John A. Hoffnagle (left) and C. Michael Jefferson (right) at the IBM Almaden Research Center in San Jose, CA.
Each year, the Rudolf Kinglsake Medal and Prize honors the most noteworthy original paper in the SPIE journal Optical Engineering. Recipients of the 2003 award are C. Michael Jefferson and John A. Hoffnagle of IBM Almaden Research Center (San Jose, CA) for their paper "Beam shaping with a plano-aspheric lens pair" (November 2003). The PDF file of the paper can be downloaded at no charge.
Jefferson received a PhD in physics from the University of California, Santa Barbara, in 1984. He has been a member of the research staff at the IBM Almaden Research Center since 1985. He is an avid rock climber and sailboat racer, and has completed three solo trans-Pacific races from San Francisco, CA, to Kauai, HI.
Hoffnagle received a Doctor of Natural Sciences degree from Eidgenoessische Technische Hochschule (ETH) Zurich (Swiss Federal Institute of Technology; Zurich, Switzerland) in 1982. From 1982 through 1985 he was a research assistant at ETH, and since 1985 he has been with IBM Almaden Research Center. He is married with two small children, who take up much of his free time. He enjoys searching through used bookstores for uncommon editions by a handful of favorite authors, especially Vladimir Nabokov.
Tim Lamkins and Rich Donnelly recently interviewed Jefferson and Hoffnagle for oemagazine. This an extended version of this interview that was published in oemagazine.
SPIE: Please tell me how you first became interested in science, and optics in particular.
Hoffnagle: When I was eight years old or so, a friend of my father's, an engineer at the Jet Propulsion Laboratory in Pasadena, paid us a visit and described some of the projects that he was working on. These involved probes for planetary exploration and of course he had wonderful photographs from JPL of spacecraft launches and of the probes themselves. I immediately wanted to learn more about space and science generally. Since then my interests have changed, but I remember that encounter with space exploration as the starting point.
As for optics specifically, I didn't have any more than the usual introduction to the subject as a special case of E&M until I began graduate school and chose to work on an experiment looking for parity violation in atomic cesium. Step one was to build a jet stream dye laser to excite the extremely forbidden 6S-7S transition in Cs. That beast took over my life for years, and I have been working with lasers in one way or another ever since.
Jefferson: I was turned on to physics as a child by Sputnik and the great importance that became attached to the role of science in the nation's interest. I was (at that time), a rather isolated "nerdish" kid, with thick glasses, no athletic ability (that later changed), and perhaps, more interest in "academic" things than most of my peers. The social rejection I received from other kids for having these interests somewhat polarized me, and focused me even more tightly on having a scientific career. As I grew up, got contacts, learned to socialize with people, became a competitive athlete (swimming), and so on, the original (negative) forces driving me towards physics diminished, to be replaced by a burning passion to solve problems and to build things. I became interested in optics during my PhD thesis, where I used HeNe lasers to measure scattered light from critical systems near their critical points. After grad school, I went to work at the IBM Almaden Research Center as a Research Staff Member, and after a few years working on Magnetic Recording, started a research effort on Time Domain Spectral Hole-Burning. This effort used dye lasers, argon lasers, and many optical techniques for manipulating, controlling, and measuring laser light, and during this time I really got some experience in optical design issues.
SPIE: Using a pair of plano-aspheric lenses, you've shown that you can convert a Gaussian beam into a radially-symmetric flat-top beam over a wide spectrum of wavelengths. What applications and interests drew you to investigate this problem, and what drew you to aspheres as opposed to GRIN lenses or other possible systems for a solution?
Hoffnagle: Several years ago I was working on two separate projects at IBM that both seemed to cry out for beam shaping, and the combination of requirements probably had some impact on the final implementation of our beam shaping optics. One project was holographic data storage, in which binary data was encoded optically by illuminating a spatial light modulator (SLM) with 532 nm light. For best system performance all the pixels of the SLM should be illuminated uniformly, and at the same time it is important to use the available laser power as efficiently as possible, since that directly affects the achievable data rate. At the same time, I was starting work on a project to use deep ultraviolet interferometric lithography to investigate the limits to which one can push the spatial resolution of modern photoresists. Because the photoresist response is by design extremely nonlinear, uniform illumination is required if you want to get uniform feature sizes. We envisioned exposing some fairly large areas, the output of my frequency-doubled argon-ion laser was limited, so once again beam shaping was the obvious solution to squeeze the most performance from the tools at hand. It is interesting that both projects involved optical interference, so that achieving a uniform wavefront was as important as getting uniform irradiance.
It took a while to arrive at the design that is described in our paper. I first spent some time designing radial-gradient absorptive elements based on neutral-density filter glass, but these are very lossy and in any case it was hard to see how to implement that idea in a practical way in the deep UV. The idea of using GRIN materials didn't occur to us at the time since then I have learned of the work of David Shealy and colleagues on GRIN lenses for beam shaping but here, too, I imagine there would be a materials problem in the ultraviolet. Eventually I came across the 1980 Applied Optics paper by Patrick Rhodes and David Shealy, describing an aspheric lens pair, which immediately struck me as a beautiful solution. I played around with some numbers and discovered that a single prescription would work at both 532 nm and 257 nm simply by adjusting the lens spacing, which made this approach look even better. There were some details, though, which we felt had not been satisfactorily addressed by Rhodes and Shealy, or in the earlier works of Frieden and Kreuzer. One point was that we wanted to produce a beam that would propagate for a meter or so without being appreciably distorted by diffraction, which led to the conviction that the output profile had to be designed with a precisely defined roll-off, not a "top-hat" profile. Concern about diffraction also led us to extend the lens design well beyond the 1/e^2 beam diameter where earlier designs were truncated. At this time we also learned about MRF and started talking with Don Golini and Paul Dumas at QED about the practical matters that distinguish a real, physical artifact from a solution of an integral equation. We designed the lenses to be convex, which I believe was crucial to making them manufacturable, discussed what dimensions would be practical, tweaked the design to keep the asphericity of the surfaces within reason, and finally emerged with prototypes that did what the equations said they should do.
Jefferson: The Gaussian to Flattop beam reshaper arose from the IBM effort to develop digital holographic storage techniques. We had a team of about 6 PhD researchers, all of whom were old laser jocks of one sort or another. We naturally divided up the work according to our individual backgrounds and interest spectrums, and had at it. My role in the team was as the designer of the optical systems we built as Holographic Storage Engines. Around that time I found the Zemax lens design software program, and taught myself to design real optics with it. As the project progressed, it became apparent that we really needed to illuminate our Spatial Light Modulators (SLM) with a uniform intensity. THe errors in the reconstructed digital holograms tended to cluster in the corners, as these were the most dimly illuminated, and also because the lens system aberrations were the worst there. The lens designs could not be improved within our budget, but my colleague John Hoffnagle thought that perhaps it would be possible to reshape the laser beam intensity to help flatten the profile. John is a scholar, while I am more of a photon mechanic. Our skills are almost exactly complementary. He studied the history and literature of the problem, and identified a concave-convex reshaper made with aspheric surfaces, originally proposed by Kreuzer many years before, as the best of the possibilities in the literature. We studied this design, and tried to find a vendor to build one. We soon came to the conclusion that it was both poorly suited to our needs and impossible to build in the size we needed and with our budget. This is where John and I re-invented the refractive beam shaper. We identified three important things which were lacking in the Kreuzer design, and we found a way to incorporate them into a novel new design. These were that essentially ALL of the laser power must pass through the reshaping optics without truncation by a hard aperture, which causes diffraction; that the rolloff of the intensity at the edge of the "flat top" output profile be done in such a manner to minimize diffraction effects and to allow the beam to propagate well(i.e. a "Flattop"is a BAD profile), and lastly, that the shape of the two lenses both be convex. This last criterion is really a key element of the success of the optics, both from theoretical and practical perspectives, as John has shown with some elegant mathematics. By making the surfaces both convex, they become monotonic, with no re-entrant curvature. This makes it possible to polish them in a practical and cost effective manner, using Magneto-Rheological finishing for example. John had re-written Kreuzer's equations and added his own twists, and we picked a design point for the reshaper based on the needs of the Holographic Storage Engine which we were constructing. We chose to use 99.7% of the laser power (because e6 is a nice round number...), and chose the rolloff profile to be the Fermi-Dirac function from solid state physics because we were physicists, and had never heard of Super-Gaussians. Anyway, we wrote down the reshaping mapping, and while John solved it analytically, I used Zemax to derive the surfaces. The two approaches came out essentially the same. We were very fortunate to meet Harvey Pollicove, who was running a program for encouraging the manufacturing of aspheric lenses, and he introduced us to Don Golini of QED Technologies, who was developing a commercial MRF tool. Don found our design for a pair of aspheric lenses to be an interesting challenge, which seems to have been helpful in the development of the tools they sell today. He and his colleagues gave us a great deal of feedback about the practicalities of manufacturing different designs for the reshaper, and their input taught us a lot about how to make such a design practical.
SPIE: Given the low cost of fused silica and the wide array of uses for illumination with a flat-top profile, do you see this as a boon to MRF in the mass production of relatively inexpensive aspheres for beam shapers, or is the needed figure error too restrictive to insure negligible imperfections in uniformity and phase without high expense in manufacturing?
Jefferson: It is hard to see at this time the impact that the reshaper may have on both laser technology and the development of aspheres as common optical elements. I think it has been, and will be fairly substantial. We were very lucky to have the right problem at the right time, and to find the right people to build it. Aspheres used to be basically inaccessible to the regular scientific customer, as the figuring and measuring were hugely expensive. MRF has had a big impact on this, and the precision and thru-put of the tools is increasing rapidly as CNC techniques and advanced metrology become better developed. John and I identified many applications which a commercially available and affordable reshaper technology would have an impact upon. Although the surface figure precision is very stringent relative to a spherical optic, we have seen that it is well within the state of the art at the moment to mass produce these devices at an affordable price. We have licensed the technology to Newport Corporation, which is now selling off the shelf reshapers for not much more than any other piece of precision optics. As researchers use these tools in such applications as laser rod pumping, flow cytometry, and other endeavors which really benifit from flat intensity profiles, the use of the tool will blossom. The current Newport version is essentially the same as the one that John and I designed and built for our (obsure) requirements. It is very easy to redesign the surfaces, and in the next few years we expect to see new, application specific design points appear. This is a commercially viable technology which has become so due to the huge advances in the state-of-the-art in lens manufacture which have occurred in the last 8 years or so. There really are no technical barriers to practical use of this technique on a wide scale. One caveat: this is a precision remapping of rays from the input to the output of the lens system. It is very much a Garbage-in, Garbage-Out situation. The input beam (which need not be Gaussian) must accurately match the designed transformation profile if the desired output profile is to be obtained. These systems are sensite to alignment, but experience has shown that these sensitivites are manageable and practical. Beam reshapers are precision tools, not onions, and if treated with appropriate respect for the issues will perform well.
Hoffnagle: I am sure there are many other applications for this kind of beam shaping. (Incidently, the lens design equations are general enough to accomodate other output profiles than flat-topped ones, which may be of some practical value.) Of course, different applications may place different requirements on the accuracy of the output profile and the cost of the optics. Generating a really uniform output beam does require accurately figured surfaces; generating a uniform beam that propagates well is more demanding still. Our experience shows that figuring these surfaces is a tractable job. It is also worth noting that there is a lot of ongoing work devoted to the manufacturing and testing of aspheric surfaces. I think it is reasonable to expect that fabricating optics like these will become more routine in the future.
SPIE: You have done extensive research in holographic data storage, receiving an IBM Research Award for your contributions. Can you describe how you have utilized this beam-shaping method in encoding data?
Jefferson: As mentioned above, the original use of the reshaper was to create a flat illumination profile for our Spatial Light Modulators. This was a very successful use of them, and although vast amounts of work went into the system issues, coding of the data, error correction, and so on, these sophisticated strategies for data reconstruction would have been (as other research groups have found) partly defeated by the non-uniform profile of a Gaussian beam. The strategy of expanding the Gaussian laser beam and only using the central region suffers from the fatal flaw , as John showed analytically, that the power thruput is about equal to the flatness error. Thus a flatness of 5% allows only about 5% of the laser beam power to be used. The Flat top region of one of our reshapers is around 70% power thruput. The lost power means a big decrease in signal to noise ratio, which in a real storage system is always basically on the edge anyway. The only way to compensate for lowered SLM illumination intensity is to reduce the data density, a VERY undesireable compromise.
As the Holgraphic Storage Project came to a close, John became involved with Deep UV lithography, and used a reshaper to flatten a 257 nm laser beam, setting a world record for grating pitch in the process. Photoresists are very non-linear in their response to light, and to illuminate an area of several square mm or more uniformly is very difficult. The reshaper does the job nicely. We found out an amazing thing about our design: If implemented in silica, one prescription is usable with virtually no errors other than a need to recollimate the output beam (easily done by changing the spacing between the lenses) from the deep UV to the far IR. John has published the elegant proof of the reasons for this, but it makes the commercial practicality far greater, as a single prescription and appropriate coatings allow use at many wavelengths, and over substantial spectral bandwidths (narrow pulse lasers). We have also found efficient ways to achromatize the reshaper over a 2:1 range of wavelengths, allowing multi-spectral operation.
Hoffnagle: In the holographic data storage system, it was important to uniformly illuminate the SLM that transformed digital data to optical form. At the same time, we needed to use our laser power as efficiently as possible, since the data rate depended on having enough photons to get acceptable signal to noise. Shaping the beam gave us the uniformity that we needed at the least cost in terms of wasted laser output.
SPIE: What do you consider to be some of your most important professional achievements?
Hoffnagle: Our work on holographic data storage is one project that I am proud of, on its own merits as well as because it led to the beam shaping work, which has been such an enormous thrill to me. I would also mention liquid immersion interferometric lithography. About 5 years ago it occurred to me that we could take advantage of the wavelength reduction of light in glass to reduce the pitch of interferometrically exposed gratings. It turns out that this idea has been invented many times in the past hundred years, but even so the results with deep UV illumination and modern photoresist were striking. With a 257 nm laser we wrote gratings of 45 nm lines and spaces in commercial photoresist. Even today, that is close to the limit of what can be done using optical lithography. Going further back, my thesis and the follow-on work I did in Zurich represents some really high sensitivity spectroscopy.
Jefferson: My PhD thesis was one of the early studies of highly non-linear, stochastic thermodynamic behaviour. It was (as all PhD theses should be) pretty state-of-the-art at the time, and tied some exacting measurements (me) to some very difficult and novel calculations of fully non-linear behaviour (Rolf Petschek). Non-linear dynamics was a hot topic at the time, and I am proud of the work I did on this subject. It was real science.
Much of my professional career at IBM has been in the instrument building field, but I have made a few contributions of which I am proud. I built a device to study the effects of magnetic recording at very small spacings by developing a servo-controlled actuator system which would position a magnetic recording hear 20 nm above a disc moving at a speed of 10 m/sec, and moving vertically as much as 25 microns. It was an interesting project...
I built the Monticello Teststand, which was a precision magnetic recording test system, also used extensively for magneto-optical recording development. This teststand was quite novel at the time, and was widely used within IBM as a key development tool.
I spent many years working on the use of TIme Domain Spectral Hole-burning as a data storage and processing technique. In collaboration with several excellent postdocs, researchers at MSU Bozeman, and my colleagues here at IBM, we actually measured the effect of phase shifting reference and data pulses on the stored ground state gratings, discovered that the Stark Effect could be used to selectively suppress photon echoes, made the first demonstration, with W. Randall Babbitt, of a continuous real time correlator using TDHB techniques, discovered chirp nutation, and elucidated the relationship of optical frequency chirped pump signals to the resulting excitation profile in the excited state.
As part of the Holographic Storage Team we set many world records for density of stored and recalled date, created the paradigm for bit error rate still widely used today, developed both optical technique and data analysis for reading and recalling high density digital data. This program was highly successful in understanding holographic storage materials and in using them to store and retreive data. This was a team effort, but I am proud of the role that my design skills played in this project.
As previously mentioned, as part of the holographic storage activity, John Hoffnagle and I developed the Beam Reshaper. This is a hugely satisfying accomplishment if for no other reason than we have made a significant contribution to a field worked in by so many talented and knowledgeable professionals. To find a new solution in such a field, especially a solution that is both elegant and practical, will be something that John and I will forever cherish. That project was really FUN! (at least in retrospect...)
SPIE: What interesting projects are you working on now?
Jefferson: I am supporting some deep UV lithography work that John is doing, building optically based testers for studying phase change materials, and starting to figure out how to make some high speed electrical measurements of some specialized structures we are building in a new (unspecified...) project in the lab.
Hoffnagle: Deep UV interferometric lithography still takes up most of my time. Almaden Research Center has some of the best lithographers and photoresist experts anywhere the people who have made modern microelectronics possible and it is a particular pleasure to work with them. We want to extend our capabilities from 257 nm to 193 nm, with liquid immersion at this wavelength as well. I am also trying to better understand how we can quantify the spatial response of the photoresist. A couple years back we took the approach of attributing a modulation transfer function to the photoresist film, essentially treating it on the same footing as one would treat any other element of a precision imaging system. There is probably still more that can be done along those lines.
SPIE: You were both surprised at receiving the Kingslake Award ï¿½ has it changed your outlook or your approach to research in any way?
Hoffnagle: It is a bit soon to talk about a change in outlook, but certainly receiving the award prompted me to take a fresh look at the beam shaping work that we and others have done. The more I think about beam shaping, the more it appears to me as a microcosm of optics and laser physics. There is the geometrical optics that goes into the lens design equations; there are caustic surfaces between the lenses, in our design at least, where geometric optics fails; there are limitations imposed by physical optics; there are issues of beam propagation; and since ultimately we are trying to improve the beam shape for some application, there is the tricky matter of how to describe "beam quality" as quantitatively and objectively as possible.
Jefferson: The Kingslake Award was a great surprise, and an honor I am humbled to receive. I have always viewed myself as sort of a high tech mudwrestler; someone that you throw into the pit when there is something really nasty in there, and who, after a lot of loud noise, howling, banging sounds, and so forth, usually emerges later, tattered and grimy, but with the job done. The Kingslake award is about elegance and exposition, characteristics more likely to be associated with John Hoffnagle than with me. I think that the beam reshaper is elegant, and the exposition of it has been a valuable contribution to the field of beam reshaping generally. So the Kingslake Award is something I am very proud to have received. I received the news about it while I was at sea, sailing between Hawaii and California. You could have knocked me down with a feather. I can't say that it has changed my life much at this point, but I do think about it, and the implications, and feel that it has the effect of making me try just a little harder each day, as if to live up to the expectations that it implies. I think that it may have more effect in the future.
SPIE: Has Rudolf Kingslake's legacy influenced you in your work?
Jefferson: I am not a professional Optics designer by trade. I have learned about Rudolf Kingslake by reading his books on optical design. I have found them very helpful at helping me understand lens design. Since I am not particularly analytical by nature, understanding things in a physical way is very helpful. Many optics texts dive headlong into reams of matematical exposition, which although technically appropriate perhaps, lacks physical insight into what those nasty little photons are thinking as they navigate the various lenses in an optical design. I use CAD tools to design optics. I sometimes describe myself as an idiot with a CAD program. What saves me is that I do have a very good insight into the essence of good design- a feel if you will, for a design that is practial and elegant. Rudolf Kingslake's text books have helped me understand these isues as they relate to optical design, as opposed to merely feeding me mathematical results.
Hoffnagle: I am not a lens designer myself but fortunately my colleague, Mike Jefferson, has studied lens design systematically. Of the two of us, he is the expert on real lens design. (The beam shaping problem is unusual in that the surfaces of the aspheric doublet have an exact, analytic solution that can be expressed as a single quadrature, which is the kind of problem that I can solve, but this is a very special, simple case.) Of course, even as a non-specialist I knew of Kinglake's contributions to lens design and I am deeply honored to have won the award named for him.
SPIE: Talk about your family and what you like to do in your spare time. For example, do you have any hobbies, favorite travel destinations, pets, home projects?
Hoffnagle: I am married with two young children who do their best to fill any spare time that I might have. They do that well. Until last year, for instance, I didn't realize that a 6-inning baseball game played by 7 year-olds takes about as long to complete as a major league game. On the few occasions when I can find the time, I like to hunt through used book stores in search of uncommon editions by any of a handful of favorite authors, Vladimir Nabokov being the main one.
Jefferson: As I mentioned previously, as a child I was rather nerdish, with thick glasses and a bookworm, reclusive personality. Around age 11 I got contact lenses, and they changed my life. I took up competitive swimming, and swam seriously for 30 years. I started rock climbing in nearby Yosemite at age 18, and have been doing it for 36 years. I have climbed El Capitan in Yosemite several times, and done first ascents of big walls in Canada. I did judo seriously for 7 years, but had to stop when I went into grad school. I did triathlon for many years, and completed Ironman Canada twice (1990, 1991). In 1991 I started racing sailboats shorthanded, and have done three SingleHanded TransPac races from San Francisco to Kauai (1992,1996,2000). I have well over 20,000 nautical miles of blue water sailing, mostly single or double handed. I own two sailboats a Yamaha 33 and a Garcia Passoa 47. I wintered over in Antarctica at South Pole Station in 1975-76 as Station Science leader. I play guitar and greatly enjoy building things. I take great delight in operating machine tools in the shop, and often build my own fixtures and mechanisms for research. I really enjoy Telemark skiing, and spend most winters flailing down ridiculous slopes at ski areas and in the back country, screaming and waving my arms in the air. I backpack, drink beer, and like the blues. I try to avoid ingesting vegetables whenever possible.