Study of the mechanism of single event burnout in lateral depletion-mode Ga₂O₃ MOSFET devices via TCAD simulation

With the continuous development of aerospace and nuclear energy, the demand for electronic devices that operate reliably in high-radiation environments is increasing.^1–3 Gallium oxide (Ga₂O₃), a novel wide-bandgap semiconductor material, exhibits high electron mobility and an ultra-wide bandgap.^4–7 This makes it promising for applications involving high breakdown voltage (BV), substantial output current, and high power factor of merit (PFOM). So it has potential in power device applications both on Earth and in space.^8,9 However, in the context of space applications, these devices could be exposed to various radiation sources such as high-energy protons, electrons, and heavy ions. These radiation sources can degrade the electrical characteristics of Ga₂O₃ devices, and even lead to catastrophic damage.^10–12 

At present, the irradiation effect of Ga₂O₃-based FET devices has been studied by conducting irradiation experiments involving protons, γ-rays, and other particles.^13–15 Yang et al.¹⁴ used high-energy protons to investigate performance degradation in back-gate two-dimensional β-Ga₂O₃ nanoribbon FET devices. Their results show that the conductivity and effective carrier mobility decrease with the increase of proton irradiation fluence. In addition, Wong¹⁵  et al. utilized ⁶⁰Co γ-ray to irradiate Ga₂O₃ MOSFET devices with a field plate (FP) structure. The results showed that the γ-ray irradiation does not produce severe defects in the β-Ga₂O₃ channel, and the carrier concentration and mobility remain unchanged. However, there are few reports on heavy ion irradiation of Ga₂O₃ MOSFET.

Ma et al.'s research¹³ used the Technology Computer Aided Design (TCAD) platform to simulate the transient characteristics and studied the single event effect (SEE) of β-Ga₂O₃ MOSFET. It is found that the peak value of drain transient current increases with an increase in V_DS, linear energy transfer (LET), and the incident angle of heavy ions. Furthermore, based on the research foundation of Si semiconductor materials, it is revealed that heavy ion irradiation is prone to the SEE, among which the single-event burnout (SEB) phenomenon is the most concerning.

It is reported that SEB can be attributed to the parasitic transistors in Si and SiC VDMOS devices. The difference between Si and SiC is that Ga₂O₃ MOSFET is a field effect transistor without a PN junction structure, and there is no parasitic transistor structure inside. Due to the deep acceptor level caused by doping,^16,17 it is very difficult to form a P-type channel in Ga₂O₃ because of the relatively high ionization energy (up to 860 meV).¹⁸ In conclusion, the mechanism of device failure induced by heavy ion irradiation on β-Ga₂O₃ MOSFETs remains to be investigated.

This work investigated the radiation effects of heavy ions on lateral depletion-mode Ga₂O₃ MOSFET devices using the TCAD platform,¹⁹ discussed the SEB sensitive position of the device, and combined it with electrical parameters such as electric field, carrier concentration, and current density. The physical mechanism of SEB in lateral depletion-mode Ga₂O₃ MOSFET is preliminarily revealed.

Based on the experiments conducted by Higashiwaki,²⁰ a model for a 370 V depletion-mode Ga₂O₃ MOSFET device has been established. The specific thicknesses and doping concentrations of each region of the device are given in Table I.²⁰ 

As shown in Fig. 1, this power device has a lateral structure, with a single crystal β-Ga₂O₃ substrate. A 300 nm n-type Ga₂O₃ epitaxial layer doped with a concentration of 1.1 × 10¹⁷ cm⁻³ is grown on this substrate. The source and drain contacts are formed by ion implantation of Si, resulting in a Gaussian donor doping with a concentration of 5 × 10¹⁹ cm⁻³ and a thickness of 150 nm. The separation between the source and the drain regions measures 20 μm. At the top, the gate electrode has a length of 2 μm with a 20 nm thick Al₂O₃ dielectric layer. The Ga₂O₃ material parameters are given in Table II.

In this simulation, the main physical equations include Poisson equation, carrier transport equation, and carrier continuity equation. Similarly, the main models consider the SRH composite model, CVT model, Fermi–Dirac statistics model, bandgap narrowing model, and impact model. The Auger generation–recombination (AGR) was not assumed in the TCAD simulation. Finally, based on the aforementioned typical physical equations and device models, the electrical characteristics of the devices were simulated.^22,23 We obtain the transfer characteristics curve at V_DS= 25 V [Fig. 2(a)] and the output characteristics curve with V_GS ranging from −12 to 4 V in steps of 4 V [Fig. 2(b)]. The threshold voltage V_TH is about −10 V, which is consistent with that of the previous experimental results.²⁴ 

It applies to both electrons and holes. Here, a and b are fitting parameters based on Ref.26, in which E is the electric field in V/m.

In this paper, the transient current generated by heavy ions incident on the device is utilized to determine the occurrence of the SEB effect. Under the condition of a negative bias V_GS= −20 V in the cut-off state, the drain voltage V_DS = 80 V was ramped up with a step of 5 V to search the threshold voltage of SEB in the device. We discuss the SEB sensitive region of the device for LET = 10 MeV cm²/mg. The result is shown in Fig. 3. The inset of Fig. 3 shows that the drain current changes with time under different V_DS in the device at x₀= 14 μm. After incident heavy ions, the peaks of 10⁻¹³ and 10⁻¹¹ s are mainly the transient current generated by the charge collection of the device. The main peak at 10⁻¹⁰ s is due to the conduction of the SEB channel inside the device. If there is insufficient current to maintain the state, the drain current will be restored to 0, indicating that there is no SEB in the device, as depicted by the black line. Conversely, if the drain current rises and remains consistently high, it indicates the occurrence of SEB, as illustrated by the red line.

As shown in Fig. 3, it is observed that the threshold voltage of SEB varies at different incident positions, i.e., the varying sensitivity. For the lateral depletion-mode device, the SEB-sensitive region is between the gate and the drain, with the sensitive region located at x₀= 14 μm. At this time, the SEB threshold voltage of the device is V_DS= 100 V.

The sensitive area of the device is analyzed in combination with the electric field. As shown in Fig. 4, the evolution of the electric field distribution within the device is shown when avalanche (charge) multiplication occurs at x₀= 14 μm. It can be seen from Figs. 4(a)–4(c) that as time increases, the electric field inside the device also increases, especially near the gate side.

Figure 4(d) shows the variation in electric field intensity within the channel beneath the gate. It is found that there are two electric field peaks in the device, which are two edges of the gate electrode, respectively. As avalanche (charge) multiplication occurs in the device, the electric field intensity experiences gradual enhancement, reaching its peak at t = 1 × 10⁻¹⁰ s.

Due to the application of a positive high voltage (V_DS) to the drain electrode, conduction occurs in the channel region extending from the gate's edge to the drain, effectively forming a low-resistance region on the right-hand side of the device. These voltages are mainly applied to the high resistance depletion region below the gate, resulting in a higher electric field peak near the gate. As a result, a higher electric field peak develops near the gate, making this area the most sensitive to SEB. Subsequently, we further discuss the underlying microscopic mechanisms of this phenomenon.

After the heavy ion is incident at x₀= 14 μm (the most sensitive region of SEB in the device), the bias voltage is V_DS= 110 V and V_GS= −20 V, and the electrical parameters such as carrier concentration distribution and electric field intensity at different times inside the device are analyzed.

Figure 5 shows the distribution of electron and hole concentrations in the device before and after heavy ion incidence with time. It can be seen that before the incidence of heavy ions [Fig. 5(a)], since the device is in the cut-off state V_GS= −20 V, the carrier was depleted as the gate is fully closed, and there is no conductive channel. After the heavy ion is incident on the device, a large number of electron–hole pairs will be generated along the incident tracking. The V_DS is positive, so there is an electric field from the drain to the source direction inside the device. Driven by this electric field, these electrons and holes experience a carrier drift. It can be seen from Figs. 5(b) and 5(c) that the width of the depletion layer below the gate gradually narrows and the conduction channel tends to open gradually. As seen in Fig. 5(d), the leakage channel is fully opened at 1 × 10⁻⁷ s finally.

Since the above phenomenon is related to carrier concentration and current flow density, we combine these for further analysis. Before the incidence of heavy ions, holes accumulate below the gate electrode due to the influence of the electric field [Fig. 5(e)]. After the incidence of heavy ions, the hole drifts to the gate and source driven by the electric field. As shown in Figs. 5(f)–5(h), with the ionization deposition, more and more electron–hole pairs are generated, and the carrier concentration generated along the incident track increases.

Since the mobility of the holes is slower than that of electrons, a substantial quantity of holes are not collected and accumulated in the channel layer below the gate. These positively charged holes accumulated below the gate can reduce the barrier in the region below the gate, resulting in electrons being injected from the source into the channel, and the conductive channel below the gate is gradually opened in the depletion region, which is consistent with that in Figs. 5(b)–5(d).

To further reveal the cause of large current when SEB occurs in the device, Fig. 6 shows that the time-dependent current density and impact generation rate change after heavy particle incidence. The heavy-ion-induced electron–hole pair generation rate has a peak of 1 × 10²⁴ cm⁻³ s⁻¹ at about 1 × 10⁻¹⁰s, with a $J n$ of 6.0 × 10⁵ and $J P$ of 2.8 × 10⁵ A/cm². As shown in Figs. 6(a)–6(c), the impact generation rate will gradually increase with time, and the incidence of heavy ions will generate a large number of electron–hole pairs on the path and accelerate the carrier drift of electrons and holes, which is consistent with that in Fig. 5. Particularly at t = 1 × 10⁻⁷s, the impact generation rate still exists [Fig. 6(d)]. It provides a large number of electron–hole pairs and creates a sensitive environment for SEB to occur. However, the impact generation rate on the track is not equal, and it is higher below the gate, which is consistent with the strength of the electric field distribution in the device, as shown in Fig. 4.

Because Ga₂O₃ is mainly an N-type semiconductor device, Figs. 6(e)–6(h) show the electron current density. With the incidence of heavy ions, the electron current density along the ions track begins to increase. In the process of heavy ion incidence, the increase in the impact generation rate leads to an increase in carrier concentration driven by the electric field-induced drift of electrons to the drain. Therefore, the current flow is not limited, and the electron current density increases because the channel in the substrate away from the gate is gradually opened.

It is worth noting that the electron conductive channel does not flow directly from the source to the drain along the lateral direction, which takes on a “V” shape when SEB occurs, forming a V-shaped source–substrate–drain conductive path. Meanwhile, the channel layer under the gate could still be depleted by a negative gate voltage bias in Ga₂O₃ MOSFET. We analyze that this is related to the depletion region below the gate and further discuss it theoretically.

Because the electron mobility $μ n$ is fixed in the simulation, we can see that the main factors affecting $| J → | n$ are electron concentration n and electric field $E x$ from the formula. Figure 4 shows that during the incident process of heavy ions, the electric field below the gate exists and has a maximum in the x value. Combined with the change of electron concentration in Figs. 5(a)–5(d), it can be seen that due to the incidence of heavy ions, the depletion layer below the gate becomes narrow, and the electron concentration in the substrate increases. The increase in electric field distribution and electron concentration in the substrate layer in the depletion region leads to a rapid increase in $| J → | n$⁠, as shown in Figs. 6(e)–6(h).

It can be seen from Eq. (9) that, due to the increase of n, $| J → | n$ increases rapidly, resulting in the increase of G and the generation of more electron–hole pairs. At the same time, the increase of the electron–hole pairs leads to an increase of $| J → | n$⁠, forming a positive feedback and generating a continuous high current, and the SEB effect occurs in the device.

Different from the parasitic transistor in Si and SiC VDMOS devices, the Ga₂O₃ device that is currently being studied is mainly an N-type field effect transistor without a PN junction structure,¹⁹ because it is difficult to perform P-type doping inside. Based on the special structure and physical properties of Ga₂O₃ devices, we summarized the physical process of the SEB effect on Ga₂O₃ devices and described it as shown in Fig. 7.

The field-effect transistor without the PN junction is controlled mainly by the gate, and it can regulate the current between the source and the drain regions. The device operates by utilizing the gate material's work function to place the device in a fully depleted state, i.e., the device works in a cut-off state. With the incidence of heavy ions, a large number of electron–hole pairs will be generated on the incident track. On the one hand, a large number of electrons begin to carrier drift, resulting in the depletion region in the channel gradually narrowing, which is marked by process ① in Fig. 7(a). On the other hand, more and more holes are accumulated below the gate, which makes the gate voltage gradually increase and reduces the gate control ability; the channel of the substrate layer far away from the gate is opened in the depletion region, which is marked by process ② in Fig. 7(b). The V_DS is positive, so there is an electric field pointing from the drain to the source inside the device. Driven by this electric field, these electrons tend to drift toward the drain. Therefore, the electrons from the source flow to the drain region through the substrate layer far away from the gate in the depletion region, forming a conductive channel, marked by process ③ in Fig. 7(c). Under the applied bias voltage, the electrons generated by the impact ionization generation inside the device continue to drift along the conductive channel. Finally, the source and drain are short-circuited, resulting in a continuous large current and the SEB effect occurs in the device.

In this paper, the SEB effect of lateral depletion-mode β-Ga₂O₃ MOSFET was simulated and analyzed by TCAD. It is found that the most sensitive region of SEB is located on the gate near the drain side. Combined with the distribution of electrical parameters at different times inside the device, the mechanism of SEB in Ga₂O₃ MOSFET was attributed to the special field effect transistor structure without a PN junction inside the device. A large number of electron–hole pairs will be generated on the incident track of heavy ions, and holes under the gate will gradually increase the gate voltage and reduce the gate control capability. Meanwhile, a large number of electrons began to drift. The depletion region is gradually narrowed and the channel in the substrate layer away from the gate is opened. The SEB current between the source and the drain flows through the MOSFET body, forming a V-shaped source–substrate–drain conductive path, while the channel layer under the gate could still be depleted by a negative gate voltage bias of Ga₂O₃ MOSFET.

The authors acknowledge support from the Open Project of State Key Laboratory of Intense Pulsed Radiation Simulation and Effect (No. SKLIPR2115), the National Natural Science Foundation of China (No. 12004329), the Postgraduate Research and Practice Innovation Program of Jiangsu Province (No. SJCX22_1704), and the Innovative Science and Technology Platform Project of Cooperation between Yangzhou City and Yangzhou University, China (Nos. YZ202026301 and YZ202026306).

The authors have no conflicts to disclose.

Kejia Wang: Investigation (equal); Methodology (equal); Writing – original draft (equal). Zujun Wang: Investigation (equal); Methodology (equal). Rongxing Cao: Conceptualization (equal); Supervision (equal); Writing – review & editing (equal). Hanxun Liu: Investigation (equal); Resources (equal). Wenjing Chang: Investigation (equal); Resources (equal). Lin Zhao: Formal analysis (equal). Bo Mei: Formal analysis (equal); Investigation (equal). He Lv: Formal analysis (equal). Xianghua Zeng: Conceptualization (equal); Methodology (equal). Yuxiong Xue: Conceptualization (equal); Supervision (equal).

The data that support the findings of this study are available from the corresponding author upon reasonable request.