Oral History Interview with Philip Baumeister (PB) 1929 – 2003
Conducted by Ric Shimshock (RS)

Philip Baumeister

RS: Phil, what is your schooling background? Where did you start your interest in science?

PB: I got my undergraduate degree at Stanford, and then I came back to grad school in 1953 at Berkeley. Professor Francis Jenkins had read about coating Fabry-Perot mirrors with dielectric multilayers. Silver had been used for 50 years before that, and the silver was not very reflective and tarnished. So he asked a graduate student, John Stone, if he would he would coat these mirrors as a Ph.D. thesis project. John Stone clearly understood the theory. There were some papers published, primarily in France, on what we call quarterwave stack reflectors made out of zinc sulfide and cryolite, typically. And John Stone read these, sat down at a drawing table with a T-square triangle. He used vellum paper and a 4H pencil; three months later he had produced all of the detailed drawings for a whole coating tank. The diffusion pump and the mechanical forepump were all that he really bought. I think almost everything else was homemade. All of the electrical feed-throughs were all homemade, and there was a mechanical feed-through that was John’s individual design. So about another two months later the parts all came back from the machine shop, John assembled it and produced these nice coatings for the Fabry-Perot plates. He had a motor running outside the vacuum, and he used one of these spring belts that used to be used like in the old movie projectors to rotate the plates. I think he could coat two plates at a time inside the vacuum tank there. He had an optical monitor system. This included a photomultiplier tube right inside the vacuum tank. You had to be very careful about what pressure you turned on the photomultiplier voltage; it would arc over. But a in a good vacuum, it was fine. The photomultiplier survived being inside a vacuum system. He evaporated zinc sulfide and cryolite layers. Now, John again made his own boats and he knew the trick in the spot welding. He knew that if you want to spot weld tantalum you put a very thin sheet of platinum between the tantala sheets. So when you spot weld it, that kind of helps to bond the two together and maybe overcome any oxide layer of the tantalum. So there was a little spot welder there, and he spot welded his own boats together. Then at the bottom of the boats would be some ceramic compound that sealed them, and I have forgotten what it was. But it would last about two or three runs, then the cryolite would begin eating through something and it would begin dripping; he would replace it. These were very successful, and John’s thesis is still in the library. There was a theoretical section, which John based on admittance theory. Then he showed experimental results. And so he showed these reflectors at a talk at a symposium for the Department. I got interested in it and wanted to know more about it, so that is how I got interested in thin films.

RS: So you chose this thesis area after seeing a presentation from another student?

PB: Yes, I saw John Stone, and I said, this kind of interests me, these thin film coatings. And that was actually when I was a beginning graduate student. But remember my thesis at Berkeley was not on thin films. They would not allow me to do it. So my thesis was on the optical properties of cuprous oxide. So that was a whole other project I had to do. I designed my own helium cryostat and produced all the detailed drawings on the drawing board, and did even some of the welding of the ultra-thin-wall stainless steel, 2 mil thick, that held the liquid helium. To silver solder that was no easy task, but you learn the tricks. And the whole trick was to use a copper cap and a copper bottom and keep the torch just on the copper. Then the thing would heat up and seal in the silver solder where you joined the thing together. So I built that; that took a long time. Then I finally did the measurement of the transmittance of the cuprous oxide at liquid helium temperatures.

RS: So that was reflectance or transmittance?

PB: Transmittance of the cuprous oxide.

RS: You still had an interest in thin films, though.

PB: Yes, and what we did is some of the first calculations in the United States. The places that could do calculations at the time were places that had computers, and computers were pretty rare. Things really changed in 1956 when the IBM model 650 became available, and the Lawrence Berkeley Radiation Laboratory received an IBM 650. So we purchased time on it so we very quickly programmed up the matrix method to do the calculations to compute the reflectance of a multilayer stack.

RS: And that matrix method was the one that people had used, that formalism to characterize the film properties?

PB: Yes that was from John’s thesis and John, I think, probably referenced Abelès, but I am not sure. But at any rate, it was a two-by-two matrix to get the properties of the thin film. So we did the calculations.

RS: Dielectric materials?

PB: This was non-absorbing stack, and we did that on the IBM 650. Then the next thing we thought about doing was the optimization on it. And so that was much more complicated because there was not any hardware floating point on the machine. So I wrote my own floating-point routines to do the matrix manipulation. There are some economies that you can do, like when you are doing a dot product of a row times a column, you simply keep the exponent and the fraction in separate storage locations. Then when you get to the end, you normalize the data. You used to add 50 to the exponent, then put that at the front of the representation of the floating-point number. That was the way you represented floating-point numbers in a 10-decimal digit machine.

RS: Because the early machines really didn’t have that capability of floating point?

PB: Yes, well, eventually they did. They didn’t ever have that on the radiation lab machine, the so-called hardware floating point. But they did offer it eventually. IBM did. Then of course we went to other machines, other than the 650.

RS: How did you start working with some of the companies in the Bay Area looking for optical performance of thin film stacks?

PB: What happened was, I was at Berkeley and John Stone said, we are going to have some visitors. Somehow maybe Rolf Illsley had seen John’s publications or something, perhaps a published paper he had given on his system. And they said, we want to come down and visit you. And so Rolf Illsley showed up and maybe Alfred Thelen was there on one of the visits. Anyway, that was the first time I met Rolf Illsley. So I knew that he was producing thin films in Santa Rosa. He came down to Berkeley.

RS: And they asked you to start working on some of the problems that they were facing?

PB: That happened later. Rolf had an incredible sense of where there was money and he knew that a new missile system — the sidewinder — needed a bandpass filter. To be more accurate, it needed what is now called an edge filter. And they had to have a transmission below one tenth of a percent, then over a certain range; then the transmittance could pop up at about 2.7 micrometers where there is a water band emission. The bandpass filter kept the sidewinder missile from heading toward the sun. Rolf sensed that there was big money in that, and so he got Alfred to evaporate silicone monoxide and germanium to do this bandpass. Now, the problem was it needed a two-stack design. You needed computational ability, and they didn’t have any. So, on a contract basis, I did the computations for them on the sidewinder missile.

RS: So you worked with Alfred on this, then?

PB: I worked with Alfred and Rolf, and I occasionally would drive up there and they’d pay me a consulting fee to drive from Berkeley to Santa Rosa. Sometimes they’d telephone me what they wanted for a bandpass, and I would do the calculations for them. So eventually I did the calculations for them in two locations. One was at an IBM service center in San Francisco, that being in the Ferry Building, believe it or not. And that was not very pleasant to get out there at night, and it was not a very nice part of town. Anyway, we used it just on a per-hour basis; they charged like $300 per hour to rent an IBM 650. And I had my own wired control panel, which I wanted to emulate what the radiation laboratory did. It was very expensive, but I bought one. Then I would bring a big deck of cards along. Now, one of the reasons today that OCLI numbers the layers that they do in a multilayer stack is when you had a card deck, you fed it in face down so the layers at the top of the stack would go in the machine first, and so when the stack was sitting face up on the table, you wanted it to look the same way it was in the multilayer, and that was next to air, the layer one was next to air.

RS: Not next to substrate?

PB: That is right. That is why we numbered the layers the way we did, because of the way cards are read into an IBM card reader. We did that. Then another place I did it. There was a place called Kaiser Services in Oakland, and I could rent a 650 at that place, too. By 1958 I finally got the optimization of a multilayer stack running on the 650.

RS: What type of problems were you optimizing? AR?

PB: No, we did not do AR, but I calculated primarily reflectors and edge filters.

RS: Visual and infrared materials?

PB: Mostly infrared or visible coatings. Again, expensive government projects could only pay the money to design such coatings.

RS: Did these operate at angles of incidence?

PB: They would have a cone angle coming in, as, for example, for the edge filter for the sidewinder missile.

RS: You decided to go and work at the university?

PB: Yes, so then I finally got my degree. I really enjoyed grad school. I didn’t really want to leave, but I had to. So at any rate, I got an offer at the University of Rochester, and they also had an IBM 650 computer when I arrived. So that was convenient, and I could do all my calculations there.

RS: Also the department, or were you the single focus for thin film research?

PB: No, I was their only person in a university working in thin films in the United States. Now there was a fellow who worked at the University of Rochester a little earlier. His name is Harry Polster. He just published one or two papers back in the mid 1950s, and then he left and went to work for PerkinElmer. Then, of course, I was in the same city as Francis Turner. That is a whole other story about Turner’s operation. Then Kodak had a pretty active thin film group, too. They sold infrared filters to the U.S. Government, out to 6 or 7 micrometers. They had four coating machines running at their Hawkeye plant there. Then the people who really had the big operation going were the American Optical Company. People don’t appreciate that, but American Optical was probably the preeminent optical company not doing military work in the early 1960s.

RS: So we had American Optical and we had Kodak and Bausch & Lomb and then Optical Coating Laboratory.

PB: OCLI was considered to be a dwarf back in the 1950s, then after they started producing the sidewinder missile and took that business away from everybody else, and after they started producing cold mirrors with hard coatings, people stood up and took notice of them. They became a formidable competitor of Bausch & Lomb.

RS: Did **** Denton and Ed Barr and some of these other fellows do some thin film industrial coatings?

PB: Yes, I think at the time it is safe to say that there were only three larger players in the optical coating field. The question is, how do you set up an optical monitor? And I don’t know whether Denton ever set up optical monitors. In that era, many coatings were done without any optical monitor, and I give an example. There is a company called Liberty Mirror Company, and they made a lot of automobile mirrors. And they would do coatings by giving the technician little envelopes with pre-weighed materials in them. They dumped them into the various boats and evaporated everything. And that is the way they would control the layer thicknesses. Three and four-layer coatings were done that way successfully. Barr was very successful. In 1955 he was producing bandpass filters, because I visited him in Cambridge, Massachusetts, and saw what he was doing. So he was good. Then Bausch & Lomb had always had optical monitors. Then Rolf Illsley eventually built them, too.

RS: Typically in-house?

PB: One of Rolf Illsley’s filters was very interesting, what he did. Rolf Illsley would evaporate ceric oxide, and he would evaporate it in eight seconds for a quarterwave. At the time, there was a fairly crude but okay scanning spectrometer (made by I think it was American Optical) that had a vibrating mirror in the focal plane. As that little mirror vibrates (just think of a long piece of spring steel with a mirror glued on the end of it) then that is driven by a magnetic coil vibrating like that. It was quite non-linear, but what the hell. It vibrated and scanned the spectrum across the detector. Then they took the output of that and put it on an oscilloscope, and you could see the spectral reflectance of the single ceric oxide layer move across the screen in real time, in eight seconds. That is the way they controlled the thickness of the coatings at OCLI. It was an interesting accomplishment, I thought. How OCLI got into optical monitors, too, was John Stone had an optical monitor running in 1953 when he finished his thesis.

RS: You said he had one right in the vacuum chamber.

PB: Yes, the photomultiplier was in the vacuum, and the monochromatic light source was outside the vacuum. And it went in through a window, then got reflected down on a glass plate, the monitor plate, then went directly from there up into the photomultiplier, which is inside the vacuum tank.

RS: Which color did it operate? What was the wavelength of operation for this spectrophotometer?

PB: The detector was a 931-A photomultiplier, so its responsivity was close to zero at 700 nanometers. But anything up to 700 nanometers was OK.

RS: Did you use fixed filters or a grating?

PB: He used filters initially, and later he put a prism monochrometer on the monitor system and we scanned wavelength that way. Yes, that was John’s accomplishment there.

RS: Now, some of the early attempts at optimization for the refining, was it limited to a number of layers that you could attempt?

PB: Well, I did some optimization, for Rolf Illsley’s problem again. It was the edge filter for the sidewinder missile. Since the government was paying for it, we could afford a little bit of money in the front end. So I did the design of the edge filter. We would adjust the thickness of the layers to increase the transmission up in the passband spectral region. It worked fine. Then it was a two-stack design, so we had a little bump in the center of the spectral region. And the optimization was slow; it was slow enough so you could watch the flashing lights on the front of the console, the 650, and tell exactly where it was in the optimization. I knew it was in this part of it; it was in the matrix multiplication the way they flash in a certain pattern.

RS: What method did you use for optimization? Did you use like a simplex method?

PB: I used the matrix method. I later published that when I got to the University of Rochester, the optimization method. Now the 650 also had an interesting feature. It had a set of stepper switches on the front of it. You could dial in a 10-digit number. Sometimes when you were doing optimization you could use that feature of the machine if you wanted to, you could address it and grab it, then you could change it real-time, the wavelength. You did in the right way, the coating design would be in the storage of the machine, but the wave number, you could change with the stepper switch. Then you could change in real time the wave number at which the transmittance was to be computed.

RS: So you had optimization and you had some of the characterization routines. Is that the early start of the computer code "MULTFILM"?

PB: Well, MULTFILM really came later. What happened next, is that we had a computer called the IBM 7070, which replaced the 650. It arrived in Rochester about 1960, about a year after I got there. It was fine; as a machine, it had a Fortran compiler on it. So we switched to Fortran. So we had some students write some programs for us to do the thin films. Then, of course, what happened was that all vanished when the 7070 got replaced with the IBM 360, and that is when I said, I am going to write it myself. I started writing MULTFILM in Fortran.

RS: For the 360?

PB: For the 360 IBM unit, the IBM 360 compiler. And at that time, OCLI had an IBM 1620 computer, and they did all their little calculations up in the research department on that. Bausch & Lomb was somewhat more conservative. They at one time were a real leader in the computational area, but they kind of fell behind, because I don’t think there was a driver. I think Ivan Epstein, when he was there, got the thing working completely independent of Turner; Turner was always a little bent out of shape by that. Epstein computed tables of the refractive indices of equivalent layers versus the thickness ratio of the two materials. These were published in U.S. Government reports.

RS: Did George Haas develop these early tables?

PB: Hass was the guy who funded Turner. Haas funded Turner starting probably in 1950, 1951. Turner had an "in" with Georg Hass because Turner could speak German and he got his Ph.D. in Berlin. Bausch and Lomb didn’t get a lot money, and hence the government really got its money’s worth for it. They produced some of some of the early infrared bandpass filters there. They were never very good at materials, but they did the calculations. Turner invented the multiple half-wave bandpass. That was certainly one of his inventions. You can read about this development in copies of the old Bausch & Lomb reports. Now, incidentally, you should try to get in touch with Peter Berning. He probably has a complete set of them. Maybe he could photocopy them and get them to the archives.

RS: That is a good point; I shall make note of that.

PB: Yes, because the old, so-called Fort Belvoir Reports. Fort Belvoir was the place where Haas was physically located. So he had them there.

RS: So how long were you at the University of Rochester in the thin films group?

PB: I started in 1959 and finally left in 1978. Nineteen years.

RS: How many students did you have then, roughly?

PB: I probably had 15 or 20 master’s students and maybe four or five Ph.D. students in thin films during that time. Then here I was, late 40s, trying to do a career change, not the easiest thing to do in one’s life. And so any port in a storm. I went to an aerospace company for one year, which was AeroJet. Then Alfred helped me out as an old buddy and got me back into OCLI again. That was a fun job. I enjoyed that very much.

RS: When did you start at OCLI again?

PB: October 1979.

RS: And you worked for Alfred at that point?

PB: And I reported directly to Alfred, and we did some fun things.

RS: Now you had some duties there as training of some students.

PB: Yes, we did training, but I think one of my earliest contributions to OCLI was that when I got to OCLI, I was aghast at how backward the computer department was. What they had done for years was try to save money, and of course, yes, you can do that, but you pay a price for it. And so they had their own separate little mini-computer. They were kind of offshoots of the old IBM model 1620. And the machine OCLI used at that time was produced by Varian, and the operating system I think was written at Chico State University. And it would crash once an hour.

RS: I remember that.

PB: Yes, and it was just bad. And so here suddenly IBM mainframe was coming in there and I really lobbied hard to get the scientific community to use it. I did the first Fortran compile at OCLI, I got the Fortran compiler installed -- at $500 per month rental, I might add! But we rented the Fortran compiler from IBM, and I got my computer code MULTFILM to compile okay, then I knew I could do other stuff, too. But I knew that this was the way to steer the scientific community at OCLI away from little mini-computers. The operating system of the mini-computer was so bad. People time was valuable, and this was just not a good way to spend your time.

RS: I remember us getting more time to get your deck to read than actually the computation.

PB: Oh, it was bad. It was very bad. So then we finally got terminal input. It was expensive, but we got the terminals from IBM because they had the IBM VM operating system at OCLI. So anyway, that was one of the first things that happened. And we had awfully good computing facilities at the University of Rochester. We always had big IBM mainframes. You never worried about memory space, storage space, stuff like that. OCLI was still fussing around. They had several programs, for example, one to do dielectric films, then they could use more layers in them because it didn’t take up so much space for the program. The one to do absorbing layers could do fewer layers because the program was bigger. I’m glad we got out of that mini-computer business and got over to the IBM mainframes. It was a big breakthrough for the company, I thought.

RS: As I remember, Phil, in 1979 there were, as you indicated, lots of small subsets of programs. If you wanted to do a cone angle calculation, you had to pull up that file. Then you had to do another one to do absorbing layers.

PB: That is right. Yes, they had these strange names for them, too. You had to really know what the names were, to use them. But Carol Snaveley did it, and as I said, as long as you wanted to run a tight operation and not spending much money, she was great. But I think the company deserved something better. Now, I shall say one thing related to OCLI. They converted over the older program to the IBM. They had this older program that ran on the mini and when they did that, they converted over and it was part of their own computer code ASP. I also ran my own computer code MULTFILM at OCLI. It turned out that when we compared the answers for the radiant reflectance, ASP was not giving the correct answer. The difference was in the third decimal place, but it was definitely there. It turned out that they had gotten the square root routine screwed up.

RS: Oh, goodness.

PB: So I saved their ****. If that had gone on and some customer had found that mistake, the proverbial would have hit the fan. But any rate, I saved OCLI’s **** because there was an independent computer program that you could check against.

RS: Did you do some machine automation and training of students?

PB: Yes, and we had a very small optical coating machine, very purposely, so no one would ever want to grab it and try to do production in it. That was a very deliberate decision on my part. And you could learn just as well on a small machine. And we did almost all of our maintenance on all of our electronics, wherever we could. I got people who liked to work with their hands and somehow the people at OCLI, the maintenance people, did not know how to leak-test a chamber properly. Did you notice that when you were there, Ric?

RS: Yes. It was always a struggle. We would always want to do independent checks to make sure that they were making the correct measurements. What were they measuring? Where did they put the sensors?

PB: They always put it in the wrong place as far as I am concerned!

RS: Then we would check the leak-up rate, do some simple tests, and see what we got. It seemed to be a difficult concept to transfer to the maintenance group.

PB: Yes. At any rate, if you really want to leak test a chamber properly, what you do is that you take off one of the windows and you put a spigot on there that can connect — you don’t use rubber hose, you use plastic hose — directly, using as short a distance as possible, into the diffusion pump of your leak tester. And when the pressure is smaller than 10-5, you shut down the main DP, and you let that little diffusion pump handle the whole machine, which can easily do it at 10-5. Now all of the helium goes by mass spectrometer. You get the tremendous sensitivity. The OCLI maintenance people couldn’t grasp that fact.

RS: But you had put in data logging?

PB: Yes, we were set up. We used an IEEE 533 interface, which was used in large accelerators, nuclear accelerators, and therefore it is very independent of magnetic field interference. And that is why I wanted the reliability in it. On a typical interfacing project, 80% is labor and 20% is the capital. Do not ever try to chintz on the capital equipment; it’s not a good investment.

RS: You had some of the CAMAC?

PB: Yes, CAMAC was the trade name for the interface we used. That proved to be a good one. I don’t know if there are better interfaces, now. Things have changed in 20 years. But that is what we used. And CAMAC even at the time was getting to be old; but the point is, there were still people using it. You could buy the hardware and do that. The machine was working fine. And we had a little PDP11 computer (machine), which was a modern computer to use. They had tremendous software with it, stuff like that. Now at the U of R, the problems we ran into there were really getting funding. Back in the 1960s, money was just sluicing through the whole economy from science. So I got funding from NASA to produce some filters for the ultraviolet. And I published a paper on a bandpass filter for a 184.9 nm mercury line, and we did bandpass filters for the lower wavelength in our chamber. We had a vacuum monochrometer, McPherson, and we produced filters of 121.6 nanometers for NASA. That was a good field. But then NASA got short of money in 1970. Then I finally began to scrounge some money from Los Alamos. Again, we had to produce a product. I had to actually deposit antireflection layers on some optical parametric oscillator crystals. I personally would go out to the airport at 5:00 PM to get them in FedEx so they’d get it in Los Alamos the next morning. And we didn’t have spectrophotometers; that didn’t help us, either. So we did coatings like that. For example, we did a mirror at 16 micrometers composed of germanium and lead fluoride. Lead fluoride is transparent, with really a long wavelength out in the infrared. We did coatings like that. But I could see that whole project coming to an end — and it did, right after I left. The reason was the lack of funding for our research. Then I had some ONR funding for a while to do ultraviolet mirrors. We had the best reflectance of anybody. They were kind of begrudging to give us any credit for it. So anyway, in the 1970s we were branching out with our coating machine there. Then the next thing that happened is that they changed buildings on us; we got moved to a new building. We had a chain hoist stuck right in the ceiling to hoist the metal bell jar of my machine, and we had all the water flow interlocks in the DP and things like that. And the whole thing just got dumped in another lab in the new building. I didn’t have any tech help or anything to get it back in operation. One graduate student can only do so much. That and some other things convinced me to get out of the university. It was so depressing trying to get that up and running again, my one coating machine. We had a little BA500 running, too, but we mostly did service coatings. Now [Angus] Macleod got into the big largesse that the University of Arizona got from the space-wars stuff. The strategic defense initiative. They had lots of that money down there. That did Macleod very well for years. Then when that funding finally vanished, Macleod quit in a couple of years.

RS: You started a conference that is held every three years or so.

PB: The idea was to get together every four years and give some very significant papers on thin films, rather than to keep publishing the same stuff year after year. So we decided to have, every four years, a thin film conference. 1976 was the first. Anyway, we had one of the first thin film conferences, just devoted to optical coatings.

RS: Where was the first one held?

PB: That was the Asilomar Conference, near the city of Pacific Grove, California. I think it was 1976.

RS: Who showed up there? How many people were at your first conference?

PB: I think we had a good 100 people, easy to meet the overhead that way. It was the Optical Society.

RS: It was organized under the Optical Society?

PB: That is right. We had the Optical Society of America. We had people like George Dobrowolsky on the organizing committee.

RS: Was it international?

PB: We tried to get as many international people as we could, and we were successful because Asilomar was a very nice place to visit, and that helped.

RS: The tradition is, it is ongoing now?

PB: Yes, then we kept that going; we cut it down to a three-year cycle. But it is still going today. We use the word "interference" in the title of the conference because there are so many other types of "thin films" (thin films for electronics, semiconductors, and so on).

RS: Phil, we have just seen a big bubble burst in the telecomm side of optical thin films. What is your feeling, reflecting back on that experience?

PB: It was like a meteor running through the sky, wasn’t it? Really it was a burst of light. Suddenly you realize that if you knew something about bandpass filters, you were an important person because bandpasses were the big thing then. In the year 2001, February, that the bottom fell out of it, didn’t it?

RS: Looking to see if it turns up.

PB: Yes, we’ll see what happens. I’m still publishing some papers on WDM, but we’ll see what happens on that.

RS: Where do you think we are in the world of optical thin films in relation to computer codes and computer calculations?

PB: It is like what is happening in all branches of engineering. The software is taking over. It used to be that we, as engineers, understood the basic theory that underlay the design methods. But you do not need to anymore. Literally you can take a person off the street and if they’re intelligent and innovative, they’ll learn in a couple of months how to use the software to get the results. So I just do not know whether there is a future for equivalent layers and other design techniques. We have learned through the years and we’ll find out. But the software is getting better and better. When I worked for Deposition Sciences, I gave a paper showing that many thin layers is a good starting design for an antireflection coating, like an AR that spanned a two-to-one bandwidth. And so I picked the thicknesses of the layers for the starting design out of the Santa Rosa telephone directory. And some of the optical coating designers that listened to my talk were very offended at that. They thought that I was making fun of designers. I was not making fun of them, but the system worked. And then I introduced the starting design into the computer and optimized it very quickly into an antireflection coating. So that’s really the direction I think we are going. At one time, you designed edge filters using equivalent layers. If you are working in an industrial atmosphere, you can’t mess around with that sort of stuff. You have just got to dump it in and optimize it. And the only judgment is knowing how many of the outer layers you permit to be changed by the optimization. You usually keep the inner layers constant in thickness. Antireflection coatings are now the product of computer optimization.

RS: Where do you think we are with respect to understanding the nature of films, the microstructure, what actually we deposit?

PB: I’m hoping to find out someday a definitive theory of the scattering in thin films, and I just haven’t seen it yet. I know a lot of people have worked on it, there is Claude Amra, and there are people at China Lake and maybe some other places.

RS: How about in terms of the actual different processes out there for depositing thin films?

PB: Yes, once you get something working, it is not liable to change. The tradition has always been to use vacuum evaporation. Then the one method that is probably the one that has a great future is ALD (atomic layer deposition).

RS: So this is a chemical process, then?

PB: It is physiochemistry. First, you form a monatomic layer on the surface of the substrate by the introduction of water vapor into the chamber. And then you let the precursor in, and the precursor only reacts with that monatomic layer--you only get one atomic layer each time. So number one, it is tremendously efficient at not wasting your precursor material. Presumably it can coat any surface, curved or flat. It is the one with a real future to it, I think.

RS: One of the drivers that I have seen is that optical coatings tend to be an enabling technology, and there is a motivation to install coating plants in-house at the end of some manufacturing process, as opposed to being independent, supply-based. What do you think of the different consulting people you support?

PB: Well, can you see some examples of that where people actually did that, or did they actually try to do it, during the WDM craze to guarantee their supply?

RS: I see some focus on lighting applications at GE and Philips. They no longer rely on shipping parts to a company like Deposition Sciences. They decided to do theirs in-house.

PB: In a tremendous mass production that certainly makes sense because your are trying to save pennies. You take a cold mirror. The glass alone costs 25 cents in an "MR 16" cold mirror, and so you cannot add a lot of value to that without going over price. So you do want to do that coating in-house if you possibly can.

RS: So I see that as one driver. I see another driver as the focus of a lot of the technology moving to the Pacific Rim, to China and some of the different industrial centers there. It would be difficult to get instruction in thin film research here in the United States or in North America, maybe.

PB: Is it difficult though? Because look at the educational program of the Society of Vacuum Coaters, that you have done so much for. And so people can attend these courses. There are a lot of ways that you can learn information these days, and the reason why the UCLA course that I taught starting in 1978 was so successful was that there wasn’t anything else, literally. And so the first time we taught it we had 60 people in it, in 1978. There was a pent-up demand there for that sort of information. Then the bigger companies like OCLI could educate their engineers in-house. Then they started teaching courses at the Society of Vacuum Coaters. But that had taken time to do that and get that running.

Now the experience I had at Coherent after I left OCLI in 1985 was interesting because, you see, I had been spoiled at the universit

because I could visit the library. I had a library key and at 10 o’clock at night and could get a book off the shelf, make a photocopy of it. Then I noticed that when your are isolated like up at Auburn, where it is an hour’s drive to go to [UC] Davis to even get some of the journals, not all of them, I began to appreciate how the "man in the street" had to survive without personal access to a library, like I had at the University of Rochester. Then the other thing I learned up at Auburn is how to keep production going with a minimum amount of money input. They were tight as the paper on the wall. They would rarely spend money for analytical techniques, to send something out to have an auger analysis or Rutherford backscatter. OCLI had all of these capabilities. You just kind of guessed at what you thought the problem was, if you weren’t getting your process working right. And you’d better be right, or you wouldn’t have a job in a few weeks. So with that motivation we usually did it. So it was over seven years up at Auburn, and I enjoyed that different environment up there.

RS: Do you think, Phil, that the Internet has addressed some of the problems of isolation of the individual thin film researchers, that maybe they could get their articles or information if they work just a single man in a coating shop?

sPB: That’s a good question. I don’t use the Internet enough to say anything intelligent about that.