Session: 02-14-01: 3D/4D BioManufacturing, BioMaterials, & Computational Modeling

Paper Number: 100178

100178 - Formation Mechanism of Holey Graphene/graphene Oxide in Laser Heated Graphene Oxide-Hydrogen Peroxide Solution From Deep Neural Network Interatomic Potential 

Holey graphene/graphene oxide, single- or multi-layer graphene/graphene oxide containing nanometer- to micrometer-sized holes, has found promising applications in the areas of electrochemical energy storage, water desalination, hydrogen storage, solar water splitting, etc. The synthesis of holey graphene has greatly relied on solution-based processes. A typical route is to etch reduced graphene oxide in H₂O₂ solution under a sufficiently high temperature and high pressure. However, the formation mechanism of nano-/micro-holes in graphene still remains elusive, thus demanding rigorous molecular-level simulations. To elucidate the formation mechanism of holey graphene/graphene oxide using materials/molecular simulation methods, researchers have confronted the challenge of balancing the need for precision and speed when simulating the potential energy surface (PES) and interatomic interactions. A system size of thousands to millions of atoms, or even larger, is frequently required to understand many chemical reactions that are of considerable importance, whereas the formation of holey graphene/graphene oxide is an example of such a kind of chemical reaction. In general, the dissociation of H₂O₂ on graphene/graphene oxide has been identified into two distinct processes. The first kind describes the direct interaction of H₂O₂ with graphene/graphene oxide, whereas the second type describes the dissociation of H₂O₂ molecules into OH radical species, which subsequently react with the carbon atoms in the graphene/graphene oxide layer. We have recently developed an accurate (close to the accuracy of density-functional theory in predicting atomic forces) and stable deep neural network-based interatomic potential for graphene oxide-water-H₂O₂ using data from ab initio molecular dynamics simulations (AIMD). The deep neural network potential allowed us to perform large-scale molecular dynamics (MD) simulations of the formation process of holes in graphene oxide merged in H₂O₂ solution. Specifically, we found that the local temperature of the graphene oxide layers, instead of the average temperature of the entire system (consisting of graphene oxide and H₂O₂ solution), determines the onset of the formation of nanoholes in reduced graphene oxide layers. To comprehend these implications, we carried out four different AIMD simulations of H₂O₂ reacting with the following models: 1) with a perfect graphene layer, 2) with a defected graphene layer, 3) with a graphene oxide layer, and 4) with a defected graphene oxide layer. The obtained AIMD trajectories are transformed into data sets, which are subsequently trained using deep learning artificial neural networks. Finally, the trained data sets were then converted into interatomic potentials, which were then utilized to perform large-scale MD simulations. The details of the chemical reaction between graphene/graphene oxide and H₂O₂ are also being investigated and will be presented. These include reaction pathways, preferred reaction sites, and required local thermodynamic conditions.

Presenting Author: Iyyappa Rajan Panneerselvam University of Nevada Reno

Presenting Author Biography: Dr. Iyyappa Rajan Panneerselvam obtained his Ph.D. in Chemistry, focused on computational materials, from Vellore Institute of Technology (VIT), Chennai Campus, Tamil Nadu, India, in 2018. He was a Young Scientist Research Fellow at the Asia Pacific Center for Theoretical Physics (APCTP), POSTECH Campus, South Korea from 2019 to 2021. Since May of 2021, he has been working as a Post-doctoral Research Fellow in Prof. Yan Wang's Nano-Thermal-Mechanical Engineering (NTME) Laboratory & Group, Department of Mechanical Engineering, University of Nevada, Reno, USA. Rajan has authored 23 research articles, including 6 recent corresponding authored articles in ACS, RSC, AIP, and IOP journals, and has expertise in broad areas of computational materials science, but not limited to first-principles calculations.

Authors:

Iyyappa Rajan Panneerselvam University of Nevada Reno
Yan Wang University of Nevada Reno