Gallium oxide (Ga2O3) is emerging as a viable candidate for certain classes of power electronics, solar blind UV photodetectors, solar cells, and sensors with capabilities beyond existing technologies due to its large bandgap. It is usually reported that there are five different polymorphs of Ga2O3, namely, the monoclinic (β-Ga2O3), rhombohedral (α), defective spinel (γ), cubic (δ), or orthorhombic (e) structures. Of these, the β-polymorph is the stable form under normal conditions and has been the most widely studied and utilized. Since melt growth techniques can be used to grow bulk crystals of β-GaO3, the cost of producing larger area, uniform substrates is potentially lower compared to the vapor growth techniques used to manufacture bulk crystals of GaN and SiC. The performance of technologically important high voltage rectifiers and enhancement-mode Metal-Oxide Field Effect Transistors benefit from the larger critical electric field of β-Ga2O3 relative to either SiC or GaN. However, the absence of clear demonstrations of p-type doping in Ga2O3, which may be a fundamental issue resulting from the band structure, makes it very difficult to simultaneously achieve low turn-on voltages and ultra-high breakdown. The purpose of this review is to summarize recent advances in the growth, processing, and device performance of the most widely studied polymorph, β-Ga2O3. The role of defects and impurities on the transport and optical properties of bulk, epitaxial, and nanostructures material, the difficulty in p-type doping, and the development of processing techniques like etching, contact formation, dielectrics for gate formation, and passivation are discussed. Areas where continued development is needed to fully exploit the properties of Ga2O3 are identified.

A review of Ga2O3 materials, processing, and devices

Solar-blind photodetectors are of great interest to a wide range of industrial, civil, environmental, and biological applications. As one of the emerging ultrawide-bandgap semiconductors, gallium oxide (Ga2O3) exhibits unique advantages over other wide-bandgap semiconductors, especially in developing high-performance solar-blind photodetectors. This paper comprehensively reviews the latest progresses of solar-blind photodetectors based on Ga2O3 materials in various forms of bulk single crystal, epitaxial films, nanostructures, and their ternary alloys. The basic working principles of photodetectors and the fundamental properties and synthesis of Ga2O3, as well as device processing developments, have been briefly summarized. A special focus is to address the physical mechanism for commonly observed huge photoconductive gains. Benefitting from the rapid development in material epitaxy and device processes, Ga2O3-based solar-blind detectors represent to date one of the most prospective solutions for UV detection technology towards versatile applications.

/pdf/review-of-gallium-oxide-based-solar-blind-ultraviolet-2iszxfymnr.pdf

Review of gallium-oxide-based solar-blind ultraviolet photodetectors

First‐principles electronic structure calculations on wurtzite AlN, GaN, and InN reveal crystal‐field splitting parameters ΔCF of −217, 42, and 41 meV, respectively, and spin–orbit splitting parameters Δ0 of 19, 13, and 1 meV, respectively. In the zinc blende structure ΔCF≡0 and Δ0 are 19, 15, and 6 meV, respectively. The unstrained AlN/GaN, GaN/InN, and AlN/InN valence band offsets for the wurtzite (zinc blende) materials are 0.81 (0.84), 0.48 (0.26), and 1.25 (1.04) eV, respectively. The trends in these spectroscopic quantities are discussed and recent experimental findings are analyzed in light of these predictions.

Valence band splittings and band offsets of AlN, GaN and InN.

Solid state UV emitters have many advantages over conventional UV sources. The (Al,In,Ga)N material system is best suited to produce LEDs and laser diodes from 400 nm down to 210 nm—due to its large and tuneable direct band gap, n- and p-doping capability up to the largest bandgap material AlN and a growth and fabrication technology compatible with the current visible InGaN-based LED production. However AlGaN based UV-emitters still suffer from numerous challenges compared to their visible counterparts that become most obvious by consideration of their light output power, operation voltage and long term stability. Most of these challenges are related to the large bandgap of the materials. However, the development since the first realization of UV electroluminescence in the 1970s shows that an improvement in understanding and technology allows the performance of UV emitters to be pushed far beyond the current state. One example is the very recent realization of edge emitting laser diodes emitting in the UVC at 271.8 nm and in the UVB spectral range at 298 nm. This roadmap summarizes the current state of the art for the most important aspects of UV emitters, their challenges and provides an outlook for future developments.

https://iopscience.iop.org/article/10.1088/1361-6463/aba64c/pdf

The 2020 UV emitter roadmap

Precise control of the heteroepitaxy on a low-cost foreign substrate is often the key to drive the success of fabricating semiconductor devices in scale when a large low-cost native substrate is not available. Here, we successfully synthesized three different phases of Ga2O3 (α, β, and e) films on c-plane sapphire by only tuning the flow rate of HCl along with other precursors in an MOCVD reactor. A 3-fold increase in the growth rate of pure β-Ga2O3 was achieved by introducing only 5 sccm of HCl flow. With continuously increased HCl flow, a mixture of β- and e-Ga2O3 was observed, until the Ga2O3 film transformed completely to a pure e-Ga2O3 with a smooth surface and the highest growth rate (∼1 μm/h) at a flow rate of 30 sccm. At 60 sccm, we found that the film tended to have a mixture of α- and e-Ga2O3 with a dominant α-Ga2O3, while the growth rate dropped significantly (∼0.4 μm/h). The film became rough as a result of the mixture phases since the growth rate of e-Ga2O3 is much higher than that of α-Ga2O3...

/pdf/hcl-flow-induced-phase-change-of-a-b-and-e-ga2o3-films-grown-2bq5u3w0sy.pdf

HCl Flow-Induced Phase Change of α-, β-, and ε-Ga2O3 Films Grown by MOCVD

Both β-Ga2O3 and wurtzite AlN have wide bandgaps of 4.5–4.9 and 6.1 eV, respectively. We calculated the in-plane lattice mismatch between the (−201) plane of β-Ga2O3 and the (0002) plane of AlN, which was found to be 2.4%. This is the smallest mismatch between β-Ga2O3 and binary III-nitrides which is beneficial for the formation of a high quality β-Ga2O3/AlN heterojunction. However, the valence and conduction band offsets (VBO and CBO) at the β-Ga2O3/AlN heterojunction have not yet been identified. In this study, a very thin (less than 2 nm) β-Ga2O3 layer was deposited on an AlN/sapphire template to form the heterojunction by pulsed laser deposition. High-resolution X-ray photoelectron spectroscopy revealed the core-level (CL) binding energies of Ga 3d and Al 2p with respect to the valence band maximum in individual β-Ga2O3 and AlN layers, respectively. The separation between Ga 3d and Al 2p CLs at the β-Ga2O3/AlN interface was also measured. Eventually, the VBO was found to be −0.55 ± 0.05 eV. Consequent...

/pdf/valence-and-conduction-band-offsets-of-b-ga2o3-aln-1gxaiwwk9s.pdf

Valence and conduction band offsets of β-Ga2O3/AlN heterojunction

Recently, a new paradigm of the adversarial attack on the quantized neural network weights has attracted great attention, namely, the Bit-Flip based adversarial weight attack, aka. Bit-Flip Attack (BFA). BFA has shown extraordinary attacking ability, where the adversary can malfunction a quantized Deep Neural Network (DNN) as a random guess, through malicious bit-flips on a small set of vulnerable weight bits (e.g., 13 out of 93 millions bits of 8-bit quantized ResNet-18). However, there are no effective defensive methods to enhance the fault-tolerance capability of DNN against such BFA. In this work, we conduct comprehensive investigations on BFA and propose to leverage binarization-aware training and its relaxation -- piece-wise clustering as simple and effective countermeasures to BFA. The experiments show that, for BFA to achieve the identical prediction accuracy degradation (e.g., below 11\% on CIFAR-10), it requires 19.3x and 480.1x more effective malicious bit-flips on ResNet-20 and VGG-11 respectively, compared to defend-free counterparts.

/pdf/defending-and-harnessing-the-bit-flip-based-adversarial-4pr7a1btrm.pdf

Defending and Harnessing the Bit-Flip Based Adversarial Weight Attack

The spontaneous polarization (SP) and piezoelectric (PZ) constants of BxAl1-xN and BxGa1-xN (0 ≤ x ≤ 1) ternary alloys were calculated with the hexagonal structure as reference. The SP constants show moderate nonlinearity due to the volume deformation and the dipole moment difference between the hexagonal and wurtzite structures. The PZ constants exhibit significant bowing because of the large lattice difference between binary alloys. Furthermore, the PZ constants of BxAl1-xN and BxGa1-xN become zero at boron compositions of ∼87% and ∼74%, respectively, indicating non-piezoelectricity. The large range of SP and PZ constants of BxAl1-xN (BAlN) and BxGa1-xN (BGaN) can be beneficial for the compound semiconductor device development. For instance, zero heterointerface polarization Δ P can be formed for BAlN and BGaN based heterojunctions with proper B compositions, potentially eliminating the quantum-confined Stark effect for c-plane optical devices and thus removing the need of non-polar layers and substrate...

/pdf/wurtzite-baln-and-bgan-alloys-for-heterointerface-i0fs3mjlac.pdf

Wurtzite BAlN and BGaN alloys for heterointerface polarization engineering

Recent studies show that adversarial attacks on neural network weights, aka, Bit-Flip Attack (BFA), can degrade Deep Neural Network’s (DNN) prediction accuracy severely. In this work, we propose a novel weight reconstruction method as a countermeasure to such BFAs. Specifically, during inference, the weights are reconstructed such that the weight perturbation due to BFA is minimized or diffused to the neighboring weights. We have successfully demonstrated that our method can significantly improve the DNN robustness against random and gradient-based BFA variants. Even under the most aggressive attacks (i.e., greedy progressive bit search), our method maintains a test accuracy of 60% on ImageNet after 5 iterations while the baseline accuracy drops to below 1%.

Defending Bit-Flip Attack through DNN Weight Reconstruction

Traditional Deep Neural Network (DNN) security is mostly related to the well-known adversarial input example attack. Recently, another dimension of adversarial attack, namely, attack on DNN weight parameters, has been shown to be very powerful. As a representative one, the Bit-Flip-based adversarial weight Attack (BFA) injects an extremely small amount of faults into weight parameters to hijack the executing DNN function. Prior works of BFA focus on un-targeted attack that can hack all inputs into a random output class by flipping a very small number of weight bits stored in computer memory. This paper proposes the first work of targeted BFA based (T-BFA) adversarial weight attack on DNNs, which can intentionally mislead selected inputs to a target output class. The objective is achieved by identifying the weight bits that are highly associated with classification of a targeted output through a class-dependent weight bit ranking algorithm. Our proposed T-BFA performance is successfully demonstrated on multiple DNN architectures for image classification tasks. For example, by merely flipping 27 out of 88 million weight bits of ResNet-18, our T-BFA can misclassify all the images from 'Hen' class into 'Goose' class (i.e., 100 % attack success rate) in ImageNet dataset, while maintaining 59.35 % validation accuracy. Moreover, we successfully demonstrate our T-BFA attack in a real computer prototype system running DNN computation, with Ivy Bridge-based Intel i7 CPU and 8GB DDR3 memory.

Jingtao Li

Papers

Valence and conduction band offsets of β-Ga2O3/AlN heterojunction

Defending and Harnessing the Bit-Flip Based Adversarial Weight Attack

Wurtzite BAlN and BGaN alloys for heterointerface polarization engineering

Defending Bit-Flip Attack through DNN Weight Reconstruction

T-BFA: Targeted Bit-Flip Adversarial Weight Attack