An optimized strategy for density prediction of intermetallics across varied crystal structures via graph neural network

Abstract

Intermetallic compounds are crucial in modern industry due to their exceptional properties, where density is identified as a critical parameter determining their potentiality for lightweight applications. In this study, over 7,000 density data points are collected for binary intermetallic compounds from different crystal structures. A new intermetallics graph neural network (IGNN) model is developed to perform regression and classification tasks for density prediction. Compared to traditional machine learning models, the IGNN model demonstrated superior capability in capturing crystal structure and effectively addressing challenges posed by polymorphism. The interpretability of the IGNN model classification process is enhanced through the t-distributed stochastic neighbor embedding (t-SNE) visualization method. Additionally, the IGNN model exhibited excellent performance in predicting the density of multicomponent complex intermetallic compounds, indicating its robustness and generalizability. This study presents a graph neural network (GNN) method suitable for multi-crystal structure data modeling, providing a novel computational framework for density prediction in intermetallic compounds. This advancement represents a significant contribution to this field, paving the way for more targeted material selection and application in lightweight technologies.

Keywords

Intermetallic compounds / density / machine learning / graph neural network

References

[1] Russell A.Ductility in intermetallic compounds.Adv Eng Mater2003;5:629-39

[2] Taub AI.Intermetallic compounds for high-temperature structural use.Science1989;243:616-21

[3] Dshemuchadse J.Some statistics on intermetallic compounds.Inorg Chem2015;54:1120-8

[4] Yamaguchi M,Ito K.High-temperature structural intermetallics.Acta Mater2000;48:307-22

[5] Kimura Y.Ductility and toughness in intermetallics.Intermetallics1998;6:567-71

[6] Zhu D,Wu H.Identifying intrinsic factors for ductile-to-brittle transition temperatures in Fe–Al intermetallics via machine learning.J Mater Res Technol2023;26:8836-45

[7] Ravindran P.Correlation between electronic structure, mechanical properties and phase stability in intermetallic compounds.Bull Mater Sci1997;20:613-22

[8] Kim SH,Kim NJ.Brittle intermetallic compound makes ultrastrong low-density steel with large ductility.Nature2015;518:77-9

[9] Crawley AF.Densities of liquid metals and alloys.Int Metall Rev1974;19:32-48

[10] Fleischer RL.High-strength, high-temperature intermetallic compounds.J Mater Sci1987;22:2281-8

[11] Stoloff N,Deevi S.Emerging applications of intermetallics.Intermetallics2000;8:1313-20

[12] Uenishi K.Processing of intermetallic compounds for structural applications at high temperature.Intermetallics1996;4:S95-101

[13] Paul AR,Singh D.A critical review on the properties of intermetallic compounds and their application in the modern manufacturing.Cryst Res Technol2022;57:2100159

[14] Fatima B,Acharya N.Density functional study of XRh (X=Sc, Y, Ti and Zr) intermetallic compounds.Comput Mater Sci2014;89:205-15

[15] Lu ZW,Zunger A,Ferreira LG.First-principles statistical mechanics of structural stability of intermetallic compounds.Phys Rev B Condens Matter1991;44:512-44

[16] Zhu D,Wu Y.Improved material descriptors for bulk modulus in intermetallic compounds via machine learning.Rare Met2023;42:2396-405

[17] Oliynyk AO.Discovery of intermetallic compounds from traditional to machine-learning approaches.Acc Chem Res2018;51:59-68

[18] Medasani B,Ding H.Predicting defect behavior in B2 intermetallics by merging ab initio modeling and machine learning.npj Comput Mater2016;2:1

[19] Nie M,Wang D.Reinforcement learning on graphs: a survey.IEEE Trans Emerg Top Comput Intell2023;7:1065-82

[20] Xie T.Crystal graph convolutional neural networks for an accurate and interpretable prediction of material properties.Phys Rev Lett2018;120:145301

[21] Park CW.Developing an improved crystal graph convolutional neural network framework for accelerated materials discovery.Phys Rev Mater2020;4:063801

[22] Reiser P,Eberhard A.Graph neural networks for materials science and chemistry.Commun Mater2022;3:93 PMCID:PMC9702700

[23] Dreger M,Eikerling MH.Synergizing ontologies and graph databases for highly flexible materials-to-device workflow representations.J Mater Inf2023;3:2

[24] Choudhary K.Atomistic line graph neural network for improved materials property predictions.npj Comput Mater2021;7:650

[25] Wu X,Gong Y.Graph neural networks for molecular and materials representation.J Mater Inf2023;3:12

[26] Fung V,Juarez E.Benchmarking graph neural networks for materials chemistry.npj Comput Mater2021;7:554

[27] Dai M,Liang Y.Graph neural networks for an accurate and interpretable prediction of the properties of polycrystalline materials.npj Comput Mater2021;7:574

[28] Jain A,Hautier G.Commentary: the materials project: a materials genome approach to accelerating materials innovation.APL Materials2013;1:011002

[29] Chicco D,Jurman G.The coefficient of determination R-squared is more informative than SMAPE, MAE, MAPE, MSE and RMSE in regression analysis evaluation.PeerJ Comput Sci2021;7:e623 PMCID:PMC8279135

[30] Chai T.Root mean square error (RMSE) or mean absolute error (MAE)? - Arguments against avoiding RMSE in the literature.Geosci Model Dev2014;7:1247-50

[31] Zhu D,Hou F.A transfer learning strategy for tensile strength prediction in austenitic stainless steel across temperatures.Scr Mater2024;251:116210

[32] Bradley AP.The use of the area under the ROC curve in the evaluation of machine learning algorithms.Pattern Recognit1997;30:1145-59

[33] Yoshida Laboratory. XenonPy. 2020. https://github.com/yoshida-lab/XenonPy. (accessed 2025-02-05)

[34] Fessler J.Nonuniform fast fourier transforms using min-max interpolation.IEEE Trans Signal Process2003;51:560-74

[35] Isayev O,Toher C,Curtarolo S.Universal fragment descriptors for predicting properties of inorganic crystals.Nat Commun2017;8:15679 PMCID:PMC5465371

[36] Ioffe S. Batch normalization: accelerating deep network training by reducing internal covariate shift. arXiv 2015, arXiv:1502.03167. Available online: https://doi.org/10.48550/arXiv.1502.03167. (accessed 5 Feb 2025)

[37] Du J,Wu G,Kar S. Topology adaptive graph convolutional networks. arXiv 2017, arXiv:1710.10370. Available online: https://doi.org/10.48550/arXiv.1710.10370. (accessed 5 Feb 2025)

[38] Belkina AC,Anno R,Spidlen J.Automated optimized parameters for T-distributed stochastic neighbor embedding improve visualization and analysis of large datasets.Nat Commun2019;10:5415 PMCID:PMC6882880