CoChat
Sign Up
Cited by 0Machine Learning in Materials ScienceModel Reduction and Neural NetworksNeural Networks and Applications

The DOI matches its Crossref record; 12 of the 12 references it cites were found in a registry or an index.

Modern machine learning in thin film device development

Open access
Dobrynin, A., Khaydukov, Y., Gubbins, M.·2026·Chemical Physics Reviews
DOI 10.1063/5.0231166

Published version of a preprint

Open this paper in CoChat — free account

Abstract

Recent advances in deep learning and scientific machine learning (SML) offer a wide spectrum of novel tools with the potential to radically transform various areas of academic research, industrial research and development (R&D), and manufacturing. These include inverse device design, data-driven equation discovery, rapid approximate partial solutions to differential equations for exploring the design space, automated metrology and manufacturing process control, new materials design, and the creation of fast and realistic digital twins for virtual testing and design optimization. The purpose of this review is to assist R&D practitioners, who are not specialists in artificial intelligence, in navigating this complex and dynamic landscape, enabling them to adopt modern machine learning (ML) methods in their work. We particularly emphasize the potential advantages of deep learning methods for the field of thin film device development, highlighting the main approaches and points of their applications in R&D design and process. The review is organized into several sections. First, we provide a brief overview of machine learning and deep learning, introducing basic neural network architectures and describing their possible use cases relevant to industrial R&D. In Sec. III, we introduce examples of ML approaches enabling the reduction of dependency on the amount of input data and improving generalization capabilities of neural networks through introduction of realistic inductive biases in the form of symmetries, conservation laws, physics equations, etc. Then we review some of the most successful large-scale SML models, including foundational materials simulation and generation models. Finally, we discuss existing and prospective applications of ML models in different aspects of thin film devices development.

Overview

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1063/5.0231166

Cites

12 of 12 references verified0 citation errors · 0 not checkable

Statistics

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Research works printed but not linked: 95

  1. 1997, Machine Learning
  2. R. Balestriero , M.Ibrahim, V.Sobal, A.Morcos, S.Shekhar, T.Goldstein, F.Bordes, A.Bardes, G.Mialon, Y.Tian, A.Schwarzschild, A. G.Wilson, J.Geiping, Q.Garrido, P.Fernandez, A.Bar, H.Pirsiavash, Y.LeCun, and M.Goldblum, “A cookbook of self-supervised learning,” arXiv:2304.12210 (2023). · arXiv
  3. 2009, The Elements of Statistical Learning: Data Mining, Inference, and Prediction
  4. 2022, Deep learning, reinforcement learning, and world models, Neural Networks, 152, 267, 10.1016/j.neunet.2022.03.037 · DOI
  5. 2006, Pattern Recognition and Machine Learning
  6. 1998, Reinforcement Learning: An Introduction
  7. 2019, Data-Driven Science and Engineering: Machine Learning, Dynamical Systems, and Control
  8. 1959, Some studies in machine learning using the game of checkers, IBM J. Res. Dev., 3, 210, 10.1147/rd.33.0210 · DOI · Paper page
  9. 1967, Some studies in machine learning using the game of checkers. II—Recent progress, IBM J. Res. Dev., 11, 601, 10.1147/rd.116.0601 · DOI
  10. 2018, The Deep Learning Revolution: Artificial Intelligence Meets Human Intelligence
  11. 1969, Perceptrons: An Introduction to Computational Geometry
  12. 1986, Learning representations by back-propagating errors, Nature, 323, 533, 10.1038/323533a0 · DOI · Paper page
  13. 2016, Deep Learning
  14. A. Vaswani , N.Shazeer, N.Parmar, J.Uszkoreit, L.Jones, A. N.Gomez, Ł.Kaiser, and I.Polosukhin, “Attention is all you need,” arXiv:1706.03762 (2017). · arXiv
  15. 1989, Multilayer feedforward networks are universal approximators, Neural Networks, 2, 359, 10.1016/0893-6080(89)90020-8 · DOI · Paper page
  16. 2017, Why does deep learning work? A physics perspective, J. Stat. Phys., 168, 1223, 10.48550/arXiv.1608.08225 · DOI
  17. 2019, Deep learning for audio signal processing, IEEE J. Sel. Top. Signal Process., 13, 206, 10.1109/JSTSP.2019.2908700 · DOI
  18. S. Bai , J. Z.Kolter, and V.Koltun, “An empirical evaluation of generic convolutional and recurrent networks for sequence modeling,” arXiv:1803.01271 (2018). · arXiv
  19. 1990, Finding structure in time, Cognit. Sci., 14, 179, 10.1207/s15516709cog1402_1 · DOI · Paper page
  20. 2023, What are graph neural networks?
  21. D. P. Kingma and M.Welling, “Auto-encoding variational bayes,” arXiv:1312.6114 (2022). · arXiv
  22. 2019, An introduction to variational autoencoders, Foundations and Trends in Machine Learning, 12, 307, 10.1561/2200000056 · DOI · Paper page
  23. I. Goodfellow , J.Pouget-Abadie, M.Mirza, B.Xu, D.Warde-Farley, S.Ozair, A.Courville, and Y.Bengio, “Generative adversarial networks,” arXiv:1406.2661 (2014). · arXiv
  24. 2023, Generative adversarial network-based fault detection in semiconductor equipment with class-imbalanced data, Sensors, 23, 1889, 10.3390/s23041889 · DOI
  25. L. Yang , Y.Zhang, Y.Hong, M.Xu, Y.Wang, Y.Wang, Y.Xu, C.Xu, and D.Tao, “Diffusion models: A comprehensive survey of methods and applications,” arXiv:2209.00796 (2022). · arXiv
  26. 2020, Mastering Atari, Go, chess and shogi by planning with a learned model, Nature, 588, 604, 10.1038/s41586-020-03051-4 · DOI · Paper page
  27. Konidaris, Welcome to the era of experience
  28. 2022, Magnetic control of tokamak plasmas through deep reinforcement learning, Nature, 602, 414, 10.1038/s41586-021-04301-9 · DOI · Paper page
  29. R. T. Q. Chen , Y.Rubanova, J.Bettencourt, and D.Duvenaud, “Neural ordinary differential equations,” arXiv:1806.07366 (2019). · arXiv
  30. Z. Liu , Y.Wang, S.Vaidya, F.Ruehle, J.Halverson, M.Soljačić, T. Y.Hou, and M.Tegmark, “KAN: Kolmogorov–Arnold networks,” arXiv:2404.19756 (2025). · DOI
  31. 2021, Physics-informed machine learning, Nat. Rev. Phys., 3, 422, 10.1038/s42254-021-00314-5 · DOI · Paper page
  32. 2019, Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations, J. Comput. Phys., 378, 686, 10.48550/arXiv.1711.10561 · DOI
  33. L. Fesser , L.D'Amico-Wong, and R.Qiu, “Understanding and mitigating extrapolation failures in physics-informed neural networks,” arXiv:2306.09478 (2023). · arXiv
  34. 2021, Learning nonlinear operators via DeepONet based on the universal approximation theorem of operators, Nat. Mach. Intell., 3, 218, 10.1038/s42256-021-00302-5 · DOI
  35. Z. Li , N.Kovachki, K.Azizzadenesheli, B.Liu, K.Bhattacharya, A.Stuart, and A.Anandkumar, “Fourier neural operator for parametric partial differential equations,” arXiv:2010.08895 (2020). · arXiv
  36. C. Rackauckas , Y.Ma, J.Martensen, C.Warner, K.Zubov, R.Supekar, D.Skinner, A.Ramadhan, and A.Edelman, “Universal differential equations for scientific machine learning,” arXiv:2001.04385 (2020). · DOI · Paper page
  37. 2022, Graph neural networks for materials science and chemistry, Commun. Mater., 3, 93, 10.1038/s43246-022-00315-6 · DOI
  38. 2023, Graph neural networks for molecules, 21
  39. M. M. Bronstein , J.Bruna, T.Cohen, and P.Veličković, “Geometric deep learning: Grids, groups, graphs, geodesics, and gauges,” arXiv:2104.13478 (2021). · arXiv
  40. P. W. Battaglia , J. B.Hamrick, V.Bapst, A.Sanchez-Gonzalez, V.Zambaldi, M.Malinowski, A.Tacchetti, D.Raposo, A.Santoro, R.Faulkner, C.Gulcehre, F.Song, A.Ballard, J.Gilmer, G.Dahl, A.Vaswani, K.Allen, C.Nash, V.Langston, C.Dyer, N.Heess, D.Wierstra, P.Kohli, M.Botvinick, O.Vinyals, Y.Li, and R.Pascanu, “Relational inductive biases, deep learning, and graph networks,” arXiv:1806.01261 (2018). · arXiv
  41. M. Geiger , T.Smidt, B. K.Miller, W. M. C.Foulkes, K. L. R.Morgan, J. T. P.Overduin, J. W. J.Saukko, M.Korpela, A.Rizzi, and S.Hey, “e3nn: Euclidean neural networks,” arXiv:2207.09453 (2022). · arXiv
  42. 2020, Machine learning for interatomic potential models, J. Chem. Phys., 152, 050902, 10.1063/1.5126336 · DOI
  43. T. Le , F.Noé, and D.-A.Clevert, “Equivariant graph attention networks for molecular property prediction,” arXiv:2202.09891 (2022). · arXiv
  44. F. A. Hölzl , D.Rueckert, and G.Kaissis, “Bridging the gap: Differentially private equivariant deep learning for medical image analysis,” arXiv:2209.04338 (2023). · arXiv
  45. S.-O. Kaba and S.Ravanbakhsh, “Equivariant networks for crystal structures,” arXiv:2211.15420 (2022). · DOI
  46. 2018, Crystal graph convolutional neural networks for an accurate and interpretable prediction of material properties, Phys. Rev. Lett., 120, 145301, 10.1103/PhysRevLett.120.145301 · DOI · Paper page
  47. 2021, Atomistic line graph neural network for improved materials property predictions, npj Comput. Mater., 7, 185, 10.1038/s41524-021-00650-1 · DOI
  48. 2022, A generative approach to materials discovery, design, and optimization, ACS Omega, 7, 25958, 10.1021/acsomega.2c03264 · DOI
  49. 2024, Physics-informed data augmentation of experimental erosion data through generative adversarial networks, 10.1115/FEDSM2024-130858 · DOI
  50. 2024, Deep reinforcement learning for inverse inorganic materials design, npj Comput. Mater., 10, 287, 10.1038/s41524-024-01474-5 · DOI
  51. 2023, A newcomer's guide to deep learning for inverse design in nano-photonics, Nanophotonics, 12, 4387, 10.1515/nanoph-2023-0527 · DOI
  52. 2000, Gene ontology: Tool for the unification of biology, Nat. Genet., 25, 25, 10.1038/75556 · DOI · Paper page
  53. 2023, Uncovering new families and folds in the natural protein universe, Nature, 622, 646, 10.1038/s41586-023-06622-3 · DOI
  54. J. Gasteiger , J.Groß, and S.Günnemann, “Directional message passing for molecular graphs,” arXiv:2003.03123 (2022). · arXiv
  55. 2013, Commentary: The materials project: A materials genome approach to accelerating materials innovation, APL Mater., 1, 011002, 10.1063/1.4812323 · DOI · Paper page
  56. 2025, Accelerated data-driven materials science with the Materials Project, Nat. Mater., 24, 1522, 10.1038/s41563-025-02272-0 · DOI
  57. H. Yang , C.Hu, Y.Zhou, X.Liu, Y.Shi, J.Li, G.Li, Z.Chen, C.Shuizhou, C.Zeni, M.Horton, R.Pinsler, A.Fowler, D.Zügner, T.Xie, J.Smith, L.Sun, Q.Wang, L.Kong, C.Liu, H.Hao, and Z.Lu, “MatterSim: A deep learning atomistic model across elements, temperatures and pressures,” arXiv:2405.04967v2 (2024). · arXiv
  58. C. Zeni , R.Pinsler, D.Zügner, A.Fowler, M.Horton, X.Fu, S.Shysheya, J.Crabbé, L.Sun, J.Smith, S.Erhard, T.Greaves, C.Liu, H.Qian, N.Thiry, D.Gunceler, M.Riechert, J.Li, H.Hao, Z.Lu, R.Tomioka, and T.Xie, “MatterGen: A generative model for inorganic materials design,” arXiv:2312.03687 (2023). · arXiv
  59. 2025, A generative model for inorganic materials design, Nature, 639, 624, 10.1038/s41586-025-08628-5 · DOI · Paper page
  60. 2022, A universal graph deep learning interatomic potential for the periodic table, Nat. Comput. Sci., 2, 718, 10.1038/s43588-022-00349-3 · DOI · Paper page
  61. Y. Shi , S.Zheng, G.Ke, Y.Shen, J.You, J.He, S.Luo, C.Liu, D.He, and T.-Y.Liu, “Benchmarking Graphormer on large-scale molecular modeling datasets,” arXiv:2203.04810 (2022). · arXiv
  62. 2018, The Alexandria library, a quantum-chemical database of molecular properties for force field development, Sci. Data, 5, 180062, 10.1038/sdata.2018.62 · DOI
  63. 2023, Accurate medium-range global weather forecasting with 3D neural networks, Nature, 619, 533, 10.1038/s41586-023-06185-3 · DOI · Paper page
  64. D. Li , Z.Wang, Y.Chen, R.Jiang, W.Ding, and M.Okumura, “A survey on deep active learning: Recent advances and new frontiers,” arXiv:2405.00334 (2024). · arXiv
  65. 2023, Efficient wafer defect patterns recognition using deep convolutional neural network, 220, 10.1109/cai54212.2023.00102 · DOI
  66. 2024, Automated industrial load station inspection, 10.1109/IST63414.2024.10759268 · DOI
  67. 2025, Predictive quality of in-fabrication products in smart manufacturing using graph-based deep learning, 145, 10.5220/0013855100003982 · DOI
  68. 2023, Wafer map failure pattern classification using geometric transformation-invariant convolutional neural network, Sci. Rep., 13, 8127, 10.1038/s41598-023-34147-2 · DOI
  69. 2020, Quantifying defects in thin films using machine vision, npj Comput. Mater., 6, 111, 10.1038/s41524-020-00380-w · DOI
  70. 2020, A general approach to maximise information density in neutron reflectometry analysis, Mach. Learn., 1, 035002, 10.1088/2632-2153/ab94c4 · DOI
  71. 2021, Towards automated analysis for neutron reflectivity, Mach. Learn., 2, 035006, 10.1088/2632-2153/abe7b5 · DOI
  72. 2021, Towards reflectivity profile inversion through artificial neural networks, Mach. Learn., 2, 025034, 10.1088/2632-2153/abe564 · DOI
  73. 2021, Machine learning for neutron reflectometry data analysis of two-layer thin films, Mach. Learn., 2, 035001, 10.1088/2632-2153/abf257 · DOI
  74. 2021, Neural network analysis of neutron and x-ray reflectivity data: Pathological cases, performance and perspectives, Mach. Learn., 2, 045003, 10.1088/2632-2153/abf9b1 · DOI
  75. P. S. Savchenkov , K. V.Nikolaev, V. I.Bodnarchuk, A. N.Pirogov, A. V.Belushkin, and S. N.Yakunin, “Multimodal reconstruction of TbCo thin film structure with Bayesian analysis of polarized neutron reflectivity,” arXiv:2405.01243 (2024). · DOI
  76. 2018, A hybrid multiscale modelling framework for predictive simulations and optimization of thin film deposition processes, Comput. Chem. Eng., 119, 465, 10.1016/j.compchemeng.2018.08.029 · DOI
  77. 2019, Integrated multiscale modeling and control of atomic layer deposition processes: Linear parameter-varying and neural network-based approaches, Chem. Eng. Res. Des., 151, 131, 10.1016/j.cherd.2019.09.005 · DOI
  78. 2019, Machine-learning-assisted thin-film growth: Bayesian optimization in molecular beam epitaxy of SrRuO3 thin films, APL Mater., 7, 101114, 10.1063/1.5123019 · DOI
  79. 2020, Quantified structure zone diagram for sputtered nitride thin films enabled through generative machine learning, Commun. Mater., 1, 15, 10.1038/s43246-020-0017-2 · DOI
  80. 2025, Quantitative analysis of the prediction performance of a convolutional neural network evaluating the surface elastic energy of a strained film, Comput. Mater. Sci., 249, 113657, 10.1016/j.commatsci.2024.113657 · DOI
  81. 2025, Data science of the in silico crystallization, Acta Mater., 287, 120762, 10.1016/j.actamat.2025.120762 · DOI
  82. 2025, Morphology prediction for polymer blend thin films using machine learning, Soft Matter, 21, 5284, 10.1039/D5SM00335K · DOI
  83. 2025, Autonomous closed-loop exploration of composition-spread films for the anomalous Hall effect, npj Comput. Mater., 11, 329, 10.1038/s41524-025-01828-7 · DOI
  84. 2025, High-throughput materials exploration system for the anomalous hall effect using combinatorial experiments and machine learning, npj Comput. Mater., 11, 269, 10.1038/s41524-025-01757-5 · DOI
  85. 2021, Tackling photonic inverse design with machine learning, Adv. Sci., 8, 2002923, 10.1002/advs.202002923 · DOI
  86. 2024, Machine learning surrogate for 3D phase-field modeling of ferroelectric tip-induced electrical switching, npj Comput. Mater., 10, 197, 10.1038/s41524-024-01375-7 · DOI
  87. 2021, Self-supervised learning and prediction of microstructure evolution with convolutional recurrent neural networks, Patterns, 2, 100243, 10.1016/j.patter.2021.100243 · DOI
  88. 2021, Accelerating phase-field-based microstructure evolution predictions via surrogate models trained by machine learning methods, npj Comput. Mater., 7, 3, 10.1038/s41524-020-00471-8 · DOI
  89. 2023, Prediction of microstructure evolution at the atomic scale by deep generative model in combination with recurrent neural networks, Acta Mater., 259, 119295, 10.1016/j.actamat.2023.119295 · DOI
  90. 2024, Digital Twins, Simulation, and the Metaverse
  91. 2022, GraphVAMPNet, using graph neural networks and variational approach to Markov processes for dynamical modeling of biomolecules, J. Chem. Phys., 156, 184103, 10.1063/5.0085607 · DOI
  92. 2007, Giant tunnel magnetoresistance in magnetic tunnel junctions with a crystalline MgO (0 0 1) barrier, J. Phys. D, 40, R337, 10.1088/0022-3727/40/21/R01 · DOI
  93. 2018, Matminer: An open source toolkit for materials data mining, Comput. Mater. Sci., 152, 60, 10.1016/j.commatsci.2018.05.018 · DOI
  94. 2020, Maml - materials machine learning package, GitHub.
  95. 2019, Deep Learning for the Life Sciences

Metadata

Authors
Dobrynin, A., Khaydukov, Y., & Gubbins, M.
Affiliations
Seagate (Ireland) (IE)
Corresponding author
Dobrynin, A.
ORCID
Venue
Chemical Physics Reviews
Publisher
AIP Publishing
Published
2026-03-01
Volume
7
Issue
1
ISSN
2688-4070 (ISSN-L)
DOI
10.1063/5.0231166
OpenAlex
W7138329027
Cited by
0
References
12
FWCI
0.00
Type
journal article
Language
en
Open access
Locations
Chemical Physics Reviews (publishedVersion)
Topics
  • Physical Sciences › Materials Science › Materials Chemistry › Machine Learning in Materials Science
  • Physical Sciences › Physics and Astronomy › Statistical and Nonlinear Physics › Model Reduction and Neural Networks
  • Physical Sciences › Computer Science › Artificial Intelligence › Neural Networks and Applications
Keywords
thin film devices, scientific machine learning, deep learning, physics-informed neural networks, inductive bias, materials design, digital twin, manufacturing process control, neural network architecture

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review