{"id":"1fc58eff-aeb4-440e-9cc6-425b7b66bcd6","arxiv_id":"2505.06444","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":9,"one_line_summary":"A shallow-lattice 87Sr clock reaches roughly 118 s coherence time and 1.5e-18 fractional instability at 1 s, with decoherence traced to Raman scattering and spectator-atom collisions.","lead":"This paper reports a strontium optical lattice clock with an atomic coherence time of about two minutes and a measured single-clock instability of 1.5e-18 at one second. The result comes from reducing lattice photon scattering and collision effects in a shallow, low-density Wannier-Stark lattice.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Extrapolation to U0=0 relies on fits at U0>50 Er; a bias here would undermine the 'fundamental limit' interpretation of the 118 s coherence.","rationale":"The reader's weakest_assumption focuses on the zero-density and zero-lattice-depth extrapolation. I agree that the zero-lattice-depth extrapolation is the load-bearing step, but I would separate it from the zero-density extrapolation: the headline 118(9) s is measured directly at Nsite=9, not extrapolated to zero density; the density fit only serves to extract γ0. The critical step is the comparison of γ0 to (Γ_R U + Γ_nat + Γ_BBR)/2 and to (Γ_nat + Γ_BBR)/2 as U→0. Both bands rely on Γ_R and Γ_nat obtained from Fig. 1, where the fit is restricted to U0>50 Er. The authors' own justification for this restriction—'Raman scattering induced loss and lattice intensity noise'—means the shallow region is exactly where the model may be unreliable. Since the headline coherence time is at 11 Er, the interpretation that the clock is near the spontaneous-emission limit is an extrapolation. The independent confirmation of the lifetime by previous measurements mitigates the lifetime bias but does not validate the slope Γ_R below 50 Er. If Γ_R were different at shallow depths, the blue band in Fig. 2(d) would shift and the 118 s could be farther from the fundamental limit than claimed. I also note the internal inconsistency in the quoted coherence-time uncertainty (118(18) s in the text vs 118(9) s in the abstract and Fig. 2 caption); this should be clarified, but it does not change the core concern. A targeted measurement at U0=30-40 Er or a re-analysis including the excluded data would settle whether the extrapolation is valid. I therefore recommend keeping the reader's CONDITIONAL verdict (UNCHANGED), with the condition that the extrapolation be justified or the claim be softened.","tokens_in":10202,"tokens_out":12717,"duration_ms":125623,"concrete_test":"Measure Γ_L(U0) at intermediate depths (e.g., U0 = 30 and 40 Er) with the same population-decay protocol, or include the previously excluded U0<50 Er data with an explicit model for Raman-induced loss and intensity noise. If the new points deviate from the linear fit in Fig. 1(c) by more than the quoted uncertainty, the U→0 extrapolation of Γ_L(0) and the derived 174(28) s lifetime are not supported. Alternatively, measure the Ramsey coherence time γ0 at U0 ≈ 5 Er and test whether it follows (Γ_R U + Γ_nat + Γ_BBR)/2 within 2σ.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that the 118(9) s coherence time at U0=11Er 'approaches the fundamental limit' depends on the gray band in Fig. 2(d), which is (Γ_nat + Γ_BBR)/2. The natural lifetime 1/Γ_nat = 174(28) s and the Raman slope Γ_R = 5.6(3)×10^-4 s^-1 Er^-1 used in the blue band come from the population-decay fit in Fig. 1 and Table I, where the authors deliberately restrict the fit to U0 > 50 Er (Fig. 1 caption: 'to avoid complications of the model such as Raman scattering induced loss and lattice intensity noise'). Since the headline coherence time is measured at U0 = 11 Er, far below this fit range, the comparison of γ0 to the model is an extrapolation. If the shallow-lattice dynamics introduce additional decoherence or a nonlinear U-dependence (e.g., delocalization, off-site interactions, or intensity noise), then the extrapolated Γ_L(0) and Γ_R could be biased, shifting the gray and blue bands. The observed γ0 at U0=11 Er would then no longer be 'approaching' the claimed fundamental limit. The agreement of the extracted 174(28) s with previous independent lifetime measurements [20,34,35] is reassuring, but it does not validate the linear-in-U model below 50 Er, because the lifetime was inferred from the same extrapolated intercept.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports a systematic study of decoherence and stability in a 87Sr Wannier-Stark optical lattice clock. The authors measure lattice-depth-dependent population decay rates, extract a 3P0 clock-state lifetime of 174(28) s, and study Ramsey contrast decay as a function of dark time, atom number per site, and lattice depth. At U0=11Er and Nsite=9 they report a coherence time of about 118 s, which they argue approaches the spontaneous-emission and blackbody-radiation limit. Using imaging spectroscopy, they also report a single-clock instability of 1.5e-18 at 1 s under Tdark=7 s with 313 realizations. The paper attributes the observed decoherence to lattice Raman scattering, collisions involving spectator atoms, and density-dependent interactions, and supports the density-dependent rates with DDTWA simulations.","tokens_in":10554,"tokens_out":5564,"duration_ms":52760,"significance":"If the results hold, the demonstrated 118 s coherence time at a usable atom number is a major step toward spontaneous-emission-limited operation of an optical lattice clock, and the 1.5e-18 at 1 s single-region instability is a notable precision result in a shallow lattice. The paper is careful in many ways: density-dependent rates are extracted from in-situ images, the instability is a measured quantity rather than an extrapolated one, the presence of an unexplained atom-number-independent noise floor is explicitly acknowledged, and DDTWA provides an independent theoretical benchmark. The main qualification is that the 'fundamental limit' interpretation rests on a linear extrapolation in lattice depth from fits deliberately restricted to U0>50Er, applied at U0=11Er and at U0=0.","major_comments":[{"comment":"The coherence time at U0=11Er and Nsite=9 is reported as 118(18) s in the body text (the paragraph following Fig. 2(a)) but as 118(9) s in the abstract and in the Figure 2(a) caption. Because this quantity is the headline result of the paper, the inconsistency must be resolved; please state which uncertainty is correct and update all occurrences consistently.","section":"Coherence time, Fig. 2(a) caption, and Abstract"},{"comment":"The lifetime 1/Γnat = 174(28) s and the Raman slope ΓR = 5.6(3)e-4 s^-1 Er^-1 used in Fig. 2(d) come from population-decay fits that the authors deliberately restrict to U0 > 50Er, while the comparison with γ0 is made at U0 = 11Er and the fundamental-limit claim involves the extrapolation to U0→0. The caption to Fig. 1 states that U0 < 50Er data are excluded to avoid complications such as Raman-scattering-induced loss and lattice intensity noise, but those same effects could introduce a nonlinear dependence of ΓL and ΓR below 50Er. The agreement of the extracted Γnat with previous lifetime measurements is reassuring but does not validate the linear model below 50Er, because the lifetime is inferred from the same extrapolated intercept. Please provide an additional test of the shallow-depth regime, for example population-decay or coherence measurements at intermediate depths such as 20-40Er, or alternatively soften the claim that the 118 s coherence time approaches the fundamental limit.","section":"Population decay, Fig. 1 caption, Table I, and Fig. 2(d)"},{"comment":"The fit model C(Tdark)=C(0) exp[-(γ Tdark)^α] is used to define the coherence time as γ^-1, but the main text does not report α or C(0) for the key data points. If α differs significantly from 1, γ^-1 is not the mean decay time of the stretched exponential, so the interpretation of 'coherence time' depends on a convention that should be stated. Please report the fitted α and C(0), or explicitly define γ^-1 as the 1/e time.","section":"Coherence time, stretched-exponential model"}],"minor_comments":[{"comment":"The prefactor κ in the two-body loss term is not defined in the main text; please define it or refer to the supplement at the point of first use.","section":"Eq. (1)"},{"comment":"The nonlinearity of γ versus Nsite is accounted for by splitting the fit range, but the resulting systematic uncertainty added to γ0 and γN is not reported numerically; please state these extra uncertainties explicitly.","section":"Fig. 2(b) and Fig. 2(e)"},{"comment":"The text says the density-dependent decoherence is 'supported by' DDTWA but does not provide a quantitative agreement metric in the main text; please give the discrepancy or a goodness-of-fit measure, or state where in the supplement it appears.","section":"Fig. 2(e) and DDTWA"},{"comment":"The statement that Eq. (2) is a good approximation to the classical Cramer-Rao bound 'for large atom numbers and not so small ϕ' should be made more precise; a citation to the relevant bound would help the reader judge when the approximation is valid.","section":"Imaging spectroscopy, Eq. (2)"},{"comment":"The figure captions would be clearer if the number of independent experimental realizations per data point and the meaning of the shaded uncertainty bands (68% confidence interval of the fit versus propagated uncertainty of the model) were stated explicitly.","section":"Figures 1 and 2"}],"recommendation":"major_revision","confidential_remarks":"The work is experimentally substantial and likely of interest to the atomic-clock community. The main reservations are the inconsistent reporting of the headline coherence time and the extrapolation issue around U0<50Er; both need to be addressed before publication. I do not see evidence of unsupported novelty claims: the instability result is measured and the excess noise is acknowledged. The manuscript appears to fit the journal's scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The two things to know: the measured numbers are real—118(9) s coherence at U0=11 Er and 1.5e-18 at 1 s single-clock instability—and the paper's main defense of them, the quantum model for spectator-atom collisions, is sound. The soft spots are the extrapolation used to claim the coherence time approaches the natural lifetime limit, and an internal inconsistency in the reported error bar.\n\nWhat's genuinely new: a measured two-minute coherence time in a many-atom lattice clock, with a clear density-dependence study and a dissipative TWA simulation that captures the spectator-atom mechanism. The instability measurement via imaging spectroscopy is careful, and the authors are honest that the observed noise floor sits 50% above the QPN estimate with an unidentified source. That's good experimental practice.\n\nThe soft spots, in proportion. First, the body text reports γ^{-1}=118(18) s for the same dataset while the abstract and Fig. 2 caption say 118(9) s. That needs to be reconciled before this goes to press. Second, the 'fundamental limit' interpretation rests on extrapolating the lattice-depth dependence from fits restricted to U0>50 Er down to U0=0. The headline coherence measurement itself is at 11 Er and doesn't rely on that extrapolation, but the claim that γ0 matches (Γ_nat+Γ_BBR)/2 does. The authors cite Raman loss and intensity noise as reasons for the fit restriction; if those effects have a nonlinear U-dependence below 50 Er, the agreement could be fortuitous. The agreement of the extracted 174(28) s lifetime with independent measurements is reassuring, but it doesn't validate the linear model at shallow depth. This weakens the interpretation but not the data. Third, the missing supplemental material makes it hard to check the DDTWA details and the ellipse-fitting tests; that's a practical issue for the referee, not a red flag.\n\nWho should read it: anyone working on optical lattice clock stability, coherence engineering, or quantum sensing with clocks. It's a solid experimental paper from a leading group, with a new measured result and a plausible mechanism. I'd send it to a serious referee. The referee should ask for the error-bar fix, a caveat on the extrapolation, and the supplemental material.\n\nRecommendation: engage with it.","headline":"Measured 118 s coherence and 1.5e-18 instability are solid; the 'fundamental limit' extrapolation and an internal error-bar mismatch need attention before publication.","tokens_in":11105,"tokens_out":2555,"would_cite":true,"duration_ms":24157,"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":"In a shallow Wannier-Stark strontium lattice clock, atomic coherence reaches 118(9) seconds at low density and the single-clock instability reaches $1.5\\times10^{-18}$ at 1 second.","keywords":["optical lattice clock","strontium-87","Wannier-Stark lattice","atomic coherence","Raman scattering","blackbody radiation","clock instability","quantum projection noise"],"falsifier":"Measure the zero-density Ramsey contrast decay rate $\\gamma_0$ at lattice depths between 10 and 40 $E_r$ and compare it with the extrapolated line: if $\\gamma_0$ deviates upward as $U_0$ drops, the lifetime inference is biased. Independently, measure the ${}^3P_0$ lifetime with a single atom in a magic-wavelength trap; agreement with 174(28) s would confirm the population-decay analysis, while a significantly different value would break the link between the measured Raman pumping rate and the fundamental lifetime limit.","tokens_in":10005,"feed_emoji":"🕰️","tokens_out":11285,"duration_ms":98257,"temperature":0.7,"pith_summary":"An optical lattice clock built from strontium-87 atoms in a shallow, gravity-tilted Wannier-Stark lattice holds atomic coherence for 118(9) seconds at nine atoms per site, close to the fundamental limit set by spontaneous emission and blackbody radiation. The same system reaches a single-clock fractional instability of $1.5\\times10^{-18}$ at one second of averaging. The paper establishes how lattice Raman scattering and atomic collisions set these limits: scattered photons create spectator atoms that collide with clock atoms, and this density-dependent decoherence can be pushed down by operating at shallow lattice depth and low density. It also reports a ${}^3P_0$ clock-state lifetime of 174(28) seconds, complementing the coherence measurement. The result matters because it shows that a many-atom clock can approach the atomic lifetime limit while preserving the atom number needed to suppress quantum projection noise.","feed_headline":"Clock coherence reaches 2 minutes; instability 1.5e-18 at 1 s","feed_subtitle":"Strontium atoms in a shallow lattice approach the fundamental spontaneous-emission limit.","key_machinery":"The argument runs through three devices: a rate-equation model for population decay that separates single-body loss ($\\Gamma_e,\\Gamma_g$), Raman pumping ($\\Gamma_L$), and two-body inelastic loss ($\\tilde\\Gamma_{ee}$); a linear density extrapolation $\\gamma=\\gamma_0+\\gamma_N N_{\\mathrm{site}}$ that isolates the single-atom coherence rate; and an extrapolation of $\\gamma_0$ in lattice depth $U_0$ to zero. A Wannier-Stark lattice, a one-dimensional optical lattice tilted by gravity with roughly 260 $\\mu$m site spacing, lets the clock run at only a few recoil energies $E_r$, suppressing Raman scattering, while in-situ imaging spectroscopy with ellipse fitting converts spatial contrast into a frequency-difference estimate for the instability. A dissipative discrete truncated Wigner approximation supplies the theoretical $\\gamma_N$ that supports the spectator-atom collision picture.","core_discovery":"The paper's central claim is that in a shallow Wannier-Stark lattice of ${}^{87}\\mathrm{Sr}$ at $U_0=11E_r$ and $N_{\\mathrm{site}}=9$, the Ramsey coherence time reaches 118(9) s, and the inferred zero-density, zero-lattice-depth decoherence rate matches $(\\Gamma_{\\mathrm{nat}}+\\Gamma_{\\mathrm{BBR}})/2$, so the clock runs at the fundamental atomic limit. Guided by this understanding, a single clock region demonstrates $\\sigma_y(1\\,\\mathrm{s})=1.5\\times10^{-18}$ under $T_{\\mathrm{dark}}=7$ s with 313 realizations. The density-dependent decoherence is traced to spectator atoms produced by lattice Raman scattering, which undergo strong s-wave collisions with atoms in the clock state; a dissipative discrete truncated Wigner simulation reproduces the measured $\\gamma_N$. The paper argues that the combination of shallow lattice depth and moderate atom number optimally balances Raman scattering, collisional decoherence, and quantum projection noise.","pith_inferences":["The agreement with $(\\Gamma_{\\mathrm{nat}}+\\Gamma_{\\mathrm{BBR}})/2$ is the strongest evidence for the paper's thesis, but it rests on a fit that excludes lattice depths below $50E_r$; direct measurements there would settle whether the extrapolation is valid.","If spectator atoms are the main density-dependent decoherence channel, then detecting and removing atoms that have Raman-scattered before the Ramsey readout could recover long coherence at high density.","The atom-number-independent instability floor indicated by $b=1.8\\times10^{-18}$ suggests a technical noise source in imaging or ellipse fitting; removing it would likely push the single-clock instability below $10^{-18}$ at 1 s.","Because the interaction nulling near $11E_r$ already controls $\\gamma_N$, applying spin squeezing in the same shallow lattice could reduce quantum projection noise without the density cost it normally brings."],"forward_implications":["The single-clock instability follows $\\sigma_y(\\tau)=1.5\\times10^{-18}/\\sqrt{\\tau/\\mathrm{s}}$ for averaging times up to $10^3$ s, so longer integration continues to improve precision.","The measured $\\gamma_N$ shows that spectator atoms, not direct excited-state collisions, dominate density-dependent decoherence in deeper lattices; removing those atoms would allow larger atom numbers without paying a coherence penalty.","At the zero-lattice-depth limit the decoherence rate is $(\\Gamma_{\\mathrm{nat}}+\\Gamma_{\\mathrm{BBR}})/2$, meaning shallow-lattice clocks are limited by the atom's own lifetime and the thermal environment rather than by lattice photons.","Operating at $U_0\\approx11E_r$ with $N_{\\mathrm{site}}\\approx9$ balances coherence, atom number, and quantum projection noise, providing a practical operating point for high-stability clocks."],"supporting_citations":[{"why":"Establishes the Wannier-Stark lattice Hamiltonian and the interaction nulling near $U_0\\approx10E_r$ that the shallow-lattice operation relies on.","marker":"[18]"},{"why":"Supplies the lattice Raman scattering rates and the blackbody radiation scattering rate used to set the zero-depth coherence limit.","marker":"[20]"},{"why":"Provides the systematic uncertainty budget and apparatus context for the clock, including the BBR shift evaluation used in the lifetime subtraction.","marker":"[27]"},{"why":"Introduces the imaging spectroscopy and ellipse-fitting method used to extract the frequency difference and the single-clock instability.","marker":"[28]"},{"why":"Gives the two-body inelastic loss coefficient and the density-dependent collision framework used in the population rate equation.","marker":"[31]"},{"why":"Provides an earlier measurement of the ${}^3P_0$ lifetime that the paper's 174(28) s result is compared with.","marker":"[34]"},{"why":"Provides an independent determination of the ${}^3P_0$ lifetime from the electric dipole matrix element, used as an agreement check.","marker":"[35]"},{"why":"Provides the laser coherence baseline that motivates the randomized-phase Ramsey contrast measurement because atomic coherence outlives the interrogation laser.","marker":"[38]"},{"why":"Supplies the sub-10 mHz laser linewidth context that bounds the interrogation laser's coherence in the same measurement.","marker":"[39]"},{"why":"Contains the DDTWA simulation and supplementary data/model details used to reproduce the density-dependent decoherence coefficient.","marker":"[41]"}],"fun_headline_variants":["Strontium clock hits 2-min coherence, 1.5e-18 instability","2-min coherence, 1.5e-18 instability in Sr lattice clock","Wannier-Stark clock: 118 s coherence, 1.5e-18 at 1 s","Clock at atomic limit: 118 s coherence, 1.5e-18 instability","Sr clock: 2-min coherence, 1.5e-18 instability at 1 s"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The weakest load-bearing premise is that the straight-line relation between $\\gamma_0$ and lattice depth $U_0$, fitted only for $U_0>50E_r$, remains valid down to $U_0=0$; if shallow-lattice dynamics change below $50E_r$, the agreement with $(\\Gamma_{\\mathrm{nat}}+\\Gamma_{\\mathrm{BBR}})/2$ and the inferred 174(28) s lifetime would be biased.","fun_headline_variants_meta":{"raw":{"variants":["Strontium clock hits 2-min coherence, 1.5e-18 instability","2-min coherence, 1.5e-18 instability in Sr lattice clock","Wannier-Stark clock: 118 s coherence, 1.5e-18 at 1 s","Clock at atomic limit: 118 s coherence, 1.5e-18 instability","Sr clock: 2-min coherence, 1.5e-18 instability at 1 s"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001151,"raw_usage":{"total_tokens":4755,"prompt_tokens":911,"completion_tokens":3844,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":527,"completion_tokens_details":{"reasoning_tokens":3723}},"tokens_in":527,"tokens_out":3844,"duration_ms":24075,"temperature":1.0,"reasoning_tokens":3723,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T22:42:19.552979+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the zero-density Ramsey contrast decay rate $\\gamma_0$ at lattice depths between 10 and 40 $E_r$ and compare it with the extrapolated line: if $\\gamma_0$ deviates upward as $U_0$ drops, the lifetime inference is biased. Independently, measure the ${}^3P_0$ lifetime with a single atom in a magic-wavelength trap; agreement with 174(28) s would confirm the population-decay analysis, while a significantly different value would break the link between the measured Raman pumping rate and the fundamental lifetime limit.","supporting_citations":[{"cited_title":"Aeppli, A","cited_arxiv_id":null,"evidence_quote":"Establishes the Wannier-Stark lattice Hamiltonian and the interaction nulling near $U_0\\approx10E_r$ that the shallow-lattice operation relies on."},{"cited_title":"D¨ orscher, R","cited_arxiv_id":null,"evidence_quote":"Supplies the lattice Raman scattering rates and the blackbody radiation scattering rate used to set the zero-depth coherence limit."},{"cited_title":"Aeppli, K","cited_arxiv_id":null,"evidence_quote":"Provides the systematic uncertainty budget and apparatus context for the clock, including the BBR shift evaluation used in the lifetime subtraction."},{"cited_title":"Ozeri, C","cited_arxiv_id":null,"evidence_quote":"Gives the two-body inelastic loss coefficient and the density-dependent collision framework used in the population rate equation."},{"cited_title":"Yasuda and H","cited_arxiv_id":null,"evidence_quote":"Provides an earlier measurement of the ${}^3P_0$ lifetime that the paper's 174(28) s result is compared with."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides an independent determination of the ${}^3P_0$ lifetime from the electric dipole matrix element, used as an agreement check."},{"cited_title":"Lisdat, J","cited_arxiv_id":null,"evidence_quote":"Provides the laser coherence baseline that motivates the randomized-phase Ramsey contrast measurement because atomic coherence outlives the interrogation laser."},{"cited_title":"Oelker, R","cited_arxiv_id":null,"evidence_quote":"Supplies the sub-10 mHz laser linewidth context that bounds the interrogation laser's coherence in the same measurement."},{"cited_title":"Fundamental limit of phase coherence in two-component Bose-Einstein condensates","cited_arxiv_id":"2004.02492","evidence_quote":"Contains the DDTWA simulation and supplementary data/model details used to reproduce the density-dependent decoherence coefficient."}],"review_version":1}