Study on the conductive behavior in three-dimensional high-density carbon system

Exploring the relationship between structural evolution and electrical conductivity in three-dimensional (3D) high-density carbon (HDC) is essential for understanding its functional properties. In this work, a deep-learning-based machine learning potential was employed to construct atomistic HDC models with densities ranging from 2.2 to 2.8 g cm−3. Molecular dynamics simulations were performed to investigate the evolution of the sp2/sp3 hybridization ratio and the medium-range order parameter (ηMRO). Conductive sites were identified to quantify their average density (ρsites), and a two-step screening strategy was used to extract effective sites within nanocrystalline regions. Monte Carlo simulations were conducted to evaluate the relative electrical conductivity, enabling analysis of the density–structure–relative conductivity relationship. The results show that ηMRO exhibits a non-monotonic trend with increasing density, while ρsites shows an opposite variation. The electrical conductivity is governed by the coupling between transport channel continuity (ηMRO) and hopping accessibility (ρsites), leading to a non-monotonic dependence on density. An optimal conductivity window is identified at 2.256–2.4 g cm−3, where the synergistic balance between the two parameters minimizes electron transport resistance. This work reveals a dual-order-parameter conduction mechanism and advances the understanding of the structure–transport relationship in high-density carbon, providing useful guidance for the design of high-performance carbon-based materials.