(4jw) Multi-Scale Computational Modeling: Toward Fundamental Design in Electrocatalysis for Sustainable Chemical Production

My research interest is to develop and implement multi-scale computational models for electrocatalysis to advancing sustainable chemical manufacturing using renewable electricity—including but not limited to CO₂ reduction, NO₃^- reduction, and biomass oxidation. My lab’s research theme will revolve around the rational design of the electrochemical interface, comprising three sub-themes:

1) Modulating the aqueous electrochemical double-layer using organic additives.

2) Investigating the dynamics of catalyst surfaces and adsorbates under applied voltages.

3) Exploring programmable external biases to potentially impact electrocatalytic kinetics.

Concurrently, significant efforts will be dedicated to developing and validating state-of-the-art computational methods capable of accurately describing the electrochemical interface across multiple time and length scales. These methods will be the workhorse for my research. In this abstract for my faculty-candidate poster at the AIChE 2024 Annual Meeting, I will briefly outline these three research sub-themes/topics.

The first research sub-theme will entail incorporating organic additives into conventional aqueous electrochemical interfaces—water and monovalent ions as the electrolyte, and d-block metals as the electrode—such as surfactants, self-assembled monolayers, and co-solvents to modulate the physical characteristics of the electrochemical double-layer (EDL), and in turn alter electrocatalysis kinetics. While experimental investigations have explored these additives across various electrocatalytic systems, a molecular-level understanding is lacking due to limited computational studies. Current mean-field descriptions of the EDL provide insights into average-thermalized ions distribution, solvent dielectric constant, and surface charge density. However, my work in this area shows that the molecular details of the EDL are not as well-defined as the mean-field counterpart: interfacial water exhibits highly unique structure, dynamics, and dielectric polarizability; ions spatially rearrange at the interface due to water (de)solvation and metal surface-induced electrostatics. My work furthermore made clear that quantum mechanical-based approaches such as density functional theory (DFT) fall short in resolving the EDL structure due to limitations in time and length scales. Classical force fields (cFF) are better suited for simulating EDL dynamics, especially when incorporating organics that are spatially bulky with slow relaxation time. I expect the hydrophobic organics to locally disrupt the water hydrogen-bonding network and alter the solvation of reaction species near or absorbed on the catalyst surface. Furthermore, organics can lower the overall dielectric constant of the EDL, altering interfacial capacitance and surface charge density. Short-range effects such as Van der Waals and steric interactions may also play a role in (de)stabilizing surface adsorbates. Collectively, I hypothesize that these organic additives can modulate the interfacial EDL, consequently affecting electrocatalytic reaction rates.

To investigate the dynamics of the catalyst surface and adsorbates under electrocatalytic conditions, I will leverage machine learning force fields (MLFF), capitalizing on the rapidly growing field of artificial intelligence. Modeling the restructuring of catalyst active sites at a solid-liquid interface under an applied voltage remains a significant challenge, given that these sites are conventionally treated as static and well-ordered. Additionally, current kinetic models of (electro)catalysis often overlook the diffusion of reactants on the surface. Therefore, having a clearer picture of the dynamic and fluxional nature of the electrocatalyst and adsorbates holds great potential for advancing the field. In my postdoctoral training, I am implementing an recently developed MLFF to simulate copper oxide electrocatalyst restructuring. This entails training a neural network on DFT calculations to parameterize a mathematical force field capable of dynamically exploring new non-ordered structures not present in the training data set. With my experience in analyzing complex molecular dynamics trajectories—such as those in mixed solvents and solid-liquid interfaces—I aim not only to produce a molecular “movie” of catalyst restructuring, but also formalize a quantification of restructuring based on statistical thermodynamics. For example, analogous to hydrogen bonding networks in liquid water, one can define an localized average percolating size of the catalyst nanoparticles, along with a relaxation timescale for hopping between low-energy configurations. Regrading adsorbate diffusion, an average diffusivity can be directly measured using MLFF to encode the translational mean-squared distance of the adsorbate, which can have a non-negligible contribution to the overall surface reaction kinetics. Furthermore, the translation entropy of an adsorbate might be modified from a simple 2D ideal gas to include an extra configurational term from the non-ordered surface. This research will compliment experiment studies where catalyst restructuring is prominent, while also striving to establish a theoretical framework for encoding (electro)catalysis dynamics.

Tethering to the emerging area of programmable dynamic catalysis—that is, the perturbation of a catalytic system by temporally/spatially varying (thus, programmable) external bias forces—I will explore three programmable biases for electrocatalysis through computational approaches: alternating voltages, surface plasmon resonance, and sound waves. Compared to a constantly applied voltage, alternating voltages can dynamically modulate surface adsorption/desorption, the EDL structure, and ionic species diffusion, all of which can potentially impact electrocatalytic kinetics. Modeling these transient effects of voltage switching necessitates near-equilibrium molecular dynamics and grand-canonical DFT (GC-DFT) to probe for electrolyte dynamics and voltage-dependent adsorption/desorption, respectively. Next, surface plasmon resonance, involving electron collective fluctuations on a catalyst surface (such as on Ag, Au, and Cu) as optical response to electromagnetic light waves can, in principle, impact the electrochemical interface. Electron clouds spilling toward or away from the catalyst surface (during each phase of the resonance) can alter electronic interaction with adsorbates as well as ionic response in the electrolyte dynamically. Modeling this phenomenon requires coarse-grained molecular dynamics with a continuum-based polarization scheme for nanoparticle optical response, alongside GC-DFT and molecular force fields for electronic and ionic responses, respectively. Exploring the unconventional combination of sound waves and electrocatalysis, I will employ molecular dynamics to simulate mechanical waves in liquid. The mechanical sound waves may perturb the hydrogen bonding network and water translational/rotational dynamics, subsequently affecting ion hydration and dielectric screening at solid-liquid interfaces. For all of the above external biases exploration, I will begin by building parts of the multi-scale model on toy systems, then incorporate them either sequentially (output of different models are combined post-processing) or concomitantly (the models exchange input/output during simulation to yield a final output).

I do not believe in a “panacea” computational model—an all-encompassing solution capable of simulating all fundamental physics of a given system. Instead, I will advocate for creatively combining models of different time and length scales in my research, ranging from electronic DFT to molecular force fields to coarse-grained and continuum models. This multi-scale modeling approach can alleviate the drawbacks and leverage the strengths of each constituent model. With experience in systematically developing and implementing multi-scale models for increasingly complex (electro)catalytic systems, I am well-equipped to pursue the outlined research topics and contribute to the computation-guided rational design of electrocatalysis for sustainable chemical productions.

Teaching Interests

Growing up and going to school across the pre- and post-internet time, I am excited about integrating both old school and new school teaching styles—i.e., a multi-style teaching approach—into my classroom. The attention span of the younger generations is growing shorter, disfavoring the slow and methodical chalkboard style of teaching. Yet, the important fundamentals in chemical engineering (e.g., deriving the Navier-Stokes equation) often require an industrious teaching/learning process at a slower pace. On the other hand, certain supplementary tidbits and knowledge (e.g., Reynolds numbers of different fluids with laminar and turbulent flow) can perhaps benefit from the “TikTok style” of quick-firing information in videos or pictures that are edited creatively. Similar to my research philosophy, I will not adopt a one-size-fits-all teaching method, but continuously improvise and adapt different teaching approaches—from chalkboard derivation to quick-fact videos—to suit the classroom environment.

As a faculty member, I intend to develop and teach two graduate-level courses. The first course will focus on practical implementation of atomistic and molecular modeling, aligning closely with my research expertise. The second course will survey international (i.e., non-U.S) problems and technologies in renewable energy. This topic is inspired by my Ph.D. advisor, Prof. Mike Janik at Penn State University, who taught a similar course surveying the renewable energy landscape in the U.S. Growing up outside of the U.S., I’m passionate about bringing a global perspective on renewable energy, integrating socioeconomic and cultural factors to inform research and innovation in the U.S.