{"id":"3276319e-2501-4b9a-829a-5373e42f26fc","arxiv_id":"2412.09267","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Apollon's F1 beamline achieved 2 PW operation, with 45 J on target, 50 MeV protons, and 4x10^8 neutrons per shot, and demonstrated simultaneous F1/F2 operation.","lead":"The Apollon laser facility in France has completed a 2 petawatt commissioning of its main F1 beam, delivering 45 J, 22 fs pulses and generating 50 MeV proton beams from solid targets. This report provides the first full performance characterization of this new user facility, including focal spot, X-ray, neutron, and dual-beam capabilities, which is what future experimental teams need when designing campaigns.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Unmeasured full-energy temporal contrast makes the simulation-derived plasma characterization and the extrapolated neutron total non-unique; the direct beam and source measurements themselves are sound.","rationale":"The reader's conditional verdict is appropriate. The paper's headline quantities (45 J, 22 fs, 50 MeV proton cutoff, 7Be-derived Li neutron yield) rest on direct measurements and are credible. The weakest link is the unmeasured temporal contrast, which the paper itself acknowledges in Section II. The simulation chain uses an assumed prepulse and an intensity tuned to the proton data, so the derived electron spectrum, X-ray spectrum, and 8 PW projections are underdetermined: different prepulse structures combined with different fitted intensities could match the same proton spectrum. This is a genuine correctness risk for the paper's interpretive claims, though not for the measured beam and source performance. A secondary, related issue is that the total neutron yield quoted with a 6% error bar depends on a simulated 7Li-to-19F branching ratio computed from an inferred proton spectrum; the measured 7Be activity alone fixes only the Li branch. The reader focused on the temporal contrast; I agree with that focus but additionally flag its propagation to the neutron total. The proposed full-energy contrast measurement would settle whether the assumed prepulse is realistic, and the rerun without intensity tuning would show how much the inferred quantities depend on the assumption. Until then, the conditional verdict, with the derived physics treated as provisional, is the right call.","tokens_in":19746,"tokens_out":16033,"duration_ms":169330,"concrete_test":"Measure the full-energy temporal contrast at the Apollon F1 focus with a high-dynamic-range third-order cross-correlator covering the nanosecond to sub-picosecond prepulse range. Feed the measured prepulse, rather than the assumed 2 ps/1e15 W/cm2 prepulse, into the MULTI preplasma calculation and rerun the SMILEI simulations without re-tuning the intensity to match the proton spectrum. If the inferred hot-electron temperatures, X-ray spectrum, and predicted 7Li branch fraction remain within the quoted uncertainties, the concern is resolved; if they shift substantially, the simulation-based characterization and the 4.10e8 n/shot total need revised error bars or explicit qualification.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's direct measurements (on-target energy, Wizzler duration, RCF proton cutoffs, 7Be activation) support the core commissioning claims. The load-bearing weakness is the assumed temporal contrast. Section II states that contrast was only measured on the low-energy 10 Hz beam, not at full energy; Section IV then initializes the MULTI/SMILEI chain with an assumed 2 ps, 1e15 W/cm2 prepulse located 100 ps before the main pulse, and the PIC intensity is adjusted to reproduce the measured proton spectrum. The reflected-beam pattern (Fig. 5) and the 6-8 um optimum thickness are qualitative evidence for a preplasma, but they cannot determine the prepulse intensity, duration, or timing. Since the preplasma profile and the laser intensity are both effectively free parameters while only the proton spectrum is used as the constraint, the inferred electron spectrum (Fig. 10), the X-ray spectrum (Fig. 17), and the 8 PW projections (Fig. 20) are non-unique. This also propagates to the quoted 4.10e8 n/shot total, because the 76.7% 7Li branch fraction used to convert the measured 7Be activity into total neutron yield comes from a Geant4 simulation driven by the neutron-inferred proton spectrum, not by a directly measured one. The direct Li yield (3.14e8/shot) is robust; the extrapolated total is not.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper reports the second commissioning phase of the Apollon F1 beamline in the short-focus area at the 2 PW level. It presents direct measurements of the delivered beam (up to 45 J on target, ~22 fs pulse duration, focal spot of about 2.3 x 2.5 um, Strehl ratio ~51%), plus experimental characterization of plasma conditions, proton acceleration from Al foils, neutron production from a LiF pitcher-catcher, X-ray generation and radiography, and simultaneous operation with the F2 beamline. Simulation results with MULTI, SMILEI, and Geant4 are used to interpret the data and to project performance at 8 PW. The paper claims that the experimental campaign demonstrates stable ~50 MeV proton cutoffs, ~3-4 x 10^8 neutrons per shot, and good laser-to-target coupling.","tokens_in":20012,"tokens_out":5559,"duration_ms":55954,"significance":"If the direct measurements are taken at face value, this is a valuable commissioning dataset for a multi-PW user facility: the on-target beam parameters, RCF proton spectra, activation-based neutron yields, X-ray dose maps, and the F1-F2 dual-beam demonstration are concrete and useful to future users. The paper is weaker where it promotes simulation-inferred quantities (electron temperatures, X-ray spectra, total neutron yield, 8 PW projections) to the same level of confidence as the direct measurements. The underlying issue is that the full-energy temporal contrast was not measured, and the simulation chain is constrained only by the proton spectrum. The direct experimental results are sound; the model-dependent secondary claims need to be relabeled and, where possible, bracketed by sensitivity studies.","major_comments":[{"comment":"Section II states that the temporal contrast was not fully characterized at full energy and was only measured on the low-energy 10 Hz beam. Section IV then initializes the MULTI/SMILEI chain with an assumed 2 ps, 10^15 W/cm2 prepulse located 100 ps before the main pulse, and the SMILEI laser intensity is adjusted to match the measured proton spectrum (Fig. 9). Because neither the prepulse parameters nor the in-target intensity are independently constrained, the simulated electron spectrum (Fig. 10), the X-ray spectrum (Fig. 17), and the 8 PW projections (Fig. 20) are not unique; agreement with the proton spectrum cannot validate these derived quantities. The reflected-beam pattern (Fig. 5) and the 6-8 um optimum thickness are qualitative evidence for a preplasma, but they do not determine the prepulse level, duration, or timing. These simulation-based results should be presented as model-dependent estimates, ideally with a sensitivity scan over the assumed contrast, or the full-energy contrast should be measured.","section":"Secs. II and IV"},{"comment":"The quoted total neutron yield, (4.10 +/- 0.26) x 10^8 neutrons/shot, is obtained by dividing the directly measured 7Be activity by a 76.7% contribution of the 7Li(p,n)7Be reaction, and that fraction comes from a Geant4 simulation driven by a 'proton spectrum guess' with a mean energy of 4.25 MeV (green dashed curve in Fig. 9), not by the RCF-measured spectrum. The directly measured Li-based yield, (3.14 +/- 0.20) x 10^8 neutrons/shot, is robust, but the extrapolated total is model-dependent and its stated uncertainty does not include the uncertainty in the proton spectrum. This propagation should be quantified, and the model dependence should be stated wherever the total yield is quoted, including the abstract and conclusions.","section":"Sec. V.A"},{"comment":"The electron temperatures quoted in Fig. 10 (5.3 MeV below 25 MeV, 20 MeV above) are extracted from the same adjusted PIC simulation, not from a direct measurement; the raw electron spectrometer images in Fig. 11(d)-(e) are not quantitatively compared with the simulated spectrum. Consequently, the claimed agreement with the (I0 lambda^2)^(1/3) and ponderomotive scalings rests on the assumed laser intensity and preplasma profile. The two-temperature description should be explicitly labeled as simulation-based, and calibrated electron spectra or a sensitivity analysis should be provided if the authors wish to retain this as a characterization result.","section":"Sec. IV"},{"comment":"The X-ray energy spectrum in Fig. 17 is presented as part of the experimental characterization, but the RPL dosimetry in Fig. 16 constrains only the angular distribution of the dose integrated over photon energy, not the spectrum. The Bremsstrahlung spectrum computed from Eq. (2) depends on the electron distribution from the adjusted PIC run and on the refluxing efficiency eta_r, neither of which is independently measured. The total spectrum should therefore be clearly stated as a model prediction, and the qualitative source-size comparison from Fig. 18 should be described as a consistency check rather than a validation of the simulated spectrum.","section":"Sec. VI, Eq. (2)"}],"minor_comments":[{"comment":"The inline definition of a0 appears garbled as printed; the square root is missing or mis-rendered. Please write a0 = sqrt(I0 lambda^2 mu0 q_e^2 / (2 pi^2 m_e^2 c^3)) or an equivalent explicit expression.","section":"Sec. II"},{"comment":"Equation (1) is unnumbered and nshot is not defined in the text; please number the equation and define all symbols at first use, including the number of shots used in the activation series.","section":"Sec. V.A, Eq. (1)"},{"comment":"The caption says the simulated neutron spectra are 'obtained respectively from the experimental proton spectrum shown in Fig. 9 and the proton spectrum guess', but Fig. 9 contains three curves (PIC simulation, RCF-inferred spectrum, and the 4.25 MeV guess). Please specify which curves are being referenced to avoid ambiguity.","section":"Fig. 13 caption"},{"comment":"The statement that 'the intensities used in the SMILEI simulations are lower than the experimental ones' is not quantified; giving the simulation intensity value would allow readers to judge the size of the adjustment.","section":"Sec. IV"},{"comment":"The caption reports '<5% PtV energy fluctuation' but the abbreviation PtV is not defined; please spell out 'peak-to-valley' and state the number of shots used for the 5-hour stability window.","section":"Fig. 2(b)"}],"recommendation":"major_revision","confidential_remarks":"The paper is a legitimate commissioning report whose direct measurements are well suited to the journal. My main concern is that the simulation-derived secondary quantities and the total neutron yield are given equal weight to direct measurements despite relying on an unmeasured temporal contrast and an adjusted PIC intensity. I would support publication after the authors reframe the simulated spectra and projections as model-dependent, add sensitivity caveats around the assumed prepulse, and soften or bracket the total neutron yield claim. A table separating directly measured quantities from simulation-inferred ones would substantially improve the manuscript."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a solid commissioning report, and the 2 PW beam parameters are the real news. Everything that is directly measured—on-target energy, Wizzler duration, focal spot stability, RCF proton cutoffs, activation-based neutron yields—is credible and internally consistent. The 50 MeV proton cutoff from 6–8 µm Al foils and the E^0.5 scaling check out. The dual-beam F1/F2 demonstration is new and useful for users. I'd send this to a serious referee.\n\nThe soft spots are where the paper moves from beam characterization to plasma physics. Temporal contrast at full energy was not measured; Section II says so plainly. The MULTI/SMILEI chain then assumes a 2 ps, 10^15 W/cm2 prepulse at 100 ps, and the PIC intensity is adjusted to match the measured proton spectrum. Using that same simulation to produce the electron spectrum (Fig. 10), the X-ray spectra (Fig. 17), and the 8 PW projections (Fig. 20) is partly circular, because the prepulse profile and laser intensity are both free parameters and only the proton spectrum constrains them. The reflected-beam pattern and the 6–8 µm optimum thickness are qualitative evidence for a preplasma, but they don't pin down the prepulse intensity, duration, or timing. So the inferred electron temperatures (5.3 and 20 MeV) and the X-ray estimates should be labeled as model-dependent, not measured.\n\nThe neutron section has a similar issue but one step removed. The direct 7Be measurement gives (3.14 ± 0.20)×10^8 neutrons/shot from Li—that's solid. The quoted total of (4.10 ± 0.26)×10^8 depends on the 76.7% branch fraction from Geant4, which is driven by an inferred proton spectrum, not a directly measured one. The activation foils and bubble detectors are consistent with the simulated fast-neutron spectrum, so this is not a fatal flaw; it's just a reminder that the total has simulation in it.\n\nThe paper is honest about these limitations, and the direct measurements stand on their own. I'd recommend peer review with a request to either measure the contrast at full energy or tone down the simulation-derived claims. The code and data are on request only; shipping them would help, but for a facility paper that's not a blocker. This is a paper for laser-plasma experimentalists planning campaigns at multi-PW facilities, and for facility managers tracking worldwide capabilities. It deserves a proper referee.","headline":"Solid Apollon 2 PW commissioning data; the simulation-derived electron and X-ray spectra rest on an unmeasured contrast and adjusted PIC intensity, so treat those as provisional.","tokens_in":20829,"tokens_out":2110,"would_cite":true,"duration_ms":21274,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["52.38.-r","52.38.Kd"],"model":"deepseek-v4-flash","headline":"Apollon's 2 PW beam delivered ~50 MeV protons, ~4e8 neutrons per shot, and X-ray radiography-quality emission under commissioning conditions.","keywords":["petawatt laser","laser commissioning","target normal sheath acceleration","laser-driven proton acceleration","neutron generation","X-ray radiography","dual-beam operation","laser temporal contrast"],"falsifier":"A full-energy temporal-contrast measurement of the F1 beam (for example with a third-order cross-correlator covering the 100 ps window) would directly test the assumed 2 ps, $10^{15}$ W/cm$^2$ prepulse; if no such pedestal is present, the MULTI/SMILEI preplasma initialization is wrong and the simulation-based consistency between proton, electron, X-ray, and neutron spectra collapses. Alternatively, one high-contrast shot on a target about 2 µm thick that shows a proton cutoff well above 50 MeV and a conversion efficiency above 0.9% would confirm that the reported performance is contrast-limited.","tokens_in":19525,"feed_emoji":"⚡","tokens_out":16795,"duration_ms":142274,"temperature":0.7,"pith_summary":"At its second commissioning stage, the Apollon main beam (F1) was run at about 2 PW and found to deliver on-target pulses of up to 45 J in 22 fs, focused to a roughly 2.5 µm spot at an intensity near $1.8\\times10^{22}$ W/cm$^2$. The paper reports that these pulses produced stable proton beams with cutoff energies around 50 MeV from 6–8 µm aluminum foils, about $(4.10\\pm0.26)\\times10^8$ neutrons per shot from a lithium-fluoride converter, and X-ray doses about ten times higher than the facility's 1 PW beam for four times the laser energy, enough for point-projection radiography. It also demonstrates simultaneous operation of the main beam with the auxiliary 0.5 PW beam, with adjustable delay and no intrinsic jitter. The authors offer these measurements as a characterization of the beam for user experiments and as a baseline for the planned 8 PW operation in 2025.","feed_headline":"2-PW Apollon beam makes 50 MeV protons and 4e8 neutrons","feed_subtitle":"Commissioning data set user expectations for proton, neutron, and X-ray sources at 2 PW.","key_machinery":"The chain that carries the argument is the laser-to-secondary-source pipeline: a 22 fs, 45 J pulse is focused by a 1 m focal-length off-axis parabola (f/2.5) to a 2.3–2.5 µm focal spot, reaching $1.8\\times10^{22}$ W/cm$^2$ and $a_0\\approx90$, and irradiates a thin foil at 45° incidence; the sheath field at the target rear then accelerates protons. The interpretive engine is the simulation sequence: the MULTI code generates a preplasma profile from an assumed 2 ps, $10^{15}$ W/cm$^2$ prepulse 100 ps ahead of the main pulse, SMILEI PIC (with intensity adjusted downward to compensate for 2D overestimation) reproduces the measured proton spectrum and supplies the electron spectra and electromagnetic radiation, and Geant4 transports neutrons from the LiF converter to the activation, bubble, and time-of-flight detectors. The preplasma assumption is the load-bearing component: it explains the 6–8 µm optimum target thickness, the roughly 15° shift of the high-energy proton beam away from the target normal, and the lower conversion efficiency compared with high-contrast shots.","core_discovery":"The central claim, stated on the paper's own terms, is that Apollon's F1 beam at 2 PW is a functioning, user-relevant laser driver: on-target pulses of maximum energy 45 J and 22 fs duration, focused to a slightly elliptical spot of 2.3 $\\times$ 2.5 µm FWHM with about 44% of the energy in the first lobe, generate target-normal sheath accelerated proton beams with a stable ~50 MeV cutoff from 6–8 µm Al foils, and through a LiF pitcher-catcher produce $(4.10\\pm0.26)\\times10^8$ neutrons per shot. The X-ray emission, dominated by bremsstrahlung in the simulations, reached average doses around 15.5 mGy/shot at 87 cm and supported point-projection radiography with a source size much smaller than 125 µm. The authors match the measured proton spectrum with 2D PIC simulations in which the target starts with a preplasma from a 2 ps prepulse at $10^{15}$ W/cm$^2$ located 100 ps before the main pulse; the fitted electron spectrum is two-temperature (5.3 MeV and 20 MeV), and Geant4 simulations using the inferred proton spectrum reproduce the measured neutron yields and the activation of the indium and magnesium foils. The paper also demonstrates that F1 and F2 can operate simultaneously, with F2 providing proton radiography of F1-generated plasmas.","pith_inferences":["If the assumed 2 ps, $10^{15}$ W/cm$^2$ prepulse is the real limiter, then improving the full-energy temporal contrast should shift the optimum target thickness below 6 µm and raise conversion efficiency above 0.9%; this testable prediction follows from the paper's simulation setup rather than being stated by the authors.","The 8 PW predictions inherit the calibration of the 2 PW simulation chain, which rests on an unmeasured prepulse; if contrast is improved before 2025, the predicted 90 MeV proton cutoff may be conservative, since high-contrast shots typically couple more efficiently.","The roughly ten-fold increase in X-ray dose for a four-fold increase in laser energy suggests X-ray yield scales superlinearly with pulse energy at fixed contrast, which would make radiation backgrounds and shielding an increasingly important constraint at 8 PW.","Because the dual-beam demonstration used a single shared deformable mirror, F2's focal spot and pulse duration were not optimized; until a second mirror is installed, simultaneous-operation users should expect a degraded auxiliary beam."],"forward_implications":["Users can plan around stable $\\sim$50 MeV proton beams from 6–8 µm aluminum foils, with a laser-to-proton conversion efficiency of 0.7–0.9% under the tested conditions.","The LiF converter delivers about $4.1\\times10^8$ neutrons per shot, with a measured 2–10 MeV fluence near $2\\times10^7$ neutrons/sr/shot, making neutron experiments feasible at the 2 PW level.","X-ray doses near 15 mGy/shot at 87 cm and a sub-125 µm source size make point-projection radiography a practical diagnostic on F1 shots.","F1 and F2 can be fired simultaneously with adjustable delay and no intrinsic jitter, enabling dual-beam experiments such as proton radiography of F1-driven plasmas.","Extrapolating the same simulation chain to 8 PW predicts roughly a doubling of the proton cutoff energy (to about 90 MeV), an order-of-magnitude increase in bremsstrahlung with tungsten targets, and neutron energies about three times higher."],"supporting_citations":[{"why":"Prior 1 PW commissioning of the same short-focus area; it supplies the baseline beam parameters, the reflected-beam pattern comparison, and the X-ray spectra used to infer poor contrast.","marker":"[6]"},{"why":"Previous Apollon campaign with the 0.5 PW F2 beam on thin foils; it provides the high-contrast comparison, the 25 MeV proton baseline, and the conversion-efficiency reference.","marker":"[13]"},{"why":"Laser-driven proton scaling law (cutoff growing as the square root of pulse energy) used to place the 50 MeV result on the expected trend.","marker":"[19]"},{"why":"The SMILEI particle-in-cell code used for the proton spectra, electron spectra, electric-field radiation, and angular distributions.","marker":"[21]"},{"why":"Radiative-hydrodynamic calculation with MULTI that supplies the preplasma profile initialized in the SMILEI simulation.","marker":"[22]"},{"why":"Companion characterization of the Apollon neutron fields; it supplies the unfolding procedure, activation cross sections, and detector methods reused here.","marker":"[11]"},{"why":"The pitcher-catcher scheme with a lithium-fluoride converter and the 7Li(p,n)7Be yield characterization on which neutron production is based.","marker":"[29]"},{"why":"Geant4 simulations used to predict neutron spectra, angular distributions, and foil activation from the inferred proton spectrum.","marker":"[33]"},{"why":"MCNP6 code and IRDFF-II cross-section library used to compute the simulated foil activities compared with measurement in Table I.","marker":"[35]"}],"fun_headline_variants":["Apollon 2 PW beam yields 50 MeV protons and 4e8 neutrons","45 J in 22 fs: Apollon hits 2 PW with user-ready sources","Apollon's 2 PW beam: stable protons and neutrons for experiments","Commissioning shows 2 PW Apollon beam, dual-beam capability","Dual-beam debut: Apollon F1 at 2 PW with proton and neutron sources"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The full-performance chain depends on the unmeasured temporal contrast of the full-energy pulse: the simulations assume a 2 ps prepulse at $10^{15}$ W/cm$^2$ arriving 100 ps before the main pulse, and if the real prepulse differs, the fitted laser intensity and the inferred preplasma conditions lose support, invalidating the consistency between measured and simulated proton, X-ray, and neutron spectra.","fun_headline_variants_meta":{"raw":{"variants":["Apollon 2 PW beam yields 50 MeV protons and 4e8 neutrons","45 J in 22 fs: Apollon hits 2 PW with user-ready sources","Apollon's 2 PW beam: stable protons and neutrons for experiments","Commissioning shows 2 PW Apollon beam, dual-beam capability","Dual-beam debut: Apollon F1 at 2 PW with proton and neutron sources"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.0002,"raw_usage":{"total_tokens":1433,"prompt_tokens":1061,"completion_tokens":372,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":677,"completion_tokens_details":{"reasoning_tokens":265}},"tokens_in":677,"tokens_out":372,"duration_ms":4268,"temperature":1.0,"reasoning_tokens":265,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T17:05:11.682234+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A full-energy temporal-contrast measurement of the F1 beam (for example with a third-order cross-correlator covering the 100 ps window) would directly test the assumed 2 ps, $10^{15}$ W/cm$^2$ prepulse; if no such pedestal is present, the MULTI/SMILEI preplasma initialization is wrong and the simulation-based consistency between proton, electron, X-ray, and neutron spectra collapses. Alternatively, one high-contrast shot on a target about 2 µm thick that shows a proton cutoff well above 50 MeV and a conversion efficiency above 0.9% would confirm that the reported performance is contrast-limited.","supporting_citations":[{"cited_title":"\\ Zou , author C","cited_arxiv_id":null,"evidence_quote":"Prior 1 PW commissioning of the same short-focus area; it supplies the baseline beam parameters, the reflected-beam pattern comparison, and the X-ray spectra used to infer poor contrast."},{"cited_title":"Lahaye , author K","cited_arxiv_id":null,"evidence_quote":"Previous Apollon campaign with the 0.5 PW F2 beam on thin foils; it provides the high-contrast comparison, the 25 MeV proton baseline, and the conversion-efficiency reference."},{"cited_title":"Chen , author M","cited_arxiv_id":null,"evidence_quote":"Laser-driven proton scaling law (cutoff growing as the square root of pulse energy) used to place the 50 MeV result on the expected trend."},{"cited_title":"Bolton , author M","cited_arxiv_id":null,"evidence_quote":"The SMILEI particle-in-cell code used for the proton spectra, electron spectra, electric-field radiation, and angular distributions."},{"cited_title":"Derouillat , author A","cited_arxiv_id":null,"evidence_quote":"Radiative-hydrodynamic calculation with MULTI that supplies the preplasma profile initialized in the SMILEI simulation."},{"cited_title":"Moulanier , author L","cited_arxiv_id":null,"evidence_quote":"Companion characterization of the Apollon neutron fields; it supplies the unfolding procedure, activation cross sections, and detector methods reused here."},{"cited_title":"Burdonov , author R","cited_arxiv_id":null,"evidence_quote":"The pitcher-catcher scheme with a lithium-fluoride converter and the 7Li(p,n)7Be yield characterization on which neutron production is based."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Geant4 simulations used to predict neutron spectra, angular distributions, and foil activation from the inferred proton spectrum."}],"review_version":1}