The structural evolution of SiBCNZr amorphous ceramics analyzed by machine learning potential

In the present study, a suitable machine learning potential for SiBCNZr amorphous materials was used in classical molecular dynamics (CMD) for creating the large-scale SiBCNZr amorphous materials with 3375 atoms. This study focused on the effects of Zr on the amorphous structure and atomic diffusion of SiBCN materials. Moreover, the effects of different molar ratio of Zr, B and C atoms on SiBCNZr amorphous materials structure and properties were also detected. The results have revealed the structure evolution of SiBCNZr materials, in which B–N–C clusters with B–N at the core and C aggregated at outer layer formed. The Zr atoms randomly distributed instead of wrapping by the clusters. In addition, it was found that the increasing of Zr contents reduced the size and number of B–N–C clusters, in contrast to the effect of increased B content. The high contents of Zr caused the formation of Zr-rich region, having negative impact on the structural stability. Besides, the increasing of B content facilitated the formation of B–N–C clusters by affect the contents of B–C bonds in the materials, which was also detrimental to the mechanical performances. The lower contents of C in Si₂B₁C₁N₁Zr_0.15 caused the lower density and ultimate tensile stress of amorphous materials. SiBCNZr amorphous materials were broken at different parts with the different atoms molar ratio.