{"id":"06687857-8155-4e60-8ef5-27e14baed391","arxiv_id":"2607.25556","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Stepwise ionization changes hydrogen beam stopping in mirror-trap plasma by up to ~20%, but the fast-ion distribution's anisotropy changes it by less than 1%.","lead":"A new open-source tool computes how fast hydrogen atoms get ionized in hot plasma, including the effect of non-uniform fast ion populations, and applies it to two Russian magnetic-mirror fusion machines. The headline numbers: stepwise ionization matters at the ~20% level and the angular shape of fast ions barely matters (<1%) for beam penetration in these devices.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"7% non-additivity claim may be within atomic-data uncertainty: dominant ion-impact channel is unbenchmarked and uses an approximate de-excitation relation.","rationale":"The reader's weakest-assumption identification focuses on the provenance of the GDML fast-ion distribution and regime parameters, which is a valid input-provenance concern. However, the most load-bearing assumption for the central non-additivity claim is the accuracy of the ion-impact atomic data and the de-excitation approximation, because ion-impact processes dominate the capture rate (~50%) and are entirely unvalidated. The paper's own limitation statements (Klein–Rosseland approximation for protons, no benchmark for ion-impact) support this concern. The 7% non-additivity is a small effect relative to the ~10% electron-channel discrepancy, so without uncertainty propagation the quantitative claim is not secure. This does not contradict the reader's conditional verdict; it reinforces it. The suggested concrete test would directly determine whether the effect survives under plausible atomic-data variations.","tokens_in":9695,"tokens_out":5609,"duration_ms":61307,"concrete_test":"Recompute the exact vs. additive mean free paths for 40 keV atoms in the GDML regime while varying all proton-impact cross sections (excitation, ionization, charge exchange) systematically by ±10% (or by their stated uncertainties), and, independently, replace the Klein–Rosseland proton de-excitation rates with rates obtained from a more rigorous treatment (e.g., detailed balance at the actual relative kinetic energy, or cross sections from a quantum-mechanical H+H+ database). If the 7% non-additivity changes by more than a few percentage points or changes sign, the claim is not robust to atomic-data uncertainty.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claim—up to 7% reduction in mean free path due to non-additivity of species contributions—rests on the level-population model's ability to capture the coupling between electron, warm-ion, and fast-ion collisions. The paper validates only the electron-impact channel against OPEN ADAS, finding ~10% deviations (Fig. 1). Yet in the GDML application, ion-impact processes contribute ~50% of the capture rate (Section 4), and the proton-impact excitation/ionization/charge-exchange cross sections are taken from the same Janev–Smith report without independent benchmark. Moreover, de-excitation for proton collisions is computed with the Klein–Rosseland relation (Eq. 2), whose derivation assumes a light projectile (electron). The authors apply it to protons while neglecting kinetic-energy transfer, an approximation they flag in Section 2. For fast atoms (20–40 keV), the fractional energy loss per collision is indeed small, but the resulting de-excitation rates could still be significantly wrong because the detailed-balance phase-space factor for equal-mass collisions differs from the electron case. No sensitivity analysis or error propagation is given. The 7% effect is comparable in magnitude to the known 10% uncertainty in the electron channel; with an unquantified ion-channel uncertainty, the non-additivity could plausibly be an artifact of the chosen cross-section set or the Klein–Rosseland approximation. The paper does not establish that the 7% reduction is physically robust rather than within the noise of the atomic data.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript describes HIR, a steady-state collisional-radiative code for hydrogen level populations and ionization dynamics in plasma. The model includes electron and ion impact excitation/de-excitation/ionization, charge exchange, and spontaneous radiative transitions, with cross sections taken from Janev-Smith and transition probabilities from Johnson. The author compares electron-impact effective ionization rates with OPEN ADAS (agreement within about 10%), studies the role of stepwise ionization (up to 20% contribution, requiring levels up to N=8), and applies the model to the GDML and GOL-NB mirror-trap facilities. The principal quantitative claims are that stepwise ionization makes species-specific collision contributions non-additive, reducing the fast-atom mean free path by up to 7% at 40 keV in GDML, and that anisotropy of the fast-ion distribution affects the ionization dynamics by less than 1%. The ionization cost in GOL-NB plasma is also computed as a function of density and temperature.","tokens_in":9914,"tokens_out":4882,"duration_ms":54684,"significance":"The HIR package is openly available and the comparison with OPEN ADAS provides an external benchmark for the electron-impact channel. The non-additivity effect is conceptually important for neutral-beam capture modeling in mirror machines and is plausibly a real consequence of stepwise population coupling. However, the quantitative 7% claim is not fully established: the ion-impact channel, which dominates capture in the GDML application, is not independently benchmarked, and the GDML fast-ion distribution and operational parameters are asserted without provenance or a sensitivity study. If the supporting calculations are made reproducible and the ion-channel uncertainty is quantified, the paper would be a useful contribution to the atomic-physics and plasma-modeling literature.","major_comments":[{"comment":"The GDML regime parameters (Te=650 eV, Tiw=1080 eV, ne=4.5e13 cm^-3, niw=2.8e13 cm^-3, nif=1.7e13 cm^-3) and the fast-ion distribution function in Fig. 4 are presented without citation, construction details, or a numerical definition. The distribution is shown only as a velocity-space plot. Every quantitative result in Figures 5 and 6, including the 7% non-additivity claim, depends on this input. The author should provide the source simulation or reference, a tabulated/numerical definition of the distribution function, and ideally a sensitivity scan over plausible variations of the distribution and regime parameters.","section":"Section 4, Figure 4 and operational regime"},{"comment":"The electron-impact channel is benchmarked against OPEN ADAS to about 10%, but the ion-impact channel is not independently validated. In the GDML application the text states that collisions with thermal ions contribute about 50% of the capture rate. The proton-impact excitation/ionization/charge-exchange cross sections from Janev-Smith [5] have no external benchmark here, and de-excitation for proton impact is computed using the Klein-Rosseland relation (Eq. 2), which is derived for light projectiles (electrons) and ignores kinetic-energy transfer. The author notes this approximation but does not quantify its error. Since the claimed 7% non-additivity is comparable in magnitude to the known ~10% uncertainty in the electron channel and the ion-channel uncertainty is unquantified, the central quantitative claim is not yet robust. A sensitivity analysis (e.g., varying the ion-impact cross s","section":"Section 2, Eq. (2), and Section 4 (GDML ion-impact channel)"},{"comment":"The statement that 'levels up to N=8' are required for stepwise ionization is based on visual inspection of Figures 3 and 5, but no convergence criterion or quantitative threshold is given. The effect of adding levels appears to saturate only gradually, and the GDML/GOL-NB calculations use imax=11 and 13 due to motional-Stark autoionization. The central claim about the importance of stepwise ionization would be strengthened by reporting the relative change in the effective ionization rate or mean free path when the level cutoff is increased from N=7 to N=8, N=9, etc., rather than a qualitative statement.","section":"Section 3, Figure 3, and Section 4, Figure 5"}],"minor_comments":[{"comment":"The notation for spontaneous radiative transitions is inconsistent: the first sum uses Aik while the text defines Aki as the spontaneous transition probability. Presumably the first term should be sum_k n_k A_ki (transitions from upper k to lower i). Please clarify the index convention.","section":"Eq. (1)"},{"comment":"The axes in the uploaded figure are garbled (the x-axis appears as 10^40 through 10^3 and the labels are difficult to read). The caption and figure should be regenerated clearly, and the meaning of the three curves (scd89_h, scd96_h, scd12_h) should be explicitly tied to the plot.","section":"Figure 1"},{"comment":"The ALADDIN database is spelled 'Aladdin' in the text; please use the official spelling. Also, it would be helpful to state the version or retrieval date for the GitHub repository and databases referenced.","section":"Section 2"},{"comment":"The fractional-energy components E0/2, E0/3, and E0/18 are mentioned, and Fig. 6 shows energies 2.22, 13.3, 20, and 40 keV. For the reader it would be useful to explicitly connect these values to the full and fractional energies for the GDML beam (E0=40 keV).","section":"Section 4, fractional beam components"}],"recommendation":"major_revision","confidential_remarks":"The manuscript's quantitative conclusions for GDML rest on an undocumented fast-ion distribution and asserted regime parameters, and the dominant ion-impact channel is unbenchmarked. I would ask the editor to require the author to make the distribution data and regime parameter sources available, and to add a sensitivity analysis for the ion-impact cross sections and the de-excitation approximation. With those additions, the paper could become a solid contribution; as it stands, the 7% non-additivity claim is suggestive rather than established."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: worth reading if you work on neutral-beam capture in mirror machines, worth refereeing, but the authors need to pin down input provenance and add error bars before the quantitative claims can be taken at face value.\n\nWhat's actually new: the anisotropic fast-ion distribution module and the concrete numbers for GDML/GOL-NB — 20% stepwise contribution, N=8 convergence, 7% non-additivity at 40 keV, <1% anisotropy effect, and the density-dependent ionization cost. The CR method itself is standard (Janev–Smith cross-sections, Johnson A-coefficients), but the package is public and the facility-specific statements haven't appeared elsewhere. The paper is also honest: it flags the Klein–Rosseland approximation for proton de-excitation, notes inaccuracies in refs [5,6], and admits ADAS cross-section provenance is opaque.\n\nWhere it holds up: Eq. (1) is the standard steady-state CR balance, solved as stated. The electron-impact effective rate matches OPEN ADAS within ~10% across a wide temperature range — a genuine external anchor. That covers the electron channel only.\n\nWhere the soft spots are: (i) The GDML regime parameters and, more importantly, the fast-ion distribution function in Fig. 4 are asserted without any construction, simulation, or reference. Since that distribution drives the 7% and <1% numbers, this is a provenance gap, not a detail. (ii) The ion-impact and charge-exchange channels contribute ~50% of capture in GDML, but are not benchmarked at all; they rely on Janev–Smith and the Klein–Rosseland detailed-balance formula written for electrons. The stress-test concern lands here: a 7% non-additivity effect is smaller than the known 10% uncertainty in the electron channel and could be an artifact of the ion data. The paper should include a sensitivity run varying those cross-sections. (iii) No error bars on any quantitative claim. (iv) The repo is public but no commit hash or data files are given, so nothing is pinned.\n\nMinor: the GDML/GDMT naming inconsistency in the code and text should be fixed.\n\nBottom line: this is a workmanlike engineering-physics paper with a useful public code and plausible numbers, but the headline 7% is not yet demonstrated robust against the atomic-data uncertainty. Not fatal — addressable in a revision. Send it to a referee with expertise in atomic data for beam diagnostics; the author's own limitations section is a sign they know what's missing.","headline":"A useful, honest CR-model paper whose main quantitative claim (7% non-additivity) is plausible but rides on unbenchmarked ion-impact data.","tokens_in":10544,"tokens_out":2011,"would_cite":true,"duration_ms":20346,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["52.20.Hv","52.50.Gj"],"model":"deepseek-v4-flash","headline":"Stepwise ionization makes hydrogen capture in mirror-trap plasmas non-additive across plasma species, cutting fast-atom mean free paths by up to 7%.","keywords":["collisional-radiative model","hydrogen ionization","stepwise ionization","neutral beam injection","mirror trap","fast ions","mean free path","ionization cost"],"falsifier":"A beam-attenuation scan in GOL-NB across plasma densities 1e12–1e14 cm-3: the model predicts a density-driven variation of about 10% in the effective capture cross section and a ~4% excess over summed per-species cross sections at 1e14 cm-3, while an additive model predicts no such density dependence. Alternatively, recompute the GDML mean free paths with an independently documented fast-ion distribution; the 7% non-additivity figure would shift if that input is wrong.","tokens_in":9452,"feed_emoji":"⚛️","tokens_out":22232,"duration_ms":189694,"temperature":0.7,"pith_summary":"The paper claims that hydrogen atoms entering hot plasma are ionized largely through excited states rather than directly from the ground state, and that this stepwise route changes how fast-neutral beams are captured in mirror-trap fusion devices. Its collisional-radiative model, HIR, solves the steady-state populations of the excited levels and shows that stepwise ionization contributes up to 20% of the effective ionization rate, with levels up to the principal quantum number N=8 required for convergence. Because every plasma species — electrons, warm ions, and fast ions — feeds the same excited-state population, collisions with different species do not add independently: in the GDML regime the mean free path of 40 keV atoms is up to 7% shorter than an additive estimate. A second finding is that the fast-ion distribution's anisotropy changes ionization dynamics by less than 1%, so isotropic distributions are adequate for beam-capture modeling in this class of device. A reader would care because these are concrete corrections to neutral-beam and fueling calculations in existing and planned mirror traps.","feed_headline":"7% shorter fast-atom range when ionization channels couple","feed_subtitle":"Excited-state paths supply up to 20% of hydrogen ionization; per-species models overestimate atom range.","key_machinery":"The machinery is a steady-state collisional-radiative balance over hydrogen principal quantum levels: each level's population is fixed by electron- and ion-impact excitation, de-excitation, ionization, charge exchange, and spontaneous emission, with rate coefficients built from a standard fusion atomic-data cross-section set, the Klein–Rosseland detailed-balance relation for de-excitation, and the Johnson formula for radiative transition probabilities. Because all plasma species drive the same excited-state population, per-species contributions to ionization cannot be separated — this shared-population coupling is the source of the non-additivity. For fast atoms, a motional-Stark autoionizat","core_discovery":"The paper's central claim is that a hydrogen atom in hot plasma is ionized largely through excited states, coupling collisions with different plasma species into one non-additive problem. Solving the steady-state population balance, the paper finds stepwise ionization contributes up to about 20% of the effective ionization rate, with levels through N=8 needed to capture it. In the GDML and GOL-NB mirror-trap regimes, that coupling shortens the mean free path of 40 keV atoms by up to 7% versus additive estimates; an isotropic stand-in for the anisotropic fast-ion distribution changes results by less than 1%. The capture cross section and the ionization cost depend on plasma density — about 10","pith_inferences":["If the non-additivity generalizes, any plasma with several collision partners sharing an excited-state manifold — including beam-heated tokamaks with a fast-ion population — should expect per-species ionization additivity to fail at the few-percent level wherever stepwise ionization is active.","The sub-1% anisotropy insensitivity, if it holds beyond this one regime, justifies a cheap modeling shortcut: isotropicizing the fast-ion energy distribution suffices for beam-capture studies, so long as the fractional-energy beam components are retained.","A testable prediction follows from the density dependence: a beam-attenuation scan in GOL-NB across 1e12–1e14 cm-3 should reveal the ~4% non-additive excess at the high-density end, a signal an additive model cannot produce.","Because the paper validates its rate data only to about 10% against an independent atomic database, the exact sizes of the corrections (20% stepwise, 7% non-additivity) are worth rechecking against another cross-section compilation; the qualitative coupling effect is unlikely to vanish."],"forward_implications":["Fast-neutral beam capture in GDML/GOL-NB-class mirror traps should be computed with coupled multi-level populations; adding per-species rates separately overestimates the mean free path by up to 7% at 40 keV.","Stepwise ionization is largest at high density and low temperature, but at densities above 1e13 cm-3 it stays significant even above 10 keV electron temperature, so ground-state-only rates can undercount ionization by up to 20%.","Retaining levels through N=8 is enough to capture stepwise ionization for slow atoms; for fast atoms in a magnetic field the effective truncation is set higher by motional-electric-field autoionization (n=11–13 here).","Isotropic fast-ion distributions are an adequate stand-in for anisotropic two-dimensional distributions when computing ionization dynamics in GDML-type plasmas: the difference is below 1%.","In GOL-NB conditions, the effective capture cross section and the ionization cost vary with plasma density (about 10% over 1e12–1e14 cm-3) because collisional de-excitation increasingly outcompetes radiative decay as density rises."],"fun_headline_variants":["Stepwise ionization adds up to 20% to hydrogen rate","Coupled ionization paths shrink fast-atom range 7%","Excited states drive up to 20% of hydrogen ionization","Anisotropic fast ions shift ionization by under 1%","Plasma density alters hydrogen ionization cost"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"Every quoted number — the 20% stepwise share, the 7% non-additivity, the sub-1% anisotropy effect — inherits the accuracy of the adopted electron- and ion-impact cross-section fits (which the paper benchmarks only to roughly 10% against an independent atomic database) and, for the facility studies, a fast-ion distribution function and plasma parameters that are asserted without citation; if the cross sections or those inputs are off, the quantitative claims move.","fun_headline_variants_meta":{"raw":{"variants":["Stepwise ionization adds up to 20% to hydrogen rate","Coupled ionization paths shrink fast-atom range 7%","Excited states drive up to 20% of hydrogen ionization","Anisotropic fast ions shift ionization by under 1%","Plasma density alters hydrogen ionization cost"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000594,"raw_usage":{"total_tokens":2631,"prompt_tokens":765,"completion_tokens":1866,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":509,"completion_tokens_details":{"reasoning_tokens":1785}},"tokens_in":509,"tokens_out":1866,"duration_ms":14293,"temperature":1.0,"reasoning_tokens":1785,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T02:05:41.215914+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A beam-attenuation scan in GOL-NB across plasma densities 1e12–1e14 cm-3: the model predicts a density-driven variation of about 10% in the effective capture cross section and a ~4% excess over summed per-species cross sections at 1e14 cm-3, while an additive model predicts no such density dependence. Alternatively, recompute the GDML mean free paths with an independently documented fast-ion distribution; the 7% non-additivity figure would shift if that input is wrong.","supporting_citations":[],"review_version":1}