{"id":"0f5e4ea0-9b61-4002-913b-de6157083747","arxiv_id":"2506.09578","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"A high-sensitivity polarimeter maps birefringence in silicon wafers, finding position-dependent values near 10^-7 in (100) samples and -1.5x10^-6 for the (110) orientation at 1550 nm.","lead":"This paper describes a sensitive optical polarimeter that maps birefringence across silicon wafers and reports values around 10^-7 in commercial (100) samples with strong spatial variation. It matters for Einstein Telescope because crystalline silicon is a candidate substrate material and polarization non-uniformity can degrade interferometer sensitivity.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 'intrinsic' (110) birefringence claim is not secured against mount-induced stress: the channeling strip's stress state is uncharacterized, and self-weight/clamping can shift Δn by more than the quoted uncertainty.","rationale":"The reader's weakest assumption is the single-uniform-etalon model, which is a real concern for the absolute scale of all maps, but it is partially mitigated by the empirical Z measurement, internal consistency checks, and the fact that the reported (100) values are robust to the ±20% correction. The (110) stress issue is more load-bearing: it threatens a headline quantitative result by an unknown factor that could be much larger than the etalon correction, and the paper provides no evidence to rule it out. The provenance of the sample (channeling strip) raises the prior. I therefore identify this as the single most load-bearing concern, while agreeing with the reader that the overall verdict should remain conditional pending additional characterization. The proposed test is straightforward and would settle the issue.","tokens_in":13551,"tokens_out":16952,"duration_ms":179213,"concrete_test":"Measure the same (110) strip under at least two mountings: (i) the current holder and (ii) a three-point kinematic support with the beam centered at the neutral plane, and compare the position-resolved Δn maps. If the average shifts by more than ±0.15e-6, the intrinsic attribution is not established. Also record a full map along the 55-mm length: bending-dominated stress produces a systematic gradient or face-dependent sign change, whereas intrinsic spatial dispersion should be uniform. Complement with finite-element photoelastic modeling of self-weight and clamping, benchmarked against a controlled dead-weight load.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 4.2 reports Δn(110)=(-1.50±0.15)e-6 at 1550 nm as the intrinsic spatial-dispersion birefringence of silicon. The sample, however, is a 2×4.1×55 mm strip manufactured for channeling (Ref. [27]), and the paper gives no mounting, clamping, or stress characterization for it, unlike the (100) wafers for which 'without stress' holders are explicitly described. Silicon's stress-optic coefficient is C_so≈2e-11 Pa^-1 (Ref. [34]); a modest 1 MPa bending/clamping stress yields Δn≈2e-5, an order of magnitude larger than the reported value. Even the strip's own weight, if simply supported, produces stresses of order 10^4-10^5 Pa at the sampled beam positions, i.e., Δn≈1e-7-1e-6, comparable to the 10% error bar. The gravity estimate in Eq. (32) is computed only for a 96.5-mm wafer and is not applied to the strips. Because the Discussion uses this single number to set ET alignment tolerances (3 mrad for Δn<1e-8), an uncharacterized stress contribution undermines the central 'intrinsic' claim.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper describes a high-sensitivity polarimeter, based on co-rotating half-wave plates and heterodyne detection, for two-dimensional mapping of optical birefringence, and applies it to crystalline silicon samples for the Einstein Telescope. A Jones-matrix etalon model is used to invert measured ellipticity and rotation into birefringence Δn, with calibration by the Cotton-Mouton effect in air. The authors report position-dependent birefringence of order 10^-7 for (100)-oriented 1-mm silicon wafers at 1064 nm, with average values in Figures 8-10, and an intrinsic (110) birefringence Δn(110)=(-1.50±0.15)×10^-6 at 1550 nm with the fast axis along [100]. Implications for ET substrate specifications, including a 3-mrad alignment tolerance, are discussed.","tokens_in":13834,"tokens_out":4975,"duration_ms":50262,"significance":"If correct, the apparatus achieves an optical-path-difference sensitivity around 10^-12 m and provides spatially resolved birefringence data for commercially available silicon, a candidate ET substrate material. Strengths include an independent Cotton-Mouton calibration in air, use of literature values for n and k rather than free parameters for the (100) samples, and a zoom repeatability check in Figure 9 that supports the reality of the observed spatial patterns. The (110) result, however, is the highest-impact quantitative claim and is also the least supported: the mechanical stress state of the strip is not characterized, and the error budget for the inversion is not given. These gaps currently prevent the paper from fully supporting its central specifications.","major_comments":[{"comment":"The central claim that Δn(110)=(-1.50±0.15)×10^-6 is the intrinsic spatial-dispersion birefringence of silicon is not supported by the evidence given for mechanical stress. The (110) sample is a 2×4.1×55 mm strip manufactured for channeling, and unlike the (100) samples described in Section 3, no mounting, clamping, or stress state is described. With a stress-optic coefficient C_so≈2×10^-11 Pa^-1 (Ref. [34]), a modest clamping stress of ~1 MPa would produce Δn≈2×10^-5, while the strip's own weight under simply supported conditions gives stresses of order 10^4-10^5 Pa, corresponding to Δn≈10^-7-10^-6, comparable to the quoted uncertainty. Equation (32) estimates gravity effects only for the 96.5-mm wafer, not for this strip. The authors should either characterize the stress state (for example, by finite-element modeling or by measuring the birefringence under different support conditions) or re-state the result as an apparent birefringence that includes mounting-induced contributions.","section":"§4.2, Eq. (31)"},{"comment":"The paper does not provide a propagated uncertainty budget for the inversion from measured Ψ and Φ to Δn. The etalon parameter is measured with the same apparatus as Z=0.128±0.018, i.e., about 14% uncertainty, and Eqs. (15)-(18) are nonlinear in Z. The text itself notes that Ψ can differ from ψ by about ±20% at 1064 nm and ±50% at 1550 nm (Section 2.2), so the Z uncertainty should dominate the final Δn uncertainty. Yet the average values quoted for Figures 8-10 and the value Δn(110)=(-1.50±0.15)×10^-6 in Eq. (31) are presented without an error budget or a covariance analysis of Z, R, f, and the measured Fourier amplitudes. Please provide a per-point uncertainty propagation and state whether the claimed ~10^-7 level and the 10% error bar on Δn(110) survive when the Z uncertainty is included.","section":"§2.2 and §4.1, Eqs. (15)-(19), (30)"},{"comment":"The birefringence maps show no per-point error bars or confidence intervals. The claim that birefringence is position dependent in both magnitude and axis orientation requires that the observed spatial variation exceed the measurement noise at each point. The zoom repeatability check in Figure 9 supports reproducibility of the pattern, but it does not quantify the per-vector uncertainty. Because the reported values span roughly 0.5-3.0×10^-7 with a step of 2 mm, a statement of typical per-point uncertainty—from lock-in noise, spurious-ellipticity subtraction, and Z propagation—is needed to establish that the map features are not artifacts of the analysis.","section":"Figs. 8-10"},{"comment":"For the (110) sample, the paper states 'In this case we measured Z=0.257 corresponding to f=Z/R=0.84', but no uncertainty or measurement method for this Z is given, and the interference fraction f is not independently verified. The extraction of Δn(110) therefore relies on an assumed value of f that can bias the result if the beam is not fully interfering or if surface roughness varies across the strip. Similarly, the assumption ε=0 (no dichroism) is asserted in Section 3.1 without a direct experimental test. A sensitivity analysis of the extracted Δn to f and ε, or an independent measurement of these parameters, is required to support the quantitative result in Eq. (31).","section":"§4.2"}],"minor_comments":[{"comment":"The displayed equation contains an apparent typo: the text reads 'P'(t) = P_out [...] = ≈ P_0 [...]'. Please correct the equality/approximation chain, and check the dimensions of the terms.","section":"§3.1, Eq. (24)"},{"comment":"The caption contains a typo: 'Elliticity' should be 'Ellipticity'.","section":"Fig. 3 caption"},{"comment":"The table legend is partly in Italian ('Legenda'); please use English throughout, and ensure that all column headers are explicit.","section":"Table 2"},{"comment":"The sign convention leading to negative Δn and the fast axis along [100] is introduced without derivation; a sentence connecting the measured ellipticity/rotation phases to the sign of Δn would help the reader verify the result.","section":"§4.2"},{"comment":"The claimed ultimate sensitivity S_ΔD≲10^-12 m is derived indirectly as |ψ_spurious|/5; it would be useful to state the assumed noise bandwidth or integration time associated with this sensitivity.","section":"§3.2"}],"recommendation":"major_revision","confidential_remarks":"The paper fits the journal's scope and the apparatus description is valuable. The main risk is the (110) 'intrinsic' claim: the strip's stress state is completely uncharacterized, and the quoted uncertainty is too small to absorb plausible mounting and self-weight stresses. This is fixable within a revision by adding stress modeling or re-labeling the result, but it is currently load-bearing for the ET alignment-tolerance discussion. The (100) maps would also benefit from an explicit per-point error budget."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"I read this with the stress-test note next to it, and the note is right. The apparatus paper is solid, and the maps are believable—the zoom repeatability in Fig. 9 is a good check. But the paper's marquee number, the sign-resolved (110) birefringence of –1.5e-6 at 1550 nm, is presented as 'intrinsic' without characterizing the stress state of that channeling strip. A 1 MPa stress would produce Δn ~2e-5 in silicon, and even self-weight of that 2×4.1×55 mm strip gives stresses at the 10^4–10^5 Pa level, i.e., Δn of order 1e-7 to 1e-6—comparable to the quoted ±0.15e-6 error bar. The paper's own gravity estimate for the 96.5-mm wafer (3.5e-8) is not applied to the strip, and no clamping details are given for it, unlike the (100) wafers. So the 'intrinsic' label is not earned yet.\n\nWhat is genuinely new: adapting the VMB/PVLAS modulation polarimetry scheme to 2D mapping is a real step; the sign and fast-axis determination for (110) is new relative to the prior table; and the (100) maps with magnitude and axis orientation are a useful dataset for ET substrate spec work. The calibration via the Cotton-Mouton effect in air is independent, and the etalon inversion is carefully derived. The f=1 assumption for the thin wafers is plausible but not tested; the measured Z=0.128±0.018 for (100) carries 14% uncertainty that propagates into Δn. And the maps lack per-point error bars, which makes it hard to know whether the spatial variation is significant beyond the intrinsic noise floor.\n\nThe paper is honest about the gravity issue in the Discussion, but it stops short of correcting for it, and the (110) section doesn't carry the same caveat. That is the soft spot a referee should push on.\n\nOverall: this is a serious experimental contribution that deserves a proper review, not a desk reject. The apparatus works and the maps are credible. But the (110) number should be reframed as 'birefringence of a channeling strip under unknown stress' until stress characterization is added. I'd send it to review and ask for error bars, a stress check on the (110) sample, and a softened claim.","headline":"Solid apparatus paper with believable maps, but the 'intrinsic' (110) value is not stress-secured and needs a caveat.","tokens_in":14423,"tokens_out":4699,"would_cite":true,"duration_ms":47967,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["42.25.Lc","95.55.Ym"],"model":"deepseek-v4-flash","headline":"A heterodyne polarimeter can measure optical-path differences below $10^{-12}$ m in silicon, and maps of (100) wafers show position-dependent birefringence of order $10^{-7}$.","keywords":["birefringence","crystalline silicon","polarimetry","gravitational-wave interferometers","etalon interference","optical path difference sensitivity","silicon (110) intrinsic birefringence","birefringence mapping"],"falsifier":"Map the same (100) wafer at both 1064 and 1550 nm, where $Z$ and the etalon finesse differ by more than a factor of two; if the per-pass $\\Delta n$ values extracted from Equations 15-19 disagree by more than the stated uncertainties, the single-etalon inversion with $f=1$ is not correct.","tokens_in":13348,"feed_emoji":"🔬","tokens_out":15742,"duration_ms":135887,"temperature":0.7,"pith_summary":"The paper presents a polarimeter that maps the birefringence of transparent substrates with an optical-path-difference sensitivity below $10^{-12}$ m, and applies it to crystalline silicon, the leading candidate mirror material for next-generation gravitational-wave detectors. On 1-mm-thick (100)-oriented wafers at 1064 nm, the instrument finds average birefringence values of $1.32\\times10^{-7}$, $1.24\\times10^{-7}$, $1.19\\times10^{-7}$, and $7.64\\times10^{-8}$, with both the magnitude and the axis direction varying from point to point. On a (110) strip at 1550 nm, it measures an intrinsic birefringence of $(-1.50\\pm0.15)\\times10^{-6}$ with the fast axis along $[100]$, somewhat smaller than earlier determinations. The authors conclude that average values near $10^{-7}$ are likely too high for the low-frequency arm of a next-generation observatory, and that spatial uniformity of birefringence, not just its mean, must enter the substrate specification.","feed_headline":"Silicon wafers hide position-dependent birefringence of order 10^-7","feed_subtitle":"A polarimeter resolves optical path differences below 10^-12 m, sharpening substrate specs for next-gen detectors.","key_machinery":"The load-bearing object is the Jones-matrix description of the sample as a birefringent Fabry-Perot etalon (Equation 12). Interference among the multiple internal reflections converts part of the per-pass ellipticity $\\psi$ into a measurable rotation $\\Phi$; measuring both $\\Psi$ and $\\Phi$ gives the etalon phase $\\delta$ through $\\Phi/(i\\Psi) = 2Z\\sin\\delta/(1-Z^2)$, and Equation 19 recovers $\\psi$ from the transmitted power, so the birefringence $\\Delta n$ follows from $\\psi = (\\pi/\\lambda)\\Delta D$. In the instrument, ellipticity and rotation are separated by frequency: the rotating half-wave plates move the sample signal to $4\\nu_w$, the photoelastic modulator oscillates ellipticity at $\\nu_m$, and the Faraday cell oscillates rotation at $\\nu_F$, so lock-in demodulation gives $\\Psi_0$ and $\\Phi_0$ independently.","core_discovery":"The paper's core discovery is that crystalline silicon, even when cut on a nominally isotropic (100) plane, is measurably birefringent at the $10^{-7}$ level, with a pattern that varies across the wafer in both magnitude and axis orientation; the same measurement on a (110) face gives an intrinsic birefringence of $(-1.50\\pm0.15)\\times10^{-6}$ at 1550 nm whose fast axis coincides with $[100]$. The paper further establishes that the polarimeter can separate the ellipticity from the rotation produced by the sample, and that the ratio of the two determines the etalon phase, allowing the per-pass birefringence to be extracted even though interference inside the 1-mm wafer modifies the raw signal by up to tens of percent. On this basis it argues that birefringence of order $10^{-7}$ is present in commercially produced silicon and cannot be ignored when specifying substrates for low-frequency gravitational-wave interferometry.","pith_inferences":["A natural extension of the single-etalon model would be to include birefringence that varies along the beam; this could be probed by comparing maps taken with different beam diameters, since a wider beam averages over more of the transverse stress field.","The gravity-induced birefringence estimate in the paper ($\\sim 3.5\\times10^{-8}$ for a 96.5-mm wafer) implies that support geometry affects the apparent birefringence of large thin samples; future measurements on mounted test masses should correct for this or suspend the sample at its midline.","The polarimeter's ability to separate ellipticity from rotation could be turned into a dichroism microscope: any residual rotation signal after the etalon correction would be direct evidence of anisotropic absorption.","The measured (110) intrinsic birefringence could be used to set a systematic-error floor for any future silicon-based interferometer whose beam path is not perfectly aligned to the $[100]$ axis."],"forward_implications":["If these measurements are representative, next-generation interferometer substrates will need a birefringence budget that includes spatial gradients, not just a mean value, because direction-varying birefringence cannot be suppressed by aligning the input polarization.","For (110)-oriented silicon, keeping birefringence below $10^{-8}$ requires aligning the crystal's $[100]$ direction to the beam within about 3 mrad, turning crystal-axis orientation into a tolerance requirement.","The same instrument can map other substrate and coating materials with sensitivity below $10^{-12}$ m optical path, since the method is material-independent once the etalon parameters are known.","The measured average near $10^{-7}$ is above what the authors consider acceptable for the low-frequency arm of a next-generation detector, so substrate selection and mounting procedures will have to aim at lower-stress material."],"supporting_citations":[{"why":"supplies the rotating half-wave-plate modulation scheme and the treatment of spurious ellipticity from plate defects.","marker":"[7]"},{"why":"show that a birefringent etalon produces a rotation in addition to ellipticity, the basis for separating the two.","marker":"[9, 10]"},{"why":"fix the refractive index of silicon used to compute interface reflectivity at 1064 and 1550 nm.","marker":"[11-17]"},{"why":"provide the absorption coefficient at 1064 nm that sets the effective etalon parameter Z.","marker":"[18-20]"},{"why":"gives the gas Cotton-Mouton effect used to calibrate the polarimeter's ellipticity scale and axis direction.","marker":"[26]"},{"why":"earlier measurements of intrinsic (110) silicon birefringence to which this result is compared.","marker":"[28, 29]"},{"why":"recent scanning measurements of silicon birefringence used as comparison for the (100) maps.","marker":"[32, 33]"}],"fun_headline_variants":["Silicon's hidden birefringence varies at 10^-7","Position-dependent birefringence in silicon at 10^-7","Silicon birefringence maps: non-uniform at 10^-7","Silicon's birefringence: 10^-7 and not uniform","For ET, silicon's birefringence varies at 10^-7"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The analysis assumes that each illuminated spot on the sample is a single uniform birefringent etalon with known interface reflectivity, no dichroism, and full spatial overlap of the interfering beams; if stress varies through the thickness, surfaces are rough, or the beam only partially overlaps the etalon, the extracted $\\Delta n$ values are biased even though the raw ellipticity is accurate.","fun_headline_variants_meta":{"raw":{"variants":["Silicon's hidden birefringence varies at 10^-7","Position-dependent birefringence in silicon at 10^-7","Silicon birefringence maps: non-uniform at 10^-7","Silicon's birefringence: 10^-7 and not uniform","For ET, silicon's birefringence varies at 10^-7"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000592,"raw_usage":{"total_tokens":2782,"prompt_tokens":958,"completion_tokens":1824,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":574,"completion_tokens_details":{"reasoning_tokens":1721}},"tokens_in":574,"tokens_out":1824,"duration_ms":13858,"temperature":1.0,"reasoning_tokens":1721,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T04:45:08.686703+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Map the same (100) wafer at both 1064 and 1550 nm, where $Z$ and the etalon finesse differ by more than a factor of two; if the per-pass $\\Delta n$ values extracted from Equations 15-19 disagree by more than the stated uncertainties, the single-etalon inversion with $f=1$ is not correct.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"supplies the rotating half-wave-plate modulation scheme and the treatment of spurious ellipticity from plate defects."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"gives the gas Cotton-Mouton effect used to calibrate the polarimeter's ellipticity scale and axis direction."}],"review_version":1}