Multi-scale modeling of layered double hydroxides for active corrosion protection

This thesis lays the foundation for a computational workflow based on atomistic simulation techniques and machine learning to accelerate the development and evaluation of LDH-based corrosion inhibition systems. The atomistic simulation results shed more light on the anion exchange mechanism and demonstrate how the water content in the interlayer region affects the structure and stability of LDHs. The results show that the initially intercalated anions are released first during the anion exchange process, followed by new anion intercalation, and finally the change in water content stabilizes the LDH system. Furthermore, the mobility of the intercalated anions may be an intrinsic factor influencing their release during anion exchange. In addition to the atomistic simulations, machine learning techniques were used to create a machine-learned force field (ML-FF) for LDH system with almost ab initio accuracy. Despite being trained on small size LDH structures, this ML-FF has demonstrated the ability to scale up to investigate large LDH systems, providing a powerful and accurate force field for further computer-assisted design of LDH-based systems. Another aspect of the presented work is the development of a quantitative structure-property relationship (QSPR) model for the pre-selection of effective corrosion inhibitors that extends the computer-aided LDH design by another element to render it more versatile. The QSPR-based models provided robust estimates for the majority of the blindly tested compounds and the used techniques provide a framework to increase the domain of applicability of the models by increasing the size and versatility of the underlying training datasets following an active learning approach. This thesis provides a proof of concept on how atomic simulations that provide detailed information on the underlying mechanism can be synergistically combined with machine learning techniques to speed up the development of novel LDHs in future research.

Rechte

Nutzung und Vervielfältigung:


CC BY-NC 4.0

Bitte beachten Sie, dass einzelne Bestandteile der Publikation anderweitigen Lizenz- bzw. urheberrechtlichen Bedingungen unterliegen können.

Zitieren

Zitierform:
Zitierform konnte nicht geladen werden.