Waveguide amplifiers bring integration benefits

Brian Lawrence

Michael Shimazu

There is little doubt that broadband optical amplification, enabled by the erbium-doped fiber amplifier (EDFA), has transformed the fiberoptic infrastructure, sparking the explosion in fiberoptic network capacity. As optical networks evolve to offer more functions, such as optical monitoring, dynamic reconfiguration and even optical packet switching, the role of optical amplification will also change. Amplifiers will evolve from the network elements they are today to components in a more function-rich environment.

This advance involves integration of multiple active and passive devices, first to simplify the amplifier gain block and make it less expensive, and eventually to combine amplifier gain stages with other devices to offer improved or perhaps novel optical functions. A new generation of waveguide optical amplifiers offers compact and functional platforms for this integration.

The current generation of optical amplifier has certainly been a success. The EDFA has eliminated the need for many regenerators on long-haul links, and thus made dense wavelength-division multiplexing (DWDM) an attractive economic proposition. As another example, this technology has made possible radical consolidation of head-ends in hybrid fiber-coax and similar systems.

However, focusing on these particular cases in some ways detracts from what might be the greater significance of optical amplification. By improving the budgets associated with each link in a network, amplification liberates the network designer to build more function and flexibility into the optical layer.

Building blocks

Current EDFAs are well-designed systems of optoelectronic and optical components-pump lasers, isolators, pump multiplexers, and others-but amplification is just one optical function that designers require. As networks grow in complexity and more function is sought at the optical layer, it will become more desirable for optical amplifiers to resemble self-contained devices rather than systems; devices that can become building blocks for larger, more capable systems and subsystems.

Just as op-amps or transistors provide gain in electronic circuits, so optical amplifiers will become gain stages in more complex optical circuits. For this to happen, for it to become feasible to populate optical networks with multiple gain stages, the cost of optical gain must be reduced.

The development path toward compact, lower-cost, more component-like amplifiers involves integration, and here is where waveguide optical amplifiers represent a step forward.

Erbium-doped waveguide amplifiers (EDWAs) are similar to EDFAs in that they use an erbium-doped glass as the gain medium. Instead of the coil of several meters of erbium-doped fiber found in EDFAs, however, EDWAs use a highly doped waveguide that may be as short as two centimeters in length.

Aside from slight differences resulting from what is often a different glass host, erbium-doped waveguide amplifiers are similar to EDFAs in their absorption and gain spectra. The excited-state lifetime of the erbium atoms in the specialty glasses used by waveguides is typically on the same order of magnitude as the erbium in conventional doped silica fiber, giving EDWAs a comparable high performance in noise figure and cross-gain modulation (see Fig. 1). EDWAs are optically pumped at the same wavelengths as EDFAs.

Consequently, an EDWA has the same parts as an EDFA. It incorporates a pump laser, pump multiplexer, possibly isolators, and a gain-flattening filter (see Fig. 2). Because of the small size of the waveguide and its substrate, it is feasible to integrate many of these components into a small, single package, either by free-space micro-optics or within other waveguide structures.

Gain-block design

In one new approach to a gain block, a laser diode chip pumps an erbium-doped waveguide via free-space coupling (see Fig. 3). The pump multiplexer consists of an appropriately designed thin-film filter, fabricated into the waveguide substrate, that reflects the pump energy into the waveguide while remaining transparent to the signal wavelengths. Input and output optics can incorporate isolators or gain-flattening filters (see Fig. 4). The size and shape of the waveguide itself is what makes this degree of integration possible.

The amplifier gain block is therefore a compact, micro-optically integrated component. This technology offers the potential for significant cost savings. In a manner similar to commercially available passive components that may combine pump multiplexers, isolators, or tap couplers into the same package, the elimination of several fiber pigtails from the finished waveguide amplifier product greatly reduces the cost and size.

Additional cost savings can be realized from the use of a free-space multimode pump diode. Not only are multimode diodes considerably less expensive than their single-mode counterparts used in conventional EDFAs, the use of free-space coupling to the waveguide eliminates the need to couple pump energy into a single-mode fiber for delivery. This fiber coupling can account for the majority of the cost of standard pump modules for EDFAs, but is eliminated in this waveguide amplifier design.

The use of multimode pumps requires some form of multimode waveguide structure to accept the pump energy. At the same time the signal must remain in a single mode; normally the signal can be coupled to a low-order mode of the waveguide and be easily picked up at the waveguide output.

Multimode pumps can contribute more than their low cost to the economy of waveguide amplifiers. They also offer very high output powers, on the order of several hundred milliwatts. Consequently, a multimode waveguide amplifier has a high saturated-output power, making it suitable for applications where amplification of multiple optical channels is required, and affording a low cost per dB-channel.

System applications

Although their inherently shorter path lengths may limit the amount of gain that can be had from EDWAs, such devices can have substantial utility in those applications where granular or distributed gain is desired. One such application is in metropolitan DWDM systems.

Unlike long-haul links, where all signals pass through the same path and it is therefore desirable to have a small number of high-gain amplifiers, metro systems are characterized by multiple signal sources and multiple paths. In such an environment it may be desirable to have smaller chunks of gain available at multiple sites throughout the network, rather than concentrated gain at a single point. Here, the cost per dB-channel becomes a useful figure for comparison of amplifier alternatives.

So far we have discussed ways in which integration of the conventional parts of an optical amplifier can be economically achieved with a waveguide amplifier, making the amplifier more like a discrete component than an optical system. Beyond this, the further integration of other optical functions within the same compact package makes the amplifier more like a gain stage within an optical circuit or subsystem.

One option for integrating further function with waveguide amplifiers is to write other waveguide devices into the same substrate. It is easy to envision splitters, arrayed-waveguide-grating (AWG) devices, variable attenuators, switches, and even modulators monolithically integrated with the amplifier. Of course, such a strategy requires that the fabrication processes for these devices are compatible with the amplifier gain medium.

Free-space opportunities

Micro-optics provides another route to integration. Unlike EDFAs, the use of EDWAs requires free-space coupling from the fiber to the waveguide. This arrangement presents an opportunity for free-space integration of multiple functions around the waveguide gain stage.

Such functions may be passive, such as optical filtering, tapping, switching, branching, and wavelength separation. Passive devices could include thin film filters, microelectromechanical systems (MEMS), gratings, and even planar integrated optical circuits.

Some novel active components may also benefit from free-space integration with an optical gain stage. For example, many tunable lasers offer lower powers than their fixed-wavelength counterparts. The same is true of vertical cavity lasers. Integrating a compact waveguide amplifier with these sources in a master oscillator-power-amplifier architecture could yield economical and highly functional components. Similarly, high-speed photoreceivers often exhibit a lower responsivity than lower-speed devices. An integrated component with an optical preamplifier interposed between the incoming signal fiber and the receiver could become useful.

The size and shape of rare-earth-doped waveguides makes them amenable to free-space micro-optic integration with other active and passive devices. In the near term, this makes it possible to produce EDWAs that resemble components in size, cost, and function. Additional integration may be feasible and would represent a next stage in the evolution of optical amplifiers from systems to gain stages in more integrated functional units.

FIGURE 4. Free-space isolators and thin film components can be built into micro-optic trains, adding functions at a low cost.