8. Conclusions and Outlook

Two-dimensional device simulation of High Electron Mobility Transistors based on the pseudomorphic AlGaAs/InGaAs/GaAs and InAlAs/InGaAs materials system was the investigation's object in this work. The simulation actively supports the development of several HEMT processes using the device simulator  MINIMOS-NT, while mass production on 6 inch wafers is introduced and 8 inch substrates are under development for AlGaAs/InGaAs/GaAs HEMTs with ${\it l}_{\mathrm{g}}$ as low as 120 nm. For pseudomorphic high power devices, especially for the Ka-frequency band, predictive simulation studies are performed for large scale processes. The development is supported by active load-pull measurements. Special emphasis was paid to the investigation of statistical process variations with respect to the device optimization. For simulation of AlGaN/GaN HEMTs material models have been developed and first simulations have been performed, which show good agreement with measurements. For the future of process development, pseudomorphic HEMTs will strongly compete with the silicon based RF-devices. A major issue of further development will be process simplification and stabilization in order to reduce costs while preserving or even improving device performance. Metamorphic HEMTs for industrial processes are under development for low noise and high gain amplifiers. High-speed InAlAs/InGaAs HEMTs for data rates of 80 Gbit/s and beyond are being developed for the next generation optical data transmission. For sensing applications at and beyond 94 GHz, i.e., in the next atmospheric windows at 140 GHz and 220 GHz, reliable low noise and power processes with gate-lengths below ${\it l}_{\mathrm{g}}$ $\leq$ 100 nm are to be developed.

Possible directions for improvement of GaAs based III-V device simulation are the understanding of breakdown considering dynamic trap occupation. These dynamic issues of breakdown need to be addressed similarly to approaches for silicon, e.g. for electrostatic discharge. Further modeling of the III-V quaternary materials is required for InAlGaAs, InGaAsP, and AlGaAsSb. More knowledge is required for the understanding of the process influence on the SiN/barrier interface. For GaN based HEMTs more specific high field models need to be developed for the hydrodynamic transport based on MC models and improved material characterization. Spontaneous and piezoelectric polarization effects require further understanding. This includes scattering at the channel interfaces, in AlGaN barrier layers, and the metal-semiconductor contacts. The high field mechanisms for transport in the AlGaN/GaN barrier layers need to be understood in order to control and stabilize the resistances ${\it R}_{\mathrm{S}}$ and ${\it R}_{\mathrm{D}}$. The impact of surface passivation requires further investigation. For ultra high-speed low power devices the development of simulation tools for InP based HBTs and InAs/AlSb HEMTs is required. Furthermore, the simulation of S-parameters for the HBTs with InGaP/GaAs needs to be further developed. For InP based HBTs the existing model base needs to be applied parallel to the continuing process development. More computationally efficient three-dimensional electro-thermal device simulators will lead to a closer connection towards chip design, especially when thermal effects are addressed.

Although a variety of simulators have successfully demonstrated formidable agreement with measurements, the understanding of changes of transport and interface parameters due to the influence of specific process steps remains the ultimate goal of process control.