Session W10: Modeling the Electrochemical Interface II

Combining First Principles Theory and Experimental Characterization to Investigate Model Catalyst Surfaces

The electrochemical interface is relevant in applications such as water purification, corrosion, catalysis, and energy storage. The exact structure and composition of the solid surface crucially impacts ion adsorption, dissolution and intercalation and electron energy alignment between surface and reacting molecules. However, the structure of these surfaces under operating conditions is challenging to probe experimentally and the relevant chemical reaction mechanisms are often unknown. We will review the latest developments and ongoing challenges in building first-principles models for solid-liquid interfaces under electrochemical potential control. Such theoretical models of interfacial structure still rely upon approximations which can be inaccurate for surfaces in complex environments and with defects, highlighting the importance of combining them with experimental characterization methods such as X-ray reflectivity (XRR) and temperature programmed desorption (TPD).  

XRR determines the electron density of an interface with high resolution but typically relies on model-dependent fitting to invert the data and obtain the corresponding atomic structure. We integrate first principles theory with XRR measurements to gain insights into bonding and structure at the Al₂O₃(001)/water interface. TPD determines reactant desorption energy by increasing the temperature of the substrate in a high vacuum environment, but identification of the active sites contributing to peaks in the spectra remains a challenge. We use a combination of TPD and density-functional theory (DFT) to investigate the atomic-scale origin of increased reactivity of small alcohols at under-coordinated sites on Ag(111) and TiO₂/Au(111) model catalysts.   

Improving the Computational Efficiency of the Explicit-Implicit Hybrid Solvent Model for Simulations of the Electrochemical Environment

The development of ab initio methods for atomistic simulations of the electrochemical environment is essential for obtaining a mechanistic understanding of fundamental reactions. We have recently developed a hybrid solvent model, SOLHYBRID [1], that enables simulations of the electrochemical environment including both explicit and implicit solvents with the popular Vienna Ab initio Simulation Package (VASP). However, the high computational cost associated with the model prevents its application in large length and time scale ab initio molecular dynamics simulations. In this presentation, we will present our latest effort to accelerate SOLHYBRID simulations without significantly compromising its accuracy. This is achieved by reducing the frequency of solving the linearized Poisson-Boltzmann equation. We will showcase SOLHYBRID's new capabilities enabled by this development for the understanding of liquid/solid interface in electrocatalysis.

1.         D. Le, An Explicit-Implicit Hybrid Solvent Model for Grand Canonical Simulations of the Electrochemical Environment. ChemRvix 2023. http://dx.doi.org/10.26434/chemrxiv-2023-z2n4n   

A molecular-scale picture of the electrical double layer at TiO2-electrolyte interfaces

The electrical double layer (EDL) is a structure that appears at solid-liquid interfaces, which governs the chemical reactivity and physical properties of the interface and plays a critical role in numerous electrochemical, electrocatalytic, geological, and biological processes. A molecular-scale understanding of the EDL is a significant step towards better controlling and optimizing these impactful processes. However, due to the inherent complexity of the interface, the molecular-scale simulation of the interface has been a long-standing challenge. In this work, we use the advanced deep potential long-range neural network method to simulate the interface between TiO₂ and electrolyte of varying pH values with ab initio accuracy. This gives us a comprehensive molecular-scale picture of the EDL, including the surface charging mechanism, ion distribution, and water orientation in the EDL, which confirms the limitations of the widely adopted classical mean-field description of the EDL provided by the Gouy-Chapman-Stern model. Moreover, our DeepWannier neural network describes the separation of electronic and ionic centers of charge, which enables us to calculate the electrostatic potential drop at the interface. We further calculate the capacitance, a property of experimental relevance. The computed capacitance agrees semi-quantitatively with experimental results, suggesting the reliability of our molecular-scale picture of the EDL.   

First-Principles Molecular Dynamics Simulations of Indium Oxide/Water Interfaces

Indium oxide (In2O3) is commonly used as a transparent conducting electrode in photovoltaic and electrochemical cells, due to its high photocatalytic activity, chemical stability, and commercial availability. In particular, indium oxide surfaces in contact with water can produce hydroxyl radicals (•OH) from the splitting of interfacial water molecules, which are required for advanced oxidation processes. In order to optimize desired oxidation processes, it is important to understand the influence of surface hydroxylation, doping (e.g. with tin) and disorder in determining the properties of the interface with water and the ability of the solid oxide to induce water splitting. Here, we carry out first principles molecular dynamics simulations with the SCAN functional and the Qbox code to study indium oxide/water interfaces under different hydroxyl coverages (100%, 98%, 83% and 66% hydroxylation), with the goal of characterizing the structural, electronic, and vibrational properties of the aqueous interfaces. Work is in progress to carry out a detailed comparison between simulations and experiments   

Understanding the low dielectric constant of nanoconfined water capacitors

The relative out of plane dielectric constant of nano confined water in graphene capacitors 

was shown to be as low as 2.1 for capacitor thickness below 1 nm by Fumagalli et al. (Science 360, 2018). This

study motivated a plethora of theoretical studies aiming to explain this surprising behavior. It is surprising because such low

measured value is even lower than the high frequency dielectric constant of ice.

In this work we show that both experimental and theoretical studies suffer from intrinsic arbitrariness

associated to the ill defined concept of a macroscopic  dielectric constant in a microscopic 2D limit. 

We show that the two-dimensional transverse polarizability is the natural choice to characterize the

response of a nanoconfined dielectric in a capacitor. This polarizability can also be connected to

the actual experimental measurements of differential capacitance, providing a clean way of comparing

both experimental and theoretical results, eliminating the need for intermediate arbitrary choices needed

for the computation of the dielectric constant.   

Reaction and Ionic Migration at the Electrode-electrolyte Interface in Solid State Batteries from Machine Learning Molecular Dynamics

Atomistic-level understanding of the chemical reactions forming the solid-electrolyte interphase (SEI) in solid-state lithium batteries has remained challenging, primarily due to the limited resolution in experimental techniques and the insufficient accuracy in large-scale simulations. In this work, we combine on-the-fly active learning based on Gaussian Process regression (FLARE) with local equivariant neural network interatomic potentials (Allegro) to construct a machine-learning force field (MLFF) to perform large-scale long-time explicit reactive simulation of a complete symmetric battery cell with ab initio accuracy. The MLFF is validated with experimental values of mechanical properties of bulk lithium and diffusion coefficient of solid electrolyte. For the symmetric battery, we observe prominent fast reactions at the interface and characterize the dominant reaction products along with their evolution time scales, using unsupervised learning techniques based on atomic geometry descriptors. Our simulation reveals the kinetics and the passivation involved in the chemical reaction responsible for the SEI formation. The methods in this study are promising for acceleration analysis of atomistic mechanisms in complicated heterogeneous systems and provide design insights for the development of solid-state batteries.   

First principles derived potential barriers to Li-ion transport through Solid Electrolyte Interphases in batteries

Charging a Li-ion battery is dependent on the ability of Li ions to transport between the cathode and the anode. Li-ion transport is dependent upon (among other factors) the electrostatic environment the ion encounters on the Solid-Electrolyte-Interphase (SEI) which separates an anode from the surrounding electrolyte. Previous first principles work has had difficulty accurately reflecting the electrostatic potential barrier for ions moving through the SEI due to the large length scale necessary to simulate. In this work, we develop and apply the Quantum Continuum Approximation (QCA), a methodology for coupling explicit Density Functional Theory (DFT) calculations of interfaces with Poisson-Boltzmann distributions of charge in bulk insulating systems to provide an equilibrium electronic potentiostat for first-principles interface calculations. Using QCA, we calculate the electrostatic potential barrier for Li-ion transport through various SEIs on Li metal anodes. We demonstrate that the SEI potential barriers are dependent on the electronic voltage in each system, although the degree of dependency varies based on the specific SEI interface. This work suggests modifying the voltage and anode-SEI interface structure as a strategy for improved charging rates in Li-ion battery systems.   

A state-of-the-art approach to addressing the missing ingredients of molecular modeling

More environmentally friendly organic molecules are gradually replacing the toxic inorganic ones used in corrosion inhibition (CI). Molecular modeling approaches can help to unveil their acting mechanisms, paving the way for a rational design of novel inhibitors. Nonetheless, CI modeling frequently neglects crucial aspects such as the solvent and voltage effects. We present a state-of-the-art approach to tackle that problem by combining the non-equilibrium Green's functions (NEGF) formalism with the quantum mechanics/molecular mechanics (QM/MM) method to include the voltage effects while fully accounting for the solvent with atomic resolution, but at an affordable computational cost, as implemented in the SIESTA code.

We apply this approach to study the interaction between mercaptobenzimidazole (MBI) and cooper under different concentrations. Our results show that the stability of the adsorbed layer increases as we increase the coverage, forming a self-assembled monolayer. We also demonstrate that the charge redistribution due to the voltage applied to the metallic substrate leads to significant changes, also affecting the structural properties of the solvent (water). In summary, the proposed NEGF-QM/MM approach presented here might represent a significant step forward in modeling electrochemistry, which we demonstrate can be successfully applied in CI.