Microstructural atomic defects, including voids, cleavage, and inclusions, are commonly observed in alumina materials, and their impact on mechanical properties, such as fracture stress and toughness, is significant. In this paper, we introduce novel alumina models that incorporate experimentally observed void features. An atomic model is established to study the effects of micro-structural void features on fracture properties and atomic structure changes using molecular dynamics simulations. The electron backscatter diffraction and scanning electronic microscopy analysis of experimental samples are used to evaluate microstructural features that are used as inputs to the simulations (e.g., void aspect ratio, void angle). We apply an innovative Atomistic-to-Continuum (ATC) method based on Riemann sums to bridge atomic and continuum mechanics theories, evaluating the resistance of materials with atomic defects to crack propagation. The results show the greatest effects of pore angles on weakening mechanical properties such as peak strength and fracture energy density. The accuracy and efficiency of the ATC method in evaluating stress intensity factors are used to calculate the mechanical responses. Additionally, we establish a multiple layer perceptron neural network to evaluate the complex relationship between void features (aspect ratio, pore angle, relative distance) and typical fracture properties (fracture stress, critical stress intensity factor). A meta-analysis of these results from both machine learning methods and molecular dynamics simulations reveals the significant impact of each void feature on the sensitivity of typical fracture properties (peak strength, critical stress intensity factor at peak strength) and highlights the critical role of aspect ratio on fracture properties.
Read full abstract