Supercomputing System

Hongyi Li, Norihiko L. Okamoto, Takuya Hatakeyama, Yu Kumagai, Fumiyasu Oba and Tetsu Ichitsubo

Adv. Energy Mater., 8 (2018) 1801475. DOI:10.1002/aenm.201801475Jingxiang Xu, Yuji Higuchi, Nobuki Ozawa, Kazuhisa Sato, Toshiyuki Hashida and Momoji Kubo

ACS Appl. Mater. Interfaces, 9 (2017) 31816-31824. DOI:10.1021/acsami.7b07737

Maaouia Souissi, Marcel H. F. Sluiter, Tetsuya Matsunaga, Masaaki Tabuchi, Michael J. Mills and Ryoji Sahara

Sci. Rep., 8 (2018) 7279. DOI:10.1038/s41598-018-25642-y

Hiroshi Tsukahara, Kaoru Iwano, Chiharu Mitsumata, Tadashi Ishikawa and Kanta Ono

Joint MMM-Intermag conference 2019.

Parallel Large-Scale Molecular Dynamics Simulation on Effect of Porous Ni/YSZ Structure of Fuel Cell Electrode on Sintering Process of Ni Catalysts

Effects of grain boundary phases on magnetization reversal inside nanocrystalline permanent magnets

Fast Diffusion of Multivalent Ions Facilitated by Concerted Interactions in Dual-Ion Battery Systems

New understanding of complex γ-M23C6 formation in high-strength heat resistant steels

　To solve energy and environmental problems, the spread of solid oxide fuel cells capable of realizing high power generation efficiency is strongly desired. The anode of solid oxide fuel cell consists of porous composite materials of Ni and YSZ ceramics. Here, the Ni particles sintering in the porous Ni/YSZ structure during the operation is a large problem because it reduces the Ni surface area, leading to the Ni catalyst degradation.　The previous simulations with several thousands of atoms employ two Ni particles model on flat YSZ surface. This model can reveal the effect of the YSZ on the Ni sintering behaviors, however cannot elucidate the effects of the porosity and tortuosity of the porous structure, the composition ratio of Ni and YSZ, the size of Ni and YSZ particles, etc. on the Ni sintering behaviors in the Ni/YSZ structures. Therefore, the simulation results cannot contribute to the material design for sintering suppression.　In the present study, we developed a parallel large-scale molecular dynamics simulator for supercomputer, which can handle several million atoms. It enables us to clarify the effects of the porosity and tortuosity of the porous structure, the composition ratio of Ni and YSZ, the size of Ni and YSZ particles, etc. on the Ni sintering behaviors in the Ni/YSZ structures. As shown in figure, as the YSZ particle size decreases, the diffusion and movement of the Ni particles decrease, leading to the suppression of the Ni sintering. Such findings cannot be obtained by several thousand atoms simulation and then we conclude that our developed parallel large-scale molecular dynamics simulator realizes the clarification of the effects of the porous structure on the sintering behaviors, leading the theoretical design of the electrode materials in the fuel cell.

Minimum energy paths of Li- and Mg-ion diffusion in a Mo6S8 host with the Chevrel structure were computed by the first-principles ca lcula t ion combined wi th the nudged elastic band (NEB) method. (a) Activation energies computed for the diffusion processes of (orange) a Mg ion in a pristine Mo6S8 host, (green) a Li ion in a pristine Mo6S8 h o s t a n d ( b l u e ) a M g i o n i n a LiMo6S8 host, where a Li ion has been precedingly inserted. A schematic figure illustrating single Li- or Mg-ion diffusion in a Mo6S8 host is shown in the right side above the energy curves. The activation energy for Mg hopping in the LiMo6S8 host is reduced remarkably in Li-Mg dual-cation systems compared to that for Mg or Li hopping in the pristine Mo6S8 host, owing to the “concerted” motion of following Mg and leading Li ions with the Li-Mg distance being kept almost constant in the range of 3.5–4 Å during the diffusion process as indicated by the gray dotted curve. (b) Schematic figure illustrating the diffusion process of a Mg ion in a LiMo6S8 host. Mg and Li ions are colored orange and green, respectively. (i) and (v) show the initial and final states of the NEB calculation, respectively. Atomic arrangements of the inserted ions in several important configurations are illustrated as (ii)–(iv). The present results prove that dual-ion battery systems are a promising way to construct novel high-performance rechargeable batteries employing multivalent ions with an enhanced diffusion rate.

We have studied the mechanism of magnetization reversal within nanocrystalline permanent magnets using micromagnetic simulations. We simulate magnetization dynamics inside the simulation model which consists of cubic grains and grain boundary phases, as shown in (A). We consider two types of grain boundary phases as shown in (B) and (C). When the grain boundary phases are inserted only along the z-direction (Type A), the magnetization reversal regions construct the pillar shape structure as shown in (D). In contrast, the magnetization reversal region is placed randomly inside the permanent magnet having the grain boundary phases in the all-direction (Type C) as shown in (E). The magnetization reversal regions consist of different grains between Type A and C. Figure (F) and (G) show number of the grains having (red) positive and (blue) negative magnetization along the z-direction. In the case of Type A, magnetization reversal does not depend on the easy-axis orientation inside the grain. In contrast, the magnetization reversal occurs preferentially in the grains having the large tilt-angle of the easy axis from the z-direction inside the permanent magnet with Type C grain boundary phase.

The strengthening mechanism in materials is an important subject in metallurgy for many years. A material gets stronger when some obstacles are created within its polycrystalline microstructure to make it difficult for the dislocations to move. In steels, modeling these obstacles can be beyond the modeling of a single impurity atom to include the formation of complex carbide precipitations. Recently, a new understanding

23C6 carbides in high-temperature steels, which usually precipitate at the grain boundaries. Ab initio calculations combined with statistical thermodynamics approaches reveal that the site occupation of metal sites of the carbides may be incorrectly predicted when only the commonly used approach of full sublattice occupation is considered. An analysis of the electronic structure of atomic

23C6 supercell has revealed that these carbides are stabilized by the formation of super-ion in a NaCl t y pe spa t ia l a r r a ngement . Posit ively charged centered rhombic dodecahedral M(4a)M12

(48h) clusters alternate with negatively charged cuboctahedron M8

(32f)C6(24e) clusters, with additional metal atoms

in the cubic interstices. This will provide a new playground for in-depth analysis of the 23C6 in precipitate hardening and subsequently a better understanding of

the carbide microstructure.

Figure. Super-ion cluster morphology of γ-M23C6(M = Cr, Fe) stabilized by the chemical interactions of partially occupied metal sites: Incorrectly predicted by conventional model (upper figure) and correctly assessed by the present new model (lower figure).

Figure. Simulation Snapshots of Ni Sintering Behaviors in Porous Ni/YSZ Structure at 1273 K.- Ni/YSZ Multi-Nanoparticle Models with YSZ Particle Diameter of (a) 60 Å and（b）20 Å.

GreetingHead, Momoji KUBO

Supercomputing System for Computational Materials Science

Mission of Center for Computational Materials Science

Great expectations towards the computational materials science have been growing and growing in recent years in order to create prosperous and livable future society. Especially, for solving many problems facing humanity, such as addressing energy and environmental issues as well as establishing safe, secure, and sustainable society, computational materials science is expected to play key and crucial roles. The Ministry of Education, Culture, Sports, Science and Technology (MEXT) has been proceeding with “Post-K Project” since fiscal 2014, targeting the development of new post-K supercomputer which aims 100 times faster application execution performance than K supercomputer. In addition to the development of the post-K supercomputer, co-design with application software development is a characteristic of the “Post-K Project” . Under these circumstances, the supercomputer of Center for Computational Materials Science (CCMS) belongs to the second level in the domestic computational resource hierarchy where K or post-K supercomputer is in the first level and is strongly required to contribute to the development and advancement of materials science. In order to fulfil our mission above, CCMS provides computational resources to materials science community as an international collaborative research institute as well as promotes the development of application software for supercomputer and its application to materials science. CCMS also provides computational resources to the researchers who participate in the projects of Post-K Priority Issues, Post-K Exploratory Challenges, Professional development Consortium for Computational Materials Scientists (PCoMS), and Element Strategy Initiative as a project of Supercomputer Consortium for Computational Materials Science. Furthermore, CCMS cooperates with the Post-K and PCoMS projects, assists the parallelization of application software, and holds the lecture courses for application software. Since August 2018, CCMS has started the operation of new supercomputing system “MASAMUNE-IMR” , which has Cray XC50-LC as a main supercomputer. Total operation performance of this new system is 3 PFLOPS, which is 10 times faster than our previous supercomputing system of 300 TFLOPS. CCMS aims to clarify and confront the mission of CCMS in the materials science field, to promote the cooperation and integration with different research fields, and to advance the creative and innovative researches, contributing materials science as a leading supercomputer center with high originality and creativity. We would appreciate your continuous support and cooperation with CCMS.

The high-performance supercomputing system of CCMS consists of supercomputer Cray XC50-LC, supercomputer Cray CS-Storm 500GT with GPU accelerator, and other servers, in order to satisfy diverse research needs in computational materials science field. The total theoretical peak performance of the two main supercomputers is 3 PFLOPS. This system is specialized for materials science simulations, having various application software for materials design tuned for our supercomputer, and is provided to the domestic and foreign researchers of the materials science field.

Supercomputing system of Center for Computational Materials Science (CCMS) has been used by the researchers of materials science field as a collaborative research institute in Institute for Materials Research (IMR) through renovating the supercomputing systems several times since the first introduction of the supercomputing system in 1994. CCMS established frameworks as a domestic collaborative research institute in 2009 and has contributed to producing a wide variety of creative and innovative achievements in materials science field by providing the computational resources to materials science researchers. Furthermore, in 2018 CCMS is certified as an international collaborative research institute. The system renovation in 2018 makes it possible to provide new computational resources which satisfy diverse research needs such as GPU-based massively parallel computing, multi-scale simulations, and materials informatics, in addition to the conventional demands such as prediction of physical phenomena and materials design. This renovation aims at contributing to the innovation of materials research technologies for solving energy and environmental problems, to the development of device/electronics materials for creating prosperous and livable future society and for strengthening Japan's international competitiveness, and to developing new social infrastructure materials for realizing safe, secure, and sustainable society. CCMS also hopes to contribute to the advancement and promotion of computational materials science field and community in the world by releasing the leading-edge results obtained by the maximum utilization of our supercomputing system. Furthermore, CCMS provides the computational resources to the researchers who participate in the projects of Post-K Priority Issues, Post-K Exploratory Challenges, Professional development Consortium for Computational Materials Scientists (PCoMS), and Element Strategy Initiative as a project of Supercomputer Consortium for Computational Materials Science, cooperates with the Post-K and PCoMS projects, assists the parallelization of application software, and holds the lecture courses for application software, in order to provide active supports for the advancement and promotion of computational materials science field and community in the world.

Total Theoretical Peak Performance of Supercomputer: 3.03 PELOPS

System integrated by Hitachi, Ltd.

Supercomputer(Large-Scale Parallel Computing Server)

1

Storage System3

Visualization Server5 Network System6

Supercomputer(Accelerator Server)2

Parallel Computing & Informatics Server4

Materials Science Supercomputing System

■ “MASAMUNE-IMR” is a registered trademark of Tohoku University.

Nickname and Front Panel Design of Supercomputing System

▶Total Nodes 320 nodes

▶Total Theoretical Peak Performance 0.91 PFLOPS▶Total Memory Size 219.8 TiB

Cray XC50-LC

Cray CS-Storm 500GT

Computing Node 293 nodesI/O Node, etc . 27 nodes

Storage System

▶Total Disk Capacity 4.0 PB

DDN EXAScaler ES14KX

▶Total Disk Capacity 435.0 TB

DDN GRIDScaler SFA7700X

Parallel Computing & Informatics Server

▶Total Nodes 29 nodes ▶Total Theoretical Peak Performance 0.10 PFLOPS▶Total Memory Size 16.3 TiB

HPE ProLiant DL360 Gen10

HPE ProLiant DL380 Gen10

Visualization Server

Network System

HPE ProLiant DL360 Gen10

▶Total Nodes 29 nodes▶Total Theoretical Peak Performance 2.12 PFLOPS

▶Total Memory Size 21.8 TiB

CPU 0.09 PFLOPSGPU 2.03 PFLOPS

Supercomputing system network(Ethernet 10 Gbps , 1 Gbps)

Storage area network(InfiniBand EDR 100 Gbps)

Supercomputer (Large-Scale Parallel Computing Server)1

Supercomputer (Accelerator Server)2

3

4

5

6

MASAMUNE-IMR MAterials science Supercomputing system for Advanced MUlti-scale simulations towards NExt-generation - Institute for Materials Research

Supercomputing System Configuration Diagram

Nickname “MASAMUNE-IMR” is given to the supercomputing system of CCMS. This nickname is from “Masamune Date” who was the first lord of Sendai in the 17th century and dispatched the sailing ship “San Juan Bautista” in 1613 towards the world from Sendai. “Masamune Date” is drew on the front panel of our supercomputer by Sumi-E Artist OKAZU, with the letters “MASAMUNE-IMR” . This “Masamune Date” on the front panel is looking ahead to the next-generation materials science of the world from Sendai. In the nickname, “MASAMUNE-IMR” , we express our hope that creative and innovative achievements of the advanced multi-scale simulations on materials science obtained by our supercomputing system will give great impacts towards the next-generation from Sendai to the world.

Hongyi Li, Norihiko L. Okamoto, Takuya Hatakeyama, Yu Kumagai, Fumiyasu Oba and Tetsu Ichitsubo

Adv. Energy Mater., 8 (2018) 1801475. DOI:10.1002/aenm.201801475Jingxiang Xu, Yuji Higuchi, Nobuki Ozawa, Kazuhisa Sato, Toshiyuki Hashida and Momoji Kubo

ACS Appl. Mater. Interfaces, 9 (2017) 31816-31824. DOI:10.1021/acsami.7b07737

Maaouia Souissi, Marcel H. F. Sluiter, Tetsuya Matsunaga, Masaaki Tabuchi, Michael J. Mills and Ryoji Sahara

Sci. Rep., 8 (2018) 7279. DOI:10.1038/s41598-018-25642-y

Hiroshi Tsukahara, Kaoru Iwano, Chiharu Mitsumata, Tadashi Ishikawa and Kanta Ono

Joint MMM-Intermag conference 2019.

Parallel Large-Scale Molecular Dynamics Simulation on Effect of Porous Ni/YSZ Structure of Fuel Cell Electrode on Sintering Process of Ni Catalysts

Effects of grain boundary phases on magnetization reversal inside nanocrystalline permanent magnets

Fast Diffusion of Multivalent Ions Facilitated by Concerted Interactions in Dual-Ion Battery Systems

New understanding of complex γ-M23C6 formation in high-strength heat resistant steels

　To solve energy and environmental problems, the spread of solid oxide fuel cells capable of realizing high power generation efficiency is strongly desired. The anode of solid oxide fuel cell consists of porous composite materials of Ni and YSZ ceramics. Here, the Ni particles sintering in the porous Ni/YSZ structure during the operation is a large problem because it reduces the Ni surface area, leading to the Ni catalyst degradation.　The previous simulations with several thousands of atoms employ two Ni particles model on flat YSZ surface. This model can reveal the effect of the YSZ on the Ni sintering behaviors, however cannot elucidate the effects of the porosity and tortuosity of the porous structure, the composition ratio of Ni and YSZ, the size of Ni and YSZ particles, etc. on the Ni sintering behaviors in the Ni/YSZ structures. Therefore, the simulation results cannot contribute to the material design for sintering suppression.　In the present study, we developed a parallel large-scale molecular dynamics simulator for supercomputer, which can handle several million atoms. It enables us to clarify the effects of the porosity and tortuosity of the porous structure, the composition ratio of Ni and YSZ, the size of Ni and YSZ particles, etc. on the Ni sintering behaviors in the Ni/YSZ structures. As shown in figure, as the YSZ particle size decreases, the diffusion and movement of the Ni particles decrease, leading to the suppression of the Ni sintering. Such findings cannot be obtained by several thousand atoms simulation and then we conclude that our developed parallel large-scale molecular dynamics simulator realizes the clarification of the effects of the porous structure on the sintering behaviors, leading the theoretical design of the electrode materials in the fuel cell.

Minimum energy paths of Li- and Mg-ion diffusion in a Mo6S8 host with the Chevrel structure were computed by the first-principles ca lcula t ion combined wi th the nudged elastic band (NEB) method. (a) Activation energies computed for the diffusion processes of (orange) a Mg ion in a pristine Mo6S8 host, (green) a Li ion in a pristine Mo6S8 h o s t a n d ( b l u e ) a M g i o n i n a LiMo6S8 host, where a Li ion has been precedingly inserted. A schematic figure illustrating single Li- or Mg-ion diffusion in a Mo6S8 host is shown in the right side above the energy curves. The activation energy for Mg hopping in the LiMo6S8 host is reduced remarkably in Li-Mg dual-cation systems compared to that for Mg or Li hopping in the pristine Mo6S8 host, owing to the “concerted” motion of following Mg and leading Li ions with the Li-Mg distance being kept almost constant in the range of 3.5–4 Å during the diffusion process as indicated by the gray dotted curve. (b) Schematic figure illustrating the diffusion process of a Mg ion in a LiMo6S8 host. Mg and Li ions are colored orange and green, respectively. (i) and (v) show the initial and final states of the NEB calculation, respectively. Atomic arrangements of the inserted ions in several important configurations are illustrated as (ii)–(iv). The present results prove that dual-ion battery systems are a promising way to construct novel high-performance rechargeable batteries employing multivalent ions with an enhanced diffusion rate.

We have studied the mechanism of magnetization reversal within nanocrystalline permanent magnets using micromagnetic simulations. We simulate magnetization dynamics inside the simulation model which consists of cubic grains and grain boundary phases, as shown in (A). We consider two types of grain boundary phases as shown in (B) and (C). When the grain boundary phases are inserted only along the z-direction (Type A), the magnetization reversal regions construct the pillar shape structure as shown in (D). In contrast, the magnetization reversal region is placed randomly inside the permanent magnet having the grain boundary phases in the all-direction (Type C) as shown in (E). The magnetization reversal regions consist of different grains between Type A and C. Figure (F) and (G) show number of the grains having (red) positive and (blue) negative magnetization along the z-direction. In the case of Type A, magnetization reversal does not depend on the easy-axis orientation inside the grain. In contrast, the magnetization reversal occurs preferentially in the grains having the large tilt-angle of the easy axis from the z-direction inside the permanent magnet with Type C grain boundary phase.

The strengthening mechanism in materials is an important subject in metallurgy for many years. A material gets stronger when some obstacles are created within its polycrystalline microstructure to make it difficult for the dislocations to move. In steels, modeling these obstacles can be beyond the modeling of a single impurity atom to include the formation of complex carbide precipitations. Recently, a new understanding

23C6 carbides in high-temperature steels, which usually precipitate at the grain boundaries. Ab initio calculations combined with statistical thermodynamics approaches reveal that the site occupation of metal sites of the carbides may be incorrectly predicted when only the commonly used approach of full sublattice occupation is considered. An analysis of the electronic structure of atomic

23C6 supercell has revealed that these carbides are stabilized by the formation of super-ion in a NaCl t y pe spa t ia l a r r a ngement . Posit ively charged centered rhombic dodecahedral M(4a)M12

(48h) clusters alternate with negatively charged cuboctahedron M8

(32f)C6(24e) clusters, with additional metal atoms

in the cubic interstices. This will provide a new playground for in-depth analysis of the 23C6 in precipitate hardening and subsequently a better understanding of

the carbide microstructure.

Figure. Super-ion cluster morphology of γ-M23C6(M = Cr, Fe) stabilized by the chemical interactions of partially occupied metal sites: Incorrectly predicted by conventional model (upper figure) and correctly assessed by the present new model (lower figure).

Figure. Simulation Snapshots of Ni Sintering Behaviors in Porous Ni/YSZ Structure at 1273 K.- Ni/YSZ Multi-Nanoparticle Models with YSZ Particle Diameter of (a) 60 Å and（b）20 Å.

Hongyi Li, Norihiko L. Okamoto, Takuya Hatakeyama, Yu Kumagai, Fumiyasu Oba and Tetsu Ichitsubo

Adv. Energy Mater., 8 (2018) 1801475. DOI:10.1002/aenm.201801475Jingxiang Xu, Yuji Higuchi, Nobuki Ozawa, Kazuhisa Sato, Toshiyuki Hashida and Momoji Kubo

ACS Appl. Mater. Interfaces, 9 (2017) 31816-31824. DOI:10.1021/acsami.7b07737

Maaouia Souissi, Marcel H. F. Sluiter, Tetsuya Matsunaga, Masaaki Tabuchi, Michael J. Mills and Ryoji Sahara

Sci. Rep., 8 (2018) 7279. DOI:10.1038/s41598-018-25642-y

Hiroshi Tsukahara, Kaoru Iwano, Chiharu Mitsumata, Tadashi Ishikawa and Kanta Ono

Joint MMM-Intermag conference 2019.

Parallel Large-Scale Molecular Dynamics Simulation on Effect of Porous Ni/YSZ Structure of Fuel Cell Electrode on Sintering Process of Ni Catalysts

Effects of grain boundary phases on magnetization reversal inside nanocrystalline permanent magnets

Fast Diffusion of Multivalent Ions Facilitated by Concerted Interactions in Dual-Ion Battery Systems

New understanding of complex γ-M23C6 formation in high-strength heat resistant steels

　To solve energy and environmental problems, the spread of solid oxide fuel cells capable of realizing high power generation efficiency is strongly desired. The anode of solid oxide fuel cell consists of porous composite materials of Ni and YSZ ceramics. Here, the Ni particles sintering in the porous Ni/YSZ structure during the operation is a large problem because it reduces the Ni surface area, leading to the Ni catalyst degradation.　The previous simulations with several thousands of atoms employ two Ni particles model on flat YSZ surface. This model can reveal the effect of the YSZ on the Ni sintering behaviors, however cannot elucidate the effects of the porosity and tortuosity of the porous structure, the composition ratio of Ni and YSZ, the size of Ni and YSZ particles, etc. on the Ni sintering behaviors in the Ni/YSZ structures. Therefore, the simulation results cannot contribute to the material design for sintering suppression.　In the present study, we developed a parallel large-scale molecular dynamics simulator for supercomputer, which can handle several million atoms. It enables us to clarify the effects of the porosity and tortuosity of the porous structure, the composition ratio of Ni and YSZ, the size of Ni and YSZ particles, etc. on the Ni sintering behaviors in the Ni/YSZ structures. As shown in figure, as the YSZ particle size decreases, the diffusion and movement of the Ni particles decrease, leading to the suppression of the Ni sintering. Such findings cannot be obtained by several thousand atoms simulation and then we conclude that our developed parallel large-scale molecular dynamics simulator realizes the clarification of the effects of the porous structure on the sintering behaviors, leading the theoretical design of the electrode materials in the fuel cell.

Minimum energy paths of Li- and Mg-ion diffusion in a Mo6S8 host with the Chevrel structure were computed by the first-principles ca lcula t ion combined wi th the nudged elastic band (NEB) method. (a) Activation energies computed for the diffusion processes of (orange) a Mg ion in a pristine Mo6S8 host, (green) a Li ion in a pristine Mo6S8 h o s t a n d ( b l u e ) a M g i o n i n a LiMo6S8 host, where a Li ion has been precedingly inserted. A schematic figure illustrating single Li- or Mg-ion diffusion in a Mo6S8 host is shown in the right side above the energy curves. The activation energy for Mg hopping in the LiMo6S8 host is reduced remarkably in Li-Mg dual-cation systems compared to that for Mg or Li hopping in the pristine Mo6S8 host, owing to the “concerted” motion of following Mg and leading Li ions with the Li-Mg distance being kept almost constant in the range of 3.5–4 Å during the diffusion process as indicated by the gray dotted curve. (b) Schematic figure illustrating the diffusion process of a Mg ion in a LiMo6S8 host. Mg and Li ions are colored orange and green, respectively. (i) and (v) show the initial and final states of the NEB calculation, respectively. Atomic arrangements of the inserted ions in several important configurations are illustrated as (ii)–(iv). The present results prove that dual-ion battery systems are a promising way to construct novel high-performance rechargeable batteries employing multivalent ions with an enhanced diffusion rate.

We have studied the mechanism of magnetization reversal within nanocrystalline permanent magnets using micromagnetic simulations. We simulate magnetization dynamics inside the simulation model which consists of cubic grains and grain boundary phases, as shown in (A). We consider two types of grain boundary phases as shown in (B) and (C). When the grain boundary phases are inserted only along the z-direction (Type A), the magnetization reversal regions construct the pillar shape structure as shown in (D). In contrast, the magnetization reversal region is placed randomly inside the permanent magnet having the grain boundary phases in the all-direction (Type C) as shown in (E). The magnetization reversal regions consist of different grains between Type A and C. Figure (F) and (G) show number of the grains having (red) positive and (blue) negative magnetization along the z-direction. In the case of Type A, magnetization reversal does not depend on the easy-axis orientation inside the grain. In contrast, the magnetization reversal occurs preferentially in the grains having the large tilt-angle of the easy axis from the z-direction inside the permanent magnet with Type C grain boundary phase.

The strengthening mechanism in materials is an important subject in metallurgy for many years. A material gets stronger when some obstacles are created within its polycrystalline microstructure to make it difficult for the dislocations to move. In steels, modeling these obstacles can be beyond the modeling of a single impurity atom to include the formation of complex carbide precipitations. Recently, a new understanding

23C6 carbides in high-temperature steels, which usually precipitate at the grain boundaries. Ab initio calculations combined with statistical thermodynamics approaches reveal that the site occupation of metal sites of the carbides may be incorrectly predicted when only the commonly used approach of full sublattice occupation is considered. An analysis of the electronic structure of atomic

23C6 supercell has revealed that these carbides are stabilized by the formation of super-ion in a NaCl t y pe spa t ia l a r r a ngement . Posit ively charged centered rhombic dodecahedral M(4a)M12

(48h) clusters alternate with negatively charged cuboctahedron M8

(32f)C6(24e) clusters, with additional metal atoms

in the cubic interstices. This will provide a new playground for in-depth analysis of the 23C6 in precipitate hardening and subsequently a better understanding of

the carbide microstructure.

Figure. Super-ion cluster morphology of γ-M23C6(M = Cr, Fe) stabilized by the chemical interactions of partially occupied metal sites: Incorrectly predicted by conventional model (upper figure) and correctly assessed by the present new model (lower figure).

Figure. Simulation Snapshots of Ni Sintering Behaviors in Porous Ni/YSZ Structure at 1273 K.- Ni/YSZ Multi-Nanoparticle Models with YSZ Particle Diameter of (a) 60 Å and（b）20 Å.

Figure.Figure. Super-ion cluster morphology of γ-M23CC6(M = Cr, Fe) stabilized by (M = Cr, Fe) stabilized by

Japan

Sendai

TokyoNarita Int’l AirportHaneda Airport

NagoyaChubu Centrair Int’l Airport

2-1-1 Katahira, Aoba-ku, Sendai, 980-8577, JAPANTEL: +81-22-215-2411, FAX: +81-22-215-2166

https://www.sc.imr.tohoku.ac.jp/eng/E-mail: ccms-adm@imr.tohoku.ac.jp

15 min 15 min

1 hour

1 hour 1 hour 30 min

15 min

1 hour 30 min25 min

10 min

Haneda Airport (Tokyo/Haneda)

HamamatsuchoStation

ShinagawaStation

Narita InternationalAirport (Tokyo/Narita)

Chubu Centrair InternationalAirport (Nagoya)

Sendai Airport

JR Tokyo Station

JR Sendai Station

Airplane Airplane

A Taxi Stand (1F)

Institute for Materials Research, Tohoku University

Access

Tohoku Shinkansen

Tokyo Monorail Keikyu Line Narita Express

Sendai Airport Access Line

JR Yamanote Line

Tohoku Univ.North Gate

University HouseKatahira

HondaMemorial Hall

Bldg. 1

Information●

Bldg. 3

Bldg. 4

Multi-UseLaboratory

TechnicalPlant 2

Lab. of Low Temp.Mat. Science

High Field Lab. forSuperconducting Mat.

TechnicalPlant 1

P

The Open Universityof Japan

Katahira Hall

to Sendai Station

IMRNorth Gate

IMRMain Gate

Bldg. 2

N

Campus Map

Center for ComputationalMaterials Science