Simulating molecular adsorption on dielectric surfaces with classical MD, DFT and machine learning

Thumbnail Image
Files
isbn9789526411996.pdf (11.62 MB)
Journal Title
Journal ISSN
Volume Title
School of Science | Doctoral thesis (article-based) | Defence date: 2023-04-14
Unless otherwise stated, all rights belong to the author. You may download, display and print this publication for Your own personal use. Commercial use is prohibited.
Author
Date
2023
Department
Teknillisen fysiikan laitos
Department of Applied Physics
Major/Subject
Mcode
Degree programme
Language
en
Pages
86 + app. 60
Series
Aalto University publication series DOCTORAL THESES, 41/2023
Abstract
Surfaces and the interface physics at the nano-scale play a vital role in several technological and natural processes. These interface interactions have applications in geochemistry and environmental science, biomineralisation, corrosion, etc. The interaction physics can be understood by studying the molecular adsorption interactions, at various surface coverages -- from singular molecule to bulk. Atomic force microscopy (AFM) has emerged as a potent tool to characterise such molecular interactions with the surface. These AFM images are often complemented with simulation tools to further characterise the surface phenomena. This dissertation applies simulation and machine learning tools to characterise molecular adsorption on surfaces. It targets two objectives: (a) characterisation of the 2×1 surface reconstruction of the (104) calcite surface and determination of the influence of the reconstruction on the surface chemistry through molecular adsorption, and (b) development of a machine-learning (ML) workflow to predict bulk molecular -- water in particular -- interactions over surfaces, that form the hydration layers. The design of the ML workflow is split into intermediate targets in the dissertation: (a) generation of a solid-liquid interface database (b) designing of a general descriptor of the surfaces, and (c) development and training of ML techniques to rapidly predict the hydration layers over the surface. Additionally, an out-of-distribution detection ML technique is used to gauge the accuracy of the prediction of the hydration layers.
Description
Supervising professor
Foster, Adam, Prof., Aalto University, Department of Applied Physics, Finland
Keywords
simulation, machine learning, molecular adsorption, dielectric surfaces
Other note
Parts
  • [Publication 1]: Himanen, Lauri, Marc OJ Jäger, Eiaki V. Morooka, Filippo Federici Canova, Yashasvi S. Ranawat, David Z. Gao, Patrick Rinke, and Adam S. Foster. DScribe: Library of descriptors for machine learning in materials science. Computer Physics Communications, Volume 247, Article number 106949, Feb 2020. Full text in Acris/Aaaltodoc: http://urn.fi/URN:NBN:fi:aalto-201911076187.
    DOI: 10.1016/j.cpc.2019.106949 View at publisher
  • [Publication 2]: Jäger, Marc OJ, Yashasvi S. Ranawat, Filippo Federici Canova, Eiaki V. Morooka, and Adam S. Foster. Efficient machine-learning-aided screening of hydrogen adsorption on bimetallic nanoclusters. ACS combinatorial science, Volume 22, Article number 12, Pages 768–781, Nov 2020. Full text in Acris/Aaaltodoc: http://urn.fi/URN:NBN:fi:aalto-202103222460.
    DOI: 10.1021/acscombsci.0c00102 View at publisher
  • [Publication 3]: Ranawat, Yashasvi S., Ygor M. Jaques, and Adam S. Foster. Predicting hydration layers on surfaces using deep learning. Nanoscale Advances, Volume 3, Article number 12, Pages 3447–3453, May 2021. Full text in Acris/Aaaltodoc: http://urn.fi/URN:NBN:fi:aalto-202108048238.
    DOI: 10.1039/d1na00253h View at publisher
  • [Publication 4]: Heggemann, Jonas, Yashasvi S. Ranawat, Ondrej Krejcí, Adam S. Foster, Philipp Rahe. Differences in molecular adsorption emanating from the (2×1) reconstruction of calcite(104). The Journal of Physical Chemistry Letters, Volume 14, Pages 1983–1989, February 2023.
    DOI: 10.1021/acs.jpclett.2c03243 View at publisher
  • [Publication 5]: Ranawat, Yashasvi S., Ygor M. Jaques, and Adam S. Foster. Generalised deep-learning workflow for the prediction of hydration layers over surfaces. Journal of Molecular Liquids, Volume 367, Page 120571, December 2022. Full text in Acris/Aaaltodoc: http://urn.fi/URN:NBN:fi:aalto-202211096378.
    DOI: 10.1016/j.molliq.2022.120571 View at publisher
Citation
Permanent link to this item
https://urn.fi/URN:ISBN:978-952-64-1199-6
Collections
[diss] Perustieteiden korkeakoulu / SCI