{"id":"2c13b942-3396-4778-b6f2-266d3bdc5911","arxiv_id":"2509.06605","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"MBE-grown gamma-InSe/gamma-In2Se3 crystal-phase heterostructures emit near-infrared light that blueshifts by up to 250 meV as the gamma-InSe layer thins, attributed to quantum confinement.","lead":"This paper grows stacks of two different crystal forms of indium selenide on a wafer and shows they emit infrared light. The light color shifts by a controlled amount when the top layer is made thinner, pointing to a usable platform for infrared devices.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Thin-layer thicknesses are extrapolated from growth time; the one-parameter fit me = 0.09 m0 and the quantitative QSE curve in Fig. 3c are not robust to an early-growth-rate error.","rationale":"The paper has two claims: a qualitative one (near-IR emission from gamma-InSe/gamma-In2Se3 heterostructures shifts with gamma-InSe thickness) and a quantitative one (the shift is a type II quantum-size effect with me = 0.09 m0 and VBO = 0.15 eV). The qualitative claim is supported by the PL series and by phase/interface characterization (RHEED, Raman, STEM for one 12 nm structure), and I do not see an internal inconsistency that would overturn it. The quantitative claim, however, depends on an unmeasured thickness scale for the very samples that show the largest blueshift. Section 2.3 discloses the extrapolation, and no complementary measurement (e.g., STEM of the thinnest layers, in-situ growth-rate calibration, or independent effective-mass determination) is provided. This is the most load-bearing weak point because it directly enters the fitted parameter; strain relaxation and exciton binding are secondary uncertainties that would only change the model details, not the ordering of the data. The reader's CONDITIONAL verdict is appropriate: the condition is a direct thickness verification of the thinnest samples. I therefore leave the verdict unchanged.","tokens_in":11611,"tokens_out":7437,"duration_ms":70881,"concrete_test":"Measure the true gamma-InSe thickness in the two thinnest heterostructures (nominal 5 and 7 nm) by cross-sectional STEM on exactly those samples (or identically grown twins), as was done for the 12 nm structure in Fig. 2c. Compare the measured values with the growth-time extrapolated values, re-plot the emission energies from Fig. 3c against measured thicknesses, and refit me. If the 5 and 7 nm points shift by more than ~15% and the refitted mass changes by more than ~30%, the reported 0.09 m0 value and the quantitative QSE curve are not robust; if they remain on the same curve, the extrapolation concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claim is the 250 meV thickness-induced blueshift reproduced by a type II quantum-well calculation in Sec. 2.3. The load-bearing input to that calculation is the thickness axis of Fig. 3c, and the paper explicitly states: \"The layer thickness is determined from a comprehensive study involving scanning electron microscopy of the structure cross-section for layers thicker than 10 nm and the extrapolation of this quantity for the thinnest layers based on the growth time.\" The samples that exhibit the 250 meV shift are precisely the ones below 10 nm whose thickness was not directly measured. If the gamma-InSe growth rate is not constant during the first nanometers — due to nucleation, the 30 min growth interruption before switching the Se flux, or the gradual phase conversion seen in RHEED over the first ~3 nm — then the actual thicknesses of the nominal 5 and 7 nm layers could differ systematically from the plotted values. Because me = 0.09 m0 is the single fitted parameter and no error bars or raw data are given, even a 20–30% thickness error would materially change the inferred mass. The monotonic blueshift with growth time remains qualitatively consistent with confinement, so the existence of a size effect is less vulnerable; what the concern attacks is the quantitative model, the fitted mass, and the reliability of the inferred band offsets.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports MBE growth of indium selenide in different crystal phases (γ-InSe, γ-In2Se3, β-In2Se3) by controlling the In/Se flux ratio, and uses this control to fabricate γ-InSe/γ-In2Se3 phase heterostructures. The central experimental finding is an intense near-infrared emission near 1.10 eV from these heterostructures, attributed to a type-II interface transition. The authors show that this emission blueshifts by up to 250 meV when the γ-InSe layer thickness is reduced from tens of nanometers to about 5 nm, and they interpret this blueshift as a quantum size effect. A calculation with a one-parameter fit for the electron effective mass (me = 0.09 m0) and a valence band offset inferred from the thick-layer emission energy (VBO = 0.15 eV) is said to reproduce the thickness dependence. The paper also presents temperature- and excitation-power-dependent PL measurements supporting the type-II interface assignment.","tokens_in":11904,"tokens_out":4895,"duration_ms":43184,"significance":"If the quantitative interpretation is upheld, this is a valuable demonstration of MBE-grown, all-indium-selenide phase heterostructures with an interface emission tunable over a practically relevant near-infrared range. The qualitative claim—that the emission energy depends strongly on the γ-InSe layer thickness—is well supported by the measured thickness series, the absence of such a shift for thick layers, and the consistency of temperature and power dependence with a type-II transition. The paper's strengths include the multi-technique phase identification (Raman, PL, XRD, RHEED, STEM, AFM), the direct STEM evidence of an abrupt γ-InSe/γ-In2Se3 interface, and the explicit demonstration that the emission is not the ordinary γ-InSe band-edge transition. The main load-bearing weakness is the quantitative modeling: the thin-layer thicknesses are extrapolated from growth time, and the effective mass is fitted to the same data the model explains, so the numerical values (0.09 m0 and 0.15 eV) are less robust than the qualitative confinement effect.","major_comments":[{"comment":"The quantitative QSE calculation is anchored to a thickness axis that is explicitly extrapolated from growth time for the thinnest layers—precisely the layers (5 and 7 nm) that carry the 250 meV blueshift. Because the growth sequence includes a 30 min interruption and a change of Se flux, and RHEED shows that the phase conversion is not complete until after about 3 nm of growth, an early-growth-rate transient cannot be excluded. A systematic thickness error of 20–30% would materially change the fitted value me = 0.09 m0 and the inferred VBO = 0.15 eV. The authors should provide direct thickness measurements for at least the thinnest samples (e.g., STEM cross-sections) or a sensitivity analysis showing how the fitted parameters depend on assumed thickness errors, and they should include error bars on the data points in Fig. 3c.","section":"Section 2.3, Fig. 3c"},{"comment":"The valence band offset is obtained as VBO = Eg2 − Ed = 0.15 eV, treating the thick-layer emission energy as exactly equal to Eg2 − VBO. This neglects the exciton binding energy of the spatially separated electron-hole pair at the type-II interface, possible strain-related shifts (which the authors themselves invoke in Sec. 2.1 to explain the 1.23 eV emission of γ-InSe on GaAs), and any residual confinement at 30–65 nm. The inferred VBO should therefore be presented with an estimated uncertainty or as an approximate value, rather than as the definitively determined offset stated in the Conclusions.","section":"Section 2.3, VBO derivation"},{"comment":"The model calculation uses me = 0.09 m0 as the single fitted parameter and VBO derived from the same experimental data set. The paper does not provide raw data points, fit residuals, or a discussion of the parameter sensitivity. As presented, the statement that the blueshift is 'well reproduced by the calculations' is a one-parameter fit to five points whose abscissae carry unquantified uncertainty. Please show the individual PL spectra or extracted peak positions for all five thicknesses, report the fit residuals, and clarify which aspects of the fit are robust (e.g., the functional form of the confinement shift) versus parameter-dependent.","section":"Section 2.3, Fig. 3c model"}],"minor_comments":[{"comment":"The abstract contains the typo 'β-yIn2Se3'; this should read 'β-In2Se3'.","section":"Abstract"},{"comment":"The caption refers to 'indium telluride thin layers' but the paper is about indium selenide; this should be corrected.","section":"Fig. 1d caption"},{"comment":"The text and figure captions use 'fluence of 0.5 mW' and '30 mW', but milliwatts are units of power, not fluence. Please use consistent units (e.g., W/cm² or report the spot size and power) and refer to excitation power where appropriate.","section":"Section 2.3 and Fig. 4"},{"comment":"The Conclusions state that growth temperature variation 'in the range 350–400 °C' affects surface roughness, but Section 2.1 and Fig. 1a describe the range 300–450 °C; these statements should be reconciled.","section":"Conclusions vs. Section 2.1"},{"comment":"The paper mentions a 'comprehensive study involving scanning electron microscopy of the structure cross-section' for thickness calibration, but no SEM images or calibration curve are shown. Adding a representative SEM cross-section and the growth-rate calibration data would make the extrapolation procedure transparent.","section":"Section 2.3"},{"comment":"Reference [30] has an incomplete author name ('G. S˛'); the full author list should be provided.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The paper is within scope for a materials/applied physics journal and reports a useful experimental system. The qualitative observation—a 250 meV thickness-dependent blueshift of a near-infrared interface emission—is credible and important. My main concern is that the quantitative model, including the fitted effective mass and inferred band offset, rests on an extrapolated thickness axis for the thinnest samples and on a one-parameter fit with no uncertainty analysis. If the authors can provide direct thickness measurements or a convincing sensitivity analysis, I would be happy to reconsider for acceptance after revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a credible experimental paper that demonstrates something new—MBE-grown heterostructures between two different indium selenide crystal phases with a sharp interface and thickness-tunable near-infrared emission. The qualitative result (the 250 meV blueshift as the γ-InSe layer thins) is likely correct. The quantitative interpretation (effective mass 0.09 m0, VBO 0.15 eV) is shakier, because it depends on thicknesses that were extrapolated from growth time and a mass fitted to the same data.\n\nWhat the paper does well: the growth phase diagram is a useful contribution. The authors identify clean windows for γ-InSe, γ-In2Se3, and β-In2Se3 as a function of In/Se ratio and temperature, and they back the phase assignment with Raman, low-temperature PL, XRD, and STEM. The STEM image of the γ-InSe/γ-In2Se3 interface is convincing evidence of an abrupt lattice change. The PL story is also coherent: the 1.10 eV line appears only in the heterostructure, shifts monotonically with thickness, has an excitation-power dependence consistent with a type-II interface transition, and disappears at low thickness in a way consistent with the known direct-to-indirect crossover in thin InSe. That is a solid chain of evidence for an interfacial quantum-confined emission.\n\nThe soft spots are real but narrow. The thin-layer thicknesses—exactly the ones that carry the blueshift—are not measured directly; they are extrapolated from growth time using a rate calibrated on layers >10 nm. The RHEED data show the γ-In2Se3-to-γ-InSe conversion takes the first ~3 nm, so the early growth rate may not be constant. A 20–30% error in the thinnest-layer thicknesses would change the fitted mass noticeably. The paper also gives no error bars on the five PL peak positions, and the fit uses one free parameter (me) after fixing VBO from the thick-layer emission energy. None of this threatens the existence of the size effect—the blueshift with growth time is monotonic and qualitative—but it does mean the quoted mass and band offsets should be treated as rough estimates, not precise values.\n\nOne more minor point: the abstract says 'β-yIn2Se3' which looks like a typo, and the Conclusions say the temperature range is 350–400 °C while the phase diagram covers 300–450 °C. Small inconsistencies, easily fixed.\n\nWho this is for: experimentalists working on van der Waals epitaxy, 2D optoelectronics, and InSe-based devices. It is a good paper to referee seriously. I would send it to review, with the request that the authors add a direct thickness measurement or at least discuss the early-nucleation uncertainty, and include error bars on the PL energies. The qualitative advance is worth publishing; the quantitative model needs a caveat.","headline":"Credible MBE-grown InSe phase heterostructures with a tunable interface emission, but the quantitative quantum-size-effect numbers rest on extrapolated thicknesses and a fitted mass.","tokens_in":12483,"tokens_out":2601,"would_cite":true,"duration_ms":22431,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["81.15.Hi","78.55.-m","73.21.-b"],"model":"deepseek-v4-flash","headline":"A 250 meV blueshift ties indium selenide heterostructure emission to a type-II interface transition.","keywords":["indium selenide","crystal phase heterostructures","molecular beam epitaxy","quantum size effect","type-II band alignment","photoluminescence","near-infrared emitters","van der Waals semiconductors"],"falsifier":"Measure the actual thickness of the nominally 5 nm and 8 nm γ-InSe layers by cross-sectional scanning transmission electron microscopy; if the real thicknesses differ enough that the emission energies no longer follow the $0.09\\,m_0$ quantum-well curve, the quantum size effect interpretation and derived band offsets would need revision.","tokens_in":11459,"feed_emoji":"💡","tokens_out":5390,"duration_ms":42068,"temperature":0.7,"pith_summary":"The paper reports molecular beam epitaxy growth of indium selenide crystal phase heterostructures, showing that the In/Se flux ratio selects between γ-InSe, γ-In2Se3, and β-In2Se3 phases. It claims that the γ-InSe/γ-In2Se3 interface emits intensely in the near infrared at 1.10 eV, and that shrinking the γ-InSe layer below 20 nm blueshifts this emission by up to 250 meV. This thickness dependence is presented as evidence for a quantum size effect in a type-II staggered band alignment, with the optical transition occurring across the interface. A sympathetic reader would care because it demonstrates a route to large-area, lithographically patternable near-IR light sources and detectors built from a single materials family.","feed_headline":"InSe phase heterostructures tune near-IR light over 250 meV","feed_subtitle":"A stacked γ-InSe/γ-In2Se3 film emits at 1.10 eV; thinning the γ-InSe layer shifts it by a quarter of an electronvolt.","key_machinery":"The argument turns on the type-II (staggered) band alignment at the γ-InSe/γ-In2Se3 interface, which places the electron and hole in different layers so the optical transition energy is lower than either band gap. Quantum confinement in the conduction band of the γ-InSe slab raises the electron level as the layer thins, producing the blueshift; the paper models this as an electron in a potential well with $m_e = 0.09\\,m_0$, treating the valence band as unconfined because of the small valence band offset.","core_discovery":"The central claim is that the 1.10 eV photoluminescence from γ-InSe/γ-In2Se3 is an interface (type-II) transition, not a defect transition. The paper supports this by showing the emission blueshifts by about 250 meV when the γ-InSe layer thickness drops from 20 nm to 5 nm, which matches calculations of a confined electron in a quantum well with effective mass $m_e = 0.09\\,m_0$. From the large-thickness emission energy, the paper derives a valence band offset of 0.15 eV and a conduction band offset of 1.04 eV between γ-InSe and γ-In2Se3. It also reports that the emission vanishes below about 5 nm, consistent with a direct-to-indirect band gap crossover in thin γ-InSe.","pith_inferences":["The extracted effective mass of $0.09\\,m_0$ is likely the out-of-plane (c-axis) electron mass of γ-InSe; one could test it independently with magneto-optical measurements on the same heterostructures.","If thickness can be controlled to the monolayer level, the same quantum size effect could push emission across the near-infrared into the visible, extending the platform beyond the 250 meV range reported.","The phase-selection rule (In/Se flux ratio dominant) suggests that other III-VI compounds with competing polymorphs might support analogous phase heterostructures by the same MBE switching method.","A direct STEM thickness measurement on the thinnest samples would confirm whether the growth-rate extrapolation holds and would sharpen the fitted mass and offsets."],"forward_implications":["If the interpretation is right, the emission wavelength of γ-InSe/γ-In2Se3 heterostructures is set by γ-InSe thickness and can be designed from about 1.10 eV to 1.35 eV.","The derived valence band offset (0.15 eV) and conduction band offset (1.04 eV) give a concrete band diagram for further InSe device engineering.","The inability to grow β-In2Se3 on γ-In2Se3 shows that not all phase combinations are kinetically accessible, guiding future phase-heterostructure design.","The quenching of PL below 5 nm marks a thickness window for useful emitters and correlates with the direct-indirect crossover reported for exfoliated InSe.","The linear power dependence and fluence blueshift support an excitonic interface recombination channel."],"supporting_citations":[{"why":"Supplies the type-II band alignment diagram and the Raman mode assignments used to identify γ-InSe and γ-In2Se3.","marker":"[13]"},{"why":"Provides the reference Raman and X-ray data used to identify β-In2Se3 for low In/Se ratio.","marker":"[25]"},{"why":"Documents the direct-to-indirect band gap crossover in ultrathin γ-InSe, invoked to explain the PL quenching below 5 nm.","marker":"[28,29]"},{"why":"Reports indium droplet formation and nanowire growth at high In/Se flux ratios, used as the boundary condition for γ-InSe growth.","marker":"[23]"},{"why":"Establishes the ambiguity in X-ray identification among γ-InSe, β-InSe, and ε-InSe, motivating the STEM confirmation.","marker":"[12]"},{"why":"Provides the reported band gap tunability of InSe with layer count, context for the quantum size effect claim.","marker":"[11]"}],"fun_headline_variants":["Quantum size effect tunes InSe heterostructure emission by 250 meV","Type-II interface emission shifts 250 meV with InSe layer thickness","Controlling InSe layer thickness tunes near-IR emission from 1.10 eV","Quantum well effect blueshifts InSe heterostructure emission by 250 meV"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The thicknesses of the thinnest γ-InSe layers, where the largest blueshift is observed, are not measured directly but extrapolated from growth time using a rate calibrated on layers thicker than 10 nm.","fun_headline_variants_meta":{"raw":{"variants":["Quantum size effect tunes InSe heterostructure emission by 250 meV","Type-II interface emission shifts 250 meV with InSe layer thickness","Controlling InSe layer thickness tunes near-IR emission from 1.10 eV","Quantum well effect blueshifts InSe heterostructure emission by 250 meV"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000725,"raw_usage":{"total_tokens":3270,"prompt_tokens":988,"completion_tokens":2282,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":604,"completion_tokens_details":{"reasoning_tokens":2197}},"tokens_in":604,"tokens_out":2282,"duration_ms":15029,"temperature":1.0,"reasoning_tokens":2197,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T16:14:40.451199+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the actual thickness of the nominally 5 nm and 8 nm γ-InSe layers by cross-sectional scanning transmission electron microscopy; if the real thicknesses differ enough that the emission energies no longer follow the $0.09\\,m_0$ quantum-well curve, the quantum size effect interpretation and derived band offsets would need revision.","supporting_citations":[{"cited_title":"Balakrishnan, E","cited_arxiv_id":null,"evidence_quote":"Supplies the type-II band alignment diagram and the Raman mode assignments used to identify γ-InSe and γ-In2Se3."},{"cited_title":"Balakrishnan, C","cited_arxiv_id":null,"evidence_quote":"Provides the reference Raman and X-ray data used to identify β-In2Se3 for low In/Se ratio."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reports indium droplet formation and nanowire growth at high In/Se flux ratios, used as the boundary condition for γ-InSe growth."},{"cited_title":"Bergeron, L","cited_arxiv_id":null,"evidence_quote":"Establishes the ambiguity in X-ray identification among γ-InSe, β-InSe, and ε-InSe, motivating the STEM confirmation."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the reported band gap tunability of InSe with layer count, context for the quantum size effect claim."}],"review_version":2}