, at the end of my undergraduate, taking a job at NASA, at the Goddard Space Flight Center. Because I said, “Well, let me see what these kinds of things are all about.” And maybe, again, I wanted to know a little bit more about building things. [laugh] And you know I was also interested in space and kind of the NASA mission quite a bit. And of course, my senior year in 1986 was the year of the Challenger disaster.

Zierler:

Yeah.

Schoelkopf:

I remember watching that on TV with people at Princeton. And so yeah, I decided to sort of put graduate school at least temporarily on hold and took a job as a cryogenic engineer, I think was the actual job title. I was working with a group of people there in sort of a division that does space astrophysics. So these are the kinds of people at NASA who build COBE or space telescopes or things like that. And in particular, this group was developing some very, very sensitive, very low-temperature detectors, which was kind of already something I was pretty interested in, and so I went to go do that for a couple of years, which was also a great experience. I worked with a couple of, several wonderful people there, Including Harvey Moseley and John Mather. So, I wasn't involved directly in the COBE spacecraft thing, but it was kind of coming together at the same time. So--

Zierler:

You got to see it at a point when a lot of people were doubting whether this was going to actually turn into something.

Schoelkopf:

Yeah. And I mean this story may be apocryphal, but I think I remember somebody coming in and saying to Harvey or John or somebody like, “I think this bolt is sticking up 1/16th of an inch too much. Here's the blueprint. Is it okay to still launch?” You know? So, I thought that was pretty neat. And, you know, at Goddard we got to see some things, you know, giant environmental test chambers, platforms where you do the acceleration shake test of payloads to make sure they're not going to fall apart [imitates rumble noises] on the shuttle launch or whatever. And there was also this Supernova 1987 A that went off in that era. And so this group, which did a lot of x-ray astrophysics, would kind of put together a quick response sounding rocket mission, flew out of Australia to try and get some x-ray spectra shortly after the supernova went off. And so, I helped design one of the cryostat-- just a liquid nitrogen cryostat for that and got to go shake-test it and see what it would do. And those guys treated me very well, even though I kind of had a bit of a ill-defined job function. And I was a little bit of a floater. In a way, I was in the physics group, but kind of I think informally, what they wanted me to do—and I spent a fair bit of time doing this—was kind of back of the envelope estimates to make sure that some of the things that the other engineering branches—who were very specialized, but maybe not as trusted by the physicists always—that sort of things held together or made sense or that we could look into some alternative approaches that the engineers would say, “No, no, that's kind of too wacky for us, but we could sort of look into these things and see.” And you know what I learned after a couple of years was, it seemed like the physicists had the most interesting jobs.

Zierler:

Yeah. And so that was probably formative for you to think about getting back on track and looking at graduate school programs?

Schoelkopf:

That's right. So that kind of, after a couple of years of doing that, I decided to reapply for graduate schools, or apply for graduate schools and go back.

Zierler:

And did that re-focus or change whatever trajectory you might have been on? I mean, being at Princeton, you know, physics undergraduate, there was probably a strong natural momentum towards graduate school, even by the time you were a sophomore or a junior, right?

Schoelkopf:

Yeah.

Zierler:

So, I guess my question is, how might the experience at NASA have changed whatever momentum you were building toward graduate school during your time at Princeton?

Schoelkopf:

Yeah, well I mean maybe what it helped do a little bit for me, I think there are many different styles of science, and of physics. And maybe it took me a little while to recognize what my style was, or that it was a good enough style. [laugh] So there were some really smart people at Princeton, and probably they also studied harder than I did as an undergrad. And like my forte is not quickly whipping off multiple pages of math, I'm a little bit more intuitive and a little bit-- and I have, I think, a pretty good practical side, or common sense. Maybe it comes from having burned myself with the sand casting experiments. You know, the how many times does a glass blower get burned? Once.

Zierler:

Right. [both laugh]

Schoelkopf:

Right? And then never again. So... I think that just sort of gave me a little bit more focus of what it is that I wanted to do, and it assured me that like physics wasn't all about being able to do four-dimensional tensor geometry as well as Einstein—which I cannot do. And so finding that there was a lane in physics that I actually believed I had enough ability in-- and also maybe, I really got interested in this area of astronomy, which I had always loved, and devices, right? And so, one thing that really got instilled in me at Goddard was they didn't have an agenda about what science or what technology actually they were going to use. It was like, “Hey, we want to count single photons with energy resolution of a few EV coming from space. No one's ever done that. What are all the ways we could do that? What do we know about superconductivity? Properties of materials at low temperatures? Optics?” Anything was fair game that you could put together, and that kind of ethos of, it's not that we're studying this particular thing for this particular reason. It's like, I want to build a widget so what is everything that mankind knows that can help me make that widget? That's a pretty interesting game, and that I went to Caltech because I wanted to work with some of the people there doing detector development for astrophysics and things. And that was a good--

Zierler:

And who was doing that at Caltech at that time? Who was really leading the charge in detector development?

Schoelkopf:

Well, there were a few groups there. My advisor was a fellow named Tom Phillips, and my sort of unofficial advisor is Jonas Zmuidzinas, who kind of moved-- I think he was a Caltech undergrad, did his PhD at Berkeley, and then moved back to join the faculty as a real hotshot. But kind of came to Caltech and joined the faculty a couple years after I was there. So, I couldn't actually get to join his group just yet. And you know, they were pioneering this area of submillimeter astrophysics. So, it's an interesting waveband. It's radio-- Sometimes I say for short that I was trained as a radio astronomer. It's radio astronomy, but in the range where typically radio techniques don't work anymore, right? So, there were telescopes and optics in infrared. And then there's long wavelengths in radio. And then this range of hundreds of gigahertz to terahertz is kind of an awkward in-between, where the technologies were for many years very, very hard and you kind of had to nibble at it with some techniques from either side. And so they had pioneered-- Tom was one of the first people to build a SIS mixer, a superconductor tunnel junction-based mixer, because one of the first things you have to do if you want to measure a super high frequency signal is quick convert it down to a lower frequency where we can actually process it, amplify it, all that kind of stuff. And so that was in this same vein of okay, what can we do that's possible? Why did they use the superconducting tunnel junction? Well, earlier on they were using other things like Schottky diodes or whatever, and SIS just had a better, sharper non-linearity, and the devices were faster, better products, things like that. So, you could go up, up, up to a little bit higher than radio and get into the submillimeter. And so that was a very interesting thing. And this was a kind of a cool group where the typical path was you spent two or three years doing lab experiments and developing some new piece of technology, and then you built an instrument, took it to the telescope, and observed. And there's this funny thing in the community, there's sort of development—which means the gadgets—and then there's science, which is the photons from space that get published in APJ, in the Astrophysical Journal. And I actually kind of gravitated more and more towards the devices side and a little bit less towards the astronomy. I had actually been very-- I was very excited about x-ray astrophysics, and when I was at NASA my first paper, I think, I published was about x-ray pulsations from a neutron star. And the technology of the submillimeter, I really liked. The astrophysics was a little harder. It's challenging because you're looking at things like proto-stellar regions or planet formation, and at the distances we are, it's very hard to actually resolve the structure on the relevant scale. And these are complex systems and one of the things that was very neat that you could do with these SIS technologies. You could actually measure the chemical composition of the gas in these star-forming regions, right? So, one of the main things you observe in the submillimeter is vibrational or mostly rotational transitions of various light molecules. And so, you can do things like map out where the carbon monoxide is in this proto-stellar region. You can, by knowing exactly what the frequency is, of course you can measure doppler shifts and you can get some sense of what's going where. But then there's a stage of interpretation where the main thing you make is like a map. And you say, “I think the star's there.” And you can't say plus or-- you know, at the 4-sigma confidence level.

Zierler:

That's not going to work?

Schoelkopf:

It's hard. I think, you know, I still sort of say this to my students. Don't show me a pretty color scale plot. Okay, that gives me the lay of the land. Now show me a line cut. I want to know, where is the continental divide, not that there are mountains in the West. Exactly how high is the highest mountain in the-- you know. And so, I think it's sort of challenging when what you're left with is kind of a somewhat blurry map of the object and you want to infer things that come out. And there were, seemed to me like sort of arguments in the field where people would use their similar devices to make maps and then they'd kind of interpret them a little bit differently. And didn't seem like you could really falsify the other guy's theory. That the disk is going this way versus this way. So, I just didn't somehow love-- maybe I also was doing a technology project that never, I never quite got to a device that went to the telescope. I went to the telescopes a couple times and observed a few times, but my thing was a bit more speculative technology, which didn't quite make it all the way to the prime focus of the telescope. So, somehow, I just didn't love the astrophysics of the submillimeter--

Zierler:

Yeah.

Schoelkopf:

--as much. And so, by the time I was finishing my PhD, I thought, you know, I really want to do more of the physics of the devices. I want to develop the devices. But I want to understand the physics of that more and get into the physics of that a bit more. And you know, my PhD work was all about Josephson junctions and so on, and there's this really amazing thing about Josephson junctions that you have a current voltage characteristic that has fundamental constants in it, right? So, one of the really neat things you can do with these Josephson junctions, you can have a trace on the oscilloscope, it'll apply a certain frequency and you measure off a voltage which is given by H-bar omega over E. Or 2E depending on if it's single electron or Cooper pair tunneling. And seeing a fundamental constant right there on the oscilloscope? That was like, "Whoa, that's really cool." And I'd always been really fascinated by the quantum mechanics and actually in the detector development, there wasn't as much quantum mechanics there. And so, after grad school I kind of wanted to go do a postdoc, and so I ended up going to Yale to work with a guy named Dan Prober, still here in our applied physics department, who had sort of this very similar makeup. He had done detector development and he hired me to do a project that was NASA-supported that was a new kind of technology for terahertz even higher than submillimeter astrophysics or Earth sensing applications. But he had also done a lot of really neat experiments on non-equilibrium superconductivity. He—along with another guy named Rich Webb—was one of the competing group, and was one of the first people to observe sort of electron interference in nanostructures. So, sort of the Aharonov–Bohm effect in the normal metal ring and things like that. And so, I thought, oh, well this would be a really neat way to get into, what at the time we called mesascopic physics, which was kind of the quantum behavior of things in small structures. And so yeah, that's kind of the trajectory that graduate school put me on.

Zierler:

Rob, I want to, before we get too far afield, I want to ask a few more questions about Caltech.

Schoelkopf:

Sure.

Zierler:

First of all, I mean initial impressions, of course you're an East Coast guy through and through, even from your parents. Through school, through NASA. What were your impressions when you got to Pasadena, you know, broadly in terms of what life was like as a graduate student at Caltech, and perhaps more to the point, did you detect any ways generally that physics was done differently at Caltech than it was at a place like Princeton?

Schoelkopf:

Hmm. It's a good question. Yeah, they are very different kinds of institutions in terms of size, breadth, of course. You know, the physical environment, the eucalyptus trees and the mountains in the back versus the oaks and the Gothic architecture. Also, I guess, you know, at Princeton, there's an amazing amount of energy and attention that the undergraduates receive, and at Caltech, I always think of it as almost an inverted pyramid. It sort of feels like there are more graduate students than there are undergraduates, and more research scientists than there are graduate students. More professors than there are research scientists. It's not quite like that, but it still is a very different kind of makeup to the place. It was... I mean, Caltech is quite idyllic, actually. It was a very nice place to be a graduate student and do things. And it was, as you say, it was like one of my first experiences with the West Coast. And actually, just before I went to Caltech, I had a couple of friends of mine and we took a couple months off in the summer, and we drove out to Seattle and then rode our bikes back to New York. Which was the first time I had ever been west of the Mississippi.

Zierler:

Oh wow.

Schoelkopf:

And so, the West Coast had a lot of interesting appeals with the mountains and skiing and beaches and all that kind of stuff. So, it was a nice place to be. But now I'm trying to think, what are the other things that were different about it? I mean the other thing we had there, which was pretty important, was things like the Jet Propulsion Lab. And so, our group at Caltech collaborated quite a bit with JPL, and actually to make the devices I used in my thesis, I went up to JPL and spent most of my time for a couple of years kind of in their clean rooms and stuff fabricating devices and all that. So that was also, I mean I guess Princeton has other things like the plasma physics lab and so on, but there was just a, seemed like there was a lot of support. There was a good environment for doing research.

Zierler:

Who was on your committee?

Schoelkopf:

Mmm... I believe it was Tom Phillips and a professor named Mike Cross, who's a condensed matter theorist. Who taught a great course on chaos that I took. And a woman named Nai-Chang Yeh. I think they're all still there. I believe Tom is emeritus and is retired now. And I think that was it. I think there may be only three people on the committee. And we had these defenses which were private. It's very different from the way we do it here at Yale, where you have everyone in the department is invited, at least for the beginning part. There's of course a closed-door session where the professors have to prove that they're still smarter than the candidate. But Caltech it was just sort of this private thing, which felt a little bit like a let-down actually, but [both laugh]--

Zierler:

And administratively, given your research interests where there's astrophysics, there's the physics of devices, there's astronomy. How did that shake out in terms of just physically, the buildings that you were in on any given day?

Schoelkopf:

Oh. Yeah, say more? You mean like... what do the labs look like?

Zierler:

I mean because there's divides, right? I mean Kellogg is a different place than the physics department. And you have the device component as well, so I'm just curious where you were situated because you could be theoretically in a few different places.

Schoelkopf:

Right, right. Yeah no, I see what you mean. Yeah, it's an interesting thing. So the labs of our group were in Downs. And as I mentioned, a lot of the devices were physically made at JPL. And then Tom, you know, had a NSF-supported facility, the telescope. The Caltech Submillimeter Observatory, on the summit of Mauna Kea that was sort of the main thing the group did. So, there were a bunch of engineers and research scientists at Caltech that helped support that, but then there was a whole team at the telescope doing that. So, indeed it was kind of multi-local. You know, in the sense that sort of students and postdocs or whatever would design and make new instruments, which started at JPL, then they were sort of shaken out and tested in Downs lab, and then crated up and sent out to the telescope and attached to the telescope and used. And then some of the students would do more observational things. They would just go to the telescope and use the facility instruments and then-- like one of my officemates was doing a survey of all the molecules in several different objects, and so he took these long spectra with many, many, many spectra lines and would work with actually some computational chemists or... maybe they weren't computational, but molecular spectroscopists, I guess, basically. I think who were also at Caltech, that sort of had a catalogue of what all the, you know, even just calculating all the frequencies that some, you know, like methanol or something is, ethanol is... It's incredibly complicated by the thing, how it twists and tumbles and it's asymmetric and it's basically in every possible frequency you can think of. And cataloguing what are all these lines, and which molecule do they correspond to is actually quite a task. So yeah, it was kind of, I mean in that sense it was a little bit tricky, maybe, in the group, because it was easy to get lost a little bit.

Zierler:

Right.

Schoelkopf:

It's hard to have a cohesive thing when people are in multiple locations.

Zierler:

And Rob, just to zoom out a little bit, which graduate students rarely do, looking back at your dissertation, what were some of the broader research questions that were being asked in the field at that time, and how was your work responsive to those broader questions?

Schoelkopf:

Right, right. So, I mean one of the big questions was, of course, let's say on the detector or on the technology side, the question was, what's the best way to do a measurements? What are the ultimate limits with which I can do a measurement? And you know then, how can I employ other kinds of phenomena or physics to make that measurement better or faster, broader bandwidth, something like that? So, and this is another place where the sort of quantum comes in that's pretty interesting. For the Josephson junction detectors, which isn't using really the-- My thesis was about using the Josephson effect to do the down conversion from many gigahertz, or hundreds of gigahertz, to gigahertz. The standard in the lab and in the field was and remains actually using the single particle tunneling in a Josephson junction, the so-called SIS mixer. So it's just a slightly different range of parameters for the junction, and a slightly different phenomena, it's just the tunneling of electrons instead of Cooper pairs. And interestingly, and this had already been developed by the time I was in grad school, the sort of-- People built these SIS tunnel junction detectors. My understanding is the initial motivation was very crude, was like, “Look at that IV curve. It's got like a step function in it instead of a rounded turn-on like a Schottky diode, so that's going to be better and more efficient to mixing.” Which is a correct intuition. But then pretty quickly, people observed and realized that there were these quantized features in the IV characteristic, and there were-- John Tucker and Mark Feldman had a paper where they wrote up kind of a theory of mixing and frequency modulation inside these devices, based on a microscopic model of the tunneling of electrons across a superconducting device. And you could work out stuff that matched up pretty well with the experiments and prove that these devices should be quantum-limited amplifiers. So they can measure both the amplitude and the phase of the electric field, or both quadratures of the electric field. And that's important because you want to preserve that and down-convert that so you can get a few gigahertz-wide spectrum, and find all these lines and see the frequency shift and all that kind of stuff. For that you don't want a photo multiplier, right? You want something which actually preserves the wiggles of the electric field so you can really analyze it and get very fine resolution. And in principle, you can do super high-resolution spectroscopy, because you're taking whatever the signal is that's coming in, and you're multiplying it by a delta function local oscillator, right? And then you can measure for as long as you want the down-converted signal and get as fine a frequency resolution as you like. But to be able to show that this thing reaches the Heisenberg limit, right? Or should reach the Heisenberg limit. And during this era, people were starting to build these SIS detectors that got to the Heisenberg limit. I mean in fact in a funny little bit of serendipity, Dan Prober at Yale, who would later be my postdoc advisor, was one of the people who collaborated with Paul Richards and using a Tantalum junction, they sort of showed in the lab at 30 gigahertz that they could get this close to the quantum limit. And during this era when I was at Caltech, people were starting to build a whole host of different frequency channel receivers that were within a factor of a few of this Heisenberg limit from the top of the atmosphere to the tape that the astronomer would take home with them. Provided the weather was good enough.

Zierler:

Right.

Schoelkopf:

Right? And so that's pretty amazing, right? And this kind of, what's the limits on the precision or ability to make a measurement? That's something that like I really love and kind of carries over a lot into my later work as well, right. So yeah.

Zierler:

What were some of the theoretical breakthroughs or limitations at the time that were sort of most relevant for your dissertation research?

Schoelkopf:

Yeah. So as I mentioned, there were these two ways you could use a Josephson junction to do this heterodyne detection, or down-conversion of the frequency. And the SIS mode was working well and was well-understood. And people had experimented with the Josephson mode of using these things earlier in the field. And it worked, but it had some aspect—they called it excess noise. It was noisier, didn't reach the quantum limit. And the literature was very unclear about what was going on.

Zierler:

Yeah.

Schoelkopf:

And what I ended-- The rabbit hole I ended up going down on my dissertation was, well what is this excess noise, really? And part of the hope that started us on the project, this was also—So, I started grad school at Caltech in ‘88. 1986 was the advent of high-Tc. Make a long story short, to go up to higher frequencies in the submillimeter, you want a larger energy gap with your superconductor. High-Tc could be a way to do it. But no one could make, and maybe still has not made, high-Tc tunnel junctions of the quality of an iovian [?] tunnel junction that are used in an SIS device. So part of the sort of thesis of my thesis was, let's try and revisit Josephson mixing, and maybe we can make high-Tc devices that will get us into the terahertz. But what about this excess noise? And then what I ended up doing for the dissertation was not the high-Tc stuff, which really was kind of… So there were a lot of people doing experiments on high-Tc mixers in the millimeter and submillimeter there, which to my mind, were not super high quality work. Like people would stick a piece of high-Tc in and you know if you drive anything really, really hard with a local oscillator, like it'll have some non-linearity, and you'll get some mixing. And so, I used to joke you could just put a spoon in the wave guide and drive it hard enough and you'd probably get some mixing. And that was sort of where the state of the materials and device technology was at the time. So, I ended up making kind of explicitly-optimized for the Josephson mode devices using the conventional technology. Which wasn't going to be a breakthrough in terms of frequency, but which I thought we had to solve this problem of, you know, is this always going to be just too noisy, though? And I ended up doing a combination of device fabrication with some special aspects for that particular use, and building a receiver and a test cryostat in the lab. And actually, a fair bit of numerical simulations of kind of the dynamics of this driven Josephson junction to try and understand what was going on, and you know one of the things that was talked about was, like, oh they're somehow chaotic dynamics or something. And so, I did end up spending a fair bit of time thinking about chaos in driven systems and unfortunately, there isn't, as far as I can tell, there still isn't a prescription about, like, “Do this, do that, and you're guaranteed chaos can't exist.” There are lots of things that sort of say, “Yeah, look, it's over here. It's over there!” But saying, “You're safe in here.” Like, not so much. Which would— Probably chaos is bad if you're trying to build a pure frequency converting device that's very, very low noise, right? We can agree on that. So, I was able to sort of figure out through the combination of the simulations in the experiments that this excess noise was not chaos, it was not a thing associated with anything particularly mysterious in the Josephson junctions. It was more a fact that the Josephson mixtures kind of have a drawback, which the SIS mixers don't have.

Zierler:

Which is what?

Schoelkopf:

The drawback is they're biased at a finite voltage, and through the AC Josephson effect, they make a coherent signal. They make a time-varying voltage that is proportional to the frequency, and that process basically means that in addition to converting the signal down to the lower frequency where you want to measure it, you have the ability to convert noise from other sort of spurious frequency channels you wish you didn't have. And I was able to sort of show that that noise made sense and could be controlled to some extent, and that if you design the system right, you could get probably pretty close to the quantum limit. I think it's not as clean, we weren't able to do something like Tucker and Feldman and show that first principles, it gets to right up to the quantum limit. And you know, that it would have been practical to do this with high-Tc devices and things if we really wanted to. And in the end, kind of understanding that process of the excess noise, getting rid of the noise or understanding the noise, that was kind of the focus of my thesis. And I just basically kind of decided to narrow in on something that I thought I could really address.

Zierler:

Rob, how did the postdoc at Yale come together for you?

Schoelkopf:

Yeah, it was an interesting thing. So maybe there's a recurring thing where I sort of reach a stage and take a hard look at things. So that was not a great time to be thinking about academic careers.

Zierler:

Yeah. And on that point, did you ever consider one of the National Laboratories?

Schoelkopf:

I did, yeah. So, I thought about National Laboratories. I went even broader. So, I had three job offers in the end that I was considering after grad school. One was a job at Institute for Defense Analyses—kind of technical analysis for national security problems, which I found interesting and appealing. I thought it was important kind of work. It wasn't going to be lab-based work, so there was the downside about that. The other one was-- so a lot of people at the time were leaving to go into management consulting and things like that. And I didn't think that was so interesting, but there were also a lot of people going off to become quants on Wall Street. And so I did a little looking into this, and had a friend who was in currency trading, and I ended up getting an offer to be sort of, start a technical analysis division at a currency trading firm in New York. And then I had, you know, I knew Dan Prober in part because one of his former students, George Ugras, had come to Caltech and was a postdoc in our group. And he convinced me to talk to Dan and to get to know him. And Dan seemed really awesome from the start. Very hands-on, very thoughtful, very caring, and really top-notch scientist. And so, my wife and I had a very interesting decision to make. Very different lifestyles. You know, different things you could optimize for. And among other things like the signing bonus, I would have had at the currency firm was more than a year as a postdoc's salary.

Zierler:

[laugh] Right.

Schoelkopf:

And in the end I was like, "But I love physics. I love science."

Zierler:

Yeah, sure.

Schoelkopf:

And so, I went off to Yale, and I said, well you know, it's probably not going to work. But let's just give it a try. I got a couple more years to sort of keep the dream alive, and let's give that a whirl. So--

Zierler:

And you're single guy at this point? You don't have family considerations?

Schoelkopf:

No, I got married. Met my wife, Inger. She grew up in the Pasadena area, and we met through friends out there and got married about halfway through my PhD. So, we were married but didn't have a family or anything yet. And yeah, so that was a big job to convince her to leave the West Coast, where she'd always been, and come out East in the middle of January. But you know, we've come to love it here. [laugh] She says, anyway.

Zierler:

Rob, for your postdoc, to what extent did you see this as an opportunity to continue on the trajectory you were on with your dissertation, and how much of it was an opportunity for new physics, new devices, new questions?

Schoelkopf:

Yeah. It's a good question. I mean as I sort of alluded to earlier, I think, Dan had done a lot of work in detector development and things like that, but wasn't an observer. Wasn't an astronomer. But then what he did—which was not so much done in the group at Caltech—was really more condensed matter, solid state, quantum physics experiments on, as I mentioned, sort of... It was called at the time mesoscopic physics. It was sort of the ability to observe quantum phenomena or effects like quantization of a charge or quantization of flux in devices that were mesoscopic, meaning sort of the size you can make in a clean room, but very macroscopic compared to real atoms, right? And so I was conscious that I wanted to learn more about that and have the opportunity to get a bit more into that. And saw Yale as-- that position as a way to do that. And so the first project I worked on with Dan—and with a graduate student named Peter Burke—was an idea Dan had for a very, very fast bolometric device that could be used as a mixer that relied on basically making a superconducting strip that was very short, and the electrons would diffuse out very rapidly. And so, Peter and I did some experiments in the lab to kind of show this physics and show, for example, that the effective bandwidth or the speed of this bolometric detector was scaling like diffusion would with the length that you could make. And showed how small we could make the length and still have it be an effective mixer, and therefore have a large bandwidth device which would also work at very high frequencies, and I think some of these HEVs are still being developed and have been used in the community. They're called bolometers.

Zierler:

What was your postdoc advisor working on at the time?

Schoelkopf:

He was working on a couple of these things, so these hot electron bolometers, he was working with a company on sort of improved wires for superconducting magnets. So, sort of understanding high-Tc superconductivity and the pinning of vortexes using artificial lattice structures in those. He also had a project that was looking at sort of non-equilibrium effects in semiconductor nanostructures. Electron heating and quantum dots and things like that. So, it was kind of a good mix of projects. And I don't know, the group was maybe one or two postdocs and five or six students? Pretty good-sized group at the time.

Zierler:

Yeah. And in terms of superconductivity and semiconductor research, of course if you want, you can really concentrate on some of the commercial or broader public goods that these technologies might present. Were you thinking along those lines at all, or were you still very much in basic science mode?

Schoelkopf:

A little in between there. So, I mean, I think like the part of the project Dan was working on was with, I think they're here in New England somewhere, it's a manufacturer that makes the wire that goes into MRI magnets, or high field magnets for scientific research. The hot electron bolometer project was funded by NASA because it was being developed for things like missions that would fly in space and look at things like, you'd do the molecular spectroscopy to understand certain molecules that contribute to ozone decay and things like that. So in that sense, it's an applied thing, but not for industry. It's for another mission, or another group of scientists, right? And then some of these projects were just like, let's publish it in PRL because nobody's ever seen this effect in a little nanostructure. So, it was a good mix, and follows a little bit, I guess, where I've ended up.

Zierler:

And at what point did you start to see that Yale would be viable for you in terms of a tenure track offer? And I guess coupled with that is the question, what was your sense, if you were thinking strategically, about Yale's perspective on promoting from within?

Schoelkopf:

Oh yeah, I don't know that it was all that strategic. [laugh] So you know, I was a postdoc for three or a little more than three years, I think. And I actually at the end of the second year, you know, I had started to do some experiments with Dan's blessing. You know, we kind of took some of the stuff we were doing on the detectors for NASA and launched a new project which was about noise and quantum effects. About shot noise and non-equilibrium things in these small nanostructures. Which we ended up publishing in PRL. And I went out on the job market, but at the time, I had a bunch of IAAA publications from my PhD work and wasn't really very well known in astronomy and wasn't applying for astronomy jobs. I had like one PRL about some condensed matter thing, which was about noise, which I learned when going out on the job market and giving talks, people didn't get. [laugh]

Zierler:

Right.

Schoelkopf:

They were like, “The noise? Why are you studying the noise? I thought noise is the thing you want to get rid of.” It's like, yeah, but it's quantum noise and it's got real information in it and so on. And this, there was a period here where the noise in mesoscopic systems became a big thing, but it was kind of not all the way there, or it hadn't made it into the broader condensed matter physics community. So I kind of was out on the job market and had several interviews, but not much success the first year. And the second year, then we had already done the RF single electron transistor, and things were going better, and there were a few offers. And I guess Yale was just looking. I don't think they created the position or the opportunity for me. I think they were interviewing others. And maybe with some trepidation on everybody's part, I was asked to interview for the position. It's not a very-- maybe, where you're kind of going with this, it's not a very traditional thing. We usually say, “It's healthier for your career that you don't stay where you were a postdoc.” Some schools back in that day, anyway, used to say, “Well, we don't tenure people from inside, come on. I mean it's much better to—" So there was kind of, in some of these fields-- Yale and Harvard and others like that are notorious for sort of you have to be sent into the wilderness as an assistant professor somewhere and then if you really publish your great book, then we'll hire you back with tenure. But it was a great department and very supportive. And there were several other faculty members in the applied physics department. Werner Wolf and Bob Wheeler, who had a long background in low-temperature physics, and were retired or retiring around this time when I came as a postdoc, and so I got to know them and they were very supportive as well. I think in the end, really, Bob Wheeler retired, which is where the position in the department came from, that the search was part of. And oh, I've sort of forgotten an interesting bit of serendipity too, that's probably worth talking about. My first exposure to low-temperature physics.

Zierler:

Ah.

Schoelkopf:

So, as a junior in high school in Chappaqua, New York, I applied for and got into a Science Saturday program at Yale University. And in Davies Auditorium they had a series of six or seven, different topic every Saturday, presentations from the faculty. And Werner and Bob from the applied physics department did one of those, which was The Road to Absolute Zero. And they had liquid nitrogen. They liquified oxygen and blew stuff up. They had liquid helium in a glass dewar that they pumped on, and they got it to go superfluid and you could see the fountain effect and the sort of liquid dripping over the edge and all that sort of stuff. And a few years after I was on the faculty, I got to do that thing again as one of the Science Saturday things. So yeah, kind of didn't really plan to go back there and didn't really-- Anyway, but so that was probably where I really sort of first got exposed to low temperature physics and started liking it. So yeah, so--

Zierler:

What was compelling about low temperature for you? What was, right off the bat, it seems like something really spoke to you.

Schoelkopf:

I think it's somehow the combination of, it's tangible and visible sort of right in front of you, but also quite exotic. You know, I mean superconductivity, what? Superfluidity? Why do the properties of things change so much when you lower the temperature? Like it's quite amazing, and there's many of these things which are manifestations also of the quantum world—which again, like as scientists, as physicists, when we learn about this one of the sort of strangest parts or something is like, well how come we don't need quantum mechanics to get up in the morning and go to work? And in the regular world, right? But it's really somehow underpinning everything and then you do something exotic, and suddenly now the current keeps running around and around and around forever. And these quantize. Like you can measure off a thing that has this H-bar in it. I mean, okay, fine, you can-- hydrogen arc lamp and you can see H-bar there too, but yeah, somehow there's a leap where you just, you don't see it sort of, you just say, “Wow, there must be hydrogen inside.” And I draw a cartoon of a proton, an electron, and calculate that. Okay, it all fits perfectly, right, it's great, but you don't see it being weird. And you know that kind of thing, I guess, really carries over into the whole quantum physics and quantum information and all of that, right? I mean we can see this thing being in the super position, like we can see the entanglement until we measure it. What? That's crazy.

Zierler:

Rob, you became full professor only five years after joining the faculty, which is a pretty meteoric rise. What was your sense in terms of, what chord did your research strike in terms of really impressing the people who would have been behind that recognition in that timeframe?

Schoelkopf:

Yeah. Well, you know, I think there's a lot of right place at right time in all of this. You know, we didn't yet talk about so much of the postdoc work and kind of how we got into the whole microwaves for quantum information and all that sort of stuff. So, we can talk about that, but you know...

Zierler:

I didn't realize that. I thought that that came more after you joined the faculty.

Schoelkopf:

No, it really... So as I mentioned, I had that first year where job searching didn't go so well. Because people, I think we had some pretty neat results about noise in nanostructures, but people didn't get it and I didn't have the track record. And the time through Dan, we had this collaboration with Chalmers University in Göteborg, Sweden, and there was a great student in Per Delsing's group, Professor Delsing's group there, named Peter Valgren, who came over and collaborated with us, and brought over some single electron transistors and devices that we didn't yet have the ability to make at Yale. And we were sort of initially doing this as a follow-on of the quantum noise studies. You know, looking for colored noise or correlations in the transit of electrons through these small nanostructures. But there was this really interesting thing that happened, which was I kind of came back from the APS meeting that year and it was kind of clear I wasn't getting callbacks. And I just said, “Ah! Let's just go in the lab.” And Peter and I were kind of thinking about how to make a measurement of noise in these nanostructures that would be at sort of the frequency corresponding to the passage of singular electrons. So at high over e.—Which if you think about it, unfortunately e is really small, so for any measurable current, that's at kind of a fast frequency. So we were trying to figure out how to get a few gigahertz signals of this drip-drip-drip of the electrons out of the small device and into a measurement apparatus. And of course, I knew all these microwave techniques and everything, and we had outfitted for the quantum noise experiments, for the experiments, this tiny little dilution refrigerator, which was the only fridge in New Haven, I think, at the time that Dan had got one co-ax[ial] cable into it and one HEMT amplifier that was from NRAO in Virginia. And we're sort of doing some of the first things to listen to these nanostructures instead of just measuring an IV curve, or as other people had done, sort of spectroscopy on-- It was fairly common to do spectroscopy on these nanostructures, meaning you would irradiate it by just sort of weakly coupling in some high frequency signal from outside, and you'd look to see things like the Shapiro steps or the SIS steps. Some feature that's corresponding to the particular frequency that you're putting in, but in transport in the DCID curve of the thing that you would measure. And here we were trying to do the reverse. Put current through and see a spontaneous generation of a frequency coming out. It's kind of the dual Josephson effect that we were looking for, this single electron oscillations. And we just kind of realized that the way we were doing this with this very high impedance device made it pretty hard. And that we were wire-bonding over to the sort of bond pad on the device, and I said, “Well, we better think about that as a microwave circuit.” You know, the wire bond is an inductor, which is something Jonas and Tom drilled into me, like, “No, you can't make a direct connection at very high frequencies. It's going to have some parasitic reactances and stuff with it.” And I'm like, “Oh. Darn, that's going to transform the impedance. Wait, that's going to transform the impedance.” And so, we figured out we could take a single electron transistor, which is like a super sensitive electrometer—you know, it's the quantum-limited device that's kind of the dual of a SQUID magnetometer. They had been made and used for many years, but because they were sort of hundreds of kiloohms or megaohms devices, biased with a battery and you're reading from the bottom of a dilution refrigerator on a long capacitive cable... The measurements were done at 100 Hertz and things like that. And we realized we could send a microwave signal in, bounce it off the transistor, and see that the reflection coming back, and that would tell us whether the transistor was on or off, or what was the charge that was coupled to the gate. And all we needed was a quality factor of 30 tank circuit to do that, and match it perfectly and get very high sensitivity. And of course, quality factor of 30 at a gigahertz means you've got... a gigahertz over 30, you've got many, many megahertz of bandwidth. And dang, it worked. Like we were able to make this very high-speed transistor. We could get away from the ubiquitous 1/f noise that had limited everybody else. It was like both a big improvement in the speed, and in the sensitivity. And you know, it also kind of allowed us to-- I mean, people had already done this, but for me I was able to go back through and say, “Okay, so what are the fundamental limits on doing electrometry?”

Zierler:

Yeah.

Schoelkopf:

And so that kind of was a pretty interesting avenue, and then that was a much more compelling research angle. I mean, I've got a great tool now to do these different things, and that was also, talk about right time. So this is...

Zierler:

2001?

Schoelkopf:

19... Yeah, I guess the paper came out in 2... no, no. Earlier than that. Sorry. So, I joined the faculty in ‘98, the SET paper came out probably in ‘98, but we did the experiment in ‘97. And so, Peter Shor on quantum factorization and quantum error correction is ‘94-95.

Zierler:

Yeah.

Schoelkopf:

And all of a sudden, a lot of people are talking about, how can you make qubits? And Peter Valgren brought back with him a PhD dissertation from Michel Devoret's group in Saclay by Vincent Boucher, which was about the Cooper pair box. And they had already realized that you could use something like the single electron transistor to make a device that would have two quantum states that were associated with a specific charge. No Cooper pairs or one additional Cooper pair. And that this thing could potentially be manipulated electrically and have a long coherence time. They showed that there were superpositions of these charge states possible in the devices, and the question was, well, but if it's got a finite coherence time, how are you going to measure that state of the thing? I know how. The SET. You have, you know, that's the... And they were doing SET measurements, they were just doing them sort of at very low frequencies, and you had to look at kind of time average properties. And so, the idea of doing dynamics on these things was just like a really great time to be able to do that. And so yeah, then basically we started out by collaborating with the group in Sweden to make these devices and using the SETs to be able to start measuring these kind of qubits. There were a lot of groups gearing up and starting to do these things, but we had a very different set of techniques.

Zierler:

Rob, just so I understand, in terms of it being a two-step process, it's, you know, step one is just confirming that the thing works, right? Is it immediately obvious what you're able to measure as a result of confirming that it works? Or is that a more open-ended question and it requires discussion with your collaborators? You know, sort of head-scratching about, now what do we do with this now that it works? Can you describe a little bit that connection between stage one and stage two?

Schoelkopf:

By it meaning the SET?

Zierler:

Right, exactly.

Schoelkopf:

Yeah. Well right, so that first paper was very much like, here's a technique and this is the device performance we get out of it, right? So, we could measure things like its sensitivity by kind of calibrating it with an externally applied signal. So, we actually in that paper didn't use it to measure individual charges, or to actually do the experiment we had sort of been intending to do, which was now run current through a wire and have the electrometer there, kind of watching the electrons one by one as they go through. But, you know, I guess the point is, you know you start with, “Okay, I have a measurement technique. I want to show proof of concept. I want to understand its performance and its noise.” You see all the themes coming back, right? And then you can say, “All right, so now what are the panoply of things that I can use this new sledgehammer for?” And so, I mean, it sounds a little too contrived, but I think it's actually kind of true, I mean it's like instead of doing astronomy, now I had a detector that allowed me to see other things that people couldn't see potentially, and it was pretty easy to show, given where the scientific community was moving, that there would be a lot of neat things to do with this. And so, I guess to kind of answer your question about career trajectories and stuff, like I mean it was a lot of work in the beginning to get a group set up and to start doing these things. But you know, showing success on that, it was clear we were going to have a lot of interesting things to do. And you know, I think this, given that now there was sort of the whole motivation of potential quantum information things, eventually maybe in the far-off future, that was motivating a lot of people to look at dynamics and high speed things and also for these solid state qubits, you control them by using microwave pulses and so on, so it was kind of like a perfect storm for me, and so there was a lot of excitement and interest in doing that, and I think it was clear that really interesting physics, at least, was going to come out of it. But you know, I like to point out that like when we started the lab, no one had built a qubit yet.

Zierler:

Right.

Schoelkopf:

I mean there were ion traps and atomic clocks and those are qubits, but no one had built a solid-state qubit of any sort yet. And the first superconducting qubit experiments were either these experiments of Michel’s [group] in Saclay, or there was also the seminal experiment by Yasu Nakamora and Jaw-Shen Tsai in Japan, showing kind of the first oscillations between states. But then pretty quickly, we were able to kind of get in the game and do things like excite the system to its excited state and measure the analog of the spontaneous emission by just watching it with the electrometer and seeing when it would fall back down. And then the other thing that was kind of fun in this phase was like, thinking about okay, the quantum limits of the SET as a detector, but also when you take the measurement apparatus, which is the single electron transistor, and couple it to this artificial atom. Like now, what does Heisenberg say about that? And so if you're observing the qubit, what does it do and how do you know you're not observing the qubit and how rapidly will you collapse the state, let's say, when you turn on the measurement and all these sort of things. And we could see that this was going to be a really cool playground for doing all this kind of stuff, right? So that we were going to really be able to re-explore a lot of quantum optics and quantum information—or quantum measurement I guess was really what we called it at the time. Like, you know, that this would be a real sort of playground for doing experiments in kind of basic quantum mechanics. In a wholly different way. In a very different domain where it's electrical parameters and electrical controls, and where it's also macroscopic variables. So when you're measuring the charge, it's one Cooper pair. But it's also like a few micron piece of metal containing millions and millions of atoms that is either at zero volts or 50 microvolts. Right? Like that's something which isn't supposed to happen in the real world, right? And so that was just the very fun possibility, and it was a fun time.

Zierler:

And Rob, to set that narrative looking ahead to quantum information, in what ways-- I mean, there's a duality to this. It's obvious that this is where these things are heading, but it's also clear as you said that you're nowhere near getting there yet. So sort of early 2000s, can you just sort of set the stage for where you're seeing quantum information on the horizon, and why that horizon is still distant from that vantage point?

Schoelkopf:

Yeah well, I mean, at the time we were just wanting to see whether it would be possible to make a qubit.

Zierler:

Yeah.

Schoelkopf:

And then of course in these systems, it's always been clear if you can make good qubits and you can manipulate the quantum states and all this sort of stuff, it's going to be, yeah, both a really neat physics playground, and also a much more attractive technology than things like individual atoms in a vacuum chamber with a laser for each atom, as a potential technology. But the thing that was really unclear was the extent to which these kinds of artificial systems would really be well-described by a simple Hamiltonian or really a good, a well-behaved quantum system, right? So, my first postdoc, Konrad Lehnert—who's now at JILA in Colorado—we used to go around and give talks and we were sort of, you know, most of the talk would be explaining, like, how is this network of Josephson junctions like a spin? And then we'd show some measurements, and then we would say, “This is the Hamiltonian. And we really need it.” Right? Like in the beginning it was just like, “Oh my God, there's actually energy levels that track with these externally, electrically-controlled parameters that we can understand.” We were just trying to make a model. A single spin with an axial and a z field. And what was really unclear was, yeah, but is that thing going to ever be coherent? Is it going to have-- Surely this thing must be so many particles, so many degrees of freedom, like nuh-uh, no way this is ever going to reach the coherence that an atom could have. And kind of one of the miracles of the field is that, actually, Josephson junctions and superconductors are pretty darn good.

Zierler:

Yeah.

Schoelkopf:

They're not just extra-good conductors. They are superconductors, right?

Zierler:

[laugh] Right.

Schoelkopf:

And there are these collective phenomena and these collective degrees of freedom, like the voltage or the current, which are actually very decoupled from all of the other microscopic parts of the system. You know, it sits on a silicon wafer with billions and billions of atoms, and there are lots of dangling bonds on the surface and stuff, like surely that's ever going to work. And in the beginning, these devices had very short lifetimes. The first experiment, the lifetime of the superposition was like a nanosecond. And these days, it's up to a millisecond. It has improved a millionfold. That's this thing that sometimes gets called Schoelkopf's Law, because we had to propose in a funding thing, you know, they said, “Well, is there really any evidence of progress on this?” And I said, “Well, let's make this plot.” My gosh, you know, it's going to be every two or three years when we think of the new thing and make the new style of qubit. So that was kind of one of the big questions, was just would it really allow you to make superpositions? Could you do all these things that had been done now? Ran the experiment's entanglement over the-- You know, two qubits, entanglement of many qubits, detection of quantum errors, running algorithms, all these kinds of things.

Zierler:

Rob, could you talk a little bit about how your lab environment changed from being a postdoc to being a faculty member?

Schoelkopf:

Sure. Yeah, and you know, the really important story there too is how we grew the collaboration at Yale into what it is today.

Zierler:

Yeah, right.

Schoelkopf:

So, after the RFSET, I kind of got out in the community a little bit more. I was able to meet Michel Devoret, who's this amazing scientist, and whose work I had read even as a graduate student, you know? And their group in Saclay with Daniel Esteve and others had done a lot of really neat things on single electron devices, the capacitance standard and current pumps, current standards, and on the Cooper pair box. And I started in 1998. To my surprise, I think everyone's surprise, Michel said, “You know, I'd like to take a sabbatical from Saclay. How about I come for a year to New Haven and the applied physics department?” And he arrived sort of before my lab was even set up. So, he and Konrad and my first graduate student-- and we'd sort of get sack lunches and we'd come sit in the alcove across from the lab and we'd say, “Which crate are we going to unpack today?” [both laugh] And Michel and I were talking about things like, what is really the limits of these electrometers and things, and I was learning a lot of this mesoscopic physics, which he was really an expert in. And he came to New Haven, I guess for a few reasons. Interestingly, his father was also an academic, and they had spent a sabbatical year when he was a teenager in New Haven, so he kind of knew the place and actually liked it. And toward the end of that sabbatical, he sort of let it be known that he liked universities and didn't have that in Saclay, where he was sort of in a national lab, and well, you know. And so the department made him an offer, and he came and joined the faculty a year or two later. At which point, the most typical response from others in the field was, “Ooh, what? You did what? Like, he would never leave Paris! What? How did you do that?” Right? And we were just having a great time, and we set up when he came, we set up some joint labs, and expanded the number of fridges and everything we had. And started off on this collaboration, which we continue to today, and has been a great joy for me. The other important thing that happened was I got to me Steve Girvin. Who's a theorist, but...

Zierler:

He's in it.

Schoelkopf:

Not the typical kind of theorist. So, he's the winner of the Buckley Prize for his theory of, among other things, the quantum Hall effect. I mean, he's got real live theorist chops that he likes to talk to experimentalists.

Zierler:

Uh-huh, right.

Schoelkopf:

And we met sort of remotely and ended up writing a proposal about 1/f noise that didn't get funded in the end. And I knew a little bit about quantum optics, and with the SET and the sort of noise or what was called electron counting statistics-- So part of the interest in mesoscopic physics of looking at the fluctuations of the current was related to kind of photon counting statistics, which is used very widely in quantum optics. You can tell if it's a single atom emitting because you never get two photons at the same time. So, you can do similar things. So, I had already pitched for my faculty position, that I'm going to build these devices and we're going to do quantum optics-inspired experiments on solid state systems, and I even had in there, we're going to do these Cooper pair box qubits, although that's kind of crazy. And I also said well, my main bread and butter I'm going to make tenure on is going to be detector development using SETs and superconductors and things like that. And Steve Girvin came, I remember him coming to campus to give a seminar and us having our first face-to-face meeting. And he came for 45 minutes or an hour, sat in my office, and said, “Hey, I've heard about you from Per Delsing,” because he had actually worked in Chalmers earlier in his career. And... “We should talk about these Cooper pair boxes and microwaves, because I think it's exactly like an ion trap system, where you have atoms coupled to bosonic modes. You know, phonons moving.” And I said, “Yes! I know.” And we kind of cartooned up circuit QED on the blackboard in that one meeting.

Zierler:

Whoa. That's an exciting day.

Schoelkopf:

Yeah, yeah, yeah. And so, then Steve let it be known that he was interested in potentially leaving Indiana and again, a few people really tried for him, and we were able to recruit him to Yale. And so, then it became this incredibly fertile and creative thing. Steve taught me a lot of quantum optics. We worked out the whole translation of atomic physics, or cavity QED, you know, kind of the coupling of single photons to single atoms, translated in the electrical domain. And very rapidly, you can show that this is a system with which if only there can be high-quality factors or sharp lines or long coherence times, which we know we can get with superconducting microwave resonators, right? And which we thought we could get with qubits, that this system would be like the atomic physicist's dream, because the coupling of the light and the matter is kind of silly. It's four orders of magnitude stronger in kind of relative terms to the overall frequency scale or whatever. Atoms are great. They're very, very coherent, but that's because they're teeny, tiny little things, and they have these teeny, tiny little dipoles, and they don't interact with the environment at all. So you start with a great coherence, but it's very hard to get a lot of transfer of information or to entangle multiple ones together. And it was clear that this was just going to be so much better in the circuit domain. And we wrote up a kind of joint paper. Alex Blais was the first author on that, proposing this system. And there's a funny story about this, which is we sent that proposal to Nature. They said, “We're not taking any more, these quantum computing proposals, none of them is ever going to work. We're not taking any more of those.” We sent it to Science. “No.” We sent it to Physical Review Letters, rejected. We sent it to Phys Rev A, and eventually the long paper was published in Phys Rev A. Now that paper has like, I don't know, 2,500 citations today?

Zierler:

Yeah, yeah.

Schoelkopf:

So, Alex did okay on that one, right? But it's an interesting sort of thing. And then very shortly thereafter, we were able to do these experiments, led by Dave Schuster and Andreas Wallraff and my group, that showed that strong coupling of Cooper pair box to single photon, and we could immediately show many of these phenomena of cavity quantum electrodynamics, but in this novel circuit domain, and it was also clear that, okay, now if you can entangle your solid state cubit, which is the analog of the atom, with a microwave photon that's either trapped or steered around on a chip or in other devices using the microwave engineering, you've got a whole host of things that you can do. And you can control the spontaneous emission, you can make single photon sources, you can entangle things and build a quantum bus. And this all sort of just got this incredible momentum from the combination of the brain trust of the three of us and the environment that we were in.

Zierler:

Rob, I asked the question, when we were talking about graduate school, about the relevance of theory to your work, but it seems like the more appropriate question now is, in what way was this research sort of up-ending theoretical understandings of quantum at this point?

Schoelkopf:

Yeah. It's a good question. So, in some regards, and you may still find some people out there in the field, people will say, “Well, I mean this is pretty standard physics.” Or textbook physics maybe is a better way of saying it. And indeed, many of the things we've done in circuit QED are, they become almost the textbook definition of like, how is the Jaynes-Cummings Hamiltonian work? And it's a great thing you can write down. I mean some of the things like the Hamiltonian for the transmon, I give as a final exam in my undergraduate quantum mechanics class often. Right, because you can just say, “Well, what if you had an x^4th term in your oscillator and work this out.” Maybe the big theoretical thing is the-- are all the things that didn't happen. You know, the fact that there isn't a new physics that prevents quantum mechanics from working when you have 10 to the 10 Cooper pairs all changing their chemical potential. Or being simultaneously in two chemical potentials because there's a superposition in your transmon. What I think we've done over the years, though, is gotten into the physics of quantum measurement and illuminating and elaborating on what exactly are the right rules and ways to understand these things? You know, a lot of the mysteries or confounding kind of gedanken experiments that people sweat about in quantum mechanics, they turn out to be largely about poor semantics or about people assuming that of course a certain thing is true when it's not really true. Or what I like to say sometimes is, it's about not being clear where the dividing line is that you start to say it's classical and it's not really doing quantum mechanics anymore, right? Because we think that you just should apply quantum mechanics all the way up until you and I are having this conversation. It's just that that's exhausting, and we can't solve it. And so, we kind of give up and say, “It'll probably just work according to hydrodynamics and classical physics at this point.” So, what I think is interesting, and where the field is going, is that there are things about many body systems, or many degrees of freedom in entangled states, that maybe we don't know what's going to be new and different about it, but it's certainly uncharted territory. You know, there's again this funny thing that we can do in our undergraduate labs, like a Bell inequality or a simple superposition experiment with photons or other things. You can show these basic phenomena of quantum mechanics. Again, we believe that as soon as you put 14 particles in a box, they're in some complicated, entangled state, but the number of experiments which have observed end-way entanglement are very, very few. And it's the builders of quantum computers and quantum information processing devices that are leading the way on that, right? And in a way, one of the things that'll happen in this field is we start to kind of do more in terms of complicated algorithms or error correction. Like we're building complicated many body states and things with very special symmetry properties or topological properties from the ground up, right? And so, I think there's a lot that still can come, but I mean I guess the theoretical surprise is that any of this would actually hold true, and I guess-- I mean there's also a lot of things that we've explored with circuit QED because of the much stronger coupling you need. You can see and sometimes stumble over and really reveal phenomena that would be higher order in coupling strengths and things like that. By virtue of having a new regime where you're in a much stronger coupling limit. You know, except for the Rydberg atom microwave cavity QED, like circuit QED is the only thing that gets into this very strong dispersive limit, where adding a single quantum to one degree of freedom in the system shifts the energy of all the other components by a very large amount compared to their line width. And so, you can use that to do these amazing things like non-demolition measurement of a state. So you can actually, you know, force the qubit to decide. Is it zero or one? Without scrambling the state. Like it will choose zero or one with the probabilities that we tell everyone they should use to calculate in your basic quantum mechanics, and then it's in that state and it stays in that state. And when you come to quantum error correction, it's making that measurement which collapses the computer into, ah, the error did actually occur. And now I know that it occurred, and so it's not really an error anymore. As opposed to, well maybe there's a superposition if there has been an error and there hasn't. So, you know these are the things that we can really get into that I think are kind of unprecedented, and really kind of beautiful textbook or whatever testbeds for this kind of quantum description. So.

Zierler:

Rob, and also to return to an earlier question about at any point you thought about industry or about the relevance of your research on commercial ventures. Can you talk about the origins of quantum circuits? Where did that come about and what were your objectives as you were sort of putting this idea together?

Schoelkopf:

Right, right. Yeah, well I mean it's been an amazing arc over the last 20-some years. Going from the kind of crazy idea that maybe these things could be qubits to seeing national-level, very large initiatives in countries all around the world. Big companies getting into this and doing their own development and so on. Somewhere along the line here, it became clear to us that these things really work. And we're going to have super-connecting quantum computers that are going to do computations that no one else can do or that you could never have foreseen. And that's going to happen in the next few years. I think it's going to happen a lot faster than people think, because there's this thing where people want to extrapolate in a straightforward way, what the application of a certain amount of force is going to do on a problem using all the things we know right now. When I look at our last 20 years of science, we've almost never solved the hard problems the first way we thought we were going to solve them. We have always found that actually there's a much easier way to solve them. A good example is the coherence of qubits, let's say. So, it's known that even a few broken Cooper pairs can spoil the performance of your superconductor, or even a small amount of dielectric loss in your substrate or whatever will give some damping that comes in. And even a tiny coupling to the outside world, because these things have these ginormous dipole moments and they want to emit, right? They can lose their information by spitting it into the environment. But what we found is that they're very simple tricks—you can just make these three-dimensional structures for instance that reduce by orders of magnitude the amount of energy that lives in a surface. And so, you don't need to actually make a perfect surface. You want to make a more perfect surface of all your devices and all those sorts of things, but you can also just make them much less important. So, these are kind of, there are clever design tricks, and there are efficient ways to solve some of these problems that you can go from idea to breakthrough in very short order, right? And so, I think there's a lot of energy going into all of this. Everyone believes that this is going to happen in the next several years. I think with some of the things we're doing these days in the lab and some of the ideas that are out there, there are more efficient ways of building these machines than people realize. And coming back to your original question, we kind of realized that it's necessary to do some scaling and some engineering, or turning these things into products, which some people are trying to do, but they're not doing it the way we would want to do it. Or what we think is the best way. And that building a technology or a product is not what a university physics lab does. And shouldn't do.

Zierler:

Yeah. On that note--

Schoelkopf:

I think the field is in an interesting stage right now of slightly awkward adolescence, right?

Zierler:

Yeah.

Schoelkopf:

And so to do the cutting-edge science you want to do, you have to build more complicated devices and systems. But there's a limit to how much systems engineering physics graduate students can or should do. Right? And over the years, we've made amazing progress in part because Michel's group has done a lot of really cool work on parametric conversion using Josephson junctions, which allow us to make amplifiers which reach the Heisenberg limit in the gigahertz range, which we weren't able to do before. And allow us to do the measurement of qubits in things. And every one of our experiments has multiple of these amplifiers inside it—which are now sort of a turnkey part of the system. But we kind of still have to keep making them in the lab. And there's just a limit to how much of that one can do in the university environment. So, the reason to start the company was I could see that there were things where a little bit of effort by some software engineers or some mechanical engineers or some conventionally-trained electrical engineers can do things that we need to go to the next level that aren't really practical to train each and every graduate student in and to implement in a physics lab. And frankly, we also realized that what we've created—because you see IBM and Google and all these machines that are out there—are transbonds and circuit QED with dispersive readout using parametric amplifiers. Like, it's very much technology that our team at Yale played a huge role in creating.

Zierler:

Right.

Schoelkopf:

So, there's some huge value that we've created, which I feel really excited and really proud about. But we also have a chance to participate in some of that activity and to try and add our own spin to it. So, Quantum Circuits, Inc. is a venture-backed company. Our lead investors are Canaan and Sequoia. It's about 25 people located in Science Park here in New Haven, and what we're trying to do with that is create a synergistic structure where we have now the right organizations that can talk frequently and where I can help participate in all of these aspects. Where everybody's doing the thing they're best suited for. Right? So, we have federally supported research by graduate students who publish all their results and do the high risk, high reward long-term kind of research which is necessary, because it's the seed corn for this field. You can't create a field... I mean, like you can't create a semiconductor industry if people don't keep doing semiconductor research. Right?

Zierler:

[laugh] Yeah.

Schoelkopf:

But on the other hand, universities don't create their own laptop processor chips. That's crazy, right, that's daft. Why would you do that? And so, there are different roles for the different organizations, and so I think what we're trying to do is find the best and most efficient way of doing those things. And really what I think we've figured out at Yale in the last few years, is there's elegant, simple tricks that allow you do to things like quantum error correction and making inherently robust operations and gates, which is what's really going to be needed in the next few years to get to useful applications in the quantum computing field. That we know how to do in a very different architecture than what the big boys are already trying to sort of ram through in a kind of brute force, I would say, way. That's maybe a bit unkind, but they've been doing some really elegant engineering of these transmon circuit QED systems, but you don't want to build the entire computing infrastructure out of the first kind of transistor, let's say, that someone would come up with, right? So I think what we're focusing on at QCI, at Quantum Circuits, is kind of developing and productizing and being able to scale up the kind of more recent Yale approach to quantum information processing with superconductors. Which has taken a very different turn. We have the cavities now, as the information carriers which serve as sort of hardware-level logical qubits that allow error correction that saves, you know, orders of magnitude in terms of the complexity and the resources you need, and so developing that out, which I think is going well and is going to be pretty exciting, is hopefully going to add a lot to the field.

Zierler:

And Rob, so to really understand the different goals, right? So in the news of course, it's very exciting to hear Google or IBM, they're on the verge of quantum computing, and yet you've very clearly laid out that you have a different vision of how this is going to play out in the next few years. So, I wonder if you could explain a bit more if the goals between what you're doing and what a company like Google is doing, are they the same goals where you are applying very different means, or are you actually aiming for different end-use products?

Schoelkopf:

Yeah. Well, I think one of the reasons that quantum computing is so exciting is that it's this completely different paradigm and there are lots of different ways that it can be used. The near-term things include most probably doing things like computational chemistry or simulations of very hard scientific problems, because some of these things, if they're inherently quantum, they are exponentially hard to do on conventional computers for the same reason quantum computers have this enormous power.

Zierler:

I mean, let's just pluck two things out of the headlines, for example. What would a quantum computer do for us that would make COVID-19 sort of more manageable? Just for example.

Schoelkopf:

Yeah, so I mean I think it's still a bit early to talk about very specific use cases or things like that. I think what a lot of people are expecting in the field is that using quantum computers to eventually design new drugs is a thing that can happen, right? I think a lot of people then have this vision of, you're just going to type into the quantum computer, “What is the cure for COVID?” It's going to be a little bit different than that, right? I think the next big thing for the field is what people sometimes call “quantum advantage.” It's a bit different than quantum supremacy. Quantum advantage is, I learned something using a quantum computer I couldn't have known any other way. So, if that just tells like a chemist something about the properties of some molecules they're playing with, or lets a drug company like maybe decide that certain way of doing things is less appealing than another one. They don't have the resources to follow every possible route to every possible drug. They need to make choices. So even if they get a little bit of insight using a quantum computation about which one is the better one, that could be a very valuable thing, right? So, to get back to your original question, I think... the long-term goal that everyone has is really large-scale, general purpose quantum computers which can do a whole host of different applications. And the question is, how do we get from here to there? So, it's clear you need much larger quantum computers than we have so far. It's clear you also need quantum computers that work much better than the ones we have so far. They need to be more robust and have error correction built in. They need to be fault-tolerant quantum computers. And we know that's possible. It only takes some thousands or millions of qubits. And that's sort of proof that theoretical quantum computer scientists have done to show that there's a bound which is not infinity on the size of the computer you need to be fault tolerant or robustly address some very complicated problem. That's a very different thing than saying, “This is the smallest quantum computer that does quantum advantage. And what it looks like.” So in the field right now, there's this two-pronged thing. We want to develop error correction, fault tolerance, and robustness to the point where we can eventually do all possible quantum computations. That'll be really, really exciting. In the meantime, though, we've got to find a way to do things that offer real quantum advantage on the right timescale for people. And so that's going to be a combination of picking the right problem picking the right algorithm, maybe special-purpose hardware to some extent. And what I guess we’re doing that's different-- I mean, I think most of the big boys are saying, “Well, transmons and circuit QED... It works. It works probably almost well enough that if we scaled it to millions of qubits and it worked the same or a little better, in principle we can turn on all this quantum computer science theorems and everything and say that it will be a good enough quantum computer to do really useful tasks.” So, you know, I think they're kind of taking the strategy of, “Well, we've got an engineering description of what we need to build. Let's go build it.” It's like building a particle accelerator or something. It might take a long time, it's a really massive effort, but it's “only engineering.”

Zierler:

And of course--

Schoelkopf:

What I think is kind of difficult for the field, though, is you don't want to ask the quantum designer, “This is what my hardware is, go engineer the crap out of it.” And you might make that engineering task way too hard. And on the flip side, you don't want the engineers saying, “Well I really wish it would be-- make my life much easier is, the quantum hardware needs to look just like this.” You need to do what's now starting to be called co-design. You need to be able to kind of understand the linkages between these things. How do we make qubits that are high performance but relax the cost and the complexity of the control system?

Zierler:

If the basic goals, even if the means and some of the motivations are very different, are you amenable, at least theoretically, to closer collaboration with a Google or an IBM, if at the end of the day, everybody recognizes that the sooner we get to useful quantum computing, the better?

Schoelkopf:

Sure. And I think some of that goes on already. It goes on in many different ways. Of course, a lot of the people out there moving and shaking the space are people who know this field, because probably they've worked with us at some point in time, right? [both laugh] And this is great to see. It's another, as your career goes on, you love all the papers you can point to and the awards and things, but you also love the fact that there are all these people out there doing great science and you had a little small hand in helping them along or making that happen. So, we'll sort of see. I mean one of the tensions of course when something becomes commercially relevant is, people want to be having the right competitive position. [laugh] And so, it's important, I think, that this field finds the right balance. I mean, there are other disciplines which are somewhat cautionary tales, and there's been some written about this. Machine learning is a really big thing. It's really, there are-- People who know machine learning are in huge demand. And it's very hard to recruit and retain faculty in machine learning, because they just get siphoned off by industry. And then that creates a problem for industry, which is there's no one out there training people, so they can hire more of them. And so, we need to find the right balance, and I think it's going pretty well. This is going to continue to be a big effort and require a lot of resources in a lot of different ways and a lot of different avenues. So, the government has a big role in it—stepping up to do that—because it knows that it has to have fundamental research being done because that's something that in the modern era, companies don't do fundamental research that's really got a very long-time horizon. At least, there are very few that can do that. They also can't create the workforce, or the trained people, and so that's where academia has a big role to play. And academia also has a big role to play in terms of figuring out the better ways of doing things. And so I think there's going to be room for many different approaches and many different successes here. It's nice to talk about the race for quantum computer and who's going to win, but it's probably not going to be just one winner. And it may not even be just one technology, although I think everyone agrees that superconducting circuits are the leading technology. A vast majority of the machines you can access and use today out there are based on this approach, right?

Zierler:

Well, Rob, now that we've gotten to the point in the discussion where we're sort of prognosticating about the future, I want to ask in the last part of our talk, one aspect of your career we really haven't focused on at all, and that is your work as a teacher to undergraduates and a mentor to graduate students. And particularly, I'm interested in the kinds of courses that you're teaching undergraduates for whom five, ten years into the future, this is going to be their generation when quantum computing becomes a part of their reality. So, I wonder how you sort of operationalize that understanding in terms of the kinds of things they need to know about where we are currently with quantum computing and what impact this might have in terms of their own career, business or academic interests, as they chart their own future.

Schoelkopf:

Mm-hmm, yeah, and I guess there's kind of a couple of aspects to that. There's traditional lecturing and undergraduate education, and then there's maybe mentorship and graduate training and things. So, I mean I teach a variety of different courses. It kind of rotates around. That's the way it works in most universities. But one thing I teach from time to time is an undergraduate first quantum mechanics class. And over the years, I have evolved that curriculum a little bit, and some of my other colleagues here at Yale have done similar things. In my one semester first quantum mechanics class, we don't really do the hydrogen atom. When I was taught quantum mechanics, that was like, “Okay, so this is what the course is about. We're going to explain to you the hydrogen atom. We're going to explain perturbation theory by calculating the fine-structure and the hyper-fine-structure in the hydrogen atom.” Which is a great thing. I mean it's a nice thing you can write down and you can work this out and lots of textbooks go through this, and it has the feature that it's kind of historically accurate. “We measured spectra from atoms and went, 'Huh? Why are there lines?'” And then we saw Zeeman shifts and fine-structure and we said, "Huh? Why is that there? I thought they were just electrons and protons. And, oh, there's a spin thing.” But it's not very modern. It's not very tactile or tangible or real-world, or how does what we figured out in 1920s Germany academic physics, why is that the pinnacle of my undergraduate physics education? And it also has an interesting feature, which is it's kind of-- because it's a three-dimensional thing with a potential, we have to do YLMs, like the math is just a lot more cumbersome. So, you do 1D-- So what I do in the intro quantum course is, we do spins first. We start with the spin. And the Stern–Gerlach experiment as the historical first incarnation of that, but then talk about qubits right away. And then we do the harmonic oscillator, which is the quantum harmonic oscillator, and then we talk about circuit QED. And you can do almost all of this stuff introducing time-independent, time-dependent perturbation theory and all that kind of stuff without really ever going into the hydrogen atom. Now, some of my colleagues [laugh], there are some physicists who say, “What? Shocking! You're creating physics undergraduate degree holders who are culturally illiterate.” They say they don't know how big the hyper-fine splitting in hydrogen is. But on the other hand, I think it's got a lot more appeal and maybe what I think you can argue is it's getting to the essence of the quantum mechanics without the historical baggage and as much of the math, as we've experienced in this discussion, resonates with me. And I think this is true now in a growing number of courses. So, a lot of people are trying to revamp physics in general and not have it be so historically based and have the interesting stuff come last. In some regards, fields like biology and stuff are better that way. You can be in a first or second semester biology class and get your own DNA sequenced, or you can talk about some things that feel much more modern and much more cutting-edge much more rapidly... it's partly the difference in the field. I mean, you also don't have to teach them everything about mitochondria first before you talk about genetics and DNA. Whereas to some extent, you need some E&M before you can do quantum, because-- I don't know. But yeah, I think that's happening. There are other courses here at Yale were advanced first year physics classes where they cover qubits and spins and could even give them homework problems to solve on one of the online quantum computers, things like that.

Zierler:

And Rob, are you especially interested in welcoming undergraduates into your lab work who express the interest?

Schoelkopf:

Yeah sure, and we get a huge amount of interest now because people read about it and see it and this modern generation, of course, they've grown up with computers and they are the most amazing objects they come across in their everyday lives. So the physics behind that or the newest, not-yet-here version of computers is very alluring and very interesting. We continue to have the undergraduates involved. I mean it's, the real experiments we do are pretty hard and pretty technically complicated. It's difficult to get undergraduates involved in them so intimately. You know, takes weeks to put the thing together and to cool it down in the fridge and so on. It doesn't always fit with their timeframe. But yeah, I think that's an important thing. And we have some specialized curricula also that kind of addresses some of the new topics of quantum computer science or the microwave design of these things or the quantum physics of noise and quantum circuits, I guess we would just call the field now.

Zierler:

And Rob, on the graduate student side, particularly this idea that it's satisfying for you to see some of the people who you've trained who are going out and working in other universities and in industry that will help make quantum computing a reality. So, I'm curious, who have been some of your most successful graduate students over the years, and what do you see as some of the commonalities that makes for a successful graduate school experience in a quantum computing kind of environment?

Schoelkopf:

Oh, wow okay. That's a good set of questions there. Let's see. Can we break that down? Let's start with one of those aspects.

Zierler:

Sure.

Schoelkopf:

So, let's see, what were all the things you had there? [laugh]

Zierler:

So, it's personally satisfying to you to sort of see your intellectual DNA out there in the broader world of this move toward quantum computing. So, who have been some of those graduate students and postdocs over the years that really represent that for you?

Schoelkopf:

Yeah well, I mean I guess this is a place where I get to say which of my proteges. It's a bit like asking someone which of your children do you love more? [both laugh] So I mean there are quite a few of them and I think they have gone in a few different directions. Some of them are out there in academia, some of them are out there in industry, and even a few at national labs and things like that. So, I mean go back to the early days, I mean there's some of my earliest graduate students like John Teufel and Dave Schuster. They're out doing very well. John is sort of a leader in quantum electromechanics and kind of combining mechanics and quantum systems. He's at NIST in Boulder and has worked with Konrad Lehnert, who was my first postdoc who's a professor at JILA and CU. Dave Schuster is a professor in Chicago. Runs a very successful group there doing superconducting circuits and also other things. On the other side, people from matter include Jerry Chow and Julie Wyatt, now Julia Love. So, Jerry is one of the leaders of the IBM effort. And Julia is, I believe her title is Director of Quantum or something, at Microsoft. And another person who's out there and doing great things is Jake Embetta, who was a postdoc with Steve Girwin but worked with us very closely on these early circuit QED things, and he's really the director of the technical part of IBM's program, which is very large and very active. There's a lot of people. It's hard to name them all off. Another person who's doing very well is Andrew Houck, who's at Princeton, and is the head of their quantum institute and is the deputy director of this national quantum initiative hub that we put together just recently, which is Yale, Brookhaven, Princeton, MIT, and IBM. And a cast of thousands, I guess. So, I've been really lucky to have these key scientific collaborators, peers, like Michel and Steve and then to have really smart and talented postdocs and students who come in and can do things that surprise even me. And I think I'm not sure exactly what it is, but something about the way we run things and the lab culture or whatever seems to help foster people developing the independent skills. So, I think kind of in my PhD work we had a group that was very top-notch but was pretty separate. People didn't really work together on projects. They were sort of one man, one thing. And it made you strong, but it wasn't the best and fastest way to get stuff done. We liked to work in small teams within this larger collaboration or consortium we have at Yale, and I think maybe we just give people enough independence to define what they are doing and so on. That it seems to work well in terms of developing their ability to become independent researchers and lead their own groups and all that kind of thing. So yeah, I think--

Zierler:

Which touches on, it touches on the second part of my question, where I asked about the commonalities. What have you seen among your successful graduate students? They're obviously very different people, very different perspective, very different goals. But there are probably some things generally that they share. And I ask that within the context of-- I mean it's a good window into how the people behind quantum computing work, you know? The kinds of things that successful graduate students in particle physics, it's a very different skillset, right? So in quantum computing, what are the kinds of things in terms of the skillsets that people have, their creativity, their sense of adventure? You know, what are those characteristics that sort of orient towards a successful career in this push in quantum computing?

Schoelkopf:

Yeah. That's a good question. I'm not sure, I think there's a single phenotype that's most productive or... You know, I think everybody has strengths and weaknesses, and I think maybe a little bit for me, my career didn't take off until I got to Yale, and I recognized I had the right support and kind of recognized that I could work in a particular way and I kind of had to just relax and let it go and assume that good things will happen, but be careful about making sure that the experiment doesn't fail, and then kind of letting the creativity come? Like maybe as a graduate student, I was a little bit too verklempt and I had ideas but I'm like, “Oh, that's going to be too hard. I can't get that done in time in a PhD.” You're very worried about getting stuff done. I think some people's strength is that they're super original, super creative. Some people's strength is they're really, really diligent and disciplined, and so their experiment works the first time every time because they don't let a detail slip. And some people are just deep in the mathematics and have insight there or can rapidly do the calculations. I mean, there's sort of all different types. That's something I even do with new graduate students doing internships or something, is I often give them an internship project which is orthogonal to whatever it is they've had experience with before to just sort of try and give them the confidence to do the other thing, you know? So they won't get pigeon-holed. “Well, I was always doing computer simulations of this or that.” “Well okay, then don't do computer simulations. Let's have you build a widget.” And they don't always have to be good at it. But at least it exposes you to those other aspects. I think one of the things I love about experimental physics and what we do is that it requires so many different skills, and so many bits of knowledge. And you never know which bit of knowledge it's going to require. Maybe today I need to remember what the differential contraction between sulfur and aluminum is. My chips are all shattering. And oftentimes if you're working in a lab, you don't even know, like, well I'm going to go in and try to solve this problem and then I'm not even sure what I'm going to be doing in the afternoon. You kind of have to jump around and sometimes you get stuck and you have to jump to a different thing and work on a different thing. Another thing that becomes more important now is being able to manage complexity and actually being a good programmer. That's especially true, maybe, because we're in “quantum computing.” But I think it's also just a modern trend that you no longer-- I mean, when I was a graduate student, you literally kind of like would flip a switch, read a meter, and write by hand in a table in your notebook. Literally, right? And nowadays you actually are all up in the-- everything is computer-controlled, and you're always programming some complicated protocol of coordinating multiple instruments and different things, and being able to program rapidly and not get lost in that, and being able to manipulate the data in the right way is... But some of it still comes down to, well, am I thinking about the right thing right now? Am I really-- I've got a bug and I need to fix it, and I can do lots of things over here, but it seems like it's when I wave my hand over here that the bug comes and goes, and that must be where I need to focus my attention over there, even though... I can manipulate this robustly all day every day and see phenomena about the bug, but I'm not going to find the bug and get rid of it. So I don't know. I feel like that's a very poor answer. It's a good question.

Zierler:

[laugh] Well Rob, if that question wasn't hard enough, I'll save my hardest question for the last.

Schoelkopf:

Okay.

Zierler:

Which is, you know, a forward-looking question. I don't want to ask you about the next five years, because it seems like you have a pretty-- I mean, with all the caveats of who knows what's going to happen, but at least in terms of your goals or using powers of extrapolation, you have a pretty good idea of how rapidly things are going to advance in quantum computing within the next several-- in the short term, right? So my question is, it's the what's after that question. What then in a world where useful quantum computing is a reality, where it's integrated in our technology and our medical systems and all of the rest, right? What next? What next for you personally in terms of the science you want to pursue at that point? And what do you think the new questions might be most compelling, as a result of quantum computing just being a thing in our world now?

Schoelkopf:

Yeah, well, I mean... I think there's a few aspects of that. I mean, it would be wonderful to see that this thing which started as a basic science-driven question, “I wonder if...” turned into something that impacts the everyday lives of people. That's one of the reasons, I think, when you really look at it, that I wanted to stick with science and do that. It fit, it's fun and that I want to be good at it, but I also like the idea that as a scientist, you're investigating things. As a physicist, we don't always think about how it's going to translate into the real world. And it's pretty rare. I feel like it's more common if you're doing genetics research, just to pick something, you might well discover the gene for a disease and see it cured in your lifetime. That's not so surprising. I think for physicists, starting with a fundamental physics question and then seeing that turn into a world-changing piece of technology or whatever, it's a little bit less common--

Zierler:

I mean for example, does quantum computing possibly get us closer to what dark matter is? Just for example.

Schoelkopf:

Yeah. Maybe? I mean there's now interest in theoretical questions about how is entanglement related to black holes and cosmology and things like that. I don't know that I really follow it, but I think it's pretty interesting. We are finding ways, aside from building quantum computers, that the concepts of quantum information are helping to explain some other-- or giving alternate ways of tackling some of these questions. Some of these problems. So, you know, that can be a benefit or a real change, even before we have the quantum computers. I think most of us believe that the quantum computers are going to be really enabling for a lot of science applications first. That was also, by the way, quite a bit how conventional computing went. I mean they were sort of a host of problems like weather prediction and, okay, also things like nuclear weapons and things like that, but were topics where you needed the computational power. So, I think there'll be things in chemistry in many body physics. One of the things that our new hub is going to look at his how do we devise quantum algorithms that can address problems like lattice QCD and stuff like that. So, there may be ways to sort of differently address some of those problems using quantum computers. I think part of our belief and our hope, right, is that we will find ways to use quantum information and quantum computers that people haven't really grokked yet. My colleague Steve Girvin always likes to talk about the laser, right? So the laser was sort of I think developed because it would be a really great light source, but the maser and really the concepts that went into the laser, those were just mysteries in quantum mechanics. What happens if you have an inversion of the population in some gas or some ruby crystal or something like that? And probably people anticipated death rays and maybe scientific things for doing better spectroscopy and stuff like that, but did we anticipate eye surgery and fiberoptic communication?

Zierler:

Right.

Schoelkopf:

No. And I think that's one of the things that's appealing about physics or science as opposed to technology or later stage, more applied research, is that it could end up anywhere. But it requires a little bit of a leap of faith that even if it's not going to directly translate or I'm not going to see it in my lifetime, or if it ends up being irrelevant, somehow it's a thing we're adding on the pile of what our species knows and eventually it could be useful for something. To have that translate in our lifetime has been a little bit of a challenge, and something that we had to scramble and learn about and maybe we're still feeling our way as a field, and me personally as the right way to kind of do the research both on the fundamental and the applied side and the translational or commercialization side. But I think it's certainly going to be very interesting and very exciting, and I think the science is going to be going strong. I mean there's-- maybe we didn't talk directly about this, but there was a period—I think it's passing maybe now—but there was a period a few years ago where again as people were sort of like, “Hey, this is going to work. You can build quantum computers.” I even had a graduate student who said, “Well, I was really glad I got to work in your lab, Rob, because just a few years from now, it's all going to be done. This won't even be here. I'll be over at IBM or Google or something like that.” And sure, there's a lot of stuff going on over there, but no. I mean that's a bit of a misunderstanding of how the generation of knowledge and creation of technology and all of that goes. You don't investigate fundamental questions, tie them up with a bow, hand them to engineers who do engineering stuff or whatever they do, it's sort of magic, but I don't know. That's what physicists say sometimes. They both denigrate engineers and give them special magical powers that they can somehow make stuff work even though we couldn't really get it. And then eventually it goes into companies, and it's done like that. And maybe as a graduate student, you're trying to do a bit of knowledge, put it in a thesis, tie it up with a bow, and have your defense committee say, “Yay.” And then maybe you go off into industry or something like that. So, there's temptation to see that process. But when you do research, you ask questions and you sort of answer them, but it always leads to 27 more questions that are just as equally interesting, and that you can't always foresee which way they're going to go, which now, okay, I understand why they might be interesting to answer. And I understand a way I can address them where I can test my theories even more and so it's just going to... These things don't go away, right? We were all doing quantum physics quite actively before we had the algorithm, and the idea of building quantum computers. And we'll see where it goes.

Zierler:

So really what you're saying is, the real theme of what you're saying is, the most important thing is, you're keeping an open mind. That's the bottom line. No matter where this goes, you're priming yourself not to be surprised, because we don't even know what those questions are until quantum computing gets to that point?

Schoelkopf:

Yeah. I mean, I think we know-- The really good thing, I think, and I've been lucky to find this in my career, is you know enough to know that there's a good reason to explore this avenue.

Zierler:

Yes, right. Right.

Schoelkopf:

Right? And I think what I've found in research is it's very important not to say... And research isn't like, “I'm going to go to Chicago. I'm going to turn left, I'm going to go down the street, and I'm going to go here.” You can't plot out every move. You're like, “West. West must be cool. I've heard there are buffalo and stuff. Let's go west.” And first you try rivers, then you build highways. There are different ways to get there, right? And you just have to have some intuition, some insight, that that's going to lead somewhere good, and then you have to adapt along the way. You have to say, “Well, Chicago's not exactly where I wanted to go. I want to see all of that.” And there's benefit in getting to destinations that are adjacent to the one you thought you were aiming to get to. So yeah, I don't know, I guess it's been a long discussion. Maybe I'm losing my ability to put together a thought very coherently, but... I think we know enough, to sort of get back to your question, I think we know enough about where we're going to know that there's lots of really good things coming. But I think the flavor I would like to give is that there are probably the Rocky Mountains and another ocean and all of that out ahead, even though we know that there's fertile farmland in Ohio. [laugh]

Zierler:

Right, right. [laugh]

Schoelkopf:

Sorry, I'm reading Hamilton right row, so I'm thinking about American history or something.

Zierler:

Well, Rob, on that note, I mean, it's a great answer because it's just, it's whatever it is, it's going to be exciting. Wherever this is headed, it's obvious that it's going to be exciting, and part of the excitement is the mystery. So I want to thank you for spending this time with me. I had a lot of fun with you, and...

Schoelkopf:

Thank you. Yeah, it's great. Good questions.

Zierler:

I'm really happy that we had this opportunity to use this format to treat these topics, which are so often understood in the media so superficially, right? That this is really a unique format to really flesh these things out and give you the opportunity to convey these ideas to the quite broad audience that comes to the Niels Bohr Library. So, Rob, thank you again for spending this time with me. I really appreciate it.

Schoelkopf:

Awesome. Thanks a lot.