高分子 Vol.71 No.12
>> Japanese >> English
特集 今さらだけど、今なら聞ける高分子科学のための統計学
展望 COVER STORY: Highlight Reviews
立体規則性分布およびモノマー連鎖分布の統計的扱い方
Statistical Treatment for Distributions of Stereosequence and Monomer Sequence in Vinyl (Co)polymers
平野 朋広
Tomohiro HIRANO
<要旨> 高分子材料の物性は一次構造の影響を強く受ける。最も重要なものは立体規則性とモノマー連鎖である。精密重合系の著しい発展とともに高度に構造規制されたポリマーが合成され、精密構造解析の重要性が再認識されている。本稿では、ビニルポリマーの立体規則性およびモノマー連鎖分布を統計的に扱う際の基礎的な考え方について述べる。
Keywords: Stereosequence / Chain-End Control / Bernoullian Statistics / Markovian Statistics / Monomer Sequence / Terminal Model / Monomer Reactivity Ratio
ページトップへ▲
高分子の統計力学的取り扱い
Statistical Mechanical Approach to Polymers
畝山 多加志
Takashi UNEYAMA
<要旨> 統計力学は統計的手法を用いた物理の一分野であり、高分子の物性を調べるために広く用いられている。統計学的な側面を強調しつつ統計力学の基本的な考え方と高分子への応用を説明する。また、基本的な統計力学の手法が使えないような特殊な高分子系を扱うための最近の方法について紹介する。
Keywords: Statistical Mechanics / Statistics / Most Probable Distribution / Transient Potential
ページトップへ▲
材料開発における四つのインフォマティクス
Four Informatics in Material Science
岩崎 悠真
Yuma IWASAKI
<要旨> 近年、データ駆動材料科学は非常に速いスピードで開発が進められており、さまざまな技術が登場している。そこで本記事では、『マテリアルズインフォマティクス』、『プロセスインフォマティクス』、『計測インフォマティクス』、『物理インフォマティクス』の四つの視点からデータ駆動材料科学を俯瞰する。
Keywords: Materials Informatics / Process Informatics / Measurement Informatics / Physics Informatics
ページトップへ▲
トピックス COVER STORY: Topics and Products
分子量分布を利用した物性制御技術
Improvements of Rheological and Mechanical Properties of Semi-Crystalline Polymers by Controlling Molecular Weight Distribution
木田 拓充
Takumitsu KIDA
<要旨> Molecular weight distribution (MWD) is one of the most important molecular parameters affecting the rheological and mechanical properties of polymers. The addition of a small amount of an ultra-high-molecular-weight (UHMW) component resulted in the enhancement of the orientation of the crystalline structure during a flow-induced crystallization due to the long relaxation time of the stretching of the UHMW component. Moreover, the regularity of the crystalline structure was also improved by adding the UHMW component because the highly oriented UHMW chains acted as nucleating agents. In the case of the solid-state mechanical properties, the addition of the UHMW component led to improve the strength and strain hardening. The improvements of the strength and the strain hardening were caused by the increase in the number of tie molecules connecting more than 6 lamellar crystalline layers because the latter tie molecules acted as stress transmitters between the lamellar cluster units.
Keywords: Molecular Weight Distribution / Rheology / Mechanical Properties / Morphology / Tie Molecule
ページトップへ▲
少数データに基づくモデルの予測性能および適用範囲を考慮したデータ解析・機械学習
Data Analysis and Machine Learning with Small Data Considering Predictability and Applicability of Mathematical Models
金子 弘昌
Hiromasa KANEKO
<要旨> Chemical and chemical engineering data are used to design molecules, materials, and processes with machine learning. In molecular design, a model is constructed between properties and activities of molecules, and molecular descriptors, and new chemical structures are designed based on the model. In material design, a new model is constructed between the properties and activities, and characteristics of materials and experimental and manufacturing conditions of materials. In constructing a model with a small number of data, it is important to consider not only the predictive ability of the models but also the applicability domain (AD) of the models. When the design of experimental conditions based on a model and experiments are repeated, Bayesian optimization (BO) is effective for exploring exterior regions of experimental conditions. In this article, AD, BO, and applications of AD and BO are explained, and then, the web service to perform AD and BO is introduced.
Keywords: Small Data / Machine Learning / Applicability Domain / Bayesian Optimization / Data Chemical LAB / Molecular Design / Material Design / Process Design
ページトップへ▲
物質探索におけるベイズ統計の活用
Bayesian Approach to Exploring Novel Materials
本郷 研太
Kenta HONGO
<要旨> Materials Informatics (MI) has attracted enormous attention in materials science. About a decade ago, computational material design relied only on materials simulations, considering a small search space limited to tens to hundreds of compounds. On the other hand, MI such as high-throughput virtual screening (HTVS) approaches based on machine learning techniques expands its search space into that in the order of hundred thousand of compounds. The Bayesian statistics has been applied to explore novel materials and expanded its search space larger than the HTVS approach. This Bayesian material exploration has been applied to search novel polymers with high thermal conductivity (high-κ), where the transfer learning model has been used to construct the prediction model of thermal conductivity with a small dataset of thermal conductivity. To further search monomer units for high-κ polymers, first-principles phonon simulations have been also employed to evaluate thermal conductivities of polymer crystals taken from the Polymer Genome database.
Keywords: Materials Informatics / Bayesian Statistics / Inverse Problems / Machine Learning / Natural Language Processing / Thermal Conductivity / First-Principles Phonon Simulations
ページトップへ▲
シミュレーションを用いた高分子材料の開発
Development of Polymer Materials Using Coarse Grained Molecular Dynamics Simulation
冨永 哲雄・森下 和哉
Tetsuo TOMINAGA, Kazuya MORISHITA
<要旨> We describe two examples of applications of coarse grained molecular dynamics (CGMD) simulation to develop new polymer materials. The first is end-modified SBR (Styrene–Butadiene Rubber) and the other is hydrogenated SBR. The characteristics of end-modified SBR is high dispersibility of filler in rubber, and in order to incorporate this into the simulation model we used data of synchrotron radiation experiments. Since hydrogenated SBR has a high entanglement density, we added the angle potential to the Kremer-Grest model, which is widely used in CGMD, in order to construct polymer models with different entanglement densities. The mechanism of physical properties of the respective materials could be described by these models. CGMD simulation is revealed to be useful for the clarification of the mechanism of physical properties of polymer materials from these examples, and we expect it will further develop as a tool indispensable for development of polymer materials.
Keywords: Course Grained Molecular Dynamics Simulation / Polymer Materials / End-Modified SBR / Hydrogenated SBR / Filler Dispersion / Entanglement Density
ページトップへ▲
グローイングポリマー Polymer Science and I: A Personal Account
国境と分野の垣根を越えて
Crossing Borders and Disciplines
成田 明光
Akimitsu NARITA
<要旨> The participation in a Marie Curie Initial Training Network “SUPERIOR” during my Ph.D. studies in Germany provided me precious experiences to approach common goals through international and interdisciplinary collaborations, without restrictions of borders and disciplines. Such collaborative efforts have been essential for the characterization of scarcely soluble graphene nanoribbons.
ページトップへ▲
高分子科学最近の進歩 Front-Line Polymer Science
DNAナノ構造体に配置された酵素の機能
Enzyme Function on DNA Nanostructure
中田 栄司・森井 孝
Eiji NAKATA, Takashi MORII
<要旨> In the cell, enzymes are spatially organized, either in close proximity on the cell membrane or confined inside a micro-compartment. Such environments are belived to play key roles in enabling the extraordinary efficiency and specificity of metabolic enzymatic reactions. Inspired by the nature systems, individual or multi enzyme compleses have been interested to be spatially organized on a platform. Because of the structual programmability and accurate addressability of DNA, DNA nanostructure are ideal characteristics for the platform of assembling enzymes with the nanoscale precisions. Especially, a typical exampe of DNA nanostructures, DNA origami provides ideal platforms for the assembly of various functional macromolecules including enzymes. The most demanding task in constructing enzyme modified DNA nanostructure is the method to assemble enzyme on DNA nanostructure at high yields while retaining their activity. Here, an example of assembling methods and its applications were introduced.
Keywords: DNA Nanostrucutre / Enzyme / DNA-Protein Conjugation / Cascade Reaction
ページトップへ▲