Predicting Stability of Modified Oxide Surfaces with Functional Atomic-Layers for Nano-engineered Catalysts through First Principles Calculations and Statistical Learning

Restricted (Penn State Only)
Jonayat, A S M
Graduate Program:
Mechanical Engineering
Doctor of Philosophy
Document Type:
Date of Defense:
June 25, 2018
Committee Members:
  • Adri C.T. van Duin, Dissertation Advisor/Co-Advisor
  • Adri C.T. van Duin, Committee Chair/Co-Chair
  • Michael J. Janik, Committee Member
  • Richard Yetter, Committee Member
  • Lasse Jensen, Outside Member
  • catalysis
  • monolayer metal oxide
  • mixed metal oxides
  • DFT
  • machine learning
  • compact sensing
  • feature selection
Multicomponent metal oxides (MMOs) are of significant interest because of their tunable catalytic properties. They can form different structures – core shell particles, coatings on substrate, or bulk mixtures. Two specific types of MMOs – monolayer metal oxides and surface confined mixed metal oxides - are the focus of this work. Despite the growing interest in MMOs, our understanding of their stability to date has been limited; only a few experiments have been undertaken of such systems. To date, discovery of these systems has mainly been through empirical procedures. The large number of possible combination makes it very difficult to discover stable MMOs and systems of interest may be metastable, making experimental discovery more difficult. In this dissertation, Density Functional Theory was used along with ab initio thermodynamics to find possible descriptors of (meta)stable monolayer metal oxides and surface confined mixed oxides. A thermodynamic framework is developed to predict phase diagrams of monolayer metal oxide stability with respect to oxide particles of different sizes and pressures-temperatures. Finally, we show that statistical and Machine Learning algorithms can be useful to not only predict, but also discover underlying physical rules that dictate the stability of the monolayer coating/oxide support combinations.