NCAFM2023 Programme Booklet

MOLECULAR CHARACTERIZATION OF SOLID LIQUID INTERFACES BY MACHINE LEARNING APPROACHES

Farzin Irandoost 1 , Yashasvi Ranawat 1 and Adam S. Foster 1,2

1 Department of Applied Physics, Aalto University; Helsinki, Finland, 2 Nano Life Science Institute (WPI-NanoLSI), Kanazawa University; Kanazawa, Japan

Email: farzin.irandoost@aalto.fi

The interaction between liquids and solid surfaces is studied in a wide variety of natural and technological processes like mineral dissolution, growth, and weathering within geochemistry, biomineralization and biomimetic crystallization, and optimizing electrochemical industrial processes. The structure and dynamics of the liquids at the interface completely differ from the bulk, and there are many factors that govern the behaviour e.g., changes in hydrogen bonding and the influence of other solutes and impurities. Atomic Force Microscopy is already established as an extremely powerful tool for nano-scale imaging of biological systems in solution [1], but higher resolutions are needed for understanding the wide spectrum of processes occurring in solution [2]. The contrast mechanism in solution is dominated by a complex interplay between the tip and surface hydration layers, providing a significant challenge for simulations in solid-liquid environments. Our most recently developed framework [3,4] offers a rapid prediction of hydration layers over a given surface and we build upon this to develop a method that also incorporates the role of the tip in predicted images. The concept first uses unsupervised learning to identify characteristic contrast patterns on benchmark surfaces and then links them to free energy simulations including explicit tips. Ultimately, this will be used to train a deep learning infrastructure to predict tip-surface interactions in solution just from a descriptor based on the local atomic and chemical environment.

References [1] Y. F. Dufrene et al., Nat. Nanotechnol. 12 (2017) 295 [2] T. Fukuma and R. Garcia, ACS Nano 12 (2018) 11785

[3] Y. S. Ranawat, Y. M. Jaques and A. S. Foster, Nanoscale Adv. 3 (2021) 3447 [4] Y. S. Ranawat, Y. M. Jaques and A. S. Foster, J. Mol. Liquids 367 (2022) 120571

128