Experimental Study of the Comparison of the Synergistic Effect of Total Ionizing Dose and Neutron Single Event on Si/SiC MOSFETs

Abstract

A comparative study on the synergistic effect of the total ionizing dose and neutron single event effect on a SiC MOSFET and Si MOSFET was performed based on the ⁶⁰Co γ source and the high-pressure multiplier 14 MeV neutron source at the China Institute of Atomic Energy. First, a γ-ray total ionizing dose experiment was performed on these two devices, and the differences in the total ionizing dose damage of the SiC and Si MOSFETs were analyzed. Then, neutron single event effect experiments were performed to investigate the effects of different doses on the single event effect for the devices. The results indicate that the unhardened SiC MOSFET has stronger resistance to the total ionizing dose compared with hardened Si MOSFET. During the 14 MeV neutron irradiation experiment, no single event burnout was observed in either device, but single event transients were observed. Even though the hardened Si MOSFETs are capable of suppressing single event transient currents at a higher drain bias, the trapped charge concentration of SiC MOSFETs due to irradiation is smaller than that of Si MOSFETs, which improves their resistance to the total ionizing dose and makes them less affected by the synergistic effect of the total ionizing dose and neutron single event effects. The research results can provide some guidelines for the radiation hardening technology of power devices used in aerospace and nuclear industries.

1. Introduction

The extensive deployment of silicon (Si)-based power devices in power electronic systems has been observed to exhibit a growing tendency for their performance to degrade or even fail in extreme environmental conditions, including high temperature, high voltage, and strong radiation field [1]. In contrast, silicon carbide (SiC) has emerged as a prominent material in power electronics systems. As a third-generation semiconductor material, silicon carbide (SiC) has a bandgap and thermal conductivity 3 times that of Si, and a critical breakdown field 8~10 times that of Si [2]. It can be reasonably deduced that SiC power devices possess enhanced tolerance to irradiation, and as a consequence, represent a prospective option for applications in extreme radiation environments, such as in the aerospace and nuclear industries. Power MOSFETs are widely used power devices in the field of power electronics, which are typically employed as switching and circuit control devices in electronic systems, and any fluctuations in their performance may give rise to malfunctions, or, in extreme instances, system failure. It is therefore essential to conduct a comprehensive investigation into the radiation-resistant properties of Si/SiC MOSFETs in order to enhance their dependability and functionality for applications in the aerospace and nuclear industries.

Heavy ions, protons and electrons in the space radiation environment, as well as neutrons and γ-rays in the reactor radiation environment, contact SiC MOSFETs, triggering a variety of radiation effects, affecting the reliability of the device [3,4,5]. According to the radiation damage mechanism and radiation-induced degradation performance of electronic devices, radiation effects can be divided into three categories, namely, single event effect (SEE), total ionizing dose (TID) and displacement damage (DD). Prior research has demonstrated that SiC MOSFETs exhibit enhanced resilience to the total ionizing dose in comparison to Si MOSFETs [6,7,8,9]. Nevertheless, SiC MOSFETs remain susceptible to single event effects, including single event burnout (SEB), Single Event Gate Rupture (SEGR), single event transient (SET), and others [10,11,12,13,14,15]. Furthermore, both TID and DD are cumulative radiation effects. Consequently, the degree of damage to the electronic device will increase gradually over time. This will also result in a significant difference in the single event effect of the device, depending on the degree of cumulative radiation suffered. This phenomenon is also known as the synergetic effect [16].

Considering that the practical environment of electronic devices is mostly a complex field where many kinds of particles exist at the same time, the appearance of a synergistic effect is an inevitable consequence. A number of studies have been carried out to investigate the synergistic effect between the total dose effect and the single event effect. However, at present, the research in this area is still dispersed, and no comprehensive consensus has been reached. For Si/SiC MOSFET power devices, the study of the synergistic effect of TID and SEE is relatively rare, and most of these studies are theoretical simulations and mainly focus on the synergistic effect of TID and heavy ion SEE, while the experimental study of the synergistic effect of TID and neutron SEE have hardly been reported.

In 2013, Busatto et al. [17] investigated the charge collection mechanism of the TID-SEGR synergistic effect in Si MOSFET devices. Their findings revealed that the TID induces a trap charge at the gate oxide layer, which reduces the tunneling limit of electrons at this layer. As a consequence, a leakage path is formed at the gate and the threshold voltage for the self-eroding gate resistor (SEGR) is reduced significantly.

In 2022, Wu Lei et al. [18] investigated the TID-SEB synergistic effect of SiC MOSFETs by simulation and experiment, and found that the trap charge induced by the TID at the gate oxide layer generates a strong electric field at the gate–source junction. When high-energy particles are incident on the device, the leakage current rises higher and the local lattice temperature rises faster, which makes the device more susceptible to SEB, and based on this, a radiation hardening method for electronic devices is proposed by increasing the thickness of the epitaxial layer and the doping concentration of the gate–source junction.

In 2023, Cao et al. [19] investigated the TID-SEGR synergistic effect of SiC MOSFETs using TCAD simulation and found that the effect of TID on SEGR is mainly manifested in the generation of oxide trap charges that generate additional electric fields inside the gate oxide layer. When energetic particles are incident on the device, the generated holes accumulate at the SiC-SiO₂ interface and, together with the negative charge induced by the gate, form an electric field, which is added to the electric field generated by the trap charge, thus increasing the strength and spatial inhomogeneity of the gate–oxygen electric field, which subsequently increases the sensitivity of the device to SEGR.

With the steady advancement of deep space exploration, space nuclear reactors have become the best power option because of their small size, high energy density, high output power and long endurance. However, nuclear reactors inevitably produce a large number of neutrons and γ-rays. These particles contact the reactor control system and the aircraft electronic system will trigger radiation effects, resulting in electronic device reliability degradation or even functional failure. Power MOSFETs are the core devices of nuclear reactor power conditioning systems and aircraft power systems, and the influence of radiation effects is critical. It is of great significance to investigate the radiation effect of power MOSFET devices in the neutron and γ-ray mixed radiation environment. In this paper, based on the high voltage multiplier 14 MeV neutron source and Co-60 γ-ray source at the China Institute of Atomic Energy (CIAE), we carry out the study of TID and neutron SEE synergistic effect for SiC MOSFET and Si MOSFET devices. The variation rule of TID with γ-ray dose and the variation rule of neutron SEE with neutron fluence of the two devices are investigated; additionally, the effect of TID on neutron SEE and the underlying mechanism are analyzed.

2. Experiments

Power MOSFETs are one of the most widely used power devices due to their low gate drive power, fast switching speed and excellent parallel operation capability.

In this experiment, two n-channel planar gate power MOSFET devices with different materials and parameters were selected, namely, a commercial 650 V SiC MOSFET device from company A and a 500 V Si MOSFET device from company B. The devices from company B have undergone radiation hardening for SEE, and a buffer layer is formed at the epitaxial layer/substrate interface through a retarded epitaxial process to improve their resistance to SEE. The main parameters of the devices are listed in

Table 1. Device parameters.

Manufacturer	A	B
materials	SiC	Si
seal inside	TO-247	TO-254AA
VDSS/V	650	500
Vth/V	2.3	3.5
IDSS/nA	<1000	<1000
RDS(on)/mΩ	120	<320
gm/S	5.0	>6.0

, and eight samples were selected for each device, numbered A1~A8 and B1~B8, respectively.

2.1. γ-Ray Irradiation Experiment

The TID experiments were carried out with the Co-60 γ-ray at CIAE with a γ-ray dose rate of 200 krad(Si)/h at room temperature. Each device sample was tested for electrical parameters before the experiment to ensure that the samples functioned properly. The γ-ray irradiation doses for each sample were as shown in

Table 2. The γ-ray irradiation dose of all samples.

γ-Ray Radiation Dose	SiC 650 V	Si 500 V
500	A1~A3	--
300	--	B1~B3
200	A4~A6	B4~B6
0	A7~A8	B7~B8

.

Typically, the TID damage of MOSFETs under gate voltage bias conditions is significantly higher than that of the drain voltage bias or zero bias [8,9]. In order to ensure that the threshold voltage of the Si devices selected in this paper is not changed to a negative value, which would affect the results of the neutron irradiation experiments, the drain voltage bias conditions were specifically selected, i.e., drain bias is applied up to the rated voltage for all samples and the gate–source junction is shortened. In the experiment, SiC device samples A1~A6 were irradiated first (see Figure 1). When the accumulated dose reached 200 krad(Si), A4~A6 were removed, and A1~A3 continued to be irradiated to 500 krad(Si). After all SiC devices were irradiated, they were replaced with Si device samples B1~B6, where B4~B6 were irradiated to 200 krad(Si) and B1~B3 were irradiated to 300 krad(Si). Each sample was stored in a dry ice environment immediately after irradiation. Samples A3 and A6 and B3 and B6 were tested for electrical parameters after irradiation to assess the degradation of the device performance at different doses of γ-ray irradiation; the remaining samples were irradiated with neutrons at the shortest possible time interval (the actual interval time was less than 25 h), to prevent the samples from annealing due to the prolonged placement time, which would affect the results of synergistic effect tests.

2.2. Neutron Irradiation Experiment

The neutron source used for the experiment was a monoenergetic 14 MeV neutron from the high-voltage multiplier at CIAE, with an average neutron flux of 5.1 × 10⁶ n·cm⁻²·s⁻¹. Except for the samples A3 and A6 and B3 and B6, which were used for the test of degradation of electrical parameters after γ-ray irradiation, the remaining samples were subjected to neutron irradiation tests, as shown in Figure 2.

Before neutron irradiation, the off-state leakage current of each sample and the stability of the test instrument were tested. When the test results showed that the off-state leakage currents of all samples were stable without any abnormality, the neutron irradiation was then started. During the irradiation period, the gate–source junction was kept short for all samples, and the drain bias voltage was started from 200 V. For every 10 min of irradiation (with a neutron fluence of about 3.1 × 10⁹ n·cm⁻²), the bias voltage was increased by 100 V until the rated voltage reached 650 V for SiC MOSFETs and 500 V for Si MOSFETs. In the whole process of irradiation, the off-state leakage current of the sample is monitored, and the upper limit is set at 10⁻⁴ A. If the off-state leakage current of the device was higher than 10⁻⁴ A, it was judged to be an SEB phenomenon and the power supply was cut off immediately; at the same time, we obtained the variation curves of off-state leakage current with neutron fluence under different bias voltages as well as different γ-ray dose conditions, so that we could analyze the neutron SEE regularity and the effect of TID on the neutron SEE of the device accordingly.

3. Experimental Results and Discussion

3.1. Total Ionizing Dose

The results of the electrical parameters for samples A3 and A6 and B3 and B6 after the γ-ray irradiation experiment are presented in

Table 3. Electrical parameter results of SiC MOSFET samples after γ-ray irradiation.

DUT	Test items	Vth/V	IDSS/nA	gm/S
	Test conditions	VDS = VGS IDS = 1.86 mA	VDS = 650 V VGS = 0 V	VDS = 20 V ID = 6.76 A
SiC A6	0 krad(Si)	2.565	34.28	1.873
	200 krad(Si)	2.340	34.86	1.976
	Percentage change	−8.77%	1.66%	5.50%
SiC A3	0 krad(Si)	2.460	38.90	1.878
	500 krad(Si)	2.106	39.08	1.901
	Percentage change	−14.39%	0.46%	1.22%

and

Table 4. Electrical parameter results of Si MOSFET samples after γ-ray irradiation.

DUT	Test items	Vth/V	IDSS/nA	gm/S
	Test conditions	VDS = VGS IDS = 1.0 mA	VDS = 400 V VGS = 0 V	VDS = 15 V ID = 11.7 A
Si B6	0 krad(Si)	3.756	8.426	17.49
	200 krad(Si)	1.987	317.300	12.72
	Percentage change	−44.44%	3665.73%	−27.27%
Si B3	0 krad(Si)	3.758	46.09	17.76
	300 krad(Si)	1.499	1319.00	12.64
	Percentage change	−60.11%	2761.79%	−28.83%

. The threshold voltage Vth and off-state leakage current I_DSS of the two devices are degraded, and the degree of degradation increases with the increase in the irradiation dose. This is due to the fact that the trap charge accumulated in the gate oxide layer of the two types of devices increases with the irradiation dose, causing a depletion zone to appear below the gate, which leads to the degradation of the gate control capability. Comparing the degradation of the threshold voltage and off-state leakage current of the two devices, it is found that the degradation of the SiC device at irradiation doses up to 500 krad(Si) is still less than half of the degradation of the Si device at 200 krad(Si). The analysis shows that SiC MOSFET power devices are more resistant to the TID than hardening Si MOSFET power devices for a variety of reasons, among which the γ-ray irradiation-induced oxide trap charge is one of the main reasons for threshold voltage degradation. The density of oxide trap charges induced by irradiation in SiC devices is much smaller than that in Si devices, which substantially improves the tolerance of SiC devices to the TID, which is also in agreement with related research papers [20,21].

There is no significant change in the transconductance g_m for the SiC devices, whereas a notable shift is observed in the transconductance of the Si devices. The transconductance is proportional to the device carrier mobility, whereas the interfacial state trap charge induced by γ-irradiation at the oxide interface leads to the scattering of channel carriers, which reduces the carrier mobility, and the influence relationship is shown in Equation (1) [22]:

$$\mu ={\displaystyle \frac{{\mu }_0}{1+\alpha \Delta Q_{it}}}$$

(1)

where μ₀ and μ are the carrier mobility before and after γ-ray irradiation, respectively, ΔQ_it is the change in the interfacial state trap charge, and α is an empirical parameter. Due to the large intrinsic interfacial defects in SiC devices, the interfacial trap charge induced by irradiation is not enough to cause a significant change in the mobility, which in turn does not lead to a significant decrease in the transconductance, whereas the interfacial trap charge in Si devices has a large increase, and the decrease in the mobility is relatively obvious, which reacts in the device parameter as a significant decrease in the transconductance.

3.2. Total Ionizing Dose and Neutron Single Event Synergistic Effects

The images of the leakage current with neutron fluence for all the device samples at different voltages and γ-ray doses are given in Figure 3 and Figure 4. The images show that most of the off-state leakage currents (hereinafter referred to as leakage currents) of the SiC devices show obvious and rapid decreases after applying different drain voltages, and tend to stabilize after the neutron fluence reaches 1~1.5 × 10⁸ n·cm⁻², whereas this kind of phenomenon is relatively insignificant in the Si devices. The probable cause of this phenomenon may be the instability of the built-in electric field of the device during the removal of the device sample from the dry ice environment and the return of the temperature to room temperature, and then the leakage current gradually decays to a stable state, so that the subsequent analyses only consider the stable state of the device. The leakage currents of all the samples did not exceed 10⁻⁴ A, i.e., no SEB occurred. Instead, the observed phenomenon was SET, and the number and maximum height of the transient pulses for SET vary significantly with the device material, drain bias and γ-ray dose. Therefore, the discussion of the synergistic effect of the TID and neutron SEE is mainly based on single event transients.

Figure 3 illustrates the variation in the leakage current with drain bias voltage in SiC devices. After the neutron fluence reaches 1.5 × 10⁸ n·cm⁻², a slight increase in the leakage current is observed with an increase in the drain bias voltage at different γ-ray irradiation doses, and the increase in the leakage current at the rated voltage is not more than 100 nA compared with that at 200 V. It is noted that the leakage current at a γ-ray irradiation dose of 200 krad(Si) is always larger than that at 0 and 500 krad(Si), and before the drain bias reaches the rated voltage, the leakage current at a γ-ray irradiation dose of 500 krad(Si) is greater than that at 0 krad(Si). However, at the rated voltage, the leakage current at 500 krad(Si) is actually lower than that at 0 krad(Si). The exact cause of the non-monotonic trends is not clear, but the phenomenon was indeed observed experimentally, and in all the experimental samples. We are still analyzing the cause of this phenomenon, and have not yet found a reasonable and reliable theoretical explanation. In contrast, no similar phenomena were observed in the Si device shown. In Figure 4, the neutron fluence is 1.5 × 10⁸ n·cm⁻², the leakage current increased with the drain bias voltage at all γ-ray irradiation doses, and the leakage current at 300 krad(Si) was always greater than that at 200 and 0 krad(Si). When the γ-ray dose is 200 and 500 krad(Si), the increase in the leakage current at rated voltage in Figure 4d is significantly increased compared with that in Figure 4a–c, and the increase is about 550 nA and 1150 nA, respectively, compared with that in Figure 4a, which is more than that of the SiC devices. Therefore, the gate of the Si devices displays a more substantial weakening of its ability to control the drain current.

In terms of the neutron SET, the neutron transient pulses of the SiC devices are gradually obvious with the increase in the drain bias at γ-ray doses of 0 and 200 krad(Si), but the peak of transient pulses decreases at a dose of 500 krad(Si). In the SiC device leakage current curves, the threshold voltage corresponding to the appearance of the obvious pulse phenomenon decreases with the increase in the γ-ray dose, and the first transient pulse at a γ-ray dose of 200 krad(Si) occurs at a drain bias of 300 V, which is at the neutron fluence of 1.1 × 10⁹ n·cm⁻² in Figure 3b; meanwhile, the first transient pulse at a γ-ray dose of 0 krad(Si) occurs at a drain bias of 400 V, which is at a neutron fluence of 2.6 × 10⁸ n·cm⁻², as shown in Figure 3c.

In Figure 4, the SET phenomenon of the Si device is insignificant until the drain bias reaches the rated voltage, and the transient pulse does not change obviously with the increase in the γ-ray dose, only at Figure 4d a significant transient pulse appears at the rated voltage. This is due to the fact that the radiation hardening used in the Si device plays a great role in suppressing the SET phenomenon, and the doping density of the buffer layer added by the retarded epitaxial process is between the epitaxial layer and the substrate, which can effectively reduce the electric field strength at the epitaxial layer/substrate interface and inhibit the carrier collision ionization rate and the transient currents here, which is also in agreement with the results of related studies [23]. When the drain bias voltage is low, the device carrier collision ionization rate is always limited to a low level, and no obvious transient pulse occurs, while when the drain bias voltage reaches the rated voltage, the elevation of the carrier collision ionization rate by the high electric field strength has already exceeded the limitation of the reinforcement measures, and the height of the transient pulse appears to be significantly increased. In contrast, the unhardened SiC devices show obvious transient pulses when the drain bias voltage is less than half of the rated voltage, as shown in Figure 3b,c.

In Figure 3f, under the rated voltage, the transient pulse height of the SiC device at a γ-ray dose of 0 krad(Si) decreases after a neutron fluence of 1.8 × 10⁹ n·cm⁻², and the phenomenon increases to about 5 × 10⁸ n·cm⁻² at 200 krad(Si), and the pulse height never reaches the level of the former two at 500 krad(Si). The underlying cause of this phenomenon can be attributed to the fact that secondary particles resulting from neutron radiation will produce new composite centers, which reduces the carrier density and mobility in the devices. When the density of the composite center accumulates to a certain value, it will cause the peak transient current to be significantly reduced, and it will overlap with the reduction in the carrier mobility caused by the TID, so the phenomenon will occur sooner after the increase in the γ-ray dose. We used the Geant4 simulation to obtain the distribution of the yield, energy and LET value of the secondary particles produced by the 14 MeV neutron incident SiC devices, as shown in Figure 5. All these secondary particles, to a certain extent, increase the composite center inside the device and affect the carrier concentration and mobility. It can also be seen from Figure 5 that the secondary products have higher average energies for lighter ions such as H and He ions, while heavier ions generally have lower average energies, and the average energies for all types of ions are generally less than half of the maximum energy. This suggests that secondary particles with higher energies and LET values that could potentially trigger SEBs account for only a very small percentage of the secondary particles.

In order to further compare the effects of TID damage on the SET of the two devices, the maximum heights of the transient pulses of the two devices after the neutron fluence reaches 1.5 × 10⁸ n·cm⁻² at the rated voltage were measured, and the results are shown in

Table 5. The maximum SET pulse height of SiC/Si MOSFETs at rated voltage.

γ-Ray Radiation Dose	Maximum Pulse Height/nA (SiC Devices)	Maximum Pulse Height/nA (Si Devices)
0	55.026	81.618
200	56.420	48.527
300	--	81.337
500	13.107	--

. The maximum pulse height of the SiC devices at different γ-ray doses is about 55 nA, which is smaller than that of the Si devices, with a maximum of 80 nA. The trend of the maximum pulse height of the SiC devices with increasing γ-ray doses is also different from that of the Si devices. The maximum heights of the SiC devices at 0 and 200 krad(Si) are almost unchanged, and there is an obvious decrease in the maximum height at 500 krad(Si), whereas the maximum heights of the Si devices at 200 krad(Si) decrease compared to that at 0 krad(Si), but the decrease is still not as large as that of the SiC devices at 500 krad(Si), and the maximum height at 300 krad(Si) instead rises back to the level at 0 krad(Si). We argue that the SET pulse height is determined by the concentration of carriers generated by the ionization of secondary particles produced by neutron nuclear reactions. And the accumulation of composite centers reduces the carrier concentration and mobility of the device, thus reducing the SET pulse height. But the maximum SET pulse height in the Si devices was highest at 200 krad(Si) than at 0 and 300 krad(Si); the cause of the non-monotonic trends is not clear. Analyzing the phenomenon that the transient pulse height of the SiC devices decreases with the increase in the neutron fluence at different γ-ray doses, as well as the statistical results of the maximum height of the transient pulse of the SiC devices, it can be concluded that the neutron SET tolerance of the SiC devices is improved after a certain dose of γ-ray irradiation. Therefore, we predict that there exists the possibility of γ-ray pre-irradiation hardening of SiC MOSFETs.

4. Conclusions

In order to explore the influence of the power MOSFET synergistic effect under the mixed radiation environment of neutrons and γ-rays, a comparative experimental study was conducted to assess the γ TID and neutron SEE for a commercial SiC MOSFET and a hardened Si MOSFET. In the total dose effect experiment, the electrical parameters of the two devices before and after γ-ray irradiation are compared, and it is found that the degradation of electrical parameters such as the threshold voltage and off-state leakage current of the SiC MOSFET are significantly lower than those of Si MOSFET, and the SiC power device has a better ability to tolerate the TID. In the neutron SEE experiment, both devices did not generate SEB, but SET occurred. The transient pulse height and the threshold voltage, where obvious transient phenomena occur, are affected by the TID for both types of devices, and the leakage current growth of SiC devices is much lower than that of Si devices with the increase in the γ-ray dose. The transient pulse peak value of the SiC devices at a rated voltage is also lower than that of the Si devices at a higher γ-ray dose, namely, the SiC MOSFET power devices are less affected by the synergistic effect of the TID and neutron SEE than the Si MOSFET power devices are. The possible reason for this is that the concentration of trap charge induced by irradiation in the SiC devices is smaller than that in the Si devices, which improves the capability of the devices to tolerate the TID, and also effectively reduces the effect of the TID on the neutron SET.