{"id":"e2ef7be0-d5c5-49e3-a73e-9e7f02a96967","arxiv_id":"2509.08733","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Future e+e- colliders can probe the axion-photon-dark photon coupling down to about 10^-4 GeV^-1 for dark photon masses near 10 GeV via single-photon events with missing energy.","lead":"Future lepton colliders such as ILC, CEPC, and FCC-ee could detect a very weakly coupled interaction between axions, photons, and dark photons through a clean single-photon-plus-missing-energy signal. The projected sensitivity reaches a coupling of about 10^-4 GeV^-1 for dark photon masses around 10 GeV, and the dark photon mass could be measured from the edge in the recoil mass distribution.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Projected 1e-4 GeV^-1 reach rests on a statistics-only significance formula; at the quoted luminosities a few-percent background systematic shifts the 5-sigma contour by a factor of 3-7 in the coupling.","rationale":"The paper's main deliverable is a projected 5-sigma reach on g_a_gamma'_gamma; the abstract and conclusion both rest on that number. The significance is defined as N_s / sqrt(N_b) with no systematic term. At the design luminosities the selected background is large: after the cuts of Eqs. (7)-(8), the irreducible gamma nu nu background leaves N_b ~ 10^6 events at ILC and several times 10^6 at CEPC/FCC-ee, so the statistical error is only ~10^3 events. A 1% normalization uncertainty on this background is ~10^4 events and dominates the denominator; at 5% it is ~5x10^4 events. Because N_s scales as g^2, the 5-sigma contour moves upward by a factor of roughly 3-7 in g. Thus the advertised 'down to 10^-4 GeV^-1' is a statistics-only projection, and the paper provides no evidence that the required sub-percent background control is achievable. This is the most load-bearing concern because it directly invalidates the headline number if systematics are at the few-percent level, as is common in collider analyses. The reader's formal weakest assumption, Br(gamma' -> a gamma) ~ 1, is a stated and internally consistent choice in the epsilon << g m_gamma' region used for the projections; a moderate reduction of the branching ratio would shift g_min by only sqrt(2), whereas a few-percent systematic shifts it by several. The missing polarization study is a genuine inconsistency but affects an auxiliary claim in the abstract, not the unpolarized reach. A single re-evaluation with background nuisance parameters would settle whether the concern lands, so the appropriate verdict remains CONDITIONAL, in agreement with the reader.","tokens_in":7194,"tokens_out":26245,"duration_ms":578123,"concrete_test":"Recompute the Fig. 5 contours with a profile-likelihood significance Z = N_s / sqrt(N_b + (sigma_b N_b)^2 + (sigma_sig N_s)^2), using the authors' own MadGraph N_s and N_b values, taking sigma_b = 0%, 1%, 5% and sigma_sig = 2%, and keeping all cuts and luminosities identical. If the minimum g at m_gamma' = 20 GeV moves above 1e-4 GeV^-1 for CEPC or above 3e-4 GeV^-1 for ILC when sigma_b = 1%, the headline sensitivity is not stable and the paper must either include systematics or weaken the abstract.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claim is the 5-sigma sensitivity contour in Fig. 5, computed as Z = N_s / sqrt(N_b) with no systematic uncertainty term. After the cuts in Eqs. (7)-(8), the dominant irreducible background e+e- -> nu nu gamma leaves N_b ~ 10^6 events at ILC (L = 2 ab^-1) and several times 10^6 at CEPC/FCC-ee, so sqrt(N_b) ~ 10^3. A background normalization or shape uncertainty of only 1% contributes ~10^4 events to the error, an order of magnitude above the statistical error; at 5% it is ~5x10^4 events. Since the signal yield scales as g^2, the minimum probed coupling shifts upward by sqrt(1 + (sigma_sys N_b)^2 / N_b), roughly a factor of 3 at 1% systematics and 7 at 5%. The abstract's claim of sensitivity 'down to 10^-4 GeV^-1' is therefore not robust unless systematics are controlled below about 1%, and the paper neither provides such a control nor justifies its neglect. This is the weakest load-bearing point because it is the step that converts event counts into the advertised reach. By contrast, the branching-ratio assumption Br(gamma' -> a gamma) ~ 1 is explicitly stated and internally consistent in the chosen region epsilon << g m_gamma', and the missing polarization study is a completeness issue rather than a threat to the unpolarized sensitivity.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the effective interaction vertex a-γ-γ' described by a dimension-five operator plus kinetic mixing, focusing on single-photon plus missing-energy signals at LEP II and at future e+e- colliders (ILC, CEPC, FCC-ee). The authors derive cross sections and decay widths, extract a LEP II bound from published L3 single-photon data, simulate signal and background with MadGraph, propose cuts on recoil mass and photon polar angle, and project 5σ sensitivity contours in the (m_γ', g_{aγγ'}) plane. They also propose determining m_γ' from the sharp edge of the recoil-mass distribution. The stated central result is that future colliders can probe g_{aγγ'} down to about 10^-4 GeV^-1 for m_γ' in the O(10-100) GeV range, under the assumption that the dark photon decays dominantly to γ plus an invisible, very light axion.","tokens_in":7460,"tokens_out":9006,"duration_ms":87338,"significance":"If the projected reach is robust, the paper identifies a clean and potentially powerful search channel for a well-motivated but less-studied coupling, and it provides a useful LEP II constraint derived from published data with explicit assumptions. The recoil-mass edge method is a nice observable for mass reconstruction. However, the central quantitative claim depends on a statistics-only significance estimate and on a branching-ratio assumption that is not mapped onto the plotted parameter space; the polarization statement in the abstract is not supported by the body of the paper. With these points addressed, the work would be a solid phenomenological contribution.","major_comments":[{"comment":"The projected sensitivity is computed as Z = N_s / sqrt(N_b) with no systematic uncertainty term. After the cuts in Eqs. (7)-(8), the dominant e+e- -> nu nu gamma background leaves on the order of 10^6 events at the quoted luminosities, so sqrt(N_b) is about 10^3 while a 1% normalization uncertainty contributes about 10^4 events to the error budget. Because the signal yield scales as g^2, a 1-5% background systematic shifts the 5-sigma contour upward by roughly a factor of 3-7 in g. The paper neither provides a systematic budget nor explains why systematics are negligible. Since Fig. 5 and the abstract's 10^-4 GeV^-1 claim are the central results, this is a load-bearing issue that must be addressed, for example by including a systematic term in Z or by quoting contours for several assumed background uncertainties.","section":"Section III.B, significance formula and Fig. 5"},{"comment":"The abstract accompanying the manuscript states that longitudinal beam polarization at the ILC can enhance the signal significance by a factor of four and provides the strongest projected reach. No polarization analysis appears anywhere in Sections II-IV, and the abstract in the full text omits this statement. This is a discrepancy between the claimed result and the actual content. Either the polarization study must be added, or the abstract claim must be removed.","section":"Abstract (submitted version) and Section III"},{"comment":"The projected contours in Fig. 5 are computed under the assumption Br(γ' -> a γ) ≈ 1, realized for ε ≪ g_{aγγ'} m_γ'. However, Fig. 5 is plotted in the (m_γ', g_{aγγ'}) plane without specifying the value of ε used or the region of that plane in which the branching-ratio assumption actually holds. Since Br depends on the ratio of the widths in Eqs. (2) and (5), the same point in Fig. 5 can give significantly different signal yields depending on ε. The paper should state the benchmark choice of ε for the contours or superimpose the region where Br(γ' -> a γ) ≈ 1 is valid.","section":"Section II and Fig. 5"},{"comment":"The recoil-mass edge relation m_γ' = sqrt(s - (M_recoil^max)^2) follows for the three-body process e+e- -> a γ' (on shell) -> a a γ. The two-body off-shell s-channel contribution e+e- -> γ'* -> γ a, which the authors also list as a signal channel, produces a photon with essentially fixed energy and recoil mass near m_a, not an edge at sqrt(s - m_γ'^2). The text and Fig. 3 do not state whether the simulated recoil distribution includes both channels and, if so, which contribution dominates after the cuts. The mass-reconstruction claim needs this separation to be made explicit.","section":"Section III.B, Eq. (6) and Fig. 3"}],"minor_comments":[{"comment":"The caption lists m_γ' = 200 GeV while the text in Section III.B refers to m_γ' = 220 GeV; these should be made consistent.","section":"Fig. 3 caption"},{"comment":"The phrase 'progress (2) dominates' should read 'process (2) dominates'.","section":"Section III.A"},{"comment":"The signal and background efficiencies after cuts are quoted only in relative terms ('background reduced by ~65%', 'signal efficiency 65% to 80%'); reporting the absolute cut flow and the resulting N_s and N_b would make the significance estimate reproducible.","section":"Section III.B"},{"comment":"The condition for γ' -> a γ dominance is stated verbally ('if the numerical values for ε and g are of the same size'); a quantitative comparison of Γ(γ' -> a γ) with Γ(γ' -> l+l-) would help the reader identify the valid parameter region.","section":"Section II"}],"recommendation":"major_revision","confidential_remarks":"The abstract attached to the submission appears to differ from the abstract printed in the full text, with the former containing a polarization claim not present in the latter. The editor may wish to verify which version is intended. The main technical concern is the statistics-only significance estimate, which is the basis for the headline reach; this is fixable in revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a solid, straightforward pheno paper on the axion–photon–dark-photon coupling at e+e- colliders. It computes a LEP II constraint from published L3 single-photon data and projects ILC/CEPC/FCC-ee reach in the (m_gamma', g_a_gamma'_gamma) plane, plus a recoil-mass edge to extract the dark photon mass. The LEP bound and the future sensitivity contours are genuinely new numbers; the recoil-mass method is simple and useful. The paper is honest about its key assumption: it focuses on m_gamma' > 10 GeV with a very light axion, and Br(gamma' -> a gamma) ~ 1 is internally consistent for the chosen parameter region.\n\nThe biggest soft spot is the significance calculation. Eq. (6) uses Z = N_s / sqrt(N_b) with no systematic term. After cuts, the SM background is ~10^6 events at these luminosities, so pure Poisson error is ~10^3. A 1% background normalization uncertainty is an order of magnitude larger than the statistical error, and that shifts the 5-sigma reach on the coupling by roughly a factor of 3; 5% systematics pushes it to a factor of 7. The abstract's 'down to 10^-4 GeV^-1' is therefore not robust without demonstrated control of the background shape and normalization. This is load-bearing for the central quantitative claim, not a footnote.\n\nSecond, the abstract promises that longitudinal beam polarization at the ILC enhances significance by a factor of four, but I found no polarization analysis anywhere in the text. That is a clear internal inconsistency, and the authors should either add the analysis or drop the claim.\n\nMinor issues: the relation to ref. [23] is not spelled out, and ref. [4] is an 'in preparation' paper — fine as motivation, but should be flagged. The LEP derivation uses reasonable assumptions (80% efficiency, 1% theory uncertainty, CLs) and appears internally consistent.\n\nWho is this for? Dark-sector searchers at lepton colliders will want the LEP bound and the recoil-mass method, and the projected reach is a useful target. The paper deserves a serious referee, but only after the systematics treatment and the polarization claim are fixed. I would send it to review with a note that the significance calculation needs a systematic term or an explicit explanation of why it can be neglected.","headline":"Solid single-photon reach study for axion–dark-photon couplings at future e+e- colliders, but the advertised sensitivity ignores background systematics and the abstract promises a polarization analysis the text does not deliver.","tokens_in":8059,"tokens_out":3337,"would_cite":false,"duration_ms":27628,"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":"Future e+e- colliders could probe the axion-photon-dark photon coupling down to 10⁻⁴ GeV⁻¹ for dark photon masses around 10-230 GeV, using single-photon events with missing energy.","keywords":["axion","dark photon","single-photon plus missing energy","future e+e- colliders","ILC","CEPC","FCC-ee","recoil mass"],"falsifier":"Run the single-photon search at $\\sqrt{s}=250$ GeV with $2\\,\\mathrm{ab}^{-1}$ of data, require $|\\cos\\theta_\\gamma|<0.75$, and reject recoil masses within $[m_Z-2\\Gamma_Z,\\,m_Z+2\\Gamma_Z]$; if the observed event count matches the SM $\\nu\\bar\\nu\\gamma$ background within uncertainties, all $g_{a\\gamma'\\gamma}$ values on the paper's $5\\sigma$ contour are excluded. The claim fails if the background systematic uncertainty is larger than the roughly 1 percent assumed, so a measurement of the $\\nu\\bar\\nu\\gamma$ background rate at the 1 percent level is the decisive check.","tokens_in":6902,"feed_emoji":"⚛️","tokens_out":16306,"duration_ms":116223,"temperature":0.7,"pith_summary":"Future electron-positron colliders can test a dark sector in which a light axion and a dark photon interact with the Standard Model photon through the dimension-five operator $\\frac{1}{2} g_{a\\gamma'\\gamma} a F_{\\mu\\nu}\\tilde F'^{\\mu\\nu}$. The paper's central claim is that the resulting single-photon-plus-missing-energy signal would let the ILC, CEPC, and FCC-ee probe the coupling $g_{a\\gamma'\\gamma}$ down to about $10^{-4}\\,\\mathrm{GeV}^{-1}$ for dark photon masses between roughly 10 and 230 GeV, about an order of magnitude below existing LEP II limits. The signal stands out from the Standard Model $\\nu\\bar\\nu\\gamma$ background through a sharp drop-off in the recoil-mass distribution, which also provides a direct measurement of the dark photon mass. These machines would therefore complement hadron-collider searches for dark photons decaying to muon pairs.","feed_headline":"Single-photon searches at e+e- colliders could probe 10⁻⁴ GeV⁻¹","feed_subtitle":"Future ILC, CEPC and FCC-ee runs could probe the axion-photon-dark photon coupling down to 10⁻⁴ GeV⁻¹.","key_machinery":"The central object is the dimension-five interaction term $\\frac{1}{2}g_{a\\gamma'\\gamma}aF_{\\mu\\nu}\\tilde F'^{\\mu\\nu}$, which couples the axion to one Standard Model photon and one dark photon. The paper combines this with the kinetic-mixing parameter $\\varepsilon$; in the regime $m_{\\gamma'}\\gtrsim 10$ GeV, $m_a=1$ MeV, and $\\varepsilon$ comparable to $g_{a\\gamma'\\gamma}m_{\\gamma'}$, the decay $\\gamma'\\to a\\gamma$ dominates, making single-photon-plus-missing-energy the leading signature. The discriminating observable is the recoil mass $M_{\\mathrm{recoil}}^2 = s - 2\\sqrt{s}E_\\gamma$, whose maximum value sets a sharp edge used both to separate signal from the $\\nu\\bar\\nu\\gamma$ background and to measure $m_{\\gamma'}$; the analysis also applies a $Z$-mass recoil veto and a photon polar-angle cut $|\\cos\\theta_\\gamma|<0.75$.","core_discovery":"The paper argues that whenever the dimension-five axion-photon-dark photon vertex exists, the processes $e^+e^-\\to\\gamma'\\to\\gamma a$ and $e^+e^-\\to\\gamma^*\\to a\\gamma'$ followed by $\\gamma'\\to\\gamma a$ produce a single photon plus missing energy. For dark photon masses $m_{\\gamma'}\\gtrsim 10$ GeV and an axion mass of 1 MeV, the dark photon decay is dominated by the channel $\\gamma'\\to a\\gamma$, so nearly every signal event has one hard photon and large missing energy. Using LEP II single-photon data, the paper derives an upper limit $g_{a\\gamma'\\gamma}\\lesssim 10^{-3}$ GeV$^{-1}$ for $m_{\\gamma'}\\lesssim 100$ GeV, and it projects that ILC, CEPC, and FCC-ee will be sensitive down to about $10^{-4}$ GeV$^{-1}$ for $m_{\\gamma'}$ between 10 and 230 GeV at the $5\\sigma$ level, with longitudinal beam polarization at the ILC improving the significance by a factor of four. The recoil-mass distribution exhibits a sharp edge at $M_{\\mathrm{recoil}}^{\\max}$, from which the dark photon mass is reconstructed as $m_{\\gamma'} = \\sqrt{s - (M_{\\mathrm{recoil}}^{\\max})^2}$.","pith_inferences":["If the branching ratio $\\mathrm{Br}(\\gamma'\\to a\\gamma)$ is not close to one, the projected $10^{-4}$ GeV$^{-1}$ reach would degrade; mapping the reach as a function of $m_a$ and $\\varepsilon$ would show how robust the claim is.","The recoil-mass edge technique is more general than this model: it could be used to measure the mass of any invisibly decaying resonance produced with a single photon at a lepton collider, such as a $Z'$ or a heavy neutral lepton.","Because the two production channels scale as $\\varepsilon^2$ and $g_{a\\gamma'\\gamma}^2$ respectively, combining a single-photon measurement with a $\\gamma'\\to\\ell^+\\ell^-$ search at the same machine could separate the kinetic-mixing and axionic couplings."],"forward_implications":["A null result from the ILC single-photon search with $2\\,\\mathrm{ab}^{-1}$ at $\\sqrt{s}=250$ GeV would exclude $g_{a\\gamma'\\gamma}\\gtrsim 10^{-4}$ GeV$^{-1}$ for dark photon masses of order 20-200 GeV, roughly an order of magnitude below the LEP II bound.","The sharp edge in the recoil-mass spectrum gives a direct, model-independent way to measure the dark photon mass in single-photon events without reconstructing the invisible axion.","Longitudinal beam polarization at the ILC would increase the signal significance by a factor of four, making the ILC's projected reach the strongest of the three colliders considered.","For dark photon masses above about 230 GeV, these $e^+e^-$ machines lose sensitivity in single-photon events because the dark photon cannot be produced on shell at $\\sqrt{s}=240$-$250$ GeV.","LEP II data already place the strongest existing constraint in this model, excluding $g_{a\\gamma'\\gamma}$ above roughly $10^{-3}$ GeV$^{-1}$ for $m_{\\gamma'}\\lesssim 100$ GeV."],"supporting_citations":[{"why":"Introduces the axion-photon-dark photon dimension-five interaction and its ultraviolet origin with heavy fermions charged under both gauge groups.","marker":"[3, 4]"},{"why":"Provide the LEP II single-photon event counts and selection criteria used to derive the existing constraints.","marker":"[30, 31]"},{"why":"Specifies the ILC center-of-mass energy and integrated luminosity used for the projected sensitivity.","marker":"[24]"},{"why":"Specifies the FCC-ee energy and luminosity used for the projected sensitivity.","marker":"[25]"},{"why":"Specifies the CEPC energy and luminosity used for the projected sensitivity.","marker":"[26]"},{"why":"Supply the Z-pole precision constraints that exclude large kinetic mixing and shape the allowed parameter region.","marker":"[32, 33]"},{"why":"Used to simulate signal and Standard Model background event rates and distributions.","marker":"[34]"}],"fun_headline_variants":["Future e+e- colliders could uncover axion-dark photon coupling at 10⁻⁴","Single photons at ILC, CEPC, FCC-ee to probe axion-photon-dark photon vertex","Axion-photon-dark photon: e+e- colliders set sights on 10⁻⁴ GeV⁻¹","Polarized ILC beams could quadruple axion-dark photon search sensitivity","Recoil mass sharp edge reveals dark photon mass at e+e- colliders"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The projections assume that for dark photon masses above about 10 GeV and an axion mass of 1 MeV, the dark photon decays almost exclusively to a photon plus an invisible axion, so every signal event gives one hard photon and missing energy; if that branching fraction is not near 100 percent, or the axion does not escape the detector, the projected $10^{-4}$ GeV$^{-1}$ sensitivity collapses.","fun_headline_variants_meta":{"raw":{"variants":["Future e+e- colliders could uncover axion-dark photon coupling at 10⁻⁴","Single photons at ILC, CEPC, FCC-ee to probe axion-photon-dark photon vertex","Axion-photon-dark photon: e+e- colliders set sights on 10⁻⁴ GeV⁻¹","Polarized ILC beams could quadruple axion-dark photon search sensitivity","Recoil mass sharp edge reveals dark photon mass at e+e- colliders"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001125,"raw_usage":{"total_tokens":4715,"prompt_tokens":1021,"completion_tokens":3694,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":637,"completion_tokens_details":{"reasoning_tokens":3588}},"tokens_in":637,"tokens_out":3694,"duration_ms":24377,"temperature":1.0,"reasoning_tokens":3588,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T16:01:02.054956+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the single-photon search at $\\sqrt{s}=250$ GeV with $2\\,\\mathrm{ab}^{-1}$ of data, require $|\\cos\\theta_\\gamma|<0.75$, and reject recoil masses within $[m_Z-2\\Gamma_Z,\\,m_Z+2\\Gamma_Z]$; if the observed event count matches the SM $\\nu\\bar\\nu\\gamma$ background within uncertainties, all $g_{a\\gamma'\\gamma}$ values on the paper's $5\\sigma$ contour are excluded. The claim fails if the background systematic uncertainty is larger than the roughly 1 percent assumed, so a measurement of the $\\nu\\bar\\nu\\gamma$ background rate at the 1 percent level is the decisive check.","supporting_citations":[{"cited_title":"Abadaet al","cited_arxiv_id":null,"evidence_quote":"Specifies the FCC-ee energy and luminosity used for the projected sensitivity."}],"review_version":2}