pith. sign in

arxiv: 2605.03081 · v1 · submitted 2026-05-04 · ❄️ cond-mat.mtrl-sci

Building a physics-aware AI ecosystem for solid-state hydrogen storage materials

Seong-Hoon Jang , Yiwen Yao , Chuanyu Liu , Linda Zhang , Di Zhang , Xue Jia , Hung Ba Tran , Eric Jianfeng Cheng
show 62 more authors
Ryuhei Sato Yusuke Ohashi Toyoto Sato Yusuke Hashimoto Mark Allendorf Nongnuch Artrith Marcello Baricco Andreas Borgschulte Darren P. Broom Ang Cao Benjamin W. J. Chen Lixin Chen Ping Chen Eun Seon Cho Stefano Deledda Zhao Ding Martin Dornheim Michael Felderhoff Yaroslav Filinchuk George E. Froudakis Mingxia Gao Thomas Gennett Zaiping Guo Ikutaro Hamada Jason Hattrick-Simpers Bj{\o}rn C. Hauback Michael Hirscher Torben R. Jensen Baohua Jia Hyoung Seop Kim Takahiro Kondo Kentaro Kutsukake Xiao-Yan Li Tongliang Liu Piao Ma Jianfeng Mao Rana Mohtadi Hyunchul Oh Mark Paskevicius Chris J. Pickard Astrid Pundt Long Qi Anibal Ramirez-Cuesta Hiroyuki Saitoh Kaihang Shi Aloysius Soon Chenghua Sun Chris Wolverton Hiroshi Yabu Weijie Yang Zhenpeng Yao Xuebin Yu Jianxin Zou Shouyi Hu Panpan Zhou Xi Lin Zhigang Hu Zhenhao Zhou Pengfei Ou Jiayu Peng Shin-ichi Orimo Hao Li
This is my paper

Pith reviewed 2026-05-08 17:39 UTC · model grok-4.3

classification ❄️ cond-mat.mtrl-sci
keywords hydrogen storage materialsphysics-informed AIclosed-loop discoverydigital twininverse designmaterials optimizationexperimental feedback
0
0 comments X p. Extension

The pith

A unified framework embeds physical constraints and experimental feedback into AI for adaptive hydrogen storage materials discovery.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Hydrogen storage is limited by complex multiscale physics in thermodynamics, kinetics, and material structure, while existing AI methods suffer from scattered data and inconsistent physical predictions. The paper proposes a single framework that links data infrastructure, physics-based models, and AI inverse design inside a closed loop that feeds back experimental results. This setup lets the system adjust designs while staying true to physical laws. A reader would care because it sketches a route to faster, more reliable finding of materials that could make hydrogen energy practical.

Core claim

The paper claims that integrating coherent data infrastructure, physics-grounded modeling, and AI-driven inverse design inside a closed-loop paradigm, with physical constraints and experimental feedback embedded, produces adaptive and physically consistent optimization and thereby opens a route to autonomous, digital-twin-enabled discovery of hydrogen storage materials.

What carries the argument

Closed-loop discovery paradigm that unifies coherent data infrastructure, physics-grounded modeling, and AI-driven inverse design while embedding physical constraints and experimental feedback.

If this is right

  • Optimization of hydrogen storage materials becomes adaptive and remains consistent with physical laws.
  • Autonomous discovery of new storage materials becomes feasible.
  • Digital-twin representations of the materials can be maintained and updated with real experiments.
  • The same closed-loop structure can be reused for other multiscale materials problems.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the loop works, it would shorten the cycle from computational prediction to lab-verified material by orders of magnitude.
  • The same structure could be tested on battery electrode or catalyst discovery where similar data and physics gaps exist.
  • Success would require public sharing of the data infrastructure so other groups can plug in their own experimental results.

Load-bearing premise

Fragmented data, limited physical consistency, and weak experimental integration can be fixed simply by creating one unified framework, even without detailing the exact mechanisms or showing prior success.

What would settle it

A side-by-side laboratory test in which materials proposed by the physics-aware AI system show no better match to measured thermodynamics, kinetics, or microstructures than materials found by conventional or non-physics AI methods would disprove the central claim.

Figures

Figures reproduced from arXiv: 2605.03081 by Aloysius Soon, Andreas Borgschulte, Ang Cao, Anibal Ramirez-Cuesta, Astrid Pundt, Baohua Jia, Benjamin W. J. Chen, Bj{\o}rn C. Hauback, Chenghua Sun, Chris J. Pickard, Chris Wolverton, Chuanyu Liu, Darren P. Broom, Di Zhang, Eric Jianfeng Cheng, Eun Seon Cho, George E. Froudakis, Hao Li, Hiroshi Yabu, Hiroyuki Saitoh, Hung Ba Tran, Hyoung Seop Kim, Hyunchul Oh, Ikutaro Hamada, Jason Hattrick-Simpers, Jianfeng Mao, Jianxin Zou, Jiayu Peng, Kaihang Shi, Kentaro Kutsukake, Linda Zhang, Lixin Chen, Long Qi, Marcello Baricco, Mark Allendorf, Mark Paskevicius, Martin Dornheim, Michael Felderhoff, Michael Hirscher, Mingxia Gao, Nongnuch Artrith, Panpan Zhou, Pengfei Ou, Piao Ma, Ping Chen, Rana Mohtadi, Ryuhei Sato, Seong-Hoon Jang, Shin-ichi Orimo, Shouyi Hu, Stefano Deledda, Takahiro Kondo, Thomas Gennett, Tongliang Liu, Torben R. Jensen, Toyoto Sato, Weijie Yang, Xiao-Yan Li, Xi Lin, Xuebin Yu, Xue Jia, Yaroslav Filinchuk, Yiwen Yao, Yusuke Hashimoto, Yusuke Ohashi, Zaiping Guo, Zhao Ding, Zhenhao Zhou, Zhenpeng Yao, Zhigang Hu.

Figure 1
Figure 1. Figure 1: Data infrastructure for solid-state hydrogen storage materials. (a) view at source ↗
Figure 2
Figure 2. Figure 2: Machine-learning interatomic potentials (MLIPs)-enabled multiscale modeling for view at source ↗
Figure 3
Figure 3. Figure 3: The dual-stream generative model (GM)-large language model (LLM) architecture and agent-driven iterative inverse design for the constrained discovery of high-performance hydrogen storage materials. (a) GMs: Property-Guided Structural Design. Generative models (e.g., MatterGen) navigate the continuous latent space to propose novel atomic configurations view at source ↗
Figure 4
Figure 4. Figure 4: Closed-loop discovery framework for hydrogen storage materials. (a) view at source ↗
read the original abstract

Hydrogen storage remains a central bottleneck for scalable hydrogen energy systems due to the multiscale and coupled nature of the thermodynamics, kinetics, and microstructural evolution of hydrogen storage materials (HSMs). Although artificial intelligence (AI) has accelerated materials discovery, current approaches remain constrained by fragmented data, limited physical consistency, and weak integration with experimental validation. Here, we propose a unified framework that integrates coherent data infrastructure, physics-grounded modeling, and AI-driven inverse design within a closed-loop discovery paradigm. By embedding physical constraints and experimental feedback, this approach enables adaptive, physically consistent optimization, thereby establishing a pathway toward autonomous, digital-twin-enabled discovery of HSMs.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 1 minor

Summary. The manuscript proposes a unified 'physics-aware AI ecosystem' for solid-state hydrogen storage materials (HSMs) that integrates coherent data infrastructure, physics-grounded modeling, and AI-driven inverse design inside a closed-loop discovery paradigm. By embedding physical constraints and experimental feedback, the framework is claimed to overcome fragmented data, limited physical consistency, and weak experimental integration, thereby enabling adaptive optimization and a pathway to autonomous, digital-twin-enabled HSM discovery.

Significance. A successfully implemented version of the proposed ecosystem could meaningfully advance AI-assisted materials discovery for hydrogen storage by enforcing physical consistency and closing the experiment–model loop. The conceptual integration of data, physics, and inverse design is timely given the multiscale challenges in HSMs, but the manuscript supplies no concrete mechanisms, equations, architectures, or validation to demonstrate feasibility.

major comments (3)
  1. [Abstract] Abstract: the central claim that embedding physical constraints and experimental feedback 'enables adaptive, physically consistent optimization' and 'establishes a pathway toward autonomous... discovery' is unsupported; the manuscript provides neither the thermodynamic/kinetic equations with embedded constraints nor the closed-loop update protocol (e.g., how experimental data revise model parameters or loss terms).
  2. [Framework description (main text)] Framework description (main text): no data schemas, sources, or curation protocols are specified for the 'coherent data infrastructure,' nor are any ML architectures, physics-informed loss functions, or inverse-design algorithms described that would enforce physical consistency.
  3. [Abstract and main text] Abstract and main text: the assertion that the approach resolves 'fragmented data... and weak integration with experimental validation' lacks any concrete mechanism (e.g., active-learning loop, digital-twin interface, or benchmark against existing HSM datasets) or preliminary result demonstrating prior success in analogous systems.
minor comments (1)
  1. The manuscript would benefit from at least one concrete example (even schematic) of how a physics constraint would be imposed on an HSM property prediction or how feedback from a synthesis experiment would update the AI model.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the detailed and constructive feedback on our manuscript proposing a physics-aware AI ecosystem for solid-state hydrogen storage materials. We address each major comment below, clarifying the conceptual scope of the work and indicating where revisions have been made to the manuscript.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central claim that embedding physical constraints and experimental feedback 'enables adaptive, physically consistent optimization' and 'establishes a pathway toward autonomous... discovery' is unsupported; the manuscript provides neither the thermodynamic/kinetic equations with embedded constraints nor the closed-loop update protocol (e.g., how experimental data revise model parameters or loss terms).

    Authors: The manuscript presents a high-level conceptual framework rather than a detailed implementation study. The claims summarize the intended benefits of the integrated components described in the main text. We have revised the abstract to employ more measured language (e.g., 'is designed to enable') and added a brief discussion of how physical constraints might be incorporated via physics-informed approaches. revision: partial

  2. Referee: [Framework description (main text)] Framework description (main text): no data schemas, sources, or curation protocols are specified for the 'coherent data infrastructure,' nor are any ML architectures, physics-informed loss functions, or inverse-design algorithms described that would enforce physical consistency.

    Authors: We agree that greater specificity on the data infrastructure strengthens the presentation. The revised manuscript now references established data schemas (e.g., those aligned with the Materials Genome Initiative) and high-level curation protocols based on FAIR principles. Specific ML architectures and loss functions remain outside the scope of this perspective paper, as the focus is on the overall ecosystem architecture. revision: yes

  3. Referee: [Abstract and main text] Abstract and main text: the assertion that the approach resolves 'fragmented data... and weak integration with experimental validation' lacks any concrete mechanism (e.g., active-learning loop, digital-twin interface, or benchmark against existing HSM datasets) or preliminary result demonstrating prior success in analogous systems.

    Authors: The closed-loop discovery paradigm outlined in the framework supplies the mechanism for addressing fragmentation and validation gaps via iterative experimental feedback. We have expanded the main text to include a description of an active-learning loop and digital-twin interface, along with citations to analogous successes in related materials domains. No new benchmarks are added, consistent with the manuscript's role as a conceptual proposal. revision: partial

Circularity Check

0 steps flagged

No circularity: conceptual proposal without derivations or fitted quantities

full rationale

The manuscript is a high-level proposal paper whose central claim is the existence of a 'unified framework' and 'pathway toward autonomous, digital-twin-enabled discovery' (abstract). No equations, ansatzes, fitted parameters, or derivation steps appear in the provided text. The strongest claim reduces to an architectural sketch rather than any self-referential reduction of the form 'X derives Y where Y is defined from X.' No self-citations are invoked as load-bearing uniqueness theorems, and no renaming of known results occurs. The derivation chain is therefore absent; the paper is self-contained as a non-quantitative suggestion and receives the default non-circularity finding.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 1 invented entities

The proposal rests on standard domain assumptions about hydrogen storage materials and the limitations of current AI approaches, with the framework itself as the main invented element lacking independent evidence.

axioms (2)
  • domain assumption Hydrogen storage remains a central bottleneck due to the multiscale and coupled nature of thermodynamics, kinetics, and microstructural evolution.
    Directly stated as the opening premise in the abstract.
  • domain assumption Current AI approaches are constrained by fragmented data, limited physical consistency, and weak integration with experimental validation.
    Presented as the motivation for the new framework.
invented entities (1)
  • physics-aware AI ecosystem no independent evidence
    purpose: To integrate coherent data infrastructure, physics-grounded modeling, and AI-driven inverse design within a closed-loop discovery paradigm.
    Introduced as the core proposed solution without prior independent validation or falsifiable predictions.

pith-pipeline@v0.9.0 · 5729 in / 1252 out tokens · 27515 ms · 2026-05-08T17:39:37.341784+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

80 extracted references · 4 canonical work pages · 1 internal anchor

  1. [1]

    The renaissance of hydrides as energy materials

    Mohtadi, R & Orimo, S. The renaissance of hydrides as energy materials. Nat. Rev. Mater. 2, 16091 (2017)

    2017

  2. [2]

    Johnson, N. et al. Realistic roles for hydrogen in the future energy transition. Nat. Rev. Clean Technol. 1, 351–371 (2025)

    2025

  3. [3]

    Webb, C. J. et al. Diverse hydrogen chemistry with perspectives for energy storage. Chem. Commun. 62, 4477–4495 (2026)

    2026

  4. [4]

    E., O’Keeffe, M

    Furukawa, H., Cordova, K. E., O’Keeffe, M. & Yaghi, O. M. The chemistry and applications of metal-organic frameworks. Science 341, 1230444 (2013)

    2013

  5. [5]

    O., Idrees, K

    Chen, Z., Kirlikovali, K. O., Idrees, K. B., Wasson, M. C. & Farha, O. K. Porous materials for hydrogen storage. Chem 8, 693–716 (2022)

    2022

  6. [6]

    L., Mardel, J

    Sutton, A. L., Mardel, J. I. & Hill, M. R. Metal-organic frameworks (MOFs) as hydrogen storage materials at near-ambient temperature. Chem. Eur. J. 30, e202400717 (2024)

    2024

  7. [7]

    Zhang, R. et al. Balancing volumetric and gravimetric capacity for hydrogen in supramolecular crystals. Nat. Chem. 16, 1982–1988 (2024)

    1982

  8. [8]

    Materials for hydrogen storage

    Züttel, A. Materials for hydrogen storage. Mater. Today 6, 24–33 (2003)

    2003

  9. [9]

    R., Zuttel, A

    Orimo, S., Nakamori, Y., Eliseo, J. R., Zuttel, A. & Jensen, C. M. Complex hydrides for hydrogen storage. Chem. Rev. 107, 4111–4132 (2007)

    2007

  10. [10]

    Hirscher, M. et al. Materials for hydrogen-based energy storage - Past, recent progress and future outlook. J. Alloys Compd. 827, 153548 (2020)

    2020

  11. [11]

    Jiang, M. et al. Applicability and limitations of hydrogen affinity as a descriptor for designing high-entropy alloys for hydrogen storage. Chem. Mater. (2026)

    2026

  12. [12]

    G., Banat, F., & Cheng, C

    Gebretatios, A. G., Banat, F., & Cheng, C. K. A critical review of hydrogen storage: Toward the nanoconfinement of complex hydrides from the synthesis and characterization perspectives. Sustain. Energ. Fuels. 8, 5091–5130 (2024)

    2024

  13. [13]

    Witman, M. D. et al. A bulk versus nanoscale hydrogen storage paradox revealed by material- system co-design. Adv. Funct. Mater. 34, 2411763 (2024)

    2024

  14. [14]

    Wang, Q. et al. AI agents for solid electrolytes: Opportunities, challenges, and future directions. AI Agent 1, 10 (2025)

    2025

  15. [15]

    Zhao, C. & Li, H. AI agents: Opportunity, hype, and the way through. AI Agent 2, 3 (2026). 33

    2026

  16. [16]

    Zhang, D. et al. Digital materials ecosystem: from databases to AI agents for autonomous discovery, Chem. Sci. 17, 5782-5804 (2026)

    2026

  17. [17]

    & Cheng, D

    Wang, X., Li, Z., Zhang, D., Li, H., Xu, H. & Cheng, D. Catalysis AI agent guides discovering the universal design principle of Cu-based single-atom alloy catalysts for CO2 electroreduction. Angew. Chem. Int. Ed. e24612 (2026)

    2026

  18. [18]

    Zhang, D. et al. Accelerating catalyst materials discovery with large artificial intelligence models. Angew. Chem. Int. Ed. e26150 (2026)

    2026

  19. [19]

    Gao, Z. et al. Catalytic strategies and mechanisms for enhancing MgH2 solid-state hydrogen storage. Chem Catal. 6, 101692 (2026)

    2026

  20. [20]

    Cheng, M. et al. Artificial intelligence-driven approaches for materials design and discovery. Nat. Mater. 25, 174–190 (2026)

    2026

  21. [21]

    Curtarolo, S. et al. AFLOW: An automatic framework for high-throughput materials discovery. Comput. Mater. Sci. 58, 218–226 (2012)

    2012

  22. [22]

    E., Kirklin, S., Aykol, M., Meredig, B., & Wolverton, C

    Saal, J. E., Kirklin, S., Aykol, M., Meredig, B., & Wolverton, C. Materials design and discovery with high-throughput density functional theory: The open quantum materials database (OQMD). JOM 65, 1501–1509 (2023)

    2023

  23. [23]

    Sbailò, L., Fekete, Á., Ghiringhelli, L. M. & Scheffler, M. The NOMAD Artificial-Intelligence Toolkit: turning materials-science data into knowledge and understanding. npj Comput. Mater. 8, 250 (2022)

    2022

  24. [24]

    Horton, M. K. et al. Accelerated data-driven materials science with the Materials Project. Nat. Mater. 24, 1522–1532 (2025)

    2025

  25. [25]

    Cavignac, T. et al. AI-Driven expansion and application of the Alexandria database. JPhys Mater. (2026)

    2026

  26. [26]

    Chung, Y. G. et al. Advances, Updates, and analytics for the computation-ready, experimental metal-organic framework database: CoRE MOF 2019. J. Chem. Eng. Data 64, 5985–5998 (2019)

    2019

  27. [27]

    npj Comput

    Rosen et al., High-throughput predictions of metal-organic framework electronic properties: Theoretical challenges, graph neural networks, and data exploration. npj Comput. Mat. 8, 112 (2022)

    2022

  28. [28]

    Bobbitt, N. S. et al. MOFX-DB: An Online Database of Computational Adsorption Data for Nanoporous Materials. J. Chem. Eng. Data 68, 483–498 (2023). 34

    2023

  29. [29]

    Allendorf, M. D. et al. An assessment of strategies for the development of solid-state adsorbents for vehicular hydrogen storage. Energy Environ. Sci. 11, 2784–2812 (2018)

    2018

  30. [30]

    Mendoza-Cortés, J. et al. High H2 uptake in Li-, Na-, K-metalated covalent organic frameworks and metal organic frameworks at 298 K. J. Phys. Chem. A 116, 1621–1631 (2012)

    2012

  31. [31]

    Kim, W.-T. et al. Machine learning-assisted design of metal-organic frameworks for hydrogen storage: A high-throughput screening and experimental approach. Chem. Eng. J. 507, 160766 (2025)

    2025

  32. [32]

    G., Trikalitis, P

    Livas, C. G., Trikalitis, P. N. & Froudakis, G. E. MOFSynth: A computational tool toward synthetic likelihood predictions of MOFs. J. Chem. Inf. Model. 64, 8193–8200 (2024)

    2024

  33. [33]

    P., Gkagkas, K

    Sarikas, A. P., Gkagkas, K. & Froudakis, G. E. RetNeXt: A pretrained model for transfer learning across the MOF adsorption space. J. Chem. Inf. Model. 66, 2110–2116 (2026)

    2026

  34. [34]

    Jang, S.-H. et al. Digital Hydrogen Platform (DigHyd): A rigorously curated database for hydrogen storage materials empowered by AI-assisted literature mining. arXiv:2603.14139 (2026)

    work page arXiv 2026

  35. [35]

    Zhang, D. et al. “DIVE” into hydrogen storage materials discovery with AI agents. Chem. Sci. 17, 3031–3042 (2026)

    2026

  36. [36]

    D., Bon, B., Senkovska, I

    Evans, J. D., Bon, B., Senkovska, I. & Kaskel, S. A universal standard archive file for adsorption data. Langmuir 37, 4222–4226 (2021)

    2021

  37. [37]

    Broom, D. P. & Hirscher, M. Improving reproducibility in hydrogen storage material research. ChemPhysChem 22, 2141–2157 (2021)

    2021

  38. [38]

    Li, K. et al. A critical examination of robustness and generalizability of machine learning prediction of materials properties. npj Comput. Mater. 9, 55 (2023)

    2023

  39. [39]

    Jang, S.-H. et al. Physically interpretable descriptors drive the materials design of metal hydrides for hydrogen storage. Chem. Sci. 6, 23111–23120 (2025)

    2025

  40. [40]

    GoodRegressor: A hierarchical inductive bias for navigating high-dimensional compositional space

    Jang, S.-H. GoodRegressor: A hierarchical inductive bias for navigating high-dimensional compositional space. arXiv:2510.18325 (2025)

    work page arXiv 2025

  41. [41]

    Jang, S.-H. et al. A unified descriptor framework for hydrogen storage capacity and equilibrium pressure in interstitial hydrides. arXiv:2604.11660 (2026)

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  42. [42]

    Li, C. et al. Picturing the gap between the performance and US-DOE’s hydrogen storage target: A data-driven model for MgH2 dehydrogenation. Angew. Chem. Int. Ed. 63, e202320151 (2024)

    2024

  43. [43]

    Zhou, P. et al. Machine learning in solid-state hydrogen storage materials: Challenges and perspectives. Adv. Mater. 37, 2413430 (2025). 35

    2025

  44. [44]

    & Trattner, A

    Klopčič, N., Grimmer, I., Winkler, F., Sartory, M. & Trattner, A. A review on metal hydride materials for hydrogen storage. J. Energy Storage 72, 108456 (2023)

    2023

  45. [45]

    Voskuilen, T. G. & Pourpoint, T. L. Phase field modeling of hydrogen transport and reaction in metal hydrides. Int. J. Hydrog. Energy 38, 7363–7375 (2013)

    2013

  46. [46]

    Unke, O. T. et al. Machine learning force fields. Chem. Rev. 121, 10142–10186 (2021)

    2021

  47. [47]

    Zhang, L., Han, J., Wang, H., Car, R. & E, W. Deep potential molecular dynamics: A scalable model with the accuracy of quantum mechanics. Phys. Rev. Lett. 120, 143001 (2018)

    2018

  48. [48]

    A., Wood, M

    Cusentino, M. A., Wood, M. A. & Thompson, A. P. Machine learned interatomic potentials for gas-metal interactions. Model. Simul. Mater. Sci. Eng. 33, 015007 (2024)

    2024

  49. [49]

    Kulichenko, M. et al. Data Generation for Machine Learning Interatomic Potentials and Beyond. Chem. Rev. 124, 13681–13714 (2024)

    2024

  50. [50]

    Yuan, E. C.-Y. et al. Foundation models for atomistic simulation of chemistry and materials. Nat. Rev. Chem. 10, 212–230 (2026)

    2026

  51. [51]

    Angeletti, A. et al. Hydrogen diffusion in magnesium using machine learning potentials: a comparative study. npj Comput. Mater. 11, 85 (2025)

    2025

  52. [52]

    & Huang, S

    Wang, N. & Huang, S. Molecular dynamics study on magnesium hydride nanoclusters with machine-learning interatomic potential. Phys. Rev. B 102, 094111 (2020)

    2020

  53. [53]

    Ito, K. et al. Predicting hydrogen diffusion in nickel-manganese random alloys using machine learning interatomic potentials. Commun. Mater. 6, 195 (2025)

    2025

  54. [54]

    Shuang, F. et al. Decoding the hidden dynamics of super-Arrhenius hydrogen diffusion in multi-principal element alloys via machine learning. Acta Mater. 289, 120924 (2025)

    2025

  55. [55]

    & Ikeda, Y

    Kumar, P., Körmann, F., Edalati, K., Grabowski, B. & Ikeda, Y. Hydrogen diffusion in TiCr2Hx Laves phases: A combined ab initio and machine-learning-potential study. Acta Mater. 308, 122048 (2026)

    2026

  56. [56]

    & Ikeda, Y

    Kumar, P., Körmann, F., Grabowski, B. & Ikeda, Y. Machine learning potentials for hydrogen absorption in TiCr2 Laves phases. Acta Mater. 297, 121319 (2025)

    2025

  57. [57]

    W., Wood, B

    Qi, J., Ko, T. W., Wood, B. C., Pham, T. A. & Ong, S. P. Robust training of machine learning interatomic potentials with dimensionality reduction and stratified sampling. npj Comput. Mater. 10, 43 (2024)

    2024

  58. [58]

    Sato, R. et al. Surface melting-driven hydrogen absorption for high-pressure polyhydride synthesis. Proc. Natl. Acad. Sci. U.S.A. 122, e2413480122 (2025). 36

    2025

  59. [59]

    & Wang, L.-W

    Zhang, B., Asta, M. & Wang, L.-W. Machine learning force field for Fe-H system and investigation on role of hydrogen on the crack propagation in -Fe. Comput. Mater. Sci. 214, 111709 (2022)

    2022

  60. [60]

    & Mori, H

    Ito, K., Otaki, T., Yokoi, T., Hyodo, K. & Mori, H. Machine learning interatomic potential reveals hydrogen embrittlement origins at general grain boundaries in -iron. Commun. Mater. 7, 30 (2025)

    2025

  61. [61]

    & Cangi, A

    Tahmasbi, H., Ramakrishna, K., Lokamani, M. & Cangi, A. Machine learning-driven structure prediction for iron hydrides. Phys. Rev. Mater. 8, 033803 (2024)

    2024

  62. [62]

    & Oda, T

    Kwon, H., Shiga, M., Kimizuka, H. & Oda, T. Accurate description of hydrogen diffusivity in bcc metals using machine-learning moment tensor potentials and path-integral methods. Acta Mater. 247, 118739 (2023)

    2023

  63. [63]

    & Shiga, M

    Kataoka, Y., Haruyama, J., Sugino, O. & Shiga, M. Predictive evaluation of hydrogen diffusion coefficient on Pd(111) surface by path integral simulations using neural network potential. Phys. Rev. Res. 6, 043224 (2024)

    2024

  64. [64]

    & Alibakhshi, A

    Steffen, J. & Alibakhshi, A. Hydrogen diffusion on Ni(100): A combined machine-learning, ring polymer molecular dynamics, and kinetic Monte Carlo study. J. Chem. Phys. 161, 184116 (2024)

    2024

  65. [65]

    Tran, H. B. et al. Tuning stability of AB3-type alloys by suppressing magnetism. Chem. Mater. 38, 497–507 (2026)

    2026

  66. [66]

    M., Fenocchio, L., Cacciamani, G

    Palumbo, M., Dematteis, E. M., Fenocchio, L., Cacciamani, G. & Baricco, M. Advances in CALPHAD methodology for modeling hydrides: A comprehensive review. J. Phase Equilib. Diffus. 45, 273–289 (2024)

    2024

  67. [67]

    Alvares, E. et al. Multiscale modeling of metal-hydride interphases—quantification of decoupled chemo-mechanical energies. npj Comput. Mater. 10, 249 (2024)

    2024

  68. [68]

    W., Colas, K

    Heo, T. W., Colas, K. B., Motta, A. T. & Chen, L.-Q. A phase-field model for hydride formation in polycrystalline metals: Application to δ-hydride in zirconium alloys. Acta Mater. 181, 262–277 (2019)

    2019

  69. [69]

    Alekseeva, S. et al. Grain-growth mediated hydrogen sorption kinetics and compensation effect in single Pd nanoparticles. Nat. Commun. 12, 5427 (2021). 37

    2021

  70. [70]

    Dyck, A. et al. Hydride formation in open thin film metal hydrogen systems: Cahn-Hilliard- type phase-field simulations coupled to elasto-plastic deformations. Mech. Mater. 203, 105258 (2025)

    2025

  71. [71]

    Zeni, C. et al. A generative model for inorganic materials design. Nature 639, 624–632 (2025)

    2025

  72. [72]

    Wilmer, C. E. et al. Large-scale screening of hypothetical metal-organic frameworks. Nature Chem. 4, 83–89 (2012)

    2012

  73. [73]

    E., Zhu, M., Evangelopoulos X

    Cisse, A., Cooper, M. E., Zhu, M., Evangelopoulos X. & Cooper, A. I. Can we automate scientific reasoning in closed-loop experiments using large language models? Digit. Discov. 5, 1132–1160 (2026)

    2026

  74. [74]

    Wang, H. et al. Training-free active learning framework in materials science with large language models. arXiv:2511.19730 (2025)

    work page arXiv 2025

  75. [75]

    dos Santos, E. C. et al. Explore the ionic conductivity trends on B12H12 divalent closo-type complex hydride electrolytes. Chem. Mater. 35, 5996–6005 (2023)

    2023

  76. [76]

    Wang, Q. et al. Unraveling the complexity of divalent hydride electrolytes in solid-state batteries via a data-driven framework with large language model. Angew. Chem. Int. Ed. 64, e202506573 (2025)

    2025

  77. [77]

    Zou, Y. et al. El Agente: An autonomous agent for quantum chemistry. Matter 8, 102263 (2025)

    2025

  78. [78]

    Kusne, A. G. et al. On-the-fly closed-loop materials discovery via Bayesian active learning. Nat. Commun. 11, 5966 (2020)

    2020

  79. [79]

    Bennett, J. A. et al. Autonomous reaction Pareto-front mapping with a self-driving catalysis laboratory. Nat. Chem. Eng. 1, 240–250 (2024)

    2024

  80. [80]

    & Wang, L

    Cao, Z. & Wang, L. Reinforcement fine-tuning for materials design. Phys. Rev. B 113, 024106 (2026). 38 ToC

    2026