{"id":"842eea82-6b1f-4836-8210-0dd979e8278f","arxiv_id":"2511.22308","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Thin 20 µm LGAD sensors achieve 16.6 ps timing resolution in a test beam, and 12.2 ps with a two-plane tracker.","lead":"Thin silicon sensors called LGADs measured particle arrival times as fast as 16.6 picoseconds in a beam test, with two sensors together reaching 12.2 ps. This is one of the fastest timing results for this sensor family, pointing toward future 4D particle trackers.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Irradiated-sensor 20 ps claim rests on an unsupported assertion that timing resolution is temperature-independent across the -50 to -35°C drift; this needs demonstration before the radiation-hard headline is accepted.","rationale":"The reader's verdict is CONDITIONAL, and I agree that the central 16.6 ps result for 20 µm sensors and the 12.2 ps two-plane tracker are credible and well-documented. The MCP resolution is extracted in situ with a robust three-plane formula, the systematics on CFD and charge are studied, and the results are consistent with Weightfield2 simulations and expected jitter/ionisation trends. The 45 µm β-source substitution is a presentation problem that affects the abstract and the thickness-scaling plot, but it does not bear on the 20 µm or tracker results; I agree it should be addressed but do not consider it load-bearing for the central claim. The weakest point is the unsupported claim that temperature does not affect σ_t for irradiated sensors. Section 3 documents a 8–15°C temperature drift during data taking, and Section 3.1 explicitly treats gain as temperature-dependent via a 2 V/°C correction. The timing resolution depends on the signal rise time and slew rate, which are also temperature-dependent, so the assertion in Section 4 is not a logical consequence. Without a demonstrated temperature independence, the irradiated-sensor headline of ≈20 ps is not fully supported. This does not invalidate the non-irradiated results, but it is a real gap in one of the three headline claims. A concrete re-analysis of the existing data by temperature bin would settle it; if the effect is smaller than the uncertainty, the claim stands. Hence the reader's CONDITIONAL verdict remains appropriate, and no change is needed.","tokens_in":12240,"tokens_out":14614,"duration_ms":129638,"concrete_test":"Re-analyze the existing irradiated 30 µm beam data by splitting events according to the recorded DHT11 temperature into cold (≤ -45°C) and warm (> -45°C) subsets for each fluence. Recompute σ_t(DUT) for each subset at matched gain/charge (using the bias-temperature correction already applied to the gain). If the two σ_t values differ by more than the combined statistical and fit uncertainties, the Section 4 assertion is false. As a complementary check, run Weightfield2 simulations at -50°C and -35°C at identical gain to quantify the expected change in rise time and jitter contribution.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 4 states: \"The change in temperature does not affect the σ_t(DUT) measurements, and hence no bias correction is applied\" for the irradiated 30 µm sensors. Yet Section 3 reports that these sensors were operated at temperatures varying from -50°C to -42°C, and for the 1.5×10^15 sample from -50°C to -35°C. Section 3.1 applies a 2 V/°C bias correction to gains because the gain is temperature-sensitive. Since σ_t depends on jitter (slew rate/noise) and ionisation, both of which are affected by temperature through drift velocity and gain, the assertion of temperature independence is not self-evident. No measurement or simulation is provided to support it. If σ_t varies by even a few ps across this temperature range, the quoted ≈20 ps for irradiated 30 µm sensors, and the comparison across fluences at different temperatures, become unreliable. This is load-bearing because the radiation-hardness performance is a central claim of the EXFLU study.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports timing-resolution measurements of thin LGAD sensors from the FBK EXFLU production, performed at the DESY Test Beam Facility with 4 GeV/c electrons. Non-irradiated sensors of 20, 25, 30, 35, and 45 µm thickness were tested; the best single-sensor resolution is 16.6 ± 0.7 ps for the 20 µm sensors, and a two-plane tracker using two 20 µm sensors reaches 12.2 ps. Irradiated 30 µm sensors were measured at fluences from 0.4 × 10^15 to 2.5 × 10^15 n_eq/cm² and achieve about 20 ps timing resolution. The analysis uses constant-fraction discrimination at 30%, Gaussian fits to the DUT–MCP time difference, an in-situ extraction of the MCP reference resolution from three pairwise combinations, and explicit systematic checks on amplitude extraction and interpolation. The paper also presents the jitter and ionisation contributions to the timing resolution and a thickness-scaling relation for the minimum charge required to reach a given resolution.","tokens_in":12472,"tokens_out":4386,"duration_ms":38244,"significance":"If the results hold, the 16.6 ps single-sensor resolution for a 20 µm LGAD and the 12.2 ps two-plane tracker represent state-of-the-art timing performance for very thin LGADs in a beam test, with direct relevance to 4D tracking at HL-LHC and FCC-hh. The paper's strengths include a clean beam-test methodology: the MCP resolution is obtained by solving the quadrature system from three pairwise combinations rather than assumed, the CFD threshold is optimized, and the systematic checks on interpolation and amplitude extraction are explicit. The radiation-hardness claim for irradiated 30 µm sensors at 20 ps up to 2.5 × 10^15 n_eq/cm² is significant for extreme-fluence applications, provided the temperature-dependence question is resolved. However, two limitations—the unsupported assertion of temperature independence for the irradiated-sensor data and the substitution of a β-source measurement for the 45 µm beam-test point—currently weaken the broader scaling and radiation-hardness conclusions.","major_comments":[{"comment":"The 45 µm timing result (26.4 ± 1.7 ps in Table 2) is not from the 4 GeV/c beam, as stated in Section 3: \"the 45µm results are replaced with low-noise data obtained using a 45µm sensor tested with a 90Sr β source.\" The abstract and Table 2 present this value as part of the beam-test series without flagging the substitution, and Figure 13 mixes the β-source point with beam points. While the figure caption acknowledges the β source, the paper provides no cross-check that the β-source measurement is equivalent to the beam measurement for timing resolution. The two setups differ in particle type (4 GeV/c electrons vs. endpoint β electrons), particle flux, readout chain (single-plane vs. triple-plane), and noise conditions. This issue affects the quantitative thickness-scaling conclusions in Figures 13 and 14, which use the 45 µm point. The authors should either provide a beam measurement usi","section":"Section 3, Table 2, and Figure 13"}],"minor_comments":[{"comment":"The abstract in the full text states \"four device-under-test (DUT) planes,\" but Section 3 describes two DUT planes (DUT1 and DUT2). The arXiv abstract correctly says \"two.\" This inconsistency should be corrected.","section":"Abstract (full text version)"},{"comment":"The arXiv title reads \"Timing resolution from beam tests on thin LGADs down to 16.6 ps,\" while the full-text title is \"Timing resolution from very thin LGAD sensors tested on particle beam down to 12 ps.\" The title should be harmonized.","section":"Title"},{"comment":"The caption states \"The uncertainties in the charge estimates are not shown.\" For a quantitative scaling claim, charge uncertainties should be displayed or at least a representative value given.","section":"Figure 14"},{"comment":"The units \"pWb or mV·ns\" for signal area are unconventional; \"mV·ns\" (or V·s) is clear, but \"pWb\" appears to be a typographical artifact. Please clarify.","section":"Section 3.1, Equation (4)"}],"recommendation":"major_revision","confidential_remarks":"The core 16.6 ps measurement appears sound and is a valuable result for the LGAD community. However, the paper's two broader claims—the radiation-hardness performance at ~20 ps and the thickness-scaling trend—rest on the unsupported temperature-independence assertion and on mixing a β-source measurement with beam data for the 45 µm point. Both issues are fixable with additional analysis or with a revised presentation that explicitly separates the affected points. I recommend major revision rather than rejection because the central result is defensible and the required work is within the scope of a revision. I would also encourage the editor to ensure the final version clearly discloses the β-source substitution in the abstract, as the current abstract could mislead readers into thinking all values come from the beam test."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Hey,\n\nThe headline here is that the core result is real: a 20 µm LGAD hitting 16.6 ± 0.7 ps in a 4 GeV electron beam, and a two-plane tracker at 12.2 ps. That's a genuine step below the usual 25–35 ps range for sub-55 µm LGADs, and the measurement looks careful: CFD at 30%, Gaussian fits, in-situ MCP resolution extraction from the three pairwise combinations, and explicit checks on interpolation and amplitude-fitting systematics. I don't see a circularity problem — the MCP resolution is solved from the quadrature system, not fitted to a target. The non-irradiated part of the paper deserves a serious referee.\n\nThe soft spots are both presentation and support. First, the 45 µm number (26.4 ps) in Table 2 and Figure 13 is not from the same beam setup — the paper says it's replaced with β-source data from a single-plane setup. The abstract presents it as part of the beam-test series without flagging the substitution, and the thickness-scaling plot mixes apparatuses. That's a real issue for anyone trying to use those points for scaling. It does not touch the 16.6 ps result.\n\nSecond, the irradiated-sensor claim of ~20 ps up to 2.5×10^15 n_eq/cm² rests on an assertion that temperature variation across -50 to -35°C does not affect σ_t. They don't show a measurement or simulation for that, and they do apply a 2 V/°C bias correction to gain because it is temperature-sensitive. Jitter and ionisation both plausibly depend on gain and drift velocity, so the assertion is not self-evident. This is load-bearing for the radiation-hardness part of the paper's story. It's not a fatal flaw — maybe the effect really is sub-ps — but it needs demonstration before I'd quote the 20 ps figure as robust.\n\nThe citation pattern looks normal: prior work from the same FBK EXFLU program is cited, and the new measured values do improve on the referenced state of the art. The authors are clear about what is new.\n\nFor peer review: yes, send it. The central measurement deserves referee time, and the referee can push for the clarifications: flag the β-source substitution in the abstract and show the temperature dependence (or justify the existing claim) for the irradiated data.\n\nBest.","headline":"Solid beam-test result: 16.6 ps timing in 20 µm LGADs is credible and new, but the paper overreaches in the abstract and thickness scaling by mixing in a β-source 45 µm point, and the irradiated-sensor temperature claim needs support.","tokens_in":13070,"tokens_out":2742,"would_cite":true,"duration_ms":24421,"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":"Ultra-thin 20 µm LGAD sensors reach 16.6 ps timing resolution in a 4 GeV/c electron beam, and two sensors used together as a tracker reach 12.2 ps.","keywords":["Low-Gain Avalanche Diodes (LGADs)","timing resolution","beam tests","thin silicon sensors","radiation hardness","4D tracking","acceptor removal","single-event burnout"],"falsifier":"Re-measure the same 45 µm EXFLU sensor in the same 4 GeV/c electron beam with the identical readout chain and CFD settings, and compare the resulting σ_t(DUT) with the 26.4 ps β-source value; a mismatch beyond the quoted uncertainties would break the thickness-scaling conclusion.","tokens_in":12096,"feed_emoji":"⚡","tokens_out":4834,"duration_ms":44432,"temperature":0.7,"pith_summary":"This paper reports beam-test measurements of thin Low-Gain Avalanche Diodes (LGADs) designed for high-radiation environments. The central result is that a 20 µm-thick sensor achieves a single-device timing resolution of 16.6 ± 0.7 ps, and two such sensors used together as a tracker reach 12.2 ps. Neutron-irradiated 30 µm sensors keep timing resolution around 20 ps up to a fluence of 2.5 × 10^15 n_eq/cm², which is the regime needed for 4D tracking at future colliders. The authors attribute the improvement to thinner sensors producing faster signals and smaller ionisation fluctuations. A caveat: the reported 45 µm point (26.4 ps) was measured with a radioactive β source rather than the beam, so the thickness-scaling trend mixes two apparatuses.","feed_headline":"20 µm LGADs time particles to 16.6 ps in beam","feed_subtitle":"A two-plane tracker of the same ultra-thin sensors reaches 12.2 ps, and irradiated 30 µm sensors hold about 20 ps.","key_machinery":"The central object is the thin LGAD: an n-in-p silicon sensor with a shallow boron gain implant co-implanted with carbon, with substrate thickness varied from 20 to 45 µm. The load-bearing mechanism is the timing-resolution relation σ_t² = σ_jitter² + σ_ionisation². Thinning the substrate reduces the signal rise time, directly lowering the jitter term, and also reduces the ionisation-fluctuation term because the faster rising edge is less affected by non-uniform energy deposition. The two-plane tracker result is obtained by averaging the constant-fraction-discriminator arrival times of two 20 µm sensors, demonstrating the expected 1/√2 improvement. The micro-channel plate reference resolutio","core_discovery":"The paper claims that in the EXFLU family of thin LGADs, timing resolution improves steadily as the substrate is thinned, without sacrificing radiation tolerance. The best single-sensor result is σ_t(DUT) = 16.6 ± 0.7 ps for a 20 µm sensor, and averaging the arrival times of two 20 µm planes yields 12.2 ps. Irradiated 30 µm sensors, at fluences from 0.4 to 2.5 × 10^15 n_eq/cm², still deliver about 20 ps, with the most irradiated sample reaching 20.5 ps before the single-event burnout limit. The authors explain the improvement via the decomposition σ_t² = σ_jitter² + σ_ionisation²: thinner substrates shorten the rise time linearly (from ~600 ps at 45 µm to ~300 ps at 20 µm) and reduce the ion","pith_inferences":["If the 45 µm β-source point is not equivalent to the beam measurements, the precise slope of the thickness-scaling curve is uncertain; the robust comparison is among the beam-tested thicknesses 20, 25, 30, and 35 µm.","The 12.2 ps two-plane result suggests that a larger tracker of several thin LGAD layers could push overall time resolution toward ~10 ps or below, provided the readout electronics preserve the fast rise time.","The observed thickness-independent signal amplitude implies that further thinning to 15 µm—already fabricable—may yield even lower timing resolution, but would face lower collected charge and stronger Landau fluctuations.","Because the MCP reference contributes only about 5 ps, the quoted 16.6 ps is not reference-dominated; an independent faster reference could confirm the DUT-only resolution."],"forward_implications":["Sub-25 ps single-plane timing is achievable in LGADs thinner than 35 µm, with 20 µm sensors reaching 16.6 ps.","A two-plane tracker of 20 µm LGADs reaches 12.2 ps, verifying the expected 1/√2 improvement from combining planes.","Thinning the sensor from 35 µm to 20 µm cuts the minimum charge needed for sub-30 ps timing from ~6 fC to ~2 fC, a factor greater than 2.5.","30 µm sensors retain ~20 ps timing after neutron irradiation up to 2.5 × 10^15 n_eq/cm², with the limit set by single-event burnout at about 405 V.","Signal amplitude at a given gain is independent of substrate thickness, so thinner sensors can improve timing without losing signal amplitude."],"fun_headline_variants":["Beam tests: 20 µm LGADs hit 16.6 ps timing","Thinner LGADs yield better timing: 16.6 ps at 20 µm","LGAD timing resolution hits 16.6 ps in thin sensors","Two 20 µm LGADs combine for 12.2 ps timing"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The thickness-scaling trend rests on comparing the 45 µm point measured with a 90Sr β source to the other points measured in the 4 GeV/c electron beam; if those two measurements are not equivalent, the apparent improvement with thinning is partly an artifact of the changed setup.","fun_headline_variants_meta":{"raw":{"variants":["Beam tests: 20 µm LGADs hit 16.6 ps timing","Thinner LGADs yield better timing: 16.6 ps at 20 µm","LGAD timing resolution hits 16.6 ps in thin sensors","Two 20 µm LGADs combine for 12.2 ps timing"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000723,"raw_usage":{"total_tokens":3221,"prompt_tokens":1024,"completion_tokens":2197,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":768,"completion_tokens_details":{"reasoning_tokens":2120}},"tokens_in":768,"tokens_out":2197,"duration_ms":15369,"temperature":1.0,"reasoning_tokens":2120,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T19:46:36.625629+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-measure the same 45 µm EXFLU sensor in the same 4 GeV/c electron beam with the identical readout chain and CFD settings, and compare the resulting σ_t(DUT) with the 26.4 ps β-source value; a mismatch beyond the quoted uncertainties would break the thickness-scaling conclusion.","supporting_citations":[],"review_version":1}