月別: 2019年3月

[Published online Journal of Computer Chemistry, Japan Vol.17, 225-227, by J-STAGE]
<Title:> 赤外分光法と新規判別分析によるセルロース繊維種の判別
<Author(s):> 坂本　幸祐, 菅野　麻奈美, 吉村　季織, 高柳　正夫
<Corresponding author E-Mill:> s187475q(at)st.go.tuat.ac.jp
<Abstract:> As a new method for discrimination analysis applicable to data containing a large number of variables such as spectra, FDOD (Fisher’s Discriminant Orthogonal Decomposition) was developed by incorporating regularization and orthogonal decomposition into ordinary FDA (Fisher’s Discriminant Analysis). Here, FDOD was applied to distinguish infrared absorption (IR) spectra of rayon and modal. Since rayon and modal are very similar recycled fibers, their IR spectra cannot be discriminated by score plots based on principle component analysis (PCA). By using FDOD, their discrimination with a correct answer rate of 94% has been achieved.
<Keywords:> 判別分析, 正則化, 直交分解, 赤外分光法, 再生繊維
<URL:> https://www.jstage.jst.go.jp/article/jccj/17/5/17_2018-0058/_article/-char/ja/

[Published online Journal of Computer Chemistry, Japan Vol.17, 228-231, by J-STAGE]
<Title:> 特徴ベクトルから分子構造を生成するアルゴリズム
<Author(s):> 陸　明浩, 安田　耕二
<Corresponding author E-Mill:> yasudak(at)imass.nagoya-u.ac.jp
<Abstract:> Machine learning for predicting properties of molecules and designing molecules with desired properties requires suitable numerical representation of molecular structure. In this study we propose to represent a molecular graph as a high dimensional feature vector by counting the number of substructures of a molecule, and report conditions that these vectors should satisfy. The vectors are expanded by the lattice basis extracted from the feature vectors of a large molecular dataset. In addition, an efficient algorithm to reconstruct molecular structures from the feature vector is presented. The latent space provides a continuous and concise representation of molecular graphs, and the representation allows smooth interpolation between two molecules. The proposed method has wide application to studies of quantitative structure activity relationships and machine learning of molecules.
<Keywords:> KEYWORDS Machine learning, Feature vector, Molecular graph reconstruction, De Bruijn graph, Lattice basis
<URL:> https://www.jstage.jst.go.jp/article/jccj/17/5/17_2018-0056/_article/-char/ja/

[Published online Journal of Computer Chemistry, Japan -International Edition Vol.5, -, by J-STAGE]
<Title:> Revisiting the Nature of Si-O-Si Bridging
<Author(s):> Fumiya NORITAKE
<Corresponding author E-Mill:> fnoritake(at)yamanashi.ac.jp
<Abstract:> Non-empirical calculation based on the Schr dinger equation is an appropriate tool for investigating the relationship among Si-O-Si angle, Si-O bond length, Si-O bond strength, and electronic structure. However, past studies could not reach a consensus about the equilibrium structure of the C_2v pyrosilisic acid molecule. Moreover, the structure of disiloxane, the simplest siloxane molecule, could not be reproduced using non-empirical molecular orbital calculations. In this study, I checked the reproducibility of various model chemistries and basis sets, and found that employing the post-Hartree-Fock method and a larger basis set (at least, aug-cc-pVTZ) is necessary for accurate calculation of the disiloxane molecule. In contrast to past studies on molecular orbitals, the present study reveals no significant occupancy in the Si 3d orbitals. The total energy landscape of the C_2v pyrosilisic acid molecule is calculated by using the coupled cluster method concerning three excited electrons and the aug-cc-pVTZ basis set. The stable bond length for Si-O_br is 1.604 , and the stable Si-O-Si angle is 159.449°. There are gentle curves around the stable angles for each bond length comparing with bond length direction. The stable angle for each bond length decreased with increasing Si-O_br bond length. The weakening of the Si-O_br bond with decreasing Si-O-Si bond angle can be explained by the decrease in the bond index and the increase in the orbital energy for Si-O_br σ-bond. Consequently, hybridization of the valence electrons of the bridging oxygen with decreasing Si-O-Si angle weakens the Si-O_br σ-bond. Electrostatic potential favors a straight configuration because of the repulsion between the SiO₄ tetrahedra, while the valence electrons of the bridging oxygen favor a bent configuration. These two competing behaviors can explain the bent configuration of pyrosilisic acid without considering d-p π bonding.
<Keywords:> SIlicate, Pyrosilisic Acid, Disiloxane, Benchmark, Molecular Orbital Calculations
<URL:> https://www.jstage.jst.go.jp/article/jccjie/5/0/5_2018-0016/_html

[Published online Journal of Computer Chemistry, Japan -International Edition Vol.5, -, by J-STAGE]
<Title:> Development of PolyParGen Software to Facilitate the Determination of Molecular Dynamics Simulation Parameters for Polymers
<Author(s):> Makoto YABE, Kazuki MORI, Kazuyoshi UEDA, Minoru TAKEDA
<Corresponding author E-Mill:> kazuki.mori.013(at)ctc-g.co.jp
<Abstract:> In the case that the parameters to describe the force field, such as bond angles and charges, cannot be added to the library of a molecular dynamics (MD) simulation, self-development of the force field should be considered by performing quantum mechanics calculations and/or utilizing an automatic parameter generation tool. However, these techniques are not suitable for macromolecules with a large number of atoms. Typically, the force field of an oligomer containing three unit structures (a unit at both ends and a repeating unit at the center) is calculated and converted to polymer form (both ends + central part × n). Considering this, we recently developed the program o2p, which is a semi-automated program designed to set up the force field for polymers with repeating structures. However, it is difficult to apply this method to macromolecules with complex repeating structures. Thus, in this project, we developed PolyParGen, a new open-source automatic force field generation program for Gromacs that can relatively easily and reliably simulate the MD of complex macromolecules. The proposed program (1) divides the structure of the polymer into substructures with a number of atoms within the limit of the handling size for the automatic parameter generation tool program; then, (2) acquire the parameters for each divided substructure, and finally, (3) combine the parameters of these substructures to obtain the parameters for the whole polymer. By automating these processes, it is possible to acquire a parameter of a polymer having complicated structures. This program was evaluated by simulating the polymers P3EHT and F-P3EHT in chloroform. In agreement with previous reports, fluorination was found to cause F-P3EHT to adopt an extended structure, thereby indicating the effectiveness of the proposed program.
<Keywords:> Molecular dynamics simulation, Macromolecules, Parameters, Gromacs, Polymer
<URL:> https://www.jstage.jst.go.jp/article/jccjie/5/0/5_2018-0034/_html

[Published online Journal of Computer Chemistry, Japan Vol.17, 215-216, by J-STAGE]
<Title:> 分子軌道法を用いた多層グラフェン間への水素吸蔵特性の研究
<Author(s):> 菅谷　大智, 斎藤　秀俊, 内田　希
<Corresponding author E-Mill:> sugaya(at)stn.nagaokaut.ac.jp
<Abstract:> The activated carbon synthesized from rice husks has high hydrogen storage on 77K. It is inferred that activated carbon has a lot of defect on the surface compared with graphene. However, the actual structure and hydrogen storage mechanism have unknown by experiments. The purpose of this study is to clarify the hydrogen storage mechanism for multilayered graphene. The sp² hybrid system carbon structure that introduced a defect was prepared and set a hydrogen molecule between the layers. MOPAC PM7 involving the dispersion term for semiempirical MO was used for calculating heat of formation. From the potential curve between graphene layer where hydrogen was inserted, it is suggested that when a defect made on the edge, the structure becomes more stable.
<Keywords:> graphene layer, Carbon defect, Hydrogen storage, Semiempirical MO, Dispersion force
<URL:> https://www.jstage.jst.go.jp/article/jccj/17/5/17_2018-0047/_article/-char/ja/

[Published online Journal of Computer Chemistry, Japan Vol.17, 217-218, by J-STAGE]
<Title:> Rh触媒による3-フェニルチオフェンとスチレンのカップリング反応における位置選択性の理論的研究
<Author(s):> 林　慶浩, 佐藤　哲也, 三浦　雅博, 川内　進
<Corresponding author E-Mill:> skawauch(at)polymer.titech.ac.jp
<Abstract:> Regioselectivity in the Rh-catalyzed coupling reaction of 3-phenylthiophene with styrene was investigated by quantum chemistry calculations. The coupling reaction experimentally reported that styrene is selectively coupled to the phenyl group of the phenylthiophene. Results of the reaction path search and the natural bond orbital analysis indicate that the coupling position is determined by the preference of the styrene insertion into the 5-membered rhodacycle intermediate, which is produced by the double C H bond cleavage of 3-phenylthiophene in presence of the Rh-catalyst.
<Keywords:> カップリング反応, C-H官能基化, ロジウム触媒, 反応機構, DFT計算
<URL:> 
https://www.jstage.jst.go.jp/article/jccj/17/5/17_2018-0063/_article/-char/ja/