In components, its good to be on the receiving end

A discussion with Mark Itzler of JDS Uniphase

Conard Holton

Mark Itzler is chief technology officer and vice president of research at JDS Uniphase`s West Trenton facilities. Prior to his current position, he was director of research and manager of device design at Epitaxx before it was acquired by JDS Uniphase in 1999. Itzler was presented the Technology Breakthrough Award for his development of high-speed avalanche photodiodes. He holds a Ph.D. in physics from the University of Pennsylvania and completed post-doctoral work at Harvard University.

WDM Solutions: JDS Uniphase had been making detectors for a long time. What are your major product lines related to WDM?

Itzler: A lot of our focus recently has been on long haul, so we have devoted significant resources to developing OC192 detectors: pin [p-intrinsic-n] and APD [avalanche photodiodes]-based analog front-end receivers. This grows out of a substantial business in 2.5-Gbit/s receivers, which right now is the core of our metro business and which we anticipate growing quite significantly. In addition, we see 10-Gbit/s receivers entering the metro market.

In optical-network monitoring, we have a large legacy business in fairly standard pigtail photodiodes for basic monitoring applications. We are also seeing-as do many others in the industry-an enormous opportunity in the monitoring area, and so we are working on devices useful for WDM channel monitoring. The focus is on the photodiode array as the core element, but with the addition of the optical functionality necessary for channel monitoring.

Another legacy area for us is test and measurement. We have been making large-area photodetectors for quite some time, and so we still have a substantial market for instrumentation devices.

WDM Solutions: Can you give us a brief introduction to the primary detector types that you manufacture and the materials issues involved?

Itzler: For the last 15 years or so, we have been making photodiodes with an indium phosphide-based structure-it`s a double heterojunction structure with an indium gallium arsenide absorption layer that is really the key to the device because that is the semiconductor material that absorbs wavelengths that are relevant for telecom in the 1- to 1.7-µm region, And it`s lattice-matched to indium phosphide, so we are able to make very-low dark current, high-reliability devices using that platform. The structure is optimized in various ways for different applications.

More recently we have developed the APD based on the same material system. The great benefit of this device is that it provides internal gain through the avalanche process. However, it is a much more difficult device to make.

WDM Solutions: Have you seen a strong demand for the APD?

Itzler: Yes, absolutely-and in particular, at 10 Gbit/s. We also see it at 2.5 Gbit/s as well. It`s really a question of tradeoffs in the system design. The APD is a rather convenient and inexpensive way to get extra sensitivity in a highly integrated design, which is much better than getting it through optical amplification or additional electrical amplification.

The measure of cost is relative. If you look at, for instance, optical amplification as the means of counteracting loss in a system link, you can come up with a rough number of what it costs per dB. That number is almost certainly below $1000 per dB but still in the range of several hundred dollars per dB, if not higher.

If I look at the fact that I can increase the sensitivity of my receiver by 8 dB at 2.5 Gbit/s or as much as 5 or 6 dB at 10 Gbit/s, there is quite a lot of value in that extra sensitivity-not to mention the increase in dynamic range that it adds to the receiver. So it is really a question of whether you can add an APD to get enough additional sensitivity and, for instance, eliminate one of your optical amplifiers. It is a no-brainer when that opportunity exists.

WDM Solutions: A pin receiver does not have this capability so you are not going to get any kind of gain. Does this mean you have to use an APD?

Itzler: That`s right. The standard pin photodiode has an intrinsic quantum limitation, meaning that one photon in will, at best, give you one electron hole-pair out. So you really can`t do any better than that, even if your quantum efficiency is 100%.

WDM Solutions: Did you need a new manufacturing facility or line for the APDs?

Itzler: No. One of the approaches that we took to our APD was to design a device that is as consistent with our current manufacturing technology as possible. And so, in fact, we`ve managed to avoid somewhat more difficult processes, such as epitaxial regrowth or ion implantation. These things are used where they must be, but our devices are based entirely on zinc diffusion technology, which we have been using for the last 10 years. It has allowed us to put a device into manufacturing that has the basic processing technology already well established.

WDM Solutions: Do you think that the pin photodiode and the APD will co-exist at various speeds or is it just going to become all APD at some point at higher speeds?

Itzler: Certainly there will be a co-existence. What the mix will be, I think, depends on quite a number of factors. Cost is a major one. I think, in most cases, if somebody could have an APD-based receiver for the same cost as a pin-based receiver, they would go for the APD. There are some disadvantages to it. It does require a substantially higher bias. APDs typically are biased anywhere from 30 to 50 V, depending on the bit rate, whereas pins are normally down at the 5-V range or lower. The other issue is that APDs have a temperature sensitivity that has to be compensated for. So if there are fairly nondemanding applications where even those requirements are an impediment, APDs would not be the preferred device.

WDM Solutions: You`ve made a number of large additions to your manufacturing facilities. Tell us a little about the assembly process. How much is automated? How much can you automate?

Itzler: I think you can break our operation into two fairly distinct pieces. Obviously as a photodiode manufacturer, wafer fabrication is one of our core technologies and, in particular, fabrication of indium phosphide devices. There is so much potential for indium phosphide as a material system-both for optoelectronics and for very-high-speed electronics-that we are seeing lots of growth in the basic processing technology.

The other major aspect of our current business is optoelectronic packaging. We have already made substantial progress in automating processes which resemble standard electronic packaging such as die mounting and wire bonding. Handling and aligning optical fibers is much more challenging, and one recent focus has been the development of techniques for automating the most critical of these operations. Ultimately, we believe that volume manufacturing of these products can and must be highly automated, but products have to be designed with this end in mind.

WDM Solutions: What sort of package or product do you present to your customers?

Itzler: At least from our division, we deliver components. We are working with other divisions within JDS Uniphase to come up with solutions that address customers` needs at a higher level. But as a division, we are focused on adding more value to the components that we make so that, for instance, the high-speed receivers accept not just the high-speed chip but integrate the analog electronics that are necessary.

We are increasingly involved in integrating optical functionality in front of the photodiode, so instead of having several separate components-a fiber tap, a custom filter, and then finally an optical monitor as a detector device-we are integrating the functions of all three components into a single device. That is the trend that our customers are looking for.

WDM Solutions: What is the next generation of component technologies that are needed?

Itzler: There is no question that photonic integrated circuitry is something that people are excited about. But then there is the reality of light being different from electrons-dealing with photons is nontrivial compared to electrons. I think the solutions that are going to come about most readily are those that involve an evolution through hybrid integration.

Now what the platform is for this integration is not clear. You can certainly imagine integrating a lot of functionality into indium phosphide. That is what our detectors are based on, that`s what emitters are based on, and you can put plenty of waveguiding and other light manipulating functionality in indium phosphide. There are efforts all over the world to do so. But of course there are other platforms.

If you can develop waveguides that essentially get light where you want it to go and do it on a silicon platform or perhaps in a MEMS-based platform, there is an infrastructure that exists for processing silicon-based devices that gives them a big leg up. I think the jury is still out on what platform is the right one and will be for quite some time. It`s going to depend a lot on the application.

WDM Solutions: What are you focused on in R&D?

Itzler: Company-wide there is a greater focus on integrating functionality, whether it be optical or electrical. There is a growing emphasis on development programs going on between divisions.

In terms of the kinds of challenges that we face, there is no question that there is tremendous opportunity given the way the industry is growing. I think one of the biggest challenges for those of us who are already in the market and experiencing tremendous growth-and it`s a theme heard fairly often-is keeping the focus on new technology development.

JDS Uniphase manufactures several types of detectors at its West Trenton, NJ, facility.