Research

1. Machine Learning-Accelerated Materials Intelligence

Motivation: Traditional trial-and-error materials discovery is too slow and resource-intensive. We need a paradigm shift toward data-driven intelligence to rapidly navigate vast chemical spaces. 

Methodology

• Develop robust Machine Learning Interatomic Potentials (MLIPs) for energy and environmental materials.

• Build large language model (LLM)-based AI agents such as ChatGPT, Claude, Gemini or Cursor to automatically extract literature data and recommend optimal synthesis routes.

• Apply generative AI (i.e. diffusion models) for the inverse design of novel materials.

Expected Outcomes: This autonomous framework will drastically reduce discovery time and cost, yielding theoretically validated materials ready for experimental realization. 

2. Computational Discovery of Next-Generation Sensor, Battery, Electrocatalyst, and Semiconductor Materials 

Motivation: Next-generation electronics and sustainable energy demand breakthrough materials. We aim to overcome current performance limits through targeted, atomistic-level improvements for the energy and environmental materials.

Methodology

• Sensors: Optimize the "4S" criteria (Sensitivity, Selectivity, Speed, Stability) for advanced materials.

• Batteries: Develop high-conductivity solid-state electrolytes and construct "all-in-one" computational models (cathode-electrolyte-anode) for all-solid-state batteries.

• Electrocatalysts: Design highly efficient, regenerable catalysts for Oxygen and Chlorine Evolution Reactions (OER/CER).

Expected Outcomes: Realize ultra-precise sensors, reliable all-solid-state batteries, and highly active catalysts for sustainable energy applications.

3. Designing Multi-Dimensional and Multi-Functional Composite Materials 

Motivation: Single-component systems often hit a performance ceiling. Synergistic composites made of two or more material types are essential for next-generation applications. 

Methodology: We leverage our computational techniques to construct various composites (e.g., MOF@MOF, MOF@TMD, MOF@CNT), optimizing selections from vast material libraries. We conduct in-depth atomistic studies, such as density functional theory (DFT), molecular dynamics (MD) and monte carlo (MC), to uncover the exact mechanisms behind their interfacial interactions. 

Expected Outcomes: By understanding the fundamental sources of these synergistic properties, we establish a rational design strategy to create multifunctional architectures that outperform traditional materials.