{"id":"3243920a-5e0f-4e50-81a6-7bebf59cf2d9","arxiv_id":"2605.14777","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"An erbium-167-doped lithium-niobate microring stores telecom photons at 23.3% on-chip efficiency, routes them electro-optically at 20 MHz, and preserves time-energy entanglement with an 11σ witness violation.","lead":"A Nanjing University team built a telecom-band quantum memory on an erbium-doped lithium-niobate chip that stores light at 23.3% on-chip efficiency, routes it electrically at up to 20 MHz, and preserves photon entanglement. A generalist should read it because memory, efficiency, and fast frequency control now sit together on one programmable integrated platform, a realistic building block for quantum repeaters.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 23.3% efficiency and impedance-matching narrative depend on a Fano-fit decomposition of cavity losses; the ion-free reference in Fig. 1D has Q=1.78e5 while the power sweep in Fig. 1F asymptotes to Q=1.5e5, leaving incomplete-saturation or wavelength-bias unaddressed.","rationale":"The reader's weakest assumption—the Fano decomposition of cavity losses—is the most load-bearing concern because all derived quantities (C, K, the theoretical efficiency curve) inherit it. I read the relevant sections carefully and found that the paper actually uses a separate ion-free reference from Fig. 1D (Q=1.78e5) rather than assuming full saturation at the highest power in Fig. 1F (Q≈1.5e5). This means the reader's specific phrasing about a 'high-power' reference being biased is not exact, but the underlying worry remains: the ion-free reference must be verified to be truly ion-free and wavelength-matched to the storage resonance. The discrepancy between Q=1.78e5 and Q=1.5e5 is a concrete signal that the saturation picture is incomplete. However, the measured 23.3% efficiency is a direct photon-count result and is not invalidated by this; the theoretical model and the impedance-matching narrative could shift, but the core experimental achievement is robust. The entanglement witness calculation is arithmetically sound, the F-dependence of efficiency is broadly consistent, and the other points raised by the reader (missing supplement, crosstalk error bars, AFC lifetime fit, relation to ref. 25) are less load-bearing. Therefore I agree with the CONDITIONAL verdict and recommend no change.","tokens_in":13172,"tokens_out":19944,"duration_ms":175944,"concrete_test":"Re-measure the transmission of the exact resonance used in the Fig. 1F power sweep on DEV1 at on-chip powers extending to at least 10–100× the current maximum, and fit Q_loaded(P) with a saturable-absorber model: 1/Q(P) = 1/Q_ionfree + A/(1 + P/P_sat). Extract the asymptotic Q_ionfree and compare it with the 1.78×10^5 used in the paper. If Q_ionfree differs by more than 5%, recompute Eq. (1) with the corrected κ_ions and κ_loss; if the predicted efficiency moves by more than 2 percentage points, the reported impedance-matching interpretation and the 24.6% prediction require revision, and the 23.3% measured efficiency should be re-benchmarked against the corrected model.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central efficiency and cooperativity claims rest on the decomposition {κ_ext=991 MHz, κ_loss=119 MHz, κ_ions=1778 MHz} extracted from Fano fits. The paper uses the Q=1.78×10^5 resonance in Fig. 1D as the ion-free reference, but the power-dependent measurement in Fig. 1F (same nominal resonance) only reaches Q≈1.5×10^5 at the highest power, implying either incomplete ion saturation or a wavelength/condition mismatch between the two references. If the true ion-free Q at the storage resonance is lower than 1.78×10^5 (i.e., κ_loss is larger or κ_ions is smaller than reported), then the effective cooperativity C and the 'close to critical coupling' narrative shift, and Eq. (1) would no longer predict the 24.6% efficiency said to match the measured 23.3%. Because the 23.3% figure itself is directly measured from photon counts, this does not invalidate the headline efficiency, but it does undermine the impedance-matching explanation and the quantitative theoretical model that ties the observed efficiency to the cavity parameters. The supplementary controls (S2, S3) are cited but not available for inspection here.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports an integrated quantum memory in isotopically purified 167Er3+-doped thin-film lithium niobate (TFLN) microring resonators. The authors demonstrate cavity-enhanced atomic frequency comb (AFC) storage with a measured on-chip efficiency of 23.3±0.5% for a 100-ns delay, a persistent AFC lifetime of 277.6±52.6 s, electro-optic frequency-selective routing with crosstalk below 10^-4 at modulation rates up to 20 MHz, and storage/retrieval of time-energy-entangled telecom photons, with an entanglement witness violated by more than 11 standard deviations. The central physical picture is that the cavity is impedance-matched via the erbium ensemble absorption, and the storage efficiency is described by Eq. (1), which combines the cavity coupling ratio, cooperativity, comb finesse, and spectral preparation efficiency. The measured efficiency is quoted as matching the model prediction of 24.6% when independently measured parameters are inserted.","tokens_in":13360,"tokens_out":10813,"duration_ms":101756,"significance":"If correct, this work would be a significant advance: it integrates efficient cavity-enhanced quantum storage in the telecom band with fast on-chip electro-optic programming in a single monolithic TFLN platform, addressing a long-standing gap for spectrally multiplexed quantum networks. The direct measurements—23.3% on-chip efficiency, sub-10^-4 crosstalk routing, and entanglement preservation without background subtraction—are impressive and important. The internal cross-check of Eq. (1) using independently measured cavity parameters (κ_ext, κ_loss, κ_ions, η_spectral) strengthens the central efficiency claim. The demonstration of entanglement storage with a raw-data witness violation is a notable strength. However, the quantitative interpretation of the efficiency rests on a Fano-fit decomposition of cavity losses whose internal consistency is not fully established (see major comments).","major_comments":[{"comment":"The ion-free cavity loss rate is stated to be determined from Fig. 1D with Q_loaded = 1.78×10^5, while the power sweep of the same resonance in Fig. 1F asymptotes to Q_loaded ≈ 1.5×10^5 at the highest input power. This 16% discrepancy is not explained. If the high-power data represent the true ion-saturated limit, then κ_total = κ_ext + κ_loss ≈ 2π×1.30 GHz instead of 2π×1.11 GHz, and the deduced κ_ions becomes ≈2π×1.57 GHz rather than 2π×1.78 GHz. Recomputing Eq. (1) with these values yields an efficiency of roughly 13%, not the 24.6% quoted as the model prediction. The paper must clarify the measurement conditions of Fig. 1D (e.g., whether it was taken at a wavelength outside the inhomogeneous absorption profile, after optical pumping, or at a different probe power), and provide a power-dependent model consistent with both the Q and extinction-ratio data across the full range.","section":"Device Design, Fig. 1D/F and Eq. (1)"},{"comment":"The theoretical curve in Fig. 2C is said to assume η_spectral = 0.95 and the cavity parameters extracted from Fano fits, but the manuscript does not report the Fano fit parameters, the complex-coupling phase, the uncertainties in κ_ext, κ_loss, and κ_ions, or the sensitivity of the predicted efficiency to these values. Without this information, the claimed agreement between the measured 23.3±0.5% and the model cannot be independently verified. Please provide the full fitting details, a sensitivity analysis (e.g., the predicted efficiency as a function of κ_ions/κ_total within its confidence interval), and explicitly state whether η_spectral or any other parameter was adjusted to match the data. This is load-bearing because the impedance-matching narrative and the quantitative model tie the measured efficiency to the cavity parameters.","section":"Quantum Storage, Eq. (1) and Fig. 2C"}],"minor_comments":[{"comment":"The AFC lifetime is given as '277.6(52.6) s' in the abstract but '277.6±52.6 s' in the body; please use a consistent uncertainty notation.","section":"Abstract and body"},{"comment":"The displayed equation is not cleanly typeset; the bracket structure is ambiguous. Please format it properly with clear parentheses and ensure it matches the standard Afzelius–Simon form.","section":"Eq. (1)"},{"comment":"The caption lists 'theoretical efficiency curve (blue solid line), coupling parameter K (green dashed line), and effective cooperativity C′' but does not specify which curve corresponds to which axis or provide a legend in the main text. Please label the curves or add a legend.","section":"Fig. 2C caption"},{"comment":"The statement 'It is intriguing that the coherence of the erbium ensemble remains unperturbed during the reconfiguration of cavity resonances' is an interpretation; please either provide a control measurement comparing echo efficiency with a static cavity or rephrase it as 'the retrieved echo indicates that the stored coherence survives the cavity reconfiguration'.","section":"High-speed routing section"},{"comment":"The term 'on-chip efficiency' is used without a formal definition in the main text. Please state explicitly how the off-resonance input reference is defined and how coupling losses are excluded.","section":"Methods/on-chip efficiency definition"}],"recommendation":"major_revision","confidential_remarks":"The direct measurements (23.3% efficiency, 10^-4 crosstalk, 11σ entanglement witness) are compelling and appear credible. The main obstacle is the internal inconsistency between the ion-free reference (Q=1.78e5, Fig. 1D) and the high-power saturation asymptote (Q=1.5e5, Fig. 1F), which affects the model prediction in Eq. (1) and the impedance-matching interpretation. I recommend major revision; if the authors can resolve this discrepancy and provide the full Fano-fit parameters and a sensitivity analysis, the paper could become a strong candidate for acceptance. I would also like to see the supplementary materials, as many claims rely on sections S1–S16."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Hi,\n\nThis one is worth your time. The headline 23.3±0.5% on-chip storage efficiency is a direct photon-count measurement, and the 11-sigma entanglement-witness violation reproduces from their quoted g^(2)(0) and visibility. That part is solid. What is genuinely new: an integrated 167Er:TFLN microring that combines cavity-enhanced AFC storage at telecom with 277-s persistent shelving, electro-optic routing up to 20 MHz, and storage of time-energy entangled photons. Earlier integrated erbium devices (Craiciu et al.) didn't reach impedance matching, and natural-abundance Er:LN was stuck below 3%. This platform changes that.\n\nThe main theoretical curve also cross-checks: plugging their κ values, finesse, and spectral preparation into Eq. (1) gives ~24.6%, matching the measured efficiency without fitting. That's a good sign, but note the parameters are not fully independent—they come from the same cavity-loss decomposition that feeds the model.\n\nThe soft spot is exactly the one the stress test flagged. They use Q=1.78e5 from Fig. 1D as the ion-free reference, but their own power sweep in Fig. 1F saturates around Q=1.5e5 at the highest input power. Either the ions aren't fully saturated at that power, or the two references are taken under different conditions. That discrepancy directly affects the extracted κ_ions and therefore the cooperativity and the 'close to critical coupling' language. It does not move the measured 23.3%, so the core result survives, but the impedance-matching explanation is softer than presented. A referee should ask for the raw data and a clear statement of how the ion-free reference was obtained.\n\nFour smaller things. Sixteen supplementary sections are cited but not available for checking. The <1e-4 crosstalk figure is quoted without an error bar. The 277.6-s lifetime is a single-exponential fit over less than one lifetime of data, so the ±52.6 s error likely understates the uncertainty. And the entanglement protocol's relation to the group's earlier PRL [25] is never spelled out, which makes it hard to separate new from reused.\n\nThis is a strong experimental paper that deserves a serious referee. Send it to review, with a request for the supplement and a direct explanation of the Q discrepancy. I'd cite it.","headline":"Strong integrated erbium-TFLN memory; measured efficiency and entanglement witness hold up, but the impedance-matching model relies on a cavity-loss decomposition with a real Q discrepancy.","tokens_in":14109,"tokens_out":3604,"would_cite":true,"duration_ms":34567,"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":"A single erbium-doped microring on thin-film lithium niobate stores telecom photons with 23% on-chip efficiency for 100 ns, routes them by fast electrical control, and preserves their time-energy entanglement.","keywords":["quantum memory","atomic frequency comb","erbium","thin-film lithium niobate","telecom band","cavity impedance matching","electro-optic tuning","time-energy entanglement"],"falsifier":"Measure the storage efficiency under a second, independently calibrated cavity coupling condition (e.g., a device with a different bus-ring gap) and compare with the prediction of Eq. (1) using the extracted κ_ions = 1778 MHz and κ_loss = 119 MHz; a systematic disagreement would indicate that the loss decomposition is wrong. Alternatively, a power-dependent transmission curve that does not asymptotically flatten at high power would show that saturation is incomplete.","tokens_in":12856,"feed_emoji":"⚛️","tokens_out":7022,"duration_ms":65639,"temperature":0.7,"pith_summary":"This paper reports a single integrated device—an erbium-doped thin-film lithium niobate microring—that combines three functions previously split across separate experiments: efficient storage of telecom photons (23.3% on-chip efficiency for 100 ns), fast electrical frequency selection and routing (up to 20 MHz with crosstalk below 10^-4), and preservation of time-energy entanglement during storage. The authors show that cavity impedance matching between the microring and the erbium ensemble boosts storage efficiency, while the material's electro-optic response allows the cavity resonance to be tuned electrically at high speed. This establishes erbium-doped TFLN as a practical programmable light–matter interface for spectrally multiplexed quantum networks. The demonstration uses a short 100-ns delay, and the authors state that longer storage is the next key challenge.","feed_headline":"One microring stores telecom photons at 23% and routes them electrically","feed_subtitle":"Erbium-doped lithium niobate chip unites 23% storage, fast electrical routing, and entanglement in one device.","key_machinery":"The device is a racetrack microring resonator in isotopically purified 167Er3+-doped thin-film lithium niobate, coupled to a bus waveguide. Three mechanisms carry the argument: (1) atomic frequency comb (AFC) storage, where a periodic absorption spectrum is burned into the inhomogeneously broadened erbium transition and re-emits the photon after a fixed delay; (2) cavity impedance matching, where the external coupling rate balances the sum of intrinsic loss and ion absorption to maximize retrieval; and (3) the electro-optic (Pockels) effect, which tunes the cavity resonance by applied voltage. The AFC provides the storage protocol, the cavity provides the efficiency enhancement, and the elec","core_discovery":"The central claim is that a 167Er3+-doped thin-film lithium niobate microring can serve as a programmable telecom quantum memory. By preparing a persistent atomic frequency comb using hyperfine shelving states (comb lifetime 277.6 s) and matching the cavity's external coupling to the ion ensemble loss, the device stores photons with 23.3±0.5% on-chip efficiency for 100 ns. Using the Pockels effect, the cavity resonance is shifted electro-optically, enabling frequency-selective storage and routing at rates up to 20 MHz with inter-channel crosstalk below 10^-4. The memory also stores time-energy-entangled telecom photons, violating an entanglement witness by more than 11 standard deviations, c","pith_inferences":["If the impedance-matching model survives independent checks, the same design recipe (isotopically purified rare-earth doping + TFLN microring + electrodes) could be applied to other rare-earth ions to build memories at different wavelengths.","The high-speed cavity tuning demonstrated here suggests a new control primitive: shifting the cavity frequency mid-storage without disturbing the ion coherence could be used for on-demand release or spectral shaping, beyond simple routing.","A direct test of the saturation assumption—measuring storage efficiency while varying the cavity coupling strength—would sharpen the confidence in the reported efficiency numbers."],"forward_implications":["Spectrally multiplexed quantum networks become feasible with a single chip that stores many frequency channels and routes them electrically.","The memory can lock onto fixed-frequency sources (such as the entangled photon source used here) by electro-optic tuning, removing the need for slow thermal or mechanical stabilization.","Temporal multiplexing of up to 18 modes within 200 ns is demonstrated on the same device, increasing the effective rate of a quantum repeater node.","With reduced propagation losses and deeper hole burning, the authors project storage efficiencies exceeding 70%."],"fun_headline_variants":["Microring stores telecom photons with 23% efficiency on a chip","Erbium microring stores photons with 23% efficiency and 20 MHz routing","Chip-based quantum memory: 23% storage, 20 MHz routing, and entanglement","Erbium-doped microring: 23% efficient storage and 20MHz electrical routing"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The reported 23.3% efficiency and the extracted ion-cavity cooperativity assume that at the highest probe power the erbium ions are completely saturated, so the measured high-power cavity loss is purely the ion-free loss; any residual ion absorption at that reference power would shift all derived efficiencies.","fun_headline_variants_meta":{"raw":{"variants":["Microring stores telecom photons with 23% efficiency on a chip","Erbium microring stores photons with 23% efficiency and 20 MHz routing","Chip-based quantum memory: 23% storage, 20 MHz routing, and entanglement","Erbium-doped microring: 23% efficient storage and 20MHz electrical routing"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000791,"raw_usage":{"total_tokens":3317,"prompt_tokens":733,"completion_tokens":2584,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":477,"completion_tokens_details":{"reasoning_tokens":2492}},"tokens_in":477,"tokens_out":2584,"duration_ms":16657,"temperature":1.0,"reasoning_tokens":2492,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T14:01:49.526645+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the storage efficiency under a second, independently calibrated cavity coupling condition (e.g., a device with a different bus-ring gap) and compare with the prediction of Eq. (1) using the extracted κ_ions = 1778 MHz and κ_loss = 119 MHz; a systematic disagreement would indicate that the loss decomposition is wrong. Alternatively, a power-dependent transmission curve that does not asymptotically flatten at high power would show that saturation is incomplete.","supporting_citations":[],"review_version":2}