{"id":"8e55d251-aaeb-42c7-9871-0ffaf8ba5527","arxiv_id":"2505.24644","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Removing the top sulfur layer of monolayer MoS2 triggers spontaneous sulfur migration from the bottom layer through the molybdenum plane, creating a stable MoSx structure on simulation timescales.","lead":"Computer simulations show that when an entire top sulfur layer is removed from monolayer molybdenum disulfide, sulfur atoms from the bottom layer push through the metal plane and the sheet remains intact. This suggests a way to design sulfur-depleted 2D materials, with a substrate defect-size threshold that controls whether the sheet reorganizes or seals itself.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Classical MD evidence for the central S-migration claim rests on an unidentified ReaxFF parameter set; Ref. 39 is a hydrocarbon force field, so the large-system and 4 nm2 threshold conclusions are not reproducible as reported.","rationale":"The reader's weakest_assumption identified the missing ReaxFF parameterization and its consequences for the migration mechanism and the substrate-size threshold. My independent reading reaches the same conclusion: the central large-scale claim has no traceable force-field basis. I do not see a more load-bearing issue. The AIMD 3x3 result is a genuine independent data point, so the qualitative small-cell migration is plausible, but it cannot carry the size-dependent threshold and long-time stability statements. Because the omission is concrete and fixable, and because the central qualitative result is supported by two methods in the small-cell limit, the appropriate verdict remains the reader's CONDITIONAL. I would not move to REJECT, since the missing parameter file may well exist and simply was not deposited. I also would not ACCEPT, because the strongest claim cannot be checked without that file or an equivalent rerun with a published Mo/S ReaxFF parameterization. The honest recommendation is to keep the paper conditional pending the parameter disclosure or the proposed rerun.","tokens_in":8831,"tokens_out":3669,"duration_ms":50099,"concrete_test":"Run the suspended-monolayer 3x3 and ~10 nm NVT/NPT simulations with a published, explicitly Mo/S-fitted ReaxFF parameter set (e.g., Ostadhossein et al., J. Phys. Chem. C 2017) using the same thermostat, barostat, and 100 ps protocol, and compare the time to first S hop, the number of migrated S atoms after 100 ps, and whether a defect-area threshold near 4 nm2 still appears. If migration does not occur or the threshold shifts substantially, the central claim is force-field-dependent and cannot stand without the original parameter file.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The abstract's strongest claim is that after complete top-layer S removal, bottom-layer S atoms spontaneously migrate through the Mo plane to form a stable MoSx alloy. For the suspended monolayer, the large 10-nm and 100-nm systems, the NVT/NPT equilibration behavior, the substrate-size threshold of about 4 nm2, and the elevated-temperature 2 ns runs are all produced exclusively by classical ReaxFF MD. The Computational Details cite Ref. 39, Chenoweth et al., which is a ReaxFF parameterization for hydrocarbon oxidation, not for Mo-S interactions, and no parameter file is provided anywhere in the paper or supplementary material. Without the actual Mo/S ReaxFF parameters, the key quantity governing the proposed mechanism, the barrier for an S atom to hop through the Mo plane, cannot be checked. If that barrier is misdescribed, the spontaneous migration, the line-dislocation morphology, and the size threshold would not transfer. The AIMD runs give independent support only for a 3x3 cell over a few picoseconds; they do not validate the large-area stability or the defect-size threshold. Thus the load-bearing weakness is not the physics of the small-cell AIMD result but the fact that the large-scale claim rests on an unidentified, non-citable force field. This is a fixable reproducibility gap, not evidence that the mechanism is wrong, but it blocks verification of the strongest claim as written.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper uses classical ReaxFF MD and AIMD to study monolayer MoS2 with one complete sulfur layer removed. The authors report that S atoms from the remaining layer spontaneously migrate through the Mo plane within about 10 ps, producing line-like S dislocations and a locally mixed MoSx structure; they further report that this behavior persists in systems up to 100 nm, that supported monolayers show migration only above a defect area of roughly 4 nm2, and that at 600-1000 K Mo atoms from the depleted layer adsorb onto a pristine MoS2 substrate in AB/AA stacking regions. The central claim is that the S-depleted monolayer is dynamically stable and reorganizes via S migration rather than remaining a static MoS sheet.","tokens_in":9115,"tokens_out":7359,"duration_ms":84572,"significance":"If the central claim is correct, the work provides a concrete atomic mechanism for the response of MoS2 to severe chalcogen depletion, with implications for defect engineering and Janus/alloy synthesis. The paper has real strengths: it combines two independent simulation methods, shows the same S-migration event in a small 3x3 cell in both ReaxFF and AIMD, extends the classical simulations to large (10 nm and 100 nm) systems, includes a substrate geometry with variable defect size, and supplies MD videos in the Supporting Information. The simulations do not fit any parameter to the target result, so circularity is not a concern. However, the large-scale and stability conclusions rest on an unidentified ReaxFF parameter set and on very short single trajectories, so the significance is conditional on those points being fixed.","major_comments":[{"comment":"The classical MD results, including the large 10 nm and 100 nm NVT/NPT runs, the defect-area threshold of about 4 nm2, and the 2 ns elevated-temperature runs, are all produced with ReaxFF, but the Mo-S parameter set is not identified. Reference 39 is Chenoweth et al., a ReaxFF parameterization for hydrocarbon oxidation, not for MoS2, and no parameter file is given in the paper or the Supplementary Material. Because the migration barrier for an S atom to cross the Mo plane is the key physical quantity for the main claim, the reader cannot check whether the force field describes that barrier correctly. This is a reproducibility gap that blocks verification of the abstract's strongest statement; the authors should supply the actual parameter file and cite the correct ReaxFF parametrization for Mo/S, or rerun the central simulations with a published MoS2 ReaxFF set and compare.","section":"Computational Details, ReaxFF paragraph"},{"comment":"The claim that 'consistently, for both methodologies, a single S atom moves first, followed by two others' appears to be based on one classical trajectory and one AIMD trajectory. No replicate runs with different initial velocities, no ensemble averages, and no quantitative order parameter are reported. Given that the migration event occurs on a roughly 10 ps timescale, the sequence of single versus multiple S hops could easily be thermal noise; the paper should report at least several independent trajectories and show that the migration sequence and the final distribution of S atoms on L1 are reproducible.","section":"Results, second paragraph and Figure 2"},{"comment":"The 'remarkable stability' of suspended MoS is supported only by about 100 ps of equilibration, and by 20 ps for the 100 nm video, while the observed migration events occur within about 10 ps. A 100 ps trajectory is too short to establish kinetic or thermodynamic stability, and the AIMD validation covers only a 3x3 cell for a few picoseconds. The paper should provide longer trajectories for at least the representative system sizes and report the time evolution of a stability measure, such as the number of S atoms on the depleted side or the potential energy, to demonstrate a plateau rather than a transient.","section":"Results, NVT/NPT equilibration and Figures 1-3"}],"minor_comments":[{"comment":"The AIMD simulations do not state the temperature, the thermostat, the total simulation time, or the number of MD steps; the caption of Figure 2 only says 'a few picoseconds.' These details are needed to judge the cross-method comparison.","section":"Computational Details, AIMD paragraph"},{"comment":"The defect-area threshold of approximately 4 nm2 is derived from a single simulation at 1 nm2 compared with a single simulation at 4 nm2; at minimum one intermediate size and repeated runs would make the threshold statement quantitative.","section":"Results, substrate simulations and Figure 4"},{"comment":"It is not stated whether the classical NPT simulations used semi-isotropic or fully anisotropic pressure coupling, nor how the z-dimension, which must contain a free surface, is treated; this matters for the reported structural contraction and corrugation heights.","section":"Results, NPT ensemble paragraph"},{"comment":"The term 'line dislocations' is used loosely for the one-dimensional arrays of S atoms that migrate through the Mo layer; these are not crystallographic dislocations in the usual sense, so a more neutral term such as 'S-rich line defects' or 'domain boundaries' would be clearer.","section":"Introduction and Results, line dislocation terminology"},{"comment":"The y-axis label of Figure 5(a) is described only as 'number of Mo atoms from MoS added to the MoS2 layer'; the axis unit/range and the exact simulation time points should be printed on the figure or given in the caption.","section":"Figure 5"}],"recommendation":"major_revision","confidential_remarks":"The central mechanism is plausible and the small-cell AIMD/ReaxFF comparison is a useful internal consistency check, but the missing ReaxFF parameter identification is a serious reproducibility issue that must be fixed before the large-scale claims can be verified. I do not see circularity or parameter fitting to the target result; the concern is transferability of the force field. If the authors can supply a valid MoS2 ReaxFF parameter set and add statistical repeats for the small-cell and threshold claims, the paper would be suitable for publication. The manuscript is within the journal's scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my take. The paper reports something I hadn't seen: strip the entire top sulfur layer off monolayer MoS2 and, on MD timescales, sulfur atoms from the bottom layer spontaneously hop through the Mo plane, leaving line dislocations and a MoSx-like arrangement. They get this in both ReaxFF and small-cell AIMD, so I consider the qualitative event real, not a force-field artifact. The substrate-size threshold (around 4 nm^2 for migration to occur) and the high-temperature Mo adsorption into AB/AA stacking on the underlying MoS2 are also useful observations for anyone thinking about defect engineering or Janus TMD synthesis.\n\nThe paper is clearly enough written and the videos are a nice complement. The coordination analysis gives some structural context for what the migration does to the lattice.\n\nNow the soft spots. The main one is reproducibility. For the large suspended monolayers (10 nm and 100 nm) and for the substrate threshold, everything rests on ReaxFF, but the paper never says which Mo-S ReaxFF parameter set was used. The cited Ref. 39 is the Chenoweth hydrocarbon oxidation force field, not a MoS2 parameterization. That's not a minor omission: the barrier for an S atom to pass through the Mo plane is the load-bearing quantity, and if that barrier is wrong, the threshold and the morphology would change. I'm not saying the mechanism is wrong — the AIMD sees the same hop in a 3x3 cell — but the large-area and threshold statements are not checkable as written. This is fixable: a parameter file or a correct reference, plus maybe a barrier calculation with a credible force field or DFT-NEB.\n\nSecond, the stability language outruns the simulation times. \"Equilibrium\" and \"stable\" after 100 ps NVT, a few ps AIMD, and 2 ns for the high-T runs. No replicate trajectories, no error bars. That's okay for a qualitative observation, but the abstract's strong claim should be softened.\n\nThird, small stuff: the \"MoS\" naming is slightly awkward but understandable. The substrate restraint force constant is given. Nothing else seems inflated.\n\nBottom line: the paper deserves a serious referee. It should not be desk-rejected. My recommendation would be major revision: identify the ReaxFF parameters, tone down the stability claims, and ideally add at least a couple of independent trajectories for the small systems. This is a useful contribution to the defect-engineering literature, not a field-opener.","headline":"A real observation — S migration through the Mo plane after full top-layer removal — but the large-scale claims rest on an unidentified ReaxFF parameter set; fixable, but not reproducible as written.","tokens_in":9644,"tokens_out":2712,"would_cite":true,"duration_ms":30507,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"After complete removal of its top sulfur layer, monolayer MoS2 remains stable because bottom sulfur atoms migrate through the molybdenum plane to the depleted side, forming a MoSx alloy.","keywords":["MoS2","sulfur vacancies","molecular dynamics","ab initio molecular dynamics","transition metal dichalcogenides","structural stability","defect engineering","MoSx alloy"],"falsifier":"A density-functional-theory calculation of the energy barrier for a single sulfur atom to hop through the molybdenum plane in a sulfur-depleted MoS2 monolayer would settle the mechanism: if the barrier is more than a few electronvolts, room-temperature migration within tens of picoseconds cannot occur, and the observed classical result would be a force-field artifact.","tokens_in":1460,"feed_emoji":"⚛️","tokens_out":1496,"duration_ms":92760,"temperature":0.7,"pith_summary":"The paper asks what happens to a monolayer of MoS2 when every sulfur atom on one side is removed. Using classical reactive molecular dynamics and ab initio molecular dynamics, it finds that the remaining sheet does not fall apart: bottom-layer sulfur atoms spontaneously cross the molybdenum layer within tens of picoseconds, partially refilling the empty top side and producing a MoSx alloy with line-like dislocation boundaries. The same migration occurs in large suspended sheets and on a MoS2 substrate once the depleted patch is larger than about 4 nm2; smaller defects locally contract and block migration. The result matters for controlled defect engineering because it predicts that deliberately desulfurized TMD monolayers can reorganize into alloys by internal sulfur transport, without foreign species.","feed_headline":"Top-sulfur loss in MoS2 triggers S migration, not collapse","feed_subtitle":"Classical and ab initio simulations show the sheet survives as a MoSx alloy, with a size threshold for repair.","key_machinery":"The central object is the fully top-sulfur-depleted monolayer, called MoS in the paper: a single molybdenum plane with only one sulfur plane remaining. The mechanism carrying the argument is sulfur trans-layer migration: in regions where top sulfur has been removed, Mo atoms pull closer together, shortening Mo-Mo distances and broadening the Mo-Mo coordination; sulfur atoms at those sites then move across to the empty top side and form boundary lines, until enough sulfur has crossed to stabilize the sheet. The evidence combines classical reactive molecular dynamics at 300 K with ab initio molecular dynamics on a small cell, and both methods show the same migration sequence: one sulfur atom moves first, followed by two others, after which the system reaches equilibrium.","core_discovery":"The paper shows that a defect-engineered MoS2 monolayer with one complete sulfur layer removed, stoichiometrically 'MoS,' is structurally stable. In both classical reactive MD and ab initio MD, sulfur atoms from the remaining bottom layer start crossing through the Mo plane within a few picoseconds and continue until a partial restoration of the top layer is reached, creating a MoSx alloy with boundary lines of sulfur atoms rather than a uniform sheet. The migration is driven by structural stability: molybdenum atoms in sulfur-depleted regions contract toward each other, which opens pathways for sulfur to cross, and the system equilibrates with corrugation and shortened Mo-Mo distances. On a pristine MoS2 substrate, the repair only happens when the defect area is at least about 4 nm2; below that, local indentation prevents sulfur from moving. At elevated temperatures, molybdenum atoms from the depleted layer migrate onto the substrate and bond in AB or AA stacked islands, so heat transfers metal from the MoS layer to the support.","pith_inferences":["The paper leaves implicit that the same self-repair mechanism could be used to pattern MoSx regions laterally by varying the size of the desulfurized patch, without needing a second chalcogen source.","The ~4 nm2 threshold is a testable prediction: damage spots below this size should heal by local contraction, while larger spots should show sulfur crossover between layers.","If the migration is real, high-temperature operation of defective MoS2 devices could slowly transfer molybdenum and sulfur between adjacent layers or substrates, changing stacking order and local electronic properties over time.","A direct computational check is to repeat the 300 K classical simulation with a different published Mo-S interaction model to see whether the sulfur-crossing mechanism is parameterization-independent."],"forward_implications":["A fully top-layer-sulfur-depleted MoS2 monolayer remains intact over nanosecond timescales, with sulfur redistribution rather than structural collapse.","The equilibrated sheet is a MoSx alloy with local composition variations and line-dislocation boundaries, not a uniform MoS stoichiometry.","On a MoS2 substrate, sulfur migration is suppressed for defect areas below about 4 nm2 but proceeds for larger patches, giving a size threshold for controlled repair.","At elevated temperatures, molybdenum atoms from the depleted layer detach and adsorb onto the substrate in AB or AA stacked islands, so heating degrades the MoS layer by metal transfer.","In constant-pressure simulations, the MoS monolayer contracts in-plane and develops pronounced corrugation, indicating that compaction is part of the stabilization response."],"supporting_citations":[{"why":"Supplies the classical molecular-dynamics engine used for all large-scale and substrate-supported simulations.","marker":"[38]"},{"why":"Supplies the reactive force-field method whose molybdenum-sulfur parameterization carries the classical migration results.","marker":"[39]"},{"why":"Supplies the plane-wave density-functional-theory package used for the ab initio MD check of sulfur migration.","marker":"[42]"}],"fun_headline_variants":["MoS2 self-heals after total top-sulfur removal","Sulfur atoms migrate to repair depleted MoS2","MoS2 forms stable alloy when top sulfur is stripped","Size threshold governs MoS2 self-repair","Heat alters defect repair in sulfur-depleted MoS2"],"cache_read_input_tokens":11776,"weakest_assumption_plain":"The main result assumes the classical molybdenum-sulfur interaction model used for the large simulations correctly describes sulfur atoms moving through the molybdenum layer, and the paper does not identify which parameter set produced that model, with the ab initio check limited to a tiny cell for a few picoseconds.","fun_headline_variants_meta":{"raw":{"variants":["MoS2 self-heals after total top-sulfur removal","Sulfur atoms migrate to repair depleted MoS2","MoS2 forms stable alloy when top sulfur is stripped","Size threshold governs MoS2 self-repair","Heat alters defect repair in sulfur-depleted MoS2"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000405,"raw_usage":{"total_tokens":2099,"prompt_tokens":927,"completion_tokens":1172,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":543,"completion_tokens_details":{"reasoning_tokens":1093}},"tokens_in":543,"tokens_out":1172,"duration_ms":14136,"temperature":1.0,"reasoning_tokens":1093,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T12:16:24.694955+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A density-functional-theory calculation of the energy barrier for a single sulfur atom to hop through the molybdenum plane in a sulfur-depleted MoS2 monolayer would settle the mechanism: if the barrier is more than a few electronvolts, room-temperature migration within tens of picoseconds cannot occur, and the observed classical result would be a force-field artifact.","supporting_citations":[{"cited_title":"C.; Goddard, W","cited_arxiv_id":null,"evidence_quote":"Supplies the reactive force-field method whose molybdenum-sulfur parameterization carries the classical migration results."},{"cited_title":"B.; Calandra, M.; Car, R.; Cavazzoni, C.; Ceresoli, D.; Cococcioni, M.; others Advanced capabilities for materials modelling with Quantum ESPRESSO","cited_arxiv_id":null,"evidence_quote":"Supplies the plane-wave density-functional-theory package used for the ab initio MD check of sulfur migration."}],"review_version":1}