{"id":"85313dbb-89bb-496b-af6c-7f5d1669694f","arxiv_id":"2502.03361","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"PEEM reveals exciton-polariton dispersion in WSe2 waveguides, with an energy splitting of about 95 meV for the transverse electric mode.","lead":"Using photoemission electron microscopy, researchers measured the dispersion of light-guiding modes in thin WSe2 flakes and found an energy splitting near the exciton resonance, a signature of strong light-matter coupling. The work demonstrates a new way to study exciton-polaritons that could extend to real-time tracking of energy propagation.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The reported ΔE = 94.8 meV rests on 'turning points' in k(E) that are not intrinsic to planar-waveguide polariton dispersion; they may be artifacts of FFT-peak switching under strong absorption.","rationale":"The paper is a good-faith experimental demonstration with independent FDTD support, openly available data, and a plausible physical picture. The reader's weakest assumption concerns the mapping between the PEEM wave-pattern period and the true mode wavevector; that assumption is supported by prior PEEM work and by the FDTD pipeline, so I do not treat it as the most load-bearing issue. The more specific weak point is the quantitative extraction of ΔE from 'turning points' in a k(E) curve that, for a planar-waveguide polariton dispersion, should be monotonic. The local maximum and minimum in Fig. 4(f) are therefore more likely artifacts of FFT peak switching or of the normalization/filtering procedure than genuine dispersion extrema. Because the central claim includes a specific Rabi splitting of about 95 meV, this extraction procedure needs independent validation. The paper already includes one such validation on FDTD data, which is encouraging, but it does not establish robustness against the experimental complications (strong absorption near resonance, potassium-induced changes, finite ROI, row normalization). The proposed coupled-mode fit and synthetic-pipeline test would settle whether the turning-point method measures the coupling strength or merely the energy at which the FFT maximum jumps between branches. This is a refinement of the reader's conditional verdict rather than a reason to reject: the qualitative demonstration of PEEM-based dispersion measurement can stand even if the precise splitting value changes, but the quantitative claim should be conditional on the extraction method being validated.","tokens_in":17681,"tokens_out":11423,"duration_ms":123466,"concrete_test":"Using the published Zenodo data for the 30 nm flake, reproduce the FFT dispersion in Fig. 4(b) and the k(E) curve in Fig. 4(f). Then fit the extracted k(E) with a coupled-oscillator model E±(k) = (E_c(k)+E_ex)/2 ± sqrt((E_c(k)-E_ex)^2 + Ω^2)/2, with E_c(k) a linear bare-waveguide dispersion and E_ex fixed at the WSe2 A-exciton energy, allowing for a complex k to represent absorption. Compare the fitted Ω with 94.8 meV. Additionally, generate synthetic wave patterns from the FDTD fields with a known input Ω, apply the exact experimental pipeline (ROI truncation, high-pass filter, per-energy normalization, FFT peak tracking, Savitzky-Golay smoothing, turning-point detection), and vary ROI length, noise level, and linewidth.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central quantitative result, ΔE = 94.8 meV, is extracted in Fig. 4(f) from a local maximum and minimum in the k(E) curve obtained by tracking, for each photon energy, the maximum of the normalized FFT signal and then applying a Savitzky-Golay filter. For a planar waveguide coupled to a single exciton, the polariton branches in the first quadrant are monotonic: the lower branch has k → ∞ as E approaches E_ex from below, and the upper branch starts at k = 0 above E_ex and increases. There are no intrinsic turning points in k(E). The apparent extrema must therefore reflect the positions where the FFT maximum switches between the lower branch, the upper branch, or the k/k0 = 1 laser feature, combined with strong absorption, the finite ROI, row-by-row normalization, and smoothing. The energy separation between such switching points depends on detection thresholds, window length, and filter parameters, not directly on the Rabi splitting. The authors' internal check on FDTD data (84.4 meV from turning points vs. 81.5 meV from an intensity-based method) is reassuring but does not establish that the turning-point estimator is unbiased for the experimental conditions, where linewidths, potassium-induced doping, and noise differ. If the observed extrema are analysis artifacts, the claimed coupling strength is not measured.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports the use of photoemission electron microscopy (PEEM) to image wave patterns of propagating waveguide modes in exfoliated WSe2 flakes and to extract their dispersion relations around the A-exciton resonance (E_ex ≈ 1.6 eV). For a laser polarization parallel to the excitation edge, the authors observe a TE-mode whose dispersion bends near the exciton, and they interpret this as the lower polariton branch of an exciton-polariton. For perpendicular polarization, they report an essentially uncoupled TM-mode. Supporting finite-difference time-domain (FDTD) simulations use literature optical constants and reproduce the main experimental dispersions. From the experimental dispersion of a thinner flake, the authors report an energy splitting of ΔE = 94.8 meV, obtained from turning points in the photon-energy dependence of the FFT-peak wavevector after Savitzky-Golay smoothing; the same procedure applied to simulated data gives 84.4 meV, consistent with an independent intensity-based estimate of 81.5 meV. The paper concludes that PEEM can measure exciton-photon coupling in TMDC waveguides and suggests time-resolved PEEM as a future extension.","tokens_in":17904,"tokens_out":4292,"duration_ms":40735,"significance":"If the ΔE value is reliable, the paper establishes PEEM as a viable far-field method for measuring exciton-polariton dispersions in TMDC waveguides, complementing SNOM and cathodoluminescence. The main strengths are the independent FDTD validation with literature optical constants (no free parameters), the clear polarization selectivity between TE and TM modes, the open data availability, and the demonstration of a quantitative extraction of a coupling energy. The claims are falsifiable, and the central experimental observation — a dispersive bending near the exciton resonance — is supported by the simulations. However, the quantitative central number, ΔE = 94.8 meV, currently rests on an estimator whose unbiasedness for the experimental conditions is not established, and the supporting experimental details (flake thickness, upper-branch signal-to-noise) are incomplete.","major_comments":[{"comment":"The experimental splitting ΔE = 94.8 meV is extracted from turning points in a smoothed k(Eph) curve of FFT-peak maxima (Fig. 4(f)). No uncertainty or robustness analysis is provided: there are no error bars, no variation of the Savitzky-Golay window, no variation of the ROI or normalization scheme, and no test against alternative estimators. As the authors themselves note, the branches of a planar waveguide polariton are monotonic in k(E); the observed extrema therefore reflect the switching of the FFT maximum between branches or between a branch and the k/k0 = 1 laser feature. The internal check on FDTD data (84.4 meV vs 81.5 meV) is encouraging, but it is a single simulation dataset and does not demonstrate that the turning-point estimator is unbiased in the experimental conditions, where absorption, potassium-induced doping, pulse-duration variation, and noise differ. Please add an explicit robustness analysis — e.g., injecting noise into the FDTD field data, varying the smoothing window and ROI, and reporting the resulting spread of ΔE — or fit the two branches (and the k/k0 = 1 feature) simultaneously to estimate ΔE with a confidence interval. Without this, the central quantitative claim is not supported.","section":"Results and Discussion, Fig. 4"},{"comment":"The experimental upper polariton branch in Fig. 4(b) is only visible after per-energy normalization, and the paper gives no quantification of its significance (signal-to-noise ratio, peak amplitude relative to background, or reproducibility across repeated measurements). Because the turning-point method relies on the upper branch's position, it is essential to demonstrate that this branch is not a normalization artifact, particularly in the energy range where the k/k0 = 1 laser feature overlaps. Please provide a quantitative measure of the upper-branch visibility (e.g., an intensity profile along the upper branch, or a comparison with the simulated data at the same signal level), and state the detection threshold used to identify the branch.","section":"Results and Discussion, Fig. 4"},{"comment":"The flake used for the ΔE measurement is described only as 'significantly smaller,' and its AFM thickness is never given in the main text or in the figure caption. The FDTD simulations in Fig. 4 are for 30 nm, and the SI text seems to assume this value, but the experimental flake may not be exactly 30 nm. This is a crucial sample parameter for reproducibility and for the validity of the simulation comparison and the coupling-threshold discussion. Please provide the measured thickness of this flake, and if it differs from 30 nm, discuss how the comparison is affected.","section":"Results and Discussion, Fig. 4 caption"}],"minor_comments":[{"comment":"The phrase 'transversal electric and magnetic modes' should be 'transverse electric and magnetic modes'.","section":"Abstract"},{"comment":"There is a typo in 'Vext = 10 kV for this study' (missing space); also, the abbreviation 'Vext' should be defined (extractor voltage).","section":"Experimental Details"},{"comment":"The statement 'there seems to be a threshold for the waveguide thickness' based on simulations should be explicitly labeled as a simulation prediction, not an experimental observation, since the experiments in the paper do not test the threshold directly.","section":"Results and Discussion"},{"comment":"The blue experimental curve in Fig. 4(f) ends at high energy without a clear indication of the error or of the number of data points; consider plotting the raw FFT maxima as markers along with the smoothed curve.","section":"Fig. 4(f)"},{"comment":"Ref. 44 is the software package 'Tidy3D'; please include a version number or the date of access, and consider adding a note in the text about the simulation code's availability and validation (the paper already mentions the mesh-convergence check, which is good).","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the journal's scope and the experimental platform is interesting. The main issue is that the quantitative claim of ΔE = 94.8 meV lacks uncertainty quantification and an explicit test against the alternative explanation of FFT-peak switching. This is fixable with additional analysis (robustness tests, error bars, or a simultaneous-branch fit), so I recommend major revision rather than rejection. I also note the absence of the AFM thickness for the key flake, which must be added."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear [Name],\n\nThe paper does something genuinely new: it shows that PEEM can image the interference wave pattern of waveguide modes in WSe2 flakes and, by tuning photon energy, extract the dispersion of TE and TM modes around the A-exciton. The observation that the TM mode does not couple to the exciton in a 55 nm flake, while the TE mode does, is a nice result and is backed by independent FDTD simulations using literature optical constants, not fitted parameters. The data is open on Zenodo, and the authors are candid about limitations, including the difficulty of seeing the upper branch.\n\nThe soft spot is the central quantitative claim: ΔE = 94.8 meV. The extraction uses 'turning points' in the k(E) curve from the FFT maximum at each photon energy, after smoothing. As your stress-tester notes, the true polariton branches are monotonic; any local extrema in k(E) must come from peak-tracking switching between branches or the light line, plus absorption and normalization. The authors' FDTD validation is a reasonable check—they recover 84.4 meV from turning points versus 81.5 meV from an intensity-based method—but this is a single simulation with known parameters, not a robustness study. It doesn't rule out that the experimental result, measured under different linewidths and with potassium doping, is biased by the analysis procedure. There are no error bars, and the upper branch is only visible after normalization, so the whole value rests on a weak feature.\n\nI wouldn't call this fatal, because the qualitative demonstration of PEEM for polariton dispersions holds up, and the polarization-selective coupling is interesting in its own right. But the ΔE value should be treated with caution until the method is shown to be robust. A referee should ask for error bars, a sensitivity analysis of the turning-point extraction (filter width, ROI choice, alternative methods), and ideally a measurement on a flake where the upper branch is stronger.\n\nI'd bring this to the reading group, and I'd send it for peer review rather than desk-reject it. The technique advance is worth refereeing even if the quantitative result needs strengthening.","headline":"A promising technical demonstration of PEEM for TMDC polariton dispersions, but the headline Rabi splitting rests on a peak-tracking heuristic that needs robustness checks before being believed.","tokens_in":18464,"tokens_out":5525,"would_cite":true,"duration_ms":52157,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["71.36.+c","79.60.-i"],"model":"deepseek-v4-flash","headline":"PEEM can measure exciton-polariton dispersion in WSe2 waveguides and extract the coupling energy.","keywords":["Exciton-polaritons","WSe2","Photoemission electron microscopy","Waveguide modes","Strong coupling","Rabi splitting","Two-photon photoemission","Dispersion relation"],"falsifier":"A direct falsifying test would be to measure the same flake with an independent momentum-resolved probe, such as near-field scanning optical microscopy or angle-resolved reflectance, and compare the extracted $k/k_0$ peak positions and the inferred $\\Delta E$; a systematic mismatch at particular photon energies would show that the photoemission fringe period is not a faithful readout of the mode wavevector. Repeating the measurement with freshly evaporated potassium and after it has decayed would test for work-function-induced distortions.","tokens_in":17489,"feed_emoji":"🔬","tokens_out":7573,"duration_ms":66752,"temperature":0.7,"pith_summary":"The paper aims to establish photoemission electron microscopy (PEEM) as a direct probe of exciton-polaritons in thin transition-metal dichalcogenide waveguides. It claims that the fringes seen in two-photon photoemission images of exfoliated WSe2 flakes encode the in-plane wavevector of propagating waveguide modes, so a Fourier transform of the images yields the mode dispersion. From the lower and upper polariton branches of that dispersion the paper extracts the exciton-photon coupling strength, reporting $\\Delta E = 94.8$ meV for the TE-mode polariton in a 30 nm flake. If this works, PEEM could trace polariton propagation and energy exchange in real time with pump-probe methods.","feed_headline":"PEEM maps exciton-polariton dispersions in WSe2 waveguides","feed_subtitle":"The method reads waveguide-mode wavelengths from photoemission fringes and yields a 94.8 meV coupling splitting.","key_machinery":"The central mechanism is two-photon photoemission interference imaging: the electric field of a propagating waveguide mode beats against the second photon of the excitation pulse, imprinting fringes in the photoelectron yield whose period equals the mode wavelength. A fast Fourier transform of the PEEM image along the propagation direction converts each fringe pattern into a peak at the in-plane wavevector $k$, and scanning the photon energy $E_{\\mathrm{ph}}$ maps the dispersion $E_{\\mathrm{ph}}(k/k_0)$, with $k_0$ the vacuum wavevector. The TE or TM character of each feature is assigned from the dominant electric-field component in FDTD simulations, and the coupling strength is read from the local maximum and minimum (turning points) of the polariton branches after Savitzky-Golay smoothing. Lowering the work function with potassium so that 2PPE works with the same pulses that excite the polaritons is the enabling preparative step.","core_discovery":"The authors demonstrate that when a tunable near-infrared laser is focused onto the edge of a thin WSe2 flake, the launched transverse electric and transverse magnetic waveguide modes interfere with the second photon of the two-photon photoemission process and create a spatial wave pattern in the photoelectron yield. Fourier-transforming these patterns as a function of excitation photon energy reconstructs the dispersion relation of the modes. In a 55 nm flake the TE mode shows the lower-polariton bending around the A-exciton at $E_{\\mathrm{ex}} \\approx 1.6$ eV, while the TM mode remains a straight line, showing no coupling at that thickness; FDTD simulations reproduce both behaviours and trace the difference to how much of the mode field is confined inside the waveguide. In a 30 nm flake both polariton branches are visible, and the turning points of the dispersion yield $\\Delta E = 94.8$ meV for the TE-mode exciton-polariton. The central discovery is therefore a new application: PEEM can image strong-coupling dispersions in TMDC waveguides with polarization-selective excitation of TE and TM modes.","pith_inferences":["Editorial inference: the same fringe images contain propagation-length information, so fitting the fringe decay as a function of photon energy could extract the imaginary part of the polariton wavevector, not just the dispersion.","Editorial inference: a natural transfer test is to apply the method to other TMDCs or heterobilayers and compare the PEEM-derived splitting with an independent angle-resolved reflectivity measurement on the same flake.","Editorial inference: the supplementary three-photon photoemission data suggest the method can work without alkali doping at higher intensities, which if developed further would remove the potassium layer as a possible perturbation of the near-field.","Editorial inference: a direct quantitative comparison of acquisition time, mode selectivity, and energy range against SNOM would make the authors' claim of practical advantage testable."],"forward_implications":["PEEM becomes a parallel-imaging alternative to SNOM and cathodoluminescence for measuring waveguide-polariton dispersions in TMDCs.","Laser polarization gives a direct control knob: parallel polarization couples to the TE mode and shows strong coupling, while perpendicular polarization launches the TM mode, which stays uncoupled below a thickness threshold.","The reported $\\Delta E = 94.8$ meV provides a quantitative benchmark that connects PEEM data to coupling-strength estimates from near-field and cathodoluminescence studies.","Because PEEM records a full field of view at once, the same geometry can be extended with pump-probe excitation to follow polariton propagation and energy exchange in the time domain.","The thickness thresholds (TE coupling by 20–30 nm, TM coupling only near 60 nm) give a practical design rule for future waveguide-polariton samples."],"supporting_citations":[{"why":"Supplies the prior nano-optical imaging of WSe2 waveguide modes whose light-exciton interaction signatures this paper extends to PEEM.","marker":"(14)"},{"why":"Provides the SNOM benchmark of exciton-polariton transport in MoSe2 waveguides against which the 94.8 meV splitting is compared.","marker":"(15)"},{"why":"Supplies the cathodoluminescence study of WSe2 flakes whose larger splittings are explained by exciton density and Cherenkov effects.","marker":"(16)"},{"why":"Establishes normal-incidence PEEM for imaging surface plasmon polaritons, the interference-imaging approach adapted here.","marker":"(20)"},{"why":"Demonstrates photonic near-field imaging in multiphoton PEEM, providing the basis for converting photoemission fringes into wavevectors.","marker":"(27)"},{"why":"Shows subwavelength visualization of light in thin-film waveguides with photoelectrons, a direct methodological precedent.","marker":"(29)"},{"why":"Supplies the finite-difference time-domain solver used to simulate the waveguide dispersions and field components.","marker":"(44)"},{"why":"Provides the optical constants for multilayer WSe2 used as input for the FDTD simulations.","marker":"(45)"}],"fun_headline_variants":["PEEM sees polaritons in WSe2 waveguides","PEEM images WSe2 polariton couplings directly","Photoemission microscope reveals polaritons in WSe2","PEEM measures 94.8 meV polariton splitting in WSe2","Waveguide exciton-polaritons imaged by PEEM in WSe2"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is that the spatial period of the photoemission wave pattern equals the true wavelength of the propagating waveguide mode, so a Fourier transform of the PEEM image yields the mode's real in-plane wavevector without wavelength-dependent phase shifts or distortions from the two-photon process or the potassium layer.","fun_headline_variants_meta":{"raw":{"variants":["PEEM sees polaritons in WSe2 waveguides","PEEM images WSe2 polariton couplings directly","Photoemission microscope reveals polaritons in WSe2","PEEM measures 94.8 meV polariton splitting in WSe2","Waveguide exciton-polaritons imaged by PEEM in WSe2"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000302,"raw_usage":{"total_tokens":1745,"prompt_tokens":955,"completion_tokens":790,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":571,"completion_tokens_details":{"reasoning_tokens":698}},"tokens_in":571,"tokens_out":790,"duration_ms":6924,"temperature":1.0,"reasoning_tokens":698,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-09T04:57:39.774268+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct falsifying test would be to measure the same flake with an independent momentum-resolved probe, such as near-field scanning optical microscopy or angle-resolved reflectance, and compare the extracted $k/k_0$ peak positions and the inferred $\\Delta E$; a systematic mismatch at particular photon energies would show that the photoemission fringe period is not a faithful readout of the mode wavevector. Repeating the measurement with freshly evaporated potassium and after it has decayed would test for work-function-induced distortions.","supporting_citations":[],"review_version":1}