UC San Diego

Metals and alloys are the most viable solution for structural application, despite its very limited range of occupancy in the material property space. Metallic composites have long been conceptualized as a way to extend the range of structural properties, albeit with limited success in practice due to their strength-ductility trade-off. In recent years, nano-structuring has also emerged as a promising tool to obtain properties not attainable through alloying or metallic composites due to its unusually large proportion of nano-interfaces, both single phase such as in the nanocrystalline and multi-phase like in Nanocomposites. Hence to materialize the promise of nanostructured materials and incorporate them in the manufacturing process, a modeling technique is required that is based on the atomistic scale deformation mechanism near the nano-interfaces. In this work a comprehensive atomistic study is conducted to identify the key atomistic mechanisms in nano-crystalline and nanocomposite materials and a machine learning assisted multi-scale modeling technique is implemented to understand large scale implication of the atomistic mechanisms. To that end, molecular dynamic simulation is performed to study the effect of nano-interfaces in the nano-crystalline Magnesium (Mg) and the Aluminum-Silicon Carbide (Al-SiC) Metal Matrix Nanocomposites (MMNCs) materials. A series of machine learning based surrogate model is then trained which are subsequently used in a continuum scale model based on Finite Element Method (FEM). The atomistic scale results reveal the anisotropic deformation in nano-crystalline Mg is highly dependent on the grain size. Deformation in Al-SiC MMNC is accommodated through three subsequent mechanisms namely, defect free, dislocation dominant and nano-interface separation. Multiscale modeling reveals that propagation path of the dislocation dominant mechanism correlates closely to the shear-band formation path. The broader implications of the atomistic findings and the multiscale modeling outcome opens up the possibility to device a multiscale modeling framework for nanostructured materials where conventional dislocation theory is not appropriate to upscale atomistic scale information