Abstract
This thesis is concerned with the improvement of semiconductor laser characteristics using valence band engineering. We first show that the combination of strain and quantum confinement can confer considerable advantages to long wavelength lasers. With sufficient built-in strain, the highest hole subband has a low effective mass and is well separated from the lower bands. The low effective mass reduces the carrier density needed for population inversion and leads to the virtual elimination of two important loss mechanisms: Auger recombination and intervalence band absorption. We propose a specific strained-layer 1.55mum structure that can reduce the threshold current density and its temperature dependence and increase the luminescent efficiency. The presence of strain can also lead to an enhancement of the relaxation oscillation frequency due to the higher differential gain when compared to lattice-matched structures. The linewidth enhancement factor is also predicted to be reduced. Such strained-layer lasers could be of major significance for long distance optical communication. However, the long term stability of these structures, although promising, has still to be fully assessed. In view of this, we suggest that (111) growth of unstrained structures could provide the light-hole cap to the valence band needed for laser operation. We find that the threshold current density in thin (111) lasers could be reduced while the polarisation selection of TE modes could be improved compared to equivalent (001) lasers. Finally, we consider the effects of crystal orientation and of strain on the exciton binding energy.