Date of Award

August 2021

Document Type


Degree Name

Doctor of Philosophy (PhD)


Chemical Engineering

Committee Member

Rachel Getman

Committee Member

Sapna Sarupria

Committee Member

Christopher Kitchens

Committee Member

Lindsay Shuller-Nickles


Aqueous-phase heterogeneous catalysis has many applications, including biomass reforming, Fischer-Tropsch synthesis, and electrocatalysis. Formulation of accurate kinetic models for these systems is necessary not only to gain mechanistic understanding but also to have quantitative prediction of the activity and selectivity under reaction conditions. However, the molecular modeling of the solid-liquid interfaces, especially the adsorption and the surface intermediate reaction, which are important steps in aqueous phase heterogeneous catalysis, are still under development. In order to better understand these systems, in this work, we developed novel mathematical and molecular methods to study the catalytic adsorption and the surface adsorbate solvation free energies at the liquid-Pt(111) interface.

The molecular modeling of the adsorption at the liquid-solid interface is challenging because the pathway of how molecules approach and stick to the catalytic surface with the presence of the solvent is less studied. Most previous work applied gas phase adsorption theory to the liquid phase, which did not capture the solvent influence. The method we developed in this work combined both molecular dynamics (MD) simulation and the random walk model, to simulate a methanol solute diffusing in the water solvent and adsorbing to the Pt(111) catalytic surface. The presence of water solvent affected the diffusion of the methanol that led to a different adsorption phenomenon than in the gas phase due to ever-changing water configurations. The molecular dynamics simulation was first carried out to track the motions of the methanol molecule, and then a Continuous-Time Random-Walk mathematical model was used to model its trajectory, where different diffusion and adsorption states were extracted from the model. The sticking coefficient which represents the fraction of impinging molecules that stick to the surface was calculated from the model and can serve as an important property in future kinetics models.

Calculating the solvation free energies of catalytic adsorbates at the liquid-solid interface is also challenging because of both the many configurations of the liquid solvent and the energy accuracy due to bond breaking and forming need to be considered. Most previous studies used implicit solvation or ice to represent the solvent molecules which cannot capture the dynamics of the solvent-adsorbates interaction. In this work, we developed a multiscale sampling (MSS) method to calculate the surface adsorbate solvation free energy which combined the density functional theory (DFT) and the molecular dynamics simulation. This method not only captured the more accurate energy from quantum simulation but also the configurational dynamics of the solvent to obtain the entropy. We compared our methods with the implicit solvation and the results indicated that explicit quantum-based methods are needed when adsorbates form chemical bonds and/or strong hydrogen bonds with the solvent. We also investigated the energetic and entropic contributions to the solvation free energy. We found that adsorbates that exhibit strong energies also exhibit strong and negative entropies, and we attributed this relationship to hydrogen bonding between the adsorbates and the solvent molecules, which provides a large energetic contribution but reduces the overall mobility of the solvent.

The calculation of adsorbate solvation free energies using the MSS method requires multiple steps and also considerable amount of time due to the presence of explicit solvent molecules when calculating the DFT energies. This limits our ability to study more adsorbates and learn the essential factors that influence the solvation free energy. To solve this problem, we developed a pipeline which automates the simulation procedure and allows us to collect more data including 90 reactive intermediates of methane, methanol, C2 and glycerol decomposition. With this data set, we find a linear relationship between the DFT calculated energy and the MD calculated entropy, and the time-consuming energy calculation can be estimated from the entropy so that the solvation free energy can be evaluated just from entropy. Another way to bypass the tedious calculation of the DFT energy is to study the structure-property relationship, where the structure of the simulated model is used to predict the energy property. For this work, we implemented the machine learning method and applied Coulomb Matrix and Bag of Bond descriptors and obtained initial results.



To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.