Due to the current development of various machine learning models, there is an increasingly rapid and efficient exploration and synthesis of new material systems [1]. However, it is not yet determined whether these new alloys can prevail against already established and well investigated materials. This requires various - sometimes extremely time-consuming - mechanical experiments to determine so called high-fidelity mechanical characteristics, such as stress-strain behavior or the fatigue life of a material. In contrast, nanoindentation can be used to determine a multitude of selected material properties such as hardness and Young's modulus [2], as well as strain rate sensitivity, wear resistance and creep properties [3-5] within a rather short time.

The goal of this work is to study a model material (Cu) in different microstructural states (single crystal, coarse grained, coarse grained and work hardened, ultrafine-grained) using nanoindentation as well as subsequent AFM investigations of the surface topography. Using the Young’s Modulus and Berkovich hardness (at 7% plastic strain) from nanoindentation plus the pile-up volume (related to strain hardening rate), the prediction of (compressive) flow curves should be possible. For this we train machine learning models of various complexity on the obtained data as well as macroscopic or microscopic (pillar) compressive stress-strain curves. The trained model will then be applied to predict stress strain curves of similar fcc materials (such as brass, bronze, aluminium or nickel), allowing for a quick screening of flow properties of newly developed materials using high-throughput nanoindentation.

References
[1] D.S. Gianola, Curr. Opin. Solid State Mater. Sci., 2023, 27, 101090.
[2] W.C. Oliver, G.M. Pharr, J. Mater. Res, 1992, 6, 20.
[3] V. Maier-Kiener, K. Durst, JOM, 2017, 11, 2246-2255.
[4] K. Durst, V. Maier-Kiener, Curr. Opin. Solid State Mater. Sci., 2015, 19, 340-353.
[5] B. Bushan, Wear, 1995, 181-183, 743-758.