{"id":"81046eae-230d-46ff-9a40-b9df4f643e0c","arxiv_id":"2507.16758","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Scanning Josephson tunneling microscopy reveals frustrated Josephson interference between an Nb tip and two superconducting condensates in FeSe, with anti-correlated superfluid density modulations.","lead":"Using a scanning Josephson tunneling microscope, the authors watched how paired electrons tunnel between a niobium tip and two different superconducting condensates in iron-based FeSe. The two condensates interfere with each other, producing frustrated Josephson coupling and anti-correlated superfluid modulations in space.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The two-band premise is load-bearing twice: the frustration inequality must be tested against the true global minimum gap, and the n_i maps are just squared fitted gaps, so the -0.72 anti-correlation may be a fit artifact. A three-band reanalysis of the same DOS and images would settle both.","rationale":"The paper does something hard and largely follows its own logic: the s++ control (Supp. Fig. 11) would be violated if the data were s-wave, and the calibration against Nb-Nb junctions is a reasonable internal standard. My concern is not about the honesty of the fits but about how much of the conclusion is carried by the N=2 assumption. The first signature (lambda_e<1) is a real consequence of the data only if Delta_eff,e is the global minimum effective gap; FeSe's multiple electron pockets make this non-obvious. The second signature (anti-correlated n_i) is even more directly tied to the fit, because Eq. 12 makes n_i proportional to the squared fitted gap for that band. If the fit is well-constrained by the spectra, the anti-correlation could be physical; if the two components trade spectral weight, it would appear even for a single underlying condensate. The proposed three-band reanalysis is the decisive check. I therefore keep the conditional verdict; if the reanalysis confirms robustness, the paper would justify acceptance, and if not, the central claim would need to be downgraded to a model-dependent interpretation.","tokens_in":12599,"tokens_out":10516,"duration_ms":110754,"concrete_test":"Reanalyze the raw deconvoluted dI/dV maps with a three-band model (Gamma hole plus two electron pockets with independent effective gaps), compute the global Delta_eff,min(r), and re-test I_c(r)R_n(r) < (pi/4) Delta_eff,min(r) at every pixel; also recompute the n_h-n_e cross-correlation from the three-band fitted gaps and compare with -0.72. If the inequality fails at any pixel or the anti-correlation weakens substantially, the central claims are not robust. As a control, bootstrap the existing two-band fit from many starting points to check whether the anti-correlation coefficient is stable or fit-driven.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claim is Eq. 8/5: I_c R_n < (pi/4) Delta_eff,min. In the paper this bound is evaluated with the minimum of the two fitted effective gaps (Gamma hole, X electron). FeSe has additional Fermi-surface sheets and nematically split electron pockets; if any unmodeled pocket has Delta_eff smaller than the fitted Delta_eff,e, the observed inequality can be satisfied without sign-changing interference. The test is therefore conditional on the N=2 model being exhaustive. The claim is also weakened internally: in Eq. 8, lambda_1 = Delta_eff/Delta_eff,h is always <1 whenever Delta_eff,h > Delta_eff,e and epsilon>0, so only lambda_2<1 carries evidential weight. For the multi-condensate visualization, Eq. 12 combined with Eq. 2 gives n_i proportional to (Delta_eff,i)^2; the n_i maps are therefore deterministic functions of the two fitted gap maps, not independent superfluid-density measurements. The reported cross-correlation -0.72 is exactly the correlation of the fitted hole- and electron-gap maps. If the two-gap DOS fit has compensating parameters, the anti-correlation in Fig. 4 could be an artifact. Thus the same unverified premise (exactly two channels with the fitted gaps) is load-bearing for both the frustration signature and the visualization claim.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports scanning Josephson tunneling microscopy (SJTM) measurements on Nb–FeSe junctions at 0.3 K. The authors simultaneously measure the maximum Josephson current I_J, the junction resistance R_n, and quasiparticle tunneling spectra at the same atomic locations. They fit the FeSe density of states with a two-band anisotropic gap model, extract effective gaps Δ_eff,h and Δ_eff,e for the hole (Γ) and electron (X) pockets, and form the ratios λ_i = Δ_eff/Δ_eff,i. They find λ_1, λ_2 < 1, which they interpret as quantitative evidence of frustrated Josephson tunneling caused by the sign-changing s± order parameter of FeSe. Under the assumption of exactly two tunneling channels with χ_1 = 0 and χ_2 = π, they invert the total I_c and R_n into channel-resolved currents and resistances (Eqs. 9–10), image superfluid densities n_i, and report a cross-correlation of –0.72 between the two condensate maps. A two-component Ginzburg-Landau model with inter-band Josephson coupling (η_1) and gradient inter-band scattering (ν) is used to argue that anti-correlated superfluid modulations require νq^2 + η_1 > 0.","tokens_in":12944,"tokens_out":6142,"duration_ms":62545,"significance":"If correct, this work would be the first atomic-scale observation of frustrated Josephson coupling in an s± superconductor and the first simultaneous visualization of two superconducting condensates by SJTM. The experimental effort is impressive: simultaneous Josephson and quasiparticle spectroscopy at 0.3 K, internal calibration against Nb-Nb junctions, and a valuable internal check in which the s++ assignment produces unphysical negative channel resistances (Supplementary Fig. 11). The quantitative inequality test based on Eq. (5) is a promising and falsifiable approach. However, several load-bearing assumptions and missing uncertainty analyses must be addressed before the central claims are established.","major_comments":[{"comment":"The paper presents λ1 < 1 and λ2 < 1 as two independent pieces of evidence, but λ1 < 1 carries no evidential weight. From Eq. (8), Δ_eff = (Δ_eff,h − ε Δ_eff,e)/(1 + ε), so for any ε > 0 and Δ_eff,h > Δ_eff,e, λ1 = Δ_eff/Δ_eff,h is strictly less than 1 by construction. Only λ2 < 1 is a nontrivial test of frustration. The central quantitative claim should be reformulated around λ2 and Eq. (5), and the text in §2 and the Fig. 2i caption should be corrected accordingly.","section":"§2, Eq. (8)"},{"comment":"Both the frustration test and the channel-resolved inversion assume that FeSe is fully described by exactly two tunneling channels with the fitted effective gaps Δ_eff,h and Δ_eff,e. FeSe has additional Fermi-surface sheets and the electron pockets are nematically split; if any unmodeled sheet has an effective gap smaller than Δ_eff,e, then Eq. (5) can be satisfied without sign-changing interference, and the solution of Eqs. (9)–(10) is not the unique physical decomposition. Please provide a three-band (or at least a sensitivity) analysis of the same data, or a quantitative justification for why all other sheets contribute negligibly to Cooper-pair tunneling.","section":"§2, Eqs. (5), (9)–(10)"},{"comment":"No error bars are reported for I_c, R_n, Δ_eff,i, or λ_i. The central claim that “both indices are smaller than unity” is a statistical statement, and the histograms in Fig. 2i do not show uncertainties. The authors should propagate uncertainties from the measurement noise, the Nb-tip calibration, the DOS deconvolution, and the fitting parameters, and show that λ2 < 1 (the nontrivial condition) is robust under these variations. Without this, the quantitative inequality test is not fully supported.","section":"Figs. 2i and 3h"},{"comment":"The visualization claim is presented more strongly than the data support. From Eq. (2) and Eq. (12), n_i(r) ∝ [I_c,i(r) R_n,i(r)]^2 ∝ Δ_eff,i(r)^2, so Figs. 4a,b are essentially the squares of the two fitted gap maps, not independent measurements of superfluid density. The reported cross-correlation of –0.72 is therefore a property of the two-gap fit output and may reflect compensating parameters in the fit. To support the anti-correlation and the Ginzburg-Landau interpretation of Eq. (16), show that the anti-correlation persists when the fit is varied within its confidence region or when additional bands are included.","section":"§4, Eq. (12)"}],"minor_comments":[{"comment":"The caption mislabels the second and third resistance panels as “b” and “c”; they should be “e” and “f” to match the figure.","section":"Fig. 3 caption"},{"comment":"The phrase “anti correlated” is missing a hyphen and should be “anti-correlated” throughout. Also, the term “multi-condensate visualization” should be calibrated to reflect that the condensate maps are derived from the gap fits (see Major Comment 4).","section":"Abstract and main text"},{"comment":"The notation for the free energy per unit length in Eq. (16) is difficult to parse; please rewrite it with conventional integral notation.","section":"§3, Eq. (16)"},{"comment":"The fitted parameters Δ1,max, Δ2,max, a1, and a2 are reported, but the text does not state explicitly how Δ_eff,i is computed from these parameters. The definition ⟨Δ_i(k)⟩ over the Fermi surface is not sufficient for a twofold-symmetric gap; please give the explicit formula.","section":"Fig. 2c and Supplementary Note 3"}],"recommendation":"major_revision","confidential_remarks":"This is a high-profile claim, and the missing error analysis and the N=2 sensitivity analysis are the key obstacles. The paper may be acceptable after a major revision that addresses the tautological nature of λ1<1, the uniqueness of the two-channel decomposition, and the interpretation of the n_i maps."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First, the punchline: this is a real experimental advance, but the two big claims are both riding on the same two-band assumption, and that assumption is shakier than the text admits. The raw data and the internal s++ check are the strongest parts; the interpretation is more conditional than the paper says.\n\nWhat's actually new: the paper applies scanned Josephson tunneling microscopy to FeSe, which nobody has done, and it reports a clean, reproducible sublinear I_c vs 1/R_n relation (Fig. 1g) that deviates from the s-wave expectation. That is a raw observation, not fitting-dependent. The internal consistency check is also good: assuming s++ pairing yields unphysical negative channel resistances, which supports the s± interpretation. The simultaneous measurement of Josephson current and quasiparticle spectra at atomic scale is technically demanding, and the authors deserve credit for pulling it off.\n\nWhere I worry: the quantitative test of Eq. 5 uses Delta_eff,min taken from just two fitted gaps. FeSe has additional Fermi surface sheets and nematic-split electron pockets. If any of those carries an effective gap smaller than the fitted X-pocket gap, the inequality I_c R_n < (pi/4) Delta_eff,min can be satisfied without any sign-changing interference. So the central claim \"first direct demonstration of frustrated Josephson tunneling\" is conditional on the N=2 model being exhaustive, and the paper doesn't show that. Also, as written, lambda_1 < 1 is nearly automatic whenever the hole gap exceeds the electron gap, so only the lambda_2 < 1 test carries weight.\n\nThe multi-condensate visualization has a similar problem. From Eqns. 9-12, the n_i maps are deterministic functions of the same fitted gap maps and the measured I_c and R_n. They are not independent superfluid-density measurements; they are those fitted gaps squared, rescaled. The -0.72 anti-correlation in Fig. 4 is therefore only as convincing as the two-gap fit. If the fit has compensating parameters, the anti-correlation could be an artifact. The paper does not provide error bars on the extracted I_c or on the fitted gap parameters, so the reader can't tell how much of the variance in n_i is meaningful.\n\nThe GL discussion (Eqs. 13-16) is fine as qualitative rationalization, but it is not tested against any alternative, so it doesn't add evidential weight.\n\nBottom line: the experimental work is solid and the s++ check is a good falsifier, but the headline claims are not as firm as the abstract and main text imply. The paper deserves a serious referee, but the referee should demand error propagation throughout, a sensitivity test to a three-band model, and a release of the fitting residuals and maps. I'd read it again in a reading group, but I would not cite it yet as evidence for frustrated Josephson coupling.","headline":"A serious, well-executed SJTM study whose two headline claims are more conditional than the text lets on, because both load the same under-tested N=2 model.","tokens_in":13479,"tokens_out":2667,"would_cite":false,"duration_ms":27829,"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":"Atomic-scale Josephson tunneling between Nb and FeSe shows destructive interference from sign-changing gaps, giving the first condensate-resolved superfluid images.","keywords":["Josephson junction","FeSe","s± wave pairing","scanned Josephson tunneling microscopy","multi-band superconductivity","superfluid density imaging","inter-band scattering","0-π transition"],"falsifier":"Measure $I_c(\\mathbf{r})$ and $R_n(\\mathbf{r})$ on a clean FeSe region with a Nb tip while independently extracting the two gap maps, and test whether every pixel satisfies $I_c R_n \\ge (\\pi/4)\\Delta_{\\mathrm{eff,min}}$ once the two measured gaps are used. If the inequality holds at all pixels, the destructive-interference interpretation fails. A second check is to repeat the identical protocol on a known $s^{++}$ two-band superconductor: the model predicts no sub-unity $\\lambda_i$ and no anti-correlation.","tokens_in":12418,"feed_emoji":"⚛️","tokens_out":9477,"duration_ms":90459,"temperature":0.7,"pith_summary":"The authors aim to show that a Josephson junction between a superconducting Nb tip and the multiband superconductor FeSe exhibits frustrated Josephson coupling, the quantum interference of Josephson currents flowing through the hole and electron pockets whose order parameters have opposite signs. They report that the normalized critical current falls below the single-band bound, $I_c R_n < (\\pi/4)\\Delta_{\\mathrm{eff,min}}$, and that both gap-ratio indices $\\lambda_1,\\lambda_2$ are smaller than one, which is quantitative evidence of destructive interference. They then use an exact two-channel inversion to image the two condensates separately, finding that their superfluid densities are anti-correlated in space. If correct, this is the first direct observation of frustrated Josephson coupling together with simultaneous atomic-scale visualization of two superconducting condensates, a capability the paper argues opens new directions for multiband superconductivity.","feed_headline":"Frustrated Josephson coupling appears atom by atom in FeSe","feed_subtitle":"Suppressed critical currents and anti-correlated superfluid densities reveal FeSe's sign-changing gaps at atomic scale.","key_machinery":"The central object is the $N=2$ inversion of the Ambegaokar-Baratoff equations. In a junction with two bands and an s-wave tip, the total critical current is the phase-sensitive sum $I_c = I_{c,1}\\cos\\chi_1 + I_{c,2}\\cos\\chi_2$, with $\\chi_2-\\chi_1 = \\pi$ for $s^\\pm$ pairing. Combining this with the asymmetric A-B formula $I_{c,i} R_{n,i} = (\\pi/4)\\Delta_{\\mathrm{eff},i}$ and the parallel-resistance relation $1/R_n = 1/R_{n,1}+1/R_{n,2}$ makes the four channel quantities exactly solvable from measured $I_c$, $R_n$, $\\Delta_{\\mathrm{eff},1}$, and $\\Delta_{\\mathrm{eff},2}$. The two-condensate Ginzburg-Landau free energy with inter-band Josephson and gradient-coupling terms then explains the anti-correlation: the observed near-$\\pi$ spatial phase between superfluid-density modulations indicates that inter-band scattering contributes significantly relative to the Josephson coupling.","core_discovery":"The paper's central claim is that a superconducting Nb tip tunnel-coupled to FeSe measures a normalized Josephson critical current below the single-band lower bound, $I_c R_n < (\\pi/4)\\Delta_{\\mathrm{eff,min}}$ with both $\\lambda_1 \\equiv \\Delta_{\\mathrm{eff}}/\\Delta_{\\mathrm{eff},1}$ and $\\lambda_2 \\equiv \\Delta_{\\mathrm{eff}}/\\Delta_{\\mathrm{eff},2}$ below unity, and that this suppression is the signature of destructive interference between Josephson currents from the hole pocket at $\\Gamma$ and the electron pocket at $X$, whose gaps have opposite signs. Using the exactly solvable $N=2$ case, the paper decomposes the measured total critical current and resistance into per-channel values, imaging the two condensates' superfluid densities $n_i(\\mathbf{r})$ separately. These images show spatially anti-correlated modulations (cross-correlation $-0.72$), which the authors attribute to the competition between inter-band Josephson coupling and inter-band scattering, and the channel-resolved transparency ratio $\\varepsilon(\\mathbf{r}) = R_{n,1}/R_{n,2}$ varies with junction resistance in a way that suggests approach to a $0$-$\\pi$ transition.","pith_inferences":["A natural next test is to repeat the protocol on a known $s^{++}$ two-band superconductor or on a single-band control; the model predicts both $\\lambda$ indices should stay at or above unity and no anti-correlation should appear.","The same $N=2$ inversion could be applied to other two-band systems, but for materials with more than two bands the channel-resolved images would require a generalized inversion with extra constraints; the published two-band result may not transfer directly.","The Ginzburg-Landau argument implies that the anti-correlation should grow stronger at shorter modulation wavelengths, so measuring the cross-correlation as a function of Fourier-filter bandwidth is a testable prediction the paper does not explicitly make."],"forward_implications":["FeSe-Nb junctions provide a local, quantitative test of $s^\\pm$ pairing: the inequality $I_c R_n < (\\pi/4)\\Delta_{\\mathrm{eff,min}}$ is the predicted signature of destructive interference and would be absent for sign-preserving two-band pairing.","The channel-resolved inversion converts scanned Josephson tunneling microscopy into a condensate-resolved microscope, letting future experiments image superfluid densities per band near vortices, twin boundaries, and impurities.","The observed anti-correlation of $n_1(\\mathbf{r})$ and $n_2(\\mathbf{r})$ implies that inter-band scattering is not negligible on atomic length scales and should be included in models of FeSe's local superconducting response.","The decrease of $I_c R_n$ with increasing $\\varepsilon = R_{n,1}/R_{n,2}$, together with the sublinear $\\sqrt{I_J}$ versus $1/R_n$ scaling, indicates that sufficiently transparent SJTM junctions could reach the long-sought $0$-$\\pi$ transition."],"supporting_citations":[{"why":"Provides the Josephson-current formulas for a two-band s++ versus s± superconductor coupled to an s-wave electrode, including the 0-π transition criterion used to interpret the sublinear scaling.","marker":"[6]"},{"why":"Supplies the asymmetric Ambegaokar-Baratoff relation and the inequality $I_c R_n \\ge (\\pi/4)\\Delta_{\\mathrm{eff,min}}$ that the FeSe data are compared with.","marker":"[3]"},{"why":"Establishes the s± picture for iron-based superconductors and frames FeSe as the prototypical case studied here.","marker":"[9]"},{"why":"Provides the orbital-selective, twofold-symmetric hole and electron gap structure used to fit the deconvoluted FeSe density of states.","marker":"[25]"},{"why":"Introduces atomic-scale scanning Josephson spectroscopy, the experimental basis for measuring local Josephson currents and gaps.","marker":"[10]"},{"why":"Demonstrates SJTM visualization of superfluid density in a single-band superconductor and the linear $\\sqrt{I_J} \\sim 1/R_n$ relation that FeSe is shown to deviate from.","marker":"[13]"},{"why":"Gives the Ivanchenko-Zil'berman phase-diffusion model connecting the measured maximum Josephson current to the critical current $I_c$.","marker":"[26]"},{"why":"Supplies the two-condensate Ginzburg-Landau free energy with inter-band Josephson and gradient coupling used to explain the anti-correlated superfluid modulations.","marker":"[30]"}],"fun_headline_variants":["Atomic-scale view of frustrated Josephson coupling in FeSe","FeSe shows frustrated Josephson coupling at atomic scale","SJTM images anti-correlated superfluid densities in FeSe","Frustrated Josephson tunneling visualized atom by atom in FeSe","Multi-condensate imaging reveals 0-pi transition tendency in FeSe"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that FeSe is effectively a two-band superconductor at the measured energies, so that the total junction current and resistance can be uniquely split into one hole-pocket channel and one electron-pocket channel.","fun_headline_variants_meta":{"raw":{"variants":["Atomic-scale view of frustrated Josephson coupling in FeSe","FeSe shows frustrated Josephson coupling at atomic scale","SJTM images anti-correlated superfluid densities in FeSe","Frustrated Josephson tunneling visualized atom by atom in FeSe","Multi-condensate imaging reveals 0-pi transition tendency in FeSe"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000231,"raw_usage":{"total_tokens":1508,"prompt_tokens":990,"completion_tokens":518,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":606,"completion_tokens_details":{"reasoning_tokens":429}},"tokens_in":606,"tokens_out":518,"duration_ms":5128,"temperature":1.0,"reasoning_tokens":429,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T15:02:52.205718+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure $I_c(\\mathbf{r})$ and $R_n(\\mathbf{r})$ on a clean FeSe region with a Nb tip while independently extracting the two gap maps, and test whether every pixel satisfies $I_c R_n \\ge (\\pi/4)\\Delta_{\\mathrm{eff,min}}$ once the two measured gaps are used. If the inequality holds at all pixels, the destructive-interference interpretation fails. A second check is to repeat the identical protocol on a known $s^{++}$ two-band superconductor: the model predicts no sub-unity $\\lambda_i$ and no anti-correlation.","supporting_citations":[{"cited_title":"Josephson effect between a two-band superconductor with s++ or s± pairing symmetry and a conventional s-wave superconductor","cited_arxiv_id":null,"evidence_quote":"Provides the Josephson-current formulas for a two-band s++ versus s± superconductor coupled to an s-wave electrode, including the 0-π transition criterion used to interpret the sublinear scaling."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the asymmetric Ambegaokar-Baratoff relation and the inequality $I_c R_n \\ge (\\pi/4)\\Delta_{\\mathrm{eff,min}}$ that the FeSe data are compared with."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the orbital-selective, twofold-symmetric hole and electron gap structure used to fit the deconvoluted FeSe density of states."},{"cited_title":"T., Feldman, B","cited_arxiv_id":null,"evidence_quote":"Introduces atomic-scale scanning Josephson spectroscopy, the experimental basis for measuring local Josephson currents and gaps."},{"cited_title":"X., Sharma, R","cited_arxiv_id":null,"evidence_quote":"Demonstrates SJTM visualization of superfluid density in a single-band superconductor and the linear $\\sqrt{I_J} \\sim 1/R_n$ relation that FeSe is shown to deviate from."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the Ivanchenko-Zil'berman phase-diffusion model connecting the measured maximum Josephson current to the critical current $I_c$."},{"cited_title":"& Speight, M","cited_arxiv_id":null,"evidence_quote":"Supplies the two-condensate Ginzburg-Landau free energy with inter-band Josephson and gradient coupling used to explain the anti-correlated superfluid modulations."}],"review_version":1}