Mo Atom Rearrangement Drives Layer-Dependent Reactivity in Two-Dimensional MoS2

Jiaxuan Wen; Jing Tang; Keji Lai; Kelvin Huang; Liangbo Liang; Sahar Sharifzadeh; Shaopeng Feng; Tianshu Li; Tina Mihm; Xi Ling

REVIEW 4 major objections 4 minor 4 references

Mo Atom Rearrangement Drives Layer-Dependent Reactivity in Two-Dimensional MoS2

T0 review · 4 major / 4 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read Thinner MoS2 reacts slower because its Mo atoms must rearrange to form the stable MoN phase.

desk verdict The layer-dependent nitridation of MoS2 is real, well measured, and worth publishing; the claim that Mo atom rearrangement is the driver is plausible but not yet demonstrated. read the letter →

arxiv 2509.04648 v1 pith:4TV2SU56 submitted 2025-09-04 cond-mat.mtrl-sci physics.chem-ph

Zifan Wang , Jiaxuan Wen , Tina Mihm , Shaopeng Feng , Kelvin Huang , Jing Tang , Tianshu Li , Liangbo Liang

show 3 more authors

Sahar Sharifzadeh Keji Lai Xi Ling

This is my paper

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

keywords 2Dmaterialslayer-dependentreactivitychemicalRamanspectroscopytransmissionelectronmicroscopeDFTcalculationmolybdenumnitrideMoatomrearrangement

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

The paper reports that when MoS2 reacts with ammonia to become molybdenum nitride, the reaction rate depends strongly on layer number, and the reason is not the usual electronic-structure story. Monolayers are substantially less reactive than few-layer and multilayer flakes, and TEM shows the product from monolayers is a fragmented network of δ-MoN nanocrystals with voids rather than a continuous film. The authors attribute this to geometry: the stable δ-MoN phase has two close-packed Mo layers per unit cell, so a single MoS2 layer must vertically move about half its Mo atoms to build that structure, and that rearrangement carries an energetic cost. Bilayer and trilayer flakes need less out-of-plane migration, so they convert faster and form more continuous films, with 4L giving nearly void-free MoN. The claim matters because it suggests solid-state reaction rates in 2D materials can be governed by the metal-sublattice rearrangement required to form the product phase, independent of binding energies or band structure.

What carries the argument

The load-bearing object is the δ-MoN crystal structure: a hexagonal close-packed Mo sublattice with two Mo atomic layers per unit cell and N atoms in ABAB interstitial sites. Because the product phase has twice as many Mo layers as a monolayer precursor, the transformation from 1L MoS2 must follow an 'N-to-2N' path in which half of the Mo atoms migrate out of plane. This gives a quantitative signature—lateral coverage 0.5 × (2.89 Å/3.19 Å)² ≈ 41%—that the measured coverages for 1L and 2L precursors match.

What would settle it

One decisive experiment: convert monolayer MoS2 to MoN on a clean, Mo-free support and use cross-sectional scanning transmission electron microscopy to count Mo layers in the product. If the product is one Mo layer thick with about 82% lateral coverage, the vertical-rearrangement mechanism is wrong; if it is two Mo layers with about 41% coverage, the mechanism is supported. A complementary test is to grow monolayer MoS2 on a surface that can supply Mo atoms and see whether the inertness disappears.

Watch

Extended reading notes

Core claim

The paper's central claim is that the layer-dependent reactivity of MoS2 in nitridation to molybdenum nitride is controlled by how much the Mo sublattice must rearrange to reach the stable product phase, not by electronic structure or NH3 adsorption. δ-MoN packs two hexagonally close-packed Mo layers per unit cell, so a monolayer MoS2 flake must move about half its Mo atoms into a new out-of-plane layer. That migration is costly, which is why 1L MoS2 converts slowly and incompletely, leaving fragmented δ-MoN nanonetworks at about 44% lateral coverage; 2L still needs rearrangement, while 3L and 4L need little, yielding continuous films. DFT shows NH3 binding is nearly layer-independent, rulin

Load-bearing premise

Everything rests on the structural premise that the stable δ-MoN phase always contains two Mo atomic layers per unit cell, so a monolayer MoS2 must shed half its Mo atoms; if a stable one-Mo-layer MoN existed, or if extra Mo atoms could be supplied, the predicted coverage and barrier would fail.

Editorial extensions

If this is right

Nitridation of 1L and 2L MoS2 produces discontinuous δ-MoN nanonetworks, while 3L and 4L produce continuous films, explaining the conductivity contrast observed by microwave impedance microscopy.
The measured MoN coverages from 1L and 2L precursors (44.5% and 46.9%) match the prediction that half the Mo atoms migrate out of plane, supporting the N-to-2N transformation model.
DFT binding-energy calculations show nearly no layer dependence for NH3 adsorption on pristine or defective MoS2, so the reactivity trend cannot be explained by adsorption energetics.
Monolayer MoS2 takes roughly three times longer than bilayer MoS2 to fully convert (42 min vs 18 min), marking it as exceptionally inert in this reaction.
The general principle is that in solid-state 2D reactions, the extent of metal-sublattice rearrangement needed to form the stable product phase can control both reaction rate and product morphology.

Reading between the lines

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

As an extension beyond the paper's claims, the same mechanism should apply to other atomic-substitution reactions where the stable product has a different number of metal layers per unit cell than the 2D precursor; the measured lateral coverage ratio could serve as a fingerprint of the product's out-of-plane metal count.
A practical, untested corollary is that suppressing vertical Mo migration—for example by supplying Mo from the substrate or applying strain—could make monolayer MoS2 react as quickly as thicker flakes.
The 44% coverage value from monolayer MoS2 is a clean quantitative marker that could be checked from a single TEM image by another lab, independent of the time-course measurements.
If the barrier is purely kinetic, the layer-dependence should compress at high temperature: monolayer MoS2 should eventually convert fully but retain the nanonetwork morphology; the paper's 680 °C conversion is consistent with that, though no quantitative rate comparison was made.

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 is a solid experimental paper with a mechanistic story that outruns its evidence. The layer-dependent nitridation rates in MoS2 and the shift from continuous MoN film at 4L to voided nanonetwork at 1L/2L are new, and they are supported by an unusually broad set of measurements: optical contrast, Raman maps, MIM conductance, and TEM. The best part is the parameter-free coverage check: with δ-MoN's two-Mo-layer unit cell, a monolayer precursor should end up at roughly 41% lateral coverage if it doubles its Mo layers, and the measured 1L/2L values are 44.5% and 46.9%. That is independent, quantitative support for the core geometry, not just a narrative. The DFT section also does a real job of ruling out NH3 binding energy as the source of the layer dependence, so the paper clears away the obvious electronic-structure alternative.

The weakness is that the central causal claim—Mo atom rearrangement drives the reactivity—is inferred, not demonstrated. No transition-state search or molecular dynamics is done for Mo migration; the DFT is only adsorption energies on static surfaces. No cross-sectional imaging shows that the monolayer product actually has two Mo layers. The 3L coverage argument is fitted rather than predicted, and the "half of the Mo atoms" wording does not survive atom counting: going from three Mo layers to four requires one extra layer's worth of atoms, i.e. one-third, if the region is closed. The 4L discrepancy is absorbed by an unquantified strain term. Coverage numbers come from thresholded TEM images without error bars, and the fully converted TEM samples were reacted at 680 °C while the kinetics were measured at 650 °C, so the morphology and rate data come from different reaction conditions.

None of this undermines the experimental finding, but it does mean the mechanism is a plausible interpretation, not a demonstrated cause. The paper would be much stronger if the authors computed or measured the migration barrier, or at least explicitly framed the rearrangement story as a hypothesis consistent with the data. The experiments are important enough, and the 1L/2L coverage match is genuinely probative, so I would send it to peer review and ask for the mechanistic claims to be softened or backed up before acceptance. People working on 2D-to-3D conversion chemistry will get real use from this.

Referee Report

4 major / 4 minor

Summary. The paper reports layer-dependent reactivity of MoS2 in a nitridation atomic-substitution reaction: monolayer MoS2 converts much more slowly than bilayer, trilayer, and thicker flakes. Using optical imaging, Raman intensity maps, and MIM conductance maps, the authors show that conversion time decreases with thickness in the 1L-4L regime. TEM shows that 1L and 2L precursors produce discontinuous MoN nanonetworks with voids, while 3L and 4L produce continuous films. The authors propose that the thermodynamically stable δ-MoN phase requires two Mo layers per unit cell, so ultrathin precursors must undergo vertical Mo atom rearrangement; the energy cost of this rearrangement is proposed to explain the layer-dependent kinetics. DFT calculations rule out NH3 binding energy as a source of layer dependence. A geometric coverage model predicts 41% lateral coverage for 'N-to-2N' conversion, matching ~44-47% measured coverage for 1L and 2L.

Significance. If the mechanism is correct, this is a conceptually important result: it would show that in solid-state 2D transformations, the extent of metal-sublattice rearrangement required to form the stable product phase can dominate reactivity, beyond electronic-structure arguments such as Marcus theory. The experimental dataset is strong: the central fact of slower/thinner conversion is supported by multiple independent probes, and the DFT section cleanly excludes NH3 binding as a thickness-dependent factor. The 1L/2L coverage matching a parameter-free geometric model is a notable positive. However, the causal claim that Mo atom rearrangement is the driver is not yet demonstrated because the barrier is not calculated or measured and the structural layer count of the product is inferred, not observed.

major comments (4)

[Results and Discussion – Fig. 3d coverage model] Fig. 3d and the accompanying paragraph: the 3L interpretation is quantitatively inconsistent. The text says "approximately half of the Mo atoms in 3L MoS2 undergo vertical rearrangement to form a four-Mo-layer MoN," but a 3L-to-4L conversion conserves atoms only if 1/4 of the Mo atoms (one layer out of three) migrate, giving ~62% coverage. With the measured 72.25% coverage, the implied mixture is roughly half of the area following 3L-to-4L and half following N-to-N, corresponding to ~1/8 of Mo atoms migrating. Please correct this statement and specify the mixture model with the conservation equation.
[Results and Discussion – Mo rearrangement mechanism and DFT] The central claim that the layer-dependent reactivity is due to the "high energy cost associated with Mo atom diffusion and migration" is not supported by any calculation or measurement of that barrier. The DFT section (Fig. 4) only computes NH3 binding on static surfaces; no NEB, AIMD, or experimental activation energy is reported. Without a barrier estimate, the delayed conversion in 1L/2L could equally arise from partial conversion, etch-induced material loss, or strain-driven islanding. Please add a direct barrier calculation or an experiment isolating the Mo-migration step, or soften the causal claim accordingly.
[Results and Discussion – TEM characterization (Fig. 3b)] No cross-sectional or thickness-sensitive measurement confirms that the 1L-derived product actually contains two Mo layers and the 2L product four. The coverage ratios (44.5% and 46.9% vs 41%) are consistent with the geometric model but not probative: the same numbers could result from incomplete conversion or material loss. Provide cross-sectional TEM/STEM or AFM thickness of the converted films to verify the layer count.
[Results and Discussion – 4L discrepancy] The 4L coverage exceeding both model predictions is attributed to "residual strain in the MoN film that is not fully released," but no strain measurement or calculation is given. Since this is the only term that prevents the model from being falsified for 4L, please quantify the strain (e.g., from FFT d-spacing in HRTEM, or from DFT-relaxed lattice constants) or otherwise bound its magnitude.

minor comments (4)

[Fig. 2 caption] "layered-dependent" should be "layer-dependent."
[Fig. 4 caption and text] "absorption" should be "adsorption" in the Fig. 4 caption, and "chemsorption" in the main text is a typo for "chemisorption."
[Methods – coverage estimation] The coverage threshold is described as "empirically determined." Please report the threshold value and provide a sensitivity analysis, since the quantitative conclusions depend on the binarization choice.
[Results and Discussion – coverage definitions] The terms "N-to-N" and "N-to-2N" are used before being defined; define them explicitly at first use in the text.

Circularity Check

2 steps flagged · score 4.0 of 10

One load-bearing self-citation for the 3L pathway; 1L/2L core mechanism has independent geometric and DFT support.

self citation load bearing [Results and Discussion, Figure 3d coverage analysis (paragraph beginning 'For the 3L region...')]
"Our previous work has shown that the MoN structure contains four Mo atom layers exhibits a higher thermodynamic stability compared to three-layer variants,38 which further support this “3L-to-4L” transformation preference in the 3L case."

The 3L coverage (72.25%) sits between the N-to-N (82.1%) and 3L-to-4L (~62%) model values. To decide that the 3L region preferentially forms a four-Mo-layer MoN, the paper invokes a stability ordering taken from its own prior work (ref 38) rather than deriving it from calculations reported in this manuscript. That prior result is not reproduced or independently verified here, and it is load-bearing: without the self-cited stability preference, the 3L data are equally consistent with a mixture of pathways, and the paper's mixed-rearrangement explanation would lack independent justification. This is a self-citation used to select the interpretation of a central data point.
other [Results and Discussion, Figure 3d coverage analysis (paragraph beginning 'For the 3L region...')]
"For the 3L region, the measured MoN coverage is 72.25%, which lies between the values predicted in the ‘N-to-N’ (82.1%) and ‘3L-to-4L’ (~62%) models. We believe in this case, approximately half of the Mo atoms in 3L MoS2 undergo vertical rearrangement to form a four-Mo-layer MoN, while the other half retain their original plane following the ‘N-to-N’ model."

The fraction 'approximately half' is not obtained from an independent calculation; it is chosen post hoc so that a two-component average (roughly midway between 82.1% and 62%) lands near the experimentally measured 72.25%. The paper then presents this as a structural transformation pathway that explains the 3L data, but the pathway is effectively parameterized by the data point it is supposed to explain. This is fitting the observed coverage and repackaging it as a mechanistic model rather than making a falsifiable prediction. Separately, atom conservation would require one third, not one half, of the Mo atoms to move to convert three layers into four, so the stated fraction is also internally inconsistent.

full rationale

The central monolayer/bilayer argument is not circular: the N-to-2N coverage prediction (41.0%) is a parameter-free geometric consequence of the assumed δ-MoN two-layer structure and the lattice constants, and it is compared with independently measured TEM coverages (44.5%, 46.9%). The DFT NH3-binding calculations are also independent and rule out an electronic-structure explanation. However, the 3L interpretation depends on a self-cited stability ordering (ref 38) and on a post hoc 'approximately half' rearrangement fraction that is adjusted to match the measured 72.25% coverage, making that branch of the argument partially circular. These issues affect the 3L (and to a lesser extent 4L) data, not the core 1L/2L reactivity observation. The paper also does not directly compute or measure the Mo-migration barrier, but that is an evidence gap rather than a circularity. Overall score 4: some self-citation and one fitted rationalization, while the central claim retains independent content.

Assumptions & free parameters 2 free parameters · 5 assumptions · 0 invented entities

The paper introduces no new particles or forces. Its explanatory burden rests on structural assumptions from the literature (the two-Mo-layer delta-MoN unit cell; the stability of four-layer over three-layer MoN) and on one ad hoc assumption about strain relaxation for the 4L product. The only genuinely fitted numerical free parameters are the 3L rearrangement fraction and the unspecified binarization threshold. The absence of a computed Mo migration barrier means the central kinetic claim relies on these structural premises rather than on a direct calculation.

free parameters (2)

Fraction of Mo atoms vertically rearranged in 3L MoS2 conversion = ~0.5
Chosen post hoc so that the blended N-to-N and 3L-to-4L coverage model matches the measured 72.25% coverage for the 3L region; no independent measurement.
Binarization threshold for TEM coverage estimation = Empirical, unspecified
The SI says an 'empirically determined threshold' is applied to binarize TEM images before coverage calculation; the threshold value is not given and the coverage percentages have no error bars.

assumptions (5)

domain assumption Delta-MoN crystallizes with a unit cell of two hexagonal close-packed Mo layers and ABAB N stacking
Stated in the Results (refs 51-57). This structure is the basis of the N-to-2N coverage model: a monolayer MoS2 would need to double its Mo layer count, losing half its lateral area.
domain assumption MoN structures with four Mo layers are more thermodynamically stable than three-layer variants
Invoked to justify the 3L-to-4L preference; this ordering is taken from the authors' previous work (ref 38) rather than computed or measured here.
domain assumption Mo atoms are conserved locally during conversion, and all lateral area loss corresponds to vertical Mo layer formation
The coverage model equates measured area loss with Mo atoms moving out of plane; alternative sinks such as lateral diffusion, sublimation, or substrate incorporation are not considered.
ad hoc to paper Lattice mismatch strain is fully relaxed in 1L-3L MoN products but not in the 4L-derived film
Used to explain why the measured 4L coverage exceeds the 82.1% N-to-N prediction ('This discrepancy is likely due to residual strain in the MoN film...'). Not quantified or independently supported.
domain assumption PBE DFT with fixed experimental lattice vectors and a 10 Angstrom vacuum is adequate to compare NH3 adsorption across 1L-4L MoS2
Used to rule out NH3 binding as the source of the layer-dependent reactivity. No van der Waals correction is applied, which could matter for the weakly physisorbed (20-36 meV) configurations.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Mo Atom Rearrangement Drives Layer-Dependent Reactivity in Two-Dimensional MoS2." pith.science (2026). https://pith.science/paper/4TV2SU56

@misc{pith2026250904648,
  author       = {Pith},
  title        = {Pith review of: Mo Atom Rearrangement Drives Layer-Dependent Reactivity in Two-Dimensional MoS2},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4TV2SU56}},
  note         = {Machine review of arXiv:2509.04648}
}

read the original abstract

Two-dimensional (2D) materials offer a valuable platform for manipulating and studying chemical reactions at atomic level, owing to the ease of controlling their microscopic structure at the nanometer scale. While extensive research has been conducted on the structure-dependent chemical activity of 2D materials, the influence of structural transformation during the reaction remains largely unexplored. In this work, we report the layer-dependent chemical reactivity of MoS2 during a nitridation atomic substitution reaction and attribute it to the rearrangement of Mo atoms. Our results show that the chemical reactivity of MoS2 decreases as the number of layers is reduced in the few-layer regime. In particular, monolayer MoS2 exhibits significantly lower reactivity compared to its few-layer and multilayer counterparts. Atomic-resolution transmission electron microscope (TEM) reveals that MoN nanonetworks form as reaction products from monolayer and bilayer MoS2, with the continuity of the MoN crystals increasing with layer number, consistent with the local conductivity mapping data. The layer-dependent reactivity is attributed to the relative stability of the hypothetically formed MoN phase which retain the number of Mo atomic layers present in the precursor. Specifically, the low chemical reactivity of monolayer MoS2 is attributed to the high energy cost associated with Mo atom diffusion and migration necessary to form multi-layer Mo lattices in the thermodynamically stable MoN phase. This study underscores the critical role of lattice rearrangement in governing chemical reactivity and highlights the potential of 2D materials as versatile platforms for advancing the understanding of materials chemistry at atomic scale.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

4 extracted references · 1 canonical work pages

[1]

(1) Shan, J.-Y .; Morrison, N.; Chen, S.-D.; Wang, F.; Ma, E. Y . Johnson-Noise-Limited Cancellation-Free Microwave Impedance Microscopy with Monolithic Silicon Cantilever Probes. Nat Commun 2024, 15 (1),

work page 2024
[368]

(3) Kresse, G.; Furthmüller, J

https://doi.org/10.1016/S0734-189X(87)80186-X. (3) Kresse, G.; Furthmüller, J. Efficient Iterative Schemes for Ab Initio Total- Energy Calculations Using a Plane-Wave Basis Set. Phys Rev B 1996, 54 (16), 11169–11186. https://doi.org/10.1103/PhysRevB.54.11169. (4) Kresse, G.; Furthmüller, J. Efficiency of Ab -Initio Total Energy Calculations for Metals and...

doi:10.1016/s0734-189x(87)80186-x 1996
[5043]

(2) Pizer, S

https://doi.org/10.1038/s41467- 024-49405-8. (2) Pizer, S. M.; Amburn, E. P.; Austin, J. D.; Cromartie, R.; Geselowitz, A.; Greer, T.; ter Haar Romeny, B.; Zimmerman, J. B.; Zuiderveld, K. Adaptive Histogram Equalization and Its Variations. Comput Vis Graph Image Process 1987, 39 (3), 355–

doi:10.1038/s41467- 1987
[6293]

https://doi.org/10.1038/ncomms7293

doi:10.1038/ncomms7293

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.

[1] [1]

(1) Shan, J.-Y .; Morrison, N.; Chen, S.-D.; Wang, F.; Ma, E. Y . Johnson-Noise-Limited Cancellation-Free Microwave Impedance Microscopy with Monolithic Silicon Cantilever Probes. Nat Commun 2024, 15 (1),

work page 2024

[2] [368]

(3) Kresse, G.; Furthmüller, J

https://doi.org/10.1016/S0734-189X(87)80186-X. (3) Kresse, G.; Furthmüller, J. Efficient Iterative Schemes for Ab Initio Total- Energy Calculations Using a Plane-Wave Basis Set. Phys Rev B 1996, 54 (16), 11169–11186. https://doi.org/10.1103/PhysRevB.54.11169. (4) Kresse, G.; Furthmüller, J. Efficiency of Ab -Initio Total Energy Calculations for Metals and...

doi:10.1016/s0734-189x(87)80186-x 1996

[3] [5043]

(2) Pizer, S

https://doi.org/10.1038/s41467- 024-49405-8. (2) Pizer, S. M.; Amburn, E. P.; Austin, J. D.; Cromartie, R.; Geselowitz, A.; Greer, T.; ter Haar Romeny, B.; Zimmerman, J. B.; Zuiderveld, K. Adaptive Histogram Equalization and Its Variations. Comput Vis Graph Image Process 1987, 39 (3), 355–

doi:10.1038/s41467- 1987

[4] [6293]

https://doi.org/10.1038/ncomms7293

doi:10.1038/ncomms7293