PREDICTION ON PHOTOELECTRIC CONVERSION EFFICIENCY OF ORGANIC PHOTOVOLTAIC MATERIALS USING END-TO-END DEEP LEARNING
-
摘要: 碳中和背景下,亟需开发高效清洁的新型能源,以减少对化石能源的依赖。有机光伏材料作为一种可将太阳能或其他光能直接转化为电能的材料,日益成为一种具有重大应用前景的低碳能源材料。在探索新的高性能有机光伏材料的过程中,机器学习虽然能够提高材料设计效率,但其预测能力极大受制于描述符的开发和选取。利用循环神经网络、卷积神经网络、图神经网络等算法,构建端到端的深度学习模型预测有机光伏材料光电转化效率,所建模型可直接从SMILES符号、分子图像、分子图网络中提取化合物结构信息,而无须人为开发和选取描述符。所得模型不仅能够较为准确地预测有机光伏材料的光电转化效率(其中最优模型五折交叉验证结果和测试集预测结果决定系数均>0.73),而且能够识别影响转化效率的关键结构特征。该研究结果可为新型环境功能材料设计提供理论参考。Abstract: Under the context of carbon neutrality,there is an urgent need to develop high-efficiency and clean energy to reduce the dependence on petrochemical energy.As a material that can directly convert solar energy or other light energy into electrical energy,organic photovoltaic materials have become an increasingly promising low-carbon energy material with great application prospects.In the process of exploring high-performance organic photovoltaic materials,although machine learning can improve the efficiency of material design,its predictive ability is still greatly restricted by the development and selection of descriptors.In the present study,the authors applied recurrent neural network,convolutional neural network,and graph neural network to build end-to-end deep learning models to predict the photoelectric conversion efficiency,and the constructed models could directly extract chemical features from SMILES symbols,molecular images,and molecular graph networks without the need of descriptors calculation and selection.The resulted models could not only accurately predict the photoelectric conversion efficiency of organic photovoltaic materials (the optimal model obtained R2>0.73 for both 5-fold cross validation and external validation),but also identify the key structural features that affect the conversion efficiency.The results could provide theoretical guidance for the design of new environmental functional materials.
-
[1] LIU Q S, JIANG Y F, JIN K, et al. 18% efficiency organic solar cells[J]. Science Bulletin, 2020, 65(4):272-275. [2] LI C, ZHOU J D, SONG J L, et al. Non-fullerene acceptors with branched side chains and improved molecular packing to exceed 18% efficiency in organic solar cells[J]. Nature Energy, 2021, 6(6):605-613. [3] CAI Y, LI Y, WANG R, et al. A well-mixed phase formed by two compatible non-fullerene acceptors enables ternary organic solar cells with efficiency over 18.6%[J]. Advanced Materials, 2021, 33(33):e2101733. [4] YAN J C, YAN X L, HU S, et al. Comprehensive interrogation on acetylcholinesterase inhibition by ionic liquids using machine learning and molecular modeling[J]. Environmental Science&Technology, 2021, 55(21):14720-14731. [5] YAN X L, SEDYKH A, WANG W Y, et al. Construction of a web-based nanomaterial database by big data curation and modeling friendly nanostructure annotations[J]. Nature Communications, 2020, 11(1):2519. [6] YAN X L, ZHANG J, RUSSO D P, et al. Prediction of nano-bio interactions through convolutional neural network analysis of nanostructure images[J]. ACS Sustainable Chemistry&Engineering, 2020, 8(51):19096-19104. [7] NAGASAWA S, AL-NAAMANI E, SAEKI A. Computer-aided screening of conjugated polymers for organic solar cell:classification by random forest[J]. The Journal of Physical Chemistry Letters, 2018, 9(10):2639-2646. [8] SAHU H, RAO W N, TROISI A, et al. Toward predicting efficiency of organic solar cells via machine learning and improved descriptors[J]. Advanced Energy Materials, 2018, 8(24):1801032. [9] SUN W B, ZHENG Y J, YANG K, et al. Machine learning-assisted molecular design and efficiency prediction for high-performance organic photovoltaic materials[J]. Science Advances, 2019,5(11):eaay4275. [10] HACHMANN J, OLIVARES-AMAYA R, ATAHAN-EVRENK S, et al. The harvard clean energy project:large-scale computational screening and design of organic photovoltaics on the world community grid[J]. Journal of Physical Chemistry Letters, 2011, 2(17):2241-2251. [11] SCHARBER M C, MVHLBACHER D, KOPPE M, et al. Design rules for donors in bulk-heterojunction solar cells-towards 10% energy-conversion efficiency[J]. Advanced Materials, 2006, 18(6):789-794. [12] SUN W B, LI M, LI Y, et al. The use of deep learning to fast evaluate organic photovoltaic materials[J]. Advanced Theory and Simulations, 2019, 2(1):1800116. [13] 张珂,冯晓晗,郭玉荣,等.图像分类的深度卷积神经网络模型综述[J].中国图象图形学报, 2021, 26(10):2305-2325. [14] SIMONYAN K, ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[J]. Arxiv,2014,abs/1409.1556. [15] HUANG G, LIU Z, VAN DER MAATEN L, et al. Densely connected convolutional networks[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017:2261-2269. [16] 杨丽,吴雨茜,王俊丽,等.循环神经网络研究综述[J].计算机应用, 2018, 38(增刊2):1-6,26. [17] CHUNG J, GULCEHRE C, CHO K H, et al. Empirical evaluation of gated recurrent neural networks on sequence modeling[J]. Arxiv, 2014,sbs/1412.3555.. [18] HOCHREITER S, SCHMIDHUBER J. Long short-term memory[J]. Neural Computation, 1997, 9(8):1735-1780. [19] 毕常遥,袁晓彤.基于Adam局部优化的分布式近似牛顿深度学习模型训练[J].计算机应用与软件, 2021, 38(10):278-283. [20] 王健宗,孔令炜,黄章成,等.图神经网络综述[J].计算机工程, 2021, 47(4):12. [21] XIONG Z P, WANG D Y, LIU X H, et al. Pushing the boundaries of molecular representation for drug discovery with the graph attention mechanism[J]. Journal of Medicinal Chemistry, 2020, 63(16):8749-8760. [22] DAVID W. SMILES, a chemical language and information system. 1. Introduction to methodology and encoding rules[J]. Journal of Chemical Information&Computer Sciences, 1988, 28(1):31-36. [23] BAGHER A M. Comparison of organic solar cells and inorganic solar cells[J]. International Journal of Sustainable and Green Energy, 2014, 3(3):53-58.
点击查看大图
计量
- 文章访问数: 303
- HTML全文浏览量: 60
- PDF下载量: 6
- 被引次数: 0