GOX 2023 Session AC+DI+HM+TM-MoA: Characterization/Modeling II

The physics of GaAs (zincblende structure) and Gallium nitride (wurtzite structure) led to disruptive technologies driven by extreme properties such as small effective mass, large direct bandgap, and piezoelectric polarization. Gallium reappears in a monoclinic-structure oxide with enormous prospects for applications in power electronics for the future of green energy. Numerous new phenomena hitherto unknown for traditional semiconductors occur in monoclinic symmetry semiconductors such as non-parallel phonon-plasmon scattering, hyperbolic shear polaritons, splitting of associated transverse and longitudinal phonon modes, non-degenerate highly anisotropic fundamental excitonic band-to-band transitions, direction-dependent band alignments, and complex defect spin interactions within the highly anisotropic host lattice. The influences of composition, strain, doping, and defects are discussed for Ga2O3 and related alloys, and special emphasis is paid to new semiconductor phenomena, and consequences for thin film growth and device designs are pointed out. Methods such as generalized ellipsometry, the optical Hall effect, Terahertz electron paramagnetic resonance ellipsometry, and density functional theory computations are employed for characterization and analysis.

β-Ga₂O₃ is an ultrawide bandgap semiconductor that has attracted interest for use in high-power electronics due to its theoretical breakdown field of 8 MV/cm and availability of native bulk substrates. Integration of high-quality insulator layers is critical to realizing the material breakdown fields of β-Ga₂O₃ devices. Recent work has demonstrated the advantages in using hybrid extreme-k/low-k insulator stacks to enable high average electric fields in the β-Ga₂O₃ [N.K. Kalarickal et al. IEEE EDL (2021)]_. In this work, we study such hybrid insulator stacks. Four different insulator stacks were used to fabricate metal-oxide-semiconductor capacitors (MOSCAPs) on top of (010) β-Ga₂O₃ Sn-doped substrates. The first set of samples consisted of a 20 nm Al₂O₃ low-k layer and BaTiO₃ (BTO) layers of 20 and 35 nm, respectively. The second two samples used 24 nm of SiO₂ with BTO layers of 35 nm and 50 nm, respectively. Al₂O₃ and SiO₂ were deposited by plasma-assisted atomic layer deposition after a Piranha/HF clean [A.E. Islam et al. IEEE TED (2022)]. BTO layers were deposited by RF sputtering. All thicknesses were confirmed by ellipsometry of coloaded Si. The backside and top contacts were deposited by e-beam evaporation to complete fabrication. The samples were analyzed in terms of breakdown strength, C-V characteristics and supported reverse field in the Ga₂O₃. The sample with 50 nm of BTO on 24 nm SiO₂ resulted in a larger oxide capacitance, likely due to the increased permittivity of the BTO due to higher thermal budget. For breakdown measurements, a leakage current of 1 mA/cm² was defined as the breakdown limit for all devices. The field in the Ga₂O₃ was calculated by extracting the effective donor density from C-V and under the assumption of no charge in the dielectric layers. The sample with 20 nm Al₂O₃ and 20 nm BTO showed the best performance, with a forward breakdown electric field of 5.7 MV/cm in the Al₂O₃ and reverse breakdown field of 6.8 MV/cm in the β-Ga₂O₃. This is one of the highest reverse breakdown fields reported to date for Gallium Oxide in such a junction and shows the potential of hybrid low-k/high-k stacks to enable extreme fields in ultra-wide bandgap semiconductors. In summary, we have studied the properties of hybrid low-k/high-k insulator stacks on β-Ga₂O₃ and shown that such stacks can enable excellent reverse breakdown fields in β-Ga₂O₃ power devices. We acknowledge funding from Department of Energy / National Nuclear Security Administration under Award Number(s) DE-NA0003921, and AFOSR GAME MURI (Award No. FA9550-18-1-0479, project manager Dr. Ali Sayir).

We demonstrate successful blistering of He-implanted (001) β-Ga₂O₃, bonding to (0001) 4H-SiC, and initial results towards large-area transfer of (001) β-Ga₂O₃ to SiC. Compatible with large-scale processing, exfoliation is an important step in controlled-thickness transfer of films for heterogeneous integration. Furthermore, integration of β-Ga₂O₃ to high thermal conductivity materials will be crucial for thermal management of high-power devices. Two-inch (001) β-Ga₂O₃ wafers were implanted with He⁺ at an energy of 160 keV and a dose of 5×10¹⁶ cm^-2, mimicking our previous parameters used to successfully exfoliate (010) β-Ga₂O₃.¹ Strain fringes were observed after the implant which corresponded to a maximum strain of ~1.7%, which is higher than the ~1% strain when implanting (010) β-Ga2O3¹ likely due to the larger Poisson’s ratio of (001) β-Ga₂O₃.² The implanted substrates were then bonded to (0001) SiC at room temperature using a ~5 nm Ti interlayer to assist with the bond. Both unbonded implanted substrates and the bonded structures were first annealed at 200 °C for 10 hours to simultaneously initiate He bubble nucleation at the projected range (~0.68 μm) and strengthen the bond. Then, even after annealing at 500 °C for 10 hours to initiate bubble growth, the (001) β-Ga₂O₃ did not transfer to the (0001) 4H-SiC. Unlike what was observed for (010) β-Ga₂O₃,¹ blistering did not even occur at this temperature. Annealing at 800 °C for up to 12 hours resulted in ~10 μm blisters for the unbonded substrates, which is typically an indication that large wafer-area transfer can be achieved if bonding was done prior to annealing.³ However, the β-Ga₂O₃ substrate did not wafer split from the bonded structure, and instead only small area transfers up to ~200 μm were achieved. Only ~7% of the total bonded area transferred while the entire structure remained bonded. Strategies to improve transfer will be presented, including initiating a cold split prior to exfoliation and refined annealing strategies to improve He bubble nucleation and larger bubble growth at the projected range. These are promising results towards achieving large wafer-scale (001) β-Ga₂O₃ composite wafers, and when combined with subsurface damage-free chemical mechanical polishing,⁴ these composite wafers would be suitable for subsequent devices and/or epitaxial growth.

References

1. M.E.Liao,et al., ECS J.Sol.State Sci.Technol., 8(11) P673 (2019)

2. K.Adachi,et al., J.Appl.Phys., 124 085102 (2018)

3. M.Bruel,et al., Jpn.J.Appl.Phys., 36 1636 (1997)