3.6.1 Statistical Analysis and Data Extraction Scheme of Devices

The extraction of simulation parameters performed in this chapter is based on long term statistical analysis. It relies on several different devices from different processes, as described in Chapter 5. In [50,51] a data extraction scheme was presented which is used and further developed: One consistent set of parameter values was found to describe various HEMT types based on the same technology. In this work simulation parameters are not extracted from one process or even measurements from one device and singularly fitted to high precision, but from a number of processes, that yield a mean value and ranges for deviations. Thus, several different technologies can be simulated.

As the most important transport parameter, the mobility, is strictly correlated to typical measured carrier sheet densities ${\it n}_{\mathrm{sheet}}$. Data for ${\it n}_{\mathrm{sheet}}$ versus In content $x$ in the channel are compiled in Fig. 3.23. The data are taken from GaAs based pseudomorphic AlGaAs/InGaAs HEMTs for $x$$\leq$ 0.3, from metamorphic InAlAs/InGaAs HEMT for $x$$\leq$ 0.53, and from lattice matched and pseudomorphic InAlAs/InGaAs/InP HEMTs for higher $x$ values. Double channel HEMTs enter the figure by an average value [322].Changes of the mobility data sheet densities ${\it n}_{\mathrm{sheet}}$ during MBE growth are estimated from Hall mobility measurements, so that typical shifts are well known due to a high throughput of MBE grown wafers.

Figure 3.23: Typical carrier concentrations versus channel In content in In$_x$Ga$_{1-x}$As HEMTs obtained by Hall data at ${\it T}_\mathrm{L}$= 300 K.

Figure 3.24: Active doping density $n_{ac}$ versus nominal doping density $n_{nom}$.