pith. sign in

arxiv: 2605.00609 · v1 · submitted 2026-05-01 · ❄️ cond-mat.mtrl-sci

Coordination Engineering of Dual-Atom Catalysts for Overall Water Splitting: Mechanistic Insights from Constant-Potential First-Principles and Machine Learning

Jiahang Li , Suhang Li , Chong Yan , Jiajun Yu , Qinzhuang Liu , Ruo-Ya Wang , Dongwei Ma This is my paper

Pith reviewed 2026-05-09 19:42 UTC · model grok-4.3

classification ❄️ cond-mat.mtrl-sci
keywords dual-atom catalystscoordination engineeringwater splittingHEROERdensity functional theorymachine learningelectrocatalysis
0
0 comments X

The pith

Coordination engineering identifies 24 dual-atom catalysts with low overpotentials for overall water splitting.

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

The paper tries to establish that coordination engineering of dual-atom catalysts on graphene, by varying metal pairs and C/N ratios, can produce bifunctional materials effective for both hydrogen and oxygen evolution in water splitting. Using a screening approach with constant potential calculations and kinetic analysis on 228 structures, it identifies 24 top candidates based on cobalt with nickel, copper or cobalt, showing performance close to iridium oxide for oxygen evolution and low barriers for the key steps, plus good hydrogen evolution in most cases. Machine learning then maps out how the coordination affects the results. A sympathetic reader would care because this points to a way to create affordable catalysts for clean hydrogen production from water.

Core claim

Through construction of 228 graphene-supported dual-atom catalyst models with varying metal pairs and C/N coordinations, a three-step screening using constant-charge and constant-potential DFT together with PCET kinetic analysis identifies 24 highly active OER candidates with mixed C/N coordination, particularly CoNi, CoCu, and Co2 combinations. These exhibit overpotentials comparable to IrO2 and PCET barriers below 0.40 eV, with 22 also displaying high HER activity. Machine learning analysis reveals coordination-dependent structure-performance relationships, with bifunctionality arising from optimized adsorption of OER intermediates and hydrogen binding.

What carries the argument

Coordination engineering of TM1TM2-C6-xNx dual-atom catalysts, where mixed C/N coordination around the metal pair tunes adsorption of OER intermediates and hydrogen binding strength for HER.

Load-bearing premise

The constant-potential DFT calculations and PCET kinetic analysis accurately predict real-world catalytic performance and that the model systems represent actual synthesized catalysts.

What would settle it

Experimental measurement of overpotentials for a synthesized CoNi dual-atom catalyst with mixed C/N coordination showing values much higher than IrO2 would falsify the prediction.

Figures

Figures reproduced from arXiv: 2605.00609 by Chong Yan, Dongwei Ma, Jiahang Li, Jiajun Yu, Qinzhuang Liu, Ruo-Ya Wang, Suhang Li.

Figure 1
Figure 1. Figure 1: (a) Structural models of TM1TM2-C6-xNx DACs (TM1TM2 = Co2, Fe2, CoCu, CoNi, FeCu, FeNi, and FeCo). The coordination environments span C6 to N6 (C6, C5N, C4N2, C3N3, C2N4, CN5, and N6). A single configuration is identified for C6 and N6. For mixed C/N coordination environments, C5N and CN5 exhibit three (heteronuclear) and two (homonuclear) configurations; C4N2 and C2N4 have nine and six configurations, res… view at source ↗
Figure 3
Figure 3. Figure 3: (a) Pearson correlation heatmap of 18 features. (b) Comparison of DFT calculated and BO–RF predicted ηOER. (c) Feature importance of ηOER calculated by the BO–RF model. (d) Comparison of DFT calculated and SISSO predicted ηOER. 0.0 0.1 0.2 0.3 0.4 -1.0 -0.6 -0.2 0.2 0.6 1.0 0.0 0.4 0.8 1.2 1.6 0.0 0.4 0.8 1.2 1.6 SISSO predicted ηOER (V) DFT calculated ηOER (V) R² = 0.88 RMSE = 0.10 V d 0.0 0.4 0.8 1.2 1.6… view at source ↗
Figure 5
Figure 5. Figure 5: (a) Hydrogen adsorption free energy (ΔG*H) of TM1TM2-C6-xNx DACs with high OER activity (TM1TM2 = CoNi, CoCu, and Co2) at URHE = 0.00 V. The pink shaded region highlights near-optimal ΔG*H window (between −0.2 and +0.2 eV). (b) Volcano plots of exchange current density (i0) as a function of ΔG*H for all adsorption sites in TM1TM2-C6-xNx DACs (TM1TM2 = CoNi, CoCu and Co2). −0.3 −0.2 −0.1 0.0 0.1 0.2 0.3 0.4… view at source ↗
read the original abstract

The rational design of bifunctional electrocatalysts for the hydrogen evolution reaction (HER) and oxygen evolution reaction (OER) is essential for achieving efficient and cost-effective overall water splitting. Atomically dispersed transition-metal catalysts, including single-atom catalysts and dual-atom catalysts (DACs), have emerged as a prominent class of heterogeneous catalysts, in which coordination engineering plays a decisive role in tuning catalytic performance. Herein, we explore coordination-engineered bifunctional overall water splitting electrocatalysts using graphene-supported DACs (TM1TM2-C6-xNx) as model systems. By tuning C/N coordination and dual-metal combinations (Fe, Co, Ni, and Cu), a library of 228 structures was constructed. A three-step screening strategy, combining constant-charge and constant-potential density functional theory with kinetic analysis of proton-coupled electron transfer (PCET), identifies 24 highly active candidates (TM1TM2 = CoNi, CoCu and Co2) with mixed C/N coordination for OER. These catalysts exhibit overpotentials comparable to that of IrO2 and low PCET barriers (lower than 0.40 eV), among which 22 also show high HER activity. Machine learning reveals clear coordination-dependent structure-performance relationships. Such bifunctionality arises from coordination engineering that enables the simultaneous optimization of OER intermediate adsorption and the hydrogen binding strength for HER. This work establishes coordination engineering as an effective strategy for designing high-performance bifunctional dual-atom electrocatalysts for overall water splitting.

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

2 major / 2 minor

Summary. The manuscript reports a computational screening of 228 graphene-supported dual-atom catalysts (TM1TM2-C6-xNx, TM = Fe, Co, Ni, Cu) for bifunctional HER/OER. A three-step workflow (constant-charge DFT screening, constant-potential DFT refinement, and PCET kinetic analysis) identifies 24 promising structures, primarily CoNi, CoCu, and Co2 variants with mixed C/N coordination, that exhibit OER overpotentials comparable to IrO2, PCET barriers below 0.40 eV, and (for 22 of them) favorable HER activity. Machine learning is used to extract coordination-dependent structure–performance relationships, with the central claim that coordination engineering simultaneously optimizes OER intermediate adsorption and HER hydrogen binding.

Significance. If the predictions are robust, the work provides a concrete, coordination-tuned library of bifunctional DAC candidates and demonstrates a logically sequenced computational pipeline that combines constant-potential electrochemistry with kinetic PCET analysis. The ML-derived relationships offer transferable design rules that could accelerate experimental realization of low-cost overall-water-splitting catalysts.

major comments (2)
  1. [Methods/Results] Methods and Results sections: the manuscript provides no error bars, convergence tests, or sensitivity analysis on the choice of DFT functional, implicit-solvent parameters, or electrode-potential window used in the constant-potential calculations. Because the final ranking of the 24 candidates rests directly on the computed overpotentials and PCET barriers, these omissions are load-bearing for the reliability of the screening outcome.
  2. [Results] Results, Table of screened candidates: no experimental validation or direct comparison to measured overpotentials for any of the predicted CoNi/CoCu/Co2 structures is presented. While computational screening papers need not contain new experiments, the absence of even a single benchmark against known DACs or IrO2 under identical computational settings weakens the claim that the identified candidates are “highly active.”
minor comments (2)
  1. [Abstract/Introduction] The abstract and introduction use “parameter-free” language for the ML model; the text should clarify whether any hyperparameters were tuned on the generated dataset.
  2. [Figure captions] Figure captions for the ML correlation plots should explicitly state the training/test split and the cross-validation procedure used to generate the reported R² values.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive and detailed comments. These points help clarify the robustness of our computational screening. We provide point-by-point responses below and will revise the manuscript accordingly.

read point-by-point responses

  1. Referee: [Methods/Results] Methods and Results sections: the manuscript provides no error bars, convergence tests, or sensitivity analysis on the choice of DFT functional, implicit-solvent parameters, or electrode-potential window used in the constant-potential calculations. Because the final ranking of the 24 candidates rests directly on the computed overpotentials and PCET barriers, these omissions are load-bearing for the reliability of the screening outcome.

    Authors: We agree that additional validation strengthens the reliability of the screening. In the revised manuscript, we will add convergence tests for k-point sampling, plane-wave cutoff, and vacuum thickness in the Methods section. We will also perform and report sensitivity analyses on the implicit-solvent dielectric constant and the electrode-potential window (e.g., testing ±0.1 V around the target potentials) for a representative subset of structures, including the top 24 candidates and several lower-performing ones for comparison. Error bars on overpotentials and PCET barriers will be estimated from these tests and included in the Results. Full sensitivity across all 228 structures is computationally prohibitive, but the subset analysis will directly support the ranking of the identified candidates. revision: partial

  2. Referee: [Results] Results, Table of screened candidates: no experimental validation or direct comparison to measured overpotentials for any of the predicted CoNi/CoCu/Co2 structures is presented. While computational screening papers need not contain new experiments, the absence of even a single benchmark against known DACs or IrO2 under identical computational settings weakens the claim that the identified candidates are “highly active.”

    Authors: We acknowledge that direct benchmarking under identical settings improves credibility. The original manuscript already compares our computed OER overpotentials to experimental literature values for IrO2 and several reported DACs. To address the referee’s specific concern, we will add a new subsection and table in the revised Results that recomputes overpotentials and PCET barriers for selected literature-known DACs (e.g., Co2 and CoNi variants with documented structures) using exactly the same constant-potential DFT protocol, functional, and solvent model as our screening workflow. This provides a direct, apples-to-apples computational benchmark. As this is a computational study, new experimental measurements cannot be added, but the added benchmarks will reinforce the claim that the 24 candidates are highly active relative to both IrO2 and existing DACs. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper constructs a library of 228 graphene-supported DAC structures by varying C/N coordination and metal pairs (Fe, Co, Ni, Cu). It then applies a three-step screening workflow consisting of constant-charge DFT, constant-potential DFT, and PCET kinetic barrier calculations to identify 24 OER-active candidates (CoNi, CoCu, Co2 with mixed coordination) whose computed overpotentials and barriers are compared to IrO2. Machine learning is applied afterward to the DFT-generated data to extract coordination-dependent trends. None of these steps reduces to a self-definition, a fitted parameter renamed as a prediction, or a self-citation chain; the performance metrics are direct outputs of the independent first-principles calculations on the enumerated structures. The derivation is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claim rests on standard assumptions in computational catalysis regarding the accuracy of DFT for surface energetics and the representativeness of the model systems.

axioms (2)
  • domain assumption Constant-potential density functional theory calculations reliably predict the energetics and kinetics of electrocatalytic reactions on dual-atom sites
    Invoked as the basis for the three-step screening strategy and identification of active candidates.
  • domain assumption The graphene-supported TM1TM2-C6-xNx models adequately capture the behavior of real dual-atom catalysts
    Used to construct and evaluate the library of 228 structures.

pith-pipeline@v0.9.0 · 5603 in / 1292 out tokens · 43375 ms · 2026-05-09T19:42:29.736719+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

13 extracted references · 13 canonical work pages

  1. [1]

    Ultrastable supported oxygen evolution electrocatalyst formed by ripening-induced embedding

    (1) Shi, W.; Shen, T.; Xing, C.; Sun, K.; Yan, Q.; Niu, W.; Yang, X.; Li, J.; Wei, C.; Wang, R.; Fu, S.; Yang, Y .; Xue, L.; Chen, J.; Cui, S.; Hu, X.; Xie, K.; Xu, X.; Duan, S.; Xu, Y .; Zhang, B. Ultrastable supported oxygen evolution electrocatalyst formed by ripening-induced embedding. Science 2025, 387 (6735), 791-796. (2) An, B.; Li, X.; Li, L.; Dai...

    work page 2025

  2. [2]

    Lattice strain engineering in Ru-based electrocatalysts for efficient acidic overall water splitting and Ru dissolution suppression

    (9) Duan, X.; Wen, N.; Liu, S.; Li, H.; Jiao, X.; Chen, D.; Xia, Y . Lattice strain engineering in Ru-based electrocatalysts for efficient acidic overall water splitting and Ru dissolution suppression. ACS Catal. 2025, 15 (12), 10119-10129. (10) Niu, X.; Chen, Y .; Sun, M.; Nagao, S.; Aoki, Y .; Niu, Z.; Zhang, L. Bayesian learning-assisted catalyst disco...

    work page 2025

  3. [3]

    Bifunctional electrocatalytic activity of two- dimensional metallophthalocyanine-based metal–organic-frameworks for overall water splitting: A DFT study

    (18) Das, P.; Ball, B.; Sarkar, P. Bifunctional electrocatalytic activity of two- dimensional metallophthalocyanine-based metal–organic-frameworks for overall water splitting: A DFT study. ACS Catal. 2023, 13 (24), 16307-16317. (19) Qiao, B.; Wang, A.; Yang, X.; Allard, L. F.; Jiang, Z.; Cui, Y .; Liu, J.; Li, J.; Zhang, T. Single-atom catalysis of CO oxi...

    work page 2023

  4. [4]

    Design Principles of single-atom catalysts for oxygen evolution reaction: From targeted structures to active sites

    (20) Jiang, F.; Li, Y .; Pan, Y . Design Principles of single-atom catalysts for oxygen evolution reaction: From targeted structures to active sites. Adv. Mater. 2024, 36 (7), 2306309. (21) Rouger, J.; Cavaliere, S.; Jaouen, F. Single -atom catalysts for oxygen evolution reaction in acidic media. Curr. Opin. Electrochem. 2025, 49, 101606. (22) Zhuo, H. -Y...

    work page 2024

  5. [5]

    M.; Kuang, P.; Yu, J

    (34) Zhao, W.; Luo, C.; Lin, Y .; Wang, G.-B.; Chen, H. M.; Kuang, P.; Yu, J. Pt –Ru dimer electrocatalyst with electron redistribution for hydrogen evolution reaction. ACS Catal. 2022, 12 (9), 5540-5548. (35) Kumar, A.; Bui, V . Q.; Lee, J.; Wang, L.; Jadhav, A. R.; Liu, X.; Shao, X.; Liu, Y .; Yu, J.; Hwang, Y .; Bui, H. T. D.; Ajmal, S.; Kim, M. G.; Ki...

    work page 2022

  6. [6]

    H.; Kang, J.; Kim, M

    (36) Jeong, T.; Kim, K.; Kim, B.-H.; Choi, S.-I.; Choi, C. H.; Kang, J.; Kim, M. Ligand engineering of Co-N4 single-atom catalysts for highly-active and stable acidic oxygen evolution. Adv. Sci. 2025, 12 (27), 2502230. (37) Rong, C.; Shen, X.; Wang, Y .; Thomsen, L.; Zhao, T.; Li, Y .; Lu, X.; Amal, R.; Zhao, C. Electronic structure engineering of single-...

    work page 2025

  7. [7]

    (51) Ouyang, R.; Curtarolo, S.; Ahmetcik, E.; Scheffler, M.; Ghiringhelli, L. M. SISSO: A compressed-sensing method for identifying the best low -dimensional descriptor in an immensity of offered candidates. Phys. Rev. Mater. 2018, 2 (8), 083802. (52) Wang, B.; Yang, X.; Xie, C.; Liu, H.; Ma, C.; Zhang, Z.; Zhuang, Z.; Han, A.; Zhuang, Z.; Li, L.; Wang, D...

    work page 2018

  8. [8]

    Synergistic effect of photonic crystals and oxygen vacancies on photoelectrochemical water splitting of TiO2 nanotube

    (56) Meng, M.; Qin, W.; Li, C.; Xu, K.; Xu, L.; Li, J.; Ma, L.; Liu, K.; Li, J.; Qin, N.; Zhu, Y . Synergistic effect of photonic crystals and oxygen vacancies on photoelectrochemical water splitting of TiO2 nanotube. J. Nanoelectron. Optoe. 2020, 15 (2), 226-230(5). (57) Talib, S. H.; Lu, Z.; Yu, X.; Ahmad, K.; Bashir, B.; Yang, Z.; Li, J. Theoretical in...

    work page 2020

  9. [9]

    (70) Osella, S.; Goddard Iii, W. A. CO2 reduction to methane and ethylene on a single- atom catalyst: A grand canonical quantum mechanics study. J. Am. Chem. Soc. 2023, 145 (39), 21319-21329. (71) Sun, F.; Tang, Q.; Jiang, D.-e. Theoretical advances in understanding and designing the active sites for hydrogen evolution reaction. ACS Catal. 2022, 12 (14), ...

    work page 2023

  10. [10]

    Understanding synergistic catalysis on Cu- Se dual atom sites via operando X-ray absorption spectroscopy in oxygen reduction reaction

    (76) Sun, Z.; Zhang, H.; Cao, L.; Liu, X.; Wu, D.; Shen, X.; Zhang, X.; Chen, Z.; Ru, S.; Zhu, X.; Xia, Z.; Luo, Q.; Xu, F.; Yao, T. Understanding synergistic catalysis on Cu- Se dual atom sites via operando X-ray absorption spectroscopy in oxygen reduction reaction. Angew. Chem. Int. Ed. 2023, 62 (13), e202217719. (77) Pei, J.; Yang, L.; Lin, J.; Zhang, ...

    work page 2023

  11. [11]

    The upper -right inset shows representative adsorption configurations of key OER intermediates

    (a) OER overpotentials (ηOER) of TM1TM2-C6-xNx DACs, (TM1TM2 = Co2, Fe2, CoCu, CoNi, FeCu, FeNi, and FeCo ). The upper -right inset shows representative adsorption configurations of key OER intermediates. Linear relationships of ΔG∗OH (b), ΔG∗OOH (c), and ηOER (d) versus ΔG∗O for all TM1TM2-C6-xNx DACs. 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 −0.4 0.0 0.4 0.8 1.2...

    work page arXiv

  12. [12]

    (b) Same as in (a), but for Co2-C4N2-25

    (a) OER free energy diagram and electronic structure analysis of key intermediates on Co 2-N6, including the free energy profile (left), projected density of states (PDOS) of *OH/*O and Co 3d orbitals, and crystal orbital Hamilton population (COHP) analysis of Co –O interactions (middle and right). (b) Same as in (a), but for Co2-C4N2-25. (c) HER free ene...

    work page 1996

  13. [13]

    S.; Lee, S

    (27) Ram, S.; Lee, A. S.; Lee, S. -C.; Bhattacharjee, S. Advanced multifunctional electrocatalysts: Integrating DFT and machine learning for OER, HER, and ORR reactions. Chem. Mater. 2025, 37 (10), 3608-3621

    work page 2025