1. Introduction

In recent years, Ga₂O₃ semiconductor materials with excellent performance have entered the public field of vision, with high dielectric constant, breakdown voltage, and Baliga’s Figure of Merit [1-4], which is more advantageous than Si, SiC, GaN materials. In addition, single crystals can be grown using low-cost and low-loss technology [5-8], thus reducing production costs for mass production, which has strong practical application value.

Because the current research on Ga₂O₃ MOSFETs is relatively few and the preparation technology is immature, many studies use simulation software to simulate the performance of the components. By calculating the simulation process steps and simulating the electrical characteristics of the power device, Silvaco TCAD can roughly predict the experimental results and provide guidance for the actual experiments. This makes the experiment cost greatly reduced, and can speed up the experiment process, aslo avoiding the occurrence of large errors. At present, many scholars have used simulation methods to conduct a large number of studies on the electrical characteristics and reliability of Ga₂O₃ MOSFET. Man Hoi Wong et al. used Silvaco Atlas to simulate the electrical characteristics of Ga₂O₃ MOSFET at different temperatures to explore the influence of its self-heating effect on thermal resistance [9]. Inhwan Lee et al. proposed non-field plate (NFP) devices and field plate (FP) Ga₂O₃ MOSFET devices, obtained the static characteristics of the devices in Silvaco TCAD software, and conducted their switching performance in Silvaco mixed-mode with inductive load test circuits [10]. Junsung Park et al. used TCAD simulation to study the electrical performance of multi-gate vertical β-Ga₂O₃ MOSFETs and proposed a simulation setting based on β-Ga₂O₃ devices, and found that the I_D- V_GS simulation curve fitted well with the practice [11]. These studies have proved the accuracy and reliability of Silvaco TCAD.

This paper is divided into five parts: In the second part, it will explain how to use Silvaco TCAD to simulate according to the simulation idea, the simulation model and the simulation process sequence. In the third part, Ga₂O₃ MOSFET will be simulated to explore the effect of its structure on the electrical characteristics. In the fourth part, the research on the reliability of Ga₂O₃ MOSFET in recent years will be described. In the fifth part, the full text is summarized.

2. Simulation method

Before using Silvaco TCAD for simulation, it is necessary to obtain the specific structure of the simulation device and related data. Next, a variety of parameters are set, including material parameters such as electron affinity, band gap width, state density, saturation speed, etc. Model parameters, such as mobility, impact ionization parameters, etc. Numerical calculation methods, such as newton iteration method, gummel iteration method and block iteration method. After setting, print and tonyplot functions can be used to display the electrical characteristics of the device, as well as internal carrier concentration distribution and electric field distribution [13].

We chose the device model that used Silvaco Atlas to simulate wide-band gap Ga₂O₃ MOSFET for the first time in 2013 [12].

Figure 1. Device structure and doping [12]

The substrate is a semi-insulating single-crystal β-Ga₂O₃, above which there is a 300nm thickness N-type Ga₂O₃ channel layer with a dopant concentration of 7×10¹⁷ cm^-3, about half of which can be considered activated. The source and drain contact regions were formed by multiple silicon implants with a deep box profile of 150 nm and a concentration of 5×10¹⁹ cm^-3, of which 3×10¹⁹ cm^-3 is activated. The distance between the source and drain box regions is 20 um, while forming a 20 nm thick Al₂O₃ gate dielectric film with a 2 um long Ti/Pt/Au metal gate at the top. According to the analysis, the effective mass of electrons is 0.28m₀, the local conduction band density is N_C=3.72×10¹⁸ cm^-3, the electron affinity of β-Ga₂O₃ is 4.0 eV, and the work function of the gate is 5.23 eV [12].

The simulated I_D-V_D curve and I_D-V_G curve are shown in Figure 2 and Figure 3. Figure 2 shows the relationship between I_D and V_D under different V_GS, while Figure 3 shows the relationship between I_D and V_G. It can be seen that the simulation curve fits well with the experimental curve, which proves that the simulation calculation of Silvaco Atlas has high accuracy.

3. Simulation and analysis of electrical characteristics of Ga₂O₃ MOSFET

The electrical characteristics of Ga₂O₃ MOSFET can be affected by a variety of factors, such as the field plate, the environment, doping, fin shape, etc. This section will introduce the characteristics of its breakdown voltage and on-resistance.

Zeng Ke et al. proposed a new field plate design using composite dielectric, and used ALD deposited film in the high field region, which greatly improved the breakdown voltage of MOSFET power devices [14].

As shown in Figure 4, in order to avoid the breakdown outside the channel, a composite field plate is added on the top of the device, the upper part of which is composed of a higher density 50nm ALD SiO₂ layer, the lower part is composed of 350nm PECVD SiO₂ layer [14], and the l_FP in the figure is optimized. In addition, a dielectric liquid Fluorinert (FC-770) is used instead of air, which has a dielectric constant 4 times that of air and can effectively prevent air breakdown.

The actual breakdown voltage of the device was measured using a specific device. As shown in Figure 5, When dielectric liquid Fluorinert is used, the breakdown voltage is greatly increased relative to air, reaching an extremely high 1850V. In addition, the dielectric constant also correlates with the electric field strength in the device. Measurements on a device L_gd=0.8μm showed a breakdown voltage of 324V at a -32V gate bias under a fluorine noble gas, resulting in an average calculated electric field strength of 4.4±0.2MV/cm, higher than the 3.8MV/cm record reported in reference [15]. Meanwhile, in the electric field simulation of Silvaco Atlas (FIG. 6), the peak electric field intensity near the drain edge of the device is 6.1 MV/cm, which is higher than the actual value of 4.4 MV/cm. Although there is a certain error, the successful prediction of the rising trend of the electric field strength is of guiding significance to the experiment.

Junsung Park et al. found that the fin shape of a β-Ga₂O₃ vertical MOSFET has an effect on the on-resistance of the device, and used Silvaco TCAD to simulate the electrical performance. The structure of the proposed β-Ga₂O₃ vertical MOSFET is shown in Figure 7.

The experiment made changes to the channel width (W_ch) and fin height (L_g) of the fin structure, and measured the on-resistance of the device at different W_ch and L_g, as shown in Figure 8(a). Considering the DIBL characteristics of the device, when L_g= 1.0 μm, W_ch= 0.3μm, the overall performance of the device reaches the optimal value, and can achieve a very low DIBL (< 5 mV/V) while maintaining a low R_on per unit area.

At the same time, TCAD simulation was carried out on the changes of channel current and drain current of the device. FIG. 8(b) shows that the TCAD simulation curve fits well with the experimental results, and the predicted threshold voltage (V_th) of the improved fin increases from 0.92V to 1.44V, indicating that its conductive ability has been improved. FIG. 8(c) simulates the I_D-V_DS curve, and finds that the maximum drain current is also increased, and the on-resistance at V_DS=0V is reduced from 6.55 mΩ·cm² to 4.78 mΩ·cm². Silvaco TCAD accurately predicted the change in the on-resistance of the device.

4. Simulation and analysis of reliability of Ga₂O₃ MOSFET

Ga₂O₃ MOSFET also has some reliability problems, such as performance reduction caused by self-heating effect, switching loss caused by device switching process, etc. This chapter will show the application of Silvaco TCAD in reliability research.

Due to the highly localized power dissipation and heat generation of MOSFET in the submicron region of the gate drain edge channel, as well as the self-heating effect caused by the low thermal conductivity of Ga₂O₃, Ga₂O₃ MOSFET power devices will degrade performance at high temperatures. Reference [9] shows a method of using TCAD to study the self-heating effect of MOSFET devices. The MOSFET device structure used in the study is shown in Figure 9.

After parameter setting, the device was actually measured and Silvaco TCAD simulated. FIG. 10(a) shows the I_DS-V_DS curve at different temperatures with DC voltage. It can be found that the simulation curve fits well with the experimental results, which proves the reliability of TCAD.

Next, the device is subjected to a 5-μs pulsed drain current and the I_DS-V_DS curve is measured and simulated (Figure 10b). At this time, it can be found that the experimental results fit well with the simulation results at low voltage, but the I_DS of the experimental results is lower when the V_DS simulation results is high, indicating that when V_DS increases, the self-heating effect in the Ga₂O₃ channel becomes non-negligible and the thermal resistance of the device increases. This has also been verified in literature [16], where the saturation current of VLD transistors using SH (self-heating effect) model is significantly lower than that without SH model, indicating that the low thermal conductivity of Ga₂O₃ has a greater impact on device performance. These studies prove that the direct influence of the self-heating effect on the electrical characteristics of Ga₂O₃ MOSFET devices can be obtained through TCAD simulation.

Switching loss refers to the energy loss of a MOSFET during switching, which can affect its efficiency, temperature and reliability. Literature [10] found that the switching loss of the device can be reduced through field plate and width optimization.

Table 1. Switching loss of NFP and switching loss of FP under different T_FPOX [10]

The device structure of NFP and FP is shown in FIG. 11. Band Gap energy is set at 4.8eV, Effective density of states in the condition band at 300k is at 4.45e-18 cm^-3. Electron Affinity is set to 4.0eV and Electron Mobility is set to 118 cm²/V*s [10]. FP devices with T_FPOX=1 μm can be found to have the lowest switching loss. However, compared with commercial SiC MOSFETs, although Ga₂O₃ devices have lower conduction loss, the switching loss is still high, so it is necessary to optimize the width of the devices. Silvaco TCAD is used to simulate the switching loss curve of the device width from 1.05e6μm to 2.0e6μm under different nominal power, as shown in FIG. 12.

It can be concluded that the optimal device widths at 200, 300 and 500kVA are 1.4e6, 1.8e6 and 2.0e6 μm, respectively, which have the smallest power loss at these device widths.

5. Conclusion

In summary, under the background that Ga₂O₃ MOSFETs are difficult to prepare and few studies have been conducted, Silvaco TCAD simulation software can be used to explore the characteristics of power devices with low cost and high efficiency. In this paper, the simulation method of Silvaco TCAD is introduced, and it is applied to the study of the electrical characteristics and reliability of Ga₂O₃ MOSFET. It is found that the electric field strength of the power device designed by the new field board can predict the change of its breakdown voltage, and the simulation software successfully predicts the rising trend of the electric field strength. In addition, the channel width and fin height of the β-Ga₂O₃ vertical MOSFETs can be optimized by simulation to reduce the on-resistance of the devices. At the same time, it can also be applied to explore the reliability of self-heating effect and switching loss of the device. These studies help to improve the performance and durability of MOSFETs, such as reducing power consumption and heating. But in addition to these studies, the material parameters of Ga₂O₃, device model setting, process simulation research and other aspects of the exploration is still less. It is expected that the parameters and models used by Silvaco TCAD could be more abundant and accurate in the future, so as to simulate Ga₂O₃ MOSFETs more accurately, and be applied in more studies.