摘要/Abstract

采用分子模拟高通量筛选的方法研究了6013种实验已经合成的金属-有机框架（MOFs）对天然气五元混合物（CH₄，C₂H₆，C₃H₈，H₂S和CO₂）中H₂S和CO₂的吸附分离性能.为了综合吸附量和选择性这两项指标，我们首先比较了三种权衡方法[权衡α法（Tradeoff between S_{H2S+CO2/C1-C3} and N_H2S+CO2，TSN），标准值法（Standard normal method，SNM）和权衡β法（Tradeoff between selectivity and capacity，TSC）].接着，针对四种MOF描述符[最大孔径（LCD），孔隙率（φ），比表面积（VSA）和吸附热（Q_st⁰）]，通过Pearson相关系数分析了每种描述符分别对三种权衡变量的相关性，结果显示TSC法与四种MOF描述符的相关性最佳.然后，使用多元线性回归方法定量地分析了四种MOF描述符分别对TSC的影响程度；而决策树模型则被用于规划性能高效MOFs的设计路径.最后，20种性能最优MOFs从数据库中脱颖而出，它们将为净化天然气技术的发展提供坚实的理论指导.
关键词: 分子模拟, 金属-有机框架, 吸附, H₂S, CO₂
In this work, the adsorption performance of 6013 computation-ready, experimental metal-organic frameworks (CoRE-MOFs) for the capture of H₂S and CO₂ from natural gas mixture (CH₄, C₂H₆, C₃H₈, H₂S and CO₂) is calculated by high-throughput screening of grand canonical Monte Carlo (GCMC) simulation in 298 K and 10 bar. For the comprehensive consideration of both adsorption capacities and selectivities of H₂S+CO₂, first, we compare three different tradeoff methods (α tradeoff method (Tradeoff between S_{H2S+CO2/C1-C3} and N_H2S+CO2, TSN), standard normal method (SNM), β tradeoff method (Tradeoff between selectivity and capacity, TSC)). The effect of selectivity on the new tradeoff variables are appropriately reduced by these tradeoff methods, because some of selectivities are very high. Thus, the new tradeoff variables can comprehensively evaluate the adsorption performance of CoRE-MOFs. Moreover, the correlation of each MOF descriptor (including the largest cavity diameter (LCD), void fraction (φ), surface area (VSA) and isosteric heat (Q_st⁰)) with three tradeoff variables are analyzed by Pearson correlation coefficient, respectively. The LCDs are calculated by Zeo++ software, but the φ and VSA are simulated by RASPA using probes of He and N₂, respectively. The Q_st⁰ of each adsorbate gas are calculated at infinite dilution condition using NVT-MC method. All GCMC simulations for the screening are carried out using RASPA software. The results show that TSC has the best correlation with four MOF descriptors and the linear model could sufficiently describe the relationship between TSC and four MOF descriptors. Pearson correlation coefficients of four descriptors were -0.613, -0.717, -0.673 and 0.536 on TSC, respectively. Multiple linear regression is applied to quantitatively determine the influencing degree of four descriptors on performance, respectively. Among the four descriptors, Q_st⁰, φ, and LCD have larger standardized regression coefficients compared with VSA. This indicates that Q_st⁰, φ, and LCD are more useful in describing the performances of the MOFs. Thus, these three descriptors are used in the decision tree modeling to define an effective path for screening high-performance MOFs. It is concluded that a maximum probability (77.6%) of finding the good MOFs can be obtained from the three descriptors. Finally, the 20 best MOFs stand out from the whole database, and find that the alkali or alkaline earth metals in MOFs could effectively enhance the separation performance of H₂S and CO₂. The microscopic insights and guidelines by this computational study can provide significant theoretical guidance for the development of adsorbent for the purification of natural gas.
Key words: molecular simulation, metal-organic frameworks, adsorption, H₂S, CO₂


PDF全文下载地址:

点我下载PDF