4.4.2 Device Reliability

Based on these investigations it is now possible to explain the base current degradation of an InGaP/GaAs HBT which is strongly stressed under conditions far from normal operating conditions. In this case the base current degradation in the middle voltage range can be explained by a decreasing surface charge density along the interface between ledge and passivation from $10^{12}$ cm$^{-2}$ to $4.10^{11}$ cm$^{-2}$. This might be due to compensation of the negative surface charges by H+ ions which are known to be present in the device due to the epitaxial manufacturing processes [205,206]. In Fig. 4.35 a comparison of measured and simulated forward Gummel plots at V $_{\mathrm {CB}}$ = 0 V is shown. Filled and open symbols denote measured characteristics of the non-degraded and degraded device, respectively. The lines show the corresponding simulation results. The good agreement also for stressed devices demonstrates

the applicability of physically-based device simulation to device reliability issues. The electron current density corresponding to a surface charge density of $4.10^{11}$ cm$^{-2}$ is presented in Fig. 4.36. Fig. 4.37 shows the corresponding electron distribution in the ledge at x = 1.6 $\mu$m, 2.0 $\mu$m, and 2.4 $\mu$m, and the hole distribution at x = 2.0 $\mu$m. Note that the upper part of the ledge is now not completely depleted, thus allowing a base leakage current.

Several other effects supposed to lead to a strong increase in the base leakage current, e.g. spreading out of the base contact at the metal/GaAs interface, increased recombination-generation in the InGaP layer, degradation of the SiN/GaAs interface (see e.g. [207], [208] and references therein) are also analyzed. The simulation results show that such effects cannot be the dominant reason for degradation of the current gain. The decrease in the collector current at high level injection is suggested to be due to increased emitter resistance which could occur due to emitter contact detachment, indium segregation in the metal layer, or dislocations at the InGaAs/GaAs interface (see e.g. [208]). Our simulations show, that contact detachment leads to an electron current crowding in the remaining contact area which leads to insignificant changes. Only an emitter contact detachment of more than 80%, which is slightly probable, can explain the measured values (see Fig. 4.38). Indium segregation in the metal is found to be a possible reason, as the emitter contact resistance increases, while the decrease of the indium content in the cap has no significant influence on the emitter resistance.

Figure 4.35: Forward Gummel plots at V $_{\mathrm {CB}}$ = 0 V: Comparison between measurement (symbols) and simulation (lines) before (filled) and after (open) HBT aging.

Figure 4.36: Electron current density [A/cm$^2$] at V $_{\mathrm {BE}}$=1.2V: Simulation with a surface charge density of $4.10^{11}$ cm$^{-2}$

Figure 4.37: Electron and hole distribution in the ledge: Simulation with a surface charge density of $4.10^{11}$ cm$^{-2}$

Figure 4.38: Electron current density [A/cm$^2$]: Simulation of emitter contact detachment