Mechanical properties of molybdenum diselenide revealed by molecular dynamics simulation and support vector machine

Abstract

Despite the spurring interests in two-dimensional transition metal dichalcogenide (TMDC) materials, knowledge on the mechanical properties of one of their important member, i.e., molybdenum diselenide (MoSe₂) is scarce and remains an open topic. In this work, the mechanical properties of h-MoSe₂ and t-MoSe₂ were systematically investigated using classical molecular dynamics (MD) simulations combined with machine learning (ML) techniques. The effects of chirality, temperature and strain rate on fracture strain, fracture strength and Young's modulus were characterized in both armchair and zigzag directions. For h-MoSe₂, the fracture strengths were 13.6 and 13.0 GPa for armchair and zigzag chiralities, respectively, at 1 K and strain rate of 5 × 10⁻⁴ ps⁻¹; the corresponding fracture strains were 0.23 and 0.27. The Young's moduli in armchair and zigzag directions exhibited similar values of 100.9 and 99.5 GPa, respectively. For t-MoSe₂, much lower fracture strengths of 6.1 and 6.3 GPa, fracture strains of 0.13 and 0.15, and Young's moduli of 83.7 and 83.0 GPa were predicted under the same conditions. A total of 700 MD simulation cases were calculated under different impact factors and initial conditions, which were subsequently fed into the support vector machine (SVM) algorithm for ML modeling. After training, the ML model could predict the mechanical properties of both MoSe₂ types given only the input features such as chirality, temperature and strain rate.