神经网络全局势函数在多相催化中的应用

doi:10.16085/j.issn.1000-6613.2020-0430

摘要/Abstract

摘要：

当今的多相催化研究需要新的技术和方法从原子尺度上表征活性中心结构和反应中间体。本文作者课题组近期开发了理论模拟新技术来探索催化剂活性位点结构，即基于神经网络势函数的大规模原子模拟（LASP）软件中实现的全局神经网络势函数计算方法。本文介绍了该方法可以显著降低催化体系的计算代价，而维持与密度泛函理论同一级别的计算精度，从而解决多相催化中的许多复杂问题。本文对神经网络势函数方法的实现细节和目前已实现的应用场景进行了详细介绍。神经网络势函数可以用来预测材料晶体结构，理解高压氢化条件下TiO₂表面的结构演化和确定三元氧化物ZnCrO晶相中合成气制甲醇活性位点。最后文章分析了神经网络势函数的局限性和今后可能的三个研究方向，即材料性质预测、多元素体系神经网络势函数构造和化学反应拟合。

关键词: 机器学习, 神经网络, 密度泛函理论, 随机势能面行走, LASP软件

Abstract:

Heterogeneous catalysis demands new techniques and methods to characterize the structures of catalytic active centre and reaction intermediates from atomic scale. The recently developed global neural network potential (NNP) to explore catalyst structure is introduced, which has been implemented in the software of Large-scale Atomic Simulation with neural network Potential (LASP). The technical details of the NNP and its recent applications in heterogeneous catalysis are discussed. NNP can significantly reduce the calculation cost with comparable calculation accuracy to the ab-initio methods, with which many complex problems in heterogeneous catalysis can be solved. The success of NNP function in predicting the crystalline phase of materials, understanding the surface structure evolution of TiO₂under high pressure of hydrogen and determining the active sites of ternary oxide are illustrated as examples. Finally, the limitations of NNP are discussed and its future research directions are pointed out as the estimation of material properties, the construction of NNP for multi-element system and the fitting of chemical reaction.

Key words: machine learning, neural networks, density functional theory, stochastic surface walking, LASP software

中图分类号:

马思聪, 刘智攀. 神经网络全局势函数在多相催化中的应用[J]. 化工进展, 2020, 39(9): 3433-3443.

Sicong MA, Zhipan LIU. Global neural network potential applications in heterogeneous catalysis[J]. Chemical Industry and Engineering Progress, 2020, 39(9): 3433-3443.

参考文献

1	MA Sicong, SHANG Cheng, LIU Zhipan. Heterogeneous catalysis from structure to activity via SSW-NN method[J]. Journal of Chemical Physics, 2019, 151(5): 050901.
2	WALES David J, DOYE Jonathan P K. Global optimization by basin-hopping and the lowest energy structures of Lennard-Jones clusters containing up to 110 atoms[J]. The Journal of Physical Chemistry A, 1997, 101(28): 5111-5116.
3	OAKLEY Mark T, JOHNSTON Roy L, WALES David J. Symmetrisation schemes for global optimisation of atomic clusters[J]. Physical Chemistry Chemical Physics, 2013, 15(11): 3965-3976.
4	MARTO? K R, LAIO Alessandro, PARRINELLO Michele. Predicting crystal structures: the Parrinello-Rahman method revisited[J]. Physical Review Letters, 2003, 90(7): 075503.
5	Roman MARTO? K, DONADIO Davide, OGANOV Artem R, et al. Crystal structure transformations in SiO₂ from classical and ab initio metadynamics[J]. Nature Materials, 2006, 5(8): 623.
6	GLASS Colin W, OGANOV Artem R, HANSEN Nikolaus. USPEX—evolutionary crystal structure prediction[J]. Computer Physics Communications, 2006, 175(11/12): 713-720.
7	OGANOV Artem R, MA Yanming, GLASS Colin W, et al. Evolutionary crystal structure prediction: overview of the USPEX method and some of its applications[J]. Psi-k Newsletter, 2007, 84: 142-171.
8	WANG Yanchao, LV Jian, ZHU Li, et al. Crystal structure prediction via particle-swarm optimization[J]. Physical Review B, 2010, 82(9): 094116.
9	WANG Yanchao, LV Jian, ZHU Li, et al. CALYPSO: a method for crystal structure prediction[J]. Computer Physics Communications, 2012, 183(10): 2063-2070.
10	KIRKPATRICK Scott, Daniel GELATT C, VECCHI Mario P. Optimization by simulated annealing[J]. Science, 1983, 220(4598): 671-680.
11	Vladimír ?ERN. Thermodynamical approach to the traveling salesman problem: an efficient simulation algorithm[J]. Journal of Optimization Theory and Applications, 1985, 45(1): 41-51.
12	GOEDECKER Stefan, HELLMANN Waldemar, LENOSKY Thomas. Global minimum determination of the Born-Oppenheimer surface within density functional theory[J]. Physical Review Letters, 2005, 95(5): 055501.
13	SHANG Cheng, ZHANG Xiaojie, LIU Zhipan. Stochastic surface walking method for crystal structure and phase transition pathway prediction[J]. Physical Chemistry Chemical Physics, 2014, 16(33): 17845-17856.
14	ZHANG Xiaojie, LIU Zhipan. Variable-cell double-ended surface walking method for fast transition state location of solid phase transitions[J]. Journal of Chemical Theory and Computation, 2015, 11(10): 4885-4894.
15	ZHANG Xiaojie, SHANG Cheng, LIU Zhipan. From atoms to fullerene: stochastic surface walking solution for automated structure prediction of complex material[J]. Journal of Chemical Theory and Computation, 2013, 9(7): 3252-3260.
16	GUAN Shuhui, ZHANG Xiaojie, LIU Zhipan. Energy landscape of zirconia phase transitions[J]. Journal of the American Chemical Society, 2015, 137(25): 8010-8013.
17	ZHU Shengcai, XIE Songhai, LIU Zhipan. Design and observation of biphase TiO₂ crystal with perfect junction[J]. The Journal of Physical Chemistry Letters, 2014, 5(18): 3162-3168.
18	KOHONEN Teuvo. An introduction to neural computing[J]. Neural Networks, 1988, 1(1): 3-16.
19	MCCULLOCH Warren S, PITTS Walter. A logical calculus of the ideas immanent in nervous activity[J]. Bulletin of Mathematical Biology, 1943, 5(4): 115-133.
20	SUMPTER Bobby G，NOID Donald W. Potential energy surfaces for macromolecules. a neural network technique[J]. Chemical Physics Letters, 1992, 192(5/6): 455-462.
21	FISCHER Thomas H, PETERSEN Wesley P, LüTHI Hans Peter. A new optimization technique for artificial neural networks applied to prediction of force constants of large molecules[J]. Journal of Computational Chemistry, 1995, 16(8): 923-936.
22	RAFF L M, MALSHE M, HAGAN M, et al. Ab initio potential-energy surfaces for complex, multichannel systems using modified novelty sampling and feedforward neural networks[J]. Journal of Chemical Physics, 2005, 122(8): 084104.
23	MANZHOS Sergei, WANG Xiaogang, DAWES Richard, et al. A nested molecule-independent neural network approach for high-quality potential fits[J]. The Journal of Physical Chemistry A, 2006, 110(16): 5295-5304.
24	J?rg BEHLER, PARRINELLO Michele. Generalized neural-network representation of high-dimensional potential-energy surfaces[J]. Physical Review Letters, 2007, 98(14): 146401.
25	J?rg BEHLER. Atom-centered symmetry functions for constructing high-dimensional neural network potentials[J]. Journal of Chemical Physics, 2011, 134(7): 074106.
26	J?rg BEHLER. Neural network potential-energy surfaces in chemistry: a tool for large-scale simulations[J]. Physical Chemistry Chemical Physics, 2011, 13(40): 17930-17955.
27	J?rg BEHLER. Perspective: machine learning potentials for atomistic simulations[J]. Journal of Chemical Physics, 2016, 145(17): 170901.
28	J?rg BEHLER. First principles neural network potentials for reactive simulations of large molecular and condensed systems[J]. Angewandte Chemie: International Edition, 2017, 56(42): 12828-12840.
29	HUANG Sida, SHANG Cheng, ZHANG Xiaojie, et al. Material discovery by combining stochastic surface walking global optimization with a neural network[J]. Chemical Science, 2017, 8(9): 6327-6337.
30	HUANG Sida, SHANG Cheng, KANG Peilin, et al. Atomic structure of boron resolved using machine learning and global sampling[J]. Chemical Science, 2018, 9(46): 8644-8655.
31	BOTTOU Léon. Stochastic gradient learning in neural networks[J]. Proceedings of Neuro-N?mes, 1991, 91(8): 12.
32	FLETCHER Roger, POWELL Michael J D. A rapidly convergent descent method for minimization[J]. The Computer Journal, 1963, 6(2): 163-168.
33	POLAK Elijah. Computational methods in optimization: a unified approach[M]. Academic Press, 1971: 22.
34	HAGAN Martin T, MENHAJ Mohammad B. Training feedforward networks with the Marquardt algorithm[J]. IEEE Transactions on Neural Networks and Learning Systems, 1994, 5(6): 989-993.
35	ZHANG Hengzhong, BANFIELD Jillian F. Structural characteristics and mechanical and thermodynamic properties of nanocrystalline TiO₂[J]. Chemical Reviews, 2014, 114(19): 9613-9644.
36	WANG Lianzhou, SASAKI Takayoshi. Titanium oxide nanosheets: graphene analogues with versatile functionalities[J]. Chemical Reviews, 2014, 114(19): 9455-9486.
37	SASAKI Takayoshi, WATANABE Mamoru, HASHIZUME Hideo, et al. Macromolecule-like aspects for a colloidal suspension of an exfoliated titanate. Pairwise association of nanosheets and dynamic reassembling process initiated from it[J]. Journal of the American Chemical Society, 1996, 118(35): 8329-8335.
38	AKIMOTO J, GOTOH Y, OOSAWA Y, et al. Topotactic oxidation of ramsdellite-type Li_0.5TiO₂, a new polymorph of titanium dioxide: TiO₂ (R)[J]. Journal of Solid State Chemistry, 1994, 113(1): 27-36.
39	LATROCHE M, BROHAN L, MARCHAND R, et al. New hollandite oxides: TiO₂ (H) and K_0.06TiO₂[J]. Journal of Solid State Chemistry, 1989, 81(1): 78-82.
40	SAKAO Mitsumasa, KIJIMA Norihito, AKIMOTO Junji, et al. Synthesis, crystal structure, and electrochemical properties of hollandite-type K 0.008 TiO₂[J]. Solid State Ionics, 2012, 225: 502-505.
41	PEREZ-FLORES J C, BAEHTZ C, KUHN A, et al. Hollandite-type TiO₂: a new negative electrode material for sodium-ion batteries[J]. Journal of Materials Chemistry A, 2014, 2(6): 1825-1833.
42	ZHU Guannan, WANG Yonggang, XIA Yongyao. Ti-based compounds as anode materials for Li-ion batteries[J]. Energy & Environmental Science, 2012, 5(5): 6652-6667.
43	PUHLF R P, VOIGT A, WEBER R, et al. Microporous TiO₂ membranes with a cut off ＜500Da[J]. Journal of Membrane Science, 2000, 174(1): 123-133.
44	Jelena SEKULI?, ELSHOF Johan E TEN, BLANK D H A. A microporous titania membrane for nanofiltration and pervaporation[J]. Advanced Materials, 2004, 16(17): 1546-1550.
45	XU Qunyin, ANDERSON Marc A. sol-gel route to synthesis of microporous ceramic membranes: preparation and characterization of microporous TiO₂ and ZrO₂xerogels[J]. Journal of the American Ceramic Society, 1994, 77(7): 1939-1945.
46	CHEN Xiaobo, LIU Lei, YU PETER Y, et al. Increasing solar absorption for photocatalysis with black hydrogenated titanium dioxide nanocrystals[J]. Science, 2011, 331(6018): 746-750.
47	CHEN Xiaobo, SHEN Shaohua, GUO Liejin, et al. Semiconductor-based photocatalytic hydrogen generation[J]. Chemical Reviews, 2010, 110(11): 6503-6570.
48	HU Yunhang. A highly efficient photocatalyst—hydrogenated black TiO₂ for the photocatalytic splitting of water[J]. Angewandte Chemie: International Edition, 2012, 51(50): 12410-12412.
49	NALDONI Alberto, ALLIETA Mattia, SANTANGELO Saveria, et al. Effect of nature and location of defects on bandgap narrowing in black TiO₂ nanoparticles[J]. Journal of the American Chemical Society, 2012, 134(18): 7600-7603.
50	GUO Yao, CHEN Shunwei, YU Yaoguang, et al. Hydrogen-location-sensitive modulation of the redox reactivity for oxygen-deficient TiO₂[J]. Journal of American Chemical Society, 2019, 141(21):8407-8411.
51	MOLSTAD Melvin Carl, DODGE Barnett F. Zinc oxide-chromium oxide catalysts for methanol synthesis[J]. Industrial & Engineering Chemistry, 1935, 27(2): 134-140.
52	KUNG Harold H. Methanol synthesis[J]. Catalysis Reviews: Science and Engineering, 1980, 22(2): 235-259.
53	JIAO Feng, LI Jinjing, PAN Xiulian, et al. Selective conversion of syngas to light olefins[J]. Science, 2016, 351(6277): 1065-1068.
54	WAUGH K C. Methanol synthesis[J]. Catalysis Today, 1992, 15(1): 51-75.
55	DUMITRU Raluca, MANEA Florica, Cornelia P?CURARIU, et al. Synthesis, characterization of nanosized ZnCr₂O₄ and its photocatalytic performance in the degradation of humic acid from drinking water[J]. Catalysts, 2018, 8(5): 210.
56	PIERO Gastone DEL, TRIFIRO Ferruccio, VACCARI Angelo. Non-stoicheiometric Zn-Cr spinel as active phase in the catalytic synthesis of methanol[J]. Journal of the Chemical Society, Chemical Communications, 1984(10): 656-658.
57	BERTOLDI Massimo, FUBINI Bice, GIAMELLO Elio, et al. Structure and reactivity of zinc–chromium mixed oxides. Part 1.—The role of non-stoichiometry on bulk and surface properties[J]. Journal of the Chemical Society, Faraday Transactions, 1988, 84(5): 1405-1421.
58	ERRANI E, TRIFIRO F, VACCARI A, et al. Structure and reactivity of Zn-Cr mixed oxides. Role of non-stoichiometry in the catalytic synthesis of methanol[J]. Catalysis Letters, 1989, 3(1): 65-72.
59	SONG Huiqing, LAUDENSCHLEGER Daniel, CAREY John J, et al. Spinel-structured ZnCr₂O₄ with excess Zn is the active ZnO/Cr₂O₃ catalyst for high-temperature methanol synthesis[J]. ACS Catalysis, 2017, 7(11): 7610-7622.
60	MA Sicong, HUANG Sida, LIU Zhipan. Dynamic coordination of cations and catalytic selectivity on zinc-chromium oxide alloys during syngas conversion[J]. Nature Catalysis, 2019, 2(8): 671-677.

[1]	徐晨阳, 都健, 张磊. 基于图神经网络的化学反应优劣评价[J]. 化工进展, 2023, 42(S1): 205-212.
[2]	俞俊楠, 俞建峰, 程洋, 齐一搏, 化春键, 蒋毅. 基于深度学习的变宽度浓度梯度芯片性能预测[J]. 化工进展, 2023, 42(7): 3383-3393.
[3]	张建伟, 许蕊, 张忠闯, 董鑫, 冯颖. 基于卷积神经网络的撞击流反应器浓度场混合特性[J]. 化工进展, 2023, 42(2): 658-668.
[4]	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.
[5]	潘艳秋, 李鹏飞, 高石磊, 俞路. 基于MIC筛选规则和BP神经网络的变换装置建模及产品预测[J]. 化工进展, 2022, 41(S1): 36-43.
[6]	孙德贇, 胡艳宏, 刘鹏, 唐茂, 胡泽, 柳召刚, 吴锦绣. 不同铈盐体系（硝酸盐、硫酸盐、氯化盐）中CTAB与Ce³⁺的相互作用机理[J]. 化工进展, 2022, 41(6): 3212-3220.
[7]	李国友, 张新魁, 才士文, 贾曜宇, 宁泽. 基于改进KFDA与DE优化SOM的故障诊断模型及其化工过程诊断[J]. 化工进展, 2022, 41(4): 1793-1801.
[8]	李欣铜, 陈志冰, 魏志强, 李苏桐, 陈旭, 宋凯. 具有输入数据注意力机制的卷积神经网络用于氟化工产品质量预测[J]. 化工进展, 2022, 41(2): 593-600.
[9]	李帅, 刘明言, 马永丽. 基于BP人工神经网络预测地热井中流体的结垢位置[J]. 化工进展, 2022, 41(11): 5761-5770.
[10]	王英, 冉进业, 张今, 杨鑫, 张浩. 基于深度时间序列特征融合的西安市2015—2020年供暖季雾霾重污染过程预警[J]. 化工进展, 2022, 41(10): 5685-5694.
[11]	练成, 程锦, 黄盼, 陶浩兰, 杨洁, 刘洪来. 新能源化工热力学[J]. 化工进展, 2021, 40(9): 4711-4733.
[12]	李芳国, 张贝克, 高东. HAZOP知识图谱构建方法[J]. 化工进展, 2021, 40(8): 4666-4677.
[13]	范峥, 田润芝, 林亮, 韩彦忠, 郭阳, 豆龙龙, 景根辉, TYOOR Agi Damian. 利用基于PSO算法的径向基人工神经网络优化重催干气脱硫[J]. 化工进展, 2021, 40(6): 3107-3118.
[14]	张梦轩, 刘洪辰, 王敏, 蓝兴英, 石孝刚, 高金森. 化工过程的智能混合建模方法及应用[J]. 化工进展, 2021, 40(4): 1765-1776.
[15]	朱礼涛, 欧阳博, 张希宝, 罗正鸿. 机器学习在多相反应器中的应用进展[J]. 化工进展, 2021, 40(4): 1699-1714.