{"id":"1f86a354-9966-43ca-b26e-468cb0dcc36d","arxiv_id":"2508.20743","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Single-photon emitters in bilayer MoTe2 show near-infrared emission and a post-selected Hong-Ou-Mandel visibility up to about 40 percent, the first such measurement for a TMD emitter.","lead":"This paper reports a repeatable way to make single-photon emitters in bilayer MoTe2 that emit near-infrared light (1090-1200 nm), with high purity and voltage-tunable wavelength. The authors also report the first two-photon interference from such a TMD emitter, though the indistinguishability is modest and relies on temporal post-selection.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Raw HOM visibility is only ~1.9σ from zero; post-selected >30% values lack error bars, leaving the central indistinguishability claim statistically unsupported.","rationale":"After reading the manuscript, I find the reported single-photon purity, fast lifetimes, and Stark tuning to be credible and well-supported. The HOM measurement, however, is the linchpin of the abstract's headline claim. The raw visibility of 10.5±5.4% is only marginally different from zero, and the post-selected values lack confidence intervals. The reader's weakest assumption focused on the comparability of filtered vs. raw visibilities; I agree that such a comparison is problematic, but the more fundamental issue is that the filtered values may themselves be fluctuations. Without error bars, there is no way to know whether the >30% values are real or artifacts of choosing a narrow window around the central peak where statistical noise is large. A bootstrap or Monte Carlo analysis of the raw data would settle this. If the raw visibility does not reach statistical significance, the paper should not claim a demonstration of indistinguishability, and the 'highest reported' comparison becomes irrelevant. The verdict CONDITIONAL remains appropriate: the result is plausible and could be confirmed by re-analysis, but the strong claim is not yet supported.","tokens_in":19075,"tokens_out":6991,"duration_ms":69464,"concrete_test":"Perform a bootstrap resampling of the raw coincidence histograms underlying Fig. 5a. Compute the full-window visibility (12.5 ns) and the sub-200 ps visibility with 95% confidence intervals. Also, define the temporal filter a priori based on the independently measured coherence time (τc ≈ 63 ps) and recompute the visibility for a fixed window (e.g., ±63 ps and ±200 ps). If the full-window confidence interval includes zero, or if the filtered windows do not show a statistically significant positive visibility, the central indistinguishability claim is not established.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim — first demonstration of photon indistinguishability from a TMD emitter in the near-infrared — rests on a measured HOM visibility of V = 10.5 ± 5.4% (Fig. 5b). This is only ~1.9σ above zero, below the conventional 3σ threshold for a detection. The headline 'up to V_HOM ~40%' comes from post-selecting integration windows below 200 ps (Fig. 5c), but no error bars are given for these filtered values, and the window choice appears post hoc, introducing selection bias. Moreover, the g(2)(0) of the same emitter under HOM excitation conditions is reported as below 0.35 (Supporting Note S6), indicating substantial multi-photon and/or background emission that directly reduces the achievable HOM dip. If the raw dip is not statistically robust, then the claim of indistinguishability is not supported, and the comparison to previous TMD measurements (e.g., WSe2, ~2%) is moot. The paper's other results (single-photon purity, Stark tuning, narrow linewidths) are plausible and credible, but the headline claim requires a statistically solid HOM measurement.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports a deterministic fabrication route to quantum emitters in bilayer MoTe2, combining strain engineering with electron-beam defect activation. The emitters show narrow (nominally resolution-limited) emission lines around 1090–1200 nm, strong linear polarization, sub-nanosecond lifetimes, single-photon purity with g(2)(0) < 0.2, and Stark tuning over ~3 meV. The central novelty claim is the first demonstration of two-photon (HOM) interference from a TMD emitter in the near-infrared, with a raw visibility of V = 10.5 ± 5.4% and a post-selected visibility above 30% for sub-200 ps integration windows.","tokens_in":19365,"tokens_out":2901,"duration_ms":32421,"significance":"If the HOM claim were robust, this would be an important step: it would establish MoTe2 as a telecom-adjacent single-photon platform and report the first measured photon indistinguishability for any TMD emitter in the near-infrared. The paper has clear strengths: a reproducible fabrication protocol, systematic comparison of above-band and quasi-resonant excitation, data on encapsulation and electrostatic biasing, and a direct measurement of two-photon interference. The single-photon purity, Stark tuning, and lifetime data are internally consistent and credible. However, the headline indistinguishability claim is statistically fragile, and the paper's own supporting data (g(2)(0) < 0.35 under HOM excitation) weaken it further.","major_comments":[{"comment":"The raw HOM visibility is V = 10.5 ± 5.4%, i.e. only ~1.9σ above zero. This does not meet the conventional 3σ threshold for a detection. The abstract and Discussion ('first demonstration of photon indistinguishability', 'highest reported indistinguishability for any TMD') are therefore not supported by the raw data. The authors should either acquire substantially more statistics, report the result as an upper limit/preliminary indication, or present a clearly justified Bayesian or multi-measurement error analysis.","section":"§VI, Fig. 5b"},{"comment":"The post-selected visibility >30% for integration windows below 200 ps is presented without error bars, and the choice of window appears to be made after seeing the data. This is a selection effect. Moreover, comparing this filtered value to raw, unfiltered visibilities from other TMD sources (e.g., WSe2 in ref. 33) is not a like-for-like comparison; temporal filtering removes the very multi-photon and re-excitation events that lower the raw visibility. The claim of 'highest reported indistinguishability' should either be based on the unfiltered value or clearly labeled as a post-selected, non-on-demand figure of merit.","section":"§VI, Fig. 5c"},{"comment":"The authors note that under the pulsed quasi-resonant excitation used for HOM, the fitted g(2)(0) is below 0.35, and they attribute the reduced visibility partly to elevated g(2)(0). A g(2)(0) as high as 0.35 implies substantial multi-photon or background emission, which directly caps the achievable HOM visibility. The paper should quantify this effect, e.g., by applying a background-correction model to the HOM signal, or by explicitly stating that the raw HOM value is an upper bound contaminated by non-single-photon events. Without this, the interpretation of the 10% raw visibility is ambiguous.","section":"Supporting Note S6 and §VI"},{"comment":"The linewidth-to-lifetime ratio R is a central figure used to claim 'approaching the transform-limited regime' (R down to ~55). However, the measured linewidths are at the instrument resolution limit (Wexp ~150 µeV vs. resolution ~140–160 µeV), so Wexp cannot actually resolve changes in the true linewidth. The observed reduction of R from 100 to 55 is driven entirely by the measured lifetime shortening, not by a measured narrowing of the emission line. The claim should be rephrased as an upper bound on R, and the resolution-limited nature of Wexp should be explicitly propagated into the uncertainty on R. The current wording overstates what the data establish.","section":"§V, Supporting Note S11"}],"minor_comments":[{"comment":"Typo: 'yeilds' should be 'yields'.","section":"§IV"},{"comment":"In the description of the bias dependence, 'Start effect' should be 'Stark effect'.","section":"§V"},{"comment":"The cross-polarized data are 'slightly time-shifted for clarity'. This should be stated in the caption and the shift amount given, otherwise the reader may misread the coincidence peak positions.","section":"Fig. 5a"},{"comment":"The plot has no error bars and no indication of how the integration windows were chosen. At minimum, error bars from Poissonian counting statistics should be added.","section":"Fig. 5c"},{"comment":"The IQE values (0.81, 0.66, 0.37) are extracted from bunching fits and on/off times using a specific blinking model. The model dependence should be stated more explicitly, and the uncertainty in the IQE estimates should be given.","section":"§III"},{"comment":"Ref. 41 is cited as 'Tunable and low-noise wse2 quantum emitters' (arXiv:2507.03355). If this is the source of the WSe2 HOM visibility ~2% comparison, the comparison should be made explicit in the text.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The paper's strongest selling point is the HOM measurement, but the raw visibility is not statistically significant and the post-selected value is not directly comparable to previous reports. If the authors cannot provide more data or a rigorous error analysis, the claim of 'first demonstration of photon indistinguishability' should be withdrawn or heavily qualified. The remaining characterization (purity, lifetimes, Stark tuning, reproducibility) is solid and would support a more modest publication focused on the emitter platform."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is the first Hong-Ou-Mandel interference measurement on a MoTe2 quantum emitter, and that is the real (and only) headline. The rest of the paper is solid, careful characterization of a strain-engineered, defect-activated bilayer MoTe2 single-photon source in the 1090–1200 nm range. What they do well: reproducible fabrication across samples, clean polarization, sub-nanosecond lifetimes, g(2)(0)<0.1 under several excitation conditions, clear Stark tuning, and a sensible comparison of hBN-encapsulated vs. bare emitters. They also explicitly flag their own limitations (e.g., coherence time partly limited by setup resolution), which is good practice.\n\nThe soft spot is exactly what the stress-test note says: the raw HOM visibility is 10.5±5.4%, which is ~1.9σ from zero. That alone is not a convincing detection. The \"up to 40%\" number comes from post-selecting integration windows below 200 ps, and those filtered values have no error bars. Since the paper claims \"highest reported indistinguishability for any TMD QE\" on the strength of that filtered number, the claim is statistically unsupported as presented. A referee should ask for a careful error propagation on the filtered visibility, a non-post-hoc window choice, and a clear statement of how this filtered value compares to the raw visibilities quoted for other TMD emitters.\n\nA smaller but real concern: the linewidths used for the R ratios are at the instrument resolution limit (~150 µeV vs. ~140–160 µeV resolution). So the claim of R~55 \"approaching the transform limit\" is an upper bound, not a measured quantity. The IQE values and the HOM coherence time are extracted from fits to the same data they are used to explain, which is common but worth labeling.\n\nNone of this invalidates the central fabrication and single-photon results. The paper is honest, the data look internally consistent, and the HOM measurement, even if marginal, is a real first. With statistical tightening and a modest revision, the indistinguishability claim could become defensible. I would send it to peer review; a referee can push on the numbers without rejecting the work.","headline":"First HOM measurement in MoTe2, but the headline indistinguishability claim rests on a marginal raw visibility and error-bar-free post-selection.","tokens_in":19891,"tokens_out":3457,"would_cite":true,"duration_ms":34912,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["42.50.Ar","42.50.Ex","78.67.-n"],"model":"deepseek-v4-flash","headline":"Bilayer MoTe2 quantum emitters demonstrate the first two-photon interference in the near-infrared, with Hong-Ou-Mandel visibilities reaching ~10% and up to ~40% with temporal post-selection.","keywords":["single-photon sources","MoTe2","quantum emitters","Hong-Ou-Mandel interference","photon indistinguishability","strain engineering","near-infrared emission","Stark tuning"],"falsifier":"Measure the HOM visibility at an excitation power where the pulsed g(2)(0) is below 0.05 (so re-excitation is negligible) and compare the full 12.5-ns integration window with the sub-200-ps filtered window; a persistent gap between them would show the filtered visibility is inflated by discarding multi-photon events that a triggered source must suppress.","tokens_in":18989,"feed_emoji":"💡","tokens_out":5788,"duration_ms":55578,"temperature":0.7,"pith_summary":"This paper tries to establish that bilayer MoTe2, a molybdenum-based 2D semiconductor emitting in the near-infrared, can host deterministic single-photon emitters whose photons are partially indistinguishable. Using strain wrinkles created by nanopillars plus electron-beam defect activation, the authors make emitters in the 1090–1200 nm range with high linear polarization, sub-nanosecond lifetimes, high single-photon purity, and Stark tunability. The central result is the first Hong-Ou-Mandel two-photon interference measurement from MoTe2 emitters, with a visibility of about 10% and up to about 40% when only photons within a sub-200 ps time window are selected. This is claimed as the highest photon indistinguishability reported for any TMD-based quantum emitter and the first in the near-infrared, a step toward telecom-compatible single-photon sources.","feed_headline":"MoTe2 emitters hit 40% photon indistinguishability","feed_subtitle":"Nanopillar-strained MoTe2 yields near-infrared single photons with record 2D-emitter indistinguishability.","key_machinery":"The central objects are strain-engineered nanopillars in a three-pointed star shape that create nanowrinkles in bilayer MoTe2, where localized exciton states form, and the Hong-Ou-Mandel interferometer with a 12.5-ns delay line used to measure photon indistinguishability via coincidence peak areas. The temporal filtering analysis (varying the integration window of the zero-delay peak) is the mechanism that exposes the fast coherence component and produces the headline visibility values.","core_discovery":"The authors demonstrate that bilayer MoTe2 quantum emitters, created deterministically by nanopillar-induced strain and e-beam defect activation, produce triggered single photons in the 1090–1200 nm range with purity g(2)(0) < 0.1, lifetimes as short as 130 ps, and Stark shifts over ~3 meV. Most notably, pulsed excitation generates pairs of consecutively emitted photons that interfere in a Hong-Ou-Mandel interferometer, yielding a raw visibility of 10.5 ± 5.4% and over 30% when the coincidence integration is temporally filtered to below 200 ps. They interpret the strong time-filter dependence as consistent with theoretical predictions and as evidence that coherence is limited by fast dephasi","pith_inferences":["The sub-200 ps post-selected visibility of ~40% may not translate to the usable indistinguishability required for linear-optics quantum computing, which relies on interference of simultaneous photons from independent sources; a two-source HOM test would be a stricter benchmark.","The same strain-and-defect fabrication method could plausibly be transferred to other molybdenum-based TMDs or multilayer MoTe2 to target the 1310 nm and 1550 nm telecom windows, since emission energy shifts with layer count.","Bias-induced shortening of the radiative lifetime at nearly constant linewidth suggests electric-field control of oscillator strength could become a general coherence-engineering tool for 2D emitters.","The paper's comparison to prior WSe2 emitters uses the filtered visibility value; a fair platform comparison would require identical integration windows across materials."],"forward_implications":["If emission can be shifted further with thicker flakes or modified strain, MoTe2 sources could reach the telecom O- and C-bands, making them directly useful for fiber-based quantum networks.","The ~3 meV Stark tuning provides a concrete path to spectrally matching two spatially separated emitters, a prerequisite for remote entanglement and interference between independent sources.","The strong time-filter dependence of the HOM visibility implies that suppressing fast dephasing—via cavity Purcell enhancement or resonant excitation—could substantially raise the usable indistinguishability.","The reproducible fabrication across multiple samples indicates a scalable route to arrays of tunable near-infrared single-photon emitters.","The measured linewidth-to-transform-limit ratio R ~ 55 shows these emitters approach transform-limited coherence without cavity integration, suggesting headroom for improvement with photonic structuring."],"supporting_citations":[{"why":"Provides the previous TMD HOM visibility (~2% in WSe2) that this work compares against as the indistinguishability benchmark.","marker":"[33]"},{"why":"Supplies the nanopillar strain-engineering method and the R = Wexp/Wrad coherence metric used throughout.","marker":"[41]"},{"why":"Demonstrates nanowrinkle strain localization and e-beam defect activation that the MoTe2 fabrication adapts.","marker":"[34]"},{"why":"Reports earlier MoTe2 quantum emitters with much longer lifetimes and broader linewidths, serving as the baseline this work improves on.","marker":"[42]"},{"why":"Provides the Hong-Ou-Mandel interferometer implementation and analysis, including temporal filtering of the zero-delay peak.","marker":"[53]"},{"why":"Supplies the fitted model for pulsed two-photon interference with bunching used to extract the HOM visibility.","marker":"[54]"},{"why":"Gives the theoretical prediction that HOM visibility depends strongly on the temporal integration window, motivating the post-selection analysis.","marker":"[56]"}],"fun_headline_variants":["MoTe2 emitters achieve 40% post-selected photon indistinguishability","Best-ever 2D emitter indistinguishability: 40% from MoTe2","Strained MoTe2 produces near-IR single photons with record coherence","Bilayer MoTe2 delivers triggered photons with 40% HOM visibility","Near-infrared MoTe2 single-photon source reaches 40% indistinguishability"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The headline indistinguishability claim rests on the assumption that the temporally filtered Hong-Ou-Mandel visibility (sub-200 ps window, values above 30% without error bars) is a fair, directly comparable measure of single-photon indistinguishability, equivalent to the unfiltered raw visibilities quoted for other TMD emitters.","fun_headline_variants_meta":{"raw":{"variants":["MoTe2 emitters achieve 40% post-selected photon indistinguishability","Best-ever 2D emitter indistinguishability: 40% from MoTe2","Strained MoTe2 produces near-IR single photons with record coherence","Bilayer MoTe2 delivers triggered photons with 40% HOM visibility","Near-infrared MoTe2 single-photon source reaches 40% indistinguishability"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000388,"raw_usage":{"total_tokens":1942,"prompt_tokens":863,"completion_tokens":1079,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":607,"completion_tokens_details":{"reasoning_tokens":975}},"tokens_in":607,"tokens_out":1079,"duration_ms":10780,"temperature":1.0,"reasoning_tokens":975,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T14:51:42.848014+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the HOM visibility at an excitation power where the pulsed g(2)(0) is below 0.05 (so re-excitation is negligible) and compare the full 12.5-ns integration window with the sub-200-ps filtered window; a persistent gap between them would show the filtered visibility is inflated by discarding multi-photon events that a triggered source must suppress.","supporting_citations":[{"cited_title":"Drawer, V","cited_arxiv_id":null,"evidence_quote":"Provides the previous TMD HOM visibility (~2% in WSe2) that this work compares against as the indistinguishability benchmark."},{"cited_title":"Tunable and low-noise WSe$_2$ quantum emitters for quantum photonics","cited_arxiv_id":"2507.03355","evidence_quote":"Supplies the nanopillar strain-engineering method and the R = Wexp/Wrad coherence metric used throughout."},{"cited_title":"Paralikis, C","cited_arxiv_id":null,"evidence_quote":"Demonstrates nanowrinkle strain localization and e-beam defect activation that the MoTe2 fabrication adapts."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reports earlier MoTe2 quantum emitters with much longer lifetimes and broader linewidths, serving as the baseline this work improves on."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the Hong-Ou-Mandel interferometer implementation and analysis, including temporal filtering of the zero-delay peak."},{"cited_title":"Vannucci, J","cited_arxiv_id":null,"evidence_quote":"Gives the theoretical prediction that HOM visibility depends strongly on the temporal integration window, motivating the post-selection analysis."}],"review_version":1}