journal article·2024-10-03

Custom Hybrid Activation Function over Delocalization Error for Gap State Predictions of A2CeZrO6 (A = Ba2+, Sr2+, Ca2+, Mg2+) Proton Conductors: A First-Principles and Machine Learning Approach

D. Vignesh, Ela Rout

The Journal of Physical Chemistry C· Volume 128· Issue 39
Publisher page

Abstract

The crystal symmetry and band structure are two essential identifiers for a desirable proton-conducting electrolyte in fuel cell technology. However, structural and electrophysical inconsistency among cerium-based polymorphs originates due to multifacet issues ranging from cationic disorder (Ce 4+ → Ce 3+ ) to proton–polaron interaction. As a result, the distorted symmetry of Ca 2 CeZrO 6 and Mg 2 CeZrO 6 compounds displays an ambiguous trend in the forbidding gap due to differences in the spatial arrangement of atoms, chemical bonding, and hybridization. Besides material-driven limitations, the widely used density functional theory (DFT) simulations also face the band gap problem due to the lack of matrix element effects, final state effects, finite temperature offsets, and delocalization error by lower generation functionals (local density approximation (LDA) and generalized gradient approximation (GGA)). However, accurate estimations with higher rung functionals (hybrid (HSE06)) are often accompanied by critical limitations, including material-dependent response and exhaustive computations. Meanwhile, the sensitivity of an effective Hubbard parameter ( U eff ) within the semiempirical DFT+ U approach also limits their practical utility among complex oxides. As a result, we illustrate a mapping between the chemo-structural fingerprints and electronic band energy of distinct proton conductors using a state-of-the-art machine learning (ML) approach as a statistical learning and regression technique to combat the empirical approximations and derivative discontinuity of DFT simulations. In this study, we showcase the performance efficiency and limitations (training instability and generalization loss) of the designed artificial neural network (ANN) using a single activation function with a plausible solution via the custom hybrid activation function. According to our outcomes, the feature engineering of the hybrid activation function assists in overcoming numerical instability, ill conditionality, and dying gradient via adaptive neuro-controllers (tunable parameters) and nonlinear optimization of the loss landscape. As a result, the customized network was further used to model the potential energy surface (PES) of distinct proton conductors under their operating regimes (300–700 °C) to examine the proton landscape and transport characteristics with promising accuracy.
Source: openalex_inverted_index · Quality: baseline

Research context

OpenAlex topics: Machine Learning in Materials Science · X-ray Diffraction in Crystallography · Electronic and Structural Properties of Oxides

SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.

Affiliations

Birla Institute of Technology Mesra, Ranchi 835215, India

Funding

Department of Science and Technology, Ministry of Science and Technology, India

References (85)

Show source-provided references
1. Vignesh D. · 2023 · 1
2. 10.1016/j.ceramint.2022.01.10610.1016/j.ceramint.2022.01.106 ↗
3. 10.1016/j.cocom.2022.e0076310.1016/j.cocom.2022.e00763 ↗
4. 10.1016/j.matpr.2022.08.37310.1016/j.matpr.2022.08.373 ↗
5. 10.1039/C6CP04065A10.1039/C6CP04065A ↗
6. 10.1016/j.jpowsour.2019.22732710.1016/j.jpowsour.2019.227327 ↗
7. 10.1016/j.commatsci.2023.11258310.1016/j.commatsci.2023.112583 ↗
8. 10.1142/S0218625X0700949910.1142/S0218625X07009499 ↗
9. 10.1103/PhysRevB.99.16512010.1103/PhysRevB.99.165120 ↗
10. 10.1038/s41524-020-00360-010.1038/s41524-020-00360-0 ↗
11. 10.1063/5.013726410.1063/5.0137264 ↗
12. Neamen D. A. · 2011
13. Sze S. M. · 2021
14. 10.1021/acs.energyfuels.2c0392610.1021/acs.energyfuels.2c03926 ↗
15. 10.1103/PhysRevB.50.1795310.1103/PhysRevB.50.17953 ↗
16. 10.1103/PhysRevB.59.175810.1103/PhysRevB.59.1758 ↗
17. 10.1103/PhysRevApplied.10.02402610.1103/PhysRevApplied.10.024026 ↗
18. 10.1063/5.016081710.1063/5.0160817 ↗
19. 10.2166/wcc.2021.33610.2166/wcc.2021.336 ↗
20. 10.30534/ijeter/2020/5586202010.30534/ijeter/2020/55862020 ↗
21. Claesen, M.; Simm, J.; Popovic, D.; Moreau, Y.; De Moor, B. Easy Hyperparameter Search Using Optunity. arXiv, 2014. 10.48550/arXiv.1412.1114.
22. 10.1016/j.ssc.2022.11506210.1016/j.ssc.2022.115062 ↗
23. 10.30919/esmm5f104210.30919/esmm5f1042 ↗
24. 10.1103/PhysRevB.104.17410210.1103/PhysRevB.104.174102 ↗
25. 10.1038/s41467-023-43725-x10.1038/s41467-023-43725-x ↗
26. 10.1016/j.ceramint.2013.07.06110.1016/j.ceramint.2013.07.061 ↗
27. 10.1021/acs.energyfuels.2c0065010.1021/acs.energyfuels.2c00650 ↗
28. 10.1016/j.ssi.2021.11557110.1016/j.ssi.2021.115571 ↗
29. 10.1103/PhysRevB.59.1499210.1103/PhysRevB.59.14992 ↗
30. 10.1088/1367-2630/aac7f010.1088/1367-2630/aac7f0 ↗
31. 10.1103/PhysRevB.80.19510910.1103/PhysRevB.80.195109 ↗
32. 10.1103/PhysRevB.59.175810.1103/PhysRevB.59.1758 ↗
33. 10.1103/PhysRevB.57.150510.1103/PhysRevB.57.1505 ↗
34. 10.1021/acs.jctc.0c0037410.1021/acs.jctc.0c00374 ↗
35. 10.1103/PhysRevB.103.07510910.1103/PhysRevB.103.075109 ↗
36. 10.1016/j.jssc.2022.12341910.1016/j.jssc.2022.123419 ↗
37. 10.1088/0953-8984/9/4/00210.1088/0953-8984/9/4/002 ↗
38. 10.1103/PhysRevB.73.19510710.1103/PhysRevB.73.195107 ↗
39. 10.1103/PhysRevB.77.11512310.1103/PhysRevB.77.115123 ↗
40. 10.1103/PhysRev.140.A113310.1103/PhysRev.140.A1133 ↗
41. 10.1103/PhysRevB.26.544510.1103/PhysRevB.26.5445 ↗
42. 10.1103/PhysRevB.79.15510710.1103/PhysRevB.79.155107 ↗
43. 10.1103/PhysRevB.76.02430910.1103/PhysRevB.76.024309 ↗
44. 10.1080/14786435.2017.132813310.1080/14786435.2017.1328133 ↗
45. 10.48550/arXiv.1808.0519610.48550/arXiv.1808.05196 ↗
46. 10.1063/1.489045810.1063/1.4890458 ↗
47. Rappoport J. D. · Comput. Inorg. Bioinorg. Chem. · 2008 · 594 · 1
48. 10.1103/PhysRevResearch.2.01321410.1103/PhysRevResearch.2.013214 ↗
49. 10.1103/PhysRevB.79.17410710.1103/PhysRevB.79.174107 ↗
50. 10.1016/S0167-2738(97)00085-410.1016/S0167-2738(97)00085-4 ↗
51. 10.1016/S0167-2738(99)00070-310.1016/S0167-2738(99)00070-3 ↗
52. 10.1088/1361-648X/abc40710.1088/1361-648X/abc407 ↗
53. 10.1039/D0CS01074J10.1039/D0CS01074J ↗
54. 10.1063/1.367208010.1063/1.3672080 ↗
55. 10.1016/j.cplett.2007.04.02010.1016/j.cplett.2007.04.020 ↗
56. 10.1021/acs.jpclett.8b0091910.1021/acs.jpclett.8b00919 ↗
57. 10.1103/PhysRevMaterials.2.07380310.1103/PhysRevMaterials.2.073803 ↗
58. 10.1016/j.trechm.2020.02.00510.1016/j.trechm.2020.02.005 ↗
59. 10.1038/s41524-023-01000-z10.1038/s41524-023-01000-z ↗
60. 10.1016/j.neucom.2020.07.06110.1016/j.neucom.2020.07.061 ↗
61. 10.1016/j.ijleo.2020.16446910.1016/j.ijleo.2020.164469 ↗
62. 10.12928/telkomnika.v14i4.395610.12928/telkomnika.v14i4.3956 ↗
63. 10.1007/s00521-020-05582-310.1007/s00521-020-05582-3 ↗
64. 10.1007/s44196-023-00186-w10.1007/s44196-023-00186-w ↗
65. 10.1504/IJCSE.2020.10686610.1504/IJCSE.2020.106866 ↗
66. 10.1007/s11042-019-08453-910.1007/s11042-019-08453-9 ↗
67. 10.1016/j.jocs.2020.10110710.1016/j.jocs.2020.101107 ↗
68. Masters, D.; Luschi, C. Revisiting Small Batch Training for Deep Neural Networks arXiv 2018, pp 1–18 https://arxiv.org/abs/1804.07612.
69. Ortega, L. A.; Cabañas, R.; Masegosa, A. R.Diversity and Generalization in Neural Network Ensembles. In International Conference on Artificial Intelligence and Statistics 2022; pp 11720–11743.
70. Salman, S.; Liu, X.Overfitting Mechanism and Avoidance in Deep Neural Networks arXiv 2019 https://arxiv.org/abs/1901.06566.
71. Johnson, R.; Zhang, T. Inconsistency, Instability, and Generalization Gap of Deep Neural Network Training. In Advances in Neural Information Processing Systems; Curran Associates, Inc., 2023.
72. 10.1109/TNNLS.2021.310604410.1109/TNNLS.2021.3106044 ↗
73. 10.1016/j.ifacol.2020.12.134210.1016/j.ifacol.2020.12.1342 ↗
74. 10.1016/j.neucom.2022.06.11110.1016/j.neucom.2022.06.111 ↗
75. Ramachandran, P.; Le, Q. V. Diversity and Depth in per-Example Routing Models. In International Conference on Learning Representations, 2018.
76. Lu, L.; Shin, Y.; Su, Y.; Karniadakis, G. E. Dying Relu and Initialization: Theory and Numerical Examples. arXiv Prepr. arXiv1903.06733, 2019. https://arxiv.org/abs/1903.06733.
77. 10.1109/72.80909510.1109/72.809095 ↗
78. 10.1109/TAP.2021.313853510.1109/TAP.2021.3138535 ↗
79. 10.1080/23311916.2018.144434510.1080/23311916.2018.1444345 ↗
80. 10.1021/acs.jpcc.1c0071810.1021/acs.jpcc.1c00718 ↗
81. 10.1021/acs.jpcc.7b0742110.1021/acs.jpcc.7b07421 ↗
82. 10.1021/acs.jpcc.0c0459410.1021/acs.jpcc.0c04594 ↗
83. 10.1038/ncomms1583010.1038/ncomms15830 ↗
84. 10.1016/j.ijhydene.2024.01.04710.1016/j.ijhydene.2024.01.047 ↗
85. Vignesh, D.; Rout, E. Defect Scattering and Thermionic Landscape of Acceptor Doped BaCeO3 Polymorph for Intermediate Temperature Fuel Cell Technology. Ceram. Int. 2024, No. February. 10.1016/j.ceramint.2024.02.214.10.1016/j.ceramint.2024.02.214 ↗