Group III nitrides, particularly GaN, have seen widespread use in various devices due to their superior properties compared to silicon-based counterparts. GaN-based high-electron mobility transistors (HEMTs) exhibit higher efficiency, operate at higher voltages, and feature faster switching transitions than their silicon counterparts. Furthermore, GaN HEMTs possess better thermal stability and show greater resistance to radiation, making them particularly attractive for applications in space and high-energy physics. The European Space Agency (ESA) is already leveraging GaN HEMTs in its Porba-V and Biomass satellites, demonstrating improved data transmission rates and reliability. However, the extreme radiation environments encountered in these applications present significant challenges. While the effects of low- to medium-energy ions on GaN are reasonably well-understood, the high-energy radiation regime remains less certain. Space applications are especially interested in high atomic number Z and high energy (HZE) ions, a significant component of galactic cosmic radiation (GCR) and solar radiation. Despite lower abundance than protons or gamma radiation, HZE ions (such as Fe and Si) have high ionization power and fluxes, potentially accelerating degradation and causing catastrophic device failure. This communication focuses on the fundamental interactions of strongly ionizing particles with GaN, specifically its response to swift heavy ion (SHI) radiation. SHI radiation, with energies in the MeV range per nucleon, primarily loses energy through electronic interactions, inducing ionization spikes and latent track formation. Previous studies presented conflicting findings about the effects of strongly ionizing radiation on GaN; some suggesting amorphization while others demonstrated a defective but crystalline structure even at high fluences. This discrepancy highlights the need for further investigation into the underlying mechanisms.

Several studies have investigated the effects of ion irradiation on GaN. Some have used transmission electron microscopy (TEM) to study GaN films irradiated with SHIs, concluding that SHIs are capable of amorphizing the crystal. This contrasts with other work using Rutherford backscattering spectrometry/channeling (RBS/C), which showed that even with high fluences, a defective but still crystalline structure remained. This conflicting evidence underscores the complexity of damage formation and annihilation mechanisms in GaN under high-energy irradiation. Existing RBS/C models, like Poisson and Gibbons models, do not adequately account for the observed behavior, suggesting complex interactions and recovery processes are at play. This research aims to resolve this discrepancy by employing a multi-faceted approach combining experimental techniques and advanced computational modeling to analyze the interaction of SHIs with GaN.

This study employs a two-temperature model-molecular dynamics (TTM-MD) simulation scheme to investigate the interaction of SHIs with GaN. The TTM simulates the initial femtoseconds of the SHI interaction, depicting the ionization spike as a hot electron gas that relaxes its energy into the lattice. This lattice energy profile is then input into a molecular dynamics (MD) simulation to capture the structural dynamics, such as phase transitions and defect dynamics. The TTM parameters for GaN, including heat capacity and conductivity, were carefully considered, including the incorporation of an effective mass theory for semiconductors. A key parameter in the TTM, the electron relaxation time, was estimated to be 85 fs. The MD simulations were performed using the PARCAS code with the Albe-Nord potential, and the simulations were conducted for both bulk and surface regions to account for surface effects. The thickness of the GaN films used in the simulations were matched to the experimental conditions. The experimental work involved irradiating 3-µm thick GaN films on Al2O3 substrates with various SHIs: 185 MeV Au ions (at the Australian National University Heavy Ion Accelerator Facility), and 70 and 45 MeV Xe ions (at the Grand Accélérateur National d'Ions Lourds (GANIL)). Different fluences were employed to study the impact of the radiation dose. Multiple experimental techniques were used to characterize the irradiated samples: high-resolution transmission electron microscopy (HRTEM) and scanning transmission electron microscopy (STEM) in high-angle annular dark-field (HAADF) mode provided high-resolution images of the ion tracks. Electron energy-loss spectroscopy (EELS) was used to determine the density profiles within the tracks. RBS/C measurements were performed to quantify the damage. To simulate the experimental conditions, the RBSADEC code, a Monte Carlo implementation of a binary collision algorithm was applied using the final atomic structures obtained from the MD simulations. This allows for the accurate prediction of RBS/C spectra, taking into account the complex track morphologies such as extended defects, density gradients, and voids.

The simulations revealed a robust recrystallization effect as a primary mechanism for GaN's resistance to SHI radiation. The initial SHI impact induces a solid-liquid phase transition, forming a cylindrical molten track. The subsequent cooling and high-pressure conditions lead to rapid recrystallization starting from the track's edge and proceeding inwards. However, the formation of extended defects interrupts complete recrystallization, leading to a damaged yet highly crystalline shell surrounding a smaller amorphous core. This observation aligns exceptionally well with experimental TEM-HAADF images, demonstrating a remarkable agreement between the simulations and experimental results in terms of track dimensions, morphology, and defect structures. The dislocation extraction algorithm (DXA) confirmed the presence of dislocation edges and stacking faults within the recrystallized region, further correlating simulations and experimental cross-sectional TEM-HAADF images. The EELS data corroborated the simulation results, showing excellent agreement between experimental and predicted density profiles within the ion tracks. Near the surface, a pressure gradient and sputtering resulted in the formation of voids and a hillock. The simulated cross-sectional TEM-HAADF images again reflected the presence, shape, and density of these voids. A crucial finding is the efficient recrystallization that also occurs when SHIs impact pre-damaged material. This significantly affects the damage evolution with fluence, as the amorphous core of an existing track is recrystallized even by an indirect overlap from a second SHI impact. The recrystallization cross-section (radius ≈ 6.5 nm) is larger than the initial melting cross-section (radius ≈ 4.4 nm). This extensive recovery mechanism helps explain why the GaN samples do not fully amorphize even at high fluences. The RBS/C measurements demonstrated that even at the highest fluence (1 × 10¹³ cm⁻²), the sample retained its single-crystalline nature. Simulations using the RBSADEC code with the track morphologies generated from the MD simulations, including surface voids and density gradients, showed excellent agreement with the experimental RBS/C spectra. Furthermore, simulations on lower energy SHIs (70 and 45 MeV Xe ions) showed similar recrystallization dynamics, with a reduced track size and fewer extended defects compared to the 185 MeV Au ions. RBS/C data confirmed the minimal damage caused by the Xe ions at a fluence of 2 × 10¹² cm⁻².

The remarkable agreement between the simulations and the experimental TEM, EELS, and RBS/C results validates the employed TTM-MD modeling approach and highlights its predictive power. The simulations successfully unravel the key mechanism behind GaN's radiation resistance, identifying the strong recrystallization effect as the dominant factor. This effect not only reduces the damage within a single track but also efficiently recovers damage from previous ion impacts, mitigating the cumulative effect of high fluences. The model's ability to accurately predict the formation of various defects, including point and extended defects, amorphous cores, density gradients, and voids, is significant. This provides a valuable tool for predicting the behavior of GaN under extreme radiation conditions, reducing reliance on extensive and expensive experimental characterization. The results for 185 MeV Au ion irradiation are comparable to findings in GaN HEMTs irradiated with 1540 MeV Bi ions. The energy loss of the Fe component of HZE radiation (peaking at ~14 keV/nm) is similar to the 45 MeV Xe ions in this study, suggesting the model can be extended to predict properties like failure rates and dose thresholds in space applications. Moreover, the model can provide insights for SHI-based material engineering.

This research successfully demonstrates the efficacy of a combined experimental and computational approach to understand GaN's remarkable resistance to strongly ionizing radiation. The strong recrystallization effect, revealed through TTM-MD simulations and corroborated by experimental data, is the key mechanism mitigating radiation damage. This model accurately predicts various defect types and morphologies, and can be used to design and test future radiation-hardened GaN devices for challenging applications like space technology. Future work could focus on refining the TTM parameters for even greater accuracy and exploring more complex radiation scenarios, including mixed ion species and energy distributions.

The TTM-MD simulations rely on several approximations, including the use of equilibrium thermodynamic quantities to model a non-equilibrium process. While the results indicate significant agreement with experimental observations, minor deviations could arise due to limitations of the TTM. Furthermore, the MD simulations have a time limitation that might not fully capture the long-term evolution of point defects, particularly their mobility at lower temperatures. The RBSADEC simulations assume a certain distribution of surface voids to account for experimental resolution, which introduces some uncertainty in the quantitative analysis. Lastly, the current model does not include effects such as surface oxidation or other chemical interactions that could influence GaN behavior under prolonged radiation exposure.