Comparison of Proton and Swift Heavy Ion-Induced Traps in GaN and β-Ga₂O₃

GaN and Ga₂O₃ have the potential for insertion into high-radiation environments due to their large bandgaps, strong bonds, superior power and/or RF devices compared with incumbent technologies, and potential radiation hardness compared with incumbent and other next generation materials, but a complete understanding of the degradation mechanisms is not well established. To wit, we investigated 946 MeV Au swift heavy ion irradiation of n-type ammonia molecular beam epitaxy (MBE)-grown GaN diodes and compared this with 1.8 MeV proton irradiation results and also with the same radiation conditions of n-type hydride vapor epitaxy (HVPE)-grown β-Ga₂O₃ diodes as well. The GaN diodes used Ti/Al/Ni/Au Ohmic contacts, 440 x 440 μm semi-transparent 80 Å Ni Schottky contacts, and were grown on a GaN-on-sapphire template with a heavily doping Ohmic contact layer and lightly-doped (n = 3 x 10¹⁶ cm^-3) layer where the defects were characterized for highest sensitivity. The β-Ga₂O₃ sample details were essentially the same, but the structure was a 1 x 10¹⁶ cm^-3 Si-doped 10 μm thick hydride vapor phase epitaxial (HVPE) layer was grown on an (001) n+ Sn-doped β-Ga₂O₃ substrate by Novel Crystal Technology. Using deep level transient and optical spectroscopies (DLTS/DLOS) to characterize the traps throughout the GaN bandgap, the Au irradiation resulted in two broad DLTS peaks with energies of ~E_C-0.14 and ~E_C-0.18 eV at the lowest fluence. At the higher fluences, these two traps/peaks merged into one very broad peak, but it could be fit well using extracted peak temperatures for both the ~E_C-0.14 and ~E_C-0.18 eV traps from the lowest fluence sample. It was observed both trap concentrations increased with Au fluence and the extracted trap introduction rates of ~7.0 x 10⁵ and ~4.4 x 10⁵ cm^-1, respectively. These two broad peaks are indications that these trap energies are not discrete energies but traps with energies centered around 0.14 and 0.18 eV with an energy range of ~31 meV. This contrasts the proton irradiation results where trap concentrations for traps at E_C-0.13, 0.16, 1.25, 2.5, and 3.25 eV increased with proton fluence, but the traps energies remained discrete [Z. Zhang, Appl. Phys. Lett. 103, 042102 (2013)]. The 1.8 MeV proton damage in the top ~20 μm will generally be isolated point defects while the swift heavy ion irradiation will produce localized regions of high disorder in an otherwise undamaged material. The discreteness of the trap energies from proton irradiation are consistent with the proton-induced isolated point defect formation while the broad peaks from Au irradiation will result in slightly different bond lengths and/or angles that would produce a range of trap energies, which is consistent with the DLTS results. The Au irradiation-induced carrier removal rate (CRR) was 2.5 x 10⁶ cm^-1. Comparing this with β-Ga₂O₃ where the rate was 8.0 x 10⁶ cm^-1 reveals interesting differences especially when comparing with the measure proton CRRs where both GaN and β-Ga₂O₃ have very comparable rates of 270 and 350 cm^-1 [Z. Zhang, Appl. Phys. Lett. 103, 042102 (2013), J. McGlone, J. Appl. Phys. 133, 045702 (2023)]. These different relative responses for GaN and β-Ga₂O₃ indicate the radiation-induced damage is complex and still needs to be better understood including field and doping effects, for example. The next steps are to determine if any of these traps lead to increased gate leakage current or contribute to degradation and single events in transistors based on these materials. This work was supported in part by the Air Force Center of Excellence in Radiation Effects, award No. FA9550-22-1-0012, and from AFOSR award No. FA9550-22-1-0165.