A DISCUSSION WITH DAVID PAYNE OF SOUTHAMPTON PHOTONICS
BY CONARD HOLTON
David Payne is the cofounder and chairman of Southampton Photonics (Southampton, England) and a director of the University of Southampton Optoelectronics Research Centre. In the mid-1980s, he led the team that invented the erbium-doped fiber amplifier. He was also the cofounder and director of York Technology and a director of GeoSensor Corp. David Payne was awarded the Benjamin Franklin Medal by the Franklin Institute for his work on the EDFA and is a fellow of the Royal Society and the Optical Society of America. He has a doctorate in optical fiber communications from the University of Southampton.
WDM Solutions: Southampton Photonics was recently spun out of the Optoelectronics Research Centre. What are you major technical directions and why?
Payne: We have always had a reputation in the Centre for developing the latest and greatest new component technologies, often based on new material developments. And we are one of the very few universities in the world that have fiber and planar manufacturing capabilities. So it was our intention to use these assets and develop a company that would be a major global supplier of the very latest new component technologies. Initially the technology has been acquired by agreement from the Optoelectronic Research Centre. Ultimately, of course, the company's technology will have a life of its own.
WDM Solutions: So components and subsystems are your focus?
Payne: That is right. The photonics market, up to now, has been dominated by new component developments, and I can give you many examples of that, starting way back with the fiber itself, the laser diode before then, and in no particular order, the optical amplifier, the Bragg grating filters, the AWG [arrayed waveguide]—all of these had a profound impact upon the direction that optical telecommunications is currently taking. I don't think for a moment that we've seen the last of component developments. And these then produce an earthquake within the market, and everybody moves off in a new direction
WDM Solutions: The EDFA has been the great enabler of WDM. When you and your colleagues first announced it in 1987, was it possible to foresee what was coming—the tremendous bandwidth and the multiple channels that could amplified?
Payne: Yes. For a good 18 months before the announcement, we had been working on rare-earth fiber lasers, and we made statements in a number of publications that an optical fiber amplifier would be ideal for very-high bandwidth optical fiber systems.
When we actually developed the amplifier, as with all new inventions, it took up our whole laboratory and needed three PhDs to drive it. It used a huge gas laser as a pump source and consequently total power consumption of the amplifier was about 100 kW.
We did realize the potential, but we were a little daunted by the amount of work that was required to reduce it to what we have today, which is a small, very highly controlled, very precise amplifier at a reasonable price.
WDM Solutions: Now that we have it down to its current size and shape, where do you see it going?
Payne: I think the EDFA is, without a doubt, here to stay. There are several other technologies, but it is hard to envision a better amplifier than the EDFA. And just remaining on the historical track for a moment, one can but wonder at the luck of actually picking on the one dopant that produces the perfect amplifier at 1.55 µm when all the other dopants do not produce a reasonable amplifier in silica base. So physics was kind to us.
What can be done to improve it? Well obviously, flatter gain, wider bandwidth. We have seen a lot of very good work from all the major labs and suppliers on developing the S-band and L-band amplifiers because, of course, the whole of our market is about bandwidth, and that means capacity. So there is considerable revenue associated with developing wider and wider spectrum WDM systems.
We are getting pretty close to the limits of what can be done. People are developing new glass hosts, such as the telluride glasses, in the hope of squeezing that extra little bit of bandwidth out of the EDFA. We have the praseodymium amplifier, which is based around 1.3 µm, and the thulium amplifier for the S band. It is a sad observation, but all of these other rare-earths require a different type of fiber, a different glass—and technologically that has yet to be fully developed as a viable alternative to the beauty of a silica host.
So where these amplifiers have been demonstrated, they have not been a great success commercially. This brings us to the other types of amplifiers, the semiconductor and waveguides amplifiers, and the Raman amplifiers. I think it would be a mistake to assume that any one amplifier will dominate the whole market. There is always room for other amplifiers, although I think most of the market projections suggest that their total market will be relatively small compared to the EDFA. Also in many cases, they are used in conjunction with the EDFA.
I think the Raman amplifier, which is causing a lot of excitement at the moment, fits that description exceeding well. A lot of people are interested in developing combinations of Raman and EDFA amplifiers to stretch or to maximize the span between amplifiers. The semiconductor amplifier, on the other hand, I think, has an entirely different potential. It provides functionality which you don't get from the EDFA—namely its fast switching capability. That, by the way, is also its disadvantage. It means you get channel interactions, which are quite hard to overcome. But I foresee a great future for the semiconductor optical amplifier, with integrated packages of multiple amplifiers for use in switching applications.
WDM Solutions: Perhaps wavelength conversion as well?
Payne: Yes, wavelength conversion is another possibility. But I don't think we've seen the last of the technologies for wavelength conversion yet. I think there are many other possibilities that are up and coming, and the market will have to decide for itself what is best.
There is one amplifier I have left out so far— the waveguide amplifiers. This is a fairly controversial area, and these comments often apply to many planar devices. One should never forget the beautiful circular symmetry of an optical fiber, which makes it relatively insensitive to polarization effects. If you then go to a planar structure, you inevitably end up with polarization birefringence, which everybody fights to overcome. The AWG designers fight to overcome it, so do the amplifier designers. Every other planar waveguide structure has similar problems.
WDM Solutions: The metro market is the target for waveguide amplifiers because of the lesser power requirements and the integration capabilities.
Payne: Well it is debatable whether there is lesser power required in the metro market. Again, I think the market has yet to decide precisely what it wants. But in many applications, relatively high power is required because it's effectively a splitting loss compensator, and in metro you would like to to split the signal many ways. So I think there is a debate, and certainly I would agree with you that low cost is the key for the whole amplifier market. It remains to be seen whether the planar approach can deliver that while producing the right performance.
In examining the issues of cost and performance, one must look at the manufacturing and assembly process. I think what Southampton Photonics is offering is a very new concept which addresses a lot of problems. Design for manufacture is key and the photonics market hasn't done that now to any great extent. In our amplifier we've developed a way in which the pump launching, which is often the difficult part, can be very much simplified.
We believe that ultimately this will lead to major cost reductions in manufacturing, while at the same time providing versatility so that, for example, you can put multiple amplifiers in a package and minimize the number of pumps. You can also get high power—something in excess of a watt is what we are currently offering in power output. And in a versatile package, for example, containing 8 or 16 gain blocks.
WDM Solutions: Can you describe your Southampton facilities and plans for the future?
Payne: The strategy of the company is to be a major manufacturer, and we are currently building the necessary facilities. We have already gotten the first stage of our facilities on line, and delivered prototype products to customers just before Christmas. We expect to be in major volume production by third or fourth quarter of this year—but with supplies well before then.
In the longer term, by which I mean two years out, the company is considering its options because we will need to continue to scale up for major production. Precisely where this capacity will be located is a subject of debate. One notes that most of world market in these components is in America. We actually have our corporate headquarters in Cupertino, CA, and are incorporated in Delaware.
WDM Solutions: To what extent is it possible to automate amplifier production? I've seen amplifier lines at major component manufacturers and there are a lot of people involved, a lot of hand assembly.
Payne: I think it is not just amplifiers which suffer from this problem. You are quite right—photonics assembly lines have large numbers of people assembling things by hand. It is nothing like the sort of line that you would see, for example, developing photonics components for CD players, which is totally automated. I think we all have some lessons to learn and there is an increasing awareness throughout the industry that this must be done if we are to get the cost reductions needed in this marketplace. For example, Southampton Photonics is starting to tool up our Bragg grating facilities for multiple manufacturing to minimize staff involvement.
WDM Solutions: What do you see coming in terms of breakthroughs or really innovative technologies?
Payne: We all need a crystal ball for that. It's perhaps easier to ask what the market wants and predict that that is what will appear. There are a couple of observations that can be made. We've covered photonic integrated circuits partly in my comments on planar activities. The semiconductor engineers have been working on this for a long time, and getting three components onto an integrated circuit is still considered to be pretty impressive.
I will probably get shot for saying this, but I believe we will never have integration in photonics at the level that we have in electronics. One should not try to fight physics. A photon is a large thing, and photonic interactions are weak compared to electronic ones. This translates into relatively large-size devices, so one has to recognize that up front.
Furthermore, it is an extraordinary fact that we have an analog network—analog in the sense that what we put in degrades as it goes down and is never reshaped in the optical domain. Whereas in electronics, you do regeneration all of the time. Any digital electronics has automatic reshaping within it. We don't have that in photonics, which is incredible.
So we end up with dispersion compensation and all sorts of things like that because we are working effectively in the analog regime. I think converting that into digital by means of optical level switches will come. In other words, the all-optical 3R regenerator will come, and I think that will profoundly change the way we transmit information in optics.
WDM Solutions: Would you identify some research trends for us to keep our eye on?
Payne: I would watch some of the very nonlinear glasses or crystals which people are using as switches because they are extremely fast. I think the challenge there is to get the switching thresholds down. This can be done by using very, very small mode sizes, for example. One of the projects at the Optoelectroncs Research Centre—which is being closely watched by Southampton Photonics—is the so-called holey fibers, which enable you to get an incredibly small mode-spot size so that potentially you could reduce the threshold of switching by using a combination of small spot size and a highly nonlinear new material.
Another very interesting area to watch is what is generically called IP over optics. That is, can we use Internet protocol? Can we recognize packet headers and route them in the optical domain without having to go through a router and do it all in the electronics domain.
It is not an easy thing to do optically. But it is work that we have been doing at the Optoelectronics Research Centre—which is now funded by Southampton Photonics—on the recognition of packet headers using a form of OCDMA [optical code domain multiple access] technology. It has been very, very successful to date as a potential new technology. Using the Bragg grating capability to be able to design passive filters that are so precise that they can actually recognize the unique footprint of an IP header in the optical domain. It is very exciting.
WDM Solutions: What do you see as the biggest obstacles you are facing in terms of developing new technologies?
Payne: Well I think the first law of optics is that you do not have the material you want. We wish we had the optical equivalent of silicon, but we don't. And that is usually the obstacle. I think the really successful companies will be those that have a strong materials focus. As has been observed by many people, most optical materials are optical used, not necessarily because they are the best way of doing things but because they happen to be the technology that we have or that we can manufacture.
A perfect example is the use of silica for planar waveguides. Silica is not the best material by any means, but we have the technology so that is what we do. I think that we will see, particularly in the planar and the integration and switching areas, new materials emerging with far better parameters, which will give us major performance enhancements.
WDM Solutions: Your earlier success in erbium is evidence of that, I suppose?
Payne: Well, I think that is absolutely right. I think most of the great advances in optical telecommunications can be traced back to the availability of components, which in turn were caused by materials research and development. As they always say, the past is the best predictor of the future.
"I foresee a great future for the semiconductor optical amplifier, with integrated packages of multiple amplifiers for use in switching applications."