{"id":"d9382649-fbaf-4f47-9a72-1b406ff53a1a","arxiv_id":"2506.21047","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Using charge-pair distribution functions from MeV ultrafast electron diffraction, the authors track valence-electron and hydrogen dynamics in NH3 photodissociation in real space and time.","lead":"This study shows that analyzing pair distances between all charges, not just atoms, in ultrafast electron diffraction data can follow how valence electrons and hydrogen atoms move during ammonia photodissociation. The approach may let researchers watch coupled electron and nuclear motion in many molecules.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The assignment of the time-zero 2–4 Å negative ΔCPDF to valence-electron motion assumes negligible nuclear motion, but the 130-fs IRF convolves in early N–H elongation; a quantitative NN decomposition of the convolved signal is needed.","rationale":"I read the paper as a proof-of-principle: CPDF is conceptually better than IAM-based PDF for light molecules, and the ab initio simulation reproduces the measured momentum-space PD (including the low-s inelastic feature where IAM fails) well enough to make the overall interpretation plausible. The real-space assignment nevertheless has a specific soft spot. The t=0 frame is not instantaneous: with a 130-fs IRF, the measurement integrates over positive delays comparable to the 70-fs time at which the authors already detect N–H elongation. The paper's argument that only eN pairs can produce a negative 2–4 Å band assumes no NN change at t=0, but this is only true for the instantaneous FC geometry, not for the convolved experimental observable. This does not refute the central claim; it makes the strongest version of it ('captured electron motion prior to substantial structural alterations') conditional on a quantitative decomposition. The reader's verdict of CONDITIONAL already captures this, so I recommend no change. A single computational check—NN-only contribution after IRF convolution—can settle the point.","tokens_in":12676,"tokens_out":11307,"duration_ms":139526,"concrete_test":"Using the same MD trajectories and inversion parameters (s-range 1–10 Å^-1, α=0.08 Å^2 in Eq. 3), compute the NN-only contribution to ΔCPDF at the nominal t=0 point after convolution with the 130-fs IRF, integrating |ΔCPDF_NN(r)| over r∈[2,4] Å and comparing with the total negative band integrated over the same window. If the NN-only integral is below ~5% of the total, the electron interpretation is supported; if it exceeds ~20%, the time-zero feature cannot be assigned predominantly to electron motion and the claim should be qualified.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that CPDF analysis of MeV UED data enables disentanglement of valence-electron and hydrogen dynamics. The decisive real-space evidence is the negative ΔCPDF band at 2–4 Å near time zero, interpreted as new long-range eN pairs from the diffuse excited electron cloud while the nuclei have not yet moved (Figs. 3(b), 4(a)). This interpretation requires the NN contribution to be zero in the measured time-zero frame. That condition is not guaranteed by the measurement: the instrument response is ~130 fs FWHM, and the theoretical ΔCPDF in Fig. 3(b) is explicitly convolved with a 130-fs Gaussian. Hydrogen moves quickly; the paper's own MD/ab initio analysis (Fig. 2(c)–(d)) needs N–H bond lengths near 1.28 Å to reproduce the 70-fs predissociation signal, and that delay lies within the IRF tail of the nominal t=0 point. Thus 'the signal from NN pair is zero' is an instantaneous Franck–Condon statement, not a statement about the convolved observable at t=0. If the NN-only part of the convolved ΔCPDF contributes non-negligibly in the 2–4 Å window, the time-zero band is partly structural and the electron-motion claim is overstated. The paper does not report this decomposition or an uncertainty estimate for it.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This Letter reports a MeV ultrafast electron diffraction (UED) study of the photodissociation of NH3 with an instrument response of approximately 130 fs FWHM. The authors use a charge-pair distribution function (CPDF) analysis, which retains the full scattering signal including inelastic contributions, and invert it to obtain time-resolved difference CPDF maps. They compare the experimental maps with ab initio fewest-switches surface-hopping simulations and with independent-atom-model PDF calculations. The central claim is that the CPDF analysis allows simultaneous real-space, real-time tracking of valence-electron and hydrogen dynamics: a negative ΔCPDF band at 2–4 Å near time zero is assigned to the diffuse excited electron cloud before nuclear motion, and later features are assigned to umbrella vibration, N–H bond breaking, and ground-state recovery.","tokens_in":12969,"tokens_out":5501,"duration_ms":60081,"significance":"If the central claim holds, this is a significant methodological advance. The use of CPDF to incorporate inelastic electron scattering and electron-density redistribution addresses a known limitation of standard PDF analysis for low-Z molecules. The enhanced temporal resolution allows the authors to observe the dynamics of NH3, a benchmark system, without deuteration. The comparison with independent ab initio MD simulations, using no fitted parameters to force agreement, is a genuine strength, and the decomposition into ee, eN, and NN contributions is a valuable theoretical tool. The conclusions, however, depend on quantitative support that the manuscript currently lacks in several places, most importantly the attribution of the time-zero 2–4 Å feature to electron motion.","major_comments":[{"comment":"The interpretation of the time-zero negative band at 2–4 Å as electron motion assumes that 'the nuclei has not yet moved' and hence that the NN contribution is zero. However, the measured ΔCPDF in Fig. 3(a) and the theoretical ΔCPDF in Fig. 3(b) are necessarily averaged over the ~130 fs FWHM instrument response, and the paper's own analysis requires an N–H bond length near 1.28 Å at 70 fs to reproduce the predissociation feature (Fig. 2(c)–(d)). Early N–H elongation within the IRF therefore contributes to the convolved signal at nominal t=0, so the instantaneous Franck–Condon decomposition in Fig. 4(a) does not by itself prove that the NN part of the convolved ΔCPDF is negligible in the 2–4 Å window. Please provide the ee/eN/NN decomposition of the convolved theoretical ΔCPDF at time zero, with an uncertainty estimate for the NN contribution; without this, the assignment of the 2–4 Å negative band to valence-electron motion is not uniquely supported.","section":"Fig. 4(a) and End Matter on CPDF inversion and IRF"},{"comment":"The experimental ΔCPDF is presented without error bars or a noise-level threshold, and the claimed 'excellent agreement' is not quantified. The central real-space evidence is a small negative band at 2–4 Å near time zero, so the reader cannot judge whether this feature is statistically significant relative to scan-to-scan fluctuations. Please report the uncertainty of the experimental ΔCPDF (e.g., standard error propagated from the 4700 scans) and a quantitative agreement metric (e.g., a noise-weighted residual between experiment and ab initio ΔCPDF).","section":"Fig. 3(a) and comparison with Fig. 3(b)"},{"comment":"The CPDF inversion uses a fixed s-range (1–10 Å⁻¹) and damping parameter α=0.08 Ų, but no validation is shown that this inversion faithfully recovers the true CPDF of NH3 without truncation artifacts. Because both experimental and theoretical maps are processed identically, the differential comparison is internally consistent, but the real-space distances and the relative signs of the bands in the 2–4 Å region, which are central to the disentanglement claim, depend on the fidelity of the inversion. A forward-model test (e.g., inverting a simulated intensity from a known CPDF with the same s-range and α) would substantiate the real-space interpretation.","section":"End Matter, Eq. (3)"}],"minor_comments":[{"comment":"The caption states that the total ΔCPDF is scaled by a factor of 6, but the text does not explain why this scaling is applied; including unscaled curves or stating the scaling explicitly would help the reader assess the relative contributions of ee, eN, and NN pairs.","section":"Fig. 4 caption"},{"comment":"The definitions do not specify the normalization of P_uv(r) or the charge convention for electrons (Z = −1) versus nuclei; please make these conventions explicit.","section":"End Matter, Eqs. (1)–(2)"},{"comment":"The phrase 'excellent agreement' should be supplemented with a quantitative comparison, since no goodness-of-fit statistic is reported for the measured versus simulated PD curves.","section":"Discussion of Fig. 2(d)"},{"comment":"There is a typo: 'photexcited ammonia' should be 'photoexcited ammonia'.","section":"Main text, summary paragraph"},{"comment":"The extrapolation procedure for the elastic signal relies on a smooth low-order polynomial fit, but no uncertainty in the extrapolated values is reported; since these values are used to extract the inelastic decay constant, a brief sensitivity statement would be useful.","section":"Fig. 6 and End Matter on inelastic extraction"}],"recommendation":"major_revision","confidential_remarks":"The manuscript fits the scope of the journal and the methodological claim is of broad interest, but the time-zero assignment needs the requested convolved decomposition before the central claim can be considered established. The requested analysis is feasible with the existing simulation data, so this is a revision-level issue rather than a rejection. I do not see a novelty-disclosure concern relative to Ref. [24]; the improved time resolution and the CPDF-based real-space decomposition are new contributions, and the earlier work is properly cited."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear [Colleague],\n\nThe headline is that this paper demonstrates something real: applying the charge-pair distribution function to gas-phase MeV UED to get simultaneous information about valence electrons and hydrogen motion in ammonia. The 130 fs resolution on NH3 itself (not ND3) is an advance over Champenois et al., and the CPDF inversion from the full scattering signal is a reasonable route around the inelastic-scattering contamination that broke the earlier PDF analysis. The experimental work is substantial, and the comparison with ab initio surface-hopping simulations is honest—no fitted parameters force the agreement, and the same inversion parameters are used for theory and data.\n\nWhat I like: the low-s region that IAM misses is reproduced by ab initio theory, giving the enhanced band below 2 Å^-1 a plausible electronic origin. The decomposition into ee, eN, and NN contributions in Fig. 4 is a nice use of theory to interpret the features, and the authors are upfront about the ground-state-recovery discrepancy.\n\nThe soft spots are real but not fatal. There are no error bars on the ΔCPDF maps and no quantitative goodness-of-fit for the 'excellent agreement.' More importantly, the central time-zero claim—the negative band at 2–4 Å being purely electronic—rests on the assertion that the NN signal is zero because the nuclei have not moved. That is true for an instantaneous Franck–Condon geometry, but the measured t=0 is convolved with a 130 fs IRF. The paper itself needs N–H bonds near 1.28 Å to explain the 70 fs predissociation peak, which is inside the IRF tail. The theoretical ΔCPDF in Fig. 3(b) is convolved, but the decomposition in Fig. 4(a) appears instantaneous; the authors never show the convolved NN contribution in the 2–4 Å window. So the electron-motion attribution is plausible but not as clean as the abstract implies. A referee should ask for a quantitative decomposition of the convolved signal at t=0, not just the instantaneous one.\n\nThis deserves a serious referee. The experiment is hard, the method is promising, and the data look genuine. With error bars and a clearer treatment of the IRF convolution, the central claim could be solid. As is, I'd accept with major revision.","headline":"A solid feasibility study of CPDF on gas-phase ammonia UED, but the time-zero electronic assignment needs a convolved decomposition to fully land.","tokens_in":13481,"tokens_out":2742,"would_cite":true,"duration_ms":28596,"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":"MeV ultrafast electron diffraction with charge-pair analysis resolves simultaneous valence-electron and hydrogen motion in photoexcited ammonia.","keywords":["charge-pair distribution function","ultrafast electron diffraction","valence electron dynamics","ammonia photodissociation","real-time imaging","femtosecond dynamics","nonadiabatic dynamics","conical intersection"],"falsifier":"A direct test would be to repeat the measurement on deuterated ammonia (ND3) under the same 130-fs instrument response: if the 2-4 Å negative band at time zero is caused solely by the diffusing electron cloud, it should appear with nearly the same magnitude and decay in ND3; if it shifts or diminishes because the heavier deuterons move more slowly, a nuclear contribution would be exposed. Alternatively, a frozen-nucleus ab initio calculation of the ΔCPDF at time zero, with all atoms held at their ground-state equilibrium positions, should reproduce the observed band if the electronic assignment is correct.","tokens_in":12468,"feed_emoji":"⚛️","tokens_out":6421,"duration_ms":61158,"temperature":0.7,"pith_summary":"The paper seeks to establish that a charge-pair distribution function (CPDF) analysis of mega-electron-volt ultrafast electron diffraction data can separate, in real space and real time, the motions of valence electrons from those of hydrogen nuclei during a photochemical reaction. Using ammonia photodissociation as the test case, the authors show that with roughly 130-femtosecond temporal resolution the full scattering signal, elastic and inelastic, can be inverted into three charge-pair components: electron-nucleus, nucleus-nucleus, and electron-electron. Because electrons and hydrogen nuclei carry the same unit charge, the electron-scattering signal carries comparable sensitivity to both species, which is what makes simultaneous imaging possible. If correct, the approach turns UED from a structure-only probe into a tool for watching electronic and nuclear dynamics as they couple in real time.","feed_headline":"Imaging electron and hydrogen motion in ammonia in real time","feed_subtitle":"A charge-pair analysis of ultrafast electron diffraction separates valence-electron motion from N-H bond breaking.","key_machinery":"The charge-pair distribution function (CPDF), defined as $\\mathrm{CPDF}(r) = \\sum_{uv} Z_u Z_v P_{uv}(r)$, where the sum runs over electron-nucleus, nucleus-nucleus, and electron-electron pairs, is the central object. It generalizes the pair distribution function by including electron-nucleus and electron-electron pairs alongside nucleus-nucleus pairs, and it is recovered from the measured scattering intensity by a Fourier sine transform over the full $s$ range with a damping term. The key property exploited in this work is that electron and hydrogen-nucleus charges are equal in magnitude, making the scattering signal comparably sensitive to valence-electron and hydrogen motion, so that the three CPDF components can be separated by simulation and assigned to distinct reaction stages.","core_discovery":"The central claim is that, for a molecule as light as ammonia, the time-resolved percent-difference electron diffraction signal can be inverted into a CPDF that exhibits distinct, assignable features for valence-electron motion and hydrogen dynamics. The experiment captures, in succession, the redistribution of the excited electron cloud in the Franck-Condon region, the umbrella-mode vibration of the hydrogens, the breaking of the N-H bond along adiabatic and non-adiabatic dissociation pathways, and the vibration of hot ground-state molecules. The assignment of the early-time negative CPDF band at 2-4 Å to the diffusing excited electron cloud rests on the fact that nuclei are still at their equilibrium positions at time zero, so only new electron-nucleus pairs can create long-range negative signals. The authors further show that an independent-atom-model PDF analysis fails to reproduce the measured low-angle enhancement, while the ab initio CPDF calculation reproduces both the momentum-space and real-space features, supporting the claim that valence-electron information is being retrieved.","pith_inferences":["If the electronic assignment of the early band is correct, the same analysis applied to a series of photoexcited hydrides such as water or methane should show a similar time-zero negative eN feature whose range reflects the size of the excited orbital, which could be checked with existing UED data.","Combining CPDF analysis with coincident ion or fluorescence detection could tag the reaction channel per event and verify the channel-decomposed CPDFs predicted by the simulation.","The method implicitly assumes that the independent-atom form factor model is inadequate for light molecules; a quantitative comparison of CPDF-derived electron densities with quantum-chemistry densities would turn the claim into a calibrated measurement of excited-state charge distributions."],"forward_implications":["For molecules where valence-electron redistribution is strong and light atoms dominate the nuclear motion, independent-atom-model PDF analysis fails; the CPDF route extends ultrafast electron diffraction to such systems.","The retrieved time-zero electron signal and its roughly 96 fs decay provide a direct measure of the S1 state lifetime in ammonia, comparable to velocity-map imaging values.","Because the electron-electron CPDF equals the difference in radial distribution measured by x-ray scattering, electron and x-ray experiments on the same system could isolate the eN and NN contributions.","The observed excess in the 2.5-4 Å region after about 200 fs indicates that surface-hopping simulations underestimate the ground-state recovery channel, giving a quantitative target for theory."],"supporting_citations":[{"why":"Previous ammonia UED study at about 500 fs resolution that could not track predissociation in real space; the baseline this work improves on.","marker":"[24]"},{"why":"Introduced the CPDF formalism for structure retrieval in liquid-phase electron scattering, which this paper adapts to gas-phase molecules.","marker":"[33]"},{"why":"Showed that inelastic UED signal encodes electron-correlation changes upon S0-to-S1 excitation, providing the foundation for retrieving electronic dynamics.","marker":"[21]"},{"why":"Demonstrated the double-bend achromat compressor that delivers roughly 50 fs electron pulses, enabling the 130 fs instrument response used here.","marker":"[9]"},{"why":"Supplies the quasi-diabatic two-state Hamiltonian for NH3 used to run the surface-hopping molecular dynamics.","marker":"[25]"},{"why":"Provides the framework for including excited electronic states in total isotropic scattering calculations used in the ab initio signal.","marker":"[27]"},{"why":"Used to perform the fewest-switches surface-hopping dynamics that produce the molecular trajectories.","marker":"[38]"},{"why":"Used to compute ab initio elastic and inelastic scattering signals from the trajectories at the CASSCF level.","marker":"[39]"}],"fun_headline_variants":["Charge-pair imaging tracks valence electrons and hydrogens in ammonia","Ultrafast diffraction disentangles electron and hydrogen motion","Real-space view of valence electron and proton dynamics","Ammonia photodissociation probed with charge-pair analysis","Electron and hydrogen motion resolved in ammonia by UED"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The interpretation of the early-time negative signal at 2-4 Å as purely electronic assumes that the hydrogen nuclei have not yet moved appreciably within the first 100 femtoseconds, so that no new nucleus-nucleus pairs exist at those distances.","fun_headline_variants_meta":{"raw":{"variants":["Charge-pair imaging tracks valence electrons and hydrogens in ammonia","Ultrafast diffraction disentangles electron and hydrogen motion","Real-space view of valence electron and proton dynamics","Ammonia photodissociation probed with charge-pair analysis","Electron and hydrogen motion resolved in ammonia by UED"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000233,"raw_usage":{"total_tokens":1460,"prompt_tokens":882,"completion_tokens":578,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":498,"completion_tokens_details":{"reasoning_tokens":496}},"tokens_in":498,"tokens_out":578,"duration_ms":6026,"temperature":1.0,"reasoning_tokens":496,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T22:35:29.527473+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct test would be to repeat the measurement on deuterated ammonia (ND3) under the same 130-fs instrument response: if the 2-4 Å negative band at time zero is caused solely by the diffusing electron cloud, it should appear with nearly the same magnitude and decay in ND3; if it shifts or diminishes because the heavier deuterons move more slowly, a nuclear contribution would be exposed. Alternatively, a frozen-nucleus ab initio calculation of the ΔCPDF at time zero, with all atoms held at their ground-state equilibrium positions, should reproduce the observed band if the electronic assignment is correct.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Previous ammonia UED study at about 500 fs resolution that could not track predissociation in real space; the baseline this work improves on."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Introduced the CPDF formalism for structure retrieval in liquid-phase electron scattering, which this paper adapts to gas-phase molecules."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Showed that inelastic UED signal encodes electron-correlation changes upon S0-to-S1 excitation, providing the foundation for retrieving electronic dynamics."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Demonstrated the double-bend achromat compressor that delivers roughly 50 fs electron pulses, enabling the 130 fs instrument response used here."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the quasi-diabatic two-state Hamiltonian for NH3 used to run the surface-hopping molecular dynamics."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the framework for including excited electronic states in total isotropic scattering calculations used in the ab initio signal."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Used to perform the fewest-switches surface-hopping dynamics that produce the molecular trajectories."},{"cited_title":"Richter, P","cited_arxiv_id":null,"evidence_quote":"Used to compute ab initio elastic and inelastic scattering signals from the trajectories at the CASSCF level."}],"review_version":1}