{"id":"41dd18a0-9fd5-4967-9253-d6490d8eefe5","arxiv_id":"2608.03402","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Two cascaded silicon nitride micro-rings on one chip amplify and measure squeezed light before off-chip loss, yielding 4.6 dB observed squeezing despite 7 dB downstream loss.","lead":"This paper reports a silicon nitride chip with two cascaded micro-rings: the first makes squeezed light, the second amplifies and measures it before it leaves the chip. The team saw 4.6 dB of quantum noise reduction even with 7 dB of loss after the chip, showing on-chip amplification can make quantum measurements loss-tolerant.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 4.6 dB claim fails an internal consistency check: the reported max/min levels imply a +4.3 dB common-mode R2 gain shift between the Pump1-off reference and the dual-pump measurement, invalidating Eq. (6).","rationale":"The reader's conditional verdict is well-founded, but the reported numbers point to a more specific, load-bearing flaw. I found no circularity or fabrication concerns: the theory is standard and the 98% fringe demonstrates coherent two-stage operation. The issue is entirely about experimental reference calibration. Under Eq. (6), any common-mode gain offset between the reference and the measurement directly contaminates R_det; near OPO threshold the gain is steeply power-dependent, and the >30 dB Pump1 rejection is insufficient to rule out interference or thermal effects. The internal inconsistency, max plus min equals +8.6 dB, is quantitative evidence that the equal-gain assumption fails, so the 4.6 dB cannot currently be attributed to R1 squeezing. A targeted seed-gain calibration would settle this. I moved the verdict to UNVERDICTED because the key claim is unverified, not because the approach is conceptually wrong.","tokens_in":10784,"tokens_out":28600,"duration_ms":254295,"concrete_test":"Inject a weak 1542 nm seed directly into R2 through the F2 add port (bypassing R1) and measure the phase-sensitive R2 gain at the working Pump2 setting with Pump1 off and then with Pump1 on at its normal operating power. If the seed gain or the amplified-vacuum noise level changes by more than 0.5 dB between the two configurations, the R2-alone trace in Fig. 6(B) is not a valid denominator and the 4.6 dB value is a calibration artifact. In the same re-run, also verify that V_min times V_max divided by V_ref squared equals unity, to rule out a common-mode gain offset.","verdict_should_be":"UNVERDICTED","load_bearing_attack":"Equations (4)-(6) require that R2's gain G be identical in the 'vacuum' reference (Pump1 off) and the squeezing measurement (both pumps on); otherwise R_det measures G_on^2/G_off^2 times the true R. The paper never checks this. The data themselves are inconsistent with equal gains: for a pure two-mode squeezed state, the linear noise variances in the squeezed and anti-squeezed quadratures relative to the same amplifier reference must be reciprocal, so in dB the max and min values should sum to approximately 0. The reported +13.2 dB and -4.6 dB sum to +8.6 dB, a +4.3 dB common-mode offset. Attributing this offset to a gain change would imply 8.9 dB of R1 squeezing, exceeding the paper's own 5.2 dB escape-efficiency bound; attributing it to phase-insensitive noise cannot explain the 13.2 dB anti-squeezing without also destroying the 4.6 dB minimum. The most plausible explanation is that turning on Pump1 changes R2's gain, either through thermal crosstalk, pump-induced refractive shifts (the paper notes operation close to OPO threshold where gain is extremely sensitive), or residual Pump1 leakage (filters specified only as '>30 dB') interfering with Pump2. If R2's gain is phase-modulated by Pump1 leakage, the red trace in Fig. 6(B) would be phase-dependent even with vacuum input, producing a false 'squeezing'. Thus the central loss-tolerance claim is not established.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes and implements an amplifier-assisted on-chip quantum measurement scheme using two cascaded silicon nitride micro-ring resonators. A first ring (R1) generates a two-mode squeezed state via four-wave mixing, and a second ring (R2) acts as a high-gain parametric amplifier that measures the squeezing before off-chip loss. The authors report 4.6 dB of detected squeezing with an estimated 7 dB downstream loss, and claim the first monolithic SU(1,1) interferometer with an estimated 5 dB signal-to-noise enhancement. The theoretical basis (Eqs. 1-6) is standard, and the device fabrication and linear characterization are described in detail.","tokens_in":11077,"tokens_out":12474,"duration_ms":107439,"significance":"If the 4.6 dB squeezing claim is correct, the work demonstrates a practical route to loss-resilient quantum measurement in integrated photonics, which would be a valuable advance. The paper includes a clean derivation of the loss-tolerant detection principle, a plausible device architecture, and a direct measurement of the noise reduction. However, the central experimental claim currently rests on an unverified assumption about the stability of the amplifier gain, and the data themselves show an internal inconsistency that must be resolved before the claim can be accepted.","major_comments":[{"comment":"Equation (6) assumes that the gain G of R2 is identical in the squeezed-input measurement (both pumps on) and in the vacuum-reference measurement (Pump 1 off). The reported data are inconsistent with this assumption. For a pure two-mode squeezed state measured with the same amplifier gain, the maxima and minima of the phase-dependent noise in dB must sum to 0 dB relative to the same vacuum reference. The reported values of +13.2 dB and -4.6 dB sum to +8.6 dB. This common-mode offset implies that the reference level (the orange trace in Fig. 6(B)) does not correspond to the same amplifier response as the red trace, or that an additional phase-insensitive noise contribution is present. If the offset is interpreted as a gain increase of R2 when Pump 1 is turned on, the inferred squeezing from R1 would be -13.2 dB, exceeding the paper's own escape-efficiency bound of -5.2 dB; if interpreted as added noise, it would raise the minimum above the observed value. The paper provides no measurement of the R2 gain stability between the two configurations, despite the acknowledged sensitivity of the gain near the OPO threshold. The authors must provide a direct calibration of the R2 gain (or an equivalent reference with both pumps on) to validate Eq. (6).","section":"Quantum noise measurement and observation of squeezing, Fig. 6(B), Eq. (6)"},{"comment":"The paper attributes the discrepancy between the observed -4.6 dB and the -5.2 dB escape-efficiency upper bound to coupling loss between the two rings and detuning in the squeezer ring. This explanation is not consistent with the observed anti-squeezing of +13.2 dB. Loss between the rings would reduce both the squeezing and the anti-squeezing relative to the ideal values, but the measured anti-squeezing is 8.6 dB higher than the reciprocal of the measured squeezing. Therefore, the stated loss mechanism cannot account for the data. The manuscript needs to explicitly address the origin of the common-mode offset and to show that it does not affect the squeezing estimate.","section":"Discussion, second paragraph"},{"comment":"The claimed 5 dB SNR enhancement of the SU(1,1) interferometer is an indirect estimate that depends on the same unverified parameters. The estimate uses an R2 classical gain of 17 dB and an output noise level '5 dB above shot noise' from Fig. 6(B). If the R2 gain changes when Pump 1 is on, or if the noise reference is invalid, both the phase-signal gain (10 dB) and the noise penalty (5 dB) are incorrect. The paper also acknowledges that the quantum enhancement is not directly measured. To support this secondary claim, the authors should either measure the SNR enhancement directly or clearly state that the value is conditional on the validity of the calibration.","section":"On-chip SU(1,1) Interferometer with Injected Seed"}],"minor_comments":[{"comment":"The text states that the R2 working pump power is 16.3 mW, equal to the measured OPO threshold of 16.3 mW, while also claiming the pumps remain below threshold; clarify how detuning changes the threshold and whether the working point is indeed below the detuned threshold.","section":"Gain characterization of OPAs"},{"comment":"The origin of the ±0.4 dB uncertainties is not described; specify whether these are standard deviations over repeated phase scans, and the averaging procedure.","section":"Quantum noise measurement and observation of squeezing, Fig. 6(B)"},{"comment":"Equation (5) introduces L as the overall loss but does not explicitly define it as an intensity loss between 0 and 1; please define the variable and state its relationship to the dB losses quoted in the text.","section":"Principle of loss-tolerant, amplifier-assisted measurement, Eq. (5)"},{"comment":"The phrase 'the first monolithic SU(1,1) interferometer' should be supported with a comparison to prior integrated implementations, or softened to avoid a potentially contested novelty claim.","section":"Abstract and Introduction"}],"recommendation":"major_revision","confidential_remarks":"The referee report identifies a serious consistency issue in the central data. I recommend that the editor require the authors to provide a direct calibration of the R2 gain with Pump 1 on and off, and to reanalyze the squeezing value accordingly. If the calibration shows that the gain changes, the 4.6 dB claim is likely invalid. Given the paper's otherwise clean theoretical framework, a major revision with additional experiments seems appropriate."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague — you should know two things about this paper. The device is genuinely new: two cascaded SiN microrings, the second acting as a high-gain parametric amplifier to measure the squeezing from the first before off-chip loss. That is a nice piece of integration and the first monolithic SU(1,1) interferometer on this platform. The theory part is standard and clean. But the headline number — 4.6 dB of squeezing from R1 — does not survive a simple consistency check.\n\nFor a two-mode squeezed state amplified by a phase-sensitive amplifier, the max and min noise relative to the same vacuum reference should be reciprocal: in dB they sum to zero. The paper reports +13.2 dB max and -4.6 dB min, which sum to +8.6 dB. The ratio alone (17.8 dB) implies an input squeezing of about 8.9 dB, exceeding the paper's own 5.2 dB escape-efficiency upper bound. So the data are internally inconsistent with the interpretation. The most plausible resolution is that R2's gain is not the same in the 'vacuum reference' (Pump1 off) as in the dual-pump measurement. Thermal crosstalk from R1 could easily change R2's detuning and gain, especially since they operate just below OPO threshold. The paper does not check R2 gain stability between the two configurations, so the 4.6 dB figure is not a faithful measure of R1's squeezing. The loss-tolerance claim depends on Eq. (6), which assumes identical R2 gain in both traces; that assumption appears false.\n\nThere are also softer issues: the SU(1,1) SNR enhancement is estimated, not measured; loss calibration is given without uncertainties; and the abstract attributes the 4.6 dB to the first ring although inter-ring loss is included. None of these are fatal on their own, but the consistency problem is.\n\nStill, the concept is worth taking seriously. If the gain-stability issue can be settled — e.g., by measuring R2 gain with both pumps on, or by showing that the max/min sum to zero after proper reference — the loss-tolerant measurement paradigm on a chip would be a real step forward. As written, the central claim is not established.\n\nI would send this to review, because the device is novel and the flaw is addressable with additional measurements. A referee should require the reciprocity check and a direct measurement of R2 gain in the dual-pump condition. If the authors can provide that, the paper could be a solid contribution.","headline":"Nice device concept, but the 4.6 dB claim fails a reciprocity check, so the loss-tolerance result is not established as written.","tokens_in":11651,"tokens_out":8104,"would_cite":false,"duration_ms":73018,"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":"By placing a second silicon nitride micro-ring amplifier before the lossy chip-to-fiber interface, this paper demonstrates on-chip measurement of 4.6 dB of two-mode squeezing even when downstream losses exceed 7 dB.","keywords":["squeezed light","silicon nitride photonics","microring resonator","parametric amplification","on-chip quantum measurement","SU(1,1) interferometer","loss-tolerant detection","continuous-variable quantum optics"],"falsifier":"Turn on R1 below threshold while keeping Pump 2 fixed and monitor the R2 gain at 100 MHz with a weak seed; if the R2 gain or its phase response shifts by more than the quoted uncertainties when Pump 1 is switched on, the 4.6 dB value does not isolate R1's squeezing and the reference must be remeasured under identical on-chip conditions. A simpler check is to record the R2-alone noise trace before and after a period of R1 pumping and see whether the 10 dB-above-shot-noise level remains unchanged.","tokens_in":10576,"feed_emoji":"💡","tokens_out":6624,"duration_ms":55333,"temperature":0.7,"pith_summary":"This paper tries to establish that squeezed light generated in an on-chip micro-ring can be measured on the chip itself, before fiber-coupling losses destroy the quantum correlations. The scheme places a second, matched silicon nitride ring after the squeezer and runs it as a high-gain parametric amplifier; the amplifier lifts the fragile quantum noise to a high level that downstream loss cannot pull back down. With both rings driven just below their oscillation thresholds, the authors observe a 4.6 dB noise reduction relative to the equivalent vacuum reference, even though total loss from chip to detector is about 7 dB. They also show that the two rings form a monolithic SU(1,1) interferometer with a 98% fringe visibility and an estimated 5 dB signal-to-noise enhancement over a linear interferometer. If correct, the work removes one of the central obstacles to practical chip-based quantum sensing.","feed_headline":"Chip reads out 4.6 dB of squeezing through 7 dB of loss","feed_subtitle":"A second ring amplifies quantum noise before fiber coupling, making the measurement loss-resistant.","key_machinery":"The load-bearing element is the second ring operated as a phase-sensitive, high-gain non-degenerate optical parametric amplifier whose pump acts as the equivalent local oscillator. In the high-gain limit, the quadrature output follows $\\hat{X}_s^{\\rm out}(\\theta)\\approx G\\hat{X}_+(\\theta,\\phi_p)$, where $\\hat{X}_+$ is the sum of the signal quadrature and the corresponding idler quadrature; comparing squeezed-input to vacuum-input noise gives $R=\\langle\\Delta^2\\hat{X}_+\\rangle_{\\rm sq}/\\langle\\Delta^2\\hat{X}_+\\rangle_{\\rm vac}$. Because the amplified noise is about $2G^2R$ before any loss, the vacuum entering through loss $L$ is negligible, and the detected ratio $R_{\\rm det}\\approx R$. This noiseless amplification before detection is what carries the loss tolerance; the cascaded filters and heaters merely prepare matched operation of the two rings.","core_discovery":"The central claim is that a high-gain optical parametric amplifier can perform the quantum measurement of squeezing on-chip, making the result insensitive to downstream loss. For two-mode squeezed light generated by ring R1, ring R2 acts as a non-degenerate amplifier with gain $G$; in the large-gain limit the detected noise reduction $R_{\\rm det}$ becomes equal to the input squeezing $R$ and independent of the loss $L$ (Eq. 6). Using this architecture, the paper reports $4.6\\pm 0.4$ dB of measured squeezing and $13.2\\pm 0.4$ dB of anti-squeezing at 100 MHz, with only about 70% escape efficiency in the squeezer, and estimates a 5 dB quantum SNR enhancement of the resulting SU(1,1) interferometer. The experiment also shows the measured noise level of R2 alone sits about 10 dB above shot noise, consistent with roughly 17 dB on-chip gain, confirming the amplifier operates in the required high-gain regime.","pith_inferences":["A natural extension the paper does not explore is using the same amplifier-assisted readout to characterize other two-mode states, such as EPR-entangled or frequency-bin entangled light, on the same chip; the derivation only assumes a two-mode input and high gain.","A testable prediction is that as Pump 2 gain $G$ is increased, $R_{\\rm det}$ should stay approximately constant while the absolute noise floor rises by $20\\log_{10}G$; measuring that plateau would confirm the loss-tolerance mechanism.","The 5 dB SNR estimate relies on comparing signal gain with output noise, so a chip with on-chip phase modulation and a linear interferometer reference would turn this estimate into a direct measurement, which the authors say is planned.","Actively locking Pump 2's phase and the ring temperatures would likely reduce the $\\pm 0.4$ dB uncertainty in the reported squeezing and improve the stability of the monolithic interferometer."],"forward_implications":["On-chip squeezing can be characterized without monolithically integrated detectors, because the measurement happens in the same nonlinear platform as the source.","Chip-to-fiber coupling loss of 5 dB or more no longer caps the observed squeezing; the same device should preserve a fixed $R_{\\rm det}$ as long as the amplifier gain stays high.","The cascaded-ring device is a functional SU(1,1) interferometer on a chip, so phase sensing with quantum-enhanced SNR could be done in a compact CMOS-compatible circuit.","The 4.6 dB level is bounded by the squeezer's roughly 70% escape efficiency, so improving the ring coupling ratio should directly raise the measured squeezing toward the theoretical limit.","The measurement establishes a way to verify non-classical correlations at the point of generation, which is relevant for chip-based quantum sensors and continuous-variable quantum computing."],"supporting_citations":[{"why":"Supplies the founding idea that an amplifier placed before loss can make interferometric quantum-noise measurements loss-insensitive.","marker":"[20]"},{"why":"Provides the input-output relation for parametric-amplifier-assisted homodyne detection that the paper's Eq. (1) and noise-reduction derivation use.","marker":"[11]"},{"why":"Demonstrates amplifier-assisted detection of pulsed squeezing on a thin-film PPLN chip, the closest prior implementation this scheme extends to silicon nitride rings.","marker":"[13]"},{"why":"Introduces the SU(1,1) interferometer with two parametric amplifiers that the cascaded rings realize in monolithic form.","marker":"[14]"},{"why":"Shows quantum metrology enhancement in parametric-amplifier-based interferometers and its tolerance to losses, used for the SNR estimate.","marker":"[16]"},{"why":"Gives the theoretical phase-sensitivity enhancement beyond the standard quantum limit for nonlinear interferometers, the basis for the estimated 5 dB enhancement.","marker":"[18]"},{"why":"Demonstrates near-degenerate quadrature-squeezed vacuum generation on a silicon-nitride chip, establishing the source platform and escape-efficiency considerations.","marker":"[3]"},{"why":"Reports broadband squeezed vacuum from a nanophotonic device, providing the background on escape efficiency that bounds the observed squeezing.","marker":"[4]"}],"fun_headline_variants":["On-chip amplifier measures squeezed light before loss corrupts","Loss-tolerant chip measures 4.6 dB squeezing on-chip","Second ring amplifies squeezing measurement on chip","On-chip quantum measurement beats coupling loss","Chip squeezes, then amplifies to read quantum noise"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The measurement is only as good as the assumption that the noise of the second ring with Pump 1 off is the right empty-input reference for the output of the first ring, meaning that switching on the first ring does not change the second ring's gain or phase through thermal or refractive crosstalk.","fun_headline_variants_meta":{"raw":{"variants":["On-chip amplifier measures squeezed light before loss corrupts","Loss-tolerant chip measures 4.6 dB squeezing on-chip","Second ring amplifies squeezing measurement on chip","On-chip quantum measurement beats coupling loss","Chip squeezes, then amplifies to read quantum noise"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000476,"raw_usage":{"total_tokens":2399,"prompt_tokens":1023,"completion_tokens":1376,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":639,"completion_tokens_details":{"reasoning_tokens":1301}},"tokens_in":639,"tokens_out":1376,"duration_ms":10069,"temperature":1.0,"reasoning_tokens":1301,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T14:50:26.374005+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Turn on R1 below threshold while keeping Pump 2 fixed and monitor the R2 gain at 100 MHz with a weak seed; if the R2 gain or its phase response shifts by more than the quoted uncertainties when Pump 1 is switched on, the 4.6 dB value does not isolate R1's squeezing and the reference must be remeasured under identical on-chip conditions. A simpler check is to record the R2-alone noise trace before and after a period of R1 pumping and see whether the 10 dB-above-shot-noise level remains unchanged.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the founding idea that an amplifier placed before loss can make interferometric quantum-noise measurements loss-insensitive."},{"cited_title":"Li,et al., Measuring continuous-variable quantum entanglement with parametric-amplifier- assisted homodyne detection.Physical Review A101(5), 053801 (2020)","cited_arxiv_id":null,"evidence_quote":"Provides the input-output relation for parametric-amplifier-assisted homodyne detection that the paper's Eq. (1) and noise-reduction derivation use."},{"cited_title":"Nehra,et al., Few-cycle vacuum squeezing in nanophotonics.Science377(6612), 1333– 1337 (2022)","cited_arxiv_id":null,"evidence_quote":"Demonstrates amplifier-assisted detection of pulsed squeezing on a thin-film PPLN chip, the closest prior implementation this scheme extends to silicon nitride rings."},{"cited_title":"Yurke, S","cited_arxiv_id":null,"evidence_quote":"Introduces the SU(1,1) interferometer with two parametric amplifiers that the cascaded rings realize in monolithic form."},{"cited_title":"Hudelist,et al., Quantum metrology with parametric amplifier-based photon correlation interferometers.Nature communications5(1), 3049 (2014)","cited_arxiv_id":null,"evidence_quote":"Shows quantum metrology enhancement in parametric-amplifier-based interferometers and its tolerance to losses, used for the SNR estimate."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the theoretical phase-sensitivity enhancement beyond the standard quantum limit for nonlinear interferometers, the basis for the estimated 5 dB enhancement."},{"cited_title":"Zhao,et al., Near-degenerate quadrature-squeezed vacuum generation on a silicon-nitride chip.Physical Review Letters124(19), 193601 (2020)","cited_arxiv_id":null,"evidence_quote":"Demonstrates near-degenerate quadrature-squeezed vacuum generation on a silicon-nitride chip, establishing the source platform and escape-efficiency considerations."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reports broadband squeezed vacuum from a nanophotonic device, providing the background on escape efficiency that bounds the observed squeezing."}],"review_version":2}