Characterizing and Controlling the Impact of Dislocations on the Reliability of InAs Quantum Dot Lasers
Dissertation Defense
Eamonn Hughes
WEDNESDAT, November 29th AT 10:00AM
ZOOM MEETING - Click Here to Join
Photonics is effectively the light-based analog of electronics, involving the manipulation of light at small scales, for applications ranging from high-bandwidth optical transceivers and optical computing to LIDAR and biomedical sensing and diagnostics. Integrating on a silicon platform, termed silicon photonics, is a key path to massively expanding the scale and reducing the cost of such devices. Due to its indirect band gap, silicon is largely infeasible as a light source, so the well-developed III-V material system often fulfills this role. One method to integrate the two is by directly growing III-V on silicon. This generates many defects due to material property mismatches, but many can be eliminated or mitigated through defect engineering. Quantum dot (QD) active regions help limit the impact of remaining defects, partly through their three-dimensional confinement of carriers and strain fields. Still, reliability of these devices, such as InAs QD datacom lasers, falls short of application requirements.
One key source of this limited reliability is the presence of previously unexplained misfit dislocations near strained indium-containing active regions. These are demonstrated to form by an unconventional mechanism, specifically during post-growth cooldown rather than as the layers themselves are grown, due to a combination of thermal stress from the substrate and dislocation pinning selectively in the active region. By simply extending the pinning effect beyond the upper and lower QD layers using thin strained indium-alloyed layers, termed “trapping layers,” the misfit dislocations can be displaced away from the active layers with >95% effectiveness. QD lasers with trapping layers exhibit substantially lower threshold current densities, higher slope efficiencies, and improved reliability by up to 100× compared to their standard equivalents. Trapping layers are also demonstrated to improve defect densities for a separate III-V/Si integration scheme involving heterogeneous bonding of a GaAs template on silicon followed by regrowth of a III-V laser.
Laser reliability on silicon with trapping layers and even on nominally defect-free native substrates, however, is still short of requirements at elevated temperatures. Degradation in these devices is shown to be driven by evolving point defect densities, which leave signatures as nanoscale dislocation loops visible in transmission electron microscopy. To address this gradual mode of degradation, additional post-growth annealing procedures along with laser fabrication changes should be pursued. Further gains may also be attained by optimizing trapping layer design to improve effectiveness and minimize carrier recombination at trapped misfit dislocations.
Committee Chair: John Bowers