{"id":"5c4c1bf5-4ed4-4d31-80c7-173889d4383d","arxiv_id":"2506.21279","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"CuInP2S6 shows a 23% thickness-driven refractive index change and a birefringence of about 1.24 in the blue-ultraviolet range, making it a promising electro-optic platform.","lead":"Researchers measured the optical properties of the 2D ferrielectric material CuInP2S6 as they thinned a single flake from bulk down to 14 nm, finding an anomalously large thickness-driven change in refractive index (up to 23%) and a very strong difference between in-plane and out-of-plane refractive indices (about 1.24) in the blue-ultraviolet range. The results suggest thickness could serve as a tuning parameter for electro-optic devices.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central claims rest on ellipsometric extraction without independent thickness verification or confidence limits; a correlation artifact between fitted thickness and refractive index could produce the reported 23.2% variation and record birefringence.","rationale":"The reader's weakest assumption identifies exactly the load-bearing point: the extracted optical constants may be fitting artifacts rather than intrinsic material properties, especially given the fitted thickness and sequential Ar etching. My review agrees with this assessment. The paper does provide some mitigating evidence: a thoughtful multi-sample model, EMA roughness corrections, control measurements on LiNbO3 and SiO2, and Raman spectra showing no obvious structural damage. These are real strengths and deserve credit. However, the central numerical claims are quoted without error bars, and no independent thickness measurement is provided anywhere in the manuscript or supplementary material. For a 22 nm anisotropic film, the OOP refractive index is notoriously difficult to determine unambiguously, and a 23.2% change in n_OOP at a wavelength near the band edge is precisely the kind of signal that can be produced by small errors in the fixed or fitted parameters (thickness, roughness, EMA fraction). The birefringence claim of 1.24, claimed as the largest in the blue-UV, inherits the same uncertainty. Since the concern is concrete, technically specific, and addressable though not yet resolved, the appropriate verdict remains CONDITIONAL. I therefore recommend UNCHANGED relative to the reader's verdict. If the proposed AFM-constrained test or synthetic-data test fails, the claims would need to be withdrawn or substantially weakened, but on current evidence a conditional acceptance with a data-release and validation requirement is the honest position.","tokens_in":20487,"tokens_out":2298,"duration_ms":31000,"concrete_test":"Repeat the etch-and-measure cycle on a fresh CIPS flake, independently measuring thickness at each step with AFM, then re-extract n_IP and n_OOP with thickness fixed to the AFM values. If the resulting n(t) curves, δn ≈ 23.2%, and |Δn| ≈ 1.24 are reproduced within 10%, the concern is resolved. As a complementary check, generate synthetic ellipsometric data from a single-thickness, fixed-n CIPS stack at all measured thicknesses with the same noise level, run the identical CompleteEase model, and ask whether an apparent thickness-dependent δn as large as 23.2% emerges solely from thickness-index correlation.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claims of δn ≈ 23.2% (OOP, λ = 280 nm) and |n_OOP − n_IP| ≈ 1.24 (λ = 339.5 nm) depend entirely on wavelength-dependent optical constants n_IP and n_OOP extracted from spectroscopic ellipsometry at each thickness. In an anisotropic layered film on SiO2/Si, thickness and refractive index are strongly correlated in the Fresnel model; for films near 22 nm, sensitivity to n_OOP is especially limited. The Methods describe a multi-step model (Cauchy-Urbach, spline, Cody-Lorentz/Gaussian, EMA roughness, thickness non-uniformity) and a multi-sample analysis, but no confidence intervals, parameter-correlation matrices, or fit-quality landscapes are reported. Thickness is fitted, not measured independently: no AFM step-height data are shown for any etching step. The argon-beam etch sequence itself could modify the surface—Cu depletion, disorder, or a thin altered layer—and the Raman check presented in the paper samples bulk phonon modes, which are not surface-sensitive and cannot exclude such a layer. The controls on LiNbO3 and SiO2 are useful, but they use the same fitting pipeline, so a systematic model bias would affect them too. Therefore the claims that the thickness dependence is intrinsic and that the birefringence is a record are not yet established; the numerical results could be partly or wholly artifacts of the extraction model.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript reports variable-angle spectroscopic ellipsometry (VASE) measurements on mechanically exfoliated CuInP2S6 flakes that are progressively thinned by argon-beam etching. The authors extract in-plane and out-of-plane optical constants from ~615 nm down to ~14 nm and report two central claims: (i) an 'anomalous' thickness-dependent refractive-index change, largest in the out-of-plane direction with δn ~23.2% at λ = 280 nm over t ∈ [22, 170) nm, and (ii) a giant birefringence |n_OOP − n_IP| ~1.24 at t ~22 nm and λ = 339.5 nm, claimed to be the largest intrinsic birefringence in the blue-ultraviolet regime of any known material. The thickness dependence is attributed to changes in the Cu(I) ferroelectric polarization and ionic mobility. Control VASE measurements on LiNbO3 and SiO2 and thickness-dependent Raman spectroscopy are included to support the interpretation.","tokens_in":20875,"tokens_out":5754,"duration_ms":63873,"significance":"If the extracted optical constants are intrinsic material properties, the results are significant: they introduce thickness as a new tuning parameter for the refractive index of a 2D ferroelectric and report a record birefringence in the blue-ultraviolet range. The experimental effort is substantial and includes multi-angle ellipsometry, a multi-sample fitting approach, control measurements on LiNbO3 and SiO2, and careful low-power Raman measurements. However, the central quantitative claims rest entirely on a heavily parameterized ellipsometric model without independent thickness verification or uncertainty analysis, and the sequential argon-beam etching introduces a plausible route to surface modification that the Raman check may not detect. The significance is therefore conditional on resolving these extraction and surface-integrity concerns.","major_comments":[{"comment":"The central quantitative claims (Table S1: δn = 23.18% for OOP at λ = 280 nm in the 22–170 nm regime; Fig. 5b: |Δn| ≈ 1.24 at t ≈ 22 nm) are extracted from a spectroscopic-ellipsometry model with many free parameters: Cody-Lorentz and Gaussian oscillator parameters for both axes, thickness at every etch step, EMA roughness fraction, and thickness non-uniformity. No confidence intervals, parameter-correlation matrices, or fit-quality landscapes are reported. For a ~22 nm anisotropic film on SiO2/Si, thickness and refractive index are strongly correlated in Fresnel fitting, so the reported thickness dependence—and the birefringence derived from the same fit—could be partly or wholly a numerical artifact of the model. Please provide (i) independent thickness measurements (e.g., AFM step heights) at least at the endpoints and ideally at each etch step, (ii) confidence intervals and correlation analysis for the extracted n and t, and (iii) a quantitative comparison (MSE, AIC/BIC) against a model with thickness-independent optical constants.","section":"Methods – Modelling; Table S1"},{"comment":"The conclusion that argon-beam etching does not cause significant damage to CIPS's crystal structure is based on the absence of new Raman peaks in the 100–400 cm−1 range. Raman with 488 nm excitation probes the bulk phonon spectrum and is not surface-sensitive; a thin altered layer (Cu depletion, cation disorder, or an amorphous/partially oxidized surface layer) could easily escape detection while having a large effect on the ellipsometric spectra, especially for the thinnest flakes where the surface-to-volume ratio is highest. This is a direct alternative explanation for the thickness-dependent optical constants. Please add surface-sensitive characterization (XPS or AES depth profiling, cross-sectional TEM, or at least a surface-layer term in the optical model) or repeat the measurement on several flakes with different initial thicknesses to verify that the trend is independent of etch history.","section":"Argon beam etching; Fig. 1c"},{"comment":"The statement that alternative models 'did not yield better fitting results. Therefore, it implies that, in fact, CIPS' optical properties actually change with thickness' overreaches: failure of a few alternative models to improve the fit does not establish that the thickness dependence is intrinsic, particularly when thickness itself is a fitted parameter. To make this point load-bearing, the fit statistics for the competing models (including number of parameters and information criteria) should be reported, and a model with thickness-independent optical constants but an independently measured thickness should be shown to be inconsistent with the data.","section":"Methods – Modelling (final paragraph)"}],"minor_comments":[{"comment":"The symbol δn is defined in Table S1 as a relative change, (nmax − nmin)/nmax × 100, so the abstract's 'δn ∼ 23.2%' should be described as a relative change in refractive index rather than an absolute change, to avoid confusion with the birefringence Δn used elsewhere.","section":"Abstract; Table S1"},{"comment":"Typo: 'analysied' should be 'analysed'.","section":"Methods – Modelling"},{"comment":"Typos: 'sheering angle' should be 'shearing angle'; 'accelaration beam voltage' should be 'acceleration beam voltage'; and 'thickness ranget' should be 'thickness range t'.","section":"Methods – Argon beam etching"},{"comment":"The page range '1309–11310' appears to be a typo, likely '1309–1310'.","section":"Reference [14]"},{"comment":"The statement that data are available from the corresponding authors upon reasonable request is acceptable, but for record claims it would be helpful to deposit the raw ellipsometric spectra and fitting code in a public repository.","section":"Data and materials availability"},{"comment":"The Raman spectra are normalized to the ~375 cm−1 peak; please state whether normalization is applied before or after averaging and whether spectra are offset for display, as the plotted intensities could be misread.","section":"Fig. 1c"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is well within the scope of the journal and the experimental effort is substantial. My main concerns are the ellipsometric extraction and etch-induced surface modification; if the authors can supply independent thickness verification, uncertainty analysis, and surface-sensitive characterization, the claims would be convincing. I do not see issues with novelty disclosure or citation practice."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: the paper gives a new thickness series of optical constants for CuInP2S6, and the claimed giant birefringence is worth attention. But the headline numbers are only as good as the ellipsometric fit, and the paper never shows the fit uncertainty.\n\nWhat's genuinely new is the systematic thickness dependence of n_IP and n_OOP from ~600 nm down to 14 nm. Earlier work measured bulk birefringence; nobody had mapped the evolution through the ferrielectric-to-paraelectric transition region. The experimental design is thoughtful: three angles, multi-sample analysis, roughness and thickness-nonuniformity corrections, and control measurements on LiNbO3 and SiO2 through the same pipeline. The Raman frequency shifts across the transition are a useful independent probe that something structural happens around 60-90 nm.\n\nThe soft spot is exactly where the stress-test lands. In a thin anisotropic film on SiO2/Si, fitted thickness and refractive index trade off against each other, especially for the out-of-plane component near 22 nm. No confidence intervals, no parameter correlation matrix, and no independent thickness measurement (AFM step heights, for example) are reported. The sequential argon-beam etch is another confounder; Raman is not surface-sensitive enough to rule out a thin altered layer. These concerns don't invalidate the paper; they shift it from 'established' to 'promising'. The 23.2% change and |Δn| ~1.24 may well be real, but the current evidence doesn't prove it.\n\nThe interpretation that Cu(I) polarization and ion mobility drive the thickness dependence is reasonable and honestly flagged as incomplete. The comparison against other birefringent materials is useful, though 'largest of any known material' is a strong claim from a limited set.\n\nFor a reader working in 2D ferroelectrics or nanophotonics, this is worth reading as a careful experimental study with a potentially important result. I would accept it for peer review; a serious referee should ask for error bars, an independent thickness measurement, and ideally surface-sensitive characterization before publication. Once those are in, the record claim may hold, but not yet.\n\nSo: worth engaging, with clear revision requirements.","headline":"Thickness-dependent ellipsometry on CuInP2S6 shows a promising but unproven giant birefringence and anomalous index change; the measurement needs error bars and independent thickness checks before the record claims can be trusted.","tokens_in":21324,"tokens_out":2539,"would_cite":false,"duration_ms":28691,"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":"The paper claims that thinning CuInP2S6 from bulk to about 22 nm changes its refractive index by up to 23.2% and yields a blue-ultraviolet birefringence of about 1.24, the largest of any known material in that range.","keywords":["CuInP2S6","ferrielectric","2D ferroelectric","refractive index","birefringence","spectroscopic ellipsometry","thickness dependence","electro-optics"],"falsifier":"Measure the same CuInP2S6 flakes with atomic force microscopy after each etch, then refit the ellipsometry data with thickness fixed to the AFM value; if the 23.2% change in refractive index collapses when thickness is fixed, the anomaly is a fitting artifact rather than a material property. Alternatively, compare etched flakes with unetched exfoliated flakes of matched thickness: if the unetched flakes do not reproduce the anomalous trend, the argon-beam process is responsible.","tokens_in":20281,"feed_emoji":"🔬","tokens_out":11045,"duration_ms":103358,"temperature":0.7,"pith_summary":"This paper reports that the refractive index of the layered ferrielectric crystal CuInP2S6 changes by as much as 23.2% as the crystal is thinned from bulk to about 22 nm at room temperature, an effect far larger than the usual finite-size corrections seen in ordinary dielectrics. It also reports a giant intrinsic birefringence, $\\Delta n = |n_{\\mathrm{OOP}} - n_{\\mathrm{IP}}| \\approx 1.24$ at 339.5 nm for a 22 nm flake, which the authors state is the largest of any known material in the blue-ultraviolet regime. The paper attributes the thickness-driven optical changes to changes in the Cu(I) polarization component, coupled to the mobility and site occupancy of Cu(I) ions. If correct, flake thickness becomes a practical tuning parameter for optical response in a broad spectral range, beyond the ultra-thin limit.","feed_headline":"Thinning a 2D ferrielectric crystal shifts refractive index by 23%","feed_subtitle":"The same flakes show the largest intrinsic blue-UV birefringence yet reported, adding thickness as an optical tuning knob.","key_machinery":"The central object is the layered van der Waals ferrielectric CuInP2S6, in which Cu(I) cations occupy multiple sites and contribute both the ferroelectric polarization and ionic conductivity. The load-bearing observable is the anisotropic pair of optical constants $n_{\\mathrm{IP}}$ and $n_{\\mathrm{OOP}}$ (with the corresponding extinction coefficients), extracted from variable-angle spectroscopic ellipsometry at each thickness. The headline quantities are the birefringence $\\Delta n = n_{\\mathrm{OOP}} - n_{\\mathrm{IP}}$ and the relative thickness-induced change $\\delta n = (n_{\\mathrm{max}} - n_{\\mathrm{min}})/n_{\\mathrm{max}}$. The proposed mechanism coupling thickness to optics is a change in the Cu(I) polarization contribution, inferred from Raman mode shifts and broadening, especially the Cu$^+$ mode near 311 cm$^{-1}$, accompanying the thickness-driven structural transition from monoclinic Cc to trigonal P31c.","core_discovery":"The authors find that CuInP2S6, a layered van der Waals ferrielectric, shows an anomalous thickness-dependent optical response in the range $t \\in [22, 170)$ nm at room temperature, with the out-of-plane refractive index changing by up to $\\delta n \\sim 23.2\\%$ at $\\lambda = 280.0$ nm. They further find that the intrinsic birefringence $|n_{\\mathrm{OOP}} - n_{\\mathrm{IP}}|$ is large over the entire measured range and reaches about 1.24 at $t \\sim 22$ nm and $\\lambda = 339.5$ nm, which they state is the largest of any known material in the blue-ultraviolet regime. The paper connects these optical changes to the Cu(I) ferrielectric polarization contribution, whose ionic mobility and site occupancy change with thickness, and it supports this connection with Raman measurements showing thickness-driven shifts in Cu$^+$, P-P, S-P-P, and S-P-S modes. The authors propose that the effect is generalizable to other ferroelectrics such as LiNbO3 and Bi2FeCrO6, though weaker than in CuInP2S6.","pith_inferences":["If the effect is intrinsic, etching a single flake into terraces of different thickness would create a lateral refractive-index gradient, enabling graded-index optics or waveguides without changing material composition.","A direct testable extension is to drive Cu(I) ions with an in-plane electric field at fixed thickness and look for a reversible shift in $n_{\\mathrm{OOP}}$; the paper cites the sister compound CuCrP2S6 showing such tuning but does not demonstrate it for CuInP2S6.","The wavelength crossing points in $n(t)$ and $\\kappa(t)$ imply that CuInP2S6 could be designed to be index-matched or birefringence-sign-switching at chosen wavelengths, an opportunity the paper does not explore.","A natural next check is to measure flake thickness independently, for example by atomic force microscopy, and refit the ellipsometry data with that thickness fixed; this would test how much of the reported thickness dependence is model-driven."],"forward_implications":["Thickness becomes a control knob for the optical constants of CuInP2S6 across the 22-170 nm range, not only in the few-nanometre limit.","A 22 nm CuInP2S6 flake offers an intrinsic blue-UV birefringence of about 1.24, larger than quartz, calcite, rutile, hBN, and transition-metal dichalcogenides, without metasurface structuring.","Because the optical changes track the Cu(I) polarization and ionic subsystem, electrical or ionic control of Cu(I) occupancy is a plausible route to electro-optic modulation in CuInP2S6.","The same thickness-dependent refractive-index behaviour appears, more weakly, in LiNbO3 and Bi2FeCrO6, suggesting a general ferroelectric phenomenon rather than a CuInP2S6-specific accident.","The largest thickness-induced index change occurs near 280 nm, a wavelength at which CuInP2S6 already functions as a UV photodetector, pointing to combined detection and polarization-control functionality."],"supporting_citations":[{"why":"Establishes the thickness-driven structural phase transition and loss of in-plane polarization below the critical thickness, which the paper uses to locate the t ~ 90 nm optical anomaly.","marker":"[45]"},{"why":"Defines the ferrielectric ordering and the three Cu(I) sites that couple polarization to ionic mobility.","marker":"[23]"},{"why":"Documents the ionic conductivity of CuInP2S6 and CuCrP2S6, underpinning the ferro-ionic mechanism.","marker":"[42]"},{"why":"Shows that in-plane Cu(I) migration induces out-of-plane polarization switching, supporting the proposed link between ion motion and optical constants.","marker":"[43]"},{"why":"Provides the bulk electro-optic coefficient and small in-plane anisotropy of CuInP2S6 that motivate the electro-optic application.","marker":"[61]"},{"why":"Demonstrates Cu(I)-motion-tuned refractive index in the sister compound CuCrP2S6, the closest prior analogue of the proposed mechanism.","marker":"[93]"},{"why":"Supplies the anisotropic ellipsometry modelling approach used for van der Waals materials and comparison birefringence data for transition-metal dichalcogenides.","marker":"[77]"},{"why":"Gives the thickness-independent refractive index of SiO2 films used as a control showing ordinary dielectric behaviour.","marker":"[56]"},{"why":"Reports thickness-dependent optical properties in ferroelectric Bi2FeCrO6, the main literature comparison for ferroelectric index modulation.","marker":"[59]"},{"why":"Identifies 280 nm as a key ultraviolet photodetection wavelength for CuInP2S6, where the largest index change is observed.","marker":"[50]"}],"fun_headline_variants":["Thickness tunes 2D ferrielectric's index by 23%, birefringence hits 1.24","Largest blue-UV birefringence: 1.24 in 2D ferrielectric","2D ferrielectric's refractive index swings 23% with thickness","CuInP2S6: 23% index shift, 1.24 blue-UV birefringence"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the thickness-dependent refractive indices extracted from spectroscopic ellipsometry are intrinsic properties of CuInP2S6, not artifacts of the fitting model's trade-off between thickness and index or of argon-beam etching altering the surface.","fun_headline_variants_meta":{"raw":{"variants":["Thickness tunes 2D ferrielectric's index by 23%, birefringence hits 1.24","Largest blue-UV birefringence: 1.24 in 2D ferrielectric","2D ferrielectric's refractive index swings 23% with thickness","CuInP2S6: 23% index shift, 1.24 blue-UV birefringence"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001214,"raw_usage":{"total_tokens":5063,"prompt_tokens":1077,"completion_tokens":3986,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":693,"completion_tokens_details":{"reasoning_tokens":3877}},"tokens_in":693,"tokens_out":3986,"duration_ms":24640,"temperature":1.0,"reasoning_tokens":3877,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T22:26:59.302090+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the same CuInP2S6 flakes with atomic force microscopy after each etch, then refit the ellipsometry data with thickness fixed to the AFM value; if the 23.2% change in refractive index collapses when thickness is fixed, the anomaly is a fitting artifact rather than a material property. Alternatively, compare etched flakes with unetched exfoliated flakes of matched thickness: if the unetched flakes do not reproduce the anomalous trend, the argon-beam process is responsible.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the ferrielectric ordering and the three Cu(I) sites that couple polarization to ionic mobility."},{"cited_title":"Journal of Materials Chemistry C 8, 6966–6971 (2020) https://doi.org/10.1039/d0tc01620a","cited_arxiv_id":null,"evidence_quote":"Shows that in-plane Cu(I) migration induces out-of-plane polarization switching, supporting the proposed link between ion motion and optical constants."},{"cited_title":"Journal of Physics D: Applied Physics 43 (2010) https://doi","cited_arxiv_id":null,"evidence_quote":"Gives the thickness-independent refractive index of SiO2 films used as a control showing ordinary dielectric behaviour."},{"cited_title":"Vacuum 216 (2023) https://doi","cited_arxiv_id":null,"evidence_quote":"Reports thickness-dependent optical properties in ferroelectric Bi2FeCrO6, the main literature comparison for ferroelectric index modulation."},{"cited_title":"Applied Physics Letters 117, 131102 (2020) https://doi.org/10.1063/5.0022097","cited_arxiv_id":null,"evidence_quote":"Identifies 280 nm as a key ultraviolet photodetection wavelength for CuInP2S6, where the largest index change is observed."}],"review_version":1}