Search

Characterization of defects in semiconductor wafers is essential for the development and improvement of semiconductor devices, especially power devices. X-ray topography (XRT) using synchrotron radiation is one of the powerful methods used for defect characterization. To achieve detailed characterization of large-size semiconductor wafers by synchrotron XRT, we have developed nuclear emulsion plates achieving high resolution and wide dynamic range. We have shown that higher-resolution XRT images could be obtained using emulsions with smaller iodobromide crystals and demonstrated clear observation of threading edge dislocations in a SiC epitaxial layer having small contrast. Furthermore, we demonstrated XRT image acquisition for almost all of a 150-mm SiC wafer with one plate. Our development will contribute to advances in electronic materials, especially in the field of power electronics, in which defect characterization is important for improving the performance and yield of devices.

For the nondestructive characterization of SiC wafers for power device application, birefringence imaging is one of the promising methods. In the present study, it is demonstrated that birefringence image contrast variation in off-axis SiC wafers corresponds to the in-plane shear stress under conditions slightly deviating from crossed Nicols according to both theoretical consideration and experimental observation. The current results indicate that the characterization of defects in SiC wafers is possible to achieve by birefringence imaging.

The complete automation of materials manufacturing with high productivity is a key problem in some materials processing. In floating zone (FZ) crystal growth, which is a manufacturing process for semiconductor wafers such as silicon, an operator adaptively controls the input parameters in accordance with the state of the crystal growth process. Since the operation dynamics of FZ crystal growth are complicated, automation is often difficult, and usually the process is manually controlled. Here we demonstrate automated control of FZ crystal growth by reinforcement learning using the dynamics predicted by Gaussian mixture modeling (GMM) from small numbers of trajectories. Our proposed method of constructing the control model is completely data-driven. Using an emulator program for FZ crystal growth, we show that the control model constructed by our proposed model can more accurately follow the ideal growth trajectory than demonstration trajectories created by human operation. Furthermore, we reveal that policy optimization near the demonstration trajectories realizes accurate control following the ideal trajectory.

In this study, an algorithm was constructed for the automatic detection of basal plane dislocations (BPDs) propagating in SiC epitaxial layers in photoluminescence images, and its performance was evaluated. The BPDs are the origin of the degradation of SiC bipolar devices caused by the expansion of stacking faults. The present automatic detection algorithm, based on the template-matching method, was confirmed to have high accuracy and precision, and we succeeded in visualizing the BPD density in 150-mm SiC epitaxial wafers. We confirmed that the template-matching method is applicable for the detection of crystalline defects with geometrically fixed shapes such as BPDs in SiC epitaxial layers.

4H-SiC has been commercialized as a material for power semiconductor devices. However, the long-term reliability of 4H-SiC devices is a barrier to their widespread application, and the most important reliability issue in 4H-SiC devices is bipolar degradation. This degradation is caused by the expansion of single Shockley stacking-faults (1SSFs) from basal plane dislocations in the 4H-SiC crystal. Here, we present a method for suppressing the 1SSF expansion by proton implantation on a 4H-SiC epitaxial wafer. PiN diodes fabricated on a proton-implanted wafer show current–voltage characteristics similar to those of PiN diodes without proton implantation. In contrast, the expansion of 1SSFs is effectively suppressed in PiN diodes with proton implantation. Therefore, proton implantation into 4H-SiC epitaxial wafers is an effective method for suppressing bipolar degradation in 4H-SiC power-semiconductor devices while maintaining device performance. This result contributes to the development of highly reliable 4H-SiC devices.

To design a time-dependent control recipe which can ensure consistently suitable growth conditions in an unsteady growth system with dynamic environmental changes, an adaptive control method based on high-speed machine learning prediction models was proposed and applied to the solution growth of SiC crystals. This approach comprised three parts, namely, a quasi-unsteady computational fluid dynamics (CFD) model for thermal and flow field simulation, machine learning models for approximating the simulation results and giving instant prediction, and an optimization algorithm for searching the optimal growth conditions. First, the evolution of the flow, temperature and carbon concentration fields over 50 h unsteady growth following an original recipe with fixed control parameters was analyzed by CFD simulation. Then, adaptive control was applied to design a time-dependent growth process with a 100-timestep sequence. The hybrid of machine learning models and CFD simulation accelerated the entire design and optimization process by 300 times, compared with CFD simulations alone. The adaptive control facilitated superior performance compared with the fixed recipe, where the single SiC crystal thickness increased by ∼30% and the growth interface was more uniform. Further, crucible dissolution and polycrystal precipitation were suppressed by ∼50%, enabling longer growth time and more stable growth. It is the first time that the importance of adaptive control during long-term SiC solution growth is discussed, and the method proposed in this study demonstrated the potential for real-time optimization in the future.

In the design of a crystal growth system, the ability to efficiently regulate intertwined geometrical parameters is crucial for its successful development and commercialization. However, the traditional experimental and computational methods consume tremendous amounts of time and resources. To address this problem, a machine learning approach was developed in this study to accelerate the geometry optimization process. It was found that the combination of machine learning with a genetic algorithm could generate various possible solutions through a global search at a relatively high speed, which lie outside the solution range of the experimental optimization methods that are currently used. By applying this technique, an optimal geometrical design was obtained for a 150 mm top-seed solution growth system, indicating that the proposed method represents an innovative and attractive strategy for the development of crystal growth systems with superior characteristics.

Li metal anodes are plagued by low coulombic efficiency due to their interfacial instability. Many approaches were proposed to cope with this problem; however, little attention has been given to the current collector of Li anodes. In this study, we investigate the crystal orientation dependence of the cycling stability of Li anodes on single-crystal Cu(111), (101), and (001) and polycrystalline Cu current collectors. Cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) show that (111) and (001) achieved high current efficiency and low interfacial resistance, while (101) and polycrystalline Cu exhibited low cyclabilities. X-ray photoelectron spectroscopy (XPS) and auger electron spectroscopy (AES) analysis revealed that the thickness of the solid electrolyte interphase (SEI) varies with the Cu crystal orientation, and the SEI is the thinnest on the single-crystal Cu(111). This tendency can be explained by the orientation dependence of the surface energy of Cu, which corresponds to the chemical activity of the surfaces. Our result advocates the importance of considering Cu orientation for interfacial engineering of Li metal anodes.

Thermal conduction was investigated in titanium-chromium oxide natural superlattices with an ordered arrangement of crystallographic shear planes, which were nearly pristine interfaces for thermal phonons. Thermal conductivity exhibited a clear maximum value at around 50 K, which indicates interface roughness as small as 20 pm based on the calculation of thermal conductivity by a modified Debye-Callaway model. With an ordered arrangement of nearly pristine interfaces at nanoscale, the wave contribution to phonon transport was estimated to be predominant at lower temperatures (less than 100 K) and as much as 30 % even at room temperature. We found that the interface roughness greatly influences the thermal conduction in natural superlattice systems, which may lead to variations in the reported value of thermal conductivity. We have demonstrated that the wave contribution of thermal conductivity become apparent in bulk crystal having nano scale periodic structure with pristine interfaces.

Bipolar degradation in SiC bipolar devices, in which stacking faults (SFs) expand to accommodate the movement of partial dislocations during forward bias application, is one of the critical problems impeding the widespread implementation of SiC power devices. Here we clearly demonstrate that the movement of partial dislocations can be suppressed by proton implantation, which has good compatibility with semiconductor processing, through investigation of the contraction behavior of SFs in SiC epitaxial layers subjected to proton implantation.

The resolution of spectroscopy, which delivers valuable insights and knowledge in various research fields, has sometimes been limited by the number of multi-channel detectors employed. For example, in Raman spectroscopy using charge coupled device (CCD) detectors, the resolution is limited by the number of the CCD arrays and it is difficult to achieve spectroscopic data acquisition with high resolution over a wide range. Here we describe a methodology to increase the resolution as well as signal-to-noise (S/N) ratio by applying Bayesian super-resolution in the analysis of spectroscopic data. In our present method, first the hyperparameters for the Bayesian super-resolution are determined by a virtual experiment imitating actual experimental data, and the precision of the super-resolution reconstruction is confirmed by the calculation of errors from the ideal values. For validation of the super-resolution of spectroscopic data, we applied this method to the analysis of Raman spectra. From 200 Raman spectra of a reference Si substrate with a resolution of about 0.8 cm− 1, super-resolution reconstruction with resolution of 0.01 cm− 1 was successfully achieved with the promised precision. From the super-resolution spectrum, the Raman scattering peak of the reference Si substrate was estimated as 520.55 (+ 0.12, -0.09) cm− 1, which is comparable to the precisely determined value from previous works. The present methodology can be applied to various kinds of spectroscopic analysis, leading to increased precision in the analysis of spectroscopic data and the ability to detect slight differences in spectral peak positions and shapes.

The conversion of threading screw dislocations (TSDs) to defects on basal planes during SiC solution growth caused by advancing macrosteps is a key factor for high-quality crystal growth. To study this effect for large growth areas, a 2 inch 4H-SiC crystal was grown on the C face of a 1° off-axis seed crystal using the top-seeded solution growth method. Step-flow growth towards the [110] direction occurred over the entire growth surface. Inhomogeneities in the surface morphology were observed. The step height increased monotonically along the step-flow direction. The non-uniform step height resulted in a spatial distribution of TSD conversion. Little conversion took place in the upstream area of step flow with small steps. In contrast, conversion happened frequently in the centre of the crystal and in the downstream area, with a conversion ratio as high as 80% in these areas. According to a numerical simulation, the solution flow direction was outward near the growth surface. The solution flow affected the steps adversely in the upstream and downstream areas, resulting in a spatial distribution of step height and TSD conversion ratio. This study revealed that the solution flow direction is a key factor for the growth of large high-quality crystals.

Recently, quality of free-standing gallium nitride (GaN) substrate has been improved. For example, single crystal GaN substrate with threading dislocation density (TDD) of 104 cm-2 is available. Meanwhile, there is a progress in observation method of dislocation in GaN. In addition to conventional etch pits method, X-ray topography, and CL observation, recently, multiphoton PL microscope which can observe dislocation three-dimensionally is actively being carried out. However, these methods are not simple observation because these are destructive observation or require a lot of labor and time. In this study, we attempted crystal evaluation of GaN with a newly developed birefringence microscope, which can be used with the same convenience as the conventional optical microscope. GaN is a material having birefringence, so that strain around dislocation can be seen with birefringence microscope as retardation[1]. Therefore, we tried observing commercially available GaN substrate with TDD of 106, 105, 104 cm-2. As results, an useless random contrast image was obtained with the substrate with TDD of 106 cm-2. With the substrate with TDD of 105 cm-2, distinctive contrasts that seem to coincide with the dislocation cores were obtained, but it was an unclear image. On the other hand, with the substrate with TDD of 104 cm-2, we could obtain contrasts that seems to be independent dislocation cores and surrounding strain field. It seems that the strain field of each dislocation overlapped because the distance between dislocations was too close in substrate with TDD of 106 and 105 cm-2. For the substrate with TDD of 104 cm-2, comparison with the X-ray topography image taken at BL8S2 of Aichi SR was carried out, and it was confirmed that the dislocation was exist at the position of the contrast in the image taken by birefringence microscope. However, we also found that there is no contrast at some dislocation positions. This is considered to be caused by that no strain occurs in the planar direction when the threading dislocation is pure screw dislocation. In some of the dislocations, butterfly type contrast which is theoretically expected to appear around edge dislocation was observed. We consider that this novel birefringent method is very useful method to discriminate edge component and its direction of Burgers vector, and also to discriminate screw dislocation in combination with other methods.

The contrast features of synchrotron X-ray topographic images of screw-type basal plane dislocations (BPDs) in on-axis 4H-SiC wafers have been studied. Screw BPD images are categorized into two types: one exhibiting a white line bordered by black lines and the other a pure black line contrast. Similar images for off-axis specimens and the corresponding ray-tracing simulations demonstrate that these contrasts can be attributed to the depth of the screw BPDs below the crystal surface. The correlation of the contrast features between simulations and the screw BPD topography images can be used to estimate the depth.

We observed the behavior of double Shockley stacking faults (DSFs) in 4H-SiC crystals with nitrogen concentrations of 1.0×1019-2.6×1019 cm−3 over an extensive temperature range (1380-1910 K) by in-situ synchrotron X-ray topography. For a nitrogen concentration of 2.6×1019 cm−3, the expansion velocity of the DSFs exponentially increased with temperature for temperatures in the range from 1370 to 1650 K. In contrast, at a nitrogen concentration of 1.0×1019 cm−3, this velocity decreased above 1610 K and the DSFs shrank above 1730 K. The DSF energy, considered to be the driving force for DSF expansion, was quantitatively estimated from the radius of the curvature of bowed-out partial dislocations (PDs) pinned by threading screw dislocations (TSDs), showing a positive temperature dependence and lying in the range from −0.6 to 0.8 mJ/m2 at a nitrogen concentration of 1.0×1019 cm−3. The DSF expansion and shrinkage behavior can be understood by the simple temperature and nitrogen concentration dependence of the DSF energy.

Evaporation of the volatile solute from the liquid phase is a common problem for the crystal grown from solution, especially for the growth under high temperature. In the present study, a numerical model was constructed to quantitatively investigate the evaporation process in crystal growth. This model was applied in top-seeded solution growth (TSSG) of SiC crystal to simulate the transport of aluminum (Al), which is important for crystal surface morphology but easy to vaporize. The transport path of Al in the growth system was determined by analyzing the possible reactions on different boundaries. Accordingly, an improved structure was proposed to suppress the evaporation loss of Al during long-term growth, and was compared with the original case both numerically and experimentally. The simulation results showed that the improved structure could effectively decrease the Al loss by over 70%, and meanwhile had almost no influence on the thermal and flow environment in the solution. For the experimental results, the improved case presented much lower spontaneous nucleation possibilities and higher step height on the crystal surface, which matched well with the features of high Al addition in literatures. Therefore, the improved structure proposed in the present study was proven to be effective to enhance the composition stability of solution during long-term SiC solution growth, and this numerical method could be applied in the growth of other crystals facing the similar problem.

We have investigated thermal conduction in bulk titanium oxide natural superlattices with crystallographic shear (CS) structures, in which dense planar faults are introduced with different periodicities, prepared by reductive annealing of rutile TiO2 and crystal growth by the floating zone method. High-angle annular dark-field (HAADF) scanning transmission electron microscopy (STEM) revealed that (132)rutile and (121)rutile CS planes with interspacings of 2.7 and 1.0 nm were introduced in the mother rutile structure. Time-domain thermoreflectance (TDTR) revealed that the thermal conductivity decreased by the introduction of CS planes, but that the decrease is not monotonic with increasing density of CS planes. Calculation of the thermal conductivity and the mean free path for phonons revealed that a crossover from incoherent to coherent thermal conduction took place, and coherent interfaces with nanoscale periodicity were formed as thermodynamically stable phases in bulk titanium oxide natural superlattices.

For solution growth of silicon carbide, it is significant to understanding the evolutionary mechanism of step bunching. This study infers that solute’s incorporation into steps and transport together determines step bunching progress. The occurrence of step bunching is due to the depletion of solute in the region with high step density, caused by a high step kinetic coefficient. On the other hand, by promoting the transport of the solute in the solution, the step speed becomes uniform, thereby the step bunching can be prevented. Furthermore, we proposed a non-dimensional Damkohler number for crystal growth in step-flow mode. It correlates incorporation rates with bulk diffusion rates and can build a phase map of growth rates and step bunching stability. Several solvents are located in the phase map, demonstrating the possible usage of the phase map as a pointer for solvent designing.

The improvement mechanism of sputtered AlN films by high temperature annealing in nitrogen ambient has been investigated. Sputtered AlN films were annealed at 1100–1700 °C and their microstructures were observed by scanning transmission electron microscopy. A two-layer structure consisting of columnar domains with different sizes was observed in the as-grown sputtered AlN films. The first layer with 10 nm thickness located at the AlN/sapphire interface, comprised columnar domains with diameters of nm order and was compressively strained owing to the lattice mismatch between AlN and sapphire. The diameter of columnar domains in the other layer was approximately 25 nm. The columnar domains split into irregularly shaped and coalesced at an elevated annealing temperature, resulting in improved crystal quality. When the annealing temperature was 1700 °C, the domain boundaries in AlN films were almost annihilated and the full width at half maximum of the (0 0 0 2)- and (1 0 −1 2)-plane X-ray rocking curves were improved to as low as 49 and 310 arcsec, respectively. The polarity switched from N-polar to Al-polar after about 4–10 AlN layers in the growth direction. The oxygen element content slightly increased at the polarity inversion boundary, which may have caused the polarity to switch.

We have found that large chiral symmetry breaking in chiral crystallization can be achieved by irradiating a several milliwatts focused laser to a plasmonic nanolattice immersed in a stagnant NaClO3 saturated aqueous solution. Several hundreds of chiral crystals with the same handedness showed up in the solution after the laser irradiation in contrast to spontaneous crystallization. In situ microscopic observation for the early stage of the crystallization in the vicinity of the focal spot revealed that microbubble generation followed by large supersaturation increase, in which supersaturation reaches 360%, promotes several numbers of crystal nucleation in the vicinity of the bubble as “mother” crystal. The generation of the microbubble induced Marangoni convection, the velocity of which reaches several hundreds of micrometers per second, crushing the first appearing chiral crystal into pieces by microfluidic shear. Namely, secondary nucleation caused by microfluidic shear amplified the number of “daughte...

The dislocation behavior during bulk crystal growth on the 4H-SiC (000-1) C-face using the solution method was investigated. A 2 inch wafer with a 4° off angle was fabricated from a bulk crystal grown by the TSSG method, and the dislocations in the crystal were evaluated using synchrotron X-ray topography and TEM observation. From the topograph images, it was found that the TSD density remarkably decreased as the growth progressed. Furthermore, the TEM observation suggested that TSD decreases as the threading dislocations convert to in-plane defects toward the center of the crystal. Conventionally, it was considered that conversion of threading dislocations hardly occurs in solution growth on the C-face. However, it is thought that this phenomenon was not observable because the conversion efficiency is remarkably low. We speculate that dislocations may be transformed by suddenly making macrosteps during bulk growth.

We achieved the growth of extremely-high quality SiC crystal with two-step solution method with specially-designed seed crystals. The two-step growth consists of 1st step growth on Si-face for the reduction of threading dislocations and 2nd step growth on C-face for the reduction of basal plane dislocations and thickening. In this method, we can make the dislocation density extremely low, while the polytype easily changes during growth due to the absence of spiral hillocks originating from threading screw dislocation (TSD). In this study, we prepared specially designed seed crystals for both 1st and 2nd growth steps to provide steps continuously. In the seeds, a few TSDs exist at the upper-side of the step structure. Consequently, we demonstrated the suppression of the polytype transformation during the C-face growth with extremely low-dislocation-density crystal. Accordingly, we successfully obtained extremely low-dislocation density 4H-SiC with TSD, TED and BPD density of 11, 385 and 28 cm-2.

Accelerating the optimization of material processing is essential for rapid prototyping of advanced materials to achieve practical applications. High-quality and large-diameter semiconductor crystals improve the performance, reliability and cost efficiency of semiconductor devices. However, much time is required to optimize the growth conditions and obtain a superior semiconductor crystal. Here, we demonstrate a rapid prediction of the results of computational fluid dynamics (CFD) simulations for SiC solution growth using a neural network for optimization of the growth conditions. The prediction speed was 107 times faster than that of a single CFD simulation. The combination of the CFD simulation and machine learning thus makes it possible to determine optimized parameters for high-quality and large-diameter crystals. Such a simulation is therefore expected to become the technology employed for the design and control of crystal growth processes. The method proposed in this study will also be useful for simulations of other processes.

The expansion of double Shockley stacking faults (DSFs) in an n-type 4H-SiC substrate with a nitrogen concentration of 3.9 × 1019 cm−3 was investigated using in situ synchrotron X-ray topography. DSF expansion was observed to be suppressed and immobilized above 1590 K, along with the partial dislocation (PD) shape being changed from a straight to zig-zag configuration. For a different heating process (higher heating rate), the PDs could continue to expand, even above 1590 K. Ex situ topography experiments revealed that the DSFs close to the specimen surface expanded widely, although those expanding toward the specimen interior became immobile. One possible mechanism for this immobilization was proposed, where the core structural changes from a Si-core to the C-core by non-conservative motion induced by the interaction between the PDs and point defects (C interstitials).

A more sophisticated simulation model is developed based on the principle of ray-tracing to simulate the grazing-incidence synchrotron X-ray topographic contrast of dislocations lying on the basal plane including basal plane dislocations and deflected threading screw and mixed dislocations in off-axis 4H-SiC crystals. The model incorporates effects of surface relaxation as well as the photoelectric absorption to predict dislocation contrast. Compared to conventional ray-tracing images, surface relaxation effects dominate dislocation contrast for diffraction near the crystal surface. The simulated dislocation contrast gradually weakens with increasing depth of the diffracted beam position within the crystal due to photoelectric absorption. The distinctive features of the net simulated dislocation images obtained by aggregating through the effective penetration depth correlate well with contrast features observed on the experimental topographic images. Depth analysis reveals that in some cases the diffracted X-rays from regions below the dislocation can contribute additional contrast features previously not considered.

We have investigated the morphology of AlN whiskers on a polycrystalline AlN substrate by using Fe-Al alloy melts under the different synthesis conditions. Formation density of the AlN whiskers increases and the diameter of the whisker decreases with increasing Al content of the Fe-Al alloy melt. Most of the AlN whiskers were zigzag shape with the hexagonal cross section. The longitudinal direction was the [0001] direction. The facet with the zigzag shape was the {1 1 ¯ 01} or {1 1 ¯ 0 1 ¯ } pyramidal plane and the period of the zigzag facet was almost constant. The average diameter of the whiskers and the period of the zigzag facet decrease with increasing Al content. At the initial stage of the whisker formation, the island of AlN formed on the AlN substrate and the pyramidal facet grows via step-flow growth. From the observation, we discussed the possible mechanism for the formation of the zigzag-shape AlN whiskers.

We propose a two-step silicon carbide (SiC) solution growth method for dislocation reduction to produce the high-quality silicon carbide SiC crystals. The two-step growth consists of the growth on a Si face and a C face. Firstly, seed crystal with low threading dislocation density was prepared by the growth on a Si face utilizing the threading dislocation conversion. Secondly, the growth on the C face was conducted on the prepared seed crystal with low threading dislocation density to reduce the density of basal plane dislocations and keep smooth growth surface. We demonstrate that the two-step growth leads to the reduction of the density for all types of dislocations by two orders of magnitude compared to the initial density of the seed crystal.

We report the results of microscope measurements examining the phase transition process of a cationic lipid, Dimethyldioctadecylammonium bromide (DDAB) supported lipid bilayer (SLB). Due to lateral fluidity and strong electrostatic interaction with DNA, SLB serves as a fluid substrate for assembling 2D lattices of DNA functionalized nanoparticles (DNA-NPs): lipid molecules work as carriers for transporting DNA-NPs. By fluorescence microscopy and atomic force microscopy (AFM), two types of phase transitions, which correspond to liquid crystalline-gel and liquid crystalline-interdigitated gel (L β I) ones, were observed in DDAB SLB during cooling. In thermal equilibrium at room temperature both gel and L β I phases have enough adsorbed amounts of DNA-NPs which indicate that both domains have enough surface charge densities for adsorbing DNA-NPs, however, during phase transition DNA-NPs preferably distributed into L β I phase.

In order to design a solvent for high-purity SiC solution growth, the impurity incorporation and the carbon solubility of various solvent materials have been investigated. Among the transition metal elements, the impurity elements of Cr, Ti, V and Hf are more readily incorporate during the solution growth than the other transition metal elements. The thermodynamic calculation revealed that the Y-Si solvent has relatively large carbon solubility, which is comparable to the Cr-Si and Ti-Si solvents often used in the solution growth of bulk SiC crystals. From these results, the Y-Si solvent is expected to be a suitable solvent for the high-purity SiC solution growth. Furthermore, we have demonstrated that the Y-Si solvent can achieve lower incorporation of metal impurity in the grown crystal than the Cr-Si solvent maintaining the growth rate.