InP-based InGaAs/InAlAs pseudomorphic HEMTs (pHEMTs) were long considered to be the best devices for high-frequency communication applications because of their excellent RF and low-noise performance: e.g. in the beginning of the decade structures with a gate length of 25 nm, 45 nm and a cut-off frequency of 526 GHz, 400 GHz, respectively, were reported [71,72]. However, InP pHEMTs have some issues, including high cost of InP substrates, low mechanical stability, and last but not least poor breakdown performance due to the low band gap of the InGaAs channel. An alternative are GaAs pHEMTs with lower In content in the channel, and therefore superior breakdown performance but limited by the lower mobility and velocity saturation. The advantages of both technologies are combined in metamorphic HEMTs. Those use a strain-relaxed, compositionally-graded buffer to accommodate the lattice mismatch between the substrate and the top layer. The technology offers very good RF performance, the lowest noise figures, and high gain performance. Aggressive gate scaling and several optimization techniques such as zig-zag formed T-shaped gate electrodes helped to push the cut-off frequency to 440 GHz [73,74] (see Fig. 2.8). Interest in these devices is still strong and numerous efforts are devoted to improve the performance. As an example, Su et al. suggest a dilute antimony channel in order to improve the interface quality and channel confinement . Another focus point of research is the design of an enhancement mode device, which faces several issues. The narrow-gap channel enables high impact ionization rates, which combined with the high Schottky gate leakage current limit the input dynamic range and increase the noise figure. Therefore, attempts to restrain the gate leakage current in E-mode devices through MOS structures are ongoing . An advantage of the MOS technique is the improved RF performance of the devices due to the lower gate-source and gate-drain capacitances .