2016 Fiscal Year Annual Research Report
「ベイズ最適化を活用した」分子自己組織化による ナノ構造制御
Publicly Offered Research
| Project Area | Exploration of nanostructure-property relationships for materials innovation |
|---|
| Project/Area Number |
16H00879
|
|---|
| Research Institution | Kyoto University |
|---|
Principal Investigator |
Packwood Daniel 京都大学, 物質－細胞統合システム拠点, 講師 (40640884)
|
|---|
| Project Period (FY) |
2016-04-01 – 2018-03-31
|
| Keywords | molecular self-assembly / surface / first-principles / Bayesian optimisation / machine leanring |
|---|
| Outline of Annual Research Achievements |
We succeeded to develop a computational method based on Bayesian optimization (BO) to study self-assembly of molecules on metal surfaces. Our method implements BO into the structure optimization steps of a density functional theory (DFT) calculation. For the case of two organic molecules adsorbed to a copper surface, we found that BO could correctly identify the energetically optimum arrangement of molecules on the surface within tens of energy calculations. Our methodology ('BO-DFT') is therefore a first step towards first-principles studies of molecular self-assembly processes.
Submitted manuscript:
Daniel Packwood and Taro Hitosugi. Rapid prediction of molecule arrangements on metal surfaces via Bayesian optimization.
|
| Current Status of Research Progress |
Current Status of Research Progress
2: Research has progressed on the whole more than it was originally planned.
Reason
In order to connect to real molecular self-assembly processes, the important effect of surface temperature must be taken into account. We are currently extending our BO-DFT methodology to predict the arrangement of molecules which minimizes the free energy of the surface. To achieve this, we are incorporating entropy into BO-DFT, using a technique develop by us in another project (Packwood, Han, Hitosugi. Nature Communications 8, 2017, 14463). We call the resulting method BO-FE-DFT (‘FE’ means free energy). We have developed all the necessary theory for the BO-FE-DFT method, and have made a preliminary code for performing these calculations. However, further work is needed to achieve a satisfactory computational performance.
|
| Strategy for Future Research Activity |
For the first half of FY2016, we will work to improve the efficiency of BO-FE-DFT by carefully examining our theory and computer code. We expect to be able to make good predictions of the outcome of the molecular self-assembly process by the end of October 2016.
During the second half of FY2016, we will use BO-FE-DFT to predict a suitable molecule precursor for bottom-up fabrication of a novel type of graphene nanoribbon. We will aim for a graphene nanoribbon with an edge structure different from those fabricated by other groups. By making such a prediction, we will demonstrate how the BO-FE-DFT method can be used to facilitate the bottom-up fabrication of novel nanomaterials.
|
