Hongping Zhao, The Ohio State University
Masataka Higashiwaki, National Institute of Information & Comm Tech
Robert Kaplar, Sandia National Laboratories
Julien Pernot, University of Grenoble
Taiyo Nippon Sanso
EL04.01: AlGaN Materials and Devices
Sunday PM, April 18, 2021
10:30 AM - *EL04.01.01
AlN Nanowire pn Heterojunctions—A New Paradigm Towards UV-C LEDs
Bruno Daudin1,2,Rémy Vermeersch1,2,Alexandra-Madalina Siladie1,2,Gwénolé Jacopin3,Ana Cros4,Nuria Garro4,Eric Robin1,5,D. Calliste6,Pascal Pochet6,Fabrice Donatini3,Julien Pernot3
Université Grenoble Alpes1,CEA, IRIG-PHELIQS "Nanophysics and Semiconductors" Group2,Institut Néel, Université Grenoble Alpes, CNRS3,Universidad de Valencia4,CEA, INAC-MEM, LEMMA5,CEA, INAC-MEM, L-SIM6Show Abstract
The desired realization of efficient solid state deep UV emission devices is currently limited by the difficulty to achieve efficient p-type doping of AlGaN layers with high AlN molar fraction, due to the high ionization energy of single Mg impurity. However, if now using AlN nanowires (NWs), it will be shown that Mg/In codoping leads to an increased Mg solubility limit by more than one order of magnitude. Optimal electrical activation of acceptor impurities was achieved by electron irradiation . Next, the formation of AlN NW p-n junction was assessed by electron beam induced current (EBIC) experiments, putting in evidence the electrical field associated with the junction. Current-voltage characteristics in forward bias conditions have established that the current was varying as Vn, (with n larger than 6) before activation. By contrast, following acceptor activation, a space charge limited current regime was observed. Finally an AlGaN active region consisting of either a short period AlN/GaN superlattice or of an AlGaN section was inserted in the pn junction. The structural, electrical and electroluminescent properties of the resulting NW-based UVC LED will be discussed.
 A. M. Siladie et al, Nano Lett. 2019, 19, 8357−8364
10:55 AM - EL04.01.02
Growth of (Al,Ga)N/GaN Heterostructures at Temperatures Below 600 °C by Metalorganic Vapor Phase Epitaxy
Caroline Reilly1,Nirupam Hatui1,Shuji Nakamura1,Steven DenBaars1,Stacia Keller1
University of California, Santa Barbara1Show Abstract
As the role of the nitrides in electronic applications expands, integrating group-III nitrides with other materials such as silicon based integrated circuits is an attractive route. Although bonding processes exist, the ability to grow III-N materials directly on already processed wafers would greatly widen the design space, an approach which is currently hampered by the high temperatures associated with nitride growth by metalorganic vapor phase epitaxy (MOVPE). For GaN and AlN, typical growth temperatures can exceed 1000 °C – too high to be compatible with many processed wafers or other sensitive substrates. High quality material growth, with low impurity incorporation and good morphology, is then desirable at low growth temperatures for integration with other materials. In this work, the MOVPE growth of (Al,Ga)N layers at temperatures below 600 °C was explored using a flow modulation epitaxy (FME) scheme. The latter was applied in order to compensate for the lower surface mobility of adsorbed species at these very low deposition temperatures.
All experiments were conducted via atmospheric pressure MOVPE on semi-insulating GaN-on-sapphire templates. Low temperature (LT) layers were grown using a FME scheme consisting of pulsing the group III precursor(s), trimethylaluminum (TMAl) and triethylgallium (TEGa), while continuously flowing ammonia. The LT layers were analyzed by Hall measurements, atomic force microscopy (AFM), X-ray diffraction (XRD), secondary ion mass spectrometry (SIMS), and X-ray photoelectron spectroscopy (XPS).
The FME growth scheme used in this study was previously optimized for GaN deposition with TEGa to allow for step flow growth at 550 °C. Upon the addition of TMAl, however, the surface no longer showed steps and instead a layer by layer growth mode was observed via AFM. The transition into a layer by layer growth mode with the addition of TMAl is consistent with the known lower surface mobility of Al adatoms. Despite losing the step flow growth mode, the RMS roughness of the AlGaN layer was still sub-nanometer at 660 pm in comparison to 405 pm observed for the step-flow grown GaN, sufficiently smooth to be used for further studies.
By using the same growth scheme exclusively with TMAl, layers with different thickness were grown varying the number of cycles between 94 and 563. The thinnest layer showed an RMS roughness of 470 pm and the thickest layer 535 pm. Layers with 282, 375, 469, and 563 cycles were additionally analyzed by XRD. The layer peak became sharper and shifted away from the substrate peak as the number of cycles was increased. This was the opposite trend of what may be expected if the shift in the XRD peak was solely due to increasing relaxation of the layer with increasing thickness. From SIMS and XPS data, it was confirmed that this shift was due to unintentional gallium incorporation, which was more pronounced in the thinner samples. Ga incorporation into AlN has been seen for higher temperature AlN growth on GaN following a similar trend to that seen herein.
Electrical characterization of the samples with varying thickness revealed that the two thinnest samples were too resistive to be measured. For all other samples, the sheet resistance decreased with increasing AlN thickness. The highest mobility measured was 400 cm2/(V●s) with a sheet charge of 1.65 x 1013 cm-2 for the sample with 469 cycles. Additionally, a sample with 8 nm LT GaN underneath a LT (Al,Ga)N layer exhibited a mobility of 230 cm2/(V●s) and a sheet charge of 1.60 x 1013 cm-2. More details on the effects of growth temperature, thickness, and composition on the characteristics of the (Al,Ga)N layers will be presented at the conference. This work is partially supported by the Solid State Lighting and Energy Electronics Center and Intel Corporation.
11:10 AM - EL04.01.03
Influence of Chemical Potentials on Strain Development in Si-Doped Al0.7Ga0.3N
Yan Guan1,Shun Washiyam1,Dolar Khachariya1,Pegah Bagheri1,Ji Hyun Kim1,Pramod Reddy2,Ramon Collazo1,Zlatko Sitar1
North Carolina State University1,Adroit Materials Inc.2Show Abstract
AlGaN-based electronics and optoelectronics require Si doping for n-type conductivity. However, tensile stress is generated in AlGaN layers grown on foreign substrates with increasing Si concentration. High tensile stress could cause wafer bowing and cracking, which are detrimental to the device performance. Though several models have been proposed, the mechanism of the strain development by Si doping is still in debate. We have suggested that tensile stress is generated by dislocation climb. More vacancies and its complexes with Si are formed by the Fermi level shift brought by Si doping. Thus, the tensile stress is expected to be enhanced by lowering defect formation energy. This work reports the influence of chemical potentials on VIII-nSi complexes and the resulting strain development in Si-doped Al0.7Ga0.3N.
Si-doped Al0.7Ga0.3N layers were grown on unintentionally doped Al0.7Ga0.3N layers on top of c-plane AlN/sapphire substrates by MOCVD. Trimethylaluminum (TMA), triethylgallium (TEG), ammonia (NH3), and SiH4 were used as Al, Ga, N, and Si precursors, respectively. NH3 flow rate was varied from 0.3 slm to 0.7 slm and 1.5 slm to control chemical potentials. SiH4 flow rates were changed to control the Si concentration from 1E18 cm-3 to 1E19 cm-3. TMA and TEG flow rates were adjusted to control the growth rate and composition. Growth temperature and pressure were kept constant at 1100°C and 20 Torr, respectively. X-ray diffraction (XRD) was employed to determine the composition and strain development. Photoluminescence (PL) and Hall measurements were performed to characterize the influence of VIII-nSi complexes on electrical and optical properties.
First, the influence of growth rate on strain evolution was characterized. Growth rate was varied by metalorganic flow rates while SiH4 flow rate, Al composition and NH3 flow rate were kept constant at 30 sccm ([Si]~1e19 cm-3), 70% and 0.7 slm, respectively. Although strain increases linearly with thickness under the same growth condition (i.e., when growth time is varied), net strain was decreased from -0.19 % to -0.44 % (negative strain means compressive strain) as growth rate was increased from 600 nm/h to 900 nm/h for one hour growth. There are two possible reasons for the strain reduction by increasing growth rate: (1) as Si impurities are incorporated into the AlGaN layer in a mass transport limited process, [Si] is decreased at a higher growth rate; (2) metalorganic flow rate was increased to increase growth rate, and growth condition became more III-rich. Both lead to an increase in the formation energy of VIII-nSi complexes. Therefore, growth rate needs to be constant to characterize the role of chemical potentials in strain development.
When AlGaN layers were grown at a growth rate of 600±50 nm/h, an increase in strain towards a tensile state was clearly observed. Strain was increased from -0.41 % to -0.19 % and -0.04 % with increasing NH3 flow rate. As the growth environment becomes more N-rich, the formation energy of VIII-related defects decreases, resulting in the increase in the VIII-related defect concentration. Under an equivalent growth rate, the Si concentration is constant for samples grown at different NH3 flow rate. Thus, the observed high tensile strain at high NH3 flow rate does not originate from the Si substitutional impurities by themselves, eg. formation of SiNx or lattice expansion. Under N-rich conditions, more VIII-related defects are expected to be involved in surface mediated dislocation climb. The increase in the [VIII-nSi] was also confirmed from an increase in the VIII-nSi peak PL intensity and an abrupt decrease in the carrier concentration by self-compensation. In summary, we demonstrate that tensile stress induced by dislocation climb is governed by vacancy concentration rather than Si concentration, supporting and adding further details to the mechanisms previously discussed for the observation of strain in Si doped III-nitride layers.
11:25 AM - EL04.01.04
Late News: Assessment of N-Type and P-Type Doping in (Al,Ga)N Heterostructures by Scanning Probe Microscopy Techniques
Albert Minj1,Ming Zhao1,Benoit Bakeroot1,2,Lennaert Wouters1,Kristof Paredis1,Thomas Hantschel1,Stefaan Decoutere1
imec1,Centre for Microsystems Technology (CMST), imec and Ghent University2Show Abstract
The current state of dopant assessment for the optimization of the wide band gap (Al,Ga)N-based heterostructures for high-frequency and high-power applications relies heavily on quantitative chemical analysis techniques such as secondary-ion mass spectrometry. In such heterostructures, the determination of P-type carrier density of the cap layer, control of background concentration and assessment of polarization-induced confined carriers are necessary for the realization of optimal working devices. None of these can be completely inferred from the chemical analysis owing to several material and growth issues including the poor activation of Mg, the presence of O impurities and the amphoteric nature of carbon impurities. Here, as regions of interest in a typical epitaxial stack for GaN electronic devices consist of multiple triangular quantum wells with electron/hole confinement and doped layers, it demands electrical characterization at a nanoscale. In this context, the exploitation of the Schottky behaviour of nano-size metal-semiconductor junction formed between a metallic scanning probe microscopy (SPM) probe and (Al,Ga)N-nitride surface can be promising for carrier assessment by SPM techniques.
The two scanning probe microscopy techniques, scanning capacitance microscopy (SCM) and Conductive-atomic force microscopy (C-AFM), which are today routinely used for the characterization of Si- and Ge-based semiconductor structures with nanometer spatial resolution are investigated for their applicability on (Al,Ga)N heterostructures with energy band gap values larger than 3.4 eV. Our experiments make use of rectifying property of the electrical conduction at the nanoscale in both the modes. In this paper, it will be shown that by combining the two techniques, the nature of free carriers originating from extrinsic n-/p-type dopants and polarization-induced confined carriers, two-dimensional electron gas (2DEG) and hole gas (2DHG), can be revealed across (Al,Ga)N/Silicon heterostructures consisting of n-/p-doped GaN and AlGaN transition layers (TLs) . Here, a simple phase shift of approximately 180° consistently observed between all n-type and p-type regions in SCM phase maps verifies the absolute and correct determination of the type of carriers in accordance with TCAD band simulation. We also found that by fabricating Ti/Al-based low resistance back contacts, the issues related to back-to-back Schottky configuration and surface charging were mitigated and it concurrently allowed C-AFM measurements at low DC biases. Our back-contact methodology also aided the SCM signal enhancement at low magnitudes of AC bias comparable to that used in conventional semiconductors like Si and Ge. A large SCM signal strength for AC bias ≤ 1 V was seen on both extrinsically (Si, Mg) doped regions and on the polarization induced 2D carrier gases near the interface, while no signal was recorded otherwise on undoped regions such as unintentionally doped GaN and AlGaN regions in the transition layers. Because of the rectifying property of the tip-sample junction for both n-type and p-type doped (Al,Ga)N regions, the type of the carriers can also be determined from the C-AFM measurements through selection of the polarity of the DC bias. Thanks to the small tip-sample contact area, the C-AFM technique eventually allowed resolving the 2DHG and 2DEG regions separated by an only 5 nm AlN layer. As the contact area can be further reduced with the use of sharper probe tips , this analysis can be extended to even narrower barriers. Our paper illustrates the high potential of the two techniques to study device heterostructures for high power and RF applications and multi-quantum well heterostructures as in optoelectronic devices.
 A. Minj, M. Zhao, B. Bakeroot, and K. Paredis, Appl. Phys. Lett. 118, 032104 (2021)
 T. Hantschel, M. Tsigkourakos, L. Zha, T. Nuytten, K. Paredis, B. Majeed, and W. Vandervorst, Microelectron. Eng. 159, 46 (2016)
11:40 AM - EL04.01.05
Late News: Hanbury Brown and Twiss Correlations in a SEM—Imaging Carrier Lifetime at the Nanoscale in Ultra Wide Band Gap Materials
Sylvain Finot1,Vincent Grenier2,Vitaly Zubialevich3,Catherine Bougerol1,Pietro Pampili3,Joël Eymery4,Peter Parbrook3,Christophe Durand2,Gwénolé Jacopin1
University Grenoble Alpes, CNRS, Grenoble INP, Institut Néel1,University Grenoble Alpes, CEA, IRIG, PHELIQS, NPSC2,Tyndall National Institute, University College Cork3,University Grenoble Alpes, CEA, IRIG, MEM, NRS4Show Abstract
In order to establish the relationship between luminous efficiency and material characteristics, it is essential to precisely measure the relative contribution of radiative and non-radiative recombinations. This is usually obtained through the measurements of the luminescence decay time as a function of temperature. To do so, the commonly used technique is time-resolved photoluminescence spectroscopy . However, this technique does not allow nanoscale resolution, which is the relevant scale for defect characterizations. In addition, when dealing with ultra-wide bandgap materials, it required expensive pulsed deep UV lasers.
To go beyond these limitations, fast electrons can be used as a highly localized excitation source for luminescence measurements. This technique is called cathodoluminescence (CL) spectroscopy. Recently, picosecond time-resolved CL technique has been developed to reach at the same time high spatial and high temporal resolutions [2,3]. However, it is difficult to obtain a high brightness pulsed electron gun, which leads to reduced spatial resolution, low CL intensity or photocathode aging issues.
In this work, to circumvent these limitations, we took advantage of the specific statistics of electron/hole pair generation by fast electrons. Indeed, in a secondary electron microscope, the interaction of the incident electron with the semiconductor generates almost instantaneously (< 1 ps) a bunch of electron/hole pairs (typically > 300). These electron/hole pairs can then radiatively recombine, according to their carrier lifetime. Hence, by studying the autocorrelation function of the CL intensity (g2(τ)), a strong bunching is expected at τ = 0 (g2(τ = 0)>> 1). More importantly, by fitting g2(τ), we access the local carrier lifetime without the need for an expensive pulsed electron gun .
Thus, to measure the g2(τ) of the CL signal, we built a Hanbury Brown and Twiss (HBT) interferometer to analyze the CL photon statistics. As an illustration, we applied this technique to the study of AlGaN/GaN core-shell quantum wells on GaN microwires grown by metalorganic vapor phase epitaxy . Due to the lattice mismatch between the GaN core and the AlGaN shell, there is a formation of cracks. These cracks locally affect the material properties (strain, energy shift, CL intensity change…). Therefore, to quantify the impact of cracks on the luminous efficiency at the nanoscale, we recorded the carrier lifetime in this region. As expected, we observe a reduction of carrier lifetime near the crack, which indicates the creation of dangling bonds and point defects in the crack region. The CL results are then compared with transmission electron microscopy.
 T. Langer et al., Appl. Phys. Lett. 103, 202106 (2013).
 S. Chichibu et al., Jpn. J. Appl. Phys. 59, 020501 (2019).
 W. Liu et al., Appl. Phys. Lett. 109, 042101 (2016).
 S. Meuret et al., ACS Photonics 3, 1157 (2016).
 S. Finot et al., Appl. Phys. Lett. 117, 221105 (2020).
11:55 AM - *EL04.01.06
UV-VIS Photodetectors Using AlGaN High Electron Mobility Transistors with GaN Nano-Dot Floating Gates
Andrew Armstrong1,Alexandra Brianna Klein1,Andrew Allerman1,Albert Baca1,Mary Crawford1,Jacob Podkaminer2,Carlos Perez1,Michael Siegal1,Erica Douglas1,Vincent Abate1,Francois Leonard1
Sandia National Laboratories1,Current Address: 3M Corporate Research Laboratory2Show Abstract
Visible-blind and solar-blind photodetectors were demonstrated using AlGaN-channel high electron mobility transistors (HEMTs) with GaN nanodots as an optically-active floating gate. The effect of the photo-induced floating gate voltage was large enough to switch a HEMT from the off-state an on-state with illumination. Responsivity > 108 A/W was observed at room temperature with the HEMTs biased in the off-state in the dark for low dark current and low dc power dissipation. The absorption threshold was shown to be controlled by the AlN mole fraction of the HEMT channel layer, enabling the same device design to be tuned for either visible-blind or solar-blind detection. The influence of GaN nanodot position in the HEMT on the dynamic range of the photodetector was investigated. The responsivity and temporal response of the detectors as a function of optical intensity was investigated, and a gain-bandwidth product > 109 Hz was observed. The dependence of device performance on optical intensity was successfully modeled by treating the nanodots as a two-level system with characteristic rates of carrier capture and recombination.
EL04.02: Nitrides Materials and Devices
Sunday PM, April 18, 2021
1:00 PM - EL04.02.01
Late News: A Comprehensive Study of MOVPE Growth on 200 mm GaN-on-SOI for Monolithic Integrated GaN ICs
Ming Zhao1,Karen Geens1,Xiangdong Li1,Nooshin Amirifar1,Stefaan Decoutere1
With the mass production of discrete devices based on GaN-on-Si technology is just around the corner, the monolithic integrated GaN ICs (integrated circuits) has become a hot topic in recent years. This is largely attributed to its great potential in increasing the switching speed and efficiency, reducing the form factor and enabling a higher level of design flexibility, etc. We have proposed and experimentally validated the approach of using GaN-on-SOI combained with deep trench isolation as a technology platform to mainly eliminate the “back gating” effect which is associated with GaN-on-Si technology . On the other hand, the MOVPE growth on SOI substrate, as a critical enabling technique, is far less studied compared to GaN-on-Si. In this work, we present a comprehensive study on multiple aspects of the MOVPE growth on 200 mm SOI substrates, including SOI substrate and buffer design. We further demonstrate the growth of HEMT device structures for voltage rating of 200 V, 650V and even beyond.
We found that the epitaxial growth on SOI substrates behaved very differently from the growth on regular Si substrate. The SOI wafer deformed much more strongly upon the lattice mismatch introduced stress. Further in-depth study pointed out that the phenomena is very likely associated with the strain partition effect. We have proven that the Si (111) device layer thickness is a tuning knob to alter the strain partition effect, which is in line with the strain partition thoery. In addtition, using Si (111) instead of Si (100) handling wafer can also to certain extent mitigate the strong wafer bowing of SOI substrate, which results from the higher stiffness of the Si (111) due to its higher biaxial modulus. Following our superlattice based GaN-on-Si buffer scheme but with dedciated modification, we have obtained a 200 mm GaN-on-SOI buffer platform with low wafer bow (<50 µm) and suitable for 200 V application. The vertical buffer breakdown voltagae (voltage at which the leakage current reaches 1 µA/mm2 at 25 °C and 10 µA/mm2 at 150 °C) is ~400 V at 25 °C and >300 V at 150 °C. A high operating voltage of ~440 V is extrapolated (corresponding to a lifetime of 10 years at 175 °C) based on time-dependent dielectric breakdown (TDDB) measurements. The buffer also shows a low buffer dispersion of < 10% at up to 150 °C using the “back-gating” method. This buffer platform has enabled the fabrication of E-mod p-GaN HEMT devices and the further demonstration of a functional 48V-to-1V buck converter with monolithic integrated half bridge and on chip drivers. However, it became extremely challenging to extend this buffer platform to a higher voltage, i.e. 650 V and beyond, by thickness scaling even using Si (111) handling wafer in the SOI substrate. It is very difficult to control the high in situ wafer deformation and to limit the post growth wafer bow <50 µm. We therefore proposed a novel buffer scheme by taking advantage unique properties of superlattice structure and the strain partition effect of SOI substrate. This buffer scheme consists of multiple superlattice blocks of different average Al% which allows to introduce alternatively the compressive and tensile stress along the buffer growth. Based on this buffer scheme we demonstrated a 4.8 µm buffer for 650 V application and a 6 µm buffer for a even higher voltage rating.
 X. Li et al., Proc IEEE Int. Electr. Dev. Meeting, San Francisco, CA, USA, 2019, pp. 4.4.1-4.4.4. and references therein.
1:15 PM - EL04.02.02
Characterization of III-Nitride Epitaxial Layers by Scanning Spreading Resistance Microscopy
Sizhen Wang1,Jinho Kang1,Bingjun Li1,Jung Han1
Yale University1Show Abstract
Scanning spreading resistance microscopy (SSRM) is an AFM-based technique, it detects the resistance of current spreading layer just below the tip-semiconductor contact region. This innovative nanoscale characterization techniques find its wide application in 2D carrier mapping in semiconductor material and device characterization because of its high resolution and ease of quantification. Although SSRM has been used to study GaN carrier concentration, AlGaN-GaN heterostructure, AlN-GaN quantum well, and related defects in III-nitride, there is still lack of research on dopant calibration and quantification on GaN materials. Here we presented a systematic research on GaN dopant calibration under variable tip stress condition using SSRM and apply this technology to characterize the detail structure of distributed Bragg reflector (DBR) of vertical cavity surface emitting laser (VCSEL).
First, n-GaN sample with staircase doping of Si, Ge was used to calibrate with variable tip force (range from 0.32 µN to 6.40 µN). Although the measured resistance log(R) decreased with increased tip force, the log(R) decreased monotonically as the dopant concentration (Si, Ge) increased, almost four order of magnitude doping range ( ~4x1016cm-3 for UID to 1x1020cm-3) can be distinguished for n-GaN thin film. And linear fitting curve was extracted to estimate the carrier concentration in heavy doping region. Then, the SSRM technology was used to reveal the fine DBR structure which consist of 25 pairs of Si-doped (43nm, 5x1018cm-3) and Ge-doped GaN (55nm, 9x1019cm-3), and the best contrast of 2D SSRM image of DBR structures was obtained with -0.1 V DC bias voltage and 2.56 µN tip force. Finally, the SSRM was applied to investigate the in-plane inhomogeneous incorporation of Ge dopants into GaN which was grown on Sapphire. 2D SSRM mapping results showed that as the Ge doping concentration was more than 3.0x1019cm-3, Ge incorporation become inhomogeneous. It was estimated that the light doping zones had only 8.90x1018/cm3 Ge dopant for sample with 6.0x1019/cm3 intentional Ge doping, while for sample with 1.4x1020/cm3 intentional Ge doping, the light doping zone was estimated only doped with 1.06x1019/cm3 Ge dopants. Using the ratio of nominal doping concentration to the minimum doping concentration as an indicator, its value changed from 6.69 to 13.23 as Ge doping of GaN epitaxial layer increased from 6.0x1019/cm3 to 1.4x1020/cm3. The inhomogeneous doping became worse with heavy Ge doping. And this issue might be related to anisotropic dopant incorporation efficiency at different facets which were exposed during GaN epitaxial growth.
1:30 PM - EL04.02.03
Structural and Piezoelectric Properties of ScxAl1-xN-GaN Heterostructures Grown by
Molecular Beam Epitaxy
Joseph Casamento1,Celesta Chang1,Yu-Tsun Shao1,John Wright1,David Muller1,Huili Xing1,Debdeep Jena1
Cornell University1Show Abstract
Alloying with Scandium (Sc) into the III-nitride platform of InN, GaN, and AlN has gained tremendous interest in recent years, from fundamental research to transformative changes in technological applications. This is because isoelectronic alloying with Sc has been demonstrated to increase the piezoelectric response and induce ferroelectric behavior in ScxAl1-xN alloys. [1,2] However, most studies of the fundamental properties ScxAl1-xN have been focused on sputter deposited materials, where challenges such as sub-optimal crystallinity, microstructural instabilities, and chemical inhomogeneities have been encountered. Molecular beam epitaxy (MBE) aims to enable solutions to these issues via epitaxial deposition on single crystalline substrates in an ultra-high vacuum environment and thereby better elucidate the fundamental properties of ScxAl1-xN.
In this work,  we report on the epitaxial growth studies of ScxAl1-xN of various Sc compositions (x = 0.18 to 0.40) grown by MBE on n+ bulk GaN substrates and their corresponding structural and piezoelectric properties. Aluminum (Al) was supplied with a conventional Knudsen effusion cell and Sc was supplied via electron beam evaporation in a tungsten (W) crucible. Growth temperature was maintained at 600C, measured by a thermocouple. In-situ reflection high energy electron diffraction (RHEED) was used to assess the epitaxial nature of the films.
Phase purity and epitaxial relationship was further assessed using X-ray Diffraction (XRD). Films with a surface roughness of less than 1 nm were confirmed by atomic force microscopy (AFM). Out-of plane piezoelectric coefficients (d33,eff) were analyzed with piezoresponse force microscopy (PFM) using a conductive AFM cantilever. High-angle annular dark field scanning transmission electron microscopy (HAADF-STEM) was utilized to study atomic level defects and to confirm the metal polar orientation of ScxAl1-xN grown on metal polar GaN.
Post-growth XRD measurements show a wurtzite 002 diffraction peak for the ScxAl1-xN layers at all Sc compositions studied (x = 0.18 to 0.40). XRD full-width-half maximum (FWHM) values ranged from 0.09 to 0.13 degrees (324-468 arcseconds) for x = 0.18 to x = 0.33, indicating superior crystalline quality. RHEED images suggested pure wurtzite phase ScxAl1-xN but decreasing crystallinity due to primary azimuthal streak broadening as Sc content increased. PFM results showed an increase in piezoelectric coefficient for ScxAl1-xN (x=0.18, 15 pm/V) relative to AlN (6 pm/V) and a decrease thereafter as Sc concentrations increased. Insight to this counterintuitive trend was given via HAADF-STEM, which showed the presence of zinc-blende inclusions bound by partial dislocations in the ScxAl1-xN (x=0.40) film. This contrasts with the RHEED images suggesting single-phase wurtzite crystals. The zinc-blende crystal structure is not thermodynamically stable and may arise from kinetic factors during the deposition process. The presence of zinc-blende inclusions likely hinders piezoelectric performance and is deleterious to potential ferroelectric behavior as zinc-blende point group prohibits ferroelectricity via symmetry arguments. This work gives insight into the fundamental behavior of epitaxial ScxAl1-xN and should provide guidance toward future optoelectronic applications based on these heterostructures. Future work will involve growth optimization and continued structural and electrical characterization to enhance piezoelectric performance and include ferroelectric behavior these epitaxial heterostructures.
 Akiyama, M et al. Adv. Mater. 21, 593 (2009).
 Fichtner, S et al. J. Appl. Phys. 125,114103 (2019).
 Casamento, J et al. Appl. Phys. Lett. 117, 112101 (2020).
1:45 PM - EL04.02.04
Dielectric Properties, Electronic Transitions and Infrared Active Optical Phonon Modes of Molecular Beam Epitaxy ScxAl1-xN Determined by Spectroscopic Ellipsometry
Alyssa Mock1,Alan Jacobs2,Eric Jin1,Matthew Hardy2,Marko Tadjer2
NRC Research Associateship Programs1,U.S. Naval Research Laboratory2Show Abstract
Significant interest has grown recently in alloying ultra-wide bandgap AlN with ScN in order to obtain unique and tunable properties. Such alloys could have impactful applications in high electron mobility transistor structures potentially enabling more efficient, compact, and powerful amplifiers critical to global communications infrastructure. Other possible applications include high frequency and broadband acoustoelectric filters and resonators. Further, the ferroelectric behavior of ScAlN with high scandium content has begun to garner notable attention of late with potential in sensor applications.
In our work, we employ spectroscopic ellipsometry to investigate (0001) wurtzite ScxAl1−xN (0.00 ≤ x ≤ 0.20) thin films grown by molecular beam epitaxy on c-plane sapphire substrates. We present the dielectric function obtained in the spectral region from 4-6.4 eV and report on the thicknesses and extracted critical point parameters from our best match model analysis. We find that the critical points associated with electronic transitions shift to lower energy as a function of increasing ScN alloy fraction. This is particularly interesting for tunability of band offsets between ScAlN and GaN in heterostructures, for example, critical for the development of next generation RF devices as the conduction band offset impacts the confinement and electron density in the two-dimensional electron gas that forms the channel of the device.
We also present the infrared dielectric functions and dielectric loss functions in the spectral region from 400-1200 cm-1. From these we obtain infrared active phonon mode properties, which are fundamental material properties important for future device design. We find that all phonons shift to lower wavenumber as a function of scandium incorporation and we also see evidence of a decrease in crystal quality which is confirmed by x-ray diffraction measurements. Further, we report the high frequency and static dielectric constants as well as their evolution with scandium content. We calculate the Born effective charge for each composition and find an increase with scandium content. Thus, we attribute the softening of the phonon modes with increasing scandium content to an increase in the effective metal ion mass and a corresponding increase of the dynamic effective charge. This work elucidates fundamental properties of MBE grown ScAlN thin films and contributes to improved understanding of Sc-based alloys currently under development for electronic and optoelectronic devices.
2:00 PM - EL04.02.05
Growth of High Quality MOCVD Aluminum Nitride Using N2 as Carrier Gas
Samiul Hasan1,Abdullah Mamun1,Kamal Hussain1,Dhruvinkumar Patel1,Mikhail Gaevski1,Iftikhar Ahmad1,Asif Khan1
University of South Carolina1Show Abstract
High quality, low dislocation density, AlN is required for high efficiency, high power, and long lifetime of the III-nitride based electronic and photonic devices. Several techniques have been reported to improve the quality of AlN, including substrate patterning, pulsed lateral overgrowth, high-temperature annealing, and modulation of temperature and precursors. All these growth processes have been done using ultra high purity H2 as a carrier gas. The impact of carrier gases on the growth of AlN has rarely been studied. The rationale to study N2 as a carrier gas for AlN growth in this research is to explore another carrier gas than H2, which is safer and economical. Previously, the use of N2 as a carrier gas resulted in inferior quality AlN layers. Here, we report the growth of high-quality AlN using N2 as carrier gas on a basal plane sapphire substrate using metal-organic chemical vapor deposition (MOCVD). A series of samples with N2 as carrier gas were grown with thickness varying from 1μm to 4μm by employing a two-step growth process. The process includes the formation of a thin buffer layer grown at 950° C with a high V/III ratio and a thick high-temperature growth at 1180° C with a low V/III ratio without introducing any intralayer. AlN samples were characterized using scanning electron microscopy (SEM), atomic force microscopy (AFM), high-resolution x-ray diffraction (HR-XRD), and Raman spectroscopy. SEM images show the formation of air pockets in the AlN layers, which acts as strain relief centers, reducing the dislocation densities. The XRD analysis of these samples shows the full width at half maxima of 289 arcsec of the omega scan for (10-12) plane and total dislocation density of 1.1x109 cm-2 for 4μm sample, which are the best-reported values for AlN on a sapphire substrate using N2 as the carrier gas. Raman study of a 4µm AlN sample shows low compressive stress of 0.59 GPa. A systematic study of the evolution of the quality of AlN with thickness will be presented.
2:15 PM - EL04.02.08
Late News: Polarization-Induced 2D Hole Gases in Undoped InGaN/AlN Heterostructures Grown on Single-Crystal AlN Substrates
Jimy Encomendero Risco1,Zexuan Zhang1,Reet Chaudhuri1,Masato Toita2,Debdeep Jena1,Huili Xing1
Cornell University1,Asahi Kasei Corporation2Show Abstract
We report, for the first time, the epitaxial growth, structural characterization, and transport properties of polarization-induced 2D hole gases (2DHGs) homoepitaxially grown on high-quality single-crystal AlN substrates. By harnessing the polarization discontinuity at the InGaN/AlN heterointerface, a high density of free holes (p > 5x1013 cm-2) is induced within AlN crystals without the need of acceptor impurities. This important feature, combined with the high structural quality of the homoepitaxial growth, results in the highest hole mobilities experimentally measured in bulk AlN substrates. The polarization origin of the 2D holes is experimentally confirmed by systematically increasing the Indium composition of the top InGaN layer, thereby inducing a higher density of free carriers at the InGaN/AlN heterojunction. Hole concentrations between p = 5.2x1013 cm-2 and p = 1.1x1014 cm-2 are measured at room temperature, as the indium composition is varied between 0% and 6%, showing a positive correlation with indium incorporation. X-ray diffraction reveals that over this composition range, the InGaN layers are coherently strained to the AlN lattice. Furthermore, the presence of strong interference fringes in the diffraction pattern and clear InGaN atomic steps indicates the presence of atomically smooth hetero-interfaces. Cryogenic Hall effect measurements show that hole concentrations are almost independent of temperature, ranging between p = 2.5x1013 cm-2 and p = 5.2x1013 cm-2 for indium compositions of 0% and 6%, respectively. These high hole densities, inaccessible by traditional impurity doping at 77 K, attest to the electrostatic origin of the 2D holes. Our results constitute a pivotal step towards the manufacture of nitride-based p-type field effect transistors (FETs) that can take the advantage of the high thermal conductivity and high breakdown electric fields of single-crystal AlN substrates. As an added benefit, the simultaneous demonstration of complementary 2D electron gases (2DEGs) and 2DHGs in undoped heterostructures grown on single-crystal AlN substrates, enables an unmatched level of integration for high-power and high-frequency nitride-based electronics, raising hopes for all-nitride CMOS technology.
EL04.03: Other UWBG Semiconductors
Sunday PM, April 18, 2021
4:00 PM - *EL04.03.01
Prospects of LiGaO2 and NaGaO2 as Ultrawide Band Gap Semiconductors
Walter Lambrecht1,Adisak Boonchun2,Klichchukon Dabsamut2,Dmitry Skachkov3,Santosh Radha1,Amol Ratnaparkhe1
Case Western Reserve University1,Kasetsart University2,University of Florida3Show Abstract
Inspired by the recent interest in β-Ga2O3 as ultra-wide-band-gap material, because of its combination of n-type dopability with a gap of about 4.9 eV, we propose two new even higher band gap materials and investigate their prospects for doping and transparent conducting applications in the ultraviolet range. We present a study of the phase stability, band structures, optical absorption and defect properties. LiGaO2 is a material with a wurtzite derived orthorhombic structure with space group Pna21 at ambient conditions. It consists of an ordered arrangement of the Li and Ga cations on the wurtzite lattice and can be viewed as a I-III-VI2 ternary derived from the binary II-VI compound ZnO. It has been considered in the past as ceramic material with piezo-electric properties and can be grown in bulk form by the Czochralsky method. Here we present quasiparticle self-consistent GW band structure calculations, showing that it has a direct gap of 5.6 eV (in agreement with experiment) when taking into account an estimated zero-point-motion correction of -0.2 eV. Furthermore, we find that transitions from the bottom of the conduction band to higher bands can only occur above 3.9 eV. The valence band maximum is split by crystal field splitting in three levels which leads to a polarization dependent absorption onset. The valence and conduction band effective mass tensors are determined. Defect calculations using the HSE hybrid functional show that the lowest energy of formation defect is the GaLi antisite, which is found to be a deep (negative U type) donor with a 2+/0 transition at 0.7 eV below the conduction band minimum. It is compensated by VLi1- or LiGa2- depending on chemical potential conditions. These defects pin the Fermi level deep in the gap leading to insulating behavior in intrinsic LiGaO2. However, Si and Ge dopants are shallow donors and have low energy of formation. Thus n-type doping should be feasible. Sn leads to a somewhat deeper donor. Prospects for p-type doping are less promising: NO is a deep trap with amphoteric character and Zn doping leads to self-compensation between ZnGa and ZnLi. Ga vacancies are found to have high energy of formation but can be created by high-energy particle irradiation, while Li vacancies are expected to occur in as grown material. Their EPR signals have been reported and we find excellent agreement between our calculated g-tensors and hyperfine splitting with the experiment. Finally, to avoid potential problems with Li diffusion, we also study NaGaO2 and find it to have a gap of 5.5 eV and to also be stable in the Pna21 structure. Both NaGaO2 and LiGaO2 can undergo a phase transition under pressure of about 8-14 GPa to an octahedrally coordinated phase. Among those phases, we find the R-3m phase to have lower energy than a rocksalt type phase. Even in this R-3m phase, the gap is above 5 eV but it becomes slightly indirect.
4:25 PM - EL04.03.02
Electron and Hole Mobility of Rutile GeO2 from First Principles—An Ultrawide-Band-Gap Semiconductor for Power Electronics
Kyle Bushick1,Kelsey Mengle1,Sieun Chae1,Emmanouil Kioupakis1
University of Michigan1Show Abstract
Rutile germanium dioxide (r-GeO2) is an ultrawide-band-gap semiconductor with potential applications in high-power electronic devices or transparent conductors. In these devices, understanding the carrier dynamics is paramount to evaluate the possible performance of a new material. In this work we use first-principles calculations based on density functional and density functional perturbation theory to investigate carrier-phonon coupling in r-GeO2 and predict its phonon-limited electron and hole mobilities as a function of temperature and crystallographic orientation. Our findings show that r-GeO2 has relatively high carrier mobilities, and in particular high hole mobility compared to other UWBG materials. We find the room temperature mobilities are μelec,⊥c = 244 cm2 V-1 s-1, μelec,‖c = 377 cm2 V-1 s-1, μhole,⊥c = 27 cm2 V-1 s-1, and μhole,‖c = 29 cm2 V-1 s-1, with carrier scattering dominated by low-frequency polar-optical phonons. While the electron mobility compares readily with other UWBG materials such as β-Ga2O3, the hole mobility is notably high. This finding is complimented by other recent work that reports r-GeO2 can be ambipolarly doped, enabling access to both n- and p-type devices. Paired with a wide band gap around 4.68 eV, the high carrier mobilities lead to a predicted Baliga figure of merit that surpasses an array of more established semiconductors, including Si, SiC, GaN, and β-Ga2O3. Through this work, we show that r-GeO2 is a promising UWBG semiconductor with desirable properties for both n- and p-type high-power electronic devices.
4:40 PM - EL04.03.03
Extremely Sharp Free Exciton Emission in Heteroepitaxial Cuprous Iodide Thin Films Grown by MBE
Masao Nakamura1,Sotato Inagaki2,Yoshihiro Okamura2,Makiko Ogino2,Youtarou Takahashi2,1,Licong Peng1,Xiuzhen Yu1,Yoshinori Tokura1,2,Masashi Kawasaki2,1
RIKEN1,The University of Tokyo2Show Abstract
Triggered by the successful development of perovskite solar cells, halide semiconductors have gained extensive attention in the last few years. Cuprous iodide (CuI) is a representative wide-bandgap halide semiconductor with superior optical and electronic properties, such as direct bandgap and large exciton binding energy as well as high hole-mobility. In particular, the exciton binding energy and oscillator strength in CuI are comparable with or even surpassing those in well-known wide-bandgap semiconductors like GaN and ZnO. These prominent excitonic properties are of great advantage for future quantum optoelectronic applications, including ultraviolet lasers and quantum simulators utilizing the Bose-Einstein condensate of exciton-polariton. However, the studies towards such device applications have been of little progress in CuI due to the lack of high-quality thin films so far.
Recently, we have succeeded in a dramatic improvement in the quality of CuI thin films. Thin films were grown by the molecular beam epitaxy (MBE), which had been rarely employed for the fabrication of halide films but has a successful history in the growth of high-quality compound semiconductor films. As a substrate, we adopted InAs that is almost perfectly lattice-matched with CuI. The fabricated CuI film has a genuine single-crystalline structure with excellent lattice coherence and atomically-flat surface. The low-temperature photoluminescence spectra exhibit extremely-sharp luminescence peak assigned to the free-exciton emission, which has never been observed even in bulk single crystals. These results indicate that the optical quality of CuI film has reached to the level of much more extensively studied wide bandgap semiconductors.
The successful growth of the high-quality CuI film will facilitate the optoelectronic device applications of CuI. Moreover, the present result will open a new research direction to explore novel functionalities at the atomically-sharp heterointerface of halides.
4:55 PM - EL04.03.04
Rutile GeO2—An Ultra-Wide-Band-Gap Semiconductor for Power Electronics
Sieun Chae1,Kelsey Mengle1,Kyle Bushick1,Hanjong Paik2,Nguyen Vu1,John Heron1,Emmanouil Kioupakis1
University of Michigan1,Cornell University2Show Abstract
Ultra-wide-band-gap (UWBG) semiconductors have tantalizing advantages for power electronics as their wider band gaps enable higher breakdown voltages. A handful of materials such as AlN/AlGaN, β-Ga2O3, and diamond have been developed for UWBG semiconducting devices, however, they are still facing numerous challenges, such as doping asymmetry and/or inefficient thermal conduction. In our work, we have identified rutile GeO2 (r-GeO2) to be a promising, yet unexplored UWBG (4.68 eV) semiconductor. Our first-principles calculations predict shallow ionization energies for donors such as Sb, As, and F, a phonon-limited electron mobility of 289 cm2 V-1s-1, and a breakdown electric field of 7.0 MV cm-1, which lead to a higher Baliga figure of merit than β-Ga2O3. r-GeO2 also has superior thermal conductivity (51 W m–1 K–1 (experiment)) and p-type doping property (the calculated ionization energy for Al acceptors is 0.45 eV and the calculated phonon-limited hole mobility is 28 cm2 V-1s-1). Though the thin-film synthesis of r-GeO2 has remained challenging due to its highly metastable amorphous phase, we demonstrate the first synthesis of single crystalline epitaxial thin films of r-GeO2 on a sapphire substrate using ozone-assisted molecular beam epitaxy. Also, Sb-doping of r-GeO2 was achieved for r-GeO2 single crystal nanorods. Our work motivates further exploration of r-GeO2 as an alternative UWBG semiconductor that can overcome the limitations of the current state-of-the-art UWBG materials.
5:10 PM - EL04.03.05
Late News: Thermal Imaging of Ultrawide Bandgap Devices Using MoS2 Thermoreflectance Enhancement Coatings
Riley Hanus1,Samuel Graham1,Asif Khan2
Georgia Institute of Technology1,University of South Carolina2Show Abstract
Measuring the maximum operating temperature within the channel of ultrawide bandgap transistors is critically important, since thermal management often sets operational limits such as maximum power and lifetime. Thermoreflectance imaging is an optimal choice due to sub-micron spatial resolution, sub-microsecond transient characterization, and rapid data acquisition. Unfortunately, commercially available light sources are limited to energies less than ~3.9 eV and are therefore transparent to ultrawide bandgap materials. Here, we utilize an MoS2 coating as a thermoreflectance enhancement coating which allows for the measurement of the surface temperature of (ultra)wide bandgap materials. This coating is a polycrystalline thin film with the c-axis preferentially aligned normal to the surface and is easily removable via sonication. The method is validated using electrical and thermal characterization of GaN and AlGaN devices. We demonstrate that this coating does not significantly influence the electrical and thermal behavior of the devices tested. A maximum temperature rise of 49 K at 0.59 W was measured within the channel of the AlGaN device, which is over double the maximum temperature rise obtained by measuring the thermoreflectance of the gate metal, demonstrating this methods value.
5:25 PM - *EL04.03.06
Deep UV Optical Properties of High-Mg-Content Rocksalt-Structured MgZnO
Takeyoshi Onuma1,Kanta Kudo1,Kyohei Ishii2,Mizuki Ono1,Yuichi Ota3,Kentaro Kaneko2,Tomohiro Yamaguchi1,Shizuo Fujita2,Tohru Honda1
Kogakuin University1,Kyoto University2,Tokyo Metropolitan Industrial Technology Research Institute3Show Abstract
Deep and vacuum UV light emitters have a variety of applications, such as in ozone cleaners, UV sterilization, medical care, and photolithography. Recently, the alternative use of 222-nm-DUV-light instead of conventionally-used 254-nm-DUV-light is proposed to sterilize human skin with suppressing DNA lesions based on the UV effects on the epidermis of hairless albino.  Indeed, action spectra for producing DNA lesions exhibit a peak at around 260 nm and subsequent increase in a wavelength range shorter than 220 nm and VUV region.  Enhancement of light absorption by a protein in the human epidermis in a shorter DUV and VUV regions enables selective inactivation of virus in an atmosphere. An effective wavelength range of 190 nm to 220 nm can be proposed by a tradeoff between the action spectra and the air absorption. However, most of commercially available DUV and VUV light sources are discharge-type lamps or lasers. Therefore, demands are increasing for developing solid-state DUV and VUV emitters since their cost, environmental impact, energy consumption are expected to be drastically reduced compared to the discharge-type light sources.
AlGaN alloys are the most promising materials for DUV light emitters. [3,4] However, their shortest emission wavelength is limited to be around 210 nm for AlN.  In this study, we are focusing on high-Mg-content rocksalt-structured (RS) MgZnO alloys as alternative candidate materials for DUV and VUV emitters by virtue of their wide variation of bandgap energy Eg from 4.5 eV for RS-ZnO to 7.8 eV for RS-MgO. Recently, successful growths of atomically-flat single crystalline RS-MgxZn1-xO films on (001) MgO substrates by the mist chemical vapor deposition (mist CVD) method [6,7] and observation of DUV emission [6-8] were demonstrated. Further improvements in its crystalline quality  has resulted in the predominate observation of cathodoluminescence (CL) peak at around 199-217 nm for x=0.92-0.95. [9,10] Excitation-current-density dependent CL spectra were measured, and analyses based on a rate equation model confirmed that the DUV luminescence band is attributed to the near-band-edge emission.  Electronic structure calculations suggested predominate contribution of Γ-Γ direct transition for the RS-MgxZn1-xO alloys with x>0.5. However, it is found that relatively broad tail-states expand toward deep UV spectral region. Consequently, CL peak energies show large Stokes-like shift of 0.7-0.8 eV. The Eg fluctuations and resultant exciton localization are possible origins of the large Stokes-like shift. Equivalent internal quantum efficiency is defined as spectrally integrated CL intensity at 300 K divided by that at 6 K. Relatively high values of 2-4% were obtained for the DUV NBE emission. The efficiencies were improved by employing the MgxZn1-xO/MgO multi-layered structures.  This work was supported in part by Grants-in-Aid for Scientific Research Nos. 17H01263 and 20H00246 from MEXT, Japan.
 N. Yamano et al., Photochem. Photobiol. 96, 853 (2020).  T. Matsunaga et al., Photochem. Photobiol. 54, 403 (1991).  H. Hirayama, J. Appl. Phys. 97, 091101 (2005).  M. Kneissl and J. Rass, III–Nitride Ultraviolet Emitters (Springer International Publishing, Cham, 2016).  Y. Taniyasu, M. Kasu, and T. Makimoto, Nature 441, 325 (2006).  K. Kaneko et al., Appl. Phys. Express 9, 111102 (2016).  K. Kaneko et al., J. Electron. Mater. 47, 8 (2018).  T. Onuma et al., Appl. Phys. Lett. 113, 061903 (2018).  K. Ishii et al., Appl. Phys. Express 12, 052011 (2019).  M. Ono et al., J. Appl. Phys. 125, 225108 (2019).  K. Kudo et al., in Compound Semiconductor Week 2019, Nara, May 20 (2019), MoE3-5.
EL04.04: Keynote Session: UWBG
Sunday PM, April 18, 2021
6:30 PM - *EL04.04.01
Materials and Device Engineering for AlGaN and β-Ga2O3 UWBG Electronic Devices
Siddharth Rajan1,Nidhin Kalarickal1,Towhidur Razzak1,Hao Xue1,Zhanbo Xia1,Chandan Joishi1,Hongping Zhao1,Asif Khan2,Wu Lu1
The Ohio State University1,University of South Carolina2Show Abstract
This presentation will review recent and ongoing work on ultra-wide band gap (UWBG) semiconductor electronic devices based on AlGaN and β-Ga2O3. The high breakdown field strength of UWBG semiconductors such as AlGaN and Gallium Oxide have opened exciting opportunities for achieving improved performance over well-established materials such as Si, GaN, and SiC. While these materials have several unique properties, they also share common challenges, such as relatively low mobility (compared to conventional Si and GaN), and the need for managing extreme electric fields
The first part of the presentation will discuss the development of AlGaN-based transistors for RF and mm-wave applications. Device simulations show that despite the relatively low electron mobility, the high breakdown field in AlGaN could enable performance enhancement for high frequency transistors. We will then discuss key recent developments in AlGaN transistors that have enabled significant improvement in device performance. The use of electron affinity grading  and advanced epitaxial designs has enabled UWBG AlGaN transistors with cutoff frequency over 40 GHz , and current density over 900 mA/mm. Introduction of high-permittivity materials in lateral device structures prevents premature gate breakdown, while improving field uniformity in transistor gate-drain regions . We will discuss the application of this idea in BaTiO3/AlGaN lateral diodes, where we achieved a breakdown field of 8.5 MV/cm  (breakdown voltage of 160 V), which represents the highest field achieved in a lateral device.
In the second part of the presentation, we will discuss recent work toward development of high-performance β-Ga2O3 lateral and vertical devices. We will discuss the growth of [endif]-->-Ga2O3and related alloys, and our investigation of fundamental electronic properties in these structures. The use of delta doping enables several key lateral electronic device structures. We will first give an outline of the relatively new mechanisms involved in Si delta doping using molecular beam epitaxy , and then discuss our work on growth of ��-(Al,Ga)2O3/��-Ga2O3 modulation-doped structures, including recent investigation into high sheet charge density modulation-doped structures . We will then outline details of recent device demonstrations, including investigation of advanced passivation methods to reduce current collapse , scaled delta-doped transistors with cutoff frequency of 27 GHz , and modulation-doped field effect transistors state-of-art power switching figure of merit of 586 MW/cm2.
The authors acknowledge funding from the Air Force Office of Scientific Research (AFOSR Grant No. FA9550-17-1-0227, Program Manager Dr. Kenneth Goretta) the DARPA DREaM program (No. ONR N00014-18-1-2033, Program Manager Dr. Young-Kai Chen, monitored by the Office of Naval Research, Program Manager Dr. Paul Maki), Air Force Office of Scientific Research under Award No. FA9550-18-1-0479 (GAME MURI, Program Manager Dr. Ali Sayir), NSF ECCS-1809682, and the Department of Energy / National Nuclear Security Administration under Award Number(s) DE-NA0003921.
 Razzak, Towhidur, et al. Applied Physics Letters 115.4 (2019): 043502.
 Xue, Hao, et al., Applied Physics Express 12.6 (2019): 066502.
 Xia, Zhanbo, et al.," IEEE Transactions on Electron Devices 66.2 (2019): 896-900
 Razzak, Towhidur, et al., Applied Physics Letters 116.2 (2020): 023507
 Kalarickal, Nidhin Kurian, et al., Applied Physics Letters 115.15 (2019): 152106
 Kalarickal, Nidhin Kurian, et al., Journal of Applied Physics 127.21 (2020): 215706
 Joishi, Chandan, et al., IEEE Transactions on Electron Devices (2020)
 Xia, Zhanbo, et al., IEEE Electron Device Letters 40.7 (2019): 1052-1055.
6:54 PM - *EL04.04.02
Current Status of Deep Level Defects in β-Ga2O3
Steven Ringel1,Hemant Jagannath Ghadi1,Joseph McGlone1,Rachel Adams1,Aaron Arehart1,Zixuan Feng1,A F M Anhar Uddin Bhuiyan1,Yuxuan Zhang1,Hongping Zhao1,Zhanbo Xia1,Nidhin Kalarickal1,Siddharth Rajan1,Alexander Senckowski2,Man Hoi Wong2
The Ohio State University1,University of Massachusetts Amherst2Show Abstract
Beta phase gallium oxide (β-Ga2O3) is a strong contender as the basis for next generation high power and RF devices, and there has been significant progress and interest in developing this material system. Investigating the presence and properties of crystalline defects and their manifestation as bandgap states within the ~ 4.8 eV β-Ga2O3 bandgap has been paramount to improve material quality. With β-Ga2O3 rapidly advancing as a forward-looking semiconductor device technology, the exploration of deep level defect states has been extensive and is now broadening beyond initial baseline studies that continue to guide optimization of crystal growth and material quality. These directions include the characterization of the influence of high energy particle radiation, which represents envisioned applications of β-Ga2O3 devices used in harsh environments, direct investigations of deep level defect impacts on β-Ga2O3 transistor characteristics, and the characterization of intentional defect engineering such as using compensating deep acceptors for achieving semi-insulating layers that are needed in some device applications.
This presentation will first review the state of knowledge regarding deep level defect distributions in β-Ga2O3 based on the application of deep level optical spectroscopy (DLOS), deep level transient (thermal) spectroscopy (DLTS), and admittance spectroscopy to β-Ga2O3 materials and devices. This combination of techniques enables quantitative characterization of defect states throughout the ultrawide β-Ga2O3 bandgap. A comparison of deep level distributions in energy and concentration will first be discussed based on materials grown using edge-defined film growth (EFG), metalorganic chemical vapor deposition (MOCVD), low pressure chemical vapor deposition (LPCVD), hydride vapor phase epitaxy (HVPE) and molecular beam epitaxy (MBE). Clear differences are observed depending on growth method, with the lowest total trap concentrations being found for MOCVD and HVPE materials. A deeper exploration of MOCVD materials reveal significant and inequivalent dependences of individual trap state concentrations on growth temperature, implying that the low total trap concentration in MOCVD material could be reduced further. Deep acceptor doping by Fe, Mg and N is also compared as a first step toward exploring the comparative efficacy of each impurity in enabling ideal semi-insulating behavior for β-Ga2O3, along with reviewing how Fe impacts the behavior of β-Ga2O3 transistors by virtue of its primary deep level position in the bandgap near EC – 0.8 eV. Finally, the β-Ga2O3 response to proton irradiation will be discussed, detailing the dominant radiation-induced states that are formed and the subsequent radiation hardness.
7:18 PM - *EL04.04.03
CVD Diamond P-I-N Structures for High Temperature Electronics, Radiation Detection and Electron Emission
Robert Nemanich1,Franz Koeck1,Mohamadali Malakoutian2,Harshad Surdi1,Jason Holmes1,Manpuneet Benipal3,Ricardo Alarcon1,Srabanti Chowdhury4,Stephen Goodnick1
Arizona State University1,University of California, Davis2,Advent Diamond Inc.3,Stanford University4Show Abstract
Diamond is a semiconductor with extreme and unique properties, which enable applications for high temperature electronics, radiation detectors, and electron emitters for ultra-high voltage vacuum switches and traveling wave tube cathodes. The fundamental structure for each of these applications is the p-i-n diode prepared from epitaxial, doped diamond layers. The availability of high-quality diamond substrates and the development of plasma CVD of epitaxial doped and undoped diamond have enabled fabrication of diodes with high breakdown field. CVD growth of high purity intrinsic diamond has shown room temperature mobility of both electrons and holes > 5000 cm2/V-s. Boron and Phosphorus doped layers achieve p-type and n-type character, respectively. Highly doped layers > 1019 cm-3 demonstrate low resistance through hopping conduction, which supports low resistance contacts. Epitaxial diamond PIN diodes show high current density injection mode transport described as space charge limited current. The unipolar hole current density through a Schottky diode intrinsic drift layer shows a V2 dependence as described by the Mott-Gurney expression where the differential resistance decreases as the voltage increases. Similarly, the differential resistance of a p-i-n diode shows a decrease of on-resistance with V indicative of bipolar conductivity modulation. At higher forward bias, velocity saturation is expected to limit the current in both unipolar Schottky and bipolar p-i-n diodes. The injection limited current transport of the undoped intrinsic diamond layer can enhance the predicted Baliga’s figure of merit for diamond relative to other wide bandgap semiconductors. Results on diamond p-i-n alpha particle detectors show essentially 100 % charge collection efficiency at an applied bias of <20V. Moreover, background from x-ray and gamma photons does not substantially i,pact the count rate. Similalry, efficient electron emission is observed from forward biased diamond p-i-n diodes. Here, improved n-type contact structures substantially enhances the emission. The tremendous progress in diamond applications is limited by materials challenges. As materials research progresses, new device concepts are being developed based on the outstanding, extreme and unique properties of diamond materials.
Acknowledgement: financial support by ARPA-E, DOE, NASA, NSF and ONR.
7:42 PM - *EL04.04.04
Diamond Electronics for Quantum Sensing
Tokyo Institute of Technology1Show Abstract
Diamond is an excellent host for spin-based qubits, and the spin in diamond has excellent properties. Nitrogen-vacancy (NV) center in diamond is one of the most promising candidates for quantum sensing. The energy levels of NV centers are sensitive to magnetic fields, electric fields, strain, and temperature, enabling scalable applications from the atomic to the macroscopic range .
Core technologies on material, devices, quantum control technology, sensor systems, and applications for life science and energy electronics are introduced.
Selectively-aligned NV ensemble formed by distinctive CVD-growth for scalable applications.
The heteroepitaxial growth of NV-contained diamond on Si substrate for large area and on-chip integration.
- Sensing devices using pn junctions.
- Multi-scale and multi-modal sensor systems for life-science and energy electronics applications [2-4].
This work was supported by MEXT QLEAP Grant Number JPMXS0118067395.
The author would like to thank Lab and Q-LEAP members for their contributions and helpful discussion.
 L. Doherty et al., Phys. Rep. 528, 1 (2013).
 A. Hoang et al., APL 118,044001 (2021)
 Y. Hatano et al., APL 118, 034001 (2021)
 Bang Yanget et al., Physical Review Applied 14, 044049 (2020).
8:06 PM - *EL04.04.05
UV-C Laser Diode with Distributed Polarization Doped P-Cladding Layer
Maki Kushimoto1,Ziyi Zhang1,2,Tadayoshi Sakai1,Naoharu Sugiyama1,Leo Showalter3,Yoshio Honda1,Chiaki Sasaoka1,Hiroshi Amano1
Nagoya University1,Asahi Kasei Corporation2,Crystal IS, Inc.3Show Abstract
UV laser diodes are candidates for health care applications such as water and air sterilization, medical sensing, and curing. Especially AlGaN-based UV optical devices are compact, have a long lifetime, and is environmentally friendly compared with conventional UV light sources. Recently, UVC and UVB LDs [1,2] are realized a long time after the UVA LD is reported . These were achieved by overcoming the several issues: low hole injection efficiency, high operation voltage, and optical loss caused by crystal defects. In particular, the the non-dopes distributed polarization doping (DPD)  structure of the p-clad layer contributes to the lasing by improving the hole injection efficiency. The average Al composition of 85% and hole concentrations in the mid-17th power range can be expected to be obtained by this method. The p-clad layer with a non-doped DPD structure has other advantages. The high Al composition in the region close to the active layer acts as an EBL layer that suppresses the electron overflow. Mg doping is not performed in this layer, so that absorption of light emission in the p-cladding layer can be suppressed. This mean that non-doped DPD is well matched method with the UVC-LD structure for p-type conductivity control.
The device structure we have prototyped consists of n, p-AlGaN cladding layer, guiding layer, single quantum well and p-contact layer. The p-type DPD cladding layer is constituted by continuously reducing the Al composition from AlN to AlGaN with 70% Al composition. From the evaluation of the optical properties, it was found that the internal loss could be reduced to less than 10 /cm with the appropriate DPD layer thickness . After electrode formation, the wafer cleavage and the deposition of HfO2/SiO2 distributed Bragg reflector were carried out. As a result, a pulsed LD operation of a 270 to 280 nm was observed at the room-temperature. Lasing was observed at a forward current density of 14 kA/cm2.
We also demonstrated LDs by the combined of dry and wet etching method as LD mirror facet fabrication. This method generally known to fabricate LDs grown on heterogeneous substrates, and it is attractive for mass production and monolithic integration. Since DBR formation was a major issue in this method, we focused on ALD with good coverage. As a result, the lasing was achieved with this combined method as well as the conventional cleaved method.
The authors would like to acknowledge Mr. Kazuhiro Nagase, and Dr. Naohiro Kuze of Asahi Kasei Corporation for their invaluable discussion and considerable support. The authors would also like to acknowledge Dr. Nishii and working members of C-TEFs for their great contribution to development of laser diode process.
1. Z. Zhang, M. Kushimoto, T. Sakai, N. Sugiyama, L. J. Schowalter, C. Sasaoka, and H. Amano, Appl. Phys. Express 12, 124003 (2019).
2. K. Sato, S. Yasue, K. Yamada, S. Tanaka, T. Omori, S. Ishizuka, S. Teramura, Y. Ogino, S. Iwayama, H. Miyake, et al., Appl. Phys. Express 13, 031004 (2020).
3. H. Yoshida, Y. Yamashita, M. Kuwabara, and H. Kan, Appl. Phys. Lett. 93, 241106 (2008).
4. D. Jena, S. Heikman, D. Green, D. Buttari, R. Coffie, H. Xing, S. Keller, S. DenBaars, J. S. Speck, and U. K. Mishra, et al., Appl. Phys. Lett. 81, 4395 (2002).
5. Z. Zhang, M. Kushimoto, T. Sakai, N. Sugiyama, L. J. Schowalter, C. Sasaoka, and H. Amano, JJAP 59 0904011 (2020).
Hongping Zhao, The Ohio State University
Masataka Higashiwaki, National Institute of Information & Comm Tech
Robert Kaplar, Sandia National Laboratories
Julien Pernot, University of Grenoble
Taiyo Nippon Sanso
EL04.05: GaN Power Electronics I
Monday AM, April 19, 2021
8:00 AM - *EL04.05.01
Wide Bandgap Semiconductor Based Power Electronic Devices for Energy Efficiency
Isik Kizilyalli1,Eric Carlson2
U.S. Dept. of Energy1,Booz Allen Hamilton2Show Abstract
Abstract – Development of advanced power electronics with increased functionality, efficiency, reliability, and reduced form factor are required in an increasingly electrified world economy. Fast switching power semiconductor devices are key to increasing the efficiency and reducing the size of power electronic systems as a significant portion of the losses in the systems is dissipated in the power semiconductor devices. However, the prevailing power semiconductor devices, which are based on the semiconductor silicon, are fast approaching their operational limits due to the intrinsic Si material properties. Wide-bandgap (WBG) semiconductors, such as gallium nitride (GaN) and silicon carbide (SiC), with their superior electrical properties are enabling a new generation of power semiconductor devices that offer higher efficiency and higher power conversion densities in a wide range of applications. The U.S. Department of Energy’s Advanced Research Project Agency - Energy (ARPA-E) has invested in WBG semiconductors to enable a new generation of power semiconductor devices that far exceed the performance of silicon-based devices. In 2017 ARPA-E launched the PNDIODES program to address a major barrier to fabricating GaN power electronic devices which is the lack of a viable selective area doping processes. At the launch of the program, the selective area doping processes commonly used for other semiconductor materials, such as ion implantation or solid state diffusion, had not produced satisfactory p-type regions or p-n junctions in GaN due to the thermodynamic decomposition of GaN at high temperatures which has limited the device fabrication processes. The selective area doping processes in GaN resulted in poor electrical performance not sufficient for power electronic applications. The goal of the PNDIODES program is to develop transformational advances and mechanistic understanding in the process of selective area doping in the group III-Nitride wide-bandgap semiconductor material systems in order to demonstrate arbitrarily placed, reliable, contactable, and generally useable p-n junction regions. The projects selected for funding as part of the PNDIODES program are developing innovative selective doping processes for GaN. These include cutting-edge high temperature annealing processes to activate implanted dopants and remove the damage caused by ion implantation while overcoming the thermodynamic limits of GaN decomposition, low temperature solid-state diffusion of dopants using pioneering mechanisms to increase the diffusion driving force, non-traditional selective area doping process using low damage patterned etching followed by GaN regrowth to form buried selectively doped regions, and nuclear transmutation techniques to convert host atoms to dopants in-situ through exposure to neutrons and/or high energy photons. The PNDIODES projects are using/developing advanced nanoscale characterization techniques to investigate the local optical, chemical, structural, and electrical properties of the selectively doped regions to gain a mechanistic understanding of the processes being developed. These include Photo and Cathodoluminescence, Raman Spectroscopy, Atom Probe Tomography, Secondary Electron Emission, X-ray Topography, Rutherford Backscattering Spectrometry/Channeling, Electron Paramagnetic Resonance, Nuclear Magnetic Resonance, Scanning Spreading Resistance Microscopy, Scanning Capacitance Microscopy, Photoconductivity Spectroscopy, and Electron Holography. The progress and challenges of selective area doping processes being developed under the PNDIODES program is reviewed along with the mechanistic understanding being generated. Material and processing challenges, including reliability concerns, for GaN power devices are also described. A glimpse into the future trends in device development, system integration, and commercialization is offered.
8:25 AM - EL04.05.02
Materials and Technology Issues of Vertical GaN Power FinFETs
Ahmad Zubair1,Joshua Perozek1,John Niroula1,Tomas Palacios1
Massachusetts Institute of Technology1Show Abstract
By 2030, about 80% of all US electricity is expected to flow through power-electronic devices. To enable this vision, the next generation of power electronics will require much higher efficiency and smaller form-factor than today’s silicon-based systems. III-Nitride semiconductors form an ideal material system for this thanks to the combination of excellent transport properties and the high critical electric field enabled by their wide bandgap.
Both lateral and vertical III-Nitride devices are being investigated. In this paper, we will study vertical GaN FinFETs, as the vertical fin channel offers excellent electrostatic and threshold voltage control, eliminating the need for epitaxial regrowth1 or p-type doping2 unlike other vertical power transistors. Vertical GaN FinFETs with 1200 V breakdown voltage (BV) and 5-A current rating have been demonstrated recently on free-standing GaN substrates3. Besides, the high current density of these devices, in combination with minimum parasitics, allow these devices to achieve beyond-state-of-the-art switching performance. This talk will discuss the recent progress of GaN vertical power FinFETs on native GaN substrates, highlighting the device and materials-level opportunities as well as some of the challenges to push the performance limits in these devices. In addition, the talk will review recent efforts on GaN vertical power FinFETs on non-GaN substrates4. In spite of their promising performance, the commercialization of vertical GaN FinFETs on native GaN substrates has been limited by the high cost ($50-$100/cm2) and small diameter (2-4 inch) of free-standing GaN substrates. The use of Si or engineered substrates with a matched thermal expansion coefficient could potentially reduce the substrate cost by 1000×, although key trade-offs between leakage currents and performance need to be carefully studied.
Acknowledgements - This research was supported by the ARPA-E SWITCHES and PNDIODES programs, monitored by Dr. Isik Kizilyalli.
 H. Nie et al., IEEE EDL vol. 35, no. 9, p.939-941 (2014).
 T. Oka et al., Appl Phys. Exp. vol. 7 no. 2, (2014).
 Y. Zhang et al., IEDM Tech. Dig., p. 9.2.1, (2017).
 A. Zubair et. al., Device Research Conference (2020).
8:40 AM - EL04.05.03
Laser-Assisted MOCVD GaN Epitaxy for Vertical Power Device Applications
Yuxuan Zhang1,Zhaoying Chen1,Zixuan Feng1,Hongping Zhao1
The Ohio State University1Show Abstract
GaN and its alloys have been widely utilized in optoelectronics, photonics, and electronics. Due to its large band gap (3.4 eV), strong critical electric field (3.4 MV/cm) and high electron mobility (>1000 cm2/Vs), the Baliga’s figure of merit (BFOM) of GaN is more than 500X and 3X higher than that of Si and SiC, exhibiting great potential for high power electronics. Vertical GaN PN diode with 5kV breakdown voltage has been demonstrated . To further improve the device performance, thicker drift layer with lower controllable doping is desired. For MOCVD GaN growth, TMGa and NH3 are typically used as Ga and N precursors. However, one of the challenges associated with current MOCVD technology is the low NH3 cracking efficiency (<1%), due to its high chemical stability even under high temperature (1000 °C). Carbon is considered as the most common impurity which originates from TMGa precursor. It is most likely to be incorporated on nitrogen site to form a deep acceptor (CN) in n-type GaN , which compensates effective doping, thus increase on-resistance and leakage current in vertical GaN power devices. Traditionally, C can be suppressed by increasing growth temperature (increase NH3 cracking efficiency) and V/III ratio (increase NH3 partial pressure) . Recent study  indicated that NH3 can be efficiently cracked by CO2 laser in which the photon energy is coupled with N-H wagging mode. With laser-assisted MOCVD (LA-MOCVD), GaN growth window is expected to be expandable to lower temperature regime with faster growth rate and lower impurity incorporation.
In this work, we performed a systematic study on the growth of GaN via LA-MOCVD growth technique. A tunable CO2 laser with wavelengths between 9.201 to 10.365 µm was utilized as external excitation source in a commercial MOCVD system. From our studies, NH3 has a strong interaction with the CO2 laser, while neither N2 nor H2 has much absorption of the laser beam. The MOCVD GaN growth condition is widely tuned, including the growth temperature, chamber pressure, NH3 flow rate, and type of carrier gas, to understand the effect of CO2 laser beam on the GaN growth. The preliminary results indicate that LA-MOCVD growth can suppress C incorporation in GaN. Meanwhile, the CO2 laser can also lead to strong gas phase reaction between precursors, which in turn leads to the reduction of GaN growth rate. LA-MOCVD growth of GaN at lower temperature regime indicates the suppression of V-pits formation as compared to the conventional GaN growth without the laser. Smooth surface morphology of LA-MOCVD GaN was observed under optical microscope, AFM and SEM, with low RMS of 0.34 nm measured by AFM. Therefore, LA-MOCVD can become a novel growth technique to grow high quality GaN suitable for high power vertical devices.
In summary, for the first time, we performed CO2 based LA-MOCVD GaN growth in a commercial MOCVD system. LA-MOCVD growth conditions are systematically mapped and comprehensive material characterization is performed. Results from this work indicate the great potential of using the novel LA-MOCVD growth method to achieve GaN epitaxy for vertical power devices with high breakdown voltages.
Acknowledgment: The authors acknowledge the funding support from Advanced Research Projects Agency-Energy (ARPA-E), U.S. Department of Energy (DE-AR0001036).
 H. Ohta, N. Asai, F. Horikiri, Y. Narita, T. Yoshida, and T. Mishima, Jpn. J. Appl. Phys. 58, SCCD03 (2019).
 J. L. Lyons, A. Janotti, and C. G. Van de Walle, Phys. Rev. B 89, 035204 (2014).
 D. D. Koleske, A. E. Wickenden, R. L. Henry, and M. E. Twigg, J. Cryst. Growth 242, 55 (2002).
 H. Rabiee Golgir, Y. Gao, Y. S. Zhou, L. Fan, P. Thirugnanam, K. Keramatnejad, L. Jiang, J. F. Silvain, and Y. F. Lu, Cryst. Growth Des.14, 4248 (2014).
8:55 AM - EL04.05.04
GaN P-N Power Diodes with > 1.5 kV Breakdown Voltage
Vishank Talesara1,Zhaoying Chen1,Yuxuan Zhang1,Hongping Zhao1,Wu Lu1
The Ohio State University1Show Abstract
In this work, we report 1.5 kV GaN PN diodes on an HVPE substrate. The GaN layers were grown on a 2” HVPE GaN substrate using the Metal Oxide Chemical Vapor Deposition (MOCVD). The dislocation density of the substrate ranges between 2×106 cm−2 in the less dense regions to 1×107 cm−2 in regions that show the presence of dense dislocation clusters. The device structure consists of a 20 nm p+ GaN contact layer(Mg: 1×1020 cm−3), followed by a 500 nm p-GaN layer (Mg: 1×1018 cm−3), grown on an 8 μm n-GaN drift region (Si: 1.2×1016 cm−3) and 1.6 μm n+ GaN buffer layer (Si: 1.5×1018 cm−3) on a thick bulk HVPE GaN substrate. The hole concentrations in the p and p+ GaN layers are 1×1017 cm−3 and 5×1017 cm−3, respectively, measured by Hall measurements. CV measurements show that the drift layer has an electron concentration of 5.1×1015 ~ 1.2×1016 cm−3. According to the analytical calculations, this device structure design has a theoretical punch-through breakdown voltage value of ~ 2.0 kV. For electrical field management, three layers of guard ring structures were designed and implemented by nitrogen ion implantation at the dose of 3.2×1013 ions/cm−2 at five ion energies which results in a depth of ~520 nm. Based on numerical simulations, the electrical field peaks at the edge of each guard ring and this device design should result in a breakdown voltage of 1670 V at a critical field of 3.4 MV/cm. For device fabrication, Ti/Al-based metal scheme was used to form the contact to n-GaN as the back cathode. Pt/Ni/Au anodes were annealed at 350 °C for p-contacts with a specific contact resistivity of 0.6 mΩcm2. The devices were then passivated by a 500 nm SiN layer after nitrogen ion implantation. The device size varies from 100 to 600 µm. The devices exhibit a breakdown voltage of 1530 V due to a clear avalanche process. The measured breakdown voltage has a remarkable agreement with numerical simulations. This suggests that the devices have an excellent breakdown efficiency > 75%. The reverse leakage current density is in the µA/cm2 range till 500 V. The turn-on voltage is 4.9 V at 100 A/cm2 and the device exhibited 1 kA/cm2 current density at 6.0 V. The specific on-resistance for this device is 1.2 mΩ cm2. This measured breakdown voltage and on-resistance lead to a figure of merit of 2.1 GW/cm2. Considering the p-contact resistivity, this suggests that the n-drift layer has an electron mobility of 480 cm2/Vs. The devices have an ideal factor of 2.7 at 2.3 V, likely due to a resistive layer at the bulk GaN substrate and epitaxy interface. More statistical analysis of device performance will be presented at the conference. Overall, this work represents the state-of-the-art device performances of GaN power diodes.
This material is based upon work supported by the U.S. Department of Energy's Office of Energy Efficiency and Renewable Energy (EERE) under the Advanced Manufacturing Office, FY18/FY19 Lab Call.
9:25 AM - EL04.05.06
Low-Defect Etched-and-Regrown PN Diodes by In Situ Tertiarybutylchloride (TBCl) Etching
Bingjun Li1,Sizhen Wang1,Jung Han1
Yale University1Show Abstract
Gallium Nitride (GaN)-based devices are very attractive candidates for next-generation high-power applications, due to the intrinsic material properties. However, the lack of selective-area doping technique in GaN greatly limits the design flexibility and device performance. Due to the difficulty of post-ion-implantation annealing process in GaN material, selective-area etching (SAE), followed by selective area growth, is an alternative and promising approach to overcome this obstacle. Plasma-etching, as a well-developed technique to create smooth and high aspect-ratio trenches, induces serious optical and electronic defects and impurities on the treated surface. Therefore, we introduced a chlorine-based metal-organic (MO) precursor, tertiarybutylchloride (TBCl), into the MOCVD reactor to replace the role of plasma etching.
TBCl etching was also performed on 1.5µm unintentional-doped (UID) GaN templates grown on bulk GaN substrate to mitigate the effect of dislocation-mediated etching. Four samples are compared here. Sample A is a template. Sample B~D are templates etched by Cl-based plasma, TBCl and a combination of both, respectively. Photoluminescence (PL) showed strong near-band-edge emissions only from Sample A, C and D (Fig. 3(a)). And only Sample B had obvious Cl peak from x-ray photoelectron spectroscopy (XPS) (Fig. 3(b)). Both PL and XPS confirm that TBCl etching is able to remove the impurity (except Si and Al) and damage induced by plasma etching and does not introduce damage itself.
Besides, electrical properties were also evaluated. Four planar PN diodes were grown and fabricated. Device A is a continuous PN diode, while Device B~D were etched-and-regrown PN diodes. Three different etching were performed on 1.5µm UID drift layer grown on GaN substrate: plasma etching followed by AZ400K cleaning for Device B; plasma etching (ICP)+AZ400K cleaning+150nm TBCl etching for Device C; and 150nm TBCl etching for Device D. 300nm UID and 400nm p-GaN was regrown on the etched templates. Forward and reverse I-V characteristics were compared. No obvious difference is observed when they were forward biased. However, leakage behaviors are quite distinct. The TBCl-etched diode (Device D) shows the best leakage behavior among all three regrown devices, while ICP+TBCl-etched diode has the highest leakage current. From SIMS and CV measurement comparisons, Device C has the highest Si impurity concentration, which could explain the worst leakage behavior. On the other hand, a high concentration of deep traps was observed on ICP-etched diodes from CV measurement.
This work is supported by the Advanced Research Projects Agency-Energy (ARPA-E), U.S. Department of Energy, under Award Number DE-AR0000871 as part of the PNDIODES program managed by Dr. Isik Kizilyalli.
9:40 AM - EL04.05.07
Scanning Capacitance Microscopy of GaN p-n Junction Through Planar and Selective Area Doping
Sizhen Wang1,Bingjun Li1,Jung Han1
Yale University1Show Abstract
Scanning capacitance microscope (SCM) integrates a high-frequency capacitance sensor into the conductive atomic force microscope (C-AFM) to measure the change of capacitance of a well-known metal-oxide-semiconductor (MOS) or Schottky diode structure under an AC driving voltage. The dC/dV amplitude and phase of the output signal tells the doping type and carrier concentration, respectively. SCM has demonstrated the capability to delineate the p-n junction and form a 2-D dopant mapping on Si substrate accurately. In addition, doping characterization using SCM on GaN were also reported. Here, we demonstrated SCM characterization of a selective-area grown (SAG) lateral GaN p-n junction on bulk GaN substrate for the first time, which is a key building block of high-performance GaN-based high-power devices.
To quantify the doping concentration in a 2-D mapping, dC/dV amplitude and phase need to be calibrated with the known doping level at the beginning. Therefore, a sample consisting of multiple GaN layers with Si, Ge doping concentration ranging from unintentional doped GaN to 1x1020/cm3 were used for the calibration. And the dC/dV amplitude peak was selected as parameter for calibration, which can reduce noise impact and eliminate the contrast reverse issue. By applying different AC driving voltage, near four order of magnitude doping of GaN can be calibrated on bulk GaN. Similarly, p-GaN layers with Mg concentration range from 1x1019 to 8x1019/cm3 were also calibrated. Based on those calibration studies, we proposed an efficient operation procedure to characterize semiconductor samples with scanning capacitance microscopy and apply this procedure to investigate the continuously grown diodes, planar regrown diodes, and selective area regrown GaN diodes.
Several key results of applying SCM to characterize vertical GaN diode were summarized here: (1) to identify the p-n junction location, dC/dV phase is preferred to be used, because comparing to dC/dV amplitude, the dC/dV phase is much less susceptible to DC bias and AC driving signal. (2) SCM revealed that a thin n-type conducting layer existed at the regrown interface of planar regrown diode. This n-type layer, with increased conductivity comparing to UID-GaN, was related to ICP plasma damage. (3) For SAG regrown diodes, typical p-GaN layer was verified with SCM by applying proper DC bias condition, it was also found at the trench bottom (or regrowth interface), a typical n-type thin layer existed, which should not in ideal case. The possible n-type doping impurity might come from the out-gassing of the SiO2 mask during the growth or other reactor parts made of quartz, which needs further investigation.
9:55 AM - EL04.05.08
Late News: Surface Photovoltage Study of the Bulk Photovoltaic Effect in Carbon-Doped Gallium Nitride
Igal Levine1,Ivan Gamov2,Marin Rusu1,Klaus Irmscher2,Christoph Merschjann1,Eberhard Richter3,Markus Weyers3,Thomas Dittrich1
Helmholtz-Zentrum Berlin1,Leibniz-Institut für Kristallzüchtung2,Ferdinand-Braun-Institut GmbH, Leibniz-Institut für Höchstfrequenztechnik3Show Abstract
Surface photovoltage (SPV) of carbon-doped gallium nitride single crystals (GaN:C) was investigated by Kelvin probe as a function of carbon doping.1 GaN:C crystals are highly resistive due to compensating deep defect states. Depending on the carbon doping, SPV signals much larger than expected in relation to the band gap of GaN were detected setting on at photon energies of 2.5-2.6 eV due to excitation from defect states. Under constant illumination, the large SPV signals saturated at values above 20 V and even overshot abruptly by up to several volts above the saturation value when switching off the light source. The sign of the large SPV signals changed from positive to negative when flipping the crystal from the Ga-polar to the N-polar surface. Therefore, the large SPV signals on GaN:C are not related to a conventional Dember photovoltage, which is based on different mobilities for photogenerated electrons and holes, but rather to a bulk photovoltaic effect (BPVE) caused by a structure dependence of directed momentum transfer. It is proposed that (tri)carbon complexes are key for directed momentum transfer in GaN:C and the resulting observed BPVE.
(1) Levine, I.; Gamov, I.; Rusu, M.; Irmscher, K.; Merschjann, C.; Richter, E.; Weyers, M.; Dittrich, T. "Bulk Photovoltaic Effect in Carbon-Doped Gallium Nitride Revealed by Anomalous Surface Photovoltage Spectroscopy". Phys. Rev. B 2020, 101 (24).
EL04.06: GaN Power Electronics II
Monday PM, April 19, 2021
10:30 AM - *EL04.06.01
Design and Fabrication of Edge-Termination for Achieving Reliable GaN P-N Diodes for High Voltage Applications
Srabanti Chowdhury1,Andrew Allerman2,Robert Kaplar2,Andrew Armstrong2,Jeramy Dickerson2,Alan Jacobs3,Karl Hobart3
Stanford University1,Sandia National Laboratories2,U.S. Naval Research Laboratory3Show Abstract
Electrification drives us closer towards our goal of a sustainable energy system. Highly efficient and reliable ways of converting, delivering and conditioning power calls for innovations at every level, from system to materials. Silicon’s role in the electrification has led to exploring wide- bandgap (WBG) materials. SiC led power electronics have revealed the large scope of improvement that one can expect to see out of WBGs due to an enhancement of the critical electric field. GaN, originally pushed for medium voltage (650V-900V) devices using the well-known HEMT configuration, has a lot more to offer in power electronics through its vertical configuration. Over the last decade a tremendous amount of progress has been made with the support from ARPA-E and ONR which resulted in the improvement of the epitaxial material, understanding of the underlying physics, optimization of design and processing of the devices, accurate failure analysis, and finally, laying the groundwork for mass production.
Under the current ARPA-E Open plus program our team is developing high voltage (towards 20 kV) GaN PN diodes for use as devices to protect the electric grid against electromagnetic pulses . A common goal shared by both DOD and DOE is to have a foundry component in the US to successfully scale and produce vertical GaN devices for circuit implementation. Under the current program, we are bringing together all the research components starting from material development and wafer metrology to reliability and failure analysis. These are the essential components for successful manufacturing of GaN devices leading to its transfer to the foundry.
Achieving a reliable p-n junction definitely is the first step towards high voltage (1.2kV and up) power devices. In this talk we will focus on various approaches of edge termination to realize a robust p-n junction. Bevel and junction termination extension (JTE) approaches are being pursued on p-n diodes to study their efficacies in field mitigation. Using the bevel termination technique for 1.2kV diodes, we are studying the reverse blocking strength of these diodes, by closely monitoring their avalanche capabilities. The foundry effort is utilizing as a primary path various combinations of nitrogen-implanted isolation structures, guard rings, and JTEs.
The team, through a tight research collaboration, is working towards identification of the key challenges of edge- terminations and the best solution in GaN, to serve various voltage classes.
The authors gratefully acknowledge the support of ARPA-E’s OPEN+ Kilovolt Devices Cohort managed by Dr. Isik Kizilyalli, and support of ONR managed by Mr. Lynn Petersen.
 R. Kaplar, et al. IEDM 2020
10:55 AM - EL04.06.02
Variations in GaN Substrates and the Structural Evolution of Defects in Homoepitaxial GaN Layers
Yekan Wang1,Michael Liao1,Kenny Huynh1,Andrew Allerman2,Mark Goorsky1
University of California, Los Angeles1,Sandia National Laboratories2Show Abstract
In this study, the structural defects in 40 mm homoepitaxial GaN grown on what is commonly referred to as dot-core GaN substrates were evaluated using synchrotron double crystal x-ray topography (XRT). 8.05 keV X-ray energy was used. The first crystal was an asymmetric Si (333) beam expander and the sample (2nd crystal) was oriented for diffraction of the (114) reflection. Films were obtained from single exposure at particular positions on the rocking curve and continuous exposure by rocking the sample (step size ~0.008°) about the  axis through the Bragg condition.
For as-received dot core substrates, regions of high distortion around the cores are observed. These defective regions are not confined within the core regions. The effect of tilt and distortion distribute to further affect regions between the cores as well. Meanwhile, variations in the substrates are observed. The substrates curvature for three as-received dot core substrates from the same vendor is quite different, with a radius of curvature of 25m, 17m, and 7m respectively. The defect structure surrounding the cores in the three substrates is also different. The distortions around the cores in two of the three substrates have a few hundred um white round (non-diffracting) regions around the cores while the other substrate shows irregular distorted regions. A few ‘butterfly’ shape patterns are present around some cores, which result from localized tilting of the lattice planes. Each side of the butterfly wing consists of contour lines representing regions of similar tilt. Rotating the sample every 60° and taking rocked images along different (1124) reflections, we found the ‘butterfly’ patterns follow the 60° rotation, suggesting that the lattice tilt around the cores occurs radially. The sample with 40 um epilayer grown on a dot core substrate shows a change in the defect structure around the cores. Defective regions after epitaxial growth are more concentrated in the core areas. A hypothesis is that defects observed on the bare substrate propagate into the overlaying epilayer. However, during epitaxial growth, structural distortions are redirected towards the centers of cores. The surface shows haze and specular regions with variation in AFM roughness (1.3 nm vs 0.4 nm). More defective regions observed using XRT appears to be correlated with haze on the surface. Variation of distortions across the wafer can also be identified. There are circular spot features with bright contrast, corresponding to highly defective regions of tilt after the epitaxial growth. Between cores and over some cores, there is undistorted, low dislocation density (~105 cm-2) material with uniform grey contrast, corresponding to higher quality GaN (hundreds of um-size). Devices fabricated on such regions are expected to have better performance. In conclusion, the structural evolution of defects after the epitaxial growth is closely related to variations in the substrates. As a result, wafer screening and quality assessment using non-destructive techniques such as x-ray topography before and after the epitaxial growth are promised to help improve the overall quality of the homoepitaxial GaN structures.
The authors would like to acknowledge support through the ARPA-E PNDIODES program under contract DE-AR0001116 at UCLA. This research used resources of the Advanced Photon Source, a U.S. Department of Energy (DOE) Office of Science User Facility operated for the DOE Office of Science by Argonne National Laboratory under contract no. DE-AC02-06CH11357. The synchrotron X-ray topography measurements were carried out at 1-BM Beamline of the Advanced Photon Source, Argonne National Laboratory.
11:10 AM - EL04.06.03
Application of Synchrotron X-Ray Topography to Characterization of Selective Area Doping Processes for the Development of Vertical GaN Power Devices
Yafei Liu1,Hongyu Peng1,Tuerxun Ailihumaer1,Shanshan Hu1,Balaji Raghothamachar1,Michael Dudley1
Stony Brook University, The State University of New York1Show Abstract
Selective area p-type doping of GaN is required for the development of vertical GaN devices that will help realize the potential of WBG semiconductor GaN in power electronics [1, 2]. Approaches being investigated include implantation followed by activation annealing by different methods, selected area etching and regrowth of p-type regions , diffusion doping and neutron transmutation . Using synchrotron X-ray topography  and complementary tools like HRXRD and Raman spectroscopy, the effect of substrate choice, implantation and annealing conditions have been evaluated on the structural quality and strain in the epilayers and regrown material. The choice of bulk GaN substrates plays an important role in the eventual extended defect configurations in the active layers. Ammonothermal-grown GaN substrate wafers show the best quality among all the wafers [6, 7]. These wafers, which are free of basal plane dislocations (BPDs), have low curvature and threading mixed dislocations (TMDs) dominant among the threading dislocations (TDs). Patterned HVPE GaN reveal a starkly heterogeneous distribution of dislocations with large areas containing low threading dislocation densities in between a grid of strain centers with higher threading dislocation densities and BPDs . The strain level of HVPE GaN substrates is very high, and the dislocation density is around 105-106 cm-2, which is much higher than 104 cm-2 of ammonothermal samples and dislocation-free areas in the patterned HVPE samples. During epitaxial growth by CVD for implantation purposes, defects in substrates are shown to replicate into the epilayer and typically no new defects are observed to be introduced at the interface. On implantation, damaged layers are generated in the epilayer as revealed by satellite peaks in double axis rocking curves. The radiation fluence and energy determine the extent of damage. Depending on annealing conditions most of the damage is healed. However, the annealing temperatures greater than 1100 C can result in introduction of inhomogeneous strains and dislocation generation. While etching by TBCl is shown to be sensitive to certain types of threading dislocations, any thermal treatment is shown to introduce basal plane dislocations. Further investigations are underway to analyze the regrowth interface for the nucleation of new defects. Results will be discussed with implications for vertical device fabrication and expected impact on device performance.
 I.C. Kizilyalli, A.P. Edwards, H. Nie, D. Disney, D. Bour, IEEE Transactions on Electron Devices 60 (2013) 3067-3070.
 I.C. Kizilyalli, A.P. Edwards, H. Nie, D. Bour, T. Prunty, D. Disney, IEEE Electron Device Letters 35 (2014) 247-249.
 B. Li, S. Wang, M. Nami, J. Han, Journal of Crystal Growth 534 (2020) 125492.
 R. Barber, Q. Nguyen, J. Brockman, J. Gahl, J. Kwon, Scientific Reports 10 (2020) 1-8.
 B. Raghothamachar, M. Dudley, G. Dhanaraj, X-Ray Topography Techniques for Defect Characterization of Crystals, in: G. Dhanaraj, K. Byrappa, V. Prasad, M. Dudley (Eds.), Springer handbook of crystal growth, Springer Science & Business Media, 2010, p. 1425.
 Y. Liu, S. Hu, H. Peng, T. Ailihumaer, B. Raghothamachar, M. Dudley, ECS Transactions 98 (2020) 21.
 Y. Liu, B. Raghothamachar, H. Peng, T. Ailihumaer, M. Dudley, R. Collazo, J. Tweedie, Z. Sitar, F.S. Shahedipour-Sandvik, K.A. Jones, Journal of Crystal Growth 551 (2020) 125903.
 B. Raghothamachar, Y. Liu, H. Peng, T. Ailihumaer, M. Dudley, F.S. Shahedipour-Sandvik, K.A. Jones, A. Armstrong, A.A. Allerman, J. Han, H. Fu, K. Fu, Y. Zhao, Journal of Crystal Growth 544 (2020) 125709.
11:25 AM - EL04.06.04
Late News: Microstructure Impacts on Performance of GaN Power Switching Devices
Brett Setera1,Aris Christou1
University of Maryland1Show Abstract
GaN power switching devices with thick epitaxial layers (3-15 microns) suffer from a high density of crystal defects, and specifically threading dislocations(TDs) in the epi-layer. Presence of dislocations leads to non-ideal performance and higher costs manifested in device voltage derating, limited die sizes, poor wafer yields, operating parameter instability, and limited device types. Microstructure analysis of defect type and location is crucial to correlating structure with device performance.
Experimental results with 10 μm GaN epilayer grown on HVPE(hydride vapor phase epitaxy) + GaN indicate that threading dislocations can propagate through the epilayer to the contact, resulting in increased gate leakage current density. Cathodoluminescence (CL) imaging correlated with ECCI as well as AFM allow the conclusion to be reached that the main threading defects are the edge dislocations and dislocation loops which tend to be either tilted, exposing the edge dislocations, or parallel to the growth plane. Utilizing the three aforementioned non-destructive techniques allows measurement of as-grown defects before the device undergoes performance testing. Transmission x-ray topography analysis showed that the GaN is characterized by a nearly uniform distribution of strain centers which are likely bundles of threading screw and edge dislocations. Strain free centers are then regions of relatively low dislocation densities. The threading dislocations are part of dislocation loops which are present in the GaN substrate and migrate through the interface. Cross-sectional TEM showed partial dislocation loops through the thickness of the GaN sample.
The two terminal GaN-on-GaN (HVPE) device was able to withstand 2.77 MV/cm before the contact to contact leakage current produced a gate – source short with an epilayer dislocation density of 106 cm-3.
11:40 AM - EL04.06.05
MOCVD GaN Epitaxy with Fast Growth Rates
Yuxuan Zhang1,Zhaoying Chen1,Wenbo Li1,Aaron Arehart1,Steven Ringel1,Hongping Zhao1
The Ohio State University1Show Abstract
Gallium nitride (GaN) has been considered as a promising candidate for power electronic devices due to its wide bandgap (3.4 eV), high critical electric field (3 MV/cm), and high electron mobility (>1000 cm2/V s). As a key parameter to evaluate the performance of power devices, Baliga’s figure of merit (BFOM) of GaN is more than 500X higher than silicon (Si) and 3X higher than silicon carbide (SiC) . Vertical GaN PN diode with 5kV breakdown voltage has been demonstrated . Although tremendous progress has been made in the past few years, challenges still exist to achieve vertical GaN power devices with VBR>10 kV, which requires the epitaxy of high quality GaN films with thick drift layer and low controllable doping with high mobility. Typical MOCVD GaN films with high mobilities were grown with growth rate of 2-3 µm/hr. Thus, it is important to develop and understand MOCVD GaN growth with fast growth rate, high crystalline quality, and controllable low carrier concentration.
In this work, the effect of increasing growth rate by increasing TMGa on the impurity incorporation, charge compensation, surface morphology and transport properties are systematically studied. Under optimized MOCVD GaN growth with a growth rate of 2 µm/hr, high quality GaN with stable low net charge density at 4×1015 cm-3 was demonstrated. [Si] and electron concentration (Nd-Na) as a function of SiH4 flow was studied. Stronger plummet of Nd-Na was observed with reduction of SiH4 flow when Nd-Na is lower than 1×1016 cm-3, due to prominent compensation effect at low doping range. According to capacitance-voltage (CV) and secondary ion mass spectroscopy (SIMS) analysis, the [C] level was at 7×1015 cm-3 (SIMS detection limit). Low Nd-Na was achieved at 4×1015 cm-3 based on CV. Deep level transient spectroscopy / deep level optical spectroscopy (DLTS / DLOS) results show minimal electron trap concentration at 2.5×1015 cm-3, mainly from carbon related deep level defects. By increasing the TMGa flow rate, GaN with fast growth rate at 5.2 µm/hr was achieved with stable Nd-Na at 1.5×1016 cm-3 and [C] at 2×1016 cm-3. Large area atomic force microscope (AFM) images on fast growth rate sample show smooth morphology with clear atomic strips and low surface roughness (RMS=0.647 nm). The Mn incorporation from the semi-insulating ammonothermal substrate into epilayer was observed. Controlled experiments show that Mn incorporation rate into epilayer is highly related to the growth rate, and Mn incorporation mechanism shares a similar behavior as Fe. SIMS showed that both Mn and Fe incorporation can be suppressed with faster growth rates. Room temperature Hall measurement showed that with electron concentration at around 1.5×1016 cm-3, electron mobility decreased from 852 cm2/Vs to 604 cm2/Vs as growth rate increasing from 2 µm/hr to 5.2 µm/hr.
In summary, we studied the doping and compensation of n-type GaN as a function of growth rate. Low compensation level at 2 µm/hr was confirmed by both DLTS/DLOS and CV. Increasing GaN growth rate led to the increase of carbon concentration while suppressing Mn and Fe incorporation. In order to achieve high quality thick GaN drift layer with controllable doping concentration at low- to mid- 1015 cm-3 range for high voltage vertical power devices, the key challenge is to reduce carbon incorporation in the GaN epilayer. The results from this work provide insights for GaN vertical power electronics.
Acknowledgment: The authors acknowledge the funding support from Advanced Research Projects Agency-Energy (ARPA-E), U.S. Department of Energy (DE-AR0001036), and U.S. Department of Energy's Office of Energy Efficiency and Renewable Energy (EERE) under the Advanced Manufacturing Office, FY18/FY19 Lab Call.
 B. J. Baliga, Fundamentals of Power Semiconductor Devices (Springer Science & Business Media, 2010).
 H. Ohta, N. Asai, F. Horikiri, Y. Narita, T. Yoshida, and T. Mishima, Jpn. J. Appl. Phys. 58, SCCD03 (2019).
11:55 AM - EL04.06.06
Influence of Surface Treatments on the Structure and Properties of GaN Layers
Jiaheng He1,Guanjie Cheng1,Maggie Chen1,Zhirong Zhang1,Sam Frisone1,Alexandra Zimmerman1,Fabian Naab1,Sizhen Wang2,Bingjun Li2,Jung Han2,Rachel Goldman1
University of Michigan–Ann Arbor1,Yale University2Show Abstract
Although silicon-based electronics are used to power light-emitting diodes and electric vehicles, their utility in high power applications is limited by slow switching and high on-state resistance. The most promising alternatives are vertical GaN devices, but these involve etching and selective-area re-growth which enhance the displacement of surface and near-surface Ga and N atoms. To understand processing-structure-property relationships relevant to vertical GaN devices, we are examining the influences of various surface treatments on the structure and properties of GaN layers. Utilizing ion beam analysis, we have examined the influence of ambient exposure and dry etching on the structure and properties of regrown p-i-n GaN structures. We show that dry etching improves the crystallinity of p-i interfaces, but also introduces interfacial H. In addition, we report on the influence of dry etching and metal-organic (MO) precursor treatment on the structure and properties of GaN substrates and epitaxial GaN layers. For these studies, we visualize the crystal symmetry and orientation using 2D planar ion channeling maps and angular yield profiles collected with a fully-automated 5-axis goniometer recently attached to the RC43 endstation of the 1.7 MeV Tandetron at the Michigan Ion Beam Laboratory. Our preliminary results suggest that the MO precursor reduces the density of displaced surface Ga atoms. To quantify the concentration and distribution of displaced atoms, we will compare 2D ion channeling maps with 2D Monte Carlo-Molecular Dynamics simulations using Flux 7.9.6. We will also present 2D maps of elastic recoil detection analysis spectra to evaluate the spatial distribution of H.
We gratefully acknowledge the support of ARPA-E through AWD0000191.
12:10 PM - EL04.06.07
X-Ray Diffraction Studies of GaN p-i-n Structures for High Power Electronics
Alexandra Zimmerman1,Jiaheng He1,Guanjie Cheng1,Davide del Gaudio1,Jordan Occena1,Fabian Naab1,Mohsen Nami2,Bingjun Li2,Jung Han2,Rachel Goldman1
University of Michigan–Ann Arbor1,Yale University2Show Abstract
Although silicon-based electronics are used to power light-emitting diodes and electric vehicles, their utility in high power applications is limited by a low breakdown voltage. The most promising alternative power devices consist of vertical GaN devices, which often require regrown active regions. Here, we report on x-ray diffraction studies of the crystallinity of the GaN p-i-n structures prepared with and without ex-situ ambient exposure and/or chemical etching. To quantify the mosaicity and threading dislocation (TD) densities at the p-i interfaces, we quantify and compare the full-width at half-maximum (FWHM) of both phi and omega x-ray diffraction scans. For the "in-situ" GaN structure, the screw-type TD densities are lowest, the edge-type TD densities are highest, and the interfacial near-band edge (NBE) and donor-acceptor pair (DAP) cathodoluminescence (CL) emissions are highest. Interestingly, elastic recoil detection analysis (ERDA) and Rutherford backscattering spectroscopy reveal minimal interfacial [H] but the highest fraction of displaced Ga atoms, suggesting efficient incorporation of MgGa. On the other hand, for the ex-situ structures, minimal interfacial [H] is also observed, along with a high screw-type TD and low interfacial NBE and DAP CL emissions. Finally, for the etched/regrown structures, ERDA reveals the highest interfacial [H], along with moderate screw- and edge-type TD, moderate DAP CL, and significant yellow CL emission. The relationship between interfacial [H], displaced Ga, CL emission features, and screw- and edge-type TD densities will be discussed.
We gratefully acknowledge the support of ARPA-E through AWD0000191.
EL04.07: Ga2O3 I
Monday PM, April 19, 2021
1:00 PM - *EL04.07.01
Defects in b-Ga2O3—An Ultra High-Resolution Scanning/Transmission Electron Microscopy Imaging and Spectroscopy
Nasim Alem1,Adrian Chmielewski1
The Pennsylvania State University1Show Abstract
Group III wide band gap (WBG) oxides are considered ideal materials systems for power electronics at extreme conditions. Due to their high band gap, this family of materials has a high breakdown voltage and high resistivity to electric field and temperature. Among WBG oxides, β-Ga2O3 is considered an ideal candidate for high power electronics due to its high band gap and its high breakdown electric field compared with GaN and 4H-SiC. Yet, defects currently limit the high performance of β-Ga2O3 crystal in high power electronic applications. While there have been a number of studies on the characterization of the defects in β-Ga2O3, little is known about the underlying atomic scale physics and chemistry of such defect complexes within the crystal, at the interfaces, and how they can affect the macroscale electronic properties. In addition, little is known about the stability and transition dynamics of the crystal under extreme conditions, i.e. thermal or electrical biasing, ultimately leading to its failure.
In this work we use advanced scanning/transmission electron microscopy imaging and electron energy loss spectroscopy to understand the atomic and chemical structure of b-Ga2O3 and epi thin film of b-Ga2O3/b-(AlxGa1-x)2O3. We present the atomic and chemical structure of the defects, interfaces, and its structural transformation in the bulk b-Ga2O3 and epi thin film of b-Ga2O3/b-(AlxGa1-x)2O3. In addition, nanoscale modulations in the electronic structure and the band gap in the epi thin films of b-Ga2O3/b-(AlxGa1-x)2O3 is discussed. This presentation will further cover the nanoscale transition dynamics of b-Ga2O3 lattice under extreme environments such as high temperature and electric field. This fundamental study can bridge the gap between atomic scale structural modifications and macroscale materials functionality not only in b-Ga2O3 but also in the family of WBG oxides.
1:25 PM - EL04.07.02
Mg Acceptor Doping in MOCVD (010) β-Ga2O3
Zixuan Feng1,A F M Anhar Uddin Bhuiyan1,Nidhin Kalarickal1,Siddharth Rajan1,Hongping Zhao1
The Ohio State University1Show Abstract
Metalorganic chemical vapor deposition (MOCVD) of β-Ga2O3 thin films have been demonstrated with record-high room temperature and low-temperature mobilities that approach the theoretically predicted limit. [1,2] The extracted low acceptor concentration (Na < 1015 cm-3) is extremely encouraging for its potential application in high power electronics. Among various impurities in MOCVD grown β-Ga2O3, Si represents the dominant one that contributes to the conductivity in the unintentionally doped (UID) films. It is commonly observed the Si spike at the growth interface between Ga2O3 substrate and the epitaxial layer. The existence of interface charges not only severly impacts charge transport characteristics , but also detrimentally affects device performance, such as causing buffer leakage current in lateral power devices.  In addition, controllable charge compenstation serves as a key component in device designs, e.g., forming current blocking layer. Without effective p-type β-Ga2O3, an alternative route is to use semi-insulating layer to engineer the electric field in devices. Thus far, there are limited reports on the epitaxy of semi-insulating β-Ga2O3.
Among various acceptors in β-Ga2O3, Mg represents one of the most promising candidates with relatively shallow acceptor level and the lowest formation energy as compared to other cation-site acceptors from DFT calculation.  In this study, Mg in-situ doping in MOCVD β-Ga2O3 was conducted for the first time. Trimethylgallium (TEGa) and O2 were used as Ga, O precursors and Ar as the carrier gas. Mg doping was introduced by using Cp2Mg as precursor. Chamber pressure was set at 60 Torr in this study. The MOCVD growth temperature for β-Ga2O3 was expanded to the range from 650 °C to 900 °C. The growth was conducted on commercial Fe-doped (010) β-Ga2O3 substrates. From secondary ion mass spectroscopy (SIMS), in-situ Mg doping concentration was tuned in the range of 1018 cm-3 to 1020 cm-3 by the variation of Cp2Mg molar flow. H impurity concentration exhibits an obvious companion with Mg doping concentration, indicating possible Mg-H complex configurations in as-grown MOCVD Mg-doped β-Ga2O3.
In addition, we analyzed the SIMS diffusion characteristics of Mg acceptor under different growth temperature of 700 °C to 900 °C. Experimental results indicate Mg incorporation has a minimum dependence on the growth temperature. Instead, the Mg diffusion has a strong dependence on the growth temperature. The diffusion barrier energy of Mg in MOCVD β-Ga2O3 was estimated at ~0.9 eV based on numeric analysis of the SIMS profile. Capacitance-voltage (C-V) on lateral Schottky diode structures, with an Mg-doped buffer layer and a Si-doped channel layer, also verified the depletion of interface charge, demonstrating the effective charge compensation by in-situ Mg doping.
In summary, we demonstrated in-situ Mg acceptor doping in MOCVD growth of (010) β-Ga2O3 thin films. The growth conditions for Mg-doped MOCVD β-Ga2O3 were established with controllable doping between 1018 cm-3 to 1020 cm-3 and a wide growth temperature regime, ranging between 700 °C and 900 °C. The as-grown thin films were characterized to be electrically insulating despite the Mg-H chemical companion. Mg diffusion was observed strongly dependent on the growth temperature. With low growth temperature, Mg diffusion can be significantly suppressed while maintaining high crystalline quality. The demonstration of in-situ Mg-doping in MOCVD β-Ga2O3 can provide new routes for high-performance device design and device fabrication.
Acknowledgment: The authors acknowledge the funding support from the Air Force Office of Scientific Research No. FA9550-18-1-0479 (AFOSR, Dr. Ali Sayir).
 Z. Feng et al., Appl. Phys. Lett., 114, 250601 (2019).
 Z. Feng et al., Phys. Status Solidi RRL 14, 2000145 (2020).
 M. H. Wong et al., Jpn. J. Appl. Phys. 55, 1202B9 (2016).
 J. L. Lyons, Semicond. Sci. Technol. 33, 05LT02 (2018).
1:40 PM - EL04.07.03
Acceptors in Gallium Oxide
Matthew McCluskey1,Jani Jesenovec1,Jacob Ritter1,Christopher Pansegrau1,John McCloy1
Washington State University1Show Abstract
Monoclinic gallium oxide (β-Ga2O3) is an ultra-wide bandgap semiconductor with potential applications in power electronics. Semi-insulating substrates are required for most practical devices such as metal-oxide-semiconductor field effect transistors. This presentation will discuss recent experimental studies on Czochralski-grown β-Ga2O3 single crystals doped with Mg or Zn acceptors. These dopants result in semi-insulating material and are likely compensated by oxygen vacancies and shallow donors. Ir impurities originating from the crucible form deep donors that also compensate acceptors. The Ir4+ oxidation state gives rise to an absorption threshold in the visible/UV part of the spectrum and an IR absorption peak at 5150 cm-1. Acceptors are passivated by hydrogen, an omnipresent contaminant, resulting in IR absorption peaks corresponding to O-H vibrational modes near 3300 cm-1.
1:55 PM - EL04.07.04
High-Mobility Low-Temperature Metalorganic Vapor Epitaxy Grown (010) β-Ga2O3 Homoepitaxial Films and Its Application to Realize Low Resistance Ohmic Contacts
Arkka Bhattacharyya1,Praneeth Ranga1,Saurav Roy1,Sriram Krishnamoorthy1
The University of Utah1Show Abstract
Metalorganic vapor phase epitaxy (MOVPE) has emerged as the leading growth technique that allows for the growth of high-quality β-Ga2O3 epitaxial films with mobility values closer to the theoretical limit (~ 200 cm2/Vs). In the first part of this work, we report on the growth of high-mobility β-Ga2O3 homoepitaxial thin films grown at a temperature more than 200°C less than the conventional growth temperature window for metalorganic vapor phase epitaxy . Low-temperature β-Ga2O3 thin films grown at 600°C on Fe-doped (010) bulk substrates exhibit remarkable crystalline quality, which is evident from the measured room temperature Hall mobility of 186 cm2/Vs for the unintentionally doped films. N-type doping is achieved by using Si as a dopant, and a controllable doping in the range of 2×1016–2×1019cm-3 is studied. Si incorporation and activation is studied by comparing the silicon concentration from secondary ion mass spectroscopy and the electron concentration from temperature-dependent Hall measurements. The films exhibit high purity (low C and H concentrations) with a very low concentration of compensating acceptors (~2×1015cm-3) even at this growth temperature. Additionally, an abrupt doping profile with a forward decay of ~5 nm/dec (10 times improvement compared to what is observed for thin films grown at 810°C) is demonstrated by growing at a lower temperature due to suppression of Si dopant segregation .
In the second part of this work, we use this low-temperature growth regime to achieve degenerately n-type Si-doped (>1020cm-3) β-Ga2O3 layers to realize high-quality low resistance Ohmic contacts. We perform selective area regrowth of heavily-doped n+ β-Ga2O3 Ohmic contacts to lightly doped channel layers using a novel Ni/SiO2 mask which is patterned by a combination of dry and wet etching technique. The Ohmic metal stack of Ti/Au/Ni (20nm/100nm/30nm) was selectively evaporated using photolithography patterning and lift off process. The doping in the n+ β-Ga2O3 layers was estimated to be ~2×1020 cm-3. TLM measurements on the n+ β-Ga2O3 layers show Rsh, Rc and ρc values to be 45 Ω/sq, 0.08 Ω.mm and 8.2 ×10-7 Ω.cm2 respectively. These results indicate that these MOVPE-grown n+ β-Ga2O3 layers were able to maintain good material quality even in this doping regime and is comparable to or better than that reported by other techniques such as MBE and PLD, . Fully MOVPE-grown FET performance will be presented. The growth of high mobility films with an expanded growth window, MOVPE technique for Ga2O3 growths can now be considered more versatile. The ability to perform selective area regrowth of ohmic contacts expands the capability of the MOVPE technique in terms of device processing as well. This initial result of the low temperature growth and selective area regrowth using MOVPE shows the promise of this approach to realize high-quality channel layers as well as low resistance contacts for several lateral and vertical device topology.
ACKNOWLEDGEMENT: This work was supported by the Air Force Office of Scientific Research under Award No. FA9550-18-1-0507 (Program Manager: Dr. Ali Sayir). We also acknowledge the II–VI foundation Block Gift Program for financial support. This work was performed in part at the Utah Nanofab sponsored by the College of Engineering and the Office of the Vice President for Research.
 A. Bhattacharyya et al." Low-temperature homoepitaxy of (010) β-Ga2O3 by metalorganic vapor phase epitaxy: Expanding the growth window", Appl. Phys. Lett. 117, 142102 (2020)
 P. Ranga et al. “Delta-doped β-Ga2O3 films with narrow FWHM grown by metalorganic vapor-phase epitaxy,” Appl. Phys. Lett. 117, 172105 (2020)
 K. D. Leedy et al., Appl. Phys. Lett. 111, 012103 (2017)
 Z. Xia et al., IEEE Electron Device Letters, vol. 39, no. 4, pp. 568-571, April 2018
2:10 PM - EL04.07.05
MOCVD Epitaxy of β-(AlxGa1−x)2O3 Films on (100) and (-201) β-Ga2O3 Substrates with Al Compositions up to 52%
A F M Anhar Uddin Bhuiyan1,Zixuan Feng1,Jared Johnson1,Hsien-Lien Huang1,Jinwoo Hwang1,Hongping Zhao1
The Ohio State University1Show Abstract
β-Ga2O3 has gained a remarkable attention in high power electronic applications due to its higher bandgap energy (~ 4.85 eV), predicted high breakdown field strength (~ 8 MV/cm), and availability of high-quality native substrates with different orientations. β-(AlxGa1−x)2O3 alloys, due to its bandgap tuning capability up to ~8.8 eV, can take advantage of its higher breakdown voltage in not only vertical power devices but also in high-performance lateral devices through device scaling. Recently, modulation doped field effect transistors have been demonstrated with promising transport properties by forming 2-dimentional electron gas in β-AlGaO/GaO heterostructures with limited Al composition in β-AlGaO layer (x < 20%). High quality β-AlGaO epitaxy with higher Al compositions is required to form heterostructures with large band offset and thus high 2D electron density. While theoretical studies have predicted that the solubility limit of Al2O3 in β-Ga2O3 is as high as ~70%, very limited Al incorporation in (010) β-(AlxGa1−x)2O3 films has been observed experimentally (x < 27%), as targeting for higher Al composition resulted in phase segregation .
In this work, for the first time, we have studied the growth of β-(AlxGa1−x)2O3 thin films on (100) and (-201) oriented β-Ga2O3 substrates via metalorganic chemical vapor deposition (MOCVD). Trimethylaluminum (TMAl), Triethylgallium (TEGa), and pure O2 were used as Al, Ga and O precursors, respectively. Argon (Ar) was used as the carrier gas. By the systematic tuning of TEGa/TMAl molar flow ratio, pure β-phase (AlxGa1−x)2O3 films with Al compositions up to 52% were achieved on (100) β-Ga2O3 substrates . X-ray diffraction (XRD) and X-ray spectroscopy (XPS) were used for determining the bandgaps and the Al compositions. Two-dimensional twin boundary defects in the β-(AlxGa1−x)2O3 films with different Al compositions were investigated by utilizing atomic resolution scanning transmission electron microscopy (STEM) imaging. Coherent growth of high quality (100) β-AlGaO/GaO superlattice (SL) structures with abrupt interfaces and uniform Al distribution were observed with Al compositions as high as 50%. Step flow growth with lower RMS roughness values (< 1.2 nm) were observed for higher Al composition samples. A mechanism was proposed for the step-flow growth of high-Al content films by considering Al adatoms as preferred nucleation sites for (AlxGa1−x)2O3 growths. The growth of β-(AlxGa1−x)2O3 films on (-201) oriented β-Ga2O3 substrates also revealed higher Al incorporation in pure β-phase (x ≤ 48%) without causing phase segregations . The bandgap energies for (-201) β-(AlxGa1−x)2O3 films extracted from XPS spectra ranged between 5.20 ± 0.06 eV (x = 21%) and 5.72 ± 0.08 eV (x = 48%). The surface morphologies showed elongated features with granules along  direction, which were suppressed with the increasing Al content. The influence of the chamber pressure and temperature on Al incorporations and the surface morphologies were studied.
In summary, we have demonstrated MOCVD growth of (100) and (-201) β-(AlxGa1-x)2O3 films with high-Al compositions. Results from this work provide great promises for future device technologies based on this emerging ultrawide band gap semiconductor material system.
Acknowledgment: The authors acknowledge the funding support from the Air Force Office of Scientific Research No. FA9550-18-1-0479 (AFOSR, Dr. Ali Sayir) and the National Science Foundation (Grant No. 1810041, No. 2019753).
Bhuiyan et al., APL Mater. 8, 031104 (2020).
Bhuiyan et al., Cryst. Growth Des. 20, 6722 (2020).
Bhuiyan et al., Appl. Phys. Lett. 117, 142107 (2020).
2:25 PM - EL04.07.06
Carrier Density in Ga2O3 Measured by Infrared Absorption
Etienne Gheeraert1,Ernesto Gribaudo1,Toshimitsu Ito2
Université Grenoble Alpes1,AIST2Show Abstract
Gallium oxide Ga2O3 is a very promising material for the next generation wide bandgap semiconductor devices. Even if impressive electronic devices have already been fabricated, demonstrating the high potential of this new semiconductor, the knowledge about its electronic properties is still weak. The purpose of this work is to explore dopants and carriers by infrared absoorption in gallium oxide. In regular semiconductor, such technique allows to identify at low temperature the ground level of dopants as shallow level centers, their excited states and photoionisation. At high temperature, when all the dopants are ionized, absorption by free carriers can by observed. All these information about the semiconductor electronic properties make infrared absorption a powerful technique.
In this work β-Ga2O3 crystals were grown by the floating-zone (FZ) method using no crucibles. By the zone-refining that were executed by repeating the FZ procedure, very higher purity compared to that of regular growth method was obtained. Crystal were doped with Si with a concentration estimated from SIMS measurements from 2x1017 cm-3 to 1x1019 cm-3. Infrared was recorded in transmission from 5K to 300K.
All the spectra exhibit a strong absorption continuum from 0.2 eV, with no mesurable transmission below this value, and extending beyond 1.2 eV for high doping values. No absorption peak, that could have been assigned to a transition from the dopant ground level to an excited state was observed, whatever the temperature and the doping level.
The absorption intensity in the continuum, at 0.5 eV, is clearely correlated to the Si concentration. A linear relation is proposed, allowing the measure of the Si concentration in Ga2O3 in a non destructive way. This continuum is assigned to free electron infrared absoprtion.
The fact that no absorption peak is observed, even at 5K, and that the free absorption is observed whatever the temperature, suggest that silicon is ionized without any ionization energy. This could be attributed to an impurity band formed by very shalow silicon centers.
2:40 PM - EL04.07.07
Late News: Ge Doping of Epitaxial β-Ga2O3 Films by MOCVD
Fikadu Alema1,George Seryogin1,Alexei Osinsky1,Andrei Osinsky1
Agnitron Technology Incorporated1Show Abstract
We report on Ge doping of epitaxial Ga2O3 films by MOCVD using germane (GeH4) balanced in nitrogen as a Ge source. The incorporation efficiency of Ge into Ga2O3 films grown using triethylgallium (TEGa) and trimethylgallium (TMGa) sources was studied by varying substrate temperature, GeH4 flow rate, growth rate, and oxygen flow rates using SIMS. As expected, Ge incorporation increased with GeH4 flow rate, but it incorporated well into films grown using TEGa than that grown using TMGa despite similar growth conditions. Hall Effect measurement and SIMS showed a strong dependence of Ge incorporation on the growth temperature. By decreasing the substrate temperature over a range of ~200 oC, the film's Ge concentration has increased by two orders of magnitude. The Hall Effect measurement also showed Ge doped films with RT electron mobilities ranging from ~40 to ~150 cm2/Vs with the corresponding carrier concentration of n=3x1019 to n=2x1016 cm-3 for Ge doped films. Temperature-dependent Hall Effect measurement will be discussed to investigate the donor activation energy and compensation concentration in representative TMGa and TEGa grown films. The results will be compared with similar films doped by Si. In this talk, the effect of Ge doping on the surface roughness and the incorporation of carbon and hydrogen impurities into the films will be discussed.
EL04.08: Ga2O3 II
Monday PM, April 19, 2021
4:00 PM - *EL04.08.01
Electro-Thermal Co-Design of a Gallium Oxide MODFET
The Pennsylvania State University1Show Abstract
To extend further the electrical performance envelope of wide bandgap (WBG) electronics based on gallium nitride (GaN) and silicon carbide (SiC), novel device structures based on ultra-wide bandgap (UWBG) semiconductors such as β-phase gallium oxide (Ga2O3) are being actively developed. While UWBG electronics target for higher power handling capabilities and smaller device footprints, the thermal conductivity of Ga2O3 is an order of magnitude lower than those for GaN and SiC. Therefore, device self-heating has become a major challenge to accomplish the successful transition from WBG electronics to the UWBG device technology. In this talk, the electro-thermal co-design process for a (AlxGa1-x)2O3/Ga2O3 modulation-doped field-effect transistor (MODFET) will be demonstrated. First, the steady-state and transient thermal response of an operational MODFET is characterized using a recently developed 2D material-assisted Raman thermography technique. Second, a multi-physics device modeling scheme is used to reproduce the temperature-dependent electrical output characteristics as well as the self-heating behavior in response to the electrical inputs. Finally, this coupled electro-thermal model is used to design a device-level thermal management solution, taking advantage of a low thermal resistance composite substrate.
4:25 PM - EL04.08.02
Thermal Annealing of a-Ga2O3 on the Millisecond Time Scale and the Observation of γ-Phase Inclusions
Katie Gann1,Ming-Chiang Chang1,Aine Connolly1,Duncan Sutherland1,Maximillian Amsler1,R. Bruce van Dover1,Michael Thompson1
Cornell University1Show Abstract
β-Ga2O3 has shown significant promise as a wide bandgap, high-breakdown field semiconductor with extensive device applications. While the β-phase is the thermodynamically stable phase, there are a variety of polymorphs in the system, including the hexagonal α-phase with the widest bandgap, a defect spinel γ-phase, and an -phase with predicted piezoelectric properties. These alternate polymorphs, while metastable, readily form under various processing conditions. Laser spike annealing (LSA), with heating times from 150 μs to 10 ms, has been shown to be effective in kinetically trapping metastable phases from a deposited amorphous phase. By varying the time and temperature, a processing-phase diagram for metastable phases can be rapidly developed using high throughput characterization methods.
The metastable phase formation sequence from amorphous Ga2O3, on a non-crystalline neutral substrate, was determined for annealing (dwell) times from 250 μs to 10 ms (τdwell), and peak annealing temperatures (Tpeak) between ambient and 1400 oC. Amorphous Ga2O3 films, 170 nm thick, were sputter deposited from a Ga2O3 target in an Ar(90%)/O2(10%) ambient onto 100 mm oxidized (20 nm thermal SiO2) Si wafers. Samples were annealed with 617 time/temperature conditions to fully characterize the time and temperature phase space using a CO2 laser spike annealing system. The resulting thermally induced structural transformations were characterized with a variety of techniques, including optical imaging, reflectance spectroscopy, X-ray diffraction, and transmission electron microscopy. For temperatures below ~650 oC, no transformations were observed. Between 650 and 800-850 oC, and for all dwell times explored, the spinel γ-phase was observed in the quenched samples. At higher temperatures above 850-950 oC, the equilibrium β-phase was observed in the quenched samples. No other phases were observed under any processing conditions. TEM micrographs show that the γ-phase nucleated near the center of the deposited film homogenously with minimal heterogenous nucleation at either the SiO2 interface or surface. We postulate that this phase is always the first to nucleate, with β-phase only nucleating heterogeneously off of the γ-phase at higher temperatures. This is supported by observations of extremely large grains of the β-phase at short annealing times, suggesting nucleation is limited with significant grain growth during the subsequent heating and cooling times of the thermal process. The low surface energy implied by the homogenous nucleation also suggests a possible explanation for the widely observed γ-phase inclusions observed in MBE grown β- and α-phase films.
4:55 PM - EL04.08.04
Thermal Transport Across Metal/β-Ga2O3 Interfaces
Jingjing Shi1,Chao Yuan1,Shangkun Wang1,Riley Hanus1,Zhe Cheng1,2,Samuel Graham1
Georgia Institute of Technology1,University of Illinois at Urbana-Champaign2Show Abstract
β-Ga2O3 is a very promising material to be applied in power and radio frequency devices because of its exceptional properties. However, the heat dissipation of its devices will be limited by its ultra-low thermal conductivity. Previous study showed that its device could achieve high power density with double-side cooling strategy and thin Ga2O3 layer. Therefore, the thermal transport across β-Ga2O3 interfaces becomes very important because the boundary resistance would be a main source of the total device resistance. In this work, we study the thermal transport at β-Ga2O3/metal interfaces which play important roles in heat dissipation and as electrical contacts in β-Ga2O3 devices. A theoretical Landauer approach was used to model and elucidate the factors that impact the thermal transport at these interfaces. Experimental measurements using time-domain thermoreflectance (TDTR) provided data for the thermal boundary conductance (TBC) between β-Ga2O3 and a range of metals used to create both Schottky and ohmic contacts. From the modeling and experiments, the relation between metal cut-off frequency and the corresponding TBC is observed. Moreover, the effect of metal cut-off frequency on TBC is seen as the most significant factor followed by chemical reactions between the metal and the β-Ga2O3. Among all metal/Ga2O3 interfaces, for Schottky contacts, Ni/Ga2O3 interfaces show the highest TBC, while for ohmic contacts, Cr/Ga2O3 interfaces show the highest TBC. While there is a clear correlation between TBC and the phonon cutoff frequency of metal contacts, it is also important to control the chemical reactions at interfaces in order to maximize the TBC in this system.
5:10 PM - EL04.08.05
Computational Fermi Level Engineering and Doping-Type Conversion of Ga2O3 via Three-Step Processing
Anuj Goyal1,Andriy Zakutayev1,Vladan Stevanovic2,Stephan Lany1
National Renewable Energy Laboratory1,Colorado School of Mines2Show Abstract
Ga2O3 is being actively explored for power electronics, deep-ultraviolet optoelectronics, and other applications due to its ultra-wide bandgap and low projected fabrication cost of large-size and high-quality crystals. N-type doping of Ga2O3 can be achieved and tuned, but p-type doping faces fundamental obstacles due to deep character of acceptor levels and polaron transport of resulting holes. However, successful engineering of Ga2O3 based devices requires critical control of doping density, Fermi level position, and free carrier concentration, providing opportunities for predictive process simulation. We use first-principles defect theory and defect equilibrium calculations to simulate a 3-step growth-annealing-quench protocol for hydrogen assisted Mg doping in Ga2O3, taking into account the hydrogen-oxygen-water gas phase equilibrium. We predict type conversion to a net p-type regime following O-rich annealing after growth under reducing conditions in the presence of H2. This process is similar to the Mg acceptor activation by H removal in GaN. We show that there is an optimal temperature that maximizes the net acceptor density during the equilibrium annealing step for a given Mg doping level. Quenching of non-equilibrium annealed samples then results in a Fermi level EF below mid-gap down to about EV +1.5 eV, creating a significant number of uncompensated neutral MgGa0 acceptors. The resulting free hole concentration in quenched samples is very low (109 – 1012 cm-3) due to deep energy level of these Mg acceptors, but this type converted Ga2O3 material can create a significant built-in field in a p-n junction with an adjoining n-type material. Additionally, the electron concentration is greatly suppressed in such acceptor-doped Ga2O3, which could enable the use as a current blocking layer to fabricate normally-off (enhancement-mode) vertical Ga2O3 based metal-oxide-semiconductor field effect transistors (MOSFETs), and as a guard ring for edge termination in MOSFETs and Schottky barrier diodes (SBDs) with increased breakdown voltage.
5:25 PM - *EL04.08.06
Thermal and Electrical Properties of Ga2O3 FETs and Diodes
University of Bristol1Show Abstract
Ga2O3 offers exciting new opportunities for power electronics with its ultra-wideband gap; numerous materials related challenges such as its low thermal conductivity, deep level traps and also the challenge to achieve “good” p-doping remain, however, these need to be resolved to be able to exploit Ga2O3’s full potential. I will review our latest results in this field, amongst them device temperature measurements and cooling mitigation strategies, trap generation during device stress also nitrogen implantation induced trap states, superjunction designs.
EL04.09: Ga2O3 III
Tuesday AM, April 20, 2021
8:00 PM - EL04.09.01
Late News: Prospects for Donor-Doping AlGO Alloys
John Lyons2,Joel Varley1,Darshana Wickramaratne2
Lawrence Livermore National Laboratory1,U.S. Naval Research Laboratory2Show Abstract
Gallium oxide has emerged as a promising ultrawide-bandgap semiconductor partly due to its ability to be n-type doped. Alloying Ga2O3 with Al2O3 can further widen its band gap, potentially enabling novel device designs. But this alloying quickly leads to an increase in the position of the conduction-band minimum, and it is not clear whether (AlxGa1-x)2O3 (“AlGO”) alloys will also be n-type dopable. Here we systematically explore the properties of group-IV (C, Si, Ge, and Sn) and transition metal (Hf, Zr, and Ta) substitutional dopants in AlGO alloys using first-principles hybrid functional calculations. In Ga2O3, all of these dopants act as shallow donors, but in Al2O3 they are deep defects characterized by the formation of either DX centers or positive-U (+/0) levels. Combining our calculations of dopant charge-state transition levels with information of the AlGO alloy band structure, we estimate the critical Al composition at which each dopant transitions from being a shallow to a deep donor. We identify Si as being the most efficient dopant to achieve n-type conductivity in high Al-content AlGO alloys, acting as a shallow donor over the entire predicted stability range for AlGO solid solutions.
This work was partially performed under the auspices of the U.S. DOE by the Lawrence Livermore National Laboratory under Contract No. DE-AC52-07NA27344 and supported by the Critical Materials Institute, an Energy Innovation Hub funded by the U.S. DOE, Office of Energy Efficiency and Renewable Energy, Advanced Manufacturing Office. JLL and DW were supported by the Office of Naval Research through the Naval Research Laboratory’s Basic Research Program.
8:15 PM - EL04.09.02
Control of Surface Morphologies of M-Plane α-Al2O3 Homoepitaxial Films Using Plasma-Assisted Molecular Beam Epitaxy
Riena Jinno1,Hironori Okumura1
University of Tsukuba1Show Abstract
Rhombohedral α-Al2O3 (sapphire) (Eg=8.8 eV) and its alloys with α-Ga2O3 (Eg~5.3-5.6 eV) have attracted for high power electronic devices and optical devices with function in the deep ultra-violet region due to the large predicted critical electric field and wide bandgap energy . Insulating α-Al2O3 substrates have been widely used for the thin film growth of different materials because of the large-area and low-cost α-Al2O3 bulks grown by melting methods. However, the reports on homoepitaxial growth of α-Al2O3 thin films are limited . In this study, we investigated the surface morphologies of α-Al2O3 thin films, which were grown by plasma-assisted molecular beam epitaxy (PAMBE) under various growth conditions.
The aluminum-oxide thin films were grown on a m-plane α-Al2O3 substrate with a miscut angle of 2 ° toward  by PAMBE. The radio frequent power and oxygen flow rate were fixed at 150 W and 1.0 sccm, respectively. The thermocouple substrate temperature (Tsub) and Al beam equivalent pressure (BEP) were changed in the ranges of 130-830 °C and 4.0x10-6-1.8x10-5 Pa, respectively. The surface morphology was observed by atomic force microscopy.
The amorphous AlOx films were grown for Tsub<330 °C, while the single-crystal α-Al2O3 films were homoepitaxially grown for Tsub>530 °C as revealed by reflection high energy electron diffraction and x-ray diffraction. The growth rate of the α-Al2O3 films increased with increasing the Al BEP, and showed a larger slope at the Al BEP higher than 1.3x10-5 Pa. The growth rate was independent on the substrate temperature between 130 and 830 °C. The α-Al2O3 surfaces exhibited root mean square (RMS) roughness smaller than 0.6 nm for the Al BEP lower than 1.3x10-5 Pa, while the α-Al2O3 surfaces suddenly increased the RMS roughness larger than 10 nm for the Al BEPs higher than 1.3x10-5 Pa. The boundary between the smooth and rough surface morphologies corresponded with the singular point where the slope of growth rate versus the Al BEP changed. We consider that the Al BEP lower than 1.3x10-5 Pa agrees with the O-rich condition and that the Al-rich condition probably contributes to the rough surface morphologies and/or nonstoichiometry of the α-Al2O3 epitaxial films. From these results, the O-rich growth condition with higher growth temperature than 530 °C is suitable for the homoepitaxial growth of α-Al2O3.
This work was supported by NEDO Feasibility Study Program Uncharted Territory Challenge 2050.
 S. Fujita, et. al., Jpn. J. Appl. Phys. 55(2016) 1202A3.
 T. Maeda, et al., J. Cryst. Growth, 177(1997) 95-101.
8:30 PM - EL04.09.03
Killer Defects in β-Ga2O3 Schottky Barrier Diodes Observed by Ultrahigh Sensitive Emission Microscopy and Synchrotron X-Ray Topography
Makoto Kasu1,Sayleap Sdoeung1,Kohei Sasaki2,Katsumi Kawasaki3,Jun Hirabayashi3,Akito Kuramata2
Saga University1,Novel Crystal Technology, Inc.2,TDK Corporation3Show Abstract
β-Ga2O3 with a bandgap of 4.5–4.8 eV is expected to serve as a high-power semiconductor surpassing the capabilities of SiC and GaN. Recently, we demonstrated the production of > 20 A-class β-Ga2O3 Schottky barrier diodes (SBDs) with a low on-state resistance of 6 mΩ.cm2. However, the reliability of SBD plays a crucial role in the commercialization of SBDs in power system circuits. It is especially important to identify killer defects that can cause an increase of reverse leakage current and/or a decrease in the breakdown voltage.
The sample was halide vapor phase epitaxy (HVPE)-grown ~10-μm-thick epitaxial layer on EFG-grown β-Ga2O3 (001) substrate. Pt/Ti/Au Schottky electrodes with diameters of 50–1000 μm were formed on the surface, and Ti/Au ohmic contacts were formed on the back surface. First, we observed light emission patterns from SBDs in operation using ultrahigh sensitive emission microscopy. The emission microscopy equipment consists of an ultrahigh sensitive electron-multiplying CCD camera and a probe station enabling the observation of light emission patterns of an SBD in real time.
In the reverse bias conditions, we detected emission patterns from SBDs with a high leakage current. The patterns were observed by AFM and synchrotron X-ray topography, and their cross-sections were observed by STEM. The killer defects were found to be stacking faults and polycrystalline defects in the HVPE epitaxial layer. Their properties will be reported.
8:45 PM - EL04.09.05
Growth and Characterization of MOVPE-Grown Low Sheet Resistance β-(AlxGa1-x)2O3/β-Ga2O3 Heterostructure Channels
Praneeth Ranga1,Arkka Bhattacharyya1,Adrian Chmielewski2,Rujun Sun1,Saurav Roy1,Mike Scarpulla1,Nasim Alem2,Sriram Krishnamoorthy1
The University of Utah1,The Pennsylvania State University2Show Abstract
In this work, we study the growth and characterization of β-(AlxGa1-x)2O3/ β-Ga2O3 heterostructure channels with record low sheet resistance. In the past few years, β-Ga2O3 has emerged as a promising material for next-generation power electronics applications. The high bandgap of β-Ga2O3 leads to a very high predicted breakdown field of 6 – 8 MV/cm, larger than that of GaN and SiC. However, room temperature mobility of low-doped β-Ga2O3 is still limited to 200 cm2/V.s due to severe polar optical phonon scattering. Ab-initio calculations show that mobility of modulation-doped β-(AlxGa1-x)2O3/ β-Ga2O3 channels could exceed that of uniformly-doped β-Ga2O3 layers at charge densities nearing 5 x 1012 cm-2 . This is attributed to enhanced screening of LO phonons modes at high 2DEG sheet charge densities. Currently, the sheet charge density of single β-(AlxGa1-x)2O3/ β-Ga2O3 heterostructure is limited to ~5 x 1012 cm-2 , which is in turn limited by the Al composition of the MBE-grown β-(AlxGa1-x)2O3. To achieve such high sheet charge densities, a high Al content β-(AlxGa1-x)2O3 barrier along with a sharp delta sheet is necessary. Recently, high-quality β-Ga2O3 films with mobilities close to 200 cm2/V.s have been realized MOVPE. Growth of (100) β-(AlxGa1-x)2O3 layers up to 52% has been realized using the MOVPE technique . In addition, n-type doping and delta doping were also achieved [4,5]. However, the FWHM of the Si delta sheet was found to be larger than that of MBE-grown β-Ga2O3. By optimizing the growth conditions MOVPE-grown β-(AlxGa1-x)2O3/ β-Ga2O3 heterostructures with high charge and mobility can be potentially attained.
Growth of delta-doped (010) β-Ga2O3 films is performed by Agnitron Agilis MOVPE reactor with TEGa, O2 and silane (SiH4) as precursors and argon as a carrier gas. Delta doping of β-Ga2O3 is achieved by interrupting the growth of β-Ga2O3 and supplying silane to the reactor. Multiple samples are grown under varying growth temperatures to characterize the spread of silicon donors β-Ga2O3. CV measurements are used to characterize the sheet charge density and FWHM of the silicon delta sheet. SIMS characterization revealed that surface segregation is the key mechanism, which leads to large FWHM Si delta sheets. By reducing the growth temperature to 600 C, CV measured FWHM of 3.2 nm is achieved, which is close to the FWHM of an ideal delta sheet. Similar growth conditions are utilized to realize a sharp delta sheet in β-(AlxGa1-x)2O3/ β-Ga2O3 heterostructure. Two delta-doped β-(AlxGa1-x)2O3/ β-Ga2O3 heterostructure are grown with different β-(AlxGa1-x)2O3 spacer thicknesses (2- 4 nm) under identical growth conditions. MOVPE based low-temperature n+ regrowth process is utilized to achieve ohmic contacts to the channel. Hall measurements showed a sheet charge density of 6.4 x 1012 to 1 x 1013 cm-2 and mobility of 125 – 111 cm2/V.s. FET devices fabricated using heterostructure channels show a peak current of 22 mA/mm and transconductance of 7 mS/mm. Sheet resistance of 5.3 kΩ/square is realized at room temperature, which is the lowest value for a single heterostructure in current literature. CV and TLM characterization revealed similar charge density and sheet resistance values, confirming the properties of the heterostructure channel. These results show that high-quality β-(AlxGa1-x)2O3/ β-Ga2O3 heterostructures with mobility exceeding 100 cm2/Vs can be realized using MOVPE technique.
References:  Kumar.A et.al Journal of Applied Physics 128.10 (2020): 105703.  Kalarickal N.K et.al Journal of Applied Physics 127.21 (2020): 215706.  Bhuiyan AFM et.al Crystal Growth & Design 20, 6722-6730 (2020).  Ranga.P et al. Applied Physics Express 13.4 (2020): 045501.  Ranga.P et al. Applied Physics Express 12.11 (2019): 111004. Acknowledgments: This material is based upon work supported by the Air Force Office of Scientific Research under Award No. FA9550-18-1-0507 and monitored by Dr. Ali Sayir.
9:15 PM - EL04.09.06
Structural Analysis of Ion Implanted and Exfoliated (010) and (-201) β-Ga2O3
Michael Liao1,Yekan Wang1,Kenny Huynh1,Fengwen Mu2,Tiangui You3,Wenhui Xu3,Zhe Cheng4,Jingjing Shi4,Tadatomo Suga5,Xin Ou3,Samuel Graham4,Mark Goorsky1
University of California, Los Angeles1,Institute of Microelectronics of Chinese Academy of Science2,Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences3,Georgia Institute of Technology4,Meisei University5Show Abstract
The evolution of defects in both He+ and H+ ion implanted β-Ga2O3 for exfoliation was assessed using triple-axis x-ray diffraction and transmission electron microscopy. First, (010) and (201) substrates were He+ ion implanted simultaneously at -20 °C with an ion energy of 160 keV and a dose of 5 × 1016 cm-2. Strain profiles for both implanted orientations were obtained from modeling the symmetric X-ray diffraction curves with dynamical simulations and the differences are attributed to the structural anisotropy of β-Ga2O3. After annealing at 200 °C to initiate He blister nucleation followed by 500 °C for up to 96 hours to induce He blister growth, the crack formation at the ion projected range beneath the surfaces also differed between the (010) and (201) implanted substrates. The annealed (010) substrates showed large continuous cracks at the projected range relatively parallel to the substrate surface. These cracks correspond to blistering directly above at the surface – a clear sign of exfoliation. In contrast, while the (201) annealed substrate showed disjointed cracks at the projected range, which would also be compatible with exfoliation, did not exhibit surface blistering. Furthermore, the cracks in this substrate were tilted at a series of angles ranging from 0° to 54° from the surface towards the (201) planes. Eighty percent of the cracks observed were tilted 54° – which is parallel to the (100) plane, a primary cleavage plane in β-Ga2O3. Note that the (100) is 90° from (010) and angled cracks were not observed in a previous effort1 nor in this work. This directional dependence of the crack formation in the (201) substrate is a consequence of the anisotropic properties of the monoclinic structure, suggesting the (100) cleavage plane plays an important role in crack propagation for exfoliation. After annealing at 500 °C for 96 hours, the strain was removed from the (010) substrates while the (201) substrates still exhibited residual strain. In a second set of samples that used H+ implantation at 35 keV, residual strain was also observed in annealed and exfoliated (201) β-Ga2O3 layers. These H+ implanted (201) substrates were exfoliated and bonded to (0001) 4H-SiC substrates. Exfoliation due to blister growth of the implanted ions with annealing typically corresponds to the removal of the implant-induced strain. However, as was the case for the He+ implant, diffraction measurements showed the presence of strain in the (201) β-Ga2O3 layers even after exfoliation that was only removed after further annealing. The analysis of the evolution of this strain is important because strain plays a significant role in important characteristics such as electrical and thermal transport. Subsequent annealing at 800 °C for 30 mins (with a ramp up rate of 5 °C/min), resulted in relieving this residual strain. Corresponding thermal measurements showed that the thermal conductivity of these β-Ga2O3 improved after annealing and removing the residual strain.
1. M.E. Liao, et al., ECS J. of Solid State Sci. and Technol., 8(11), P673 (2019).
The authors M.E.L., Y.W., K.H., Z.C., J.S., S.G., and M.S.G. would like to acknowledge the support from the Office of Naval Research through a MURI program, grant No. N00014-18-1-2429.
9:30 PM - EL04.09.07
Mg Acceptor Level in Ga2O3 as Studied by Photo-Induced Electron Paramagnetic Resonance
Mary Zvanut1,Suman Bhandari1
The University of Alabama at Birmingham1Show Abstract
Gallium oxide is a wide band-gap (4.6-4.9 eV) semiconductor that has potential for power electronics. Doping Ga2O3 with Mg makes it semi-insulating material, which can be an integral part of power devices. For such applications, knowledge of the Mg-related defect level is essential. Theory predicts a Mg acceptor level (Mg-/0) ~1.2 eV above valence band maximum (VBM) whereas thermal measurements using electron paramagnetic resonance (EPR) place the level ~0.65 eV above VBM [1, 2]. To address the disagreement between the reported defect levels, we investigate optical absorption of neutral Mg acceptors (Mg0) using photo-induced EPR. In this approach, we consider relaxation of the defect during optical excitation and determine the defect level and associated relaxation energy. Previously, using a similar technique, we have determined the Fe2+/3+ level and associated relaxation energy in Ga2O3 . In this work, by examining two different Mg-doped Ga2O3 samples, which were grown by the Czochralski method at two different institutions, we show that the presence of other defects could affect the experimental results and possibly lead to misinterpretation. The crystals are irradiated with energies between 0.6 and 4.7 eV after Mg0 is generated with an LED (4.4±0.2 eV), and the effect on the amount of Mg0 is investigated at 130 K. The steady state EPR results show that, for one set of samples, the amount of Mg0 starts to decrease near 1.7 eV whereas for the other set of samples, the decrease begins at approximately 1 eV. The decrease in the amount of Mg0 indicates that Mg0 becomes Mg- by either of the two processes: when electrons are excited 1) from valence band to Mg0 or 2) from other defects to conduction band and subsequently captured by Mg0. For samples with photo-threshold near 1.7 eV, no new defects appear nor do other existing defects change when the decrease of Mg0 is initiated. Therefore, we disregard the second process for these samples. Furthermore, a preliminary analysis of the optical cross section spectrum obtained from photo-EPR yields values for the defect level and relaxation energy that agree reasonably well with the predicted defect level for Mg-/0 when a relaxation energy of ~1 eV is considered. On the other hand, for the set of samples with the 1 eV photo-threshold, an unknown defect ‘X’ increases when Mg0 starts decreasing, suggesting that the defect ‘X’ could be responsible for the decrease of Mg0. We suggest that the 1 eV threshold could represent a defect level for ‘X’, which lies approximately 1 eV below conduction band minimum. In the talk, we will show the analysis of the optical cross section spectrum for both sets of samples and discuss ambiguities caused by the presence of other defects. In addition, we will present a detailed analysis of the optical cross section spectrum to supplement our preliminary results suggesting that the ~1.7 eV photo-threshold likely represents the Mg-/0 level accompanied by a large relaxation energy.
NSF Grant: DMR-1904325 supports the work at UAB. We would like to acknowledge Jacob Leach, Kyma Inc. and Kevin Stevens, NG Synoptics, and Matthew D. McCluskey, Washington State University for the Mg-doped Ga2O3 samples.
1. J. L. Lyons, Semicond. Sci. Technol. 33 (2018) 05LT02 (5pp)
2. Lenyk et al., Appl. Phys. Lett. 116, 142101 (2020)
3. Bhandari et al., J. Appl. Phys. 126, 165703 (2019)
10:00 PM - EL04.09.09
Recessed Gate Enhancement-Mode Ultrawide Bandgap AlxGa1-xN Channel MOSHFET with Drain Current 0.48 A/mm and Threshold Voltage +3.6 V
Shahab Mollah1,Kamal Hussain1,Abdullah Mamun1,Mikhail Gaevski1,Grigory Simin1,MVS Chandrashekhar1,Asif Khan1
University of South Carolina1Show Abstract
For ultra-wide bandgap (UWBG) devices to truly exploit their potential for power electronics applications, it is essential to develop enhancement mode (normally-off) devices. In the past, normally off devices have been made using p-gate, fluorine treatment and recessed gate techniques. Recently, enhancement mode devices were reported by Douglas et al using p-AlGaN gate  and by Klein et al using fluorine treatment . In the past, we have shown that the performance of normally-on UWBG AlGaN HFETs can be significantly improved using insulating gate MOSHFET design with Al2O3 and ZrO2 gate dielectrics . Here we use stack of these two dielectric materials to make high-current recessed gate enhancement mode UWBG Al0.60Ga0.40N/Al0.40Ga0.60N MOSHFETs.
The epitaxial layers for the devices were grown by low pressure metal organic chemical vapor deposition on AlN (3 µm)/sapphire template on which a graded composition back-barrier AlxGa1-xN layer (x from 1 to 0.4) was grown. The back-barrier design enables a reduction in leakage currents by screening the substrate-epilayer growth interface. It also leads to a tighter confinement of the 2-DEG which improves the ON-OFF ratios, drain-currents, and the sub-threshold swing factor. It was followed by the Al0.40Ga0.60N channel layer of 185 nm thickness. On top of this a 170 Å thick n- Al0.60Ga0.40N barrier layer was grown which was followed by a 200 Å highly doped composition graded AlxGa1-xN layer with x-value changing from 0.60 to 0.3 towards the surface to facilitate the ohmic contact. The n-doping of this layer compensates the positive charges resulting from the reverse composition grading. The 2DEG sheet resistance was ~1900 Ω/square. Annealed source-drain ohmic contact metallization (Zr/Al/Mo/Au) yielded contact resistance of 1.7 Ω-mm. Gate recess etching through the top barrier, resulted in the positive shift of the threshold voltage by ≈ 12 V compared to devices without gate-recess (D-mode). Then a 25 nm thick ZrO2-Al2O3 insulator stack was deposited in the recess region using Atomic Layer Deposition (ALD) to suppress leakage gate current and to enable positive gate bias operation. Gate electrode was made using Ni/Au metallization. The device exhibited a threshold-voltage (VTH) of +3.6 V and drain source saturation current (IDS) as high as 480 mA/mm at a gate voltage VGS = +12 V. The peak extrinsic transconductance in the Al2O3-ZrO2-MOSHFET is GM ≈ 70 mS/mm.
To determine the factors leading to the high drain current we extracted the gate voltage dependencies of electron sheet density NS and electron mobility µ. The extracted sheet carrier density is NS ≈ 1.6 ×1013 cm-2. µ is as high as 1600 cm2/(V.s) at low gate voltage near threshold VG=2V and decreases down to 200 cm2/(V.s) at VG +12V. The on-state field effect mobility is comparable to our depletion mode devices . The E-mode MOSHFET shows a hysteresis of 0.5 V between forward and reverse sweep of gate voltage, higher than that of D-mode MOSHFET (0.2 V) indicating higher trap charges at semiconductor/oxide interface or bulk. For our E-mode MOSHFET, we estimated a sub-threshold slope (SS) value of 128 mV/decade and an ON/OFF ratio of more than 1.5x108, while the SS value for D-mode device is ~100 mV/decade. This slight increase of SS is consistent with the increased hysteresis in the transfer curve. The gate leakage currents at VG = -20V was 30 pA indicates the high quality of the oxide layer. Our studies suggest that UWBG AlGaN MOSHFETs are very promising candidates for next generation power electronic devices.
 E. A. Douglas et al., J. Vac. Sci and Technol. B, 37, 021208( 2019).
 B. A. Klein et al., Appl Phys. Lett., 114, 112104(2019).
 S. Mollah et al., phys. status solidi (a) 217, 7, 1900802 (2020).
 S. Mollah et al., Semicond. Sci. Technol., 34,125001( 2019).