| 研究実績の概要 |
This project aims at a computational method to simulate metal complex self-assembly on metal substrates. Such simulations will allow us to study 2D magnetic ordering in molecular layers.
This computational method requires three components: (1) an inter-molecular interaction potential, (2) a molecule-substrate interaction potential, and (3) a method for optimising the monolayer structure using these potentials.
For (1), we constructed a large database of metal complex interactions from first-principles calculations and built an interaction potential using Bayesian machine learning. The accuracy of this method is reasonable, but must be improved for making real predictions. Data for (2) has been collected, but the potential remains in-progress. For (3), a Monte Carlo algorithm has been created
|
| 今後の研究の推進方策 |
First half of FY2020:
(1) We aim to improve the accuracy of the inter-molecular interaction potential by improving how the training data are encoded in the machine learning algorithm. (2) We will also aim for an accurate surface-molecule interaction potential by judicious encoding of the training data.
Second half of FY2020:
We will perform self-assembly simulations using the improved potentials and the Monte Carlo algorithm, first for the case of symmetric metal complexes, and then for the case of asymmetric metal complexes. Whenever possible, literature data will be used to confirm the accuracy. After such predictions are made, we will perform first-principles calculations to determine the magnetic ordering in the monolayer, and predict ways in which magnetic ordering might be controlled.
|
