{"id":"0921e701-887a-4369-bc04-f6fad644317c","arxiv_id":"1908.11605","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"high","formal_verification":"none","parameter_count":3,"one_line_summary":"Thin LGAD sensors could deliver signals above 1 fC at fluences up to 1e17 n_eq/cm2 by shifting charge multiplication from the gain layer to the radiation-doped bulk.","lead":"Very thin silicon detectors with a built-in charge-multiplying layer may keep working at the extreme radiation levels of future hadron colliders. The paper proposes that as radiation erases this layer, the damaged silicon itself can start multiplying charges, keeping signals readable.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Ionization quenching at 5-10e16 n_eq/cm2 is asserted from data only to 3e15; the 'will assure' conclusion overstates this extrapolation.","rationale":"Good-faith reading: this is a short proposal from an experienced LGAD group; the physics arguments about acceptor removal, thin-sensor depletion, and field shaping are standard, and the paper is explicit that it is a proposal. The load-bearing step is not the geometry or the 1 fC arithmetic but the persistence of impact ionization into the 5-10e16 n_eq/cm2 range. The manuscript itself flags this as 'one key question to be addressed in future R&D' and cites measurements only to 3e15 n_eq/cm2; the factor-of-30 extrapolation is a correctness risk, not a disagreement with consensus. The Poisson saturation model in Section 1 is suggestive but does not constrain λ: scattering by radiation-induced defects, not the probability of an untouched 1 Å^2 square, determines when α(E)^{-1} < λ fails. My concern therefore matches the reader's weakest assumption; it is a testable empirical premise. Because the paper is framed as an R&D proposal and the authors acknowledge the open question, CONDITIONAL with high correctness risk is the right verdict; my read does not move it.","tokens_in":5714,"tokens_out":4442,"duration_ms":44212,"concrete_test":"Irradiate thin (25-50 µm) LGADs to 1e15, 3e15, 1e16, 3e16, and 1e17 n_eq/cm2, and measure collected charge as a function of bias voltage up to 500-600 V at -30°C. If the internal gain extracted from these curves falls to unity for fluences at or above 1e16 before breakdown, or if the bias needed to restore gain exceeds the sustainable voltage, the 'more than 1 fC at 5-10e16' claim fails. A complementary TCAD model with NIEL-scaled defect scattering and impact ionization would test the same premise analytically at all five fluences.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim requires impact ionization to persist at fluences of 5-10e16 n_eq/cm2, but the only cited evidence in Section 4 extends to 3e15 n_eq/cm2, a factor of 17-33 short. Irradiation introduces scattering centers that shorten the carrier mean free path λ; once λ < α(E)^{-1} at any field, multiplication is quenched. The paper's Section 1 Poisson saturation model with a hand-picked 1 Å^2 target does not close this gap: it estimates how much area has been hit, not the density of electrically active defects that scatter hot carriers, and at 5e16 cm^-2 the model itself leaves only about 0.7% of area untouched. Section 4 explicitly calls gain quenching 'one key question to be addressed in future R&D', and the conclusion's 'will assure the delivery of more than 1 fC' is stronger than the body's 'looks possible'. Without a gain-versus-fluence measurement or a radiation-damage simulation that includes impact ionization, the load-bearing premise remains an extrapolation, not a demonstrated result.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"Cartiglia, Sadrozinski, and Seiden propose using 25–50-µm-thick Low Gain Avalanche Diodes (LGADs) as tracking sensors at fluences of 5–10×10^16 n_eq/cm^2. They argue that at low fluence the gain layer provides a gain of about 5–10, while at high fluence the irradiation-increased bulk doping, combined with a bias voltage around 500 V, will induce charge multiplication in the bulk, preserving signals above 1 fC. The paper reviews known non-linear radiation-damage effects, thin-sensor advantages, stable electric-field configurations for internal gain, irradiation effects on doping and ionization, and a trench-isolated high-density LGAD design for near-100% fill factor. It contains no new measurements and no quantitative device simulation; the central claim is presented as a physics-based extrapolation and a design goal.","tokens_in":5895,"tokens_out":7541,"duration_ms":69757,"significance":"The paper gives a clear and honest summary of why thin LGADs are interesting for extreme fluences, and it explicitly identifies the main unknown: whether impact ionization survives at fluences above about 3×10^15 n_eq/cm^2. Its proposal to rely on both gain-layer and bulk multiplication is physically plausible and falsifiable, and the trench-isolation idea addresses a real fill-factor problem. The contribution is useful as a roadmap for future R&D, but as it stands it does not demonstrate its central quantitative claim: there is no gain-versus-fluence curve, no field simulation, and no measurement beyond 3×10^15 n_eq/cm^2. The value of the paper is therefore prospective rather than evidential.","major_comments":[{"comment":"The central quantitative claim—that more than 1 fC will be delivered up to 5–10×10^16 n_eq/cm^2—is not established by the material in the paper. Section 4 states that “current studies have proven that up to fluences 3·10^15 n_eq/cm^2 impact ionization is not quenched,” and then calls the dependence of gain on irradiation “one key question to be addressed in future R&D.” Section 5 concludes only that the possibility of gain in the 5–10×10^16 interval “looks possible,” yet the Conclusion asserts that the interplay of gain-layer and bulk multiplication “will assure” more than 1 fC. No calculation of gain versus fluence, no simulation of the electric-field and avalanche evolution, and no measured data at fluences above 10^16 n_eq/cm^2 are presented. The factor-of-17-to-33 extrapolation from 3×10^15 n_eq/cm^2 is load-bearing and needs either a quantitative model with realistic defect densities and impact-ionization coefficients, or an explicit softening of the conclusion to a conjecture.","section":"Sections 4, 5, and 6"},{"comment":"The saturation argument uses a Poisson model with a_o = 1 Ų, a value chosen without justification even though the paper itself notes that damaged clusters extend over tens of ångströms. The choice determines the quantitative conclusion: for a_o = 1 Ų, the probability that a particle hits an untouched square after 5×10^16 cm^-2 is exp(-5) ≈ 0.7%, whereas clusters with ten times larger area would make the same probability exp(-50). More importantly, the model computes the geometric overlap of damaged regions, not the density of electrically active scattering centers that could shorten the carrier mean free path λ below α(E)^{-1}; the latter is what determines whether impact ionization is quenched. The model therefore cannot carry the weight of the claim that multiplication persists to 5–10×10^16 n_eq/cm^2.","section":"Section 1, Fig. 2"},{"comment":"The proposed operating scenario depends on a thin sensor that can be over-depleted at about 500 V after irradiation, with a high-density trench-isolated electrode structure and near-100% fill factor. None of these requirements is demonstrated in the paper. The trench design is supported only by a workshop reference [9]; no measured breakdown voltage, no pixel-isolation data, and no post-irradiation results are shown. Because the entire scenario fails if the sensor cannot hold the bias needed for bulk multiplication, this is a second load-bearing gap rather than a cosmetic detail.","section":"Section 5"}],"minor_comments":[{"comment":"Equation (1) as written, N(x) = ∫ N(o) e^{α(E)x} dx, is dimensionally inconsistent and not the correct integral form for avalanche gain; the intended expression is N(x) = N0 exp(∫_0^x α(E(x')) dx') or equivalently dN/dx = αN. Please correct the notation.","section":"Section 3, Eq. (1)"},{"comment":"For a_o = 1 Ų and fluence 1×10^16 cm^-2, the model gives exp(-1) ≈ 0.37, not the 30% stated in the text; please either quote the correct value or explain the fit or approximation used for the curve.","section":"Section 1, Fig. 2"},{"comment":"The sentence defining Eq. (2) contains typos and a garbled clause (“andc”, “the initial (after a fluence φ) acceptor density”), which makes the definition of c and the acceptor densities hard to parse.","section":"Section 4, Eq. (2)"},{"comment":"The abstract states “above 1·10^16 n_eq/cm^2” while the title and Section 5 specify 5–10×10^16 n_eq/cm^2; these ranges should be harmonized.","section":"Abstract"},{"comment":"The statement that impact ionization is “not quenched” up to 3×10^15 n_eq/cm^2 is given without a specific citation; please add the reference to the measurement or measurements that support this claim.","section":"Section 4"},{"comment":"The captions refer to tones of red indicating doping levels, but the grayscale rendering in the posted version makes these levels hard to distinguish; explicit labels or a colormap legend would improve clarity.","section":"Figures 5 and 7"}],"recommendation":"major_revision","confidential_remarks":"The manuscript reads more like a proceedings-style research proposal than a completed study. I would support publication only if the journal accepts forward-looking detector-simulation papers; in that case the authors should either add a quantitative plausibility estimate (for example, a simple gain-versus-fluence calculation using measured or conservative α(E) and defect densities) or explicitly reframe the conclusion as a conjecture. The authors’ reliance on their own earlier LGAD radiation studies is not inappropriate because those references do contain the underlying data, but the new extrapolation needs independent support."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear Colleague,\n\nThe paper is a short design proposal from a group that knows LGADs. The genuinely new idea is to let irradiation move charge multiplication from the gain layer into the bulk, so that a 25–50 micron sensor could keep delivering around 1 fC up to 5–10e16 n_eq/cm2. That is worth stating clearly, and the paper does it well. It explains why thin sensors are attractive at such fluences (lower leakage, lower depletion voltage, better trapping tolerance) and why the LGAD field configuration is the right one to achieve controlled gain.\n\nThe soft spot is the quantitative claim. The conclusion says the interplay of gain layer and bulk multiplication 'will assure the delivery of more than 1 fC' at 5–10e16. The body only says it 'looks possible', and Section 4 identifies gain quenching as a key open R&D question. The gap matters. The evidence for unquenched ionization stops at 3e15, a factor of 20 short of 5e16. The saturation model in Section 1 is a toy: hitting area with 1 A^2 squares does not directly tell you how many electrically active defects will shorten the carrier mean free path. There is no gain-versus-fluence calculation, no simulation, and no irradiation data.\n\nThis is not a fatal flaw if the paper is read as a roadmap. The authors are candid about the open questions, and the idea is testable. But a referee should ask for the conclusion to be softened or, better, for a quantitative model that connects defect density to lambda and alpha. The depletion voltage and geff values used are also extrapolated, which should be flagged.\n\nThe references are fine; the self-citations point to their own previous measurements and reviews, which is legitimate here.\n\nWho should read this? People working on 4D tracking or future hadron collider detectors. It is a useful pointer for an R&D direction, not a demonstration. I would send it to peer review rather than desk reject, but with major revision requested. The right outcome is a publication that clearly labels the extrapolation and states what needs to be measured.","headline":"A clear, well-aimed proposal for LGADs at 5–10e16 that would be stronger if its central 'will assure 1 fC' conclusion were backed by a quantitative model or data.","tokens_in":6480,"tokens_out":4173,"would_cite":true,"duration_ms":36714,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["29.40.Gx","29.40.Wk","61.82.Fk"],"model":"deepseek-v4-flash","headline":"Very thin LGADs can keep tracking signals above 1 fC at fluences of 5–10×10^16 n_eq/cm2.","keywords":["Low Gain Avalanche Diode","LGAD","silicon tracking detectors","radiation damage","charge multiplication","impact ionization","thin silicon sensors","future hadron collider"],"falsifier":"Take a 25–50 µm LGAD, irradiate it to $10^{16}$–$10^{17}\\,n_{\\rm eq}/{\\rm cm}^2$, and measure collected charge versus bias: if the signal stays below 1 fC at the highest sustainable voltage, or if gain does not increase as bulk multiplication turns on, the central claim fails.","tokens_in":5444,"feed_emoji":"⚡","tokens_out":14115,"duration_ms":118779,"temperature":0.7,"pith_summary":"This paper argues that very thin Low Gain Avalanche Diodes (LGADs), 25–50 µm thick, can still produce the ≥1 fC signal needed by readout chips after irradiation to $5\\text{--}10\\times10^{16}\\,n_{\\rm eq}/{\\rm cm}^2$, roughly 30 times the fluence at which silicon's response has been directly tested. The idea is to split the amplification between two stages: at lower fluence the implanted gain layer multiplies the signal by 5–10, and as irradiation erases that layer and dopes the bulk, the bias voltage shifts multiplication into the bulk itself. If the argument holds, future hadron-collider trackers would not need thick silicon sensors, avoiding the leakage current, trapping, and high depletion voltage that make thick sensors impractical at such fluences. The authors present the proposal as an extrapolation whose load-bearing assumption is that impact ionization is not quenched in the $5\\text{--}10\\times10^{16}$ range.","feed_headline":"Thin silicon sensors could track at 30 times the tested fluence","feed_subtitle":"Gain shifts from the doping layer to the bulk, keeping thin trackers above the 1 fC readout threshold.","key_machinery":"The load-bearing object is the electric-field profile of an LGAD, in which a thin buried gain layer creates a short, flat, bias-controlled region where the field approaches the critical ionization strength of roughly 250–300 kV/cm. The argument follows how equation (2), $N_A(\\varphi)=g_{\\rm eff}\\varphi+N_A(0)e^{-c\\varphi}$, evolves that profile: acceptor removal erases the gain layer while acceptor creation dopes the bulk, moving the high-field region deeper. Charge gain is computed as $N(x)=\\int N(0)e^{\\alpha(E)x}\\,dx$, with $\\alpha(E)^{-1}$ the drift distance needed to ionize; the mechanism works only while this distance is shorter than the carrier mean free path $\\lambda$. A separate enabling piece is the proposed shallow-trench termination, which removes the interpad dead area and would allow 50×50 µm² pixels with near-100% fill factor.","core_discovery":"The paper claims that a 25–50 µm thin LGAD can deliver more than 1 fC of signal at fluences of $5\\text{--}10\\times10^{16}\\,n_{\\rm eq}/{\\rm cm}^2$ by switching its amplification site as radiation damage accumulates. It models the damage with $N_A(\\varphi)=g_{\\rm eff}\\varphi+N_A(0)e^{-c\\varphi}$: irradiation removes dopants from the gain layer while creating acceptor-like defects in the bulk, moving the high-field region from the gain layer into the bulk. Up to $0.1\\text{--}0.3\\times10^{16}$ the gain layer alone gives a stable factor of 5–10; above that, a bias near 500 V triggers impact ionization in the bulk and provides the same gain. A 20–25 µm sensor is singled out because it can remain fully depleted at $10^{17}\\,n_{\\rm eq}/{\\rm cm}^2$, and its small initial signal of about 0.3 fC requires only moderate gain to reach the 1 fC threshold. The authors are explicit that current evidence for unquenched impact ionization reaches only $3\\times10^{15}$.","pith_inferences":["The paper's own evidence stops at $3\\times10^{15}\\,n_{\\rm eq}/{\\rm cm}^2$, so the $5\\text{--}10\\times10^{16}$ prediction is a testable extrapolation; measuring gain versus bias on thin LGADs irradiated past $10^{16}$ would settle the bulk-multiplication handoff directly.","If the damage-saturation picture in Section 1 is correct, leakage current and effective doping should plateau rather than keep rising with fluence, which would make the proposed operating point easier to reach than a linear extrapolation predicts.","Because the mechanism relies only on field strength and carrier scattering, the gain-layer-to-bulk transfer could be tested in simpler thin silicon diodes before committing to full LGAD fabrication."],"forward_implications":["Tracking layers at future hadron colliders could be built from 25–50 µm silicon instead of 100–200 µm, cutting material while keeping signals above the ~1 fC threshold set by front-end chips.","Thin sensors would remain fully depleted after extreme irradiation, avoiding the steep depletion-voltage rise and field distortion that disable thick sensors at high fluence.","The detector could keep working after its gain layer is erased, because bias-induced bulk multiplication takes over; the bias voltage becomes the control knob for sensor lifetime.","A shallow-trench LGAD with 50×50 µm² pixels would combine high radiation tolerance with close to 100% fill factor, solving the traditional dead-area problem of gain-terminating structures."],"supporting_citations":[{"why":"Documents the non-linear radiation-damage behaviour and field stability at high fluence that motivate extrapolation beyond current tests.","marker":"[1]"},{"why":"Supplies the acceptor-creation coefficient and depletion-voltage framework used to argue that 20–25 µm sensors remain fully depleted.","marker":"[4]"},{"why":"Sets the ~1 fC minimum charge that the readout chip requires from the sensor.","marker":"[5]"},{"why":"Supplies the impact-ionization model, the gain equation, and the critical-field description used throughout the paper.","marker":"[6]"},{"why":"Defines the LGAD design and gain-layer geometry that the proposal modifies.","marker":"[7]"},{"why":"Underpins the acceptor-removal and acceptor-creation model in equation (2) for irradiated LGADs.","marker":"[8]"},{"why":"Provides the high-density shallow-trench design needed to reach near-100% fill factor in small pixels.","marker":"[9]"}],"fun_headline_variants":["Thin LGADs shift gain to bulk to track at extreme fluence","Thin silicon trackers reach 1 fC at 10^16 via dual gain","Gain layer fades, bulk gain takes over in thin LGADs","Thin trackers could double gain at 500V for 5e16 fluence"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument stands or falls on the assumption that impact ionization is not quenched at $5\\text{--}10\\times10^{16}\\,n_{\\rm eq}/{\\rm cm}^2$; the paper's evidence for unquenched multiplication reaches only $3\\times10^{15}$, roughly a factor of 30 lower.","fun_headline_variants_meta":{"raw":{"variants":["Thin LGADs shift gain to bulk to track at extreme fluence","Thin silicon trackers reach 1 fC at 10^16 via dual gain","Gain layer fades, bulk gain takes over in thin LGADs","Thin trackers could double gain at 500V for 5e16 fluence"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001126,"raw_usage":{"total_tokens":4727,"prompt_tokens":1034,"completion_tokens":3693,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":650,"completion_tokens_details":{"reasoning_tokens":3604}},"tokens_in":650,"tokens_out":3693,"duration_ms":30671,"temperature":1.0,"reasoning_tokens":3604,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T10:10:27.004296+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a 25–50 µm LGAD, irradiate it to $10^{16}$–$10^{17}\\,n_{\\rm eq}/{\\rm cm}^2$, and measure collected charge versus bias: if the signal stays below 1 fC at the highest sustainable voltage, or if gain does not increase as bulk multiplication turns on, the central claim fails.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the acceptor-creation coefficient and depletion-voltage framework used to argue that 20–25 µm sensors remain fully depleted."},{"cited_title":"Balbuena and et al., RD50 S tatus R eport 2008 - R adiation hard semiconductor devices for very high luminosity colliders ,","cited_arxiv_id":null,"evidence_quote":"Sets the ~1 fC minimum charge that the readout chip requires from the sensor."},{"cited_title":"de Maria and et al., Results and perspectives from rd53 on the next generation readout chips for hl-lhc silicon pixel detector phase 2 upgrades, TREDI2019, 2019","cited_arxiv_id":null,"evidence_quote":"Supplies the impact-ionization model, the gain equation, and the critical-field description used throughout the paper."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the LGAD design and gain-layer geometry that the proposal modifies."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Underpins the acceptor-removal and acceptor-creation model in equation (2) for irradiated LGADs."}],"review_version":1}