Machine learning aided atomic structure identification of interfacial ionic hydrates from atomic force microscopy images

TL;DR

This paper presents a machine learning approach that accurately identifies atomic structures of interfacial ionic hydrates from AFM images, enabling cost-effective and precise structural predictions of complex systems.

Contribution

It introduces a novel machine learning method, including transfer learning, for atomic structure identification from AFM images, advancing analysis of complex interfacial systems.

Findings

Machine learning achieves precise atomic structure identification from AFM images.

Transfer learning enables cost-effective structure prediction using available data.

Method facilitates analysis of complex interfacial ionic hydrate systems.

Abstract

Relevant to broad applied fields and natural processes, interfacial ionic hydrates has been widely studied by ultrahigh-resolution atomic force microscopy (AFM). However, the complex relationship between AFM signal and the investigated system makes it difficult to determine the atomic structure of such complex system from AFM images alone. Using machine learning, we achieved precise identification of the atomic structures of interfacial water/ionic hydrates based on AFM images, including the position of each atom and the orientation of water molecules. Furthermore, it was found that structure prediction of ionic hydrates can be achieved cost-effectively by transfer learning using neural network (NN) trained with easily available interfacial water data. Thus, this work provides an efficient and economical methodology which not only opens up avenues to determine atomic structures of more complex systems from AFM images, but may also help to interpret other science-wide studies involving sophisticated experimental results.