Downturn brings opportunity for integrated InP devices
With capital expenditure dropping from $90 billion in 2000 to $70 billion in 2001, a shakeout of both technologies and suppliers is inevitable. Analysts at JP Morgan H&Q Equity Research have described the predicament that manufacturers of components for optical communication systems now face (see Compound Semiconductor July 2001, p69). Because there are a wide variety of technologies, there is no dominant manufacturing process, material system or generally applied set of core building blocks. Compare this to the semiconductor industry where silicon is the material, CMOS is the manufacturing process and transistors are the building blocks. In the supply-driven market that existed until the end of 2000, the lack of dominance by any material or technology was not a problem. Cost was less of an issue and so the incentive to change manufacturing processes in order to improve productivity was practically non-existent. In some cases prices even rose when in the semiconductor industry they might have fallen.
With network operators under great pressure to reduce investment and operating costs, the component vendors who supply them find themselves in a demand-driven market with a great incentive to optimize manufacturing processes and find that all important dominant technology and material.
Cost, footprint and power The tighter requirements of telecommunication system suppliers focus in three specific areas: cost, footprint and power consumption. The downward pressure on the unit cost of equipment is intensified by even more new systems being deployed in metropolitan and access networks, where larger quantities of systems and components are needed. Over the past year the focus has been mostly on long-haul connections: coast-to-coast, transatlantic and transpacific backbones. Now these backbones are in place (and often under-utilized) it is time to expand the feeder networks. Because of the large volumes involved this is a segment where every cent of unit price counts.
The volume factor creates a demand for physically smaller system sizes. Building space impacts investment and operational cost, especially in metropolitan environments. Network operators would like to see the systems that are currently filling their racks shrink to the size of a single system board. For this to happen boards will have to be reduced to packages, and packages in turn will have to fit on a single chip.
In some areas the third requirement, reduced power consumption, is even more important than the other two, and the cost of power is not the only issue. The recent energy crisis in California, as well as some incidents in Europe, are warnings of what can happen when the power supply falls out of step with an increasing demand for energy. In many large cities network operators are limited in their location selection by the availability of electrical power. Since an upgrade or expansion of the power grid can take many years, the only realistic way out is to reduce the power consumption of telecommunication systems.
One answer to the demands of today s telecommunication network operators is a technology that allows the monolithic integration of electronic functions as well as both active and passive optical functions. A growing force of industry watchers and players believe that indium phosphide is the most promising platform for subsystem-on-a-chip integration.
Indium phosphide fits the bill For over 20 years, researchers have been investigating practical methods to realize the benefits of InP. The first attractive feature of InP is that it allows the construction of extremely compact devices, which is an important consideration for mass production. Small devices can be realized in this material because of the relatively high refractive indices of InP and its ternary and quaternary derivatives, such as InGaAs and InGaAsP. A high refractive index allows a smaller radius of curvature leading to components that are at least 10 to 100 times smaller than current state-of-the-art silica-on-silicon technology.
Another characteristic of InP is its direct bandgap, leading to very easy and fast quantum transitions when photons are either absorbed or emitted. This, combined with its strong electro-optic effects, makes InP a suitable material for applications that require short switching times, such as packet switching.
It is fortuitous that the bandgaps of InP and its derivatives translate to the wavelength range that is used for optical communications. Furthermore, it can be tuned anywhere between 0.92 µm and 1.65 µm by varying the In(GaAs)P composition. This also influences its absorption and electro-optical characteristics, enabling the construction of transparent waveguide structures, photodetectors, modulators, switches, lasers and semiconductor optical amplifiers (SOAs) for a wide range of wavelengths. SOAs can find application not only as optical amplifiers, but also - because of the non-linear properties of InP-based SOAs - for all-optical signal regeneration, a function that is in high demand. Finally, the derivates of InP are extremely well suited for the creation of high-speed electronic devices.
Of course none of this is new. InP had proven itself long ago as a suitable material for both electronic components and lasers. But it was only recently that the technology required for integrating various components on a single chip has become available. An example of what can be achieved by integration can be seen in figure 2. The integrated WDM optical cross-connect (the right-hand image) is the world s smallest to date at 1.5 mm x 3.3 mm2 and was made in 2000 by the optoelectronic devices group at Delft/ Eindhoven University of Technology.
