56016 research outputs found
Sort by
????????? ??? ???????????? ??? ????????? ????????? ????????? ????????? ???????????? ?????? ????????? ?????? ?????? ??????
No full textDepartment of Nuclear EngineeringZirconium alloys have been highly utilized as one of the major materials in nuclear industry due to great mechanical property and low thermal neutron absorption cross-section of zirconium. Various types of nuclear fuel cladding and the structural materials in nuclear power plants are manufactured by Zr alloys: Zircaloy-4 and Zr-Nb alloy (HANA, Zirlo etc.) for fuel cladding in a pressurized water reactor (PWR), and Zr-2.5Nb for pressure tube and Zircaloy-2 for Calandria tube in Canada Deuterium Uranium (CANDU) type reactors. However, the radiological activation products can be a great concern for lower radioactive level and volume reduction of total irradiated Zr-alloy waste. Because they make high radioactivity with small mass of activation products, such as Nb-94, Co-60, Ni-63, etc. The main objective of this dissertation is the development of advanced Zr recovery method from irradiated Zr-alloy wastes. If the highly purified Zr can be retrieved from irradiated Zr-alloys, the waste classification of retrieved Zr can be classified as LLW. Therefore, it can be used for conditional recycling in the nuclear industry or disposed into disposal facility in ROK because of the weak beta-ray by Zr-93. Also, the volume of an intermediate-level waste (ILW) would be significantly reduced.
For volumetric decontamination of Zr-alloy, various methods, e.g., halogenation and electrorefining, have been developed. There are two halogenation methods which are simple and have high efficiency using iodine and chlorine, representatively. Also, several electrorefining technologies based on the types of molten salts have been proposed. Among them, halogenation methods produce a lot of secondary wastes and issues for operation costs. Electrorefining technology would be a good option due to the compact system and very low secondary waste theoretically. Also, the chloride based electrorefining technology seems to be more promising than fluoride or fluoride-chloride based electrorefining containing inherent characteristics of salts, high operation temperature and fluoride corrosion. However, there are several obstacles in previous chloride based electrorefining aimed to directly recover Zr metal from the molten salt system. Electrochemical behavior of Zr in chloride salt is very complicate with four valence states. It makes co-deposition of Zr metal and ZrCl on the cathode. Also, electrodeposited Zr metal has powdery and sparse morphology causing the detachment from the cathode readily. It can lower the process efficiency and require additional process to collect the detached Zr metal.
In this dissertation, a two-step Zr metal recovery method combining electrochemical and pyrolysis method using ZrCl as medium was designed and experimentally verified. The electrolysis of ZrCl at 723 K applying -1.1 V (vs. Ag/AgCl) on the cathode was performed to investigate the morphology of ZrCl and produce ZrCl. The ZrCl has a flat-layered morphology, which has a good adhesion quality. It was also confirmed that a single chemical form, ZrCl, can be obtained as electrodeposits. Acquired ZrCl was thermally decomposed at various conditions into Zr metal. Zr metal was successfully achieved in 1,073 K and 1.7-1.8 torr conditions with more than 24 hours of operation time, and 1,373 K and 1.7-1.8 torr conditions with more than 12 hours, with high salt distillation effect as well. By analyzing produced Zr metals by crystallographic analysis, the mechanism of thermal decomposition was revealed, 4ZrCl???ZrCl4 + 3HCP-Zr??? FCC-Zr. The phase transition of Zr metal was also observed during the thermal decomposition reaction, HCP-Zr to FCC-Zr.
The anodic dissolution response in electrochemical system was also investigated to better understand the electrorefining process in long term operation. The anode potential was maintained at the beginning of electrorefining between -0.9 to -0.6 V. In this region, Zr can be a selectively dissolved into the molten salts. Zr dissolution mechanism at the early stage of electrorefining was explained by percolation mechanism. However, there was dramatic increase of anode potential, where the co-dissolution of Zr and Nb can occur. This region is called the passivation region, where passivation products hinder the dissolution of alloy elements. In passivation region, the passivation products of Zr and Nb, K2ZrCl6 and K3NbCl6, was observed. Also, another by-products, Nb metal and NbZr alloy were also observed in high viscosity salt layer. After the passivation region, the anode potential was recovered in the similar potential range of early stage. It was referred that the continuous dissolution of alloy element makes the detachment of high viscosity salt layer, thus, base alloy can be exposed to the molten salts.
In conclusion, a new concept of two-step Zr recovery process was developed and verified including electrochemical response on anode. It is expected that the process can be further developed and applied to recover Zr metal in various electrochemical system, such as, electrorefining of irradiated Zr-alloy wastes and spent metal fuel.clos
Dipolar effect in Janus colloidal crystal and membrane encapsulating dense bacteria
No full textDepartment of PhysicsColloid has been employed to emulate atoms in materials for decades and was applied to understand various phenomena such as nucleation, melting process, etc. It successfully proved the fundamental difference in the melting process between two dimensions and three dimensions and gave an insight into the kinetics of nucleation because we can observe them at the single-particle level. However, for several decades, they have mainly focused on the translational degree of freedom, which is just a single factor to describe real materials. For example, components of polar fluids molecules retain dipoles individually and this is critical in nature. For example, water molecules??? dipoles induce a distinct solid structure in ice, and it makes the ice less dense than water. In addition, hydrogen fluoride has strong dipolar interaction between them, so they behave like a chain even in liquid. Therefore, to simulate components of materials more approximately, the rotational degree of freedom also should be considered as important as the translational degree of freedom.
To compensate for the above problems in colloidal physics, we studied how to add a rotational degree of freedom to the experimental system. For this purpose, we selected the Janus sphere, whose hemispheres have different electric and optical properties, to represent the dipolar interaction between them. The interaction is controlled by the external electric field. For the first project, we studied their rotational degree of freedom in Janus colloidal crystal, which is translationally ordered in hexagonal lattices, to separate it with a translational degree of freedom. In the study, we find that the three-dimensional orientational arrangement on the Janus colloidal crystal is dependent on the applied electric field. Moreover, we discovered a phase transition exists across the electric field, which means orientations of Janus spheres play some roles to determine their phases in the Janus colloidal crystal. The result reminds us of the importance of the rotational degree of freedom when it comes to defining material phases.
For the next step, we did another experiment to study the consequences of the coupling between the rotational and translational degree of freedom. We built a system where Janus colloids are transformed from liquid to solid, which is like the nucleation process, using illumination. In the experiment, we observed characteristics of the interface between solid and liquid, which is unpredicted by the previous experiment using isotropic particles. Furthermore, it induced different growth mechanisms of crystals from the prediction of conventional nucleation theory. Based on these results, we expect the coupling of translational and rotational degrees of freedom make the previously known observations without considering components??? orientations much more complicated.
Lastly, we studied GUVs encapsulating active matter inside. Soft boundaries are ubiquitous in biological cells, which are influenced by the dynamics of inner matter intricately. GUVs have been harnessed as a model system to mimic biological cells. However, membrane response to active flow generated by inner material is still elusive because the membrane is constantly deformed to interact with the inner active matter. To study it, we enveloped bacteria enough to make active turbulence inside GUVs. Here, we observed two types of collective motions of bacteria groups such as vortex and dipolar flow. Then, we measured the active fluctuations of membranes exposed to such dynamics. Finally, we surveyed the mechanical property of active matter using the surface extension rate and curvature vectors on the membrane. It is found that the elasticity of the active matter is radially asymmetric, which indicates non-reciprocity. We expect that this experiment provides a model system to help people understand membrane responses to active flow and measure the odd elasticity of active matter.clos
Feasibility of Vitrification of Spent Uranium Catalyst Used in Acrylonitrile Production
No full textDepartment of Nuclear Engineeringclos
Freeze-vulnerable plastic crystal to cryogenic liquid electrolyte for lithium batteries by deep freezing-point depression
No full textSchool of Energy and Chemical Engineering (Chemical Engineering)clos
Generating Adversarial Examples Based on Low-Dimensional Compression Models for Robust Adversarial Training
Get PDFGraduate School of Artificial IntelligenceIt is well known that deep neural networks are vulnerable to adversarial examples, which are malevolent inputs designed to fool the network by adding small but fatal perturbations. Recent work has demonstrated this phenomenon focused on perturbation scales and overfitting. These attempts showed that adversarial training provides robustness to adversarial examples with various attack methods. However, there is a lack of explanation for the fact that the algorithm for generating adversarial examples does not utilize the dimensions of the model. We delve into adversarial examples produced accordingly by changing the size and dimensions of the model. Although it is generally known that the larger the size of the model, the greater the capacity and thus the better at learning. We achieve improvements over the baselines even at larger model capacities such as Wide-ResNet(34-10/40-14), and demonstrate results by proving the reason why the accuracy decreases in relatively small models. That means a simple linear model can have adversarial examples if its input has sufficient dimensionality. Using our cross-modeling methods, we can get higher robust accuracy(96.02%) and it means that the larger model, the better robustness. We expect to see these arguments from a different perspective through the variation in standard accuracy and robust accuracy according to the size of the model by introducing the practical demonstrations. Moreover, this view yields insight into diffusion models that require models that are not too large to close the gap between the data and the model.ope
???????????? ????????? ????????? ?????????, ????????? ????????? ????????? ?????? ??????
No full textDepartment of Mechanical EngineeringLayer jamming has been one of the most promising aspects of varying the stiffness of a structure due to its lightweight and compact structure, enabling its application to various fields such as wearable robotics and surgical robotics. Especially, a remarkable class of variable stiffness is reconfigurable layer jamming, which can be reused to various 3-dimensional shapes utilizing its thin form factor. Here, a novel method to enrich the reconfigurability of layer jamming method by increasing the stretchability will be proposed. The design procedures of kirigami patterns that make high stretchability while maintaining its reconfigurability are described and experimentally evaluated. The performances of the proposed design were experimentally compared to that of without kirigami patterns. To verify reconfigurability of our design, its applications as a variable stiffness skin for various human joint and load-bearing structures with various shapes and sizes were demonstrated.clos
Human Perception on Robot???s Face and Color Expression using Linear Dynamic Affect Expression Model
No full textDepartment of DesignWith the rapid rise of the industrial revolution, robots are becoming increasingly popular and have been actively implemented in various areas. As a result, social robots are getting more and more popular. To increase the acceptance of social robots in private and public places, robots should be more sympathetic and expressive. The face is one of the most popular ways to express a robot???s emotion. However, it can be improved by adding colors and blinking modalities. In this study, we will dive deep and try to understand whether colors and blinking modalities have a significant role. We will create a mini-social robot that can express its emotions efficiently. Additionally, we will implement the presented Linear Dynamic Affect-Expression Model (LDAEM) to make it more dynamic and interactive.
Moreover, we conducted an experiment and determined which modality plays a significant role in each emotion. As a result, we discovered that anger could express better with colors, sadness and surprise with eye color, disgust with color and blinking, and fear with blinking. Furthermore, data show that blinking and eye color change are significantly better than just face modality.
In addition, we interviewed participants and gained qualitative data to discover which emotions can convey efficiently using face, color, and blinking. We found that in happiness, face modality is good enough, but possibly tuning color might improve the robot???s expressions. Exaggerating the blinking rate might increase quality too. Finally, we discussed comments from participants where we can find good insights for future worksclos
Integration of Conductive 2D MOF and Porous 3D MOF toward Chemiresistive Sensor
Get PDFDepartment of ChemistryMetal-organic frameworks (MOFs), porous crystalline materials composed of metal clusters and organic ligands, have attracted attention due to their high designability resulting from the various combinations of metals and organic ligands. The numerous structures and pore environments caused by adjusting the metal nodes and organic linkers allow MOFs to be applied as promising materials in various fields such as gas separation, catalysts, bio-applications, and sensors. MOFs as chemiresistive sensors have good gas accessibility owing to their large surface area compared to other materials with sensing properties, and the selectivity toward to target material would be expected by various pore sizes and environments. In this thesis, we conducted a study on MOF-based chemiresistive sensors targeting H2S, one of the harmful gases used for air pollution monitoring and disease diagnosis.
In our previous research, we synthesized a dimensionally well-integrated heterostructure that combines the properties of conductivity from 2D MOF and porosity from 3D MOF. The 2D MOF@3D MOF core-shell composite (Ni-HHTP@UiO-66-NH2, HHTP=2,3,6,7,10,11-hexahydroxytriphenylene), showed a higher response than the 2D MOF/3D MOF physical mixture toward 5 ppm H2S in room temperature air, and the 2D MOF/3D MOF physical mixture was better than the pristine 2D MOF in the same condition. To find composite pairs with higher performance, we noted that the physical mixture exhibited a higher response than the pristine 2D MOF. Therefore, we tried to find some combinations of 2D MOF and 3D MOF with high sensitivity through physical mixture screening. The 2D MOF candidates used for the screening were M-HHTP and M-HITP (M=Ni, Co, Cu, HITP=2,3,6,7,10,11-hexaiminotriphenylene), which were successfully synthesized among the conductive 2D MOFs calculated from the simulation group. 3D MOF candidates were selected based on the uptake and binding energy with the target gas. The synthesis of these candidates used for the screening was characterized by scanning electron microscopy (SEM) and powder X-ray diffraction (PXRD). As a result of physical mixture screening, a pair of Cu-HHTP and Zn-MOF-74 showing a high response of 27.5 to H2S was obtained.
Core-shell (MOF-on-MOF) composites that demonstrate synergistic effects or protect the core MOF have been reported. Therefore, core-shell MOF synthesis can be used as a strategy for property design. Even a lattice-matched pair obtained by calculation to make a composite might not be substantialized core-shell composite experimentally. There are also cases in which successfully synthesized core-shell composites even for pairs with lattice-matching errors. The combination of Cu-HHTP and Zn-MOF-74 obtained through the physical mixture screening is also not easy to synthesize. Accordingly, it is worth finding out the parameters affecting the synthesis of core-shell composite. By controlling the defects of the core MOF, crystal size, and the coverage of the shell MOF, we studied their effect on core-shell MOF composite synthesis. The coordination of 2D MOF and 3D MOF is confirmed by electron microscopy (EM) and electron paramagnetic resonance (EPR) analysis. And the coverage of shell MOF characterized through EM and inductively coupled plasma optical emission spectroscopy (ICP-OES) data.ope
Investigating the novel functions of ZNF212 and PWWP2B in DNA damage response pathway
Get PDFDepartment of Biological SciencesThere are several types of DNA damage that can occur due to endogenous cellular processes or exogenous DNA damaging agents. If the DNA damage is not repaired and accumulated, it causes increase of genomic instability and the risk of cancer and other critical diseases. Therefore, DNA damage response (DDR), including cell cycle arrest, damage repair and apoptosis, is a key pathway involved in maintaining genome integrity. There are numerous kinds of proteins involved in DDR pathwayshowever, the function of many proteins has not yet been studied. In my dissertation, I found the putative target proteins involved in DDR pathway and investigate its functions.
In the first part, I identify the zinc finger protein 212, ZNF212, as a novel binding partner for TRAIP. TRAIP is a key factor involved in the homologous recombination (HR) and DNA interstrand crosslink (ICL) repair. However, the exact functions of TRAIP in these processes in mammalian cells are not fully understood. I find that ZNF212 colocalizes with TRAIP in sites of DNA damage and the recruitment of TRAIP or ZNF212 to sites of DNA damage is mutually interdependent. I show that depletion of ZNF212 causes defects in the DDR and HR-mediated repair in a manner epistatic to TRAIP. In addition, an epistatic analysis of Zfp212, the mouse homolog of human ZNF212, in mouse embryonic stem cells (mESCs), shows that it appears to act upstream of both the Neil3 and Fanconi anemia (FA) pathways of ICLs repair. We find that human ZNF212 interacted directly with NEIL3 and promotes its recruitment to ICL lesions. Collectively, I identify ZNF212 as a new factor involved in the DDR, HR-mediated repair and ICL repair though direct interaction with TRAIP.
In the second part, our group find PWWP2B is one of the most frequently mutated genes in 25 Korean gastric adenocarcinoma patients using whole-exome sequencing. I find that PWWP2B interact with UHRF1, a key factor regulating the pathway choice of double strand break (DSB) repair. PWWP2B also has a role in DNA double-strand break repair. PWWP2B moves to sites of DNA damage through its interaction with UHRF1. Depletion of PWWP2B enhances cellular sensitivity to ionizing radiation (IR) and impairs IR-induced foci formation of RAD51. PWWP2B participates in HR via promoting DNA end-resection. Taken together, these data show that PWWP2B facilitates the recruitment of DNA repair machinery to sites of DNA damage and promotes HR-mediated DNA DSB repair. Impaired PWWP2B function might thus cause genome instability and promote gastric cancer development.
These findings will contribute to an improved understanding of the DNA damage response and repair pathway.ope
Modeling and Evaluation of the Volumetric Errors for the 5-Axis Machine Tool
Get PDFDepartment of Mechanical EngineeringCNC Machine tools are among the most important means of production in metalworking industries and/ have been most widely used. Since an increasing demand for machinery parts with geometric complexity in high efficiency, multi-axialization and multi-functionalization emerged as technology trends in the machine tools field. As a result, 5-axis machine tools have been extensively used in various manufacturing applications requiring higher machining accuracy. In reality, demand in aerospace, medical, electric vehicle, and precision&semiconductor industries are driving. Based on the order composition of machine tools, the proportion of five-axis machine tools has become large remarkably, and also this trend is expected to continue in the future.
While this high flexibility to machine complex parts efficiently could have led the 5-axis machine tool to become a great solution in the metalworking industry, at the same time, 5-axis machine tools have encountered challenges that have to overcome hurdles that come from this flexibility and freedom. As is well known, machine tools' accuracy is one of the most important indicators for the performance of machine tools, and 5-axis machine tools accompany more complexities and require more elements and more assembly processes, so encounter more accuracy problems indispensably.
Overcoming these challenges and under the motivation to high-accuracy 5-axis machine tools, modeling and evaluating the volumetric errors for a 5-axis machine tool was set as this research objective. Specifically, in order to model the volumetric error, a study on the kinematic structures and identification for systematic error is carried out beforehand, and based on this, the goal aims to establish an error model of the 5-axis machine tool. Then, the error model established for a 5-axis machine tool is applied to a practical machine tool, and the errors propagating sensitively to volumetric error are determined as key errors. Also, the resultant effect of the individual errors and key errors is evaluated in advance through estimation of volumetric error. To estimate the volumetric error in virtual, stochastic estimation with random variables was conducted to obtain tool points??? coordinates within a workspace, and statistical analysis has carried out. Through these analysis processes, whether machine tools' volumetric error is enhanced under the condition when key errors were specified and assembled is confirmed.
Lastly, to confirm the utility of modeling and evaluation for volumetric error in advance, the demonstration was conducted on two actual 5-axis machine tools, and each machine tool had been manufactured under the management to be assembled with the tolerance used in virtual evaluations as same as possible. Depending on whether or not the specified key error tolerance range is fulfilled, it has clearly confirmed the superiority and inferiority of 5-axis machine tools' volumetric error. And also by comparing estimation and experimental results, it is found that the approach and methods used in this study were useful and could be applied to the actual 5-axis machine tool manufacturing process.ope