{"id":"055aca15-fd5b-487c-8dfb-80a16e327c21","arxiv_id":"2608.08221","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":7,"one_line_summary":"A glue-free bonded erbium:CaWO4-on-lithium-niobate chip achieves millisecond optical coherence, collective cooperativity of 6.7, and 5-second optical phase storage at telecom wavelengths.","lead":"Researchers bonded an erbium-doped calcium tungstate crystal directly onto a lithium niobate photonic chip, without glue, and showed that the erbium ions keep narrow optical lines. The device combines strong light-matter coupling, millisecond-scale memory times, and electrical tuning at telecom wavelengths, a combination needed for quantum network nodes.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Coherence and strong coupling are measured on different Er3+ sub-ensembles; the combined claim assumes, without direct evidence, that ring-coupled ions retain the bus-waveguide 289 Hz effective linewidth.","rationale":"The reader's weakest_assumption identifies exactly the principal gap in the argument. My independent read found no additional load-bearing flaw: the cooperativity extraction is conservative (the authors note G is a lower bound because of optical pumping), the spectral-diffusion model is standard, and the 5 s visibility measurement is internally consistent. The concern is a missing measurement rather than an internal inconsistency, but the paper's own Supplementary Note 8 confirms that the two sub-ensembles have different local environments, so the coherence transfer is genuinely insecure. This supports a conditional acceptance pending a direct measurement of the coherence of the resonator-coupled ions.","tokens_in":34109,"tokens_out":4344,"duration_ms":48690,"concrete_test":"Burn a spectral hole through the ring resonator on transition A at |B|=0.2 T, using the electro-optic tuning to hold the resonator on the spin-preserving transition, and measure the hole width and its recovery versus waiting time. Compare the extracted homogeneous linewidth and spectral-diffusion parameters with the bus-waveguide values (Γ0=254±7 Hz, R=86±18 Hz, ΓSD=1.5±0.2 kHz). If the ring-coupled hole width is substantially larger than ~300 Hz, or if the spectral diffusion is faster, the central combined claim does not hold as stated.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section II.A characterizes optical coherence with 3PPE on ions coupled to the bus waveguide, and the authors explicitly avoid the resonator to exclude Purcell modification and strong-coupling nonlinearities. Section II.B measures C=6.7 on ions coupled to the ring. The central claim—millisecond coherence and strong coupling in one device—requires that the ring-coupled sub-ensemble, which sits closer to the bonded interface and is exposed to electrode fields and strain, has the same ~254 Hz homogeneous linewidth. Supplementary Note 8 shows the two sub-ensembles differ measurably: Γinh for resonator-coupled ions is smaller, and the authors attribute this to a spatially varying local environment. That established spatial variation undermines the automatic transfer of coherence values. No measurement in the paper probes homogeneous linewidth or spectral diffusion of the ring-coupled ions; the avoided-crossing line shapes are dominated by Γinh ~ 332 MHz and cannot constrain a ~kHz-scale homogeneous linewidth. The 5 s phase-storage experiment is also performed on the bus-coupled ensemble, so it does not rescue the combined claim. Without evidence that ring-coupled ions retain millisecond coherence, the headline assertion of simultaneous strong coupling and millisecond coherence is conditional.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports the heterogeneous integration of a 50 ppm Er3+:CaWO4 crystal directly bonded to an electro-optically tunable thin-film lithium niobate microring, with no adhesive interlayer. Using three-pulse photon echoes on ions coupled to the bus waveguide at 0.2 T, the authors extract a homogeneous linewidth Γ0 = 254 ± 7 Hz, an effective linewidth of 289 ± 34 Hz (T_M = 1.10 ± 0.13 ms), a spectral diffusion rate R = 86 ± 18 Hz, and a saturated spectral diffusion width ΓSD = 1.5 ± 0.2 kHz. Electro-optically tuning the ring through the erbium transition yields an avoided crossing described by a collective cooperativity C = 6.7 ± 0.4 at zero field, consistent with an independent estimate C = 6.0 ± 0.5 based on simulated mode participation and absorption-derived dipole moment. A 3PPE-based protocol stores optical phase for 5 s with visibility 0.935 ± 0.015. The paper concludes that strong collective coupling, millisecond optical coherence, and in situ spectral tuning are simultaneously available in one integrated device.","tokens_in":34389,"tokens_out":11544,"duration_ms":112707,"significance":"If the combined claim holds, this is a significant step for integrated telecom quantum memories: it would be the first device to combine C > 1 collective coupling with sub-kHz optical linewidth and seconds-scale phase storage in a scalable platform. The paper's independent estimate of the cooperativity from mode simulation and measured oscillator strength is a genuine cross-check rather than a restatement of the fit, and error bars on fitted parameters are reported throughout. The spectral diffusion analysis follows established models and is internally consistent, and the explicit statements about what remains to be demonstrated (high-efficiency AFC on quantum signals, Discussion; extracted G being a lower bound under optical pumping, Supp Note 7) are appropriate. The main weakness is that the coherence and the strong coupling are measured on different erbium sub-ensembles; whether the ring-coupled ions share the millisecond coherence is not established by the data as presented.","major_comments":[{"comment":"The central claim that the device simultaneously provides strong coupling and millisecond coherence requires transferring the coherence results from the bus-waveguide sub-ensemble to the ring-coupled sub-ensemble, and this transfer is not demonstrated. Section II.A deliberately probes ions coupled to the bus waveguide to avoid Purcell modification and strong-coupling nonlinearities, while Section II.B measures C = 6.7 on ions coupled to the ring. Supplementary Note 8 shows that the two sub-ensembles have measurably different inhomogeneous broadening (for example, the ring-coupled Γinh at 0 T is about 332 MHz, while the bus-coupled ensemble is broader) and attributes the difference to a spatially varying local environment. Because the avoided-crossing line shapes are dominated by the ~300 MHz inhomogeneous width, they cannot constrain a kHz-scale homogeneous linewidth. I request either a direct coherence measurement on the ring-coupled sub-ensemble (for example, a two-pulse or three-pulse echo through the resonator with excitation weak enough to avoid strong-coupling effects) or a quantitative argument with supporting data that the documented local-environment differences do not affect the homogeneous linewidth. Until this is provided, the headline should be qualified to state that the bus-coupled ions retain millisecond coherence while the ring-coupled ions exhibit strong coupling in the same bonded device.","section":"II.A, II.B and Supplementary Note 8"},{"comment":"The 5 s optical phase-storage visibility of 0.935 is obtained with the same bus-waveguide 3PPE arrangement used in Section II.A, not with the resonator-coupled ensemble. It therefore demonstrates long-lived shelving and phase retention for the bus-coupled ions, but it does not provide evidence that the strongly coupled ring-coupled sub-ensemble can store optical phase for seconds. The Discussion's statement that the platform simultaneously offers strong collective coupling, bulk-level optical coherence and on-chip resonator tuning should be qualified accordingly, or supported by a resonator-coupled storage measurement.","section":"II.C and Discussion"},{"comment":"The combined claim is anchored at the field where coherence is optimized (|B| = 0.2 T), but the reported strong-coupling parameters in Supplementary Table 3 are for 0, 0.1, 0.4, and 1.0 T. The text asserts that avoided crossings are observed for all fields between 0 and 1 T; given the emphasis on 0.2 T in the abstract and Fig. 2, the 0.2 T avoided-crossing fit and the corresponding C, G, and Γinh values should be reported explicitly to support operation of the same device at the coherence-optimal field.","section":"II.B and Supplementary Table 3"}],"minor_comments":[{"comment":"The abstract calls the 289 ± 34 Hz value the effective homogeneous linewidth, while Section II.A derives it as the empirical linewidth 1/πT_M from the spectral diffusion model; please use consistent terminology (effective linewidth for 1/πT_M and homogeneous linewidth for Γ0).","section":"Abstract and Section II.A"},{"comment":"In Fig. 2e, the left and right axes display the same dataset with reciprocal scalings; the caption and text should state this explicitly so readers do not interpret the two curves as independent measurements.","section":"Figure 2e"},{"comment":"The mode-volume axis in Supplementary Figure 7b is labeled in µm^2, which is dimensionally inconsistent with a three-dimensional mode volume; please correct the unit or clarify that the simulation is two-dimensional.","section":"Supplementary Note 7, Fig. S7b"},{"comment":"The main sample accumulated about one month of post-bond annealing while sample b was annealed for approximately 12 hours; the manuscript should state explicitly whether this difference affects the quoted bond strength, Q values, or the comparison between the two samples.","section":"Supplementary Note 3"},{"comment":"The heterodyne detection setup is described in the supplement as based on Ref. [27], which appears to be the same work as Ref. [44] in the main text; please harmonize the citation or add a cross-reference for clarity.","section":"Supplementary Note 5 and Ref. [44]"}],"recommendation":"major_revision","confidential_remarks":"I recommend major revision. The experimental work is careful, the error analysis is thorough, and the independent cooperativity estimate is a real strength. The single substantive gap is the sub-ensemble coherence transfer; if the authors can supply direct resonator-coupled coherence data or a convincing bound, the paper would be acceptable for this journal. There is no concern about novelty or scope. A minor fit-to-journal issue: the 0.2 T strong-coupling point should be added to support the simultaneous-claims narrative."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Worth a serious look. The device is new: adhesive-free bonded Er3+:CaWO4 on a TFLN microring, with C = 6.7, a 289 Hz effective linewidth, and 5 s phase storage. The strongest part is the internal consistency: the fitted cooperativity matches an independent estimate from mode simulation and absorption oscillator strength, so C is not a free-parameter artifact. The coherence analysis is standard and carefully done, with spectral diffusion parameters separated and checked against a linearized model. The literature comparison in Supplementary Table 6 is genuinely useful.\n\nThe real soft spot is the one the stress-test note identifies: coherence is measured on bus-coupled ions (Section II.A), and the 5 s storage also uses that ensemble, while strong coupling is measured on ring-coupled ions. These are different sub-ensembles, and Supplementary Note 8 shows the ring sees narrower inhomogeneous broadening—evidence that the local environment differs. The paper assumes the ring-coupled ions share the millisecond coherence, and that is an assumption, not a measurement. It is a plausible assumption, since the coherence of the bonded host should not depend strongly on which evanescent tail you sit in, but the headline 'simultaneously offers' outruns the data. This is fixable: a cavity-enhanced echo or low-power transmission linewidth on the ring-coupled ensemble would close the gap, or at least the paper could explicitly discuss why the transfer is safe.\n\nMinor issues: the spin temperature is fit rather than directly measured, which is acceptable given the good fit to two physical mechanisms. The self-citation for the heterodyne method is not circular; the method is described independently.\n\nOverall, the paper is honest, includes its limitations, and the central result is likely correct. It deserves peer review. I would recommend revision, with the required change being either a direct coherence measurement on the ring-coupled ensemble or a softening of the combined claim to 'coherence preserved in the bonded host and strong coupling in the ring.' If the authors can close that sub-ensemble gap, this becomes a clean demonstration of the platform.","headline":"Strong paper with a real new combination, but the headline 'simultaneous' claim relies on an unmeasured transfer of coherence between bus- and ring-coupled ions.","tokens_in":34975,"tokens_out":2770,"would_cite":true,"duration_ms":29704,"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 telecom-wavelength device can now combine millisecond erbium coherence, strong collective coupling, and electro-optic spectral tuning, by bonding a coherence-optimized crystal directly onto a lithium niobate microring.","keywords":["erbium","rare earth","quantum memory","collective cooperativity","thin-film lithium niobate","direct bonding","spectral diffusion","telecom C-band"],"falsifier":"Run a three-pulse photon echo on the ring-coupled ions themselves, for example by detecting the echo re-emitted into the ring, and compare the effective linewidth with the $289$ Hz measured on the bus; a substantially broader linewidth or a Purcell-broadened value would falsify the claim that one device holds both strong coupling and millisecond coherence.","tokens_in":33904,"feed_emoji":"📡","tokens_out":8068,"duration_ms":69115,"temperature":0.7,"pith_summary":"The paper aims to show that a quantum-network node at telecom wavelengths does not have to choose between good erbium coherence and integrated photonics. It bonds a bulk Er$^{3+}$:CaWO$_4$ crystal, a host chosen for its dilute, weakly magnetic nuclear-spin bath, directly onto an electro-optically tuneable thin-film lithium niobate microring, with no adhesive layer. On that single chip it reports an effective homogeneous linewidth of $289\\pm34$ Hz (memory time $T_\\text{M}=1.10\\pm0.13$ ms), a collective cooperativity of $C=6.7\\pm0.4$ seen as an avoided crossing, and 5-second optical phase storage at $0.935$ visibility via $^{183}$W superhyperfine shelving. If correct, this combines strong light-matter coupling, millisecond coherence, and in situ spectral tuning in one telecom device, a combination previously split across different platforms.","feed_headline":"Erbium chip keeps millisecond coherence and strong coupling","feed_subtitle":"Direct bonding keeps the crystal's coherence, while the tuneable microring provides the coupling and long-lived storage.","key_machinery":"The load-bearing element is adhesive-free direct bonding of an Er$^{3+}$:CaWO$_4$ crystal to a high-$Q$ electro-optically tuneable thin-film lithium niobate microring, which places the ions in the evanescent field while preserving the host's weakly magnetic nuclear-spin environment. Coherence is extracted with the three-pulse photon echo, fitted to the spectral-diffusion model $\\Gamma_\\mathrm{eff}=\\Gamma_0+\\tfrac12\\Gamma_\\mathrm{SD}(R\\tau_{12}+1-e^{-R\\tau_{23}})$. Strong coupling is quantified with a coupled resonator-ensemble transmission model assuming a Lorentzian inhomogeneous profile and the collective cooperativity definition $C=4G^2/(\\kappa_\\mathrm{tot}\\Gamma_\\mathrm{inh})$. Long storage rides on superhyperfine coupling to $^{183}$W nuclear spins, which act as long-lived shelving states in a three-pulse interference protocol.","core_discovery":"The central claim is that heterogeneous integration can preserve the coherence of an optimised rare-earth host while inheriting the scalability and tunability of thin-film photonics. In the bonded device, three-pulse photon echoes give a homogeneous linewidth $\\Gamma_0=254\\pm7$ Hz at 0.2 T, with spectral diffusion rate $R=86\\pm18$ Hz saturating at $\\Gamma_\\mathrm{SD}=1.5\\pm0.2$ kHz, yielding an empirical linewidth $\\tilde\\Gamma_\\mathrm{eff}=289\\pm34$ Hz. Electro-optically sweeping the ring through the spin-preserving transition resolves an avoided crossing fitted with collective cooperativity $C=6.7\\pm0.4$. The same ensemble stores optical phase in spectral gratings for 5 s with visibility $V=0.935\\pm0.015$, using the $^{183}$W nuclear-spin bath as shelving states.","pith_inferences":["We infer the decisive test the paper leaves open: carrying out the three-pulse echo on the ring-coupled sub-ensemble, rather than the bus-coupled one, would confirm that strong coupling and millisecond coherence inhabit the same ions.","The observed difference in inhomogeneous broadening between bus- and ring-coupled ions suggests strain variations across the bonded interface; if those variations also broadened the homogeneous linewidth, the quoted coherence would be an overestimate for the strongly coupled mode.","The same direct-bonding recipe could plausibly transfer to other coherence-optimised hosts, with electro-optic tuning serving as in situ frequency alignment for multi-node networks; quantifying bonding-induced loss would directly predict how much cooperativity and memory efficiency remain on the table."],"forward_implications":["A sub-kilohertz effective linewidth sustained over seconds makes kilohertz tooth spacings plausible in an atomic frequency comb, corresponding to memory delays two orders of magnitude beyond the microsecond range demonstrated in directly doped devices.","Operating the device in the overcoupled regime, where $C=1$ suffices, would give a high-efficiency impedance-matched echo memory with on-demand storage around a millisecond.","Because the field needed is only 0.2 T, a permanent magnet replaces multi-tesla magnets for comparable coherence, easing deployment as network nodes.","Enrichment in $^{167}$Er combined with zero first-order Zeeman points in CaWO$_4$ could lift the roughly one-second hyperfine coherence ceiling seen elsewhere, pointing to minute-scale spin-wave storage."],"supporting_citations":[{"why":"Supplies the three-pulse photon echo technique and spectral-diffusion model from which $\\Gamma_0$, $R$, and $\\Gamma_\\mathrm{SD}$ are extracted.","marker":"[34]"},{"why":"Provides the asymmetric plasma activation bonding method that produces the covalent, adhesive-free Er$^{3+}$:CaWO$_4$\\textendash TFLN interface.","marker":"[33]"},{"why":"Prior flip-chip integration of Er$^{3+}$:Y$_2$SiO$_5$ on TFLN with $C=0.36$, the baseline this work's $C=6.7$ improves on.","marker":"[30]"},{"why":"Prior Er$^{3+}$:Y$_2$SiO$_5$ on silicon carbide with $C=1.9$, the strong-coupling baseline for heterogeneous rare-earth integration.","marker":"[31]"},{"why":"Provides the coupled resonator-ensemble model with a Lorentzian inhomogeneous profile used to fit the avoided crossing and define $C$.","marker":"[41]"},{"why":"Establishes the impedance-matched memory criterion $C=1$ that motivates the device design and efficiency target.","marker":"[43]"},{"why":"Demonstrates minute- to hour-scale spectral holes in bulk Er$^{3+}$:CaWO$_4$ via $^{183}$W superhyperfine coupling, underpinning the 5 s shelving storage.","marker":"[27]"},{"why":"Reports a 1.8 kHz effective linewidth in directly doped Er:LiNbO$_3$, the integrated-coherence benchmark this device surpasses.","marker":"[19]"}],"fun_headline_variants":["Erbium chip: 1 ms coherence, C=6.7, 5 s storage","Telecom erbium device with millisecond coherence and strong coupling","Chip-scale erbium keeps coherence, enables strong coupling and storage","Ms coherence and strong coupling achieved in telecom erbium chip","Integrated erbium: strong coupling, millisecond coherence, 5s memory"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The millisecond coherence is measured on erbium ions coupled to the bus waveguide, while the strong coupling is measured on ions coupled to the ring, and the central demonstration assumes those two sub-ensembles share the same coherence.","fun_headline_variants_meta":{"raw":{"variants":["Erbium chip: 1 ms coherence, C=6.7, 5 s storage","Telecom erbium device with millisecond coherence and strong coupling","Chip-scale erbium keeps coherence, enables strong coupling and storage","Ms coherence and strong coupling achieved in telecom erbium chip","Integrated erbium: strong coupling, millisecond coherence, 5s memory"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000988,"raw_usage":{"total_tokens":4234,"prompt_tokens":1035,"completion_tokens":3199,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":651,"completion_tokens_details":{"reasoning_tokens":3108}},"tokens_in":651,"tokens_out":3199,"duration_ms":21288,"temperature":1.0,"reasoning_tokens":3108,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T00:15:41.222023+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run a three-pulse photon echo on the ring-coupled ions themselves, for example by detecting the echo re-emitted into the ring, and compare the effective linewidth with the $289$ Hz measured on the bus; a substantially broader linewidth or a Purcell-broadened value would falsify the claim that one device holds both strong coupling and millisecond coherence.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Demonstrates minute- to hour-scale spectral holes in bulk Er$^{3+}$:CaWO$_4$ via $^{183}$W superhyperfine coupling, underpinning the 5 s shelving storage."}],"review_version":1}