{"id":"8b94576f-8e32-4593-b4ec-94685384cc01","arxiv_id":"1908.04360","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Hydrogen interstitials on the metal plane of TMDC monolayers act as nearly ideal one-electron donors with no in-gap defect states, and diffusion barriers in selenides and tellurides are high enough to keep them in place.","lead":"Density functional theory predicts that hydrogen atoms settle into the metal plane of six atomically thin semiconductors and donate exactly one free electron per atom, with no defect states in the band gap. This offers a possible route to stable, position-selective n-type doping for two-dimensional electronic devices.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The one-electron-per-H claim rests on an untested PBE level ordering: the H antibonding state sits ~3 eV above the PBE CBM, and no hybrid/GW calculation checks that it remains above the true CBM.","rationale":"The reader's weakest-assumption analysis identifies the same issue, and I agree it is the most load-bearing. The diffusion-barrier concern is real but secondary: the reported Ebarr values are lower bounds and the abstract overstates stability for sulfides, yet the conclusion already restricts 'large barriers' to selenides and tellurides. The interstitial-site finding has independent code-level support from the VASP check, so the structural claim is less fragile. The donor-level ordering, however, is the heart of the novelty claim (one electron per H, no gap states) and is checked only at PBE-D2 level. The paper's own argument in Sec. III.B that the 3 eV margin protects against gap underestimation is reasonable but not a substitute for a hybrid/GW calculation, because the relevant quantity is the alignment of a localized defect resonance with the host CBM, not the gap width alone. The proposed HSE06/G0W0 check on two representative compounds (one sulfide and one telluride) would settle whether the claimed mechanism is an artifact of PBE. If the check passes, the CONDITIONAL verdict should become ACCEPT; until then, CONDITIONAL is appropriate.","tokens_in":13339,"tokens_out":9076,"duration_ms":99580,"concrete_test":"Run HSE06 (or G0W0) calculations for H at the hollow site in MoS2 and MoTe2 using the same 5x5 supercell and the paper's relaxed geometries, and determine whether the H-induced antibonding state remains above the CBM and whether the integrated occupation still places one electron per H in the conduction band. If the antibonding state drops below the CBM in either compound, the one-electron-per-H claim and the no-in-gap-state claim fail for that compound; if it remains above, the central claim is supported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that interstitial H donates one electron to the conduction band with no in-gap state depends on the PBE-D2 level ordering shown in Fig. 5(b) and Sec. III.B: the H-induced bonding state is 7-8 eV below the VBM and the antibonding state is about 3 eV above the CBM. The paper asserts that the known PBE band-gap underestimation cannot affect this conclusion. That assertion controls only the size of the gap, not the absolute position of the defect resonance relative to the host bands. PBE self-interaction and delocalization error can shift a localized antibonding state relative to the CBM by more than the band-gap correction, so a hybrid or GW calculation could in principle place the antibonding state below the CBM. In that case the donated electron would occupy a defect level rather than the conduction band, invalidating the one-electron-per-H claim and creating an in-gap state. The VASP cross-check (footnote 45) verifies only the binding-energy site preference for MoS2 and MoTe2, not the electronic level ordering. The headline doping mechanism is therefore supported by a single exchange-correlation approximation, and the error-magnitude assumption is not independently tested.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper reports first-principles DFT-PBE-D2 calculations of hydrogen in 1H-phase monolayer (and some bulk and bilayer) Mo/W dichalcogenides. It claims that H is most stable at the interstitial hollow site on the transition-metal plane, that diffusion barriers increase from sulfides to selenides to tellurides, and that each interstitial H donates exactly one electron to the conduction band while producing no in-gap defect state. The proposed mechanism is formation of bonding and antibonding states between the H 1s state and nearby metal s/d orbitals, with the antibonding state lying about 3 eV above the CBM, so that one electron transfers to the CBM. The paper also compares H2 adsorption and Te-vacancy-related defects, concluding that interstitial hydrogen is a promising stable n-type dopant for patterned TMDC devices.","tokens_in":13541,"tokens_out":6401,"duration_ms":65302,"significance":"If the one-electron-per-H doping and absence of in-gap states are correct, the result is a design-relevant and falsifiable prediction for n-type doping of TMDCs, with practical implications for ALD-based patterned doping. The paper has several genuine strengths: it carefully counts states and electrons; it estimates BSSE; it cross-checks the site preference with VASP for MoS2 and MoTe2; it uses large 5x5 supercells with a 100-angstrom vacuum; and it provides wavefunction and charge-density analysis of the doping mechanism. The main risks are the functional dependence of the donor-level position and the lack of true diffusion saddle-point calculations.","major_comments":[{"comment":"The central one-electron-per-H conclusion rests on the PBE-D2 result that the H antibonding state lies about 3 eV above the PBE CBM. The paper argues that the well-known band-gap underestimation cannot affect this conclusion, but that argument is incomplete: Kohn-Sham eigenvalue errors from self-interaction and delocalization can shift a localized resonance relative to the host band edges by an amount that is not controlled by the gap error alone. A hybrid-functional (HSE) or G0W0 calculation for at least one representative compound, such as MoTe2 or MoS2, is needed to verify that the antibonding state remains above the true CBM and that no occupied defect state enters the gap. As written, the headline doping mechanism is supported by a single exchange-correlation approximation.","section":"Section III.B, Fig. 5(b)"},{"comment":"The quantity called the diffusion barrier, E_barr, is defined as the binding-energy difference between the hollow site and the M-top or X-top site; it is not a saddle-point barrier. Without nudged-elastic-band or equivalent minimum-energy-path calculations, the values 0.14-0.93 eV are only lower bounds, and for MoS2 and WS2 the reported lower bounds (0.143 and 0.353 eV) are below the room-temperature scale. The claim of substantial barriers and negligible diffusion should be supported by actual barrier calculations for at least the sulfides and one telluride.","section":"Section III.A, Table I"},{"comment":"The 5x5-supercell binding energies are obtained by transplanting atomic coordinates relaxed in a 3x3 supercell without a stated re-relaxation in the 5x5 cell. At hydrogen concentrations of 1/9 versus 1/25, the optimal local relaxation around H is not guaranteed to be the same, and the residual strain can bias binding-energy differences by amounts comparable to the reported MoS2 barrier. The authors should either fully relax the 5x5 supercell or demonstrate that the transplanted geometry and a fully relaxed one give the same energies.","section":"Section II, Table I"}],"minor_comments":[{"comment":"The text uses 'adsorbed hydrogen atoms' for the lowest-energy configuration, but that configuration is an interstitial site on the transition-metal plane; consider using 'interstitial hydrogen' to avoid confusion with surface adatoms.","section":"Abstract and Introduction"},{"comment":"The notation 'h doping' should be replaced by 'hole doping' or 'p-type doping' for consistency with the rest of the text.","section":"Table IV and Fig. 8"},{"comment":"The term 'diffusion barrier height' for E_barr should be replaced by 'site-energy difference' or an explicit lower-bound qualifier until true minimum-energy-path barriers are computed.","section":"Section III.A"},{"comment":"The VASP cross-check is useful but should be expanded in the main text with the functional, supercell size, and the computed site-energy differences so that its scope is clear.","section":"Footnote 45"},{"comment":"The manuscript does not state explicitly whether spin polarization was allowed in the DFT calculations; since a neutral H impurity introduces an unpaired electron if localized, this choice should be stated and checked.","section":"Section II"}],"recommendation":"major_revision","confidential_remarks":"The paper addresses an important practical problem and the electron-counting analysis is careful. The two main risks are (i) the PBE-only verification of the donor-level position and (ii) reported barriers being site-energy differences rather than true barriers; both are fixable by additional calculations. The 5x5/3x3 geometry transplant is a further point to clarify. I would not reject; the revision should include hybrid/G0W0 tests and NEB barriers for representative systems."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Punchline: this is a useful, carefully done DFT study, and the central claim is plausible: an interstitial hydrogen sitting on the metal plane of a TMDC monolayer dopes one full electron per H without introducing an in-gap state. The metal-plane hollow site, the one-electron count, and the bonding/antibonding mechanism are new relative to the earlier literature, which focused on chalcogen-site adsorption and reported much lower doping efficiencies. I would send it to review.\n\nThe paper does several things well. The electron-counting argument is laid out carefully, including a check that the number of states below the VBM is unchanged after adsorption. The wave-function analysis of the bonding and antibonding states is clear, and the DFT-D2 binding energies, BSSE estimate, and the VASP cross-check for MoS2 and MoTe2 give reasonable confidence that the site ordering is right. The comparison with Te vacancies and H2 adsorption is a useful add-on. The diffusion barriers are presented honestly as binding-energy differences, and the paper explicitly notes that they are lower bounds on the true maximum barrier.\n\nThe soft spots are proportionate. The one-electron-per-H result depends on the PBE-D2 level ordering: the antibonding defect state sits about 3 eV above the PBE CBM, and the authors argue that the usual band-gap error cannot bring it down. That argument is plausible but not rigorous—self-interaction error in PBE can shift a localized state relative to the host bands by more than the gap correction, and the VASP check verifies only geometry, not the level ordering. A hybrid or GW calculation on at least one sulfide, one selenide, and one telluride would be the appropriate response in review. Also, the abstract says 'substantial' barriers for all three families, but the sulfide barriers are 0.14–0.35 eV, which is modest; the real stability claim is for selenides and tellurides. That is an overstatement, not a fatal one.\n\nWho should read this? People working on doping, contacts, and ALD-based processing of 2D devices. It gives a concrete, testable prediction: interstitial H on the metal plane acts as a shallow donor, and the diffusion data suggest where position-selective doping is feasible. The paper deserves a serious referee; the main queries should be the exchange-correlation check and a request to soften the blanket barrier claim in the abstract.","headline":"Genuinely new DFT result with a plausible one-electron-per-H doping mechanism; the claim rests on PBE level ordering that deserves a hybrid-functional check before the strongest version is trusted.","tokens_in":14096,"tokens_out":2967,"would_cite":true,"duration_ms":30405,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Interstitial hydrogen atoms dope TMDC monolayers with one electron each and create no in-gap states.","keywords":["hydrogen doping","transition metal dichalcogenides","n-type doping","first-principles DFT","interstitial hydrogen","diffusion barrier","defect states","MoTe2"],"falsifier":"A hybrid-functional or many-body GW calculation of H at the hollow site in a selenide or telluride monolayer that places the H-metal antibonding state below the conduction-band minimum, or any state inside the band gap, would overturn the one-electron-per-H donor picture. In experiment, Hall measurements on H-doped monolayer MoTe$_2$ showing an electron density far below the hydrogen concentration, or a thermally activated carrier density, would contradict the claim.","tokens_in":13068,"feed_emoji":"⚛️","tokens_out":7378,"duration_ms":75468,"temperature":0.7,"pith_summary":"This paper argues that a hydrogen atom placed at a specific interstitial position in a monolayer transition-metal dichalcogenide is a near-ideal n-type dopant: it sits stably on the metal plane, donates one electron per atom to the conduction band, and creates no electronic state inside the band gap. The result matters because patterned, stable doping is the missing ingredient for building monolayer TMDC devices, which otherwise tend to be uniformly one carrier type. The paper also shows that the dopant is effectively immobile at room temperature in selenides and tellurides, so a written doping pattern would survive. A simple orbital picture explains the extreme efficiency.","feed_headline":"A single interstitial hydrogen atom dopes TMDCs with one electron","feed_subtitle":"Selenides and tellurides keep the hydrogen pinned, and no defect states appear in the band gap, preserving mobility.","key_machinery":"The load-bearing object is the hexagonal hollow site on the transition-metal plane, where interstitial H binds to three metal atoms. Its H 1s orbital hybridizes with metal 5s and 4d orbitals to produce a filled bonding state roughly 7–8 eV below the valence-band maximum and an empty antibonding state roughly 3 eV above the conduction-band minimum; because the antibonding state sits above the conduction-band minimum, the third electron (two from host valence states plus one from H) falls into the conduction band instead. Diffusion barriers are estimated from the binding-energy difference between the hollow site and the saddle-point sites (M-top or X-top), which is a lower bound on the maximum barrier along the path.","core_discovery":"Hydrogen atoms adsorbed into 1H-phase monolayer TMDCs ($MX_2$ with $M = \\mathrm{Mo}, \\mathrm{W}$ and $X = \\mathrm{S}, \\mathrm{Se}, \\mathrm{Te}$) are most stable at the center of the hexagonal hollow site lying directly on the transition-metal plane, where each H atom donates exactly one electron to the conduction bands and creates no defect state inside the band gap. The diffusion barrier for leaving that site grows from sulfides to selenides to tellurides, reaching 0.67–0.93 eV in the selenides and tellurides, so the dopants are effectively immobile at room temperature. The doping mechanism is the formation of a bonding state far below the valence-band maximum and an antibonding state about 3 eV above the conduction-band minimum from the H 1s orbital and metal orbitals; the antibonding state remains empty because it lies above the conduction band, and the electron that would occupy it instead goes into the conduction band. For comparison, H$_2$ molecules dope p-type, and H atoms at Te vacancies compensate the p-doping of the vacancy.","pith_inferences":["The same orbital-counting argument implies a ceiling on useful hydrogen dose: once neighboring H atoms pair into H$_2$, the doping switches from n-type to p-type, so the maximum n-type carrier density is set by the H-pairing distance rather than by the adsorption site.","Because the global minimum is reached only after passing a shallow local minimum a few ångströms off the metal plane, low-temperature deposition may leave hydrogen at the wrong site; an annealing step or elevated substrate temperature should be needed to realize the one-electron doping in practice.","The donor mechanism should transfer to other two-dimensional semiconductors whose conduction band has strong metal-d character: an interstitial s-orbital impurity will dope efficiently when its antibonding partner lies above the conduction-band minimum. Testing hydrogen in other $MX_2$ compounds would map the scope of this design rule."],"forward_implications":["Hydrogen atoms can locally convert p-type MoTe$_2$ to n-type, allowing p-n junctions or more complex circuits within a single TMDC nanosheet.","In selenide and telluride monolayers, barriers of 0.67–0.93 eV mean patterned hydrogen doping will not blur by diffusion at room temperature.","Because no defect state lies in the gap and the donor state is above the conduction-band minimum, conduction electrons see little added scattering and the carrier density equals the hydrogen density, independent of temperature.","Hydrogen molecules, which form when two H atoms meet, dope p-type instead, so preserving the n-type effect requires keeping hydrogen atoms isolated from one another.","Hydrogen atoms at tellurium vacancies neutralize the vacancy's hole doping, enabling net p-to-n conversion in realistic samples that contain such vacancies."],"supporting_citations":[{"why":"Supplies the exchange-correlation functional used for all structural and electronic-structure calculations.","marker":"[35]"},{"why":"Supplies the norm-conserving pseudopotential scheme that defines the ionic potentials.","marker":"[36]"},{"why":"Supplies the localized-orbital DFT implementation in which the main calculations are performed.","marker":"[37]"},{"why":"Supplies the plane-wave DFT implementation used as an independent numerical check of the main code.","marker":"[43, 44]"},{"why":"Documents the cross-check that confirms the hollow site on the metal plane is the lowest-energy configuration for monolayer MoS2 and MoTe2.","marker":"[45]"},{"why":"Previous work on H adsorption sites and Te-vacancy doping that this paper compares against for the lowest-energy site and p-type behavior.","marker":"[15]"},{"why":"Previous prediction of about 4% doping efficiency for H adsorbed to sulfur atoms, the baseline the one-electron-per-H result exceeds.","marker":"[14]"},{"why":"Earlier study that discussed hydrogen doping only superficially, framing the need for a detailed dopant-site and diffusion analysis.","marker":"[17]"},{"why":"Establish that donor levels above the conduction-band minimum occur in bulk InN and ZnO, the classification this paper applies to interstitial H in TMDCs.","marker":"[30, 33]"},{"why":"Experimental report of hydrogen release during atomic-layer deposition, the proposed fabrication route for patterned H doping.","marker":"[28]"}],"fun_headline_variants":["Each H atom donates one electron to TMDCs, no gap states","Hydrogen adatoms: single-electron donors for TMDCs without defects","Stable H interstitials give TMDCs one electron, zero gap states","H doping in TMDCs: one electron per atom, no defect states"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the generalized-gradient-approximation functional places the hydrogen-metal bonding state about 7–8 eV below the valence-band maximum and the antibonding state about 3 eV above the conduction-band minimum, so its well-known band-gap underestimation cannot shift either state into the gap or bring the donor level below the conduction-band minimum.","fun_headline_variants_meta":{"raw":{"variants":["Each H atom donates one electron to TMDCs, no gap states","Hydrogen adatoms: single-electron donors for TMDCs without defects","Stable H interstitials give TMDCs one electron, zero gap states","H doping in TMDCs: one electron per atom, no defect states"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001201,"raw_usage":{"total_tokens":4990,"prompt_tokens":1025,"completion_tokens":3965,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":641,"completion_tokens_details":{"reasoning_tokens":3881}},"tokens_in":641,"tokens_out":3965,"duration_ms":24398,"temperature":1.0,"reasoning_tokens":3881,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:45:13.485411+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A hybrid-functional or many-body GW calculation of H at the hollow site in a selenide or telluride monolayer that places the H-metal antibonding state below the conduction-band minimum, or any state inside the band gap, would overturn the one-electron-per-H donor picture. In experiment, Hall measurements on H-doped monolayer MoTe$_2$ showing an electron density far below the hydrogen concentration, or a thermally activated carrier density, would contradict the claim.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the localized-orbital DFT implementation in which the main calculations are performed."},{"cited_title":"Our VASP results also show the hy- drogen atom adsorbed at the hollow site on the metal plane is the lowest energy configuration, confirming our SIESTA results","cited_arxiv_id":null,"evidence_quote":"Documents the cross-check that confirms the hollow site on the metal plane is the lowest-energy configuration for monolayer MoS2 and MoTe2."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Previous work on H adsorption sites and Te-vacancy doping that this paper compares against for the lowest-energy site and p-type behavior."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Previous prediction of about 4% doping efficiency for H adsorbed to sulfur atoms, the baseline the one-electron-per-H result exceeds."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Earlier study that discussed hydrogen doping only superficially, framing the need for a detailed dopant-site and diffusion analysis."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Experimental report of hydrogen release during atomic-layer deposition, the proposed fabrication route for patterned H doping."}],"review_version":1}