Injection of CO2 into coal is known to improve the yields of coal-bed methane gas. However, the technology of CO2 injection-enhanced coal-bed methane (CO2-ECBM) recovery is still in its infancy with an unclear mechanism. Density functional theory (DFT) calculations were performed to elucidate the mechanism of CO2 adsorption-assisted CH4 desorption (AAD). To simulate coal surfaces, different six-ring aromatic clusters (2 x 2, 3 x 3, 4 x 4, 5 x 5, 6 x 6, and 7 x 7) were used as simplified graphene (Gr) carbon models. The adsorption and desorption of CH4 and/or CO2 on these carbon models were assessed. The results showed that a six-ring aromatic cluster model (4 x 4) can simulate the coal surface with limited approximation. The adsorption of CO2 onto these carbon models was more stable than that in the case of CH4. Further, the adsorption energies of single CH4 and CO2 in the more stable site were-15.58 and -18.16 kJ/mol, respectively. When two molecules (CO2 and CH4) interact with the surface, CO2 compels CH4 to adsorb onto the less stable site, with a resulting significant decrease in the adsorption energy of CH4 onto the surface of the carbon model with pre-adsorbed CO2. The Mulliken charges and electrostatic potentials of CH4 and CO2 adsorbed onto the surface of the carbon model were compared to determine their respective adsorption activities and changes. At the molecular level, our results showed that the adsorption of the injected CO2 promoted the desorption of CH4, the underlying mechanism of CO2-ECBM. (C) 2016 Published by Elsevier B.V.