1. Introduction
AlGaN/GaN HEMTs have shown exceptional promise for high-frequency, high-voltage, and high-power applications[1, 2]. However, the high power dissipation of GaN-based HEMTs can result in substantial self-heating effects and parasitic effects induced by traps, which have detrimental effects on the electrical properties[3-5]. Drain current lag and collapse are major factors limiting these applications. It has been widely accepted that self-heating and traps are the two reasons for the drain current lag and collapse[6-9]. Information about the effect of traps and self-heating on the drain current transient response and the contribution of each factor is crucial for the optimization of the performance of AlGaN/GaN HEMTs.
The self-heating related thermal effects have been simulated by Turin et al.[10], Hu et al.[11] and Zhang et al.[12]. Meneghesso et al.[13] studied the trapping phenomena by changing the ambient temperature. The relationship between the channel temperature rise and drain current is not investigated in detail. Hu et al. [9] studied the thermal effects on drain current lag and current collapse by 2D numerical simulation. However, studies that include both the buffer trapping effects and the self-heating in the drain current transient response simulation are still lacking. Although Miccoli et al. [8] simulated the trapping effects and thermal effect on the drain transient response of AlGaN/GaN HEMTs, lattice heat flow simulation and electrical simulation are separated, their method only considered the process of heat conduction and did not consider electro-thermal coupling.
In this paper, we present the physical-based 2D numerical simulations of trapping and self-heating in the drain current transient response under drain turn-on pulse voltage. The intrinsic mechanisms of traps in GaN-buffer and self-heating effects on the drain current are investigated by trapped charge density evolution under the pulse-on drain bias. In addition, the contribution of traps in GaN-buffer and self-heating effects are analyzed through channel electron density.
2. Device description and physical models
The schematic of the device structure for the simulation is given in Fig. 1(a). For the device, the thickness of intrinsic GaN and Al
The physical-based two-dimensional numerical simulation is performed by Sentaurus Device software, which is developed by Synopsys Inc. The thermodynamic model (also known as non-isothermal model) is applied to calculate the self-heating effect. The thermodynamic model is based on the drift-diffusion model and the effect of temperature gradient on the carrier transport is taken into account, which can be used to calibrate the uneven heat distribution from the self-heating effect. The equation of lattice heat flow is included, as are the basic Poisson equation and the continuity equation. The current of the device end and lattice temperature can be obtained from the self-consistent calculation of the above equations. Other related models are also used, such as the mobility model caused by surface scattering, field-dependent drift velocity model, the SRH (Shockley's-Read-Hall) composite model etc.
The traps of GaN buffer layer and AlGaN barrier layer are used in the simulation because they play an important role in charge transfer. The traps density of GaN is
There exists a thermal boundary resistance between the GaN and the SiC [14] because the real interface between the epitaxial film and the growth substrate usually contains nucleation layers with a high density of defects and impurities. Here, we use a rather low thermal conductivity of 3.7 W/(K
The material parameters[8-10] for the simulation of transient temperature are listed in Table 1. Default values are used for other material parameters in the software.
![]() |
Mixed-mode simulations are selected because the device is operated under the pulse on mode. The diagram of the circuit connection is shown in Fig. 1(b). The drain is applied with a voltage source, which is pulsed from 0 V to
3. Results and discussion
Figure 2 shows the simulated
The drain current transient response of the AlGaN/GaN HEMTs device under different
Besides the channel temperature, the drain current is also affected by the traps in the buffer. Figure 3 shows the simulation of
Since the traps in buffer are under trapping and de-trapping, they will affect the channel electron density. For this reason, the trapped charge density and channel electron density at the drain side gate edge are calculated. The inset of Fig. 3 shows the spatial distributions of trapped charge density evolution under the pulse-on drain bias. It is can be seen that the thickness of fully filled by channel electrons increases when the drain voltage is pulsed from 0 to 10 V. The percentage of deep traps filled by electrons increases with time. The trapped charge density does not change in a time of
The injection of electrons from the channel into the buffer will lead to a change of the electron density in the channel. The GaN layer electron density distribution at 50 ns is shown in Fig. 4(a). The results for the evolution of channel electron density are shown in Fig. 4(b). A reduction in channel electron density is found as the drain bias pulsed from 0 to 10 V. It then increases to a steady state as the drain bias stays at 10 V. The difference between our simulation result and the results from drift--diffusion model is due to the involvement of lattice temperature.
There are two reasons for the filling of traps by electrons. One is that the channel electrons acquire enough energy when the electric field intensity increases. The hot electrons in the 2DEG channel generated under high drain bias could be injected into the adjacent epitaxial buffer layer, where they can be captured by donor like traps[16, 17]. The other reason is that the filled acceptor traps can emit electrons due to the increase of temperature and the 2DEG density increases as the temperature rises, which is the so called a de-trapping process[18]. In this case, the drain bias is pulsed from 0 to 10 V in 500 ns. The excess channel electrons can spill over in all directions under the drain bias and acquire enough energy. A number of electrons injects from the channel into the buffer at the gate edge and are captured by the traps. Although the temperature in the channel of device rises and the de-trapping exists, the trapping process is greater due to the change of electric field intensity. As the drain bias reaches a steady state, the electric field intensity will not change but the temperature of the channel rises rapidly[12]. The de-trapping process due to the increase of temperature becomes increasingly important until the trapping and de-trapping are in a dynamic equilibrium. The analysis is consistent with the experimental results, the 2DEG density increases as the temperature rises, which is studied by Maeda, et al.[18]. Therefore, the electron density of channel first decreases, and it then increases.
The evolution of channel electron density of the AlGaN/GaN HEMTs device under different rise time is simulated, as it is shown in Fig. 5. All of the results show a similar trend under different rising times, indicating that the electron density of channel decreases first and then increases. Moreover, the minimum of the channel electron density increases with the increase of rising time. This phenomenon is related to the effects of the electric field and temperature on the filling of traps by electrons.
4. Conclusion
In summary, the effects of self-heating and traps on drain current of AlGaN/GaN HEMTs are studied by transient simulation. The simulated drain current transient response is in good agreement with both the reported experimental results and the theoretical models. The time of drain current reaching a steady state is dependent on the thermal time constant, which indicates that temperature is the main factor for the drain current lag because the electrical properties are affected by the channel temperature. The electron density of channel decreases first and then increases. This is related with the effects of electric field and temperature on the filling of traps by electrons.