Four-phonon scattering significantly reduces the predicted lattice thermal conductivity in penta-graphene: A machine learning-assisted investigation

•Machine learning potential (MLP) with high accuracy is trained for 2D PG.•Intrinsic thermal conductivity (298–45 Wm−1K−1, 300–900 K) of 2D PG is obtained by considering the three-phonon and four-phonon scatterings.•Four-phonon scattering plays a significant role in reducing the thermal conductivity of PG.•The contribution of diffusive modes to the thermal conduction of PG is negligible.