High-throughput screening of zeolite materials for CO2/N2 selective adsorption separation by machine learning

doi:10.16085/j.issn.1000-6613.2022-0539

Abstract

Abstract:

At present, in the determination of gas adsorption performance and material design screening, the traditional experiments consume time and effort, so Grand Canonical Monte Carlo (GCMC) method in molecular mechanics has been widely used. However, the growing number of materials makes the GCMC method more and more computationally intensive, and a framework for screening adsorption materials based on machine learning (ML) method was proposed to solve this problem. The framework included three stages: the building of ML model, material selection of idealized PSA process model and validation using GCMC method. Firstly, artificial neural network models were established, and the structure descriptors "natural building unit (NBU)" of zeolite materials was proposed to predict the adsorption capacity under certain conditions. For CO₂ and N₂, two multi-layer feed-forward neural networks with different topological structures were built. Secondly, the ideal adsorbed solution theory (IAST) can predict the mixture adsorption isotherms of CO₂/N₂ (mole fractions is 0.14/0.86) from pure-component adsorption isotherms, and then 11 "best" zeolite materials were selected by some adsorbent evaluation metrics. Four zeolite materials (MON, ABW, NAB and VSV) were selected and calculate their adsorption data using GCMC method. The results proved that their adsorption capacity for N₂ was much lower than that for CO₂, so they had high adsorption selectivity for the two gases and can well separate CO₂ from binary mixtures.

Key words: machine learning, neural networks, CO₂ capture, zeolite, grand canonical Monte Carlo

摘要：

目前，针对气体吸附性能的测定及材料设计筛选，传统的实验法耗时耗力，因此分子力学方法中的巨正则蒙特卡洛（GCMC）方法已被广泛应用于该领域中，但日益增长的材料数目使得GCMC方法的计算量越来越高。为解决这一问题，本文提出了一种基于机器学习（ML）方法的吸附材料的筛选框架，包含ML模型的建立、理想化PSA工艺模型筛选材料及GCMC方法的验证三个阶段。首先，建立人工神经网络模型，提出了沸石材料的结构描述符“天然构造单元（NBU）”对特定条件下的气体吸附量进行预测。对于CO₂和N₂气体，分别构建了两个拓扑结构不同的多层前馈神经网络。其次，通过理想吸附溶液理论（IAST）将纯组分的吸附等温线转化为摩尔分数为0.14/0.86的CO₂/N₂二元混合物吸附等温线，并根据一系列吸附材料评估指标筛选出11种最佳沸石材料，并从中选出4种沸石（MON、ABW、NAB和VSV）计算其GCMC的吸附数据。结果表明，它们对N₂的吸附能力远低于CO₂，因此对两种气体的吸附选择性较高，能够很好地从二元混合物中分离CO₂。

关键词: 机器学习, 神经网络, 二氧化碳捕集, 沸石, 巨正则蒙特卡洛

CLC Number:

TQ021.8

WANG Lu, ZHANG Lei, DU Jian. High-throughput screening of zeolite materials for CO₂/N₂ selective adsorption separation by machine learning[J]. Chemical Industry and Engineering Progress, 2023, 42(1): 148-158.

王璐, 张磊, 都健. 机器学习高效筛选用于CO₂/N₂选择性吸附分离的沸石材料[J]. 化工进展, 2023, 42(1): 148-158.

Figures/Tables 19

评估指标	公式
摆动能力 $∆ Q C O 2$ /mol·kg^-1	$∆ Q C O 2$ = $Q C O 2 a d s$ $- Q C O 2$ ^des
吸附选择性 $S a d s, C O 2 / N 2 a d s$	$S a d s, C O 2 / N 2 a d s$ $= Q C O 2 a d s / Q / N 2 a d s y C O 2 / y N 2$
吸附材料选择参数 $S S P, C O 2 / N 2$	$S S P, C O 2 / N 2 = (S a d s a d s, C O 2 / N 2) 2 ∆ Q C O 2 (S a d s d e s, C O 2 / N 2) ∆ Q N 2$
吸附材料性能分数 $A P S C O 2 / N 2$ /mol·kg^-1	$A P S C O 2 / N 2 = S a d s, C O 2 / N 2 a d s$ $∆ Q C O 2$
再生能力R	R $=$ ( $∆ Q C O 2 / Q C O 2 a d s$ ) $× 100$ %

评估指标	公式
摆动能力 $∆ Q C O 2$ /mol·kg^-1	$∆ Q C O 2$ = $Q C O 2 a d s$ $- Q C O 2$ ^des
吸附选择性 $S a d s, C O 2 / N 2 a d s$	$S a d s, C O 2 / N 2 a d s$ $= Q C O 2 a d s / Q / N 2 a d s y C O 2 / y N 2$
吸附材料选择参数 $S S P, C O 2 / N 2$	$S S P, C O 2 / N 2 = (S a d s a d s, C O 2 / N 2) 2 ∆ Q C O 2 (S a d s d e s, C O 2 / N 2) ∆ Q N 2$
吸附材料性能分数 $A P S C O 2 / N 2$ /mol·kg^-1	$A P S C O 2 / N 2 = S a d s, C O 2 / N 2 a d s$ $∆ Q C O 2$
再生能力R	R $=$ ( $∆ Q C O 2 / Q C O 2 a d s$ ) $× 100$ %

参数	值
模拟类型	MonteCarlo
平衡数	50000
循环周期数	50000
力场类型	GarciaPerez2006
晶格单位	3×3×3
组分	CO₂	N₂
分子类型	TraPPE	TraPPE
摩尔分数	0.14	0.86
平移概率	0.5	0.5
旋转概率	0.5	0.5
重插概率	0.5	0.5
交换概率	1	1

数据集	CO₂	N₂
训练集	4704	4848
验证集	588	606
测试集	589	604
可用	5881	6058
空白值	257	80
统计	6138	6138

神经元个数	训练集		验证集		外部测试集		RMSE统计	R²统计
神经元个数	RMSE	R²	RMSE	R²	RMSE	R²	RMSE统计	R²统计
5	0.4227	0.9968	0.1995	0.9940	0.1601	0.9963	0.7823	0.9957
7	0.3028	0.9992	0.1281	0.9968	0.1252	0.9984	0.5562	0.9981
9	0.1216	0.9996	0.1256	0.9972	0.0834	0.9989	0.3305	0.9986
11	0.3681	0.9993	0.2662	0.9973	0.1219	0.9979	0.7562	0.9982
8	0.2360	0.9980	0.2051	0.9913	0.1171	0.9974	0.5581	0.9956
10	0.3638	0.9966	0.2632	0.9941	0.2611	0.9948	0.8881	0.9952

神经元个数	训练集		验证集		外部测试集		RMSE统计	R²统计
神经元个数	RMSE	R²	RMSE	R²	RMSE	R²	RMSE统计	R²统计
7	0.0130	0.9991	0.0571	0.9972	0.0316	0.9966	0.1018	0.9976
8	0.0791	0.9983	0.0507	0.9959	0.0565	0.9958	0.1863	0.9967
9	0.0326	0.9996	0.0219	0.9978	0.0270	0.9970	0.0814	0.9981
10	0.0522	0.9997	0.0332	0.9984	0.0429	0.9975	0.1283	0.9985
11	0.1516	0.9979	0.0703	0.9962	0.0514	0.9955	0.2733	0.9965
12	0.0317	0.9997	0.0372	0.9985	0.0271	0.9975	0.0960	0.9986
13	0.0176	0.9997	0.0228	0.9983	0.0284	0.9978	0.0689	0.9986
14	0.0289	0.9998	0.0214	0.9985	0.0270	0.9975	0.0773	0.9986
15	0.1264	0.9990	0.0710	0.9941	0.0680	0.9958	0.2654	0.9963

Ⅰ组	Ⅱ组	Ⅲ组	Ⅳ组
ABW、APC、LOV、MER、MON、NAB、PHI、PUN	AEI、AFY、RHO、SAV	AEN、AFG、AFI、AFN、ASV、ATT、BIK、ESV、HEU、MTW、MVY、NSI、RRO、RWR、UEI、YUG	GIS、SIV、WEI

沸石	$∆ Q C O 2$	R	$S a d s, C O 2 / N 2 a d s$	$S S P, C O 2 / N 2$	$A P S C O 2 / N 2$
MON	4.494542	79.02956	191.4463	5732.045	860.4634
ABW	3.279881	81.96941	133.9630	2715.545	439.3827
APC	3.089023	81.20606	131.6266	2570.451	406.5975
PHI	3.488620	81.73907	102.9267	2181.423	359.0721
PUN	4.489790	89.73103	78.15195	1797.634	350.8858
NAB	4.182718	94.98367	76.82922	1403.537	321.3550
MER	3.277813	83.61190	89.87160	1762.165	294.5823
LOV	3.614247	86.22495	63.68264	770.8370	230.1648
机器学习模型补充数据筛选得到的沸石：
VSV	4.588631	94.35575	83.24993	1971.854	382.0032
RSN	4.073608	91.19332	72.17017	1260.089	293.9930
KFI	3.447833	89.05406	62.91092	871.5884	216.9064

沸石	$∆ Q C O 2$	R	$S a d s, C O 2 / N 2 a d s$	$S S P, C O 2 / N 2$	$A P S C O 2 / N 2$
MON	4.494542	79.02956	191.4463	5732.045	860.4634
ABW	3.279881	81.96941	133.9630	2715.545	439.3827
APC	3.089023	81.20606	131.6266	2570.451	406.5975
PHI	3.488620	81.73907	102.9267	2181.423	359.0721
PUN	4.489790	89.73103	78.15195	1797.634	350.8858
NAB	4.182718	94.98367	76.82922	1403.537	321.3550
MER	3.277813	83.61190	89.87160	1762.165	294.5823
LOV	3.614247	86.22495	63.68264	770.8370	230.1648
机器学习模型补充数据筛选得到的沸石：
VSV	4.588631	94.35575	83.24993	1971.854	382.0032
RSN	4.073608	91.19332	72.17017	1260.089	293.9930
KFI	3.447833	89.05406	62.91092	871.5884	216.9064

References 40

1	CHAO Cong, DENG Yimin, DEWIL Raf, et al. Post-combustion carbon capture[J]. Renewable and Sustainable Energy Reviews, 2021, 138: 110490.
2	Seongbin GA, JANG Hong, LEE Jay H. New performance indicators for adsorbent evaluation derived from a reduced order model of an idealized PSA process for CO₂ capture[J]. Computers & Chemical Engineering, 2017, 102: 188-212.
3	WU Luogang, LIU Jiaqi, SHANG Hua, et al. Capture CO₂ from N₂ and CH₄ by zeolite L with different crystal morphology[J]. Microporous and Mesoporous Materials, 2021, 316: 110956.
4	顾玉明. 氮气在分子筛中吸附性质的理论研究[D]. 南京: 南京大学, 2019.
	GU Yuming. Theoretical study on adsorption properties of nitrogen in zeolites[D]. Nanjing: Nanjing University, 2019.
5	Youn Sang BAE, SNURR Randall Q. Development and evaluation of porous materials for carbon dioxide separation and capture[J]. Angewandte Chemie International Edition, 2011, 50(49): 11586-11596.
6	LEPERI Karson, CHUNG Yongchul G, YOU Fengqi, et al. Development of a general evaluation metric for rapid screening of adsorbent materials for postcombustion CO₂ capture[J]. ACS Sustainable Chemistry & Engineering, 2019, 7(13): 11529-11539.
7	TONG Minman, LAN Youshi, YANG Qingyuan, et al. High-throughput computational screening and design of nanoporous materials for methane storage and carbon dioxide capture[J]. Green Energy & Environment, 2018, 3(2): 107-119.
8	WIERSUM Andrew D, CHANG Jong San, SERRE Christian, et al. An adsorbent performance indicator as a first step evaluation of novel sorbents for gas separations: application to metal-organic frameworks[J]. Langmuir, 2013, 29(10): 3301-3309.
9	CHUNG Yongchul G, GÓMEZ-GUALDRÓN Diego A, LI Peng, et al. In silico discovery of metal-organic frameworks for precombustion CO₂ capture using a genetic algorithm[J]. Science Advances, 2016, 2(10): e1600909.
10	RAHMATI Mahmoud, MODARRESS Hamid. Grand canonical Monte Carlo simulation of isotherm for hydrogen adsorption on nanoporous siliceous zeolites at room temperature[J]. Applied Surface Science, 2009, 255(9): 4773-4778.
11	LIN Shiru, WANG Yekun, ZHAO Yinghe, et al. Machine-learning-assisted screening of pure-silica zeolites for effective removal of linear siloxanes and derivatives[J]. Journal of Materials Chemistry A, 2020, 8(6): 3228-3237.
12	毛海涛. 混合机器学习与机理建模的香精产品设计[D]. 大连: 大连理工大学, 2020.
	MAO Haitao. Fragrance product design with A hybrid machine learning and mechanistic modeling approach[D]. Dalian: Dalian University of Technology, 2020.
13	WANG Sunchong. Interdisciplinary computing in Java programming[M]//Springer Science+Business Media. New York: Kluwer Academic Publishers, 2003: 81.
14	LIU Yue, ZHAO Tianlu, JU Wangwei, et al. Materials discovery and design using machine learning[J]. Journal of Materiomics, 2017, 3(3): 159-177.
15	MALEKIAN Arash, CHITSAZ Nastaran. Concepts, procedures, and applications of artificial neural network models in streamflow forecasting[M]//Advances in Streamflow Forecasting. Amsterdam: Elsevier, 2021: 115-147.
16	ROSENBLATT F. The perceptron: a perceiving and recognizing automaton[R]. New York: Cornell Aeronautical Lab, 1957.
17	WERBOS Paul J. Applications of advances in nonlinear sensitivity analysis[M]//System Modeling and Optimization. Heidelberg: Springer-Verlag, 2005: 762-770.
18	RUMELHART David E, HINTON Geoffrey E, WILLIAMS Ronald J. Learning internal representations by error propagation[R]. Defense Technical Information Center, 1985.
19	BROOMHEAD D, LOWE D. Multivariable functional interpolation and adaptative networks[J]. Complex Systems, 1988, 2: 321-355.
20	ELMAN Jeffrey L. Finding structure in time[J]. Cognitive Science, 1990, 14(2): 179-211.
21	YANN LeCun, YOSHUA Bengio. Convolutional networks for images, speech, and time-series[J]. The Handbook of Brain Theory and Neural Networks, 1995: 255-258.
22	HINTON Geoffrey E, OSINDERO Simon, Yee Whye TEH. A fast learning algorithm for deep belief nets[J]. Neural Computation, 2006, 18(7): 1527-1554.
23	IYER Shachit, DEMIREL Salih E, M M Faruque HASAN. Combined natural gas separation and storage based on in silico material screening and process optimization[J]. Industrial & Engineering Chemistry Research, 2018, 57(49): 16727-16750.
24	SIMON Cory M, SMIT Berend, HARANCZYK Maciej. pyIAST: ideal adsorbed solution theory (IAST) Python package[J]. Computer Physics Communications, 2016, 200: 364-380.
25	ANUROVA Natalia A, BLATOV Vladislav A, ILYUSHIN Gregory D, et al. Natural tilings for zeolite-type frameworks[J]. The Journal of Physical Chemistry C, 2010, 114(22): 10160-10170.
26	Structure Commission of the International Zeolite Association (IZA-SC). Database of Zeolite Structures[DB/OL]. [2022-04-01]. .
27	GOODFELLOW I, BENGIO Y, COURVILLE A. Deep learning[M]. Cambridge: The MIT Press, 2016.
28	MCCULLOCH Warren S, PITTS Walter. A logical calculus of the ideas immanent in nervous activity[J]. The Bulletin of Mathematical Biophysics, 1943, 5(4): 115-133.
29	MYERS A L, PRAUSNITZ J M. Thermodynamics of mixed-gas adsorption[J]. AIChE Journal, 1965, 11(1): 121-127.
30	HAND D W, LOPER S, ARI M, et al. Prediction of multicomponent adsorption equilibria using ideal adsorbed solution theory[J]. Environmental Science & Technology, 1985, 19(11): 1037-1043.
31	BABARAO Ravichandar, HU Zhongqiao, JIANG Jianwen, et al. Storage and separation of CO₂ and CH₄ in silicalite, C₁₆₈schwarzite, and IRMOF-1: a comparative study from Monte Carlo simulation[J]. Langmuir: the ACS Journal of Surfaces and Colloids, 2007, 23(2): 659-666.
32	CESSFORD Naomi F, SEATON Nigel A, Tina DÜREN. Evaluation of ideal adsorbed solution theory as a tool for the design of metal-organic framework materials[J]. Industrial & Engineering Chemistry Research, 2012, 51(13): 4911-4921.
33	ROTHER J, FIEBACK T. Multicomponent adsorption measurements on activated carbon, zeolite molecular sieve and metal-organic framework[J]. Adsorption, 2013, 19(5): 1065-1074.
34	FRENKEL D, SMIT B. Understanding molecular simulation second edition from algorithms to applications computational science series vol 1[M]. Academic Press, Inc, 2001.
35	WILMER Christopher E, FARHA Omar K, Youn Sang BAE, et al. Structure-property relationships of porous materials for carbon dioxide separation and capture[J]. Energy & Environmental Science, 2012, 5(12): 9849-9856.
36	SUMER Zeynep, KESKIN Seda. Ranking of MOF adsorbents for CO₂ separations: a molecular simulation study[J]. Industrial & Engineering Chemistry Research, 2016, 55(39): 10404-10419.
37	DUBBELDAM David, CALERO Sofía, ELLIS Donald E, et al. RASPA: molecular simulation software for adsorption and diffusion in flexible nanoporous materials[J]. Molecular Simulation, 2016, 42(2): 81-101.
38	GARCÍA-PÉREZ E, PARRA J B, ANIA C O, et al. A computational study of CO₂, N₂, and CH₄ adsorption in zeolites[J]. Adsorption, 2007, 13(5/6): 469-476.
39	HASSE DAVID J, KULKARNI SUDHIR S, SANDERS EDGAR S JR, et al. Method of obtaining carbon dioxide from a carbon dioxide-containing gas mixture by means of a membrane and condensing: EP2512623[P]. 2016-11-02.
40	袁俊鹏, 刘秀英, 李晓东, 等. 沸石分子筛对CH₄/H₂的吸附与分离性能[J]. 物理学报, 2021, 70(15): 209-218.
	YUAN Junpeng, LIU Xiuying, LI Xiaodong, et al. Molecular simulation for adsorption and separation of CH₄/H₂ in zeolites[J]. Acta Physica Sinica, 2021, 70(15): 209-218.

[1]	DAI Huantao, CAO Lingyu, YOU Xinxiu, XU Haoliang, WANG Tao, XIANG Wei, ZHANG Xueyang. Adsorption properties of CO₂ on pomelo peel biochar impregnated by lignin [J]. Chemical Industry and Engineering Progress, 2023, 42(S1): 356-363.
[2]	CHEN Chongming, CHEN Qiu, GONG Yunqian, CHE Kai, YU Jinxing, SUN Nannan. Research progresses on zeolite-based CO₂ adsorbents [J]. Chemical Industry and Engineering Progress, 2023, 42(S1): 411-419.
[3]	XU Chenyang, DU Jian, ZHANG Lei. Chemical reaction evaluation based on graph network [J]. Chemical Industry and Engineering Progress, 2023, 42(S1): 205-212.
[4]	WANG Shengyan, DENG Shuai, ZHAO Ruikai. Research progress on carbon dioxide capture technology based on electric swing adsorption [J]. Chemical Industry and Engineering Progress, 2023, 42(S1): 233-245.
[5]	GE Yafen, SUN Yu, XIAO Peng, LIU Qi, LIU Bo, SUN Chengying, GONG Yanjun. Research progress of zeolite for VOCs removal [J]. Chemical Industry and Engineering Progress, 2023, 42(9): 4716-4730.
[6]	YANG Ying, HOU Haojie, HUANG Rui, CUI Yu, WANG Bing, LIU Jian, BAO Weiren, CHANG Liping, WANG Jiancheng, HAN Lina. Coal tar phenol-based carbon nanosphere prepared by Stöber method for adsorption of CO₂ [J]. Chemical Industry and Engineering Progress, 2023, 42(9): 5011-5018.
[7]	WANG Lanjiang, LIANG Yu, TANG Qiong, TANG Mingxing, LI Xuekuan, LIU Lei, DONG Jinxiang. Synthesis of highly dispersed Pt/HY catalyst by rapid pyrolysis of platinum precursors and its performance for deep naphthalene hydrogenation [J]. Chemical Industry and Engineering Progress, 2023, 42(8): 4159-4166.
[8]	PAN Yichang, ZHOU Rongfei, XING Weihong. Advanced microporous membranes for efficient separation of same-carbon-number hydrocarbon mixtures: State-of-the-art and challenges [J]. Chemical Industry and Engineering Progress, 2023, 42(8): 3926-3942.
[9]	BAI Yadi, DENG Shuai, ZHAO Ruikai, ZHAO Li, YANG Yingxia. Exploration on standardized test scheme and experimental performance of temperature swing adsorption carbon capture unit [J]. Chemical Industry and Engineering Progress, 2023, 42(7): 3834-3846.
[10]	WANG Shuaiqi, WANG Congxin, WANG Xuelin, TIAN Zhijian. Solvent-free rapid synthesis of ZSM-12 zeolite [J]. Chemical Industry and Engineering Progress, 2023, 42(7): 3561-3571.
[11]	WANG Jiuheng, RONG Nai, LIU Kaiwei, HAN Long, SHUI Taotao, WU Yan, MU Zhengyong, LIAO Xuqing, MENG Wenjia. Enhanced CO₂ capture performance and strength of cellulose-templated CaO-based pellets with steam reactivation [J]. Chemical Industry and Engineering Progress, 2023, 42(6): 3217-3225.
[12]	LU Shijian, ZHANG Yuanyuan, WU Wenhua, YANG Fei, LIU Ling, KANG Guojun, LI Qingfang, CHEN Hongfu, WANG Ning, WANG Feng, ZHANG Juanjuan. Health risk assessment of nitrosamine pollutant diffusion in a million ton CO₂ capture project [J]. Chemical Industry and Engineering Progress, 2023, 42(6): 3209-3216.
[13]	GU Shiya, DONG Yachao, LIU Linlin, ZHANG Lei, ZHUANG Yu, DU Jian. Design and optimization of pipeline system for carbon capture considering intermediate nodes [J]. Chemical Industry and Engineering Progress, 2023, 42(6): 2799-2808.
[14]	YIN Chengyang, HOU Ming, YANG Shuang, MAO Di, LIU Junyan. Research progress in transition metals modified Cu-SSZ-13 zeolite denitration catalysts [J]. Chemical Industry and Engineering Progress, 2023, 42(6): 2963-2974.
[15]	REN Zhongyuan, HE Jinlong, YUAN Qing. Research progress on intercrystalline defects control and remediation technologies for zeolite membranes [J]. Chemical Industry and Engineering Progress, 2023, 42(5): 2454-2463.

High-throughput screening of zeolite materials for CO₂/N₂ selective adsorption separation by machine learning

机器学习高效筛选用于CO₂/N₂选择性吸附分离的沸石材料

RichHTML

PDF (PC)

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 19

References 40

Related Articles 15

Recommended Articles

Metrics

Comments