{"id":"581cc72b-2c97-416f-b298-cb11e3050441","arxiv_id":"2509.04638","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"4H-SiC LGADs achieve 61 ps time resolution with 90Sr beta particles, comparable to Si LGADs, with limited collected charge identified as the current bottleneck.","lead":"A silicon carbide low gain avalanche detector timed beta particles from a strontium-90 source with a 61 picosecond resolution, about as fast as standard silicon detectors. The measurement supports silicon carbide as a radiation-hard candidate for precision timing in future particle collider 4D trackers.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 61 ps headline rests on an unverified quadrature subtraction: the in-stack Si reference resolution and its correlation with the SiC detector are assumed, and the quoted 76 ps and 41 ps inputs do not even yield 61 ps.","rationale":"The reader's weakest assumption correctly identifies the deconvolution from an external, standalone Si reference as the linchpin of the central claim. My review reinforces that concern and adds a concrete internal inconsistency: the stated inputs (σ_ΔT = 76 ps, σ_Si = 41 ps) do not produce 61 ps, which suggests either a different reference value or a typo. The raw ΔT measurement itself is plausible and the paper gives useful charge-collection and bias-dependent trends, so this is not a reason for rejection. But the headline '61 ps, comparable to standard Si LGADs' depends on an uncontrolled systematic that can be checked with existing data. The paper should be accepted only conditionally on that check, which is exactly the reader's verdict. I therefore see no need to change the verdict, but I want the condition to be explicit and tied to the energy-correlation/reference-drift issue.","tokens_in":6938,"tokens_out":7908,"duration_ms":87866,"concrete_test":"Re-analyze the stored waveforms in narrow bins of Si LGAD collected charge (or pulse amplitude). Within a bin, the common beta-energy variation is suppressed, so compute σ_ΔT and the implied σ_SiC per bin. If the binned σ_SiC is inconsistent with the global 61 ps by more than the bin-to-bin scatter (or if σ_ΔT shows a trend with amplitude), energy correlation contaminates the quadrature extraction. Additionally, report the exact σ_Si value used in the subtraction and reconcile it with the quoted 41 ps and the arithmetic mismatch that 76 ps and 41 ps give 64 ps, not 61 ps.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section IV extracts the headline 61 ps via σ_SiC = sqrt(σ_ΔT² − σ_Si²), using σ_ΔT = 76 ± 2 ps at 40% CFD and σ_Si ≈ 41 ps from a previous standalone measurement. The load-bearing premises are that the BNL Si LGAD retains 41 ps resolution in the stacked, coincidence-selected beta setup, and that the timing errors t_Si and t_SiC are uncorrelated. Neither premise is established. The 90Sr spectrum is continuous and only betas energetic enough to traverse both detectors are analyzed; the same particle deposits correlated energies in the two layers, which can create covariance through residual amplitude-dependent time walk or amplitude-modulated jitter. Positive covariance makes quadrature subtraction underestimate the true σ_SiC; negative covariance overestimates it. There is also an internal arithmetic inconsistency: sqrt(76² − 41²) = 64 ps, not the quoted 61 ps; reproducing 61 ps would require σ_Si ≈ 45 ps in situ, which would itself mean the reference assumption has already failed. The qualitative conclusion may survive, but the exact headline number is not secured.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports timing measurements of a 300 µm diameter 4H-SiC LGAD using beta particles from a 90Sr source. The authors stack a 50 µm BNL Si LGAD as a timing reference above the 4H-SiC LGAD, record coincident pulses on a 2.5 GHz oscilloscope, extract times with 40% constant-fraction discrimination, and fit the distribution of ΔT = t_SiC − t_Si with a Gaussian. They quote σ_ΔT = 76 ± 2 ps and, using a previously measured Si reference resolution of about 41 ps, deconvolve the Si contribution in quadrature to obtain a 4H-SiC LGAD time resolution of 61 ps. They also compare collected charge and timing as a function of bias voltage, concluding that the present timing is limited by the relatively small collected charge in the SiC device rather than by carrier drift. The central claims are that this is the first demonstration of MIP-like timing in a 4H-SiC LGAD at the tens-of-picosecond level and that charge generation, not drift velocity, is the current bottleneck.","tokens_in":7258,"tokens_out":3207,"duration_ms":34295,"significance":"If the reported 61 ps value is robust, the result is significant for 4D tracking and timing detector R&D: it would demonstrate that 4H-SiC LGADs can time minimum-ionizing-like particles at a level comparable to standard Si LGADs, with the added potential of higher voltage tolerance and faster carrier drift. The paper's strength is that it is a direct two-detector differential measurement with a realistic beta source, and the charge-versus-timing trend in Fig. 10 independently supports the qualitative interpretation that signal charge is the limiting factor. The study also builds on the authors' earlier UV-TCT work, which already indicated excellent intrinsic timing. However, the headline number rests entirely on a quadrature subtraction whose inputs are not internally consistent and whose reference value is not verified in the stacked configuration. The qualitative conclusion may survive those concerns, but the exact numerical claim needs further support.","major_comments":[{"comment":"The arithmetic in the central extraction is inconsistent. With σ_ΔT = 76 ± 2 ps and σ_Si = 41 ps, quadrature subtraction gives sqrt(76² − 41²) = 64 ps, not 61 ps as quoted in the abstract and Section IV. To obtain 61 ps one would need σ_Si ≈ 45 ps in the stacked measurement. The paper must correct this arithmetic or clearly state which inputs produce 61 ps; as written, the headline value does not follow from the stated numbers.","section":"Section IV"},{"comment":"The reference resolution σ_Si ≈ 41 ps is taken from a previous standalone determination and applied to the Si LGAD inside the stacked beta setup. No in-stack calibration or cross-check is presented. The through-hole PCBs, scattering from the stack, and coincidence selection for particles that traverse both detectors can all change the reference jitter. Please provide an in-situ reference measurement (e.g., a third detector, a split-signal self-calibration, or quantitative simulation) or otherwise bound the systematic shift. Without this, the deconvolved 61 ps number is not secured.","section":"Sections III–IV"},{"comment":"The quadrature relation σ_ΔT² = σ_Si² + σ_SiC² assumes that the timing errors of the two detectors are uncorrelated. For a continuous 90Sr spectrum, the same beta particle deposits correlated energies in the two layers; residual amplitude-dependent time walk or amplitude-modulated jitter can produce covariance between t_Si and t_SiC. Positive covariance would make the true σ_SiC larger than the quadrature estimate. The authors should estimate or bound this correlation, for instance by repeating the analysis in narrow energy/amplitude bins, or justify why the correlation is negligible.","section":"Section IV"},{"comment":"No total uncertainty is quoted for the final 61 ps value. The statistical uncertainty on σ_ΔT is only 2 ps, but the uncertainty on σ_Si is not stated, and the correlation issue above is unquantified. An error budget is needed before the number can be compared with Si LGADs. In addition, Fig. 10 shows points without error bars, so the proposed charge-scaling trend is not quantitatively assessable.","section":"Section IV, Figs. 9–10"}],"minor_comments":[{"comment":"The caption says panel (a) uses a '10%–90% fraction,' while the text says the minimum is reached for fractions between 30% and 50%. Please clarify what fraction is plotted and how the CFD fraction is defined.","section":"Fig. 8 caption and Section IV"},{"comment":"The axes and legends appear truncated in the figure, and the units are not visible. Please add a complete axis label with units and error bars, and state the charge integration method used.","section":"Fig. 10"},{"comment":"There is a typo: 'due to due to the lower ionization energy deposition' should be 'due to the lower ionization energy deposition.'","section":"Section IV"},{"comment":"The gain of 7–8 is taken from the authors' own UV-TCT work in Ref. [16]. This is not circular because the present time resolution is directly measured, but stating that the gain value comes from a separate measurement and is not re-derived here would improve transparency.","section":"Section II and IV"},{"comment":"The phrase 'time resolutions of 61 ps were achieved' is used in both the abstract and conclusion, but the paper reports one operating point. Please specify the bias voltage and CFD fraction when quoting the value.","section":"Abstract/Conclusion"}],"recommendation":"major_revision","confidential_remarks":"The manuscript fits the scope of IEEE TNS and reports a potentially important first result for SiC LGAD timing with MIP-like particles. The main issue is not the quality of the measurement but the load-bearing deconvolution step: the arithmetic inconsistency and the unverified in-stack reference resolution need to be addressed before the 61 ps headline can be accepted. I would ask the authors to either provide an in-situ reference calibration, a corrected value with full uncertainty, or a softened claim consistent with the actual inputs."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is the first MIP-like timing measurement of a 4H-SiC LGAD, and that alone makes it worth knowing. The headline 61 ps is plausible but not fully secured: the paper quotes sigma_dT = 76 ± 2 ps and subtracts a 41 ps Si reference in quadrature, but sqrt(76^2 − 41^2) = 64 ps, not 61 ps. Reproducing 61 ps would need sigma_Si ≈ 45 ps in the stacked setup. That 45 ps might be right, but it is assumed, not measured, and the paper gives no final uncertainty on 61 ps. Those are reporting gaps, not necessarily wrong physics.\n\nWhat is actually new and good: the device itself is interesting—a 300 µm 4H-SiC LGAD with gain 7–8, a 75 µm drift layer, and a field plate. The charge-vs-time-resolution comparison in Fig. 10 supports the claim that charge collection, not carrier drift, currently limits timing. The Landau-convolution charge spectra and bias scans are straightforward and honestly presented. The authors do not overclaim: the conclusion explicitly says charge collection is the bottleneck and further optimization is needed.\n\nSoft spots, in proportion: (1) The in-stack calibration of the Si reference is missing; t_Si and t_SiC may be correlated through amplitude-dependent walk or jitter, since the same beta deposits correlated energies in both layers. Quadrature subtraction assumes zero covariance, which is never checked. (2) The CFD fraction is chosen from the data (40%), and no uncertainty is propagated from that choice into the quoted resolution. (3) This is a single device, single source, no beam test, so the generality is limited. (4) The arithmetic issue above is minor in magnitude but worth correcting—it flags that the reference assumption has already shifted by 4 ps. These are all addressable in revision. The central qualitative result—that SiC LGADs can time MIP-like particles at tens of picoseconds—probably survives even if 61 becomes 64 ± 5.\n\nWho this is for: the 4D-tracking and wide-bandgap detector community. It is a serious, useful data point, not a breakthrough claim dressed up. A careful referee can ask for in-situ reference calibration or a second independent timing method.\n\nRecommendation: accept for peer review. The soft spots are fixable, the measurement is genuinely new, and the paper is written with appropriate caution. I would bring it to the reading group.","headline":"First beta-source timing of a 4H-SiC LGAD, but the 61 ps headline rests on an unverified 41 ps reference and the internal arithmetic doesn't quite close; worth refereeing nonetheless.","tokens_in":7793,"tokens_out":1799,"would_cite":true,"duration_ms":15475,"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":"4H-SiC LGADs time beta particles from a 90Sr source to 61 ps, matching silicon LGADs and pointing to charge collection as the next optimization target.","keywords":["4H-SiC","LGAD","time resolution","90Sr beta source","minimum ionizing particles","4D tracking","charge collection"],"falsifier":"Build a coincidence setup with two 4H-SiC LGADs of identical design and measure their time difference directly; if the per-device resolution derived from that measurement disagrees with the 61 ps obtained by subtracting the silicon reference, the reference resolution or the correlation assumption is wrong. Alternatively, measure the silicon LGAD's resolution in the stacked geometry with a second silicon LGAD to check whether the 41 ps value holds there.","tokens_in":6895,"feed_emoji":"⚡","tokens_out":6068,"duration_ms":54060,"temperature":0.7,"pith_summary":"This paper reports the first timing measurement of 4H-SiC Low Gain Avalanche Detectors (LGADs) with beta particles from a 90Sr source, which stand in for minimum-ionizing particles. It claims a time resolution of 61 ps, comparable to standard silicon LGADs and better than earlier silicon-carbide PIN detectors. The authors argue that the remaining gap to silicon is set by limited charge generation—lower ionization yield, modest internal gain, and a partially depleted drift layer—rather than by carrier drift, which is faster in SiC. If correct, the result opens a path to timing detectors for future 4D tracking that tolerate higher voltages and harsher radiation than silicon.","feed_headline":"SiC detector times beta particles to 61 picoseconds","feed_subtitle":"Silicon carbide LGADs now match silicon timing on beta particles, with charge collection the next hurdle.","key_machinery":"The measurement uses a stacked coincidence setup: a 90Sr beta source sends particles through a thin silicon LGAD (the timing reference) and then through the 4H-SiC LGAD, both read out by fast transimpedance amplifiers and a high-speed oscilloscope. Arrival times are extracted with constant-fraction discrimination at 40% of pulse amplitude for both devices, and the spread of the time difference is fit with a Gaussian. Because the two detectors are triggered by the same beta, the silicon reference's known 41 ps jitter can be removed in quadrature to isolate the SiC device's resolution. The SiC LGAD itself is a P++/N+/N- diode with a 0.5 µm gain layer, a 75 µm drift layer, and a field plate, op","core_discovery":"The paper's central claim is that a 300 µm diameter 4H-SiC LGAD operated at 500 V times beta particles from 90Sr with 61 ps resolution, extracted by measuring the spread of arrival-time differences against a 50 µm silicon LGAD with known 41 ps resolution and subtracting in quadrature. This places SiC LGADs in the same timing class as silicon LGADs (~50 ps) for MIP-like particles. The paper further claims that the current resolution is limited by the amount of charge the device collects: beta particles deposit less energy in SiC than in Si, the internal gain is only 7–8, and the 75 µm drift layer is not fully depleted at the available bias. It concludes that improving charge collection—throug","pith_inferences":["The 61 ps number depends on the assumption that the silicon reference's 41 ps resolution is unchanged inside the stacked geometry and that the two detectors' timing jitters are uncorrelated. Since both see the same beta track, correlated energy deposition could bias the quadrature subtraction; a coincidence measurement between two identical 4H-SiC LGADs would settle this without relying on an exte","If charge collection is indeed the limiting factor, the paper's charge-versus-resolution curve predicts a concrete scaling: devices with larger gain or thicker depleted drift layers should move down that curve, providing a direct target for the next fabrication round.","The through-hole stack reduces scattering, but multiple scattering in the first detector still changes the beta's direction and energy at the second; a Monte Carlo of energy deposition in the stack could quantify how much this affects the extracted resolution.","SiC's wide bandgap implies lower leakage current and likely better radiation tolerance at high fluence; testing the timing resolution after irradiation is the natural next step, but is not part of this paper."],"forward_implications":["4H-SiC LGADs can serve as timing detectors for minimum-ionizing particles in future 4D trackers, with tens-of-picosecond precision.","Increasing internal gain and fully depleting the drift layer should improve the 61 ps resolution, because charge collection, not drift speed, is the stated bottleneck.","At equal collected charge, SiC LGADs should match or outperform Si LGADs, since they run at higher bias and have faster carrier drift.","The higher voltage tolerance and low leakage of SiC make this technology a candidate for operation in extreme or high-radiation environments."],"supporting_citations":[{"why":"Supplies the constant-fraction discrimination method used to extract timing, and a SiC PIN comparison point at roughly 100 ps.","marker":"[6]"},{"why":"Provides the silicon LGAD used as the timing reference and its previously measured ~41 ps resolution.","marker":"[10]"},{"why":"Reports the same 4H-SiC LGAD device, its gain of 7–8, and previous UV-TCT timing below 35 ps, which this beta measurement extends to MIP-like particles.","marker":"[16]"},{"why":"Describes the fabrication and characterization of the 4H-SiC LGADs used in this measurement.","marker":"[15]"},{"why":"Establishes the ~50 ps time-resolution benchmark for standard silicon LGADs that the 61 ps result is compared against.","marker":"[17]"},{"why":"Establishes that minimum-ionizing particles deposit roughly two-thirds the charge in 4H-SiC as in silicon, underpinning the charge-limitation argument.","marker":"[4]"},{"why":"Provides an earlier fast SiC PIN detector timing result on the order of hundreds of picoseconds, against which the improvement is measured.","marker":"[5]"}],"fun_headline_variants":["SiC LGADs reach 61 ps timing, matching Si","Beta timing: SiC LGADs hit 61 ps at 500V","4H-SiC LGADs: 61 ps resolution, limited by charge","SiC detector times betas at 61 ps, next step: charge","61 ps timing from SiC LGADs on beta particles"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The result rests on the reference silicon LGAD timing exactly as well inside the stacked beta setup as it did standalone, and on the two detectors' timing errors adding in quadrature with no correlation between them.","fun_headline_variants_meta":{"raw":{"variants":["SiC LGADs reach 61 ps timing, matching Si","Beta timing: SiC LGADs hit 61 ps at 500V","4H-SiC LGADs: 61 ps resolution, limited by charge","SiC detector times betas at 61 ps, next step: charge","61 ps timing from SiC LGADs on beta particles"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000252,"raw_usage":{"total_tokens":1382,"prompt_tokens":715,"completion_tokens":667,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":459,"completion_tokens_details":{"reasoning_tokens":569}},"tokens_in":459,"tokens_out":667,"duration_ms":5692,"temperature":1.0,"reasoning_tokens":569,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T05:56:31.545197+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Build a coincidence setup with two 4H-SiC LGADs of identical design and measure their time difference directly; if the per-device resolution derived from that measurement disagrees with the 61 ps obtained by subtracting the silicon reference, the reference resolution or the correlation assumption is wrong. Alternatively, measure the silicon LGAD's resolution in the stacked geometry with a second silicon LGAD to check whether the 41 ps value holds there.","supporting_citations":[{"cited_title":"Time Resolution of the 4H-SiC PIN Detector,","cited_arxiv_id":null,"evidence_quote":"Supplies the constant-fraction discrimination method used to extract timing, and a SiC PIN comparison point at roughly 100 ps."},{"cited_title":"Develop- ment of a technology for the fabrication of Low-Gain Avalanche Diodes at BNL,","cited_arxiv_id":null,"evidence_quote":"Provides the silicon LGAD used as the timing reference and its previously measured ~41 ps resolution."},{"cited_title":"Ultra-fast 4h-sic lgad with etched termination and field plate,","cited_arxiv_id":null,"evidence_quote":"Reports the same 4H-SiC LGAD device, its gain of 7–8, and previous UV-TCT timing below 35 ps, which this beta measurement extends to MIP-like particles."},{"cited_title":"Characterization of 4H-SiC Low Gain Avalanche Detectors (LGADs)","cited_arxiv_id":"2408.12744","evidence_quote":"Describes the fabrication and characterization of the 4H-SiC LGADs used in this measurement."},{"cited_title":"Characterizing the Timing Performance of a Fast 4H- SiC Detector With an 241 Am Source,","cited_arxiv_id":null,"evidence_quote":"Provides an earlier fast SiC PIN detector timing result on the order of hundreds of picoseconds, against which the improvement is measured."}],"review_version":1}