First-principles and machine learning investigation of the structural and optoelectronic properties of dodecaphenylyne: a novel carbon allotrope

We report the computational discovery and characterization of Dodecaphenylyne (DP), a novel carbon allotrope with a unique geometric structure. The structural, dynamic, mechanical, electronic, and optical properties of DP were evaluated using density functional theory (DFT) and a machine learning interatomic potential trained specifically for this material. The formation energy of -7.98 eV per atom indicates high dynamic stability, further supported by the absence of imaginary phonon modes and the preservation of structural integrity up to 1000 K in ab initio molecular dynamics (AIMD) simulations. Both DFT and AIMD simulations were performed within the generalized gradient approximation using the PBE functional. Mechanical analysis reveals high in-plane stiffness with directional dependence: Young's modulus values of 469.09 GPa and 600.41 GPa along the x and y directions, respectively. Electronic band structure and projected density of states analyses confirm the DP semiconducting character. Calculations of carrier mobility using the deformation potential theory reveal pronounced anisotropy, with maximum values reaching up to 30.6 × 10⁴ cm² V^-1 s^-1 (electrons, e) and 8.4 × 10⁴ cm² V^-1 s^-1 (holes, h), much higher than the observed for other 2D materials. DP also exhibits anisotropic optical absorption in the visible and ultraviolet spectrum, highlighting its potential for optoelectronic applications.