Functions

First-Principles (FP) simulations make use the basic laws of quantum mechanics to give an insight of material properties such as electronic, mechanical, magnetic properties without making assumptions in empirical models or using fitting parameters. DFT (Density Functional Theory) calculation is one of the most popular methods in the FP simulation field. Current computational power enables to calculate properties of several hundreds to thousands of atoms with DFT.

SIESTA is a registered trade-mark of SIMUNE.
SIMUNE Website >>

• 

  SIESTA interfacial energy tool

  This tool gives an easy way to evaluate interaction energies between a surface and a molecule with SIESTA. It is possible to estimate surface adsorption energies from the interaction energy curve. Based on the interaction energies, force field parameters for molecular dynamics simulation can be estimated. Case Study: Example of Calculation with SIESTA

• 

  SIESTA modeler

  This modeler gives an easy way to calculate stable structures, elastic moduli, reaction paths and so on using SIESTA. The activation energy, which is used as an input parameter of the reaction modeling, can be obtained from the reaction path analysis. Case Study: Cross-Linking of Epoxy Resin by the Monte Carlo method using activation energyDetail information of SIESTA modeler

Estimate the phase separation structure and the interface geometry of materials with various molecular structure and block copolymers by using the mean field method (SUSHI) and the dissipative particle dynamics method (COGNAC-DPD).

• 

  Estimate calculation of the interaction parameters

  Estimate the interaction parameter (χ parameter). (1) Estimate the SP value (solubility parameter) by using the group contribution method, QSPR, or molecular dynamics to obtain the χ parameter. (2) Obtain the χ parameter based on the contact energy between the Coarse Grained segments by using force field calculations or fragment molecular orbital calculations, and the coordination number. (3) Back calculate the χ parameter based on the experimental phase diagram. Case study: Estimation of χ Parameters from Phase Diagrams

• 

  KRI-NIWA method

  The KRI-NIWA method is a new group contribution method developed by KRI, Inc. that can estimate SP values more precisely than the traditional group contribution method. Estimate χ parameters based on the estimated SP values.

• 

  χ parameter estimation using  Monte Carlo method (FMO)（FCEWS）

  Interaction parameters (χ parameters) are estimated based on results of fragment molecular orbital calculations (FMO) using ABINIT-MP. Since the parameters are determined based on electronic state, general purpose calculations are possible even for molecular species that are difficult to evaluate in classical force field, for example, in the case of charge transfer being essential. FCEWS is a system developed by Mochizuki Lab., in Rikkyo University.
  Detail of FCEWS

• 

  Generate a mesh from the phase separation structure

  Convert the phase separation structure to an FEM mesh. Generate a mesh along the interface and export the volume fraction distribution into a finer mesh. (For the latter function, please refer to the description for MUFFIN as well.)

• 

  Initial concentration distribution

  By combining multiple geometries (box, sphere, and cylinder), a concentration distribution or an obstacle domain can be created.

Finite element simulations are performed using phase-separated structures obtained by mean-field methods and other methods. It is also possible to simulate complex flow fields such as particulate dispersion systems using the particle method.

RVE analysis function in J-OCTA

VSOP-PS（Particle Simulation）

• 

  Import the phase separation structure

  Import the phase separation structure obtained by the mean field method or DPD as an elastic modulus distribution on the FEM mesh. Estimate the stress distribution and the energy distribution on the interface and the mean elastic modulus that are obtained by the elastic simulation using MUFFIN.

• 

  Meshing

  Generate an FEM mesh along the interface by allocating simple spheres inside the rectangular domain. The result can be used to simulate the virtual phase separation structure and the composite materials.

• 

  Multi-layer fluid simulation

  Build models to perform a simulation with a micro-fluid that contains multiple polymers and/or electrolytes. These models are applicable to the calculation of thin film formation with phase separation (e.g., solvent evaporation) and other related phenomena.

• 

  External engine

  Ansys LS-DYNA link : The phase separation structure in voxel mesh which is output from SUSHI (SCF), DPD or CGMD can be used to run Ansys LS-DYNA structural analysis (uniaxial tensile, biaxial tensile, and shear deformation) and thermal analysis. The material can be elastic, elastoplastic, hyperelastic, or viscoelastic. * This function will be released in next version. Refer Ansys LS-DYNARefer the interface to Ansys LS-DYNA

• 

  Representative volume element (RVE) for composite materials

  COGNAC can create a structure (Representative Volume Element(RVE)) of a rectangle region which is highly filled (over 50 vol%) with particles, fibers, and discoid shaped fillers. The resulting structure can be transferred to Digimat-FE and be output as an STL file and a voxel mesh. These files can be used for FEM simulation for composite materials. Refer Digimat-FERefer RVE modeler

• 

  Process Simulation (particulate dispersion systems, droplets, coatings)

  Fluid dynamics analysis can be performed using the MPS (particle method) functionality included in VSOP-PS. It targets complex fluid phenomena such as flow fields in particulate dispersion systems and evaporation phenomena from droplets and coating films (under development).Case StudyVSOP-PS（Particle Simulation）

Machine Learning functions for Materials & Process informatics. Prediction of physical properties from the molecular or crystal structure, and inverse analysis is possible. Other useful functions are implemented.

Review article : Materials & Process Informatics

Property estimation function using machine learning "MI-Suite (Materials Informatics Suite)"

• 

  Quantitative Structure Property Relationship（QSPR）

  This function can estimate ranges of material properties of polymers from input molecular constructions. This function outputs a list of material properties including the density, thermal expansion coefficient, glass transition temperature, Poisson's ratio, and dielectric constant. This is a powerful tool for supporting the molecular design of an organic polymer. Case Study: Estimation of Property Values Using QSPR

• 

  Machine Learning based QSPR（ML-QSPR）

  Machine learning using GCN, CGCN, XGBoost, etc. is used to learn the relationship between various parameters such as molecular structures and processes and their physical properties. The system then predicts the physical property values under unknown molecular structures and conditions. It also includes a transfer learning function. Molecular structures in various formats such as SMILES are supported. We also provide learned libraries and a database of physical properties.MD modeling APICase study

• 

  mol-infer (Inverse QSPR)

  J-OCTA interfaces with mol-infer, which is developed by Nagamochi Laboratory at Kyoto University in Japan. The inverse analysis, i.e., a method to predict molecular structure from physical properties is available. The inverse operation is solved by Mixed Integer Linear Programming (MILP). Detail information >>

• 

  Analysis & public DB connection

  The latest functions related to data science, such as the ability to obtain physical property data by connecting to public DBs such as Materials Project and PubChem, and the ability to extract common parts from multiple molecular structures, are constantly being updated.

• 

  Molecular descriptor calculation function

  This function calculates and outputs molecular descriptors required for machine learning, using SMILES and MOL formats as input.
  · 2D descriptors
  · 3D descriptors
  · Fingerprints　
  · Graph structures
  

• 

  MD-GAN

  This function predicts the motion of molecules over a long period of time by machine learning using the results of short-time molecular dynamics (MD) calculations. This method was developed by Yasuoka Laboratory at Keio University in Japan. For example, it is possible to predict diffusion coefficients quickly, whereas long-time MD calculations are usually required. Detail information >> Case study