{"id":"cda1c28e-dd32-4d16-a7c4-79cb659f5e81","arxiv_id":"2411.12705","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Upgraded CLASS 90 GHz detectors reach 94% stability to 15% normal resistance, median telescope efficiency 0.65 on new wafers, and array NET 9.7 µK√s, about half the previous value.","lead":"This paper reports that swapping four of seven detector modules in the CLASS 90 GHz cosmic microwave background telescope for redesigned superconducting sensors improved detector stability and cut the array's noise-equivalent temperature from 19 to 9.7 microkelvin root-seconds. The upgrade also raised median telescope optical efficiency from 42% to 60%, roughly doubling how quickly the survey can integrate down to a given sensitivity.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claim that detector efficiency exceeds 0.90 rests on an assumed receiver optics absorption/reflection efficiency of 0.91; if the true optics loss differs, the headline detector-efficiency number changes materially and is not directly measured.","rationale":"The reader's weakest_assumption identifies exactly the same load-bearing concern: the detector-efficiency >0.90 claim relies on an assumed receiver optics efficiency. I considered other candidate concerns. The efficiency improvement from 0.42 to 0.60/0.65 is confounded by replacement of the four worst original wafers with upgraded ones, but the paper is transparent about this and separately reports the upgraded-wafer-only median, so it is a manageable caveat rather than a central flaw. The use of the simulated bandpass instead of the measured FTS bandpass is justified by an internal consistency check (the measured bandpass would imply telescope efficiency >1); this is conservative for the efficiency and NET claims rather than inflating them. The NET calculation uses the per-detector measured efficiency, so it is not affected by the optics decomposition. The single most load-bearing point is that the headline detector-efficiency number changes from about 0.92 to 0.83 if the absorption/reflection assumption is relaxed to the physically conservative value of unity, and to an unphysical value if the earlier 0.58 estimate is used. Since this is precisely an unmeasured, assumed quantity, the claim should be presented as conditional. The paper's other headline results are direct measurements and are credible. Therefore the CONDITIONAL verdict is appropriate and no verdict change is needed; the concern should be resolved by a direct optics-efficiency measurement before the detector-efficiency claim is taken at face value.","tokens_in":23153,"tokens_out":6550,"duration_ms":66243,"concrete_test":"Measure the receiver optics efficiency directly: place a calibrated cryogenic blackbody source of known radiometric temperature at the receiver input and compare the detected power increment to the expected Rayleigh-Jeans power for the measured passband and beam solid angle. Alternatively, measure the transmission and reflection of each window, IR filter, and lens at 90 GHz at operating temperature with a vector network analyzer, and remeasure the spill efficiency from near-field beam maps. If the derived epsilon_abs deviates from 0.91 by more than about 0.05, the 'detector efficiency exceeds 0.90' claim is not supported and should be restated as scenario-dependent or as an upper limit.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim that detector efficiency exceeds 0.90 is not directly measured. In Section 6, the telescope efficiency for the upgraded wafers is measured to be 0.65, but the detector efficiency is inferred as eta_det = eta_telescope / (epsilon_spill * epsilon_abs), using epsilon_spill = 0.78 from Iuliano (2020) and epsilon_abs = 0.91, which the text explicitly calls 'a plausible scenario' adopted because the earlier absorption/reflection estimate of 0.58 was deemed likely overestimated. With the assumed values, eta_det = 0.65/(0.78*0.91) ~ 0.92. If epsilon_abs were actually 1.0 (no loss), eta_det would be 0.83, below the claimed 0.90. If the old estimate of 0.58 were correct, the implied detector efficiency would exceed unity (1.44), which is unphysical — so either the old optics model or the measured telescope efficiency is biased. No direct measurement of epsilon_abs is presented; the paper relies on an argument that the earlier value was too pessimistic. This does not weaken the directly measured telescope efficiency (0.65), the stability result (94%), or the array NET (9.7 uK sqrt(s)), but it makes the '>0.90' detector-efficiency headline contingent on an unverified assumption rather than a measurement.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports the design, in-lab characterization, and on-sky performance of the upgraded CLASS 90 GHz focal plane, in which four of seven detector wafers were replaced during the 2022 austral winter. The design changes include revised magic-tee and crossover terminations, a direct normal-metal thermal contact between the MoAu TES and the Pd heat capacity, and revised bias-line filtering. The central reported results are: 94% of upgraded detectors are stable down to 15% of their normal resistance; the median telescope optical efficiency for the four upgraded wafers is 0.65 with a 68% interval of ±0.06; the full-array NET improved from 19 to 9.7 µK√s; and, under an assumed receiver optics efficiency, the implied detector efficiency exceeds 0.90. The paper also presents measurements of electrothermal parameters, bandpasses, noise, and yield, and discusses discrepancies with design targets.","tokens_in":23364,"tokens_out":6693,"duration_ms":69209,"significance":"If the results hold, this is a valuable instrument paper: the upgrade substantially improves the sensitivity and stability of a deployed CMB polarimeter, and the design changes are likely relevant to future TES bolometer arrays. The paper's strengths are its direct measurements: I-V curve stability statistics, laboratory and on-sky noise spectra, and Jupiter photometry. The 94% stability figure, the median telescope efficiency of 0.65 for the upgraded wafers, and the array NET of 9.7 µK√s are credible as reported. The main caveats are that the 'detector efficiency exceeding 0.90' headline is an inference resting on an assumed receiver-optics efficiency, and that the array-level improvement comparison includes a selection effect from replacing the poorest original wafers. These issues are fixable by reframing or additional analysis, and they do not undermine the directly measured performance gains.","major_comments":[{"comment":"The headline claim that detector efficiency exceeds 0.90 is not directly measured. The measured quantity is the telescope efficiency η = 0.65 from Jupiter observations. The detector efficiency is inferred as η_det = η / (ε_spill ε_abs) with ε_spill = 0.78 from Iuliano (2020) and ε_abs = 0.91, which the text in Section 6 explicitly calls 'a plausible scenario' and for which no independent measurement is presented. The result is highly sensitive to this assumption: if ε_abs = 1.0, then η_det ≈ 0.83, below the claimed threshold; if the older estimate ε_abs = 0.58 were used, the implied η_det would exceed unity, which is unphysical and indicates that either the older optics model or the measured telescope efficiency is biased. The Abstract and Conclusions should either present η = 0.65 as the direct measured result with the detector-efficiency value explicitly labeled as scenario-dependent, or provide a direct measurement of the optics absorption/reflection efficiency.","section":"Section 6, Eq. (4), Abstract"},{"comment":"The array-level improvements quoted in the Abstract compare the original seven-wafer array with a new array that contains the four upgraded wafers and only the three best original wafers. The paper itself notes in Section 7 that the retained originals have NEP ≈ 33 aW√s whereas the full original array had NEP ≈ 47 aW√s. Therefore the improvement from 19 µK√s to 9.7 µK√s, and similarly the telescope-efficiency improvement from 0.42 to 0.60, partly reflects the decision to replace the least-well-performing modules rather than the detector design change alone. The paper should separate the design improvement from the selection effect, for example by reporting the NET of the four replaced original wafers before and after the upgrade, or by stating plainly in the Abstract and Conclusions that the quoted array comparison includes replacement of the four poorest original wafers.","section":"Abstract, Table 3, Section 7"}],"minor_comments":[{"comment":"In the sentence 'Setting n = 3.77 would shift the average measured κ from 15.2 nW/K4 to the target 10 nW/Kn', the final unit 'nW/Kn' should read 'nW/K4' (or 'nW/K^n').","section":"Section 4.4"},{"comment":"The statement that 'telescope optical efficiencies are nearly all above 60%' is stronger than the data support: Table 3 gives wafer 1 a median efficiency of 0.61 with a 68% interval extending down to 0.21, and Figure 7 shows a substantial tail of upgraded detectors with efficiencies below 0.4. A phrase such as 'the median telescope efficiency is above 60%' would be accurate.","section":"Section 9"},{"comment":"The Abstract says 'Given our efficiency estimate for the receiver optics,' while Section 6 describes the corresponding input as 'a plausible scenario.' The language should be consistent and should make clear that the detector-efficiency value is conditional on an assumed optics model.","section":"Abstract vs Section 6"},{"comment":"The fitted parameter Poffset introduced in Section 7 does not appear in the parameter index in Table 4; it should be included with its estimation method so that the index is complete.","section":"Table 4"},{"comment":"The measured FTS bandpass shows a roughly 50% decrement above 100 GHz that the paper attributes to the test setup, and the simulated bandpass is preferred for the bandwidth Δν. This choice is stated in Section 5, but it is also load-bearing for Eq. (4) and Eq. (11); a cross-reference should be added at those points so that readers immediately see that the absolute efficiency calibration depends on the simulated bandwidth.","section":"Section 5 and Eq. (4)"}],"recommendation":"major_revision","confidential_remarks":"The direct measurements are credible and well matched to an ApJS instrument paper. The two major concerns are both fixable in revision: the detector-efficiency >0.90 claim should be reframed as scenario-dependent or supported by a direct optics-loss measurement, and the array-level improvement claims should be separated from the selection effect of replacing the worst wafers. I do not see a need for a fully new measurement campaign to address these issues."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"What you should know: this is the full characterization of the upgraded CLASS 90 GHz focal plane, and the main measured results hang together. The 94% stability fraction, the 0.65 median telescope efficiency for the four upgraded wafers, and the array NET improvement from 19 to 9.7 µK√s are direct measurements, well documented, and mutually consistent. The design changes themselves (stepped-impedance Nb-PdAu terminations, unified Pd heat capacity with direct MoAu contact, revised bias filters) appeared in the authors' N22/N23 conference papers; what is new here is the complete in-lab and on-sky validation, including yield statistics, spatial uniformity, intra-pair efficiency scatter, and the discussion of alternative electrothermal scenarios. That is a real contribution.\n\nThe soft spots are real but manageable. The most important is the claim that detector efficiency exceeds 0.90. That number is not measured; it is telescope efficiency (0.65) divided by an assumed receiver optics efficiency (spill 0.78, absorption/reflection 0.91), and the paper itself calls that a 'plausible scenario.' If absorption/reflection were actually 1.0, detector efficiency would be about 0.83; if the old 0.58 estimate were right, it would exceed unity, which is unphysical. The paper is transparent about this, but the >0.90 claim should be softened or given error bars. Relatedly, the headline improvement from 0.42 to 0.60 telescope efficiency compares the new array (upgraded wafers plus the three best original wafers) to the old full array. Part of that gain is selection. The retained original detectors have lower median efficiency, so the 0.42→0.60 number overstates the upgrade's effect; the fairer comparison is the upgraded wafers' 0.65 versus the old array's 0.42, which still looks good but should be stated cleanly.\n\nThe bandpass handling is a smaller issue: the measured FTS bandpass shows a high-frequency decrement the authors argue is a test artifact, and they use the simulated bandpass for efficiency and NEP calculations. The Jupiter cross-check and the second FTS give them some cover, and they say why, but it is a modeling choice layered on top of a measurement. The NEP model includes a fitted Poffset, but the reported NET comes from measured noise spectra, so that free parameter does not drive the headline. The alternative scenarios for the κ and heat-capacity discrepancies are clearly labeled, and the authors correctly note that the key observational parameters are scenario-independent.\n\nWho gets value: anyone building or upgrading TES arrays, and CMB instrument people generally. The yield, stability, and efficiency numbers will be cited. The citation pattern looks honest, with the design details properly traced to N22/N23 and D22. The paper deserves a serious referee. The issues are not fatal; they are about framing and transparency. I would send it to review, with the expectation of minor-to-moderate revision.","headline":"Solid detector-upgrade paper with credible measured gains; the '>0.90 detector efficiency' claim is the one number to treat with caution.","tokens_in":24081,"tokens_out":1992,"would_cite":true,"duration_ms":19697,"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":"Redesigned TES bolometers nearly double the sensitivity of the CLASS 90 GHz camera and show detector efficiency above 90 percent.","keywords":["transition-edge sensor","bolometer","90 GHz","cosmic microwave background","polarimetry","detector efficiency","noise-equivalent temperature","CLASS"],"falsifier":"Measure the receiver optics efficiency directly by placing a calibrated cryogenic blackbody source at the receiver window and comparing the detected power with that from a full-aperture source; if the product of spill and absorption/reflection efficiencies comes out significantly above 0.71, the inferred detector efficiency would drop below 0.90, whereas a lower product would push it higher.","tokens_in":22884,"feed_emoji":"🔭","tokens_out":3022,"duration_ms":30172,"temperature":0.7,"pith_summary":"This paper reports that a detector redesign successfully fixed the main weaknesses of the CLASS 90 GHz focal plane: unstable bias ranges and low optical efficiency. Four of seven detector wafers were replaced with versions featuring revised terminations, a unified thermal mass with direct metal contact to the TES, and stronger bias filtering. The upgrade made 94 percent of detectors stable down to 15 percent of normal resistance, raised the median telescope efficiency from about 0.42 to 0.65 among upgraded wafers, and halved the array's noise-equivalent temperature from 19 to 9.7 µK√s. These results matter because improved detector efficiency and stability translate directly into more sensitive measurements of the cosmic microwave background polarization.","feed_headline":"Upgraded 90 GHz detectors halve noise, hit 65% efficiency","feed_subtitle":"Four redesigned TES wafers push detector efficiency above 90 percent and widen the stable bias range across the CLASS focal plane.","key_machinery":"The central mechanism is a set of three coordinated changes to the TES bolometer pixel: (1) stepped-impedance Nb-to-PdAu terminations that replace the previous tapered gold terminations and avoid superconducting proximity effects; (2) a single contiguous palladium film that provides the bolometer heat capacity and is connected directly to the MoAu TES bilayer through a normal-metal contact, suppressing internal thermal fluctuation noise; and (3) revised bias-line filters that extend RF chokes onto the long support legs. These changes are what make the detectors both more efficient at absorbing incoming radiation and more stable in bias, and they carry the argument from design to measured performance.","core_discovery":"The paper establishes that the upgraded 90 GHz transition-edge-sensor (TES) bolometers achieve their design optical efficiency and much better stability than the original detectors. Based on observations of Jupiter and a modeled receiver optics efficiency (spill 0.78, absorption plus reflection 0.91), the measured telescope efficiency of 0.65 implies a detector efficiency exceeding 0.90. The upgrade also produced a wide, overlapping voltage bias range across each wafer, with 94 percent of detectors staying on their superconducting transition down to 15 percent of normal resistance, enabling nearly the entire array to be biased simultaneously. The array noise-equivalent temperature improved from 19 to 9.7 µK√s, a factor of two gain in sensitivity that the authors attribute mainly to lower dark NEP and higher optical efficiency.","pith_inferences":["The inference that detector efficiency exceeds 0.90 is sensitive to the assumed spill (0.78) and absorption/reflection (0.91) efficiencies; if the true optics efficiency is higher, the detector efficiency would be correspondingly lower, and if it is lower, the detector efficiency would be even higher than claimed.","The design choices validated here—stepped-impedance absorbing terminations and a direct normal-metal thermal link to the TES—are likely transferable to the 150/220 GHz CLASS detectors and to other kilopixel TES arrays that face the same proximity-effect and internal-fluctuation-noise issues.","The paper's alternative scenarios for reconciling measured thermal parameters (a 10 mK higher critical temperature or a thermal conductance index of 3.77 instead of 4) could be tested by re-measuring Tc with a calibrated thermometer or by directly measuring the short silicon beam's conductance; such a test would refine thermal design rules for ballistic phonon beams.","A direct calibration of the receiver optics throughput (e.g., placing a chopped source at the receiver window and comparing with a full-aperture source) would turn the inferred detector efficiency into a measured quantity and would also clarify whether the 0.78 spill estimate or the 0.91 absorption/reflection estimate needs revision."],"forward_implications":["If the claimed detector efficiency above 0.90 holds, then the remaining loss in telescope efficiency sits in the receiver optics, so replacing or upgrading the three original detector wafers should push the full array toward the 0.65 efficiency level and further reduce the array NET.","A wide common bias range across each wafer means nearly all detectors on a wafer can be biased at their optimal transition point, improving the effective yield and the fidelity of polarization maps without redesigning the readout.","The observed uniformity of optical efficiency across upgraded wafers indicates the microwave circuit design is robust to fabrication variation, simplifying production of additional wafers.","The lower dark NEP of the upgraded detectors, combined with higher optical efficiency, yields background-limited performance at lower optical loading, which is advantageous for observing the faint large-angular-scale CMB polarization.","The authors suggest that yield (around 60 percent) can be improved by routing bias leads to multiple wafer edges, a straightforward change that would increase the number of working detectors in future modules."],"supporting_citations":[{"why":"Provides the baseline performance of the original 90 GHz array (optical efficiency 0.42, NET 19 µK√s) that the upgraded detectors are compared against.","marker":"D22"},{"why":"Reports initial in-lab characterizations of the upgraded detectors, including electrothermal parameters and dark NEP that this paper extends.","marker":"N22"},{"why":"Describes the Jupiter observation procedure and preliminary optical efficiency results that this paper analyzes in full.","marker":"N23"},{"why":"Supplies the receiver optics efficiency estimates (spill 0.78; absorption and reflection 0.58 or 0.91 in scenarios) used to separate telescope efficiency into detector and optics parts.","marker":"Iuliano 2020"},{"why":"Provides the formalism for deriving thermal time constant and heat capacity from measured optical time constants and I-V curves.","marker":"Appel et al. 2022"},{"why":"Supplies the Jupiter brightness temperature at 90 GHz (172.8 K) used as the calibration standard for optical efficiency.","marker":"Bennett et al. 2013"},{"why":"Supplies the standard TES electrothermal relations (saturation power, thermal conductance, NEP) used throughout the parameter analysis.","marker":"Irwin & Hilton 2005"}],"fun_headline_variants":["CLASS 90 GHz upgrade: detector efficiency >90%, noise down to 9.7","New TES wafers push CLASS 90 GHz sensitivity up 2x","CLASS 90 GHz: 94% stable detectors, detector efficiency >90%","CLASS 90 GHz upgrade doubles sensitivity with redesigned TES"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The claim that detector efficiency exceeds 0.90 rests on an assumed split of the telescope efficiency into a spill efficiency of 0.78 and an absorption/reflection efficiency of 0.91; if the real receiver optics efficiency differs from this scenario, the implied detector efficiency changes.","fun_headline_variants_meta":{"raw":{"variants":["CLASS 90 GHz upgrade: detector efficiency >90%, noise down to 9.7","New TES wafers push CLASS 90 GHz sensitivity up 2x","CLASS 90 GHz: 94% stable detectors, detector efficiency >90%","CLASS 90 GHz upgrade doubles sensitivity with redesigned TES"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000363,"raw_usage":{"total_tokens":1975,"prompt_tokens":980,"completion_tokens":995,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":596,"completion_tokens_details":{"reasoning_tokens":912}},"tokens_in":596,"tokens_out":995,"duration_ms":9019,"temperature":1.0,"reasoning_tokens":912,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T17:14:08.716439+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the receiver optics efficiency directly by placing a calibrated cryogenic blackbody source at the receiver window and comparing the detected power with that from a full-aperture source; if the product of spill and absorption/reflection efficiencies comes out significantly above 0.71, the inferred detector efficiency would drop below 0.90, whereas a lower product would push it higher.","supporting_citations":[{"cited_title":"2020, PhD thesis, Johns Hopkins University","cited_arxiv_id":null,"evidence_quote":"Supplies the receiver optics efficiency estimates (spill 0.78; absorption and reflection 0.58 or 0.91 in scenarios) used to separate telescope efficiency into detector and optics parts."}],"review_version":1}