SEMICONDUCTOR LASERS: "Inverted" ridge suggests high power output

BOSTON, MA--A semiconductor laser with a novel "inverted" microstrip ridge structure shows potential for both high-power output and very-high-frequency operation at the important communication wavelength of 1.55 µm. The laser, developed by a team consisting of Daniel Tauber and others at the University of California, Santa Barbara, KDD (Saitama, Japan), AT&T Bell Laboratories (Holmdel, NJ), and the University of Delaware (Newark, DE) was reported at the OSA Integrated Photonics Research meeting in May.

The ridge-waveguide indium gallium arsenide phosphide (InGaAsP) quantum-well laser is fabricated with an unusual layer of gold between it and the InP substrate. The laser has an n-doped ridge and operates at 1.55 µm. Most InGaAsP ridge-waveguide lasers have p-doped ridges and an n-doped cladding layer below the active region. In the new device, the inverted geometry and the 1-µm-thick gold layer allow it to be driven continuous-wave at currents of more than 400 mA. This amount of current does not cause the severe heating that occurs in conventional p-doped ridge-waveguide lasers and limits their performance.

Tauber suggests that the laser will have applications in data communications and possibly as a pump laser for other devices. To date, the group has demonstrated threshold currents equivalent to those of conventional p-doped ridge lasers in 10-µm-wide ridge microstrip lasers but is still working to lower the thresholds for narrower ridge lasers.

Conventional metal-organic chemical-vapor-deposition (MOCVD) growth of devices with the p-doped side toward the substrate is difficult because zinc, which is the p-dopant, is highly mobile. Tauber says, "If you grow the p-layers first, [zinc] will diffuse up into the rest of the layers during the remainder of the growth and degrade the quality of the laser material."

The microstrip laser was fabricated by growing the layers for a conventional p-doped ridge quantum-well laser on an InP substrate. After growth, a thick gold layer is evaporated onto the grown structure. The gold layer atop the laser structure is pressed against another gold-metallized InP substrate at 300°C for four hours. This step fuses the gold layers together, after which an acid etch is used to remove the original substrate. The MOCVD-grown layers, now p-doped side down, are then processed into a ridge-waveguide laser. This technique could be used to create other n-doped ridge devices (that is, with the p-doped side down) on InP substrates.

The inverted geometry of the ridge waveguide alters several device characteristics, including the maximum power of the laser. Tauber explains that "n-ridge devices have both microwave and thermal advantages due to the capability of making low electrical resistance n-ridges. . . . The low resistance contributes to low electrical parasitics and minimal resistive heating in the ridge." The heat in the active region limits the laser output power. "If the heat generated in that ridge, which flows through the active region, is very small, this effect is minimized and higher power operation is possible," says Tauber.

The thick gold layer that acts as the bottom electrode for the laser also helps dissipate heat from the active region. This metal layer is novel because, using standard growth techniques, crystalline material cannot be grown on top of gold. "The gold fused layer also contributes to about a 10% reduction in the thermal resistance of the device," says Tauber.

Such n-doped ridge lasers also have advantages over p-doped ridge lasers for high-frequency modulation. The device has the microwave structure of a microstrip line, which offers good high-frequency microwave performance. Conventional structures tend to be grown on doped substrates, which attenuate high-frequency signals and can limit bandwidth at extremely high frequencies. Although these effects are largest in a matched transmission line system (and in general semiconductor lasers are not driven in a matched system), this issue is still important as researchers attempt to push the modulation bandwidth to higher and higher frequencies.

Tauber points out, "The highest-frequency lasers to date are built in the coplanar electrode geometry on a semi-insulating substrate," which is an alternate geometry that also solves the microwave attenuation problem. Further work will focus on optimizing and demonstrating the ultimate performance capabilities of the microstrip structure.