Fabrication and evaluation of In0.52Al0.48As/In0.53Ga0.47As/InP quantum cascade laser
InP-based quantum cascade (QC) lasers were fabricated following the design reported by Faist et al. with the initial aim of applying them in gas sensing applications. The lasers were characterised by our collaborators, Cockburn et al. Unfortunately, the performance of our QC lasers (QCLs) did not live up to expectations, and the gas sensing objectives became replaced with resolving the issues of their poor performance. This was achieved through a mixture of laser fabrication, characterisation, and optical and thermal waveguide modelling. The devices fabricated included mesa-etched QCLs, shallow-etched QCLs, and novel native-oxide defined QCLs.2D thermal modelling using a commercial finite element modelling package was carried out to solve the 2D non-linear thermal diffusion equation for all of the structures listed above, as well as for InP-clad and buried heterostructure configurations. The temperature elevations, distributions and heat flow vectors were calculated under high, but not unrealistic thermal power generation in the active waveguide core. The relative effectiveness at dissipating heat was judged using these results. The modelling indicated the presence of high temperatures and thermal gradients across the active waveguide core under continuous wave (CW) operation. The thermal resistance derived through the use of the thermal modelling agreed very well with that calculated by Faist et al. from experimental data. The optical modelling, which was instrumental in resolving the anomalous behaviour of our QC lasers, comprised of two parts. First, modelling of the dielectric permittivity gave values of n and a for InP, In0.52Al0.48As and In0.53Ga0.47As as a function of the free-carrier density and wavelength. The calculations were made using single and multiple-oscillator models with a free-carrier contribution in the form of a classical Drude expression. The dependencies of the electron mass and electron mobility on the free electron density were taken in the calculations. These values were used to perform 2D optical modelling of the waveguide using a commercial, fully vectorial waveguide mode solver. This yielded the effective index, confinement factor, facet reflectivity, waveguide loss and far-field distribution for each mode. These, in turn, were used to calculate the threshold gain for each mode. Perhaps the most important findings were the prediction of the existence of higher order transverse modes, that these modes can have low values of gth (comparable or better than that of the fundamental mode of the waveguide) and that the collection efficiency will vary dramatically from mode to mode. The modelling also indicated that the performance of the QCLs would suffer greatly with an increase in the InP substrate doping level, even just from 1x1018 cm-3 to 3x1018 cm-3.