Research

My research interests mainly focus on developing and exploiting multiscale modeling methods such as density functional theory (DFT), combined with data-driven techniques like machine learning (ML) to help understand, design, and discover next-generation heterogeneous catalysts and functional materials for a broad range of pressing scientific issues related to sustainable chemical production and energy conversion and storage. My past research works include significant contributions to a wide variety of interdisciplinary projects, for example:

1) Exploiting ML neural network to develop electric dipole-related descriptors for catalytic surface-adsorbates interactions (J. Am. Chem. Soc., 2020, 142, 7737-7743). The proposed descriptors were recommended as a “promising new type of catalytic descriptor” on Science, 2020, 368, 727-728 as the “editor’s choice”.

2) For the first time, establishing interpretable machine-learned mathematical formulas to quantitatively determine catalytic-related properties from vibrational spectroscopy (J. Am. Chem. Soc., 2022, 144, 16069–16076). This study dramatically broadens the utility of spectroscopic tools in the context of catalysis.

3) Exploiting high-throughput DFT + ML to accelerate perovskite-type catalysts discovery for a wide range of chemical looping applications, such as air separation, CO2/H2O splitting, and methane partial oxidation (Energy Environ. Sci., 2022, 15, 1512–1528). The effectiveness of this approach has been experimentally validated, with many of the predicted perovskites outperforming the previous benchmark by a factor of >2.

The long-term vision of my future research group is to integrate computational modeling (e.g., DFT) and data-driven techniques (e.g., machine learning) to help understand and design advanced catalysts. Our research will cover a broad range of pressing fundamental and applied scientific issues, providing in-silico solutions to the major challenges pertaining to sustainable chemical production and energy conversion & storage.

Research Skills

  • Computational chemistry methods based on density functional theory (DFT) (VASP, Gaussian, CASTEP)

  • Python, Shell, C programming

  • Python-based Machine Learning packages, including Scikit-learn and TensorFlow.

  • SISSO: a compressed-sensing method for identifying the best low-dimensional descriptor in an immensity of offered candidates.

  • Ab initio real-time electronic and nuclear dynamic simulation package (PYXAID) and ab-initio nonadiabatic molecular dynamics program (Hefei-NAMD).

  • First-principles based stochastic surface walking (SSW) package HOWTOs and Large-scale Atomic Simulation with neural network Potential (LASP) developed by Fudan University.

  • Experiences in high-throughput calculations with big data analysis and machine learning training and predictions.

  • Experiences in studying electronic structures, exciton kinetics and reaction mechanisms in complex surface and interface of photo- and electro-catalytic systems

  • Experiences in tuning bandgaps and metal-insulator transition of solid materials, especially in periodic boundary condition (PBC)

  • Experiences in studying the catalytic behaviors of amorphous molten salt using ab initio molecule dynamics (AIMD)

  • Experiences in the simulation of molecular light absorption and luminescence using time-dependent density functional theory (TDDFT).

  • Familiar with Linux environments, experienced skills with work-related software (e.g. Origin, Material studio, VMD, Microsoft office suites)