using a second-nearest neighbor sp s tight-binding

A ten-band sp3s* second-nearest neighbor tight- binding model has been used to model the electronic structure of various AI,Gal-IAs quantum cascade laser gain media. The results of the simulations have been compared with experimental emission wavelength data, and it has been shown that the model is able to predict bounds on the photon energy at the peak in the gain coefficient spectrum to within at worst 21% of the experimental value. It is believed that the accuracy of the predictions can be improved by better analysis of the electronic structures. Comparison of the results of the calculations with results from a two-band k.p model shows that the tight-binding model is able to find the X-like states simultaneously with the I'-like states. Two methods have been used to estimate the electric field at laser threshold and it is found that neither method offered any substantial advantage over the other. The effects of increasing and decreasing all the layer thicknesses in the gain medium by one monolayer have also been investigated. The main application of QCLs is gas sensing since they have been made to emit at wavelengths in the range of at least -3.5 pm 141 to -67 pm (5), which overlaps the region of the electromagnetic spectrum containing molecular absorption bands. Optimization of the maximum operating temperature, threshold current, output power and careful control of the emission wavelength range are required to exploit fully the capabilities of QCLs in this area. However, the large space available for QCL gain medium design means that an accurate tool is required to select those designs that should be carried forward for expensive and time-consuming growth, fabrica- tion, testing and, ultimately, production. This paper presents the results of simulations performed using the NanoElectronic Modeling 3.0.2 software package (NEMO) 161, which is a candidate for such a tool. NEMO was developed by the Applied Research Laboratory of Raytheon TI Systems and others as a comprehensive quantum device mod- eling package, and is based on the non-equilibrium Green's function formalism. Previously, NEMO has been used for the simulation of resonant tunneling diodes (7)-(9). Here, one of its sp3s* tight-binding models has been used to make predictions of the photon energies, Epeak, at which the gain coefficients of several three-well GaAs/AI,Gal-,As QCL gain media are maximized. The predictions are then compared to experimental results. With a few caveats (lo), ( 111, sp3s* tight-binding models offer the possibility of modeling the electronic structure of a 111-V heterostructure where transport can take place via any valley. They are also able to model accurately the conduction band non-parabolicity for r-like states. Both of these capa- bilities are potentially important in a QCL, where quantum confinement pushes the resonant states far above the bulk conduction band edge of the well material. The atomic-like basis states used in a tight-binding model should be better suited to modeling the electronic structure of a QCL than the bulk basis states used in a k . p model. The latter set of states should be reserved for modeling heterostructures with layer thicknesses much greater than a monolayer, where the electronic structure is only weakly perturbed from that of the bulk. 11. NOMENCLATURE