{"id":"71e96f6f-682b-4e22-a9d1-43411903c275","arxiv_id":"2412.06434","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"True QCD axion dark matter cannot induce observable neutron-antineutron oscillations, because its Goldstone nature forces competing axionless baryon-number-violating effects that are already ruled out.","lead":"This paper studies whether dark matter made of a very light field carrying baryon number two could make neutrons oscillate into antineutrons, and finds that the QCD axion cannot do it. It shows that a generic scalar or axion-like particle could in principle produce a resonant signal, so the idea is not dead but the target should be axion-like particles.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No-go for the ε=0 branch rests on the un-derived 'Naive' ILL bound λ<10^-9 on production/decay mixing; if the actual limit on an initial antineutron component is weaker, axion-induced oscillations reappear within reach.","rationale":"After reading the manuscript in good faith, the structural no-go for QCD axions is internally consistent: the Goldstone-boson/current-algebra argument of Secs. 5.1-5.2 shows that a shift-symmetric ∆B=2 derivative coupling cannot exist without companion mass terms, so either vacuum mixing ε or production/decay mixing λ appears. The vacuum branch is solidly excluded by the ILL ε bound. The decay/production branch is excluded only by the one-line 'Naively' transfer in Sec. 6. This is not an internal inconsistency, but it is a genuinely load-bearing gap: the ILL experiment was optimized for a t^2 oscillation signal from a pure neutron beam, whereas an initial antineutron admixture yields a constant-rate signal with different acceptance. The value of λ is not directly measurable from the published oscillation limit without a dedicated recast. The reader's weakest_assumption identifies exactly this point, and the recommended CONDITIONAL verdict is appropriate: the paper should either derive the λ bound from the ILL data or cite a direct bound on wrong-B production/decay. My read does not change the verdict; it reinforces the condition.","tokens_in":31282,"tokens_out":17715,"duration_ms":196758,"concrete_test":"Re-analyze the ILL search [6,7] with a beam state n' = cosα n + γ5 nC sinα, i.e., an initial antineutron fraction P0=sin^2 α≈λ^2/4 that is constant along the flight path, whereas the searched oscillation signal grows as t^2. Using the published exposure, annihilation detector efficiency, and zero-candidate/background information, derive the 90% CL upper limit on P0 and thus on λ. If the resulting bound is λ≲3×10^-9, the ε=0 no-go survives; if it is weaker than ~10^-7, the axion-induced ε0 can reach ~10^-21 eV and the paper's central conclusion requires revision.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Sec. 6 excludes the ε=0 branch (ϕΣ=π/2, mL=mR) by bounding λ=R(U)13=¯m/mn<10^-9 with 'Naively, the ILL search sets λ<10^-9' and no derivation. The ILL limit ε<0.8×10^-23 eV is a vacuum-oscillation limit that assumes a pure neutron beam at t=0 and a t^2-growing signal; it does not automatically constrain the constant antineutron admixture produced in n'=cosα n+γ5 nC sinα states. For the mixing in question, the wrong-B contamination probability is ~λ^2 and the axion-induced amplitude is ε0≈v_a√(2ρ_DM)/mnׯε. If the true 90% CL bound on this admixture were only ~10^-12 instead of ~10^-18, then λ~10^-6 and ε0~10^-21 eV, which is within two orders of magnitude of the current ILL limit and could be resonantly enhanced by magnetic tuning. No derivation, exposure estimate, or detector-acceptance argument for the transferred bound appears in the paper, so this is the load-bearing step for the claim that true QCD axions leave no room for axion-induced oscillations.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the possibility that dark matter carrying baryon number B = -2 induces neutron-antineutron oscillations through a very light scalar or axion field, with a resonant (Rabi) enhancement when the scalar mass matches the neutron-antineutron energy splitting. The authors construct a general low-energy Lagrangian for baryonic axion models in which the PQ symmetry is aligned with baryon number, then develop a detailed diagonalization of the neutron Dirac and Majorana mass terms into a standard n-nbar basis. They track the effect of this diagonalization on electromagnetic and weak interactions, on derivative axion couplings, and on anomalous triangle contributions, and they derive a final effective Lagrangian in Sec. 6. The central conclusion is that for true QCD axions the Goldstone nature forces axionless n-nbar mixing to appear either in vacuum or in production/decay, and existing experimental limits leave no room for observable axion-induced oscillations; only generic scalars or axion-like particles could produce the resonant signal.","tokens_in":31562,"tokens_out":14394,"duration_ms":160278,"significance":"If the no-go is correct, it is an important negative result: it rules out QCD axion dark matter as the source of resonantly enhanced neutron-antineutron oscillations and identifies axion-like particles or generic scalars as the only viable candidates. The technical machinery in Secs. 3-5, including the Takagi diagonalization, the transformation of currents, and the anomaly matching, is presented in detail and appears internally consistent. The paper also makes a falsifiable prediction: true QCD axions cannot induce observable resonant n-nbar oscillations, while ALPs can. The main caveat is that the exclusion of the epsilon=0 branch depends on a bound that is asserted rather than derived, which is the decisive point for the abstract's central claim.","major_comments":[{"comment":"The exclusion of the epsilon=0 branch rests on the sentence 'Naively, the ILL search sets lambda < 10^-9' and the resulting bound epsilon0 < 10^-36 eV. This transfer is not derived. The ILL limit epsilon < 0.8 x 10^-23 eV is a vacuum-oscillation limit: it assumes a beam that is initially pure neutron and a signal growing like t^2. It does not automatically bound the time-independent wrong-B admixture lambda at the production/decay vertices that controls this branch. Since epsilon=0 is exactly the branch on which axion-induced oscillations could be resonantly enhanced, the authors should derive the corresponding limit from the ILL exposure or from another experiment, including the relevant acceptance and backgrounds, or provide a citable derivation. Without that step, the abstract's central claim is not established.","section":"Sec. 6, after Eq. (109)"},{"comment":"The non-relativistic reduction used to obtain Eq. (109) is acknowledged in Sec. 7 to be incomplete: the partial_t a gamma5 term is discarded because it mixes small components, and the Conclusion states that 'further work would be needed to develop a systematic procedure.' Because the paper is a no-go statement, this is a load-bearing assumption: if the leading non-relativistic off-diagonal term were not the sigma.grad a term but an unsuppressed partial_t a term, the estimate for epsilon0 would change. The authors should either provide the systematic Foldy-Wouthuysen reduction or state explicitly that the discarded term is suppressed by additional powers of p/m or m_a/m, making Eq. (109) an estimate valid up to O(1). As written, the robustness claim in Sec. 6 goes beyond what the paper itself establishes.","section":"Sec. 6, Eqs. (107)-(109), and Sec. 7"},{"comment":"The construction in Sec. 2 assumes that the PQ symmetry is aligned with baryon number (phi carries B = -2), and the Conclusion acknowledges that 'some intricate ways to break the PQ symmetry could evade this conclusion, but they remain to be devised.' The abstract nevertheless states the no-go for 'true QCD axion models' without this qualification. Since the paper does not prove that every QCD axion coupled to Delta B = 2 must fall into the analyzed class, please qualify the abstract and Conclusion to 'baryonic QCD axion models' or give a general argument for why the analyzed class is exhaustive.","section":"Sec. 2 and Sec. 7"}],"minor_comments":[{"comment":"There is a typo in the term '1/2 m_R e^{-ia/v} m_R \\bar n_R n_R^C'; the second m_R should be removed.","section":"Eq. (9)"},{"comment":"The text states a probability sin^2(2 alpha) for detecting a positron, but Eq. (56) gives n' = cos alpha n + gamma5 n^C sin alpha, which naively yields an amplitude sin alpha and probability sin^2 alpha; please clarify whether the quoted quantity refers to a different observable.","section":"Sec. 4.2"},{"comment":"The phrase 'Naively, the ILL search sets lambda < 10^-9' should either be replaced by the derivation requested in the major comments or be accompanied by a reference; as written, the word 'Naively' signals an unverified input in a quantitative bound.","section":"Sec. 6, paragraph after Eq. (109)"},{"comment":"The Conclusion uses 'In our opinion' and 'most models' where the technical results in Secs. 3-6 support a more precise statement; consider aligning the wording with the actual scope of the analysis.","section":"Sec. 7"}],"recommendation":"major_revision","confidential_remarks":"The paper is a strong technical contribution, and the diagonalization, current transformation, and anomaly analysis are self-consistent and clearly presented. The main issue for the editor is the un-derived transfer of the ILL oscillation bound to the production/decay mixing parameter lambda in Sec. 6. If the authors can supply a rigorous derivation or a citable existing limit, I would be inclined to accept after revision. If not, the no-go should be reformulated as conditional on that bound, which would substantially weaken the abstract's claim. The paper fits the journal's scope well, and the reliance on previous work by the same group is standard tool usage rather than circularity."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read Brugeat–Smith, 2412.06434. The paper earns its keep. What is actually new: for axions that carry B = −2 and solve strong CP, the Goldstone nature forces the axion's ΔB = 2 coupling to be accompanied by axionless n–nbar mixing, either in vacuum (ε) or in production/decay vertices (λ). Both are experimentally constrained, so resonant axion-induced oscillations are out. That closes a loop earlier literature had left open. The EFT machinery—Takagi diagonalization to a standard n–nbar basis, the current rotation R(U), and the anomaly matching for ΔB = 2 currents—is careful and mostly self-contained; the step-wise diagonalization in Sec. 3 and the current transformations in Sec. 5 are the real substance, and they hold up.\n\nThe weak spots are as stated. The exclusion of the ε = 0 branch leans on a single sentence in Sec. 6: \"Naively, the ILL search sets λ < 10^−9,\" with no derivation. ILL's 0.8 × 10^−23 eV limit is on vacuum oscillations from a pure neutron beam, and it does not automatically bound a constant wrong-B admixture in the beam. That is a genuine gap. But the stress-test's stronger claim—that a weaker λ bound would put axion-induced oscillations within reach—does not survive arithmetic. In the ε = 0 scenario, ε0 = v_a √(2ρ_DM)/(4 m_n) × εbar with εbar = λ m_n/v, so ε0 = λ v_a √(2ρ_DM)/(4 v). For λ ≲ 1 and v ~ 10^12 GeV, that is ≲ 10^−27 eV, about four orders below the ILL constant-mixing limit and far below any plausible future sensitivity. The no-go does not actually depend on the un-derived ILL transfer; it is driven by the PQ scale. The paper just uses the transfer to make the bound look stronger than it needs to be.\n\nSecond caveat: the non-relativistic reduction of the axial ΔB = 2 terms is explicitly left incomplete. The γ5 off-diagonal mixes large and small components, and the authors discard the time-derivative term at leading order. They flag this in the conclusion. It is a technical loose end, not a reason to doubt the structural result.\n\nCitation pattern is fine: prior work by the same group on UV completions and triangle functions is tool usage, not circularity. No fitting to target conclusions. The central no-go is derived from Goldstone structure plus external limits.\n\nWho gets value: axion phenomenologists and anyone working on baryon-number violation or dark-matter couplings. A serious referee can usefully push on the λ bound and the non-relativistic treatment; the paper should go to review.","headline":"Solid EFT no-go: true QCD axions cannot produce observable dark-matter-induced n-nbar oscillations; one un-derived bound and an uncompleted non-relativistic reduction are real but not fatal.","tokens_in":32116,"tokens_out":8541,"would_cite":true,"duration_ms":83868,"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":"If dark matter carries baryon number two, it could resonantly drive neutron-antineutron oscillations—but a true QCD axion cannot be the driver.","keywords":["neutron-antineutron oscillations","dark matter","QCD axion","axion-like particles","baryon number violation","Rabi resonance","Goldstone boson","Majorana mass terms"],"falsifier":"Derive the actual constraint on the vertex mixing parameter $\\lambda$ from the free-neutron beam search's production and detection setup; if $\\lambda$ is not forced below $10^{-9}$, the exclusion of the $\\varepsilon=0$ branch fails. Alternatively, a neutron-beam measurement that finds positron production or fixed, time-independent $\\Delta B=2$ mixing would confirm the companion axionless effect.","tokens_in":31038,"feed_emoji":"🧲","tokens_out":16722,"duration_ms":143299,"temperature":0.7,"pith_summary":"Neutron-antineutron oscillations are the low-energy signature of baryon-number violation by two units. The paper starts from the observation that if dark matter is a very light field carrying baryon number two, its coherent oscillations can turn the tiny $n$-$\\bar n$ mixing into a resonant, time-dependent effect, greatly enhancing the signal for a given coupling. It then asks whether the QCD axion, the most minimal dark-matter candidate, can play this role. After a systematic analysis of baryonic axion models, it concludes that the axion's Goldstone boson nature forces additional, axion-independent $n$-$\\bar n$ mixing effects, in vacuum or in neutron production and decay, that are already excluded by experiment; hence resonant oscillations could only be induced by a generic scalar or axion-like particle.","feed_headline":"QCD axions can't drive neutron-antineutron oscillations","feed_subtitle":"Goldstone structure forces plain n-nbar mixing that experiments rule out, leaving generic scalars or ALPs.","key_machinery":"The central object is the $U(2)$ transformation $U$ that brings the general neutron mass matrix (Dirac mass plus left- and right-handed Majorana masses) into the standard oscillation basis through Takagi's factorization, combining baryonic and chiral rephasings with a Bogoliubov-like rotation $n\\to n\\cos\\alpha+\\gamma_5 n^C\\sin\\alpha$. The load-bearing piece is the induced $3\\times3$ rotation matrix $R(U)$ acting on the derivative currents: it mixes the baryon-number vector current with the two $\\Delta B=2$ axial currents, and its entries $R(U)_{13}$ and $R(U)_{23}$ fix the size of the axionic $n$-$\\bar n$ coupling. The Goldstone-boson counterpart is the exponential parametrization $(1+ia/v+\\dots)$ of the axion couplings, which is what forces companion axionless $\\Delta B=2$ mixing to appear.","core_discovery":"The paper's central claim is that true QCD axion models, in which the Peccei-Quinn symmetry is merged with baryon number, cannot be the source of resonant $n$-$\\bar n$ oscillations. Because the axion is a Goldstone boson, its $\\Delta B=2$ couplings must enter with the exponential factor $(1+ia/v+\\dots)$; after a baryonic reparametrization the axion couples to the baryon-number current, and its effect is entangled with the diagonalization of the neutron mass matrix. That diagonalization leaves imprints: the derivative coupling $\\partial_\\mu a\\, J^\\mu_i R(U)_{i3}$ is suppressed by $R(U)_{13}=(m_L^2-m_R^2)/(4m_D\\varepsilon_s)$ and $R(U)_{23}=-m_L m_R\\sin 2\\phi_\\Sigma/(2m_D\\varepsilon_s)$, and the $n$-$\\bar n$ mixing at production and decay vertices is equally small. Either the vacuum oscillation parameter $\\varepsilon$ is bounded by the free-neutron beam limit $\\varepsilon<0.8\\times10^{-23}$ eV, or the vertex mixing parameter $\\lambda$ is bounded below $10^{-9}$, so the paper concludes that axion-induced oscillations are phenomenologically impossible for the QCD axion and only a generic scalar or axion-like particle could induce them.","pith_inferences":["One testable extension: re-analyzing the old free-neutron beam data for production and decay mixing would either harden the $\\lambda<10^{-9}$ transfer or reopen the $\\varepsilon=0$ branch of the axion no-go.","The same Takagi-plus-anomaly machinery transfers directly to neutrinos, where merging the Majoron and axion mechanisms would couple the axion to lepton-number and $\\Delta L=2$ axial currents; the neutrino mass hierarchy would make the phenomenology rather different.","For axion-like dark matter the resonance probability scales as $m_\\phi^{-3}$, so sub-micro-eV ALPs with a direct $\\Delta B=2$ derivative coupling are the most sensitive targets for future ultracold-neutron or beam searches.","A time-correlation search in a polarized-neutron beam, looking for spin-dependent oscillations modulated at the dark-matter mass, would directly test the ALP scenario the paper leaves open."],"forward_implications":["If the no-go is right, no QCD axion model aligned with baryon number can produce observable resonant neutron-antineutron oscillations; the vacuum-oscillation and decay-mixing constraints jointly cover the parameter space.","The axionic $\\varepsilon_0$ is suppressed to roughly $10^{-36}$ eV once the free-neutron beam bound is transferred to the vertex mixing parameter, far below any planned sensitivity.","A signal of wrong-sign positrons or antineutrons in a neutron beam would not by itself prove vacuum oscillations, because production and decay can mix $n$ and $\\bar n$ even when $\\varepsilon=0$.","A generic scalar or axion-like dark matter particle with direct $\\Delta B=2$ couplings remains viable and could give a resonant, time-dependent signal; future beam searches should target such particles rather than the QCD axion.","Axial $\\Delta B=2$ couplings break the usual two-state Schrodinger reduction; the leading non-relativistic effect is a spin-dependent gradient coupling, so polarized neutrons would be needed to search for it."],"supporting_citations":[{"why":"Supplies the free-neutron beam limit $\\varepsilon<0.8\\times10^{-23}$ eV, which the paper transfers to $\\lambda<10^{-9}$ for production and decay mixing.","marker":"[6,7]"},{"why":"Provides the basis for bringing general Dirac-plus-Majorana neutron mass terms into the standard oscillation basis.","marker":"[28]"},{"why":"Gives the explicit step-wise $U(2)$ Takagi diagonalization used to construct the $U$ matrix and the $R(U)$ rotation.","marker":"[29]"},{"why":"Supplies explicit UV baryonic axion models in which the PQ symmetry merges with baryon number, whose mixing pattern is the target of the no-go.","marker":"[36]"},{"why":"Provides the anomalous triangle identities used to compute current divergences and the induced Jacobian terms.","marker":"[43]"},{"why":"Establishes that exponential and derivative axion parametrizations are equivalent, the Goldstone-boson property that forces axionless mixing.","marker":"[41, 42]"},{"why":"Gives the local dark-matter density used to normalize the oscillating axion field and estimate $\\varepsilon_0$.","marker":"[11]"},{"why":"Supplies the neutron magnetic moment and beta-decay couplings that define the standard weak-interaction basis.","marker":"[5]"}],"fun_headline_variants":["QCD axions ruled out for n-nbar oscillations","Axion-induced n-nbar oscillations impossible for QCD axion","Goldstone nature kills axion n-nbar oscillations","Only non-QCD scalars can cause n-nbar oscillations","QCD axion excluded as n-nbar oscillation source"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The no-go for the $\\varepsilon=0$ branch depends on transferring the free-neutron beam oscillation bound to the parameter $\\lambda$ that controls $n$-$\\bar n$ mixing at production and decay vertices; the paper asserts $\\lambda<10^{-9}$ without deriving the bound from that experiment.","fun_headline_variants_meta":{"raw":{"variants":["QCD axions ruled out for n-nbar oscillations","Axion-induced n-nbar oscillations impossible for QCD axion","Goldstone nature kills axion n-nbar oscillations","Only non-QCD scalars can cause n-nbar oscillations","QCD axion excluded as n-nbar oscillation source"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000658,"raw_usage":{"total_tokens":3042,"prompt_tokens":1007,"completion_tokens":2035,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":623,"completion_tokens_details":{"reasoning_tokens":1962}},"tokens_in":623,"tokens_out":2035,"duration_ms":12996,"temperature":1.0,"reasoning_tokens":1962,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T19:40:11.513737+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Derive the actual constraint on the vertex mixing parameter $\\lambda$ from the free-neutron beam search's production and detection setup; if $\\lambda$ is not forced below $10^{-9}$, the exclusion of the $\\varepsilon=0$ branch fails. Alternatively, a neutron-beam measurement that finds positron production or fixed, time-independent $\\Delta B=2$ mixing would confirm the companion axionless effect.","supporting_citations":[{"cited_title":"Neutron-Antineutron Oscillation as a Signal of CP Violation","cited_arxiv_id":"1506.05096","evidence_quote":"Provides the basis for bringing general Dirac-plus-Majorana neutron mass terms into the standard oscillation basis."},{"cited_title":"Parity-doublet representation of Majorana fermions and neutron oscillation","cited_arxiv_id":"1609.03203","evidence_quote":"Gives the explicit step-wise $U(2)$ Takagi diagonalization used to construct the $U$ matrix and the $R(U)$ rotation."}],"review_version":1}