{"id":"70b6eb6f-2ac1-483a-b85d-5eafb6e3efc5","arxiv_id":"2608.05512","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Two 39K Feshbach resonances shift by up to 7.5 G in a 1063.9 nm optical trap because the associated Feshbach molecules have unexpectedly large dynamic polarizabilities, four to seven times the atomic sum.","lead":"Researchers found that two Feshbach resonances in potassium-39 atoms shift by as much as 7.5 gauss when the atoms are held in a 1063.9 nm optical trap, far more than expected. The cause appears to be an unusually strong response of the temporarily bound molecule pairs to the trapping laser light, which could affect many cold-atom experiments and enable fast optical control of atomic interactions.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The quantitative central claim is hostage to the unpublished R1 differential magnetic moment: replacing δµ = −2.5 μB with the published −1.9 μB changes αmol/αat from −12.5 to ≈ −9.1, preserving the anomaly but materially altering the 'four to seven times' claim.","rationale":"The reader's weakest assumption is also the one I would identify: the quantitative conversion from measured shifts to molecular polarizabilities is linearly sensitive to δµ, and the R1 value in Table I is the least secure entry because it is unpublished and the text in Sec. II explains why the slope method can be systematically biased for an intermediate-strength resonance. My agreement is substantial, with one nuance: the sensitivity test shows that adopting the published δµ does not erase the anomaly — the sign remains negative and the magnitude remains several times the pair polarizability — so the concern weakens the quantitative claim without overturning the core observation. I therefore do not request a change to the CONDITIONAL verdict. The experimental core is independently supported: the shift vanishes as trap depth goes to zero, the green-point dataset in Fig. 6 with temperature held approximately constant separates trap-depth dependence from temperature dependence, and six other resonances show null or small shifts. The reasons to keep the verdict conditional rather than accept are the unpublished δµ, the simplified spherical trap-averaging model in the Appendix, and the paper's own concession in Sec. V that 'a full theoretical treatment is required to quantitatively account for the observed position shifts'. The R3 frequency dependence also deviates from the simple single-dispersion model in a way the authors attribute to unspecified spectral structure. These are not internal contradictions; they are missing external checks. A focused sensitivity analysis of Eq. (5) against the published R1 δµ is the single most informative check to run, and it would settle whether the abstract's quantitative 'four to seven times' framing can stand as written.","tokens_in":20288,"tokens_out":13048,"duration_ms":123410,"concrete_test":"Recompute the R1 differential magnetic moment using the published coupled-channel models of refs. [5] and [25] with the same singlet and triplet potentials, then re-extract αmol/αat from the measured slopes in Fig. 7 and Table II via Eq. (5) for δµ ∈ {−2.5, −1.9, −1.5} μB. Repeat the R3 extraction for δµ = +1.2 μB ± 20%. If the literature δµ leaves αmol for R1 below about −4, the central anomaly survives and the paper needs only a revised quantitative wording; if the sign or magnitude crosses that threshold, the 'four to seven times' claim and the quoted detunings must be corrected.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The quantitative extraction of the molecular polarizabilities, which is the heart of the claimed anomaly, is directly proportional to the differential magnetic moment δµ through Appendix Eq. (5): αmol/αat = 2 + δµ⟨ΔB⟩/Uav − (3/2)kBT/Uav. For R1, Table I lists δµ = −2.5 μB as '[tw]', from unpublished coupled-channel calculations [42], while the published value from ref. [5] is −1.9 μB. The paper itself states in Sec. II that for the broad intermediate-strength R1 resonance 'the slope of the apparent linear region can lead to systematic error in the measured molecular magnetic moment at the Feshbach resonance center', which is exactly why the two values differ. Re-evaluating Eq. (5) with the published δµ = −1.9 μB changes αmol/αat for R1 from −12.5(1.6) to roughly −9.1, a change of about 27%. The qualitative anomaly survives — the polarizability remains large and negative — but the abstract's quantitative statement 'four to seven times', the asymmetry between the R1 and R3 values, and the inferred +40 GHz and +50 GHz detunings in Sec. V all inherit this systematic uncertainty. The R3 value is similarly anchored to the unpublished δµ = +1.2 μB, for which no independent published check is cited. Because Table II quotes only statistical errors, this unquantified systematic from δµ is the weakest point in the extraction. This concern is distinct from the observation of the shifts themselves, which is supported by the zero-depth extrapolation, the temperature-controlled dataset, and the null results for six other resonances.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports that two 39K Feshbach resonances, at 33.6 G and 39.9 G, shift by up to +7.5 G in a 1063.9 nm optical dipole trap, while six other studied resonances show essentially no shift. The shifts scale with trap depth, approach zero at zero depth, and persist in a spin-polarized sample at fixed temperature; a second ODT laser frequency changes the shifts. The authors attribute the effect to a differential ac Stark shift caused by anomalously large dynamic polarizabilities of the associated Feshbach molecules. Using Eq. (5) of the appendix, they extract molecular polarizabilities αmol/αat of −12.5(1.6) and +9(1) at 281.7665 THz and interpret these as evidence for near-resonant a3Σ+u → b3Σ+g transitions near the ODT wavelength.","tokens_in":20510,"tokens_out":8765,"duration_ms":80220,"significance":"If the quantitative extraction holds, this is a striking result: it contradicts the usual assumption that weakly bound Feshbach molecules have polarizabilities close to the sum of the two atomic polarizabilities, and it has practical consequences for experiments using the 33.6 G resonance in 1063.9 nm ODTs, including compact BEC machines. The paper's observational case is strong: the zero-depth extrapolation, the fixed-temperature dataset in Fig. 6, the six unshifted resonances in Table II, and the two-frequency comparison are all in the right direction and are explicitly presented. The data are publicly available. The main weakness is that the quantitative polarizability values depend on differential magnetic moments δµ taken from unpublished coupled-channel calculations, with only statistical errors quoted in Table II; this limits the precision, but not the qualitative existence, of the claimed anomaly.","major_comments":[{"comment":"The extracted polarizabilities inherit an unquantified systematic uncertainty from the differential magnetic moments δµ, and Table II quotes only statistical errors. For R1, Table I lists δµ = −2.5 μB from the authors' own coupled-channel calculations [42], while the published value from ref. [5] is −1.9 μB; the paper itself states in Sec. II that the slope of the apparent linear region can lead to systematic error for this resonance. Re-evaluating Eq. (5) with δµ = −1.9 μB changes αmol/αat from −12.5(1.6) to roughly −9.1, a ~27% change. The qualitative anomaly survives, but the abstract's 'four to seven times' statement, the asymmetry between R1 and R3, and the detunings inferred in Sec. V all shift. For R3, δµ = +1.2 μB is also from [tw] with no independent published check. The authors should either propagate a defensible systematic uncertainty for δµ, compare explicitly with published values, or restate the quantitative claims with appropriate caveats.","section":"Appendix Eq. (5); Table I; Table II"},{"comment":"The inferred near-resonant detunings of about +40 GHz and +50 GHz for the two ODT frequencies rest on a single dispersion-shaped line with an assumed natural linewidth of twice the atomic linewidth, and only two frequency points are available. The observed R3 behavior (a small decrease in αmol when the frequency is increased by 13.1 GHz) is not consistent with the simple red-detuned dispersion picture and is attributed to unspecified residual structure. This part of the interpretation should be presented as a model-dependent estimate rather than a determined quantity, or supported by additional frequency points or a calculated line shape.","section":"Sec. V; Table II"},{"comment":"The extraction via Eq. (5) uses Uav derived from a spherically symmetric 3D Gaussian trap model with a truncated Boltzmann distribution, whereas the actual crossed ODT is formed by two beams with different waists crossing at 70° and has measured anisotropic trap frequencies. A systematic error in Uav maps directly onto αmol/αat. The authors should quantify how sensitive the extracted polarizabilities are to this modeling choice, either by using the measured trap geometry or by varying the assumed density distribution.","section":"Appendix; Fig. 9"}],"minor_comments":[{"comment":"There is a typo: 'esentially' should be 'essentially'.","section":"Sec. IV B"},{"comment":"The axis labels appear garbled ('Temp averaged (μK)' and 'Trap Depth, /g80K'); the temperature versus trap-depth relation should be labeled clearly.","section":"Fig. 10"},{"comment":"The R6 row reports a shift of +4 mG/µK and αmol/αat = 1.8 without uncertainties, inconsistent with the stated one-standard-error convention for all other rows; also, the R2 row gives αmol/αat = +10(4) despite a shift of +2(7) mG/µK, and no δµ is listed for R2 or R4, so the reader cannot reproduce those entries from Eq. (5).","section":"Table II"},{"comment":"The phrase 'four to seven times the sum of the polarizabilities of the two incoming potassium-39 atoms' is slightly indirect: the Table II values are quoted relative to a single atom, so the factors relative to the two-atom sum are −6.25 and +4.5 for R1 and R3. The wording should be made unambiguous.","section":"Abstract and Sec. VI"},{"comment":"Reference [29] has inconsistent quotation marks and should be reformatted; reference [60] should be cited in the text rather than appearing only as a URL in the data availability statement.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The experimental observation appears solid and well controlled, and the qualitative anomaly is likely real. The main issue is that the quantitative central claim is tied to unpublished δµ values from private communications, with no systematic error budget. I would be comfortable with acceptance after the authors either supply the coupled-channel details and uncertainty for δµ, propagate the published alternative values, or explicitly downgrade the quantitative claims to 'order of magnitude' rather than precise factors of four to seven."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This one is worth knowing about before you next tune 39K in a 1064 nm trap. The experimental observation is solid: the 33.6 G and 39.9 G Feshbach resonances shift by up to +7.5 G with ODT depth, the shift tracks trap depth and extrapolates to the known zero-depth positions, a temperature-held run separates trap depth from temperature, and six other resonances show no shift. The data also resolve the predicted R0B component of the 25.9 G resonance, which is a nice by-product. The extraction is not circular: the shifts are independent observables, and Eq. (5) is a standard first-order relation. The data availability statement is explicit, which helps.\n\nThe soft spot is quantitative, not qualitative. The extracted polarizabilities are directly proportional to the differential magnetic moment δµ. For R1 the paper uses δµ = −2.5 μB from an unpublished private communication, while the published value is −1.9 μB; using the published value moves αmol/αat from −12.5 to about −9.1. That still leaves an anomalously large polarizability, but the 'four to seven times' claim and the inferred detunings in Sec. V inherit this systematic. The R3 value is anchored to another unpublished δµ. Table II quotes only statistical errors, so the headline numbers overstate what is actually pinned down. The trap averaging is also a simplified spherical model with an assumed thermal correction, though it is reasonable for this purpose.\n\nThe mechanism—enhanced polarizability from a near-resonant a3Σ+u → b3Σ+g transition—is plausible but rests on an unpublished upper-state potential and an assumed linewidth. The paper itself concedes that a full theoretical treatment is required, and the R3 frequency dependence does not follow the simple dispersion model. Those are real limitations, and they are flagged rather than hidden.\n\nWho gets value? Anyone doing 39K BEC or precision scattering-length tuning in 1064 nm traps. The warning about shifts is directly useful, and the anomaly itself is worth chasing. My recommendation: send it to a serious referee. The experimental core deserves publication; the referee should ask the authors to quantify the δµ systematic, make the coupled-channel inputs available or cite a published check, and soften the 'four to seven times' claim accordingly.","headline":"A convincing measurement of large ODT-depth-dependent Feshbach shifts in 39K, with the headline polarizability magnitude softer than the quoted errors suggest because it depends on an unpublished differential magnetic moment.","tokens_in":21232,"tokens_out":2646,"would_cite":true,"duration_ms":26608,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Potassium-39 Feshbach resonances shift by up to 7.5 G under 1063.9 nm optical dipole trap light, the authors show, because the weakly bound Feshbach molecule has a surprisingly large dynamic polarizability at the trap wavelength.","keywords":["Feshbach resonances","potassium-39","optical dipole trap","ac Stark shift","dynamic polarizability","Feshbach molecules","ultracold atoms","molecular spectroscopy"],"falsifier":"Measure the shift of the 33.6 G resonance while tuning the single-frequency trap laser across 1063–1066 nm: if the shift does not trace a dispersion-shaped polarizability curve with a sign change and a magnitude that grows toward the molecular resonance, the near-coincidence explanation is wrong. Alternatively, measure δµ of the R1 molecular state by two-photon or radio-frequency association spectroscopy in zero light; if it comes out at −1.9 μB rather than −2.5 μB, the extracted αmol for the 33.6 G resonance drops by 25–30% and the 'four to seven times' claim is overstated.","tokens_in":19969,"feed_emoji":"🧲","tokens_out":7369,"duration_ms":56960,"temperature":0.7,"pith_summary":"This paper reports that two Feshbach resonances in potassium-39, the workhorse 33.6 G resonance and the mixed-spin 39.9 G resonance, move to higher magnetic field by up to 7.5 G when the atoms sit in a 1063.9 nm optical dipole trap, with the shift scaling with trap depth and vanishing as the trap is turned off. The authors argue that the shift is a differential ac Stark shift caused by the weakly bound Feshbach molecule having a surprisingly large dynamic polarizability at the trap wavelength, about four to seven times the sum of the two atomic polarizabilities, and negative for the 33.6 G resonance. They trace the enhancement to a near-coincidence between the 1063.9 nm trap frequency and a transition from the last vibrational level of the metastable a3Σ+u potential to a level of the b3Σ+g potential of K2. If correct, the result means that weakly bound molecules do not always inherit the sum of atomic polarizabilities, with practical consequences for any experiment that tunes the 33.6 G resonance inside an optical trap.","feed_headline":"Optical trap shifts 39K Feshbach resonances by 7.5 G","feed_subtitle":"The shift traces to molecular polarizabilities four to seven times the atomic value, opening a window on Feshbach molecules.","key_machinery":"The engine of the argument is the relation δµ ∆B = (δα/αat) Uav + ∆Uth, which ties the resonance shift ∆B to the differential polarizability δα = 2αat − αmol between the atom pair and the molecule, the differential magnetic moment δµ, and the trap-averaged ac Stark shift Uav. The experiment controls Uav by varying the 1063.9 nm trap power while keeping temperature roughly fixed, and the thermal term is subtracted using a truncated-Boltzmann model of the trapped cloud. The unusually large αmol is then explained by a near-degeneracy: the trap-laser frequency 281.7665 THz lies within roughly 50 GHz of the a3Σ+u(v''=26) → b3Σ+g(v'=4) molecular transition, whose Franck–Condon overlap is large, so the dynamic polarizability is dominated by a nearby molecular resonance and can exceed, even reverse sign relative to, the sum of the atomic polarizabilities.","core_discovery":"The central claim is that the positions of the potassium-39 Feshbach resonances at 33.6 G and 39.9 G shift by as much as +7.5 G in a 1063.9 nm optical dipole trap, and that this shift is not a temperature effect but a differential ac Stark shift between the incoming atom pair and the quasi-bound Feshbach molecule. From the measured shift versus trap-averaged Stark shift, the paper extracts molecular polarizabilities αmol = −12.5(1.6) αat for the 33.6 G resonance and αmol = +9(1) αat for the 39.9 G resonance at 281.7665 THz, with smaller magnitudes at a frequency shifted by +13.1 GHz. The magnitude is four to seven times the usual sum-of-atoms value for weakly bound Feshbach molecules, and the negative sign for the 33.6 G resonance is the signature of a resonance on the blue-detuned side of a molecular transition. The paper attributes the enhancement to a near-coincidence between the trap-laser frequency and the a3Σ+u (v''=26) → b3Σ+g (v'=4) transition of the K2 molecule, so that the trap light sits within about 50 GHz of the molecular resonance.","pith_inferences":["If the near-resonance interpretation is right, the same 1064 nm trap should produce large, controllable Feshbach resonance shifts in other alkali species whose triplet transitions fall near common fiber-laser wavelengths; the fact that only two of eight 39K resonances show the effect is then a selection rule sensitive to the hyperfine symmetry of the molecular state.","The paper's quantitative extraction leans on a differential magnetic moment taken from its own coupled-channel calculation; redoing the extraction with an independently measured δµ, for example from radio-frequency association spectroscopy, would settle whether the 'four to seven times' enhancement is as large as claimed or partly an artifact of the magnetic-moment input.","A direct test would be to tune a single-frequency fiber laser across the a3Σ+u(v''=26) → b3Σ+g(v'=4) region near 1063.9 nm while monitoring the 33.6 G shift; a dispersion-shaped polarizability curve with a zero crossing would map the molecular resonance and would also locate the magic frequency where the trap no longer shifts the resonance.","The existence of a negative molecular polarizability for the 33.6 G Feshbach molecule suggests that a suitably blue-detuned trap could cancel the atomic polarizability, potentially creating a state-dependent optical lattice for selectively trapping molecules over atoms."],"forward_implications":["Any 39K experiment that tunes the 33.6 G or 39.9 G resonance inside a 1064 nm optical dipole trap must account for a depth-dependent shift of up to several gauss, which is larger than naive temperature-shift estimates by more than an order of magnitude.","The 33.6 G resonance, widely used for creating 39K Bose–Einstein condensates, is only usable at its nominal position in very shallow traps or when the trap light is frequency-tuned away from the molecular resonance.","The extracted polarizabilities imply that the trap light can resonantly drive a3Σ+u → b3Σ+g transitions in the Feshbach molecules, so the trap itself acts as a near-resonant molecular light source.","Because the 33.6 G resonance has negative molecular polarizability, increasing trap depth pushes the resonance upward in field; the sign and size of this shift can serve as a direct diagnostic of the molecular-state character of a resonance.","The dependence of the shift on ODT laser frequency, observed as a decrease for R1 when the laser is shifted by +13.1 GHz, confirms that the enhancement is frequency-selective rather than a generic property of all Feshbach molecules."],"supporting_citations":[{"why":"Supplies the original 39K Feshbach resonance positions, widths, and the first differential magnetic moment δµ = −1.9 μB for the 33.6 G resonance, the reference against which the paper's own δµ is compared.","marker":"[5]"},{"why":"Provides the updated experimental position 33.64(15) G for the 33.6 G resonance used to calibrate the zero-light intercept.","marker":"[10]"},{"why":"Gives a precision position 33.5820(14) G for the R1 resonance, anchoring the assignment and zero-field extrapolation.","marker":"[17]"},{"why":"Supplies theoretical resonance positions used for identifying R3 and other resonances and for comparing the paper's coupled-channel results.","marker":"[25]"},{"why":"Provides the potassium atomic polarizability at 1063.9 nm (598.7 a.u.) that serves as the reference for extracting αmol/αat.","marker":"[43]"},{"why":"Supplies the method of determining a dynamic polarizability relative to potassium, which the extraction procedure copies.","marker":"[44]"},{"why":"Reports the observed a3Σ+u → b3Σ+g bandhead near 1095 nm, the key evidence that the trap-laser frequency is nearly resonant with a molecular transition.","marker":"[47]"},{"why":"Confirms the a3Σ+u → b3Σ+g bandhead near 1096 nm and provides theoretical spectra used to assign the v'=4 upper level.","marker":"[48]"},{"why":"Gives the vibrational level spacing of the b3Σ+g potential used to place the transition within ~50 GHz of the trap laser.","marker":"[52]"},{"why":"Supplies the adopted b3Σ+g interaction potential used for the level assignment and the potential curves in Fig. 8.","marker":"[54]"}],"fun_headline_variants":["Optical trap shifts 39K Feshbach resonances by 7.5 G","39K Feshbach shifts reveal 4-7x molecular polarizability","Trap light shifts 39K Feshbach resonances by 7.5 G","Differential Stark shift drives large 39K Feshbach moves"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The extracted polarizabilities depend on the differential magnetic moment δµ = −2.5 μB for the 33.6 G resonance taken from the authors' own unpublished coupled-channel calculations; if the earlier literature value −1.9 μB is used instead, the extracted molecular polarizability for that resonance changes by roughly 25–30%.","fun_headline_variants_meta":{"raw":{"variants":["Optical trap shifts 39K Feshbach resonances by 7.5 G","39K Feshbach shifts reveal 4-7x molecular polarizability","Trap light shifts 39K Feshbach resonances by 7.5 G","Differential Stark shift drives large 39K Feshbach moves"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000889,"raw_usage":{"total_tokens":3940,"prompt_tokens":1155,"completion_tokens":2785,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":771,"completion_tokens_details":{"reasoning_tokens":2698}},"tokens_in":771,"tokens_out":2785,"duration_ms":20913,"temperature":1.0,"reasoning_tokens":2698,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-08T11:44:30.485242+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the shift of the 33.6 G resonance while tuning the single-frequency trap laser across 1063–1066 nm: if the shift does not trace a dispersion-shaped polarizability curve with a sign change and a magnitude that grows toward the molecular resonance, the near-coincidence explanation is wrong. Alternatively, measure δµ of the R1 molecular state by two-photon or radio-frequency association spectroscopy in zero light; if it comes out at −1.9 μB rather than −2.5 μB, the extracted αmol for the 33.6 G resonance drops by 25–30% and the 'four to seven times' claim is overstated.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the original 39K Feshbach resonance positions, widths, and the first differential magnetic moment δµ = −1.9 μB for the 33.6 G resonance, the reference against which the paper's own δµ is compared."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the updated experimental position 33.64(15) G for the 33.6 G resonance used to calibrate the zero-light intercept."},{"cited_title":"Tanzi, C","cited_arxiv_id":null,"evidence_quote":"Gives a precision position 33.5820(14) G for the R1 resonance, anchoring the assignment and zero-field extrapolation."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies theoretical resonance positions used for identifying R3 and other resonances and for comparing the paper's coupled-channel results."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the potassium atomic polarizability at 1063.9 nm (598.7 a.u.) that serves as the reference for extracting αmol/αat."},{"cited_title":"Ravensbergen, V","cited_arxiv_id":null,"evidence_quote":"Supplies the method of determining a dynamic polarizability relative to potassium, which the extraction procedure copies."},{"cited_title":"Kiruga, C","cited_arxiv_id":null,"evidence_quote":"Reports the observed a3Σ+u → b3Σ+g bandhead near 1095 nm, the key evidence that the trap-laser frequency is nearly resonant with a molecular transition."},{"cited_title":"Vadla, R","cited_arxiv_id":null,"evidence_quote":"Confirms the a3Σ+u → b3Σ+g bandhead near 1096 nm and provides theoretical spectra used to assign the v'=4 upper level."},{"cited_title":"Magnier, M","cited_arxiv_id":null,"evidence_quote":"Gives the vibrational level spacing of the b3Σ+g potential used to place the transition within ~50 GHz of the trap laser."},{"cited_title":"Vexiau, private communication (2026)","cited_arxiv_id":null,"evidence_quote":"Supplies the adopted b3Σ+g interaction potential used for the level assignment and the potential curves in Fig. 8."}],"review_version":1}