Inference and Prediction of Nanoindentation Response in FCC Crystals: Methods and Discrete Dislocation Simulation Examples
Nanoindentation, a common technique for probing the mechanical properties of crystalline materials, exhibits both surface and bulk-dominated responses that are linked to the parent crystal's elasticity and plasticity. For FCC crystals, the nanoindentation response is primarily controlled by the indented-grain's crystalline orientation and dislocation density. However, the indentation properties, such as hardness, have ambiguous significance at the nanoscale due to tip-specific size effects. Through extensive discrete dislocation dynamics simulations of flat-punch nanoindentation on FCC single crystals, we show that a one-to-one correspondence between an indented location and the pre-existing dislocation density is possible with the help of statistical, machine-learning algorithms. We cluster and classify various experimentally plausible ensembles with varying dislocation density and/or crystalline orientation, using nanoindentation force-depth curves or post-indent surface displacement images. We use this classification to statistically predict the nanoindentation response of a given location, through the average response of classified load-displacement curves. In addition, we demonstrate that our simulation results qualitatively agree with experimental data on common FCC single crystals. Ultimately, we propose a novel pathway to uncover the depth dependence of local surface material properties for more complete, non-destructive, microstructural characterization, through nanoindentation.