The activation of the -doping and the backside -doping, as given in Fig. 3.1, in the device depends mainly on the materials system and the thickness of the spacer between the doping and the channel [276]. In order to estimate the available carrier concentration to be assumed for the simulation in relation to the nominally amount of Si doping introduced by MBE growth, Fig. 3.24 shows an idealized function of the active of MBE grown nominal doping density. Without any processing for a given quantum well the active concentration rises linearly with up to 410 cm, after that the density saturates at values of about 5.510 cm. For bulk GaAs in [148] the active concentration is strictly proportional to up to 410 cm in GaAs independent of the growth temperature. When applying semiconductor processes to the gate area, effects reducing the carrier densities in the channel are found. As an explanation chemical or semiconductor surface induced mechanisms [122,127] have been suggested to effectively reduce the number of available carriers. It was found by inverse modeling for the processes simulated in this work, that if semiconductor processes with etch parameters reducing the activation are applied, is reduced by a fixed amount, rather than further reducing the slope of the active concentration in Fig. 3.24.
For the extraction of the backside doping inverse modeling suggests a doping efficiency of 40%-60% for the technologies pseudomorphic AlGaAs/InGaAs HEMTs. Thus, the backside doping is less efficient in comparison with the upper doping as given in Fig. 3.24, since it is subject to compensation mechanism from the backside buffer, e.g. due to acceptors [16].