Emission reduction of CO2 is an urgent global environmental problem. The photocatalytic reduction of CO2 has attracted lots of interest as a novel method for producing organic fuel, and this technology does not require the consumption of additional energy. So in this paper, the photocatalytic reduction processes of gas phase CO2 in the different optofluidic microreactors were simulated to analyze the reaction mechanism and the reactants transfer process. Firstly, the applicability of the classical Langmuir-Hinshelwood equilibrium adsorption model and the self-defined dynamic mass transfer model for the numerical simulation of optofluidic planar reactor were initially verified. The results of the simulation based on the Langmuir-Hinshelwood model were not consistent with the experimental results when the flow rate were high, while those of the simulation based on the self-defined dynamic mass transfer model were consistent with the experimental results. The result indicates that the mass transfer process of the reactants is the dominant factor which affects the reaction efficiency of the photocatalytic process in the microreactor. Then, the corresponding Sherwood number was calculated according to the mass transfer coefficient under different inlet flow rates, and the mass transfer characteristic correlation was established. Finally, the planar microreactor was optimized to be an inverted convex cylinders microreactor, and the numerical simulation of photocatalytic reduction process of gas phase CO2 was carried out by using self-defined dynamic mass transfer model. It was found that the inverted convex cylinders in the microreactor could significantly enhance the transfer effect of reactants during the reaction process and improve the conversion efficiency of CO2.