固体表面能量分布不均匀性对于CDFT表征多孔材料孔径分布的影响机制

经典密度泛函理论（CDFT）在气体吸附表征多孔材料孔径分布（PSD）领域的前沿研究热点和有效的解决方案。但现有的技术仍有描述受限流体的CDFT过剩自由能泛函展开不准确以及多孔材料表面能量假设模型过于简略的局限，对PSD表征结果准确性影响很大。针对CDFT过剩自由能的问题，本项目提出采用深度学习方法用以描述过剩自由能的卷积神经网络的研究方法。完善CDFT过剩自由能展开后，针对孔材料表面能量假设不准确的问题，提出了二维表面能量分布模型用以研究界面间二元混合流体的热力学平衡研究。探索二维模型参数对于吸附等温线的预测结果的影响关系，阐明多孔材料的微观几何结构对于描述表面能量分布不均匀的影响机制，揭示表面能量分布不均匀假设描述气固界面间二元混合流体的热力学平衡规律。研究成果可为表征材料结构科学、微观下受限流体热力学提供与深度学习相结合的新思路。

Classical density functional theory (CDFT) is an advanced research and effective solution for gas adsorption characterization of pore size distribution (PSD) in porous materials. However, there are still limitations in describing the CDFT excess free energy functional expansion of confined fluids and the limitation of the surface energetical heterogeneity for porous materials, which has great influence on the accuracy of their PSD. To solve the problem of excess free energy of CDFT, this project proposes a deep learning method to study convolution neural network describing excess free energy. After the CDFT excess free energy is expanded, a two-dimensional surface energy distribution model is proposed to study the thermodynamic balance of binary mixtures between interfaces in order to improve the surface energetical heterogeneity of porous materials. The influence of 2-dimensional model parameters on the prediction results of adsorption isotherms is explored. The influence mechanism of the microscopic structure of porous materials on describing surface energetical heterogeneity is elucidated, and the thermodynamic equilibrium law of the two component mixture between gas and solid interfaces is described by revealing the surface energetical heterogeneity. The research results can provide a new idea, which is the technology by deep learning, to investigate the characterization of material structure science, thermodynamics of microscopic confined fluid.