Symposium EL04—Ultrawide Bandgap Materials, Devices and Systems

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 [1]. 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.
[1] A. M. Siladie et al, Nano Lett. 2019, 19, 8357−8364

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.

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 V_III-nSi complexes and the resulting strain development in Si-doped Al_0.7Ga_0.3N.
Si-doped Al_0.7Ga_0.3N layers were grown on unintentionally doped Al_0.7Ga_0.3N 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 V_III-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 V_III-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 V_III-related defects decreases, resulting in the increase in the V_III-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 V_III-related defects are expected to be involved in surface mediated dislocation climb. The increase in the [V_III-nSi] was also confirmed from an increase in the V_III-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.

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) [1]. 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 [2], 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.

References
[1] A. Minj, M. Zhao, B. Bakeroot, and K. Paredis, Appl. Phys. Lett. 118, 032104 (2021)
[2] T. Hantschel, M. Tsigkourakos, L. Zha, T. Nuytten, K. Paredis, B. Majeed, and W. Vandervorst, Microelectron. Eng. 159, 46 (2016)

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 [1]. 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 [4].

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 [5]. 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.

References
[1] T. Langer et al., Appl. Phys. Lett. 103, 202106 (2013).
[2] S. Chichibu et al., Jpn. J. Appl. Phys. 59, 020501 (2019).
[3] W. Liu et al., Appl. Phys. Lett. 109, 042101 (2016).
[4] S. Meuret et al., ACS Photonics 3, 1157 (2016).
[5] S. Finot et al., Appl. Phys. Lett. 117, 221105 (2020).

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.

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 [1]. 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.

[1] 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.

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.

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 Sc_xAl_1-xN alloys. ^[1,2] However, most studies of the fundamental properties Sc_xAl_1-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 Sc_xAl_1-xN.

In this work, ^[3] we report on the epitaxial growth studies of Sc_xAl_1-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 (d_33,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 Sc_xAl_1-xN grown on metal polar GaN.
Post-growth XRD measurements show a wurtzite 002 diffraction peak for the Sc_xAl_1-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 Sc_xAl_1-xN but decreasing crystallinity due to primary azimuthal streak broadening as Sc content increased. PFM results showed an increase in piezoelectric coefficient for Sc_xAl_1-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 Sc_xAl_1-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 Sc_xAl_1-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.

[1] Akiyama, M et al. Adv. Mater. 21, 593 (2009).
[2] Fichtner, S et al. J. Appl. Phys. 125,114103 (2019).
[3] Casamento, J et al. Appl. Phys. Lett. 117, 112101 (2020).

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 Sc_xAl_1−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.

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 H_2, 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.

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.

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 Ga_Li 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 V_Li^1- or Li_Ga^2- 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: N_O is a deep trap with amphoteric character and Zn doping leads to self-compensation between Zn_Ga and Zn_Li. 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.

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 μ_elec,⊥c = 244 cm² V^-1 s^-1, μ_elec,‖c = 377 cm² V^-1 s^-1, μ_hole,⊥c = 27 cm² V^-1 s^-1, and μ_hole,‖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.

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.

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.

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.

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. [1] 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. [2] 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. [5] 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 E_g from 4.5 eV for RS-ZnO to 7.8 eV for RS-MgO. Recently, successful growths of atomically-flat single crystalline RS-Mg_xZn_1-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 [9] 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. [10] Electronic structure calculations suggested predominate contribution of Γ-Γ direct transition for the RS-Mg_xZn_1-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 E_g 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 Mg_xZn_1-xO/MgO multi-layered structures. [11] This work was supported in part by Grants-in-Aid for Scientific Research Nos. 17H01263 and 20H00246 from MEXT, Japan.
[1] N. Yamano et al., Photochem. Photobiol. 96, 853 (2020). [2] T. Matsunaga et al., Photochem. Photobiol. 54, 403 (1991). [3] H. Hirayama, J. Appl. Phys. 97, 091101 (2005). [4] M. Kneissl and J. Rass, III–Nitride Ultraviolet Emitters (Springer International Publishing, Cham, 2016). [5] Y. Taniyasu, M. Kasu, and T. Makimoto, Nature 441, 325 (2006). [6] K. Kaneko et al., Appl. Phys. Express 9, 111102 (2016). [7] K. Kaneko et al., J. Electron. Mater. 47, 8 (2018). [8] T. Onuma et al., Appl. Phys. Lett. 113, 061903 (2018). [9] K. Ishii et al., Appl. Phys. Express 12, 052011 (2019). [10] M. Ono et al., J. Appl. Phys. 125, 225108 (2019). [11] K. Kudo et al., in Compound Semiconductor Week 2019, Nara, May 20 (2019), MoE3-5.

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 [1] and advanced epitaxial designs has enabled UWBG AlGaN transistors with cutoff frequency over 40 GHz [2], 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 [3]. We will discuss the application of this idea in BaTiO₃/AlGaN lateral diodes, where we achieved a breakdown field of 8.5 MV/cm [4] (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 [5], 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 [6]. We will then outline details of recent device demonstrations, including investigation of advanced passivation methods to reduce current collapse [7], scaled delta-doped transistors with cutoff frequency of 27 GHz [8], 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.


[1] Razzak, Towhidur, et al. Applied Physics Letters 115.4 (2019): 043502.
[2] Xue, Hao, et al., Applied Physics Express 12.6 (2019): 066502.
[3] Xia, Zhanbo, et al.," IEEE Transactions on Electron Devices 66.2 (2019): 896-900
[4] Razzak, Towhidur, et al., Applied Physics Letters 116.2 (2020): 023507
[5] Kalarickal, Nidhin Kurian, et al., Applied Physics Letters 115.15 (2019): 152106
[6] Kalarickal, Nidhin Kurian, et al., Journal of Applied Physics 127.21 (2020): 215706
[7] Joishi, Chandan, et al., IEEE Transactions on Electron Devices (2020)
[8] Xia, Zhanbo, et al., IEEE Electron Device Letters 40.7 (2019): 1052-1055.

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 E_C – 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.

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.

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 [1].
Core technologies on material, devices, quantum control technology, sensor systems, and applications for life science and energy electronics are introduced.
- Materials
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.
[1] L. Doherty et al., Phys. Rep. 528, 1 (2013).
[2] A. Hoang et al., APL 118,044001 (2021)
[3] Y. Hatano et al., APL 118, 034001 (2021)
[4] Bang Yanget et al., Physical Review Applied 14, 044049 (2020).

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 [3]. 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) [4] 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 [5]. 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.
Acknowledgments
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.
References
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).

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.

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.

[1] H. Nie et al., IEEE EDL vol. 35, no. 9, p.939-941 (2014).
[2] T. Oka et al., Appl Phys. Exp. vol. 7 no. 2, (2014).
[3] Y. Zhang et al., IEDM Tech. Dig., p. 9.2.1, (2017).
[4] A. Zubair et. al., Device Research Conference (2020).

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 [1]. 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 (C_N) in n-type GaN [2], 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) [3]. Recent study [4] 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).
References:
[1] H. Ohta, N. Asai, F. Horikiri, Y. Narita, T. Yoshida, and T. Mishima, Jpn. J. Appl. Phys. 58, SCCD03 (2019).
[2] J. L. Lyons, A. Janotti, and C. G. Van de Walle, Phys. Rev. B 89, 035204 (2014).
[3] D. D. Koleske, A. E. Wickenden, R. L. Henry, and M. E. Twigg, J. Cryst. Growth 242, 55 (2002).
[4] 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).

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.

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.

Acknowledgements
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.

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.

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).

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 [1]. 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.

[1] R. Kaplar, et al. IEDM 2020

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 [100] 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.

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 [3], diffusion doping and neutron transmutation [4]. Using synchrotron X-ray topography [5] 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 [8]. 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.
References:
[1] I.C. Kizilyalli, A.P. Edwards, H. Nie, D. Disney, D. Bour, IEEE Transactions on Electron Devices 60 (2013) 3067-3070.
[2] I.C. Kizilyalli, A.P. Edwards, H. Nie, D. Bour, T. Prunty, D. Disney, IEEE Electron Device Letters 35 (2014) 247-249.
[3] B. Li, S. Wang, M. Nami, J. Han, Journal of Crystal Growth 534 (2020) 125492.
[4] R. Barber, Q. Nguyen, J. Brockman, J. Gahl, J. Kwon, Scientific Reports 10 (2020) 1-8.
[5] 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.
[6] Y. Liu, S. Hu, H. Peng, T. Ailihumaer, B. Raghothamachar, M. Dudley, ECS Transactions 98 (2020) 21.
[7] 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.
[8] 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.

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.

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) [1]. Vertical GaN PN diode with 5kV breakdown voltage has been demonstrated [2]. Although tremendous progress has been made in the past few years, challenges still exist to achieve vertical GaN power devices with V_BR>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 (N_d-N_a) as a function of SiH₄ flow was studied. Stronger plummet of N_d-N_a was observed with reduction of SiH₄ flow when N_d-N_a 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 N_d-N_a 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 N_d-N_a 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.

References:
[1] B. J. Baliga, Fundamentals of Power Semiconductor Devices (Springer Science & Business Media, 2010).
[2] H. Ohta, N. Asai, F. Horikiri, Y. Narita, T. Yoshida, and T. Mishima, Jpn. J. Appl. Phys. 58, SCCD03 (2019).

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.

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 Mg_Ga. 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.

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-(Al_xGa_1-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-(Al_xGa_1-x)₂O₃. In addition, nanoscale modulations in the electronic structure and the band gap in the epi thin films of b-Ga₂O₃/b-(Al_xGa_1-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.

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 (N_a < 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 [2], but also detrimentally affects device performance, such as causing buffer leakage current in lateral power devices. [3] 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. [4] 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).

References:
[1] Z. Feng et al., Appl. Phys. Lett., 114, 250601 (2019).
[2] Z. Feng et al., Phys. Status Solidi RRL 14, 2000145 (2020).
[3] M. H. Wong et al., Jpn. J. Appl. Phys. 55, 1202B9 (2016).
[4] J. L. Lyons, Semicond. Sci. Technol. 33, 05LT02 (2018).

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.

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 [1]. 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 [2].

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 R_sh, R_c and ρ_c 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[3], [4]. 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.

REFERENCES:

[1] 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)
[2] P. Ranga et al. “Delta-doped β-Ga₂O₃ films with narrow FWHM grown by metalorganic vapor-phase epitaxy,” Appl. Phys. Lett. 117, 172105 (2020)
[3] K. D. Leedy et al., Appl. Phys. Lett. 111, 012103 (2017)
[4] Z. Xia et al., IEEE Electron Device Letters, vol. 39, no. 4, pp. 568-571, April 2018

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-(Al_xGa_1−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-(Al_xGa_1−x)₂O₃ films has been observed experimentally (x < 27%), as targeting for higher Al composition resulted in phase segregation [1].

In this work, for the first time, we have studied the growth of β-(Al_xGa_1−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 (Al_xGa_1−x)₂O₃ films with Al compositions up to 52% were achieved on (100) β-Ga₂O₃ substrates [2]. 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-(Al_xGa_1−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 (Al_xGa_1−x)₂O₃ growths. The growth of β-(Al_xGa_1−x)₂O₃ films on (-201) oriented β-Ga₂O₃ substrates also revealed higher Al incorporation in pure β-phase (x ≤ 48%) without causing phase segregations [3]. The bandgap energies for (-201) β-(Al_xGa_1−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 [010] 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-(Al_xGa_1-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).

References:
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).

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.

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.

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 (Al_xGa_1-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.

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 (τ_dwell), and peak annealing temperatures (T_peak) 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.

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.

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 E_F below mid-gap down to about E_V +1.5 eV, creating a significant number of uncompensated neutral Mg_Ga⁰ 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.

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.

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 (Al_xGa_1-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.

Rhombohedral α-Al₂O₃ (sapphire) (E_g=8.8 eV) and its alloys with α-Ga₂O₃ (E_g~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 [1]. 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 [2]. 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 [0001] by PAMBE. The radio frequent power and oxygen flow rate were fixed at 150 W and 1.0 sccm, respectively. The thermocouple substrate temperature (T_sub) 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 AlO_x films were grown for T_sub<330 °C, while the single-crystal α-Al₂O₃ films were homoepitaxially grown for T_sub>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.
[1] S. Fujita, et. al., Jpn. J. Appl. Phys. 55(2016) 1202A3.
[2] T. Maeda, et al., J. Cryst. Growth, 177(1997) 95-101.

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.

In this work, we study the growth and characterization of β-(Al_xGa_1-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-(Al_xGa_1-x)₂O₃/ β-Ga₂O₃ channels could exceed that of uniformly-doped β-Ga₂O₃ layers at charge densities nearing 5 x 10¹² cm^-2 [1]. This is attributed to enhanced screening of LO phonons modes at high 2DEG sheet charge densities. Currently, the sheet charge density of single β-(Al_xGa_1-x)₂O₃/ β-Ga₂O₃ heterostructure is limited to ~5 x 10¹² cm^-2 [2], which is in turn limited by the Al composition of the MBE-grown β-(Al_xGa_1-x)₂O₃. To achieve such high sheet charge densities, a high Al content β-(Al_xGa_1-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-(Al_xGa_1-x)₂O₃ layers up to 52% has been realized using the MOVPE technique [3]. 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-(Al_xGa_1-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-(Al_xGa_1-x)₂O₃/ β-Ga₂O₃ heterostructure. Two delta-doped β-(Al_xGa_1-x)₂O₃/ β-Ga₂O₃ heterostructure are grown with different β-(Al_xGa_1-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-(Al_xGa_1-x)₂O₃/ β-Ga₂O₃ heterostructures with mobility exceeding 100 cm²/Vs can be realized using MOVPE technique.
References: [1] Kumar.A et.al Journal of Applied Physics 128.10 (2020): 105703. [2] Kalarickal N.K et.al Journal of Applied Physics 127.21 (2020): 215706. [3] Bhuiyan AFM et.al Crystal Growth & Design 20, 6722-6730 (2020). [4] Ranga.P et al. Applied Physics Express 13.4 (2020): 045501. [5] 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.

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.

References:
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.

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 Fe^2+/3+ level and associated relaxation energy in Ga₂O₃ [3]. 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.

Acknowledgement:
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.

Reference:
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)

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 [1] and by Klein et al using fluorine treatment [2]. 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[3] and ZrO₂ gate dielectrics [4]. Here we use stack of these two dielectric materials to make high-current recessed gate enhancement mode UWBG Al_0.60Ga_0.40N/Al_0.40Ga_0.60N 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 Al_xGa_1-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 Al_0.40Ga_0.60N channel layer of 185 nm thickness. On top of this a 170 Å thick n- Al_0.60Ga_0.40N barrier layer was grown which was followed by a 200 Å highly doped composition graded Al_xGa_1-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 (V_TH) of +3.6 V and drain source saturation current (I_DS) as high as 480 mA/mm at a gate voltage V_GS = +12 V. The peak extrinsic transconductance in the Al2O3-ZrO₂-MOSHFET is G_M ≈ 70 mS/mm.

To determine the factors leading to the high drain current we extracted the gate voltage dependencies of electron sheet density N_S and electron mobility µ. The extracted sheet carrier density is N_S ≈ 1.6 ×10¹³ cm^-2. µ is as high as 1600 cm²/(V.s) at low gate voltage near threshold V_G=2V and decreases down to 200 cm²/(V.s) at V_G +12V. The on-state field effect mobility is comparable to our depletion mode devices [3]. 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 V_G = -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.
References:
[1] E. A. Douglas et al., J. Vac. Sci and Technol. B, 37, 021208( 2019).
[2] B. A. Klein et al., Appl Phys. Lett., 114, 112104(2019).
[3] S. Mollah et al., phys. status solidi (a) 217, 7, 1900802 (2020).
[4] S. Mollah et al., Semicond. Sci. Technol., 34,125001( 2019).

Formation and control of electron spin in diamond is attracting much attention for next-generation quantum information and quantum sensing devices. With this fact in mind, we provide a guideline for the growth of homoepitaxial diamond films that possess higher crystalline quality, higher chemical purity, and a higher carbon isotopic ratio and nitrogen doping. A custom-built microwave plasma-assisted chemical vapor deposition system was constructed to achieve these requirements. To improve both the purity and crystalline quality of homoepitaxial diamond films, an advanced growth condition was applied: higher oxygen concentration in the growth ambient. Under this growth condition for high-quality diamond, a thick diamond film of >30 µm was deposited reproducibly while maintaining high purity and a flat surface [1, 2]. Then, combining this advanced growth condition for non-doped diamond (100) and (111) film with a unique doping technique that provides parts-per-billion order doping, single-color centers of either nitrogen-vacancy or silicon-vacancy centers that show excellent properties were formed [3, 4]. These advanced growth techniques are expected to accelerate the research fields of quantum information and quantum sensing devices using diamond.

The author would like to thank Dr. K. Ichikawa, Dr. C. Shinei, Dr. T. Kageura, Dr. S. Koizumi, Dr. K. Watanabe, Dr. S. Onoda, for helpful discussion and supporting characterization.

[1] T. Teraji, J. Appl. Phys. 118, 115304 (2015).
[2] T. Teraji, T. Yamamoto, K. Watanabe, Y. Koide, J. Isoya, S. Onoda, T. Ohshima, L. J. Rogers, F. Jelezko, P. Neumann, J. Wrachtrup and S. Koizumi, Phys Status Solidi A 212, 2365 (2015).
[3] L.J. Rogers, K.D. Jahnke, T. Teraji, L. Marseglia, C. Müller, B. Naydenov, H. Schauffert, C. Kranz, J. Isoya, L.P. McGuinness and F. Jelezko, Nature Communications, 5, 4739 (2014).
[4] P. Siyushev, M. Nesladek, E. Bourgeois, M. Gulka, J. Hruby, T. Yamamoto, M. Trupke, T. Teraji, J. Isoya and F. Jelezko, Science 363, 728 (2019).

1. Introduction
Surface finishing is a fundamental process to obtain the maximum performance from a material. Chemical Mechanical Polishing (CMP) following to mechanical polishing is widely used for semiconductor materials, such as Si, GaAs, SiC, and GaN. Diamond has traditionally been, however, mechanical-polished with diamond abrasive because it is the hardest material and ultimately inert. It is thought that nanoscale subsurface damage induced by mechanical polishing with diamond abrasive must be removed by CMP. Although there are many reports on diamond CMP, a detailed mechanism of diamond CMP has not been clarified yet. We proposed a new diamond CMP technology with chromium oxide abrasive. In this study, the time-dependent effect of the CMP treatment with chromium oxide slurry on the diamond substrate was investigated in the development of ideal diamond substrate with atomically flat and damage-less surfaces.
2. Experimental
We prepared four pieces of mechanically polished heteroepitaxial diamond substrates (KENZAN Diamond®) with diamond abrasive in order to obtain atomically flat surface prior to CMP. We applied CMP with chromium oxide slurry to three of them. The chromium oxide slurry was composed of 20 wt% of Cr₂O₃ abrasive, 70 wt% of water, and 10wt% of dispersant. The diamond substrate without CMP was also prepared as a reference sample. The surface morphology during/after CMP were observed with AFM and white-light interference microscopy.
We applied re-growth on the diamond substrates to evaluate the subsurface damage removal by CMP. The diamond thin film was grown on the polished substrates under the same growth condition. Surface morphology after re-growth was observed by SEM.
3. Results
We measured the surface roughness with white-light interference microscopy from a macroscopic viewpoint. The surface roughness drastically changed during CMP. In the initial stage from 0 to 5 hours, scratches appeared, and the surface roughness became extensive. Additional CMP up to 100 hours eliminated the scratches to attain a low surface roughness again. This phenomenon is known as a surface morphology change because of the subsurface damage removal among other hard-to-polish materials, such as sapphire, SiC, and GaN.
We observed the surface morphology with AFM from a microscopic viewpoint. An atomic step structure was observed on the 75-hours CMP substrate. The average step height was 0.09 nm, which is in good agreement with the atomic step height (a quarter of lattice constant) of the diamond (001). It is thought that the atomic step structure appears when CMP completely removes the subsurface damage induced by mechanical polishing.
We re-grown a diamond film on the polished diamond substrates in order to confirm the subsurface damage removal. The surface pits on the homoepitaxially grown film became less with increasing the CMP time. No pits were observed in the SEM image of the 100-hours CMP substrate. It is reasonable that the surface pits after homoepitaxial growth were reduced with increasing the CMP time because of the removal of the subsurface damage by CMP.
4. Summary
Surface morphology and surface roughness change after/during CMP with the chromium oxide slurry were observed to clarify the effect of CMP on the subsurface damage of the diamond substrate.
Drastic surface roughness change during CMP was observed with white-light interference microscopy.
Atomic steps were observed with AFM on the 75-hours CMP substrate.
We observed the pits reduction with increasing CMP time in the homoepitaxially grown film.
All these results indicate that the CMP removed the subsurface damage induced by the mechanical polishing. The diamond CMP with the chromium oxide slurry has the same effect on the subsurface damage removal as other reported diamond CMP technologies. Besides, this is the first time to report on the atomic step structure. We expect this result contributes to clarifying the mechanism of the diamond CMP.

Diamond semiconductors, owing to a wide bandgap of 5.47 eV and superior properties compared to SiC and GaN, can be used to develop high-power devices. Previously, we had proposed and demonstrated an NO₂ adsorbed hydrogen-terminated diamond field-effect transistor. [1] In a subsequent study, we demonstrated a diamond metal–oxide–semiconductor field-effect transistor (MOSFET) exhibiting a high drain current of 1.3 A/mm. [2] Kitabayashi and Kawarada et al. reported a diamond MOSFET with a high off-state breakdown voltage of 2021 V. [3] Diamond MOSFETs exhibiting either high current or high voltage have been reported predominantly until now. However, for a high radio–frequency power operation, a device with a combination of both high current and high voltage is necessary.
In this study, we fabricated NO₂ p-type doped diamond MOSFETs showing high current, high voltage, and the highest output power on a (001) heteroepitaxial diamond layer called the Kenzan diamond®. The heteroepitaxial diamond layer was grown on an Ir buffer layer on a (11-20) sapphire substrate. For delamination of the diamond layer from the substrate, the microneedle technique was used.
The heteroepitaxial diamond showed the lowest threading dislocation density and full width at half maximum of the (004) X-ray diffraction rocking curve of 1.4 × 10⁷ cm^-2 and 113.4 arcsec, respectively. Further, the NO₂ p-type doping was performed to generate high-density hole carriers in the diamond layer. Additionally, an Al₂O₃ double layer, optimized to both the hole channel and gate breakdown, was deposited on the NO₂ adsorbed diamond layer. To minimize the source resistance, the gate contact was formed adjacent to the source contact. The diamond MOSFET with a gate length of 1.4 µm exhibited a high drain-current density of −776 mA/mm with a negligible gate leakage current (less than 0.1 µA/mm). MOSFETs with a gate-to-drain length of 4.8 µm showed a high off-state breakdown voltage of −618 V and a specific on-resistance of 2.63 mΩ cm². The experimental Baliga’s figure-of-merit for the available output power density and maximum output power density were 145 MW/cm² and 12.3 W/mm, respectively. These values represent the highest values observed in diamond-based devices so far.
[1] M. Kubovic and M. Kasu, et al., Appl. Phys. Exp. 2, 086502 (2009).
[2] K. Hirama and M. Kasu, et al., Jpn. J. Appl. Phys. 51, 090112 (2012).
[3] Y. Kitabayashi and H. Kawarada, et al., IEEE Electron Device Letters 38, 363 (2017).

The huge earthquake and tsunami of 2011 in Japan caused the Fukushima Daiichi Nuclear Power Plant to lose all electricity and suffered a severe accident. The meltdown and subsequent hydrogen explosion caused harsh environmental conditions in the reactor containment vessels and the reactor buildings, with high radiation, high temperatures, and high humidity at the same time, which severely damaged the environmental monitoring equipment installed. Melted fuel, so called debris, are still in the damaged reactors and safe decommissioning operations are required under controlled conditions.
In order to increase the safety of next generation reactors and decommissioning operations, it is necessary to realize electronic circuits with semiconductor devices that can withstand the harsh environmental conditions. Diamond devices are known to realize long-term reliability at high temperatures [1] and are expected to be used as semiconductor devices for electronic circuits that can withstand severe accidents due to their high radiation hardness >10MGy [2]. The author and his colleague have developed active and passive devices based on diamond [2,3]. They have also succeeded in detecting pseudo-signals of radiations by a preamplifier circuit using diamond FETs under elevated temperatures. For the integration of diamond detector systems into nuclear reactors requires the development low noise circuits and mass production technology for the devices.
The author would like to thank Dr. J. H. Kaneko, Dr. A. Chayahara, Dr. S. Ohmagari and Dr. H. Kawashima for their great supports on experiments and discussion. The research is partially supported on the projects by Collaborative Laboratories for Advanced Decommissioning Science (CLADS) and MEXT, Japan.
References [1] K. Ikeda, et al., Appl. Phys. Express 2 (1), 011202 (2009). [2] H. Umezawa, et al., Proc. 29th ISPSD, 379-382 (2017). [3] S. Suzuki, et al., Thin Solid Films 680, 81-84 (2019).

The negatively charged nitrogen-vacancy centre (so-called NV^- centre) in diamond is one of the most promising systems for applications in quantum technologies because of the possibility to optically manipulate and read out the spin state of this defect, even at room temperature. These specific spin properties have paved the way to ultra-sensitive, high-performance and innovative quantum sensors (magnetometers, gyroscopes, spectrum analysers, etc.) which could open up perspectives in investigating properties that would remain inaccessible by conventional devices.
For the development of these applications, it is necessary to have diamond films of very high crystalline quality in which the density, environment, orientation and spatial localization of the introduced colour centres are perfectly controlled which is now possible thanks to microwave Plasma-Assisted Chemical Vapour Deposition (PACVD) [1]. Nevertheless, obtaining high densities of NV centres (> 500 ppb) remains particularly difficult due to the low conversion efficiency between substitutional nitrogen N_S and NV centres. In addition, the localization of these NV centres very close to the surface by in situ doping remains very challenging. For all of these reasons, the development of versatile implantation systems allowing both introducing nitrogen using N⁺ or creating vacancies with He⁺ ion bombardment is seen as a promising route towards this aim. In this paper, we present a single cavity ECR (electron cyclotron resonance) source module for producing relatively high energy ion (10-50 keV) beams using different gas sources. Using an electro-optic system, the beam size can be shaped and fitted to the surface to be implanted. We will show that the creation of high-density NV centres can be achieved near the surface using 2 approaches: (1) irradiating high purity CVD diamonds with N⁺ and (2) irradiating highly nitrogen doped CVD diamond layers with He⁺.

[1] J. Achard. V. Jacques. A. Tallaire. Chemical vapour deposition diamond single crystals with nitrogen-vacancy centres: a review of material synthesis and technology for quantum sensing applications. Journal of Physics D: Applied Physics. 53 (2020) 313001. http://dx.doi.org/10.1088/1361-6463/ab81d1

Diamond-based structures with luminescent defects are prominent platforms for photonics, sensing, imaging, and various other applications¹. Typically, high sensitivity systems with nitrogen vacancy (NV) centres in single crystal diamond (SCD) are demonstrated. However, challenges in scalable SCD processing and inferior optical properties of NV centres fuel the investigation for alternative emitters in micro- and nanocrystalline diamond (NCD) that offer wafer scale fabrication. A potential alternative is the Group IV defects that include silicon and germanium color centres².
In this study, we report in-situ fabrication of germanium vacancy (GeV) centres in polycrystalline diamond. NCD layers are grown on Ge substrates via the microwave plasma enhanced chemical vapour deposition technique under 1% CH₄ in H₂, 3000 W and 45 Torr process conditions. The deposition temperature is maintained at (700 ± 10)°C due to the proximity to Ge’s melting point (937°C). The solid source Ge doping occurs via the gas phase by plasma etching of the substrate, followed by Ge-vacancy complex formation in the deposited diamond layer^3,4. We demonstrate that Ge substrate does not form carbide layer. It is evidenced from electron energy loss spectroscopy that carbide phases are absent, instead an amorphous carbon layer of 2 nm forms at the diamond/substrate interface. Therefore, we were able to fabricate ~ 9 µm thick self-separating freestanding GeV incorporated diamond films with dimensions up to 1 × 1 cm². Scanning electron microscope analysis reveal facetted morphology of the NCD films while diamond formation is verified by Raman measurements. Room temperature photoluminescence (PL) measurements reveal 602 nm peak associated with the GeV formation in diamond. We also present time-of-flight secondary ion mass spectroscopy (ToF-SIMS) results that confirm Ge incorporation in NCD and discuss the non-uniformities in Ge incorporation in the film with PL mapping supported by the high resolution ToF-SIMS elemental imaging technique.

References

1. Aharonovich, I., Greentree, A. D. & Prawer, S. Diamond photonics. Nat. Photonics 5, 397–405 (2011).
2. Bradac, C., Gao, W., Forneris, J., Trusheim, M. E. & Aharonovich, I. Quantum nanophotonics with group IV defects in diamond. Nat. Commun. 10, 1–13 (2019).
3. Iwasaki, T. et al. Germanium-Vacancy Single Color Centers in Diamond. Sci. Rep. 5, 1–7 (2015).
4. Ralchenko, V. G. et al. Observation of the Ge-Vacancy Color Center in Microcrystalline Diamond Films. Bull. Lebedev Phys. Inst. 42, 157–164 (2015).

Diamond as a semiconductor exhibits excellent thermal conductivity (20W/cmK)[1] and therefore is very suitable for high power electronics where self-heating is often the primary reason for failure. Owing to a high thermal conductivity, very high forward current densities can theoretically be rectified through diamond Schottky-pn or Schottky-pin diodes before self-heating limitations. Although a high forward current density at 60 kA/cm² was demonstrated by Makino et. Al [2], practical diodes start self-heating and fall short of the theoretical limits. It therefore becomes important to understand what the practical limitations in diamond diodes are, especially the effects of traps and recombination centers that limit the forward current density.
In this research we demonstrate a practical diamond Schottky-pin diode with a 114 kA/cm² forward current density at 17 V forward bias and fit theoretical drift-diffusion simulations to the measured data. The simulations were done with bulk trap densities of mainly three different energy levels uniformly distributed in the i and n regions of the diode. The trap energy and capture cross-sections for all trap levels were considered from Bruzzi et. al. [3] and Ščajev et. al. [4]. A hopping mobility is calculated for both electrons and holes to account for nearest neighbor and variable range hopping. We also observe velocity saturation in the i and n diode regions where the electric field gets large due to depletion from the Schottky metal contact as well as the p-i-n junction. Therefore, an electric field dependence in the mobility model is also considered. The simulations fit very well with the measured data and provide insight into different regions in the IV characteristics. Different regions in the IV characteristics are analyzed in detail by individually turning on the trap levels, field dependence on mobility and surface recombination effects. An ideal case for a theoretically possible current density before self-heating is also made.
References:
[1] Isberg, Jan, et al. "High carrier mobility in single-crystal plasma-deposited diamond." Science 297.5587 (2002): 1670-1672.
[2] Makino, Toshiharu, et al. "Diamond Schottky-pn diode without trade-off relationship between on-resistance and blocking voltage." Physica Status solidi (a) 207(9) (2010): 2105-2109.
[3] Bruzzi, Mara, et al. "Deep levels and trapping mechanisms in chemical vapor deposited diamond." Journal of Applied Physics 91(9) (2002): 5765-5774.
[4] Ščajev, P., et al. "Features of free carrier and exciton recombination, diffusion, and photoluminescence in undoped and phosphorus-doped diamond layers." Diamond and Related Materials 57 (2015): 9-16.

Ultra wide bandgap (UWBG) materials such as monocrystalline diamond exhibit the best figure of merits for power semiconductor devices: the ultra-wide bandgap of 5.5eV translates into the highest electric field above 10MV/cm, which offers a thinner drift region with an associated higher doping level than with other Wide Bandgap (WBG) materials for the same breakdown voltage. Consequently, power semiconductor devices based on UWBG and diamond in particular have theoretically the lowest specific on state resistance, which leads to smaller active area and lower conduction and switching losses. Many different diamond power device architectures have been previously introduced, but the implementation of those in actual power converters was only reported in a limited number of communications [1-3]. Although the measurement of on state resistance is straight forward, the measurement of switching losses is challenging with the existing and future High Voltage (>600V) diamond power devices. First, diamond devices need to be packaged in order to be integrated in, at least, one power commutation cell (one transistor and one diode, or two transistors, with the same breakdown voltage). This packaging should also be compatible with a high junction temperature operation (above 150°C) to fully demonstrate the performances of diamond devices based on bulk conduction. Second, the active area of diamond devices can be relatively small compared to other available power devices. Consequently, the maximum switching speed of diamond power devices can be extremely fast, and the intrinsic parasitic capacitors much smaller than 1pF. Third, the total current rating of diamond power devices can be rather small (µA to mA), which makes it difficult to probe the fast large signal variation of low drain currents, also with high voltage swings. Finally, the required gate to source driving voltages of diamond power transistors can be different from their Silicon and WBG alternatives, with large gate to source voltage swing [5]. For all these reasons, a specific experimental approach must be developed to adapt the implementation of diamond power devices in power converters, and to be able to accurately extract the large signal switching behavior of those devices.

In this work, we present the specific experimental characterization boards we designed and fabricated to achieve the fastest switching of diamond Schottky Barrier Diodes, and a flexible converter to extract switching losses of WBG and UWBG power transistors. The classical Double Pulse method is compared to the opposition method, where switching losses can be accurately measured without having to probe drain or source currents with fast dynamics. The converter is validated for high voltage inputs above 600V, load currents above 20A, switching frequencies above 100kHz, and flexible gate driver boards to adapt to specific gate voltages. Preliminary results on power switching with diamond MESFETs are also presented and discussed.

The research leading to these results has been performed within the GreenDiamond project (http://www.greendiamond-project.eu/), received funding from the European Community's Horizon 2020 Program (H2020/2014-2020) under grant agreement n° 640947 and is partially funded by French ANR Research Agency under grant ANR-16-CE05-0023 #Diamond-HVDC.

[1] Aboulaye Traoré et al 2017 Jpn. J. Appl. Phys. 56 04CR14
[2] V.D. Blank et al, Power high-voltage and fast response Schottky barrier diamond diodes, Diamond and Related Materials, Volume 57, 2015, Pages 32-36,
[3] G. Perez et al., "Diamond Schottky barrier diodes for power electronics applications," 2018 IEEE Energy Conversion Congress and Exposition (ECCE), Portland, OR, 2018, pp. 1956-1963
[4] C. Masante, J. Pernot, J. Letellier, D. Eon and N. Rouger, "175V, > 5.4 MV/cm, 50mOhm.cm2 at 250°C Diamond MOSFET and its reverse conduction," 2019 IEEE ISPSD, Shanghai, China, 2019, pp. 151-154

Diamond transistors are promising for high temperature, high power and high frequency applications in harsh environments as a result of this material extraordinary properties, such as high electron and hole mobility (4500 cm²V⁻¹s⁻¹ and 3800 cm²V⁻¹s⁻¹), low dielectric constant (5.7), high breakdown field (10 MV/cm), high thermal conductivity (10 W/cm K) or high radiation hardness. However, no shallow dopants are found and p-type doping is acquired using boron (0.37 eV) and n-type using phosphorous (0.6 eV). Because of its specificities, it is often more convenient to find original ideas for diamond transistors rather than using the classical designs. A normally-off inversion Diamond MOSFET was demonstrated in 2016. Nonetheless, fabricating a power inversion MOSFET requires a control of diamond growth and etching that has not been yet reached. Instead, most of Diamond transistors are normally-on MOSFETs, based on the 2D hole gas emerging due to transfer doping in H-terminated surface. But also, new transistors concepts like the diamond deep depletion MOSFET have shown promising results based on bulk boron doped diamond conduction.
In this work, we demonstrate a normally-off high breakdown voltage p-diamond reverse blocking MESFET. The source is fabricated using selective p+ diamond growth and Ti/Pt/Au Ohmic contacts, while the gate and drain Schottky contacts are made of Molybdenum on O-terminated Diamond. The major challenge in the fabrication of a normally off-MESFET is to target the correct layer thickness and doping concentration, in order to have a closed channel that can be opened applying a BIAS to the Schottky gate. Despite, the progress made in diamond growth allow us for the first time to design and fabricate such a device. The Drain and Gate Schottky contacts have been characterized showing a high ideality factor of 2.1 at room temperature, tentatively explained by an interface oxidized layer. The density of states of the interface is estimated as _~10¹³cm^-2 by the conductance method. The devices are normally off displaying undetectable gate leakage current. An on-state current level of 10^-4 mA is reached in the ON-state at room temperature and 10^-2 mA at 450 K. A breakdown voltage of <1kV was measured for the device.

Ultra wide bandgap (UWBG) materials such as monocrystalline Diamond are foreseen as the next-generation power semiconductor devices [1], whereas wide bandgap (WBG) materials such as SiC and GaN are already pushing towards more efficient and more integrated power converters [2,3]. Diamond has outstanding physical properties such as the highest thermal conductivity (20 W.cm^-1K^-1 at room temperature), a large critical electric field (10 MV.cm^-1), a wide doping range and a high carrier mobility in both P and N type doping types. Diamond devices are also maturing, with many improvements and breakthroughs in the recent years: Deep Depletion and inversion MOSFETs [4,5], 2DHG FETs with high current and high breakdown voltage capabilities [6], JFETs, high voltage and high temperature MESFETs, Schottky, PiN and Schottky PiN diodes [1]. Nonetheless, diamond devices based on bulk diamond conduction (e.g. MOSFETs, JFETs, MESFETs, Schottky diodes) suffer from the incomplete ionization of Boron dopants at typical operating junction temperatures (e.g. 300K-400K), which increases consequently the ON state resistance of such unipolar devices. Moreover, the active area of diamond can be limited by defect density and substrate size, which limits the high-current capability of diamond power devices in a short term.

How could one introduce as soon as possible diamond power devices in power modules to take benefits of their outstanding performances, while mitigating their weaknesses?

In this work, we propose to combine boron-doped diamond power transistors based on bulk conduction in a hybrid association with WBG power devices such as SiC MOSFETs. Other hybrid associations between Si and SiC power devices have been previously proposed such as in [7], but UWBG such as diamond was never considered. A small area diamond power transistor is associated in parallel with a SiC MOSFET which is smaller than a SiC-only solution, and both SiC and diamond devices having the same breakdown voltage (1.7kV in this work). This hybrid association offer an attractive combination of the Positive and Negative Temperature Coefficients for the ON state resistance of SiC and Diamond power devices. This hybrid association is modeled, discussed and benchmarked with SiC-only (analytic and real datasheet) and Diamond-only solutions, considering representative specifications at the power converter level. Preliminary studies show a reduction of SiC active area by two. Particularly, the optimal sizing and the current sharing between the paralleled power devices are investigated as a function of operating conditions and junction temperatures. The thermal management solutions and electro-thermal coupling are also discussed based on analytical and numerical models.

[1] N Donato et al 2019 J. Phys. D: Appl. Phys. 53 093001
[2] D. Johannesson et al., "Assessment of 10 kV, 100 A Silicon Carbide mosfet Power Modules," in IEEE Transactions on Power Electronics, vol. 33, no. 6, pp. 5215-5225, June 2018.
[3] Y. Zhang and T. Palacios, "(Ultra)Wide-Bandgap Vertical Power FinFETs," in IEEE Transactions on Electron Devices, vol. 67, no. 10, pp. 3960-3971, Oct. 2020.
[4] T. Pham et al., "Deep-Depletion Mode Boron-Doped Monocrystalline Diamond Metal Oxide Semiconductor Field Effect Transistor," in IEEE Electron Device Letters, vol. 38, no. 11, pp. 1571-1574, Nov. 2017.
[5] Matsumoto, T. et al. Inversion channel diamond metal-oxide-semiconductor field-effect transistor with normally off characteristics. Sci Rep 6, 31585 (2016).
[6] M. Iwataki et al., "Over 12000 A/cm2 and 3.2 mΩ cm2 Miniaturized Vertical-Type Two-Dimensional Hole Gas Diamond MOSFET," in IEEE Electron Device Letters, vol. 41, no. 1, pp. 111-114, Jan. 2020.
[7] A. Deshpande and F. Luo, "Practical Design Considerations for a Si IGBT + SiC MOSFET Hybrid Switch: Parasitic Interconnect Influences, Cost, and Current Ratio Optimization," in IEEE Transactions on Power Electronics, vol. 34, no. 1, pp. 724-737, Jan. 2019.

Monocrystalline diamond is known as a material that possesss an extraordinarily combination of extreme properties. While this is an ideal starting point to become the next-gen electronic device material, several hurdles remain to be taken. The use of microwave plasma enhanced chemical vapour deposition (MW PE CVD) in combination with a a wide parameter space enables the production of films with a wide window of doping concentrations and thicknesses that range from a few hundreds of nanometers to over a millimeter thick. Nonetheless, many times the device performance of electronic applications based on such layers remains far below the theoretical predictions. This is partly due to a lack in fundamental understanding in the relation between (varying) substrate quality, process conditions, dopant incorporation, and final crystalline properties. The latter will not only influence the actual dopant mobility and lifetime, but also the level of leakage currents in device structures based on such films.

Here, a detailed study on the methane concentration dependence of the plasma gas phase on the surface morphology and boron incorporation in monocrystalline CVD diamond layers is presented. Both heavily doped p⁺ films, with boron concentrations above the Mott transition (~ 3 x 10²⁰ / cm³), as well as lightly doped p^- films with acceptor concentrations down to 1×10¹⁴ / cm³ will be discussed. A combined use of several characterisation techniques to assess the electronic, optical, and structural properties, including Hall measurements, Raman and absorption spectroscopy, AFM, and TEM, will shed light on the beforementioned relation between process conditions and subsequent film quality. The combination of said p⁺/p^- layers in pseudo-vertical Schottky barrier diodes leads to devices that show a very low leakage current, down to 10^–12 A, with forward current levels up to ~ 10³ A/cm².

Finally, the experimental findings on impurity incorporation are corroborated by first principles calculations that give insight in the relation between methane concentration and preferential boron incorporation sites, also providing a route to understanding impurity incorporation in diamond on a general level, potentially of great importance for color center formation [1].

[1] R. Rouzbahani, S.S. Nicley, D.E.P. Vanpoucke, F. Lloret, P. Pobedinskas, D. Araujo, K. Haenen, “Impact of methane concentration on surface morphology and boron incorporation of heavily boron-doped single crystal diamond layers”, Carbon 172 (2021), 463-473.

Diamond has been exhaustively studied during the last few decades as semiconductor since it is theoretically the best candidate to meet the power electronics trade-off. However, the fabrication of devices based on diamond presents a set of challenges, for example the diamond available bulk size and surface termination quality and the fact of not having a native oxide among others, such problems prevent devices high performance and reliability. The present study is focused on the electrical and microstructural characterization of zirconium dioxide to be employed in diamond-based electronic devices for field effect transistor applications. Most of diamond Metal Oxide Semiconductor Capacitors (MOSCAPs) in literature were fabricated using Al₂O₃ or SiO₂ due to properties such as the wide band gap and their alignment with the diamond band setting. Based on those gate dielectrics, it has been recently demonstrated the first inversion channel normally-off Metal Oxide Semiconductor Field Effect Transistor (MOSFET) and a Deep Depletion MOSFET (D3MOSFET) showing great perspectives for the future of power electronics. However, in case of 2D hole gas channel, low channel mobility due to high density of interface states is still an issue so more effective passivation techniques are required to reduce losses of the performance. In this context, zirconia appears as a good candidate for the gate dielectric because of its high dielectric constant (with a minimum crystalline average value of about 19.7) , high band gap (around 6eV) and its chemical and thermal stability favouring the oxide-diamond interface quality. In this work, the devices under test are MOSCAPs and Metal Oxide Metal capacitors fabricated on a p-type oxygen-terminated (001) diamond layer using an insulating layer of ZrO₂ grown by Atomic layer deposition (ALD). I-V, C-V and C-f measurements with temperature ranging between room temperature to 125 °C have been performed in order to extract information such as the doping concentration of the p- diamond layer, the MOSCAP flat band voltage, the oxide charge density or the interface state density. On the other hand, microstructural characterization through different techniques have been performed to determine the crystallinity degree and oxide thickness through High resolution electron microscopy (HREM) and the relate Fast Fourier transform (FFT) or X-ray diffraction (XRD), the oxide band gap has been analysed by Electron energy loss spectroscopy (EELS) and the stoichiometry analysis of the oxide has been based on Energy-dispersive X-ray spectroscopy (EDX). Finally, helped by finite element calculation of MOSCAP band diagram, the correlation between the electrical response and the microstructural nature of the device will be established and discussed.

A super-high dielectric constant AlO_x/TiO_y nanolaminate film is grown on hydrogenated crystalline diamond (H-diamond) to enable superior diamond-based metal-oxide-semiconductor (MOS) capacitors and metal-Oxide-Silicon-Field-Effect-Transistors (MOSFETs). In order to minimize or suppress leakage current, a nanometer thick AlO_x film is inserted at the AlO_x/TiO_y nanolaminate / H-diamond interface. The maximum values for the capacitance density and dielectric constant related to the summation of individual AlO_x and nanolaminate are 1.06 μF/cm² and 68.7, respectively. Capacitance density and dielectric constant for the AlO_x/TiO_y nanolaminate are as high as 5.22 μF/cm² and 308, respectively. Electrical properties of four H-diamond MOSFETs with gate lengths increasing from 2.4 μm to 10.1 μm were investigated. All of them showed p-type behavior and distinct pinch-off characteristics with drain current maxima of −47.4, −43.3, −26.6, and −24.6 mA/mm, respectively. On/off ratios and threshold voltages for the MOSFETs are higher than 10⁴ and lower than 0.55 ±0.10 V, respectively. The low threshold voltages indicate that the AlO_x/TiO_y nanolaminate gate based MOSFETs can switch between ON and OFF stages at very low gate voltages. Effective mobilities of the H-diamond channel layers for all the MOSFETs raised firstly and dropped subsequently with increasing voltages, which can be explained by the effect of mobility limiting factors.

1. Introduction
Many techniques have already been used to characterize carriers’ properties in diamond detectors by doing Time of Flight (ToF) measurements. The most common is to use alpha particles, which deposit their energy near the diamond surface [1, 2]. But some other techniques like using biphotonic absorptions [3] and neutrons [4] have already been carried out. Some experiments based on a pulsed laser [5] have found very high mobility at room temperature for electrons and holes. This high mobility is very interesting for fast timing and high rate detection applications. Time of flight using Electron Beam Induced Current (ToF-EBIC) is a new way to determine and study the carrier mobilities.
2. Time of Flight using Electron Beam Induced Current
The experiments were conducted through the following procedure: a CVD single crystal diamond is put in a scanning electron microscope chamber under vacuum, and biased by means of an external voltage source. A pulsed electron beam is focused on a small area of the diamond. The diamond works as a solid ionization chamber: electrons will lose their energy by creating electron-hole pairs in a small volume located close to the entrance surface. Because of the electric field, the carriers would drift to the negative electrode (holes) and to the positive electrode (electrons). The signal is extracted from the electrodes, amplified by a fast preamplifier (CIVIDEC C2-HV [6]) and the waveforms are digitized by an oscilloscope. Depending of the beam energy, the electrons are stopped in the first 1 to 5 μm from the diamond surface, which is quite small compared to the diamond bulk thickness (around 500 μm). Thus, one type of carrier is instantly collected (~10-100 ps) while the other one will drift a few nanoseconds to reach the opposite electrode. Then, it is possible to study separately the drift of electrons and holes so as to evaluate their mobilities for various bias voltages. In this work, thanks to this ToF-EBIC technique, diamond surface mapping for different depths (up to 5 μm) will be reported, charge carrier life time will be evaluated and carrier mobilities will be studied at different temperatures.

This work is supported by IDEX Université Grenoble Alpes and by DIAMTECH IN2P3.

[1] M. Pomorski, et al., phys. stat. sol. (a), 203: 3152-3160 (2006). https://doi:10.1002/pssa.200671127
[2] H. Pernegger et al., Journal of Applied Physics 97, 073704 (2005). https://doi.org/10.1063/1.1863417
[3] C. Dorfer et al., Appl. Phys. Lett. 114, 203504 (2019). https://doi.org/10.1063/1.5090850
[4] C. Weiss, et al. Eur. Phys. J. A 52, 269 (2016). https://doi.org/10.1140/epja/i2016-16269-8
[5] J. Isberg, et al., Science 297, 1670 (2002). https://doi.org/10.1126/science.1074374
[6] CIVIDEC https://cividec.at/

This work centers on AlScO₃, one of the 14 candidate oxides identified in our recent computational survey of semiconductors for power electronics [1]. The survey utilized intrinsic materials properties entering Baliga figure of merit (BFOM) to gauge performance. AlScO₃ is a distorted ABO₃ perovskite (metastable high-pressure phase) typically investigated for phase transformation under high-temperature and -pressure conditions. However, its electronic properties and doping tendencies are rarely explored. Our first-principle calculations predict an indirect electronic band gap around 8.0 eV (HSE06+G₀W₀), which is much larger than those of typical ultrawide bandgap semiconductors. Such wide bandgap makes it a promising candidate semiconductor for high-temperature or/and high-voltage applications. A functional device requires the underlying semiconductor to be dopable and we addressed this issue by performing first-principle defect calculations. Specifically, we adopted the well-established supercell approach to perform defect calculations and the total energies are calculated using the standard HSE06 hybrid functional with the bandgap corrected by the many-body perturbation theory (G₀W₀). Our calculations of native defects point to the cation and oxygen vacancies as the dominant point defects. The exothermic nature of the cation vacancies for Fermi energies close to the conduction band minimum at both cation- and oxygen-rich conditions suggest n-type insulating behavior. On the other hand, at the oxygen-rich conditions, the anion vacancies allow p-type doping. We further investigated the formation energies of potential extrinsic p-type dopants, like Mg and Zn, and found them to be deep acceptors implying moderate p-type doping. Lastly, we examined the hole self-localization and found stable small hole polarons with ionization energies of the order of 0.1 eV. Ionization energy of this magnitude implies activated transport that should be largely overcome at moderate to high-temperature. Our computational results also call for further investigations into the electronic properties and the search for shallower acceptor dopants of this promising p-type ultrawide bandgap oxide.

[1] P. Gorai et al. Energy Environ. Sci., 2019,12, 3338-3347

Beta-Gallium oxide (β-Ga₂O₃) has garnered tremendous research interests for high power, high voltage, and high frequency applications due to its ultra-wide bandgap (~4.8 eV), high critical breakdown field of 8 MV/cm, and large Baliga's figure of merit. (Al_xGa_1-x)₂O₃/Ga₂O₃ delta-doped high electron mobility transistors (HEMTs) have been experimentally demonstrated with high quality electron channel and good electron mobility. Delta-doping regulates the electrical characteristics of (Al_xGa_1-x)₂O₃/Ga₂O₃ HEMTs since it is responsible for the formation of two-dimensional electron gas (2DEG) in the channel. With epitaxial growth techniques such as MOCVD and MBE, delta-doping position and concentration can be precisely controlled. However, the effects of delta doping on device performance of (Al_xGa_1-x)₂O₃/Ga₂O₃ HEMTs are not comprehensively investigated.

In this work, we explore the design and optimization of (Al_xGa_1-x)₂O₃/Ga₂O₃ delta-doped HEMTs via TCAD SILVACO simulation, which can serve as critical guidance for experimental realizations. To date, there are very few reports on the simulation of (Al_xGa_1-x)₂O₃/Ga₂O₃ HEMTs, partly due to the immature status of Ga₂O₃ research and the lack of accurate material properties. In the simulations, we first calibrated our model with experimental electrical characteristics of (Al_xGa_1-x)₂O₃/Ga₂O₃ HEMTs such as their transfer and transconductance curves. Then a comprehensive series of design and optimization of delta doping in the HEMTs was conducted, including delta-doping concentration, delta-doping position, delta-doping width, and multiple delta-doping channels.

The simulated HEMT structure consisted of Ga₂O₃ substrate, 130 nm unintentionally doped (UID) Ga₂O₃ layer, and 27 nm (Al_xGa_1-x)₂O₃ layer, and two n⁺-Ga₂O₃ contact regions for the source and drain. The delta-doping region was modeled by a thin n-Ga₂O₃ slab with a donor concentration of 1e18 cm⁻³, which is positioned inside the (Al_xGa_1-x)₂O₃ layer at 4 nm distance from the Ga₂O₃/(Al_xGa_1-x)₂O₃ interface. With increasing delta-doping concentration, the device transconductance and drain current increased significantly. When the delta-doping concentration was over 1e19 cm⁻³, the device exhibited large leakage from the delta-doping region because the conduction band at the delta doping position was lower than that of the channel. The position of delta doping also played a critical role. It was found that the closer the delta doping to the interface, the larger the device transconductance and drain current. The largest breakdown voltage was obtained when the delta doping was 2 nm from the channel. We also simulated the effect of delta doping diffusion across the interface, which is a great concern during growth. The current collapse was observed due to the diffused impurities into the Ga₂O₃ layer, which decreased the large built-in electric field generated by the interface barrier. In addition, multiple delta-doping channels can also significantly impact device threshold voltage and breakdown voltages.

In short summary, these results provide critical references for the future development of high performance Ga₂O₃ high power and high frequency electronics.

Gallium oxide (Ga₂O₃) and its alloys are attracting great attention due to their potential in technological applications such as gas sensors, transparent conductive oxides for solar cells, UV detectors, magnetic sensors and devices, and high-power electronic devices. Doping Ga₂O₃ and/or alloying with foreign divalent elements, such as Mg²⁺, Be²⁺, Zn²⁺, and Fe²⁺ are common ways to achieve high insulating materials needed in these applications. Although there have been few studies on the effects of iron in Ga₂O₃, no systematic investigation about the stability and electronic properties of (Ga_1-xFe_x)₂O₃ have been reported so far.
In this work, we present the results of Fe doping in Ga₂O₃ using density functional theory-based calculations performed for different Fe concentrations ( and ). The role played by oxygen vacancies is also investigated. The effects of strain in the pristine oxide reveals that the band gap decreases with increasing lattice constant. This behavior is also observed in the Fe-doped systems with increasing . The theoretical results are compared with recent experimental measurements using XRD measurements and obtained in (Ga_1-xFe_x)₂O₃ samples grown with various Fe contents and in different growth conditions, allowing to get insight into the stability of the structure.

This work was supported in part by NSF through an DMR-MRI Grant under Award 1726970 and in part by DOD through a HBCU/MI grant (W911NF2010298).

As we move toward a more renewable, variable, and distributed electricity grid, we will need to find power electronic (PE) semiconductors that can withstand high power and high voltage applications. Ultra-wide band gap (UWBG) materials are promising candidates as they tend to exhibit large breakdown fields which increases the operating voltage limit for a given device size. However, currently explored UWBG materials either lack the thermal properties needed to dissipate heat at high power (Ga₂O₃), are difficult to synthesize as quality single crystals at scale (diamond, c-BN), or are difficult to dope to desired carrier concentrations (AlGaN). Using density function theory (DFT) calculations, we conduct a high-throughput computational search to identify promising UWBG semiconductors beyond those currently used or studied for n-type power electronic devices. We consider >1,500 total oxides, nitrides, sulfides, carbides, silicides, and borides. First principles generalized gradient approximation (GGA) calculations and semi-empirical models are used to evaluate the carrier mobility and critical breakdown field which are inputs to the well-known Baliga figure of merit. We use this metric along with the lattice thermal conductivity identify top potential candidates with performance better than the current market materials for high power (SiC-4H and GaN). The electronic structures of a sub-set of the most-promising oxide candidates are further assessed using low-throughput, higher-accuracy HSE06 hybrid calculations. Our results find more than 28 promising candidates that have higher lattice thermal conductivity than Ga₂O₃ and higher n-type Baliga figure of merit than SiC-4H and GaN which are the current materials used in high-power PE applications today. Of these candidates, 65% are oxides and 35% are nitrides, with chemistries covering binaries to quaternaries. We also conduct a qualitative assessment of n-type dopability of the top candidates by evaluating the governing material properties, considering acceptor defects as the limiting factor. Given the large number of oxides in the top candidates, these UWBG materials are promising for use in high-temperature electronic devices as well as high-power PE applications. This work provides guidance for selecting candidates with confidence for further experimental investigation. Portions of this work have been published by Gorai et al. in: Energy Environ. Sci., 2019, 12, 3338.

An external lattice perturbation may lead to localized height variations (ripples) in two-dimensional (2D) materials during the growth process, and as a result, affect their functionality. Here, we conduct a systematic first principle study of the effect of rippling on the structural and electronic properties of 2D layers (hBN, graphene) and the resulting 2D/Diamond heterostructures. We find the most favorable alignment of the layer on the hydrogen-terminated diamond (H-diamond (100)) substrate using a combined sampling and a machine learning approach. Rippling of sheets can arise from templating on a variety of substrates. The rippled structures we studied on H-diamond were found to have a stronger vdW interaction between the constituent layer and the substrate. A higher degree of charge transfer was observed in the heterostructures composed of a rippled hBN (graphene) layer, as compared to their planner counterparts. The transferred charges are confined within the vdW-gap between the 2D layer and H-diamond (100) surface. The higher thermodynamic stability, improved charge transfer, and additional tunability of the electronic properties in the rippled 2D/Diamond heterostructures make them better choices for application in diamond-based field effect transistors.

The transparent wide bandgap semiconductor, β-Ga₂O₃ has gained considerable attention due to its suitability to a wide range of applications. For example, its wide bandgap and small electron effective mass are suitable for power electronic devices. Interestingly, even though this is not a van der Waals material, it can be peeled along (100) surface by a scotch tape method like in graphene. The resulting ultra-thin layers are used for high power device fabrications. The theoretical calculation has shown that out of the two possible (100) surface terminations, (100)B surface is the most stable one, which is also experimentally observed from scanning tunneling microscope study (STM). One of the interesting properties of this material is that thin layers preserve the pristine bulk-like electronic properties, making it even more promising for applications in power devices. It is then of paramount importance to understand why the exfoliation occurs facilely even though the interlayer interaction in β-Ga₂O₃ is not van der Waals. This presentation will show the exfoliation phenomenon from the first principle thermodynamic total energy calculations and the mechanisms behind the easy exfoliation along (100)B surface. Lastly, we will show a theoretical prediction of a new phase of Ga₂O₃ obtained from the stacking of the exfoliated layers of β-Ga₂O₃. From the stability calculations, this phase is comparable to β-Ga₂O₃, as well as dynamically stable. From the hybrid density functional calculations, the new phase's bandgap is a little higher than the β-phase. The electronic and thermal properties of this new phase will be briefly presented and compared with β-Ga₂O₃. The methodology used and developed for this study can be utilized in general to understand bond breaking and forming in other wide bandgap materials as well. This understanding will give us better control to fabricate thin-film 2D devices, in general.

β-Ga₂O₃ is very promising to be applied in power electronic devices because of its superior properties, like the ultra-wide bandgap, high breakdown electric field, and the relatively low-cost growth of its single crystal. However, heat dissipation issue would limit the performance and reliability of β-Ga₂O₃ devices considering its extremely low thermal conductivity. A top-side cooling or double-side cooling strategy would be necessary to efficiently dissipate heat from the active region of β-Ga₂O₃ devices. As a result, thermal transport at the metal/β-Ga₂O₃ contacts is very important for device thermal management, and a thorough understanding of the thermal boundary conductance (TBC) at metal/β-Ga₂O₃ interfaces is needed. Anisotropy effects have been observed in the thermal conductivity of β-Ga₂O₃ but haven’t been studied for TBC at β-Ga₂O₃ interfaces. In this work, we perform both experiments and theoretical simulations to study how the material orientation will affect the TBC at metal/β-Ga₂O₃ interfaces at different temperatures. The Al/β-Ga₂O₃ interface is studied as our model system because of the isotropic property of Al, which will not introduce other anisotropic mechanism into interfacial thermal transport. Both E-beam evaporated and molecular-beam epitaxy (MBE) grown Al/β-Ga₂O₃ interfaces are studied to minimize the influence of other factors (like growth method related defects at interfaces) on TBC. We apply the time-domain thermoreflectance (TDTR) method to measure the TBC at Al/β-Ga₂O₃ interfaces with different β-Ga₂O₃ substrate orientations (-201), (010), and (001), and the TBCs show strong anisotropy effect. In our theoretical calculation, to capture the anisotropy effect, a full first Brillouin zone k-space Landauer formula is applied. With the phonon information from ab initio calculations of the full first Brillouin in reciprocal lattice, there is no isotropic assumption and the influence of orientation can be captured and analyzed. Our modeling results are consistent with our experimental measurements that the TBC at (010) β-Ga₂O₃/Al interface is the highest, while the TBC at (-201) β-Ga₂O₃/Al interface is the lowest, and the difference is more than 50%. Our work shows that the anisotropy in β-Ga₂O₃ not only affects its thermal conductivity but its TBC as well. To capture the anisotropy effect, a modal approach with phonon information of the full first Brillouin zone is needed to accurately predict the TBC at β-Ga₂O₃ interfaces. The anisotropic TBC effect should be considered for the future thermal design of β-Ga₂O₃ devices.

Gallium oxide has attracted a tremendous attention in recent years as a promising wide-bandgap semiconductor for power electronics and deep-UV optoelectronics. The optical properties of monoclinic β-Ga₂O₃ have been actively investigated both theoretically and experimentally, and a large range of the reported band gap values (from 4.3 eV to 5.1 eV) have been explained based on anisotropic onset of the optical absorption that originates from the crystallographic differences of the monoclinic gallium oxide phase.

Amorphous metal oxides have several advantages over their crystalline counterparts: the lack of lattice periodicity with defined crystal orientations, no grain boundaries, and an increased number of degrees of freedom for metal-Oxygen (MO) polyhedra packing in the disordered material lead to a uniform low-strain crack-resistant morphology, smooth surfaces over large areas, and in the case of ionic oxide semiconductors, a high carrier mobility. The effect of disorder on the optical properties of oxides is two-fold: on one hand, the fundamental band gap is determined by the nearest-neighbor interactions, i.e., the metal-oxygen bond strength, that is usually preserved upon amorphization; and on the other, disorder and structural defects may cause electron localization near the band edges, and these so-called tail states may contribute to the near-edge absorption.

In this work, computationally-intensive ab-initio molecular-dynamics liquid-quench simulations are combined with accurate electronic structure calculations using a hybrid functional in order to determine the structure and properties of amorphous Ga₂O₃ as a function of its density. Specifically, we calculate the coordination of individual Ga and O atoms in a large statistical ensemble of ten independent MD realizations at five different densities and predict changes in the coordination distribution as the density decreases to an optimal value for the amorphous oxide. In contrast to monoclinic β-Ga₂O₃ where half of the Ga atoms are octahedrally-coordinated with oxygen atoms and the other half is tetrahedrally coordinated, the number of the 6-fold GaO polyhedra is completely suppressed in amorphous oxide. These changes do not affect the electronic states in the conduction band which remains delocalized upon amorphization of stoichiometric Ga₂O₃. However, presence of 5-coordinated Ga and the resulting Ga coordination morphology determine the strong localization of the non-bonding O-p-states at the top of the valence band that contributes to the near-edge optical absorption. The calculated band gap of stoichiometric amorphous Ga₂O₃ is similar to the one in the crystalline phase, yet, the optical properties are isotropic in the disordered material. The effects of oxygen stoichiometry on the structure and properties of amorphous Ga₂O_3-x are also discussed.

Engineering a functionalised diamond surface that is stable and reproducible is of technological importance to future diamond electronic devices such as Schottky junctions and electron-emitting electrodes for field emission, thermionic energy converters, secondary electron multipliers.
One of the most interesting properties of diamond is the negative electron affinity (NEA) which means that the vacuum level lies below the conduction band, which makes diamond an efficient source of electrons. The NEA can appear when the surface carbon atoms are negatively charged which has been achieved by terminating the diamond surface suitably. Maier et al. [1] found EAs of 0.5 eV, -1.3 eV and 1.7 eV for the bare, H- and O-terminated (100) surfaces, respectively. A monolayer or sub-monolayer coverage of some of the electropositive group I and II metals and first-row transition metals (TMs) on the bare and oxygenated diamond surface have been found to be a suitable candidate for diamond surface termination. Transition metals like Cu, Ti, Ni and Co metals have been found theoretically to impart NEA to the diamond surface however their ability to do so depends upon their capacity to form a carbide on the diamond surface in which only Ti and V have been able to show higher NEA value than others. Most of these metals, however, do not form a stable monovalent bond with carbon which was solved by using an oxygen terminated diamond surface instead of a bare diamond surface. The key here is to induce stronger bonding between the metal layer and the underlying diamond surface and hence a large surface dipole. Recently Al was found to impart NEA of -1.47 eV at a layer thickness of 1 ML bare and -1.36 eV at 0.25 ML to the oxygenated surfaces of the diamond (100) [2].
Sn as termination on the bare and oxygenated surface of diamond has so far not been investigated although Sn is less electronegative than carbon and oxygen. Sn is also expected to form the carbide bonds and oxide on the surface of the diamond well. There has been a study on the same group elements (Si and Ge) termination of the diamond where they have been shown to bond well with the surface of diamond[3][4]. We wanted to study Sn as a potential NEA imparting termination on the bare and oxygenated surface of the diamond (100). DFT calculations were performed using CASTEP to model the system which contained 0.25 ML (Quarter Mono Layer), 0.5 ML (Half Monolayer) and 1 ML (Full Monolayer) Sn in different configurations on the bare and oxygenated surface of the diamond. It was found that Sn gives NEA on most of the absorption sites, largest being (-1.91 eV with an adsorption energy of -5.94 eV) in case of 0.5 ML (HML) Sn terminated oxygenated diamond surface. Sn on the bare surface gave NEA in all the configurations with lesser adsorption energies than the oxygenated diamond surface which again shows the important role of oxygen to improve the bonding between a metal atom and diamond surface and hence the properties.
References
[1] L. Maier, F. and Ristein, J. and Ley, “Electron affinity of plasma-hydrogenated and chemically oxidized diamond (100) surfaces,” Phys. Rev. B, vol. 64, no. 16, p. 165411, 2001.
[2] M. C. James, A. Croot, P. W. May, and N. L. Allan, “Negative electron affinity from aluminium on the diamond (1 0 0) surface: A theoretical study,” J. Phys. Condens. Matter, vol. 30, no. 23, p. 235002, May 2018, doi: 10.1088/1361-648X/aac041.
[3] M. J. Sear et al., “Germanium terminated (100) diamond,” J. Phys. Condens. Matter, vol. 29, no. 14, p. 145002, Apr. 2017, doi: 10.1088/1361-648X/aa57c4.
[4] A. Schenk et al., “Formation of a silicon terminated (100) diamond surface,” Appl. Phys. Lett., vol. 106, no. 19, 2015, doi: 10.1063/1.4921181.

ß-Ga₂O₃ is a wide bandgap (WBG) semiconductor material which has been extensively studied for power, optical and RF electronics due to its large bandgap of 4.7 - 4.9 eV and high breakdown field of ~ 8MV/cm. Most of the demonstrated ß-Ga₂O₃ devices are unipolar like high electron mobility transistors (HEMTs) and Schottky barrier diodes, due to the lack of p-type Ga₂O₃. To overcome this constraint, other p-type materials, such as GaN and NiO, have been studied to produce Ga₂O₃ based p-n heterojunctions. GaN is also a WBG semiconductor which attracted especial interest in power devices and has a clear crystallographic relationship to β-Ga₂O₃ for epitaxial growth. Metalorganic chemical vapor deposition (MOCVD) is an industrial standard tool which can produce both β-Ga₂O₃ and GaN. The growth of high-quality GaN on Ga₂O₃ by MOCVD has been recently reported. β-Ga₂O₃/GaN p-n heterojunction by mechanically exfoliation has also showed decent rectifying behaviors. However, due to the absence of effective edge termination the breakdown capability of the β-Ga₂O₃ based p-n heterojunctions is still in an early stage. Here we report the design and simulation of mesa edge terminated vertical β-Ga₂O₃/GaN heterojunction based p-n diode using SILVACO TCAD simulator. The heterojunction was first calibrated with conduction and valance band offsets and the device with electrical properties of the materials. The device structure consisted of n⁺-Ga₂O₃ n-contact layer, 5 µm unintentionally doped (UID) β-Ga₂O₃ drift layer and 500 nm p-GaN layer. The ideal breakdown voltage (BV) of the heterojunction was 1.37 kV, while the BV of the reference device without any mesa edge termination fall drastically to 300 V due to the electric field crowding at the device edge. Therefore, it is crucial to design and optimize the mesa edge termination of the β-Ga₂O₃/GaN p-n heterojunction to improve the BV.
In this work, namely three mesa edge termination structures: beveled mesa, step mesa, and deeply-etched mesa were investigated. The key parameter in determining the device BV for beveled mesa is the mesa angle θ, which was varied from 15° to 90° and obtained the highest BV of 1224 V for θ =15°. For the edge termination with step mesa the width (W), depth (D), and the number of steps define the mesa structure. It was found that the device BV increased with increasing number of steps. The combination of (W, D) values of (0.5,0.5) and (0.5,1.0) and (1.0,1,0) was investigated, where all the values are in µm. The highest value of BV was obtained for W = 0.5 µm, D = 1.0 µm and the number of steps = 5. It was found that smaller W and larger D gave better BV. To clearly understand the edge termination effect, electric field distribution of devices at -300 V with different edge termination mechanism were studied. Without any effective edge termination, the peak electric field at the junction edge was ~ 4.2 MV/cm. Introducing the mesa edge termination the electric field crowding was alleviated, and the peak electric fields at the edges were reduced to 1.8 MV/cm and 2.4 MV/cm for the beveled and step mesa edge terminations, respectively. Deeply-etched mesa structure was also investigated for mesa depths of 0.5, 1.0, 2.0, to 3.0 µm which showed a more relaxed uniform electric field at the junction edge. The peak electric fields were decreased from 2.1, 1.8 ,1.5 to 1.2 MV/cm, when the depth changes from 0.5 to 3.0 µm.
In brief, this work summarizes an extensive design and optimization of mesa edge terminations for the wide bandgap vertical β-Ga₂O₃/GaN p-n heterojunction diode structure. Beveled mesa, step mesa and deeply-etched mesa structures were exploited to gain knowledge on how the mesa edge terminations can mitigate the electric field crowding at the junction edge and increase the device BV. β-Ga₂O₃ based bipolar high voltage and high power devices can use this work as a guidance to improve their performance.

The use of wide-bandgap semiconductor materials like SiC and GaN can dramatically improve the performance of electronic devices at high powers and temperatures. However, the propensity of extended defects in these materials does challenge their implementation in commercial electronic and optical applications. Spectroscopic and microscopic tools for identifying and characterizing these defects typically offer either spectroscopic or microscopic information exclusively and/or may be destructive. Here, we show how extended defects within 4H-SiC manifest in the nanoscale infrared phonon response probed by scattering-type scanning near-field optical microscopy (s-SNOM), a nondestructive method capable of simultaneously collecting topographic and spectroscopic information with frequency-independent nanoscale spatial resolution (≈20 nm).

We correlate the s-SNOM response of various defects in 4H-SiC with UV-photoluminescence, secondary electron and electron channeling contrast imaging, and transmission electron microscopy. We identify evidence of step-bunching, recombination-induced stacking faults, and threading screw dislocations, and also demonstrate the interaction of surface phonon polaritons with extended defects. Phonon-enhanced infrared nanospectroscopy and spatial mapping via s-SNOM thus offer significant insights into extended defects within emerging semiconductor materials and devices. It thus serves as an important diagnostic tool to help advance material growth efforts for electronic, photonic, phononic, and quantum optical applications.

[1] B. Hauer, C. E. Marvinney, et al. Advanced Functional Materials 30, 1907357 (2020).

Radiation and particle detectors are a widely used technology for scientific, defense, environmental, industry, and other applications. However, many radiation detectors currently available on the market are bulky and many use fragile components. There is a great interest, therefore, in the development of solid state particle detectors. In this work, we will describe how we have designed, built, and tested breadboard versions of diamond diode-based particle detectors for a range of particle and radiation species, including: alpha particles, beta particles, neutrons, x-rays, and protons. This new class of radiation detectors are compact, rugged, low power, cost effective, easy to use, and reliable. Furthermore, for some species these detectors are capable of energy measurements/spectroscopy for particle species/isotope identification, and real time dosimetry. We will describe our progress in bringing these detectors to the commercial market.