Tohoku University Repository (TOUR) / 東北大学機関リポジトリ

Tohoku University博士（情報科学）thesi

Yi, Linya

English

Grammaticalization of Verbs in Mandarin Chinese: An Analysis in Terms of Cartography and Upward Reanalysis

LI, GUANDA

Bioinspired Deep Reinforcement Learning for Energy-efficient Control of Redundant Robotic Systems

Tohoku University博士（工学）要約のみthesi

CANDALOT, ARTHUR

論文内容要約 
 Since the beginning of the Space Age, space exploration has been considered the peak of technological 
advancements. The numerous achievements in the field have greatly deepened our understanding of the 
universe, especially the history of our Solar System. Scientific progress derived from space exploration has 
accelerated innovation in many fields and the development of technologies impacting our everyday lives, 
from mattresses to the internet. However, the will to explore space was not extinguished by the end of the 
Cold War, and ambitious plans to push the boundaries of space explorations are being undertaken. Current 
technology has facilitated access to Earth's orbit and neighboring planets with affordable rockets, opening 
the spot for commercial and private missions in space. Recent investigations of the surface of Mars and the 
Moon have proven the presence of water and valuable materials, which greatly increases the possibility of 
establishing extraterrestrial settlements that could shelter humans for extended periods. Such settlements 
can also provide intermediate stations to increase the reach of robotic missions on farther planets and 
asteroids. Even so, our current capabilities do not permit such endeavors, and progressive exploration of 
areas of interest, as well as experimentation with resource collection, is a crucial step toward space 
expansion. To do so, advanced robotics platforms must be developed to overcome the limitations of previous 
rovers, such as reliable locomotion on uneven terrain and the ability to climb over rocky terrain. The 
presented work contributes to this challenge by proposing the Tendon-Actuated Soft Capture (TASC) 
Gripper, a compliant system to allow legged robots such as the SCAR-E hexapod to operate in extreme 
terrain, such as the neighborhood of craters and underground caves. 
The dissertation is divided into five chapters covering a different aspect of the TASC Gripper's potential 
applications. The first chapter provides an overview of space exploration and technological advances in 
  カンダロ アーサー 
氏 名   CANDALOT Arthur 
研究科，専攻の名称 東北大学大学院工学研究科（博士課程）航空宇宙工学専攻 
学 位 論 文 題 目          Compliant End-Effector for Dexterous Space Applications 
  (器用な宇宙応用のためのコンプライアントエンドエフェクタ) 
論 文 審 査 委 員          主査 東北大学教授 吉田 和哉 東北大学教授 平田 泰久 
  東北大学教授 林部 充宏 教授 Carol Luna MARTINEZ 
（University of Luxembourg） 
mobile robotics. It emphasizes the importance of developing innovative robotic systems to enable deeper 
exploration and resource exploitation in space. This chapter summarizes the research objectives and 
introduces the SCAR-E hexapod as the gripper's intended platform. 
 
Chapter 2 details the engineering design of the TASC Gripper, supported by observations of geologic 
formations in space and analysis of existing prehension methods. The gripper is equipped with microspines 
to provide effective anchoring on uneven, rocky surfaces. Its compliant tendon-driven structure allows its 
fingers to adapt to a variety of surface geometries, increasing contact and gripping force while offering 
reliability and versatility. Experimental testing evaluates the gripper’s performance under various 
conditions, such as surface curvature, actuation power, and spine configurations. The results demonstrate 
the gripper's performance and robustness for rock-grasping and extraterrestrial applications. 
 
Chapter 3 focuses on the terramechanics of the TASC Gripper and observes its capability to provide stable 
footing for legged robots in extraterrestrial soil. The research introduces a hybrid simulation model 
combining multi-body dynamics and discrete element methods to simulate the gripper’s interaction with 
granular materials. Experimental validation confirms the accuracy of the simulation models, enabling the 
estimation of sinkage in extraterrestrial conditions. Complementary experiments demonstrated the impact 
of diverse parameters, such as terrain steepness, entry speed, contact area, and leg motion on the sinkage 
and traction performance of the gripper. A conclusion is then made on the key points to consider when 
developing legged robots for off-road locomotion.  
 
Chapter 4 explores the potential of the TASC Gripper for orbital manipulation, particularly for capturing 
space debris in zero-gravity environments. The gripper’s compliant features are coupled with microfibrillar 
adhesive materials, allowing the soft capture of smooth debris objects. Experiments conducted in emulated 
zero-gravity conditions at the Zero-G Lab validate the gripper’s stability in a free-floating state and its 
capability of capturing space debris without inducing collisions. This application highlights the versatility of 
the TASC Gripper and demonstrates the applicability of a soft gripper for space debris capture. 
 
Chapter 5 concludes the thesis with a summary of the findings, highlighting the TASC Gripper’s 
contributions to space robotics. Lastly, the possible future expansion of the conducted research is discussed, 
offering multiple avenues to explore regarding climbing, soft terrain locomotion, and orbital applications. 


Compliant End-Effector for Dexterous Space Applications

OKUIZUMI, HIROYUKI

論文内容要約 
   This study investigated the drag phenomena associated with fundamental bluff body shapes–circular 
cylinders, square cylinders, and spheres–through experimental studies conducted using a 1-m Magnetic 
Suspension and Balance System (MSBS). The MSBS enables experiments to be performed without support 
interference, eliminating the need for complex support interference correction methods traditionally employed in 
wind tunnel testing, such as the mirror image method or base pressure corrections. These methods are often 
challenging to implement and require complicated correction procedures. On the other hand, the MSBS allows for 
straightforward, focused experiments. The wind tunnel tests were conducted for the following specific target 
shapes: flow-aligned circular cylinders, flow-aligned square cylinders, and a rotating sphere. 
The aerodynamic characteristics of a circular cylinder aligned with the flow axis and explores suitable 
blockage correction methods were investigated. Although a circular cylinder is geometrically simple, its flow 
characteristics are notably complex and unsteady. The boundary layer separates at the leading edge, creating a 
large wake separation region that oscillates at low frequencies (bubble pumping). Further downstream, vortices 
exhibit modes such as spiraling and flapping. These phenomena have been extensively studied, and it is well 
known that the aerodynamic characteristics of a cylinder depend on its length-to-diameter ratio (L/D). Notably, the 
leading-edge separated shear layer reattaches approximately 1.5 times the cylinder’s diameter downstream, and 
the drag coefficient reaches a minimum for L/D values between 1.7 and 1.8. While significant progress has been 
made in understanding the flow field and aerodynamic characteristics of circular cylinders, some aspects remain 
unexplored. For example, due to the large separation region in the wake, circular cylinders are particularly 
susceptible to blockage effects. However, appropriate correction methods, especially for conditions without support 
interference, have not been well established, except for the study by Sawada et al. (2011) on a cylinder with L/D = 
4 using an MSBS. In this study, we examined suitable blockage correction methods for circular cylinders under 
support-interference-free conditions using 1-m MSBS. Special attention was given to the presence or absence of 
           おくいずみ ひろゆき 
氏 名          奥 泉  寛 之 
研究科，専攻の名称 東北大学大学院工学研究科（博士課程）航空宇宙工学専攻 
学 位 論 文 題 目           Experimental Study of Aerodynamic Drag about Fundamental Bluff 
Bodies Using a Magnetic Suspension and Balance System（磁力支持天秤装置を用いた基本鈍頭
物体形状の空力抵抗に関する実験的研究） 
           
論 文 審 査 委 員          主査 東北大学教授 大西 直文 東北大学教授 大林 茂 
             東北大学教授 永井 大樹 教授 野々村 拓 
                              （名古屋大学） 
reattachment of the leading-edge separated shear layer, which significantly affects aerodynamic characteristics. 
Using the MSBS for experiments on blockage effects provides a major advantage. When mechanical supports are 
used, reducing the blockage ratio of the model typically increases the relative size of the support components, 
leading to inaccuracies in drag measurements. In contrast, the MSBS eliminates support interference, enabling 
precise drag measurements for models with small blockage ratios. The test models were controlled using five 
degrees of freedom (DOF). Two circular cylinder models with different L/D ratios were tested. The L/D = 1.0 model 
did not exhibit shear layer reattachment, whereas the L/D = 2.0 model did. Drag measurements were conducted at 
two Reynolds numbers and multiple blockage ratios to estimate the drag coefficient at zero blockage CDL. The 
blockage ratios tested were limited to two for the L/D = 1.0 model and three for the L/D = 2.0 model. A broader 
range of measurements would likely improve the accuracy of CDL predictions. For L/D = 1.0 (without 
reattachment), applying Maskell’s (1963) wake blockage correction with εw = 2.75 resulted in corrected drag 
coefficient values within ±0.03 of CDL. Adding Lock’s (1929) solid blockage correction further reduced the 
discrepancy to within ±0.01. For L/D = 2.0 (with reattachment), wake blockage correction alone was insufficient 
due to the influence of reattachment. Therefore, both wake blockage and Lock’s solid blockage corrections were 
required. Lock’s solid blockage correction incorporates two parameters: λ (related to the model shape) and τ 
(related to the cross-sectional shape of the wind tunnel test section). Under the test conditions, which involved a 
cylindrical model and an octagonal test section, using λ for a Rankine ovoid model and τ for a circular test section 
reduced the difference between the corrected drag coefficient and CDL to less than ±0.02. The findings of this study 
demonstrate the effectiveness of blockage correction methods for circular cylinders under support-interference-free 
conditions. Furthermore, they highlight the importance of considering the effects of leading-edge shear layer 
reattachment when applying blockage corrections to aerodynamic measurements. 
A square cylinder aligned with the airflow generates a flow field characterized by large-scale separation 
similar to spheres and circular cylinders. However, its rectangular edges and surfaces distinguish it from the 
circular cylinder, making it a highly intriguing shape for study. In practical applications, this geometry is often 
seen in drone payloads, helicopter underslung loads, and cargo containers. Despite its relevance to society, 
research on its aerodynamic characteristics has been relatively limited, with most previous studies relying on 
measurements involving struts.  This study investigated the effects of L/D and the presence or absence of flow 
reattachment on the aerodynamic characteristics of low-aspect-ratio square cylinders using the 1-m MSBS. The 
use of the MSBS enabled the evaluation of aerodynamic characteristics without support interference in the wake 
of the square cylinder. In contrast, struts, which have been commonly used in previous research, were expected to 
generate weak longitudinal vortices that could significantly interfere with the wake of the square cylinder. The 
sensing method for square cylinders developed by Horiguchi et al. (2021) was employed, and six DOF control was 
used to ensure that the roll angle remained fixed. Aerodynamic measurements were conducted on square 
cylinders with L/D of 1.0, 1.2, 1.68, and 2.0. A pressure telemetry system installed inside the model was used to 
measure base pressure. To confirm the reattachment position, surface oil-flow visualization was performed on 
models with L/D=1.68 and 2.0. Using six DOF control allowed for stable surface flow visualization by maintaining 
consistent side surface positioning during testing. The drag coefficient was found to be independent of Reynolds 
numbers in the range of 0.5 × 105 to 1.3 × 105, and regarding the dependency on fineness ratio, two distinct regions 
were observed. For L/D  ≤  1.0, the drag coefficient remained around 1.1, while for L/D ≥  1.68, the drag 
coefficient dropped to approximately 0.8. The base pressure coefficient decreased monotonically from L/D = 1.0 to 
1.68. While bluff bodies are generally dominated by pressure drag, and thus, the drag coefficient would be 
expected to follow a similar trend as the base pressure coefficient, the results showed a discrepancy. This suggests 
the need for further investigation into the correlation between changes in base pressure and drag coefficients. The 
lift-curve and moment-curve exhibited slope reversals between L/D=1.2 and 1.68, corresponding to quadratic 
changes in the drag and base pressure coefficients concerning changes in the angle of attack. Flow visualization at 
an angle of attack of 0° showed reattachment for both L/D=1.68 and 2.0. The reattachment occurred at 
approximately 90% of the cylinder’s axis length in both cases. Prior studies on circular cylinders aligned with the 
flow axis reported reattachment positions at approximately 1.5 times the cylinder diameter without the 
axis-length dependency observed in square cylinders. This study highlights the distinct aerodynamic 
characteristics of square cylinders compared to circular cylinders, particularly in the dependency of reattachment 
position on axis length. Future investigations into the correlation between drag and base pressure coefficients and 
further experiments with varying fineness ratios are necessary to deepen the understanding of these behaviors. 
Rotating spheres are commonly observed in various sports, including baseball, soccer, tennis, and table tennis. 
When subjected to airflow, a rotating sphere generates Magnus force, which causes the ball to curve, resulting in a 
"breaking ball." Breaking balls are essential in competitive play, influencing game strategy and enhancing 
entertainment value. Additionally, under certain rotational speeds and Reynolds number (Re) conditions, the 
occurrence of negative Magnus force has been reported. However, the specific conditions for its emergence vary 
significantly between experimenters and measurement methods, leading to discrepancies between wind tunnel 
and free-flight experiments. Possible causes of these discrepancies include support interference from struts or 
wires in wind tunnel tests, inaccuracies in estimating aerodynamic forces from image analysis in free-flight 
studies, and the difficulty of observing transition phenomena, such as drag crisis, in dynamic free-flight conditions 
where velocity and other parameters continuously change. This study aimed to investigate the conditions under 
which negative Magnus force is generated using a MSBS, which eliminates support interference. Previous wind 
tunnel studies on rotating spheres have commonly employed struts or piano wires for support. However, struts can 
asymmetrically modify the flow field around the sphere, and sting supports have been reported to significantly 
influence drag coefficients in the supercritical Reynolds number regime. Conducting experiments without 
mechanical supports reduces these unwanted influences, leading to more accurate aerodynamic force 
measurements. A wind tunnel testing method for rotating spheres was developed for the 1-m MSBS. Conducting 
wind tunnel experiments without support interference allows for the exclusion of support-related factors, enabling 
more precise investigation of Magnus force phenomena. However, the MSBS was not originally designed to 
support rotation about the axis perpendicular to the wind direction, necessitating the development of new 
experimental techniques. The proposed method adopted a magnetic suspension approach, in which the main 
magnet’s ψ angle was passively controlled while the rotation angle about the z-axis of the spherical shell was not 
measured. To accommodate this approach, a dedicated sensor subsystem was developed specifically for testing 
rotating spheres. The developed methodology was applied to a spherical model with a diameter of 141.8 mm. The 
system successfully achieved stable rotation at 2000 min-1 in still air. Wind tunnel experiments were conducted at 
Re = 6.5 × 104 and spin parameter (SP) values below 0.6, without any support interference. By applying a low-pass 
filter to the aerodynamic force coefficients obtained in the experiments, the presence of Magnus force was 
confirmed. A decrease in positive Magnus force corresponding to changes in SP was observed. However, the 
accuracy of aerodynamic force measurements was affected by several factors, including calibration errors, 
uncertainty in the main magnet’s ψ orientation due to passive control, and imbalances in the rotating component’s 
center of gravity. Additionally, the experiments were limited to two trials due to instances of the model dropping 
during tests.  This study shows the method should be used to investigate additional conditions, such as Re = 6.5 
×104 with SP ≤ 0.6 and various SP values within the Reynolds number range of 6.5 × 104 to 2.0 × 105. This study 
successfully developed a wind tunnel testing method for rotating spheres using the 1-m MSBS, highlighting its 
potential for investigating Magnus forces, including conditions that generate negative Magnus force, without 
support interference. Future research should focus on improving calibration accuracy, addressing rotational 
imbalances, and enhancing system reliability to expand the experimental scope and improve the robustness. 
The results of this research are expected to improve our understanding of the drag characteristics of bluff 
bodies and help improve the design and performance of bluff bodies in various fields, including aerospace, 
automotive, and sports engineering. Future studies may focus on elucidating the aerodynamic characteristics of 
more complex bluff body shapes, conducting dynamic wind tunnel experiments using MSBS, and integrating 
numerical simulations for detailed analysis of flow fields. 


Experimental Study of Aerodynamic Drag about Fundamental Bluff Bodies Using a Magnetic Suspension and Balance System

SUZUKI, SOICHIRO

 
論文内容要約 
As a replacement for chemical rocket propulsion in satellite launches, one of the beamed energy propulsion 
systems, microwave rocket, has been proposed to reduce the high cost of satellite launches. This system uses 
high-power millimeter waves emitted from ground-based oscillators. Within the microwave rocket, the 
ionization front of the millimeter-wave discharge plasma propagates, generating thrust force through the 
pressure of the induced shock wave. The propagation speed of the ionization front is a critical factor in 
determining the thrust performance of the microwave rocket. However, the past numerical model could not 
maintain the discharge in the experimental condition, and the propagation speed was underestimated even 
when the artificially high beam power densities were applied. Previous experimental and numerical studies 
emphasized the importance of interactions between electromagnetic waves, plasma, and neutral fluids in the 
subcritical millimeter-wave discharge process. These interactions are mediated by the electromagnetic-wave 
interference, radiation transport, and chemical reaction processes. This study investigated these interactions 
to elucidate the propagation process of the millimeter-wave discharge in a microwave rocket using numerical 
and experimental approaches.  
At first, the ionization-front propagation process was numerically investigated using a one-dimensional (1D) 
numerical simulation coupling the electromagnetic-wave propagation, reaction-diffusion of plasma, 
compressible neutral fluid, detailed chemical reactions, and radiation transport. Three distinct discharge 
structures were identified at electric-field intensities exceeding 0.2 MV/m depending on the electric-field 
intensity of the incident millimeter-wave beam. At intensities above 1.4 MV/m, the ionization front propagated 
supersonically while forming a discrete structure with a quarter wavelength interval. The discrete structure 
           すずき そういちろう 
氏 名          鈴 木 颯 一 郎 
研究科，専攻の名称 東北大学大学院工学研究科（博士課程）航空宇宙工学専攻 
学 位 論 文 題 目          Study on Interactions between Electromagnetic Wave and Fluids in 
Subcritical Millimeter Wave Discharge (亜臨界ミリ波放電における電磁
波流体間相互作用の研究) 
論 文 審 査 委 員          主査 東北大学教授 大西 直文 東北大学教授 永井 大樹 
             東北大学教授 高奈 秀匡 東北大学准教授 高橋 聖幸 
was formed when the new plasma spot was induced by the electron impact-ionization triggered by seed 
electrons generated through radiation transport at the antinode of the standing wave. As a result, the new 
plasma spot was induced a quarter wavelength ahead of the preceding ionization front. At intermediate 
intensities (0.3–1.2 MV/m), the ionization front propagated continuously because the high reduced electric-
field intensity enhances the electron impact ionization at the gas expansion region due to the gas heating. The 
heating process was fast gas heating through the de-excitation of the electronically excited particles. Below 
0.3 MV/m, the ionization front propagates continuously, but the heating process was dominated by the 
vibrational-translational relaxation process, with the collisional ionization between heavy particles sustaining 
the discharge. The propagation speeds was in agreement with the experimental results at field intensities 
above 1.4 MV/m and below 0.3 MV/m; however, it was underestimated at intermediate intensities where the 
microwave rocket operates.  
The experimental supersonic ionization-front propagation was successfully reproduced in the numerical 
simulation only when the ionization front propagated discretely. This suggests that the ionization front may 
have propagated discretely in the experiment. However, the discrete propagation could not be observed using 
previous high-speed camera imaging due to the short timescale and the dense spatial distribution of the 
plasma spots. To validate this proposed hypothesis, an interferometric discharge structure identification 
(IDSI) method was proposed and numerically demonstrated by a 2D numerical simulation. When the 
ionization front propagated continuously, the time-varying waveform of the standing-wave intensity 
measured at a fixed observation point appeared smooth, and its peak frequency obtained via the Fourier 
transform was proportional to the propagation speed. Conversely, when the ionization front propagated 
discretely, high-frequency components were superimposed onto the waveform. These high-frequency time 
variations resulted from the formation of new plasma spots. These findings indicate that the discharge 
structure identification and the ionization-front propagation speed measurement can be achieved without a 
high-speed camera if the time-variation of the standing-wave intensity is analyzed.  
Subsequently, the IDSI method was experimentally demonstrated using a 28 GHz gyrotron. This was 
achieved by comparing high-speed camera observation with the standing-wave intensity measured by a 
rectenna, which converts millimeter-wave power into direct-current power. When continuous ionization-front 
propagation was observed via the high-speed camera, the time-variation waveform measured by the rectenna 
remained smooth. In contrast, the waveform exhibited rapid variations, and its Fourier spectrum contained 
high-frequency components when a streamer structure was induced, which was consistent with the prediction 
of the simulation. Further observations on shorter timescales revealed that the waveform variations were 
synchronized with the timing of the streamer elongation. These results demonstrate the experimental validity 
of the IDSI method using a gyrotron, confirming its applicability for observing millimeter-wave discharge 
plasma in a microwave rocket.  
 


Study on Interactions between Electromagnetic Wave and Fluids in Subcritical Millimeter Wave Discharge

兪, 帆緯

Development of a Reliability Evaluation Method for Flaw Detection using Eddy Current Testing under Low Signal-to-Noise Ratio

大瀧, 貴史

 
          おお たき  たか し 
氏 名          大 瀧 貴 史 
授 与 学 位          博士（工学） 
学位授与年月日          令和7年3月２５日 
学位授与の根拠法規 学位規則第４条第１項 
研究科，専攻の名称 東北大学大学院工学研究科（博士課程）応用物理学専攻 
学 位 論 文 題 目          励起子分子の量子もつれに由来する光学応答の理論研究 
指 導 教 員 東北大学教授 松枝 宏明 
論 文 審 査 委 員          主査 東北大学教授 松枝 宏明 東北大学教授 小野 円佳 
             東北大学教授 大兼 幹彦 准教授  大畠 悟郎 
                                （大阪公立大学） 
論文内容要約 
複数の粒子間に働く量子力学的な相関である量子もつれは、量子情報分野における重要な性質の一つである。
量子力学に基づいて安全性が保証される量子情報通信では、量子もつれの関係にある光子対を活用することによ
る長距離通信の実現が期待されている。量子もつれ光子対の生成法の一つとして、半導体中の励起子分子を用い
る方法がある。半導体 CuCl において二光子吸収で生成された励起子分子は偏光に関する量子もつれを示すこと
が知られている。これは励起子分子を構成する励起子間の量子もつれに由来する。この原理を逆に用いて、励起
子分子から放出される光子対の偏光状態を調べることで、励起子分子に内在する量子もつれが実験的に調べられ
ると期待される。このような背景から、誘導放出トモグラフィとポンプ・プローブ分光法を組み合わせることで、
時刻𝜏に励起子分子から放出される光子対の偏光状態に関する密度行列?̂?DMS(𝜏)を測定する実験手法である密度
行列分光法(Density matrix spectroscopy; DMS)が開発されている。DMSでは、ポンプ光とプローブ光を遅延時間𝜏だけ
時間差をつけた上で試料に照射し、放出される四光波混合(Four-wave mixing; FWM)信号の偏光ごとの強度を調べる。
ポンプ光は二光子吸収により励起子分子を生成するように設定する。プローブ光の偏光を変えて FWM信号の偏
光ごとの強度の測定を繰り返すことで、誘導放出トモグラフィに基づいて?̂?DMS(𝜏)を得ることができる。
?̂?DMS(𝜏)から計算される量子もつれダイナミクスは、励起子分子における量子もつれダイナミクスを反映してい
ると期待される。一方で、DMS に対する理論研究は存在せず、?̂?DMS(𝜏)と励起子分子の状態|𝜓bi(𝜏)⟩の間の関係
は不明であり、DMS で得られた量子もつれダイナミクスが励起子分子の量子もつれダイナミクスをどのように
反映しているかは不明である。 
以上のような背景から、本論文では DMS の理論的基盤の構築を目的とした。まず、解析計算によって
?̂?DMS(𝜏)と|𝜓bi(𝜏)⟩の間の関係式を導出した（第 2章）。次に、励起子分子の量子もつれダイナミクスを調べる上
での DMS の有効性を検証するために、励起子分子モードが固有状態として自然に現れる二次元電子・正孔系の
格子模型に対する光学応答の数値計算を行った。この模型において二光子吸収で生成された励起子分子から偏光
もつれ光子対が生成されることを確認し（第3章）、その後にDMSのシミュレーションを行った（第4章）。DMS
によって得られる量子もつれダイナミクスと励起子分子における量子もつれダイナミクスをそれぞれ計算して比
較することで、それらの間の関係を明らかにした。第 5章では本研究を総括し、展望として DMSの拡張の可能
性について述べた。各章の概要を以下に記す。 
第2章：励起子分子に対する密度行列分光の理論的定式化 
DMS を定式化するために、FWM に由来する分極電流の横成分と励起子分子状態|𝜓bi(𝜏)⟩の間の関係式を摂動
的に求め、この分極電流の横成分をソースとして放出される FWM光の電場をマクスウェル方程式に基づいて計
算した。FWM における位相整合条件の効果を適切に取り入れるために、試料の厚みが光の波長と比べて十分に
厚い場合と十分に薄い場合とでそれぞれ計算を行った。得られた FWM光の表式に対して偏光に関する誘導放出
トモグラフィを考えることで、DMS で得られる密度行列?̂?DMS(𝜏)と励起子分子状態|𝜓bi(𝜏)⟩の間の関係式を得た。
具体例としてCuClのバルク結晶に対するDMSについて考察したところ、励起子分子状態を特定の波数（試料中
のプローブ光とFWM光の波数）の横波励起子対へ射影した状態がDMSで得られることを明らかとした。 
第3章：二次元電子・正孔系の励起子分子を用いた量子もつれ光子対生成 
異種サイトが交互に並んだ二次元正方格子上の電子・正孔系をファブリ・ペロー型の光共振器中に置いた系
を考える。模型としてイオン性拡張ハバード模型を導入し、共振器中の光子と電子・正孔系の間の相互作用をパ
イエルス位相の形で導入した。十分に大きなシステムサイズの系（𝑁 ∼ 102サイト）に対する励起子分子波動関
数を数値計算するために、トランスファー積分が小さいときに有効な二次摂動を適用した。本論文で用いた電
子・正孔系のパラメータでは、二光子吸収スペクトルの低エネルギー部分に隣接した三つの励起子分子ピークが
現れる。考えている二次元系の四回対称性に由来して、これらのピークに対応する励起子分子モードは四回回転
?̂?4の固有状態となる。それぞれのモードの半分の光子エネルギーを持つ直線偏光の光子が二つある状態を初期
状態とし、時間依存ランチョス法により時間発展計算を行った。二光子吸収で生成された励起子分子は光子対へ
変換されるが、これらの光子対のうち初期状態の光子対とは異なるモードの光子対の偏光状態に注目する。この
光子対は、初期時刻付近ではもつれのない偏光状態だが、時間経過に伴って二光子共鳴させた励起子分子モード
と同じ四回対称性の最大量子もつれ状態へと近づいていくことが確認された。また、励起子分子中の励起子間の
擬角運動量に関する量子もつれを導入し、光子対の偏光量子もつれダイナミクスと励起子間の擬角運動量に関す
る量子もつれダイナミクスを比較したところ、一定以上の時間が経つと両者のダイナミクスが対応することが明
らかとなった。このことは、励起子分子から放出される光子対の偏光もつれが励起子分子における（擬）角運動
量の量子もつれを反映したものになるという、従来の解釈と整合するものである。 
第4章：二次元電子・正孔系の励起子分子に対する密度行列分光 
第 3章と同じ電子・正孔系に対して DMSのシミュレーションを行った。ポンプ光は直線偏光とし、第 3章で
注目した三種類の励起子分子モードを同時に励起するように設定し、モード間の干渉によって励起子分子の量子
もつれを振動させる。プローブ光は偏光を可変とし、線形吸収がある周波数領域を避けて導入する。ポンプ光と
プローブ光は二次元面に対して垂直に入射する。様々なプローブ光の偏光条件に対して光誘起ダイナミクスを時
間依存Lanczos法で数値計算することでDMSのシミュレーションを行った。励起子分子に内在する量子もつれを
評価する際には、励起子分子をアップスピンの正孔とダウンスピンの電子からなる励起子Aとダウンスピンの正
孔とアップスピンの電子からなる励起子Bに分割したときのエンタングルメントエントロピーを用いた。励起子
分子状態|𝜓bi⟩と DMSで得られる状態|𝜓proj⟩におけるエンタングルメントエントロピーをそれぞれ𝑆bi, 𝑆projと定
義する。両者のエンタングルメントエントロピーのダイナミクスを求めたところ、その振動の様子がある程度似
ていることが確認された。この類似の強さを定量化するために振動成分Δ𝑆bi , Δ𝑆projに対してコサイン類似度
cos𝜃を計算したところ、0.80 < cos𝜃 < 0.83となった。DMS では励起子分子状態中の特定の波数の励起子対
（本シミュレーション条件では波数が共に𝟎の励起子対）の成分に関する情報が得られ、励起子対の波数に関す
る量子もつれの情報は得られないため、波数に関するエンタングルメント𝑆𝑘 ≔ −∑ 𝑊𝐤 ln𝑊𝐤𝐤 を導入した
（𝑊𝐤：励起子分子状態中の波数±𝐤の励起子対のウェイト）。𝑆𝑘の振動成分Δ𝑆𝑘をΔ𝑆projに加えたところ、Δ𝑆biと
近い振動波形になることが確認された（Δ𝑆proj + Δ𝑆𝑘とΔ𝑆biの間のコサイン類似度：0.9871 < cos𝜃 < 0.9882）。
このことは、励起子分子のエンタングルメントエントロピーの分解公式𝑆bi = ∑ 𝑊𝐤𝐤 𝑆bi,𝐤 + 𝑆𝑘と関係している
（𝑆bi,𝐤：励起子分子状態中の波数±𝐤の励起子対成分のエンタングルメントエントロピー）。右辺第一項目に現れ
る加重平均の振動成分がΔ𝑆bi,𝐤=𝟎と近いものであることとΔ𝑆bi,𝐤=𝟎 ≃ Δ𝑆projが成り立つことがシミュレーション
結果から確認されており、これらと分解公式を組み合わせることでΔ𝑆bi ≃ Δ𝑆proj + Δ𝑆𝑘が得られる。このこと
は、DMSによって(𝑆bi − 𝑆𝑘)のダイナミクスを近似的に調べられることを意味しており、励起子分子の量子もつ
れダイナミクスを調べる上で DMS が有効な手法であると結論付けた。𝑆𝑘のダイナミクスを別の実験で得ること
ができれば、励起子分子の完全な量子もつれダイナミクスを得ることができる。 
本シミュレーションで考慮した複数の励起子分子モードを同時に励起することによる量子もつれの干渉は、
GaNやZnOのようなウルツ鉱構造の直接遷移型半導体に存在する複数種類の励起子分子モードを二つ以上同時に
励起することによって誘起できると予想される。それに対してDMSを適用することで、(𝑆bi − 𝑆𝑘)の振動ダイナ
ミクスが実験的に得られると予想される。   
第5章：総括 
本章では各章の内容をまとめた上で、今後の DMSの拡張の可能性として光ドレスト状態との組み合わせと二
探針走査型近接場光学顕微鏡との組み合わせの二点について述べた。前者については、通常の DMSでは情報が
得られない一光子禁制な励起子を含んだ励起子分子を調べる上で有効となりうる。後者については、励起子分子
半径よりも微細な空間分解能を持つ二探針走査型近接場光学顕微鏡が実現されれば、通常の DMSでは得ること
のできない波数に関するエンタングルメント𝑆𝑘のダイナミクスを調べる上で有効となりうる。 


Japanese

励起子分子の量子もつれに由来する光学応答の理論研究

折原, 秀治

ラマン分光法を用いた化学強化ガラスの応力測定装置の開発に関する研究

テイ, ハクゴウ

論文内容要約 
Iron and steel have been one of the most important structural materials for industrial and 
commercial applications due to the high abundance of Fe on Earth, the relative ease for refining 
and processing iron ores and the versatile mechanical properties of iron and steels. However, the 
carbon-intensive nature of steelmaking industry makes it necessary to develop strategies for 
reducing carbon emission associated with iron and steel materials. One of such strategies is to 
develop steels with excellent mechanical properties, as it helps to reduce the net consumption of 
the materials. 
On this note, for ferritic steels, the grain boundary segregation of C, a prominent interstitial 
solute in iron and steel, is known to enhance the mechanical properties by increasing bcc Fe grain 
boundary strength, which indicated by the increase in Hall-Petch coefficient with increased C 
segregation amount. Moreover, the segregation of C can relieve the grain boundary embrittlement 
induced by the grain boundary segregation of impurities such as P, which is indicated by the 
lowering of the ductile-to-brittle transition temperature (DBTT) with the segregation of C. 
As a consequence, for developing ferritic steels with superior mechanical properties, it is crucial 
to understand the grain boundary segregation behaviors of C atoms and the factors affecting such 
behaviors. Previous experimental studies have shown that C segregation at bcc Fe grain 
boundaries is affected by the grain boundary characters. Precisely, the C segregation amount 
increases with the increase of misorientation angle in low-angle region and stays relatively 
constant in high-angle region except several coincidence site lattice (CSL) special misorientations 
indicating very low segregation amount of C. Furthermore, the existing atomistic simulations on 
grain boundary segregation of C in CSL grain boundaries demonstrated that different grain 
boundary segregation sites associate with different segregation energies, which indicates the 
existence of a segregation energy spectrum instead of a single value of segregation energy that is 
assumed in classical thermodynamics models on grain boundary segregation. Meanwhile, studies 
on other substitutional segregants in other metallic systems (such as Mg in Al) have explicitly 
shown that the atomistic factors of spectral segregation energy and solute-solute interaction can 
lead to significant negative deviations from the segregation amount predicted by the classical 
thermodynamic models. However, currently there has been no existing study that clearly 
determines the effect of the atomistic factors on the C segregation at bcc Fe grain boundaries. 
Particularly, there have been no simulation study that aims to explain and predict the segregation 
behavior of C atoms at bcc Fe random grain boundaries. 
In this study, persistent homology, a form of topological data analysis, is applied to identify the 
           Zheng, Bohao 
氏 名          鄭 伯 豪 
研究科，専攻の名称 東北大学大学院工学研究科（博士課程）金属フロンティア工学専攻 
学 位 論 文 題 目          Topological features of α-Fe grain boundary structure and its 
connections with C segregation 
（α-Fe 粒界構造のトポロジー特徴と炭素偏析との関連） 
           
論 文 審 査 委 員          主査 東北大学教授 古原 忠  東北大学教授 久保 百司 
             東北大学教授 佐藤 裕  東北大学教授 宮本 吾郎 
                               
“open space” within the atomistic models of bcc Fe grain boundaries as the world’s first attempt the 
atomic structural analysis of grain boundaries in metallic systems. Furthermore, this study 
establishes the connection between topological features of various interstitial sites and their 
corresponding site-specific segregation energy of C in bcc Fe grain boundaries. Furthermore, this 
work also determines the quantitative relationship between the pairwise C-C interaction at bcc Fe 
grain boundaries and the C-C distance. Finally, this study investigates the segregation behavior of 
C at random grain boundaries under the influence of C-C interaction. 
Therefore, the structure and brief content of this thesis is as the followings: 
Chapter 1 presents the motivation, the background and the objectives of this study. Chapter 2 
describes the atomistic simulations methods for obtaining bcc Fe grain boundary models with 
various grain boundary characters and the grain boundary structural analysis method via 
persistent homology adopted in this work.  Specifically, molecular mechanics (MM) is adopted to 
obtain the large-scale atomic models of bcc Fe grain boundaries with various character, these 
include the symmetric tilt CSL boundaries, tilt low-angle grain boundaries and high-angle random 
grain boundaries. As for the computation of C segregation energy and C-C interaction energies, 
first-principles density functional theory (DFT) computation is adopted to ensure that the 
site-specific segregation energies as well as pairwise C-C interaction energies are obtained with 
adequate accuracy. 
Next, Chapter 3 is the investigation on the atomic-scale topological features of α-Fe grain 
boundaries. By conducting persistent homology analysis on large-scale bcc Fe grain boundary 
models, it is found that persistent homology can identify various open space units within the 
investigated grain boundaries. It is also found that for low-angle grain boundaries, the positions of 
the grain boundary open space units agree well with the dislocation lines identified via existing 
methods. Furthermore, systematical investigation on grain boundaries with various grain 
boundary characters indicates that the “birth” and “death” parameters of the identified grain 
boundary open space units are positively related with the grain boundary energy and thus 
negatively with grain boundary coherency. Compared with symmetric tilt CSL grain boundaries, 
random grain boundaries contain more varied types of open space units and wider structural 
width. These results demonstrate the ability of atomic structural analysis by persistent homology. 
Subsequently, Chapter 4 focuses on the effects of grain boundary topological features on C 
segregation energy. By combining the grain boundary atomic structure analysis ability of 
persistent homology with first-principles DFT calculations, the C segregation energies at various 
CSL grain boundaries of bcc Fe are investigated. It is demonstrated that centroids of CSL grain 
boundary open space units are stable segregation sites for C atoms, and the associated segregation 
energies show a negative correlation with the topological feature parameters (births and deaths) of 
the open space units. Furthermore, by identifying the interstitial segregation sites in C-segregated 
grain boundary structures by persistent homology, this work establishes quantitative relationship 
between the C-C pairwise interaction energy and C-C distance by DFT computation of system 
energy with the second C atoms placed at identified sites. 
Chapter 5 is the combined simulation-experimental investigation on C segregation at various 
bcc Fe grain boundaries. By integrating the grain boundary structural analysis of persistent 
homology, the relationship between grain boundary open space units and segregation energy and 
the distance-based pairwise C-C interaction obtained in this work, grand canonical Monte Carlo 
(GCMC) simulation scheme is set up to investigate the segregation behavior of C in grain 
boundaries at elevated temperature with various grain boundary characters and different bulk C 
contents. Such simulations results are also compared against the experimental observations on 
actual grain boundaries obtained via atom probe tomography (APT). Therefore, this chapter also 
contains the details on the experiment methods. It is shown such GCMC simulation scheme is able 
to qualitatively replicate the experimental results of C segregation at the corresponding actual bcc 
Fe grain boundaries regarding the effects of grain boundary character and bulk C content. It is 
found that the grain boundary character dependency of C segregation behavior, which indicates 
more C segregation in less coherent grain boundaries, is caused by the difference in their grain 
boundary topology and thus the different spectra of site-specific segregation energies. Meanwhile, 
the bulk content dependency of C segregation behavior, which indicates smaller magnitude of 
increase of segregation amount with the increase of bulk C content, is due to both the spectral 
distribution of segregation energies at different grain boundary sites and the repulsive C-C 
interaction in grain boundary region. 
Finally, Chapter 6 is the summary and the overall conclusions of this work. 
To sum up, this study developed a novel method for structural analysis of grain boundary 
atomic structure in bcc Fe via persistent homology, and a simulation scheme able to qualitatively 
predict the C segregation behavior at bcc Fe grain boundaries with respect of bulk C content and 
grain boundary characters, which helps clarifying the atomistic origins of grain boundary 
character and bulk C content dependency of segregation of C in bcc Fe grain boundaries, as 
indicated by the differences between the segregation energy spectra of different boundaries and the 
C-C interactions.  
However, there are still relatively large quantitative differences between the segregation 
amount of C atoms at bcc Fe grain boundaries given by the simulation scheme presented in this 
study and that observed experimentally on actual boundaries. Specifically, the simulations in this 
work generally yielded less C interfacial excess than the experimental observations. For the future 
work on developing quantitative predictions of C segregation at actual grain boundaries, it is 
suggested that more accurate representation of grain boundary structures, segregation energies 
and C-C interaction at elevated temperatures, especially the description of many-body C-C 
interaction at grain boundaries should be developed. Furthermore, it is suggested that a more 
flexible simulation scheme that is able to consider the grain boundary structural changes due to C 
segregation during simulation should be constructed.  
It is hoped that this study provides new insights to understanding and thus engineering of C 
segregation at bcc Fe grain boundaries. 
 


Not a member yet

Grammaticalization of Verbs in Mandarin Chinese: An Analysis in Terms of Cartography and Upward Reanalysis

Bioinspired Deep Reinforcement Learning for Energy-efficient Control of Redundant Robotic Systems

Compliant End-Effector for Dexterous Space Applications

Experimental Study of Aerodynamic Drag about Fundamental Bluff Bodies Using a Magnetic Suspension and Balance System

Study on Interactions between Electromagnetic Wave and Fluids in Subcritical Millimeter Wave Discharge

Development of a Reliability Evaluation Method for Flaw Detection using Eddy Current Testing under Low Signal-to-Noise Ratio

励起子分子の量子もつれに由来する光学応答の理論研究

ラマン分光法を用いた化学強化ガラスの応力測定装置の開発に関する研究

Topological Features of α-Fe Grain Boundary Structure and Its Connection with C Segregation

X線顕微分光法を用いたLiイオン電池正極活物質の化学状態ナノイメージング

81,990

87,477