Pith. sign in

REVIEW 3 major objections 4 minor 55 references

Going Beyond the d-band Center to Design Intermetallic Catalysts for Nitrogen Reduction: A High-Throughput DFT and Machine Learning Study

T0 review · 3 major / 4 minor · reviewed 2026-08-01 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper argues that s- and p-band electronic descriptors, not the traditional d-band center, control how strongly nitrogen-reduction intermediates bind on intermetallic surfaces, and that the Fe–Co compounds Fe9Co7 and Fe3Co are the best

desk verdict Useful new IMC adsorption dataset and an honest ML descriptor analysis, but the 'beyond d-band' claim is not yet proven for partially filled d-band alloys—needs ablation and convergence checks. read the letter →

arxiv 2607.17463 v1 pith:R3XQUACQ submitted 2026-07-20 cond-mat.mtrl-sci

Parastoo Agharezaei , Kulbir Kaur Ghuman This is my paper · ORCID
classification cond-mat.mtrl-sci
keywords nitrogenreductionreactionintermetalliccompoundsd-bandcenters-andp-banddescriptorsdensityfunctionaltheorymachinelearningscalingrelationshipsammoniasynthesis
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This study screens 47 bimetallic intermetallic compounds for the nitrogen reduction reaction, computing adsorption energies of N2, N2H, and NH3 at roughly 1,200 surface sites. It claims that the conventional d-band center is not the dominant descriptor on these multimetallic surfaces; s- and p-band centers and fillings, along with the size and electronegativity of atoms neighboring the adsorption site, carry more predictive weight, especially for N2 and N2H. The authors argue this reflects adsorption-induced sp–d hybridization that operates even in transition metals with partially filled d-bands, not just in p-block or nearly filled d-band systems. If right, the compact 20-feature descriptor set and the Fe–Co candidates Fe9Co7 and Fe3Co give a concrete, noble-metal-free path toward ammonia synthesis catalyst design.

What carries the argument

The carrying mechanism is a descriptor-based machine-learning pipeline built from per-site DFT data: for each adsorption configuration, 72 features encode s-, p-, and d-band centers and fillings (local and global), atomic charges of neighboring atoms, intrinsic atomic properties of the binding atom (electronegativity, electron affinity, covalent radius, ionization energy, and related quantities), and six aggregated statistics over shells of up to 20 neighboring atoms. Feature selection reduces the set to 20; gradient-boosted tree models then predict adsorption energies with mean absolute errors of 0.26 eV for N2, 0.39 eV for N2H, and 0.17 eV for NH3. A feature-attribution analysis ranks desc

What would settle it

Recompute the adsorption energies for Fe3Co and Fe9Co7 with spin-polarized DFT, a denser k-point mesh, and converged slab thickness; if the site ordering or the N2/N2H/NH3 adsorption windows shift by more than a few tenths of an eV, or if the predicted sp–d hybridization peaks in the projected density of states disappear, the central descriptor claim and the catalyst ranking would fail.

Watch

Extended reading notes

Core claim

The central claim: on intermetallic surfaces, adsorption strength for nitrogen-reduction intermediates is controlled more by s- and p-band features and the local atomic environment than by the d-band center. Using DFT energies for about 1,200 adsorption configurations and ML models reduced to 20 features, the authors find that local s- and p-band centers, the maximum covalent radius in the neighbor shell, and shell electronegativity dominate predictions for N2 and N2H, while the d-band center is not among the top features. Mechanistically, they argue that adsorption induces sp–d hybridization—co-localized, simultaneous modification of sp and d states in the projected density of states—even i

Load-bearing premise

The load-bearing premise is that the DFT adsorption energies—computed with a single k-point for geometry optimization, a generalized-gradient functional, and no reported spin-polarized treatment or slab-thickness convergence—are accurate enough to rank surface sites on magnetic Fe, Co, and Ni surfaces, where errors of a few tenths of an eV could change both the catalyst ranking and the feature-importance conclusions.

Editorial extensions

If this is right

  • Fe9Co7 and Fe3Co emerge as promising non-noble NRR catalysts because they fall in the adsorption-energy windows for N2, N2H, and NH3 simultaneously.
  • The d-band center alone is insufficient for screening intermetallic catalysts; s- and p-band features and local environment descriptors should be included in descriptor sets.
  • The decoupling of NH3 from the N2–N2H scaling relationship on multimetallic sites opens the possibility of independently optimizing nitrogen activation and ammonia desorption.
  • A compact 20-feature set with simple, computationally efficient ML models can predict adsorption energies accurately enough for rapid screening of additional intermetallic compounds.
  • The role of s- and p-orbitals extends beyond p-block elements and nearly filled d-band metals to transition metals with partially filled d-bands, via adsorption-induced sp–d hybridization.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Editorial inference: if the sp–d hybridization picture generalizes, descriptor-based screening for other reactions on intermetallics—such as CO2 electroreduction, hydrogen evolution, or oxygen reduction—should include s- and p-band features rather than assuming the d-band center is sufficient.
  • Editorial inference: because most of the roughly 1,200 configurations contain Al, Zn, Cu, or Pt, elements where s- and p-band roles are already expected, the strongest independent evidence for the partially-filled-d-band claim is the projected-DOS analysis of just Fe3Co and Fe9Co7; extending the claim broadly would require testing more transition-metal-only intermetallics.
  • Editorial inference: the broken NH3–N2H scaling relationship suggests a testable design rule—increase the asymmetry of nearest-neighbor electronegativity at active sites to decouple activation from product release.
  • Editorial inference: an experimental test on synthesized Fe9Co7 or Fe3Co nanoparticles, measuring N2 adsorption isotherms or ammonia synthesis rates, would directly probe whether the computed adsorption windows translate into catalytic activity.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The manuscript presents a high-throughput DFT and machine-learning study of NRR-relevant adsorption energies (*N2, *N2H, *NH3) on 47 bimetallic intermetallic compounds, totaling about 1,200 adsorption configurations. The authors compute adsorption energies with PBE, construct a 72-dimensional descriptor set combining intrinsic atomic properties, shell statistics, and s/p/d band centers and fillings, and train multiple ML models; XGBoost yields test MAEs of 0.26 eV (*N2), 0.39 eV (*N2H), and 0.17 eV (*NH3). SHAP analysis identifies max(covalent radius), s- and p-band descriptors as more important than the d-band center, and the authors interpret this as evidence of s/p–d hybridization that extends beyond p-block and filled-d-band systems. They further analyze scaling relations, find that *N2H–*NH3 scaling is broken, and propose Fe9Co7 and Fe3Co as the most balanced catalysts. The paper includes open code on GitHub and a detailed supporting information section.

Significance. If the DFT energies and the descriptor interpretation are reliable, this work would be a useful contribution to NRR catalyst screening by broadening the descriptor space beyond the d-band center and proposing specific non-noble intermetallic candidates. The manuscript's strengths include a comparatively exhaustive enumeration of adsorption sites, a transparent ML pipeline with a sealed test set, open data/code availability, and a physically interpretable feature set. However, the central 'beyond the d-band center' claim is currently supported by a global SHAP analysis that is confounded by dataset composition, and the quantitative DFT basis is not sufficiently validated for magnetic transition-metal surfaces. These issues are load-bearing for both the headline mechanistic claim and the catalyst ranking.

major comments (3)
  1. [Section 2.1] The DFT protocol is not sufficiently validated for the systems studied. Bulk and surface optimization uses a 1×1×1 Monkhorst–Pack k-point grid, and no slab thickness, vacuum gap, or spin-polarization settings are reported for Fe-, Co-, and Ni-containing surfaces. For magnetic transition metals, adsorption energies can shift by several tenths of an eV with k-point sampling and spin treatment; this uncertainty directly affects the Fe–Co ranking in Fig. 4 and the electronic-descriptor trends in Figs. 6 and 8. The authors should provide convergence tests in k-points, slab thickness, and vacuum, specify spin treatment, and quantify the resulting error bars on the adsorption energies.
  2. [Section 3.2] The SHAP-based conclusion that s- and p-band characteristics are 'more prominent' than the d-band center for *N2 and *N2H is confounded by dataset composition. The authors themselves note that ~800 of ~1200 configurations contain Al, Zn, Cu, or Pt—elements for which s/p dominance is expected from prior work. The global SHAP ranking therefore does not establish that s/p features are decisive for partially-filled d-band transition metals. The only evidence offered for the Fe–Co case is the qualitative PDOS analysis of Fig. 9, which covers two surfaces and is not quantified. Additionally, no d-band-only baseline model was trained, so the 'more prominent' claim is not measured as an improvement over a model using only d-band descriptors. Please add a d-band-only baseline and a subset analysis restricted to Fe/Co/Ni/V systems, or temper the claim accordingly.
  3. [Section 2.3] Outlier removal is performed on the full dataset before the train/test split: 'extreme outliers... are identified as samples beyond three standard deviations from the mean (|z|>3) of the target distribution' and 'were removed from the dataset for each adsorbate.' Since the mean and standard deviation are computed from the entire dataset, the test set influences which samples are removed, which is a data-leakage mechanism. This can inflate the reported test MAE/R². The outlier removal threshold and any filtering should be fit only on the training folds, or the authors should demonstrate that the model conclusions are unchanged when outliers are retained.
minor comments (4)
  1. [Fig. 4] In Fig. 4(a), 'Al3N2' appears to be a typo for 'Al3Ni2'. Please check the label.
  2. [Fig. 6] The notation 'εb,local' and 'fb,local' is used in the caption but not defined there; define these symbols explicitly, including the energy window (-10 to 10 eV) used in Eq. (2)-(3).
  3. [Section 3.1] The 'asymmetry index' is introduced as the difference between maximum and minimum electronegativity of the four nearest neighbors, but it is not fully defined: is it the spread of electronegativities or the difference between the two extremes? Clarify the calculation and whether it is averaged over sites.
  4. [Section 2.3] The feature-selection procedure uses SelectKBest with F-regression, which is univariate and may not capture interactions that tree-based models use. This limitation is not discussed; a brief note would be helpful.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity; the ML target and descriptors are independent, with only a non-load-bearing self-citation.

full rationale

The derivation chain is an empirical regression: DFT adsorption energies (Eq. 1) are the labels, while s-, p-, and d-band centers/fillings, Bader charges, and atomic properties are independent features. The reported MAEs are genuine out-of-sample results from a sealed 20% test set after repeated CV feature selection, so no predicted quantity reduces to a fitted value by construction. SHAP feature ranking is a post-hoc explanation of a fitted model, not a fitted input renamed as a prediction. The only self-citation, Ref. [25], is used for a qualitative remark that N2 adsorption is sensitive to its local environment, and this is also directly supported by the paper's own site-resolved distributions (Fig. S1a); it is not load-bearing for the ML models or the Fe-Co sp-d hybridization claim, which is based on direct PDOS analysis. Dataset-composition concerns about the global SHAP s/p ranking are external-validity limitations, not circularity.

Assumptions & free parameters 7 free parameters · 5 assumptions · 0 invented entities

All downstream conclusions rest on PBE-DFT energetics computed with a 1x1x1 k-point grid and on ML/SHAP interpretation of those energies. The free parameters listed are modeling choices that materially affect the adsorption-energy ranking, the selected 20 features, and therefore the central claim that s/p descriptors outperform the d-band center.

free parameters (7)
  • k-point grid for DFT = 1x1x1 Monkhorst-Pack for bulk/surface optimizations; 3x3x3 for DOS
    Chosen without convergence tests; sparse k-point sampling can shift adsorption energies and band centers, especially for metallic Fe/Co/Ni surfaces.
  • Outlier removal threshold = |z| > 3
    Samples beyond three standard deviations are removed per adsorbate before ML training (Section 2.3); this post-hoc data selection affects the target distribution and model evaluation.
  • Adsorption energy windows for catalyst ranking = N2 [-1.0, 0.0] eV; N2H [-2.5, -1.5] eV; NH3 [-1.0, 0.0] eV
    Used to judge which IMCs are 'balanced' (Section 3.1, Figure 4); the windows are imported from literature and are arbitrary for the final Fe9Co7/Fe3Co recommendation.
  • Optimal neighbor-shell level k = k=10 for N2, k=3 for N2H, k=19 for NH3
    Selected by CV R2; the feature set and SHAP results depend strongly on the chosen shell size, and the optimal k differs per adsorbate.
  • Band-center integration window = -10 to +10 eV
    Defines band centers and fillings via Eqs. 2 and 3; the finite window choice affects the numerical values of all s-, p-, and d-band descriptors.
  • Feature-selection thresholds = VarianceThreshold=0.01; correlation threshold=0.95; SelectKBest=20
    Pipeline hyperparameters (Section 2.3) that determine which 20 features reach the model and therefore which descriptors SHAP identifies as important.
  • XGBoost hyperparameters = n_estimators=200, max_depth=4, learning_rate=0.05, subsample=0.8, colsample_bytree=0.8
    Hand-set hyperparameters from SI Table S3; different hyperparameters could change CV rankings and SHAP importance values.
assumptions (5)
  • domain assumption PBE-GGA DFT with PAW pseudopotentials accurately captures adsorption energies and projected DOS for these intermetallic surfaces.
    The entire dataset and all descriptors are computed with this functional; no comparison to hybrid functionals or experimental adsorption energies is provided (Section 2.1).
  • domain assumption A 1x1x1 k-point grid and the selected single surface plane per material are sufficient to represent catalytic behavior.
    Only one XRD-predominant surface plane is used per IMC, and no k-point or slab-thickness convergence tests are reported (Section 2.1, Table S2).
  • domain assumption Spin polarization and magnetism of Fe, Co, and Ni can be neglected or are handled by default code settings.
    The authors do not state whether spin-polarized calculations were performed; for magnetic transition metals this can materially change adsorption energetics and DOS descriptors.
  • domain assumption The associative NRR pathway described by N2, N2H, and NH3 adsorption energies captures the limiting chemistry of the catalyst.
    The screening ignores other NxHy intermediates, explicit hydrogenation barriers, solvent effects, applied potential, and competition with hydrogen evolution, though the paper acknowledges HER as a general NRR challenge.
  • domain assumption SHAP contributions on the trained XGBoost model reflect causal electronic-structure effects.
    SHAP explains the model, not the physics; without an ablation against a d-band-only model or stability analysis, feature importance rankings are descriptive rather than causal.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Going Beyond the d-band Center to Design Intermetallic Catalysts for Nitrogen Reduction: A High-Throughput DFT and Machine Learning Study." pith.science (2026). https://pith.science/paper/R3XQUACQ

@misc{pith2026260717463,
  author       = {Pith},
  title        = {Pith review of: Going Beyond the d-band Center to Design Intermetallic Catalysts for Nitrogen Reduction: A High-Throughput DFT and Machine Learning Study},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/R3XQUACQ}},
  note         = {Machine review of arXiv:2607.17463}
}
read the original abstract

This study combines high-throughput DFT calculations with machine learning techniques to uncover the key descriptors governing the nitrogen reduction reaction (NRR) in intermetallic compounds (IMCs). A dataset of 47 bimetallic IMCs was constructed, and the adsorption energies of key intermediates (N2, N2H, and NH3) were systematically evaluated across all accessible surface sites, yielding approximately 1,200 data points. By incorporating intrinsic material properties along with electronic descriptors including s-, p- and d-band centers and fillings, as well as Bader charges of atoms neighboring the adsorbate, predictive ML models were developed with mean absolute errors of 0.26 eV for N2, 0.39 eV for N2H, and 0.17 eV for NH3 adsorption. Importantly, accurate predictions are obtained with only 20 key features, enabling the use of simple and computationally efficient ML models. SHAP analysis indicates that p- and s-band characteristics play a more prominent role in determining adsorption strength than the traditionally used d-band center, particularly for N2 and N2H intermediates. Beyond their established importance in systems containing p-block elements or nearly filled d-band metals, s- and p-orbitals are also found to contribute significantly to transition-metal alloys activity such as Fe-Co, driven by adsorption-induced sp-d hybridization. By challenging the d-band-centric paradigm and identifying s- and p-band descriptors as critical yet overlooked contributors, this work redefines the electronic descriptor space for intermetallic NRR catalysts and lays the groundwork for DFT-ML-guided discovery of non-noble materials for sustainable ammonia synthesis.

Figures

Figures reproduced from arXiv: 2607.17463 by the authors.

Figure 1
Figure 1. Schematic workflow for high-throughput DFT and machine learning analysis of [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. Distribution of average bonding energies for (a) adsorbed molecules and (b) [PITH_FULL_IMAGE:figures/full_fig_p012_2.png] view at source ↗
Figure 3
Figure 3. (a) Different orientations of N2 and N2H molecules on the surface of IMCs; Distribution of adsorption energies for (b) N2 and (c) N2H molecules in different orientations on IMC surfaces; Heatmaps of adsorption orientation preferences for (d) N2, and (e) N2H across IMCs. The color scale represents the fraction of adsorption sites in each alloy that adsorb the molecule in each orientation. Orientations of the adsorbed… view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Chosen IMCs that possess at least one adsorption site for each adsorbate falling [PITH_FULL_IMAGE:figures/full_fig_p017_4.png]
Figure 5
Figure 5. Figure 5: Scaling relationships between adsorption energies of key intermediates on IMCs: [PITH_FULL_IMAGE:figures/full_fig_p019_5.png]
Figure 6
Figure 6. Figure 6: Electronic featurewise adsorption energy distributions for (a) N [PITH_FULL_IMAGE:figures/full_fig_p020_6.png]
Figure 7
Figure 7. Figure 7: Cross-validation performance of all machine learning models as a function of the [PITH_FULL_IMAGE:figures/full_fig_p023_7.png]
Figure 8
Figure 8. Figure 8: SHAP feature importance analysis for the tuned best-performing models of each [PITH_FULL_IMAGE:figures/full_fig_p026_8.png]
Figure 9
Figure 9. Figure 9: Atom-resolved projected density of states (PDOS) of nearest-neighbor metal atoms [PITH_FULL_IMAGE:figures/full_fig_p029_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

55 extracted references

  1. [1]

    and Herrero, A

    Collado, L. and Herrero, A. and de la Pe. From Conventional to Emerging Ammonia Production Technologies , journal =. 2024 , pages =

    2024

  2. [2]

    Rafiqul Bari, G. A. K. M. and Jeong, J.-H. , title =. Int. J. Energy Res. , year =

  3. [3]

    and Pei, C

    Zhang, X. and Pei, C. and Zhao, Z.-J. and Gong, J. , title =. Energy Environ. Sci. , year =

  4. [4]

    and Wen, H

    Guan, Y. and Wen, H. and Cui, K. and Wang, Q. and Gao, W. and Cai, Y. and Cheng, Z. and Pei, Q. and Li, Z. and Cao, H. and He, T. and Guo, J. and Chen, P. , title =. Nat. Chem. , year =

  5. [5]

    and Xia, M

    Wang, L. and Xia, M. and Wang, H. and Huang, K. and Qian, C. and Maravelias, C. T. and Ozin, G. A. , title =. Joule , year =

  6. [6]

    Challenges and Opportunities for Renewable Ammonia Production via Plasmon-Assisted Photocatalysis , journal =

    Pu. Challenges and Opportunities for Renewable Ammonia Production via Plasmon-Assisted Photocatalysis , journal =. 2022 , volume =

    2022

  7. [7]

    and Pan, Z

    Fu, R. and Pan, Z. and Mu, X. and Li, J. and Zhan, Q. and Zhao, Z. and Mu, X. and Li, L. , title =. J. Mater. Chem. A , year =

  8. [8]

    and Jing, H

    Shao, J. and Jing, H. and Wei, P. and Fu, X. and Pang, L. and Song, Y. and Ye, K. and Li, M. and Jiang, L. and Ma, J. and Li, R. and Si, R. and Peng, Z. and Wang, G. and Xiao, J. , title =. Nat. Energy , year =

Show all 55 references
  1. [9]

    and Zuo, L

    Wen, J. and Zuo, L. and Sun, H. and Wu, X. and Huang, T. and Liu, Z. and Wang, J. and Liu, L. and Wu, Y. and Liu, X. and van Ree, T. , title =. Nanoscale Adv. , year =

  2. [10]

    and Liu, C

    Jin, Z. and Liu, C. and Liu, Z. and Han, J. and Fang, Y. and Han, Y. and Niu, Y. and Wu, Y. and Sun, C. and Xu, Y. , title =. Adv. Energy Mater. , year =

  3. [11]

    and Kahn, D

    Dixon, R. and Kahn, D. , title =. Nat. Rev. Microbiol. , year =

  4. [12]

    and Gao, L

    Li, M. and Gao, L. and White, J. C. and Haynes, C. L. and O'Keefe, T. L. and Rui, Y. and Ullah, S. and Guo, Z. and Lynch, I. and Zhang, P. , title =. Nat. Nanotechnol. , year =

  5. [13]

    and Kroneck, P

    Fritz, G. and Kroneck, P. M. H. and Steuber, J. , title =. Nat. Chem. Biol. , year =

  6. [14]

    and Akiyama, M

    Iwamoto, M. and Akiyama, M. and Aihara, K. and Deguchi, T. , title =. ACS Catal. , year =

  7. [15]

    and Liu, J

    Zhu, X. and Liu, J. and Hu, X. and Zhou, Z. and Li, X. and Wang, W. and Wu, R. and Tu, X. , title =. J. Energy Inst. , year =

  8. [16]

    and Yu, C

    Ren, Y. and Yu, C. and Tan, X. and Wei, Q. and Wang, Z. and Ni, L. and Wang, L. and Qiu, J. , title =. Energy Environ. Sci. , year =

  9. [17]

    and Zhang, H

    Chang, B. and Zhang, H. and Sun, S. and Zhang, G. , title =. Carbon Energy , year =

  10. [18]

    and Barboun, P

    Mehta, P. and Barboun, P. and Herrera, F. A. and Kim, J. and Rumbach, P. and Go, D. B. and Hicks, J. C. and Schneider, W. F. , title =. Nat. Catal. , year =

  11. [19]

    Chen, Z. W. and Lu, Z. and Chen, L. X. and Jiang, M. and Chen, D. and Singh, C. V. , title =. Chem Catal. , year =

  12. [20]

    Suryanto, B. H. R. and Du, H.-L. and Wang, D. and Chen, J. and Simonov, A. N. and MacFarlane, D. R. , title =. Nat. Catal. , year =

  13. [21]

    Montoya, J. H. and Tsai, C. and Vojvodic, A. and N. The Challenge of Electrochemical Ammonia Synthesis: A New Perspective on the Role of Nitrogen Scaling Relations , journal =. 2015 , volume =

    2015

  14. [22]

    and Fei, H

    Liu, R. and Fei, H. and Wang, J. and Guo, T. and Liu, F. and Wang, J. and Wu, Z. and Wang, D. , title =. Appl. Catal. B Environ. , year =

  15. [23]

    and Wu, J

    Gong, Y. and Wu, J. and Kitano, M. and Wang, J. and Ye, T.-N. and Li, J. and Kobayashi, Y. and Kishida, K. and Abe, H. and Niwa, Y. and Yang, H. and Tada, T. and Hosono, H. , title =. Nat. Catal. , year =

  16. [24]

    and Chen, Z

    Yin, H. and Chen, Z. and Xiong, S. and Chen, J. and Wang, C. and Wang, R. and Kuwahara, Y. and Luo, J. and Yamashita, H. and Peng, Y. and Li, J. , title =. Chem Catal. , year =

  17. [25]

    and Ghuman, K

    Agharezaei, P. and Ghuman, K. K. , title =. ACS Catal. , year =

  18. [26]

    and Pervolarakis, E

    Hutu, A.-I. and Pervolarakis, E. and Remediakis, I. N. and Kristoffersen, H. H. and Rossmeisl, J. , title =. J. Phys. Chem. C , year =

  19. [27]

    and Zhao, Z.-J

    Sun, G. and Zhao, Z.-J. and Mu, R. and Zha, S. and Li, L. and Chen, S. and Zang, K. and Luo, J. and Li, Z. and Purdy, S. C. and Kropf, A. J. and Miller, J. T. and Zeng, L. and Gong, J. , title =. Nat. Commun. , year =

  20. [28]

    Saidi, W. A. , title =. npj Comput. Mater. , year =

  21. [29]

    Zhou, H. Y. and Qu, Y. B. and Fan, Y. C. and Wang, Z. L. and Lang, X. Y. and Li, J. C. and Jiang, Q. , title =. Appl. Catal. B Environ. , year =

  22. [30]

    and Luo, M

    Wang, X. and Luo, M. and Lan, J. and Peng, M. and Tan, Y. , title =. Adv. Mater. , year =

  23. [31]

    and Tang, C

    Wei, H. and Tang, C. and Wang, Y. and Feng, Y. and Shi, H. and Cui, M. and Liu, H. and Dou, S. and Li, L. , title =. Green Energy Environ. , year =

  24. [32]

    and Xu, W

    Fan, G. and Xu, W. and Li, J. and Chen, J. and Yu, M. and Ni, Y. and Zhu, S. and Su, X. and Cheng, F. , title =. Adv. Mater. , year =

  25. [33]

    and Chen, X

    Zhou, J. and Chen, X. and Guo, M. and Hu, W. and Huang, B. and Yuan, D. , title =. ACS Catal. , year =

  26. [34]

    , title =

    Shamekhi, M. , title =. Phys. Chem. Chem. Phys. , year =

  27. [35]

    and Furthm

    Kresse, G. and Furthm. Efficient Iterative Schemes for. Phys. Rev. B , year =

  28. [36]

    Perdew, J. P. and Chevary, J. A. and Vosko, S. H. and Jackson, K. A. and Pederson, M. R. and Singh, D. J. and Fiolhais, C. , title =. Phys. Rev. B , year =

  29. [37]

    and Perdew, J

    Burke, K. and Perdew, J. P. and Wang, Y. , title =. Electronic Density Functional Theory , year =

  30. [38]

    Projector Augmented-Wave Method , journal =

    Bl. Projector Augmented-Wave Method , journal =. 1994 , volume =

    1994

  31. [39]

    and Sanville, E

    Tang, W. and Sanville, E. and Henkelman, G. , title =. J. Phys. Condens. Matter , year =

  32. [40]

    and Arnaldsson, A

    Henkelman, G. and Arnaldsson, A. and J. A Fast and Robust Algorithm for. Comput. Mater. Sci. , year =

  33. [41]

    Horton, M. K. and Huck, P. and Yang, R. X. and Munro, J. M. and Dwaraknath, S. and Ganose, A. M. and Kingsbury, R. S. and Wen, M. and Shen, J. X. and Mathis, T. S. and Kaplan, A. D. and Berket, K. and Riebesell, J. and George, J. and Rosen, A. S. and Spotte-Smith, E. W. C. and...

  34. [42]

    and Sun, S

    Li, J. and Sun, S. , title =. Acc. Chem. Res. , year =

  35. [43]

    Ong, S. P. and Richards, W. D. and Jain, A. and Hautier, G. and Kocher, M. and Cholia, S. and Gunter, D. and Chevrier, V. L. and Persson, K. A. and Ceder, G. , title =. Comput. Mater. Sci. , year =

  36. [44]

    and Wu, Q

    Chen, H. and Wu, Q. and Wang, Y. and Zhao, Q. and Ai, X. and Shen, Y. and Zou, X. , title =. Chem. Commun. , year =

  37. [45]

    and Louppe, G

    Buitinck, L. and Louppe, G. and Blondel, M. and Pedregosa, F. and Mueller, A. and Grisel, O. and Niculae, V. and Prettenhofer, P. and Gramfort, A. and Grobler, J. and Layton, R. and Vanderplas, J. and Joly, A. and Holt, B. and Varoquaux, G. , title =. arXiv , year =

  38. [46]

    and Lee, S.-I

    Lundberg, S. and Lee, S.-I. , title =. 2017 , doi =

    2017

  39. [47]

    and Liu, Y

    Qian, Y. and Liu, Y. and Zhao, Y. and Zhang, X. and Yu, G. , title =. EcoMat , year =

  40. [48]

    Tezak, C. R. and Singstock, N. R. and Alherz, A. W. and Vigil-Fowler, D. and Sutton, C. A. and Sundararaman, R. and Musgrave, C. B. , title =. ACS Catal. , year =

  41. [49]

    and Zhang, D

    Xiao, S. and Zhang, D. and Wang, G. and Zhou, T. and Wang, N. , title =. Molecules , year =

  42. [50]

    and Liu, C

    Shen, X. and Liu, C. and Zhang, Q. , title =. Mater. Adv. , year =

  43. [51]

    High-Throughput Computational Screening of Doped Transition Metal Oxides as Catalysts for Nitrogen Reduction , journal =

    H. High-Throughput Computational Screening of Doped Transition Metal Oxides as Catalysts for Nitrogen Reduction , journal =. 2023 , volume =

    2023

  44. [52]

    Exner, K. S. , title =. ChemCatChem , year =

  45. [53]

    and Ren, M

    He, B. and Ren, M. and Zhang, L. and Lv, P. and Liu, M. and Ye, S. and Jia, Y. , title =. J. Mater. Chem. A , year =

  46. [54]

    and Wan, J

    Yang, X. and Wan, J. and Zhang, H. and Wang, Y. , title =. Chem. Sci. , year =

  47. [55]

    and Huang, B

    Wu, Q. and Huang, B. and Dai, Y. and Heine, T. and Ma, Y. , title =. npj 2D Mater. Appl. , year =

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.