Indentation-driven machine learning approach for estimating rate-dependency in cohesive-frictional materials

The indentation technique offers an efficient, semi-destructive, and high-throughput approach for characterizing mechanical properties across various length scales. While extensively studied for metals and ceramics, its application to time- and pressure-sensitive materials, such as polymers, remains limited. For time-dependent materials, the strain rate sensitivity (SRS) of flow stress has been shown to correlate with the SRS of hardness, with a constraint factor governed by the hardness-to-effective elastic modulus ratio (H/E^∗). This study introduces a finite element (FE)-based parametric study to investigate the relationship between the SRS of hardness and the SRS of flow stress in materials exhibiting both time-dependent and pressure-sensitive behavior, incorporating the parameter H/E^∗. Finally, a neural network model is trained to predict the SRS of flow stress directly from the nano-indentation data, achieving an R² score of 0.91, an RMSE of approximately 0.04, and an MAE of around 0.03. The findings of this study provide an integrated data-driven and high-throughput framework for the accelerated characterization of materials.