{"id":"6118be43-ae1f-42d6-b4ad-8aff7dc302ff","arxiv_id":"2608.09003","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"The paper demonstrates 4-inch wafer-scale Ce:YIG films on silicon, a non-destructive wafer-scale magneto-ellipsometry method, and ring-resonator optical isolators across the wafer.","lead":"Researchers grew cerium-doped yttrium iron garnet, a light-rotating magnetic film, across an entire 4-inch silicon wafer and mapped its properties without cutting it. The work points toward mass-producing optical isolators and other light-routing devices on ordinary silicon wafers.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Headline Faraday rotation and loss values are not traceable to a described measurement: they appear only after an unreported 'calculations' step, so the central quantitative claims are not yet supported.","rationale":"The reader correctly flags the saturation assumption in the magneto-ellipsometry as a weakness: the applied field is only 'up to 578 Oe' while VSM gives in-plane saturation at about 525 Oe, so local partial magnetization could contaminate the 13% off-diagonal-permittivity variation. That concern is legitimate and testable. However, I regard the more load-bearing issue as the unexplained provenance of the two headline material figures, θ_F = 2318±102 deg/cm and α = 80±11 dB/cm. The paper never states the formula or model that converts the measured ellipsometric or device data into these numbers, and it does not reconcile α with the ellipsometric k values or the acknowledged defects. Because the strongest claim in the abstract is quantitative, this missing chain of inference undermines the headline more directly than the saturation concern: even if every measured spot is fully saturated, the reported material constants still depend on an unreported 'calculations' step. This does not invalidate the wafer-scale growth or the six working isolators, which are credible and independently interesting; it only means the quantitative material figures should be treated as conditional pending a disclosed and validated extraction. The reader's CONDITIONAL verdict therefore remains appropriate; my concern is an additional reason for the same verdict rather than a reason to reject the paper outright.","tokens_in":11791,"tokens_out":7896,"duration_ms":82287,"concrete_test":"Ask the authors to release the raw forward/backward transmission spectra for all six isolators (or the raw Mueller-matrix spectra) together with a complete extraction pipeline: explicit equations converting the measured quantities to θ_F and α, including assumed Ce:YIG/YIG thicknesses, refractive indices, mode-overlap factor, coupling coefficient, and round-trip loss. Then independently refit one isolator spectrum with the stated model and recompute θ_F and α for all six. If the refit changes the mean by more than the quoted ±102 deg/cm or ±11 dB/cm, or if the fit is non-unique (e.g., coupling and material loss trade off by more than 20%), the headline material figures are not established by the current evidence.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claims—θ_F = 2318±102 deg/cm and α = 80±11 dB/cm—are not linked to a described measurement. In Section 3, the magneto-ellipsometry discussion reports ε₂,real ≈ 0.0046 at 1550 nm with ~13% wafer variation and k between 2.2×10⁻⁵ and 1.63×10⁻⁴, but the only sentence connecting these data (or the six ring-resonator spectra) to the headline material numbers is 'After the calculations...' immediately following the isolator results in Figure 5. No formula, model, parameter list, or uncertainty budget is given. Eqs. (1)–(9) define the Mueller-matrix formalism but do not specify how the measured spectra or device transmission are converted to a material Faraday rotation and a propagation loss. If the values came from the ring-resonator fits, they depend on coupling coefficients, round-trip loss, mode overlap with the Ce:YIG/YIG stack, and nonreciprocal phase shift, and are not unique without a stated model. If they came from the ellipsometry data, the paper still omits the conversion and error propagation. Moreover, α = 80 dB/cm is not reconciled with the ellipsometric k values: the maximum k = 1.63×10⁻⁴ corresponds to α ≈ 57 dB/cm via α = 4πk/λ, and the difference is attributed to 'defects' (grain-boundary and CeO₂ precipitates) without quantification. Thus the abstract's 'strong Faraday effect' and 'low propagation loss' are not verifiable from the reported evidence, even though the fabrication and device demonstrations themselves are plausible.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports the growth of 4-inch wafer-scale Ce:YIG thin films on silicon by confocal RF magnetron sputtering and demonstrates wafer-scale non-destructive magneto-ellipsometry characterization, along with six ring-resonator magneto-optical isolators fabricated across the wafer. The paper claims a Faraday rotation of 2318±102 deg/cm, a propagation loss of 80±11 dB/cm, thickness uniformity of 3.5%, and isolator isolation ratios of 18.6–31 dB. The structural, compositional, and magnetic characterization (XRD, EBSD, XPS, VSM, STEM) supports the uniformity of the garnet phase and magnetic properties. However, the headline material parameters are presented without a traceable extraction procedure, and several reported uniformity metrics are mutually inconsistent.","tokens_in":12187,"tokens_out":2508,"duration_ms":24971,"significance":"If the claimed material parameters and wafer-scale uniformity are substantiated, this would be a notable advance in monolithic integration of magneto-optical garnets on silicon, relevant to nonreciprocal photonic integrated circuits. The custom magneto-ellipsometry mapping approach and the demonstration of multiple isolators across a 4-inch wafer are valuable contributions. However, the central quantitative claims—the Faraday rotation and propagation loss of the film—are not currently verifiable from the reported evidence, because the conversion from measurements to these values is not documented. The fabrication and device demonstrations themselves are plausible, but the material-level headline numbers need a rigorous, described extraction and uncertainty analysis before the paper can be accepted.","major_comments":[{"comment":"The headline values θ_F = 2318±102 deg/cm and α = 80±11 dB/cm are introduced only by the phrase 'After the calculations' with no model, equations, or parameter list. No formula connects the measured ellipsometric Mueller-matrix elements or the ring-resonator transmission spectra to these material parameters. If the values are derived from the isolator spectra, the extraction depends on coupling coefficients, round-trip loss, and the nonreciprocal phase shift model, none of which is specified. If they derive from the ellipsometry data, the conversion from ε2 to θ_F and from k to α is omitted. Without this derivation and a propagation-of-uncertainty budget, the abstract's 'strong Faraday effect' and 'low propagation loss' claims are not verifiable.","section":"Section 3 (after Figure 5) and Abstract"},{"comment":"The reported α = 80±11 dB/cm is not reconciled with the measured extinction coefficient. The maximum k = 1.63×10⁻⁴ at 1550 nm corresponds to α ≈ 57 dB/cm via α = 4πk/λ, and the difference is attributed to 'grain-boundary precipitates and CeO₂ precipitates' without any quantitative estimate. The paper also states that at 1550 nm 'the reliability of the fitting results was relatively low, as the measurement has approached the limit of the instrument's accuracy.' These statements together leave the propagation-loss claim without quantitative support. The authors should either provide a measured loss value with an uncertainty that accounts for the instrument limit, or report α only as an upper bound derived from k.","section":"Section 3, Figure 4 and α from k"},{"comment":"The magneto-ellipsometry applies a transverse magnetic field of up to 578 Oe at each spot and assumes this saturates the in-plane magnetization at all 54 measurement points. The VSM data in Section 2 give an in-plane saturation field of approximately 525 Oe, so the applied field margin is only about 10%. If any location on the wafer has a slightly higher local saturation field—due to thickness or composition variation—the extracted off-diagonal permittivity and its reported 13% variation would reflect partial magnetization rather than true material nonuniformity. The authors should demonstrate saturation at each mapped position (e.g., by measuring ε2 as a function of applied field at representative points) or use a field safely above the worst-case saturation field.","section":"Sections 2 and 3, magneto-ellipsometry saturation assumption"},{"comment":"The uniformity numbers reported in the conclusion are internally inconsistent with those in the Results. The conclusion states 'optical constant variation of 2.8%' and 'off-diagonal permittivity element variation of 13%,' but Section 3 reports refractive-index variation of 0.25–0.30% and off-diagonal element variation of 'around 13%' in one place and 'approximately 9%' in another (Figures 4C–4F and accompanying text). These discrepancies need to be resolved, because the wafer-uniformity claim is one of the paper's main selling points. Please report a single, well-defined uniformity metric (e.g., standard deviation divided by mean) for each property, with the number of points and the definition stated.","section":"Conclusion vs. Section 3, uniformity metrics"}],"minor_comments":[{"comment":"The manuscript contains numerous typographical errors, including 'opticla' (Introduction), 'geomery' (Introduction), 'metrit' (Introduction), 'Muller' for Mueller, 'demosntrate' (Conclusion), 'homogenity' (Conclusion), and 'arised' (Section 3). These should be corrected.","section":"Throughout"},{"comment":"The equations are not formatted correctly and are difficult to read; several subscript/superscript characters are garbled. In particular, Eq. (4) and Eq. (9) are not legible in the provided text. Please ensure all equations are typeset properly and define every symbol (e.g., Amp, E0, Br, Eg in Eq. (4)).","section":"Equations (1)–(9)"},{"comment":"The transmission spectra of the six isolators (Figures 5D–5I) are shown without error bars or a statement of measurement repeatability, which is important given the claim of isolation ratios spanning 18.6–31 dB. Please add measurement uncertainties or specify the spectral resolution and averaging.","section":"Figure 5"},{"comment":"The paper states the Ce3+ content is 'around 20%' but does not state the nominal Ce content of the target or the expected Ce:YIG stoichiometry. Please specify the target composition and how the measured Ce3+ fraction relates to it.","section":"Section 2 (XPS)"},{"comment":"The in-plane and out-of-plane hysteresis loops are measured 'at different positions on the wafer,' but the number of positions and their radial coordinates are not given. Please provide this information to support the uniformity claim.","section":"Section 2 (VSM)"}],"recommendation":"major_revision","confidential_remarks":"The paper addresses a timely topic and contains a credible fabrication effort, but the central material-parameter claims are not yet supported by a traceable analysis. The lack of an extraction model for θ_F and α is the main blocker; the uniformity-metric inconsistencies are secondary but need cleaning. I would recommend major revision, not rejection, because the issues are fixable within the manuscript's scope: the authors can add the missing derivation, a saturation check, and reconciled uniformity numbers."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read the Ce:YIG wafer paper. The headline is plausible but two numbers in the abstract are not yet supported: the Faraday rotation and loss are \"calculated\" after the isolator measurements, with no model described. That's the one thing to check before taking the quantitative claims at face value.\n\nWhat's actually new: first 4-inch wafer-scale Ce:YIG/YIG on silicon, and a custom wafer-scale magneto-ellipsometry mapper that gives thickness, n/k, and off-diagonal permittivity maps. The XRD, EBSD, XPS, and VSM data are consistent with uniform polycrystalline garnet across the wafer, and six ring isolators show isolation ratios 18.6–31 dB at different radii. That is a real fabrication advance over the mm-scale samples cited in the introduction.\n\nThe soft spots are real but not fatal to the core claim. The abstract's 2318±102 deg/cm and 80±11 dB/cm appear only in a sentence \"After the calculations...\" following Figure 5. No formula, parameter list, or uncertainty budget connects the ellipsometry or the device spectra to those material numbers. The ellipsometry section itself gives ε₂,real ≈ 0.0046 with ~13% variation, and k between 2.2e-5 and 1.63e-4; the max k translates to about 57 dB/cm at 1550 nm, so the reported 80 dB/cm implies additional loss attributed to \"defects\" without quantification. Also, the applied field of 578 Oe vs the VSM in-plane saturation field of about 525 Oe is a little close for comfort; if any location isn't saturated, the 13% variation includes a magnetization contribution.\n\nThe paper's own text flags the noise limitations in the ellipsometry, which is honest. But the missing derivation for the headline values is a genuine transparency gap. This is a strong fabrication demo and a useful system-level result, not yet a reliable material characterization paper.\n\nWho would get value: anyone working on integrated magneto-optics or wafer-scale heterogeneous integration. It deserves a serious referee, but I'd ask for a clear statement of how the Faraday rotation and loss were extracted from the device data or ellipsometry, and a reconciliation with the k values. I'd also want the sputtering and annealing parameters spelled out.\n\nRecommendation: engage with it in review; don't desk reject.","headline":"A real wafer-scale fabrication advance whose headline material numbers are not traceable to a described measurement.","tokens_in":12745,"tokens_out":2091,"would_cite":true,"duration_ms":19601,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["78.20.Ls","75.70.Ak","42.82.-m","81.15.Cd"],"model":"deepseek-v4-flash","headline":"Cerium-substituted yttrium iron garnet films can now be grown uniformly across a full 4-inch silicon wafer and used to make working magneto-optical isolators.","keywords":["Ce:YIG","magneto-optical isolator","wafer-scale integration","magnetron sputtering","magneto-ellipsometry","silicon photonics","ring resonator","nonreciprocal photonics"],"falsifier":"Measure the off-diagonal permittivity at the same 54 wafer positions under applied fields of 578 Oe and 1200 Oe, or map the local magnetic hysteresis at each spot. If the extracted permittivity values change systematically when the field is raised, or if any region has an in-plane saturation field above 578 Oe, the reported 13% wafer-scale variation would need to be reinterpreted as partial-magnetization artifact rather than pure material nonuniformity.","tokens_in":11611,"feed_emoji":"🧲","tokens_out":11404,"duration_ms":99289,"temperature":0.7,"pith_summary":"The paper is trying to establish that the long-standing size limit on magneto-optical garnet integration with silicon, which previously held Ce:YIG films to millimeter-scale areas, can be broken by a confocal sputtering process. A sympathetic reader would care because optical isolators are a missing component in dense photonic circuits, and wafer-scale growth is what would let them be batch-fabricated alongside silicon waveguides. The paper reports a 4-inch wafer of Ce:YIG on a YIG seed layer with thickness uniformity of 3.5%, Faraday rotation of $2318\\pm102$ deg/cm, propagation loss of $80\\pm11$ dB/cm, and six ring-resonator isolators with isolation between 18.6 and 31 dB. It also introduces a non-destructive wafer-scale magneto-ellipsometer that maps the film thickness, optical constants, and magneto-optical permittivity tensor at 54 points across the wafer.","feed_headline":"Magneto-optical isolators work across a full 4-inch silicon wafer","feed_subtitle":"Uniform Ce:YIG films deliver 2318 deg/cm Faraday rotation and six isolators at 18.6-31 dB across the wafer.","key_machinery":"The central object is the Ce:YIG/YIG bilayer on silicon: the YIG seed layer provides a crystallized template so the Ce:YIG layer can form the garnet phase after rapid thermal annealing, while the confocal geometry of the radio-frequency magnetron sputtering with the wafer rotating at 30 rpm is what distributes material uniformly over the 4-inch area. The characterization machinery is a custom-built wafer-scale magneto-ellipsometer that places permanent magnets over the focused spot, applies a transverse magnetic field of up to 578 Oe under Voigt geometry, and records Mueller-matrix spectra at 54 points. Those spectra are fitted with a Tauc-Lorentz dispersion model plus a transfer-matrix calculation to extract the diagonal ($\\varepsilon_1$) and off-diagonal ($\\varepsilon_2$) elements of the permittivity tensor. This two-part mechanism—uniform deposition plus non-destructive full-tensor mapping—is what lets the paper claim both fabrication and metrology at wafer scale.","core_discovery":"The central claim is that a two-step deposition process—a roughly 60 nm YIG seed layer crystallized at 850 °C followed by roughly 120 nm Ce:YIG crystallized at 950 °C under controlled oxygen—produces uniform, polycrystalline, predominantly garnet-phase Ce:YIG across an entire 4-inch silicon wafer. At 1550 nm the film shows a Faraday rotation of $2318\\pm102$ deg/cm, propagation loss of $80\\pm11$ dB/cm, saturation magnetization of $143.5\\pm1.5$ emu/cm³, refractive index $2.258\\pm0.007$, and an off-diagonal permittivity tensor element with real part about $0.0046$ and a wafer-scale variation near 13%. Six SiN ring-resonator isolators fabricated from center to wafer edge show isolation ratios from 18.6 to 31 dB and insertion losses from 2.8 to 4.3 dB. The paper further claims that a custom magneto-ellipsometer using a transverse magnetic field and transfer-matrix fitting can characterize the full permittivity tensor nondestructively over the whole wafer, and that together these results constitute the first wafer-scale monolithic integration of magneto-optical garnet films and nonreciprocal devices on silicon.","pith_inferences":["Beyond the paper: because confocal sputtering is composition-agnostic, the same seed-layer and annealing recipe should transfer to other garnet families such as Bi:YIG or Ce:TbIG, where wafer-scale uniformity would open different wavelength windows for nonreciprocal devices.","Beyond the paper: the 13% wafer-scale variation in the off-diagonal permittivity element predicts a spread in device isolation; one could test this by computing each isolator's expected isolation from the local Faraday-rotation map and comparing with the measured 18.6 to 31 dB range.","Beyond the paper: the saturation assumption behind the wafer map is directly testable by sweeping the applied field above 578 Oe at a subset of the 54 points; if the extracted off-diagonal permittivity continues to change with field, the reported map is partly a magnetization map rather than a pure material-uniformity map.","Beyond the paper: the CeO2 precipitates with average diameter 10.8 nm are a plausible physical origin of a substantial part of the 80 dB/cm loss, so compositional adjustment or oxygen-partial-pressure tuning that removes them should move the film's figures of merit toward the bonded-epitaxial benchmarks cited in the paper."],"forward_implications":["Optical isolators and other nonreciprocal devices can be made across a whole silicon photonics wafer rather than as individual millimeter-scale chips, removing a major bottleneck for high-density photonic integration.","The non-destructive magneto-ellipsometer gives a metrology path for production: thickness, optical constants, and magneto-optical constants can be mapped without cleaving or damaging the wafer.","With wafer-scale Ce:YIG, arrays of isolators and circulators for dense wavelength-division-multiplexed transceivers, magneto-optical in-memory computing networks, and integrated magnetometers become manufacturable.","The reported uniformity numbers—3.5% thickness and sub-0.3% refractive-index variation—provide concrete specifications that a silicon foundry could design around for magneto-optical process integration.","The main remaining materials path is clear: suppressing the observed CeO2 precipitates and grain-boundary defects in the Ce:YIG layer should raise Faraday rotation and lower propagation loss further."],"supporting_citations":[{"why":"Defines the waveguide-integrated isolator and circulator performance on silicon nitride that the wafer-scale process needs to reproduce.","marker":"[1]"},{"why":"Shows a prior Ce:YIG-on-silicon isolator made by adhesive bonding, the integration route that is limited in scale and that this work replaces with direct wafer-scale sputtering.","marker":"[11]"},{"why":"Provides reference values for Ce-substituted iron garnet saturation magnetization and film properties on silicon used to validate the measured magnetic uniformity.","marker":"[17]"},{"why":"Reports an epitaxial bonded-film isolator with about -4500 deg/cm Faraday rotation and 40 dB/cm loss, the performance benchmark for the lower-loss and higher-rotation direction.","marker":"[19]"},{"why":"Demonstrates polarization-independent magneto-optical isolators on silicon, showing the device class that wafer-scale film integration would enable.","marker":"[20]"},{"why":"Supplies the magneto-optical spectroscopic ellipsometry method for extracting permittivity tensor elements of garnet films that the wafer-scale instrument extends.","marker":"[30]"},{"why":"Gives the 200-1770 nm optical and magneto-optical spectra of Ce:YIG, used to assign the Ce3+ 4f-5d transitions that shape the off-diagonal permittivity.","marker":"[31]"},{"why":"Shows that uniform films over large areas can be deposited by magnetron sputtering with a small target, supporting the confocal-geometry uniformity argument.","marker":"[33]"},{"why":"Establishes that strongly Ce-substituted iron garnets have enhanced magneto-optical response, the basis for attributing the 800 nm off-diagonal peak to Ce3+ transitions.","marker":"[35]"}],"fun_headline_variants":["Wafer-scale Ce:YIG films enable silicon isolators","4-inch Ce:YIG films bring magneto-optical isolators to silicon","Ce:YIG isolators now cover full silicon wafers","Monolithic magneto-optical isolators scale to 4-inch silicon","Uniform Ce:YIG films yield wafer-scale isolators"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the 578 Oe transverse field applied at every measured spot saturates the in-plane magnetization of the film at all 54 points, so the variation seen in the magneto-optical map is real material nonuniformity rather than the result of some spots being only partially magnetized.","fun_headline_variants_meta":{"raw":{"variants":["Wafer-scale Ce:YIG films enable silicon isolators","4-inch Ce:YIG films bring magneto-optical isolators to silicon","Ce:YIG isolators now cover full silicon wafers","Monolithic magneto-optical isolators scale to 4-inch silicon","Uniform Ce:YIG films yield wafer-scale isolators"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001482,"raw_usage":{"total_tokens":6010,"prompt_tokens":1060,"completion_tokens":4950,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":676,"completion_tokens_details":{"reasoning_tokens":4858}},"tokens_in":676,"tokens_out":4950,"duration_ms":32879,"temperature":1.0,"reasoning_tokens":4858,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T04:18:04.864750+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the off-diagonal permittivity at the same 54 wafer positions under applied fields of 578 Oe and 1200 Oe, or map the local magnetic hysteresis at each spot. If the extracted permittivity values change systematically when the field is raised, or if any region has an in-plane saturation field above 578 Oe, the reported 13% wafer-scale variation would need to be reinterpreted as partial-magnetization artifact rather than pure material nonuniformity.","supporting_citations":[{"cited_title":"Ce:YIG/Silicon-on-Insulator waveguide optical isolator realized by adhesive bonding,","cited_arxiv_id":null,"evidence_quote":"Shows a prior Ce:YIG-on-silicon isolator made by adhesive bonding, the integration route that is limited in scale and that this work replaces with direct wafer-scale sputtering."},{"cited_title":"Europium-substituted cerium iron garnet thin films for silicon- integrated nonreciprocal photonic device applications,","cited_arxiv_id":null,"evidence_quote":"Provides reference values for Ce-substituted iron garnet saturation magnetization and film properties on silicon used to validate the measured magnetic uniformity."},{"cited_title":"Magneto-optical isolator with silicon waveguides fabricated by direct bonding,","cited_arxiv_id":null,"evidence_quote":"Reports an epitaxial bonded-film isolator with about -4500 deg/cm Faraday rotation and 40 dB/cm loss, the performance benchmark for the lower-loss and higher-rotation direction."},{"cited_title":"Silicon-based integrated polarization-independent magneto-optical isolator,","cited_arxiv_id":null,"evidence_quote":"Demonstrates polarization-independent magneto-optical isolators on silicon, showing the device class that wafer-scale film integration would enable."},{"cited_title":"Optical and magneto-optical properties of Bi substituted yttrium iron garnets prepared by metal organic decomposition,","cited_arxiv_id":null,"evidence_quote":"Supplies the magneto-optical spectroscopic ellipsometry method for extracting permittivity tensor elements of garnet films that the wafer-scale instrument extends."},{"cited_title":"Uniform film in large areas deposited by magnetron sputtering with a small target,","cited_arxiv_id":null,"evidence_quote":"Shows that uniform films over large areas can be deposited by magnetron sputtering with a small target, supporting the confocal-geometry uniformity argument."},{"cited_title":"Strong magneto‐optical enhancement in highly Ce‐substituted iron garnet films prepared by sputtering,","cited_arxiv_id":null,"evidence_quote":"Establishes that strongly Ce-substituted iron garnets have enhanced magneto-optical response, the basis for attributing the 800 nm off-diagonal peak to Ce3+ transitions."}],"review_version":1}