{"id":"7a723bde-072c-4996-8aca-64d7c8ea0d3b","arxiv_id":"2505.08773","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":1,"one_line_summary":"Double microwave dressing suppresses two- and three-body losses in ultracold NaCs by factors over 10,000 and 1,000 while allowing continuous tuning of the dipolar length across tens of thousands of Bohr radii.","lead":"Ultracold NaCs molecules dressed with two microwave fields stay collisionally stable for seconds while their dipole-dipole interaction is tuned over a wide range. The result could unblock strongly interacting molecular quantum gases, the next step toward dipolar quantum droplets and supersolids.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The quoted ad tuning range and stability window are model-derived, and the printed three-level Hamiltonian has Δπ and Δσ interchanged; the polarization calibration is also performed 2 MHz away from the operating point.","rationale":"The reader's weakest assumption correctly identifies the polarization-purity and calibration caveat as a threat to the model-dependent tuning claim. I agree with that assessment but find an additional, more elementary internal inconsistency: the printed three-level Hamiltonian in the Supplementary Material has the detunings Δπ and Δσ assigned to the wrong states. Since the paper's headline dipolar-length range and the boundaries of the low-loss window are computed from this dressed-state model, that typo must be settled before the quantitative tuning claim can be accepted. The proposed recomputation is inexpensive and decisive. At the same time, the core experimental result—a several-second molecular-gas lifetime under double microwave dressing, with two- and three-body loss coefficients at or below the stated detection limits—is measured directly and does not depend on the disputed model. The lower-bound suppression factors are therefore robust, and the appropriate verdict remains conditional rather than reject or unverified. The concerns are reparable through a corrected calculation and a clearer disclosure of the calibration detuning, so the reader's conditional verdict is unchanged.","tokens_in":16422,"tokens_out":13686,"duration_ms":149884,"concrete_test":"Recompute the |s⟩ eigenstate and d_ind for the Fig. 1D parameters using Eq. S1 with the diagonal entries corrected to −Δπ for |1,0⟩ and −Δσ for |1,1⟩, and repeat the computation with residual σ± Rabi components set to their stated bounds (sin(1°) and sin(4°) of Ωπ) at the operating Δπ. Compare the resulting ad(Ωπ/Ωσ) curve, the compensation point where 2|β|² = |γ|², and the bound-state boundaries with Figs. 1D and 4. If the compensation point shifts by more than about 0.05 in Ωπ/Ωσ, or ad changes by more than about 10% at any Ωπ/Ωσ value, the quoted tuning range and stability window require revision; if the curves are unchanged to that tolerance, the concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claim—continuous tuning of the dipolar length from 0 to about 20,000 a0 and a stability window 0.6 < Ωπ/Ωσ < 1.1—depends on the dressed-state composition |s⟩ = α|0,0⟩ + β|1,0⟩ + γ|1,1⟩ through d_ind² = d0²|α|²(2|β|² − |γ|²)/3. The Supplementary Hamiltonian (Eq. S1) is internally inconsistent: in the basis {|0,0⟩, |1,0⟩, |1,1⟩} it places −Δσ on the |1,0⟩ diagonal and −Δπ on the |1,1⟩ diagonal, although the π field (Rabi Ωπ) couples to |1,0⟩ and the σ+ field (Rabi Ωσ) couples to |1,1⟩. If the matrix as printed is the one used in the calculation, the coefficients β and γ, and hence every plotted ad value, are miscalculated. Independently of that typo, the authors state that the orthogonality and compensation calibration is performed at a π detuning about 2 MHz larger than the shielding value, with residual projections bounded only by sin(1°) and sin(4°); at the operating detuning the actual compensation may be worse than these bounds. Because neither the code nor the data is deposited, a reader cannot verify that the computed tuning curve, the compensation point Ωπ/Ωσ = 0.85, and the calculated bound-state boundaries in Figs. 1D and 4 were produced with the corrected model and with realistic impurity levels. This is the load-bearing limitation on the 'wide tunability' half of the central claim; the directly measured long lifetime and the lower-bound two- and three-body suppression factors are not affected.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper reports experiments on ultracold NaCs ground-state molecules in an optical dipole trap under double microwave dressing with both π and σ+ fields. The authors measure a sample lifetime of 6.6(5) s, close to the one-body limit of 8.5(1.5) s, and infer lower-bound suppression factors of more than 10,000 for two-body loss and more than 1,000 for three-body loss relative to undressed molecules. They further claim that the dipolar length can be continuously tuned from 0 to about 20,000 a0 (and to negative values down to -21,000 a0) across a stability window 0.6 < Ωπ/Ωσ < 1.1, and they project interaction-strength ratios rd up to 6.5 at densities that still permit a 100 ms lifetime. The experimental conclusions are based on direct decay measurements, fits to a kinetic model, and coupled-channels calculations; the quoted tuning range and stability boundaries are derived from a three-level dressed-state model.","tokens_in":16811,"tokens_out":5157,"duration_ms":48724,"significance":"The directly measured several-second lifetime and the honest reporting of suppression factors as lower bounds are significant experimental achievements for bulk ultracold molecular gases, as prior shielding demonstrations achieved only factors of 10 to 100. If the wide tunability and stability window are quantitatively correct, the work would open a realistic route to strongly dipolar many-body phases with molecules, and the projected rd values are compelling. The paper also uses a careful deep-trap method to suppress evaporative loss, spectral calibration of Rabi frequencies, and coupled-channels calculations that reproduce the measured loss rates where they can be compared. However, the tunability claim rests on model calculations whose printed Hamiltonian is internally inconsistent, and the calibration of the microwave polarization compensation is carried out away from the operating point; these issues must be resolved before the tuning range can be taken at face value.","major_comments":[{"comment":"The printed Hamiltonian matrix has the detunings interchanged: in the basis {|0,0>, |1,0>, |1,1>} the diagonal entries are -Δσ for |1,0> and -Δπ for |1,1>, whereas the main text (Fig. 1B) defines the π field (Rabi Ωπ) as coupling |0,0> to |1,0> with detuning Δπ and the σ+ field (Rabi Ωσ) as coupling |0,0> to |1,1> with detuning Δσ. As printed, the matrix cannot be the one used to obtain the dressed-state amplitudes β and γ that enter d_ind^2 = d0^2|α|^2(2|β|^2 - |γ|^2)/3, and hence all plotted ad values in Figs. 1D and 5A would be miscalculated. Please correct the matrix and explicitly confirm that the tuning curves and bound-state boundaries were computed with the correct assignment of Δπ and Δσ. Since no code or data is deposited, this correction is necessary for the reader to verify the central tunability claim.","section":"Supplementary, 'Three-level V-system', Eq. (1)"},{"comment":"The compensation of the π-field projection onto the σ+ plane is calibrated at a π detuning about 2 MHz larger than the shielding value, and the residual projections are bounded only by sin(1°) and sin(4°) in Rabi frequency. The authors note that a frequency-dependent relative phase shift leaves uncertainty in the compensation quality at the precise dressing frequency. Because the dressed-state composition, the compensation point Ωπ/Ωσ ≈ 0.85, and the bound-state boundaries in Figs. 1D and 4 all depend on the purity of the microwave polarizations, this calibration uncertainty is load-bearing for the quoted stability window 0.6 < Ωπ/Ωσ < 1.1 and the ad tuning range. Please provide a quantitative sensitivity analysis (for example, recompute ad and the loss rates for residual projections within the stated bounds) or otherwise demonstrate that the calibration uncertainty does not alter the central tunability claims.","section":"Supplementary, 'Rabi frequencies, microwave ellipticities and orthogonality'"},{"comment":"The density-dependent heating term En is introduced as a fitting parameter with no independent measurement or model (details are deferred to future work). The extracted two-body and three-body loss coefficients, which are compared with theory in Figs. 3 and 4 and used to compute nmax and rd in Fig. 5, may therefore carry a systematic uncertainty from this unconstrained term. Please show that the fitted β2B and L3B values are stable under reasonable variations of En, or include the resulting systematic error in the reported loss coefficients and in the projected densities and interaction ratios.","section":"Supplementary, 'Kinetic model'"}],"minor_comments":[{"comment":"The abstract states the dipolar length can be tuned 'from 0 to 1 µm ∼ 20,000 a0', but Fig. 5A shows a range from -21,000 a0 to +12,000 a0; please state the full range including the negative (antidipolar) values, and use '≈' rather than '∼' for the unit conversion.","section":"Abstract and Fig. 5A"},{"comment":"The heading 'Data Vailability' contains a typo and should read 'Data Availability'.","section":"Main text, after Acknowledgments"},{"comment":"For 0.6 < Ωπ/Ωσ < 1.1 the three-body loss coefficient is quoted as 2(2) × 10^-25 cm^6/s, which is 'at or below our detection limit'; it would be clearer to report this as an upper bound rather than a central value with uncertainty.","section":"Fig. 4B and related text"},{"comment":"The word 'significantly' is misspelled as 'significantly' (extra 'f'?); please check the spelling in the caption.","section":"Supplementary, Fig. S1 caption"},{"comment":"The notation for the Rabi frequency ratio appears as both 'Ωπ/Ωσ' and 'Ωπ/Ωσ' with inconsistent subscript formatting; please standardize the typography.","section":"Main text, Fig. 1 caption and text"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe headline result is real: a bulk NaCs gas that lives for 6.6(5) s under double microwave dressing, close to the one-body limit, with two- and three-body loss suppressed by at least four and three orders of magnitude. The suppression factors are honestly reported as lower bounds set by detection limits, and the lifetime measurement is direct. That part of the paper is solid.\n\nWhat's new beyond their BEC work is the demonstration that the suppression survives across a range of parameters, so you can tune the dipolar length while keeping losses low. The kinetic-model analysis is careful; the deep-trap technique to kill evaporation is a nice touch. The measured loss coefficients for single microwave dressing match their coupled-channels calculations well.\n\nThe soft spot is on the tuning half. The dressed-state Hamiltonian printed in the Supplementary (Eq. S1) has the π and σ detunings interchanged: the |1,0> diagonal is labeled -Δσ and the |1,1> diagonal -Δπ, even though the text says π couples to |1,0> and σ+ to |1,1>. If that matrix as printed is what went into the calculation, the coefficients β and γ, and therefore every ad value in Figs. 1D and 5A, are wrong. No code is deposited, so a referee cannot check. This may be just a typo in the paper, but it is exactly the kind of thing that needs to be fixed before the tunability claim is taken at face value.\n\nTwo smaller wrinkles: the orthogonality and compensation calibration is done at a π detuning about 2 MHz larger than the shielding value, with residual σ projections bounded only by sin(1°) and sin(4°). That leaves real uncertainty in the actual compensation at the operating point. And data/code are available only \"upon reasonable request,\" which is not great for a paper with this much quantitative modeling.\n\nNone of this touches the measured lifetime or the lower-bound suppression factors. The paper deserves a serious referee. Send it to review, but insist that the authors correct the Hamiltonian, deposit the code and data, and address the calibration offset. If the tuning curve reproduces after the fix, this is a major advance.","headline":"The measured lifetime and loss suppression are solid and important; the wide-tunability claim needs a clean fix of the Supplementary Hamiltonian before it can be trusted.","tokens_in":17339,"tokens_out":5228,"would_cite":true,"duration_ms":44374,"reading_group":"yes","serious_thinker":"unclear","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["67.85.-d"],"model":"deepseek-v4-flash","headline":"Dressing NaCs molecules with two microwave fields cuts two-body loss over 10,000-fold and three-body loss over 1,000-fold, with dipolar interactions tunable to 20,000 Bohr radii.","keywords":["ultracold molecules","microwave shielding","double microwave dressing","dipolar interactions","collisional losses","NaCs","dipolar length","quantum many-body phases"],"falsifier":"Measure the two- and three-body loss coefficients with the $\\pi$ field at the exact operating detuning ($\\Delta_\\pi/(2\\pi) = 10$ MHz) while characterizing the $\\pi$ field's $\\sigma^+$ and $\\sigma^-$ projections in situ at that same frequency; if $\\beta_{2B}$ and $L_{3B}$ rise above the coupled-channels predictions for the measured residual polarization, or above the $10^{-13}$ cm$^3$/s and $10^{-25}$ cm$^6$/s detection floors, the assumed shielding barrier is too strong and the claimed tuning range must be revised inward.","tokens_in":16248,"feed_emoji":"🧲","tokens_out":20263,"duration_ms":150494,"temperature":0.7,"pith_summary":"The paper reports that ultracold NaCs molecules in their rovibrational ground state, dressed by two microwave fields with $\\pi$ and $\\sigma^+$ polarization, almost stop colliding: two-body loss drops by more than a factor of 10,000 relative to undressed molecules, three-body loss by more than 1,000 relative to single microwave dressing, and the samples survive for several seconds at densities near $10^{12}$ per cubic centimeter. The same dressing that blocks the losses sets the strength of the long-range dipole-dipole force, so the dipolar length can be dialed continuously from zero to about 20,000 Bohr radii without leaving the low-loss window. If these results hold, molecular gases could finally combine the long lifetimes of atomic quantum gases with interactions strong enough to search for bulk quantum phases such as dipolar droplets, supersolids, and self-organized crystals.","feed_headline":"Molecular losses cut 10,000-fold by double microwave dressing","feed_subtitle":"Ultracold NaCs pairs live seconds with dipolar strength tunable from zero to about 20,000 Bohr radii.","key_machinery":"The load-bearing object is the microwave-dressed three-level system: the rotational ground state $|0,0\\rangle$ is coupled by a $\\sigma^+$ field to $|1,1\\rangle$ and by a $\\pi$ field to $|1,0\\rangle$, and the molecules are prepared in the highest-energy eigenstate $|s\\rangle = \\alpha|0,0\\rangle + \\beta|1,0\\rangle + \\gamma|1,1\\rangle$. This state combines two opposing interaction channels — the $\\pi$ admixture produces repulsive-in-plane 'dipolar' forces while the $\\sigma^+$ admixture produces attractive-in-plane 'antidipolar' forces — whose balance is set by the Rabi-frequency ratio $\\Omega_\\pi/\\Omega_\\sigma$. The effective squared dipole moment $d_{\\mathrm{ind}}^2 = d_0^2|\\alpha|^2(2|\\beta|^2 - |\\gamma|^2)/3$ controls the dipolar length $a_d = m\\,d_{\\mathrm{ind}}^2/(4\\pi\\varepsilon_0\\hbar^2)$, which vanishes when $|\\gamma|^2 = 2|\\beta|^2$. The same eigenstate carries a repulsive collisional barrier at an intermolecular distance near 2,500 $a_0$ that blocks short-range inelastic collisions, and coupled-channels calculations with absorbing boundary conditions predict the residual loss rates and the bound-state positions that mark the edges of the stability window.","core_discovery":"The central discovery is that the shielded eigenstate produced by double microwave dressing confers collisional stability and strong, tunable dipolar interactions simultaneously. With the $\\sigma^+$ field fixed near $\\Delta_\\sigma/\\Omega_\\sigma = 1$, adding a $\\pi$ field and varying the Rabi-frequency ratio $\\Omega_\\pi/\\Omega_\\sigma$ from 0.6 to 1.1 keeps two-body loss below the $10^{-13}$ cm$^3$/s detection floor and three-body loss near $2\\times10^{-25}$ cm$^6$/s, while the effective dipole moment passes through zero at the compensation point $|\\gamma|^2 = 2|\\beta|^2$ and reaches dipolar lengths from $-21{,}000\\,a_0$ (antidipolar) to $+12{,}000\\,a_0$ (dipolar). Measured loss coefficients agree with coupled-channels calculations where they can be compared, and losses rise only where a bound state enters the dressed potential, opening a three-body recombination channel. At the projected maximal densities near $10^{15}$ cm$^{-3}$, the ratio of dipolar to kinetic energy $r_d = |a_d|/a_{sp}$ reaches 6.5, placing the system firmly in the strongly dipolar regime.","pith_inferences":["The loss coefficients inside the stable window sit below the detection floor set by the one-body lifetime, so the quoted suppression factors of 10,000 and 1,000 are lower bounds; a trap with less laser scattering would reveal the true residual loss of double dressing.","The stability window ends where bound states enter the dressed potential, so varying the two-field frequency difference $\\Delta_\\pi - \\Delta_\\sigma$ should shift those bound states and could widen the low-loss window — a direct, testable extension of the reported tuning maps.","Near-total suppression of loss wherever no bound state exists suggests the same dressing pair can act as a fast on-off switch for collisions, letting experiments quench the shielding on microsecond timescales and watch a stable gas turn collisional.","If the residual-loss floor is as low as the detection limit suggests, evaporative cooling should reach degeneracy at strong dipolar interactions, not only at the $a_d = 0$ compensation point used for the NaCs condensate."],"forward_implications":["At densities near $10^{12}$ cm$^{-3}$, double-dressed NaCs samples live longer than 6 seconds, close to the 8.5-second one-body limit set by trap-laser scattering, so collisional loss no longer limits the sample lifetime.","Within the stability window $0.6 < \\Omega_\\pi/\\Omega_\\sigma < 1.1$, the dipolar length is continuously tunable from about $-21{,}000\\,a_0$ to $+12{,}000\\,a_0$, including the compensation point where the dipole-dipole interaction vanishes entirely.","Extrapolating the measured loss coefficients to a target lifetime of 100 ms gives maximal densities around $10^{15}$ cm$^{-3}$, comparable to atomic quantum gases, with the dipolar-to-kinetic-energy ratio $r_d$ reaching 6.5.","Because the shielding is set by the dressed eigenstate rather than a species-specific resonance, the same double-dressing scheme should transfer to other molecules with large electric dipole moments.","Together with the existing NaCs Bose-Einstein condensate, simultaneous stability and tunability open the route to strongly dipolar quantum liquids, droplets, supersolids, and self-organized dipolar crystals."],"supporting_citations":[{"why":"Universal two-body loss theory that sets the undressed loss scale of about $10^{-9}$ cm$^3$/s, the baseline from which the suppression factors are measured.","marker":"[26]"},{"why":"First experimental demonstration of microwave shielding of molecular collisions, establishing the single-field approach and its suppression benchmark.","marker":"[30]"},{"why":"Showed that microwave shielding supports evaporative cooling to quantum degeneracy, the benchmark of stability that double dressing must exceed.","marker":"[31]"},{"why":"Earlier single-field microwave shielding of NaCs whose 0.9 s lifetime is the single-dressing baseline compared against double dressing.","marker":"[32]"},{"why":"Realized the NaCs Bose-Einstein condensate at the compensated double-dressing point (dipolar length zero) that this work extends to strong interactions.","marker":"[34]"},{"why":"Theoretical proposal of microwave shielding with field-linked states that the dressing scheme builds on.","marker":"[36]"},{"why":"Supplies the coupled-channels framework and calculations used to predict residual two-body loss and bound-state positions under double dressing.","marker":"[38]"},{"why":"Supplementary material containing the kinetic loss model used to extract two- and three-body loss coefficients from the decay curves.","marker":"[40]"},{"why":"Theory of three-body recombination in microwave-dressed molecules used to explain the loss rise at the edges of the stability window.","marker":"[46]"}],"fun_headline_variants":["Double microwave dressing suppresses molecular losses 10,000x","Ultracold molecules stable for seconds with tunable interactions","Dipolar length tunable from zero to 20,000 Bohr radii","Microwave dressing yields stable, tunable dipolar molecular gas"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The results assume the two microwave fields are almost perfectly pure in polarization and exactly perpendicular, but the polarization was calibrated at a $\\pi$-field detuning about 2 MHz away from the operating point; if the residual $\\sigma^+$ and $\\sigma^-$ projections (bounded by $\\sin 1^\\circ$ and $\\sin 4^\\circ$ in Rabi frequency at calibration) are worse at the working detuning, the real shielding barrier is weaker than modeled and the usable tuning range shrinks.","fun_headline_variants_meta":{"raw":{"variants":["Double microwave dressing suppresses molecular losses 10,000x","Ultracold molecules stable for seconds with tunable interactions","Dipolar length tunable from zero to 20,000 Bohr radii","Microwave dressing yields stable, tunable dipolar molecular gas"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000844,"raw_usage":{"total_tokens":3686,"prompt_tokens":969,"completion_tokens":2717,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":585,"completion_tokens_details":{"reasoning_tokens":2646}},"tokens_in":585,"tokens_out":2717,"duration_ms":19933,"temperature":1.0,"reasoning_tokens":2646,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T21:46:43.057671+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the two- and three-body loss coefficients with the $\\pi$ field at the exact operating detuning ($\\Delta_\\pi/(2\\pi) = 10$ MHz) while characterizing the $\\pi$ field's $\\sigma^+$ and $\\sigma^-$ projections in situ at that same frequency; if $\\beta_{2B}$ and $L_{3B}$ rise above the coupled-channels predictions for the measured residual polarization, or above the $10^{-13}$ cm$^3$/s and $10^{-25}$ cm$^6$/s detection floors, the assumed shielding barrier is too strong and the claimed tuning range must be revised inward.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Universal two-body loss theory that sets the undressed loss scale of about $10^{-9}$ cm$^3$/s, the baseline from which the suppression factors are measured."},{"cited_title":"Anderegg, S","cited_arxiv_id":null,"evidence_quote":"First experimental demonstration of microwave shielding of molecular collisions, establishing the single-field approach and its suppression benchmark."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplementary material containing the kinetic loss model used to extract two- and three-body loss coefficients from the decay curves."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Theory of three-body recombination in microwave-dressed molecules used to explain the loss rise at the edges of the stability window."}],"review_version":1}