{"id":"4d25d703-fa23-459b-a42d-b8a8452f3f83","arxiv_id":"2505.05566","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"A transient Hubbard exciton fluid forms in photo-doped Sr2IrO4, signaled by spectral weight transfer from a Drude response to a 1.5 THz intra-excitonic peak.","lead":"Ultrafast terahertz measurements on the antiferromagnetic Mott insulator Sr2IrO4 show that photo-excited charge pairs (holons and doublons) bind into a transient insulating fluid of Hubbard excitons within about one picosecond. The work provides the first experimental evidence for a long-predicted excitonic state that can be controlled by magnetic exchange interactions.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 1.5 THz peak is assigned to an s-to-p Hubbard-exciton transition calculated at 2.85 THz; the factor-of-two gap is unresolved because no THz conductivity is computed from the model.","rationale":"The reader's weakest_assumption identifies exactly the point where the central claim is least secure. I start from the paper's own evidence: the tr-TDTS data show a clean Drude-to-Lorentzian transfer, fluence-independent monomolecular decay, a reduced Lorentzian fraction under beta pumping, and plausible exclusions of phase separation, trapping, and sideband scenarios. Those are real independent supports. But none of them identifies the microscopic origin of the 1.5 THz mode; that identification is carried by the ED calculation, and the calculation gives 2.85 THz for the only quoted dipole-allowed transition. A factor of nearly two in a transition energy is not a line-shape or broadening effect at the level of energy-scale comparison. The paper could have settled this by computing the model's optical conductivity, which is standard for this type of ED calculation, or by scanning parameters; it did neither. I therefore do not see a reason to move the reader's CONDITIONAL verdict: the claim is plausible and the experiments are well designed, but the quantitative link from model to observed mode is missing. I would keep the verdict CONDITIONAL rather than REJECT because the experimental phenomenology is strong and the discrepancy is a missing quantitative validation, not a demonstrated contradiction. If the proposed conductivity check fails to produce a peak near 1.5 THz for any reasonable parameter set, the assignment should be downgraded to unverified.","tokens_in":23992,"tokens_out":4592,"duration_ms":54510,"concrete_test":"Compute the THz optical conductivity sigma_1(omega) of the one-HD-pair sector of HtJV on the 26-site cluster, including all dipole matrix elements among the low-lying eigenstates, and overlay the result on the measured Delta sigma_1(omega) at t around 1 ps. If the strongest allowed intra-excitonic peak lies near 2.85 THz rather than 1.5 THz, the assignment as stated is unsupported. As a supplementary check, vary tNN, V, and J inside the reported literature ranges and determine whether any physically acceptable parameter set places the strongest allowed transition at about 1.5 THz; if none does, the energy-scale agreement is not sufficient to identify the observed mode.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—that the transient insulating response is a fluid of Hubbard excitons—rests on identifying the ~1.5 THz Lorentzian as an intra-excitonic transition. The only quantitative candidate offered in Figure 4a is the dipole-allowed s-to-p transition at 2.85 THz, nearly twice the observed frequency. The text says only that the observed peak lies within the predicted frequency scale and is consistent with an intra-HE transition; no optical conductivity, dipole matrix elements, or line shapes are computed from the same model, and no parameter scan is shown. This matters because several alternative explanations (phase separation, defect trapping, phonon sidebands, pseudogap) are excluded phenomenologically, but the positive identification of the mode as an HE internal transition is what distinguishes the claimed physics. The mismatch is not a small extrapolation: V, J, and tNN are taken from literature for Sr2IrO4, yet the predicted splitting is off by about a factor of 1.9. The supplement further shows that the number, order, and binding energies of excitonic states change with cluster size (N=20 vs N=26), so the 2.85 THz value is not demonstrated to be converged. Because the candidate p-state sits near the continuum and the paper concedes it may be unstable at nonzero momentum, the assignment to a specific bound transition is underdetermined by a single measured mode. The experimental findings remain valuable, but the central claim currently has only energy-scale-level support.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports time-resolved terahertz conductivity measurements on the antiferromagnetic Mott insulator Sr2IrO4 after pumping the Mott-gap α transition. The authors observe a fast-rising Drude response that, within about 1 ps, transfers spectral weight to a Lorentzian peak near 1.5 THz, which they interpret as an intra-excitonic transition of Hubbard excitons (bound holon-doublon pairs). Exact diagonalization of an extended single-band t-J-V model for one holon-doublon pair on a 26-site cluster yields four bound states with s, d, s, and p symmetries and a dipole-allowed s-to-p transition at 2.85 THz, which the authors cite as the leading candidate for the 1.5 THz peak. The decay of the 1.5 THz mode is exponential, fluence independent, and its time constant increases with decreasing temperature, following τ ∝ exp[ζΔ(T)/J] with ζ = 0.76(2) and Δ(T) the reported Mott gap. The Supplementary Information systematically argues against phase separation, defect trapping, two-band Drude response, electron-boson sidebands, and pseudogap formation as explanations of the data.","tokens_in":24167,"tokens_out":6202,"duration_ms":61652,"significance":"If the assignment is correct, this is the first experimental observation of Hubbard excitons as metastable quasiparticles in a two-dimensional antiferromagnetic Mott insulator and demonstrates a route to a transient insulating exciton fluid, a long-sought out-of-equilibrium state. The experimental methodology is careful: Δσ1 and Δσ2 are fit simultaneously, the spectral weight transfer is clearly demonstrated, and the fluence independence of the decay argues against bimolecular or Auger processes. The temperature dependence of the decay time provides a falsifiable, parameter-constrained consistency check with the multi-magnon recombination theory. The main weakness is the factor-of-two mismatch between the computed s-p transition (2.85 THz) and the observed 1.5 THz peak, together with the absence of a computed THz conductivity from the same model; these issues must be addressed before the central claim can be considered established.","major_comments":[{"comment":"The only quantitative candidate offered for the 1.5 THz peak is the dipole-allowed s-to-p transition at 2.85 THz, which is nearly twice the observed frequency. The paper does not compute the THz optical conductivity from the t-J-V model, nor does it present dipole matrix elements or line shapes; it compares energy levels only. Because the supplement (Section V, Figure S10) shows that the number, order, and binding energies of the excitonic states change between N=20 and N=26, the 2.85 THz value is not demonstrated to be converged with system size. This factor-of-two discrepancy is load-bearing because the positive identification of the 1.5 THz peak as an intra-HE transition is what distinguishes the claimed Hubbard exciton fluid from the alternatives excluded phenomenologically. I recommend computing the optical conductivity from the same model and showing the convergence of the s-p transition energy with system size, or a parameter scan showing that a transition near 1.5 THz is robust.","section":"Main text, \"To identify the specific intra-excitonic transition...\" paragraph and Figure 4a"},{"comment":"The manuscript explicitly states that the p-wave exciton at k=[0,0] may not lie below the continuum at all momenta and is \"potentially unstable against decay into the continuum.\" Since the assignment of the 1.5 THz peak relies specifically on the s-to-p transition, the stability of the p-state is essential to the claim that the observed mode is an intra-excitonic transition of a bound Hubbard exciton. If the p-state is unbound, the observed Lorentzian could correspond to a different process, such as a transition into a resonance. The authors should either establish that the p-state is bound in the relevant momentum range or temper the assignment accordingly.","section":"Methods, \"Numerical Calculations\" and Extended Data Figure 4"},{"comment":"The temperature dependence of the recombination time is fitted to τ ∝ exp[ζ Δ(T)/J] with ζ = 0.76(2), using the authors' own multi-magnon theory (refs 22–23). Although the fit is good, this is a consistency check with two effective parameters (ζ and the prefactor) rather than an independent test, and the numerical value of ζ depends on the accuracy of the t-J-V model and the reported Δ(T). The paper should state more clearly that this agreement supports, but does not by itself prove, the multi-magnon recombination mechanism; the claim of \"extremely strong coupling to magnon modes\" in the abstract is stronger than the evidence warrants.","section":"Figure 3b and Methods, \"Exponential Fitting\" and \"Numerical Calculations\""}],"minor_comments":[{"comment":"\"Lanzcos\" is a typo for \"Lanczos\" (appears twice in the Methods section).","section":"Methods, \"Numerical Calculations\""},{"comment":"Reference [19] is Meltzer et al. (1972) on MnF2, not a time-resolved reflectivity study of Sr2IrO4; the correct reference for the ~60 fs intraband cooling timescale appears to be [53] (Hsieh et al., PRB 86, 035128). Please check all citations in that sentence.","section":"Main text, paragraph beginning \"Previous time-resolved near-infrared reflectivity\""},{"comment":"The sign convention in the Lorentzian term is non-standard; for clarity, please state explicitly whether the imaginary part is positive or negative for absorption, and define the relationship to Δσ2(ω).","section":"Methods, Equation (3)"},{"comment":"The sentence \"Due to the symmetry of the N = 20 lattice, the spectrum is projected onto the kx + 0.5(ky) axis\" is confusing; please clarify the projection axes and the reason for the different treatment for N = 20 and N = 26.","section":"Supplementary Figure S10 caption"},{"comment":"The text says \"by t = 0.4 ps, the unbound holons and doublons have relaxed near the Hubbard band edges,\" but earlier it cites a ~60 fs intraband cooling time; consider rewording for consistency.","section":"Main text, paragraph after Figure 2g"}],"recommendation":"major_revision","confidential_remarks":"The experimental core is strong, but the central interpretation—assignment of the 1.5 THz mode to the s-p intra-excitonic transition—currently rests on an energy-scale comparison with a factor-of-two mismatch. The authors' own statements (p-state possibly unbound; single mode not sufficient; finite-size changes in the spectrum) support the need for a major revision. I would encourage the editor to seek a referee with expertise in exact diagonalization of Hubbard models to assess whether the proposed additional calculations are feasible within a revision cycle."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe thing to know: this is the first experiment to see something that looks like a Hubbard exciton fluid in a 2D antiferromagnetic Mott insulator. The tr-TDTS data in Sr2IrO4 show a clean Drude-to-Lorentzian spectral weight transfer on sub-picosecond timescales, and the 1.5 THz mode is absent in equilibrium. That alone is a notable result.\n\nWhat the paper does well: the experiment is careful. The simultaneous fits to sigma1 and sigma2 constrain the Drude-Lorentz decomposition, and the SI systematically rules out phase separation, charge trapping, two-band Drude, electron-boson sidebands, and pseudogap formation. The pump-wavelength dependence (alpha vs beta) supports an excitonic origin. The temperature-dependent lifetime and its exponential scaling with the Mott gap is a nice piece of physics, and the paper is honest that the exponent zeta is fitted, not predicted. Source data are provided, which helps.\n\nWhere it softens: the assignment of the 1.5 THz peak to a specific s-to-p intra-excitonic transition rests on exact diagonalization that puts that transition at 2.85 THz—nearly a factor of two higher. The paper only compares energy levels; it does not compute the THz conductivity or dipole matrix elements from the same model, so there is no line-shape check. The p-state also sits close to the continuum and the authors concede it may be unstable at finite momentum. That weakens the identification of this particular peak as the s-to-p transition, though not the broader claim that a bound state exists. The SI shows the level order and binding energies change with cluster size, so the 2.85 THz value is not demonstrably converged. This is the load-bearing soft spot.\n\nThe recombination analysis is somewhat circular in that it reuses the authors' own multi-magnon theory to fit zeta, but the experimental tau(T) curve is independent and the functional form is theoretically motivated. I would not call that a fatal flaw.\n\nBottom line: the central experimental observation is likely correct, and the Hubbard-exciton interpretation is plausible but only energy-scale-level. A serious referee should ask for a computed THz conductivity from the model, a parameter scan, and some discussion of whether the p-state is truly bound. The paper deserves peer review, not desk rejection.","headline":"First credible experimental evidence for a transient Hubbard exciton fluid in Sr2IrO4, but the mode assignment has a factor-of-two discrepancy that needs to be tightened.","tokens_in":24832,"tokens_out":2239,"would_cite":true,"duration_ms":24184,"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":"This paper reports the first experimental detection of a transient Hubbard-exciton fluid in a two-dimensional antiferromagnetic Mott insulator, formed when photo-excited holons and doublons bind into neutral pairs on a sub-picosecond…","keywords":["Hubbard exciton","Mott insulator","Sr2IrO4","terahertz time-domain spectroscopy","holon-doublon pair","photo-doping","exact diagonalization","magnon-assisted recombination"],"falsifier":"Resolve the full terahertz optical conductivity of the photo-doped state and compute the equivalent quantity from the t-J-V model; if the 1.5 THz mode's oscillator strength and line shape do not match the predicted intra-excitonic transition, or if a second transition appears near the predicted 2.85 THz, the assignment is falsified. Alternatively, vary the exchange energy J in situ (by strain or resonant phonon driving) and check whether the peak frequency shifts as expected for an exchange-bound pair while the Mott gap is held fixed.","tokens_in":23682,"feed_emoji":"⚛️","tokens_out":9260,"duration_ms":87297,"temperature":0.7,"pith_summary":"This paper reports the first experimental evidence that photo-excited carriers in a two-dimensional antiferromagnetic Mott insulator can bind into Hubbard excitons: neutral pairs of an empty site (holon) and a doubly occupied site (doublon) held together by the magnetic exchange coupling of the antiferromagnetic background. Using time-resolved terahertz spectroscopy on Sr2IrO4, the authors watch the pump-induced optical conductivity evolve within about a picosecond from a metallic Drude response, characteristic of unbound holons and doublons, to an insulating Lorentzian absorption near 1.5 THz, which they attribute to a transition between internal levels of the bound pair. The assignment is supported by exact diagonalization of an extended single-band Hubbard model that yields discrete bound holon-doublon states with s-, d-, s-, and p-wave symmetry and a dipole-allowed intra-excitonic transition in the same frequency range. If correct, this establishes a new out-of-equilibrium state of matter and shows that magnetic exchange alone can bind charge pairs in a Mott insulator, a step toward magnetically tunable excitonic devices.","feed_headline":"Terahertz data show Hubbard excitons binding into a fluid in Sr2IrO4","feed_subtitle":"Photodoped holon-doublon pairs turn a metallic Drude response into an insulating peak at 1.5 THz within a picosecond.","key_machinery":"The central object is the Hubbard exciton, a bound holon-doublon pair in which the binding is generated by the antiferromagnetic spin background: separating the pair leaves a string of overturned spins that produces a confining potential, so the pair is held together by magnetic exchange rather than by Coulomb attraction alone. Experimentally, the machinery is time-resolved time-domain terahertz spectroscopy in reflection geometry, whose differential spectra are converted into the pump-induced complex optical conductivity and then decomposed at every time delay into a Drude term, a Lorentzian term for the excitonic mode, and weak background Lorentzians. Theoretically, the key tool is exact diagonalization with the Lanczos algorithm of an extended single-band $t$-$J$-$V$ Hamiltonian on a 26-site square cluster within the one-holon-one-doublon subspace, which produces four discrete levels with s-, d-, s-, and p-wave symmetry at zero center-of-mass momentum and identifies the dipole-allowed s-to-p transition at 2.85 THz as the candidate for the observed 1.5 THz mode; the same model family supplies the multi-magnon recombination law $\\tau \\propto \\exp(\\zeta \\Delta/J)$ used for the temperature dependence.","core_discovery":"The central claim is that photo-doping Sr2IrO4 across the Mott gap produces a transient Hubbard-exciton fluid. Immediately after excitation the terahertz conductivity is Drude-like, meaning unbound holons and doublons; within roughly half a picosecond the spectral weight shifts into a Lorentzian peak centered near 1.5 THz, the signature of an optically allowed transition between bound internal levels of a holon-doublon pair. The authors identify the 1.5 THz mode with the dipole-allowed transition between the lowest s-wave and the p-wave Hubbard-exciton states predicted by exact diagonalization of an extended $t$-$J$-$V$ model on a 26-site cluster, whose computed transition energy is 2.85 THz. The lifetime of the excitonic peak is about one picosecond and grows exponentially with the temperature-dependent Mott gap, which the authors fit with the form $\\tau \\propto \\exp[\\zeta \\Delta(T)/J]$ using $\\zeta = 0.76(2)$, and interpret as recombination by multi-magnon emission. They take this combination of spectral transfer, lineshape, symmetry-resolved bound states, and exponential gap scaling as evidence that Hubbard excitons exist as metastable quasiparticles and can self-organize into a transient insulating fluid.","pith_inferences":["The factor-of-two difference between the calculated 2.85 THz intra-excitonic transition and the measured 1.5 THz peak is an unresolved quantitative gap; computing the full terahertz optical conductivity from the model, rather than only the level spacing, would test whether the assignment holds at the level of line shape and oscillator strength.","If the assignment is correct, the same experiment should show a second, weaker intra-excitonic transition at higher frequency (predicted around 2.85 THz or beyond) once the population of excited internal states is enhanced, for example by pumping at a higher photon energy or higher fluence.","The exponential lifetime-versus-gap law implies that the Hubbard-exciton fluid is a distinct thermalized branch; a testable consequence is that tuning J in situ (by strain, pressure, or resonant phonon excitation) should change both the 1.5 THz peak position and the lifetime in the direction predicted by the t-J-V model.","Because the experiment probes only a single mode, a decisive discriminator between the intra-excitonic interpretation and alternatives would be magnetic-field-dependent or momentum-resolved studies that follow the predicted internal-level structure."],"forward_implications":["A transient insulating Hubbard-exciton fluid can be created in a two-dimensional antiferromagnetic Mott insulator by pumping resonantly across the Mott gap, without chemical doping.","Free holons and doublons bind into Hubbard excitons on a sub-picosecond timescale, and the bound pairs decay on a roughly one-picosecond timescale through multi-magnon emission.","The Hubbard-exciton level spectrum is non-hydrogenic (s, d, s, p in increasing energy), so terahertz intra-excitonic transitions are a direct spectroscopic fingerprint of the exchange-binding mechanism.","Because the recombination time scales exponentially with the ratio of the Mott gap to the exchange energy, temperature and strain can be used to control the lifetime of the excitonic fluid.","The same tr-TDTS signature could be used to search for Hubbard-exciton fluids in other layered antiferromagnetic Mott insulators, such as the parent cuprates, where the relevant energy scales are comparable."],"supporting_citations":[{"why":"Supplies the effective t-J model and the multi-magnon recombination theory for holon-doublon pairs in Mott insulators used to interpret the decay dynamics.","marker":"[22]"},{"why":"Predicts the exponential recombination law $\\tau \\propto \\exp(\\zeta \\Delta/J)$ that the temperature-dependent lifetime is fit to.","marker":"[23]"},{"why":"Establishes particle-hole bound states (Hubbard excitons) in Mott-Hubbard insulators through exchange-mediated attraction.","marker":"[27]"},{"why":"Provides the string/confining-potential picture of holon-doublon binding in antiferromagnets that motivates the HE assignment.","marker":"[28]"},{"why":"Demonstrates the Drude-to-Lorentz spectral weight transfer in tr-TDTS as the hallmark of exciton formation, the template for interpreting the Sr2IrO4 data.","marker":"[14]"},{"why":"Uses intra-excitonic terahertz transitions to probe exciton stability in two dimensions, motivating the interpretation of the 1.5 THz Lorentzian.","marker":"[15]"},{"why":"Establishes the Jeff=1/2 single-band Mott description of Sr2IrO4 that justifies the one-band model.","marker":"[43]"},{"why":"Supplies the temperature-dependent Mott gap Δ(T) and equilibrium optical conductivity used in the recombination fit and peak assignment.","marker":"[44]"},{"why":"Provides the exchange energy J=60 meV and magnon spectrum of Sr2IrO4 used in the model parameters and recombination analysis.","marker":"[46]"}],"fun_headline_variants":["Photo-doped Sr2IrO4 hosts transient Hubbard-exciton fluid","Terahertz probe sees Hubbard excitons organize into fluid","Fleeting Hubbard-exciton fluid seen in photo-doped Mott insulator","Ultrafast terahertz captures Hubbard exciton fluid","Hubbard excitons bind into transient fluid in Sr2IrO4"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The identification of the 1.5 THz peak as an intra-excitonic transition rests on the assumption that the single-band t-J-V model with literature parameters for Sr2IrO4 predicts the low-energy Hubbard-exciton spectrum accurately enough that the computed dipole-allowed transition at 2.85 THz corresponds to the measured 1.5 THz mode.","fun_headline_variants_meta":{"raw":{"variants":["Photo-doped Sr2IrO4 hosts transient Hubbard-exciton fluid","Terahertz probe sees Hubbard excitons organize into fluid","Fleeting Hubbard-exciton fluid seen in photo-doped Mott insulator","Ultrafast terahertz captures Hubbard exciton fluid","Hubbard excitons bind into transient fluid in Sr2IrO4"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00083,"raw_usage":{"total_tokens":3686,"prompt_tokens":1068,"completion_tokens":2618,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":684,"completion_tokens_details":{"reasoning_tokens":2527}},"tokens_in":684,"tokens_out":2618,"duration_ms":18704,"temperature":1.0,"reasoning_tokens":2527,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T23:02:49.834348+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Resolve the full terahertz optical conductivity of the photo-doped state and compute the equivalent quantity from the t-J-V model; if the 1.5 THz mode's oscillator strength and line shape do not match the predicted intra-excitonic transition, or if a second transition appears near the predicted 2.85 THz, the assignment is falsified. Alternatively, vary the exchange energy J in situ (by strain or resonant phonon driving) and check whether the peak frequency shifts as expected for an exchange-bound pair while the Mott gap is held fixed.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the effective t-J model and the multi-magnon recombination theory for holon-doublon pairs in Mott insulators used to interpret the decay dynamics."},{"cited_title":"L., Kachur, I","cited_arxiv_id":null,"evidence_quote":"Predicts the exponential recombination law $\\tau \\propto \\exp(\\zeta \\Delta/J)$ that the temperature-dependent lifetime is fit to."},{"cited_title":"& Prelovˇ sek, P","cited_arxiv_id":null,"evidence_quote":"Establishes particle-hole bound states (Hubbard excitons) in Mott-Hubbard insulators through exchange-mediated attraction."},{"cited_title":"& Prelovˇ sek, P","cited_arxiv_id":null,"evidence_quote":"Provides the string/confining-potential picture of holon-doublon binding in antiferromagnets that motivates the HE assignment."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Demonstrates the Drude-to-Lorentz spectral weight transfer in tr-TDTS as the hallmark of exciton formation, the template for interpreting the Sr2IrO4 data."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Uses intra-excitonic terahertz transitions to probe exciton stability in two dimensions, motivating the interpretation of the 1.5 THz Lorentzian."},{"cited_title":"Mott-Hubbard exciton in the optical conductivity of YTiO 3 and SmTiO3","cited_arxiv_id":null,"evidence_quote":"Establishes the Jeff=1/2 single-band Mott description of Sr2IrO4 that justifies the one-band model."},{"cited_title":"& Gedik, N","cited_arxiv_id":null,"evidence_quote":"Supplies the temperature-dependent Mott gap Δ(T) and equilibrium optical conductivity used in the recombination fit and peak assignment."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the exchange energy J=60 meV and magnon spectrum of Sr2IrO4 used in the model parameters and recombination analysis."}],"review_version":1}