Hybrids produce high performance

A DISCUSSION WITH RAY CHEN OF RADIANT PHOTONICS

Ray T. Chen is chairman and chief technology officer of Radiant Photonics and is the Temple Foundation Endowed Professor of Electrical Engineering at the Microelectronics Research Center at the University of Texas at Austin. His previous experience includes director of electro-optic engineering at Physical Optics Corp. Chen received the University of Texas' 1999 Engineering Foundation Award for contributions in research, teaching, and service. He holds a Ph.D. in electrical engineering from the University of California.

WDM Solutions: What is the technical focus of Radiant Photonics?

Chen: Globally speaking, the major technical direction is to solve the bottlenecks of fiberoptic communication—wavelength bottlenecks, switching bottlenecks, optical transparency bottlenecks. To ease the wavelength bottleneck, we use a new blazed-grating-based WDM technology to build both coarse and standard muxes and demuxes up to 50 channels that can cover both single-mode and multimode fibers. To relieve the bottleneck of switching for all-optical networks, we created a new thermo-optic-based switch using both a hybrid and a monolithic approach—in other words, using silica-polymer and polymer-polymer technologies. We call this new platform TOIC, which stands for thermo-optic integrated-circuit technology. The switches and variable optical attenuators [VOAs] we plan to build are based on this technology. They are transparent to C-, L-, and S-bands. So basically you can cover the spectrum from 1300 to 1600 nm, which I think makes the technologies very attractive for many uses in both metro and longer-distance networks.

WDM Solutions: You say your thermo-optic switch is a hybrid. What do you mean by that?

Chen: To have a waveguide structure, basically you need to have a high index layer and low index claddings. You can have two approaches: the monolithic approach is when you use a different index of refraction based on the same host material, such as silica. Then you dope it with different materials to form a high-index core layer. In the hybrid approach you have different material combinations. You can use one material as the core and another as the cladding: for example, a polymer cladding, which has a higher thermo-optic coefficient. That will make switching efficiency much higher than the purely silica-based version.

WDM Solutions: What is the basic technology of your VOA?

Chen: The VOA is based on the technology of converting guided-mode to leaky-mode. By doing that, you can continuously adjust the throughput intensity by changing the conversion efficiency from 0 dB into 20 dB. It is a planar waveguide technology using conventional silicon CMOS foundry.

WDM Solutions: How do things stand in terms of manufacturing these three different types of technologies?

Chen: We are in the process of building a large clean room that can have a throughput of several hundred wafers per day. Each wafer will contain 100 to 200 devices. The VOA and switch are pretty much based on conventional silicon CMOS technology. And we can build the WDM module itself. The facility will be able to manufacture over 40 DWDM units a day of various kinds. That's the goal we are trying to achieve in the second quarter of this year.

WDM Solutions: Why are these technologies particularly relevant for metro, short-haul, or long-haul applications?

Chen: The 8 x 8 and 1 x 8 switches are targeted at vendors providing the optical networks in the metro area—so that you don't need to go to optical-electrical-optical conversion. In other words, the carrier is all-optical from the beginning to the end.

The variable optical attenuator is basically a device to equalize the channel power in core and metro-access networks, where you can have 40 different channels on one fiber.

In principle, these devices can be used for both metro and long-haul applications.

WDM Solutions: Are you looking at coarse WDM or is that just one option?

Chen: Basically we have three families of products. One is a high-channel-count dense WDM module with up to 50 channels. The second is a 6-channel coarse WDM module covering both the C- and L-bands. And the third family is an 8-channel multimode DWDM module for last-mile free-space optical communication in the metropolitan area. So to answer your question, the 6-channel coarse WDM is one of the family we are developing to meet our customers' needs. For all three families, we have potential customers.

Chen: That is a very good question. Currently there are over 250 metro-system startup companies in the world—a majority of them in the U.S. And currently installed fiber to the office in the U.S. is only 3% to 5%. In other words, a big chunk of people doesn't have fiber in place yet. So in the next few years, you will see a lot of implementation of fiber to the office. Among these 250 metro-system startups, different companies have different system architectures to release the bottleneck. So whether coarse WDM is the dominant technology depends on which company you talk to.

Some people really like coarse WDM because their system requires only up to six channels covering the C- and L-bands. Some of the metro system companies may not have 40-channel DWDM but may plan to in the future. In this case, coarse WDM cannot meet their channel count requirements. I would say that coarse WDM will play a significant role in the next few years in the metro space mainly due to the cost-effectiveness and lower-tolerance requirement of the lasers needed for the WDM interconnections.

WDM Solutions: Over the years you have done considerable research on integrating optical and electrical devices, and you have worked with DARPA and various other organizations. What do you see in the future for these devices?

Chen: This is a very global question. Basically there are three competing technologies in the WDM space that try to integrate optical and electrical devices. They are AWGs [arrayed-waveguide gratings], volume grating-based technologies such as our blazed-grating product, and thin-film filter-based products. And each one of these technologies has its customer base. For example, if you're talking about low-channel count—lower than four—I think thin-film filter devices have advantages in cost because they are basically cascading devices that add four different filters.

But when you go to higher channel counts, AWGs can win the game. Among the three technologies, I think low cost and high performance are key. In the context of the competing technology, the blazed-grating-based technology among the three has the advantages of low cost and less polarization dependence. It can upgrade or double your channel count fairly easily, which is not the case for the other competing technologies. I believe many people will use either blazed-grating-based or AWG-based system filters in the next few years—both in metro and long-haul. The 1-dB passband of blazed-grating-based DWDM is a primary concern for system users. We have recently solved this problem.

Chen: I think they will gradually become a reality. There are several companies working in this field. We are trying, for example, to integrate switches and VOA channel waveguides together. This type of technology is very attractive because it is mass-producible, although several technology bottlenecks need to be overcome. The number one bottleneck is material reliability because, as you know, we have optically transparent material trying to integrate onto a substrate—that is a different kind of requirement and requires different kinds of processing conditions, which need to meet the Bellcore standards. Number two bottleneck is packaging.

For example, you need to do the fiber coupling, you need to do the pigtail, and you need to do a lot of connections in the optical domain. I think once the material and the packaging issues are solved, this technology will prosper quickly.

WDM Solutions: Do you see any important new technologies beginning to emerge?

Chen: Right now most people focus on the C-band. Eventually, they will go to the L- and S-bands. In the future you will need to use the entire spectrum from 1300 to 1600 nm. So there are several enablers to help realize such a broadband optical network. For example, ultradense WDM can cover hundreds, even thousands, of wavelengths. In 2000, I chaired a panel discussion at Photonics West. We concluded that in five years, the bandwidth needs of each individual in this country will be about 10 Mbit/s. When you can multiply that by the total population of this country and then eventually the world, you can see how big the bandwidth demand will be. So I think we are only observing the tip of an iceberg.

The second enabler will release the bottleneck to your desktop and allow high-speed network data to your computer in a much faster format. That enabler is the optical backplane.

WDM Solutions: The core of that would be free-space interconnects, VCSELs, that sort of thing?

Chen: Well, both, either guided-wave or through free space. The optical backplane technology we developed is a hybrid structure. In other words, it is fully compatible with IEEE standardized bus protocols such as VMEbus and Futurebus. So back to your question—which one will win? Well, I think in the next few years you will see that the one that demonstrates the reliability and the robustness in a harsh environment will be the winner.

WDM Solutions: What do you see as the biggest obstacles facing Radiant Photonics and what are you doing to meet them?

Chen: Challenge number one is we want to get the best people—the best technical team and best management team. Actually, we pretty much have our management team in place at the vice-president level. Then I think the second challenge will be how fast the service providers truly implement the new optical networks because that is what really makes the market a reality. This is crucial because it will determine the volume, cost, and all the logistics needed to build a successful company.

From the technical point of view, I think the most challenging part for any startup in this space is packaging. If you trace all the Ph.D.s that are working in optical communications, none of them are really working on a packaging-based dissertation. Why is that? I think most Ph.D. students do not like to be involved in purely packaging-related research because really, from a scientific point of view, it is more of a craft.

This industry is relatively young compared to microelectronics—and photons and electrons are different animals. So it requires different packaging technologies to accommodate both together in a miniaturized packaging that electronics and optics people can accept. I think this area really needs a lot of input from managers and scientists.