NEGF Analysis of Tunnel Junctions for Long-Wavelength Emitting VCSELs

Many gases and polluting elements exhibit absorption lines in the mid-wave infrared, thus light sources emitting in this range are sought after for sensing applications. Although it is possible to reach such long wavelengths with edge emitters, their large power consumption is not compatible with the realization of portable devices. On the other hand VCSELs, already consolidated in smartphones for near-IR sensing, would be an excellent fit. Increasing the wavelength requires designing new VCSELs, based on low-gap materials such as InAs/GaSb alloys, although introducing a number of problems concerning current injection/confinement and electro-opto conversion. For the transverse confinement, a viable replacement of the oxide apertures of AlGaAs VCSELs is given by buried tunnel junctions (BTJs). The first part of the thesis is aimed at investigating injection in GaSb/InAs VCSELs by focusing on the BTJ computer-aided design. Usually, the standard simulation technique used by most of the commercial TCAD programs is based on a drift-diffusion (DD) solver, but this cannot consider rigorously any quantum effect such as band-to-band tunneling (BTBT). On the other hand, the non-equilibrium Green's functions (NEGF) approach is one of the most rigorous quantum transport pictures available, but a simulation of an entire device would be too memory- and CPU-intensive. The strength of this work lies on the development of a self-consistent NEGF-DD solver. This is based on applying NEGF only to the BTJ. Then a generation rate describing BTBT is obtained, which is finally plugged into the DD solver, applied to the entire device. The preliminary BTJ simulations show that the reachable currents are higher than the ones obtained with standard AlGaAs, due to the lower band gap and a more evident overlay between the conduction and valence band edges, specifically due to the use of a type-III broken gap heterojunction p+ GaSb / n+ InAsSb. Moreover, since the BTJ is heavily doped, the impurity scattering can narrow the band gap (BGN), which was simulated as well by extrapolating data from the existing literature for GaSb. Once a reasonable VCSEL structure has been selected, in the second part of this work we analyse the carrier transport behaviour as a function of the doping profile, even though there were convergence issues when the original BTJ doping, i.e. 2e19 cm^{-3}, was employed in such a complex structure. Indeed, having the best electrical properties available might affect the optical absorption of the device due to free-carrier induced absorption. An important issue which can strongly help the VCSEL optimization is to understand how much and where the doping density can be lowered without affecting significantly the current injection. More specifically, to complete such a doping analysis, many NEGF-DD simulations have been performed by gradually lowering the doping, either in the entire device or just in the BTJ portion. By analysing the band diagram of the device at -2 V, no significant changes can be noted when just the BTJ doping is varied, however, if the doping is uniformly reduced over the entire VCSEL, a barrier arises due to a significant voltage drop right after the BTJ. From the current density at -2 V standpoint, the two mentioned analyses are then compared. It is clear that a doping reduction in the BTJ alone does not worsen the electrical properties, but it might greatly lower the optical absorption in the VCSEL.