Files
Abstract
I develop a generic coarse-grained potential in order to study the mechanical performance of 2D materials-based cellular kirigami structures towards understanding of the relationship between mechanical properties, structure pattern, and component material. By patterning the structure lattice cell, the 2D materials-based structures show a very wide range of mechanical properties across several orders of magnitude. Moreover, results indicate that there are two distinct stress- strain stages, J-shape non-linear elasticity and linear elasticity, determined by the material structure density. Results also indicate that hole-in structures show a better performance over no-hole structures for ductility. In addition, the material effect on the mechanical performance of 2D materials-based cellular kirigami is significant, exemplified by graphene-based structures outperforming those composed of other 2D materials. Furthermore, by integrating coarse-grained molecular dynamics (CGMD) simulations with machine learning algorithms, the mechanical performance of 2D materials-based kirigami structures with mixed cellular patterns can bepredicted in order to optimize the design patterns of the kirigami structures. CGMD simulation are performed to obtain stress-strain responses of the 3 ∗ 3 grid architectured graphene kirigami under biaxial tensile tests, in which 2,483 datasets are obtained with different combinations of cellular patterns. With previous stress-strain responses, feedforward neural networks (FNN) have been applied in order to obtain prediction of mechanical performance. Result shows the R2 were able to increase up to 0.4 by optimizing the structure of the input dataset, which revealed the possibility of using machine learning method to design and predict mechanical properties of architectured 2D materials, it also specifically demonstrated that FNN could achieve an acceptable accuracy in terms of prediction by increasing the structural details and size of the training dataset . Overall, this study provides a computational basis towards the predictive design of future kirigami structures with outstanding properties and functions.