During the "gold-rush" years, life was sweet in the optical components sector. But then the bottom fell out of the telecoms market and things turned very sour, very quickly. For the past 12 months, the reality confronting component makers has been one of plummeting stock prices, inventory write-offs, multibillion-dollar losses and tens of thousands of redundancies.
Now, as the industry struggles to regain some semblance of equilibrium, it's worth remembering that significant opportunities remain. Specifically, price has emerged as the big differentiator among the massed ranks of suppliers. At the same time, more vendors acknowledge that packaging - which can equate to 80% of the cost of many components - holds the key to competitive advantage in this new-look market-place.
What standards?
In the long term, standardization must be seen as a critical enabler for the efficient manufacture of optoelectronic devices. Photonics should take its cue from the IC packaging industry, where standards bodies and trade organizations have been working together for over 30 years to develop common package types and assembly methods.
The resulting plethora of IC packaging standards fostered the development of high-volume automated-assembly equipment. After all, once component makers agree on their packaging requirements, the automated-assembly equipment only has to be designed once, rather than separately for each manufacturer.
The logic is hard to argue with: the cost of packaging for semiconductors is typically less than 5% of the total component costs. In contrast, there are no widely agreed package standards for photonic components. Even with de facto "standard" component packaging, such as the 14-pin butterfly package employed for laser (source and pump) transmitters, different suppliers often use different input/output interconnections.
That said, there's growing recognition of the importance of packaging standards. Research carried out by BPA Consulting identifies four main drivers:
• multisource agreements (MSAs)
• dominant suppliers
• trade and industry organizations
• component customers.
An MSA involves a group of suppliers agreeing on standard physical and electrical parameters for a specific component or component family. Currently, there are at least 10 different MSAs in use across the optical components market, all of which give prospective customers access to multiple supply lines. Another big plus is that MSAs can result in de facto standards for components. A good example is the small-form-factor MSA for 2.5 Gbit/s transceivers, which was agreed in January 1998.
It's also worth pointing out that MSAs allow vendors to outsource production to other MSA members, after which they can rebadge the product. In turn, this critical mass of suppliers can create new markets, so that ultimately the MSA can become the basis for future standards.
Put simply, established companies involved in MSAs can use the opportunity to become market leaders, while smaller MSA members can find it easier to compete with the main players. For these reasons, an MSA for 40 Gbit/s transponder components is already in place, despite the members' expectations that actual products will not be available in any significant volume until 2003.
To date, however, the main standards on photonic packaging originate from Telcordia (formerly Bellcore) of the US, and largely restrict themselves to the reliability and environmental requirements of fibre-optic components. Complying with these standards is a requisite for virtually all photonic packaging, although implementing the recommendations does drive up costs considerably.
Most notably, the Telcordia standards may not be realistic for very-low-cost applications, particularly for those devices with shorter lifetime expectations. With this in mind, a number of organizations, including the IPC (formerly the Institute for Printed Circuits) and the Telecommunications Industry Association in the US, are looking at the creation of revised reliability specifications for low-cost packaging applications. Research by BPA suggests that this trend will gain momentum, ultimately leading to new industry-wide reliability and packaging standards for many metro and FTTx devices sometime in 2003.
The right package?
Change is evident across all sectors of the components business - whether it's low-end discrete devices, value-added subsystems, or the packaging that goes with them. Integration of multiple components within modules is becoming commonplace, with MSAs now in place for 2.5 and 10 Gbit/s transponder modules that include all of the transmit and receive, optical and electronic driver components within the same housing. The incentive is that, in addition to reducing the length of the electrical interconnects, integration eliminates two other packages (for the modulator and driver electronics).
When it comes to very-low-cost components, semiconductor-style packaging is likely to be significant. This involves the use of techniques such as kovar leadframes (alloy structures of electrical connectors) onto which multiple devices will be assembled in one pass. In this way, devices will be packaged and protected using epoxies that will be over-moulded on the leadframe.
Back at the other end of the scale, manufacturers of next-generation 40 Gbit/s components will need a new take on packaging technology. At 40 Gbit/s transmission speeds, the electrical components within photonic packages will be in the microwave operating regime. This means that parasitic inductive, resistive and capacitive effects all have to be minimized. It's also expected that wire-bond interconnections will be exchanged for ribbon bonds, which are rectangular in cross-section and help to reduce the effect of microwave cross-talk. The only alternative is to eliminate wirebonds completely within the package and use flip-chip techniques to interconnect the semiconductor devices.
Integration will also come into play. Although the first 40 Gbit/s components will be discrete devices (such as lasers and modulators), the trend towards module-level solutions will become all-important. The upshot of this is that the 14-pin butterfly package will eventually give way to a 6- or 7-pin package with an SMC (multifibre) connector.
Within the next five years it's inevitable that component makers will look beyond the MSA model and drive through standards on all aspects of photonic packaging. Cost reduction, multisourcing and the trend towards automated manufacturing are just a few of the factors that make this unavoidable.
• Photonic Packaging - Global Trends and Markets 2000-2005 is published by BPA Consulting. For further information, email Sarah Demmon, direct marketing manager.
• This article originally appeared in FibreSystems Europe March 2002 p29