1. NCCU Academic Hub
  2. 學術產出
  3. College of Science
  4. Theses
  5. Item

學術產出-Theses

Article View/Open

Publication Export

Google ScholarTM

政大圖書館

Citation Infomation

  • No doi shows Citation Infomation
題名 機器學習預測材料特性
Predicting material properties with machine learning
作者 陳建豪
Chen, Jian-Hao
貢獻者 許琇娟
Hsu, Hsiu-Chuan
陳建豪
Chen, Jian-Hao
關鍵詞 深度學習
二維材料
材料信息學
Deep learning
Two-dimensional material
Materials Informatics
日期 2023
上傳時間 1-Sep-2023 16:28:41 (UTC+8)
摘要 二維材料具有非常特別的物理性質，在電子、光學元件及航空航太等，都具有極高的應用價值，以石墨烯為例，具有高導電性、高導熱性及高機械強度等等，除石墨烯外，包括像單層二硫化鉬(MoS2)、二硒化鉬(MoSe2)，他們因為皆為直接能隙半導體，可被製作成透明的發光二極體(light-emitting diode, LED)，且⼆維結構⾃然有利於各種類型器件的性能，如減小尺寸以防止短通道效應以及提高可穿戴設備的靈活性。
一般材料之探勘到實際應用的耗時較長，從實驗室發現一種新型材料，再到研發、驗證、直至最後的業務場景落地需要較長久的時間，而傳統材料探勘與研究通常較仰賴專業人員的研究背景與實務經驗，來提出可行的候選材料與方法，而過程相當耗時耗力，且需投入大量經費。
有鑑於此，本文基於機器學習對二維材料的性質進行預測，希望能減少二維材料在研發過程中所需的時間，並能夠輔助研究人員在前期尚未開始實際研究測量數值時，有個大致的參考數值。
本實驗所使用的深度學習算法基於以下三種，一種為多層感知機(Multilayer perceptron, MLP)，以及另外兩種分別為圖卷積神經網路(Graph Convolutional Network, CGCNN) 和殘差網路(Residual Network,ResNet)，多層感知機為一種前向傳遞類神經網路，透過每層神經元之間的資料傳遞來學習到相關的資訊，並且利用「倒傳遞」的技術達到學習(model learning)的監督式學習。另一項圖卷積神經網路其本質目的是用來提取拓撲圖的空間特徵，透過將數據轉化為拓撲圖使網路能從中提取出特徵並學習到相關資訊，可用於非歐幾里得空間的圖形，並可應用於數據預測及類別分類任務。最後一項殘差網路則為卷積神經網路的一個變體，透過將輸入與輸出特徵做結合，從而改善模型過擬合的狀況。
本實驗透過引入深度學習技術來預測材料的剝離能、晶格常數以及晶格結構，並利用不同模型比較各自差異，找出較好的結果以及訓練方法，本實驗所獲得的最好結果分別是，透過CGCNN 訓練剝離能，其誤差為0.0508eV/atom，而晶格常數之三邊長以MLP 在邊長a、b、c的誤差最小分別為0.5994Å，1.0664Å 以及1.2785Å，最後晶格結構的分類準確度以利用MLP 作為訓練模型的效果最好，其準確度為65%。
除此之外，現今電動車的高度發展，電動車電池（electric-vehicle battery）的需求提高，且續航以及安全要求也相對提升，使電動車電池之研究頗為重要，有鑑於此，本實驗也建立一自製電池材料數據庫，希望能提供給開發人員進行開發使用。
Two-dimensional materials possess extraordinary physical properties and
hold immense value for applications in electronics, optical components, aerospace, and other fields. Taking graphene as an example, it exhibits high electric conductivity, thermal conductivity, and mechanical strength. Apart from graphene, materials like MoS2 and MoSe2 belong to direct bandgap semiconductors and can be used to fabricate transparent light-emitting diodes (LEDs). Moreover, the two-dimensional structure inherently improves the performance of various device types by reducing size to mitigate short-channel effects and enhancing the flexibility of wearable devices.
The development from material exploration to practical application is often time-consuming. It involves discovering a new material in the laboratory, followed by immense research, verification, and final implementation in business scenarios. Conventional material exploration research heavily relies on the expertise and practical experience of specialized professionals to identify viable candidate materials. This process is both time-consuming and resource-intensive, requiring a significant investment of funds.
Recently, the data-driven techniques have been adopted to predict the material properties, owing to the mature machine learning algorithms and growing number of databases. This new paradigm reduces the time required for research and development in material science. Thus, this thesis applies this approach to predict properties of two-dimensional materials with the aim to accelerate the material exploration.
The deep learning models used in this thesis are based on three models:
Multilayer Perceptron (MLP), Crystal Graph Convolutional Neural Networks (CGCNN), and Residual Network (ResNet). These models were employed to predict material properties such as exfoliation energy, lattice constants, and crystal system. By introducing deep learning techniques, a comparison between different models was conducted to determine the best results and training methods.
The best results obtained are as follows. For exfoliation energy prediction,
CGCNN achieved an error of 0.0508 eV/atom. Regarding lattice constants,
MLP had the smallest error for lattice constant a with 0.5994 Å.Lastly, for
crystal system classification, MLP performed the best, achieving an accuracy of 65%.
Moreover, with the growing development of electric vehicles, the demand
for electric-vehicle batteries has increased, making the research on electric vehicle batteries very important. This thesis establishes a battery material database that contains several crystal and energetic properties of battery materials. The database could be utilized for the study with data-driven approaches in the future.
參考文獻 [1] Q. Tong, P. Gao, H. Liu, Y. Xie, J. Lv, Y. Wang, and J. Zhao, “Combining machine learning potential and structure prediction for accelerated materials design and discovery,” The Journal of Physical Chemistry Letters, vol. 11, no. 20, pp. 8710–8720, 2020.
[2] C. Gao, X. Min, M. Fang, T. Tao, X. Zheng, Y. Liu, X. Wu, and Z. Huang, “Innovative materials science via machine learning,” Advanced Functional Materials, vol. 32, no. 1, p. 2108044, 2022.
[3] K. Guo, Z. Yang, C.-H. Yu, and M. J. Buehler, “Artificial intelligence and machine learning in design of mechanical materials,” Materials Horizons, vol. 8, no. 4, pp. 1153–1172, 2021.
[4] G. H. Jeong, S. P. Sasikala, T. Yun, G. Y. Lee, W. J. Lee, and S. O. Kim, “Nanoscale assembly of 2d materials for energy and environmental applications,” Advanced Materials, vol. 32, no. 35, p. 1907006, 2020.
[5] B. Radisavljevic, A. Radenovic, J. Brivio, V. Giacometti, and A. Kis, “Single-layer mos2 transistors,” Nature nanotechnology, vol. 6, no. 3, pp. 147–150, 2011.
[6] D. Berman, B. Narayanan, M. J. Cherukara, S. K. Sankaranarayanan, A. Erdemir, A. Zinovev, and A. V. Sumant, “Operando tribochemical formation of onion-like-carbon leads to macroscale superlubricity,” Nature communications, vol. 9, no. 1, p. 1164, 2018.
[7] W. Wu, J. Wang, P. Ercius, N. C. Wright, D. M. Leppert-Simenauer, R. A. Burke, M. Dubey, A. M. Dogare, and M. T. Pettes, “Giant mechano-optoelectronic effect in an atomically thin semiconductor,” Nano letters, vol. 18, no. 4, pp. 2351–2357, 2018.
[8] H. G. Kim and H.-B.-R. Lee, “Atomic layer deposition on 2d materials,” Chemistry of Materials, vol. 29, no. 9, pp. 3809–3826, 2017.
[9] G. Carleo, I. Cirac, K. Cranmer, L. Daudet, M. Schuld, N. Tishby, L. Vogt-Maranto, and L. Zdeborová, “Machine learning and the physical sciences,” Reviews of Modern Physics, vol. 91, no. 4, p. 045002, 2019.
[10] E. Dolotov and N. Zolotykh, “Evolutionary algorithms for constructing an ensemble of decision trees,” in Analysis of Images, Social Networks and Texts: 8th International Conference, AIST 2019, Kazan, Russia, July 17–19, 2019, Revised Selected Papers 8, pp. 9–15, Springer, 2020.
[11] G. Biau and E. Scornet, “A random forest guided tour,” Test, vol. 25, pp. 197–227, 2016.
[12] Y. LeCun, Y. Bengio, and G. Hinton, “Deep learning,” nature, vol. 521, no. 7553, pp. 436–444, 2015.
[13] S. B. Kotsiantis, I. Zaharakis, P. Pintelas, et al., “Supervised machine learning: A review of classification techniques,” Emerging artificial intelligence applications in computer engineering, vol. 160, no. 1, pp. 3–24, 2007.
[14] H. Schulz and S. Behnke, “Deep learning,” KI-Künstliche Intelligenz, vol. 26, no. 4, pp. 357–363, 2012.
[15] S. Curtarolo, G. L. Hart, M. B. Nardelli, N. Mingo, S. Sanvito, and O. Levy, “The high-throughput highway to computational materials design,” Nature materials, vol. 12, no. 3, pp. 191–201, 2013.
[16] D. Jha, V. Gupta, W.-k. Liao, A. Choudhary, and A. Agrawal, “Moving closer to experimental level materials property prediction using ai,” Scientific reports, vol. 12, no. 1, pp. 1–9, 2022.
[17] M. C. Sorkun, S. Astruc, J. V. A. Koelman, and S. Er, “An artificial intelligenceaided virtual screening recipe for two-dimensional materials discovery,” npj Computational Materials, vol. 6, no. 1, p. 106, 2020.
[18] J. Zhou, L. Shen, M. D. Costa, K. A. Persson, S. P. Ong, P. Huck, Y. Lu, X. Ma, Y. Chen, H. Tang, et al., “2dmatpedia, an open computational database of twodimensional materials from top-down and bottom-up approaches,” Scientific data, vol. 6, no. 1, pp. 1–10, 2019.
[19] A. S. Rosen, S. M. Iyer, D. Ray, Z. Yao, A. Aspuru-Guzik, L. Gagliardi, J. M. Notestein, and R. Q. Snurr, “Machine learning the quantum-chemical properties of metal–organic frameworks for accelerated materials discovery,” Matter, vol. 4, no. 5, pp. 1578–1597, 2021.
[20] L. Ward, A. Agrawal, A. Choudhary, and C. Wolverton, “A general-purpose machine learning framework for predicting properties of inorganic materials,” npj Computational Materials, vol. 2, no. 1, pp. 1–7, 2016.
[21] Y. Li, R. Dong, W. Yang, and J. Hu, “Composition based crystal materials symmetry prediction using machine learning with enhanced descriptors,” Computational Materials Science, vol. 198, p. 110686, 2021.
[22] B. Sattari Baboukani, Z. Ye, K. G Reyes, and P. C. Nalam, “Prediction of nanoscale friction for two-dimensional materials using a machine learning approach,” Tribology Letters, vol. 68, pp. 1–14, 2020.
[23] N. Corriero, R. Rizzi, G. Settembre, N. Del Buono, and D. Diacono, “Crystalmela: a new crystallographic machine learning platform for crystal system determination,” Journal of Applied Crystallography, vol. 56, no. 2, 2023.
[24] H. Liang, V. Stanev, A. G. Kusne, and I. Takeuchi, “Cryspnet: Crystal structure predictions via neural networks,” Physical Review Materials, vol. 4, no. 12, p. 123802, 2020.
[25] A. Jain, S. P. Ong, G. Hautier, W. Chen, W. D. Richards, S. Dacek, S. Cholia, D. Gunter, D. Skinner, G. Ceder, et al., “Commentary: The materials project: A materials genome approach to accelerating materials innovation,” APL materials, vol. 1, no. 1, p. 011002, 2013.
[26] J. Enkovaara, C. Rostgaard, J. J. Mortensen, J. Chen, M. Dułak, L. Ferrighi, J. Gavnholt, C. Glinsvad, V. Haikola, H. Hansen, et al., “Electronic structure calculations with gpaw: a real-space implementation of the projector augmented-wave method,” Journal of physics: Condensed matter, vol. 22, no. 25, p. 253202, 2010.
[27] J. J. Mortensen, L. B. Hansen, and K. W. Jacobsen, “Real-space grid implementation of the projector augmented wave method,” Physical review B, vol. 71, no. 3, p. 035109, 2005.
[28] M. N. Gjerding, A. Taghizadeh, A. Rasmussen, S. Ali, F. Bertoldo, T. Deilmann, N. R. Knøsgaard, M. Kruse, A. H. Larsen, S. Manti, et al., “Recent progress of the computational 2d materials database (c2db),” 2D Materials, vol. 8, no. 4, p. 044002, 2021.
[29] S. Huang and J. M. Cole, “A database of battery materials auto-generated using chemdataextractor,” Scientific Data, vol. 7, no. 1, p. 260, 2020.
[30] Z. Lu, B. Zhu, B. W. Shires, D. O. Scanlon, and C. J. Pickard, “Ab initio random structure searching for battery cathode materials,” The Journal of Chemical Physics, vol. 154, no. 17, p. 174111, 2021.
[31] C. Wang, C. Yang, and Z. Zheng, “Toward practical high-energy and high-power lithium battery anodes: present and future,” Advanced Science, vol. 9, no. 9, p. 2105213, 2022.
[32] L. Zhang, C. Zhu, S. Yu, D. Ge, and H. Zhou, “Status and challenges facing representative anode materials for rechargeable lithium batteries,” Journal of Energy Chemistry, vol. 66, pp. 260–294, 2022.
[33] Y. Zhang, X. Xia, B. Liu, S. Deng, D. Xie, Q. Liu, Y. Wang, J. Wu, X. Wang, and J. Tu, “Multiscale graphene-based materials for applications in sodium ion batteries,” Advanced Energy Materials, vol. 9, no. 8, p. 1803342, 2019.
[34] B.-H. Hou, Y.-Y. Wang, Q.-L. Ning, W.-H. Li, X.-T. Xi, X. Yang, H.-J. Liang, X. Feng, and X.-L. Wu, “Self-supporting, flexible, additive-free, and scalable hard carbon paper self-interwoven by 1d microbelts: superb room/low-temperature sodium storage and working mechanism,” Advanced Materials, vol. 31, no. 40, p. 1903125, 2019.
[35] C. Vaalma, D. Buchholz, M. Weil, and S. Passerini, “A cost and resource analysis of sodium-ion batteries,” Nature reviews materials, vol. 3, no. 4, pp. 1–11, 2018.
[36] J. B. Robinson, D. P. Finegan, T. M. Heenan, K. Smith, E. Kendrick, D. J. Brett, and P. R. Shearing, “Microstructural analysis of the effects of thermal runaway on li-ion and na-ion battery electrodes,” Journal of Electrochemical Energy Conversion and Storage, vol. 15, no. 1, 2018.
[37] W. B. Park, S. C. Han, C. Park, S. U. Hong, U. Han, S. P. Singh, Y. H. Jung, D. Ahn, K.-S. Sohn, and M. Pyo, “Kvp2o7 as a robust high-energy cathode for potassiumion batteries: pinpointed by a full screening of the inorganic registry under specific search conditions,” Advanced Energy Materials, vol. 8, no. 13, p. 1703099, 2018.
[38] M. Zhou, P. Bai, X. Ji, J. Yang, C. Wang, and Y. Xu, “Electrolytes and interphases in potassium ion batteries,” Advanced Materials, vol. 33, no. 7, p. 2003741, 2021.
[39] K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 770–778, 2016.
[40] J. Singh and R. Banerjee, “A study on single and multi-layer perceptron neural network,” in 2019 3rd International Conference on Computing Methodologies and Communication (ICCMC), pp. 35–40, IEEE, 2019.
[41] T. Xie and J. C. Grossman, “Crystal graph convolutional neural networks for an accurate and interpretable prediction of material properties,” Physical review letters, vol. 120, no. 14, p. 145301, 2018.
[42] H. J. Bernstein, J. C. Bollinger, I. D. Brown, S. Gražulis, J. R. Hester, B. McMahon, N. Spadaccini, J. D. Westbrook, and S. P. Westrip, “Specification of the crystallographic information file format, version 2.0,” Journal of Applied Crystallography, vol. 49, no. 1, pp. 277–284, 2016.
[43] I. Brown, “Cif (crystallographic information file): a standard for crystallographic data interchange,” Journal of research of the National Institute of Standards and Technology, vol. 101, no. 3, p. 341, 1996.
[44] M. D. Zeiler and R. Fergus, “Visualizing and understanding convolutional networks,” in Computer Vision–ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings, Part I 13, pp. 818–833, Springer, 2014.
[45] K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 770–778, 2016.
[46] J. M. Crowley, J. Tahir-Kheli, and W. A. Goddard III, “Resolution of the band gap prediction problem for materials design,” The journal of physical chemistry letters, vol. 7, no. 7, pp. 1198–1203, 2016.
[47] J. E. Moussa, P. A. Schultz, and J. R. Chelikowsky, “Analysis of the heyd-scuseriaernzerhof density functional parameter space,” The Journal of chemical physics, vol. 136, no. 20, p. 204117, 2012.
[48] V. Venturi, H. L. Parks, Z. Ahmad, and V. Viswanathan, “Machine learning enabled discovery of application dependent design principles for two-dimensional materials,” Machine Learning: Science and Technology, vol. 1, no. 3, p. 035015, 2020.
[49] O. Isayev, C. Oses, C. Toher, E. Gossett, S. Curtarolo, and A. Tropsha, “Universal fragment descriptors for predicting properties of inorganic crystals,” Nature communications, vol. 8, no. 1, p. 15679, 2017.
描述 碩士
國立政治大學
應用物理研究所
110755006
資料來源 http://thesis.lib.nccu.edu.tw/record/#G0110755006
資料類型 thesis
dc.contributor.advisor 許琇娟zh_TW
dc.contributor.advisor Hsu, Hsiu-Chuanen_US
dc.contributor.author (Authors) 陳建豪zh_TW
dc.contributor.author (Authors) Chen, Jian-Haoen_US
dc.creator (作者) 陳建豪zh_TW
dc.creator (作者) Chen, Jian-Haoen_US
dc.date (日期) 2023en_US
dc.date.accessioned 1-Sep-2023 16:28:41 (UTC+8)-
dc.date.available 1-Sep-2023 16:28:41 (UTC+8)-
dc.date.issued (上傳時間) 1-Sep-2023 16:28:41 (UTC+8)-
dc.identifier (Other Identifiers) G0110755006en_US
dc.identifier.uri (URI) http://nccur.lib.nccu.edu.tw/handle/140.119/147298-
dc.description (描述) 碩士zh_TW
dc.description (描述) 國立政治大學zh_TW
dc.description (描述) 應用物理研究所zh_TW
dc.description (描述) 110755006zh_TW
dc.description.abstract (摘要) 二維材料具有非常特別的物理性質，在電子、光學元件及航空航太等，都具有極高的應用價值，以石墨烯為例，具有高導電性、高導熱性及高機械強度等等，除石墨烯外，包括像單層二硫化鉬(MoS2)、二硒化鉬(MoSe2)，他們因為皆為直接能隙半導體，可被製作成透明的發光二極體(light-emitting diode, LED)，且⼆維結構⾃然有利於各種類型器件的性能，如減小尺寸以防止短通道效應以及提高可穿戴設備的靈活性。
一般材料之探勘到實際應用的耗時較長，從實驗室發現一種新型材料，再到研發、驗證、直至最後的業務場景落地需要較長久的時間，而傳統材料探勘與研究通常較仰賴專業人員的研究背景與實務經驗，來提出可行的候選材料與方法，而過程相當耗時耗力，且需投入大量經費。
有鑑於此，本文基於機器學習對二維材料的性質進行預測，希望能減少二維材料在研發過程中所需的時間，並能夠輔助研究人員在前期尚未開始實際研究測量數值時，有個大致的參考數值。
本實驗所使用的深度學習算法基於以下三種，一種為多層感知機(Multilayer perceptron, MLP)，以及另外兩種分別為圖卷積神經網路(Graph Convolutional Network, CGCNN) 和殘差網路(Residual Network,ResNet)，多層感知機為一種前向傳遞類神經網路，透過每層神經元之間的資料傳遞來學習到相關的資訊，並且利用「倒傳遞」的技術達到學習(model learning)的監督式學習。另一項圖卷積神經網路其本質目的是用來提取拓撲圖的空間特徵，透過將數據轉化為拓撲圖使網路能從中提取出特徵並學習到相關資訊，可用於非歐幾里得空間的圖形，並可應用於數據預測及類別分類任務。最後一項殘差網路則為卷積神經網路的一個變體，透過將輸入與輸出特徵做結合，從而改善模型過擬合的狀況。
本實驗透過引入深度學習技術來預測材料的剝離能、晶格常數以及晶格結構，並利用不同模型比較各自差異，找出較好的結果以及訓練方法，本實驗所獲得的最好結果分別是，透過CGCNN 訓練剝離能，其誤差為0.0508eV/atom，而晶格常數之三邊長以MLP 在邊長a、b、c的誤差最小分別為0.5994Å，1.0664Å 以及1.2785Å，最後晶格結構的分類準確度以利用MLP 作為訓練模型的效果最好，其準確度為65%。
除此之外，現今電動車的高度發展，電動車電池（electric-vehicle battery）的需求提高，且續航以及安全要求也相對提升，使電動車電池之研究頗為重要，有鑑於此，本實驗也建立一自製電池材料數據庫，希望能提供給開發人員進行開發使用。		zh_TW
dc.description.abstract (摘要) Two-dimensional materials possess extraordinary physical properties and
hold immense value for applications in electronics, optical components, aerospace, and other fields. Taking graphene as an example, it exhibits high electric conductivity, thermal conductivity, and mechanical strength. Apart from graphene, materials like MoS2 and MoSe2 belong to direct bandgap semiconductors and can be used to fabricate transparent light-emitting diodes (LEDs). Moreover, the two-dimensional structure inherently improves the performance of various device types by reducing size to mitigate short-channel effects and enhancing the flexibility of wearable devices.
The development from material exploration to practical application is often time-consuming. It involves discovering a new material in the laboratory, followed by immense research, verification, and final implementation in business scenarios. Conventional material exploration research heavily relies on the expertise and practical experience of specialized professionals to identify viable candidate materials. This process is both time-consuming and resource-intensive, requiring a significant investment of funds.
Recently, the data-driven techniques have been adopted to predict the material properties, owing to the mature machine learning algorithms and growing number of databases. This new paradigm reduces the time required for research and development in material science. Thus, this thesis applies this approach to predict properties of two-dimensional materials with the aim to accelerate the material exploration.
The deep learning models used in this thesis are based on three models:
Multilayer Perceptron (MLP), Crystal Graph Convolutional Neural Networks (CGCNN), and Residual Network (ResNet). These models were employed to predict material properties such as exfoliation energy, lattice constants, and crystal system. By introducing deep learning techniques, a comparison between different models was conducted to determine the best results and training methods.
The best results obtained are as follows. For exfoliation energy prediction,
CGCNN achieved an error of 0.0508 eV/atom. Regarding lattice constants,
MLP had the smallest error for lattice constant a with 0.5994 Å.Lastly, for
crystal system classification, MLP performed the best, achieving an accuracy of 65%.
Moreover, with the growing development of electric vehicles, the demand
for electric-vehicle batteries has increased, making the research on electric vehicle batteries very important. This thesis establishes a battery material database that contains several crystal and energetic properties of battery materials. The database could be utilized for the study with data-driven approaches in the future.		en_US
dc.description.tableofcontents 1 緒論 1
1.1 材料搜尋 1
1.2 二維材料 2
1.3 機器學習 3
1.4 深度學習 3
1.5 基於深度學習的材料計算 5
2 實驗方法 8
2.1 數據集 8
2.1.1 2DMatPedia 8
2.1.2 Materials project 10
2.1.3 Computational 2D Materials Database(C2DB) 12
2.1.4 自製電池材料數據集 12
2.2 神經網路模型 15
2.2.1 多層感知器(Multilayer Perceptron, MLP) 15
2.2.2 晶體圖卷積神經網路(Crystal Graph Convolutional Neural
Networks, CGCNN) 17
2.2.3 殘差網路(Residual Network, ResNet) 19
2.3 實施細則 22
2.3.1 剝離能(Exfoliation energy) 22
2.3.2 晶格常數(Lattice Parameters) 26
2.3.3 晶體結構(Crystal System) 33
3 結果 36
3.1 剝離能 36
3.1.1 CGCNN 36
3.1.2 ResNet 41
3.1.3 決策樹 42
3.1.4 隨機森林 43
3.2 晶格常數 43
3.2.1 CGCNN 43
3.2.2 MLP 45
3.2.3 ResNet 51
3.3 晶體結構 56
3.3.1 CGCNN 56
3.3.2 MLP 58
3.3.3 自製電池材料數據庫 60
4 結論 63
5 附錄 65
5.1 CRYSPNet 介紹 65
5.2 結果比較 66
參考文獻 68		zh_TW
dc.format.extent 6955310 bytes-
dc.format.mimetype application/pdf-
dc.source.uri (資料來源) http://thesis.lib.nccu.edu.tw/record/#G0110755006en_US
dc.subject (關鍵詞) 深度學習zh_TW
dc.subject (關鍵詞) 二維材料zh_TW
dc.subject (關鍵詞) 材料信息學zh_TW
dc.subject (關鍵詞) Deep learningen_US
dc.subject (關鍵詞) Two-dimensional materialen_US
dc.subject (關鍵詞) Materials Informaticsen_US
dc.title (題名) 機器學習預測材料特性zh_TW
dc.title (題名) Predicting material properties with machine learningen_US
dc.type (資料類型) thesisen_US
dc.relation.reference (參考文獻) [1] Q. Tong, P. Gao, H. Liu, Y. Xie, J. Lv, Y. Wang, and J. Zhao, “Combining machine learning potential and structure prediction for accelerated materials design and discovery,” The Journal of Physical Chemistry Letters, vol. 11, no. 20, pp. 8710–8720, 2020.
[2] C. Gao, X. Min, M. Fang, T. Tao, X. Zheng, Y. Liu, X. Wu, and Z. Huang, “Innovative materials science via machine learning,” Advanced Functional Materials, vol. 32, no. 1, p. 2108044, 2022.
[3] K. Guo, Z. Yang, C.-H. Yu, and M. J. Buehler, “Artificial intelligence and machine learning in design of mechanical materials,” Materials Horizons, vol. 8, no. 4, pp. 1153–1172, 2021.
[4] G. H. Jeong, S. P. Sasikala, T. Yun, G. Y. Lee, W. J. Lee, and S. O. Kim, “Nanoscale assembly of 2d materials for energy and environmental applications,” Advanced Materials, vol. 32, no. 35, p. 1907006, 2020.
[5] B. Radisavljevic, A. Radenovic, J. Brivio, V. Giacometti, and A. Kis, “Single-layer mos2 transistors,” Nature nanotechnology, vol. 6, no. 3, pp. 147–150, 2011.
[6] D. Berman, B. Narayanan, M. J. Cherukara, S. K. Sankaranarayanan, A. Erdemir, A. Zinovev, and A. V. Sumant, “Operando tribochemical formation of onion-like-carbon leads to macroscale superlubricity,” Nature communications, vol. 9, no. 1, p. 1164, 2018.
[7] W. Wu, J. Wang, P. Ercius, N. C. Wright, D. M. Leppert-Simenauer, R. A. Burke, M. Dubey, A. M. Dogare, and M. T. Pettes, “Giant mechano-optoelectronic effect in an atomically thin semiconductor,” Nano letters, vol. 18, no. 4, pp. 2351–2357, 2018.
[8] H. G. Kim and H.-B.-R. Lee, “Atomic layer deposition on 2d materials,” Chemistry of Materials, vol. 29, no. 9, pp. 3809–3826, 2017.
[9] G. Carleo, I. Cirac, K. Cranmer, L. Daudet, M. Schuld, N. Tishby, L. Vogt-Maranto, and L. Zdeborová, “Machine learning and the physical sciences,” Reviews of Modern Physics, vol. 91, no. 4, p. 045002, 2019.
[10] E. Dolotov and N. Zolotykh, “Evolutionary algorithms for constructing an ensemble of decision trees,” in Analysis of Images, Social Networks and Texts: 8th International Conference, AIST 2019, Kazan, Russia, July 17–19, 2019, Revised Selected Papers 8, pp. 9–15, Springer, 2020.
[11] G. Biau and E. Scornet, “A random forest guided tour,” Test, vol. 25, pp. 197–227, 2016.
[12] Y. LeCun, Y. Bengio, and G. Hinton, “Deep learning,” nature, vol. 521, no. 7553, pp. 436–444, 2015.
[13] S. B. Kotsiantis, I. Zaharakis, P. Pintelas, et al., “Supervised machine learning: A review of classification techniques,” Emerging artificial intelligence applications in computer engineering, vol. 160, no. 1, pp. 3–24, 2007.
[14] H. Schulz and S. Behnke, “Deep learning,” KI-Künstliche Intelligenz, vol. 26, no. 4, pp. 357–363, 2012.
[15] S. Curtarolo, G. L. Hart, M. B. Nardelli, N. Mingo, S. Sanvito, and O. Levy, “The high-throughput highway to computational materials design,” Nature materials, vol. 12, no. 3, pp. 191–201, 2013.
[16] D. Jha, V. Gupta, W.-k. Liao, A. Choudhary, and A. Agrawal, “Moving closer to experimental level materials property prediction using ai,” Scientific reports, vol. 12, no. 1, pp. 1–9, 2022.
[17] M. C. Sorkun, S. Astruc, J. V. A. Koelman, and S. Er, “An artificial intelligenceaided virtual screening recipe for two-dimensional materials discovery,” npj Computational Materials, vol. 6, no. 1, p. 106, 2020.
[18] J. Zhou, L. Shen, M. D. Costa, K. A. Persson, S. P. Ong, P. Huck, Y. Lu, X. Ma, Y. Chen, H. Tang, et al., “2dmatpedia, an open computational database of twodimensional materials from top-down and bottom-up approaches,” Scientific data, vol. 6, no. 1, pp. 1–10, 2019.
[19] A. S. Rosen, S. M. Iyer, D. Ray, Z. Yao, A. Aspuru-Guzik, L. Gagliardi, J. M. Notestein, and R. Q. Snurr, “Machine learning the quantum-chemical properties of metal–organic frameworks for accelerated materials discovery,” Matter, vol. 4, no. 5, pp. 1578–1597, 2021.
[20] L. Ward, A. Agrawal, A. Choudhary, and C. Wolverton, “A general-purpose machine learning framework for predicting properties of inorganic materials,” npj Computational Materials, vol. 2, no. 1, pp. 1–7, 2016.
[21] Y. Li, R. Dong, W. Yang, and J. Hu, “Composition based crystal materials symmetry prediction using machine learning with enhanced descriptors,” Computational Materials Science, vol. 198, p. 110686, 2021.
[22] B. Sattari Baboukani, Z. Ye, K. G Reyes, and P. C. Nalam, “Prediction of nanoscale friction for two-dimensional materials using a machine learning approach,” Tribology Letters, vol. 68, pp. 1–14, 2020.
[23] N. Corriero, R. Rizzi, G. Settembre, N. Del Buono, and D. Diacono, “Crystalmela: a new crystallographic machine learning platform for crystal system determination,” Journal of Applied Crystallography, vol. 56, no. 2, 2023.
[24] H. Liang, V. Stanev, A. G. Kusne, and I. Takeuchi, “Cryspnet: Crystal structure predictions via neural networks,” Physical Review Materials, vol. 4, no. 12, p. 123802, 2020.
[25] A. Jain, S. P. Ong, G. Hautier, W. Chen, W. D. Richards, S. Dacek, S. Cholia, D. Gunter, D. Skinner, G. Ceder, et al., “Commentary: The materials project: A materials genome approach to accelerating materials innovation,” APL materials, vol. 1, no. 1, p. 011002, 2013.
[26] J. Enkovaara, C. Rostgaard, J. J. Mortensen, J. Chen, M. Dułak, L. Ferrighi, J. Gavnholt, C. Glinsvad, V. Haikola, H. Hansen, et al., “Electronic structure calculations with gpaw: a real-space implementation of the projector augmented-wave method,” Journal of physics: Condensed matter, vol. 22, no. 25, p. 253202, 2010.
[27] J. J. Mortensen, L. B. Hansen, and K. W. Jacobsen, “Real-space grid implementation of the projector augmented wave method,” Physical review B, vol. 71, no. 3, p. 035109, 2005.
[28] M. N. Gjerding, A. Taghizadeh, A. Rasmussen, S. Ali, F. Bertoldo, T. Deilmann, N. R. Knøsgaard, M. Kruse, A. H. Larsen, S. Manti, et al., “Recent progress of the computational 2d materials database (c2db),” 2D Materials, vol. 8, no. 4, p. 044002, 2021.
[29] S. Huang and J. M. Cole, “A database of battery materials auto-generated using chemdataextractor,” Scientific Data, vol. 7, no. 1, p. 260, 2020.
[30] Z. Lu, B. Zhu, B. W. Shires, D. O. Scanlon, and C. J. Pickard, “Ab initio random structure searching for battery cathode materials,” The Journal of Chemical Physics, vol. 154, no. 17, p. 174111, 2021.
[31] C. Wang, C. Yang, and Z. Zheng, “Toward practical high-energy and high-power lithium battery anodes: present and future,” Advanced Science, vol. 9, no. 9, p. 2105213, 2022.
[32] L. Zhang, C. Zhu, S. Yu, D. Ge, and H. Zhou, “Status and challenges facing representative anode materials for rechargeable lithium batteries,” Journal of Energy Chemistry, vol. 66, pp. 260–294, 2022.
[33] Y. Zhang, X. Xia, B. Liu, S. Deng, D. Xie, Q. Liu, Y. Wang, J. Wu, X. Wang, and J. Tu, “Multiscale graphene-based materials for applications in sodium ion batteries,” Advanced Energy Materials, vol. 9, no. 8, p. 1803342, 2019.
[34] B.-H. Hou, Y.-Y. Wang, Q.-L. Ning, W.-H. Li, X.-T. Xi, X. Yang, H.-J. Liang, X. Feng, and X.-L. Wu, “Self-supporting, flexible, additive-free, and scalable hard carbon paper self-interwoven by 1d microbelts: superb room/low-temperature sodium storage and working mechanism,” Advanced Materials, vol. 31, no. 40, p. 1903125, 2019.
[35] C. Vaalma, D. Buchholz, M. Weil, and S. Passerini, “A cost and resource analysis of sodium-ion batteries,” Nature reviews materials, vol. 3, no. 4, pp. 1–11, 2018.
[36] J. B. Robinson, D. P. Finegan, T. M. Heenan, K. Smith, E. Kendrick, D. J. Brett, and P. R. Shearing, “Microstructural analysis of the effects of thermal runaway on li-ion and na-ion battery electrodes,” Journal of Electrochemical Energy Conversion and Storage, vol. 15, no. 1, 2018.
[37] W. B. Park, S. C. Han, C. Park, S. U. Hong, U. Han, S. P. Singh, Y. H. Jung, D. Ahn, K.-S. Sohn, and M. Pyo, “Kvp2o7 as a robust high-energy cathode for potassiumion batteries: pinpointed by a full screening of the inorganic registry under specific search conditions,” Advanced Energy Materials, vol. 8, no. 13, p. 1703099, 2018.
[38] M. Zhou, P. Bai, X. Ji, J. Yang, C. Wang, and Y. Xu, “Electrolytes and interphases in potassium ion batteries,” Advanced Materials, vol. 33, no. 7, p. 2003741, 2021.
[39] K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 770–778, 2016.
[40] J. Singh and R. Banerjee, “A study on single and multi-layer perceptron neural network,” in 2019 3rd International Conference on Computing Methodologies and Communication (ICCMC), pp. 35–40, IEEE, 2019.
[41] T. Xie and J. C. Grossman, “Crystal graph convolutional neural networks for an accurate and interpretable prediction of material properties,” Physical review letters, vol. 120, no. 14, p. 145301, 2018.
[42] H. J. Bernstein, J. C. Bollinger, I. D. Brown, S. Gražulis, J. R. Hester, B. McMahon, N. Spadaccini, J. D. Westbrook, and S. P. Westrip, “Specification of the crystallographic information file format, version 2.0,” Journal of Applied Crystallography, vol. 49, no. 1, pp. 277–284, 2016.
[43] I. Brown, “Cif (crystallographic information file): a standard for crystallographic data interchange,” Journal of research of the National Institute of Standards and Technology, vol. 101, no. 3, p. 341, 1996.
[44] M. D. Zeiler and R. Fergus, “Visualizing and understanding convolutional networks,” in Computer Vision–ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings, Part I 13, pp. 818–833, Springer, 2014.
[45] K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 770–778, 2016.
[46] J. M. Crowley, J. Tahir-Kheli, and W. A. Goddard III, “Resolution of the band gap prediction problem for materials design,” The journal of physical chemistry letters, vol. 7, no. 7, pp. 1198–1203, 2016.
[47] J. E. Moussa, P. A. Schultz, and J. R. Chelikowsky, “Analysis of the heyd-scuseriaernzerhof density functional parameter space,” The Journal of chemical physics, vol. 136, no. 20, p. 204117, 2012.
[48] V. Venturi, H. L. Parks, Z. Ahmad, and V. Viswanathan, “Machine learning enabled discovery of application dependent design principles for two-dimensional materials,” Machine Learning: Science and Technology, vol. 1, no. 3, p. 035015, 2020.
[49] O. Isayev, C. Oses, C. Toher, E. Gossett, S. Curtarolo, and A. Tropsha, “Universal fragment descriptors for predicting properties of inorganic crystals,” Nature communications, vol. 8, no. 1, p. 15679, 2017.		zh_TW