Oxygen vacancies and local amorphization introduced by high fluence neutron irradiation in β-Ga₂O₃ power diodes

Owing to its ultra-wide bandgap (4.5–4.9 eV), high critical breakdown field (8 MV/cm), large Baliga's figure-of-merit and high thermal stability, β-Ga₂O₃ has emerged as a promising candidate material for high-efficiency power switching applications.^1,2 In recent years, β-Ga₂O₃ power devices have undergone rapid development with notable progress made in the design and fabrication of high-performance Schottky barrier diodes (SBDs) and PN heterojunction diodes.^3–5 With the support of the terminal structure, such as thermally oxidized termination,⁶ field-plated trench,⁷ and high-k field plate,⁸ the performance of the diode is gradually approaching the theoretical limit of β-Ga₂O₃. Additionally, there have been achievements in the realization of enhancement-mode transistors.^9–12 These research findings suggest promising opportunities for the application of β-Ga₂O₃ devices.

β-Ga₂O₃ power devices have potential to withstand harsh operating conditions encountered in nuclear and space missions, which makes them a promising candidate for these applications.^13,14 In radiation environments, energetic particles, such as neutrons and protons, can induce displacement damage (DD) in the devices, leading to performance degradation.^15–17, β-Ga₂O₃ has been shown to have possible better radiation hardness compared to other semiconductors such as Si and GaN. This is attributed to the higher displacement energies (E_d) of β-Ga₂O₃ atoms, where E_d (Ga) and E_d (O) are 25 and 28 eV, respectively.¹⁸ In contrast, Si has an E_d value of 13 eV,¹⁹ and GaN has E_d values of 20 and 10–20 eV for Ga and N atoms, respectively.²⁰ The higher E_d values of β-Ga₂O₃ result in a greater resistance to vacancy defects and allow it to maintain stable electrical properties even after exposure to high radiation doses. Nonetheless, further studies are needed to fully understand the effects of radiation on β-Ga₂O₃ devices and to optimize their performance for specific applications. The use of deep level optical spectroscopy (DLOS) and deep level transient spectroscopy (DLTS) enables the determination of the concentration and thermal emission rate of semiconductor deep levels by capacitance characteristics.^21,22 Thus, the evolution of defects in β-Ga₂O₃ has been studied before and after exposure to proton and neutron irradiation using DLOS and DLTS.^18,23–26 Furthermore, the researchers demonstrated that β-Ga₂O₃ transistors and diodes are able to withstand radiation under the high dose of gamma ray, suggesting their potential for use in space missions and other relevant scenarios.^27–30 However, the physical source underlying defect formation, defect structure, and microscopic morphology by irradiation remains subjects of ongoing investigation. Therefore, this paper aims to investigate the degradation mechanism of β-Ga₂O₃ SBDs under neutron irradiation.

In this work, we applied a high dose of neutron irradiation whose schematic and transmission electron microscope (TEM) cross sections are shown in Fig. 1(a). Based on the results of energy dispersive spectroscopy (EDS) tests, the diffusion of Ni did not occur within the error range (±2%). Thus, the phenomenon observed in this paper is not caused by the diffusion of metal. SBDs were produced on 10 μm epitaxial layers utilizing a halide vapor phase epitaxy (HVPE) growth method on (001) β-Ga₂O₃ bulk substrates (HVPE_SUB) with a doping concentration of 5.5 × 10¹⁸ cm⁻³ by edge-defined film-fed (EFG) purchased from Novel Crystal Technology (Japan). The anode and cathode of the SBDs were formed using Ni/Au (20/100 nm) with a radius of 100 μm and Ti/Au (20/100 nm) metal stacks, respectively, which were deposited on the front and back sides of the sample using electron-beam evaporation. Neutron irradiation experiments were carried out in the JSI research reactor of TRIGA type in Ljubljana.³¹ The neutrons used for irradiation were derived from a distribution spectrum, which, when normalized based on non-ionization energy loss, corresponds to neutron irradiation with energies ranging from 1 to 2 MeV. The total dose rate applied was 1.3 × 10¹² cm⁻² s⁻¹, resulting in three total fluence of 8 × 10¹⁴ cm⁻² (8E14), 1.5 × 10¹⁵ cm⁻² (1.5E15), and 2.5 × 10¹⁵ cm⁻² (2.5E15). The SBD subjected to an irradiation dose of 8 × 10¹⁴ demonstrates a comparatively smaller performance degradation when compared to the other two doses as shown in Fig. 1(b). Moreover, the distinction in performance degradation between the doses of 1.5E15 and 2.5E15 is already minimal. Nevertheless, the device exhibited complete failure at all three neutron fluence. This phenomenon can be attributed to the complete degradation of the device, indicating that it has reached the maximum level of damage resulting from neutron irradiation. In addition, since the damage inflicted on β-Ga₂O₃ by the three different irradiation fluences does not exhibit a consistent trend and within the margin of error can be considered indistinguishable, we only show material and device characterization at a dose of 1.5 × 10¹⁵ cm⁻² in the following text. The displacement damage dose, D^disp, is considered to account for the displacement damage events in irradiation experiments, which is a measure of imparted energy per unit mass expected from a neutron field. When assuming an equivalent neutron displacement damage cross section (⁠ $σ disp$⁠) of 92.3 MeV·mb, the D^disp is calculated to be 2.01 × 10¹¹, 3.77 × 10¹¹, and 6.28 × 10¹¹ MeV/g for three total fluence, respectively. The mean free path for scattering interaction of the neutrons (λs) was determined to be greater than 1 cm for the neutron energy range of 1–2 MeV in β-Ga₂O₃, and the neutron energy distribution of the incident and outgoing β-Ga₂O₃ SBDs was basically unchanged in this work using Monte Carlo N-particle transport code (MCNP) software.³² Therefore, if the neutrons are incident from the bottom, their effect on the experimental results would be insignificant.

The forward current of SBD after irradiation (SBD_AI) has a serious performance degradation compared to that before irradiation (SBD_BI) as shown in Fig. 2(a). After irradiation, the differential specific on-resistance (R_on,sp) of the SBD increased significantly from 3.9 to 3.5 × 10⁸ mΩ·cm² and the on/off ratio which is defined as I_{@2 V}/I_{@−2 V} decreased from 10¹¹ to 10³. On the contrary, as Fig. 2(b) shows, by the H. C. Montgomery method,³³ the carrier concentration of the device substrate did not decrease significantly before and after irradiation (HVPE_SUB_BI and HVPE_SUB_AI) owing to the substantial doping of the substrate. From this result, the epitaxial layer has a very serious electrical performance degradation, which can also be verified from the capacitance–voltage (C–V) results as shown in Fig. 2(c). According to the results of C–V measurements, the net doping concentration (N_D) of the HVPE epitaxial layer prior to irradiation is approximately 1.67 × 10¹⁶ cm⁻³. In contrast, the SBD_AI exhibits a low capacitance that remains constant across different reverse voltages, suggesting that the epitaxial layer has a very low carrier concentration. Thus, β-Ga₂O₃ SBDs with HVPE epitaxial layers were completely damaged under this irradiation doses. Both the epitaxial layer and the substrate experience a decrease in the carrier concentration. However, owing to the substantial doping of the substrate, the impact of this reduction may not be readily evident.

In addition, the block and breakdown characteristics are optimized after irradiation, which is mainly reflected in the increase in the breakdown voltage as shown in Fig. 2(d). Although the reverse characteristic does not display any degradation or even indicates an “improvement,” it does not hold any substantial significance when compared to a fully compensated epitaxial layer. Thus, the primary emphasis of this paper is to investigate the underlying physical mechanisms behind device failure caused by neutron irradiation with the aim of offering insight and guidance for future radiation-resistant device design. Some parameters of SBD_BI and SBD_AI are shown in Table I.

Neutron irradiation commonly induces DD within materials. Therefore, we characterized the crystal quality and defects of the epitaxial layer material to explore the degradation mechanism of the devices. As shown in Fig. 3(a) via x-ray diffraction (XRD), the quality of the epitaxial layer crystal has significantly deteriorated after irradiation. This deterioration is reflected in the decrease in the intensity and the increase in the full width at half maximum (FWHM) observed in Fig. 3(b). The XRD intensity of the epitaxial layer significantly decreased after irradiation, indicating a reduction in the crystal coherence length. Furthermore, a previously unobserved diffraction peak at 2θ = 37.99° emerged in the XRD spectrum.³⁴ The analysis suggests that the irradiation of the epitaxial layer results in the formation of polycrystalline or amorphous regions. This is consistent with the observed increase in the FWHM of the x-ray rocking curve peaks from the (002) plane of the (001) crystallographic orientation as shown in Fig. 3(b), which indicates a significant decrease in the crystallinity of the material. On the contrary, despite the severe degradation of crystal quality in the epitaxial layer, the substrate retains its excellent single crystal orientation, as depicted in Figs. 3(c) and 3(d). After irradiation, the substrate continues to exhibit a single crystal orientation (001), and there is no substantial increase in FWHM, confirming its retention as a well-organized single crystal material. This observation highlights the critical distinction between the substrate and epitaxy in terms of their resistance to neutron irradiation, primarily stemming from the disparity in crystal quality. HVPE epitaxy is more susceptible to the degradation of crystal quality, whereas the substrate maintains its good single crystal characteristics even after irradiation.

In order to more intuitively observe the effect of neutron irradiation on the interface and interior of the device, close-up high-resolution transmission electron microscopy (HRTEM) images, together with their corresponding fast Fourier transform (FFT) patterns, were used to analyze typical regions of the epitaxial layer (labeled by dashed boxes in the HRTEM images) as shown in Fig. 4. The analysis revealed the appearance of darker, streaky areas within the crystal, as shown in Fig. 4(b), which indicates an increased disorder of atomic arrangement. The analysis of the single crystal diffraction (SCD) spectrum of the epitaxial layer following irradiation indicated a higher level of background noise, suggesting an increase in the amorphous components of the material. This finding is consistent with the conclusion drawn from the rocking curves analysis. Additionally, the SCD revealed the appearance of a polycrystalline diffraction ring after irradiation, which is in line with the emergence of observed peaks in the XRD results. These observations provide further evidence for the transformation of the material from a single crystal to a polycrystalline structure due to the irradiation.

In order to clarify the cause of these darker streaky areas which is not found in HVPE_BI, we conducted an EDS test as shown in Fig. 4(c). The perpendicular direction to these streaks exhibits a significant variation in the ratio of Ga and O atoms, but the characteristic did not appear in the EDS results for the substrate. Therefore, under the irradiation of high fluence neutrons, the HVPE epitaxial layer would appear striped lattice damage along the injection direction of neutrons, and the width of these stripe shadows is about 2.4 nm. On the contrary, the TEM result of the substrate shows that the atoms are arranged orderly after irradiation as depicted in Fig. 4(d), which illustrates that the substrate exhibits better radiation tolerance in terms of crystal quality.

In order to find the cause of these changing atomic ratio, the x-ray photoelectron spectroscopy (XPS) measurement was carried out, and peak fitting of the Ga_3d and O_1s core levels was analyzed as shown in Figs. 5(a)–5(f). The standard binding energy of 284.8 eV was chosen for the C_1s peak as the reference point.³⁵ The Ga_3d peak located at 19.76 eV, corresponding to the peaks due to the Ga–O bonding core level and Ga elements mainly exist in the form of Ga³⁺. The remaining peaks located at 19.36 eV exist in the form of Ga²⁺. Similarly, the O_1s core levels were divided into two peaks at positions 530.54 and 531.86 eV, representing O_I and O_II, respectively. Through the result of XPS, the proportion of Ga²⁺ increases from 0.43 to 0.69, and the ratio of Ga/O increases from 0.7 to 0.8, which means that the proportion of the GaO chemical state increases most likely due to the creation of oxygen vacancies. Following irradiation of both the substrate and HVPE epitaxy, there is an observed increase in the O_II/(O_I + O_II) ratio. This increase signifies a rise in oxygen vacancies within the material. Additionally, these oxygen vacancies may act as deep-level defect to compensate carriers, resulting in a decrease in the carrier concentration. Figures 5(e) and 5(f) present the XPS results obtained from both the irradiated and unirradiated substrates of the HVPE layer. The observed outcomes exhibit a resemblance to those observed in the HVPE epitaxy. Specifically, the proportion of O_II was found to increase after irradiation, indicating a notable rise in the content of oxygen vacancies, which verifies the reason for the similar decrease in the carrier concentration of substrate and epitaxial. Furthermore, the conclusions derived from the XPS test indicate no discernible difference between HVPE epitaxy and the substrate regarding the influence of chemical composition. Therefore, the radiation tolerance of the substrate is only manifested in the structural characteristics but not from the electrical one.

Based on the analysis, the degradation of device performance is primarily attributed to the deterioration of crystal quality and the generation of oxygen vacancies. To address this issue, we conducted an annealing process on the device, subjecting it to a temperature of 300 °C for a duration of 30 min within an oxygen atmosphere.

From Fig. 7(a), the forward characteristics of the irradiated SBD exhibit noticeable restoration after annealing (SBD_AI_AA). Figure 7(b) reveals a decrease in the proportion of O_II, suggesting a significant reduction in oxygen vacancies. Moreover, the quality of the epitaxial layer after annealing has been significantly improved: the intensity of the peaks has been greatly increased and the peak position corresponding to (001) has also appeared which meant that the amorphous regions are recovered to the ordered crystal structure. However, the additional crystal orientation peak cannot be eliminated after annealing as shown in Fig. 7(c).

In summary, we carried out the neutron irradiation experiments on β-Ga₂O₃ SBDs at a dose rate of 1.3 × 10¹² cm⁻² s⁻¹. From the test results of I–V and C–V, the R_on,sp of the device increases due to the drastic reduction in the carrier concentration of the epitaxial layer. By utilizing a combination of characterization techniques such as XRD, TEM, EDS, and XPS, we can infer the damage mechanisms induced by neutron irradiation in β-Ga₂O₃ devices. Upon exposure to a high flux of neutrons, the HVPE layer experiences displacement of oxygen atoms from their original positions by nuclear interactions, leading to the formation of a substantial number of oxygen vacancies. These vacancies contribute to the formation of deep-level defects that act as compensating centers for carriers in the material. Additionally, the β-Ga₂O₃ lattice undergoes severe disruption due to the bombardment of a large number of neutrons, resulting in displacement damages that take the form of striped lattice damage oriented along the direction of incident neutrons once the system has reached thermal equilibrium. These amorphous/polycrystalline regions will lead to a decrease in the carrier mobility, combined with a decrease in the carrier concentration, which eventually leads to a severe degradation of device performance. Power devices often necessitate low carrier concentrations to achieve higher breakdown voltages. However, lower carrier concentrations lead to a more pronounced carrier removal effect caused by neutron irradiation. Addressing this challenge requires a careful balance between the desired carrier concentration for breakdown voltage requirements and the need to mitigate the detrimental effects of neutron radiation. Approaches, such as enhanced crystal quality, reduce the proportion of Ga elements, and doping of elements with substantial neutron capture cross sections, such as ¹⁰B or ⁶Li, may offer potential solutions to achieve a more favorable compromise between these conflicting requirements in power device applications.

This work was supported by the Fundamental Research Plan under Grant No. JCKY2020110B010, the National Natural Science Foundation of China under Grant Nos. 61925110, 62234007, U20A20207, 62004184, and 62004186, the Key-Area Research and Development Program of Guangdong Province under Grant No. 2020B010174002, the University of Science and Technology of China (USTC) under Grant Nos. YD2100002009 and YD2100002010, the Collaborative Innovation Program of Hefei Science Center, Chinese Academy of Sciences (No. 2022HSC-CIP024), and the Opening Project of and the Key Laboratory of Nanodevices and Applications in Suzhou Institute of Nano-Tech and Nano-Bionics of CAS. This work was partially carried out at the Center for Micro and Nanoscale Research and Fabrication of USTC.