Optoelectronics Perspectives

The optoelectronics industry is at a stage where there are several routes forward. As we drag ourselves out of the "tech wreck" and survey the telecom/datacom infrastructure business, we can see signs of recovery but no rapid ramp-up in the immediate future. This has a number of ramifications for the industry as it prepares to restaff and retool.

First, much of the future production will continue to use the existing technology — butterfly packages, fiber loops, splices, connectors. It's relatively expensive but flexible. Although automation of parts of the process is feasible and economic we will continue to have "islands of automation" in a manual assembly environment ... and that environment will increasingly be in Asia.

Second, we anticipate that transceivers and other components will increasingly use "pluggable" formats like Xpak where the optical assembly is separate from the electronic assembly. Combined with optical fiber harnesses or fiber-in-flex these promise significantly reduced costs.

Third, the work to embed optical circuits in boards will continue, but faces major technical challenges in fabrication and component reliability. Recent developments in copper circuit fabrication have seen dramatic increases in data rate capability which have raised the bar on economic and technical justifications for switching to embedded optics. Where we do see the first growth of integral optical waveguides is in the formation of modules or "supercomponents" where several active and passive devices are combined.

NEMI's work on "making optoelectronics mainstream" is appropriate to all these technologies — fiber management, signal integrity, splicing, selective soldering and board technology, plus their comprehensive roadmap review will be published in the near future. IPC has also been tremendously active with IPC-STD-040 and their roadmap (similar conclusions but from a different perspective).

So optoelectronics will continue to be a key technology in our datacom/telecom environment as that industry recovers and moves more bandwidth nearer the customer.

We tend to overplay one part of OE — telecom — but it's a big, varied industry. Most finally realize that telecom OE was the supernova that collapsed to a red dwarf. Some even know why. While it's cute to say that the "sun has set for photonics," the best days may be ahead — although not right around the corner. The other OE (often called photonics) is doing well, however. This includes sensing, imaging, data storage, detection and many other areas. The threat to our security has boosted many of the OE subsets in both the commercial and military sectors. Biomed is also hot as it adds more OE technology. But let's get back to telecom, since that's what OE means to most readers.

Just about everyone was working on long-haul and too many were doing "hero" experiments. The industry was having too much fun to ask for the customer's opinion.

It's unfortunate that a mid-course correction wasn't made when the warning signs were clearly there. Many from electronics were asking the right questions as they were trying to understand the technology and business. My favorite is, "Do you use design for manufacturability?" The OE guy would come back with, "Hey, I made it work, and you want me to build another one?" The problem then and now is still excessive cost that has not been designed out. Look at the fantastic bargains in electronics! You can buy a million transistors for the price of a candy bar (as an IC). It took electronics 25 years to build 1 billion personal computers, but it will take another five or six years to hit the 2 billion mark. This is all made possible by wringing out cost — not giving away the product.

Once the optoelectronic packaging and assembly cost is reasonable, we will have fiber-to-the-home and be willing to pay at least twice as much as for broadband cable.

Does it sound too simple? This is the model that built 1 billion PCs.

The optoelectronics industry is going through the simultaneous convulsions of dramatic contraction and technology changeover, making it very difficult to predict any market or technology outcome.

The consistent theme moving forwards is lower costs per managed bit. In some applications, that means higher speeds in the same footprint. In others, it means more wavelength-agile (tunable) devices. It also may just mean focusing on manufacturing processes in order to reduce costs, without radical redesign of the device itself.

Industry consolidation and the emergence of standards in both function, form factor and manufacturing processes will accelerate over the next two to four years to help the industry address new applications and profitably serve existing ones.

Bandwidth demand will continue to double approximately every year as a fundamental driver to the industry. Establishing a profitable means for the carriers to supply broadband to the home user is absolutely critical in every scenario for market success.

Both device and manufacturing technologies will play equally important roles as technological drivers that propel future growth. As in the semiconductor market, device integration is absolutely the best way to drastically reduce costs per bit. However, this implies the maturing of the preferred material system(s) (e.g. silicon, GaAs, InP), as well as their processing and packaging. This means standardization of process and packaging tools as well as material handling equipment. Many of these will be "borrowed" from the surface-mount and semiconductor industry, and adapted as necessary.

Device designs must evolve to take advantage of the well-established process equipment in these adjacent industries. For example, design for assembly from above- and conveyor-fed equipment (like surface mount) will be critical, vs. the complex optical assemblies we see today.

It is likely that the "best" device technology will not win, but rather that which gets engineered into the most manufacturable package first. So it is more important to look at who is doing the technology (and their manufacturing execution expertise) vs. what type of device technology they have.

The contract manufacturers will also play an important role in accelerating process development and standardization as more companies aggregate their relatively small volume demand under the roof of one outsource manufacturer.

Optoelectronic technology is a victim of its own success. The technology has become so powerful that high- bandwidth applications have not grown rapidly enough to support proliferation of increasingly high volumes of opto hardware.

As an example, let's consider a 10 Gbit/sec transceiver sytem (in the telecommunications trade called OC-192, currently the state of the art in deployed telecom systems). This data rate can support 200,000 telephone calls. During the boom of the late 1990s to 2000, it was assumed that higher-bandwidth applications would drive demand for rapid growth of this type of hardware. Although home Internet "broadband" applications are growing, the actual data rates are much lower than users expected. A telephone modem will claim it communicates at >50 kbit/sec and cable modem around 1 Mbit/sec. In reality, both of these metrics are overstated by more than a factor of 10. At these lower data rates, it doesn't take many OC-192 systems (only tens of thousands) to supply the needs of the United States.

Therefore, a key driver of demand of opto hardware is higher-bandwidth applications emerging in the consumer market at the home. Until true high- bandwidth (>1 Mbit/sec continuously) applications are delivered to the home, the demand for opto hardware will not be strong. The opto hardware industry needs this combination of high bandwidth and large volumes (e.g. 60 million homes in the United States) for robust growth.

The optoelectronics industry today is in the midst of a dramatic downturn due to the slowdown of telecommunication network expansion. The impact of this slowdown is amplified by the unprecedented telecommunications growth or "bubble" during the 1998-2001 timeframe.

The future growth of the optoelectronics industry is strongly tied to the rate of the deployment of broadband services deeper into the metropolitan region. New technology developments for enhanced performance, e.g., optical switching, increased DWDM channel densities, higher bit rates, tunable devices, and integrated optical devices, will certainly assist in this end. However, it is the technology/ products that meet or exceed the metro price requirements that will propel the optoelectronics industry.

Current levels of device performance are not the major constraint in limiting their application in the metro environment; instead, it is their high price that is the limiting factor. Independent of the performance level of optoelectronic devices, the technology that is able to drive down component costs is critical to enabling the growth of the industry, and the major cost contributor to optoelectronic devices continues to be the cost of packaging. Developing technologies that will drive down packaging costs is a requirement for future growth. Today, optoelectronic devices are still plagued by numerous manual operations, which drive costs up, not only due to the cost of labor, but also due to the cost resulting from the lack of product consistency. Automation is the common denominator for cost reduction, thus the technologies that can provide automated manufacturing solutions are critical to driving the industry forward.