Amorphous graphene (a-G), the disordered counterpart of crystalline graphene (c-G), holds promise for flexible electronics, catalysis, and energy storage, yet its atomic-scale structure and property origins remain elusive. Here, we introduce a helium potential-wall method within ab initio molecular dynamics accelerated by machine-learning force fields to enable stable melt-quench simulations of two-dimensional (2D) a-G. We generate 100 statistically representative models that ensure convergence of calculated properties. Structural analysis unveils a buckled, ${\mathit{sp}}^2$-hybridized continuous random network dominated by five- to eight-membered rings, with a mean bond length of 1.43 Å and broadened bond-angle distributions mainly spanning ${108}^{\circ }–{135}^{\circ }$. Electronically, disorder induces $p_x/p_y$ mixing near the Fermi level, resulting in localized $\pi$ states and a pronounced reduction of in-plane conductivity compared to crystalline graphene, as revealed by the Kubo-Greenwood formalism, with a consistent qualitative trend across representative configurations. Mechanically, ensemble-averaged stiffness yields a Young's modulus $\overline{E}\approx 315$ N/m ($\sim 7\%$ lower than c-G) and shear modulus $\overline{G}\approx 133$ N/m ($\sim 10\%$ lower), indicating a disorder-induced flexibilization of the network and macroscopic in-plane isotropy. This framework elucidates disorder's role in governing a-G properties and provides a transferable protocol for exploring other 2D amorphous materials.

Close

PrevNext

View figure in article

• Amorphous materials
• Graphene
• Ab initio molecular dynamics