{"id":"3f92628d-da70-4d2c-be2a-05a4f794a96d","arxiv_id":"1908.10973","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"RD53A pixel modules achieve intrinsic position resolutions of about 10.9 and 6.8 micrometers in the 50 and 25 micrometer directions when tilted by 13 degrees, measured in an 11 GeV electron testbeam.","lead":"This paper measures the position resolution of prototype RD53A pixel detector modules using an 11 GeV electron testbeam at SLAC. The results, about 10.9 and 6.8 micrometers for the 50 and 25 micrometer directions under a 13 degree tilt, will help choose the geometry of the future ATLAS and CMS pixel layers.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Track-resolution subtraction is calibrated to 3.85 μm, but the same data give 3.35 μm from unbiased residuals; this ambiguity is not propagated into Table 1, so the headline 10.86 and 6.81 μm resolutions are not yet robust.","rationale":"The reader's weakest assumption correctly identifies the simulated track-resolution subtraction as load-bearing. My concern pinpoints an internal inconsistency within that subtraction: the Mimosa26 intrinsic resolution is 3.85 μm from biased residuals but 3.35 μm from unbiased residuals, and the paper does not propagate this 0.5 μm ambiguity into the final uncertainties. This is a concrete, checkable weakness in the central claim. It does not overturn the measurement, because the non-tilted values agree with pitch/sqrt(12) and the tilted improvement is qualitatively visible in the residual plots, but it does mean the numerical values in Table 1 should be treated as conditional on the telescope calibration. The reader's CONDITIONAL verdict remains appropriate; no adjustment is needed.","tokens_in":6568,"tokens_out":7597,"duration_ms":73291,"concrete_test":"Recompute Table 1 with the Mimosa26 intrinsic resolution fixed to 3.35 μm instead of 3.85 μm, rerunning the GBL Track Resolution Calculator and the Eq. (2) subtraction, and compare the resulting DUT intrinsic resolutions and tilted/non-tilted ratios to Table 1. If the tilted 50×50 value shifts by more than the quoted ±1.09 μm, or the tilted 25×100 value shifts by more than its ±1.82 μm uncertainty, the headline resolution claims are not robust to the telescope calibration choice. As a cross-check, refit with the 4.5 μm hit-error assumption in the track fit replaced by the fitted 3.85 or 3.35 μm values to test whether Eqs. (1)–(2) are being applied consistently.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The claimed intrinsic resolutions in Table 1 are not directly measured: they are obtained from Eq. (2) by subtracting a simulated track resolution from the measured unbiased residual on each DUT. The track resolution is computed with the GBL Track Resolution Calculator [15], whose input includes the intrinsic resolution of the Mimosa26 telescope planes. That intrinsic resolution is determined in Eq. (4) by fitting the biased residuals from the same data, giving 3.85 μm. The paper's own systematics bullet immediately after Table 1 states that the unbiased residuals give an alternative estimate of 3.35 μm. The two values differ by 0.5 μm, which is a substantial fraction of a 3–4 μm telescope resolution, and the paper does not quantify the resulting shift in the DUT resolutions. Because Eq. (2) is a quadratic subtraction, an error in the track resolution propagates directly into every DUT number; the effect is largest for the 25 μm direction, where the reported 6.81 ± 1.82 μm already overlaps the non-tilted value. The 3.85/3.35 μm discrepancy also suggests that the simple relationship between biased residuals, intrinsic resolution and track resolution may not be exact under the actual hit-error assumptions used in EUTelescope (4.5 μm for Mimosa26), which is another unquantified source of error in the subtraction.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports a testbeam measurement of the intrinsic position resolution of two RD53A pixel modules, one with 50×50 µm² pitch and one with 100×25 µm² pitch, using an 11 GeV electron beam at SLAC. Tracks are reconstructed with the GBL fitter in EUTelescope using the CALADIUM Mimosa26 telescope. The DUT resolution is extracted from the unbiased residual width after subtracting a simulated track resolution, and the Mimosa26 intrinsic resolution is calibrated from the biased residuals of the telescope planes. The authors report non-tilted resolutions close to pitch/√12 and improved resolutions for a 13-degree tilt: 10.9 µm for the 50 µm direction and 6.8 µm for the 25 µm direction.","tokens_in":6801,"tokens_out":13139,"duration_ms":121202,"significance":"The measurements are relevant for the HL-LHC pixel detector development, and the non-tilted results confirm the expected binary-readout resolution. The paper includes a useful consistency check in Fig. 8 and a derivation of the biased/unbiased residual relations in Appendix A. However, the central numbers rely on a simulated track-resolution subtraction whose input is calibrated on the same data, and the uncertainties from that calibration are not propagated into the final results. The claimed 14% improvement in the 25 µm direction is not statistically significant. With additional analysis and clearer reporting, the result could become a solid reference for RD53A characterization.","major_comments":[{"comment":"The determination of the Mimosa26 intrinsic resolution is not propagated into the DUT resolutions. The parabola fit in Fig. 7 gives the minimum at 3.849 µm with a statistical uncertainty of ±3.368 µm (as printed on the figure), yet the text takes 3.85 µm as a fixed value. The systematics bullet also states that the unbiased residuals give 3.35 µm. Because Eq. (2) subtracts the simulated track resolution, which scales with this input, the DUT resolutions in Table 1 (especially the 6.81 µm tilted 25 µm value, whose error bar already exceeds its separation from the non-tilted value) are missing a potentially large source of uncertainty. Please repeat the extraction using 3.35 µm and using the ±3.4 µm range, and add the resulting variation to the systematic uncertainties in Table 1.","section":"3 (Eq. (4) and Fig. 7)"},{"comment":"The residual relations used to calibrate and subtract the track resolution assume that the track fit uses the true intrinsic resolution of each plane as its hit error (see Appendix A). In the actual analysis, EUTelescope uses a fixed 4.5 µm hit error for the Mimosa26 planes, while the fitted intrinsic resolution is 3.85 µm. The track-resolution simulator is run with the scanned intrinsic resolution, not with the 4.5 µm value used in the fit. The paper does not demonstrate that Eqs. (1) and (2) remain correct when the fit errors differ from the true resolutions. Please either verify with a Monte Carlo or simulation that this mismatch has negligible impact, or modify the simulation to use the same hit-error assumptions as the data fit.","section":"3 and Appendix A (Eqs. (1), (2))"},{"comment":"The claimed improvement in the 25 µm direction is not statistically significant. Table 1 gives the ratio of tilted to non-tilted resolution as 0.86 ± 0.30 for the 25 µm direction, which is consistent with 1.0 within one standard deviation. The abstract and conclusion state that a 6.8 µm resolution 'can be achieved' and that the tilted modules improve the 25 µm direction by 14%, without quoting this uncertainty. Please qualify these statements and report the uncertainty on the improvement.","section":"4 and Table 1"}],"minor_comments":[{"comment":"The word 'teatbeam' appears in 'was used in SLAC teatbeam'; it should read 'testbeam'.","section":"2"},{"comment":"The phrase 'for the the HL-LHC' contains a duplicated definite article.","section":"1"},{"comment":"State explicitly whether the box width d in Eq. (3) is fixed to the known pitch or is a free parameter, and report the fitted values of d and Σ for the DUTs.","section":"3 (Eq. (3))"},{"comment":"Indicate whether the quoted uncertainties are statistical only or include the systematic effects listed in the text, and show the breakdown of statistical and systematic components.","section":"Table 1"},{"comment":"Reference [15] for the GBL Track Resolution Calculator should include a full citation with the URL or DOI for the Zenodo record.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a short conference proceeding. The main technical concern is that the central values depend on a track-resolution subtraction whose input calibration uncertainty is not propagated; this is fixable with additional analysis. In addition, the claim for the 25 µm tilt improvement is not supported by the quoted errors. I see no indication of misconduct or citation issues."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a useful, workmanlike testbeam measurement of RD53A module position resolution, and the headline numbers are probably right, but the error budget has a hole the authors themselves point at without closing. Still worth refereeing.\n\nWhat's new: measured intrinsic resolution for 50x50 and 25x100 µm pitch RD53A modules, with and without a 13° tilt, at 11 GeV. Non-tilted values match pitch/sqrt(12) (14.51 vs 14.4, 28.16 vs 28.8), which is a good sanity check. The tilted values 10.86 and 6.81 µm, with claimed 26% and 14% improvement, are new numbers the ITk community will use for layout studies.\n\nWhat it does well: the write-up is concise and the reconstruction chain (EUTelescope, GBL, Millepede II) is standard but competently applied. The appendix derivation of the biased/unbiased residual relations is simple and correct, and it helps the reader see exactly what is being subtracted. The χ2 scan for the Mimosa26 intrinsic resolution is a reasonable approach, and the consistent biased-residual predictions in Fig. 8 are good internal evidence.\n\nWhere it's soft: the final DUT resolutions come from subtracting a simulated track resolution from measured unbiased residuals. That track resolution depends on the Mimosa26 intrinsic resolution, which is fitted to the same data. The paper admits that using unbiased residuals gives 3.35 µm instead of 3.85 µm for that intrinsic resolution, but it never propagates that 0.5 µm shift through Eq. (2) into Table 1. This matters most for the 25 µm direction, where the reported 6.81 ± 1.82 µm already sits close to the non-tilted 7.92 ± 1.73 µm; a 0.5 µm change in the telescope resolution could move the tilt benefit. This isn't fatal, but it is an unquantified systematic on the paper's main result.\n\nMinor: the tilted-pitch model (22.5/27.5 µm equivalent pitch) is asserted with a cartoon and not derived, and no data, configs, or code are released. For a conference proceedings this is understandable, but a full paper should include the material model and a reproducibility artifact.\n\nVerdict: worth serious peer review. The measurement is honest and the central sanity checks pass; the authors just need to close the 3.85/3.35 loop and give a proper systematic band on the tilted numbers. I would bring it to a pixel-detector reading group and would cite the numbers if working on ITk layout studies.","headline":"A workmanlike RD53A testbeam measurement whose headline numbers are probably right, but the 3.85 vs 3.35 µm Mimosa26 ambiguity needs to be propagated into the final error budget.","tokens_in":7377,"tokens_out":2098,"would_cite":true,"duration_ms":20408,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["29.40.Gx"],"model":"deepseek-v4-flash","headline":"A 13-degree tilt improves RD53A pixel resolution to 10.9 and 6.8 micrometers in a test beam.","keywords":["position resolution","RD53A","pixel detector","test beam","tilt angle","binary readout","track resolution","HL-LHC"],"falsifier":"Redo the analysis with a reference telescope whose track resolution is independently known to be much smaller than the expected DUT resolution, so the subtraction in Eq. (2) becomes negligible; if the directly measured intrinsic resolutions differ from 10.9 and 6.8 micrometers by more than the stated uncertainties, the material model or the fitted Mimosa26 resolution is wrong. Alternatively, recompute the DUT values using the 3.35 micrometer intrinsic Mimosa26 resolution obtained from unbiased residuals; a shift exceeding the quoted systematic uncertainties would indicate that the single-common-resolution assumption is inadequate.","tokens_in":6331,"feed_emoji":"🎯","tokens_out":8185,"duration_ms":71214,"temperature":0.7,"pith_summary":"This paper measures the intrinsic position resolution of RD53A pixel modules, the prototype detectors for the upgraded LHC tracking systems, using an 11 GeV electron test beam. It reports that tilting the modules by 13 degrees improves the short-pitch direction: the 50x50 module reaches 10.9 micrometers in the tilted 50-micrometer direction, and the 25x100 module reaches 6.8 micrometers in the tilted 25-micrometer direction, gains of roughly 26% and 14% over the non-tilted values. The resolutions are derived by subtracting a simulated track resolution from measured unbiased residuals, using a fitted intrinsic resolution for the telescope's reference sensors. If the result is correct, the RD53A geometry will meet the position-resolution needs of the HL-LHC innermost pixel layers.","feed_headline":"Pixel tilt sharpens tracker resolution to 6.8 microns","feed_subtitle":"A 13-degree tilt on RD53A modules cuts the 25-micrometer error by 14 percent, guiding the next pixel-layer geometry.","key_machinery":"The load-bearing identity is the unbiased-residual relation $\\sigma_{\\rm unbiased}^2 = \\sigma_{\\rm intrinsic}^2 + \\sigma_{\\rm track}^2$, which converts the measured residual width into the desired intrinsic resolution. The track resolution on each telescope plane is supplied by a simulation of the same geometry and material; the unknown intrinsic resolution of the six reference sensors is fixed by scanning a range of values and minimizing a $\\chi^2$ against the biased residuals on those planes, yielding 3.85 micrometers. The 13-degree tilt works geometrically, reducing the effective pixel pitch as seen by the incoming particle, and this shrinking projected pitch is what drives the reported improvement.","core_discovery":"The central measurement is that non-tilted RD53A modules with binary-readout clusters perform close to the ideal pitch/√12 limit, and that a 13-degree tilt effectively shrinks the projected pixel pitch, improving the intrinsic resolution of the 50-micrometer direction from 14.51 to 10.86 micrometers and of the 25-micrometer direction from 7.92 to 6.81 micrometers. The quoted resolutions include systematic uncertainties; the ratios of tilted to non-tilted resolution are 0.74±0.09 and 0.86±0.30 for the two directions. The authors establish these values by measuring unbiased residuals on the device under test and using the identity σ²_unbiased = σ²_intrinsic + σ²_track, with the track resolution obtained from a simulation tuned to the telescope geometry and a fitted common intrinsic resolution of 3.85 micrometers for the six reference sensor planes.","pith_inferences":["Because the measured gain from tilting is close to the geometric foreshortening of the projected pitch, larger tilt angles would plausibly improve resolution further until charge sharing and cluster fragmentation degrade the binary centroid estimate; the paper does not test this limit.","The difference between the fitted common intrinsic resolution of 3.85 micrometers (from biased residuals) and the 3.35 micrometers obtained from unbiased residuals suggests that a per-plane variation may be hidden; if propagated, this could widen the quoted DUT uncertainties.","The same subtraction method could serve as a general characterization pipeline for other pixel geometries, allowing different sensor pitches and thicknesses to be compared for future collider detectors."],"forward_implications":["Non-tilted RD53A modules already operate near the ideal pitch/√12 limit, so binary readout does not degrade the position measurement.","A 13-degree tilt improves the short-pitch direction resolution to 10.9 micrometers (50-micrometer direction) and 6.8 micrometers (25-micrometer direction), reductions of 26% and 14%.","RD53A modules can satisfy the position-resolution requirements for the HL-LHC innermost pixel layers, informing the geometry of the phase-II upgrades.","The methodology of unbiased residuals with simulated track-resolution subtraction transfers directly to future pixel-module test-beam campaigns."],"supporting_citations":[{"why":"Supplies the simulated track resolutions used to correct the measured unbiased residuals in Eq. (2).","marker":"[15]"},{"why":"Establishes the performance of the EUDET-type reference telescope used for track reconstruction.","marker":"[5]"},{"why":"Describes the Mimosa26 sensor whose intrinsic resolution is fitted in the analysis.","marker":"[6]"},{"why":"Provides the reconstruction chain that converts raw beam-test data into aligned hits and tracks.","marker":"[7][8][9]"},{"why":"Gives the track-fitting algorithm used during alignment and final track reconstruction.","marker":"[10]"},{"why":"Performs the global alignment of the telescope planes.","marker":"[14]"},{"why":"Defines the RD53A integrated circuit that is the device under test.","marker":"[2]"}],"fun_headline_variants":["Tilt improves pixel resolution to 6.8 microns","13-degree tilt cuts pixel error to 6.8 microns","RD53A tilt sharpens resolution to 6.8 microns","Pixel modules tilt boosts resolution to 6.8 microns","Tilted pixels hit 6.8-micron resolution"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The final numbers rest on subtracting a simulated track resolution from the measured residual width; that simulation assumes the test-beam material and beam energy are known to about 10% and that all six reference sensors share one fitted intrinsic resolution of 3.85 micrometers. If those assumptions are wrong, the subtraction changes and the quoted 10.9 and 6.8 micrometer values shift.","fun_headline_variants_meta":{"raw":{"variants":["Tilt improves pixel resolution to 6.8 microns","13-degree tilt cuts pixel error to 6.8 microns","RD53A tilt sharpens resolution to 6.8 microns","Pixel modules tilt boosts resolution to 6.8 microns","Tilted pixels hit 6.8-micron resolution"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000702,"raw_usage":{"total_tokens":3133,"prompt_tokens":872,"completion_tokens":2261,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":488,"completion_tokens_details":{"reasoning_tokens":2189}},"tokens_in":488,"tokens_out":2261,"duration_ms":16031,"temperature":1.0,"reasoning_tokens":2189,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T10:28:20.372217+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Redo the analysis with a reference telescope whose track resolution is independently known to be much smaller than the expected DUT resolution, so the subtraction in Eq. (2) becomes negligible; if the directly measured intrinsic resolutions differ from 10.9 and 6.8 micrometers by more than the stated uncertainties, the material model or the fitted Mimosa26 resolution is wrong. Alternatively, recompute the DUT values using the 3.35 micrometer intrinsic Mimosa26 resolution obtained from unbiased residuals; a shift exceeding the quoted systematic uncertainties would indicate that the single-common-resolution assumption is inadequate.","supporting_citations":[{"cited_title":"Spannagel and H","cited_arxiv_id":null,"evidence_quote":"Supplies the simulated track resolutions used to correct the measured unbiased residuals in Eq. (2)."},{"cited_title":"Baudot et al., First test results Of MIMOSA-26, a fast CMOS sensor with integrated zero suppression and digitized output, IEEE Nucl","cited_arxiv_id":null,"evidence_quote":"Describes the Mimosa26 sensor whose intrinsic resolution is fitted in the analysis."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the track-fitting algorithm used during alignment and final track reconstruction."},{"cited_title":"Blobel et al., Fast alignment of a complex tracking detector using advanced track models, Comput","cited_arxiv_id":null,"evidence_quote":"Performs the global alignment of the telescope planes."}],"review_version":1}