{"id":"0e5e1bbd-ae96-4d00-a2a9-a253954c5f8b","arxiv_id":"2507.16453","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"Photoemission evidence for spontaneously formed excitons below Tc in Ta2Pd3Te5, with an extracted Bohr radius of about 14 Å and an apparently odd-parity wave function.","lead":"ARPES measurements of the candidate excitonic insulator Ta2Pd3Te5 reveal a flat, polarization-dependent photoemission feature at the Brillouin zone center that the authors attribute to excitons in the insulating ground state. If confirmed, this would be the first direct photoemission signature of excitons below a transition temperature in a bulk solid, strengthening the case for a true excitonic insulator.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The exciton signal is isolated by subtracting second-BZ ARPES from first-BZ data, but the normalization and k-dependent matrix-element control are unspecified; the Γ-centered residual could be a subtraction artifact from the intense valence band.","rationale":"I read the paper in good faith and find the central claim genuinely interesting, but not yet secure. The reader's weakest assumption is exactly the load-bearing point: the second-zone spectrum is used as a bare valence-band baseline, and the subtraction normalization is not specified. Nothing in the manuscript checks that the relative intensity between zones is k-independent or that final-state and surface contributions are identical. Since the valence band peaks at Γ, a normalization error naturally produces a residual concentrated at Γ, precisely where the exciton feature is claimed. The energy-position claim is also fragile, with a 0.037 eV separation against a 0.058 eV FWHM, but it is secondary: if the residual is an artifact, the position is irrelevant, and if the residual survives the proposed k-dependent normalization test, the position should be re-examined with proper uncertainties. The odd-parity conclusion, despite its internal orbital-label inconsistency, is not the most load-bearing part; the existence of the exciton signal itself is. I do not see grounds to change the reader's CONDITIONAL verdict. The proposed ratio test would settle whether the central claim is an artifact or a real excitonic signal, so the verdict should remain CONDITIONAL until that check is performed.","tokens_in":15828,"tokens_out":6444,"duration_ms":78257,"concrete_test":"Use the raw ARPES maps behind Fig. 1e at 80 K. Map the second-BZ data into the first zone by a reciprocal lattice vector G and compute the ratio R(k,ω) = I_1BZ(k,ω) / I_2BZ(k+G,ω) over the window where only the bare valence bands are intense. Fit a smooth normalization surface α(k,ω) to the valence-band region, subtract α(k,ω)·I_2BZ from I_1BZ, and examine whether a Γ-centered feature with the claimed amplitude remains. If the residual disappears under a smoothly varying α, the fixed-normalization subtraction produced the signal; if a robust Γ feature persists, the subtraction concern is refuted.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The entire identification rests on the first processing step: subtracting the second-Brillouin-zone ARPES spectrum from the first-zone spectrum to expose a residual 'exciton' feature at Γ. The normalization between zones is not specified in the main text or the supplement, and the statement that the feature 'disappears completely in the second BZ' does not by itself validate the baseline because the second-zone spectrum is the baseline. This is especially delicate because the valence band has a strong intensity maximum at Γ; any zone-dependent change in matrix element, final-state diffraction, energy resolution, or surface contribution will create a residual concentrated at exactly the momentum where the claimed peak sits. The same residual is then fitted with an exciton envelope, so the fit is not independent confirmation. The odd-parity assignment and the claim that the exciton lies 0.037 eV below the VBM against a 0.058 eV FWHM both inherit this fragility. Without a specified normalization and a control for k-dependent matrix-element variation, the flat feature could be, at least partly, a subtraction artifact.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports angle-resolved photoemission spectroscopy (ARPES) measurements on Ta2Pd3Te5 below its metal-insulator transition at Tc = 365 K, claiming the observation of a nondispersive photoemission feature at the Brillouin zone center with binding energy 0.115 eV at 80 K. The authors interpret this feature as direct photoemission from spontaneously formed excitons in the insulating ground state, extract a 1s-like exciton envelope with Bohr radius ~13.6–15.1 Å, and infer an odd-parity exciton wavefunction from polarization-dependent measurements. They argue that the exciton binding energy exceeds the single-particle band gap, driving the transition, and support this with a two-band mean-field model and phonon-replica simulations.","tokens_in":15961,"tokens_out":3058,"duration_ms":35331,"significance":"If the interpretation is correct, this would be a landmark observation: direct evidence of excitons in the ground state of a three-dimensional excitonic insulator, providing real-space wavefunction information and parity symmetry. The paper is timely and relevant to the long-standing search for excitonic insulators. The main strengths are the unusually clean temperature-dependent data, the use of orbital-selective polarization, and the attempt to rule out electron-phonon and density-wave explanations through supplemental simulations. However, the central claim rests on a subtraction procedure and peak-decomposition whose robustness is not fully demonstrated, and the quantitative agreement with the exciton photoemission model is partly a fitting exercise rather than an independent prediction.","major_comments":[{"comment":"The isolation of the claimed exciton feature relies on subtracting ARPES spectra of the first Brillouin zone by those of the second zone, described as 'after normalization,' but the normalization procedure is not specified anywhere. The intensity scale, k-dependent alignment, and any energy-dependent correction between the two zones are not given. Since the valence band has a strong maximum at Γ, any zone-dependent matrix-element, final-state, or surface contribution will produce a residual concentrated at exactly the momentum where the claimed peak sits. The statement that the feature 'disappears completely in the second BZ' does not validate the baseline, because the second-zone spectrum is the baseline. Please provide the explicit normalization (e.g., integrated-energy window, valence-band intensity, or a scaling coefficient) and a control analysis showing that the residual is robust to reasonable variations in the normalization, or a comparison with a calculated k-dependent matrix-element ratio.","section":"Main text, Fig. 1e and Supplemental Sec. S9"},{"comment":"The claim that the exciton lies 0.037 eV below the valence band maximum is based on a Voigt decomposition in which the exciton peak (0.115 eV) is separated from the VB1 peak (0.078 eV) by less than the fitted full-width at half-maximum of each peak (0.058–0.060 eV). With the given linewidths, the exciton peak is not resolved from the valence band; its position and intensity are strongly correlated with the tails of VB1 and VB2. The residual fit in Table VII still contains significant VB2 intensity (0.20), further complicating the decomposition. Please demonstrate the uniqueness of this decomposition, for example by fixing the valence-band parameters to those obtained from the second-zone spectrum and showing that the fit without an exciton component fails quantitatively, and by providing confidence intervals for the exciton binding energy that account for the parameter correlations.","section":"Fig. 3c and Supplemental Tables V–VII"},{"comment":"The odd-parity assignment of the exciton wavefunction is asserted from the observation that the feature is visible for x-polarized but suppressed for yz-polarized photons, combined with the statement that 'the only odd parity wave function involved here is the valence Pd dxy orbital.' This conclusion requires a quantitative analysis of the exciton photoemission matrix element, including the degree of interband hybridization and the dipole matrix elements for the specific photon polarizations and final states. Without such a calculation, the polarization data could reflect matrix-element or final-state effects of the ordinary valence band rather than the parity of the exciton envelope. Please provide a matrix-element calculation or a symmetry analysis that explicitly includes the hybridized valence and conduction bands and predicts the relative intensities for x, y, and z polarizations.","section":"Main text, 'polarization dependence' and Supplemental Fig. S4"},{"comment":"There is a circularity in the quantitative agreement between the data and the 'exciton photoemission simulation.' The Bohr radius is extracted by fitting MDCs of the subtracted spectra with the 1s formula P ∝ 1/(1 + k^2 a0^2/4)^4, and then the same formula and fitted radius are used to generate the simulated spectral function shown in Fig. 1d and Fig. 3d. This demonstrates internal consistency but not that the signal is an exciton. The two-band model parameters in Fig. S8, including Δ = 40 meV, are likewise adjusted to reproduce the observed gap. Please clarify which quantities are predicted versus fitted, and discuss the factor-of-two discrepancy with the tight-binding calculation of the exciton radius in Ref. [47], which would provide a non-circular test of the interpretation.","section":"Supplemental Sec. 3 and Fig. 3, Fig. 1d"}],"minor_comments":[{"comment":"The sentence 'whose material realization has been elusive' is a grammatical fragment; consider revising to 'a material realization of which has remained elusive.'","section":"Abstract"},{"comment":"There is a typo: 'deviates largley' should be 'deviates largely.'","section":"Main text, second paragraph"},{"comment":"The phrase 'calulated desults' in the supplemental-materials link text is a typo; it should read 'calculated results.'","section":"Reference [50]"},{"comment":"The sentence 'the exciton photoemission signal was observed only above Tc' is confusing because the present paper reports the signal below Tc; please rephrase to clarify that this refers to the prior Ta2NiSe5 work.","section":"Introduction, comparison with Ta2NiSe5"},{"comment":"Table II lists the same direction 'Γ−X' twice; presumably the second entry should be 'Γ−Y' to match the main-text statement that the Bohr radius is larger along Γ−X than along Γ−Y.","section":"Supplemental Table II"}],"recommendation":"major_revision","confidential_remarks":"The paper addresses a high-profile and controversial topic, and its central claim is exciting. However, the evidence is not yet at the standard required for such a strong claim. The subtraction normalization, the peak-separation versus linewidth issue, and the asserted parity are load-bearing and need to be addressed with concrete additional analysis or a stated limitation. The manuscript would also benefit from an honest discussion of which results are predictions and which are fits. I would support publication after the authors provide the requested control analyses and clarifications."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear —,\n\nWhat you should know: this paper claims the first direct photoemission signature of ground-state excitons in a bulk solid, coming from Ta2Pd3Te5. The observation is a flat, momentum-localized feature at Γ, present in the first BZ but absent in the second, with a temperature evolution that tracks the gap opening. If it holds, Ta2Pd3Te5 becomes the cleanest bulk excitonic insulator candidate, with a 365 K transition and negligible lattice distortion.\n\nWhat's genuinely new: the group previously used the same exciton-photoemission formula (Rustagi-Kemper) to see a signal above Tc in Ta2NiSe5. Here the same physical signature appears below Tc, in a material whose transition is nearly distortion-free, and the authors extract a Bohr radius (~14 Å) and infer an odd-parity exciton wavefunction. That below-Tc observation is the step forward.\n\nThe paper is careful in several ways: the phonon-replica alternative is explicitly simulated and dismissed, the feature survives down to 10 K, and the second-BZ absence is a natural control—assuming the normalization between zones is correct. The raw data appear sufficient to check the fits, though I haven't seen them.\n\nThe soft spots are about the isolation of the signal. The load-bearing step is subtracting the second-BZ spectrum as a bare valence band baseline. The normalization is not specified, and since the first-BZ valence band is much more intense at Γ, any zone-dependent matrix-element or final-state variation will produce a residual concentrated exactly where the claimed peak sits. The stress-test note is right that this needs a clear answer. It's not fatal—the temperature dependence of the residual is not what a simple artifact would look like—but it's the first thing I'd ask for.\n\nSecond, the 'simulation' of the exciton spectrum uses the same fitted 1s envelope that produced the fit, so it's a consistency check, not an independent prediction. The two-band model in the supplement is parameterized to open a 40 meV gap and then used to validate the excitonic phase; that's circular to a degree.\n\nThird, the energy separation between the exciton peak and the valence band top (about 37 meV) is smaller than the fitted linewidths (about 58 meV), so peak positions need error bars. And the odd-parity conclusion rests on an orbital assignment—the text mentions dxy in one place and dxz in another—without a matrix-element calculation.\n\nNone of these individually falsifies the claim, but together they make it conditional. I'd send it to a serious referee, with the subtraction normalization as the key question.\n\nBest,\n[Your name]","headline":"First direct photoemission evidence for a ground-state exciton in a bulk solid, pending proper control of the second-BZ subtraction.","tokens_in":16699,"tokens_out":3875,"would_cite":false,"duration_ms":41912,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["71.35.-y","79.60.-i"],"model":"deepseek-v4-flash","headline":"The paper reports direct photoemission evidence for spontaneously formed excitons in the insulating ground state of the bulk crystal Ta2Pd3Te5, with binding energy exceeding the single-particle band gap.","keywords":["excitonic insulator","Ta2Pd3Te5","exciton photoemission","angle-resolved photoemission spectroscopy","odd-parity exciton","metal-insulator transition","1s exciton wavefunction","Bohr radius"],"falsifier":"Calculate the one-step ARPES intensity for the first and second Brillouin zones from the DFT band structure at the same photon energy and polarization; if the calculated first-zone/second-zone intensity ratio is not constant over the momentum-energy window used for subtraction, the residual peak is at least partly a subtraction artifact. Experimentally, repeat the measurement at different photon energies (for instance 6 eV or 21 eV) and above Tc at 400 K; a feature that changes shape or survives in the metallic phase would not be tied to the excitonic ground state.","tokens_in":15481,"feed_emoji":"⚛️","tokens_out":14367,"duration_ms":133305,"temperature":0.7,"pith_summary":"Excitons—bound pairs of an electron and a hole—have long been predicted to form spontaneously in the ground state of a narrow-gap solid, but direct spectroscopic evidence in a three-dimensional crystal has been missing. The paper reports such evidence in Ta2Pd3Te5, a layered material that undergoes a metal-insulator transition at 365 K with almost no structural distortion. Using angle-resolved photoemission spectroscopy (ARPES) with orbital-selective light polarization, the authors isolate a non-dispersing photoemission feature at the Brillouin zone center that sits below the valence band maximum, meaning the exciton binding energy exceeds the single-particle band gap and can drive the phase transition. They extract a 1s-like exciton wavefunction with a Bohr radius near 14 Å and find the feature has odd parity in one mirror plane, which they attribute to interband hybridization between valence and conduction bands in the excitonic insulating state. If correct, this is the first direct photoemission observation of excitons in the ground state of a bulk crystalline solid.","feed_headline":"In Ta2Pd3Te5, excitons are caught in the ground state","feed_subtitle":"A flat peak below the valence band edge shows bound electron-hole pairs drive the metal-insulator transition.","key_machinery":"The load-bearing object is the exciton photoemission spectral function, which in the low-temperature, zero-center-of-mass limit reduces to $P(k) \\propto |\\varphi_{1s}(k)|^2 = 1/[1 + k^2 a_0^2/4]^4$, directly relating ARPES intensity to the squared exciton envelope wavefunction in momentum space. It comes from the convolution of the exciton wavefunction with conduction-band photoemission matrix elements, with the matrix element taken as constant. This formula carries the identification: fitting the momentum distribution curves of the subtracted feature yields the Bohr radius and confirms a 1s envelope, while the energy distribution curves yield the exciton binding energy. The second essential element is the orbital-selective dipole selection rule: comparing xz- and yz-polarized spectra distinguishes the parity of the wavefunction and identifies the odd-parity valence orbital contribution through interband hybridization.","core_discovery":"The central discovery is a stationary, roughly egg-shaped photoemission signal at the Brillouin zone center of Ta2Pd3Te5 in its low-temperature insulating phase, at a binding energy of about 0.115 eV (80 K), which lies below the valence band maximum at about 0.08 eV. The signal is absent in the second Brillouin zone, where only the bare valence bands appear, and it deviates from the calculated and measured valence band dispersions. Subtracting the second-zone spectrum from the first-zone spectrum yields a residual intensity that is fit quantitatively by the exciton photoemission formula with a 1s envelope, $P(k) \\propto 1/[1 + k^2 a_0^2/4]^4$, giving anisotropic Bohr radii of about 13.6 Å along Γ−Y and 15.1 Å along Γ−X. The feature is observed for xz- and x-polarized light but suppressed for yz-polarized light; the authors read this as odd parity along the yz mirror plane, involving an odd-parity component of the valence band through interband hybridization expected in an excitonic insulator. The exciton peak shifts and transfers weight with temperature in parallel with the gap opening, and it remains below the valence band top over the whole measured range below Tc, so the authors conclude the exciton binding energy exceeds the gap and drives the metal-insulator transition.","pith_inferences":["Inference: If the odd-parity assignment is right, the same exciton should be nearly dark in two-photon absorption, which couples to even-parity states; a clear optical two-photon resonance would therefore test the parity conclusion independently of photoemission.","Inference: The subtraction procedure assumes the second-Brillouin-zone spectrum is a clean proxy for the bare valence band. A one-step photoemission calculation of matrix elements in both zones would settle whether the extracted Bohr radius and binding energy are robust; the authors do not report such a calculation.","Inference: The contrast with Ta2NiSe5—where exciton photoemission appears only above Tc—suggests the two materials sit on opposite sides of the BEC-BCS crossover of the excitonic transition, with Ta2Pd3Te5 in the condensed regime and Ta2NiSe5 in the preformed-exciton regime.","Inference: The temperature-dependent broadening of the exciton peak could be compared with a phonon-scattering model; if the width tracks a single optical phonon energy, it would provide a quantitative measure of the exciton-phonon coupling that currently limits the interpretation."],"forward_implications":["The insulating state of Ta2Pd3Te5 below 365 K is an excitonic insulator: the observed bound exciton sits below the valence band maximum, so its formation energy exceeds the single-particle gap and can open the gap without relying on a lattice-driven mechanism.","The exciton wavefunction parameters—a 1s envelope with Bohr radii near 14 Å and a binding energy near 0.1 eV—become direct experimental targets for first-principles calculations; the existing monolayer tight-binding calculation predicts roughly twice the radius, pointing to where theory must be refined.","The odd-parity exciton symmetry, paired with the system's topological band structure, puts Ta2Pd3Te5 within reach of predictions of topological excitonic condensates and related exotic ground states.","Because the transition has an undetectably small structural distortion, Ta2Pd3Te5 provides a cleaner platform than earlier candidates for studying the crossover from preformed incoherent excitons to a condensed excitonic ground state in a bulk solid."],"supporting_citations":[{"why":"Supplies the exciton photoemission formula $P \\propto |\\varphi_{1s}|^2$ used to fit the measured momentum distributions.","marker":"[43]"},{"why":"Demonstrated exciton photoemission in Ta2NiSe5 with the same orbital-selection method and gives the even-parity benchmark this work contrasts.","marker":"[44]"},{"why":"Initial candidate-material report for Ta2Pd3Te5 with prior ARPES evidence of the gap-opening transition.","marker":"[45]"},{"why":"Provides the evidence for the excitonic insulator state and the band structure and orbital characters used in the analysis.","marker":"[46]"},{"why":"Reports spontaneous gap opening and potential excitonic states in the same material, supplying the DFT band structure and two-band model context.","marker":"[49]"},{"why":"Tight-binding calculation of excitons in the monolayer whose predicted radius (about twice the measured value) is the comparison target.","marker":"[47]"},{"why":"Supplies the experimental lattice constants and the topological band structure context used in the DFT calculations.","marker":"[51]"},{"why":"Reports the absence of kz dispersion at the valence band top, used to rule out final-state effects as the origin of the zone-center feature.","marker":"[58]"}],"fun_headline_variants":["Excitons spotted in Ta2Pd3Te5 ground state","Excitonic insulator caught with excitons in ground state","Odd-parity excitons in Ta2Pd3Te5 ground state","Direct photoemission of ground-state excitons in Ta2Pd3Te5","Excitons drive metal-insulator transition in Ta2Pd3Te5"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument assumes that the second-Brillouin-zone ARPES spectrum can be normalized and subtracted from the first-zone spectrum to expose the exciton signal, so that any difference in photoemission matrix elements, final-state effects, or surface contributions between the two zones does not create the residual feature.","fun_headline_variants_meta":{"raw":{"variants":["Excitons spotted in Ta2Pd3Te5 ground state","Excitonic insulator caught with excitons in ground state","Odd-parity excitons in Ta2Pd3Te5 ground state","Direct photoemission of ground-state excitons in Ta2Pd3Te5","Excitons drive metal-insulator transition in Ta2Pd3Te5"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000684,"raw_usage":{"total_tokens":3135,"prompt_tokens":1006,"completion_tokens":2129,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":622,"completion_tokens_details":{"reasoning_tokens":2034}},"tokens_in":622,"tokens_out":2129,"duration_ms":16061,"temperature":1.0,"reasoning_tokens":2034,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T15:09:05.161779+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Calculate the one-step ARPES intensity for the first and second Brillouin zones from the DFT band structure at the same photon energy and polarization; if the calculated first-zone/second-zone intensity ratio is not constant over the momentum-energy window used for subtraction, the residual peak is at least partly a subtraction artifact. Experimentally, repeat the measurement at different photon energies (for instance 6 eV or 21 eV) and above Tc at 400 K; a feature that changes shape or survives in the metallic phase would not be tied to the excitonic ground state.","supporting_citations":[{"cited_title":"Rustagi and A","cited_arxiv_id":null,"evidence_quote":"Supplies the exciton photoemission formula $P \\propto |\\varphi_{1s}|^2$ used to fit the measured momentum distributions."},{"cited_title":"Fukutani, R","cited_arxiv_id":null,"evidence_quote":"Demonstrated exciton photoemission in Ta2NiSe5 with the same orbital-selection method and gives the even-parity benchmark this work contrasts."},{"cited_title":"Huang, B","cited_arxiv_id":null,"evidence_quote":"Provides the evidence for the excitonic insulator state and the band structure and orbital characters used in the analysis."},{"cited_title":"Zhang, Y","cited_arxiv_id":null,"evidence_quote":"Reports spontaneous gap opening and potential excitonic states in the same material, supplying the DFT band structure and two-band model context."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Tight-binding calculation of excitons in the monolayer whose predicted radius (about twice the measured value) is the comparison target."},{"cited_title":"[11,43-46,49,51,68]","cited_arxiv_id":null,"evidence_quote":"Supplies the experimental lattice constants and the topological band structure context used in the DFT calculations."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reports the absence of kz dispersion at the valence band top, used to rule out final-state effects as the origin of the zone-center feature."}],"review_version":1}