Innovative AI-Driven Discovery of Advanced Piezoelectric Materials
Interdisciplinary Areas: | Data and Engineering Applications, Innovation and Making, Future Manufacturing |
---|
Project Description
Piezoelectric materials are essential components of actuators and transducers, yet the current selection is limited to a few oxide materials that often suffer from instability, brittleness, and toxicity issues. There is a pressing need to discover new piezoelectric materials to overcome these limitations. Artificial Intelligence (AI), with its high-performance computing capabilities, has proven effective in understanding and analyzing material properties to identify new materials. Despite its widespread application in materials science, AI has not been extensively utilized for piezoelectric material discovery due to two main challenges: the limited size of available piezoelectric databases and the lack of specialized material encoders and physically informed neural networks (PINNs) to handle the complexity of piezoelectric information.
To address these challenges, we propose a multi-step approach. First, we will utilize the existing small databases containing piezoelectric information and apply efficient algorithms to quickly assess the piezoelectric properties of other existing material structures, thereby identifying promising candidates. Next, we will conduct high-throughput theoretical calculations on these potential materials, using ab initio methods to expand the piezoelectric material database and knowledge base. Finally, leveraging the enlarged database, we will employ cutting-edge AI techniques such as transformer, position encoders, and PINNs, integrated with physical information, to design specialized neural network models for new piezoelectric material discovery. This approach aims to efficiently explore the vast chemical and molecular space, leading to the rapid development and design of new piezoelectric materials.
Start Date
08/2025
Postdoc Qualifications
The candidate should have a PhD in civil engineering, material sciences, applied math, computer science, industrial engineering, chemical engineering, electrical engineering, mechanical engineering, or related fields.
Fluent programming in one of the following programming languages: Python/Julia. Familar with PyTorch or TensorFlow. Have experience in artificial intelligence research.
Co-advisors
Name: Yining Feng
email: feng109@purdue.edu
Affiliation: Lyles School of Civil and Construction Engineering
Website: https://engineering.purdue.edu/CCE/People/ptProfile?resource_id=135238
Name: Guang Lin
email:guanglin@purdue.edu
Affiliation: Departments of Mathematics, Statistics & School of Mechanical Engineering
Website: https://www.math.purdue.edu/~lin491/
Bibliography
Na Lu, Guangshuai Han, Yixuan Sun, Yining Feng, Guang Lin, Artificial intelligence guided thermoelectric materials design and discovery, Advanced Electronic Materials, 9(8): 2300042, 2023.
Guangshuai Han, Yixuan Sun, Guang Lin, Na Lu, Machine learning regression guided thermoelectric materials discovery – A review, ES Materials & Manufacturing, 14, 20-35, 2021.
Guangshuai Han, Yen-Fang Su, Siwei Ma, Tommy Nantung, Na Lu, In Situ Rheological Properties Monitoring of Cementitious Materials through the Piezoelectric-based Electromechanical Impedance (EMI) Approach, Engineered Science, 2021, 16, 259-268.
Ziqi Guo, Roy Chowdhury Prabudhya1, Zherui Han, Yixuan Sun, Dudong Feng, Guang Lin*, and Xiulin Ruan, Fast and Accurate Machine Learning of Phonon Scattering Rates and Lattice Thermal Conductivity, Nature npj Computational Material 9, 95, 2023.
Ziqi Guo, Zherui Han, Dudong Feng, Guang Lin, Xiulin Ruan, Sampling-accelerated prediction of phonon scattering rates for converged thermal conductivity and radiative properties, Npj Computational Materials, 10, 31, 2024.