Investigating the atomic-level mechanisms of negative thermal expansion in Zn(CN)₂ using machine learning-enabled molecular dynamics

Abstract The negative thermal expansion behavior (NTE) of Zn(CN) 2 over a wide range of temperature has received significant attention. Here, we develop a machine learning interatomic potential and perform large-scale molecular dynamics simulations through which we unveil the NTE mechanisms of Zn(CN) 2 . Analysis of the trajectories of atoms reveal that the C and N atoms move on the lateral side of Zn…Zn axis with different distance. The C≡N bond is not exactly parallel to the Zn…Zn axis. The bridging C and N atoms exhibit a unique motion behavior that combines lateral movement with a tilted aspect, which we term the ‘lateral + tilt’ mode. The transition pressure from cubic to orthorhombic phase under ambient temperature is found to be 2.7 GPa. Machine learning-enabled molecular dynamics open a way to better comprehend the physical origins of NTE.