Embedded Files
Peng Chen
Peng Chen
Research Associate
Department of Physics and Institute for Nanoscience and Engineering,
University of Arkansas,
731 W. Dickson St. Fayetteville, AR 72701
Research and Teaching Interests:
Materials for Next-Generation Computing, Energy Efficiency of Scaled Semiconductors, Light-Matter Interaction, Topological Magnetic/Electric Defects, Spintronics with Ferroelectrics, Quantum Ferroics, Physics-Based Artificial Intelligence Materials Modeling
Theory of Condensed Matter Physics, Computational Quantum Physics, Numerical Methods in Quantum Mechanics, Density Functional Theory for Beginners, Introduction to Scientific Computing (Linux Shell/Python/FORTRAN/MPI), Group Theory in Condensed Matter and Materials Science, Quantum Transport: Atom to Transistor, Green’s Functions Method in Quantum Physics
Previous Experience:
Postdoc, Italian Institute of Technology, Genova, Italy
Ph.D., Institute of Physics, Chinese Academy of Sciences, University of Chinese Academy of Sciences, Beijing, China
Visiting Scholar, Luxembourg Institute of Science and Technology, Luxembourg
Intern, Longxun Quantum Co., Ltd., Beijing, China
B.S., Xidian University, Xi’an, China
Research: As electronic devices continue to shrink in size, the materials powering them are approaching the near-atomic-scale limit, leading to increased power consumption requirements. Moreover, the rapid growth of artificial intelligence (AI) and big data has resulted in a massive demand for computational resources and thus a large amount of energy. For instance, the training of GPT-3, the underlying language model for ChatGPT, consumed an estimated 1,287 MWh of electricity (equivalent to approximately $167,310) which generated 552 tons of CO2 emissions [arXiv:2204.05149]. It is worth noting that the energy consumption of running ChatGPT on a monthly basis is 18 times higher than the energy used for its training. This presents a significant challenge to the future sustainable computing and energy efficiency of scaled semiconductors.
I develop and employ theoretical and computational methods including density functional theory, many-body perturbation theory, effective Hamiltonian method, and machine learning algorithms, for the discovery and design of materials that realize energy-efficient memory and computing logic. My goal is to address the energy efficiency challenges of scaled semiconductors for future sustainable computing.
My recent research interests are focused on advanced functional materials called ferroelectrics, which hold promising applications in reducing energy consumption by ultrasmall beyond-CMOS logic devices, as well as potential uses in neuromorphic computing and harvesting energy from renewable sources. Another big research passion of mine is to develop and apply AI-assisted multiscale modelings for material and nanodevice simulations, in an attempt to (i) accelerate the discovery and design of functional materials; (ii) reproduce the experimental measurements; (iii) predict experimentally verifiable material properties; (iv) understand the physics behind the phenomena.
Fig. 1 Ab-initio-based effective Hamiltonian method implemented in LINVARIANT.
Software: "LINVARIANT: Symmetry-Adapted Effective Hamiltonian for Materials Design,"
Peng Chen, Hongjian Zhao, Sergey Artyukhin, and Laurent Bellaiche, LINVARIANT: v1.0. https://doi.org/10.5281/zenodo.5951858 (2022).
AI-Copilot
AI-Copilot
The application of physics-based artificial intelligence (AI) and multiscale methods has the potential to greatly enhance the simulation and discovery of functional materials, such as post-silicon materials, optoelectronics, energy storage, and energy conversion materials, among others.
However, developing physics-based models that accurately represent real materials poses a significant challenge.
LINVARIANT involves the integration of an AI Copilot (see figure above) to assist in material modeling and the characterization of the underlying interactions responsible for their properties. It is important to note that our approach maintains a physics-based framework rather than relying solely on machine learning (ML) models, which are often considered "black boxes." This means that AI will serve as a co-pilot in our exploration of materials, complementing our investigations rather than replacing the model itself. Nevertheless, ML-type models will be accessible and can be transformed into transparent physical models.
Selected Publications
Lights can do optical cooling, trapping, tweezer, etc. We discovered and explained how light is like a squeezer when interacting with ferroelectric materials:
[1] Peng Chen†, Charles Paillard, Hongjian Zhao, Jorge Iniguez, and L. Bellaiche†, Deterministic control of ferroelectric polarization by ultrafast laser pulses, Nat. Commun. 13, 2566 (2022).
[2] Peng Chen†, Changsong Xu, Sergei Prokhorenko, Yousra Nahas, and Laurent Bellaiche†, Electrical topological defects induced by terahertz laser pulses, under review (2022).
Intrinsic electrical Dzyaloshinskii-Moriya interaction was assumed NOT to EXIST until we unveiled and demonstrated its microscopic origin. Our works provide the fertile background for the dawning of the polar topological states era:
[3] Hongjian Zhao, Peng Chen†, Sergey Prosandeev, Sergey Artyukhin, and Laurent Bellaiche†, Dzyaloshinskii-Moriya-like interaction in ferroelectrics and anti-ferroelectrics, Nat. Mater. 20, 341 (2021).
[4] Peng Chen†, Hong Jian Zhao†, Sergey Prosandeev, Sergey Artyukhin, and Laurent Bellaiche, Microscopic origin of the electric Dzyaloshinskii-Moriya interaction, Phys. Rev. B 106, 224101 (2022).
We found that topological defects, such as domain walls, can have peculiar excitation spectrums hidden in phonons. Such hidden vibrations explained the anomalous microwave conductivity of BiFeO3 under microwave electric fields:
[5] Peng Chen†, Louis Ponet, Keji Lai, Roberto Cingolani, and Sergey Artyukhin†, Domain wall-localized phonons in BiFeO3: spectrum and selection rules, npj Comput. Mater. 6, 48 (2020).
[6] Yen-Lin Huang, Lu Zheng, Peng Chen (joint-first author), Xiaoxing Cheng, Tiannan Yang, Xiaoyu Wu, Louis Ponet, Ramamoorthy Ramesh, Long-Qing Chen, Sergey Artyukhin†, Ying-Hao Chu, Keji Lai†, Unexpected Giant Microwave Conductivity in a Nominally Silent BiFeO3 Domain Wall, Advanced Materials 32 (9), 1905132 (2020).
It was puzzling that more than half of the perovskite materials adopt Pnma structure as their lowest energy state; We solved the puzzle by modeling the interactions between oxygen octahedral tilings and antipolar distortions:
[7] Peng Chen†, Mathieu N. Grisolia, Hong Jian Zhao, Otto E. Gonzalez-Vazquez, Manuel Bibes, Bang-Gui Liu, L. Bellaiche and Jorge Iniguez*, Energetics of oxygen-octahedra rotations in perovskite oxides from first principles, Editor's Suggestion Phys. Rev. B 97, 024113 (2019).
Google Sites
Report abuse