7.1.1 Wireless Applications

Especially for devices being in mass production, possible variations of the manufacturing process are most desirable to be understood. Fig. 7.5 shows the variations of the threshold voltage ${\it V}_{\mathrm{T}}$ as a function of upper $\delta$-doping concentration relative to a nominal value for a gate length of ${\it l}_{\mathrm{g}}$= 440 nm in a pseudomorphic AlGaAs/InGaAs/GaAs HEMT. It was found by simulation, that ${\it V}_{\mathrm{T}}$ is especially sensitive to the concentration of the $\delta$-doping plane directly in the gate area. In this simulation, for the deviations from the nominal value, an activation ${\it n}_{\mathrm{ac}}$/ ${\it n}_{\mathrm{nom}}$ is assumed to be unity.

Figure 7.5: Threshold voltage $V_T$ versus $\delta$-doping concentration $l_g$= 440 nm.

Figure 7.6: Saturation current $I_{Dmax}$ as a function of $\delta$-doping concentration for $V_{DS}$= 2 V.

According to Fig. 7.6 the saturation current is proportional to the $\delta$-doping. The current ${\it I}_{\mathrm{D}}$ reacts rather sensitive. The comparison of Fig. 7.5 and Fig. 7.6 allows the separation of effects. If a specific variation of ${\it V}_{\mathrm{T}}$ and ${\it I}_{\mathrm{Dmax}}$ in a wafer map is observed independently, a distinction of the contribution from ${\it n}_{\mathrm{sheet}}$ and further factors can be performed. In this case we would attribute the corresponding $\Delta$ ${\it V}_{\mathrm{T}}$ related to the change of ${\it I}_{\mathrm{Dmax}}$ and render other factors, such as changes in the gate-to-channel separation ${\it d}_\mathrm{gc}$, responsible for the unexplained part of the shifts of ${\it V}_{\mathrm{T}}$. In combination with statistical analysis given in Chapter 5, this explains the observed variations of a wafer map.