{"id":"2dcda434-cf9b-4ab9-bc91-ea2be61ef65f","arxiv_id":"2608.07260","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"A coil-free, all-dielectric cold-atom Rydberg sensor measures microwave fields with 43 dB dynamic range, sub-percent linearity, and 3 µV/cm resolution.","lead":"Physicists built a radiofrequency field sensor from cold rubidium atoms suspended in an all-optical super-molasses trap, with no magnetic coils and a metal-free glass sensor head. The device measures microwave fields through Rydberg atom trap-loss spectroscopy, achieving a self-calibrated 43 dB power range and microvolt-per-centimeter resolution.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Field-free Eq. (1) is used at low AT splittings where unresolved Zeeman structure (invoked to explain the 1.5 MHz linewidth) can shift fitted peak centers, so the low-field end of the 43 dB self-calibrated range is not demonstrated.","rationale":"The reader's weakest_assumption points to reliance on the companion super-molasses-trap preprint. That is a real external dependency, but the present manuscript gives enough in-house atom-number, temperature, and stability data for the sensing results to stand on their own if the companion is accepted. The more acute internal risk is the use of the field-free four-level expression in Eq. (1) at low MW power. The paper explicitly attributes the 1.5 MHz linewidth partly to Zeeman broadening from an ambient field of a fraction of a gauss, yet Eq. (1) assumes degenerate mJ levels. Since μB/h ≈ 1.4 MHz/G, a 0.1-0.5 G field produces differential Zeeman shifts comparable to the low-end AT splittings that define the dynamic-range floor. Unresolved and possibly asymmetric Zeeman components can move the fitted line centers, biasing the 'self-calibrated' E-field values and the claimed 3 μV/cm resolution. This does not invalidate the high-field linearity or the observed 223 MHz maximum splitting, but it means the 43 dB SI-traceable range is not fully supported at the low-field end without a Zeeman analysis or magnetic-field characterization. The proposed B-variation test would settle whether the effect is real. The verdict therefore remains CONDITIONAL, with the added condition that the authors provide a Zeeman-aware low-field calibration or demonstrate empirically that fitted peak centers are B-independent over 0-1 G.","tokens_in":12888,"tokens_out":10098,"duration_ms":99803,"concrete_test":"Re-measure the trap-loss spectrum at a fixed low MW power (AT splitting ≈ 1.5-3 MHz) while applying a small calibrated B-field along the MW propagation axis, varying B from 0 to 1 G. Fit the four peak frequencies; if the zero-field reference ω0 or the resulting Δω_AT shifts by more than the ~70 kHz/√Hz statistical floor over this B range, the symmetric-broadening assumption fails. Then recompute the field values with a full Zeeman Hamiltonian over mF/mJ sublevels; if the corrected low-field calibration differs by more than 3 μV/cm, the 43 dB SI-traceable dynamic range and low-field resolution claims require revision or an explicit B-field uncertainty budget.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central SI-traceable claim rests on Eq. (1), which maps the four Autler-Townes peak frequencies to the MW field using a field-free 4-level Hamiltonian. The paper's own explanation of the 1.5 MHz FWHM linewidth invokes 'Zeeman broadening by the ambient magnetic field on the order of a fraction of a Gauss' (Results, Trap-loss spectroscopy in the absence of cooling light). At the low-power end of the 43 dB dynamic range the AT splitting approaches this linewidth, so the Zeeman splittings of the 5S1/2(F=2) mF sublevels and the 61S1/2/61P1/2 mJ = ±1/2 states are comparable to or larger than the fitted splitting. Equation (1) then requires either resolved Zeeman lines or a model that includes their shifts and transition strengths; the paper instead assumes the unresolved Zeeman structure only broadens the fitted peaks symmetrically. A fraction of a gauss gives differential shifts on the order of 0.1-1 MHz between the relevant mF/mJ components, so the fitted peak centers can be displaced by much more than the reported 70 kHz/√Hz noise floor. If so, the low-field E-field values, the 3 μV/cm resolution, and the low-field portion of the 43 dB range are not self-calibrated as claimed.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports radio-frequency (RF) field sensing using cold 87Rb Rydberg atoms in an all-optical, coil-free super-molasses trap (SMT). Trap-loss spectroscopy on the 5S1/2(F=2) to 61S1/2 two-photon transition is performed with the cooling light off, yielding spectral features as narrow as 1.5 MHz FWHM. In the presence of a 15.973 GHz microwave field coupling the 61S1/2 and 61P1/2 Rydberg states, the spectrum splits into four Autler-Townes lines, and Eq. (1) converts the measured frequencies to the microwave electric-field amplitude using the ARC dipole matrix element. The authors report a 43 dB SI-traceable dynamic range, scale-factor linearity better than 1% over 32 dB, a short-term sensitivity of 70 kHz/√Hz (43 µV/cm/√Hz), a 3 µV/cm resolution after 300 s, long-term stability over several hours, and retrieval of the microwave-field ellipticity from the four-line pattern.","tokens_in":13149,"tokens_out":6985,"duration_ms":68128,"significance":"If the low-field calibration issue is resolved, this is a significant advance: it combines narrow cold-atom trap-loss spectroscopy with a fully dielectric sensor head, removes magnetic coils from the atom source, and offers a polarization-independent Autler-Townes readout that is in principle SI-traceable via Eq. (1). The linearity data in Fig. 4 and the long-term stability measurements in Fig. 7 are valuable quantitative contributions. The explicit use of only one free parameter in the ellipticity model, and the fact that the central frequency-to-field conversion uses an external dipole matrix element rather than a fitted scale factor, are clear strengths of the manuscript.","major_comments":[{"comment":"The low-power end of the claimed self-calibrated dynamic range is not demonstrated. The manuscript attributes the 1.5 MHz FWHM of the trap-loss features to \"Zeeman broadening by the ambient magnetic field on the order of a fraction of a Gauss.\" At the low microwave powers of Figs. 3 and 4, the Autler-Townes splitting approaches this linewidth, and the unresolved Zeeman sublevels of 5S1/2(F=2) and of the 61S1/2 and 61P1/2 states produce differential shifts on the order of 0.1–1 MHz for a fraction of a gauss. Equation (1) is derived from a field-free four-level Hamiltonian, and the paper neither reports a measurement of the ambient field nor includes Zeeman sublevels in the fit model; the asymmetric-Gaussian fits therefore do not guarantee that the fitted centroids are unbiased. Because the 70 kHz/√Hz noise floor, the 3 µV/cm resolution, and the low-field portion of the 43 dB range all rely on these centroids, the self-calibrated low-field claim is not supported. The authors should either resolve or explicitly model the Zeeman structure, measure the ambient B-field, or restrict the SI-traceable claim to splittings for which Zeeman shifts are negligible.","section":"Results, 'Trap-loss spectroscopy in the absence of cooling light' and 'RF measurements', Eq. (1)"},{"comment":"The claim of trap-loss signals \"one order of magnitude narrower than for conventional magneto-optical traps\" is not supported by an in-paper MOT measurement. The tenfold reduction shown in Fig. 2 compares the SMT with cooling light on (12 MHz) and off (1.5 MHz), not a conventional MOT. If the comparison is taken from refs. [19] or [28], the manuscript should quote those quantitative values and state the comparison explicitly; otherwise the abstract's comparative claim should be softened.","section":"Abstract and Results, first paragraph"},{"comment":"The sensor performance depends on SMT characteristics that are not independently characterized in this manuscript: the text cites ref. [25], an unpublished companion preprint with overlapping authors, for the ~10^7 atoms at 300 µK and for the long-term stability needed for the five-hour runs. Because these parameters are load-bearing for the claimed resolution, drift, and metal-free sensor-head operation, the paper should either include a concise in-situ characterization (atom number, temperature, trap lifetime, and long-term fluorescence or pointing stability) or explicitly list which parameters are inherited from ref. [25] together with their uncertainties. This matters for the transferability of the 43 dB range and the 3 µV/cm resolution to other SMT platforms.","section":"Results, first paragraph and Materials and Methods, 'Experimental design'"},{"comment":"For a self-calibrated SI-traceable measurement, the uncertainty in the dipole matrix element d appearing in Eq. (1) should be quantified. The paper gives |d| ≈ 1291.5 e a0 from ARC [34] but does not state its uncertainty, nor does it discuss how the unresolved Zeeman structure or the chosen quantization axis affects the effective d for the four measured peaks. The reported linearity (<1%) and stability figures are relative; an absolute calibration claim requires an uncertainty budget for d and for any angle-, polarization-, or Zeeman-dependent corrections.","section":"RF measurements, Eq. (1) and uncertainty budget"}],"minor_comments":[{"comment":"The y-axis label and the definition of the plotted quantity are not given in the caption; the reader must infer from Eq. (2) that the plotted quantity is the average ̅ω of the four peak positions.","section":"Fig. 5 caption"},{"comment":"Please specify the sign convention and the reference for ω_i: Eq. (1) uses the ω_i as fitted peak frequencies, while Fig. 3 is plotted against a two-photon detuning, and it should be stated explicitly that the ω_i are referenced to the zero-microwave peak position ω_0.","section":"Eq. (1) and Fig. 3"},{"comment":"\"A 3: 3 fibered coupler\" appears to be a typo; please correct to e.g. \"a 1×3 fiber splitter.\"","section":"Materials and Methods, 'Experimental design'"},{"comment":"The statement that the direct two-photon linewidth is \"ultimately limited by the lifetime of the Rydberg state, on the order of 1.5 kHz\" is in tension with the later measured 1.5 MHz; please clarify that in the present experiment saturation and Zeeman broadening dominate, as discussed in the Results.","section":"Introduction, second paragraph"},{"comment":"The confidence intervals in Fig. 7 are computed with the AllanTools function using the Greenhall equivalent-degrees-of-freedom method; please state in the caption that these are estimated 1σ intervals from that framework, so that readers do not mistake them for ordinary standard deviations.","section":"Fig. 7 and Statistical Analysis"},{"comment":"Reference [25] is an unpublished preprint; if a peer-reviewed version becomes available, it should replace or supplement the preprint citation, and the present manuscript should identify which quantitative trap parameters are taken from it.","section":"References [25]"}],"recommendation":"major_revision","confidential_remarks":"The main technical uncertainty is the Zeeman-shift bias at low microwave power, which directly affects the self-calibrated low-field portion of the 43 dB dynamic range and the 3 µV/cm resolution. If the authors resolve this by measuring the ambient field, modeling the Zeeman structure, or restricting the SI-traceable claim, I would support acceptance. The reliance on an unpublished companion preprint should be handled editorially: either the companion should be available in accepted form or the key SMT parameters should be included in this paper. The manuscript is otherwise well-focused and fits the journal's scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The coil-free cold-atom RF sensor is real, and the 43 dB dynamic range claim is plausible, but the low-field end of that range is under-supported: the same unresolved Zeeman structure invoked to explain the 1.5 MHz linewidth also shifts the fitted peak centers, and the model ignores that. This is a revision point, not a rejection.\n\nThe genuinely new thing is combining a super-molasses trap with trap-loss Rydberg spectroscopy in a fully dielectric, coil-free head, exploiting the absence of inductive loads to switch cooling off during excitation. That switch narrows the line from 12 MHz to 1.5 MHz, which is what makes the large dynamic range possible. The linearity data in Fig. 4 and the five-hour Allan deviation in Fig. 7 are clean. The four-line AT analysis with a one-parameter ellipticity fit is transparent, and the central field calibration through Eq. (1) uses an external ARC dipole matrix element, so it is not circular.\n\nThe weaknesses are proportionate. The trap itself is not characterized here; it is borrowed from a companion preprint by the same groups, so the atom number, temperature, and dielectric-head stability are assumptions. The 'ten times narrower than MOT' claim is not directly measured against a MOT in this paper; it is a comparison to earlier work. The Zeeman issue is the one that should bother a referee most. The paper explains the 1.5 MHz linewidth by Zeeman broadening from the ambient field, yet the fitting model is field-free. At splittings of a few MHz, the Zeeman splittings of the mJ/mF components are comparable to the splitting, and the symmetric-broadening assumption is not demonstrated. The linearity intercept being 0.08 +/- 0.10 MHz is reassuring, but it does not by itself bound the systematic error over the whole low-power range. Data are only 'available upon request,' which is fine but keeps independent checks limited.\n\nThis paper deserves a serious referee. The core demonstration is credible, the limitations are discussed honestly, and the advance is real. The reviewer should ask for a quantitative Zeeman analysis of the low-field AT peaks, an in-situ characterization of the SMT as used, and ideally public data. I'd accept it for review.","headline":"Credible and worth refereeing, but the unresolved Zeeman structure at low AT splittings means the self-calibrated low-field part of the 43 dB range is not yet demonstrated.","tokens_in":13794,"tokens_out":6726,"would_cite":true,"duration_ms":66212,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A coil-free cold-atom sensor reads microwave fields over a 43 dB range","keywords":["Rydberg atoms","super-molasses trap","trap-loss spectroscopy","Autler-Townes splitting","microwave electrometry","quantum sensing","cold atoms","radio-frequency field measurement"],"falsifier":"Feed a known 15.973 GHz field from a power-calibrated source into a second, identically constructed dielectric head and compare the amplitude returned by Eq. (1) with the independently measured value; any deviation beyond the claimed 1% linearity over the 32 dB range, or a shift of the fitted peak frequencies when the ambient magnetic field is varied by a fraction of a gauss, would falsify the self-calibrating assumption.","tokens_in":12685,"feed_emoji":"📡","tokens_out":12253,"duration_ms":113888,"temperature":0.7,"pith_summary":"This paper reports a radiofrequency field sensor built from cold rubidium Rydberg atoms held in a super-molasses trap, an all-optical trap that needs no magnetic coils and lets the sensor head be made entirely of dielectric materials. The key trick is to switch the cooling light off for 10 microseconds of every 200 and excite atoms from the ground state directly to a Rydberg state while watching loss of atoms from the trap. That narrows the spectral features from about 12 MHz to 1.5 MHz, roughly tenfold, and makes the measured Autler-Townes splitting a self-calibrated, SI-traceable readout of the microwave field amplitude. The paper reports a 43 dB dynamic range, better than 1% linearity over 32 dB, a resolution of 3 µV/cm after 300 seconds, and extraction of the field's ellipticity from the four spectral lines.","feed_headline":"Coil-free Rydberg sensor tracks microwaves over 43 dB","feed_subtitle":"An all-optical trap with no metal near the atoms gives 1.5 MHz lines and self-calibrated, SI-traceable field readings.","key_machinery":"The central object is the super-molasses trap: an all-optical near-resonance trap that loads about $10^{7}$ rubidium atoms at 300 µK directly from background vapor in a glass cell, with no magnetic field gradient and no metal near the atoms. The mechanism that carries the argument is trap-loss spectroscopy with the cooling light off, which removes lightshifts and ground-state Autler-Townes mixing and narrows the lines from 12 MHz to 1.5 MHz. The quantitative identity at the core is Eq. (1), which combines the four measured Autler-Townes frequencies into a single field amplitude regardless of polarization; it works because an elliptically polarized microwave field couples both circular components, each producing its own doublet, and the quadrature sum of the splittings is proportional to the total electric field through the dipole matrix element.","core_discovery":"The central claim is that a super-molasses trap, an all-optical trap producing about $10^{7}$ 87Rb atoms at roughly 300 µK in a glass-and-ceramic head, can serve as a practical microwave field sensor without any magnetic coils. By sending the two-photon Rydberg lasers only when the cooling light is off, the authors obtain trap-loss fluorescence dips with 1.5 MHz full width at half maximum. In the presence of a 15.973 GHz microwave field these dips become four Autler-Townes lines, and the field amplitude is recovered from their frequencies through the polarization-independent identity $\\Delta\\omega_{\\mathrm{AT}} = \\sqrt{\\omega_3^2+\\omega_4^2}/2 + \\sqrt{\\omega_2^2+\\omega_1^2}/2 = dE/\\hbar$, with $d$ the known dipole matrix element. The same data give the microwave detuning from the average line position, and the four-line pattern gives the ellipticity of the field. The authors report a 43 dB SI-traceable dynamic range and long-term stability that reaches 3 µV/cm resolution after 300 seconds of averaging.","pith_inferences":["The 1.5 MHz linewidth is still about three times the estimated 520 kHz Doppler width for 300 µK atoms, so the low-field limit and hence the dynamic range could be extended if the residual broadening from cooling-laser noise, polarization fluctuations, and ambient magnetic field were removed.","The paper's own stability data show the short-term noise is set by roughly 10% fluorescence fluctuations from the cooling light; stabilizing the laser frequency and polarization would likely lower the 43 µV/cm/√Hz sensitivity directly.","The trap-loading-limited response time of about a fraction of a second is an artifact of the trap-loss readout, not of the Rydberg resonance; replacing fluorescence monitoring with probe-beam absorption, as the paper proposes, should allow much faster sampling at the same or better sensitivity.","Since the super-molasses trap is all-optical, several such traps could be formed in one vacuum chamber, which points toward arrayed or imaging RF sensing; this is a future possibility raised by the authors rather than a result demonstrated here."],"forward_implications":["Microwave power can be calibrated traceably to SI with a compact, fully dielectric sensor head, because the conversion from measured splitting to field amplitude uses only atomic constants and the known dipole matrix element.","The combination of 1.5 MHz wide lines and splittings up to 223 MHz gives a 43 dB dynamic range, with an inferred upper bound near 49 dB set by coupling to the neighboring 61P3/2 state.","Long averaging reduces the frequency determination to about 6 kHz, corresponding to 3 µV/cm, and the measured power stays stable to about 1.5×10^-3 over thousands of seconds.","Because no inductive coils are present, the cooling and measurement phases can be alternated rapidly, which is what makes the narrow trap-loss lines possible and could be reused in other cold-atom sensors such as magnetometers, gravimeters, or clocks.","The four-line spectrum directly reveals the ellipticity of an applied microwave field, so amplitude and polarization information are available from a single scan."],"supporting_citations":[{"why":"The super-molasses trap: the all-optical, coil-free atom source at 10^7 atoms and 300 µK that makes the dielectric sensor head possible.","marker":"(25)"},{"why":"Trap-loss spectroscopy with cold Rydberg atoms and the Autler-Townes model for extracting microwave field amplitudes; this paper extends that protocol to the coil-free trap with the cooling light off.","marker":"(19)"},{"why":"Supplies the dipole matrix element d = 1291.5 e a0 and neighboring-Rydberg-state data used in Eq. (1) and in the ellipticity fit.","marker":"(34)"},{"why":"Provides the standing-wave picture used to model the ~10% inhomogeneous broadening of the spectral lines at high microwave power inside the glass cell.","marker":"(35)"},{"why":"The SI-traceable vapor-cell Autler-Townes baseline against which the reported sensitivity is compared.","marker":"(36)"},{"why":"Earlier cold-atom Rydberg electrometry in a magneto-optical trap, the baseline showing ~1.1 MHz EIT linewidth and 28 dB dynamic range that this work improves.","marker":"(15)"},{"why":"A three-photon vapor-cell scheme with sub-200 kHz linewidth and 17 dB dynamic range, used as a comparison for the SI-traceable range.","marker":"(12)"},{"why":"Supports the claim that cold atoms give sub-millimeter local field sampling, which mitigates inhomogeneous broadening at high RF power.","marker":"(22)"}],"fun_headline_variants":["No-coil Rydberg trap gives 43 dB microwave dynamic range","All-optical Rydberg atoms self-calibrate microwave fields","Cold Rydberg atoms sense radiofrequency without magnetic coils","43 dB range from coil-free Rydberg microwave sensor","Super-molasses Rydberg trap delivers self-calibrated RF sensing"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole demonstration assumes the super-molasses trap from the companion work performs as described, about $10^{7}$ rubidium atoms at 300 µK in a dielectric head that stays stable over five-hour runs, and that the ambient magnetic field of a fraction of a gauss only broadens the lines without shifting the fitted peak frequencies used in the calibration formula.","fun_headline_variants_meta":{"raw":{"variants":["No-coil Rydberg trap gives 43 dB microwave dynamic range","All-optical Rydberg atoms self-calibrate microwave fields","Cold Rydberg atoms sense radiofrequency without magnetic coils","43 dB range from coil-free Rydberg microwave sensor","Super-molasses Rydberg trap delivers self-calibrated RF sensing"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000747,"raw_usage":{"total_tokens":3368,"prompt_tokens":1023,"completion_tokens":2345,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":639,"completion_tokens_details":{"reasoning_tokens":2255}},"tokens_in":639,"tokens_out":2345,"duration_ms":16391,"temperature":1.0,"reasoning_tokens":2255,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T11:32:06.771206+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Feed a known 15.973 GHz field from a power-calibrated source into a second, identically constructed dielectric head and compare the amplitude returned by Eq. (1) with the independently measured value; any deviation beyond the claimed 1% linearity over the 32 dB range, or a shift of the fitted peak frequencies when the ambient magnetic field is varied by a fraction of a gauss, would falsify the self-calibrating assumption.","supporting_citations":[],"review_version":1}