Notice: We are in the process of migrating Oral History Interview metadata to this new version of our website.
During this migration, the following fields associated with interviews may be incomplete: Institutions, Additional Persons, and Subjects. Our Browse Subjects feature is also affected by this migration.
Please contact [email protected] with any feedback.
This transcript may not be quoted, reproduced or redistributed in whole or in part by any means except with the written permission of the American Institute of Physics.
This transcript is based on a tape-recorded interview deposited at the Center for History of Physics of the American Institute of Physics. The AIP's interviews have generally been transcribed from tape, edited by the interviewer for clarity, and then further edited by the interviewee. If this interview is important to you, you should consult earlier versions of the transcript or listen to the original tape. For many interviews, the AIP retains substantial files with further information about the interviewee and the interview itself. Please contact us for information about accessing these materials.
Please bear in mind that: 1) This material is a transcript of the spoken word rather than a literary product; 2) An interview must be read with the awareness that different people's memories about an event will often differ, and that memories can change with time for many reasons including subsequent experiences, interactions with others, and one's feelings about an event. Disclaimer: This transcript was scanned from a typescript, introducing occasional spelling errors. The original typescript is available.
In footnotes or endnotes please cite AIP interviews like this:
Interview of William R. Bennett by Joan Bromberg on 1987 October 26,
Niels Bohr Library & Archives, American Institute of Physics,
College Park, MD USA,
For multiple citations, "AIP" is the preferred abbreviation for the location.
In this interview, William R. Bennett discusses his work with lasers. Topics discussed include: helium neon laser; Ali Javan; Donald Herriott; Columbia University; Yale University; Air Reduction Company (Airco); Linde Air Products Company; Hugh Robinson; Lewis B. Headrick; Ora S. Duffendack; Bell Labs; Harrie Stewart Wilson Massey; Eric Henry Stoneley Burhop; Allan C. G. Mitchell; Mark W. Zemansky; Rudolf Walter Ladenburg; Sidney Millman; Technical Measurement Corporation; Ted Geballe; Conyers Herring; A. L. Schawlow; Henry Scovil; Harry Nyquist; Jim Gordon; Gardner Fox; Tingye Li; E. I. Gordon; G. D. Boyd; A. F. Turner; Bausch & Lomb Laboratories; Technical Research Group (TRG); Gordon Gould; Charlie Townes; Theodore Maiman; Deming Lewis; A. T. Forrester; George Dacey; C. Geoffrey B. Garrett; Ross McFarlane; Bruce Bogert; Willis Lamb; Paul Rabinowitz; V. P. Chebotaev; John. W. Knutson; Air Force Office of Scientific Research (AFOSR); Army Research Office, Durham (AROD); Vernon Hughes; Lloyd Wood; Marshall Harrington; Alfred P. Sloane Foundation; Institute for Defense Analyses; Bob Collins; Kurt Shuler; Walter Faust; argon laser; Spectra Physics; Bill Bridges; Grant Fowles; William Silfvast; Bill Walter; Marty Pilch.
Do you want to go through this list of questions?
Well, why don't we start with the questions and then sort of branch off where it seems appropriate? And I'll probably ask a few other questions as you talk.
All right, fine.
As I said, I thought we might just start by seeing whether you thought there were any distortions or omissions that we can talk about in the work you were doing on the helium neon laser with Ali Javan and Donald Herriott?
Well, I'm sure there are slight omissions here and there, and probably distortions. Everyone who has had experience in these areas will have his own particular perspective.
I should say, this is an early draft.
I think you've done a pretty creditable job, going through all this and trying to look at all these different things, but of course there are little things here and there that I no doubt remember better than others, and as we go along maybe I'll think of what they were. Let's see, your first question on the list here was, how would you evaluate the draft treatment I sent on the neon laser work? As I say, I think it's a fairly nice rough overall view of it. There of course was infinitely more detail involved in the actual work than ever would get into this, since particularly Ali and I were working perhaps 90 hours a week over much of that period, and taking lots and lots of data and looking at many different things, and obviously you're not going to fill up your article with all that stuff. I think the main involvement I had with that was probably adequately covered in various depositions I've given in the past and articles I've written, so I'm not sure there's any point in going through all of it. I might just give a quick picture of how things went. My own interest in this area, of course, largely dated from the research I'd done at Columbia, and also a little bit at Yale, on things having to do with excitation processes in gasses. The work I did at Columbia on my Ph.D. thesis in particular dealt with excitation transfer among these different gasses, particularly long-lived metastable states that are of significant concern, and I indeed had looked at most of the states of neon that were, in one way or another, that were involved with the later laser work, and I of course looked at a lot of problems involving excitation of the impurity atoms by helium metastables, the neutral atomic metastables, the molecular helium metastables, the metastable molecular helium ion HE2+ and so on. There are a large number of these different carriers involved and different excitation processes, and in December of 1958, I was approached by Ali, whom I'd known for many years while we were both graduate students at Columbia, and I was at that time in Yale University, about the possibility of undertaking some sort of a joint study of the possibility of getting laser action in gas discharges, and we sort of free associated for a while at that point on various possible directions we might go, and we more or less agreed at that point to write a joint paper outlining all these things. I gave Ali a copy of my thesis at Columbia, and went over some of the things I'd noticed there, and he mentioned some of the things he'd been thinking about. In particular he asked me what I thought of the possibility of neon as a medium for getting a population inversion, and also of the transfer of excitation from helium metastables to neon, and both of these things sounded like they were very fruitful to me. I had had some indirect evidence in my thesis of unusual population accumulations on the very levels that ended up being used as the upper laser states in the helium neon system. And some of the data that I'd taken, particularly at Yale in optical pumping experiments I'd done with Hugh Robinson strongly implied that excitation transfer from helium to neon was occurring. One of the things you notice, if you take a "spectroscopically pure" sample of helium, that we used to get from companies like Airco, Air Reduction Company, or Linde and other places, is that if you look at any sort of spectrum in the optical range that you can get out of those samples, either exciting them with a tesla coil or as I was doing in my thesis, exciting them at high pressures with high energy charged particles, alpha particles, positrons, electrons and so on, beta particles, one sees to a large extent a predominant emission by species other than helium. In particular, what you normally see, if you look at a helium sample of this type, are the first negative bands of N2+, that is a predominant feature of the spectrum was molecular nitrogen bands. These were enormously more enhanced than anything which was characteristic of helium itself, and I spent a lot of time tracking down the various routes of excitation that would be involved in exciting these things, and came to be aware that there were many different ways that long- lived carriers of energy could diffuse about in a high pressure gas and excite these impurities, among them the helium 2 triple S (?) metastable state, certainly also probably the 2 single S metastable state, although a less likely source at very high pressures, and in particular, that the metastable ions HE2+ and the atomic ion HE+were important carriers of energy which could excite states by charge exchange, and also, there was the role of other things such as dissociative recombination that might lead to radiating states of various types in those gasses. Anyway, there were as large number of different things that could go on, and one of the things that I was generally concerned about in that early period was just how much involvement there might be, both advantageous and deleterious, to the operation of a laser in these systems. That is, one could depend on one reaction perhaps to get an excitation of an upper level that might be suitable for laser purposes, but there might be some other processes that would go on that would populate the lower level, and thereby obscure the population inversion or detract from it.
By the way, were you just talking to Javan at this point, or were you also in touch with the people at Columbia or the people at TRG?
I was not talking to other people at Columbia or at TRG at that point. I did have a lot of close conversation with one of my colleagues on the faculty at Yale, who was Hugh Robinson, who is now at Duke University in the physics department. We had been doing optical pumping experiments for a couple of years. I went to Yale in the summer of 1957, after finishing my research at Columbia, and worked together with Hugh Robinson during much of the period between '57 and my later departure for the Bell Labs. One of the things we'd done a lot of was purify helium samples, both helium 4 and helium 3, for these optical pumping experiments. We were looking at resonances and metastable states of helium and we were also looking at free electron resonances as excited by electron exchange collisions with optically aligned helium metastable states and so on. And in the process of doing that, and also in connection with previous work I'd done purifying noble gasses, it seemed clear that when you take a typical spectroscopically "pure" sample of helium, you would start removing impurities, first thing you'd do was get rid of the nitrogen bands. That was fairly easy to do by using barium getters and methods of that sort, in ultra-high vacuum systems. However, once you got rid of the nitrogen, which one could easily get down to levels near a part in 10 to the 9th or 10 to the 10th, through this technique, you were left with other strange aspects of the spectrum, namely, the spectrum would turn from a blue color, a dark blue color, to an orange-red color, and most of the lines that were emitted there were not helium lines, and I began to wonder more and more what was being left there, and the thought that it was neon came to mind, and after having this discussion with Ali in December, I went back to Yale and actually did a little spectroscopic experiment with Hugh Robinson, where we took a high resolution visible spectrograph in the physics department at Yale, and looked at alternately a highly pure, in the sense that gettable impurities had been removed, sample of helium, and compared that with a "spectroscopically pure" sample of neon, provided by the Air Reduction Company, and we found that the two spectra were essentially the same. That is, there were neon lines in both the helium sample and the neon sample of about the same relative intensities, and very little helium. And I first mentioned that to Ali next time I talked to him, and indeed, that was the first direct experimental evidence I had seen implying that there was absolutely a mechanism for exciting very trace amounts of neon in helium. I think actually that was significantly before any of the work was done at the Bell Labs. This was just a photographic plate. It wasn't something where we controlled relative partial pressures of one gas and the other, but it did demonstrate that there was a very strong source of excitation in neon and helium samples. I had not read, at that point, any of the papers dealing specifically with that research. There was a paper published by L.B. Headrick and O.S. Duffendack in the 1930s which really covered this territory extremely well, and I had not seen that paper until much later. As a matter of fact, one of the patent lawyers at Bell Labs brought it to my attention during the deposition I made, I think it was in 1968.
It must be this one.
You have the Headrick and Duffendack paper there?
No, I meant, it must be this deposition.
Oh, yes. Anyway, that paper really reported —
I might just pull this out, in case you want to see what you said there, and what you want to add.
Well, I don't have too much to add on that specific thing, but it was brought to my attention by one of the patent lawyers, I think Art (?) was the one who did that, and in any case, it was certainly a very interesting paper, to learn of after we'd done all the research at the Bell Labs. It would have saved us a little time and effort I think if we'd known what Headrick and Duffendack had done in detail. But the main point of all this is that there had been some data in the literature, if one had done the literature research with sufficient thoroughness, which indicated exactly what was happening in these cases, and indeed Headrick and Duffendack saw enhancement not only of the lines that were attributable to transfer from the helium metastable level, but 2 triple S of the stable level, but also from the single metastable, and of course we were working with photographic plates in the visible spectrum, and I suppose to some extent, although I haven't looked at this paper in a long time, we must have made indirect inferences regarding what was going on in the infra-red.
I have certainly wondered whether all of this was just directed to getting the laser out, or whether there were scientific questions that you were pursuing at the same time.
Yes. Well...there certainly were scientific questions that were of great interest to me. As I say, I had worked for many years in this field, and I was therefore concerned with the basic physics that was involved, and studying the entire process, and I was therefore very interested in following it up, doing whatever measurements could be made that would add to knowledge of this area. One thing I started to say, in terms of background material that was available at that point, was that my introduction to collisions of the second kind and that type of excitation mechanism had largely been through a book by Burhop and Massey called ELECTRONIC AND IONIC IMPACT PHENOMENA, and I went through that very carefully when I was working on my research at Columbia, and at one point essentially knew it by memory almost. I looked up most of the papers that had seemed very interesting in relation to this subject that were cited in that major review work. But one of the things that Burhop and Massey neglected was the papers of Duffendack and his students. I didn't come across those until much later. On the other hand, Ali I think in his review of the subject was largely looking at the work on resonance radiation by Mitchell and Zemansky, which also included references in some instances to Duffendack and certainly Ladenburg's work, and probably was more oriented towards background references that were found in the Mitchell and Zemansky treatise. For some reason, I had a great deal of difficulty getting hold of Mitchell and Zemansky. It was out of print at that time. I've since picked up a few copies of it one place and another. But in any event, my own personal introduction was largely through the Burhop and Massey tome. And they cited lots and lots of work in this area, including the work of Maurer and Wolf in Germany and other places. Those were the sorts of things that I followed through. In the process of studying what had been done before, and in my own research at Columbia, I was accumulating empirical data on the probability of excitation transfer, essentially accumulating in the form of a plot of a log scale, of cross-sections for the transfer process versus energy defects, in the initial and final states. The general characteristic of this tabulation of data that I put together was that of a dart board. There wasn't any simple rule that indicated that the largest cross-sections would occur where the minimum energy defect occurred. That sounded nice but —
— energy defect is the difference —?
— the energy difference between the initial and final states, where we consider how a metastable carrier, say, a star (?) colliding with atom D in its ground state, and the excitation is transferred so A jumps to the ground state and D goes up to another state, it might be A double, D double star, plus an energy difference. Now, it was usually implied in the simple theoretical models of the problem that as that energy difference went to zero or became very small, that the transfer cross-sections could become very large, and indeed, there was a general trend in that direction, but if you actually looked at real data that had been taken over decades where this process had been studied, we found that there was an order of magnitude scatter about it, and there was no simple rule that said that having a small energy difference would necessarily give you a large transfer cross-section. And the details of what was going on there were interesting to me, and I was therefore concerned with any method we could use, in studying this kind of process quantitatively, particularly with the advent of ultra-high vacuum technique, in other situations where we could minimize the effects of impurity contamination of the experiment, which was always a big problem of the work done in the early thirties. As it's turned out in retrospect, many of these cross-sections depend very strongly on bumps in the inner molecular potential curves that make the cross-sections energy-dependent as well as very strongly dependent on the energy difference between the initial and final states, and there's an awful lot going on there that tends to affect the specific behavior of individual systems.
Is that worked out by now?
Well, some of it is and some of it isn't. I was actually quite disappointed when we started making our measurements in the helium metastable transfer cross-sections of the 2 triple S state, that is, to neon. We made some very careful measurements of the total destruction cross-section of helium metastables under various mixtures of helium and neon, and did this very carefully, especially at room temperature with relatively low excitation levels of the system, so that there would be a minimum problem of electron impact obscuring results and recombination products and so on. We did actually measure that cross-section, and it turned out to be about 3.7 x 10-17 centimeters squared, under the conditions of our experiment. We did this with only a few percent error, actually. We built a special cell that was quite large and we very accurately calibrated the volume ratios between that cell and another filling cell so that we could really control the neon pressure which we measured directly with a capacitance manometer that was calibrated with a mercury manometer, in such a way that we could vary the neon pressure over a wide range, and as I say, we found a cross-section that was about a little under 4 x 10-17centimeters squared. For the particular energy difference that was involved between the initial and final states, I was expecting a much larger cross-section. I thought it might be an order of magnitude larger, and as I say, I was rather disappointed because of that. Long after the laser had been demonstrated to work, and particularly the transitions that were excited by transfer from metastable level, the experiments were done both by myself and by Arecchi in Italy which showed that the transfer rate increased very drastically as the gas temperature went up, and I think the interpretation of that is based on a small bump in the inner molecular potential curve, when you worry about the transfer process, which represents a threshhold which the kinetic energy of the particles has to overcome in order to reach the crossing point where the transfer occurs — that is, viewing the whole thing in terms of the old (?) avoid crossing theory. In any event, what one finds is that the transfer rate goes up very substantially as you heat the gas. If you just wrap heater tape around the cell and heat it, you find that the transfer rate goes up, and indeed, it does go up to values that are more in the ballpark of what I would have expected initially. At any rate, that sort of detail is lurking around in the background, which is of very great interest theoretically, in terms of understanding this type of process, and it also has a major effect on whether or not these devices will actually work.
Did that low cross-section affect the strategy you were following at the time?
Yes. It clearly was going to provide some sort of a limit on the rate at which the excitation could be transferred to the neon system, and therefore was an unfortunate result, in terms of the kind of gain that one might expect to get. There was another area that was also of major critical concern, and that had to do with the lifetimes of the upper and lower states. That is, assuming you had a selective process for populating the upper neon level, and didn't have sources of lower state excitation of any significance, which is of course another assumption, one still needs to have a reasonable ratio in lifetimes of the upper and lower state, in order to be able to get the population inversion at all. And it turned out that the values that were available from Ladenburg's work on the lower state population were substantially in error. There had been some data taken, I think it was based on a Jamin interferometer, but it was based on a slightly indirect way of getting the transition probabilities from the lower laser levels, for the lower laser levels, of the helium-neon system, and hence a measure of the lifetimes of the lower levels. Ladenburg's data was off by a factor of 2, in the wrong direction.
Making them shorter?
It looked like they were going to be shorter by a factor of 2 than they actually were. And also, about that time, there had been some data recorded in the Russian literature by a pair of people whose names I've now forgotten, but it was something like — well, I won't try to guess at it. Actually I could look it up if you wanted. It's my paper here that appeared in the Soviet literature, which implied that the upper state lifetimes — I'm sorry, it implied they had made direct electron impact measurements of the lower state lifetimes, which gave values that were too long by a factor of something like 5, so that the data that had been available in the literature on the lower state lifetimes varied by an order of magnitude, and it was all wrong essentially, and all we could do is make approximate calculations of what these lifetimes were, and hope that things would work out. Well, we did make such calculations. I had previously made lots of calculations using the Bates and Damgaard Coulomb approximation for computing such things, and showed Ali how that sort of thing worked, and loaned him my copy of Bates and Damgaard at one point actually for him to make some of these calculations too. But the general conclusion from those calculations looked like the lifetimes must be all right, but marginally so, so it seemed the important thing to establish was, what these lifetimes actually were, and one of the experiments we therefore set up at the Bell Labs was a method of making such measurements fairly accurately. We set up a large electron gun, which is described in various of the papers published in that time, and worked out a photon counting method which is based on essentially counting how many photons we detected as a function of time after the excitation pulse to produce a certain level was turned off. And although the initial work we did on that was a little rudimentary, it turned out to indicate that there was about a five to one difference in the crucial lifetimes involved, and in the right direction to get a population inversion on a continuous basis. The question of what these lifetimes should have been, how accurately you can calculate them, of course represented an interesting problem in theory too, and it was therefore interesting to try to get as reasonable data in that direction as we could.
Where did your method stand, in terms of lifetime methods generally? I mean was this in itself a significant result, this method you used?
I think it was, actually. There had been some work done by delayed coincidence measurements in England by Herron, McWhorter, I think it was Herron, McWhorter and Roderick, if I remember correctly, and they had used a kind of rudimentary apparatus for it, in which they had a variable delaying line that they put in the system and bombarded gas under different pressures with electron impact and counted as a function the delayed coincidence target (?). It occurred to me that the only meaningful way to do that, tolerable way in terms of getting reasonable data, was actually to set up a multichannel delayed coincidence analyzer which would essentially look at all the different delay times simultaneously. The basic problem in this kind of measurement is painfully low counting rates, and if you just sit at one point in time and count and then move to another point in time and count, it just takes forever. By a parallel approach of taking the data, you can speed things up.
Sidney Millman's going out and raising $35,000 for a multichannel analyzer, that the firm of TMC —
— TMC, Technical Measurement Corporation, which is long since defunct, but they existed at that point, and had available an analyzer which made it possible to do the lifetime measurements in a more elegant fashion than just sitting and getting individual time delays and counting them.
You also need to tell that besides seeing Ted Jebal occasionally and Sid Millman, who seldom came into the lab but occasionally had a seminar in his office —
He occasionally had a review session, in which various people in his group would say what they were doing.
But you were working mostly at night because?
Taking the data on these various research problems, we mostly did in the evening. For some reason, first of all, the apparatus didn't start working right until fairly late in the day, and also the freedom from interruptions was very helpful, that occurred in the nocturnal hours, so that there wasn't much socializing when it was going on. I tended to be spending most of my time working on these problems. We occasionally had lunch with people, such as Conyers Herring and other people in the physical research group, where ideas were discussed and tried out, but the rest of the time, we were mainly improvising apparatus, taking data, and trying to analyze it. Ali and I spent a great deal of time talking to each other during this period, although not very much time talking to others. Let's see, you're asking if I — I don't know whether you were asking about interactions with people outside of the Bell Labs or not.
Well, I'd really like to get a picture of your intellectual contacts around this work, if any.
Yes. We were pretty much —
For example, Schawlow was —
He was across the hall during a lot of this period, and occasionally we'd bump into him, but he was mainly sitting in his office typing articles, and working on problems that were rather removed from what we were doing. He had a huge high resolution spectrometer that he was using for solid state spectroscopy, and working at things like ruby.
Then there was a maser group under Scoville.
Yes. I had very little interaction with Scoville. I didn't even meet Scoville until the helium neon laser was practically working, I think. He came around and was very interested in it. Of course, after the laser worked, all kinds of people came around. But (crosstalk) Harry Nyquist (?) was one of the people —
Harry Nyquist. He was still working as a consultant at the Bell Labs in that period, even though he had retired formally, I think, but as soon as the laser oscillated, he wanted to come around and seeing how it obeyed his criteria. (Specialized joke, I guess.) But in any case, we spent most of our time arguing and talking with each other about things, about the physics involved and the equipment. Occasionally we would bump into Jim Gordon. I had very little interaction with Don Herriott until about the summer of 1960. Don wasn't involved at all in these preliminary measurements of the physics in the helium neon system. He was independently setting up his apparatus for analyzing the flatness of mirrors and worrying about getting high reflectance coatings. There was a monthly seminar, I think, that met, I don't remember whether it was precisely monthly but it was roughly monthly, that met where various people talked about the different problems related to lasers, and in particular Gardner Fox and Tingye Li presented a paper at one of those seminars on their work, and that was certainly very interesting. I was delighted to hear about what results they were getting. I had been quite worried about the diffraction loss problem, and looking at the whole interferometer question, from the point of view of plain wave diffraction loss, it looked like the sort of thing that Townes had calculated. It looked very much like the loss would be very substantial, and something like a meter long and even 15 millimeters in diameter, and rather comparable with the kind of gain we were getting, that is just from the diffraction loss alone, and it was therefore very nice to hear that, first of all, from Gardner Fox and Tingye Li, that they proved that these modes did exist, and secondly, that they had very low loss compared to normal plain wave diffraction loss. That was certainly a very exciting and important aspect of the laser device part of it.
Did they change what you were doing at all?
Not really. I think if they hadn't found the existence of these modes, we would have gone ahead and tried it anyway, but it made things much nicer to know that there was some theoretical basis, even if it were merely a numerical analysis of the problem, which showed these things would be there and high Q.
Did it help you politically? You said there was some pressure, towards the summer of '60. I should think that this kind of result might have given you a political advantage?
Well, I don't think it did particularly. I mean, the Fox and Li results were very interesting and exciting results, on their own, and they were applicable to any laser system. But it is true that they did do a few calculations which were tailored a little bit towards the dimensions we were anticipating. However, the actual Fresnel numbers that were involved in the first gas laser they put together were off the plot that they had calculated. We could only sort of extrapolate their curves. It indicated that there was nothing to worry about in terms of interferometer diffraction loss, in the modes that they had studied. Later on, the calculations they did on the frequency separations between the dominant even and odd symmetric modes were very helpful, because they did indeed fit very closely to what we were observing in the beat spectra, from the output of the oscillating wave. But in the summer of 1960, other than just general background, I don't know that it had that great a political effect on the funding and things of that sort.
The other question I've got down there, just whether you considered at any point using confocal mirrors, because Boyd and Gordon were somewhat into their analysis, I think, by —
Let's see, when did Boyd arrive at Bell Laboratories? I don't think he was —
I think he came at the end of '59. Well, I wouldn't swear to it, but I think he came about when you did.
I don't remember exactly when he did come. But we didn't think of using confocal mirrors, and we didn't think of using Brewster angle windows. I was interested in the deposition by E.I. Gordon, given many years later, that he thought we were stupid not to have thought of that. The question is, of course, hindsight is always 20-20, as Charlie Townes would say. It's certainly true that life would have been easier if we'd had access to these things and known about them. Largely, we were dependent on Don Herriott's insight on the practical optical problems involved. Ali and I agreed towards the end of spring of 1960, the beginning of summer, that we didn't want to go through the headache of having to design all the Fabrey Perot? machinery that was necessary, and that we were working full time even then on understanding the gain situation, and lifetimes of cross-sections and all that sort of stuff, optical pressures. So we relied very heavily on Don Herriot to handle the interferometer question. I had never lined up a Fabrey-Perot before myself, before the helium neon laser oscillator, and indeed I found it was much easier to line the Fabrey-Perot up after you had an oscillator, than you could ever do it passively. Of course once you get the mirrors parallel, then the thing bursts into oscillation. But Don Herriott was the sort of person who could do a pretty good job lining up all these, he could do a job comparable to that which you could do with an autocollimating telescope merely by holding up a flashlight bulb and looking at momentary reflections. He was a very skilled craftsman in that type of thing, and he I guess had suggested putting parallel mirrors as being easier to deal with, and there were questions that would have been less of a severe limitation if we'd known, if we'd had more experience with the confocal situation, and certainly the Brewster angle windows would have reduced the problems of destruction of mirror coatings that did occur in the first experiments, when we baked the system out and so on.
Another thing I wanted to ask —
There also were some unknown questions though with the Brewster angle windows. That was another kind of loss that was being inserted in the cavity, and as it turned out in later experiments, I found that you could actually get in trouble by using Brewster angle windows, because of the strain that was induced by the mechanical effects, when you pumped the system out and baked it out and so on — it would occasionally show up by scattering losses increasing in the window. Indeed, there were some laser transitions that I think we didn't succeed in getting to oscillate, because we were relying on inadequate Brewster windows, in some of the work later on. But go ahead. I interrupted you.
The impression I get is that the feeling was that you might very well fail. I mean, that you and Javan as you worked were not confident that you were going to be able to get the thing to work. Is that a correct assumption?
That is certainly correct. It looked to me as if the chances were about 50-50 at that point. We had made various measurements at different times of varying reliability of what the gain might be in a real system, and we were getting numbers, the best measurements indicated numbers on the order of maybe 1 1/2 percent, with plus or minus 1 percent uncertainty. Anything that produced additional loss, particularly unknown systems where we weren't sure what the consequences would be, were things we wanted to avoid. But I remember, come to think of it, that Don Herriott had mentioned that there had been some work done recently, that is circa the summer of 1960, in France, by some people using confocal mirrors, and he was wondering if we shouldn't start worrying about that, but by then I think he had already ordered the flasks (?) that were involved, and he was imposing on one of his friends, I think it was A.F. Turner, of the Bausch and Lomb Laboratories, to make us some high reflectance multiple layer dielectric coatings on the glass, and it seemed like the safest thing at that point was to go ahead and try it, and all the data we had indicated that it would be just on the verge of working. It wasn't going to be simple, but there was a good chance that it would work. Rather than going on and doing lots of other research at that point, in a totally new direction, we thought we'd try it. And of course, once we had a laser going, it was relatively simple to make modifications and try out a new configuration, and indeed line it up.
There is one other thing I wanted to ask you about. These 90 hour weeks, was that the way you usually worked, or is that — why were you working so enormously hard? Was it the sense that lots of other teams were making lasers, or was it just the sheer fun of the chase or?
I didn't know anyone else who was working on gas lasers. I guess, later on —
Someone at the Columbia group.
Well, OK, I knew of — you're right — I was going to say, discharge lasers. The Columbia group was using optical pumping lasers.
Yes, that was quite a difference.
I wasn't worried about them. I knew that later on, that something was occurring at TRG. I guess we knew early in the game that somebody was trying to do something at TRG, although it was never clear what. I guess it was — gosh, why did I work so hard? I remember my brother-in-law asking me that at the time, saying, "You're crazy, you're killing yourself." It was very exciting. I thought there was a real possibility that the thing might work, one way or another. I was very interested in the problems that were involved, that is, the physical problems, and I'd never before had access to so much apparatus to make so many measurements in such a short time. I had lived on a budget of something like $500 for equipment, when I was a junior member of the physics department —
— $500 a year?
All together. I received the, let's see, what was it? The Higgins fund for new junior faculty members in physics, in 1957, which I had to share with Charlie Drake, who was another junior faculty member at that point, which consisted of $1000 which we divided, and we bought an oscilloscope with it, and that was sort of the extent of apparatus that I had been able to purchase pre-made. As a matter of fact, for most of my period at Columbia and even at Yale, I had built most of the apparatus that I needed. It's a totally different kind of environment for doing physics. I did all my own glass- blowing and machining a lot of the time, and wiring the circuits, and if I needed a lock-in detector, back at Columbia, that was the common situation — the first thing you did was build a lock-in detector if you wanted to do an experiment. And indeed, I built several of them, and I also built counters and indeed at one point I started to build a multichannel counter, and in fact did it for some of the work I did as a graduate student. But apparatus just wasn't available, at the scale that I saw it splashed around Bell Labs. Suddenly, I'd been going over to visit the Bell Laboratories, I became aware that, you know, every room had at least one or two oscilloscopes and probably a Varian magnet and a lot of other neat things that were readily there. I wasn't really convinced that I was going to go to the Bell Laboratories, when I went out there for the first interview. I went along more out of curiosity, to see what things were like, and suddenly I began to see just what was available, and how much importance was given to this project, and I was certainly delighted to have the opportunity to use all this equipment and do interesting things. And I think that was one of the reasons I worked hard. I guess there was some feeling that there might be some kind of a race going on out there, but I wasn't particularly concerned about what people were doing in the gas discharge area. I thought we had a substantial advantage over anyone else. Maybe that was ignorance, but on the other hand, I think it was based on some fairly solid insights. I had known some of the people who were working with Gordon Gould on their projects. I didn't think they were very well informed on the problems that were involved, to be honest.
That's a very interesting explanation, which interests me the more in that it runs counter to what I had thought. Those are always the most interesting explanations.
What did you think the explanation was?
Well, no, I mean the picture you painted, the difference in equipment, the feeling of being able to get a lot of work done, is just a new thing for me.
Yes, it was phenomenal. I'd never seen anything like it. There wasn't that kind of apparatus at Columbia, and certainly not at Yale. Trying to do any of this type of measurement at Yale was, at that period, was like setting up on a desert island. There was very little available, and essentially one had to do it all oneself.
Were you on leave from Yale?
No, I actually turned down an assistant professorship at Yale for the following academic year to go to work at Bell Labs, ultimately. As I say, when I first went down to Bell Labs, I wasn't so sure I was going to accept their offer, but then I saw all this apparatus and started thinking about what I could do with it, and that convinced me it was a pretty good idea, and I then decided I would leave Yale. In fact, I was planning at that point to stay permanently at the Bell Labs. I was told there was every probability that that would work out well. And as it was, I stayed there three years. I found that there were things lacking there that weren't necessarily made up by all this apparatus, but that was another matter.
And actually we'll come to that when we come to your return to Yale.
All right. Now, let's see, what have we got? Through the spring of '59. During that period, when I was thinking about the possibility of going to Bell Labs, I spent a lot of time, occasional weekends, talking to Ali, and we were presumably at that point writing a joint letter or a joint paper on the possibilities of obtaining population inversions in gas systems, and that persisted for quite a while. He came up to Yale a number of times, and I visited him in New York a number of times, and had correspondence and so on. And during that time, I sort of outlined a lot of different things that might be done, ranging from collisions in the cyclotron, electron impact processes, to charge exchange reactions and some other things that weren't really done until very much later. Then, towards the end of that term, I was tied up with, I was teaching 13 hours a week at Yale at that point, and towards the end of the spring term, I got a call from Ali saying that he'd received considerable pressure from Bell Labs to publish something immediately, so therefore he'd written something under his own name and sent it off to PHYSICAL REVIEW LETTERS. The explanation he gave me at that time, and I found this rather annoying, to be honest, was that there was a general fear at the Bell Labs that the whole field was going to be classified, that Gordon Gould had gotten his million dollar contract from ARPA at TRG and all his work was being classified, and that people at Bell were concerned that everything in the field, in the laser field, the optical maser field as it was called then, was going to be classified, and they therefore wanted to get some publications out, and hence, according to Ali, the explanation was that both he and John Saunders were encouraged to send off papers quickly to the PHYSICAL REVIEW LETTERS, which they did, and his paper was published at that point. By then, I had already told Yale I didn't want their assistant professorship any more, and was in a situation where I essentially went ahead and went to the Bell Labs, that summer. I've since been told by Charlie Townes, who talked to Ted Jabal about this problem, that there wasn't any worry at Bell Labs about classification. I don't know what the real story there is, but anyway that's what Ali told me.
I think that.....
Let's see, where were we?
Well, now you had bought your house. I guess you bought a house.
Yes, I had to spend some time. Well, first of all, I had a house in Hamden which I had to sell, and we then had to find a place to live near Murray Hill, so that took us most of the summer, first of all selling the one house, and trying to find another one at reasonable distance from the lab, and we did move around August or so, and took a little while to get settled, and I didn't actually start working in the lab at Bell until early in September. I guess during that period there had been the Shawanga Conference in Quantum Electronics, which I didn't go to, where Ali talked about some of these ideas. But starting in early September, we were working at a fairly feverish clip, and the process from then on — Ali had a pet scheme that he was pushing, for the measurement of gain. He described it I guess at the Shawanga Conference. It was one of these ideas which was a very clever approach to the problem, but it had basic difficulties which made it a disaster to try to apply in real life. This was a scheme where one would modulate both a source of light and a gas discharge in order to measure gain, modulate them at two different frequencies, and you could see that, from the equations involved in terms of what happens with the amplified or absorbed light, you'd find that because of the nonlinearity represented by the amplifying and absorbing medium in this system, you would generate difference in some frequencies coming out, in the modulated light wave. You know, this is very low frequency stuff, audio type modulation of the amplitude. And one would then be able to set up a detection mechanism that, in principle, that would look at either the sum or the difference of frequencies in an appropriate way to detect the presence of gain or absorption, and this whole thing involved some complicated us of phase sensitive lock-in detectors, which would, depending on the phase of the signal, you got out, tell you either that you had gain or absorption. As I say, it was a clever idea, and Ali was very fond of it, and he got a lot of apparatus to try to pursue it. The basic problem with this system — it was twofold. One of them is that any nonlinearity at all in the system would also generate some indifference frequencies. The photomultiplier in itself was nonlinear, and would generate such things. The lock-in detectors and amplifiers had some nonlinearity. So it was very easy to see a non-linearity in the system, even if there was no gas in the system. We spent an enormous amount of time trying to eliminate this source of systematic area. There was an additional problem with it, in that there were phase shifts induced in the modulated discharge, just due to things like the collision transfer process itself in the helium metastable. That is, these exponential tails that corresponded to the period after the discharge was turned off, when the excitation was being transferred in the afterglow of the discharge, resulted in a phase shift, and that phase shift made it very ambiguous, in trying to interpret whether or not there was gain or absorption in the system. Another problem with this approach was that you could never calibrate it, and you didn't know how much gain or absorption was involved, because it depended intimately on the nature of these nonlinearities which weren't well understood. So we spent an awful lot of time trying to get rid of that type of difficulty. We did see some results, during that period, which indicated gain, but in retrospect they couldn't have been gain, because the geometry was just totally bizarre, in the discharge cell that was involved, and it just didn't make sense. So I began worrying more and more about that, and in fact, there was a period in early 1960 when Ali took a vacation, went to the Virgin Islands, and during that time I designed another way of doing it, which ended up working pretty well. It was a balanced optical bridge arrangement that involved a source of light, a gas discharge source, which was modulated, at one frequency. That light was then run through a half-silvered mirror, and then one beam was sent into a spectrometer, in one photo tube. The rest of the beam went through a test cell with a discharge in it that could be turned on or off, at other frequencies, then into a second spectrometer and photo tube. Then essentially what we got out of those two photo tubes were AC signals due to the modulation of the initial source. That was fed into a passive transformer arrangement that cancelled out the two signals and permitted making a null balance bridge out of the signal that came out, and that was then fed into a lock-in detector which was driven in phase with the modulation on the lamp. This turned out to be very sensitive and reliable and highly passive. There was a minimum amount of amplification in the system that could introduce troubles. And it had the advantage that you could tell very directly what the absolute calibration was. All you had to do was stick your hand in the beam, for example, and you could see whether it was absorption or gain in the direction that the signal was going, and one could measure the percentage of it very directly that way. And when Ali got back from the Virgin Islands, we then went ahead and made some more careful measurements. I'd made some rough measurements that indicated that things were working reasonably well at that point. Then we made some more serious ones, which indicated that we could get a gain on the order of, oh, 1 1/2 or 2 percent, I forget the exact numbers. But there was a lot of noise in the signal because it was a spontaneous emission source, and the error, estimated uncertainty, was around 1 percent, so at that point... (off tape) As I was saying, the measurements had to be made in the presence of a large background of noise, and the virtue of this particular arrangement that I'd constructed was that one could definitely see whether it was gain or absorption, and what the actual magnitude of it was, and one could then modulate the gas discharge test cell at some other frequency, and see how the gain varied when you turned on and off the discharge. Anyway, it was out of that apparatus that we made the measurements that finally gave us some insight on what the situation was going to be in a real system. The work that we'd done before may or may not have had gain in it, but it was so filled with ambiguities from all these various effects that one couldn't say very much about it, and indeed, I remember one piece of data we took that for a while, we were convinced showed gain, but then we realized in retrospect that just the geometry of the cell was far too large to have permitted, under those conditions, to have been anything other than a spurious source of gain. The lifetime measurements had started around January, I think it was, in 1960, and we had been taking that data on and off while these other things were going on. One of the things you mentioned before was, why did we often work all night to take data then? Those measurements in particular with photon counting rates were highly sensitive to background noise, pulses in the wiring, — if anyone else turned on a radio frequency transmitter in the building, it probably affected the results, and we also needed total darkness for some of those measurements, and so on. Anyway, we were doing a lot of these things more or less simultaneously. We were still actually completing our best measurements on the excitation cross-sections, also, during that period, and we had about three experiments going at once, and sort of alternated between according to which had just broken or what had been made in the shop and so on. Anyway, towards the end of the spring of 1960, we thought we'd go ahead with some kind of a Fabrey-Perot design, and Don Herriott volunteered to undertake that. He had a mechanical engineer working in his group who drew up a set of plans, and sort of over-did things in some directions. We were worried very seriously about the need to make small continuous advancements in the plate angles of the Fabrey-Perot, and of course we anticipated that it was going to be hard to line these things up, and you needed to be able to sweep a moderate range of angle, and someone in Don Herriott's group, I can't remember the guy's name, actually put together a structure that had a bellows connected to the discharge tube and would permit wiggling the whole infrastructure with micrometers in orthagonal directions. The one difficulty with that design was that it was very easy to wiggle, and in fact it would go on by itself when we didn't want it to, and it wasn't very stable. You could get it lined up quickly, but then it wouldn't remain there long, and it wasn't the best design in terms of actual studies of spectroscopy and behavior of the system. But it did work in terms of getting things lined up. Anyway, the first of these assemblies was concluded some time in, I guess it must have been, I don't know the exact date but it was towards the end of the summer, maybe August.
Now, what kind of impact if any did the Maiman thing have on what was going on here?
It — well, I found it very depressing, as a matter of fact. I'd hoped our laser actually would be the first to oscillate. And I was very happy for Maiman, to see in the NEW YORK TIMES that he'd gotten these very interesting results. I think, incidentally, that the problem that Maiman worked on was perhaps even harder than the one we were doing. I was one of a number of people such as Art Schawlow who didn't believe the system was going to work because it involved inversion with respect to the ground state, and I admired his courage in being able to go ahead and show that that was possible. By now, of course, I agree that it was a nice way of doing it, but at that time, I thought it was going to be very, very difficult, to invert a system with respect to the ground state.
Did it change anything you were doing or make you go off to the beach, or (?)
No. No. No. I don't think I went to the beach at all during that time period. But I remember reading about it in the NEW YORK TIMES one morning, and on the way to work, I looked at it just before going to work, and I recall being depressed by it, because I had hoped we would get ours to work first. But it didn't really alter our course of action at that point directly in any way. There were some things going on behind the scenes at Bell Labs that I tended to be shielded from by various sub-department and department heads. Ali may have heard more about it than I did at that point.
Yes, you mentioned a little bit about that last night, and some attempts that were associated with Deming Lewis to —
Yes, apparently there was some concern expressed by various other department heads than Millman about the way in which money was being spent on the particular problem we were working on. And I guess there was some infighting going on about budget questions. It is my understanding from sort of second hand sources, that Deming Lewis was the fellow who mainly was leading the fight against us and trying to have the money, the flow of money stopped for the helium neon laser at that point, although I never heard it directly from him. I wasn't a party to these discussions about how the resources were being shared. Al Clogston came around and talked, primarily to Ali about things at that time, and I guess Ali somehow convinced him that we should have some more time to work on it. And indeed, later when the laser did work, Millman crowed quite loudly to the other department heads about the wise thing he had done in continuing support, and "See, this is the way it should be done," and so on.
He deserved it really, I think.
Yes. Although if he'd really known what was going on in some cases, I think he might have been a little worried. There's no question that there had been a fairly extravagant spending of money on apparatus that wasn't really important to the problem. And it was a little (?) I must admit. But anyway, the system did work. In the summer of 1960 at some point, Ali became enamored with the notion of using a microwave source to excite the discharge tube in the laser. He had installed a large SM kilowatt magnetron. I don't know whether it was a kilowatt or several kilowatts. Anyway, it was a pretty potent magnetron, in the basement of the building, underneath the laboratory, with holes cut in the floor, through which the wave guide came up from this magnetron source, and he therefore decided that he wanted to try exciting the laser with the magnetron, and he designed a microwave cavity that could be used for this purpose. In fact several different cavities were tried. I think actually the thing that worked most readily for microwave excitation was the slot in the end of the light guard through which he stuck the discharge tube, and he had a special water cooled quartz tube made up that would fit in this wave guide that was then put in the Fabrey-Perot interferometer, and the whole thing was set up and baked out and then filled with gas, at the sort of pressures that we'd determined before, and turned on, with the result that the whole thing melted before —
The quartz melted?
The quartz melted, yes. Anyway, I forget now exactly where it failed, but by necessity we couldn't have the water cooling jacket extend inside the wave guide, so there was a stretch of quartz that was uncooled, and the thing just couldn't withstand that sort of treatment. And that was the end of the first helium neon laser. I wasn't too happy about trying to use microwave excitation at that point, since we hadn't done any measurements whatsoever regarding what would be going on in the discharge under those conditions. What we'd used before were all 10, 15 megacycle, relatively low powered discharges, and we had looked at most of the behavior. Then in retrospect, looking at the brightness of the glow that occurred, it was quite glorious, it seemed quite clear that it was far above optimum excitation for a helium neon system. That is, as we studied in more detail later on. You start turning up the power and what happens is that the inversion basically goes through a (?) maximum and then goes down and hits zero and the whole thing becomes absorbing very quickly, and indeed the optimum power levels for the sort of geometry that we had in those early experiments was perhaps 60 to 100 watts, not kilowatts.
Did you have any sense of what interested him in trying this?
I think his idea was that it was nice to have a variety of different avenues of approach, and maybe it would work, and it would be interesting, maybe it wouldn't, but it was sort of a part of a parallel attack on the problem. But the basic trouble was that we didn't have any real data, and it wasn't very controllable, and it may have been stimulated by the race with Maiman, I don't know. I don't remember exactly. I'm not even sure what the relative time was when we tried that system, versus Maiman's disclosure. It was roughly in the same period but I don't remember exactly. Anyway, after that the system was rebuilt, using a standard 15 millimeter quartz discharge tube to be operated with a (?) discharge, and it's true that Ali made a statement somewhere along in his deposition that we built the system to have a longer discharge. Well, it's true, it was longer. It would have been longer in the first place if we didn't have that water cooled quartz thing put together. But it ended up being more related to the kind of measurements that we had made up to that point.
So that is a distortion in that —
Yes, it is a distortion. But it may have been a political distortion, at that time, to ward off the people who were threatening to cut the flow of money. Anyway, the second attempt involved the bake-out where the internal mirror coatings disintegrated, when we pumped the whole thing out. That is, because it was desired to be able to seal this thing off, we just had to use ultra-high vacuum technique and bake everything out, and seal the tube off, at least cut it off from the vacuum system supply, and the second attempt, I think it was, we went through all this agony, and carefully selected a quartz flask, and carefully made dielectric coatings on the — it was put through this torture of being baked at 450 degrees centigrade for a day or so, and we typically got vacuums of 10 to the minus 9 or so in these experiments, but the coatings didn't hold up a second time. They all flaked off and fell on the bottom of the tube inside. And it was then necessary to get another set of mirrors and another pair of coatings, which, Don was working as fast as he could on these things, but he was carefully selecting optical flasks, trying to get ones that were really very flat, typically 2/100ths of a wave across to centimeters, the sort of thing he was picking out for us, and he would send them off to Turner at Bausch and Lomb to get them coated, and there's a certain turnaround there, and I think Turner was also experimenting a little bit on different kinds of coatings, so it might get held up there. Also, Don was worrying a little bit about the possibility of designing a coating that would withstand a (?) load (or, something terminated at a node in the electric field) in the electric field, with the idea of being able to use systems such as mercury that might coat out or condense out a little bit on the coating, and if you arranged to have a node in the standing wave at the reflection from the multilayered dielectric mirror, a little degradation in the reflective intensity would occur with that design. There was another parallel approach that Ali wanted to consider at that point, which involved a mercury helium system that we'd taken practically no data out of at all, a little bit but not much. It was sort of a last ditch chance to find something that might work, if the other systems didn't. Anyway —
In all if this, it sounds as if you and Javan are the people who are doing the deciding, and Herriott is not, is that right?
Yes. Herriott was not involved in any of the problems regarding the gas discharge part of it all.
He was kind of an optical consultant?
Yes, right. He was very good at that. He was an unusual character, in that he knew a great deal intuitively about optics, and probably hadn't worked everything out and might not even have been able to in some cases, but he seemed to understand optical systems and propagated lens systems very well, and he was handling that part of it, but he really didn't get involved at all in the physics part of the medium that was producing the amplification. Anyway, he did worry about trying to get some coatings designed that would help the node at the exposed surface and might be a little bit more stable on the attack by corrosive things and metal vapor things particularly. So that kind of work was going on too in the background, and in the meantime, we got another set of mirrors together, and assembled the maser. Towards the end of November and the beginning of December, we baked it out, and I spent many days the week before it oscillated trying to get it lined up, to see if it would oscillate, and without any success. We were probably doing it, it was rather low, but we —
That's another mistake that I've gotten my —
— we were using an old collimating telescope that had a resolution of ten seconds of arc in each direction, so realizing after the fact that it took alignment within about one second of arc to really make this thing work, the chances of it randomly setting the thing, just after aligning the telescope, were about one in a hundred. And as we demonstrated, you could do it over and over and over without getting it to work. I've since, after the oscillator did work, it became obvious how you did this. It was pointed out fairly quickly, and that is, again, hindsight is 20/20, but the real technique of doing it is, you lined it up as well as you could with the two micrometers using the telescope, knowing it was going to be in the general ballpark but the odds were about one in a hundred it would work. Then you take one of the micrometers and slowly turn it, and bang on the thing with a blunt instrument like that, and this would set up all kinds of vibrations in this particular interferometer that had been designed by the mechanical engineers in Don Herriott's group, would cause it to vibrate in all different ways, and every once in a while it would flip through the range of parallelism and you'd see a burst of light being detected by the photo tubes, so you'd slowly scan the micrometer and bang on the thing with your hand judiciously, you don't want to bang too hard or you might lose it, and optimize that micrometer. Then you take the other micrometer and again banging on it now with your left hand, while turning the right hand micrometer, and very quickly you then get the thing lined up adequately so it will be in CW oscillation. But of course, it also paid to use a squarely modulated discharge while making these adjustments, because the gain in the afterglow was quite a bit larger than it was during the discharge, because of the various processes that interfered with the population inversion, electron impact in metastable states, resonance drop in the light, going from the little laser to the metastable elements, let out during the afterglow and the gains tended to be higher than the pulse action.
I guess I'm a little confused about the physics now. Since you were really demonstrating a pulsed laser by all these things, what made you think you had a continuous laser?
Well, we would (?) modulate the discharge, and then we could see, you know, the pulsed part of it would be the afterglow, of the square wave, and the continuous part of it would be while the square wave was on. Once you get it all lined up, then you could turn off the square wave modulation and leave the arc discharge on continuously and then you get continuous output. This was just a question of ease of alignment. As a matter of fact, in the spring of 1960, when we were worrying about various things that would upset the inversion, we were a little worried at that point that it might not be possible to get CW. It was so marginal that it looked like it might only work as a pulsed system. And perhaps, it might be made to go CW under certain conditions by using special diffusion of the metastables from one region to another. Having created the discharge it would diffuse out through a screen grid that would keep the electrons and ions in place but let the metastables out, and indeed, the first patent application was written up, which was actually written around April, or the first joint one anyway that Ali and I contributed to, was late spring of 1960, and that was the patent that was submitted, application that was submitted to the Patent Office in December of 1960, when the, after the laser oscillated. In that patent, we were concerned with the possibility that the system might not work CW, and we would have to rely on this enhanced gain in the afterglow to make the system oscillate in practice. Fortunately it turned out that, although we had measured CW gain in the initial experiments, it wasn't very large, and it looked like it was sort of marginal to make it oscillate continuously in the actual laser, but in the alignment process it was very helpful to look at the afterglow. And indeed, when the thing did first go into oscillation, at a time when I was out of the room, Don said that the way he noticed it was from this huge spike that occurred in the, or small spike that occurred in the afterglow. For many practical reasons, we were intending to look at lines in the nearer part of the infra-red, for most of our measurements, just because the photo tube sensitivity fell off exponentially from about l to 1.1 microns, and a lot of the actual work was actually done with lines at .967 microns and 1.18 was the one which I guess was actually the one that Don was looking at when the thing oscillated. And then, after getting that one line to go, it became a trivial matter then to go through the spectral range of the micrometer to see that the l.l52276 micron line was going very well. It was hard even to see that line, however, in spontaneous emission, just because the response of the photo tube was falling off so drastically. Indeed, we were having some — that was later on. I was trying to establish for sure which line was actually oscillating. There was another line about 2.26 angstroms away from the l.l52276 micron line, that we could not resolve with the micrometer, that we were looking at. As a matter of fact, that line under normal conditions at which the helium neon was absorbing, which made it double complicated because of the fact that you couldn't resolve it and it was actually absorbing, in the helium neon mixture case, made it very hard to see what was going on in the system. I remember noticing the difficulty. I tried to make some photographs of the spontaneous emission coming out, with a conventional photographic plate, probably Woods (?) who had a high resolution spectrometer, and that was actually with the second laser we made a little later, to try to settle that problem once and for all. I found that the photographic plates weren't very much good for trying to look at lines in that range, even though these are supposedly sensitized plates, and you can expose the thing for hours and not see anything of the lines, whereas once it's oscillating, it was relatively straightforward to make a photograph of the spectrum, and indeed by doing that, I did establish later on in the spring of 1961 that the line actually was 1.52276 microns. We had indirect reasons for assuming it was the 2 S 2 to 2 P 4 transition, because by looking at the emission from the lower state, we could see that change as the thing went into oscillation. But I wanted to try to make it absolutely certain, that we were certain, we were sure which line was oscillating the system, which as I say wasn't actually nailed down until later.
When Herriott got that oscillation, was that a surprise? Or by that time was it pretty well expected?
Oh, I don't think it was expected. I was getting pretty discouraged myself. As I say, I spent many days before that trying to get it to work. We also spent a lot of time that Monday trying to do it. I had agreed to go out to Berkeley to an APS meeting, an exciting meeting, that December, and give a talk on our lifetime measurements, which had clearly implied now that an inversion was possible, and had been told I had to pick up the tickets that Tuesday if I wanted to make sure I could make that plane, the plane that was going out, and I was out Tuesday afternoon trying to get those tickets, and as a matter of fact a blizzard started that day. It was a horrendous snowstorm came up, and although I'd spent the previous day and many days the previous week trying to get the thing to work, Tuesday afternoon I was out of the lab until after it did get to oscillate the first time. But Tuesday morning, we tried all morning to get it to work and it didn't work then either. And then I must admit, I was getting a little worried and anxious. You know, it was obviously marginal. As it turned out, one of the lines that we were monitoring during this period was one that no one else ever got to oscillate, I think. I'd have to refresh my memory as to which one it was. But that first laser oscillated in five lines. The strongest was the infra-red 5 transition line, but there was another one which was...the one that Herriott was using to line up with was the 1.182 line in vacuum, and that was not too hard to get, but there was another one, what was it? This is a — so weak that it barely oscillated in the first laser, and within a matter of days or a week, the gain dropped sufficiently so that it didn't come back, and it was never seen again until R.N. Zitter and some of his colleagues made a helium neon laser that was something like 30 feet long, and they found with that —
I often wondered about that too, but that particular system, which was 30 feet long, had enough gain to it that one line oscillated...(off tape)
Actually, this also has to do with the question on page 2 about Lamb, because I think that you know, that analysis and your interaction with Lamb —
Yes, well, let's see, one of the things that we had done with the first helium neon laser, and the research immediately after it oscillated, we started working furiously to take all the measurements we could — one of the things we wanted to know is what the inherent line width of the laser was. And it occurred to me at that point that we might be able to tell that by looking at the base band beat spectrum that was involved, and what we did was set up an RF spectrum analyzer to look at the output from the photo tube. My initial thought on it was that we could just beat — say we just had one mode and one line going, that we could beat that mode and its components with itself, and from the width of that beating experiment, we could say something about the inherent width of the laser line. Don Herriott had an oh, 10 or 20 megahertz spectrum analyzer that he brought over for us to use for that purpose, and we then took some spectra in that region, which we then looked at, and saw what appeared to be not just one peak, that I had thought that we would see, but also a lot of satellites. And these satellites tended to agree roughly with the kinds of predictions that Fox and Li had made about dominant odd, even symmetry modes being present in the system. That is, there would be the normal modes that we expected, spaced at Cover 2 L, where C is the velocity of light and L the one meter length of the Fabrey-Perot, hence we expected things on the order of 150 megahertz spacing for the dominant modes, but there were little satellite modes or beats that appeared in these spectra, even when we looked at the base band result, which obviously had an interpretation based on interference between the odd and even symmetry modes of the same latitudinal mode. And Fox and Li got quite interested in that, and they began extending their calculations to include our geometry more accurately, and indeed, these results did agree reasonably well with what their prediction was. We later got ahold of a wider band spectral analyzer that would actually permit looking at the 40 and 50 megahertz beats, and 300 megahertz ones which would correspond to alternate modes, adjacent modes, modes beating with each other, and that structure was intriguing, and it seemed like it would be very interesting to make further study of it. But you have a question very similar there, asking if we made a model to look at beats or something.
There had been a paper published by Forrester on mercury hyperfine transits, that I remember having read at that point. I should have referred to it. I thought we did. Anyway, Forrester had seen some evidence of beating effects from spontaneous emission sources that were very narrow, the hyperfine structure of mercury, I think it was, but other than that, I don't think anything else had been observed in terms of optical beats.
Because that becomes a very important technique.
And I haven't got any real feeling for what its history is, how to interpret its historical development.
Well, there had been, Forrester — there was someone from Oxford who was worrying about light beats too. I don't think it was the same kind of thing we were dealing with. Anyway, it had struck me that once you had a monochromatic source, that ordinary mixing effects that one was used to with electronic oscillators and so on ought to work, and then we could argue that the photo tube was a detector, where the current responded in a way proportional to the square of the electric field in intensity, and striking it, and if you now imagine that intensity made up of two, well, field components of different frequencies, and added together and squared, that you would get difference frequencies at least out that would be within the bandwidth of the photo tube and you could then detect them. Indeed, if these frequencies had slightly different polarization, you would then expect to see the cross-term, the dot product of the two fields, which would be linked to the difference frequency, which would exhibit properties that could be discerned, based on polarization experiments by (?) concerning the linear polarizing. It turned out that the modes in some cases were perpendicular in the field alignment, and if you then stuck a linear polarizer at 45 degrees, in the output gain, you would get a projection of two of the field components suddenly producing a beat that would appear in the spectrum and things like that. Well, anyway, as far as I know, we thought of the technique ourselves. I think I was actually the one who suggested it, and Don quickly realized that it was a good approach to the problem. He had more experience looking at base band spectra, as it might be called, than I.
I guess I'm not completely sure what a base band is.
Well, it's just zero frequency components of the beat product, and you take something that's an occupiable frequency, you know, 10 to the 15 cycles a second, but it's spread over a domain, say, 1000 megacycles. Now, you run that through a square law detector, what you expect to see are different, difference frequency components spread over a range of about a thousand megacycles, but then centered at zero frequency.
— so basically it's just (crosstalk ) — I see.
I guess it seemed to be an electronics communication terminology, and anyway, that's the initial thing we did, and we were trying to find different ways of measuring the coherence of the beam. Don had made multiple diffraction measurements and so on, and I suggested this approach, which was then followed up and produced these spectra, with all these peaks that we didn't largely understand, but they seemed roughly in the ballpark of what Fox and Li were talking about.
I just want to ask you one question on that. Now, after that, Javan uses that technique a lot. Did it spread? Did that become just a very standard technique?
Yes, I guess so. I mean, once we had an oscillator, it was very natural I think to do beating experiments with it. In fact, —
I guess I'm really asking how you want to look at that whole thing historically.
Yes. Well, I think it evolved essentially that — there had been this previous work by Forrester, which was entirely based on spontaneous emission. It was sort of a virtuoso heroic feat to be able to see anything at all under those conditions, but they did manage to see something very broad and loaded with noise. But this essentially appeared on the scene, the minute we had an oscillating CW gas laser —
So I guess Forrester had something to say about — I remember that after the laser comes out, he begins to talk about how what he had done before could be done in terms of the laser, but I don't remember whether he talked about beat frequencies.
I don't remember that either. As a matter of fact, I don't think I even knew about Forrester's previous work when we started doing this beat experiment, although I quickly heard of it. I thought we had a footnote on it. Just glancing at our first publication, I don't see it. Actually, I wrote just about everything in this paper before I left for Berkeley to give a talk on the lifetime experiments. The first version of it ended up being a little long. In fact, probably a lot long. And after I had gone to Berkeley, Javan undertook the problem of editing it and adding a sentence or two there, but essentially this is all my wording, and I thought I had had something about Forrester, but I don't see it here. It might have been in the first draft and got cut out to make it short enough to publish. As it is, this a phenomenally long PHYSICAL REVIEW LETTER. But on the other hand, there's an awful lot of research in it. Anyway, I think it essentially started at that point. When Bell Labs had a press conference in a hotel room in New York, in February lst I think it was, it was essentially the day that this article was supposed to come out, and Don Herriot and I took the laser over to this hotel room, and actually managed to set the thing up there and get it working, and we actually had a spectrum analyzer along that displayed the beats on that occasion, and various newspaper reporters looked at them and didn't understand them. But I remember a long conversation I had with Walter Sullivan at that time. We prepared fairly carefully worded descriptions of what was going on, that we handed out, which of course they didn't want to use. Sullivan seemed to be mainly concerned with how the light got out of the laser. He kept saying, "There must be a pinhole." We'd say, "No, no, it's not a pinhole, it's a coating that is partially transmitting and largely reflecting, a continuous operation where we have continuous leakage of light going out through the window." I noticed when he finally wrote the thing up in the NEW YORK TIMES he had a pinhole in there. I guess he probably didn't understand the — at any rate, we actually set that apparatus up in the hotel, and demonstrated it under the most awful conditions imaginable. I was quite interested myself in pursuing that work at that time. At that point, I was encouraged that there was an opening in George Dacey's department which I might fill, and I could have a laboratory of my own, and Art Tegibald (?) suggested that maybe it would be to my best interest to do my own work independently of Ali.
Was Javan already planning to leave for MIT at that point, or not?
I don't know whether he was or not. You'd have to ask him. He left not too long after that, but I don't remember exactly when he left. Anyway, I was told I could have lots of apparatus of my own, and one of the first things I did was design another laser, fairly quickly, that was enormously more stable than the thing Don Herriott had put together. I had decided that what one really ought to have is something that was extremely stable and rigid, rather than this sloppy arrangement. One of the problems we ran into in trying to look at beats and so on is that if you were to talk in the room, it would modulate the windows and the mirror assembly and produce audio modulation on the light beam which was contributing to the noise and distraction, when we were trying to measure the beat spectra. It was sort of nice because it could —
— think about the telephone right away.
Think about telephones and so on, "Come here, Watson, I need you," and all that sort of junk. But the publicity people at Bell Labs made a great deal of noise in print over that fact. You know, "Scientists talk over light beam." Which is sort of dismal, living with the modulation capabilities of it, but in any case, the original design just wasn't very good for this type of thing, although it was real easy to line up, so I designed a four rod structure made of ingmar (?) that, first of all, I could measure very precisely on it what the mirror separation was, and I forget exactly how accurately I knew it but it's in that Holberg (?) paper. And once that was lined up, it took a little effort to line it up but it was possible to do it by deforming the rods and putting in slight shims (?) and so on, once it was lined up it was extremely stable and remained that way for quite a while. Then I also hooked up some coils around the rods, so that I could vary the mirror separation, and hence tune the modes of the laser through the center of the line, and at that point...
The research on line width and on beam modes — sort of leading into the hole burning stuff.
OK. After going my own way from Ali, I decided immediate what I wanted to do was to build another laser, and I think I mentioned, I designed something that was fairly stable, and once it was lined up it stayed there for days, and was relatively free of modulation effects from the outside environment, and also had the ability of being able to be tuned through a magnetostrictive process, by running a current through the coils that were around the four rods holding the plates apart. I actually wrote a paper describing how this thing worked with Pete Kindleman who was working later on with me at that point, as a kind of technical assistant. He had been an undergraduate at Columbia University and had worked at the Bell Labs a couple of summers with me on these projects, and eventually came to Yale as a graduate student with me. But in any case, we did have a paper on the magnetostrictively tuned laser, described this, and then I used this laser to try to study the mode spectra in the laser. As a result of the Quantum Electronics Conference that was held in Berkeley in March, I think it was, of 1961, where we described in fair detail the aspects of the research that I've been mentioning now, I began thinking more and more about this problem of what determined the oscillator frequency, and indeed Charlie Townes gave a very interesting talk at the beginning of that conference, in which he mentioned several different kinds of experiments that people might do with the laser. But one of the things, all of which were interesting, but one of the things he mentioned was a result he pulled out of the air for an expression of the oscillator frequency with respect to the cavity frequency and line widths and Qs that would lie in the cavity, and I, as well as wondering where he got it from, he didn't derive it on the spot, but as well as wondering where he got it from, I wondered what the limitations of it were, and how accurately it really would apply to the question of what the modes really did, and things of this sort. So one of the first things I did when I got back from Berkeley was to start looking into that problem as well as I could, and in particular, the laser I'd built was one where I could measure with great precision the plate separation. I just used the machine as calipers, but it was surprising what you could do with the machine as calipers if you're very careful about it, and I could change the separation in coarse steps that were also precisely measured, and I made a number of measurements of the difference frequencies between the dominant modes of the system. I could see the dominant modes and approximately the satellites and beat structure that one expected by then. One of the things I found was that the approximate magnitudes of the shift of the laser frequency from the cavity resonance were what you'd expect, based on the estimates of the line width, and did not...and some knowledge of the cavity loss, but I also noticed that these beat frequencies were not the same, for all of the modes I could look at. That is, Townes's theory had predicted that there'd be essentially one constant separation between the dominant longitudinal modes, and I noticed as I tuned the laser through the line that there'd be variations in these different beats. In fact, the beats would split, and you could see, use the splitting of the beats to determine how the longitudinal modes were actually oscillating. You could count them up. You saw a split of the C over 2 (??) and there were three modes oscillating, and if it broke into three, there'd be four modes oscillating, and so on. And it turned out that the design I had constructed, made for this laser when it was built, produced — that the design I made for the laser meant that the beats were actually about 160 megahertz, which turned out to be the intermediate frequency range of the spectrum analyzer that I was using. It had a very nice effect. I could look simultaneously at the beats between adjacent modes, and the beats between alternate modes, and this provided a very nice confirmation of this interpretation of the number of modes that were oscillating in terms of the splitting of the beats. So this laser permitted trying out various hypotheses on what might really be going on, and I began analyzing the problem is some detail then. First of all, I realized rather quickly that the shape of the lines being Guassian primarily, as a result of the Doppler broadening, meant that Townes' theory was not applicable precisely there, because he obviously made the derivation based on a (crosstalk) — well, he was really extending the results that he had calculated with Jim Gordon and Zeiger on the ammonia maser to the optical domain; after the fact, he saw what he was doing, and then began to realize what needed to be done. So there was a little nonlinearity in the pulling, the mode-pulling problem. When the oscillation starts in the laser, the actual frequency shifts from the cavity frequency that we calculate based on something like the Fox and Li theory, to an oscillator frequency which is determined by the requirement that the closed loop phase shifts be an integral of a multiple of 2 pi in the system, that is, that the oscillator would shift to a new frequency from the cavity frequency by an amount that was determined by phase shifts in the medium. And one source of phase shifts that was important in the medium of course was the Gaussian line itself, and this had a nonlinear splitting effect on the beat spectrum that one could see, and it was about the right magnitude to explain some of the effects I was seeing. It was not adequate to explain all of them at all. And there were some power-dependent effects that were going on which would cause the beats to repel each other and attract each other in certain regions of the spectrum, which didn't have any explanation, based just on the Gaussian line shape itself. That just didn't — it was about the right magnitude, but it didn't explain the effect adequately. And I began thinking of what must be going on in the system in respect to setting up equilibrium in the presence of gain. And it was at that point that I started worrying about holes being burnt in the spectral line.
Is that an old term, or where did that term come from?
Apparently Mikel Bloembergen had brought up something like that in terms of special hole burning, I think, in some nuclear magnetic resonance experiments. I didn't know anything about Bloembergen's work at that point, and I sort of noticed that later, but —
You had consciousness of making up that phrase at the time?
No, I don't think I did, as a matter of fact. Let's see — I think, I forget exactly how it evolved. It was a question of what to call the effect, and it looked like there were small holes being burned in the line, and therefore I began using that terminology. Whether someone else suggested it or not, I don't remember at this point, as a name for it. I remember, when it came to publishing the paper, I had considerable discussion with Geoff Garrett at that point, who was now my new sub-department head when I moved to George Dacey's department, C. Goeffrey B. Garrett. I asked him, I tried out some of my ideas on him as sort of a sounding board, and when I started writing the PHYSICAL REVIEW paper on the subject, I was going to call it something less pithy, and he said, "Why not call it hole burning effects?" I said, "That's a good idea." I guess maybe Geoffrey Garrett suggested the use of that terminology, but it made certain sense, the model I was constructing of what was going on, and so forth. It turned out, I wasn't aware of it at the time, but Ali was apparently working on a similar bent. That is, he'd been worrying about the fact that the beats were split, I guess. And I gather, he had been worrying about the dispersion of the Gaussian line and its effect on mode- pulling, and I remember at one point in the spring of 1961 Geoff Garrett telling me that I was on a collision course with Ali, that he was working on the same problem I was, and that what we ought to do was have a seminar where I would present the work I was doing and he would present the work he was doing.
So you weren't really talking to him much after this intense —
— we weren't collaborating any more, after the initial publication of the results on the helium neon. I was working in a different department, and was sort of encouraged to do my own thing and so forth, which I did. In the meantime Ali, I guess, maybe he had already been thinking about going to MIT, I don't remember, but he had a technical assistant who was helping him out and I guess Ross McFarlane came to work with him at that transition period. For some reason, we didn't discuss what we were doing in spectroscopy at that point. But anyway, we then gave a joint seminar, where he discussed what he was doing and I discussed what I was doing, and I think, I'm fairly certain that what he was talking about was entirely based on the dispersion effect of the phase shifts due to the Gaussian line. He had a quite different way of looking at it. It was all complex dielectric constant, and I was thinking of it very directly in terms of phase shifts that I would calculate from the Kramers-Konig relation, based on the gain profile. In fact, I got some help with actually computing what these phase shifts would be from a guy named Bruce Bogert, who was over in Don Herriott's area. He had a nice numerical program for calculating phase shifts from any kind of arbitrary line profile, as a function of frequency, which, somehow he was interested in that kind of thing in connection with some electrical engineering problems. I don't know exactly what they were. I know he was studying problems of the cochlea? at that point, but whether this was related to this or not — but anyway, Don Herriott introduced me to Bogert, who had a nice — in fact, Bogert was a friend of my father's, I knew him that way too. He played the bassoon. But in any case, he had a nice computer program to work out phase shifts, for any numerical model of gain you wanted to present, and I said, "Here, run off this Gaussian," and he gave me the results of that numerically, which I then fed empirically through the form and showed that it was slightly different than the kind of thing, particularly near the center of the line, that you'd expect from a Lorentzian, and it had one side, that is, namely, it would tend to attract the resonance to the center of the line, the mode-pulling effect. But I was also observing some mode-pushing effects which weren't explained in this way, and it was that consideration which made me start worrying about the hole burning things. Yes, I guess actually Geoff Garrett had some useful comments around that point. I mentioned that I was worrying about the phase shifts that were due to these holes burnt in the line, and he suggested, "Why don't you try a Lorentzian model for the holes and see what the phase shift would be from that?" And gave me some results he'd calculated from the Krames-Kronig relation which turned out to be all wrong, in terms of how the phase shift varied, but I went through that and got it straight. The other thing I had done, I'd done a calculation related to this problem when I was a graduate student at Columbia. I had been working on a possible positronian spectroscopy experiment, where I was worrying about doing RF transitions between the excited states of positronium, and I had calculated essentially the same problem that was involved here, namely, the shape of an induced transition in the presence of upper and lower state relaxation processes, and the limit that the atom was both born and died in the field, and you can show through much tedious algebra and solution of Schroedinger's equation that you get a Lorentzian probability function which broadens as the power increases. That was something I just had in one of my old notebooks that I'd worked out, when I was at Columbia, and it occurred to me that it would work beautifully on this problem I was working on then. I'd never published the result at Columbia because the experiment on positronium didn't work, but I had worked out a lot of physics connected with that problem which I found very applicable to the laser question, it was really the same thing that... (off tape) ...problem I solved at Columbia when I was a graduate student had to do with the line shape that you would get from stimulated emission or absorption between a pair of levels in which you included both the upper and lower state relaxation rates or phase interruption rates and their effect on line broadening. What it showed essentially was that you'd get a, the rates which would start out as the natural width of the line, involving upper and lower state relaxation times, that the square root of something involving one plus the electric field squared; that is, that the line width would broaden due to power in the system, and because of that, it's included in the probability in affecting the atoms in the Doppler distribution would also be, it's a first order anyway Lorentzian distribution that might broaden this as the power and the mode increased. So that the notion that the whole was a Lorentzian shape was something I'd worked out, and then Geoff Garrett wondered if I might not calculate the phase shift from that Lorentzian, using the Kramers-Kronig relation. I think that's essentially the way it worked. I went back and looked at the Townes, Gordon, Zeiger and Townes article in the Columbia magazine (?) and they seemed to be doing something a little bit like that too, as a first approximation. I remember asking Jim Gordon something about it at that point, and he said, "Of course, that's naturally what we said," and it wasn't exactly the same approach to it that I was formulating, but nevertheless, it seemed to be physically and mathematically equivalent, except that I wanted to do it in the limit where the cavity width was not very large or very small compared to the line width, but more of an intermediate range. Anyway, —
Just to get in a little of the social context of this work, were you talking to Garrett very much or very little?
Yes, I saw him occasionally, I would say.
It's the same question: whom were you talking with at this point? Who were your intellectual —?
I guess I was more talking to Garrett than anyone else, for a while. It was more discussing what I'd done after the fact, after it had been done, rather than discussing it with someone while I was doing it.
And the other thing is —
— I was mainly working this out, on my own, filling up notebooks with calculations and stuff, which are probably illegible to anyone else.
And what was the relation to Bell Telephone Laboratories? On the one hand, I would assume at this point they were happy to have anyone work on lasers, period.
Oh yes. I think so. Anything you could learn about them was fair game, I think.
But on the other hand, you did bring up a frequency stabilization method in connection with that paper, and I wondered where that fit in?
Yes. That was a practical application of the hole burning analysis that I'd been in. Yes, that was certainly, also involved a beginning concern about it. The kind of question that came up, as a result of papers delivered in Berkeley in '6l, in March, were something to the effect, here you have this extremely narrow source of oscillation, it might be on the order of a cycle per second or less the inherent line width, but it's tuned to a cavity resonance which is floating all over the place, due to thermal expansion and vibrations and so on. And it is certainly true that one of the things I was worrying about at that point was, how do you make this stabilized? How could you tie it to something that was inherently more absolute than the dimensions of a particular piece of apparatus? And obviously, one thing to think in terms of was the center of the line itself, and therefore, to know how the laser frequency was determined by the parameters of the system was an important question, if you wanted to worry about frequency stabilization.
You see, I have the feeling, the hunch, which I just want to offer you, that there is a very intimate connection between technology and science in your work.
Oh, absolutely. Yes.
OK. I mean, it just sort of moves into each other, as —
— it's true. I guess I do enjoy analyzing things that have some relation to engineering problems. Although I also enjoy the pure physics too, but it's always fun if it's related to something that you can actually use and do something interesting with. That's right. Anyway, I was worrying about this question of how the frequency oscillation was determined by the characteristics of the system, and gradually evolved this hole burning model or theory, however you want to call it, that seemed capable of doing just that, and we did get results that indicated that the general magnitude of the beat splitting and movement of the beats as I tuned the laser was about that which the approximate analysis I was doing would give, and as I say, a seminar was arranged in which Ali presented what he was doing and I presented what I was doing. It seemed clear that he was worrying about the Gaussian phase shifts in his method of attack, but he wasn't worrying about the hole burning effect at all, and I was on the other hand in the middle of a quandary, because I could see this splitting and how it was varying with frequency as you tuned the laser through the line, that it was clear there were two opposite effects involved near the center of the line, one effect that was pulling it in towards the center and one effect that was pushing it out, and I concluded on the basis of those observations I made, and also playing around with this notion of holes being burned in the line and what the phases would do, that the pushing effect away from the center of the line was actually due to hole burning terms, that is, there were holes being burned in the Gaussian gain curve, and now one way of analyzing the problem is, you just write down the phases from the Gaussian itself, and then you add up the phase terms that are due to these holes that are burned in, and they would tend to move the frequency in opposite directions. And I then worked out quite a bit of that, and wrote it up in a paper that was published in the PHYSICAL REVIEW.
This is a paper that almost sounds as if you were writing it as you were thinking.
There's a kind of —
— I was. I always tend to do that.
I guess it's a personal habit. I tend to explain things to myself by writing them up in a way that I think might be understandable to someone else. Maybe it's a weakness.
No, there's a kind of evolutionary quality to that paper. It doesn't sound like the kind of paper somebody wrote when he was all finished with everything and sat down.
You're right. I tend to do that more often than I write up coherent legible notebook entries. I tend to try to write things out. Well, one of the reasons is that I type more naturally than I write with a pen. My handwriting is terrible, and I found in early life that it was much easier to type, and often I will start typing a paper while trying to explain things to myself.
Now, at this point, did you know —
— I'm surprised you noticed that. Is it that obvious?
I guess. As I say, I wasn't reading the paper all that carefully, but on the other hand, I think that's something a historian would pick up more quickly than a physicist would, because of the difference in the way we look at things. At this point, were you in touch with Willis Lamb?
No. No, Willis Lamb wasn't involved in things at that point. Ali was the one who got Willis Lamb involved in the whole problem, and I don't remember exactly when he did, but he got Willis hired as a consultant to the physical research group, in which he still was. I'm not sure when that started, but it was a long time later. I wrote that paper, I think it was in early summer of 1961. There must be a date you can see on it somewhere. I don't know when it was exactly. I thought it was the early summer of 1961. And sent it off to the PHYSICAL REVIEW, and they have their traditional long interval in which —
— November of '61 was the submission date.
November, was it? I see. Anyway I was writing it over the summer, and I was also doing some other things, so it was not —
November 16. Oh, it must be over on that side.
Here it is, all right, November 30th, 20th. Yes, I was working it out from about the end of spring to summer, and I guess, as I say, I gave a lecture or seminar on it in the early stages of it that summer. I was still feeling my way along with this model at that point. And I finally sent it to the PHYSICAL REVIEW, I guess it was about that fall. Now, I didn't learn that Willis Lamb was getting interested in this problem until certainly after I had finished writing this paper. I don't remember exactly when I did learn he was interested. He was still at Oxford, and I guess Ali had made the arrangement for him to work on the problem, and mentioned it to me at some point, and I began getting some letters from Lamb in the spring of 1961, at which point I sent him I think a preprint of this paper.
Or, do you mean the spring of '62?
'62, sorry. Spring of '62. And we had a correspondence on the subject of gas lasers sort of sporadically over that next, oh, probably from about February to July or thereabouts. He was in the process of deciding to come to Yale at that time. I guess one of the reasons he wrote me was to talk me into coming back to Yale. It wasn't clear that I was going to, at that point.
Do you have that correspondence?
That would be certainly an interesting —
I suppose that his letters are the property of Lamb, and I don't know whether he wants them distributed. I think I have carbon copies of the stuff. Anyway, Lamb sent me some material which he'd written describing what he'd been doing on the theory of the problem, and as I said, I sent him a copy of the hole burning paper. Lamb's approach to the problem was not very geared towards physical explanations of what was going on. It was clearly filled with interesting and perhaps even elegant math in places, although a lot of it was pretty tedious. But it was, you know, an interesting approach to the problem, in which he essentially started with Maxwell's equations and Shroedinger's equations and ground things out. And one of the things he produced was an equation describing a power dip in the tuning curve if you tuned a single cavity mode through the center of the line. I was quite struck by that result. It seemed very strange. I wasn't expecting it at all. And I wrote him a number of times asking him if he had a physical explanation for it, why is it doing this? He would always write back in a very mathematical way, saying, well, third order phase, that atom — you'd just sort of read off the mathematics in a summation his expressions that essentially said that the equation said that there was a dip and that was it. I asked Ali if he had any idea what the physical explanation was, and he said he didn't have any idea. He was sure Lamb didn't have either, in fact.
Ali Javan must have been at MIT by now.
I think maybe he was. At some point during that period he went to MIT and I had some brief conversation with him. Oh, I know, Willis said he was trying to get ahold of Ali, and would I please find out where Ali was and get him to answer his letters. As a matter of fact, the correspondence was sort of a joint one. Willis addressed letters to both of us and sent us each a copy, and although I just replied to Willis, and I guess Willis hadn't gotten any reply from Ali, and I guess Ali had been instrumental in getting him hired as a consultant to the Bell Labs, and he was wondering where his paychecks were and things like that, as well as data connected with the laser. He was anxious to have experimental verification of this dip, or to check it out and see if it really was there. I tried very hard to see if I could find a sign of the dip with the laser I'd made, and I spent quite a lot of time taking data with it, and it did not show a dip. Basically the problem is that the laser is too long to show a drop of this amount, about a meter long or — and it was really impossible to see any sign of a dip under the conditions that I contrived, in single mode, that is, it would tend to go two mode operation before I could really tune properly to see the dip, and there were also some questions of what the collision broadening would do to it and so on. And I sent those results back to Willis, again asking if he had any physical explanation of the dip, and again got equations as an answer.
So this is a way — we're really getting at the Lamb-Bennett-McFarlane —
Yes, well, right. Two lasers were put together, essentially the same design that I had made for the previous work I'd done —
— I didn't mean to interrupt the flow of your story, I just —
Yes, that's about that period, right. The shop in Ali's department had built two of these four rod structure lasers, exactly the same design I had come up with for the hole burning paper work, except that they were about half the length, and had a hope of seeing the dip. Ali took one of those to MIT with him, and he and Zwicke (Zucker?) looked at it there, and Ross McFarlane, who had been hired as a post-doc at Bell Labs, had the other one, and we used it there to look for the dip. Unfortunately, the lasers weren't really ready to try out very well until about the point I had to go back to, I returned to Yale, and that was, the appointment started July 1 in '62, so that McFarlane sort of continued the work on the problem that I collaborated with him on when I was still at Bell Labs. It was finished up after I got to Yale.
I see. So it was a real long distance collaboration, because Lamb was in Oxford.
Right. Yes. That's right. And in the meantime, I had been thinking about the problem a lot, and I realized at one point what I had neglected in my analysis, which meant the dip didn't show up in my equations, in the hole burning paper, and I also, I had started — let's see. I agreed to be a consultant at TRG, starting July 1, 1962.
Tell me a little bit about that, and the — well, I'd like to know —
This relates to the hole burning question, in a way. Before doing that, I had gone down to TRG to give a lecture on — I had agreed to consult on several things at TRG, but one of them involved worrying about the use of the hole burning theory that I worked out to make a stabilized laser, one that would be stable for the line centers.
They picked up on your paper and said, gee, we're really interested in stabilization, or —?
Yes. I think they had one of their many contracts there which was Signal Corps or Air Force, I forget which, to — it was the Air Force actually — to build a stabilized laser that would be locked at line center in some way. I went down there some time before July 1 to give a lecture. They wanted me to give a talk on the problem, and I described my hole burning theory, in that lecture, among a number of other things. There might have been more than one lecture. I don't remember precisely. But in the course of giving this discussion, and I was showing lots of pictures of line shapes with holes burned in them and so on, in describing the single mode case, Gordon Gould asked me, what happened to the other running wave and the standing wave? And it suddenly occurred to me that I'd totally forgotten about the other running wave, that I'd been worried really about a ring laser where the radiation was always going around in one direction, and the analysis I'd worked out was valid for that, but I had forgotten, just foolishness or carelessness, I'd forgotten to include the light going in the other direction, that would give rise to a standing wave in a plane parallel type of geometry, and I then began drawing in where the holes would be from the running wave going in the opposite direction, and began tracing out how this would vary as you tuned towards the line center, and I think it occurred both to me and to Gordon pretty much simultaneously that this was really the qualitative explanation of the Lamb dip. And I then started worrying about it in more quantitative detail, and eventually showed over the next few months that you get exactly the same equations for the frequency that Lamb was getting out of his more complicated theory, from the simple hole burning model, and one thing I do have is a copy of a letter sent to Willis on July 7th, 1962, in which I describe this effect. This was sort of part of the — the end of the correspondence really between him and me. Anyway, it was an explanation of the dip.
A photograph of this correspondence ought to go into whatever you leave.
Well, I can give you a copy of it if you want. Anyway, that was the first time I'd written it up a little bit, and I sent it to Willis, with the idea of getting his feedback on it. I guess he never replied. I guess he was in the process of moving to this country, and many weeks later, he made some comment about it, and that was about it. But I did then work out the complications that would result from this additional hole that was due to what I called the mirror image, hole in the profile due to standing wave considerations, to show that one got essentially the same results as Lamb did, and that was published in this paper that was presented at the Paris February 1963 Conference on Quantum Electronics. There are actually three different things that I talked about in that paper. Bloembergen invited me to come give a lecture there, and he suggested any of three topics, relaxation mechanisms, dissociated excitation transfer, and mode point effects, and I decided I would write about all of them, and in particular I included this generation of the shape of the Lamb dip that you would get from the hole burning model, and also some early stuff on the variation of frequency that would arise.
By the way, did that conference have any intellectual impact on you? You were there and Lamb was there, and —
Lamb was certainly there. I saw him sitting in the front row when I gave this paper. It certainly, there was a lot going on. There was too much going on, for one thing, as you can see from the two volumes that came out of it. And there were a couple of rather practical complications in my life at that time that made it hard to get an enormous amount out of the conference. One of them is that — well, it was in the middle of February, and the term was well under way at Yale, and I could only take a few days off to fly over.
Oh, you were already at Yale then?
Yes, I was at Yale.
I'm sorry, this is '63.
'63 when I presented this paper. I was in the middle of a quantum mechanics course back at Yale, and I had to get someone to fill in for a couple of lectures while I went to Paris. I brought my wife along, and Fran, who had relatives in Paris at that time that she wanted to visit and so on, she came down with some awful bug, some virus, and really got a temperature that went up to 104 or so and she had to spend all her time in bed essentially recovering from this horrible bug, and I hadn't had time before getting there to type up this paper. I'd written out notes to present the lecture with, and in fact it had illustrations but I hadn't had the slides made yet, and I got some slides made by some photographer in Paris who put them together for me, and I spent a certain amount of my time typing up the paper in the apartment we were staying in in Paris during the conference, so that I'd have it ready to give to them when the conference was over. They had said they had a firm deadline to get all the manuscripts in by the end of that conference, and I did indeed hand in the paper to those people. I put it in a giant carton filled with many other manuscripts at the end of the conference, and I assumed they were going to publish it right away, and I was really quite irritated to find that it took a year before this volume came out. Otherwise I would have sent the same results off to the PHYSICAL REVIEW LETTERS or some other place. I notice some of my contemporaries did both. Namely, they submitted a paper at that conference and also sent the same stuff in to the PHYSICAL REVIEW. But I wasn't in the habit of doing that. I just sent it to the conference where it was eventually published. But anyway, this was all presented in February of 1963 in Paris, and I remember clearly that Lamb was in the front row, along with Martin Ramsey and a couple of other people, and he didn't seem to acknowledge that it was really producing the same results that he was getting, which sort of surprised me, but in any case —
I was wondering, what were other people's reactions to all of this?
Several people wrote me later saying they were quite interested in it. Some of them were doing ring laser gyro work, and seemed to be quite interested in it and so on, but in general it didn't receive much attention, and yet it really was somehow a breakthrough in understanding how things like the Lamb dip arose from the hole burning model. When Lamb got around to publishing his paper, which was quite a bit later, he pointed out that I'd neglected one thing in here, which was, namely, the fact that the whole was initially burnt in the Doppler distribution, and one had to integrate the gain response over the hole in the Doppler profile to get the population distribution rather over the hole in the Doppler distribution to get what the gain shape would be. I had argued on heuristic grounds that the saturation would occur in the gain profile itself, because if you had steady state in a system of this sort, the gain had to be reduced to the loss, otherwise it wouldn't be steady state, it would start building up, apart from fluctuation noise due to spontaneous emission, and therefore I was adopting a model in which the hole was burnt in the gain curve directly. Well, it was actually burned in the population distribution, but the results of the hole were in the gain curve, and the essential difference between the two, in this first order approach, is that the Lorenz width varies by a factor of 2 from the gain hole to the population hole. But anyway, this was essentially the first computation of the Lamb dip from the hole burning model, and but then later I tried to clean things up a little more, and in that built the physics of gas lasers in a more complete way. That was actually presented at the Brandeis 1969 Theoretical Physics Symposium. But —
By this time of course you were physically at Yale and Lamb was physically at Yale.
Did you have much intercourse over physics, you two?
Somewhat. Probably not as much as he wanted. There was a tendency in dealing with him that he would take the attitude, "You do what I do and see what happens," and I really wasn't temperamentally inclined to do it that way. I wanted to understand these micro — in fact, I had worked out independently a new approach to the theory of these things, and I was hoping that he would give me some advice on it, being a senior man in the field, and instead he just sort of swallowed it into his version of the thing, and I found it a little disappointing. But also, it seemed clear that his interests were going in some other directions at that point. He wanted to worry about quantum statistics effects, and the laboratory space that was available in the same physical area at Yale where he was working was not adequate for the things I wanted to do, so I tended to — I set up a laboratory in another part of the campus which was a little bit removed from him, and I also at that point was getting very interested in ion laser problems, which distracted me quite a bit from trying to look at what was happening in the theory of this particular problem. But it still interested me. There was also the general question, there were so many things...(off tape)
... intellectual unrolling, and also what would have been the picture of what people were doing in terms of gas laser theories, sort of in the general field?
Yes. Well, as I say, I had started working on the hole burning model when I was at the Bell Labs, and continued to expand it as time went on, and found that I could solve most of the problems that I was interested in, based on this way of approaching the problem, whether it was the single mode helium neon laser, or something more complicated with many modes, or a laser with a saturable absorber in it and so forth, that I could calculate all the main effects fairly accurately in that way.
Is this the period when you were thinking of this as sort of semi-classical versus quantum theory, that you could do most things with it, and then there were a few things around the edges that you needed to turn to a different theory?
It may be that's a fair thing to say. The regions where it doesn't apply tend to be fairly small, and involve effects that frequently aren't dominant. Lamb's theory, of course, was a classical theory too, in the sense that it —
— yes, I just meant that —
— the, he did a more rigorous self-consistent field approach to the problem, and there are probably some areas where he had a slightly different result than what I would have done, but they aren't conspicuous. The theory that many other people worried about had to do with noise properties in lasers, and I didn't really attack that problem.
I see, that would be Lax, for example?
Yes, I guess so, and things Haken did too, but I didn't worry too much about the results that Lax and Haken were obtaining. They didn't seem obviously related to some of the things I was worrying about myself, and they were highly formal, and not sort of directly applicable to these kinds of experimental situations I was concerned with. Basically there was only so much one could do, and that didn't seem to be too fruitful at the time, in connection with things I was following up. The questions of noise in the problem were ones that generally weren't practical considerations, unless you were really going to extreme lengths to set up an apparatus which would permit seeing them, you know, operating a laser in a domain that you wouldn't normally operate it, just going out of your way to make things super-stable so that the noise questions would be observable. I'm not saying that noise isn't an interesting problem. It's just that it's one I didn't get involved in too much. Under normal circumstances, the noise that was present in a laser was induced by mechanical vibrations and local changes in the environment, and things of that sort, rather than inherent quantum limits that one would like to see.
If you were pursuing these other theoretical analyses, then I guess basically I was following up several things at once, and had about all I could do at that time to teach courses and at the same time do the other research. Now, as far as the hole burning work itself went, I think I did mention that I had agreed to do some consulting for TRG, actually on two separate problems. One of them was the frequency stabilization project, and the other had to do with some metal vapor laser work and the like.
Who were you working with, by the way, on the frequency stabilization? Who were your people at?
Well, initially Gordon Gould asked me just to analyze the problem as carefully as I could and make some suggestions of different approaches, which I did essentially myself, while working to a large extent in New Haven. Then I presented a paper on that which was published in one of their progress reports, though not in the open literature. It was decided we would follow up one of these schemes, and the people that collaborated with me on that were Jim Latourrette and Paul Rabinowitz — I've forgotten a name. There was a paper entitled — oh yes, Steve Jacobs was another one — "Dispersion Characteristics of Frequency Stabilization of the Helium Neon Gas Laser." That was number 39 on this list. That was the main result of the experimental work that we did in collaboration.
When you did experimental work, you went down to TRG?
Yes. I did. Yes. It's an awful trip. It took many hours. It was very frustrating because this place is in Sahassett, Long Island, essentially directly across the Sound from New Haven, but in order to get there, you had to drive in to New York and out the Long Island Expressway, and I went there oh, often, as much as once a week. I tried to arrange it so that I could stay there a couple of days and maybe work all night on the problem and come back and ignore it for a little while, and then go back again. I had agreed to consult with them something like 30 or 40 days a year, I forget exactly what, on this problem, which I did, and some of it was during the summer, when we — and some of it was understood to be done at New Haven instead of on Long Island. But anyway, I did go down there and spend a lot of time with them, taking data, and the basic notion was to try to make use of the dispersion calculations I'd done on the hole burning effect to stabilize a laser. In this case we used the 3.39 micron laser, and I think that was actually the first feedback stabilization scheme, based on the line center that I had tried
Did it lead them to any new technology?
Well, it provided a fairly stable laser for that period. I think it was probably the most stable one around at that point. For periods of a few days it was reasonably stable. But it had problems with the systematic drift that were due to variations in the electron density that I hadn't anticipated in the original case, and there were some phase shifts due to the electron density themselves that affected the frequency of the line, and it turned out not to be as useful as some other schemes that were based on looking at, oh, tuning dip for example, an external absorber that wasn't in the actual cell that was being subject to the discharge.
See, I guess I was —
Do you have a copy of that paper here?
Let me see it.
I would guess that they might have been doing that because they were selling lasers, and I'm wondering, when you have a relationship like that, whether you have any — what?
— as far as I know, they never sold lasers except to the government. They had a contract with, I think it was Wright Air Force Base. 39, is it? They had a contract with Wright Air Force Base, I believe, to deliver a stabilized gas laser.
It's my recollection, it was taken over by the Air Force, under one of those big grants that contracts (crosstalk )
…Project Defender grant? ARPA?
Yes, they did. And as I recall it, we were doing it specifically for the people at Wright Field, although I don't see a reference to that in the paper, but it is my further impression that after we got the thing working and tried it out for a while, that the thing was delivered to the people at Wright Field and I never saw it again. We did take some data on it for a while, and the technique looked interesting. As I mentioned before, there was a problem with an absolute shift in the electron density which was irritating, and I guess was the fundamental limitation, in terms of a real long term absolute standard, but one can get stability within a few kilocycles over many hours, as shown in this plot, and we made a couple I think and beat them together. I forget exactly to what extent we did that. Anyway, largely this was based on utilizing the dispersion curve that I'd calculated in the case of the hole burning model, applied to the helium neon transition at 3.39 microns, and that was a good line to work with, in terms of the high gain available, so you could make a very short length — it was very easy to see the Lamb dip, for example, in that case, and very easy to compute the dispersion curve that you get by dithering the gain, but the other problem with it was that the Lorentz widths in that system were rather large. In other words, they weren't negligible, infinitesimally small compared to the Doppler width. That is, they were a significant fraction of the Doppler width, and hence it was not the most sensitive stabilization one might get by application of this kind of system to a gas laser. Incidentally, there was a typographical error made by the printer in this, that I guess we never corrected in print, in equation 1, where the square brackets should have been l minus .94, instead of .94 times the whole width of the Doppler width, but in any case, it was a fairly direct clarification of some of the calculations I'd done on the problem, and sort of interesting from that point of view. We did, as I say, make a laser that was reasonably stable over short term periods and fairly satisfied the Air Force who took it away. I don't know what they did with it.
It sounds like a very nice dovetailing of your interest and theirs at that point.
The usual consultancy I think of as, it really pulls you a little bit away from your own —
Yes, it was an interesting overlap. I didn't have the facilities at Yale to do that kind of an experiment. They had some very nice equipment they'd built up at TRG, and particularly they had piezoelectric transducers that would work fairly rapidly for sweeping the frequency of a laser, which made it easy to see these sorts of things on oscilloscopes and so forth.
When you take a job like that, a consulting job, is equipment a consideration? Are the problems a consideration?
Oh, it certainly was. It was something I was interested in working on and didn't have the facilities to do it at Yale. And therefore it looked like a reasonable thing to try to do at TRG. And of course, there we had three fairly able bright PhDs in physics that worked steadily at it, whereas if I'd try to do it here at Yale, we had a graduate student who was loaded down with courses and final exams and so on, so at that point it just wasn't realistic to try to do it at Yale. This work seems to have attracted some attention fairly recently, from other people in the field who were worrying about K S 3(?). Much to my surprise, I saw some references to it fairly recently in the literature. But in any case, that was one of the consulting problems I was working on at TRG. Other than that — oh, I did do some other experiments with hole burning effects, in particular there's a fellow named Chebotaev from the Soviet Union who was very interested in some of the problems I was working on, and he came from Novosibirsk, and he arranged to get a half year exchange leave to come over here to work in my laboratory.
Now, when was that?
That was starting in about February, 1966, through half of that summer essentially.
Is this a case where he had heard of you?
And he wrote you or something? Or you had heard of their group?
I hadn't heard of him. He had heard of me and he sent me a number of his reprints and a letter, in which he asked to come visit my laboratory, and it sounded like a good idea to me. He had done some novel things. I guess most of the papers were in Russian and my Russian wasn't that good. I could sort of struggle through them. But he turned out to be an extremely bright, innovative guy and I enjoyed collaborating with him a great deal and enjoyed having him around. We did some further work on hole burning effects, in which we actually observed the hole directly that was burned in a cell of helium neon gas in discharge at 3.39 microns, and that — the trouble is, the exchange period was only for six months, and we'd just started getting the results when he was about to go back to the Soviet Union, and we didn't really get an adequate paper written up on that. We did present a paper at the Leningrad conference the following summer in 1967, and wrote up an abstract on that, and I think he wrote a paper in Russian based on it. And I also summarized the results that we presented there in the Atomic Physics Conference that was held in New York around that time.
Does that mean that there was a gap between this work at TRG and Chebotaev coming in, when you weren't much concerned with that?
I guess what happened was that the contract that TRG had with the Air Force expired; when they delivered the stabilized laser to the Air Force, that was the end of it, and there wasn't any more support for the problem at TRG.
What about at Yale? You weren't working on this generally until Chebotaev came?
I was trying to get some further work done there. I didn't seem to get too far under way until Chebotaev arrived, actually, and he was essentially working on it full time for a while with me. We got some further results there. I was looking for that reference, where I recorded the results of Chebotaev and Knutson. Knutson was one of my graduate students at the time. We wrote the things up actually for a couple of places. The most complete account of it was given in the review paper I published in the PROCEEDINGS of the Atomic Physics Conference in New York in 1969. It's number 67 on this list. And we had also planned to give a — we gave a paper at Leningrad in '67, and we also had submitted some papers to the Miami conference in 1968, which is number 66 on this list, I guess, which was largely a long abstract. I got pretty much derailed at that point for personal reasons. My brother had been killed in a train accident. That was May 13th. It was the day I was supposed to leave for the Miami conference, and in fact, I was having lunch with Nikolae Soboleev from the (?) Institute and George Schultz at that time, and we were both, all three of us were going to go to Miami together and we were going to present some of these papers, and I was going to present some of the stuff on hole burning, and I learned that my brother had been killed and I just couldn't go at all, had to go — you know. I didn't give the paper anywhere. So I stayed home and that one sort of went by the wayside.
There are a few things I wanted to clear up —
I had written some papers, that I was actually going to publish something in the PHYSICAL REVIEW LETTERS with Chebotaev, and there was also another paper on hole burning effects on the argon laser, which I wrote the manuscripts up but I never sent them to the journal. They got out of date, after my brother's death, and then I got involved with writing something up for the Brandeis 1969 Institute, and I decided the simplest thing would be to try to revise a lot of those things and include them in that article, which was published very much later for many reasons.
I want to clear up some things that we've been talking about. By this time, when you're working at TRG and they have all this equipment that you don't have, by this time I'm assuming that you have a lot more than $500 a year.
Oh yes. Sure.
This Air Force OSR contract and —
Right, I had gotten the Air Force OSR contract, fairly quickly after coming to Yale, and about a half a year or a year later I got a small contract from ARD, the Air Force — excuse me, it was a grant.
I wonder, I'd like to get a little feel for how much equipment you got from that, and whether you had any intellectual interaction with the monitors on those, or whether they were just giving you money.
I had some interaction with them. I can tell you roughly what was involved. The offer I got from Yale involved tenure, professor of physics and applied science, was also a statement that they would provide $200,000 for research equipment, and when I got here, I was told that I could have $100,000 right away but I had to go get the other $100,000, which sort of slowed me up a little bit, and the first $100,000 turned out to be part of AFS (?) that Vernon Hughes had already going, and I had given a paper at the New York Meeting of the American Physical Society, when was it? It was probably January of 1962, in which two AF OSR people had attended, Lloyd Woods and Marshall Harrington, and on that occasion, I not only described some of the dissociative excitation transfer experiments, but I also described the work on the hole burning problem that I had done, and both Lloyd Woods and Marshall Harrington seemed to be fairly impressed by this work, and wondered, if I were going to go to Yale, they expressed interest in providing a grant, except I'd have to write a proposal, which I of course did after I got to Yale or about the time I went to Yale. So I did get another $100,000 from them over the next year, I think it was. Wait a minute, no, it wasn't then, it was a year later. Let's see, I had an Alfred P. Sloane Foundation fellowship, which was about $10,000 or something like that.
Are those equipment fellowships? What are they for? Did they relieve you from teaching or something?
Well, the Sloane Foundation things you could use on anything you wanted. I did use it to buy some equipment. I forget exactly what, but it was something I needed at the time. I submitted the proposal to Air Force OSR at that time. That was around December of 1962. I also sent one to AROD. And I guess I didn't get the Air Force grant until about a year later. Those things have big time delays. I spent July of 1963 consulting for the Institute for Defense Analysis at Woods Hole, in which I had some interaction with Air Force OSR people who seemed very keen on supporting the research. I don't — what we were talking about there —
That's an advantage of working with IDA that I hadn't thought about before.
Yes, right. And by the end of that Woods Hole period, I was pretty certain I was going to get the $100,000 grant from AF OSR. I guess I didn't get it until the summer of'63 or thereabouts, or the fall. I have the records over in my other office somewhere, but that's about when it was. And somewhere along the line, AROD , Robert Lontz actually got a Yale degree in physics while I was teaching there in '57, was monitor of the contract I got from them. He expressed interest in some studies I suggested on cavity mode pulling work and stuff like that. It was a rather small grant, $30,000.
Did these contract monitors come and say, "Well, the Air Force is really interested in your work," or do they say, "The Air Force is really interested in such and such, and if your work bears on such and such," and then you may tend to emphasize such and such a little bit more than you would otherwise?
It's more the latter. In the case of the Army opposite Durham I had sent them a large proposal, about four separate parts to it on various things, on the hole burning effects, to frequency stabilization, possible velocity of light measurements, something on cavity, optical cavity modes, and they responded that they'd be interested in supporting the optical cavity modes proposal, not the other things, so they divided the total amount asked for by four, and said they had that much available if I wanted to do it. With the Air Force, the initial grant — let's see. The funding that Hughes had at AFL from which I got my initial support was a thing which, I guess he was supporting Lamb too on that, and I guess he'd gotten the grant on the basis that both Lamb and I would probably be coming, and it talked about basic frequency effects and understanding basic processes in gas lasers. It was moderately vague. The later funding from AF OSR I got was somewhat tied to a notion of working at new laser media. I had done a study at Woods Hole of power limitations on various existing gas lasers, and made a number of suggestions about the directions one might go to avoid those limitations, including collision lasers of different sorts, and it was — I guess I have an extra copy of that somewhere. It was something, I forget the exact title, it's in this.
Would you add to this conversation we just have had that there was a greater and greater tie-in, you said, as time went on, of the grant money to doing certain particular projects, and that there was an effect on research, that the research would go into an area where the money was.
Yes, that sounds true.
And that the intellectual interaction was not of importance, that is, the relation between you and the monitor wouldn't be one of the things that was influencing the cognitive content of these.
Well, when I was at Woods Hole on the IDA study, there was moderately- high level interaction there. There were a number of other active scientists there who talked to each other. I did discuss some of the things I was thinking about with them, particularly Bob Collins had been involved in this project, and I guess he was the one who actually asked me to come to the Woods Hole meeting.
Alexander Glass, was he involved in that?
I don't remember. There's probably a list of the attendees, in that research paper that was published, number 33, "Problems Related to High Powered Gas Laser Systems," and I'm sure I have an extra copy of that over in the Yale office. I didn't send it to you because it's awfully heavy.
It's more a question of whether you remember people you were specifically —
I did talk to Bob Collins. He was obviously much more concerned with solid state lasers than I was. Keith Brueckner was there, and I talked a little bit to him, and in fact, it was sort of the result of the Woods Hole study that ultimately led to that Lajolla Conference on Chemical Lasers that we had later. I think Kurt Schuler was somehow — no, I can't remember whether Kurt Schuler was at Woods Hole or whether he was at the National Bureau of Standards, but I had some interaction with Keith Brueckner, who was then at Lajolla, and it was largely through his encouragement that we got the conference scheduled there.
Does that mean that you were one of the people who wanted the Lajolla Conference to come into being or what?
I'd assumed you were taken on board after it was —
No. Actually Kurt Schuler was the one who really wanted that conference. He was a chemist.
Was he at the National Bureau of Standards?
Yes, I believe so. And he wanted to get into the laser field, and he thought that the way to get into it was to organize a conference. I remember him saying at one point, "There are two ways of getting into a new field. One of them is, you write a book about it, and the other is, you organize a conference." He asked me if I would help him put together such a conference, and I agreed to do it. And he was going to round up all the chemists he knew, and I was supposed to try to round up some physicists to come talk on things and that's essentially what we did. He knew Polanyi and W. Herzenberg and company.
Schuler I think had the idea of forming a conference first, and he talked to Charlie Townes to see what Townes thought about it. Townes said he thought it was a good idea, and I think there was some NSF and maybe ONR money that was used to help put the conference together, some of which was left over and consumed the following year, to have the summer program in theoretical laser physics questions that seemed relevant to our interests. So anyway, Kurt Schuler was the one who actually had the idea for it. He wanted to try to get into the field. He noted that I had made the first chemical laser when I was at Bell Labs with Faust and McFarlane, and it was the neon oxygen system, in the sense that the neon metastable was used to break a bond in the oxygen molecule and dissociate the two, therefore it was a chemical laser. I think that's right. It did involve exciting the neon metastable, done by an electric discharge. But —
Were you interested in chemical lasers? There was work going on — well, there was Polanyi, there was some work at Harvard, Zaire is the name that comes to mind.
Yes, Zaire was involved. Yes, he was at that conference.
Were you interested?
Yes, I was broad minded about it. I liked seeing some obvious directions it might be going. But it — not too many things seemed to arise that I thought made sense at that time, for me to get involved in it, and there were some other things that did come up which I followed instead. But indeed, the notion of chemical lasers made good sense to me, and as I say, I did do the initial work on the oxygen, neon oxygen molecule and the argon oxygen system, and had interests in such things.
See, I always think of the chemical laser as being connected with DOD because it's a way to decrease the weight of the power source, which is nice for — I don't know.
Yes, well, I guess they were willing to be interested. I think most of the DOD thinking at that time was based on systems that already existed. The general characteristic of DOD meetings on the use of lasers for defense problems was to take the latest well-established development and see how you could scale it up. That's essentially what was going on, I think, for several years in DOD. They'd say, how big a ruby laser could you make? Or how big a neon radium YAG laser? How big a gas laser? They actually went ahead and built gas lasers. There was the Huntsville, Alabama, I forget what it's called, missile base — there's some kind of a DOD —
No, it's not Redstone. In Huntsville there was a DOD laboratory that was asked to build a helium neon laser that was something like 300 feet long. It was absolutely absurd. It was practically the length of a football field. You know, the idea was to see how much power you could get out of this thing if you kept making it bigger and bigger. It was rather easy to show that it would very quickly saturate, and you'd get a certain power per unit length that would not do what they wanted done.
You really are changing some of my understandings of this, which is very useful. I mean, I sort of thought that DOD would be in there in the background behind some of these new lasers, like argon or chemical lasers, and —
Well, the DOD people couldn't really suggest anything new very specifically. They would sometimes give grants with enough generality in them that you could go out and do new things. But —
I mean, if DOD wants to do, say they want to do anti-submarine warfare, so they would like to have a laser in the blue or the green, does that have anything to do with making argon lasers?
Well, it's a funny thing that you should mention that. That topic often came up, in the naval research areas; the ONR people seemed to be interested in that, the Naval Advisory — you know, Committee was quite interested in this question, and one kept hearing reports that what the Navy really wanted was a blue-green laser, so they could try it through seawater. After we'd made our first argon ion laser, they didn't seem to be terribly interested in that. And I remember another experience I had consulting at TRG, in which we did make a green laser there, a copper laser, that's another problem I was consulting on or with, and that was supported by ONR, the whole project was supported, until we got the copper vapor laser working and started getting large amounts of power out in pulse form. Then they terminated the contract. I was left with a certain feeling of mystery, what it was that these people really wanted. But anyway, the copper vapor laser was an approach in that direction, and after that project was terminated, you know, I again kept hearing from time to time, what the Navy really wanted was a bluegreen laser, but there was a certain amount of fickleness in the way they support such things. In any case, the DOD people couldn't suggest anything new, that hadn't been done, because they didn't know what was going to be practical. They would say, "Well, it would be nice to have an ultraviolet laser," or this laser or a laser in that wavelength range, but they wouldn't know how to do it, so they wouldn't say, "Build an argon ion laser, or copper laser." I think I'm wandering from the main —
No, I'm keeping you where I want you.
You do, I think —
Anyway, I started to say that I had quite a bit of interaction with some people at IDA — during that 1963 study — and I suggested some various things that I thought might work, to them, and they gave encouragement to the AF OSR I think to give me a grant to do some work which was actually one of the things that helped fund the argon ion laser work later. But there were always time delays involved in those things, that were serious problems. I did get some loan of equipment from Vernon Hughes, and money actually to buy some equipment. I wanted to get some mirrors that would work in the visible part of the spectrum to try out various things and Vernon Hughes had another grant going on at the time, and managed to squeeze out some money to get that set of mirrors, which were the ones that I used to make argon ion lasers at Yale. That was later picked up and supported by AF OSR, but anyway.
How did that argon work come about? You got the summer study and I guess your first paper is out in the spring of '64, and by then a number of people at other places are working in various things.
Yes. I guess I was interested in any potential way of, or novel way in particular, of generating laser amplification in particularly shorter wavelengths at that time, and there had been this interesting work that had been done on mercury ions, and —
Earl Bell's work?
Earl Bell, right. I forget whether Arnold Bloom was involved in that or not. But anyway, I guess people at Spectra Physics had done something which Hg+ and I have a vague recollection that someone at RCA might have done something too, but I wasn't interested in just doing what someone else had done, and I did get interested in seeing what might happen in noble gas ion systems. I also had some interesting neutral transitions that I thought might oscillate in argon, very high line transitions, things that had been worked on particularly by my colleagues that I left behind at Bell. It seemed clear that Faust, McFarlane and Patel were cleaning up the middle and infra-red region with apparatus and programs that I'd helped launch them on that, but there were a few things that I had wanted to investigate, in particular argon. I remember, there were some high line transitions that I was trying to excite there, and it also occurred to me that because it had such a large ionization cross-section, that we might be able to do something interesting in the excited states of argon ions, but I wasn't sure what was going to happen when we tried these things out, but I did build a fairly versatile laser in which we could try very high energy capacitance pulses to discharge. It was something like 3 meters in length, maybe 2 meters long, and we had external electrodes on it which we then made some experiments with charging thick co- axial cables to produce very potent fast discharge pulses and such, and we tried this out. Initially I was hoping to excite some lines around 1.3 microns I think, and I started it up, and the set of mirrors I had had a double peak. One of them was in the blue part of the spectrum, the other was down around 1.3 microns, and the system immediately oscillated on the blue lines, of ionized argon, 4 p to 4s transitions. And I very quickly followed that up, and that was a situation where we found the gain was so high that we could get superradiant emission, and we didn't need the mirrors, and we could get a beam of light that looked like a laser coming out for the end, which we could shine down the hall and so forth, and we very quickly made fairly precise measurements of the wavelengths in that system. I was doing that largely with — it was John Knutson, Nield Mercer and Lewis Dash were the people who, were the students I was working with. Knutson had started working with me as an undergraduate, and I guess by this time he must have graduated and was in graduate school.
There's one thing I didn't understand. You said you were looking for shorter wavelengths, higher frequency lines, and we were also talking about the Navy interest in the bluegreen lines. Was this relevant? Why were you looking for high frequency lines?
Oh, I was interested in it mainly because no one had done it, really, and there was —
Because these guys were going in the other direction?
Right. It was easier to make long wavelength lasers than short wave ones, once you think about the equations involved in the gain, the Doppler broadened line versus the excitation rates and so on. You could see that the inherent gain you could get out of one of these systems with the same excitation process would be something that increases roughly as the wavelength squared, so it was easier to do it in long wavelengths than short wavelengths, and also the fact that they were cleaning up the field. I did actually build a large wide diameter laser system that I was going to use at one point to try to look at lines out to maybe 100 microns. A fellow that I had taken on to help out one summer with the research managed to smash it, and that sort of ended that direction for a while. And I was curious to see what one could do in the shorter wavelength area, so I did get some mirrors that would work at shorter wavelengths.
So really there sort of was a gap, a noticeable gap, when you didn't have the spectrum filled.
Well, certainly the CW lasers, and I guess there had been some pulsed molecular lasers that pulsed nitrogen that had been observed in transient oscillation by L.E.S. Mathias and J.T. Parker, I think, in the United Kingdom. But there hadn't any work that I recall that involved the possibility of CW oscillations in shorter wavelengths, things less than red. It's true that Rigden and White had gotten oscillation on the helium neon system in various shades of orange, by putting a prism in the cavity and selecting out different wavelengths of neon. But I did have some thought at that time that maybe the green lines in the neon might oscillate, but it looked like it was hard. But it seemed like a sensible direction to go was in the excited states of ion, in particular noble gas ions, and I also worried about some high line states.
So this is then independent of and parallel to Bridges.
It was absolutely independent and totally unaware of anything Bridges was doing. I got the argon ion laser to oscillate, and we quickly identified the lines, and narrowed them down to something like half a dozen or so that ran, and we were getting close to a kilowatt pulse power coming out in little short pulses. I had an idea of how these were being excited, which I think is actually how they were being excited, that you could explain it all on the basis of the sudden approximation of quantum mechanics, that you imagine pulling an electron out of the ground state configuration of argon and doing it fast enough so that the remaining electrons don't have a chance to adjust to a normal Eigen state, and then say, where you expand what's left, what you'd likely get is a peak configuration of ground state, you're likely to get excited peak configurations, and one of the most important ones is the upper state of the argon ion. And that seemed to make a great deal of sense, and we got it oscillating and made these measurements, and wrote up — we were essentially working day and night. We figured it was going to be a hot topic. And we were just working around the clock for several days, and we wrote the paper up, and I sent it by special delivery to APPLIED PHYSICS LETTERS in Oak Ridge. Apparently that was a drastic mistake, because they had no way of dealing with special delivery packages that arrived down there, and the thing got lost in the local mail in Oak Ridge, and was sort of floating around for quite a while, and in the meantime, I gave a colloquium at RCA in Princeton laboratories, and I was used to talking clearly, not being cagy about things, so I discussed our results with argon, and after that colloquium, Hank Gerritsen or somebody came up and said, "Have you seen this? It just came in the mail." It was an issue of APPLIED PHYSICS LETTERS with Bill Bridges' article on the same topic. So at that point, I then got in touch with the editor, and by then he had found my Letter. He said he didn't know what to do with it, because Bridges had already published the initial disclosure. He said, "We'll give you the same date of receipt on it, and put in some kind of a footnote too, but on the other hand, you ought to acknowledge Bridges' work since it's just been published." So I then rewrote the article. By then we'd gotten a lot more data. We'd gotten quasi-CW oscillation on one of the lines, with powers on the order of 10 watts or so, and we rewrote it emphasizing the quasi-CW part of it, and various other aspects of it, and then it got published, but it was after Bridges' paper came out. I also mentioned the results in the Washington meeting of APS that spring. Originally I had planned to give a different talk there on cascade laser transitions in neon, and I discovered that several different people had done the work that we had published in our abstract on cascade laser transitions of neon, and I then decided, well, I made a brief acknowledgement of that, in fact showed a slide of the different references of work that we were about to discuss, so instead of that, what I'll do is talk about our recent work with argon ion lasers, after which Gene Gordon got up and said, made some typical caustic remark, that said, "Well, we've been getting CW oscillation in argon too" and so forth. But anyway, we did not know what was going on at Hughes at all, and certainly not Convair.
Convair, was that? Oh, you mean Convert, the French group.
They didn't know it was argon in their original publication. They thought it was mercury. I guess both Bridges and the French group had been using argon as a buffer gas to excite mercury vapor ions, and there was an actual tendency for someone who didn't have a good spectrometer to think that it was the mercury rather than the argon line which was a mistake we didn't make. First of all, we didn't have any mercury in the system, which was very carefully baked and purified argon sample used, but I didn't know about those other things until after we had already sent off our first Letter. Somewhere around I have the original version of the paper. I was trying to find it last night and looked. But I did send copies of it off to various other people like Willis Lamb and Walsh at IDA and our Air Force monitors, Marshall Harrington and Lloyd Wood. But it was —
That's something sociologists really have to take note of, or historians. When they read papers they tend to take them at first value, but here's a paper whose content is really very much influenced, or its emphasis is very much influenced by publications.
That's right. If I'd sent it to some other journal, I suppose it would have been published in its original form, but since the APPLIED PHYSICS LETTERS had also published Bridges, they felt compelled to tell me that. It brings certain philosophical questions to mind, like, is there any point in two different people doing somewhat overlapping work? I think basically the answer is yes, because there's always different information.
Really three, in this case, because of the French group.
Yes. I think essentially what the French did there was not very insightful. They didn't even measure the wavelength. But on the other hand, Bill Bridges certainly did some fine work, and his colleagues of course followed it up. I guess he and G.I. Gordon also collaborated on the CW aspect of it. But essentially we had looked at both domains, within those two papers, one, the very short pulse and superradiant emission with very high peak powers of a kilowatt or so, and the other was quasi-CW emission, which we achieved by using thermionic re-heated cathodes in the discharge and very large capacitances that would discharge over a period of a millisecond or more, so that you could see that there was a fairly continuous oscillation in that period of time. We didn't have the facilities to make truly CW sustained discharges at that point, such as G. I. Gordon and his colleagues at Bell Labs put together.
There's another thing I didn't understand. When you were going at these shorter wavelengths, is there a big theoretical component of just figuring out, calculating what will give you shorter wavelengths at this point? Or is it more a matter of intuition?
I think in this case it was largely intuition. In most of these cases, nothing happens in exactly the way you expect it. In fact, you're lucky if it happens even remotely in one of the ways that you think. But there is an interesting problem in unscrambling what has just happened, that you've been presented with. One can certainly over-do the notion that you can work these things out a priori. But a useful combination of insight and experimental results, I think, is the best way to go. I mean, there were some other systems I tried at that time that didn't work. We also immediately tried neon. We thought the same thing was going to happen with neon ions. And we tried very hard to get a line around 3300 angstroms to oscillate there, and we just couldn't, and I think that was an example of a Brewster angle window that just had too much loss in it. We were using Brewster angle windows then.
So argon wasn't the first of these things you put in, just the successful one?
Argon was essentially the first, but after argon worked I immediately tried neon, thinking exactly the same thing would happen with neon, and it didn't for us but it did for Bill Bridges. He may have had heftier (?) banks to fire or something. We were —
I don't know whether he was using Brewster windows.
Oh, I'm sure he must have been. But the Brewster windows I had at that point were very thick. They were about a half inch thick quartz, and they had been sealed onto a quartz bell by our glass blower, and there obviously were strain patterns in the middle which (crosstalk ) ... the loss. The actual discharge sources we were using at that time were neon sign transformers, to fire, to drive the coaxial cables, and we were burning them out one after another. I think I had a stack of burnt out neon sign transformers outside my lab after that brief period that must have been five feet deep. We were getting them at a special rate from a local neon sign company, on Norway (?) Avenue. We didn't have the money to get a really good supply at that point. We did eventually get one later, but of course, that was a different time.
There is a comment that I asked, which is question 5 on page 2, that just I would be interested in your —
"In one article, you point out that an Hg+ is a more interesting scientific problem than Ar+." I don't remember. What article was that? Can you tell me?
Let me see if I can remember. I think it was one of those comments on atomic and molecular physics articles.
Oh, the metal vapor laser article?
Where you're saying that the excitation mechanism of Hg+is much more puzzling —
Oh, Hg+ is more — oh yes, now I understand, OK. There's a very simple way of understanding what happens with these high energy pulse electron impact excitations of argon. As I say, the sudden perturbation method works reasonably. You pull off an electron from the ground state shell, and that expands what's left, the excited states of the ion, and lo and behold, you get all the states that we got and that Bill Bridges got and many other people have gotten. But now when you want to ask what happens with Hg+, it was a little more subtle. There wasn't just a question of pulling out the outer electron and seeing what happens, but I think it was a question of inner shell excitation, of exciting one of the D shells' electrons and then expanding what's left, and Eigenstates of the ion. You don't have that article handy so I could just check what I was saying about that?
Yes, let me put this on pause ....
All right, let me read this. Let's see now, what was the question again?
Well, I was trying to provoke you. It's this question, there.
About the interest in Hg+, may be a more interesting problem than the Ar+. At the time I wrote that article, it wasn't entirely clear what was going on in Hg+, and since then actually Howard Hyman, whom I also mentioned in that article, he was one of — a graduate student whom I shared with Howard Hertzenberg — we did a fairly careful study of excitation of ions, both in cadmium, mercury and zinc, and what he concluded was that there was a very strong interaction with inner D shell electrons, which then would leave a configuration with a hole in the D shell, which would then have to be interpreted rather differently in terms of the expansion of what's left in the Eigenstates of the system, but it was my impression that the detailed calculations that Hyman did, including the effects of ionization of the D shell, inner D shell electrons, and also configuration fixing in the higher states, made good sense in terms of explaining what was going on. There's another interesting problem in many of the transitions in mercury, cadmium, zinc, in that important laser transitions are often optically forbidden, and could be explained only on the basis of configuration interaction, which should be along the lines of this here.
It's a more complicated problem, and therefore maybe more interesting, but probably more interesting wasn't a good term in that paper. It was only in the sense that it hadn't really been worked out as carefully as argon had.
Well, let's talk a little bit about those metal vapor lasers, because that's almost the only thing we haven't touched on.
Did you take that up because TRG proposed that as a consulting problem or what was happening?
I think that probably is true. I'm just trying to remember.
See, I don't know much about that. I know University of Utah was working in —
Yes, Grant Fowles had a couple of students, one of them was (William) Silvfast, another was Russ Jensen, interested in those problems. Let's see, there are probably at least two reasons why I did get involved. One of them — I had interesting charge exchange for a long time, as a source of excitation in the laser systems, and in some discussion and correspondence I had with Ali back in spring of 1959, I had mentioned charge exchange having interesting possibilities, and that we might look into some things there, and of course I worried about charge exchange excitations in my thesis at Columbia, and whether I could use that for gas (?) counter purposes and that sort of thing, and the notion of the possibility of charge exchange... (off tape) ... that Fowles and a couple of students were working on metal vapor ion laser systems, and I wondered if some of those might be ones that would be excited on a continuous basis by charge exchange with things like (?) ions and so on. I stopped at the University of Utah — well, actually, I think, yes, I was consulting for the National Bureau of Standards in Boulder one time, and I had gotten a letter from, I think it was Russ Jensen, one of Fowles' students at Utah, and on one trip I made, I think at Boulder I decided that I would try to go out and visit Jensen and Fowles and his students and see what they were doing, and I did talk to them, and Jensen in particular seemed interested in a job as a post-doc with me at Yale, and I was particularly interested in his work and what he was doing, and his thesis actually had something to do with charge exchange in an iodine laser transition, and I remember mentioning to him that I thought maybe there were some possibilities of using charge exchange to get CW oscillation in the zinc ion, which was something he had worked on, and ultimately the cadmium ion, which I guess both he and Silvfast must have worked on, I forget now who worked on what. But in any case, Jensen did come to work with me at Yale on these things, and we got some support to pursue those studies, and ultimately we did get, we did both prove that charge exchange was a dominant reaction in a couple of those systems, but also got CW oscillation in both zinc and cadmium ions, using charge exchange. It was a kind of — I made a proposal for the Air Force, that one might be able to get charge exchange, say with Hg+in something like zinc, in which an inner core electron in the zinc was grabbed by the D shell and left the ion in an excited state which then suggested it might be possible to get continuous laser oscillation out of some of those cases. As a matter of fact, that was a good example, I think, of one of the things you were talking about before. As I recall, although my memory is a little rusty on this, that the Air Force had wanted to support work on ultraviolet lasers, and I had done some thinking about the possibility of charge exchange in metal vapor systems, from the point of view of getting an ultraviolet laser, and that funded the work that Jensen did with me at Yale, which ultimately resulted in getting CW oscillation as a result of charge exchange excitation in both the systems. However, not in the ultraviolet. We got the systems to work in the short wavelengths in the visible, but we didn't get the ultraviolet ones to work, and about that time, Silvfast, who had gotten to Bell Laboratories, was working on related problems, especially in cadmium, which he interpreted in terms of Penning ionization of cadmium. Indeed, they seemed to be two different kinds of processes going on in the case of helium cadmium mixtures. There were some levels that were clearly being excited by charge exchange, some that were being excited probably by Penning ionization, so that — and I think it also turned out that the Penning process was the one that was responsible for the ultraviolet transition that's so well known in helium cadmium lasers.
But how did TRG?
Now, TRG was a different matter. When I agreed to consult for TRG, there were essentially two different kinds of problems they wanted me to work on. The first one that came up was the frequency stabilization question which I mentioned before. Then there was also, in the agreement, there was a statement that they would also like to have me work on collision lasers, for which one or two patents might be applied. At least I think that was pretty much an exact quote of the agreement I had with them. And there was a certain question in my mind as to what they meant by collision lasers. I didn't really have a clear discussion with what was in Gordon Gould's mind at that point, and I began thinking about it over the following year or so, and in particular, I assumed that what they must mean by collision lasers was lasers in which the relaxation of the lower state was caused by collision process rather than relaxation because it seemed like that would lead to a much higher powered laser in the long run. So as a result of thoughts in that direction, there actually were two patents that ultimately came about. One of them was a patent on collision lasers. I had worked out some thoughts about that, particularly three level systems which I described in a paper at the Lajolla Conference, and called them inversion mechanisms in gas lasers. It's a possible direction to go to get higher power out of gas systems, using collision depopulation of the lower laser level, and I knew that Gordon Gould had been interested in that kind of thing, or at least I'd heard that he had been, and I invited him to give a paper at La Jolla on the same topic, and we ultimately pooled our thoughts on that, and it turned into the one patent. So after it was put, explained to TRG, to try to develop a CW collision laser using various metal vapor systems — Gordon was thinking largely of metal vapors in terms of collision depopulation process, and I'd been thinking of various things — and the metal vapor systems, he was proposing and pushing at TRG, and people were working on and I was trying to help understand why they weren't working and so forth, the systems that were involved there also looked to me like they might be capable of transient pulsed oscillation with a very high efficiency, and I began worrying about transient re-inverted lasers, actually, after I had reviewed a paper by Elliott, (?) and Parker, that they had published an interesting paper. Actually they sent a paper to APPLIED PHYSICS LETTERS around July or June of 1963, in which they listed large numbers of lines that they'd made oscillate which terminated in the A triplet sigma level of nitrogen, which is inherently metastable, and it therefore meant that anything they were looking at was inherently transient in character and not CW. A lot of the lines looked like they were doubtful in terms of actual oscillation requirements, and I questioned what their criterion was for deciding whether they oscillated, and indeed the manuscript came back with half of the lines deleted. They had gone back and looked over the problem and concluded that half of them weren't really lasers. But nevertheless, there were a substantial number that still were on the list, and they were all transient inverted, and I began thinking, well, that was obviously a potent system, that they got fairly high peak power gains. There's a guy named Harry Heard who originally looked at this kind of thing, in a very sloppy manner actually, and published some stuff in the TS Record (?), and one could do it more carefully, but anyway, it looked very clear to me that what they had was a system that produced very high peak powers, for short time intervals, essentially the lifetime of the upper level or thereabouts, and that such a system might have great efficiency, if the set of levels was down near the ground state of the atom or molecule. And it occurred to me at some point that the atoms, the metal vapor atoms that the TRG people were trying to get to oscillate CW, some of them might also work better in other levels in the atoms, as one of these, I think we called it cyclic laser things.
Was TRG independent of Utah?
Oh yes, they were totally independent. Yes.
So this was also parallel.
Yes, it was a parallel thing, although I guess there was some cross-fertilization which may have been involved, perhaps inadvertently, but I've forgotten now whether Fowles actually came to the La Jolla Conference or not. He may have. I know that a number of TRG people did. At least Gordon came and gave a paper, and I think several of the other TRG people were there, and I remember, in the summer of 1964, during that conference, thinking that some of the systems that they had built apparatus to handle might be worth pulsing, instead of just firing CW, that we ought to try pulsing them. And I came back to Yale in the fall and told TRG, "Why don't you just go ahead and pulse these things, try it out and see what happens?" And both Paul Rabinowitz — let's see, no, it was Bill Walter, I guess who was involved in that to a large extent, and Marty Pilch and I forget exactly who else was involved, but maybe Paul Rabinowitz, I'm not sure, but certainly Bill Walter, and we discussed this possibility, and I suggested that, at one of the meetings I had there, that they ought to go ahead and do that, and Gordon kept saying, "No, let's not do that, let's go through the CW collision laser and not interrupt our research," and they kept putting it off, although —
Was Gordon more or less directing the research there? Was he setting the directions?
Yes, he was. He wasn't really doing much in the lab, though. He would sort of sit at his desk and give directions. I don't know whether that was because of the classification problem or something else, but —
— but by that time, he must have —
— it must have been over by then. Anyway, he wasn't really active in the research at all, actually taking data or anything. And in fact, on that occasion, he was actually hampering the process. He kept saying, "Don't distract us by doing that." And they didn't do anything on it, and I think it was the following spring, I made that trip out to Utah. I was consulting at Boulder for the NBS, I guess. And another group there. And I went out to Utah to visit Russ Jensen and Grant Fowles and Silvfast, and they mentioned, they had gotten lead to oscillate, and I mentioned that result the next time I bumped into the TRG people, and said, "You know, you should have taken my advice and started pulsing these metal lasers." Apparently at that time, Gordon Gould was on a vacation in Puerto Rico and wasn't around to say, don't do it, and Bill Walter and I think Marty Pilch and I forget who else might have been involved went ahead and set the thing up. They decided they would try lead out, because that had already been demonstrated, and they put lead in the apparatus and that worked, and then they went on to the things I'd suggested before like manganese and copper, and that worked very well. They were getting large pulses out in various parts of the spectrum and I helped them to identify the transitions that were involved, and they turned out to be the kind of transitions where you have a strongly involved electron transition in the ground state which also gave an extremely potent electron excitation cross-section, plus a strong transition to a metastable state, so you end up getting a transient inversion on the transition to the metastable, and all this was very near the ground state on some of those systems, and hence you get efficiencies that were on the order of several percent, compared to 10 to the minus 5 or so on helium neon. So that worked out pretty well, and that lead to the copper vapor laser that Bill Water followed himself, and ultimately started getting 50 kilowatt pulses out of, at about the time ONR terminated their contract. These were bluegreen pulses, which I would have thought the Navy was interested in. Anyway, that was essentially the way it worked, and I think there may be some hard feelings on the part of Grant Fowles, in that I conveyed his result on lead to the TRG people, but that was just a free exchange of information as far as I was concerned. I was talking freely about what I was doing, and I assumed it was fair for them to talk freely about what they were doing. But in any case, I'm sure that Fowles and Silvfast did the lead work first. It wasn't something that suggested that we ought to do the others, but it reminded me that I had suggested a long time ago that it seemed very timely, that the other things, in particular magnesium, copper and so forth would be important things to do. Anyway, that was the background on that. And that ultimately resulted in another patent which both Gordon and I were co-authors on, on the cyclic pulsed metal vapor laser systems.