next up previous contents
Next: 4.1.2 The Realistic Device Up: 4.1 GaAs versus SiGe Previous: 4.1 GaAs versus SiGe

4.1.1 The Test Device

A device with a conservative design is shown in Fig. 4.1. This structure was used as a reference device in [158]. It was initially used to tune simulations of AlGaAs/GaAs HBTs with MINIMOS-NT and was later adopted for testing of SiGe HBTs.

The electrical behavior at room temperature of a Si BJT, a Si HBT with SiGe narrow-bandgap base, and a GaAs HBT with AlGaAs wide-bandgap emitter was studied in a comparative way using the same geometry and HBT typical doping profiles with high base doping concentration of $10^{19}$ cm$^{-3}$. The possibility to perform such simulations was presented in [195].

Later on in [124] a material composition optimization was shown. The optimization is automatically run using the VISTA framework [196] with ten operating points distributed at equal distances of 11 nm over the 100 nm thick SiGe base starting from constant 20% Ge content constrained to 25%. The possibility to increase the maximum current gain and cutoff frequency by material optimization can be seen in Fig. 4.2 and Fig. 4.3.

Figure 4.1: Simulated HBT test structure
\resizebox{\halflength}{!}{
\includegraphics[width=\halflength]{figs/hbt0.eps}}

The cutoff frequency $f_{\mathrm{T}}$ is determined using the quasi-static approximation, which can give results close to those from a small-signal analysis, as shown in [158]

$\displaystyle f_{\mathrm{T}}$ $\textstyle =$ $\displaystyle \frac{g_m}{2\cdot \pi \cdot C_\mathrm {in}}$ (4.1)
$\displaystyle g_m$ $\textstyle =$ $\displaystyle \frac{\Delta J_\mathrm {C}}{\Delta V_\mathrm {BE}}$ (4.2)
$\displaystyle C_\mathrm {in}$ $\textstyle =$ $\displaystyle \frac{\Delta Q_\mathrm {s}}{\Delta V_\mathrm {BE}}$ (4.3)

Thus, by applying small steps of $V_\mathrm {BE}$, $f_{\mathrm{T}}$ can be calculated by
$\displaystyle f_{\mathrm{T}}$ $\textstyle =$ $\displaystyle \frac{\Delta J_\mathrm {C}}{2\cdot \pi \cdot \Delta Q_\mathrm {s}}$ (4.4)

where $\Delta Q_\mathrm {s}$ is the change of the total charge in the device.

Figure 4.2: Current gain vs. collector current
\resizebox{\halflength}{!}{
\includegraphics[width=\halflength]{figs/gain0.eps}}

Figure 4.3: Cutoff frequency vs. collector current
\resizebox{\halflength}{!}{
\includegraphics[width=\halflength]{figs/ft.eps}}

Figure 4.4: Gummel plots at ${ \mathrm {V_{CE}}}$ = 2 V for Mod. 1 and Mod. 2
\resizebox{\halflength}{!}{
\includegraphics[width=\halflength]{figs/si5.eps}}

Figure 4.5: Current gain versus collector current for Mod. 1 and Mod. 2
\resizebox{\halflength}{!}{
\includegraphics[width=\halflength]{figs/si6.eps}}

Furthermore, the impact of new mobility and bandgap narrowing models of MINIMOS-NT is studied at different temperatures and for different dopant concentrations [66]. In Fig. 4.4 the Gummel plots for SiGe HBT at 77 K and at 300 K obtained with the model of Slotboom et al. [133] (Mod.1) and with our new model (Mod.2) are compared. The significant difference in the current density values at 77 K, resulting in a higher current gain obtained by the new model (Fig. 4.5), is experimentally confirmed.


next up previous contents
Next: 4.1.2 The Realistic Device Up: 4.1 GaAs versus SiGe Previous: 4.1 GaAs versus SiGe
Vassil Palankovski
2001-02-28