Advances in graph neural networks for alloy design and properties predictions: a review

Abstract

Graph neural networks (GNNs) have become a transformative modeling paradigm in materials science, offering a data-efficient and structure-aware approach for learning from complex material systems. This review focuses on the recent progress of GNNs in alloy design and property prediction. We begin by introducing the foundational concepts of graph representations and the general architecture of GNNs, including node embeddings, message passing, and pooling strategies. The review then categorizes major types of GNNs, including supervised and unsupervised learning, with a focus on the achievements and applications of GNNs in materials modeling, and discusses their strengths and inherent limitations in the context of materials modeling. Particular emphasis is placed on the application of GNNs in the alloy domain, covering a diverse range of data types, from atomic structures and compositions to microstructural images, and target properties, such as mechanical strength, thermal stability, and phase stability. We highlight how GNNs are integrated into alloy composition optimization, multi-property prediction, and frontier research workflows. The review concludes with a summary of multi-model and multiscale approaches and outlines key challenges and future directions for constructing generalizable, physics-informed GNN frameworks for alloy discovery.

Keywords

Graph neural networks / alloys / multiscale modeling / alloy composition / machine learning / structure-property relationships

References

[1] Dobrzański LA.Significance of materials science for the future development of societies.J Mater Process Technol2006;175:133-48

[2] Fullwood DT,Adams BL.Microstructure sensitive design for performance optimization.Prog Mater Sci2010;55:477-562

[3] Darolia R.Development of strong, oxidation and corrosion resistant nickel-based superalloys: critical review of challenges, progress and prospects.Int Mater Rev2019;64:355-80

[4] Emmanuel A,Akande I.Aluminium alloys as advanced materials: a short communication.IOP Conf Ser Mater Sci Eng2021;1107:012024

[5] Yeh J,Lin S.Nanostructured high-entropy alloys with multiple principal elements: novel alloy design concepts and outcomes.Adv Eng Mater2004;6:299-303

[6] Wang L,Apelian D.Aluminium die casting alloys: alloy composition, microstructure, and properties-performance relationships.Int Mater Rev1995;40:221-38

[7] Fu H,Wang C,Xie J.Recent progress in the machine learning-assisted rational design of alloys.Int J Miner Metall Mater2022;29:635-44

[8] Pilania G.Machine learning in materials science: from explainable predictions to autonomous design.Comput Mater Sci2021;193:110360

[9] Mi X,Tang A.A reverse design model for high-performance and low-cost magnesium alloys by machine learning.Comput Mater Sci2022;201:110881

[10] Hu M,Knibbe R.Recent applications of machine learning in alloy design: a review.Mat Sci Eng R2023;155:100746

[11] Brown P.Quantum machine-learning phase prediction of high-entropy alloys.Mater Today2023;63:18-31

[12] Pei C,Zhang J.A novel model to predict oxidation behavior of superalloys based on machine learning.J Mater Sci Technol2025;235:232-43

[13] Zhang W,Ai Y.Intelligent screening model for uranium alloy corrosion substitute alloys based on machine learning.Prog Nat Sci2025;35:622-30

[14] Su J,Van Petegem S.Additive manufacturing metallurgy guided machine learning design of versatile alloys.Mater Today2025;In Press:

[15] Ma Q,Xin R.Thermodynamic calculation and machine learning aided composition design of new nickel-based superalloys.J Mater Res Technol2023;26:4168-78

[16] Jin J,Liu B.Comparative analysis of conventional machine learning and graph neural network models for perovskite property prediction.J Phys Chem C2024;128:16672-83

[17] Vatter J,Jacobsen H.The evolution of distributed systems for graph neural networks and their origin in graph processing and deep learning: a survey.ACM Comput Surv2024;56:1-37

[18] LeCun Y,Hinton G.Deep learning.Nature2015;521:436-44

[19] Leskovec J,Dasgupta A.Community structure in large networks: natural cluster sizes and the absence of large well-defined clusters.Internet Math2009;6:29-123

[20] Jamali K,Zhang R,Kimanius D.Automated model building and protein identification in cryo-EM maps.Nature2024;628:450-7 PMCID:PMC11006616

[21] Sherwood AV,Ryberg LA.Hepatitis C virus RNA is 5'-capped with flavin adenine dinucleotide.Nature2023;619:811-8 PMCID:PMC7616780

[22] Gaudelet T,Jamasb AR.Utilizing graph machine learning within drug discovery and development.Brief Bioinform2021;22:bbab159 PMCID:PMC8574649

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

[24] Chen C,Zuo Y,Ong SP.Graph networks as a universal machine learning framework for molecules and crystals.Chem Mater2019;31:3564-72

[25] Réau M,Xue LC.DeepRank-GNN: a graph neural network framework to learn patterns in protein-protein interfaces.Bioinformatics2023;39:btac759 PMCID:PMC9805592

[26] Asif NA,Chakrabortty RK.Graph neural network: a comprehensive review on non-Euclidean space.IEEE Access2021;9:60588-606

[27] Bondy A. Graph theory, 1st edition. Springer. 2008. https://link.springer.com/book/9781846289699. (accessed 31 Jul 2025)

[28] Cormen TH,Rivest RL. Introduction to algorithms. 3rd edition. MIT Press: Cambridge, MA. 2009. https://enos.itcollege.ee/~japoia/algorithms/GT/Introduction_to_algorithms-3rd%20Edition.pdf. (accessed 31 Jul 2025)

[29] Shen L,Yang T,Feng YP.High-throughput computational discovery and intelligent design of two-dimensional functional materials for various applications.Acc Mater Res2022;3:572-83

[30] Behler J.Perspective: machine learning potentials for atomistic simulations.J Chem Phys2016;145:170901

[31] Unke OT.PhysNet: a neural network for predicting energies, forces, dipole moments, and partial charges.J Chem Theory Comput2019;15:3678-93

[32] Ben Chaabene, W.; Flah, M.; Nehdi, M. L. Machine learning prediction of mechanical properties of concrete: critical review.Constr Build Mater2020;260:119889

[33] Ren Y,Xu S,Xuan W.Rapid estimation of γ' solvus temperature for composition design of Ni-based superalloy via physics-informed generative artificial intelligence.J Alloys Metall Syst2024;6:100073

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

[35] Christensen AS,Faber FA.FCHL revisited: faster and more accurate quantum machine learning.J Chem Phys2020;152:044107

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

[37] Gasteiger J,Margraf JT. Fast and uncertainty-aware directional message passing for non-equilibrium molecules. arXiv 2020, arXiv:2011.14115. https://doi.org/10.48550/arXiv.2011.14115. (accessed 31 Jul 2025)

[38] Gilmer J,Riley PF,Dahl GE.Neural message passing for quantum chemistry. In Proceedings of the 34th International Conference on Machine Learning (ICML 2017), Sydney, Australia. Aug 6-11 2017. JMLR.org; 2017. pp. 1263-72.

[39] Zhang J,Xie J,King I. GaAN: gated attention networks for learning on large and spatiotemporal graphs. arXiv 2018, arXiv:1803.07294. https://doi.org/10.48550/arXiv.1803.07294. (accessed 31 Jul 2025)

[40] Zhang M,Neumann M. An end-to-end deep-learning architecture for graph classification. In Proceedings of the 32nd AAAI Conference on Artificial Intelligence (AAAI 2018), New Orleans, USA. Feb 2-7 2018. AAAI Press; 2018. pp. 4438-45. https://muhanzhang.github.io/papers/AAAI_2018_DGCNN.pdf. (accessed 31 Jul 2025)

[41] Kipf TN. Semi-supervised classification with graph convolutional networks. arXiv 2017, arXiv:1609.02907. https://doi.org/10.48550/arXiv.1609.02907. (accessed 31 Jul 2025)

[42] Hamilton WL,Leskovec J.Inductive representation learning on large graphs. In Proceedings of the 31st Conference on Neural Information Processing Systems (NeurIPS 2017), Long Beach, USA. Dec 4-9 2017. Curran Associates Inc.; 2017. pp. 1025-35.

[43] Veličković P,Casanova A,Liò P. Graph attention networks. arXiv 2017, arXiv:1710.10903. https://doi.org/10.48550/arXiv.1710.10903. (accessed 31 Jul 2025)

[44] Chen J,Xiao C. FastGCN: fast learning with graph convolutional networks via importance sampling. arXiv 2018, arXiv:1801.10247. https://doi.org/10.48550/arXiv.1801.10247. (accessed 31 Jul 2025)

[45] Li Q,Wu XM.Deeper insights into graph convolutional networks for semi-supervised learning. In Proceedings of the 32nd AAAI Conference on Artificial Intelligence (AAAI 2018), New Orleans, USA. Feb 2-7, 2018. AAAI Press; 2018. pp. 3538-45.

[46] Oviedo F,Buonassisi T.Interpretable and explainable machine learning for materials science and chemistry.Acc Mater Res2022;3:597-607

[47] Batalović K,Paskaš Mamula B,Medić Ilić M.Predicting the heat of hydride formation by graph neural network - exploring the structure–property relation for metal hydrides.Adv Theory Simul2022;5:2200293

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

[49] Dong Z,Ji Y.SLI-GNN: a self-learning-input graph neural network for predicting crystal and molecular properties.J Phys Chem A2023;127:5921-9

[50] Huang K,Zheng Z,Shen X.Understanding and bridging the gaps in current GNN performance optimizations. In Proceedings of the 26th ACM SIGPLAN Symposium on Principles and Practice of Parallel Programmi, Virtual Event, Republic of Korea. 27 February - 3 March 2021. Association for Computing Machinery; 2021. pp. 119-32.

[51] Law JN,Gorai P.Upper-bound energy minimization to search for stable functional materials with graph neural networks.JACS Au2023;3:113-23 PMCID:PMC9875372

[52] Cui Y,Zhou W.SA-GNN: prediction of material properties using graph neural network based on multi-head self-attention optimization.AIP Adv2024;14:055033

[53] Chen C.A universal graph deep learning interatomic potential for the periodic table.Nat Comput Sci2022;2:718-28

[54] Batatia I,Simm G,Csányi G. MACE: higher-order equivariant message-passing neural networks for fast and accurate force fields. In Advances in Neural Information Processing Systems 35 (NeurIPS 2022). 2022. pp. 11423-36. https://proceedings.neurips.cc/paper_files/paper/2022/hash/4a36c3c51af11ed9f34615b81edb5bbc-Abstract-Conference.html. (accessed 31 Jul 2025)

[55] Borge-Durán I,Araya LA.Explainable GNN-derived structure–property relationships in interstitial-alloy materials.Research Square2024;

[56] Pagan DC,Benson AR.Graph neural network modeling of grain-scale anisotropic elastic behavior using simulated and measured microscale data.npj Comput Mater2022;8:952

[57] Hu G.AnisoGNN: graph neural networks generalizing to anisotropic properties of polycrystals.Comput Mater Sci2024;243:113121

[58] Hestroffer JM,Latypov MI.Graph neural networks for efficient learning of mechanical properties of polycrystals.Comput Mater Sci2023;217:111894

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

[60] Dai M,Liu X,Hu J.Graph neural network for predicting the effective properties of polycrystalline materials: a comprehensive analysis.Comput Mater Sci2023;230:112461

[61] Vlassis NN,Sun W.Geometric deep learning for computational mechanics Part I: anisotropic hyperelasticity.Comput Methods Appl Mech Eng2020;371:113299

[62] Hu Y,Lee M,Li D.A temporal graph neural network for cross-scale modelling of polycrystals considering microstructure interaction.Int J Plast2024;179:104017

[63] Thomas A,Alam M,Sack H.Materials fatigue prediction using graph neural networks on microstructure representations.Sci Rep2023;13:12562 PMCID:PMC10397301

[64] Karimi K,Mulewska K.Prediction of steel nanohardness by using graph neural networks on surface polycrystallinity maps.Scr Mater2023;234:115559

[65] Sim G,Latypov MI.FIP-GNN: graph neural networks for scalable prediction of grain-level fatigue indicator parameters.Scr Mater2025;255:116407

[66] Hansen CK,Hochhalter JD.Interpretable machine learning for microstructure-dependent models of fatigue indicator parameters.Int J Fatigue2024;178:108019

[67] Lupo Pasini, M.; Jung, G. S.; Irle, S. Graph neural networks predict energetic and mechanical properties for models of solid solution metal alloy phases.Comput Mater Sci2023;224:112141

[68] Lupo Pasini, M.; Burčul, M.; Reeve, S. T.; Eisenbach, M.; Perotto, S. Fast and accurate predictions of total energy for solid-solution alloys with graph convolutional neural networks. In Proceedings of the Smoky Mountains Computational Sciences & Engineering Conference (SMC 2021), Gatlinburg, USA. Aug 17-19 2021. Springer: Cham; 2021. pp. 79-98.

[69] Massa D,Naghdi A.Substitutional alloying using crystal graph neural networks.AIP Adv2024;14:015023

[70] Zhu D,Wu H.An optimized strategy for density prediction of intermetallics across varied crystal structures via graph neural network.J Mater Inf2025;5:8

[71] Shi X,Huang Y.Machine learning assisted screening of binary alloys for metal-based anode materials.J Energy Chem2025;104:62-8

[72] Qin Y,Radhakrishnan B.GrainGNN: a dynamic graph neural network for predicting 3D grain microstructure.J Comput Phys2024;510:113061

[73] Wang J,Oh S,Lee B.Interpretable and physics-informed modeling of solidification in alloy systems: a generalized framework for multi-component prediction.Acta Mater2025;286:120716

[74] Beaver N,Wong M.Rapid assessment of stable crystal structures in single-phase high-entropy alloys via graph neural network-based surrogate modelling.Crystals2024;14:1099

[75] Zhang H,Chen J,Chen W. Do graph neural networks work for high-entropy alloys? arXiv 2024, arXiv:2408.16337. https://doi.org/10.48550/arXiv.2408.16337. (accessed 31 Jul 2025)

[76] Long T,Pang B.An end-to-end explainable graph neural networks-based composition to mechanical properties prediction framework for bulk metallic glasses.Mech Mater2024;191:104945

[77] Yang Z.Linking atomic structural defects to mesoscale properties in crystalline solids using graph neural networks.npj Comput Mater2022;8:879

[78] Zhou J,Jiang X.Machine-learning-accelerated screening of Heusler alloys for nitrogen reduction reaction with graph neural network.Appl Surf Sci2024;669:160519

[79] Koker T,Spaeth W,Li L. Graph contrastive learning for materials. arXiv 2022, arXiv:2211.13408. https://doi.org/10.48550/arXiv.2211.13408. (accessed 31 Jul 2025)

[80] Park YJ,Hsu CW,Li J. Contrastive learning of English language and crystal graphs for multimodal representation of materials knowledge. arXiv 2025, arXiv:2502.16451. https://doi.org/10.48550/arXiv.2502.16451. (accessed 31 Jul 2025)

[81] Kipf TN. Variational graph auto-encoders. arXiv 2016, arXiv:1611.07308. https://doi.org/10.48550/arXiv.1611.07308. (accessed 31 Jul 2025)

[82] Hou Z,Cen Y.GraphMAE: self-supervised masked graph autoencoders. In Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery & Data Mining (KDD ’22), Washington, USA. Aug 14-18 2022. Association for Computing Machinery; 2022. pp. 594-604.

[83] Li J,Sun W. What’s behind the mask: understanding masked graph modeling for graph autoencoders. arXiv 2023, arXiv:2205.10053. https://doi.org/10.48550/arXiv.2205.10053. (accessed 31 Jul 2025)

[84] Xie T,Ganea OE,Jaakkola . Crystal diffusion variational autoencoder for periodic material generation. arXiv 2021, arXiv:2110.06197. https://doi.org/10.48550/arXiv.2110.06197. (accessed 31 Jul 2025)

[85] Hou Z,Cen Y. GraphMAE2: a decoding-enhanced masked self-supervised graph learner. arXiv 2023, arXiv:2304.04779. https://doi.org/10.48550/arXiv.2304.04779. (accessed 31 Jul 2025)

[86] Liu C,Zhan Y. Hi-GMAE: hierarchical graph masked autoencoders. arXiv 2024, arXiv:2405.10642. https://doi.org/10.48550/arXiv.2405.10642. (accessed 31 Jul 2025)

[87] Merchant A,Schoenholz SS,Cheon G.Scaling deep learning for materials discovery.Nature2023;624:80-5 PMCID:PMC10700131

[88] Zeni C,Zügner D.A generative model for inorganic materials design.Nature2025;639:624-32 PMCID:PMC11922738

[89] Gu GH,Noh J,Jung Y.Perovskite synthesizability using graph neural networks.npj Comput Mater2022;8:757

[90] Deng B,Jun K.CHGNet as a pretrained universal neural network potential for charge-informed atomistic modelling.Nat Mach Intell2023;5:1031-41

[91] Park CW,Vandermause J,Kozinsky B.Accurate and scalable graph neural network force field and molecular dynamics with direct force architecture.npj Comput Mater2021;7:543

[92] Vazquez G,Arróyave R.Deciphering chemical ordering in high entropy materials: a machine learning-accelerated high-throughput cluster expansion approach.Acta Mater2024;276:120137

[93] Wu S,Zhang H,Li J.Deep learning accelerates the discovery of two-dimensional catalysts for hydrogen evolution reaction.Energy Environ Mater2023;6:e12259

[94] Elrashidy A,Yan JA.Accelerated data-driven discovery and screening of two-dimensional magnets using graph neural networks.J Phys Chem C2024;128:6007-18

[95] Sibi H,Chowdhury C.Advancing 2D material predictions: superior work function estimation with atomistic line graph neural networks.RSC Adv2024;14:38070-8 PMCID:PMC11605676

[96] Rahman MH,Manganaris P.Accelerating defect predictions in semiconductors using graph neural networks.APL Mach Learn2024;2:016122

[97] Ehsan MT,Sarker A,Hasan MN. Graph neural-network framework for energy mapping of hybrid Monte-Carlo molecular-dynamics simulations of medium-entropy alloys. arXiv 2024, arXiv:2411.13670. https://doi.org/10.48550/arXiv.2411.13670. (accessed 31 Jul 2025)

[98] Tian Y,Myhill L.Data-driven modeling of dislocation mobility from atomistics using physics-informed machine learning.npj Comput Mater2024;10:1394

[99] Ghafarollahi A. Rapid and automated alloy design with graph neural-network-powered LLM-driven multi-agent systems. arXiv 2024, arXiv:2410.13768. https://doi.org/10.48550/arXiv.2410.13768. (accessed 31 Jul 2025)

[100] Qi C,Fu D,Zhou K.A hybrid knowledge-guided and data-driven method for predicting low-alloy steels performance.Comput Mater Sci2025;249:113602

[101] Wang B,Wang J. DR-Label: improving GNN models for catalysis systems by label deconstruction and reconstruction. arXiv 2023, arXiv:2303.02875. https://doi.org/10.48550/arXiv.2303.02875. (accessed 31 Jul 2025)

[102] Schütt K,Gastegger M. Equivariant message passing for the prediction of tensorial properties and molecular spectra. arXiv 2021, arXiv:2102.03150. https://doi.org/10.48550/arXiv.2102.03150. (accessed 31 Jul 2025)

[103] Satorras VG,Welling M. E(n) equivariant graph neural networks. arXiv 2021, arXiv:2102.09844. https://doi.org/10.48550/arXiv.2102.09844. (accessed 31 Jul 2025)

[104] Liang C,Ye C,Wang B.Material symmetry recognition and property prediction accomplished by crystal capsule representation.Nat Commun2023;14:5198 PMCID:PMC10457372

[105] Wen M,Munro JM,Persson KA.An equivariant graph neural network for the elasticity tensors of all seven crystal systems.Digit Discov2024;3:869-82