7.6.1 Technology E: InP Based InAlAs/InGaAs Depletion Type HEMT

InAlAs/InGaAs HEMTs based on InP substrates are the fastest three terminal devices existing. However, there are several aspects to be accounted for in order to reliably use these devices in applications. Fig. 7.35 shows the transfer curve for a split channel ${\it l}_{\mathrm{g}}$= 150 nm In$_{0.48}$AlAs/In$_{0.66}$Ga$_{0.34}$As/InP device as a function of temperature ${\it T}_{\mathrm{sub}}$. For the applied ${\it V}_{\mathrm{DS}}$= 0.75 V, the device shows a DC- ${\mit g}_{\mathrm{m}}$ of nearly 1000 mS/mm. The transfer characteristics further shows a decrease of ${\mit g}_{\mathrm{m}}$ and ${\it I}_{\mathrm{Dmax}}$ with rising temperature.

Figure 7.35: Simulated and measured transfer characteristic as a function of temperature $T_{sub}$.

Figure 7.36: Output characteristics of a dual channel InP based pseudomorphic HEMT with $l_g$= 150 nm.

Fig. 7.36 shows the corresponding output characteristics and comparison with measurements for ${\it T}_{\mathrm{sub}}$= 300 K. For the ohmic contacts Case I in Fig. 3.25 is assumed in the model based on geometry information. Very good agreement is found for this device between simulation and measurements based on the model parameter values given in Chapter 3. As a basic result, the output conductance cannot be understood without taking electrons and holes and their generation/recombination into account. Four factors are found significant for this high gain device and to be responsible for the output conductance in this static simulation analysis. The transient behavior of the generation/recombination is not included.

• Surface potential/surface charges.
• Source recess geometry ${\it l}_{\mathrm{R}}$.
• Generation/recombination.
• Impact ionization.

The output conductance ${\it g}_{\mathrm{ds}}$ is often discussed related to the term ''kink'' [281]. The analysis performed for the simulations in Fig. 7.36 show that a kink is only a very special combination of the factors to influence the conductance ${\it g}_{\mathrm{ds}}$ [281]. The factors have their influence even if a ''kink'' is not visible at all. The surface potential for In$_{0.52}$Al$_{0.48}$As is only subject to minor changes due to the pinning of the Fermi level, if a stable surface treatment and SiN passivation is provided. The complete geometry of the source side influences the current flow. First, for the pinned surface potential at the SiN/InAlAs interface, a depletion region prevails near the surface at both source and drain, which attracts holes from the material background concentration and the manifold generation processes (SRH, impact ionization, Auger). This results in a change of the source side potential near the gate and thus in changes in the currents ${\it I}_{\mathrm{G}}$ and ${\it I}_{\mathrm{D}}$. Second, and especially relevant for the RF-properties, the depth of the recessed cap ${\it d}_\mathrm{R}$ (see Fig. 7.23), in combination with the available cap doping, determines the current path in the barrier layer and thus the source resistance ${\it R}_{\mathrm{S}}$. Since the current path is involved, this again depends on the source side ohmic contacting, see Fig. 3.25. The SRH generation/recombination significantly changes when increasing ${\it V}_{\mathrm{DS}}$ between 0.5$\leq$ ${\it V}_{\mathrm{DS}}$ $\leq$ 2 V. This is due to the fact, that the shape of the space charge region changes in the bias range 0.5 V$\leq$  ${\it V}_{\mathrm{DS}}$$\leq$ 1.5 V for constant ${\it V}_{\mathrm{GS}}$. The pronounced non-symmetric shape of the space charge region under the gate for normal operation develops, since normally ${\it V}_{\mathrm{S}}$= 0$\approx$  ${\it V}_{\mathrm{G}}$ $\ll$  ${\it V}_{\mathrm{D}}$.

Figure 7.37: Simulated and measured input characteristics $I_G$ versus $V_{GS}$.

Figure 7.38: Input characteristics $I_G$ versus $V_{GS}$ as a function of temperature $T_{sub}$.

Impact ionization in the InAlAs/InGaAs device influences changes the output conductance by the following factors: an additional contribution is added to the drain current ${\it I}_{\mathrm{D}}$ due to increase electron concentration. However, this contribution is small for the typical region $1 V$ $\leq$ ${\it V}_{\mathrm{DS}}$ $\leq$ $2 V$. The generation of holes to enhance the gate-current ${\it I}_{\mathrm{G}}$ and to cause an effective potential change near the gate is much more significant. This potential shift acts like a ${\it V}_{\mathrm{DS}}$ dependent positive shift of the gate potential and, once the accumulation of holes at the source rises, leads to relatively stronger shifts of the output conductance ${\it g}_{\mathrm{ds}}$ resulting e.g. in ''kinks''. An increase of impact ionization due to increased self-heating for channel materials with high In content is a secondary effect for higher ${\it V}_{\mathrm{DS}}$ bias. Fig. 7.37 shows the input characteristics for the ${\it l}_{\mathrm{g}}$= 150 nm device with the parameter ${\it V}_{\mathrm{DS}}$. A strong increase can be observed leading to the well known bell shaped curve. The gate currents are influenced by direct/SRH generation/recombination and by impact ionization. Already the non impact ionization generation processes (due to the direct band gap of InAlAs/InGaAs) do significantly influence the gate currents. Actually, due to the space charge region at the gate, for a detailed analysis, all generation/recombination processes must be considered. In a HEMT biased at ${\it V}_{\mathrm{GS}}$ at ${\it g}_{\mathrm{m,max}}$, depleted regions (barrier- SRH, direct recombination) are adjacent to highly doped region (semiconductor caps- Auger) and regions with high carrier concentrations and current flow (channel - Auger, impact ionization). As AlGaAs, InAlAs, and InGaAs are direct semiconductors for typical material compositions used in HEMTs, direct processes are considered. Fig. 7.37 and Fig. 7.38 show, that using the models presented in Chapter 3, the gate currents can be reproduced. Fig. 7.38 shows the simulated and measured input characteristics as a function of temperature. The gate current ${\it I}_{\mathrm{G}}$ increases with increasing substrate temperature ${\it T}_{\mathrm{sub}}$ for these devices, which is typical for an In content $x\geq$ 0.53. The reason is the positive coefficient of the impact ionization, which was discussed in Chapter 3 and Chapter 6, respectively. Fig. 7.39 shows the simulated and measured S-parameters of the InP HEMT between 2 GHz and 120 GHz. Parasitic elements have been obtained by the same procedure as for GaAs devices. Good agreement is found based on the eight-element circuit presented in Fig. 4.1. The additional elements presented in Fig. 4.2 are not accounted for in this low bias ${\it V}_{\mathrm{DS}}$= 1 V case, where high field effects do not dominate the S-parameters.

Figure 7.39: Simulated (-) and measured (+) S-parameters of a dual channel InP HEMT with $l_g$= 150 nm.

Figure 7.40: Simulated and measured $f_T$  versus $V_{DS}$  bias at constant $V_{GS}$ for the $l_g$= 150 nm dual channel InP based HEMT.

Fig. 7.40 shows the simulated and measured ${\it f}_\mathrm{T}$ as a function of ${\it V}_{\mathrm{DS}}$ bias at constant ${\it V}_{\mathrm{GS}}$. A strong decrease of ${\it f}_\mathrm{T}$ can be seen, which is based on three factors of influence. The increasing gate currents ${\it I}_{\mathrm{G}}$ shown in Fig. 7.37 decrease the current gain ${\it I}_{\mathrm{D}}$/ ${\it I}_{\mathrm{G}}$. Second, ${\it C}_{\mathrm{gs}}$ increases due to the occurrence of RST, i.e., electrons leave the channel into the spacer and barrier. Thus, the center of the channel charge moves closer to the gate, which leads to an increase of ${\it C}_{\mathrm{gs}}$. A third factor is the change of generation/recombination with rising ${\it V}_{\mathrm{DS}}$ due to the change of the extension of the space charge region. This enhances hole gate currents and leads to a potential shift at the gate. The holes lead to an effective positive shift of ${\it V}_{\mathrm{GS}}$ causing higher ${\it C}_{\mathrm{gs}}$. This contributes to the observed decrease of ${\it f}_\mathrm{T}$.