Graphene, a single layer of carbon atoms tightly packed in a honeycomb lattice structure, has attracted extensive attention for both fundamental science studies and commercial applications. Due to the limitation of the exfoliation technique in producing large sheets of graphene, large-scale graphene sheets are synthesized by the chemical vapor deposition (CVD) technique. Graphene prepared by CVD is normally polycrystalline sheets composed of single-crystalline pristine graphene grains of varying orientation seamed at grain boundaries (GBs). The HR-TEM (high-resolution transmission electron microscopy) data and MD (molecular dynamics) simulations indicate that grain boundaries (GBs) are 1D line defects consisting of a regular array of edge-sharing pentagon/heptagon pairs. Grain boundaries act as defects and alter the mechanical properties of graphene. Previous studies have reported that pristine graphene is the strongest material with Young’s modulus of&nbsp;1.0 ± 0.1 TPa and strength of 130 GPa, while the inclusion of GBs reduces the Young's modulus to 600 GPa. It is important to understand the effect of grain size and grain boundaries orientation on the mechanical and fracture properties of polycrystalline graphene for its reliable use in industrial applications.Atomistic molecular dynamics (MD) simulations have been extensively used to understand the mechanical and fracture properties of two-dimensional materials. Recently, there has been a trend to utilize the data-driven computational techniques such as machine learning and deep learning to predict the behavior of materials. Due to their ability to learn the complex nature of data and make predictions, both machine learning, and deep learning techniques have been used in various industries applications to predict materials properties. In this approach, experimental or computation data are used to build a machine learning or deep learning model which can predict the material properties at a significantly lower computational cost than traditional methods such as molecular dynamics simulations.This study aims to develop a deep convolutional neural network-based method to predict the fracture strength of polycrystalline graphene from the data set generated through MD simulations. Graphene, a single atom-thicked monolayer of carbon atom arranged in a honeycombed latticed structure, displays fabulous properties which make it a great candidate for a variety of industrial applications. &nbsp;The MD simulations conducted in LAMMPS are used to predict the fracture strength of polycrystalline graphene under uniaxial tensile loading. More than 2000 data points are generated using graphene sheets of different grain sizes with grain orientations. For each configuration of the polycrystalline graphene, an atomic resolution image is created. These images and their corresponding fracture strength from the MD simulations are used as the input of the deep learning model. A deep convolution neural network model is used to predict the fracture strength of polycrystalline graphene from the atomic resolution image. The results show that our trained model can predict the fracture strength of polycrystalline graphene with high accuracy using very low computational cost.

A Deep Convolutional Neural Network-Based Method to Predict Accurate Fracture Strength of Poly-Crystalline Graphene

Paper Proceedings: https://asmedigitalcollection.asme.org/IMECE/IMECE2021/volume/85543

ASME 2021 International Mechanical Engineering Congress and Exposition (IMECE2021)

Supporting the need among researchers worldwide for additional online and multimedia content, ASME now offers the ASME Conference Video Collection. This platform will collect and provide long-term access to conference presentation videos, encompassing all technology areas within ASME's technical conference and events program. About 30,000-40,000 engineers across all mechanical engineering and related disciplines participate in ASME conferences each year. This includes authors of approximately 6,000 to 8,000 technical papers published by ASME, the majority of whom also submit video presentations for our virtual conferences. The video collection will be available as an add-on to existing product packages or as a separate subscription.