{"id":"3bf9573e-81aa-4616-8687-d470532ea2ed","arxiv_id":"2508.12887","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Thulium atoms show a 22 s Ramsey hyperfine coherence time and a 55 s dynamical-decoupling coherence time, with optical shelving into a 112 ms metastable state.","lead":"Neutral thulium atoms store quantum information in a hyperfine qubit for about 22 seconds, and for about 55 seconds when decoupling pulses are added. The paper also shows coherent transfer of the qubit to an optical metastable level, establishing thulium as a candidate for neutral-atom quantum computing.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The T2 = 55 s headline rests on an unverified contrast reconstruction (Methods G, Eqs. 9-10) that assumes a 1/2 offset and symmetric decay; a phase-resolved DD measurement is needed before this part of the abstract's record claim is accepted.","rationale":"The central scientific contribution is thulium as a viable neutral-atom qubit with direct Ramsey coherence T2* = 22 s, state preparation, readout, and metastable-state transfer; those measurements are well described and internally consistent. The one load-bearing weak point is the T2 = 55 s dynamical-decoupling claim, because it is an advertised headline number and is derived from a model-dependent reconstruction rather than from a measured fringe. The reader's weakest_assumption identifies exactly this point. I therefore do not reject the paper, but I would make acceptance conditional on either providing phase-resolved confirmation of T2 = 55 s or reframing the abstract so that T2 = 55 s is presented as a model-dependent estimated lower bound rather than a directly measured coherence time. The proposed phase-resolved Carr-Purcell measurement would settle whether the concern actually lands.","tokens_in":21636,"tokens_out":6349,"duration_ms":75749,"concrete_test":"Re-measure the Carr-Purcell sequence at B = 0.1 G with n = 8 using a phase-controlled microwave source, or by stepping the final-pulse phase/frequency over at least one full Ramsey fringe at each evolution time, fit each fringe with Eq. (3) allowing A, C, and phi_0 to vary, and compare the resulting T2 with 55 s. If the phase-resolved T2 is incompatible with 55 s, or falls below the directly measured Ramsey T2* = 22 s, the dynamical-decoupling record claim in the abstract must be revised.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The abstract advertises T2 = 55^{+59}_{-14} s as record-scale, but this value is not obtained from a directly measured Ramsey or dynamical-decoupling fringe. Methods G states that the phase of the final microwave pulse could not be controlled, so the contrast is inferred from the maximum detected population eta_max via C = 2*eta_max - 1 (Eq. 10), using Eq. (9), which assumes the fringe offset A = 1/2 and a perfectly symmetric decay around 1/2. If the true offset deviates from 1/2, for example because of the state-dependent F = 4 depolarization documented in Methods E, or if magnetic field fluctuations do not simply reduce the maximum but also shift the fringe, the inferred contrast is biased and T2 = 55 s can be substantially overestimated. The strongly asymmetric error bar (+59/-14 s) already exposes how weakly the data constrain the long-time tail. The direct T2* = 22 s Ramsey result is not affected by this concern, so the platform claim survives; only the dynamical-decoupling-based record number is at risk.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports experiments with neutral thulium atoms in an optical lattice, demonstrating microwave state preparation and state-selective readout, single-qubit rotations on the 1497 MHz hyperfine transition, Ramsey and Carr-Purcell dynamical-decoupling coherence measurements, and coherent population transfer between the ground hyperfine qubit and the 1140 nm metastable states. The headline results are a Ramsey coherence time T2* = 22(+2/-2) s at B = 0.1 G and a dynamical-decoupling coherence time T2 = 55(+59/-14) s with n = 8 intermediate pi-pulses, together with coherent optical transfer whose duration is limited by the 112 ms metastable lifetime. The paper argues that thulium combines the robust hyperfine encoding of alkali atoms with the optical-metastable toolbox of alkaline-earth-like atoms.","tokens_in":21858,"tokens_out":4676,"duration_ms":48826,"significance":"If the results hold, this is a useful experimental advance: thulium is a relatively underexplored neutral atom, and the combination of a 1497 MHz hyperfine qubit with the 1140 nm clock/metastable transition is distinctive. The state-preparation and shelving-based readout protocols are carefully described, and the direct Ramsey T2* = 22(2) s measurement at B = 0.1 G is credible: it is fitted from interference fringes with quoted 1 s.d. statistical errors and supported by an independent peak-to-peak analysis. The coherent bicolor transfer through the metastable state, with the contrast following the 112 ms lifetime limit, is also a clean result. The principal weakness is that the abstract's T2 = 55 s record-scale claim rests on an indirect contrast reconstruction (Methods G) that assumes a symmetric fringe offset of 1/2 without phase-resolved dynamical-decoupling data; this part of the paper needs strengthening or explicit qualification.","major_comments":[{"comment":"The headline T2 = 55 s is not obtained from a directly measured Ramsey or dynamical-decoupling fringe. As stated in Methods G, the phase of the final microwave pulse could not be adjusted, so the contrast is reconstructed from the maximum detected population via Eq. (10), C = 2*eta_max - 1, using Eq. (9) with the assumptions that the fringe offset A = 1/2 and that the decay is symmetric about 1/2. The supporting evidence for A = 1/2 comes from Ramsey fits at short times, but Methods F shows that at long free-evolution times the shot-to-shot scatter of eta_4 reaches about 50%, attributed to slow magnetic-field fluctuations, and Methods E documents depolarization of the F = 4 level that could shift the effective offset. A deviation of A from 1/2 biases the inferred contrast and therefore T2, and the quoted asymmetric errors (+59, -14 s) do not include this model uncertainty. Please either provide phase-resolved data for the dynamical-decoupling sequence or reword the abstract and Section IV.C so that T2 = 55 s is presented as a model-estimated quantity rather than a directly measured coherence time.","section":"Methods G / Abstract"},{"comment":"The dynamical-decoupling data extend to about T = 20 s, while the fitted T2 = 55 s lies well beyond the measurement window; the Gaussian decay in Eq. (4) is therefore extrapolated. Because the inferred contrast is still well above zero at the longest measured times, the data do not strongly constrain either the decay time or the decay shape. The chi-square-based errors in Fig. 14(c) reflect statistical uncertainty in the reconstructed contrast under the assumed model, not the systematic sensitivity to the offset assumption or to the choice of decay functional form. A conservative lower bound on T2, or the contrast value at the longest measured time, would be a more defensible headline than a fitted 55 s with asymmetric errors.","section":"Section IV.C / Fig. 6"}],"minor_comments":[{"comment":"In the sentence reporting two-body loss coefficients, the second coefficient is written as beta(mF = -4, B = 0.1 G) = 6.6(3.3)e-11 cm3/s; from the context and Fig. 4 this should be B = 0.6 G.","section":"Section IV.A"},{"comment":"The symbol T is used both for the free evolution time and for the decay constant in Eq. (9); please use distinct notation, for example T_2 or tau, to avoid ambiguity.","section":"Methods G, Eq. (9)"},{"comment":"For the B = 0.6 G peak-to-peak estimates, the text says that no significant decay is observed, yet a dashed curve with T2m = 46(2) s is drawn and said to describe the data well. Please clarify whether the B = 0.6 G empty-marker data actually constrain the decay model or merely are consistent with it.","section":"Methods F / Fig. 13"},{"comment":"Reference [55] is incomplete: it lists 'Review of Scientific Instruments 96 (2025)' without an article number or page range.","section":"References"},{"comment":"The manuscript says that more than 99% of atoms are in the ground vibrational state after 506 nm cooling, but the state-preparation efficiency is later given as about 40% after cleaning. Please state explicitly whether the 40% includes the ground-vibrational-state fraction or refers only to the total atom number retained.","section":"Section II / Fig. 2"}],"recommendation":"major_revision","confidential_remarks":"The experimental work is careful and the direct Ramsey result is solid; the main obstacle is the abstract's presentation of T2 = 55 s as a measured record-scale coherence time when the reconstruction in Methods G rests on an unverified symmetric-offset assumption. I would ask the authors to supply a phase-resolved dynamical-decoupling measurement or to clearly reclassify the 55 s value as a model-dependent estimate/lower bound. The rest of the manuscript is well within the journal's scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a decent experimental paper, and the core result—22(2) s Ramsey coherence on the 1497 MHz hyperfine transition in neutral Tm, plus shelving through the 1140 nm metastable state—looks real. The platform claim is new: previous Tm work had cooling, lattice, clock spectroscopy, but not ground-state hyperfine qubit coherence or coherent optical operations. The readout scheme with metastable shelving is a genuinely useful addition and is described in enough detail to reproduce.\n\nThe direct Ramsey measurement is the strongest part. Fitting fringes to Eq. (3) and seeing Gaussian decay to T2* = 22(2) s at 0.1 G is straightforward; the 0.6 G fit being poor is honestly discussed and attributed to magnetic field noise. The state preparation and readout calibrations are careful, and the paper is transparent about the ~1% readout crosstalk and optical transfer imperfections.\n\nThe soft spot is the headline T2 = 55 s. That value is not from phase-resolved dynamical decoupling fringes. Methods G reconstructs contrast from the maximum measured population under the assumption that the Ramsey offset sits at 1/2 and decay is symmetric. The error bar (+59/-14 s) already tells you how weakly the data constrain the tail. If the offset drifts—and the F=4 depolarization documented in Methods E means the two states do not behave symmetrically—the reconstructed contrast is biased upward. So I would treat 55 s as an order-of-magnitude lower-bound estimate, not a measured record, and the abstract's 'record-scale performance' claim leans harder on that number than the evidence supports. The 22 s Ramsey value is not affected and stands on its own.\n\nMinor: no data or code are provided, but the methods are detailed enough that a specialist could reproduce the analysis. The self-citations go to prior measurements of the hyperfine splitting and clock transition, which are external inputs; that is fine. The two-body loss analysis is a bit compressed but adequate.\n\nVerdict: I would send this to peer review. The central platform demonstration is defensible; the referee should push the authors to either re-measure T2 with controlled phase or soften the abstract so the 55 s number is presented as a model-dependent lower bound. The paper is for the neutral-atom QC and optical-clock community, and it is worth their time.\n\nFor your reading group: worth a slot, mainly to argue about Methods G.","headline":"A credible first characterization of thulium as a hyperfine-qubit platform: the direct 22 s Ramsey coherence is the solid result, while the 55 s dynamical-decoupling number is an indirect estimate that the abstract promotes more strongly than the methods support.","tokens_in":22383,"tokens_out":1688,"would_cite":true,"duration_ms":17666,"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":"Neutral thulium atoms keep a microwave qubit coherent for tens of seconds.","keywords":["thulium","hyperfine qubit","coherence time","Ramsey spectroscopy","dynamical decoupling","metastable optical states","optical lattice","quantum computing"],"falsifier":"Run the eight-pulse Carr-Purcell sequence with full phase control of the final microwave pulse and scan that phase to record the complete fringe at each evolution time; if the directly measured contrast at $T=20$ s falls below $\\exp[-(20/55)^2]\\approx 0.88$, or if the fringe baseline drifts away from one half, the symmetric-decay reconstruction has overestimated $T_2$.","tokens_in":1918,"feed_emoji":"⚛️","tokens_out":3000,"duration_ms":125618,"temperature":0.7,"pith_summary":"Thulium atoms are proposed as a neutral-atom quantum computing platform that combines a robust microwave hyperfine qubit with the optical metastable-state toolbox usually associated with alkaline-earth-like atoms. The paper establishes that a qubit stored in the $m_F=0$ sublevels of the 1497 MHz ground-state hyperfine doublet keeps its phase for $T_2^* = 22^{+2}_{-2}$ s in Ramsey measurements and for $T_2 = 55^{+59}_{-14}$ s under eight Carr-Purcell decoupling pulses, coherence that it compares with the longest values seen in neutral atoms. It also demonstrates state preparation, state-selective shelving readout, single-qubit microwave rotations with fidelity above 99%, and coherent transfer of the qubit through the 1140 nm metastable states, where the 112 ms natural lifetime sets the limit. If these results hold, thulium offers one atom with both the alkali-style hyperfine encoding and the metastable-state manipulation used for qudit, memory, and erasure-detection schemes.","feed_headline":"Thulium qubit holds its phase for 55 seconds","feed_subtitle":"A neutral atom with alkali-style hyperfine encoding and optical metastable control reaches record-scale coherence.","key_machinery":"The machinery that carries the argument is the 1497 MHz hyperfine qubit: the $m_F=0$ sublevels of the $F=3$ and $F=4$ ground-state hyperfine levels, which are first-order insensitive to magnetic field and shift only quadratically (852 Hz/G²). Ramsey spectroscopy measures the free-induction coherence, Carr-Purcell dynamical decoupling removes slow magnetic-field noise, and the 1063.5 nm optical lattice at a near-magic wavelength holds the atoms during tens-of-second interrogation. On the optical side, the 1140 nm inner-shell clock transition provides the metastable levels; the bicolor, dual-transition excitation inherited from the thulium optical clock cancels laser phase noise, making the 112 ms metastable lifetime the operative limit for optical operations.","core_discovery":"The paper's central claim is that neutral thulium ($^{169}$Tm) can serve as a quantum computing platform with a qubit encoded in the $m_F=0$ magnetic sublevels of the ground-state hyperfine doublet, split by 1497 MHz. The authors report Ramsey coherence time $T_2^* = 22^{+2}_{-2}$ s at a 0.1 G bias field and, after eight Carr-Purcell decoupling pulses, an extended coherence time $T_2 = 55^{+59}_{-14}$ s, stating that these are record-scale for neutral-atom systems. They demonstrate single-qubit operations with microwave π-pulse fidelity above 99(1)% and visibility above 0.8 after 250 Rabi oscillations. Using the 1140 nm clock transition, they shelve both qubit states in the metastable levels for state-selective readout and transfer the qubit coherently into and out of these states; in the dual-transition, bicolor configuration, laser phase noise is cancelled and the coherence time is set by the 112 ms natural lifetime of the metastable state.","pith_inferences":["Beyond the paper: if slow magnetic-field noise is indeed the dominant decoherence channel, the intrinsic Ramsey coherence of the $m_F=0$ hyperfine qubit could be substantially longer than the reported 22 s; the paper's own peak-to-peak estimate of $T_2^{*m}=46(2)$ s hints at this floor.","Beyond the paper: the bicolor metastable-state transfer is effectively an optical-clock interrogation of the qubit, so the same apparatus could function as a microwave-optical transducer linking hyperfine qubits to 1140 nm photons for quantum networking.","Beyond the paper: in single-atom optical tweezers the ensemble two-body loss and depolarization that truncate the current measurements should largely disappear, so the 55 s coherence time is more plausibly a lower bound than a ceiling for a future thulium array.","Beyond the paper: the open interval between 41 s and 114 s in the dynamical-decoupling fit is a direct invitation to repeat the measurement with phase-controlled pulses; a full fringe scan at each evolution time would confirm or revise the symmetric-decay assumption used to extract $T_2$."],"forward_implications":["Thulium can support a neutral-atom quantum computer in which the 1497 MHz transition performs hyperfine single-qubit gates and the 1140 nm transition provides shelving, mid-circuit storage, and qudit or optical-metastable-ground (omg) architecture protocols.","The shelving readout detects population that has left the qubit subspace, so the same hardware can flag erasure errors during computation.","Because the measured $T_2^*$ is limited by magnetic-field fluctuations rather than by an intrinsic atomic lifetime, passive magnetic shielding or active field stabilization should extend the coherence time beyond the reported 22 s.","Coherent transfer to the 1140 nm metastable states with sub-112 ms duration gives thulium qubits an optical interface that can be used for quantum memory, clock-based operations, or conversion between microwave and optical domains.","The very small differential polarizability of the qubit levels at the 1063.5 nm trapping wavelength supports the path from an ensemble in an optical lattice to single atoms in optical tweezers, where the observed two-body loss and depolarization should be strongly suppressed."],"supporting_citations":[{"why":"supplies the 1497 MHz ground-state hyperfine splitting and the quadratic Zeeman coefficient that set the qubit frequency and its magnetic-field sensitivity.","marker":"[25]"},{"why":"establishes the 1140 nm inner-shell clock transition, the 112 ms metastable lifetime, and the low blackbody-radiation shift that motivate the optical toolbox.","marker":"[29]"},{"why":"provides the simultaneous bicolor interrogation scheme that cancels laser phase noise in the dual-transition metastable transfer.","marker":"[30]"},{"why":"achieves ground-vibrational-state cooling in the optical lattice, the starting point from which state preparation begins.","marker":"[35]"},{"why":"supplies the two-body loss model used to fit atom lifetimes and to justify the roughly 20 s measurement window.","marker":"[45]"},{"why":"provides the peak-to-peak contrast analysis used to estimate Ramsey decoherence from maximum and minimum measured probabilities.","marker":"[58]"},{"why":"is the large tweezer-array benchmark the paper compares its coherence values against when claiming record-scale performance.","marker":"[8]"},{"why":"provides the half-minute-scale atomic coherence result in a tweezer clock that contextualizes the 22 s and 55 s measurements.","marker":"[50]"}],"fun_headline_variants":["Neutral thulium qubit sets 55-second coherence record","Thulium hyperfine qubit coheres for 55 seconds","Record-scale 55s coherence in thulium qubit","Thulium atom qubit: 55s coherence with optical control","Microwave and optical qubit states in thulium"],"cache_read_input_tokens":24576,"weakest_assumption_plain":"The longest coherence result assumes the decay is symmetric around a baseline of exactly one half, so that contrast can be reconstructed from the maximum detected population; if the baseline or symmetry is wrong, the inferred coherence time would be too long.","fun_headline_variants_meta":{"raw":{"variants":["Neutral thulium qubit sets 55-second coherence record","Thulium hyperfine qubit coheres for 55 seconds","Record-scale 55s coherence in thulium qubit","Thulium atom qubit: 55s coherence with optical control","Microwave and optical qubit states in thulium"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000275,"raw_usage":{"total_tokens":1663,"prompt_tokens":983,"completion_tokens":680,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":599,"completion_tokens_details":{"reasoning_tokens":591}},"tokens_in":599,"tokens_out":680,"duration_ms":6897,"temperature":1.0,"reasoning_tokens":591,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T17:17:24.792171+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the eight-pulse Carr-Purcell sequence with full phase control of the final microwave pulse and scan that phase to record the complete fringe at each evolution time; if the directly measured contrast at $T=20$ s falls below $\\exp[-(20/55)^2]\\approx 0.88$, or if the fringe baseline drifts away from one half, the symmetric-decay reconstruction has overestimated $T_2$.","supporting_citations":[{"cited_title":"Singh, S","cited_arxiv_id":null,"evidence_quote":"supplies the 1497 MHz ground-state hyperfine splitting and the quadratic Zeeman coefficient that set the qubit frequency and its magnetic-field sensitivity."},{"cited_title":"Saskin, J","cited_arxiv_id":null,"evidence_quote":"establishes the 1140 nm inner-shell clock transition, the 112 ms metastable lifetime, and the low blackbody-radiation shift that motivate the optical toolbox."},{"cited_title":"Urech, I","cited_arxiv_id":null,"evidence_quote":"provides the simultaneous bicolor interrogation scheme that cancels laser phase noise in the dual-transition metastable transfer."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"achieves ground-vibrational-state cooling in the optical lattice, the starting point from which state preparation begins."},{"cited_title":"Vishnyakova, E","cited_arxiv_id":null,"evidence_quote":"supplies the two-body loss model used to fit atom lifetimes and to justify the roughly 20 s measurement window."},{"cited_title":"For a 2-ms π−pulse, the Rabi frequency is Ω = 2π× 250 Hz","cited_arxiv_id":null,"evidence_quote":"is the large tweezer-array benchmark the paper compares its coherence values against when claiming record-scale performance."}],"review_version":2}