In this review, we introduce our recent applications of deep learning to solar and space weather data. We have successfully applied novel deep learning methods to the following applications: (1) generation of solar farside/backside magnetograms and global field extrapolation based on them, (2) generation of solar UV/EUV images from other UV/EUV images and magnetograms, (3) denoising solar magnetograms using supervised learning, (4) generation of UV/EUV images and magnetograms from Galileo sunspot drawings, (5) improvement of global IRI TEC maps using IGS TEC ones, (6) one-day forecasting of global TEC maps through image translation, (7) generation of high-resolution magnetograms from Ca II K images, (8) super-resolution of solar magnetograms, (9) flare classification by CNN and visual explanation by attribution methods, and (10) forecasting GOES solar X-ray profiles. We present major results and discuss them. We also present future plans for integrated space weather models based on deep learning.

This report summarizes a recent study demonstrating simple and rapid synthesis of a new Al–Mg alloy system and ultimately synthesizing a metal matrix nanocomposite, which was achieved by processing stacked disks of the two dissimilar metals by conventional high-pressure torsion (HPT) processing. The synthesized Al–Mg alloy system exhibits exceptionally high hardness through rapid diffusion bonding and simultaneous nucleation of intermetallic phases with increased numbers of HPT turns through 20, and improved plasticity was demonstrated by increasing strain rate sensitivity in the alloy system after post-deformation annealing. An additional experiment demonstrated that the alternate stacking of high numbers of dissimilar metal disks may produce a faster metal mixture during HPT. Metal combinations of Al–Cu, Al–Fe, and Al–Ti were processed by the same HPT procedure from separate pure metals to examine the feasibility of the processing technique. The microstructural analysis confirmed the capability of HPT for the formation of heterostructures across the disk diameters in these processed alloy systems. The HPT processing demonstrates a considerable potential for the joining and bonding of dissimilar metals at room temperature and the expeditious fabrication of a wide range of new metal systems.

This study quantitatively investigated the immediate effects of a photooxidative collagen cross-linking treatment with photosensitizer riboflavin (RF) and 370 nm UVA light in in vitro human corneoscleral collagen fibrils using histology, thickness, scanning electron microscopy, and atomic force microscopy analyses. Twenty 8 × 2 mm corneoscleral strips were dissected sagittally from donor tissue using a scalpel. Four parameters were investigated, including the density, thickness, adhesion force, and stiffness of corneoscleral tissues before and after the collagen cross-linking treatment. The RFUVA-catalyzed collagen cross-linking treatment led to an increase in the density of both corneal (8%) and scleral (23%) stromal collagens. However, there was no difference in corneoscleral thickness. Furthermore, RFUVA-catalyzed collagen cross-linking treatment led to an increased biomechanical response of corneosclera: 25 and 8% increases in corneoscleral stiffness, and 24 and 22% increases in corneoscleral adhesion force. The collagen cross-linking treatment through RF-sensitized photoreaction may cause structural and biomechanical changes in the collagen fibril network of the cornea and the sclera. This is due to narrowing of the interfibrillar spacing and the stromal edema.

This study investigated the inflammatory effect of intraoperative mitomycin C (MMC) on adhesion reformation in human rectus muscles. Ten consecutive patients who underwent medial rectus resection had their postoperative rectus muscles divided into two groups: control group (n = 10) and MMC group (n = 10). In the MMC group, the muscle was soaked for 2 min with MMC, prepared as a 0.2 mg/mL (0.02%) solution. The 0.02% MMC reactions were examined using histological analysis with hematoxylin-eosin (inflammatory response) and Masson's trichrome (collagen fibrils), immunoreactivities of cyclooxygenase-II (inflammatory response), and collagen type I and III, scanning electron microscopy analysis to quantify the diameter and D-periodicity of collagen fibrils, and atomic force microscopy analysis to quantify the diameter, D-periodicity, and adhesion force of collagen fibrils. The rectus muscles treated with 0.02% MMC showed a significantly increased inflammatory response (p < 0.05), increased collagen density (p < 0.0001), increased fibril diameter (p < 0.001 or p < 0.05), and decreased fibril adhesion force (p < 0.005) compared to the rectus muscles in the control group. MMC simultaneously caused an inflammatory response as well as nanostructural and biomechanical property changes in the collagen fibril network.

The preparation of BaMgAl10O17:Eu2+ (BAM) blue phosphor layer for a plasma display panel by the addition of a newly designed energetic dispersant (hereafter referred to as SAIT7) and its resulting photoluminescence (PL) efficiency is investigated. The addition of SAIT7 increases the maximum solids loading of BAM phosphor paste from 27.91 to 35.87 vol% and yields a highly packed microstructure of the phosphor layer. The PL intensity is increased by 7.57% compared to the BAM blue phosphor layer prepared without SAIT7. In conclusion, the addition of SAIT7 increased the packing density of the phosphor layer and resulted in improved luminescent properties of the phosphor layer.

Effect of SiO2 capping layer(C/L) on recrystallization of amorphous Si (a-Si) film was investigated. When a thick C/L over 500 Å was deposited on an a-Si film before crystallization, fine p-Si grains less than 100nm were obtained at full range of energy density window. However, when a thin C/L below 200 Å was used, Si-melt spouted out through C/L at over critical energy density. When Si-melt started spouting, abrupt change of grain size also occurred. These large grains could be explained by a non-uniformity of heat flow caused by Si-melt spouting. With this polycrystalline Si (p-Si) material having appreciable grain size protected by C/L, fabrication of low-cost Low Temperature Poly Si (LTPS) without additional cleaning of p-Si surface could be successfully developed.

Semi-insulating undoped GaN films were grown based on controlling the size of the nucleation sites through a special two-step growth method: First, 16 nm LT-GaN was annealed at 950 ° with a ramping time of 4 min, then the GaN was grown at this temperature for 1 min. Second, the growth temperature was increased to 1020° with a ramping time of 2 min and the GaN layer finally grown at 1020 ° for 40 min. The film grown by this sequence exhibited sheet resistance of up to 109 Ω/sq with mirror-like surface morphology. By slow ramping to 950° in the initial phase of growth, smaller grain sizes and higher nuclei densities were formed and the columnar growth mode along the c direction was dominant. The observation of higher resistance in two-step growth is believed due to the increased misorientation of nuclei when the growth proceeds during temperature ramping to 1020°. The fabricated saw filter on semi-insulating GaN exhibited a high velocity of 5342 m/s at center frequencies of 133.57 MHz and an electromechanical coupling coefficient(k2) of about 0.763 %, which was enhanced due to the improvement of surface morphology with high sheet resistance by the two- step ramping technique.

We have studied infrared photoluminescence (PL) and x-ray diffraction (XRD) of 400 nm and 1500 nm thick InAs epilayers on GaAs, and 4 nm thick InAs on graded InGaAs layer with total thickness of 300 nm grown by molecular beam epitaxy. The PL peak positions of 400 nm, 1500 nm and 4 nm InAs epilayer measured at 10 K are blue-shifted from that of InAs bulk by 6.5, 4.5, and 6 meV, respectively, which can be largely explained by the residual strain in the epilayer. The residual strain caused by the lattice mismatch between InAs and GaAs or graded InGaAs/GaAs was observed from XRD measurements. While the PL peak position of 400 nm thick InAs layer is linearly shifted toward higher energy with increase in excitation intensity ranging from 10 to 140 mW, those of 4 nm InAs epilayer on InGaAs and 1500 nm InAs layer on GaAs is gradually blue-shifted and then, saturated above a power of 75 mW. These results suggest that adopting a graded InGaAs layer between InAs and GaAs can efficiently reduce the strain due to lattice mismatch in the structure of InAs/GaAs.

Si delta-doping in the GaN layer has been successfully demonstrated by low-pressure metalorganic chemical vapor deposition at a growth temperature of 1040 . Si delta-doping concentration increases and then decreases with an increase in delta-doping time. This indicates that delta-doping concentration is limited by the desorption process owing to much higher thermal decomposition efficiency of silane at high growth temperatures of GaN. In addition, it was observed that the use of a post-purge step in the ammonia ambient reduces Si delta-doping concentration. From capacitance-voltage measurement, a sharp carrier concentration profile with a full-width at half maximum of 4.1 nm has been achieved with a high peak concentration of 9.8 1018 cm−3.

Si delta-doping in the GaN layer has been successfully demonstrated by low-pressure metalorganic chemical vapor deposition at a growth temperature of 1040°C. Si delta-doping concentration increases and then decreases with an increase in delta-doping time. This indicates that delta-doping concentration is limited by the desorption process owing to much higher thermal decomposition efficiency of silane at high growth temperatures of GaN. In addition, it was observed that the use of a post-purge step in the ammonia ambient reduces Si delta-doping concentration. From capacitance-voltage measurement, a sharp carrier concentration profile with a full-width at half maximum of 4.1 nm has been achieved with a high peak concentration of 9.8 x 1018 cm-3