Mechanistic Insights into Water-Splitting, Proton Migration, and Hydrogen Evolution Reaction in g-C3N4/TiO2-B and Li-F co-doped Heterostructures

Hanyang Ji; Qi Jiang; Shuang Qiu; Shuhan Tang; Xiaojie Liu

REVIEW 3 major objections 5 minor 5 references

Mechanistic Insights into Water-Splitting, Proton Migration, and Hydrogen Evolution Reaction in g-C3N4/TiO2-B and Li-F co-doped Heterostructures

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read This paper claims that proton migration from the TiO2-B(001) surface to the interface is the rate-determining step for hydrogen evolution, with a 1.103 eV barrier that Li–F co-doping lowers to 0.999 eV.

desk verdict Routine follow-up DFT study whose central HER efficiency claim is contradicted by its own ~1 eV proton-migration barrier and non-optimal adsorption free energies. read the letter →

arxiv 2506.05021 v1 pith:Y3UA7A3O submitted 2025-06-05 cond-mat.mtrl-sci physics.chem-ph

Shuhan Tang , Qi Jiang , Shuang Qiu , Hanyang Ji , Xiaojie Liu This is my paper

classification cond-mat.mtrl-sciphysics.chem-ph

keywords photocatalyticwatersplittinghydrogenevolutionreactiong-C3N4/TiO2-BheterostructureLi-Fco-dopingprotonmigrationbarrierCI-NEBdensityfunctionaltheoryGibbsfreeenergyofadsorption

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

Using density functional theory, this paper traces the full hydrogen evolution path on g-C3N4/TiO2-B(001) heterostructures: water adsorption and splitting on the titania surface, proton migration across the interface, and H2 formation on the g-C3N4 surface. The central claim is that the step controlling the overall rate is proton migration from the TiO2-B(001) surface to the interface, which must break and reform a polar O–H covalent bond and costs 1.103 eV in the pristine heterostructure. Li–F co-doping weakens that bond and strengthens interfacial polarization, lowering the barrier to 0.999 eV, while interfacial O–H···N hydrogen bonds make the later hop from the interface to g-C3N4 much easier (0.168 eV pristine, 0.226 eV doped). The paper concludes that both heterostructures are promising HER photocatalysts, with the doped system also requiring an interfacial proton-diffusion step of 0.701 eV that it argues does not determine the overall rate.

What carries the argument

The load-bearing machinery is the proton-transfer coordinate O29→O12→N16 (pristine) and O29→O9→O12→N16 (doped), resolved with climbing-image nudged elastic band calculations. The barrier heights are controlled by two competing interactions: a polar O–H covalent bond whose cleavage and recombination costs about 1 eV, and an O–H···N hydrogen bond to pyridine-type nitrogen that stabilizes the interface and lowers the later hop to 0.168 or 0.226 eV. Bader charge analysis, charge-density differences, and projected density of states are used to show that the proton loses about 0.6|e| to oxygen and that hybridization between the proton and its neighbors weakens at the transition state.

What would settle it

Recompute the same CI-NEB pathways with a dispersion-corrected functional, for example DFT-D3 or optB88-vdW, and check whether the 1.103 eV and 0.999 eV surface-to-interface barriers, and the minimum-energy path itself, survive; a shift of more than about 0.2 eV would overturn the rate-determining-step assignment.

Watch

Extended reading notes

Core claim

The authors establish that the rate-determining elementary step in the g-C3N4/TiO2-B(001) heterostructure HER cycle is not water splitting (0.442 eV, endothermic with no transition state) nor the final proton hop to g-C3N4 (0.168 eV), but the migration of a proton from the TiO2-B(001) surface to the interface, where oxygen acts as both proton donor and acceptor through a polar O–H covalent bond; this barrier is 1.103 eV. Li–F co-doping weakens the O–H bond and adds interfacial polarization, reducing that barrier to 0.999 eV, and introduces an interfacial diffusion step (0.701 eV) that the authors argue does not control overall HER. The thermodynamic driving force is a monotonic decrease in proton adsorption energy from the TiO2-B surface (0.041 eV) to the interface (−0.380 eV) to g-C3N4 (−0.586 eV; −0.646 eV after doping), and the computed hydrogen adsorption free energies $ΔG(*H)$ are −0.586 eV and −0.646 eV, which the paper reads as a favorable balance of adsorption and desorption.

Load-bearing premise

The whole barrier picture rests on the assumption that the computer model describes the weak forces between the stacked layers accurately enough, even though the calculation uses a standard approximation that often underestimates such forces.

Editorial extensions

If this is right

Water splitting on these surfaces is endothermic and has no transition state, so its energy cost (0.442, 0.441, and 0.338 eV) is a thermodynamic hurdle rather than a kinetic one, and Li–F co-doping cuts that cost by about 0.1 eV.
The proton's adsorption energy falls monotonically from the TiO2-B surface to the interface to the g-C3N4 surface, so after water splitting the proton is thermodynamically pulled across the junction.
The rate-limiting step is the surface-to-interface hop (1.103 eV), not the interface-to-g-C3N4 hop (0.168 eV), because the first hop requires breaking a polar O–H bond while the second is assisted by an O–H···N hydrogen bond.
Li–F co-doping lowers the rate-limiting barrier to 0.999 eV by weakening the O–H bond and enhancing interfacial polarization, and it adds a separate interfacial diffusion barrier (0.701 eV) that does not determine overall HER.
The calculated $ΔG(*H)$ values (−0.586 eV pristine, −0.646 eV doped) imply stable proton adsorption; the paper concludes this is a favorable balance while noting that stronger adsorption can hinder H2 desorption.

Reading between the lines

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

The paper stops at electronic-structure barriers; a microkinetic model translating the 1.103 and 0.999 eV barriers into turnover rates would clarify whether stirring and mild heating are truly enough to drive HER at practical rates.
Because the calculations use the PBE functional without dispersion corrections, the absolute barriers are more uncertain than the relative trend; the comparison between pristine and doped systems is likely more robust than any single barrier value.
The monotonically decreasing adsorption energies suggest that the interface itself could act as a proton reservoir; one testable design extension is to introduce dopants that stabilize the interface state without adding a 0.701 eV diffusion step.
If the rate-limiting step is O–H bond cleavage during the surface-to-interface hop, a normal kinetic isotope effect for H/D substitution should appear in the HER rate; measuring it would provide an experiment-level check of the mechanism.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Short version: this paper adds real mechanistic numbers to a system the authors already published, but the headline claim dies on the authors' own numbers. The ~1 eV proton-migration barrier is the rate-determining step they identify, and at 300 K that barrier gives a per-site rate of roughly 10^-6 s^-1 or slower. Calling it 'not sufficiently high' is not supported by any kinetic estimate in the paper. The paper would need tunneling, hot-carrier effects, or an overpotential argument to rescue it, and it provides none.

What is new: the water dissociation energies for the heterostructures (0.441 and 0.338 eV), the CI-NEB pathways that separate surface-to-interface (1.103/0.999 eV) from interface-to-g-C3N4 (0.168/0.226 eV) transport, and the ΔG(*H) values (-0.586 and -0.646 eV). These are concrete numbers absent from the group's prior work, and the charge-density/PDOS analysis of the transition states is competently done.

Soft spots, in order: first, the internal contradiction. The abstract claims 'optimal proton adsorption and desorption characteristics,' but in the HER section the authors admit that the more negative ΔG(*H) 'could impede H2 desorption.' Both cannot be true, and the computed values are far from the Sabatier optimum near zero. Second, the 1 eV barrier is treated as an acceptable hurdle without a single rate estimate. Eyring with a generous prefactor gives k ~ 10^-6 s^-1 per site at 300 K, i.e., one proton hop per site every several days; the paper's 'heating and stirring' remark does not address the photochemical context. Third, the lack of any van der Waals correction for a stacked layered interface is a real methodological weakness. I consider it secondary, because the rate problem would survive a re-relaxation, but it should be fixed if the numbers are to be trusted.

Minor items: zero-point energies are not included for the barriers, which matters for proton transfer; the adsorption-energy definition for proton includes the H2 binding energy, which is a bit unusual; and the polarization direction is taken from the authors' ref. 13, which is fine as background.

Who gets value: computational researchers working on g-C3N4/TiO2 interfaces could use the pathway analysis as a starting template, but not the conclusion that Li-F doping yields an efficient HER photocatalyst.

I would send this to review rather than desk-reject, because the mechanistic data deserve scrutiny and the contradiction is exactly what a referee should catch. But I would not accept it in current form; the efficiency claim needs either quantitative kinetic support or removal.

Referee Report

3 major / 5 minor

Summary. The paper reports DFT-PBE calculations of water adsorption/dissociation, proton adsorption and migration barriers, and HER free energies for g-C3N4/TiO2-B(001) heterostructures with and without Li-F co-doping. It concludes that both heterostructures are efficient HER photocatalysts, identifies proton migration from the TiO2-B(001) surface to the interface as the rate-determining step (barriers 1.103 eV and 0.999 eV), and reports hydrogen adsorption free energies of -0.586 eV and -0.646 eV.

Significance. If the quantitative results were reliable, the systematic CI-NEB study of proton pathways, the Bader-charge/PDOS analysis, and the comparison with literature water-splitting energies would provide a useful mechanistic reference for g-C3N4/TiO2-B interfaces. The manuscript also includes machine-checkable reproducible data availability statements and verifies transition states by vibrational analysis. However, the central efficiency claim is contradicted by the paper's own computed barriers and adsorption free energies, and the absence of dispersion corrections undermines the quantitative interface energetics. As it stands, the contribution is not reliable enough to support the stated conclusions.

major comments (3)

[Proton migration from TiO2-B(001) to g-C3N4 surface (Fig. 4)] The claim that an ~1 eV barrier (1.103 eV pristine, 0.999 eV doped) "is not sufficiently high to inhibit the HER process" is quantitatively unsupported. A transition-state-theory estimate with the kBT/h prefactor gives ~10^-19 s^-1 at 300 K; even with a generous prefactor of 10^13 s^-1, the rate is ~10^-6 s^-1 per site, meaning one proton transfer every ~12 days. The manuscript provides no rate model, no overpotential or electrochemical driving force, and no tunneling analysis. This is an internal inconsistency independent of the choice of exchange-correlation functional and directly undermines the paper's efficiency conclusion.
[HER process on different g-C3N4 surfaces; Abstract] The abstract claims the heterojunction surface exhibits "optimal proton adsorption and desorption characteristics," but the computed ΔG(*H) values are -0.586 eV and -0.646 eV, far from the Sabatier optimum near 0 eV. The text itself admits that Li-F doping "could impede H2 desorption" and that "excessively strong adsorption can inhibit H2 production." These statements directly contradict the abstract's central claim; the authors should quantify the effect of these free energies on HER rate through a microkinetic or exchange-current-density analysis rather than relying on qualitative 'optimal' language.
[Calculation methods] The methods section specifies the PBE functional with no van der Waals correction for a layered g-C3N4/TiO2-B(001) heterostructure. Dispersion interactions dominate the interlayer binding and affect the interface geometry, adsorption energies, and NEB barriers along the proton migration path. Since every calculated barrier and adsorption energy feeds directly into the mechanistic claims, the authors should provide at least a benchmark of the key IS/TS/FS energies with a dispersion-corrected method (e.g., DFT-D3, dDsC, or optB88-vdW). Without this, the quantitative barriers, especially the rate-determining 1 eV step, are not reliable.

minor comments (5)

[Water-splitting on different TiO2-B(001) surfaces] The text states that "the adsorption energy of the adsorbed water system is 0.442 eV lower than that of the dissociated system" and then later refers to a "dissociation adsorption energy of 0.177 eV, with an increase of 0.441 eV after dissociation." These statements are ambiguous and appear inconsistent; please report the reaction energy explicitly as endothermic by a single number and define all quantities used.
[Fig. 1(b)] The Ti13 site is introduced as the most stable adsorption site, but the manuscript does not define which titanium atom is labeled Ti13 in the structural figures; please indicate it clearly in the figure or caption.
[Calculation methods] The manuscript says all transition states were verified by imaginary frequency analysis, but no imaginary frequencies or vibrational analysis details are reported anywhere; please provide the values or cite the verification explicitly.
[Throughout] The manuscript contains numerous typographical errors, including "emloyed," "ehxibits," "adsoprtion," "favorbale," "hte," "remarkbale," "stesps," "chanllenge," "efficiecy," "stratey," and "the the." A thorough language edit is needed.
[Author contributions] The author list includes Hanyang Jia, but the contributions section lists "Hanyang Ji"; please correct this inconsistency.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the computed barriers, adsorption energies, and free energies are independent DFT results; self-citations are background, not load-bearing.

full rationale

The paper's derivation chain is based on direct DFT calculations: adsorption energies from total-energy differences, water-splitting energetics from optimized initial/final states, proton migration barriers from CI-NEB, and HER activity from computed ΔG(*H) values. No parameter is fitted to the conclusions, and no equation defines a predicted barrier in terms of the input data. The self-citations (refs 5, 12, 13) are used to motivate the heterostructure and to invoke an interfacial polarization direction, but the central numbers—1.103 eV, 0.999 eV, 0.168/0.226 eV, and ΔG(*H) values—are obtained from the present VASP calculations and are supported in-paper by bond-length, Bader charge, charge-density, and PDOS analyses. Even if the cited prior work were ignored, the barrier reductions and adsorption trends would still be independently computed here. The paper does contain internal numerical tensions, such as calling a ~1 eV proton-migration barrier compatible with efficient HER despite its low thermal rate, and describing ΔG(*H) values near −0.6 eV as 'optimal' while admitting they 'could impede H2 desorption.' Those are quantitative-consistency or correctness concerns, not circularity, because they do not involve an input being renamed as a prediction. Therefore no circular step is identified.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

The central results rest entirely on the accuracy of the DFT model and the preselected reaction pathways. No fitting parameters are introduced; the main free choices are the exchange-correlation functional, the absence of vdW corrections, and the choice of proton migration paths.

assumptions (4)

domain assumption PBE functional and PAW method with 400 eV cutoff and 2x2x1 k-mesh are accurate enough for the reported energies.
Methods section: used for all calculations; no convergence tests shown.
domain assumption Van der Waals interactions can be neglected for the g-C3N4/TiO2-B(001) interface.
Methods do not include dispersion corrections; layered heterostructure binding is dominated by vdW, so this is an unflagged assumption.
domain assumption The proton migration pathways O29→O12→N16 (and O9→O12 in doped) are the relevant dominant pathways.
Chosen based on adsorption energy minima; alternative pathways, coverage effects, and proton concentration are not considered.
domain assumption CI-NEB with the given settings reliably finds no transition state for water dissociation, leading to the conclusion that water splitting is thermodynamic-driven.
Water splitting section; lack of a TS could be an artifact of the search method or functional.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Mechanistic Insights into Water-Splitting, Proton Migration, and Hydrogen Evolution Reaction in g-C3N4/TiO2-B and Li-F co-doped Heterostructures." pith.science (2026). https://pith.science/paper/Y3UA7A3O

@misc{pith2026250605021,
  author       = {Pith},
  title        = {Pith review of: Mechanistic Insights into Water-Splitting, Proton Migration, and Hydrogen Evolution Reaction in g-C3N4/TiO2-B and Li-F co-doped Heterostructures},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Y3UA7A3O}},
  note         = {Machine review of arXiv:2506.05021}
}

read the original abstract

Solar water splitting has received a lot of attention due to its high efficiency and clean energy production potential. Herein, based on the band alignment principle, the g-C3N4/TiO2-B(001) heterostructure is strategically designed, then a Li-F co-doping approach is developed and implemented, leading to significant enhancement in the photocatalytic hydrogen evolution efficiency of the heterostructure systems. The decomposition of water molecule on the surface of heterostructures, the migration and diffusion of proton across the interface, and the hydrogen evolution performance are systematically studied and comprehensively analyzed. The results demonstrate that the heterojunction surface exhibits a relatively low energy barrier for water decomposition, facilitating both hydrogen evolution reaction (HER) and oxygen evolution reaction (OER). Proton transfer preferentially occurs from the TiO2-B(001) surface to the g-C3N4 surface through the interface. The presence of polar covalent bonds establishes a substantial energy barrier for proton migration from TiO2-B(001) surface to the interface, representing a rate-determining factor in the hydrogen evolution process. The formation of hydrogen bonds significantly reduces the migration energy barrier for protons crossing the interface to the g-C3N4 surface. Hydrogen adsorption free energy analysis show that that the heterojunction surface exhibits optimal proton adsorption and desorption characteristics. The synergistic combination of low water decomposition energy barrier, reduced proton migration energy barriers and exceptional HER performance endows both g-C3N4/TiO2-B(001) heterostructure and Li-F co-doped g-C3N4/TiO2-B(001) heterojunction with remarkbale potential as efficient HER photocatalyst.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

5 extracted references · 5 canonical work pages

[1]

Based on previous thermodynamics analyses, proton tends to adsorb on pyridinic-type nitrogen on the g-C3N4 surface

and efficient desorption of H2 (e.g., ΔG(*H2) > 0). Based on previous thermodynamics analyses, proton tends to adsorb on pyridinic-type nitrogen on the g-C3N4 surface. The adsorption configurations of proton on different g-C3N4 surfaces are plotted in Fig. 6(a). To further evaluate the HER performance of these three photocatalysts, the Gibbs free energy d...

work page 1972
[50]

Accurate and simple analytic representation of the electron-gas correlation energy

15 Perdew J P, Wang Y. Accurate and simple analytic representation of the electron-gas correlation energy. Physical review B, 1992, 45, 13244. 16 Perdew J P, Burke K, Ernzerhof M. Generalized gradient approximation made simple. Physical review letters, 1996, 77,

work page 1992
[952]

Mechanistic insights into water adsorption and dissociation on amorphous TiO2-based catalysts

29 Ghuman K K. Mechanistic insights into water adsorption and dissociation on amorphous TiO2-based catalysts. Science and Technology of Advanced Materials, 2018, 19, 44-52. 30 Che M. Nobel Prize in chemistry 1912 to Sabatier: Organic chemistry or catalysis?. Catalysis today, 2013, 218, 162-171

work page 2018
[1644]

Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane- wave basis set

14 Kresse G, Furthmüller J. Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane- wave basis set. Computational materials science, 1996, 6, 15-

work page 1996
[3865]

Adsorbate-substrate and adsorbate-adsorbate interactions of Na and K adlayers on Al (111)

17 Neugebauer J, Scheffler M. Adsorbate-substrate and adsorbate-adsorbate interactions of Na and K adlayers on Al (111). Physical Review B, 1992, 46, 16067. 18 Bengtsson L. Dipole correction for surface supercell calculations. Physical Review B, 1999, 59, 12301. 19 Blöchl P E. Projector augmented-wave method. Physical review B, 1994, 50, 17953. 20 Henkelm...

work page 1992

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [1]

Based on previous thermodynamics analyses, proton tends to adsorb on pyridinic-type nitrogen on the g-C3N4 surface

and efficient desorption of H2 (e.g., ΔG(*H2) > 0). Based on previous thermodynamics analyses, proton tends to adsorb on pyridinic-type nitrogen on the g-C3N4 surface. The adsorption configurations of proton on different g-C3N4 surfaces are plotted in Fig. 6(a). To further evaluate the HER performance of these three photocatalysts, the Gibbs free energy d...

work page 1972

[2] [50]

Accurate and simple analytic representation of the electron-gas correlation energy

15 Perdew J P, Wang Y. Accurate and simple analytic representation of the electron-gas correlation energy. Physical review B, 1992, 45, 13244. 16 Perdew J P, Burke K, Ernzerhof M. Generalized gradient approximation made simple. Physical review letters, 1996, 77,

work page 1992

[3] [952]

Mechanistic insights into water adsorption and dissociation on amorphous TiO2-based catalysts

29 Ghuman K K. Mechanistic insights into water adsorption and dissociation on amorphous TiO2-based catalysts. Science and Technology of Advanced Materials, 2018, 19, 44-52. 30 Che M. Nobel Prize in chemistry 1912 to Sabatier: Organic chemistry or catalysis?. Catalysis today, 2013, 218, 162-171

work page 2018

[4] [1644]

Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane- wave basis set

14 Kresse G, Furthmüller J. Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane- wave basis set. Computational materials science, 1996, 6, 15-

work page 1996

[5] [3865]

Adsorbate-substrate and adsorbate-adsorbate interactions of Na and K adlayers on Al (111)

17 Neugebauer J, Scheffler M. Adsorbate-substrate and adsorbate-adsorbate interactions of Na and K adlayers on Al (111). Physical Review B, 1992, 46, 16067. 18 Bengtsson L. Dipole correction for surface supercell calculations. Physical Review B, 1999, 59, 12301. 19 Blöchl P E. Projector augmented-wave method. Physical review B, 1994, 50, 17953. 20 Henkelm...

work page 1992