{"id":"68bbd089-ffd8-4439-bdb3-bf65ccb90113","arxiv_id":"1908.00609","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"In vertical metal/MoS2/metal junctions on ultra-flat substrates, current increases and the Fowler-Nordheim barrier height decreases with increasing layer count, while blue light reduces the current.","lead":"By pressing a tiny metal probe onto one to five layers of the semiconductor MoS2 on ultra-smooth gold or ITO, the authors found that adding layers raises the current and lowers the energy barrier at the metal contact. They also found that shining blue light lowers the current, a negative photoconductivity effect that matters for designing thin, flexible optoelectronics.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Barrier-height trend rests on an unverified Fowler–Nordheim fit: no FN plots or linearity tests are shown, so the central decreasing trend may be a model artifact.","rationale":"The experimental central claim is not the raw observation that current increases with layer count; it is the inference of a decreasing Fowler–Nordheim barrier from Eq. (1). The manuscript shows non-linear I-V curves and immediately applies Eq. (1), but it never establishes that the data obey the FN functional form. A non-linear I-V curve can arise from several mechanisms, and unless ln(I/V^2) versus 1/V is linear over a finite bias range, the extracted barrier heights are model outputs rather than measurements. This concern is upstream of the reader's stated weakest assumption about d, m*, and contact area: even with perfect values of those parameters, an invalid model would invalidate the trend. The DFT calculation in Fig. 4C provides independent theoretical support for a decreasing Schottky barrier, but it does not validate the experimental FN extraction and uses PBE plus a simple W − χ_e estimate, so it cannot rescue the experiment on its own. Because the data are not public and the Methods section omits fitting specifics, the appropriate disposition is unchanged: the paper should remain CONDITIONAL pending a demonstration of the FN regime, the requested sensitivity analysis, and correction of the electron-affinity sign error in the Conclusions.","tokens_in":11087,"tokens_out":6494,"duration_ms":71036,"concrete_test":"Obtain the layer-averaged I-V data (or raw CAFM maps) from the authors and construct Fowler–Nordheim plots, ln(I/V^2) versus 1/V, for 1L–5L on both Au and ITO. Determine the bias window over which each plot is linear and report the slopes and residuals. If no layer has a statistically linear FN segment, the barrier heights extracted from Eq. (1) are not supported and the central trend is unverified. As a secondary check, refit the same slopes using d = n×0.7 nm plus an offset δ (e.g., 0.3 nm) and a single effective mass m*/m = 0.35; if the extracted barrier no longer decreases monotonically with n, the reported trend is not robust to the assumed inversion parameters.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—that the barrier height decreases as the MoS2 layer number increases—is obtained by inverting Eq. (1), the Fowler–Nordheim tunneling expression, to extract Φ_B from each averaged I-V curve. The manuscript reports only averaged I-V curves and final barrier values (Figs. 2c–d and 3c–d); it never shows ln(I/V^2) versus 1/V plots, fitting ranges, or goodness-of-fit statistics for any layer on either substrate. Eq. (1) is only valid in the high-field FN tunneling regime, and many non-linear I-V characteristics of metal/MoS2/metal junctions can instead be governed by direct tunneling, thermionic emission, or contact-limited injection. If the data do not exhibit a linear FN segment, then every Φ_B value—and the monotonic decrease in Figs. 2(d) and 3(d)—is an artifact of applying a model whose functional form is not verified. Secondarily, even within FN theory, the conversion uses d = n × 0.7 nm and m*/m = 0.35 (1L) or 0.53 (>1L) with no check that these are appropriate for the actual tip–sample contact; a thickness offset such as an interfacial van der Waals gap of roughly 0.3 nm would make the extracted barrier decrease with n even if the true barrier were constant. The DFT trend in Fig. 4C is independent support, but it uses a different definition (W − χ_e with PBE), so it does not validate the FN extraction. Finally, the Conclusions attribute the trend to a 'decrease of the electron affinity' while Fig. 4B shows an increase; although the DFT trend can still arise from a decreasing gold work function, this sign error signals that the interpretation section was not carefully checked.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports conductive atomic force microscopy (CAFM) and photoconductive AFM measurements of 1-5 layer MoS2 on template-stripped Au and ultra-flat ITO substrates, forming vertical metal/MoS2/metal junctions. The authors report that (i) the current increases with layer number up to 5 layers, (ii) Fowler-Nordheim (FN) tunneling analysis yields barrier heights that decrease as the layer number increases, (iii) the same decreasing trend is reproduced by DFT calculations of the Au/MoS2 Schottky barrier, (iv) flake edges are less conductive than the basal plane, and (v) blue-light illumination produces negative photoconductivity. The central quantitative claim is the layer-dependent barrier height, extracted by inverting Eq. (1) using assumed layer thicknesses and effective masses.","tokens_in":11351,"tokens_out":3786,"duration_ms":40963,"significance":"If the central claim is correct, the paper provides a useful counterpoint to the opposite trend reported for MoS2 on rough ITO in Ref. 35, and it identifies substrate flatness as a key variable controlling vertical transport in TMD/metal junctions. The work is strengthened by the use of ultra-flat template-stripped Au and carefully prepared ITO, by measurements on multiple samples, and by independent DFT calculations that reproduce the decreasing barrier trend. The observation of non-transient negative photoconductivity in a vertical junction is also noteworthy. However, the quantitative barrier heights are only as reliable as the FN inversion, and the manuscript does not currently demonstrate that the measured I-V curves are in the FN tunneling regime. The strength of the conclusion therefore hinges on analysis that is not shown.","major_comments":[{"comment":"The central claim that the barrier height decreases with layer number rests entirely on inverting Eq. (1), but the manuscript shows no FN plots (ln(I/V^2) versus 1/V), no fitting ranges, and no goodness-of-fit statistics for any layer on either substrate. Eq. (1) is valid only in the high-field FN tunneling regime, and non-linear I-V characteristics of metal/MoS2/metal junctions can also be governed by direct tunneling, thermionic emission, or contact-limited injection. Unless the authors demonstrate a linear FN segment and specify the bias range used for each fit, the reported Phi_B values and their monotonic decrease are model artifacts rather than measured quantities. This is the load-bearing step for the paper's main claim.","section":"Results and Discussion, Figs. 2(c)-2(d) and 3(c)-3(d)"},{"comment":"The conversion of I-V curves into barrier heights uses d = n x 0.7 nm and m*/m = 0.35 for monolayer and 0.53 for thicker flakes, with an implicit assumption of constant tip contact area and pressure. This procedure is sensitive to systematic errors. In particular, if the actual tunneling distance includes an interfacial van der Waals gap of roughly 0.3 nm, then for a constant true barrier the extracted Phi_B scales as (d_true/d_model)^(2/3), which decreases with increasing n and would mimic the reported trend. Layer-dependent effective mass or contact area would similarly alter the extracted trend. The authors should provide a sensitivity analysis and justify the thickness and effective-mass choices, or compare the data with a direct-tunneling model to show that the trend is robust.","section":"Eq. (1) and the paragraph following it"},{"comment":"The Conclusions attribute the decreasing barrier to a 'decrease of the electron affinity as the layer number increases,' but Fig. 4B shows an incremental increase of the electron affinity with layer number. The DFT barrier in Fig. 4C is defined as W - chi_e, so it decreases partly because chi_e increases, not decreases. This internal inconsistency must be corrected. In addition, the PBE-based W - chi_e is an ideal Schottky barrier and is not the same quantity as the transport barrier extracted from FN tunneling; the DFT calculation provides supporting evidence for a decreasing trend but does not validate the FN extraction procedure.","section":"Conclusions vs. Fig. 4"}],"minor_comments":[{"comment":"The caption reports RMS roughness values of ~0.18 nm for template-stripped Au and ~1.8 nm for as-deposited Au, while the text states 0.2 nm and 2 nm; these values should be reconciled.","section":"Fig. 1(c) caption"},{"comment":"The I-V curves are described as averages over a flake or over every pixel, but no number of pixels, spatial standard deviation, or representative individual curves are provided; this information would help assess variability.","section":"Figs. 2(c) and 3(c)"},{"comment":"The proposed mechanism that illumination increases the barrier height is speculative; the manuscript shows only one I-V pair under illumination (Fig. 5b) and no bias-dependent barrier extraction under illumination, so this explanation should be framed as a hypothesis requiring further evidence.","section":"Negative photoconductivity discussion"},{"comment":"The ITO sheet resistance is reported as '~220 Ω square'; the correct unit is Ω/sq (ohms per square).","section":"Methods"},{"comment":"There are several typographical errors, including 'supplementaly information,' 'AU' for Au in the Fig. 1 caption, and 'icreases' in the ITO results section; these should be corrected.","section":"General"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know about arXiv:1908.00609. First, the central observation—current rises and Fowler-Nordheim barrier height falls as MoS2 goes from 1L to 5L on ultra-flat Au and ITO—is new and runs opposite to the rough-ITO trend of Son et al. (Ref. 35). Second, the negative photoconductivity in a vertical metal/MoS2/metal stack is non-transient, which separates it from the planar trion effect. Both results are concrete and worth taking seriously.\n\nThe paper does several things well. The substrate preparation is careful: template-stripped Au with <0.2 nm roughness and PVD ITO on quartz at <0.7 nm give a clean platform for CAFM. The current maps are internally consistent across multiple samples and two tip types. The DFT calculation in Fig. 4C produces the same decreasing barrier direction, which independently supports the empirical trend. That DFT agreement is the strongest point in the paper.\n\nThe soft spots are real but not fatal. The barrier heights are extracted by fitting the same I-V curves that establish the trend, and the manuscript shows no ln(I/V^2) vs. 1/V plots, no fitting ranges, and no goodness-of-fit statistics. Without those, a reader cannot verify that the data are in the FN regime; if the fit is poor, the Phi_B values are model artifacts. The assumed d = n*0.7 nm and m*/m = 0.35/0.53 are plausible but unchecked, and an interfacial van der Waals gap of ~0.3 nm would produce a decreasing trend even with a constant barrier. The DFT trend is independent, but it is computed as W - chi_e, so it corroborates direction rather than absolute values. The conclusion misstates the DFT result, attributing the trend to a \"decrease of the electron affinity\" when Fig. 4B shows an increase; that sign error should be fixed.\n\nThe negative photoconductivity is supported by spatial subtraction maps showing a consistent current decrease, though the barrier-modulation mechanism is presented as a plausible explanation rather than proven. That is a minor issue.\n\nWho should read this? Anyone working on metal/TMD vertical transport or optoelectronic characterization of few-layer MoS2. The layer-dependent trend and the ultra-flat substrate recipe are useful citable pieces. The paper deserves a serious referee: the observations are new, the DFT support is real, and the problems are fixable with more transparent fitting, raw data, and a corrected interpretation. Send it out.","headline":"Layer-dependent barrier trend is plausible and independently supported by DFT, but the FN extraction is under-documented and the conclusion contains a treatable sign error; worth reviewing with revisions.","tokens_in":11985,"tokens_out":2593,"would_cite":true,"duration_ms":25656,"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":"In vertical metal/MoS2/metal junctions, the tunneling barrier falls as MoS2 goes from one to five layers.","keywords":["MoS2","transition metal dichalcogenides","Fowler-Nordheim tunneling","conductive atomic force microscopy","barrier height","layer-dependent transport","negative photoconductivity","metal-semiconductor interface"],"falsifier":"Measure the same 1-5 layer MoS2 stacks in a junction with an electrode area fixed by lithography rather than by an AFM tip, and extract barrier heights without assuming a layer-independent contact area; the central claim would be falsified if the current still rose with layer count while the deduced barrier height stayed flat or rose, or if using independently measured layer-dependent effective masses inverted the computed trend.","tokens_in":10851,"feed_emoji":"🔬","tokens_out":7275,"duration_ms":67157,"temperature":0.7,"pith_summary":"This paper asks how the energy barrier at a metal/MoS2/metal junction changes as MoS2 is thinned to a few layers. Using conductive atomic force microscopy on ultra-flat gold and indium-tin-oxide electrodes, it finds that current through the vertical junction increases from one to five MoS2 layers, while the Fowler-Nordheim barrier height extracted from the I-V curves decreases. A density-functional calculation of the Au/MoS2 Schottky barrier, defined as the gold work function minus the MoS2 electron affinity, independently gives the same decreasing trend. If correct, this means layer count is a tunable lever for current injection in two-dimensional semiconductor contacts, and that earlier reports of the opposite trend may reflect substrate roughness rather than intrinsic MoS2 physics.","feed_headline":"MoS2 tunneling barrier shrinks as layers stack up","feed_subtitle":"Vertical metal/MoS2/metal junctions carry more current through 1-5 layers, against an earlier rough-substrate result.","key_machinery":"The central object is the Fowler-Nordheim tunneling equation, $$I(V)=\\frac{A_e $q^{3}$ m $V^{2}$}{8\\pi h\\Phi_B $d^{2}$ m^*}\\exp\\left(-\\frac{8\\pi\\sqrt{2m^*}\\$Phi_B^{{3/2}}$d}{3h q V}\\right),$$ which connects measured vertical current to the barrier height $\\Phi_B$. The paper inverts this expression, taking the MoS2 thickness as $d=n\\times0.7$ nm and effective mass ratio $m^*/m=0.35$ for monolayer MoS2 and $0.53$ for thicker flakes, to assign a barrier height to each layer. As an independent check, density functional theory computes the Schottky barrier at Au/MoS2 as $\\Phi_B=W-\\chi_e$, the gold work function minus the MoS2 electron affinity, and finds the same decreasing trend with layer number.","core_discovery":"The paper's central claim is that in a vertical metal/MoS2/metal junction, the barrier to electron tunneling is highest for a monolayer and falls monotonically as the MoS2 thickness is increased to five layers, so the junction current rises with layer number. The claim is established by averaging spatially resolved I-V curves pixel by pixel and applying Fowler-Nordheim tunneling theory to extract barrier heights, with an assumed layer thickness of $n\\times0.7$ nm and effective mass ratio $m^*/m=0.35$ for monolayer and $0.53$ for thicker flakes. The same decreasing barrier trend is obtained independently from density-functional calculations of the Au/MoS2 interface, where the barrier is the gold work function minus the MoS2 electron affinity; the calculated trend persists when the interfacial spacing is artificially increased by 1-3 Å. The paper also reports that the barrier is higher on ITO than on template-stripped Au, that flake edges conduct less than the basal plane, and that blue light reduces the current, a non-transient negative photoconductivity.","pith_inferences":["A natural extension would be to test whether the same decreasing-barrier trend holds for other TMDs such as WSe2 or MoSe2; the paper's mechanism predicts it should whenever the electron affinity rises with layer count.","Because the paper attributes the earlier opposite trend to substrate roughness, a direct test is to vary bottom-electrode roughness deliberately across samples and see whether the barrier trend flips sign at a roughness threshold.","The negative photoconductivity implies photoexcited carriers raise the junction barrier in the vertical geometry; if confirmed by bias- and power-dependent I-V under illumination, it would distinguish this DC effect from the picosecond trion mechanism seen in planar devices."],"forward_implications":["Layer count can serve as a design knob for injection current in vertical MoS2 contacts, with five layers delivering the most current.","The barrier-height difference between Au and ITO means the bottom electrode work function, not just the TMD itself, sets the junction's tunneling resistance.","MoS2 flake edges act as low-conductivity regions, so nanoscale devices contacting edges must expect reduced current relative to basal-plane contact.","Blue-light illumination suppresses vertical current and the suppression grows with layer number, a behavior that could be exploited as an optically controlled switch if the mechanism is confirmed.","Density-functional results indicate the trend originates in band alignment, with the electron affinity rising as layers are added, rather than in simple thickness-induced tunneling attenuation."],"supporting_citations":[{"why":"Establishes the template-stripped Au preparation that gives the sub-0.2 nm roughness the measurements rely on.","marker":"[29-32]"},{"why":"Explains why the bare Au outside the flake shows nearly zero current, defining the baseline contrast.","marker":"[34]"},{"why":"Supplies the effective-mass values used in the Fowler-Nordheim inversion and the rough-ITO result whose opposite layer trend the paper argues against.","marker":"[35]"},{"why":"Provides the metal-MoS2 interface transport analysis that motivates applying Fowler-Nordheim tunneling here.","marker":"[36]"},{"why":"Shows Fowler-Nordheim analysis applied to a metal/van-der-Waals-material/metal junction, supporting the extraction method.","marker":"[37]"},{"why":"Applies the same tunneling formalism to few-layer van der Waals crystals, the measurement class studied here.","marker":"[38]"},{"why":"Documents trion-induced negative photoconductivity in monolayer MoS2, the transient planar-device effect this paper contrasts with its DC result.","marker":"[39]"},{"why":"Reports negative photoconductivity in 2D layered materials, giving context for the observed photoresponse.","marker":"[40]"}],"fun_headline_variants":["Thicker MoS2 layers ease electron tunneling","MoS2 barrier height falls as layers accumulate","Monolayer MoS2 resists current; thick flakes pass","Layer count tunes MoS2 metal junction conductivity"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that every extra MoS2 layer simply adds 0.7 nm of tunneling distance, that the electron effective mass is known for monolayer and thicker flakes, and that the AFM tip contact area and applied pressure are identical at every layer; if any of these differ by layer, the extracted barrier heights, and therefore the central downward trend, would change.","fun_headline_variants_meta":{"raw":{"variants":["Thicker MoS2 layers ease electron tunneling","MoS2 barrier height falls as layers accumulate","Monolayer MoS2 resists current; thick flakes pass","Layer count tunes MoS2 metal junction conductivity"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000313,"raw_usage":{"total_tokens":1860,"prompt_tokens":1110,"completion_tokens":750,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":726,"completion_tokens_details":{"reasoning_tokens":687}},"tokens_in":726,"tokens_out":750,"duration_ms":7722,"temperature":1.0,"reasoning_tokens":687,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T15:44:09.216722+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the same 1-5 layer MoS2 stacks in a junction with an electrode area fixed by lithography rather than by an AFM tip, and extract barrier heights without assuming a layer-independent contact area; the central claim would be falsified if the current still rose with layer count while the deduced barrier height stayed flat or rose, or if using independently measured layer-dependent effective masses inverted the computed trend.","supporting_citations":[{"cited_title":"Uv/Ozone Treated Au for Air- Stable, Low Hole Injection Barrier Electrodes in Organic Electronics","cited_arxiv_id":null,"evidence_quote":"Explains why the bare Au outside the flake shows nearly zero current, defining the baseline contrast."},{"cited_title":"H.; Paulson, J","cited_arxiv_id":null,"evidence_quote":"Supplies the effective-mass values used in the Fowler-Nordheim inversion and the rough-ITO result whose opposite layer trend the paper argues against."},{"cited_title":"S.; Liu, X.; Yoo, W","cited_arxiv_id":null,"evidence_quote":"Provides the metal-MoS2 interface transport analysis that motivates applying Fowler-Nordheim tunneling here."},{"cited_title":"Determination of Carrier Polarity in Fowler– Nordheim Tunneling and Evidence of Fermi Level Pinning at the Hexagonal Boron Nitride/Metal Interface","cited_arxiv_id":null,"evidence_quote":"Shows Fowler-Nordheim analysis applied to a metal/van-der-Waals-material/metal junction, supporting the extraction method."},{"cited_title":"Y.; Lee, S.; Park, B","cited_arxiv_id":null,"evidence_quote":"Applies the same tunneling formalism to few-layer van der Waals crystals, the measurement class studied here."},{"cited_title":"H.; Frenzel, A","cited_arxiv_id":null,"evidence_quote":"Documents trion-induced negative photoconductivity in monolayer MoS2, the transient planar-device effect this paper contrasts with its DC result."},{"cited_title":"Negative Terahertz Photoconductivity in 2d Layered Materials","cited_arxiv_id":null,"evidence_quote":"Reports negative photoconductivity in 2D layered materials, giving context for the observed photoresponse."}],"review_version":1}