Research

Overview

The research of the Mao group centers on the development and application of theoretical and computational methods to elucidate the fundamental mechanisms involved in chemical problems ranging from molecular and enzyme catalysis to photochemical and electron transfer processes in complex environments. These problems present great challenges to theoretical and computational chemists, since they require a combination of tools that can (i) accurately and efficiently depict ground and excited state potential energy surfaces (PESs), (ii) capture dynamical processes in complex environments for a wide range of timescales, and (iii) bridge the results of computations/simulations with physical interpretations and experimental observations. Our lab is aiming to address these challenges by expanding the computational chemistry toolbox currently available with the development of new electronic structure methods, multi-scale modeling techniques, and data-driven approaches. We will apply these methods to cutting-edge chemical problems, from which we will be able to extract new design principles for various types of catalysts and elucidate intriguing photochemical processes that are pertinent to the design of functional molecules such as those for light-emitting devices and bio-imaging probes.

Harnessing noncovalent interactions in catalyst design

Noncovalent interactions (NCIs), including components such as electrostatics, steric repulsion, and dispersion, play an essential role in molecular, enzyme, and interfacial catalysis. It has thus been a common practice for chemists to make chemical modifications to tune these NCIs to achieve improved chemical reactivity and selectivity. However, these modifications do not always work as one would expect based on one's chemical intuition, and consequently catalyst design still involves a lot of trial and error. This mainly results from the competition between the different components of these interactions as well as their sensitivity to the positioning of interacting moieties and the surrounding environments.

Our lab will develop both physically motivated and data-driven approaches to help understand the role of NCIs in molecular catalysis and to facilitate the modular design of new catalysts leveraging these interactions. We will introduce a new interaction energy partitioning scheme that can precisely identify how each specific part of a catalyst contributes to the stabilization of the transition states. This will be able to tell us which part of the molecular catalyst should be modified and how this modification might look like. We will then employ machine learning (ML) models to screen the candidate catalysts with the prediction of reaction barriers using reactant state properties, including those obtained from the decomposition of substrate-catalyst interactions. In addition, to facilitate the experimental characterization of new catalysts and verification of our catalyst design, we will build quantitative or semi-quantitative maps between spectroscopic signals and physical descriptors of NCIs that can be extracted from electronic structure calcualtions.

Left: Application of ALMO-EDA to the binding of CO2 at molecular catalytic sites; Right: ML-based prediction of reaction barriers using reactant state properties

Elucidating mechanisms of organic photoredox catalysis

Photoredox catalysis is a powerful technique that uses photo-excitation to transform weakly reactive closed-shell species into highly reactive radicals. While traditionally heavy transition metal (TM) complexes, such as those of Ir and Ru, have been employed to catalyze photoredox reactions, several promising catalysts based purely on organic molecules have been proposed recently. These organic photoredox catalysts (PCs) are potentially more suitable for scalable productions because of their reduced cost and environment toxicity. Unlike TM-based catalysts that typically have long-lived triplet states, the organic PCs use their much shorter-lived S1 and/or T1 states as the redox-active species. Therefore, for the design of organic PCs it is important to know how the rate of the electron transfer (ET) or proton-coupled electron transfer (PCET) process compares to those of the competing excited-state decay processes.

Our group will develop a computational framework based on electronic diabatic states to predict the rates of excited-state ET/PCET and those competing processes for the design of purely organic PCs. Those diabatic states localize electronic excitations as well as the transferred electron/proton on specific sites such that they naturally correspond to the initial and final states of an ET/PCET process. Our research will focus on the development of specialized electronic structure methods to construct the diabatic states involved in excited-state ET/PCET processes, in particular those short-lived excited states of varying spin-multiplicities, as well as methods to evaluate the couplings between these diabatic states. These developments will allow us to predict the rates of the redox and the competing excited-state decay processes in silico for organic PCs using rate theories and/or quantum dynamics methods. Doing these calculations for a large pool of PC candidates, we look forward to establishing the relationship between the reactivity of organic PCs and their intrinsic properties as well as the important environment factors.

Left: A typical photoredox catalytic cycle; Right: The S0 → S1 excitation for a "photocatalyst-free" light-induced reaction we are currently investigating (in collaboration with the Gustafson lab)

Decoding light-driven chemical processes in complex environments

Understanding the impact of complex environments on light-driven chemical processes is of great importance to applications such as solar energy conversion, photoluminescent materials, bio-imaging, and human vision. Recent advances in time-resolved and multidimensional spectroscopy have provided new powerful tools to probe excited-state dynamics in various condensed-phase environments, which has motivated the development of theoretical approaches to unravel the chemical mechanisms underlying those measured spectra. For that purpose, on-the-fly dynamics simulations stand for a powerful tool since they can provide atomistic details for the photochemical processes. However, treating the environment at a sufficient level of accuracy entails high computational costs, especially in on-the-fly dynamics simulations where the energies and forces need to be calculated at each time step.

Our goal is to construct accurate PESs that can be utilized for efficient on-the-fly dynamics simulations of excited-state processes in condensed-phase environments. Using these simulations, we are aiming to elucidate the effects of the environment on photochemical processes and provide microscopic details for interpreting time-resolved spectroscopy. To achieve this, we will develop a hierarchy of multi-scale modeling methods ranging from quantum mechanical (QM) embedding theory to ML-assisted PESs. These developments will offer a powerful toolbox for modeling excited-state systems in complex environments at different levels of accuracy and computational cost, facilitating our understanding of how the light-driven chemical processes of interest are modulated and controlled by modifying the condensed-phase environment.

Left: Linear absorption spectra of the GFP chromophore in the gas and aqueous phases predicted using ML models; Right: the proposed hierarchy of methods to model excited-state systems in complex environments