An interview with Pascal Del’Haye, the first author of the original Nature paper on Kerr frequency combs and now a group leader at the National Physical Laboratory (NPL) in Teddington by London. We talked about the first Kerr frequency combs, the following twelve years, the spirit of fundamental research, problems with names and more.
Pascal, to the Kerr comb community you are known as the man who, as a student, did the pioneering experiments on these types of optical frequency combs. For these first experiments, what was planned and what actually happened?
Well, it was actually quite interesting. When I first visited Tobias Kippenberg’s lab, he actually only started at MPQ (Max-Planck-Institut für Quantenoptik, Garching) and there were just empty tables and no experiments yet. One or two months later I moved to Munich and started as a diploma student in his group. The original plan for my diploma thesis was to work on photon correlations based on the four wave mixing that Tobias had discovered already in Kerry Vahala’s group during his time as a PhD student. He saw these parametric sidebands generated in microresonators and my diploma thesis at that time was to look at photon correlations because the photons are generated in pairs. This is quite related to the work that Roberto Morandotti is doing right now on time-bin entanglement between photons. In the very beginning we started fabricating the first micro-toroids in the cleanrooms at LMU (Ludwig-Maximilian-Universität München). Essentially, we were just playing around and trying to reproduce some of Tobias’ earlier experiments. I had my first batches of high-Q micro-toroids fabricated at LMU and we had an erbium-doped fiber amplifier and we were just trying to see what happens when we send a lot of power into the resonators. We were quite surprised when we suddenly saw these comb-like structures with hundreds of modes being generated. And, well, then it was a bit being at the right time at the right place. At MPQ, Tobias Kippenberg’s group was part of the Hänsch division, who had invented the frequency combs, and then there was Ronald Holzwarth who had this fiber-laser frequency comb sitting around, one of the first ones or maybe even the very first one they had built at MPQ. So we could use this fiber-laser frequency comb to compare our microresonator comb with and measure the equidistance of the comb modes. At that time it was not really clear whether the microresonator comb modes were equidistant.
So, the original goal was to do correlation measurements for you diploma thesis. Was there ever any work done on this or did you immediately switch to frequency combs?
Actually, I started building a setup that was supposed to separate the signal and idler with extremely high fidelity without losing any photons. I did a few weeks of work on this before we discovered the frequency combs. Then we changed our experimental plans entirely. The high efficiency grating for the photon separation is probably still sitting somewhere in a drawer.
The main results that eventually ended up in your Nature paper, these were already achieved during your diploma thesis? When was the transition from your diploma thesis to your PhD?
Yes, these results were during my diploma thesis. Let me think. Ok, I have to look into my lab book (he checks something on his computer). So, we definitely did the first equidistance measurements during my diploma thesis (he studies his notes on his computer for a bit). Yes, the story is as follows. We did the first measurements with a multi-heterodyne comparison between the fiber-laser frequency comb and our microresonator frequency comb but we didn’t know anything about precision metrology with frequency combs so we did the measurement with an electronic spectrum analyzer with kHz resolution. We saw that all the modes were equidistant on a kHz-level and we submitted these results to Nature. Of course, being about frequency combs, it went to some hard-core metrology people and one of the reviewer responses was “[…] the precision of the measurement is not very good, only at the 10‒10 level”. That was the end of my diploma thesis and it was a good motivation to continue working as a PhD student on this project to improve the measurement precision and resubmit the paper. The first months as a PhD student I essentially worked on more precise frequency counter measurement of the equidistance.
When approximately was the first measurement of a frequency comb?
The first, the very first measurement… (checks again his lab book on his computer) …well, I saw two heterodyne beatnotes between the two frequency combs on the 22nd of August 2006 and then I have a folder that says “three beatnotes” on 24th of August 2006. So, if you count three beatnotes as a frequency comb, it was at end of August 2006. And in October 2006 we went up to nine beatnotes at the same time. That was probably more convincing.
You already talked a little bit about the reviewing process. Everyone who has published in journals with big names know that publishing can take as much time as doing the experiments. What was your experience at that time?
As far as I remember we got quite favorable reviews and it was published in the end. So everything was great but the main problem in the beginning was that the paper did not get too many citations in the first year (as a Nature paper). For me it was a good lesson. If you do really new research it is maybe not the most popular research topic in the beginning because there is nobody else working on the same subject. However, this rapidly changed in the following years and currently the paper is one of the most cited photonics papers of recent years. I was very happy to see other research groups to start work on microresonator combs. The groups of Lute Maleki and Alex Gaeta were the first ones catching up on this new topic. But, yea, I mean the beginning of this work was quite lonely in the sense that when I went to conferences I was the outsider among all the other conventional frequency comb people. It was really great to see this whole microresonator frequency comb research grow into its own field. Now there are full sessions at conferences only on this topic, which is really amazing.
And the conventional frequency comb people, were they happy about the new contribution to the field or did they see you more as the weird guy with the weird combs?
We got some complains that if a comb is not stabilized in terms of fCEO and frep then it does not count as a real frequency comb. But in general it was very friendly. The optical frequency comb community was also still quite small after all. We were not directly competing with conventional combs, so it was just an interesting new system which looked quite exiting. Regarding the question of stabilizing the frequency comb, this was then my follow-up paper after the Nature paper where we showed that we can actually control the comb modes and stabilize the frequency comb. So this was also solved in some way. More recently even with octave spanning combs with f-2f and 2f-3f self-referencing of the frequency combs.
The first stabilization was your later work during your PhD. After your time at MPQ you went straight to NIST, right?
Yes, that’s right. Because Tobias Kippenberg moved to Lausanne at the end of my PhD, I worked for a short time at Menlo Systems until my wife was finishing up her PhD in Munich. But then, half a year after my PhD defense, I moved on to NIST.
And from NIST you went straight to NPL?
Yes.
Your carrier path is then all about microresonators and frequency combs, right? Your diploma thesis, your PhD thesis, your post-doc and now your PI position.
That’s right. At NPL we started to do some more general experiments on nonlinear optics and the interaction of counter-propagating light but a lot of my research career has been dedicated to frequency combs in microresonators. Even after these many years it is still a fascinating research topic with so many practical applications and still full of surprises in terms of fundamental research.
You followed the field from the very beginning. Give me a couple of paper which you recall from the top of your head. The papers that caught your attention.
Well, we had first started off with these really small resonators with very large mode spacings where you got essentially Turing pattern combs or primary combs that did not go through some noisy regimes. After that we started to make bigger resonators to be able to measure directly the mode spacing and suddenly we had these noisy and chaotic combs. It was a bit of a rollercoaster where I thought that maybe microresonator frequency combs are not useful and cannot be used for precision metrology. At least not the ones in bigger resonators. As a relief to this problem, one of the most exciting papers was Tobias Herr’s soliton paper where he saw the soliton mode-locking of combs in bigger resonators. I think this was a very important paper.
In terms of other papers I enjoy all the first papers where combs have been demonstrated in a new material platform. Being it magnesium fluoride, Hydex, aluminum nitride, or diamond, or silicon nitride, which is certainly one of the most interesting platforms. I mean, this is probably still one of the most promising platforms for developing products for out-of-the-lab use.
And on the theory side of course there was the work about the Lugiato-Lefever model to describe the comb generation. That was also very exciting. Because I still remember, during my time as a PhD student Tobias Kippenberg was often suggesting to sit down and develop a theory how the comb generation works. At that time I had the feeling that it was a too complex problem and probably a whole PhD project by itself. That’s why I was really excited to see the work by Stephane Coen, Miro Erkintalo and coworkers in New Zealand. And seeing recent research, it feels like there could be whole conference sessions filled with the theoretical description of microresonator solitons and frequency combs, in particular with new approaches including χ(2) nonlinearities.
You mentioned the problem with the incoherence. I think the first time it showed up was in your octave-spanning comb paper where you showed this broad beatnote.
Yes, that’s right.
I sometimes ask myself this question: If the first combs by accident had been incoherent, would this whole research field exist right now?
That’s a very good question. These fused silica resonators at 1550nm are just such a magical platform. It was close to a miracle that everything came into place in the first experiments such that the comb generation worked. Also in terms of dispersion. What if we would have used a different wavelength? I think it was a really lucky coincidence that everything worked out in the beginning. We had the right resonator size, the right material, the right laser wavelength, the right dispersion. I guess I would have abandoned these experiments and continued with the originally planned photon correlation measurements if we would have used different resonators in the beginning.
Also the photon-correlation measurements had some Nature and Science papers recently. So you might have not ended up much worse in terms of publications.
My research career would have been certainly quite different.
You said, everything came into place but it is also the person who does the experiments who has the patience. What I heard about you as a PhD student is that you really had the patience and the skills to make things, which look impossible to do, work long enough to get some publishable data. Now as a PI do you still use these skills in the lab?
Unfortunately I do not have too much time anymore to do something myself in the lab. But what I tell my students is to play around in the lab but once something is working to actually make sure to get good data and to get publishable data. Even if you have to do the measurement ten times or 50 times. But maybe back to my diploma and PhD time. It was a lot about patience with these frequency counter measurements. It was a total nightmare when somebody would drop a screw driver in the lab then the whole measurement would be destroyed and you had to start all over again. According to my parents I spent many, many hours doing jigsaw puzzles as a child. Maybe this trained me for having the right amount of patience and persistence in the lab. But the other important thing at the beginning was that there were a lot of other people. It was Tobias Kippenberg who had this great feeling for what are important experiments and what to do. On the other hand there was the PhD student Albert Schliesser, with his indispensable know-how about multi-heterodyne frequency comb measurements from his diploma thesis. It was a quite interesting atmosphere in Tobias’ lab in the beginning because there was essentially just one PhD student and then there was this bunch of four or five diploma students and Master students. It was a very dynamic and inspiring atmosphere and everybody was exited to try something new. This probably helped quite a bit to find unconventional solutions to problems. Essentially in the end it was a lot of “playing around in the lab” that was leading to many groundbreaking ideas and experiments.
So that was back during your time at MPQ, then you were at NIST in Boulder now you are at NPL in London. How would you compare the atmospheres at these places?
MPQ certainly provided an outstanding research environment with many excellent scientists and a great research atmosphere. The same has been true for NIST in the group of Scott Diddams, with the difference of working at a place with more permanent scientists and a more relaxed research atmosphere. Both NIST and NPL are focused a bit more towards applied science. In particular NPL is collaborating a lot with industry for the development of new technologies into products. That is also getting interesting with microresonator frequency combs. But, for myself the transition between all these stages is to take on more and more responsibility and having, unfortunately, less time to work in the lab. In particular starting at NPL, I have been quite aware that it would be a transition into managing a research group and making decisions on research directions instead of actually doing research myself. I see this as a next step in my career, however, I hope at some point I will find again more time to be in the lab and do experiments myself.
Going to the applications of Kerr frequency combs. It has been over ten years of research now. If you take a Kerr frequency comb paper today, the introduction reads very much like the one of your initial paper. They could be used for telecom, astrocombs and so on. Of course the content of today’s papers is much more advanced. But what would be your bet, what will be the first commercial application of Kerr frequency combs?
Well, I mean one of the first commercial application you can kind of see from Tobias Kippenberg’s spin-off company Ligentec where the resonators themselves are being sold as products. Also OEwaves has been selling microresonator systems since many years. In terms of more sophisticated applications, I would think probably one of the most straight forward things would be microcombs as channel generators for telecom applications. Mostly because it is one of the applications where you do not necessarily care too much about the stability of the comb modes. So you don’t have to deal with stabilizing the combs. You can work with free-running combs which might be slightly drifting in offset frequency and mode spacing. This might be one application. The other applications could be some simple spectroscopy tools or calibration tools where you use comb modes to calibrate spectrometers for example but also compact ranging systems.
And in order for Kerr frequency combs to get there, do you see one particular critical aspect that one has to get right to go from the Kerr frequency comb research to a technology?
I think the main challenge right now is more an engineering problem. To combine the right laser source with a microresonator in a stable way and to make sure that the pump laser is resonant with the cavity. So it needs some sort of smart and robust control to generate a frequency comb after turning on the system. Even nowadays it requires a student to turn some knobs. So this essentially has to be automatized and simplified. I guess the main challenge is the combination of smart software control along with a robust hardware implementation.
As far as I can see from your recent publications there is not so much about Kerr frequency combs since you started at NPL. The publications are more about nonlinearity and non-reciprocity. Are we going to see anything on Kerr frequency combs from your lab anytime soon?
Hopefully this afternoon we will submit a paper on Kerr frequency combs where we actually look at more robust ways to generate frequency combs with a reduced sensitive to pump laser frequency fluctuations. So, yes, we have one new paper coming up hopefully in the next weeks.
You submit to Arxiv or somewhere else?
We are submitting to a journal but hopefully in the near future also to Arxiv.
I am looking forward to see the paper! Staying with Kerr frequency combs for a little bit longer. One of the first posts here at kerrfrequencycombs.org was about the name of the field. Because I found it inconvenient that people call it Kerr frequency combs or microresonator frequency combs or microcombs or some other name. Your original paper called it a Kerr frequency comb but had microresonator in the title. What is your opinion on the question of the name of the research field?
I think Kerr frequency comb is a very catchy name. One drawback I found was that for example Ti:Saph mode-locked lasers that work with Kerr-lensing are essentially also some sort of “Kerr frequency comb”. In my most recent publication I just call it microresonator frequency comb which is unfortunately a bit long for titles. I sometimes shorten it to microcombs. Recently, paper titles shifted a bit towards “microresonator solitons” or the longer version of “dissipative Kerr solitons”. Probably it would be nice for people outside the field to only have one name. But hopefully search engines are good enough to still give you all the relevant papers if you search for any of the names.
The rise of the AI (laughter). Staying with names. Your last name has a sign in there which is not typically in names. I am not entirely sure which character it is.
An apostrophe.
Is it just a plain apostrophe?
I think it is, yes.
Because if you look into similar characters in particular in Unicode, there are plenty of options. Look here. I went into ASCII and Unicode. The first two on the left exist in ASCII and in Unicode. The two on the right only exist in Unicode. Can you say which one is the correct one?
When I sign up on websites I usually us a space because it never works with any type of apostrophe. But, the version I prefer that is the one on your keyboard. I think it is some sort of quotation mark.
Yes, this is the one above on the very left. Does it look alright, the U+0027?
Yeah, yeah.
The Unicode apostrophe would be the U+2019. Which I think looks nicer typographically.
Yes, but it would probably destroy computer systems and the internet when you use it. Actually, every time I go to the doctor somewhere and they look in their list of patients for my name they usually never find me because there is always some problem with my last name.
Ok, but one can use the U+0027 character for your name?
Yes, this at least works sometimes. If it is not converted to Del'Haye, which is difficult to pronounce.
One question that is not related to Kerr frequency combs nor to your name. But it is a big topic in Europe: Brexit. Does it affect you? You are still German citizen, right?
Yes, that’s right. Well, in the UK Brexit is a big topic among scientists right now. The problem is that nobody knows what is going to happen, in particular with respect to Horizon 2020. Currently the majority of my research funding is coming from European research projects. And right now this is a big question how it will continue in the future.
One last question going back to what you said about your original Kerr frequency comb paper and the reviews then. Now there is a trend towards publishing peer reviews such that you can see the reviews online when the paper is published. For example in Nature Communications. What do you think, would it be interesting for the community to read the reviews of your Nature paper?
I would have to look into the reviews. From what I remember, the reviews from my original paper would not be so interesting scientifically. But in general it would be a good thing to have public reviews because it would help to make the reviews more constructive and more focused on what could be improved instead of making rather general remarks. What I would like to see is a general platform similar to Reddit where anyone could comment on papers. This might be a good thing or a bad thing, but at the same time it could include a voting mechanism for the quality of a comment. That might be an interesting alternative system for scientific reviews.
Some platforms like this do exist. One is PubPeer. I actually like the initiative of open peer reviews. Also the possibility to credit reviewers. We will have to see.
Yes, I think it would be a good thing to have an open review process. It might generate a friendlier atmosphere.
That’s it with my questions. Any additions from your side?
Maybe some very general things that connects back to the first part of the interview. How this whole work started in Tobias Kippenberg’s lab. It was essentially a lot of research just out of curiosity and not only driven by the plan of a research grant. I think this is important thing which I try to convey to my students that sometimes it is important to leave the beaten track and try something new. If we would not have done this in Tobias’ lab in the beginning we would not have ended up with discovering these frequency combs. It also comes from Ted Hänsch’s “Curiosity driven research”. Do you know this cartoon with the chicken (Update: now better visible here)?
Hhm, no… (quickly searching for it on the internet) …ah, I have seen it before, yes.
I always see this as an example how we did research during my time in Tobias Kippenberg’s group and I am now trying to implement this also in my own research group.
If you want to read Pascal’s PhD thesis, go ahead.
Disclaimer: Pascal and Victor have done their PhDs with the same PI, Tobias J. Kippenberg. One at MPQ one at EPFL. So there was no overlap.
This work is licensed under a Creative Commons Attribution 4.0 International License.