{"id":"c7f40f26-ab90-4ccf-a106-261509c9535c","arxiv_id":"2506.03647","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"Edge polaritons, hybrid phonon-plasmon waves guided along metal-insulator boundaries, appear in phase-separated NdNiO3 films and their character depends on boundary sharpness.","lead":"Mid-infrared near-field microscopy reveals bright phase peaks at boundaries between metallic and insulating regions in thin NdNiO3 films, attributed to edge-confined mixed phonon-plasmon polaritons. Simulations show the polaritons change character with edge smoothness, from a one-dimensional edge state at abrupt boundaries to epsilon-near-zero absorption at smooth boundaries, suggesting correlated oxides can host infrared nanophotonics.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claim of long-propagating edge polaritons rests on FEM simulations, not on any measured propagation; the experimental phase peak alone cannot distinguish a propagating edge state from a local ENZ or lossy-edge effect.","rationale":"I read the paper as claiming a genuine experimental demonstration of edge polaritons at MI boundaries, with propagation over microns and ~100 nm confinement. What would have to be true is that the measured phase peak is caused by launching a propagating edge mode and that the simulated mode properties are correct. The least secure condition is the propagation part: a static phase peak at the boundary is compatible with at least two mechanisms (ENZ-governed local response and edge-state canalization) as the paper itself states, and the data do not include a direct measurement of propagation along the edge. The propagation length and confinement are extracted from a FEM simulation whose material parameters and boundary profile are assumed rather than measured at the nanoscale. The reader's identified weakness about the effective-medium interpolation and inferred boundary width is real and related; I would sharpen it to the unmeasured propagation, since even if Eq. (1) and the mono-phase permittivities were exact, the experimental phase peak alone would still not demonstrate a long-lived edge mode. A direct edge line-scan fringe measurement would settle this. This concern does not invalidate the observation of the phase peak or the plausibility of the simulations, but it does mean the paper's strongest claim should remain conditional.","tokens_in":18335,"tokens_out":9034,"duration_ms":106145,"concrete_test":"Map s-SNOM amplitude and phase along a current-induced filament edge at 940 cm^-1, scanning the tip along the boundary starting from a natural corner or termination for at least 4 µm. If the simulated edge polariton exists with the claimed several-micron propagation, the line scan should show standing-wave fringes with period 2π/Re(q_ES) and an envelope decaying with the predicted length; fitting the profile yields both. If the signal along the boundary is flat or decays without the predicted fringe period or envelope, the long-propagation claim is experimentally unsupported. Repeat at a second laser frequency to check that the fringe period tracks the simulated edge-polariton dispersion.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The experimental evidence for the headline claim consists of a phase maximum at MI boundaries (Fig. 2b,e,h). The manuscript itself shows (Fig. 4a,b) that both a smooth boundary (ENZ absorption) and a sharp boundary (edge state) produce such a peak, and the choice between these mechanisms is made using the peak width to infer the boundary width w in Eq. (2). The 'strongly confined and long-propagating' edge polariton is then illustrated by the FEM field map of Fig. 4e, with propagation over several microns shown in the simulation only. The simulation inputs are mono-phase permittivities from far-field Drude-Lorentz fits and an assumed linear interpolation (Eq. 1). No experimental scan along the boundary, no edge-polariton standing-wave fringes, and no decay-length measurement are presented. Therefore the observed phase peak does not by itself establish a propagating edge mode; a localized edge resonance, or an artifact of harmonic normalization at an abrupt material step, would produce a qualitatively similar single-wavelength peak. The load-bearing component 'long-propagating' is currently undemonstrated in the data.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports mid-infrared s-SNOM measurements on 10-nm and 40-nm NdNiO3 films across the metal-insulator transition, either thermally induced or created by Joule-heated filaments. The central observation is a pronounced peak in the near-field phase at metal-insulator boundaries, accompanied by a characteristic flipped-S amplitude-phase correlation. The phase peak is reproduced by FEM simulations in which the local film dielectric function is linearly interpolated between far-field extracted metallic and insulating values with a boundary of variable width w. The simulations show two mechanisms: for smooth boundaries the peak is an ENZ-enhanced surface-polariton response; for sharp boundaries it is a 1D edge state with lateral confinement of about 100 nm and simulated propagation over several micrometers. The paper concludes that ignoring nonlocal polaritonic effects may bias estimates of boundary widths and argues for edge polaritons in correlated oxides.","tokens_in":18650,"tokens_out":7188,"duration_ms":82841,"significance":"If correct, the work would be significant because it extends edge-polariton physics to strongly correlated 'bad metals', a material class not previously considered for such low-loss modes, and it cautions that near-field images of phase-separated oxides should not be interpreted locally. The experimental dataset is a strength: the phase peak is reproduced across temperatures (92-95 K), two film thicknesses, and multiple devices, including electrically written filaments, and topography crosstalk is explicitly excluded. The use of a full 3D Maxwell FEM solver rather than a modal expansion, and the demonstration that smooth and sharp boundaries give distinguishable amplitude-phase correlations, are also strengths. The significance is currently limited by the fact that the headline properties of the mode, namely its propagation length and one-dimensional nature, are demonstrated only in simulation, not in the measured near-field data.","major_comments":[{"comment":"The discrimination between the ENZ mechanism and the one-dimensional edge state is made through the boundary width w in Eq. (2), but w is not independently measured. The text compares the experimental phase-peak FWHM (115-190 nm) with simulated FWHMs and concludes that the phase-separated state is 'close to the ES limit'; this effectively calibrates w to the very observable that the simulations are meant to explain. Because Fig. 4a and 4b show that both a smooth and a sharp boundary produce a phase peak, the measured peak alone does not identify the mechanism. An independent determination of w (e.g., from the temperature profile of the Joule-heated filament or from a boundary-sensitive probe) or a discrimination criterion that does not use w is needed to support the edge-state assignment.","section":"Edge polaritons and the nature of the phase peak (Fig. 4)"},{"comment":"The claim that the edge polaritons are 'long-propagating' over 'several microns' (Fig. 4e, inset of Fig. 4f) and the figure of merit 'well above 10' rest entirely on the FEM simulation. No experimental scan along the boundary, no edge-polariton standing-wave fringes, and no measured decay length are presented in the paper. The experimental observable is a local phase maximum, which cannot by itself distinguish a propagating mode from a localized edge resonance or enhanced absorption at a material step. Given that the abstract states this as a central result, the propagation claim should either be supported by real-space propagation data or explicitly downgraded to a simulated prediction.","section":"Fig. 4e and Discussion"},{"comment":"The simulation inputs are the mono-phase dielectric functions epsilon_met = -230 + 160i and epsilon_ins = 30 + 5i at 940 cm-1, obtained by Drude-Lorentz fitting of spatially averaged far-field reflectivity, combined with the linear interpolation of Eq. (1) and the piecewise-linear boundary profile of Eq. (2). The paper does not validate these assumptions at the nanoscale. If the local dielectric functions inside the phase-separated regions differ from the far-field averages, or if the linear mixing rule fails in the mixed region near f = f0, then the simulated field maps and the resulting ENZ/ES assignment could change. A robustness check, such as varying the interpolation rule within effective-medium bounds or using position-dependent spectra, would strengthen the central interpretation.","section":"Polariton dispersion and s-SNOM signal far from MI boundaries; Methods"}],"minor_comments":[{"comment":"The sentence 'the peak width for the sharp edge (170 nm in Fig. 4a)' appears to refer to Fig. 4b, since Fig. 4a shows the smooth boundary with FWHM = 400 nm; please correct the figure reference.","section":"Discussion"},{"comment":"The comparison between the experimental APC curves and the simulated intermediate-width curve (green, w = 100 nm) is only qualitative; overlaying the experimental data on the simulated APC curves would make the claimed resemblance more convincing.","section":"Fig. 2 and Fig. 4c"},{"comment":"The piecewise definition of f(x) would be clearer if the two constant regions were written explicitly as f(x) = 0 for x < -w/2 and f(x) = 1 for x > w/2, rather than using the Heaviside function in combination with the ramp.","section":"Eq. (2)"},{"comment":"The abstract's phrase 'nonlocal plasmonic effects' may be misleading: the FEM model uses local dielectric functions with a spatially varying metallicity, not a wavevector-dependent material response; consider rephrasing to 'spatially inhomogeneous' or 'boundary-mediated' effects.","section":"Introduction"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is within the journal's scope and the experimental core is solid; the main barrier is the gap between the abstract's 'demonstrate ... long-propagating' wording and the simulation-only evidence for propagation. I would ask the authors to either add propagation measurements or revise the claims accordingly."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper reports a clear, reproducible near-field phase maximum at metal-insulator boundaries in phase-separated NdNiO3 films, and attributes it to edge polaritons whose character depends on boundary smoothness. That observation is likely real: it shows up across temperatures, in both spontaneously phase-separated and current-driven filament states, and in multiple devices, with no corresponding topography. The amplitude-phase correlation analysis is careful, and the FEM simulations reproduce the qualitative features, including the crossover from an ENZ-governed response for smooth boundaries to a narrower edge-state-like peak for sharp ones. This is a new system and a new pairing of mechanisms, and it deserves attention from people working on nanophotonics in correlated oxides.\n\nThe soft spots are in the gap between the experimental observation and the headline claim. The experiment shows a phase peak, not a propagating mode. The \"strongly confined and long-propagating\" edge polariton with micron-scale propagation is a simulation result: the field maps in Fig. 4e and the propagation-length figure of merit come from FEM, not from any measured decay length, fringes, or edge-launched standing waves. The paper's own Fig. 4a,b shows that both a smooth ENZ boundary and a sharp edge produce a phase peak, so the peak alone does not discriminate; the discrimination relies on the inferred boundary width w, which is calibrated to the measured peak width. That is a fitted parameter doing real work. The dielectric inputs are also far-field fitted values extrapolated to the nanoscale, with a linear interpolation assumption. None of these are fatal, but they mean the paper's central claim is underdetermined by the current data.\n\nSmaller issues: no error bars on the phase-peak widths, no data or code release, and the discussion sometimes slides from \"simulation shows\" to what reads like an established property. Those are addressable.\n\nWho is this for: experimentalists and theorists in mid-infrared nanophotonics, near-field optics, and strongly correlated oxides. A serious referee can push for direct propagation evidence—scanning along the boundary to detect decay or fringes—or at least a much clearer separation of the ENZ versus edge-state assignment using an independent measurement of the boundary width. I would not desk-reject this; the observation is solid enough to warrant refereeing, and the interpretation, while partly simulation-dependent, is plausible and worth sharpening.\n\nRecommendation: send to review, with the expectation of major revision requesting either propagation measurements or a significantly toned-down propagation claim.","headline":"Solid near-field observation of an edge phase peak in a phase-separated nickelate, but the 'long-propagating' edge polariton is currently a simulation, not a measured fact.","tokens_in":19136,"tokens_out":2708,"would_cite":true,"duration_ms":27228,"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":"Edge polaritons of mixed phonon-plasmon character emerge at metal–insulator boundaries in thin NdNiO3 films, appearing as a one-dimensional edge state at sharp edges and as epsilon-near-zero response at smooth edges.","keywords":["edge polaritons","metal-insulator transition","phase separation","NdNiO3","correlated oxides","epsilon-near-zero","phonon-plasmon polaritons","near-field infrared nanoscopy"],"falsifier":"Measure the local dielectric function across a metal–insulator boundary in these films with nano-FTIR at sub-50-nm resolution, and check whether the phase peak survives when the measured permittivity profile across the boundary is used in the simulation; a large mismatch between the measured profile and the linear interpolation, or a phase peak with no corresponding edge-mode field map, would falsify the assignment.","tokens_in":18155,"feed_emoji":"🔬","tokens_out":12873,"duration_ms":116638,"temperature":0.7,"pith_summary":"This paper sets out to explain a sharp phase peak that scattering-type scanning near-field optical microscopy sees exactly at metal–insulator boundaries in thin NdNiO$_3$ films, in a material family usually dismissed as too lossy for plasmonics. It claims the peak is the fingerprint of edge polaritons—hybrid phonon-plasmon modes confined to the boundary—that propagate several microns while keeping their cross-section near 100 nm. The paper further claims that the electromagnetic nature of these modes depends on boundary smoothness: a one-dimensional optical edge state at abrupt edges, and epsilon-near-zero absorption when the boundary is broad. If correct, near-field maps of phase-separated correlated oxides cannot be read as purely local dielectric images, and the same materials become candidates for tunable infrared nanophotonics.","feed_headline":"Edge polaritons confine infrared light to 100-nm channels","feed_subtitle":"Near-field phase peaks at metal–insulator boundaries come from edge-confined modes, not just a local dielectric response.","key_machinery":"The load-bearing mechanism is the edge polariton: a quasi-one-dimensional electromagnetic mode that exists only at the metal–insulator boundary, carrying a mixture of phonon and plasmon character because the film's dielectric function combines phonon contributions with the free-carrier response of the metallic phase. In the simulations the boundary is parametrized by a metallicity factor $f$ between insulator and metal, Eq. (1), with a width $w$ that varies from sharp to smooth; the same interpolation places the epsilon-near-zero point at $f \\approx 0.16$. The near-field signal is computed with a point-dipole emitter and a probe-point field $E_z$ in a finite-element model, and the dispersion of surface polaritons is obtained from the momentum- and frequency-dependent reflection coefficient $r_p(q,\\omega)$. The amplitude–phase correlation curves, where the phase is plotted against the amplitude along a scan, provide the diagnostic that separates tip-excited surface-polariton interference (clockwise rotation) from the edge-polariton anomaly (counterclockwise rotation).","core_discovery":"On its own terms, the paper's central claim is that the pronounced near-field phase maximum observed at metal-insulator boundaries in phase-separated NdNiO$_3$ films is caused by edge-confined polaritons rather than by a local change in the dielectric function alone. The claim is supported by finite-element simulations in which a dipole source moves across a boundary described by a linear interpolation of the metallic and insulating dielectric functions with a metallicity parameter $f$. For an abrupt boundary the simulations show a narrow one-dimensional edge state that canalizes energy along the boundary for several microns; for a smooth boundary the response is controlled by epsilon-near-zero absorption where $\\mathrm{Re}\\,\\varepsilon$ passes through zero. The two regimes connect continuously as the boundary width is varied, and the observed phase-peak widths (about 115–300 nm in different devices and temperatures) place the experiments between the sharp-edge and intermediate regimes. The paper also attributes the oscillations in the metallic region near boundaries to surface-polariton interference excited by the tip and reflected at the boundary.","pith_inferences":["One consequence the authors leave implicit is that phase maps of other phase-separated metal-insulator-transition materials, such as VO$_2$ or manganites, should be re-examined for the same nonlocal edge contribution before boundary widths are extracted from near-field profiles.","A direct test would be to measure the local dielectric function across a boundary with nano-FTIR at sub-50-nm resolution; if the real profile deviates strongly from the linear metallicity interpolation, the simulated edge-state to epsilon-near-zero crossover would need revision.","Because the edge-state to epsilon-near-zero crossover is continuous in boundary width, strain, current, or defect engineering could tune a single device between the two regimes, effectively making the edge state switchable.","The claim that most of the edge-state field is in vacuum suggests that thinner films or substrates with lower phonon losses could push propagation lengths further, an untested path toward low-loss infrared oxide plasmonics."],"forward_implications":["If the central claim is right, a s-SNOM phase peak at a metal–insulator boundary is a nonlocal polaritonic effect, and ignoring it will overestimate the actual physical width of the boundary.","The width of the phase peak becomes a usable indicator of the boundary regime: values near the sharp-edge limit (about 170 nm in the experiments) signal a true edge state, while broader peaks (about 300 nm) signal a more smeared epsilon-near-zero-dominated boundary.","Edge polaritons appear whether the metal–insulator transition is reached thermally or by current-induced Joule heating, so the effect does not depend on the particular way the boundary is created.","The propagation length to lateral confinement ratio above 10, together with the fact that much of the field sits in vacuum, points to these oxide edges as usable infrared waveguides despite the material's high losses."],"supporting_citations":[{"why":"Supplies the observation of coexisting first- and second-order electronic phase transitions in nickelates that motivates the nanoscale phase-separated boundaries.","marker":"[3]"},{"why":"Provides the real-space mapping of sheet and edge plasmons in graphene plus the simplified dipole/probe-field simulation approach used to model edge polaritons.","marker":"[12]"},{"why":"Provides the amplitude–phase correlation and surface-polariton interference interpretation used to separate edge-polariton from surface-polariton signatures.","marker":"[13]"},{"why":"Supplies the harmonic normalization scheme (s3/s2) that removes slow spatial illumination variations from the near-field signal.","marker":"[28]"},{"why":"Introduces the amplitude–phase correlation diagnostic applied to the experimental maps.","marker":"[29]"},{"why":"Shows how electric current locally heats the film above the transition and defines the filament boundary widths used for the current-induced experiments.","marker":"[30]"},{"why":"Supplies the substrate dielectric function and infrared nanoscopy methodology used to extract the film permittivity for the simulations.","marker":"[44]"}],"fun_headline_variants":["Edge polaritons confine infrared light in oxide films","Metal-insulator boundaries host edge polaritons","Infrared light squeezed into oxide edge channels","Oxide edges guide long-propagating polaritons"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole interpretation rests on the assumption that the two coexisting phases can be described by the spatially averaged dielectric constants $\\varepsilon_{\\mathrm{met}} = -230 + 160i$ and $\\varepsilon_{\\mathrm{ins}} = 30 + 5i$ and that the boundary region is a linear mixture of them; if the local dielectric function at the nanoscale is not that mixture, the simulated edge modes and the phase-peak assignment would not hold.","fun_headline_variants_meta":{"raw":{"variants":["Edge polaritons confine infrared light in oxide films","Metal-insulator boundaries host edge polaritons","Infrared light squeezed into oxide edge channels","Oxide edges guide long-propagating polaritons"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00019,"raw_usage":{"total_tokens":1340,"prompt_tokens":944,"completion_tokens":396,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":560,"completion_tokens_details":{"reasoning_tokens":335}},"tokens_in":560,"tokens_out":396,"duration_ms":4696,"temperature":1.0,"reasoning_tokens":335,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T10:57:37.806484+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the local dielectric function across a metal–insulator boundary in these films with nano-FTIR at sub-50-nm resolution, and check whether the phase peak survives when the measured permittivity profile across the boundary is used in the simulation; a large mismatch between the measured profile and the linear interpolation, or a phase peak with no corresponding edge-mode field map, would falsify the assignment.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the observation of coexisting first- and second-order electronic phase transitions in nickelates that motivates the nanoscale phase-separated boundaries."},{"cited_title":"Nikitin, P","cited_arxiv_id":null,"evidence_quote":"Provides the real-space mapping of sheet and edge plasmons in graphene plus the simplified dipole/probe-field simulation approach used to model edge polaritons."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the amplitude–phase correlation and surface-polariton interference interpretation used to separate edge-polariton from surface-polariton signatures."},{"cited_title":"Mester, A","cited_arxiv_id":null,"evidence_quote":"Supplies the harmonic normalization scheme (s3/s2) that removes slow spatial illumination variations from the near-field signal."},{"cited_title":"Schnell, P","cited_arxiv_id":null,"evidence_quote":"Introduces the amplitude–phase correlation diagnostic applied to the experimental maps."},{"cited_title":"Luibrand, A","cited_arxiv_id":null,"evidence_quote":"Shows how electric current locally heats the film above the transition and defines the filament boundary widths used for the current-induced experiments."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the substrate dielectric function and infrared nanoscopy methodology used to extract the film permittivity for the simulations."}],"review_version":1}