{"id":"bc2aeffb-3fa3-4ba1-8ee7-f3c7ce20a019","arxiv_id":"2412.17664","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Using DFT, the authors report that arsenic substituting for molybdenum in monolayer MoSe2 is a favorable defect (formation energy 1.839 eV) that widens the band gap to 1.73 eV and introduces mid-gap defect levels, while interstitial arsenic has a very high formation energy of 17.422 eV.","lead":"This paper uses density functional theory to calculate how arsenic defects change the electronic structure of a single layer of molybdenum diselenide (MoSe2). It reports that replacing a molybdenum atom with arsenic is energetically favorable and widens the band gap, while putting arsenic in an interstitial position is extremely costly and likely unstable.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Odd-electron defect supercells (As(Mo), As-interstitial) are computed with no reported spin polarization, so the half-filled gap state anchoring the Fermi level and the p-/n-type conclusions may be a DFT artifact.","rationale":"The reader's conditional verdict is appropriate, but the most load-bearing weak point is not primarily supercell size. The manuscript reports no spin polarization or occupation smearing for defect cells that contain an odd number of electrons. In standard plane-wave DFT, a non-spin-polarized odd-electron neutral cell necessarily has a half-occupied Kohn-Sham state at the Fermi level, which is exactly what the As(Mo) band structure shows as a defect level at 0 eV. All Fermi-level-based p-/n-type assignments and the defect-level interpretation are therefore suspect. Spin polarization can also lower total energies by roughly the exchange energy of the unpaired electron, so the formation energies used to declare the As(Mo) configuration favorable are not robust. The atom-count inconsistency noted by the reader is real but does not change this conclusion: the PDOS text confirms 15 Mo + 32 Se + 1 As in the As(Mo) cell, i.e., an odd-electron system. I recommend keeping the conditional verdict and requiring the authors to report spin-polarized calculations, or to demonstrate explicitly that spin polarization does not alter their results.","tokens_in":9382,"tokens_out":7712,"duration_ms":75855,"concrete_test":"Recompute the As(Mo) and As-interstitial supercells with spin-polarized PBE using the same SSSP pseudopotentials, 4×4×1 cell, 15 Å vacuum, 2×2×1/8×8×1 k-grids, and 50 Ry cutoff, and compare total energies, formation energies, and spin-resolved DOS/band structures against the paper's values. If the ground state is magnetic (net magnetization ~1 μB per supercell) or the defect level at E_F splits, or E_f(As(Mo)) changes by more than ~0.1 eV, the central claims must be revised; if spin-unpolarized and spin-polarized results coincide, the concern is settled.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim—As(Mo) substitution is favorable (E_f = 1.839 eV, Se-rich) and produces p-type MoSe2 with E_g = 1.73 eV—rests on spin-unpolarized DFT for defect cells with an odd number of valence electrons. In a 4×4×1 stoichiometric MoSe2 supercell (16 Mo + 32 Se) the electron count is even. Replacing one Mo with As removes an even number of electrons and adds five As valence electrons, making the neutral cell odd; inserting interstitial As also gives an odd electron count. The reported As(Mo) band structure has a defect level exactly at 0 eV = Fermi level, the textbook signature of a half-occupied Kohn-Sham state produced by non-spin-polarized treatment. With spin polarization this state should split into occupied and unoccupied levels, shifting both the apparent band gap and the Fermi-level position, and lowering the total energy; hence the reported formation energies (1.839 eV and 17.422 eV) and the Fermi-level p-/n-type assignments are not established unless spin polarization was actually used and left unreported. The paper contains no mention of spin, magnetization, or smearing/occupation scheme, so this cannot be checked from the text. The atom-count typo ('36 Se and 16 Mo') is a separate issue; the later PDOS text (15 Mo, 32 Se, 1 As) confirms a 48-atom cell, so the odd-electron argument is robust.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports PBE-GGA DFT calculations with Quantum Espresso of pristine MoSe2 monolayers and three defect systems: an Mo vacancy, As substitutional on the Mo site (As(Mo)), and an As interstitial. Formation energies are computed in Se-rich conditions via Eq. (3.1), and band structures, DOS, and PDOS are used to identify mid-gap defect levels and to assign p-type (Mo vacancy, As(Mo)) and n-type (As interstitial) behavior, with band gaps between 1.5 eV and 1.73 eV. The central claims are that As(Mo) is a favorable configuration (Ef = 1.839 eV) and that it could be useful for visible-light photocatalysis and photovoltaic applications, while the As interstitial is metastable (Ef = 17.422 eV).","tokens_in":9650,"tokens_out":2909,"duration_ms":28297,"significance":"If the results are correct, the paper would provide a straightforward DFT characterization of As-doped MoSe2, a material combination not widely studied, and would offer a concrete prediction that As(Mo) substitution is energetically accessible under Se-rich growth and induces p-type character with a widened gap. The work has no fitted parameters and the formation-energy trend is internally consistent with the stated method. However, the reliability of the quantitative claims is presently undermined by the absence of spin-polarized treatment of odd-electron defect cells, by an internal inconsistency in the reported supercell composition, and by the absence of convergence checks; these issues are load-bearing for the formation energies and the band-gap/Fermi-level assignments. The paper is reportable but needs substantial revision before its main conclusions can be accepted.","major_comments":[{"comment":"The neutral As(Mo) and As-interstitial supercells contain an odd number of valence electrons (a 4x4x1 stoichiometric MoSe2 cell has an even number; replacing Mo with As removes six and adds five electrons, and interstitial As adds five), yet no spin polarization, magnetization, or smearing/occupation scheme is reported. The band structure in Figure 4(a) shows a defect level exactly at 0 eV (the Fermi level), which is the textbook signature of a half-occupied Kohn-Sham state produced by a restricted, spin-unpolarized calculation. With spin polarization this level should split into occupied and unoccupied states, lowering the total energy and shifting both the apparent band gap and the Fermi level. Consequently, the reported formation energies (1.839 eV and 17.422 eV) and the p-/n-type assignments are not established as reported; the authors must either repeat the calculations spin-polarized or explicitly justify that spin polarization does not change the conclusions.","section":"Section 2"},{"comment":"The supercell composition is stated inconsistently: the computational-details text says a 4x4x1 cell has 48 atoms, 'of which 36 are Se atoms and 16 are Mo atoms' (totaling 52), while the PDOS discussion later says the As(Mo) cell contains 15 Mo, 32 Se, and 1 As (totaling 48). A 4x4x1 pristine MoSe2 supercell is 16 Mo + 32 Se = 48 atoms. This inconsistency affects the defect concentration, the odd-electron counting argument, and the interpretation of the PDOS, and it must be corrected.","section":"Section 2"},{"comment":"No convergence tests are reported for the supercell size, k-point mesh, plane-wave cutoff, or vacuum thickness. The 4x4x1 cell and 15 Å vacuum are simply stated. For defect formation energies and mid-gap levels, finite-size interactions between periodic images can be sizable, and for a 17.422 eV interstitial formation energy the result may be particularly sensitive to the cell size. The manuscript should include at least a demonstration that the formation energy and the defect-level positions are converged with respect to cell size and vacuum thickness.","section":"Section 2"}],"minor_comments":[{"comment":"'transient metal dichalcogenides' should be 'transition metal dichalcogenides'.","section":"Keywords"},{"comment":"The text states that chemical potentials are from 'the bulk bcc phase of Mo and Se'; elemental Se is not bcc under standard conditions, so the reference phase should be specified accurately.","section":"Section 2"},{"comment":"The legend labels an 'As-3d' contribution, but arsenic has no 3d electrons in its valence manifold; this is presumably a typo for As-4p or As-4s.","section":"Figure 4(b)"},{"comment":"The caption lists '(g) and (f) top and side views of MoSe2 with As interstitial'; the duplicate '(f)' should be corrected to '(h)'.","section":"Figure 1 caption"},{"comment":"The defect levels for the As(Mo) system are described as 'a pair of defect levels' and then as 'three different defect levels' (at 0 eV, +0.35 eV, and -0.35 eV); this wording should be made consistent.","section":"Section 4"}],"recommendation":"major_revision","confidential_remarks":"The manuscript fits the journal's scope, and the DFT setup is standard, but the odd-electron spin issue is central: the reported defect levels and Fermi-level positions are likely artifacts of a spin-unpolarized treatment, and the formation energies would change upon spin polarization. The atom-count inconsistency further weakens confidence. These are fixable within the manuscript's scope by redoing the defect calculations with spin polarization and clarifying the cell stoichiometry, so I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read the Bradji/Benkhedir MoSe2 paper. Bottom line: it's a standard PBE supercell study of As impurities in monolayer MoSe2 that gives a new data point — first DFT numbers for As on a Mo site and As interstitial — but the central results are not trustworthy as reported because the defect calculations appear to have been done without spin polarization.\n\nThe genuinely new part is narrow: none of the cited papers compute As in MoSe2, so the formation energies (1.84 eV for As(Mo), 17.4 eV for interstitial) and defect levels are new in a limited sense. The authors do a reasonable job of decomposing the PDOS, and the per-atom analysis in Fig. 5 is a nice touch. The formation-energy trend is internally consistent with their stated chemical potentials, and the manuscript is concise and readable.\n\nThe problem is the spin. A neutral 4x4x1 MoSe2 supercell has 288 valence electrons; replacing one Mo (6 electrons) with As (5) gives 287, and adding interstitial As gives 293. Both are odd-electron cells. The text never mentions spin polarization, magnetization, smearing, or occupation constraints. The band structure of As(Mo) shows a defect level sitting exactly at the Fermi level, which is the textbook signature of a half-occupied Kohn-Sham state — something that should split into occupied and unoccupied states once spins are allowed to be different. That would lower the total energy and shift both the defect levels and the formation energy, and it could change the Fermi-level position that underlies their p-type conclusion. Without spin, the 1.839 eV and 17.422 eV numbers are not established, and the p/n-type labels are speculation. This is not a minor omission; it affects the paper's main claims.\n\nThere are also smaller issues: the text says the supercell has 48 atoms of which 36 are Se and 16 Mo (which is 52); the PDOS text later says 15 Mo and 32 Se with one As, consistent with a 48-atom cell, so the '36' is a typo, but it should be fixed. No convergence tests for cell size or vacuum thickness are reported. The chemical-potential description cites a bcc phase of Se, which doesn't exist; Se's ground state is trigonal. And the Fermi-level-based p/n-type classification is heuristic — proper assignment would need charge-state transition levels.\n\nWho gets value: someone doing a quick literature scan on impurity doping in TMDCs might use this as a starting data point, but the numbers shouldn't be quoted until the spin issue is resolved. A serious referee should see it, because the topic is legitimate and the fix is rerunning the defect calculations with spin polarization and reporting smearing/occupations. I'd send it back for major revision rather than desk-reject.","headline":"Routine but useful DFT data point on As-doped MoSe2; the reported numbers are unreliable because the odd-electron defect cells were apparently run spin-unpolarized.","tokens_in":10165,"tokens_out":4205,"would_cite":false,"duration_ms":38576,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["71.15.Mb","73.22.-f"],"model":"deepseek-v4-flash","headline":"Arsenic substituting for molybdenum in a MoSe2 monolayer is energetically favorable, widens the band gap to 1.73 eV, and shifts the Fermi level toward the valence band, indicating p-type behavior.","keywords":["MoSe2 monolayer","arsenic doping","density functional theory","formation energy","electronic structure","p-type semiconductor","photocatalysis","transition metal dichalcogenides"],"falsifier":"A converged supercell-size series (3×3×1, 4×4×1, and 5×5×1 with the same 15 Å vacuum) recomputing the As(Mo) formation energy and band gap would settle the central claim: if 1.839 eV or 1.73 eV shifts by more than about 0.1 eV between cell sizes, the favourability and the gap-widening are artifacts of the chosen supercell rather than properties of the isolated defect.","tokens_in":9174,"feed_emoji":"🧪","tokens_out":11716,"duration_ms":85826,"temperature":0.7,"pith_summary":"This paper asks what happens to a monolayer of MoSe2, a layered semiconductor, when arsenic is introduced in three ways: as a substitutional atom on a molybdenum site, as a simple molybdenum vacancy, and as an interstitial atom in the hollow center of a ring. Using density functional theory, it finds that the substitutional As(Mo) configuration is the favorable one under selenium-rich conditions, with a formation energy of 1.839 eV, while the interstitial configuration is metastable with a much higher formation energy of 17.422 eV. The band gap of the As(Mo) system rises from 1.41 eV in the pure monolayer to 1.73 eV, and the Fermi level shifts toward the valence band, which the authors read as p-type semiconductor behavior. Three mid-gap defect levels appear and are assigned to Mo-4d, Se-4p, and As-4s orbitals through per-atom projected densities of states. The authors conclude that As(Mo)-doped MoSe2 is a candidate for visible-light photocatalysis and photovoltaics, while the interstitial form, which shifts the Fermi level toward the conduction band, may suit infrared electronics and photocatalysis.","feed_headline":"Arsenic swap widens MoSe2's gap and makes it p-type","feed_subtitle":"DFT predicts a low-cost As-for-Mo swap that widens the gap to 1.73 eV and turns MoSe2 p-type.","key_machinery":"The analysis rests on three density-functional-theory supercell models built from a 4×4×1 repeat of the MoSe2 unit cell with a 15 Å vacuum along the out-of-plane direction: a Mo vacancy, an As substitutional at that vacancy, and an As interstitial in the hollow center of a ring. The formation energy is evaluated with chemical potentials fixed by Se-rich conditions through the formula $E_f = E_{\\rm def} - E_{\\rm pure} - \\sum_i n_i \\mu_i$, and the electronic structure is interpreted through band structures plus total and per-atom projected densities of states, which assign each mid-gap level to Mo-4d, Se-4p, and As-4s/4p orbitals.","core_discovery":"The paper's central claim is that arsenic sitting on a molybdenum site in a MoSe2 monolayer is likely to form under Se-rich conditions and improves the material's electronic profile: the formation energy is 1.839 eV, the band gap widens to 1.73 eV from the pristine 1.41 eV, and the Fermi level moves toward the valence band, making the system a p-type semiconductor. The authors interpret the mid-gap states as a mixture of the vacancy effect and the arsenic atom itself, since the smaller As atom leaves some Mo-4d and Se-4p orbitals non-bonding while adding As-4s character. They further claim that the As interstitial is metastable ($E_f = 17.422$ eV) and n-type, with the Fermi level shifted toward the conduction band and a gap of 1.5 eV.","pith_inferences":["The paper's stated atom count (48 atoms) does not match its breakdown (36 Se + 16 Mo = 52), suggesting the simulated cell may differ from the described one; if so, the nominal 6.25% As-doping concentration and possibly the defect-level positions would need revision.","Because PBE-GGA is known to underestimate band gaps, the absolute values reported here are probably lower bounds; hybrid or GW calculations would test whether the ordering of gaps across the four systems survives.","The p-type assignment for the As(Mo) system could be checked experimentally by Hall-effect or Seebeck measurements on arsenic-doped MoSe2: a positive Hall coefficient would support the Fermi-level placement, while n-type conduction would refute it.","The mid-gap defect levels that widen the gap may also act as non-radiative recombination centers, which could lower carrier lifetimes and reduce the predicted photocatalytic efficiency; time-resolved photoluminescence on As-doped samples would be a direct test."],"forward_implications":["Under Se-rich growth conditions, arsenic substituting for molybdenum in MoSe2 is energetically favorable, with a formation energy of 1.839 eV, making it a realistic doping route.","The As(Mo) system has a band gap of 1.73 eV, larger than the pristine 1.41 eV, and its Fermi level shifts toward the valence band, indicating p-type semiconductor behavior.","The mid-gap defect levels in the As(Mo) system are traced to Mo-4d, Se-4p, and As-4s orbitals, consistent with a combination of the vacancy effect and the arsenic atom.","The As interstitial is metastable with a formation energy of 17.422 eV yet shifts the Fermi level toward the conduction band, suggesting n-type behavior with a 1.5 eV gap.","If these electronic changes hold, the As(Mo) system should absorb in the visible range, making it a candidate for photocatalysis and photovoltaics, while the interstitial system could serve infrared electronics."],"supporting_citations":[{"why":"It provides the reference case of nonmetal-doped MoSe2 with n- and p-type conductivity and photocatalytic applications that the As(Mo) results are compared against.","marker":"[19]"},{"why":"It supplies the formation-energy formula and the treatment of point defects such as single and double vacancies in TMDCs.","marker":"[2]"},{"why":"It provides the formation-energy expression and the Sb(Mo) doping analogue in MoS2 used to interpret the mixed vacancy-plus-As character in the As(Mo) system.","marker":"[7]"},{"why":"It supplies the defect formation-energy methodology with chemical potentials for doped nanosheets used in equation (3.1).","marker":"[24]"},{"why":"It establishes MoSe2's semiconducting properties and supports the proposed photocatalysis and photovoltaic applications.","marker":"[13]"},{"why":"Earlier DFT studies of pristine MoSe2 provide the band structure and orbital character used to validate the pure-monolayer calculation.","marker":"[25–28]"}],"fun_headline_variants":["Arsenic swap in MoSe2 widens gap to 1.73 eV, turns p-type","DFT predicts As-for-Mo doping gives MoSe2 a p-type boost","MoSe2 monolayer: arsenic substitution enlarges gap, flips to p-type","Arsenic on Mo site makes MoSe2 p-type with wider bandgap","Substitutional arsenic in MoSe2 yields wider gap and p-type"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is that the adopted 4×4×1 supercell with a 15 Å vacuum is large enough to isolate each defect, but the paper reports no convergence test against cell size or vacuum thickness and contains inconsistent atom counts (stated 48 atoms versus 36 Se plus 16 Mo), so defect-defect interactions could change the formation energies and mid-gap levels.","fun_headline_variants_meta":{"raw":{"variants":["Arsenic swap in MoSe2 widens gap to 1.73 eV, turns p-type","DFT predicts As-for-Mo doping gives MoSe2 a p-type boost","MoSe2 monolayer: arsenic substitution enlarges gap, flips to p-type","Arsenic on Mo site makes MoSe2 p-type with wider bandgap","Substitutional arsenic in MoSe2 yields wider gap and p-type"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000277,"raw_usage":{"total_tokens":1706,"prompt_tokens":1060,"completion_tokens":646,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":676,"completion_tokens_details":{"reasoning_tokens":538}},"tokens_in":676,"tokens_out":646,"duration_ms":6096,"temperature":1.0,"reasoning_tokens":538,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T05:16:14.155183+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A converged supercell-size series (3×3×1, 4×4×1, and 5×5×1 with the same 15 Å vacuum) recomputing the As(Mo) formation energy and band gap would settle the central claim: if 1.839 eV or 1.73 eV shifts by more than about 0.1 eV between cell sizes, the favourability and the gap-widening are artifacts of the chosen supercell rather than properties of the isolated defect.","supporting_citations":[{"cited_title":"Rep., 2017,7, No","cited_arxiv_id":null,"evidence_quote":"It provides the reference case of nonmetal-doped MoSe2 with n- and p-type conductivity and photocatalytic applications that the As(Mo) results are compared against."},{"cited_title":"K., Eriksson O., Sanyal B., Phys","cited_arxiv_id":null,"evidence_quote":"It supplies the formation-energy formula and the treatment of point defects such as single and double vacancies in TMDCs."},{"cited_title":"R., Ziegen- rücker R., Kentsch U., Krasheninnikov A","cited_arxiv_id":null,"evidence_quote":"It supplies the defect formation-energy methodology with chemical potentials for doped nanosheets used in equation (3.1)."},{"cited_title":"I., Phys","cited_arxiv_id":null,"evidence_quote":"It establishes MoSe2's semiconducting properties and supports the proposed photocatalysis and photovoltaic applications."}],"review_version":1}