{"id":"e3d00cb8-b383-4c7a-a03f-05bdfda0b988","arxiv_id":"1908.09013","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A 1:10 mixture of sodium cholate and sodium chloride as a CVD growth promoter yields predominantly bilayer WSe2 flakes with lateral p-n junctions.","lead":"This paper reports a chemical recipe that grows almost pure bilayer flakes of the semiconductor WSe2, with sharp junctions between one-layer and two-layer regions. The recipe is a practical advance for building atomically thin electronic and optoelectronic devices.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 100% bilayer claim rests on an unstated optical census: no systematic layer-number verification of the >500 flakes is reported, and the same paragraph concedes non-bilayer multilayer spots. The counting and coverage protocol must be shown before this central number can be accepted.","rationale":"The reader's weakest_assumption identifies the unsupported WOxCl/cholate mechanism. That is a valid concern about explanation, but it is not the load-bearing point for the headline result: the recipe and the observed device behavior would remain important even if the mechanism were different. The load-bearing point is the quantitative claim of near-perfect bilayer selectivity, which is the basis for the claimed advance over prior work. The text gives no evidence that the >500-flake census used anything beyond optical inspection, and it explicitly acknowledges non-bilayer bright spots, creating ambiguity about what was counted. This is fixable by a more rigorous census, so it warrants a conditional rather than a reject verdict. I partially agree with the reader: their rationale flags the overclaim, but their formal weakest_assumption is the mechanism rather than the census.","tokens_in":15400,"tokens_out":9482,"duration_ms":100585,"concrete_test":"Request the raw full-area optical images used for the >500-flake count; register every WSe2 feature in the 1×2 cm2 area and classify each feature by AFM step height (about 0.8 nm per layer) or by Raman/PL maps at 248/250 cm−1 and 1.66/1.57 eV, including the bright multilayer spots. Report the bilayer flake fraction with a binomial confidence interval and recompute the surface coverage. If fewer than about 95% of flakes are bilayer, or any exclusively monolayer flake is found, the 100% bilayer claim and the stated improvement over the roughly 80% prior art must be revised.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is the quantitative selectivity statement in Results and Discussion: 'After examining >500 flakes over an area of 1×2 cm2, it was confirmed that 100% of the flakes were bilayer WSe2 ... No monolayer WSe2 flakes were observed.' For this claim to hold, the census must be both accurate and representative. The paper reports no counting protocol, no sampling plan, and no per-flake layer identification for the census. The AFM (Fig. 3a-c), Raman/PL (Fig. 3d-f), and STEM (Fig. 4) validations are performed on selected flakes only; the >500-flake count appears to rely on optical contrast, which is not calibrated over the full area. The same paragraph also concedes 'only a few multilayer regions (bright spots in Figure 2c)' — those are non-bilayer WSe2, so '100% of the flakes were bilayer' is not the same as 100% bilayer selectivity, and the reported 67.8±1.5% coverage is given without the measurement procedure. Since the paper's advance over the prior art is defined as exceeding roughly 80% bilayer purity, an unverified '100%' is the load-bearing number. If the count used a biased field of view or excluded multilayer bright spots, the headline selectivity could be materially lower.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports a CVD growth method for bilayer WSe2 using a 1:10 molar mixture of sodium cholate and NaCl as a growth promoter. The authors claim that, after examining more than 500 flakes over a 1×2 cm2 area, 100% of the flakes were bilayer WSe2 with a surface coverage of 67.8 ± 1.5% and lateral sizes of 40–80 μm. The growth produces star-like bilayer flakes with well-defined monolayer–bilayer junctions; these junctions show p-n diode rectification and ambipolar transport in back-gated FETs. The paper also reports a SAED survey of ~100 bilayer flakes showing twist angles of 0°/60° (84%), 30° (12%), and 15° (4%), and presents DFT calculations of stacking energies that predict the dominance of 0° (AB) and 60° (AA') stacking. The work is presented as a method for layer-controlled growth of TMDs, with extension to WS2.","tokens_in":15697,"tokens_out":4877,"duration_ms":47772,"significance":"If the quantitative selectivity claim is supported, this work represents a significant advance over prior reports in which bilayer content did not exceed ~80%, and it provides a scalable route to lateral 1L–2L p-n junctions for atomically thin optoelectronics. The manuscript's strengths include a detailed, reproducible growth recipe; multi-technique validation (AFM, Raman, PL, and atomic-resolution STEM) on selected flakes; twist-angle statistics; and device measurements demonstrating rectification and ambipolar transport. The DFT stacking-energy analysis is consistent with the observed twist distribution, although it is not fitted to it. The main weakness is that the headline statistics—100% bilayer selectivity and 67.8% coverage—are not backed by a transparent counting and coverage protocol, and the text contains an internal inconsistency between the '100% bilayer' statement and the admission of some multilayer regions.","major_comments":[{"comment":"The sentence 'After examining >500 flakes over an area of 1×2 cm2, it was confirmed that 100% of the flakes were bilayer WSe2' is internally inconsistent with the next sentence, which concedes 'only a few multilayer regions (bright spots in Figure 2c)'. A few multilayer regions are non-bilayer features, so the census either excluded them or the claim should be qualified as applying only to the isolated star-like flakes. The manuscript reports no counting protocol, no sampling plan, and no per-flake layer-number verification; AFM, Raman/PL, and STEM validation are performed on selected flakes only. Because the stated advance over prior work is defined as exceeding ~80% bilayer purity, this unverified '100%' is load-bearing and must be backed by a transparent census, such as optical contrast calibrated against AFM on a random subset, with the number of images and fields counted and a clear rule for handling multilayer regions.","section":"Results and Discussion, paragraph after Fig. 2c-d (main text: 'After examining >500 flakes...')"},{"comment":"The surface coverage value '67.8 ± 1.5%' is reported without a measurement procedure: no image analysis method, threshold criterion, number of optical images analyzed, or definition of uncertainty. Please provide the full protocol, representative raw images with labeled analysis, and clarify whether the coverage is the fraction of substrate covered by bilayer WSe2 or includes monolayer and multilayer regions. Without this information, the coverage number cannot be reproduced or evaluated.","section":"Results and Discussion, same paragraph (coverage value)"},{"comment":"The proposed mechanism, in which cholate anions enhance WOx adsorption and NaCl converts WOx into volatile WOxCl to increase the vapor-phase W flux, is stated as fact in the abstract and text, but no direct evidence is presented: there is no measurement of WOxCl formation, vapor-phase W species, or surface adsorption as the NaCl concentration is varied. The observed trend (monolayer with cholate, bilayer with 1:10 mixture, multilayer with pure NaCl) is consistent with the mechanism but does not uniquely confirm it. Please either add supporting measurements (e.g., vapor-phase mass spectrometry or controlled flux measurements) or explicitly present the mechanism as a hypothesis that remains to be tested.","section":"Results and Discussion, Figure 1 discussion and subsequent mechanism text"},{"comment":"The sentence 'Note that all the above mentioned morphologies for the as-grown WSe2 also showed a selectivity of ~ 100% based on the statistical analysis of >500 flakes over an area of 1×2 cm2' appears immediately after descriptions of growths at 75 mM and 25 mM NaCl that 'yielded a large amount of randomly distributed 2L regions on large 1L WSe2 flakes' and 'nearly exclusively monolayer star-shape WSe2 flakes decorated with a high density of small 2L regions', respectively. As written, the statement contradicts those descriptions. Please clarify what quantity is claimed to be ~100% selective for each promoter composition, and provide the corresponding statistics separately from the optimized growth.","section":"Results and Discussion, paragraph on NaCl concentration series"}],"minor_comments":[{"comment":"The caption uses '(d)' twice, once for the Raman spectra and once for the photoluminescence spectra; the labels should be renumbered as (d), (e), and (f).","section":"Figure 3 caption"},{"comment":"The phrase 'holy-carbon TEM grid' should be 'holey-carbon TEM grid'.","section":"Methods, Characterization"},{"comment":"The text 'E A G E R 1838412' appears to contain a typo; it should likely read 'EAGER 1838412'.","section":"Acknowledgments"},{"comment":"The reference to 'Table 1' occurs after the table is presented in the text; consider placing the table near its first mention or adjusting the order for readability.","section":"Results and Discussion (DFT energetics paragraph)"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is likely of interest to the 2D materials community, and the growth recipe and device results are valuable. My main concern is the statistical basis of the headline selectivity claim; if the authors can provide the counting protocol, reconcile the multilayer spots with the '100% bilayer' statement, and clarify the coverage analysis, I would support publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nTwo things to know up front. The paper reports a new promoter mixture—1:10 sodium cholate:NaCl—for CVD growth of bilayer WSe2, and the twist-angle statistics (84% 0/60°, 12% 30°, 4% 15° over ~100 flakes) line up with their DFT stacking energies. That is a genuinely useful contribution. But the headline claim of '100% bilayer' flakes over a 1×2 cm2 area rests on a census that is never described, and the same paragraph concedes multilayer spots. That inconsistency has to be fixed before the central number is credible.\n\nWhat the paper does well: the growth protocol is detailed enough to reproduce. The layer assignments on selected flakes are solid—AFM step heights, Raman shifts, PL peak positions, and STEM/SAED all check out. The 1L-2L boundary is atomically sharp in the STEM images. The SAED survey of ~100 flakes gives a clean twist-angle distribution, and the DFT ordering (0/60 most stable, then 30, then 15) matches the observed frequencies. The p-n diode rectification and ambipolar behavior are consistent with earlier lateral-junction reports, so not new, but the device data are decent.\n\nWhere it gets soft. The '100% bilayer' statement is load-bearing because the paper motivates itself as exceeding the ~80% selectivity of prior work. Yet there is no counting protocol: no description of how the >500 flakes were selected, whether the census used optical contrast only, or how the 67.8±1.5% coverage was computed. And the sentence 'only a few multilayer regions (bright spots in Figure 2c)' directly contradicts '100% of the flakes were bilayer.' The authors need to either exclude those spots explicitly and report a corrected percentage, or show that the spots are something other than WSe2 multilayer regions. The mechanism—cholate anions enhance WOx adsorption, NaCl produces volatile WOxCl—is plausible but entirely inferred; no WOxCl or flux measurement is presented. That's acceptable for a recipe paper if stated as a hypothesis, but it is not confirmed. The electrical statistics are also thin: one 1L-2L junction device is shown, and the rectification ratio is modest.\n\nThe DFT part is not circular. They use their own prior thermodynamic criterion and a Na-doped edge energy, but the twist-angle prediction is not fitted to the data; it emerges from the energetics. Fine.\n\nBottom line: the paper is a solid incremental advance in TMD growth, but the central quantitative claim needs to be documented or softened. I'd send it to a careful referee. The recipe and the twist-angle/DFT correlation are worth publishing after revision.","headline":"A useful bilayer WSe2 growth recipe, undercut by an unverifiable '100% bilayer' claim and a speculative mechanism; still worth refereeing.","tokens_in":16230,"tokens_out":2659,"would_cite":false,"duration_ms":26972,"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":"By balancing a molecule that sticks tungsten to the surface against a salt that volatilizes it, the paper grows WSe2 as almost pure bilayer flakes whose 1L-2L boundaries act as p-n diodes.","keywords":["two-dimensional materials","chemical vapor deposition","bilayer WSe2","growth promoter","sodium cholate","sodium chloride","monolayer-bilayer junction","p-n diode"],"falsifier":"A decisive experiment would place a clean witness substrate downstream during growth and measure the tungsten content arriving in the vapor with cholate-only, 1:10 cholate/NaCl, and NaCl-only promoters under otherwise identical conditions. The proposed mechanism predicts a monotonic increase in vapor-phase tungsten flux as NaCl is added; if the flux does not increase, or if WOxCl is absent from the vapor, the central layer-control mechanism is unsupported.","tokens_in":15240,"feed_emoji":"🧪","tokens_out":10390,"duration_ms":103039,"temperature":0.7,"pith_summary":"This paper reports a chemical vapor deposition recipe that grows tungsten diselenide (WSe2) almost exclusively as bilayer flakes over a macroscopic 1×2 cm2 area. By mixing the organic promoter sodium cholate with sodium chloride at a 1:10 molar ratio, the authors claim that every one of more than 500 inspected flakes was bilayer, with no monolayer flakes observed and a surface coverage of 67.8 ± 1.5%. The bilayer flakes contain atomically sharp monolayer-bilayer boundaries that behave as lateral p-n diodes with rectification and ambipolar transport. Because bilayer TMDs have an indirect bandgap and different band offsets from monolayers, a method that directly yields bilayer material with built-in 1L-2L junctions matters for atomically thin electronics and optoelectronics.","feed_headline":"Salt mix yields 100% bilayer WSe2 flakes","feed_subtitle":"A sodium cholate and NaCl promoter tunes tungsten vapor so every flake is bilayer and forms a lateral p-n junction.","key_machinery":"The load-bearing object is the two-component growth promoter, 25 mM sodium cholate plus 250 mM NaCl at a 1:10 molar ratio, spin-coated with the tungsten precursor onto the substrate. It operates by balancing two channels of tungsten delivery: cholate anions enhance adsorption of WOx onto the substrate, while NaCl converts some WOx into volatile WOxCl that raises the vapor-phase tungsten flux. The paper couples this to a thermodynamic criterion, $\\varepsilon_{L_1:S} - \\varepsilon_{L_1:L_2} > \\eta\\left(\\frac{\\gamma}{L_2} - \\frac{\\gamma}{L_1}\\right)$ with $\\eta = 4/\\sqrt{3}$, taken from a prior vertical-growth model, to predict which stacking orientations are stable: AB and AA' have nearly equal and most-favorable stacking energies, giving the dominant 0°/60° population, while 15° and 30° moiré stackings are less favorable and rarer.","core_discovery":"The paper claims that a 1:10 molar mixture of sodium cholate and sodium chloride, used as a growth promoter in chemical vapor deposition, produces WSe2 almost exclusively as bilayer flakes: in a survey of more than 500 flakes across a 1×2 cm2 area, all were bilayer, with no monolayer flakes and only rare multilayer spots, at a surface coverage of 67.8 ± 1.5%. The as-grown flakes contain sharp monolayer-bilayer lateral junctions, and electrical devices measured across those junctions show p-n diode rectification (rectification ratio 13.5 at a gate voltage of -80 V) and ambipolar transport. The paper also reports that roughly 84% of bilayer regions adopt 0° (AB) or 60° (AA') stacking, with 15° and 30° moiré twists in the remainder, and uses DFT stacking energies to argue that these populations follow from the relative thermodynamic stability of the orientations.","pith_inferences":["The paper's WS2 result hints that the 1:10 ratio is transferable across TMDs; a systematic ratio sweep for MoS2 and MoSe2 would show whether this is a general layer-number dial, which the paper leaves untested.","Because no direct detection of WOxCl or vapor-phase tungsten flux is reported, a surface-chemistry or molten-salt-spreading role for NaCl remains a live alternative; a downstream witness-substrate measurement of tungsten deposition with and without NaCl would settle it.","The as-grown twist-angle variety (0°, 15°, 30°) makes it possible to test whether the 1L-2L diode's band offset and rectification depend on stacking orientation, a question the paper does not address."],"forward_implications":["The same 1:10 cholate:NaCl promoter yields preferential bilayer WS2, so the approach is not limited to WSe2.","Monolayer-bilayer junctions are present as-grown, so lateral p-n diodes can be fabricated without post-growth stacking or transfer.","The dominance of AB and AA' stacking agrees with the calculated stacking energies, supporting thermodynamic control of interlayer twist during CVD.","Bilayer devices reach higher carrier mobility than monolayer devices (7.9 vs 3.5-4.3 cm2 V-1 s-1), so the layer control translates directly into a transport benefit."],"supporting_citations":[{"why":"Supplies the multiscale vertical-growth model and the thermodynamic criterion used to explain why higher vapor flux favors bilayer growth and which twist angles are stable.","marker":"20"},{"why":"Establishes that organic seeding promoters such as sodium cholate enhance precursor adsorption and promote monolayer growth.","marker":"40"},{"why":"Shows halide-assisted growth of WSe2 and WS2 through volatile oxyhalide formation, the basis for the NaCl channel.","marker":"43"},{"why":"Documents the role of NaCl in epitaxial MoS2 growth, supporting the claim that NaCl increases volatile precursor transport in the vapor.","marker":"44"},{"why":"Provides the band-offset mechanism for atomically thin lateral p-n junctions, used to interpret the 1L-2L diode behavior.","marker":"27"},{"why":"Supplies the WSe2 FET reference behavior and growth-mechanism background used to analyze the transport data.","marker":"37"},{"why":"Provides the optical-contrast method used to identify monolayer and bilayer regions in the large-area flake survey.","marker":"45"},{"why":"Gives the Raman mode positions for monolayer and bilayer WSe2 used to map the 1L and 2L regions.","marker":"50"},{"why":"Provides the photoluminescence signatures of the monolayer direct gap and bilayer indirect gap used to confirm layer identity.","marker":"51"}],"fun_headline_variants":["Salt promoter gives 100% bilayer WSe2","Pure bilayer WSe2 with lateral p-n junctions","100% bilayer WSe2 growth via salt mix","Salt-mix CVD grows only bilayer WSe2"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The paper's layer-control story depends on the idea that sodium chloride boosts tungsten delivery by forming a vapor-borne tungsten oxychloride, but the paper never directly measures that compound or the vapor flux; if the boost comes from some other effect, the recipe could still work but the stated explanation would not be established.","fun_headline_variants_meta":{"raw":{"variants":["Salt promoter gives 100% bilayer WSe2","Pure bilayer WSe2 with lateral p-n junctions","100% bilayer WSe2 growth via salt mix","Salt-mix CVD grows only bilayer WSe2"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000517,"raw_usage":{"total_tokens":2492,"prompt_tokens":918,"completion_tokens":1574,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":534,"completion_tokens_details":{"reasoning_tokens":1511}},"tokens_in":534,"tokens_out":1574,"duration_ms":11353,"temperature":1.0,"reasoning_tokens":1511,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:24:11.060326+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A decisive experiment would place a clean witness substrate downstream during growth and measure the tungsten content arriving in the vapor with cholate-only, 1:10 cholate/NaCl, and NaCl-only promoters under otherwise identical conditions. The proposed mechanism predicts a monotonic increase in vapor-phase tungsten flux as NaCl is added; if the flux does not increase, or if WOxCl is absent from the vapor, the central layer-control mechanism is unsupported.","supporting_citations":[{"cited_title":"D.; Luo, Z., Synthesis of 2d Transiti on Metal Dichalcogenides by Chemical Vapor Deposition with Controlled Layer Number and Morphology","cited_arxiv_id":null,"evidence_quote":"Gives the Raman mode positions for monolayer and bilayer WSe2 used to map the 1L and 2L regions."}],"review_version":1}