{"id":"222e6526-e4f1-40ae-bc8f-c3091b59d2fe","arxiv_id":"1908.08136","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"FeSe superconductivity is controlled by Hund's coupling strength and the proximity of the Fe dxy orbital to the Fermi level, with the monolayer on SrTiO3 naturally realizing a near-optimal combination.","lead":"This paper uses a computer model of FeSe, an iron-based superconductor, to show that the strength of a magnetic interaction called Hund's coupling, together with how close a particular electron orbital sits to the Fermi energy, controls the material's superconducting transition temperature.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Interface result depends on a single unverified Se height; a 0.02 Å shift in h could collapse the claimed five-fold lambda enhancement.","rationale":"The reader identified the fixed Se height h as the weakest assumption, and I agree that this is the single most load-bearing point. The paper's central claim is that d_xy proximity to E_F plus Hund's J controls the superconducting glue, and the application to the FeSe/STO interface hinges on h=1.40 Å placing d_xy near E_F. Because lambda is exponentially sensitive to the d_xy position (bulk lambda drops from 0.067 to 0.003 when h is reduced by 0.19 Å), and the interface h is taken from an external calculation whose accuracy the authors explicitly distrust for this quantity, the qualitative contrast between M-FeSe and M-FeSe/STO could be an artifact of the structural input. The paper does provide independent support for the general mechanism through the J-scans and doped-bulk calculations, so the work still merits conditional acceptance. However, the quantitative five-fold enhancement claim requires a structural sensitivity check; without it, the interface-specific conclusion is not yet established. Therefore the reader's CONDITIONAL verdict remains appropriate.","tokens_in":12437,"tokens_out":3464,"duration_ms":35527,"concrete_test":"Scan h for the M-FeSe/STO slab over 1.38, 1.39, 1.40, 1.41, and 1.42 Å, recomputing at each height the QSGW band structure, cRPA U and J, DMFT self-energy, and the leading Eliashberg eigenvalue lambda while keeping all other settings fixed. Also repeat the h=1.39 and h=1.41 cases with J varied by ±0.03 eV around the cRPA value. If lambda stays above 0.2 across the full 1.39–1.41 Å range and is insensitive to J shifts, the concern is resolved. If lambda falls below 0.05 at h=1.39 or h=1.41, the claimed five-fold enhancement is not robust to the accepted uncertainty in the structural input.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central contrast between M-FeSe/STO (lambda=0.34) and free-standing M-FeSe (lambda=0.002) rests entirely on the Fe 3d_xy band position, which the authors themselves state is critical and cannot be trusted from DFT: 'Its value is critical, as we have seen in the bulk case, and we cannot rely on DFT for it.' The Se height h is fixed to 1.40 Å for the interface and 1.39 Å for the monolayer, taken from the external DFT+DMFT relaxation of Ref. [44], and no sensitivity analysis is performed for the interface. The fragility is evident from the paper's own bulk results: reducing h from 1.463 Å to 1.27 Å changes lambda from 0.067 to 0.003, and moving J from 0.60 to 0.68 eV changes bulk lambda from 0.067 to 0.9. If the true h in the real M-FeSe/STO interface differs from 1.40 Å by only a few hundredths of an Ångström in the direction that pushes d_xy below the Fermi level, the predicted lambda could drop by more than an order of magnitude, eliminating the qualitative distinction between the interface and the monolayer. The paper acknowledges this sensitivity and the absence of a buffer layer, but the headline conclusion that the STO substrate restores d_xy proximity to E_F is not yet backed by a structural robustness check.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper uses a combined QSGW+DMFT+BSE+Eliashberg (QSGW++) framework with cRPA-derived U and J to study superconductivity in bulk FeSe, a free-standing monolayer (M-FeSe), and a monolayer on SrTiO3 (M-FeSe/STO). The central claim is that the superconducting pairing eigenvalue lambda is controlled by two conditions: the Fe 3dxy orbital must lie close to the Fermi level, and the Hund's coupling J must be large enough to produce incoherent 'bad metal' behavior. The authors show that bulk FeSe has strong low-energy spin fluctuations near q=(1/2,1/2), that these fluctuations are dominated by dxy, and that small increases in J strongly enhance lambda. They then argue that in M-FeSe the dxy band is pushed below EF, killing the spin fluctuations and reducing lambda to 0.002, whereas in M-FeSe/STO the substrate restores dxy near EF, giving lambda=0.34, which they interpret as explaining the high observed Tc. They also study reduced Se height and electron doping as controlled perturbations around the ab initio reference.","tokens_in":12670,"tokens_out":5521,"duration_ms":54931,"significance":"If the central claim holds, the paper provides a unified orbital-specific mechanism for superconductivity across bulk, monolayer, and interface FeSe: Hund's coupling and the proximity of dxy to EF control the intensity of low-energy antiferromagnetic spin fluctuations, which in turn control the pairing eigenvalue. The work has notable strengths: the cRPA values of U and J are not fitted to Tc; the J-scan is presented explicitly as a sensitivity study rather than a fit; the computed Im chi(q,omega) is benchmarked against inelastic neutron scattering in a companion study; and the orbital-resolved dxy dominance of the pairing glue is a concrete, falsifiable prediction consistent with ARPES and STM observations. The qualitative picture is coherent and internally consistent. However, the quantitative claims, especially the 'five-fold enhancement' for M-FeSe/STO, are fragile because lambda is extremely sensitive to the Se height h and to J, and no uncertainty or robustness analysis is provided for the interface geometry.","major_comments":[{"comment":"The central contrast between M-FeSe/STO (lambda=0.34) and free-standing M-FeSe (lambda=0.002) rests on the Se height h being 1.40 Å vs 1.39 Å, both taken from Ref. [44], but no sensitivity analysis is performed for the interface. The paper itself states about h: 'Its value is critical, as we have seen in the bulk case, and we cannot rely on DFT for it.' The bulk h-scan already shows the danger (Table II: h=1.463 Å gives lambda=0.067; h=1.27 Å gives lambda=0.003). A shift of a few hundredths of an Ångström in M-FeSe/STO, in the direction that pushes dxy below EF, could reduce lambda by more than an order of magnitude and erase the qualitative distinction between the interface and the monolayer. Please add a controlled sweep of h for M-FeSe/STO (and ideally for M-FeSe) around the adopted values, using the same cRPA U,J and QSGW++ pipeline, or provide an error estimate for h from the DFT+DMFT relaxation.","section":"Main text (paragraph 'We consider 5-ML slab...') and Fig. 2(d)"},{"comment":"The quantitative values of lambda are extremely sensitive to J: bulk FeSe goes from lambda=0.067 at J=0.60 eV to lambda=0.9 at J=0.68 eV (Table II and Fig. 2(f)). The cRPA J values are quoted without uncertainty (J=0.60, 0.67, 0.69, 0.71 eV for the four systems), and a 0.01-0.02 eV uncertainty in the interface J would change the reported M-FeSe/STO lambda=0.34 by a large factor. The qualitative trend (larger J increases low-energy Im chi and lambda up to a maximum) is supported, but the headline quantitative claim 'five times larger than bulk' is not robust unless accompanied by a J-sensitivity grid for the interface or an error estimate on cRPA J. Please either provide such a grid or soften the quantitative claim to a qualitative statement.","section":"Fig. 2(f) and Table II"},{"comment":"The calculation fails to suppress the dxz,yz hole pockets that are absent in ARPES; the paper states this suppresses lambda by only 6%, but the supporting calculation is not shown. If the Fermi surface topology is wrong, the spin susceptibility and Eliashberg eigenvalue could be affected beyond 6%. Please document how the 6% estimate was obtained (e.g., by explicitly removing those pockets or shifting bands) and show the resulting chi(q,omega) and lambda for that modified Fermi surface.","section":"Main text, paragraph on M-FeSe/STO results"}],"minor_comments":[{"comment":"Reference [44] is spelled 'Mandal' in the reference list but 'Mondal' in the main text (two occurrences); please unify the spelling.","section":"References"},{"comment":"Several references have incomplete bibliographic data or nonstandard formatting: [9] 'L. Sponza and et al', [10] lacks volume/page/article number, [13] 'E. Gull and et al.', and [46] appears twice. Please bring them into journal style.","section":"References"},{"comment":"The units 'A0' in Tables I and II should be 'Å', and the M-FeSe J value in Table I is given as 0.7 rather than 0.70 eV; please make the precision consistent with the other entries.","section":"Tables I and II"},{"comment":"The sentence 'Our study opens a paradigm for a unified understanding what controls Tc' is missing a preposition; it should read 'understanding of what controls Tc'.","section":"Abstract"},{"comment":"In the caption of Fig. 2, the phrase 'while in (b) and (c)dxy is pushed far below EF : and the system has properties similar to a normal Fermi liquid' contains a colon and spacing error; please correct the punctuation.","section":"Figure 2 caption"}],"recommendation":"major_revision","confidential_remarks":"The paper is a direct application of the authors' own QSGW++ methodology, and the novelty lies more in the FeSe/STO interface prediction than in the method itself. The main concern is not circularity but robustness: the headline lambda values are determined by structural and interaction parameters whose uncertainties are not quantified. If the authors supply the requested h-sweep for the interface and clarify the 6% hole-pocket test, the central claim could become publishable. The qualitative orbital-selective picture is plausible and well-aligned with existing experiments."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague—\n\nThis paper is worth a serious referee. The headline claim is concrete: in FeSe, high Tc requires the Fe dxy band near the Fermi level, and once that condition is met, Hund's J amplifies spin fluctuations that provide the pairing glue. The QSGW++ approach (QSGW+DMFT+BSE+Eliashberg) is state-of-the-art, and the internal trends are consistent: λ rises steeply with J up to a maximum, then falls; pushing dxy below EF collapses λ. The orbital-resolved decomposition identifying dxy as the dominant pairing channel is a genuinely new result, and the benchmark against neutron data in earlier work gives some confidence in the susceptibility.\n\nWhat the paper does well: it is transparent about its own sensitivity. The authors state flat-out that the Se height is 'critical' and that DFT cannot be trusted for it, and they acknowledge that the calculated λ is extremely sensitive to J and h. That candor helps, but it also highlights the main weakness.\n\nThe soft spot is load-bearing. The entire contrast between M-FeSe/STO (λ=0.34) and the free-standing monolayer (λ=0.002) rests on the interface h being 1.40 Å, taken from Mandal et al.'s DFT+DMFT relaxation. The paper does not test how λ for the interface varies with h. Given bulk FeSe's λ drops from 0.067 to 0.003 when h is reduced from 1.463 to 1.27 Å, a few hundredths of an Ångström in h could plausibly shift dxy below EF and eliminate the claimed five-fold enhancement. This is not a minor technicality; it is the quantitative foundation of the paper's central comparison.\n\nTwo more points. First, the paper repeatedly refers to Tc but actually computes the superconducting eigenvalue λ. The connection to Tc is never spelled out, so claims like 'eight-fold increment in Tc' are not strictly derived. Second, the references are sloppy: 'Mondal' should be 'Mandal', and refs [44] and [46] are duplicated. Minor, but in a paper whose results hinge on an external structural input, getting that citation right matters.\n\nOn circularity: I do not see a serious problem. U and J are cRPA outputs, not fits to Tc, and the J-scan is presented as a sensitivity study. The fact that the method comes from the same group is not a flaw by itself.\n\nBottom line: the qualitative mechanism—dxy proximity plus Hund's coupling controls spin-fluctuation pairing—is plausible and well supported by the trends. The quantitative interface claim needs a robustness check on h. Send it to peer review; ask for an h-scan for the interface, an error bar on U and J, and an explicit statement of the λ-to-Tc mapping. A good referee will see the value and push for the missing analyses. For anyone working on FeSe or Hund's metals, this paper is worth reading.","headline":"A serious computational study that plausibly identifies dxy proximity and Hund's J as key controls on FeSe superconductivity, but the headline interface result hangs on an untested Se height.","tokens_in":13282,"tokens_out":5752,"would_cite":true,"duration_ms":47768,"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":"FeSe superconductivity is controlled by the dxy band position and Hund's coupling J.","keywords":["FeSe","Hund's metal","superconducting pairing glue","dxy orbital","spin fluctuations","dynamical mean-field theory","Eliashberg equation","SrTiO3 substrate"],"falsifier":"Measure the selenium height in the FeSe/SrTiO3 interface with sub-0.01 Å precision (for example by surface-extended X-ray absorption fine structure or low-energy electron diffraction) and recompute the dxy band position at that height. If dxy sits more than about 100 meV below the Fermi level in the actual interface, the predicted λ = 0.34 would drop toward the monolayer value, falsifying the claim that dxy proximity is what restores superconductivity.","tokens_in":12179,"feed_emoji":"🧲","tokens_out":10883,"duration_ms":94495,"temperature":0.7,"pith_summary":"The paper asks what controls the superconducting critical temperature Tc in FeSe, from bulk crystals (Tc ≈ 9 K) to a monolayer on SrTiO3 (Tc up to about 75 K). It argues that two conditions must hold simultaneously: the iron dxy band must sit near the Fermi level, and the Hund's exchange J must be large enough to make the metal strongly incoherent. Under those conditions, low-energy spin fluctuations near the antiferromagnetic ordering vector become intense and supply the superconducting pairing glue. The claim is supported by parameter-free calculations that scan J and the Fe–Se bond length, and it explains why a free-standing monolayer should not superconduct while the same monolayer on SrTiO3 should.","feed_headline":"dxy near Fermi level and Hund's J set FeSe's Tc","feed_subtitle":"First-principles calculations show dxy proximity plus strong Hund's J turns spin fluctuations into pairing glue.","key_machinery":"The central machinery is QSGW++, a four-tier chain: QSGW (quasiparticle self-consistent GW) builds the one-particle Hamiltonian and captures nonlocal charge correlations; DMFT (dynamical mean-field theory) with a continuous-time quantum Monte Carlo impurity solver adds local spin fluctuations; a Bethe-Salpeter equation constructed from the DMFT local vertex and nonlocal bubbles yields the spin and charge susceptibilities; and the linearized Eliashberg equation turns those susceptibilities into the leading superconducting eigenvalue λ. The load-bearing ingredients for the paper's conclusion are the orbital-resolved mass renormalization 1/Z and scattering rate Γ, the intensity and dispersion of Im χ(q,ω) at q = (1/2,1/2), and the cRPA-derived values of U and J for each geometry.","core_discovery":"Using a four-tier ab initio method that combines quasiparticle self-consistent GW with dynamical mean-field theory, a Bethe-Salpeter treatment of two-particle response, and an Eliashberg solution of the pairing instability, the paper finds that the superconducting eigenvalue λ is controlled by the interplay of band structure and Hund's correlation. The Fe dxy orbital is the most strongly renormalized and incoherent of the five d orbitals, and it dominates the pairing glue as long as it lies near the Fermi energy. The glue itself is the imaginary part of the spin susceptibility Im χ(q,ω) concentrated near the antiferromagnetic wavevector (1/2,1/2): when J is raised from the ab initio 0.60 eV to 0.68 eV, the susceptibility sharpens and λ jumps from 0.067 to 0.9. Conversely, when the Fe–Se bond length is shortened so that dxy drops well below EF, the system becomes a coherent Fermi liquid with negligible spin fluctuations and λ ≈ 0. The same mechanism differentiates the two monolayer cases: in free-standing M-FeSe the dxy band is pushed about 300 meV below EF and λ = 0.002, while on SrTiO3 it returns to within 50–100 meV of EF, incoherence is restored, and λ = 0.34—five times the bulk value.","pith_inferences":["The paper's two-knob picture suggests a screening rule for other Hund's metals: an orbital with strong Hubbard correlations that sits near EF, plus a large J from reduced screening, is the recipe for high Tc; strain or pressure that moves that orbital away from EF will kill pairing even if J grows.","Because the same method predicts λ = 0.34 from spin fluctuations alone, the gap to the experimental 75 K could be closed by additional mechanisms, such as electron-phonon coupling at the SrTiO3 interface; a combined calculation including phonons is a direct test of whether spin fluctuations are sufficient.","A testable extension of the logic: electron-dope or strain bulk FeSe to move dxy closer to EF without changing J; the paper's mechanism predicts Tc should track the dxy proximity, which could be checked by ARPES plus specific-heat measurements across a doping series.","The monolayer result implies that simply isolating an FeSe layer is not enough to enhance Tc; the substrate's role is to undo the monolayer's downward shift of dxy, so interface design should focus on orbital alignment rather than only on doping or electron-phonon coupling."],"forward_implications":["Because the superconducting eigenvalue λ jumps from 0.067 to 0.9 when J is tuned from 0.60 to 0.68 eV, materials that modestly increase Hund's coupling—by reducing screening—could gain an order of magnitude in Tc.","A free-standing FeSe monolayer is predicted to be a non-superconducting good metal with dxy about 300 meV below EF; placing it on SrTiO3 restores dxy proximity and raises λ fivefold to 0.34.","The pairing instability is dominated by the intra-orbital dxy–dxy channel, so the superconducting gap should be largest on dxy-dominated Fermi-surface sheets, as the paper notes is observed.","Compressing the Fe–Se bond length pushes dxy below EF, suppresses spin fluctuations, and eliminates superconductivity, making the collapsed phase of FeSe a coherent Fermi liquid with λ ≈ 0."],"supporting_citations":[{"why":"Reports the ~75 K critical temperature for monolayer FeSe on SrTiO3 that the paper seeks to explain.","marker":"[7]"},{"why":"Supplies the numerical implementation of the QSGW+DMFT+BSE+Eliashberg method used for all spectra, susceptibilities, and eigenvalues.","marker":"[10]"},{"why":"Defines the quasiparticle self-consistent GW approximation that provides the one-particle Hamiltonian and nonlocal charge correlations.","marker":"[11]"},{"why":"Defines dynamical mean-field theory, which supplies the local vertex and spin-fluctuation treatment at the heart of the two-particle response.","marker":"[14]"},{"why":"Provides the earlier demonstration that the same QSGW+DMFT+BSE+Eliashberg chain yields parameter-free superconducting eigenvalues for Hund's materials.","marker":"[19]"},{"why":"Application to LaFe2As2 showing loss of superconductivity when bands become itinerant, used as a parallel to the compressed-bond FeSe case.","marker":"[20]"},{"why":"Constrained RPA, the scheme used to compute U and J from the QSGW band structure for each geometry.","marker":"[23]"},{"why":"The Eliashberg study that concluded spin fluctuations can account for at most a two-fold Tc increase, the comparison this paper's five-fold estimate is set against.","marker":"[42]"},{"why":"Provides the Se heights and lattice constants for the free-standing monolayer and the monolayer on SrTiO3, the structural inputs on which the main conclusion depends.","marker":"[44]"}],"fun_headline_variants":["dxy near Fermi level and Hund's J govern FeSe Tc","Hund's J sharpens spin glue to raise FeSe Tc","Monolayer FeSe/STO Tc tied to Hund's J","dxy proximity plus Hund's J sets FeSe's Tc","Incoherent dxy and Hund's J tune FeSe Tc"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire distinction between the superconducting interface (λ = 0.34) and the non-superconducting monolayer (λ = 0.002) rests on the assumed selenium height above the iron plane—1.39 Å for the free monolayer and 1.40 Å on SrTiO3, taken from an external structure optimization; if the real height differs by a few hundredths of an Ångström in the direction that pushes dxy below the Fermi level, the predicted enhancement collapses.","fun_headline_variants_meta":{"raw":{"variants":["dxy near Fermi level and Hund's J govern FeSe Tc","Hund's J sharpens spin glue to raise FeSe Tc","Monolayer FeSe/STO Tc tied to Hund's J","dxy proximity plus Hund's J sets FeSe's Tc","Incoherent dxy and Hund's J tune FeSe Tc"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001302,"raw_usage":{"total_tokens":5436,"prompt_tokens":1197,"completion_tokens":4239,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":813,"completion_tokens_details":{"reasoning_tokens":4158}},"tokens_in":813,"tokens_out":4239,"duration_ms":27833,"temperature":1.0,"reasoning_tokens":4158,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:48:30.482344+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the selenium height in the FeSe/SrTiO3 interface with sub-0.01 Å precision (for example by surface-extended X-ray absorption fine structure or low-energy electron diffraction) and recompute the dxy band position at that height. If dxy sits more than about 100 meV below the Fermi level in the actual interface, the predicted λ = 0.34 would drop toward the monolayer value, falsifying the claim that dxy proximity is what restores superconductivity.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reports the ~75 K critical temperature for monolayer FeSe on SrTiO3 that the paper seeks to explain."},{"cited_title":"Sponza and et al, Phys","cited_arxiv_id":null,"evidence_quote":"Supplies the numerical implementation of the QSGW+DMFT+BSE+Eliashberg method used for all spectra, susceptibilities, and eigenvalues."},{"cited_title":"Pashov and et al., Computer Physics Communications (2019)","cited_arxiv_id":null,"evidence_quote":"Defines the quasiparticle self-consistent GW approximation that provides the one-particle Hamiltonian and nonlocal charge correlations."},{"cited_title":"Georges, G","cited_arxiv_id":null,"evidence_quote":"Defines dynamical mean-field theory, which supplies the local vertex and spin-fluctuation treatment at the heart of the two-particle response."},{"cited_title":"Acharya, D","cited_arxiv_id":null,"evidence_quote":"Provides the earlier demonstration that the same QSGW+DMFT+BSE+Eliashberg chain yields parameter-free superconducting eigenvalues for Hund's materials."},{"cited_title":"Acharya, D","cited_arxiv_id":null,"evidence_quote":"Application to LaFe2As2 showing loss of superconductivity when bands become itinerant, used as a parallel to the compressed-bond FeSe case."},{"cited_title":"Aryasetiawan, M","cited_arxiv_id":null,"evidence_quote":"Constrained RPA, the scheme used to compute U and J from the QSGW band structure for each geometry."},{"cited_title":"Schrodi, A","cited_arxiv_id":null,"evidence_quote":"The Eliashberg study that concluded spin fluctuations can account for at most a two-fold Tc increase, the comparison this paper's five-fold estimate is set against."}],"review_version":1}