Abstract

In this work, nanoindentation tests were performed on CuZr MGs subjected to thermal-pressure treatments using molecular dynamics simulations, with pressures in the range of 0 to 20 GPa. No significant variations in the mechanical properties were observed. However, in the 20 GPa sample, shear bands (SBs) penetrated slightly more homogeneously compared to the other cases. Furthermore, a novel machine learning algorithm, called MultiSOM, revealed that the contact zone with the indenter was associated to high-shear atoms, whereas the inner and outer regions of the SBs were related to medium-shear and low-shear atoms, respectively. The 20 GPa case presented larger fractions of low-shear atoms, whereas the opposite behavior occurred in the 0 GPa case. Thus, despite the slight variation in mechanical properties, higher pressures still result in less-localized deformation. Overall, our results provide a quantitative description of plasticity in MGs using machine learning methods.