Introduction

Homogeneous plastic flow, different from shear banding, is a ductile deformation mode of metallic glasses (MGs) when the sample size is reduced to sub-100 nm1,2. Computational simulations have shown their advantages in predicting nanostructures of MGs with homogeneous plastic flow. Evolutionary algorithms3 are able to search for MG candidates with excellent glass forming ability (GFA), and thus optimize the existing MGs. Machine learning based modelings4,5,6 can predict structure heterogeneities for different MG systems and correlate their structures with mechanical properties. Multiscale modelings by molecular dynamics simulations7,8 demonstrate that heterogeneous MGs (nanoglasses or nanolaminated composites) have enhanced homogeneous deformation. In experiments, however, it is difficult to achieve very large heterogeneity as that in simulations and the mechanical strain rates are much lower (~10−3 s−1 vs. ~107 s−1). Adding crystalline phase into the amorphous matrix9,Full size image

Here, we realize this idea by using magnetron sputtering to alternately deposit 3.4-nm-thick Ti-Zr-Nb-Si-XX and 2.9-nm-thick Mg-Zn-Ca-YY nanodomains using Ti60Zr10Nb15Si15 (at.%) and Mg60Zn35Ca5 (at.%) targets, respectively. The received material shows a high yield strength of 1.6 GPa and plastic deformation of over 60% strain (including ~40% homogeneous plastic flow). During plastic flow, dynamic atomic intermixing occurs between the amorphous nanodomains, originating from atomic rearrangements via profuse shear transformation events. This behavior prevents possible interface failure and thus leads to mechanical stabilization of the whole material.