{"id":"ca001222-29ff-4c12-8f28-bfa5785238ef","arxiv_id":"2501.09728","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":8,"one_line_summary":"A triply resonant BaTiO3-on-SiO2 transducer integrated with a superconducting Nb microwave resonator demonstrates bidirectional microwave-optical conversion at 10 mK, reaching 1e-6 off-chip efficiency.","lead":"Researchers built a tiny chip that converts microwave signals to light and back using barium titrate, a soft ferroelectric crystal whose optical response is tuned with an electric field. The first integrated device of this kind works at millikelvin temperatures and could eventually link superconducting quantum computers through optical fibers.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The headline off-chip efficiency of 1e-6 rests on a fiber-to-chip loss measured only at room temperature and on a backscattering model for the broken-fiber direction; neither is verified at 8 mK.","rationale":"I read the paper as a credible first demonstration of a soft-ferroelectric BaTiO3 transducer. The qualitative claims are supported by the observed avoided crossing, microwave-resonance tuning with ferroelectric hysteresis, coherent bidirectional signals, and the heating dynamics. The reader's CONDITIONAL verdict is appropriate, and I do not regard the central scientific contribution as overturned by calibration uncertainty. My concern is narrower than a foundational flaw: the absolute off-chip efficiency is not as secure as the qualitative demonstration, because the fiber-to-chip loss is assumed from room-temperature characterization and the microwave-to-optical direction had to be inferred through a backscattering model whose parameters cannot be unambiguously separated from other reflections, as the authors themselves note. This is a reason to keep the paper conditional rather than to reject it. The missing data-availability placeholder and the absence of error bars reinforce that conditionality, but they do not change the verdict. I therefore recommend UNCHANGED, with agreement only partial because the reader bundled the backscattering uncertainty with the calibration factors, whereas I see the in-situ fiber-to-chip calibration as the more load-bearing element for the 1e-6 claim.","tokens_in":41230,"tokens_out":9446,"duration_ms":114243,"concrete_test":"Repackage the same sample, or a nominally identical device, with both facet fibers aligned and measure fiber-to-chip transmission continuously from 300 K to 8 mK. Then repeat the pulsed optical-to-microwave efficiency measurement using the same calibration protocol. If the cryogenic fiber-to-chip loss is within ±1 dB of the assumed −5 dB and the pulsed efficiency remains 1e-6 within ±1 dB, the concern is retired. If it shifts by more than 3 dB, the reported off-chip efficiency should be revised. In the same run, with the second fiber aligned, compare forward-transmission microwave-to-optical efficiency with the backward/reflection result to test the fitted backscattering model decisively.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claim, a total off-chip efficiency of 1e-6, is computed through the calibration chain of Appendix D2. That chain uses a fiber-to-chip loss 'estimated at −5 dB from room-temperature characterization' plus the empirical +11.3 dB and +15.65 dB factors. In Sec. III the authors state that one fiber broke during handling, and in Sec. IV they 'rely on residual optical backscattering' for the microwave-to-optical direction. They also explicitly admit that intra-ring forward-backward coupling cannot be clearly distinguished from residual backscattering at the bus-waveguide end facet. Because the fiber-chip interface is the least controlled element, and because the same section reports a sudden, irreversible fiber-chip efficiency degradation under high optical power, a systematic drift of even 3–5 dB between room temperature and 8 mK would directly shift the reported off-chip efficiency and weaken the consistency check with g0 = 2π × 406 Hz. The qualitative demonstration of Pockels transduction, poling, and bidirectional signals does not depend on this calibration, but the absolute efficiency headline does.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript presents an integrated microwave-optical transducer based on BaTiO3-on-SiO2, consisting of a photonic molecule (two coupled ring resonators) and a superconducting Nb microwave resonator. The authors demonstrate bidirectional continuous-wave and pulsed transduction with an off-chip efficiency of 1×10−6 at 8 mK, in-situ ferroelectric poling through a bias port located at the microwave voltage null, and a fully subtractive fabrication process with superconducting air bridges. They also study optically induced heating, distinguishing fast in-cavity dielectric heating from slower substrate and quasiparticle heating. The measured efficiencies are modeled with a linearized input-output S-matrix theory (Appendix B) using parameters from independent characterizations, including a g0 = 2π×406 Hz obtained from dc electro-optic tuning.","tokens_in":41567,"tokens_out":12911,"duration_ms":131769,"significance":"If the claims hold, this is the first integrated triply resonant electro-optic transducer using a soft ferroelectric (BaTiO3), demonstrating a viable path toward higher-Pockels-coefficient materials for quantum interconnects. The in-situ poling concept and the subtractive air-bridge fabrication are genuinely useful contributions that transfer to other ferroelectrics. The paper is strengthened by a self-contained S-matrix model, an independent dc-tuning extraction of g0 (Appendix E2), and a careful thermal characterization with a physically motivated line-shape model. However, the quantitative headline (1e-6 off-chip efficiency) rests on a room-temperature fiber-coupling calibration and an empirical power calibration chain, and the measured device was operated with a ~2 GHz detuning from the nominal triple-resonance condition. These limitations are acknowledged in the body but not fully reflected in the abstract.","major_comments":[{"comment":"The absolute off-chip efficiencies plotted in Fig. 5c are obtained via a calibration chain (Appendix D2) that includes a fiber-to-chip loss 'estimated at −5 dB from room-temperature characterization' and two empirical factors (+11.3 dB and +15.65 dB). The only cryogenic cross-check is the cw optical-heating comparison in Appendix D2, which calibrates power in the fiber but not the fiber-to-chip interface itself. Section VI reports a sudden, irreversible degradation of the fiber-chip efficiency under high optical power, demonstrating that the interface is not stable under the operating conditions used in this work. Because the headline efficiency of 1×10−6 and the comparison to state-of-the-art LiNbO3 transducers in Fig. 5c are directly proportional to these calibration factors, a systematic drift of even 3–5 dB would change the central quantitative claim by a comparable factor. I request an uncertainty budget for the reported efficiencies, and either a cryogenic calibration of the fiber-to-chip loss or a clear statement that the quoted efficiencies are upper bounds based on room-temperature coupling estimates.","section":"Sec. IV / App. D2 / Fig. 5c"},{"comment":"The microwave-to-optical direction was measured entirely in reflection after one optical fiber broke (Sec. III), relying on 'residual optical backscattering' for signal extraction (Sec. IV). The transduction efficiency for this direction is modeled with fitted forward-backward coupling rates ν1,2 (Appendix B2, Table V). The authors themselves state that intra-ring backscattering cannot be clearly distinguished from residual backscattering at the bus-waveguide end facet. Consequently, the absolute efficiency of the microwave-to-optical path and the >10 dB asymmetry between the two directions depend on an ambiguous model parameter. This does not affect the optical-to-microwave efficiency, but the bidirectional quantitative comparison should be presented with this caveat, and the abstract's unqualified 'bidirectional' efficiency should be limited to the optical-to-microwave direction or explicitly qualified.","section":"Sec. IV / App. B2"},{"comment":"The paper repeatedly describes the device as 'triply resonant', but the measured photonic-molecule splitting is 2µ = 2π×4.7 GHz (Sec. III) while the microwave resonance is at ωb = 2π×6.82 GHz (Sec. III); the residual detuning of ~2.1 GHz is acknowledged in Sec. G as the likely cause of the central-peak splitting. The transduction data in Fig. 3 were therefore taken with the microwave-to-molecule frequency mismatch present, and the authors list 'eliminat[ing] the mismatch... providing another 4 dB improvement' as a future improvement (Sec. VII). The abstract and introduction's 'triply resonant transducer' overstates the as-measured device. Please qualify the abstract (e.g., 'designed to be triply resonant') and quantify the detuning in the main text and figures.","section":"Sec. III, IV, G"}],"minor_comments":[{"comment":"The g0 value used in the consistency check combines a measured dc tuning slope with a simulated zero-point voltage; please state the systematic uncertainty in g0 and explicitly note that the dc Pockels response is an upper bound for the microwave-frequency response, as the text already suggests.","section":"App. E2 / A4"},{"comment":"The Data Availability statement says 'available through Zenodo at (to be provided)'; this must be completed before publication.","section":"Data Availability"},{"comment":"The notation ω+ and ω− is ambiguous: Section IV calls the resonance with ∆p=0 the 'lower photonic molecule resonance' while Fig. 3c assigns Stokes and anti-Stokes peaks to opposite detunings; please define the mode ordering explicitly.","section":"Sec. IV / Fig. 3"},{"comment":"The sentence 'with increasing pump power the pulse length is decreased from 1 ms to 40 µs' mixes present and past tense; also verify the stated duty-cycle arithmetic for a fixed 0.8 kHz repetition rate.","section":"Sec. VI"}],"recommendation":"major_revision","confidential_remarks":"The manuscript represents a substantial fabrication and measurement effort from a strong group. The main risk to the central quantitative claim is the absolute calibration chain; I would encourage the editor to require a clear uncertainty statement and a qualified abstract. The data availability statement is incomplete and should be corrected before acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is the first integrated microwave-optical transducer in a soft ferroelectric (BaTiO3), and that alone makes it worth a careful read. The authors built a triply resonant cavity electro-optic device with a superconducting Nb resonator, demonstrated bidirectional continuous-wave and pulsed transduction at 8 mK, and report a total off-chip efficiency around 1e-6. They also developed an in-situ poling scheme that does not add microwave loss, plus a fully subtractive fabrication process with superconducting air bridges. These are real contributions, not incremental tweaks.\n\nWhat is genuinely good: the transduction model is a full linearized S-matrix treatment with forward-backward scattering, and it reproduces the measured lineshapes well. The extracted g0 = 2π × 406 Hz is independently supported by dc electro-optic tuning (Appendix E2), so the consistency check between the model and data is not circular. The heating study is also careful: it distinguishes dielectric heating from quasiparticle heating by blue- vs red-shifts, and the conclusion that in-cavity dielectric heating dominates in the low-power quantum-relevant regime is useful for anyone building such transducers.\n\nThe soft spots are real but not fatal. The headline efficiency rests on a calibration chain that includes a fiber-to-chip loss estimated at room temperature (−5 dB) and, for the microwave-to-optical direction, a backscattering model used after one fiber broke. The stress-test concern about a few dB drift between room temperature and 8 mK is legitimate; the absolute efficiency is probably good to within a factor of 2–3, not to better. No error bars are given, and the data availability link is a placeholder. None of this undermines the qualitative demonstration or the g0 consistency, but it does mean the absolute number should be treated as approximate. The paper itself is candid about the low effective Pockels coefficient (13 pm/V vs the hoped-for 200 pm/V) and about being five orders of magnitude below optimized LiNbO3 per mW, which is the right framing for a first demonstration.\n\nWho should read this: anyone working on electro-optic transduction, ferroelectric photonics, or superconducting-optical quantum links. It deserves a serious referee. I would send it to review with the expectation of revisions: add error bars or uncertainty estimates, clarify the calibration drift risk, and point to actual data. The core results are credible and the platform is likely to matter.","headline":"First integrated BaTiO3 microwave-optical transducer with honest, calibration-limited efficiency numbers; the platform work is solid and the absolute efficiency is openly flagged as approximate.","tokens_in":42129,"tokens_out":1800,"would_cite":true,"duration_ms":21660,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper demonstrates an integrated barium titanate transducer that converts microwave and optical photons bidirectionally at millikelvin temperatures.","keywords":["microwave-optical transduction","barium titanate","Pockels effect","soft ferroelectric","superconducting resonator","triply resonant cavity electro-optics","quantum interconnect","cryogenic integrated photonics"],"falsifier":"Measuring microwave-to-optical transduction in transmission at 10 mK with both optical fibers coupled, and comparing that directly measured efficiency with the reflection-corrected value of $1\\times10^{-6}$, would settle whether the backscattering and calibration model is accurate.","tokens_in":41033,"feed_emoji":"🔁","tokens_out":9484,"duration_ms":82790,"temperature":0.7,"pith_summary":"This paper reports the first integrated microwave-optical transducer built from the soft ferroelectric barium titanate (BaTiO3), a material with a much larger electro-optic (Pockels) effect than the AlN and LiNbO3 films used in earlier devices. The authors show that a chip combining two BaTiO3 ring resonators with a superconducting niobium microwave resonator converts photons in both directions, reaching a total off-chip efficiency of $1\\times10^{-6}$ at 8 mK. They also introduce a device geometry that allows the ferroelectric domains to be poled in place after cooling without adding microwave loss, together with a fully subtractive fabrication process using superconducting air bridges. A separate heating study distinguishes fast in-cavity dielectric heating from quasiparticle and substrate heating, which matters for keeping added noise low in quantum use. If these results hold, soft ferroelectrics become a viable route to more efficient quantum interconnects between superconducting processors.","feed_headline":"Barium titanate converts microwaves to light and back","feed_subtitle":"Soft-ferroelectric chip hits 1e-6 off-chip efficiency at 8 mK, a step toward quantum processor links.","key_machinery":"The load-bearing object is the triply resonant cavity electro-optic system: a photonic molecule formed by two evanescently coupled BaTiO3 ring resonators, whose antisymmetric optical beating is phase-matched to the field of a half-wavelength superconducting niobium microwave resonator. The interaction is the Pockels-effect Hamiltonian $H_{\\mathrm{int}} = \\hbar g_0 (a_+^\\dagger a_- + a_-^\\dagger a_+)(b + b^\\dagger)$, with vacuum coupling $g_0 = G V_{\\mathrm{zpf}}$, where $G = \\partial \\omega_a/\\partial V$ is the frequency-pulling factor and $V_{\\mathrm{zpf}}$ is the zero-point voltage across the ring capacitors. A dc-bias port placed at the voltage node of the microwave mode provides in-situ ferroelectric poling without adding microwave loss, and a capacitively coupled tuning electrode is part of the microwave resonator, giving 91% electrode coverage. The transduction response is modeled by a linearized input-output S-matrix that includes forward-backward optical scattering, and the efficiency is calibrated through a chain of separately measured optical, microwave, and heterodyne gains.","core_discovery":"The central claim is a working, integrated, triply resonant electro-optic transducer in which a photonic molecule of two evanescently coupled BaTiO3 ring resonators is driven by a $\\lambda/2$-type superconducting niobium microwave resonator, with the microwave frequency matched to the optical mode splitting. Bidirectional continuous-wave and pulsed transduction is demonstrated, with a peak pulsed off-chip efficiency of $1\\times10^{-6}$ at 8 mK and linear behavior over a wide range of pump powers. The design places the dc-bias port at the voltage node of the microwave mode, so the ferroelectric can be poled in situ without loading the microwave circuit, and it capacitively couples the tuning electrode into the resonator to reach a usable electrode coverage of 91%. From the measured electro-optic tuning the authors extract a vacuum coupling rate $g_0/2\\pi = 406$ Hz, corresponding to an effective Pockels coefficient of about 13 pm/V, more than an order of magnitude below the 200 pm/V reported for BaTiO3 thin films at cryogenic temperature. The paper also claims that optically induced heating in the quantum-relevant low-power regime is dominated by fast in-cavity dielectric heating rather than straylight-driven quasiparticle heating.","pith_inferences":["Beyond the paper: if the reduced 13 pm/V Pockels coefficient is indeed strain-related, cryogenic strain engineering of the BaTiO3 film is a direct, testable route to recover the 200 pm/V value, which would make the device competitive with optimized LiNbO3 transducers at much lower pump power.","Beyond the paper: because the microwave-to-optical path was measured through residual backscattering with one fiber broken, the true on-chip microwave-to-optical efficiency could be higher than the reported off-chip value; a fully packaged two-fiber measurement would separate device efficiency from the backscattering correction.","Beyond the paper: the post-cooldown inversion of the optical tuning polarity suggests that cooling under an applied bias may deterministically set the ferroelectric domain orientation, which could be used as a fabrication protocol to maximize the low-temperature Pockels response.","Beyond the paper: the demonstrated linear transduction range and the identified heating budget imply that added-noise measurements at single-photon levels are the next decisive test; the paper's estimate of about three photons of added noise at 0 dBm peak pump power could be verified directly with a calibrated noise measurement."],"forward_implications":["If the cryogenic Pockels coefficient can be raised from the measured 13 pm/V to the 200 pm/V reported for BaTiO3 films, the transduction efficiency would rise by roughly 20 dB.","Improving the intrinsic optical quality factor from $1\\times10^5$ to the reported absorption-limited $3.8\\times10^6$ would add about 30 dB of efficiency, and the authors estimate that combining all listed improvements could bring off-chip efficiency close to unity.","In the low-power regime relevant to quantum transduction, in-cavity dielectric heating dominates over quasiparticle heating, and its sub-microsecond response means it cannot be suppressed by lowering the optical pump duty cycle.","The transducer design and the subtractive fabrication process transfer to other large-Pockels materials such as SrTiO3 and LiNbO3 with minor modifications.","Optical-to-microwave transduction remains linear in pump power from $-30$ dBm to $+12$ dBm, and the phase-coherent interference of the two optical sidebands confirms that the conversion process is coherent."],"supporting_citations":[{"why":"Establishes superconducting cavity electro-optics as a platform for coherent microwave-optical conversion, which this work extends to BaTiO3.","marker":"[22]"},{"why":"Supplies the triply resonant photonic-molecule plus microwave-resonator architecture that the present device adapts to a soft ferroelectric.","marker":"[23]"},{"why":"Provides the thin-film lithium niobate cavity electro-optic framework and efficiency-scaling benchmark for direct electro-optic transduction.","marker":"[24]"},{"why":"Gives the optimized LiNbO3 efficiency baseline ($-23$ dB/mW) against which this BaTiO3 demonstration is compared.","marker":"[25]"},{"why":"Reports the cryogenic BaTiO3 Pockels coefficient of 200 pm/V that motivates the device and sets the efficiency projection.","marker":"[36]"},{"why":"Supplies the bulk SrTiO3 Pockels values and quantum-paraelectric reference used in the material comparison.","marker":"[37]"},{"why":"Establishes that BaTiO3 needs a dc-bias field for domain poling and provides the angular-overlap model used in simulations.","marker":"[40]"},{"why":"Shows that strain affects the electro-optic response of BaTiO3 films, the explanation offered for the measured 13 pm/V value.","marker":"[43]"}],"fun_headline_variants":["BaTiO3 chip achieves two-way microwave-optical conversion","Cryogenic BaTiO3 transducer hits 1e-6 conversion efficiency","Soft ferroelectric enables bidirectional quantum link","Integrated BaTiO3: microwaves to photons and back"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The reported absolute efficiency of $1\\times10^{-6}$ and the extracted coupling rate $g_0$ rest on a chain of calibration factors measured at room temperature, including an estimated $-5$ dB fiber-to-chip loss and empirical $+11.3$ dB and $+15.65$ dB power calibrations, and on a fitted backscattering model used because one optical fiber broke before the cryogenic run.","fun_headline_variants_meta":{"raw":{"variants":["BaTiO3 chip achieves two-way microwave-optical conversion","Cryogenic BaTiO3 transducer hits 1e-6 conversion efficiency","Soft ferroelectric enables bidirectional quantum link","Integrated BaTiO3: microwaves to photons and back"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000454,"raw_usage":{"total_tokens":2344,"prompt_tokens":1068,"completion_tokens":1276,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":684,"completion_tokens_details":{"reasoning_tokens":1208}},"tokens_in":684,"tokens_out":1276,"duration_ms":13166,"temperature":1.0,"reasoning_tokens":1208,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T19:42:26.589156+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measuring microwave-to-optical transduction in transmission at 10 mK with both optical fibers coupled, and comparing that directly measured efficiency with the reflection-corrected value of $1\\times10^{-6}$, would settle whether the backscattering and calibration model is accurate.","supporting_citations":[{"cited_title":"Typically however, the sig- nal levels in transducers for quantum transduction are just a few or even single photons","cited_arxiv_id":null,"evidence_quote":"Establishes superconducting cavity electro-optics as a platform for coherent microwave-optical conversion, which this work extends to BaTiO3."},{"cited_title":"Note, that this means that both the bottom probe and mixing chamber temperature are swept simultaneously, which we call Tfridge","cited_arxiv_id":null,"evidence_quote":"Supplies the triply resonant photonic-molecule plus microwave-resonator architecture that the present device adapts to a soft ferroelectric."},{"cited_title":"Grumbling and M","cited_arxiv_id":null,"evidence_quote":"Provides the thin-film lithium niobate cavity electro-optic framework and efficiency-scaling benchmark for direct electro-optic transduction."},{"cited_title":"Bravyi, O","cited_arxiv_id":null,"evidence_quote":"Gives the optimized LiNbO3 efficiency baseline ($-23$ dB/mW) against which this BaTiO3 demonstration is compared."},{"cited_title":"Wehner, D","cited_arxiv_id":null,"evidence_quote":"Reports the cryogenic BaTiO3 Pockels coefficient of 200 pm/V that motivates the device and sets the efficiency projection."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the bulk SrTiO3 Pockels values and quantum-paraelectric reference used in the material comparison."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes that BaTiO3 needs a dc-bias field for domain poling and provides the angular-overlap model used in simulations."},{"cited_title":"Meesala, D","cited_arxiv_id":null,"evidence_quote":"Shows that strain affects the electro-optic response of BaTiO3 films, the explanation offered for the measured 13 pm/V value."}],"review_version":1}