机器学习辅助高通量筛选金属有机骨架用于富碳天然气中分离CO2

doi:10.11949/0438-1157.20241001

摘要/Abstract

摘要：

在碳达峰和碳中和目标的推动下，开发绿色化学技术，如利用海上风电电解水生产的绿色氢气和从富碳天然气中分离出来的CO₂合成绿甲醇具有重要的社会经济意义。但如何高效分离海洋富碳天然气中的二氧化碳成为其中的关键技术难点，常规的高通量筛选方法用于金属有机骨架（MOFs）分离实际天然气组分CO₂面临着模型复杂性高、求解时间长的问题。提出了一种机器学习辅助的高通量筛选策略，其在训练集和测试集上的R²值分别超过了0.98和0.92，可用于快速从富碳天然气六元混合物（N₂、CO₂、CH₄、C₂H₆、C₃H₈、H₂S）中分离出CO₂。

关键词: 金属有机骨架, 高通量筛选, CO₂分离, 机器学习, 分子模拟, 富碳天然气

Abstract:

Driven by the goal of carbon dioxide peaking and carbon neutrality, it is of great social and economic significance to develop green chemical technologies, such as the substantial use of H₂ generated by water electrolysis with offshore wind power and CO₂ separated from CO₂-rich natural gas to produce green methanol is gaining significant socioeconomic and environmental relevance. However, how to efficiently separate carbon dioxide from marine carbon-rich natural gas has become a key technical difficulty. Conventional high-throughput screening methods for metal organic frameworks (MOFs) to separate actual natural gas component CO₂ face the problems of high model complexity and long solution time. Therefore, a machine learning-assisted high-throughput screening strategy is proposed. The R² values on the training set and the test set are more than 0.98 and 0.92, respectively, which can be used to quickly and efficiently separate CO₂ from the actual natural gas of six components (N₂, CO₂, CH₄, C₂H₆, C₃H₈, H₂S).

Key words: metal-organic frameworks, high-throughput screening, CO₂ separation, machine learning, molecular simulation, CO₂-rich natural gas

中图分类号:

周印洁, 吉思蓓, 何松阳, 吉旭, 贺革. 机器学习辅助高通量筛选金属有机骨架用于富碳天然气中分离CO₂[J]. 化工学报, 2025, 76(3): 1093-1101.

Yinjie ZHOU, Sibei JI, Songyang HE, Xu JI, Ge HE. Machine learning-assisted high-throughput screening approach for CO₂ separation from CO₂-rich natural gas using metal-organic frameworks[J]. CIESC Journal, 2025, 76(3): 1093-1101.

图/表 9

图1 用于CO2分离的高性能MOFs的筛选工作流程图(1Å=0.1 nm)

Fig.1 The workflow illustration for screening high-performance MOFs for CO2 separation

表1 吸附质的Lennard-Jones势能参数和原子电荷

Table 1 Lennard-Jones potential energy parameter and atomic charge of adsorbed material

Atom	(ε/k_B)/K	σ/Å	Charge
C_CO₂	27.00	2.80	0.700
O_CO₂	79.00	3.05	-0.350
H_H₂S	50.00	2.50	0.210
S_H₂S	122.00	3.60	0
M_H₂S	0	0	-0.420
N_N₂	36.00	3.31	-0.482
N_COM	0	0	0.964
CH₄	148.00	3.73	0
CH₃	98.00	3.75	0
CH₂	46.00	3.95	0

表2 五折交叉验证的网格搜索方法确定的最优超参数

Table 2 Optimal hyperparameters identified by the grid search method using five-fold cross-validation

Hyperparameter	Hyperparameter space	Optimal value
n_estimators	range(100, 501, 50)	200
max_depth	range(5, 51, 5)	20
min_samples_split	range(2, 12, 2)	2
min_samples_leaf	range(1, 6, 1)	1
criterion	[‘gini’, ‘entropy’]	‘gini’
max_features	[‘auto’, ‘sqrt’, ‘log2’]	‘auto’
class_weight	[‘balanced’]	‘balanced’

表3 RF模型预测CO2分离R的性能

Table 3 Performance of RF models in predicting R for CO2 separation

Model	Descriptor type	Training set			Test set
Model	Descriptor type	R²	MAE/(mol·kg^-1)	RMSE/(mol·kg^-1)	R²	MAE/(mol·kg^-1)	RMSE/(mol·kg^-1)
R/%	structural	0.911	0.049	0.071	0.422	0.123	0.184
	chemical	0.983	0.019	0.031	0.898	0.047	0.078
	both	0.986	0.017	0.028	0.922	0.040	0.069

图2 预测R%值与GCMC模拟R%值在训练集和测试集的比较

Fig.2 Comparison of predicted R% values with GCMC simulated R% values

图3 MOFs在性能度量空间中的分布

Fig.3 The distribution of MOFs in performance metrics

表4 筛选出的高性能MOFs吸附剂的结构特征和分离性能

Table 4 Top 10 high-performance MOFs selected according to their structural characteristics and separation performance

CCDC	MOF	LCD/Å	VSA/(m²·cm^-3)	VF	$Δ N_{C O_{2}}$ /(mol·kg^-1)	$S_{C O_{2} / S C}$	TSN/(mol·kg^-1)	R/%
183282	XIGWUF	7.07	3242.86	0.70	7.72	39.74	12.34	88.92
1501708	ETECOX	10.31	1753.04	0.76	5.74	28.92	8.39	90.02
1432726	MAGNEQ	5.31	1670.44	0.52	4.29	50.76	7.32	85.05
808299	NAQRAA	8.82	1966.76	0.65	4.09	33.87	6.26	86.65
264131	PARHAS	6.77	1866.99	0.64	4.41	15.10	5.20	90.88
1494751	ORIVUI	11.75	1483.49	0.71	3.69	21.73	4.94	86.96
791072	RUZDAS	5.27	891.53	0.37	2.73	52.63	4.70	86.76
785296	CUSDIE	13.10	1619.70	0.64	3.65	16.64	4.46	86.62
264132	PARHEW	7.39	2084.34	0.69	3.44	12.48	3.77	91.21
239782	FIFNUE	4.75	943.07	0.46	3.25	14.52	3.77	87.19

表4 筛选出的高性能MOFs吸附剂的结构特征和分离性能

Table 4 Top 10 high-performance MOFs selected according to their structural characteristics and separation performance

CCDC	MOF	LCD/Å	VSA/(m²·cm^-3)	VF	$Δ N_{C O_{2}}$ /(mol·kg^-1)	$S_{C O_{2} / S C}$	TSN/(mol·kg^-1)	R/%
183282	XIGWUF	7.07	3242.86	0.70	7.72	39.74	12.34	88.92
1501708	ETECOX	10.31	1753.04	0.76	5.74	28.92	8.39	90.02
1432726	MAGNEQ	5.31	1670.44	0.52	4.29	50.76	7.32	85.05
808299	NAQRAA	8.82	1966.76	0.65	4.09	33.87	6.26	86.65
264131	PARHAS	6.77	1866.99	0.64	4.41	15.10	5.20	90.88
1494751	ORIVUI	11.75	1483.49	0.71	3.69	21.73	4.94	86.96
791072	RUZDAS	5.27	891.53	0.37	2.73	52.63	4.70	86.76
785296	CUSDIE	13.10	1619.70	0.64	3.65	16.64	4.46	86.62
264132	PARHEW	7.39	2084.34	0.69	3.44	12.48	3.77	91.21
239782	FIFNUE	4.75	943.07	0.46	3.25	14.52	3.77	87.19

图 4 10个高性能分离CO2 MOFs的晶体结构

Fig.4 Crystal structure of the top 10 high-performance CO2 separation

图 5 结构特征与性能指标之间的关系图：(a)PLD~Qst0~ΔN；(b) S~LCD~PLD；(c) TSN~ρ~GSA；(d)LCD~ρ~PLD（从蓝色到黄色依次表示从低到高的编码值）

Fig.5 Relationship plots of (a)PLD~Qst0~ΔNCO2， (b) S~LCD~PLD， (c) TSN~ρ~GSA， (d) LCD~ρ~PLD for 2731 MOFs with R≥85%（the color from red to blue refers to the coded value from low to high）

参考文献 34

1	唐春, 周乐懿, 李东升, 等. 绿氢制绿甲醇的技术经济可行性分析[J]. 世界石油工业, 2024, 31(1): 92-99.
	Tang C, Zhou L Y, Li D S, et al. Technical and economic analysis of green hydrogen for green methanol synthesis[J]. World Petroleum Industry, 2024, 31(1): 92-99.
2	Din I U, Alotaibi M A, Alharthi A I, et al. Green synthesis approach for preparing zeolite based Co-Cu bimetallic catalysts for low temperature CO₂ hydrogenation to methanol[J]. Fuel, 2022, 330: 125643.
3	季东, 王健, 王可, 等. 不同CO₂捕集技术的CO₂耦合绿氢制甲醇工艺研究[J]. 化工学报, 2022, 73(10): 4565-4575.
	Ji D, Wang J, Wang K, et al. Process research of methanol production by CO₂ coupled green hydrogen with different CO₂ capture technologies[J]. CIESC Journal, 2022, 73(10): 4565-4575.
4	Wang D L, Meng W L, Zhou H R, et al. Green hydrogen coupling with CO₂ utilization of coal-to-methanol for high methanol productivity and low CO₂ emission[J]. Energy, 2021, 231: 120970.
5	王江涛, 鹿晓斌. CO₂促进"甲醇经济"与"氢经济"共同发展[J]. 现代化工, 2021, 41(7): 14-18, 25.
	Wang J T, Lu X B. Together development of "methanol economy" and "hydrogen economy" driven by CO₂ utilization[J]. Modern Chemical Industry, 2021, 41(7): 14-18, 25.
6	文一如, 付佳, 刘大欢. 基于机器学习的MOFs材料研究进展: 能源气体吸附分离[J]. 化工学报, 2024, 75(4): 1370-1381.
	Wen Y R, Fu J, Liu D H. Advances in machine learning-based materials research for MOFs: energy gas adsorption separation[J]. CIESC Journal, 2024, 75(4): 1370-1381.
7	王玉杰, 李申辉, 赵之平. M-MOF-74吸附分离H₂/He混合物的分子模拟研究[J]. 化工学报, 2022, 73(10): 4507-4517.
	Wang Y J, Li S H, Zhao Z P. Molecular simulation study on adsorption and separation of H₂/He mixtures by M-MOF-74[J]. CIESC Journal, 2022, 73(10): 4507-4517.
8	李建惠, 兰天昊, 陈杨, 等. MOF复合材料在气体吸附分离中的研究进展[J]. 化工学报, 2021, 72(1): 167-179.
	Li J H, Lan T H, Chen Y, et al. Research progress of MOF-based composites for gas adsorption and separation[J]. CIESC Journal, 2021, 72(1): 167-179.
9	Aghaji M Z, Fernandez M, Boyd P G, et al. Quantitative structure-property relationship models for recognizing metal organic frameworks (MOFs) with high CO₂ working capacity and CO₂/CH₄ selectivity for methane purification[J]. European Journal of Inorganic Chemistry, 2016, 2016(27): 4505-4511.
10	Gao W Q, Zheng W Z, Yan K X, et al. Accelerating the discovery of acid gas-selective MOFs for natural gas purification: a combination of machine learning and molecular fingerprint[J]. Fuel, 2023, 350: 128757.
11	Haldoupis E, Nair S, Sholl D S. Efficient calculation of diffusion limitations in metal organic framework materials: a tool for identifying materials for kinetic separations[J]. Journal of the American Chemical Society, 2010, 132(21): 7528-7539.
12	Watanabe T, Sholl D S. Accelerating applications of metal-organic frameworks for gas adsorption and separation by computational screening of materials[J]. Langmuir, 2012, 28(40): 14114-14128.
13	Wilmer C E, Farha O K, Bae Y S, et al. Structure-property relationships of porous materials for carbon dioxide separation and capture[J]. Energy & Environmental Science, 2012, 5(12): 9849.
14	Laha S, Dwarkanath N, Sharma A, et al. Tailoring a robust Al-MOF for trapping C₂H₆ and C₂H₂ towards efficient C₂H₄ purification from quaternary mixtures[J]. Chemical Science, 2022, 13(24): 7172-7180.
15	Yuan J P, Liu X Y, Li X D, et al. Computer simulations for the adsorption and separation of CH₄/H₂/CO₂/N₂ gases by hybrid ultramicroporous materials[J]. Materials Today Communications, 2021, 26: 101987.
16	杨文远, 梁红, 乔智威. 高通量筛选金属-有机框架: 分离天然气中的硫化氢和二氧化碳[J]. 化学学报, 2018, 76(10): 785-792.
	Yang W Y, Liang H, Qiao Z W. High-throughput screening of metal-organic frameworks for the separation of hydrogen sulfide and carbon dioxide from natural gas[J]. Acta Chimica Sinica, 2018, 76(10): 785-792.
17	Chung Y G, Haldoupis E, Bucior B J, et al. Advances, updates, and analytics for the computation-ready, experimental metal-organic framework database: core MOF 2019[J]. Journal of Chemical & Engineering Data, 2019, 64(12): 5985-5998.
18	Willems T F, Rycroft C H, Kazi M, et al. Algorithms and tools for high-throughput geometry-based analysis of crystalline porous materials[J]. Microporous and Mesoporous Materials, 2012, 149(1): 134-141.
19	Li J R, Kuppler R J, Zhou H C. Selective gas adsorption and separation in metal-organic frameworks[J]. Chemical Society Reviews, 2009, 38(5): 1477.
20	Jablonka K M, Ongari D, Moosavi S M, et al. Big-data science in porous materials: materials genomics and machine learning[J]. Chemical Reviews, 2020, 120(16): 8066-8129.
21	Xue X Y, Cheng M, Wang S H, et al. High-throughput screening of metal-organic frameworks assisted by machine learning: propane/propylene separation[J]. Industrial & Engineering Chemistry Research, 2023, 62(2): 1073-1084.
22	Dubbeldam D, Calero S, Ellis D E, et al. RASPA: molecular simulation software for adsorption and diffusion in flexible nanoporous materials[J]. Molecular Simulation, 2016, 42(2): 81-101.
23	Qing W. Study on technology clusters for direct utilization of CO₂-rich natural gas and construction of hybrid system for energy and chemicals production[J]. China Pet Process Petrochemical Technol, 2020, 22(2): 1-9.
24	Cheng M, Wang S H, Zhang Z Y, et al. High-throughput virtual screening of metal-organic frameworks for xenon recovery from exhaled anesthetic gas mixture[J]. Chemical Engineering Journal, 2023, 451: 138218.
25	Shah M S, Tsapatsis M, Siepmann J I. Development of the transferable potentials for phase equilibria model for hydrogen sulfide[J]. The Journal of Physical Chemistry B, 2015, 119(23): 7041-7052.
26	Shah M S, Tsapatsis M, Siepmann J I. Identifying optimal zeolitic sorbents for sweetening of highly sour natural gas[J]. Angewandte Chemie International Edition, 2016, 55(20): 5938-5942.
27	Greathouse J A, Allendorf M D. Force field validation for molecular dynamics simulations of IRMOF-1 and other isoreticular zinc carboxylate coordination polymers[J]. The Journal of Physical Chemistry C, 2008, 112(15): 5795-5802.
28	Haldoupis E, Watanabe T, Nair S, et al. Quantifying large effects of framework flexibility on diffusion in MOFs: CH₄ and CO₂ in ZIF-8[J]. ChemPhysChem, 2012, 13(15): 3449-3452.
29	Pérez-Pellitero J, Amrouche H, Siperstein F, et al. Adsorption of CO₂, CH₄, and N₂ on zeolitic imidazolate frameworks: experiments and simulations[J]. Chemistry-A European Journal, 2010, 16(5): 1560-1571.
30	Rappe A K, Casewit C J, Colwell K S, et al. UFF, a full periodic table force field for molecular mechanics and molecular dynamics simulations[J]. Journal of the American Chemical Society, 1992, 114(25): 10024-10035.
31	Aprea P, Caputo D, Gargiulo N, et al. Modeling carbon dioxide adsorption on microporous substrates: comparison between Cu-BTC metal-organic framework and 13X zeolitic molecular sieve[J]. Journal of Chemical & Engineering Data, 2010, 55(9): 3655-3661.
32	Peng D Y, Robinson D B. A new two-constant equation of state[J]. Industrial & Engineering Chemistry Fundamentals, 1976, 15(1): 59-64.
33	Pedregosa F, Varoquaux G, Gramfort A, et al. Scikit-learn: machine learning in python[J]. Journal of Machine Learning Research, 2011, 12: 2825-2830.
34	Jung Y, Hu J H. A K-fold averaging cross-validation procedure[J]. Journal of Nonparametric Statistics, 2015, 27(2): 167-179.

编辑推荐 0

Metrics

阅读次数

全文

539

HTML			PDF

最新录用	在线预览	正式出版	最新录用	在线预览	正式出版
8	0	12	311	0	208

来源	本网站	其他网站

次数	75	464
比例	14%	86%

摘要

209

最新录用	在线预览	正式出版

71	0	138

来源	本网站	其他网站

次数	52	157
比例	25%	75%

[1]	李远华, 凌思棋, 封科军, 冯颖, 郭于菁, 谢世桓. 基于cMOFs的固定化脂肪酶微反应器的构筑及其扁桃酸催化应用[J]. 化工学报, 2025, 76(3): 1170-1179.
[2]	侯亚祺, 张玮, 张鸿, 高飞雨, 胡嘉华. 基于机器学习与粒子群算法的LBM多相流模型优化[J]. 化工学报, 2025, 76(3): 1120-1132.
[3]	冯海军, 章冰璇, 周健. 图神经网络模型预测和解释离子液体毒性的研究[J]. 化工学报, 2025, 76(1): 93-106.
[4]	张思文, 顾海明, 赵善辉. 纳米氧化铁对纤维素化学链气化的分子反应机理[J]. 化工学报, 2025, 76(1): 363-373.
[5]	王新月, 徐小虎, 张海洋, 尹春华. 维生素A醋酸酯/环糊精包合及性质研究[J]. 化工学报, 2024, 75(S1): 321-328.
[6]	代艳辉, 熊启钊, 房强, 杨东晓, 王毅, 陈杨, 李晋平, 李立博. 原位蒸汽辅助法用于一步制备多级孔Cu-BTC[J]. 化工学报, 2024, 75(9): 3329-3337.
[7]	吴哲明, 张碧云, 郑仁朝. 腈水解酶立体选择性改造及其合成布瓦西坦[J]. 化工学报, 2024, 75(7): 2633-2643.
[8]	秦晗淞, 李国梁, 闫昊, 冯翔, 刘熠斌, 陈小博, 杨朝合. 多级孔ZSM-5分子筛中油酸甲酯催化裂解吸附和扩散行为模拟研究[J]. 化工学报, 2024, 75(5): 1870-1881.
[9]	莫滨宇, 张雅馨, 刘国振, 刘公平, 金万勤. 面向一/二价离子分离的金属有机骨架膜研究进展[J]. 化工学报, 2024, 75(4): 1183-1197.
[10]	刘东飞, 张帆, 刘铮, 卢滇楠. 机器学习势及其在分子模拟中的应用综述[J]. 化工学报, 2024, 75(4): 1241-1255.
[11]	张政, 汪妩琼, 张雅静, 王康军, 吉远辉. 理论计算在药物制剂设计中的研究进展[J]. 化工学报, 2024, 75(4): 1429-1438.
[12]	周康, 王建新, 于海, 魏朝良, 范丰奇, 车昕昊, 张磊. 基于分子动力学模拟的矿物基础油泡沫破裂性能研究[J]. 化工学报, 2024, 75(4): 1668-1678.
[13]	文一如, 付佳, 刘大欢. 基于机器学习的MOFs材料研究进展：能源气体吸附分离[J]. 化工学报, 2024, 75(4): 1370-1381.
[14]	曾玉娇, 肖炘, 杨刚, 张意博, 郑光明, 李防, 汪凤玲. 基于机理与数据混合驱动的湿法磷酸生产过程代理建模与优化[J]. 化工学报, 2024, 75(3): 936-944.
[15]	陈宇翔, 刘传磊, 龚子君, 赵起越, 郭冠初, 姜豪, 孙辉, 沈本贤. 机器学习辅助乙硫醇高效吸收溶剂分子设计[J]. 化工学报, 2024, 75(3): 914-923.

机器学习辅助高通量筛选金属有机骨架用于富碳天然气中分离CO₂

Machine learning-assisted high-throughput screening approach for CO₂ separation from CO₂-rich natural gas using metal-organic frameworks

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 9

参考文献 34

相关文章 15

编辑推荐 0

Metrics

本文评价