{"id":"f985f9cb-4a98-4a24-9cac-26ad1349d694","arxiv_id":"2607.25102","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Combined IV, LBIC, and bias-dependent KPFM on few-layer MoS2 devices distinguishes Schottky from tunnel contact barriers, maps their asymmetry, and shows annealing lowers total resistance while barriers still dominate.","lead":"A correlative lab method (IV + photocurrent maps + biased Kelvin-probe imaging) sorts metal–MoS2 contact barriers into Schottky vs tunnel types and ranks their asymmetry under ambient conditions. It gives device engineers a practical way to diagnose why 2D contacts stay resistive even after annealing.","discovery_kind":"new_method","skeptic_critique":{"model":"moonshotai/kimi-k3","headline":"The Schottky-vs-tunnel typing rule is calibrated on a single diode-like device (D1) with no independent ground truth; the KPFM polarity criterion is a model-dependent fingerprint, not a validated classifier.","rationale":"My concern coincides with the reader's weakest_assumption: the uniqueness of the polarity-dependent-KPFM/gap-centered-LBIC fingerprint for Schottky barriers versus adsorbates, mid-gap states, and other confounds under ambient conditions. I sharpen it to the statistical and validation structure: the classification rule is demonstrated on N=1 diode device with no orthogonal ground truth, while the paper itself names the missing check (temperature-dependent activation analysis). This does not rise to REJECT: the workflow is genuinely multi-modal, the three techniques are mutually consistent on O1 and D1, the Table 1 arithmetic is internally exact, the annealing case study is informative, and the authors hedge the resistance decomposition as order-of-magnitude. The appropriate remedy is exactly what the reader's CONDITIONAL verdict requires — soften \"unambiguous,\" report uncertainties, deposit raw maps — plus, ideally, the activation-energy cross-check before the typing rule is generalized to other 2D semiconductor–metal systems as the Conclusion promises. Hence UNCHANGED with agreement on the load-bearing concern.","tokens_in":19563,"tokens_out":2543,"duration_ms":74713,"concrete_test":"Perform temperature-dependent IV (e.g., 200–350 K) on device D1 and extract the activation behavior at both polarities, as the authors themselves suggest (§3.4). If the dominant (left) contact yields a thermionic-emission-consistent Schottky barrier height at the polarity where KPFM shows the polarity-dependent drop, while the right contact shows temperature-insensitive (tunneling-like) transport, the typing rule is validated on this device. If the left contact's transport is not thermally activated as a Schottky barrier (e.g., fits trap-assisted tunneling through mid-gap states) or the right contact shows activated behavior, the polarity fingerprint is not unique and the \"unambiguous\" claim fails. Ideally repeat on ≥3 additional devices and on controls with engineered known barriers (e.g., h-BN interlayer tunnel contact vs. direct contact).","verdict_should_be":"UNCHANGED","load_bearing_attack":"The strongest claim — that the combined IV/LBIC/bias-KPFM analysis \"unambiguously\" identifies whether the dominant contact barrier is Schottky or tunnel type — rests, in the entire paper, on exactly one typing decision: D1's left contact is declared Schottky (polarity-dependent KPFM drop + gap-centered photovoltaic LBIC lobe) and its right contact tunnel (polarity-independent drop, no photovoltaic signal). D2's post-anneal interpretation reuses the same rule but adds no independent validation. Three conditions must hold for the claim: (i) polarity-dependent voltage drop at a contact edge is produced only by a rectifying Schottky junction; (ii) polarity-independent drop implies a vdW-gap tunnel barrier rather than, e.g., a symmetric pinned junction, contact-underlay sheet resistance, or adsorbate/trap-induced series resistance; (iii) ambient sideband-KPFM surface potentials faithfully report the junction electrostatics rather than surface dipoles, water-layer screening, photogating (which the authors themselves invoke, ref 42), or tip convolution. None of these is independently established in the paper. Notably, the authors concede the LBIC leg alone \"cannot unequivocally distinguish\" the barrier types (§3.2), and themselves propose temperature-dependent IV activation analysis as the quantitative route to barrier heights (§3.4) — i.e., the orthogonal ground-truth measurement exists and was not performed. SI Fig. 14 further shows D2's potential profile is laterally inhomogeneous (a \"triangular area\" with a distortion interpreted as yet another tunnel barrier), while main-text profiles are single-y-position lines asserted to be \"representative of all y-positions,\" so the spatial-uniqueness leg is also thinner than stated. The multi-modal internal consistency is real and the arithmetic in Table 1 checks out (α values sum to 100%; Rprop = α×Rtotal products are exact), but consistency among three model-dependent observables is not the same as uniqueness of the microscopic","agreement_with_reader":"agree"},"referee_report":{"model":"moonshotai/kimi-k3","summary":"The manuscript presents a correlative characterization framework for metal contacts to few-layer MoS2 combining (i) dark and illuminated IV, (ii) laser beam induced current (LBIC) mapping at zero and finite bias, and (iii) sideband KPFM operated under a static applied bias (with the zero-bias CPD subtracted), all under ambient conditions and without a gate. Three devices are studied: a 5L flake with ohmic-like behavior (O1) and two 2L diode-like devices (D1, and D2 before/after vacuum annealing). From mutually consistent IV asymmetry, LBIC spatial patterns (photothermoelectric on-contact vs. photovoltaic in-gap), and polarity-dependent KPFM voltage drops, the authors assign barrier types at each contact (Schottky vs. tunnel), decompose the two-terminal resistance via a three-element series model R_total = R_lc + R_sh + R_rc with fractional drops alpha read from KPFM (Table 1), and conclude that annealing lowers total resistance while contact barriers remain dominant. The abstract and conclusion claim the combined analysis \"unambiguously\" identifies whether the dominant barrier is Schottky or tunnel type.","tokens_in":19928,"tokens_out":3887,"duration_ms":161848,"significance":"If the claims are appropriately calibrated, this is a useful methodological contribution. The combination of bias-dependent KPFM with LBIC and IV on the same device, entirely under ambient conditions and without gating, is a practical, transferable diagnostic that much of the 2D-contacts community could adopt; the annealing case study (D2) demonstrates its relevance to contact optimization. The three measurements are internally consistent across all devices, the KPFM bias-subtraction protocol (forward line biased, backward line at 0 V) is sound, and the series-circuit decomposition is a standard, transparently stated model with order-of-magnitude caveats acknowledged. The paper does not ship ground-truth validation of its barrier-typing rule, however, and its strongest worded claim (\"unambiguously\") exceeds what the data support — the authors themselves concede the qualitative nature of the identification in §3.4 and point to temperature-dependent IV as the quantitative route. Published with the claims aligned to the evidence, this would be a solid, citable methods paper.","major_comments":[{"comment":"The claim that the combined analysis 'unambiguously' identifies the barrier type is not supported at the level stated, and is in tension with the manuscript's own text. §3.2 concedes LBIC alone 'cannot unequivocally distinguish' the barrier types; §3.4 retreats to 'a qualitative identification of the barrier types'; and §3.4 itself proposes temperature-dependent IV thermal-activation analysis (ref. 20) as the route to quantitative barrier heights — i.e., the orthogonal ground truth exists and was not performed. The entire typing claim rests on one polarity-dependence decision per diode device (D1: left = Schottky, right = tunnel), with D2's interpretation reusing the same rule without independent validation. Either the abstract/conclusion must be brought in line with §3.4 (qualitative, consistent-with identification), or a validation must be added — e.g., temperature-dependent IV on D1,","section":"Abstract; §3.4; Conclusion"},{"comment":"The core classifier — abrupt polarity-dependent KPFM drop at a contact edge = Schottky; polarity-independent drop = tunnel (vdW gap) — is a model-dependent fingerprint whose uniqueness is not established. A polarity-independent drop is equally consistent with a symmetric pinned junction, sheet/contact-underlay resistance (the authors themselves note the Cr underlayer extends the effective contact in KPFM), or adsorbate/trap-induced series resistance. Under ambient sideband KPFM with a ~25 nm apex, the 0 V-subtraction removes static work-function contrast but not bias- or illumination-dependent artifacts such as photogating (which the authors invoke, ref. 42) or water-layer screening. As a correctness-risk test rather than a circularity objection: the manuscript should (i) explicitly enumerate the alternative origins of a polarity-independent drop and state which observations exclude them","section":"§3.3 and Table 1 (typing rule)"},{"comment":"The conclusion that annealing reduces a right-contact tunnel barrier and thereby enables a full-width Schottky junction rests on the LBIC signal shift alone; KPFM is shown only after annealing (Fig. 5), so no before/after KPFM comparison supports the claimed redistribution of the voltage drops. Moreover, SI Fig. 14 shows D2's potential profile is laterally inhomogeneous (a triangular region with a shifted drop position, itself interpreted as an additional '(tunnel) barrier'), which complicates the use of single profile lines as representative — the Fig. 3 caption's statement that 'the profile line position is representative of all y-positions' is not demonstrated for D2 and should be qualified. Finally, the KPFM illumination power differs between datasets (5 mW in Fig. 3 vs. 1 mW in Fig. 5), limiting dark/bright comparability across devices.","section":"§4 (D2 annealing)"},{"comment":"The quantitative decomposition has internal inconsistencies. For D1, dark, −0.5 V, the alpha values (72/14/14) sum to 100% in the displayed rounding but the R_prop column (80.00/15.56/15.56) sums to 111.12 MOmega only if alpha values are 72/14/14 — however the +0.5 V dark column gives 38/8/54 (=100%) yet R_prop 12.58/2.65/17.88 sums to 33.11, consistent; please check the −0.5 V bright column (84/8/8 → 1.41/0.13/0.13 sums to 1.67, consistent) and verify all rows, since some rows appear to mix rounded and unrounded alpha. Separately, the 'Barrier Type' column labels O1's left/right contacts 'Schottky / Tunnel' while the text concludes O1 has negligible, ohmic contacts — the label should distinguish 'resistive contact (no rectifying barrier)' from the barrier types, otherwise the table contradicts the paper's own classification of O1.","section":"Table 1"}],"minor_comments":[{"comment":"The SI table of contents contains unresolved Word cross-reference errors ('Fehler! Textmarke nicht definiert.') for sections 2.5, 3.4, 4.4, and 4.5, and SI Fig. 7's caption refers to 'sample A1' instead of D1. These should be corrected before publication.","section":"Supplementary Material, table of contents and §3.2 caption"},{"comment":"The power density for global IV illumination is only an estimate ('assuming a laser spot size of roughly 50 um'), and LBIC, global IV, and KPFM use different photon energies and power densities (2.33 eV/100 uW; 2.33 eV defocused; 1.95 eV/1-5 mW). Since Table 1 combines KPFM-derived alpha with IV-derived R_total, a short table summarizing excitation conditions per technique and per figure would help readers assess the stated 'order-of-magnitude' caveat.","section":"§2 (illumination conditions)"},{"comment":"Device D1 is fabricated on glass while O1 and D2 are on Si/SiO2, and O1/D1 contacts use shadow masks vs. lithography for D2. Since substrate and contact geometry can affect pinning and thermal coupling (relevant to the PTE assignments), a sentence noting whether any systematic differences were observed or controlled for would be useful.","section":"SI §1.1"},{"comment":"The fractional-drop parameter alpha is used in Table 1 before its definition appears in the text of §3.4; define it at first use and state explicitly how the contact-edge drop is delimited (spatial window) when reading alpha from the KPFM profiles, given the ~25 nm tip convolution and the Cr-underlayer feature.","section":"§3.4 (notation)"},{"comment":"The phrase 'ohmic-like Schottky-contact' for D2's right contact after annealing is confusing — a contact is either rectifying (Schottky) or ohmic-like in its IV; please rephrase (presumably: a Schottky junction whose effective resistance after annealing is low enough that the device appears more symmetric).","section":"§4 and Conclusion"},{"comment":"Ref. 37 is a trade-magazine web article; a peer-reviewed source for the contact-resistance bottleneck claim would be preferable. Also check ref. 12 page range ('23209–2322') for a typographical truncation.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The experimental work is competent and the correlative approach is genuinely useful, but the manuscript's headline word 'unambiguously' does real work in the abstract and conclusion that the body (§3.2 caveat, §3.4 'qualitative') disowns. The typing rule is calibrated on a single device with no independent ground truth; the fix can be either added validation (temperature-dependent IV, or a control contact of known type) or an honest reframing of the claims, and I would accept either. I do not see scope or novelty concerns beyond that; the dataset is small (N=3) but appropriate for a methods demonstration."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"Punchline: this is a practical methods paper that puts IV, LBIC, and bias-under-KPFM on the same few-layer MoS2 two-terminals and gets mutually consistent pictures of ohmic-like vs diode-like contacts, including a pre/post anneal case. The real advance is not new physics; it is running KPFM under static bias (voltage-drop maps, not zero-bias work function) together with LBIC spatial signatures on the same devices under ambient conditions.\n\nWhat they do well is the correlative discipline. O1 (5L) is linear IV, PTE-dominated LBIC on the pads, and nearly linear KPFM drop—coherent low-barrier story. D1 (2L) shows asymmetric saturating IV, gap-centered photovoltaic LBIC, and polarity-dependent KPFM drops at one edge vs polarity-independent at the other—internally consistent with asymmetric barriers. Annealing on D2 cuts R_total by roughly an order of magnitude while barriers still dominate the drop; that case study is the most useful part for process people. Table 1 arithmetic is clean (α sums, Rprop = α Rtotal). Writing and figure layout are clear. Citations cover the right single-technique literature (Parzinger, Buscema, Allain, Markeev, etc.).\n\nSoft spots, in proportion: “unambiguously identify Schottky vs tunnel” is too strong. The typing rule is essentially one well-worked diode device (D1) plus reuse on annealed D2. Polarity-dependent edge drop + gap PV lobe is a standard Schottky fingerprint, not a validated classifier against adsorbates, mid-gap states, thickness steps, or ambient tip/surface effects—and the authors themselves say LBIC alone cannot distinguish and point to temperature-dependent IV for quantitative heights, which they did not run. SI lateral inhomogeneity on D2 (triangular distortion) also undercuts the “representative line” claim a bit. N=3, order-of-magnitude contact splits, data-on-request only. None of that sinks the multi-modal consistency; it just means the language should be diagnostic and comparative, not unique microscopic ID.\n\nWho it is for: groups optimizing 2D metal contacts who want an ambient workflow without cryo, UHV, or a gate. Worth a serious referee. I would engage, cite the protocol when discussing contact diagnostics, and ask authors to soften uniqueness claims, flag ambient systematics, and deposit raw maps.","headline":"Useful ambient three-technique contact diagnostic with real multi-modal consistency, but “unambiguous” Schottky-vs-tunnel typing is oversold on model-dependent fingerprints and N=3.","tokens_in":20536,"tokens_out":620,"would_cite":true,"duration_ms":19279,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.5","headline":"Combined IV curves, photocurrent maps, and biased Kelvin-probe scans can tell Schottky from tunnel barriers at few-layer MoS2–metal contacts under ambient conditions.","keywords":["molybdenum disulfide","MoS2","electrical contacts","Schottky barrier","tunnel barrier","Kelvin probe force microscopy","laser beam induced current","2D semiconductors"],"falsifier":"Find a device whose bias-KPFM and zero-bias LBIC patterns match the paper’s Schottky (or tunnel) assignment, yet temperature-dependent transport or a clean high-vacuum barrier-height measurement shows the opposite dominant mechanism; that mismatch would break the fingerprint rule.","tokens_in":20223,"feed_emoji":"⚡","tokens_out":976,"duration_ms":26045,"temperature":0.7,"pith_summary":"Metal contacts to few-layer MoS2 often dominate device resistance, but Schottky barriers, tunnel barriers from the van der Waals gap, and contact asymmetry are hard to separate with any one measurement. This paper offers a practical three-method toolkit—macroscopic current–voltage curves, micrometer-scale laser-beam-induced current maps, and nanoscale Kelvin-probe potential profiles taken under static bias—that works in air, with or without light, and without a gate. On three two-terminal devices (one ohmic-like five-layer flake and two diode-like bilayers), the joint signatures identify which contact carries the larger barrier and whether that barrier is mainly Schottky or tunnel type. The same toolkit shows that mild thermal annealing cuts total resistance sharply yet leaves contact barriers as the main remaining bottleneck. A reader who designs or optimizes 2D contacts cares because the method is transferable and does not require cryogenics or ultra-high vacuum.","feed_headline":"How to tell Schottky from tunnel barriers in MoS2 contacts","feed_subtitle":"Ambient IV, photocurrent maps, and biased Kelvin probes show which contact still limits the device","key_machinery":"Bias-dependent KPFM with zero-bias subtraction: under a static sample bias the tip maps the fractional voltage drop at each contact and across the channel; polarity dependence of that drop, cross-checked against LBIC photovoltaic versus photothermoelectric lobes and against IV asymmetry, assigns Schottky versus tunnel character and relative barrier size.","core_discovery":"When IV transport, LBIC photocurrent maps, and bias-dependent KPFM voltage profiles are read together on the same few-layer MoS2 device, they unambiguously classify the dominant metal–semiconductor barrier as Schottky or tunnel type and quantify the asymmetry between the two contacts, including how annealing redistributes those contributions while contacts still dominate resistance.","pith_inferences":["Because the series-resistor decomposition is only order-of-magnitude under mismatched illumination and bias conditions, pairing the ambient toolkit with one temperature-dependent IV run would turn qualitative barrier typing into quantitative barrier heights without losing the spatial map.","Ambient adsorbates and water layers may systematically inflate the apparent tunnel contribution; repeating the identical three-method sequence in dry inert gas would test how much of the ‘tunnel’ fingerprint is extrinsic.","The method’s ability to watch annealing move a contact from PTE-dominated to photovoltaic-dominated response suggests it could score edge-contact versus top-contact geometries on the same flake in a single fabrication lot."],"forward_implications":["Contact process splits (annealing, metal choice, doping, strain) can be diagnosed barrier-by-barrier on the same ambient two-terminal geometry without a gate.","Five-layer ohmic-like MoS2 contacts show nearly linear potential drops and photothermoelectric photocurrent at the metal edges, giving a concrete target signature for low-barrier process development.","Annealing that lowers total resistance can still leave a dominant Schottky barrier at one contact and only partially remove a tunnel barrier at the other, so resistance gains must be checked spatially.","The same correlative workflow extends directly to other 2D semiconductor–metal systems and to lateral inhomogeneities such as heterojunctions or grain boundaries."],"fun_headline_variants":["IV, LBIC, and biased KPFM classify Schottky vs tunnel MoS2 contacts","Correlative maps pinpoint which MoS2 contact barrier still dominates","Reading IV, photocurrent, and KPFM together flags barrier type and asymmetry","Annealing cuts MoS2 resistance but contacts remain the limit","Three ambient techniques resolve Schottky or tunnel barriers in few-layer MoS2"],"cache_read_input_tokens":128,"weakest_assumption_plain":"The method assumes that an abrupt, polarity-dependent voltage drop plus a gap-centered photovoltaic photocurrent uniquely fingerprints a Schottky barrier, while a polarity-independent drop fingerprints a tunnel barrier, rather than adsorbates, mid-gap states, thickness steps, or tip artifacts under ambient air.","fun_headline_variants_meta":{"raw":{"variants":["IV, LBIC, and biased KPFM classify Schottky vs tunnel MoS2 contacts","Correlative maps pinpoint which MoS2 contact barrier still dominates","Reading IV, photocurrent, and KPFM together flags barrier type and asymmetry","Annealing cuts MoS2 resistance but contacts remain the limit","Three ambient techniques resolve Schottky or tunnel barriers in few-layer MoS2"]},"model":"grok-4.5","effort":"low","cost_usd":0.002565,"raw_usage":{"total_tokens":1054,"prompt_tokens":831,"num_sources_used":0,"completion_tokens":84,"cost_in_usd_ticks":25648000,"prompt_tokens_details":{"text_tokens":831,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":139,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":831,"tokens_out":84,"duration_ms":3996,"temperature":1.0,"reasoning_tokens":139,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-31T01:10:57.018096+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"Find a device whose bias-KPFM and zero-bias LBIC patterns match the paper’s Schottky (or tunnel) assignment, yet temperature-dependent transport or a clean high-vacuum barrier-height measurement shows the opposite dominant mechanism; that mismatch would break the fingerprint rule.","supporting_citations":[],"review_version":1}