{"id":"d9886c3d-a1c6-46bc-b73d-e144e88e6bc0","arxiv_id":"2607.18420","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Electron escape probability in NEA photocathodes is read out directly from surface-photovoltage shifts caused by reverse-injecting emitted electrons back into the photocathode.","lead":"The authors demonstrate a way to measure how often electrons generated inside a photocathode escape into vacuum: they briefly reverse the electric field, send the emitted electrons back into the material, and read the change in surface voltage. If it works broadly, this gives photocathode developers a fast, direct diagnostic that does not require absorption or transport data.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Eq. 11 is only valid if reverse-injected electrons recombine at the surface exactly like photoelectrons that fail to escape; the paper verifies that they return (Fig. 3) but not that they do not enter the bulk or excite secondaries, which would bias Pe.","rationale":"The reader's weakest assumption accurately identifies the main soft spot: the microscopic equivalence between reverse-injected and photogenerated electrons in setting the surface recombination current. I agree that this is the most load-bearing concern because Eq. 11 is derived directly from it, and a quantitative error here would shift all reported Pe values. The paper's internal controls are strong but insufficient: Fig. 3's saturation of ΔQEri with retarding voltage rules out electrons missing the illuminated region and correlates the QE change with reverse injection, but it cannot distinguish surface recombination from bulk entry or secondary-carrier generation. The same applies to the observation that ΔQEri tracks Jri — it shows causation, not the specific mechanism. Other potential issues, such as field-effect changes to the surface potential or nonlinearity of QE versus V, are partially controlled by the data (e.g., ΔQEri goes to zero at the stopping voltage, and QE and UCPD show parallel logarithmic dependences) and are smaller than the unresolved microscopic assumption. Data availability is a reproduction concern, not a correctness concern. The proposed optical-power-dependence test is concrete and uses the existing apparatus; it would reveal whether the assumed Js = Jb relation holds over a range where the high-power approximation is not used, thereby settling the concern. Since the reader's verdict is already conditional on this unresolved assumption, my assessment does not change the verdict.","tokens_in":7205,"tokens_out":18113,"duration_ms":167082,"concrete_test":"Measure ΔQEri as a function of optical power I over at least two decades (e.g., 5–500 nW/cm²) using the same retarding-field soak protocol. From QE(I) and UCPD(I), determine AE0 and the ratio Jb/J0(I). Compare the data to the full (non-approximated) prediction ΔQEri = −A E0 ln[(1 + Jb/J0)/(1 + (1−Pe)Jb/J0)] with Pe as a free parameter. If the data follow this curve over the whole range — including the predicted crossover from linear (low power) to constant (high power) ΔQEri — the reverse-injection/surface-recombination equivalence is supported. A systematic deviation would demonstrate that reverse-injected electrons do not simply add to Js, and Eq. 11 is biased.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central derivation (Eqs. 4, 5, 8, 11) requires that, in the retarding field, the surface recombination current Js rises to exactly Jb, i.e., every reverse-injected photoelectron is captured at the surface and recombines there just like a photoelectron that failed to escape from the bulk. The paper asserts this ('This leads to the increase in Js and SPV toward their maximum values of Jb and Vret') but never verifies it. Physically, a returning electron arrives with kinetic energy from the negative-electron-affinity emission process and may traverse the band-bending region into the quasi-neutral bulk, generate secondary electron-hole pairs, or alter the occupancy of the surface states. If any of these channels is non-negligible, the true Vret differs from E0 ln(1 + Jb/J0), and Eq. 11 gives a biased Pe. The experimental evidence in Fig. 3 — |ΔQEri| following the reverse-injected current Jri and saturating with retarding field — demonstrates only that the QE change is caused by reverse injection and that the emitted electrons return to the illuminated region; it does not establish where those electrons end up. Since the extracted Pe values (27–32%) are then used as consistency checks, the equivalence remains an untested assumption at the heart of the method.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a direct method for measuring the electron escape probability Pe in NEA photocathodes by exploiting the surface photovoltage change induced by reverse injection of emitted electrons under a retarding field. From a three-step photoemission model and a diode-like surface recombination current, the authors derive expressions for Pe in terms of photovoltages (Eqs. 6 and 8) and, using a linear QE–SPV relation, in terms of the reverse-injection-induced QE decrease (Eq. 11). The method is demonstrated on p-GaN(Cs,O) photocathodes, yielding Pe(I*) = (27 ± 3)% and Pe,0 = (32 ± 3)% for one device and 30–60% across devices, consistent with literature.","tokens_in":7596,"tokens_out":6321,"duration_ms":54587,"significance":"If correct, the method provides a simple, direct route to Pe without requiring separate knowledge of absorption and transport parameters, which would be valuable for photocathode optimization. The derivation is internally consistent, signs are correct, and the experimental data include error bars and consistency checks: the QE decrease saturates with retarding field and follows the reverse-injected current (Fig. 3). The main strength is that Pe is not a fitted parameter; E0 and AE0 are calibrated from the same sample in the accelerating regime without using Pe. However, the method relies on an untested microscopic assumption about the fate of reverse-injected electrons, and this limits the certainty of the quantitative results.","major_comments":[{"comment":"The derivation of Vret assumes Js = Jb in the retarding field, i.e., every reverse-injected photoelectron is captured at the surface and recombines there exactly like a photoelectron that failed to escape. The paper verifies that the QE decrease saturates with the retarding field and tracks Jri (Fig. 3), which demonstrates that the effect is caused by returning electrons, but it does not establish that these electrons recombine at the surface rather than entering the quasi-neutral bulk or exciting secondary carriers. If a non-negligible fraction enters the bulk, the photovoltage is not described by Eq. (5), and Eq. (11) gives a biased Pe. This is the central load-bearing assumption of the method and should be justified by an energy/length-scale argument or tested directly (e.g., by comparing the retarding-field photovoltage Vret with E0 ln(1+Jb/J0) from independent measurements).","section":"Eqs. (4)–(5) and Fig. 3"}],"minor_comments":[{"comment":"The axis labels 'const − ln(AE0 I)' and 'const − ln(E0 I)' are ambiguous. They should read 'const − AE0 ln I' and 'const − E0 ln I' (or equivalent) to show the slope correctly.","section":"Fig. 2(c)"},{"comment":"Please specify how ΔQEri is obtained from the kinetics—whether it is the initial photocurrent immediately after switching to the accelerating field, extrapolated to zero time, or an average over a short interval. This is important for reproducibility.","section":"Fig. 2(d) description"},{"comment":"The conversion Pe,0 = Pe × QE0/QE(I*) assumes that Pab and Ptr are independent of surface photovoltage. This should be stated explicitly, since Ptr could in principle depend on band bending.","section":"Paragraph after Eq. (11)"},{"comment":"The paper would benefit from a brief discussion of the validity of Eq. (9) (linear QE–SPV relation) over the relevant voltage range, as deviations could affect the exponential in Eq. (11).","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The paper is carefully done and the extraction of Pe from Eq. (11) is algebraically sound. The main issue is that the method's central physical assumption—that reverse-injected electrons recombine at the surface exactly like unsuccessfully escaping photoelectrons—is not verified. If the authors can provide a convincing argument or experimental test (e.g., direct measurement of Vret or a comparison with an independent Pe determination), the manuscript would be suitable for publication. The presentation is otherwise clear and the data include proper error bars."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"New here is a direct route to the electron escape probability Pe in NEA photocathodes. Instead of fitting QE spectra with absorption and transport models, the authors measure the SPV change when emitted electrons are pushed back to the surface by a retarding field. The algebra from Eqs. 4–11 is clean, signs check, and on p-GaN(Cs,O) they get Pe around 30% at low power and 27% at a higher power, consistent with the reported range. That is a genuine methodological step for a practical diagnostic, and the paper is honest about its scope.\n\nThe strongest part is the internal consistency: the log-slope of QE vs I gives AE0, the measurement of ΔQEri after field reversal is shown to saturate with retarding field and to track the reverse-injected current Jri. Those checks provide real evidence that the QE change is caused by the reverse injection.\n\nThe main soft spot is the assumption that reverse-injected electrons at the surface behave exactly like photoelectrons that failed to escape—that they are captured in the band bending region and recombine at the surface, so Js rises to Jb. The paper does not directly verify that the returned electrons stay at the surface rather than traversing into the bulk or creating secondary pairs. If that equivalence fails, Eq. 11 gives a biased Pe. It is not a fatal objection: the assumption is physically plausible and matches the usual picture of NEA surface recombination, and the agreement with prior Pe values is reassuring. But the authors should either justify it more strongly or present an independent cross-check.\n\nA secondary point: Pe is treated as constant in Eq. 4 though it depends on V; they handle this by reporting Pe at a fixed I and then Pe,0, so it is not a serious error, but the exposition could be crisper.\n\nCitation practice is fine. The self-cited works provide the SPV and QE formalism; they do not supply the target result. Data are only available on request, which is a minor reproducibility limitation.\n\nOverall, this is a competent techniques paper with a clearly stated and testable central assumption. It deserves a serious referee. I would ask the authors to directly address the fate of reverse-injected electrons, and to consider depositing the data.","headline":"A clean new technique for extracting the electron escape probability from reverse-injection SPV, with one central assumption—that returned electrons recombine at the surface—that is plausible but not directly verified.","tokens_in":8058,"tokens_out":6961,"would_cite":true,"duration_ms":59073,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Electron escape probability in photocathodes can be read directly from reverse-injection photovoltage, without needing absorption or transport data.","keywords":["electron escape probability","surface photovoltage","negative electron affinity","photocathode","p-GaN(Cs,O)","reverse injection","quantum efficiency","photoemission"],"falsifier":"Measure Pe on the same p-GaN(Cs,O) cathode by an independent method—for example, from the spectral dependence of QE with independently known absorption coefficient and diffusion length—and compare with the reverse-injection value. A systematic discrepancy under varying retarding-field strength or illumination intensity would indicate the injected electrons do not simply join the surface recombination current. Alternatively, check whether ΔQEri is independent of the energy of the returned electrons by changing the retarding voltage within the saturated range.","tokens_in":7175,"feed_emoji":"⚡","tokens_out":3600,"duration_ms":28532,"temperature":0.7,"pith_summary":"This paper proposes a direct way to measure the probability Pe that an electron reaching the emitting surface of a negative-electron-affinity photocathode escapes into vacuum. The method uses the surface photovoltage change that occurs when photoemitted electrons are driven back into the cathode by reversing the external field. From the measured drop in quantum efficiency and the slope of efficiency versus log optical power, Pe follows from a single formula, Eq. 11, without needing absorption or bulk-transport data. Demonstrated on p-GaN(Cs,O), it gives Pe = 27% at a given illumination and 32% in the low-power limit, with 30–60% across samples. A sympathetic reader cares because Pe is often the limiting factor in photocathode efficiency and previously required poorly known parameters to extract.","feed_headline":"A field reversal exposes electron escape probability","feed_subtitle":"Direct measurement needs no absorption or transport data; p-GaN(Cs,O) yields 27–60% escape probability.","key_machinery":"Surface photovoltage (SPV) in a p-type NEA photocathode: under illumination, electrons that do not escape recombine at the surface with a current Js, balanced by a restoring hole current Jr(V) ≈ J0[exp(V/E0) − 1]. Reversing the field sends the emitted electrons back, raising Js to Jb and the photovoltage to Vret; the escape probability is the fraction that changes the balance between the two field directions. The measured quantity is the quantum-efficiency drop ΔQEri, linked to ΔVri by the proportionality QE ≈ QE0 − A V, so that the product AE0, obtained from the QE-vs-ln(I) slope, converts the drop into Pe.","core_discovery":"On its own terms, the paper establishes that the escape probability can be obtained from the difference between the surface photovoltage in accelerating and retarding fields. Because reverse-injected electrons add to the surface recombination current, the retarding-field photovoltage is Vret = E0 ln(1 + Jb/J0), while the accelerating-field value is Vacc = E0 ln(1 + (1 − Pe)Jb/J0). In the high-power limit the two differ by ΔVri, and since quantum efficiency drops linearly with photovoltage, Pe ≈ 1 − exp[ΔQEri/(AE0)]. The authors verify that the measured QE decrease tracks the reverse-injected current and saturates when all electrons return to the illuminated spot, then report Pe values for p-","pith_inferences":["The method implicitly assumes the injected electrons relax to the same surface state distribution as photoelectrons; a direct test would be to compare Pe from this technique with an independent measurement, such as fitting QE spectra with known absorption and diffusion length.","Because the measurement is fast and non-destructive, it could be used in situ during activation to monitor how Pe evolves as Cs and O2 are deposited.","The same photovoltage-balance argument may extend to secondary-electron emitters or cold-electron sources, where escape probability across a surface barrier is also limiting.","The observed range 30–60% across similar cathodes suggests surface morphology or activation stoichiometry, not bulk properties, controls Pe; this could be tested by correlating Pe with atomic-force microscopy or XPS on the same cathodes."],"forward_implications":["A direct measurement of Pe without absorption or transport data, removing the main source of uncertainty in three-step photoemission models.","The method can be implemented in a sealed vacuum photodiode or larger vacuum system using only photocurrent kinetics and a QE-versus-power calibration.","For p-GaN(Cs,O), measured Pe values of 30–60% set a quantitative target for optimizing the (Cs,O) activation layer and surface preparation.","The low-power formula Pe ≈ 1 − Vacc/Vret gives a check at low intensity, while Eq. 11 covers the practical high-power regime.","If the proportionality between QE and photovoltage holds, the same procedure applies to other NEA photocathodes such as GaAs and InGaAs."],"fun_headline_variants":["Reverse field measures photocathode electron escape probability","Direct escape probability from reversed-field photovoltage","Photovoltage reversal yields electron escape probability directly","No absorption data needed: field reversal gives escape probability","Reverse-injected photovoltage directly reads electron escape"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The central premise is that electrons injected back into the photocathode contribute to the surface recombination current exactly as photoelectrons that fail to escape; if they instead pass into the bulk, excite secondary carriers, or modify the surface charge, the photovoltage change would not equal E0 ln(Jb/J0) and Pe would be biased.","fun_headline_variants_meta":{"raw":{"variants":["Reverse field measures photocathode electron escape probability","Direct escape probability from reversed-field photovoltage","Photovoltage reversal yields electron escape probability directly","No absorption data needed: field reversal gives escape probability","Reverse-injected photovoltage directly reads electron escape"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000544,"raw_usage":{"total_tokens":2400,"prompt_tokens":664,"completion_tokens":1736,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":408,"completion_tokens_details":{"reasoning_tokens":1660}},"tokens_in":408,"tokens_out":1736,"duration_ms":13426,"temperature":1.0,"reasoning_tokens":1660,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T15:27:59.973081+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure Pe on the same p-GaN(Cs,O) cathode by an independent method—for example, from the spectral dependence of QE with independently known absorption coefficient and diffusion length—and compare with the reverse-injection value. A systematic discrepancy under varying retarding-field strength or illumination intensity would indicate the injected electrons do not simply join the surface recombination current. Alternatively, check whether ΔQEri is independent of the energy of the returned electrons by changing the retarding voltage within the saturated range.","supporting_citations":[],"review_version":1}