{"id":"864b6aaf-67fa-4858-9de5-543e7766f1e9","arxiv_id":"1908.07357","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Fluorinating exfoliated InSe produces stable, selenium-doped InF3 films as thin as three layers, with a direct optical bandgap of about 2.2 eV.","lead":"The authors chemically converted thin flakes of layered indium selenide (InSe) into atomically thin indium fluoride (InF3), a covalent crystal that cannot be made by exfoliation. This opens a route to two-dimensional covalent materials that have no layered bulk version.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Thinnest (1.5 nm) converted flakes lack direct phase identification; all diffraction/Raman evidence is on bulk or ~10 nm flakes, so the three-layer InF3 claim rests on AFM height alone.","rationale":"The reader's weakest-assumption analysis identifies the correct load-bearing point. The bulk conversion is credible and independently supported by XRD, SAED, HRTEM, XPS, and Raman. The thin-flake claim, however, is central to the novelty and is supported only by AFM thickness plus an indirect Raman analogy from a thicker flake. This is not an internal contradiction, but it is the least secure condition that the central claim requires: a false positive at the three-layer limit would reduce the result to a bulk conversion demonstration with a thin remnant of unknown phase. The requested test is concrete and feasible with modern registered-TEM and Raman methods, and it would discriminate between crystalline Se-doped InF3 and partially fluorinated or amorphous residues. Because the paper's bulk evidence is strong and the missing thin-flake phase identification is an addressable experimental gap rather than a demonstrated error, the appropriate verdict remains CONDITIONAL rather than ACCEPT or REJECT. The reader's verdict therefore does not change.","tokens_in":12069,"tokens_out":3522,"duration_ms":37997,"concrete_test":"Register the same individual 1.5 nm flakes used for AFM, transfer one to a TEM grid with coordinate markers, and acquire SAED/FFT plus, if possible, atomic-resolution HAADF-STEM to index R-3c InF3 planes; on the identical flake, collect a Raman spectrum (or tip-enhanced Raman) to search for In-Se modes at ~179 and ~227 cm-1. A complementary check is to prepare a cross-sectional lamella of a fluorinated trilayer InSe flake and map In, F, and Se by STEM-EDS/EELS while imaging the lattice. If the 1.5 nm object shows no InF3 diffraction or retains InSe Raman modes, the few-layer conversion claim fails; if InF3 planes and an F:In ratio near 3:1 are observed, the concern is settled in the authors' favor.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing gap sits at the thickness boundary of the central claim. The paper states conversion of InSe 'down to three layers' (2.4 nm InSe to a 1.5 nm product) and Figure S1 shows AFM of such thin flakes, but every phase-sensitive characterization of exfoliated flakes—Raman (Fig. 1g), HRTEM/FFT (Fig. 1h,i), SAED (Fig. S3), XRD (Fig. S4), XPS (Fig. 1j-l)—is performed either on bulk fluorinated crystals or on a ~10 nm flake. No diffraction, atomic-resolution imaging, or vibrational spectrum is shown for the 1.5 nm product itself. AFM height cannot distinguish crystalline Se-doped InF3 from a partially fluorinated or amorphous InSeF_x residue, or from an etched remnant; this is especially concerning because mono- and bilayer InSe flakes disappear during fluorination. The DFT calculation predicting a stable trilayer-derived InF3 slab of ~1.4 nm is consistent with the measured height but is not a measurement of composition or crystallinity. If the 1.5 nm flakes are not actually crystalline InF3, the headline claim of atomically thin covalent InF3 loses its experimental support precisely at the few-layer limit. Bulk conversion remains well supported by XRD, SAED, HRTEM, and XPS; the issue is confined to the thickness boundary that makes the paper novel.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports a chemical conversion route from exfoliated and bulk InSe to Se-doped InF3 by direct fluorination with XeF2. The authors claim that the conversion works for InSe flakes as thin as three layers (2.4 nm InSe converting to ~1.5 nm product), that the resulting covalent InF3 is doped with selenium, and that it is a direct-gap semiconductor with Eg ≈ 2.2 eV, high transparency above ~600 nm, and thermal stability. Supporting evidence includes Raman, HRTEM/FFT, SAED, XRD, XPS, electrical resistivity, optical transmission and Tauc analysis, and DFT calculations including phonon stability and Raman spectra. The paper also demonstrates scalability by fluorinating liquid-exfoliated InSe laminates into free-standing films.","tokens_in":12363,"tokens_out":4356,"duration_ms":45630,"significance":"If substantiated, this work would demonstrate a general and scalable route to atomically thin covalent solids that have no layered bulk analogues, going beyond exfoliation of van der Waals crystals and high-vacuum surface growth. The bulk conversion of InSe to InF3 is well supported by multiple phase-sensitive techniques: XRD, SAED, HRTEM/FFT, XPS, and the disappearance of InSe Raman modes. The DFT calculations add mechanistic insight, including the prediction that three layers are the minimum stable thickness and a calculated Raman spectrum consistent with the core modes. The main significance of the paper, however, lies in the few-layer limit, and at that thickness the phase identification is the weakest point of the experimental evidence.","major_comments":[{"comment":"The central claim of converting InSe down to three layers rests on AFM thickness alone for the thinnest product. The 1.5 nm flakes shown in Figure S1 are not characterized by Raman, electron diffraction, or XPS; the Raman spectrum in Figure 1g is from a ~10 nm flake, and the HRTEM/FFT, SAED, and XRD data are obtained from bulk or polycrystalline material (Methods: TEM samples were prepared by rubbing bulk InF3 crystals against a grid). AFM height cannot exclude a partially fluorinated InSeF_x residue, an amorphous product, or an etched remnant, especially because mono- and bilayer InSe flakes disappear upon fluorination. Direct micro-diffraction, nanoscale Raman, or composition mapping of the ~1.5 nm product is needed to support the headline few-layer claim.","section":"§3, first paragraph and Figure S1"},{"comment":"The direct-gap assignment is based on selecting n = 1/2 as the Tauc exponent that gives the 'best linear regime'; no quantitative criterion, uncertainty, or comparison with n = 2 (indirect) or n = 3/2 (direct forbidden) is shown. Since the 'direct bandgap of 2.2 eV' is a central result of the paper, please provide the alternative Tauc fits or a more objective linearity metric to support the choice.","section":"§4, Eq. (2) and Figure 2b inset"},{"comment":"The DFT+U calculations use Ueff = 7 eV on In d orbitals without justification or sensitivity analysis, and the Se doping concentration is set to 2.1% based on XPS. The computed bandgap and Raman spectra are then compared with experiment, but with U as a free parameter the agreement is not a parameter-free validation. Please justify the choice of Ueff or show that the main conclusions are robust to variations in U.","section":"Methods, DFT-methodology"}],"minor_comments":[{"comment":"The caption states that panel b is 'fully fluorinated bulk InSe', while the text discusses fully fluorinated bilayer InSe and concludes that three-layer InSe is the minimum stable thickness; the caption and text should be aligned.","section":"Figure S7 caption"},{"comment":"There is a stray '3.' in the sentence '... have a direct bandgap with Eg ≈ 2.2 eV 3. This value is significantly larger...' that should be removed.","section":"Page 6, after Eg ≈ 2.2 eV"},{"comment":"The phrase 'thicker layers of InSe preserved their anisotropic structure after fluorination' is ambiguous; it should clarify whether this refers to flake morphology or to the crystal symmetry of the product.","section":"Page 3, first paragraph"},{"comment":"The main text reports 'approximately 2%' selenium doping, while Figure S5 presents 6% and 9% doped samples; the relationship between the XPS quantification and the doping levels discussed in the main text should be stated more explicitly.","section":"Methods, XPS and Figure S5"},{"comment":"The main text refers to 'supplementary text 1', but the supplementary text is not numbered; please add a label so the cross-reference is unambiguous.","section":"Supplementary text 1"}],"recommendation":"major_revision","confidential_remarks":"The bulk conversion evidence is convincing and the concept is appealing. The main risk is the few-layer boundary: the 1.5 nm product lacks direct phase-sensitive characterization. I would advise the editor that the paper should either provide such evidence (e.g., Raman or diffraction from the thinnest flakes) or temper the headline claim to 'few-nanometre-thick' covalent films. The DFT U parameter also deserves justification before the DFT/experiment agreement is used as a validation."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Worth a serious look, but the strongest claim — atomically thin covalent InF3 down to three layers — is not backed by direct evidence at that thickness. The bulk conversion is well supported: XRD, SAED, HRTEM, XPS, and Raman all point to Se-doped InF3 after fluorination of bulk InSe. That part is convincing, and the combination with DFT (stability threshold at three layers, Raman spectrum) is a nice piece of work. The scalability demonstration with liquid-exfoliated laminates is a useful addition.\n\nThe soft spots are real but not fatal. The 1.5 nm flakes are identified only by AFM height; the Raman shown for a 'fluorinated InSe flake' is on a ~10 nm flake, not the thinnest product. So the central few-layer claim lacks phase-specific characterization. The stress-test note is right on this point. The paper should either provide diffraction or spectroscopy on the thinnest flakes, or soften the claim to 'covalent InF3 films at thicknesses down to ~1.5 nm by AFM, with structural evidence on thicker flakes and bulk.' The Tauc analysis chooses the direct-gap exponent by best linear fit, which is a common but somewhat circular practice. The DFT+U value (7 eV) is a free parameter, though the Raman match gives it some support. The novelty framing overstates the gap relative to ref 17, which already demonstrated transformation of a layered material into non-layered 2D Cu2SnS3 by cation exchange; the paper should acknowledge that more directly.\n\nThe citation pattern looks fine. No raw data or code are provided, which limits reproducibility but is typical for this kind of experimental paper.\n\nOverall: the concept is interesting and the bulk evidence is solid; the few-layer claim needs either more evidence or more careful wording. It deserves a serious referee and likely major revision.","headline":"A credible bulk conversion of InSe to Se-doped InF3 with solid structural evidence, but the few-layer claim rests on AFM alone and the novelty is overstated relative to prior ion-exchange work.","tokens_in":13006,"tokens_out":2278,"would_cite":true,"duration_ms":23206,"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":"Fluorinating few-layer InSe converts it into atomically thin covalent crystals of InF3.","keywords":["two-dimensional materials","covalent crystals","indium selenide","indium fluoride","fluorination","chemical conversion","van der Waals crystals","direct bandgap semiconductor"],"falsifier":"Take a fluorinated flake that started as three layers (final AFM thickness 1.5 nm) and obtain a plan-view HRTEM image with a fast Fourier transform: if the pattern does not index to the R-3c InF3 structure, or if Raman shows In-Se modes alongside the 250 cm-1 feature, then the thinnest product is not the claimed crystalline Se-doped InF3.","tokens_in":11883,"feed_emoji":"🧪","tokens_out":10116,"duration_ms":88977,"temperature":0.7,"pith_summary":"Exfoliated indium selenide (InSe) flakes, as thin as three atomic layers, can be chemically converted by fluorination into indium trifluoride (InF3), a non-layered covalent solid that cannot be made by exfoliation. The paper argues this conversion is complete and preserves the flake's 2D shape, leaving a stable Se-doped InF3 film roughly 1.5 nm thick. Optical and electrical measurements show the converted films are direct-bandgap semiconductors with $E_g \\approx 2.2$ eV, and DFT calculations explain why at least three starting layers are required. The broader claim is a general route: any cleavable van der Waals crystal whose cation reacts with a gas could become an atomically thin covalent crystal.","feed_headline":"Fluorination turns layered InSe into 2D covalent crystals","feed_subtitle":"A XeF2 reaction converts cleavable InSe into a stable, transparent 2.2 eV semiconductor that cannot be exfoliated.","key_machinery":"The load-bearing object is the chemical reaction of exfoliated InSe with gaseous XeF2, which replaces Se by F and rearranges the remaining atoms into the non-layered R-3c InF3 lattice while the overall flake outline is retained. The argument is carried by an identity chain: bulk fluorinated InSe matches InF3 in XRD, selected-area diffraction, HRTEM lattice spacings, and XPS; the same Raman signature appears in flakes as thin as 1.5 nm; and DFT says fully fluorinated mono- and bilayer InSe are dynamically unstable, fixing the observed three-layer threshold. That calculation changes the reaction from a one-off observation into a mechanism with a predicted thickness floor.","core_discovery":"The central claim is that direct fluorination of InSe with XeF2 converts the layered crystal into the non-layered, rhombohedral covalent compound InF3, and that this conversion works for flakes as thin as three layers while preserving the original 2D shape. In bulk samples the product is identified as InF3 by XRD, selected-area electron diffraction, HRTEM lattice spacings, and XPS; a residual selenium content of about 2% accounts for the difference from commercial InF3 in the Raman spectrum. For thin flakes the same Raman signature and a drop in AFM thickness from 2.4 nm to 1.5 nm are used to assign the same product. Tauc analysis of optical absorption gives a direct bandgap of $E_g \\approx 2.2$ eV, and DFT with Se impurities reproduces that gap and predicts that fully fluorinated mono- and bilayer InSe are dynamically unstable, matching the observed three-layer threshold.","pith_inferences":["The authors do not test other anion-replacement pairs, but the same logic suggests thin GaSe or SnS2 exposed to the right halogen or chalcogen source could give non-layered 2D GaF3 or similar covalent products; that is an extrapolation beyond this paper.","A direct atomic-resolution image of a flake converted from exactly three layers is absent; if taken, it would either confirm or disprove the extrapolation of the bulk InF3 assignment to the thinnest product.","Since the optical gap (2.2 eV) exceeds twice the transport activation energy (1.4 eV), the paper's own data imply mid-gap states dominate conduction; a natural follow-up is to vary the XeF2 dose and measure how transport changes with the measured 2-9% Se content."],"forward_implications":["Thin covalent crystals that cannot be exfoliated can be reached from cleavable precursors, so the set of accessible 2D materials is no longer limited to layered bulk compounds.","The conversion works on liquid-exfoliated InSe laminates, producing free-standing Se-doped InF3 films with a 2.2 eV direct gap, so the route is scalable to large areas.","Se-doped InF3 transmits visible and infrared light above 600 nm with a refractive index near 2, making it a candidate for transparent-semiconductor and optical-coating applications.","The three-layer minimum set by DFT means the method preserves a predictable lower size limit: attempts with one- or two-layer InSe will not produce InF3."],"supporting_citations":[{"why":"Provides the direct XeF2 fluorination procedure for 2D crystals and the Raman evaluation used to judge complete conversion.","marker":"6"},{"why":"Supplies the micromechanical exfoliation of InSe and the known thickness-dependent properties of the starting flakes.","marker":"18,19"},{"why":"Defines the InSe Raman modes at 117, 179, and 227 cm-1 whose disappearance signals conversion.","marker":"20"},{"why":"Documents the structural chemistry of indium fluorides, supporting the expectation that InF3 is the reaction product.","marker":"21"},{"why":"Supplies the R-3c InF3 lattice parameters used to index the HRTEM, SAED, and XRD data.","marker":"22"},{"why":"Contains the liquid-exfoliation method for InSe laminates used to demonstrate large-area conversion.","marker":"24,25"},{"why":"The plane-wave DFT code used for structure optimization, phonons, and band-structure calculations.","marker":"26"},{"why":"The exchange-correlation functional used in the DFT; the three-layer stability threshold comes from these calculations.","marker":"27"}],"fun_headline_variants":["Chemical conversion of InSe yields non-exfoliable 2D crystals","InSe to InF3: chemistry makes 2D covalent crystals","Layered InSe becomes covalent 2D after fluorination","Down to 3 layers: InSe converts to 2D covalent crystal","Fluorination transforms exfoliated InSe into 2D covalent solids"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that a 1.5 nm fluorinated flake, identified mainly by Raman similarity to bulk fluorinated InSe and by AFM thickness, is actually crystalline Se-doped InF3 and not a partially fluorinated or amorphous residue.","fun_headline_variants_meta":{"raw":{"variants":["Chemical conversion of InSe yields non-exfoliable 2D crystals","InSe to InF3: chemistry makes 2D covalent crystals","Layered InSe becomes covalent 2D after fluorination","Down to 3 layers: InSe converts to 2D covalent crystal","Fluorination transforms exfoliated InSe into 2D covalent solids"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001721,"raw_usage":{"total_tokens":6848,"prompt_tokens":1028,"completion_tokens":5820,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":644,"completion_tokens_details":{"reasoning_tokens":5724}},"tokens_in":644,"tokens_out":5820,"duration_ms":39079,"temperature":1.0,"reasoning_tokens":5724,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:20:04.551731+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a fluorinated flake that started as three layers (final AFM thickness 1.5 nm) and obtain a plan-view HRTEM image with a fast Fourier transform: if the pattern does not index to the R-3c InF3 structure, or if Raman shows In-Se modes alongside the 250 cm-1 feature, then the thinnest product is not the claimed crystalline Se-doped InF3.","supporting_citations":[],"review_version":1}