{"id":"ec12ef77-f2f1-4786-9b40-f21730c0c6c7","arxiv_id":"1909.00947","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"low","formal_verification":"none","parameter_count":4,"one_line_summary":"A 13-micrometer cutoff HgCdTe array using Teledyne's Design 2 pixel structure keeps median dark current at 1.8 electrons per second and 90.3 percent operability at 28 K with 350 mV bias.","lead":"Engineers tested four new 13-micrometer infrared detector arrays and found that one pixel design works at warmer temperatures with low dark current and large well depth. This brings future space telescopes closer to using passively cooled long-wavelength detectors instead of heavy cryogens.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Calibration-chain uncertainty is the load-bearing risk: the 1.8 e-/s, 81 ke-, and 90.3% numbers scale with the Section 4.1 capacitance/IPC/nonlinearity conversion, which lacks error bars or raw data.","rationale":"The reader's weakest assumption already identifies the calibration chain as the key fragility, and my read agrees. The single-array limitation is real but does not undermine the empirical measurement for H1RG-18509 itself; it only limits generalization of the Design-2 attribution. The dark-current model fits are explicitly fits and are not the basis for the headline numbers. The paper's direct measurement uses standard, well-described methods and is credible, so I do not change the ACCEPT verdict. The clearest soft spot is the absence of uncertainty propagation and raw data for the headline numbers, which leaves the calibration chain as the load-bearing but unverified step.","tokens_in":19367,"tokens_out":18506,"duration_ms":193774,"concrete_test":"Re-reduce the H1RG-18509 data with an independent calibration at 350 mV: obtain the node capacitance from a photon-transfer curve using the full 2-D IPC covariance model instead of the 1-8 alpha approximation, and recompute the nonlinearity correction from dark-subtracted SUTR flat-field ramps over the full bias range. Then recompute the median dark current, median well depth, and the fraction of pixels with dark current below 200 e-/s and well depth at or above 75 ke-. If the median well depth remains within 75-85 ke- and operability stays above 90%, the concern is settled; if not, the headline claim should be revised.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central numbers in Section 4.4.4 and the Abstract, median dark current 1.8 e-/s, median well depth 81 ke-, and 90.3% operability at 28 K and 350 mV, are all proportional to the electron-conversion chain of Section 4.1. For H1RG-18509 that chain is: noise-squared node capacitance with a median 34 fF at 350 mV (Table 2), IPC correction by 1-8 alpha with alpha = 1.12%, and a nonlinearity slope fitted between 20% and 80% of saturation. The Table 9 well-depth operability threshold is set in volts (355 mV) and translated as approximately 75 ke- using that median capacitance; a 10% error in capacitance or nonlinearity slope shifts the threshold by about 7-8 ke- and, because the threshold sits near the median of the well-depth distribution, can move the 90.3% operability by several percent. The dark-current slope in e-/s uses the same conversion factor, so the 1.8 e-/s number carries the same systematic risk. The paper gives no error bars on these headline values and no raw data or code, so the size of this risk cannot be assessed from the text. The methods are standard and there is no demonstrated error, but this is the least secure link in the central claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper reports the characterization of four 1024x1024, 13 micron cutoff HgCdTe detector arrays from Teledyne Imaging Systems, spanning three pixel designs (standard, Design 1, Design 2). The authors measure source-follower gain, node capacitance via the noise-squared method with interpixel-capacitance correction, signal nonlinearity, CDS read noise, and dark current and well depth per pixel using Sample-Up-The-Ramp acquisition. The central empirical result, stated in the abstract and developed in Section 4.4.4, is that for array H1RG-18509 (Design 2) at 28 K and 350 mV applied reverse bias the median dark current is 1.8 e-/s, the median well depth is 81 ke-, and 90.3% of pixels meet the operability thresholds of dark current below 200 e-/s and well depth above about 75 ke-. The other three arrays are largely inoperable at 350 mV, which the authors attribute to band-to-band and trap-to-band tunneling. The paper also presents dark-current model fits (diffusion, generation-recombination, band-to-band tunneling, and trap-to-band tunneling with a soft-breakdown activation model) and discusses a multiplexer glow that affects some data sets.","tokens_in":19633,"tokens_out":6443,"duration_ms":70952,"significance":"If the empirical result holds, it is a meaningful advance for long-wavelength HgCdTe detector technology: it suggests that a modified pixel structure can suppress quantum tunneling dark current enough to make 13 micron cutoff arrays usable at 350 mV bias and passively coolable temperatures, with clear implications for the ongoing push to 15 micron cutoff devices and for ground- and space-based LWIR astronomy. The paper's strengths are its use of standard SUTR and well-depth measurement techniques, the internal consistency of the headline numbers across Tables 9-10 and the low-curvature histogram in Figure 15, and its transparent discussion of mux glow and of the non-uniqueness of the tunneling model fits. The main weakness is that the absolute electron-scale numbers carry no reported calibration-chain uncertainties, and the causal attribution to Design 2 rests on a single array.","major_comments":[{"comment":"The headline numbers (median dark current 1.8 e-/s, median well depth 81 ke-, operability 90.3%) are directly proportional to the electron-conversion chain: noise-squared node capacitance, the 1-8*alpha IPC correction, and the fitted nonlinearity slope. No uncertainties are given for any step of this chain. In particular, the 350 mV operability threshold in Table 9 is set in volts and converted to approximately 75 ke- using the median 34 fF capacitance from Table 2; a 10% error in capacitance or in the nonlinearity slope shifts that threshold by about 7-8 ke-, and because the threshold sits near the median of the well-depth distribution it can move the quoted operability by several percent. Please provide error bars on the calibration chain and a sensitivity analysis, for example recomputing median dark current, median well depth, and operability under +/-10% variations in capacitance, alpha, and nonlinearity slope.","section":"Section 4.1 and Tables 9-10"},{"comment":"The central causal claim that the Design 2 pixel structure mitigates quantum tunneling dark current is based on a single array, H1RG-18509, from one lot split. With n=1 per pixel design and no replication across wafers or lots, the improved performance cannot be conclusively separated from sample-to-sample variability. The text should state this limitation explicitly and, if available, report PEC or mini-array data from additional Design 2 samples, or at least outline a reproducibility plan for the next development lot.","section":"Section 4.4.4 and Table 1"}],"minor_comments":[{"comment":"The operability well-depth thresholds are quoted in both mV and ke- but the conversions are not numerically consistent across arrays (for example, 155 mV corresponds to approximately 41 ke-, 38 ke-, and 37 ke- in different tables). Please state explicitly that each electron threshold is derived from that array's measured capacitance, so the apparent inconsistency is understood as a calibration difference rather than an error.","section":"Section 1.3.1 and Tables 3, 5, 7, 9"},{"comment":"For H1RG-18367, all dark-current measurements were affected by the inferred multiplexer glow, so the reported median dark currents should be labeled in the main text as upper limits on the detector dark current rather than measured detector dark current; the table footnotes already say this, but the main-text discussion should be equally explicit.","section":"Section 4.2 and Tables 3-4"},{"comment":"The model fits use 36 or 50 selected operable pixels and the text acknowledges that the trap-to-band parameters may not be unique. The mechanistic conclusions, such as band-to-band dominance at high bias, should carry this uncertainty more prominently; bootstrap uncertainties or parameter covariance estimates would strengthen the presentation.","section":"Section 5.3-5.4"},{"comment":"The curvature histograms are the key visual evidence that H1RG-18509 has nearly linear discharge behavior, but the four histograms are overlaid without a legend and with no reported sample sizes. Adding a legend and pixel counts would make the comparison substantially easier to assess.","section":"Figure 15"},{"comment":"There is a typo in the first sentence: 'the the dark current Arrhenius plot' should read 'the dark current Arrhenius plot'.","section":"Section 5.4"}],"recommendation":"major_revision","confidential_remarks":"The empirical work appears sound and the headline result is potentially important, but the missing error budget on the calibration chain is a genuine load-bearing gap for a paper whose abstract quotes absolute electron values, and the Design 2 conclusion rests on a single device. I think a revision that adds a calibration sensitivity analysis and an explicit single-sample limitation discussion would address my concerns; I do not see grounds for rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThis is a straightforward, useful detector development report. The genuinely new thing is the first comparison of three Teledyne pixel designs at 13 µm cutoff, with the Design-2 array H1RG-18509 showing median dark current 1.8 e-/s, median well depth 81 ke-, and 90.3% operability at 28 K and 350 mV reverse bias. The other three arrays collapse at 350 mV from tunneling current, so this is a meaningful step toward passively cooled LWIR astronomy.\n\nThe paper does several things well. The measurement chain is standard: SUTR dark curves, noise-squared capacitance, IPC correction, non-linearity correction. The curvature histograms in Fig. 15 and the discharge curves give direct visual support that Design-2 pixels are not suffering the runaway tunneling seen elsewhere. The dark-current modeling is explicitly a fitting exercise, not a prediction, so the headline numbers do not depend on the theory. The treatment of mux glow is careful, and the cross-hatching analysis is a nice bonus.\n\nThe soft spots are the ones you would expect. The headline electron numbers are proportional to the Section 4.1 calibration chain, and there are no error bars on the capacitance, IPC alpha, or non-linearity slope. A 10% systematic error would shift the 75 ke- threshold by about 7-8 ke- and could move the 90.3% operability by a few percent. I do not see a demonstrated error, and the methods are standard, but the uncertainty should have been propagated or at least bounded. Second, the Design-2 result comes from a single array, so wafer-to-wafer variation is uncontrolled. Third, some arrays have mux glow contamination; the authors handle it honestly, but it adds noise to parts of the I-T comparison.\n\nThe citation pattern is heavy on the Rochester team's own prior work, but that is justified: they are the group that built and tested these devices, and the NEOCam lineage is directly relevant. No invented entities, no hand-waving.\n\nOverall, the central claim holds up: H1RG-18509 mitigates band-to-band tunneling enough to be useful at 350 mV. This is not a reorganisation of the field, but it is exactly the kind of progress a mission needs. I would accept it with minor revisions; at minimum, add error bars or a sensitivity analysis for the conversion chain, and say explicitly that this is a single-array demonstration. A serious referee time is warranted.\n\nI would bring it to reading group only if there is instrumentation interest. I would cite it as evidence of the 13 µm state of the art.","headline":"A solid, engineering-significant detector paper whose headline 1.8 e-/s at 350 mV rests on a standard but unquantified calibration chain, from a single array; worth review and publication with tighter error reporting.","tokens_in":20245,"tokens_out":3273,"would_cite":true,"duration_ms":28995,"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":"A modified pixel design suppresses quantum tunneling dark current in 13-micrometer-cutoff mercury cadmium telluride arrays, reaching 90% operability at 28 K and 350 mV bias.","keywords":["HgCdTe","long-wave infrared detectors","13 micron cutoff","dark current","quantum tunneling","well depth","operability","passively cooled space telescopes"],"falsifier":"Measure the well depth of the Design 2 array by integrating a calibrated photon flux of known rate to saturation and compare with the electron count from the noise-squared/IPC/non-linearity calibration; a disagreement beyond the quoted error would rescale the 81 ke- and 1.8 e-/s headline values. A second Design 2 array from a different wafer, tested at the same 28 K and 350 mV, would show whether the suppression reproduces.","tokens_in":19078,"feed_emoji":"🔭","tokens_out":9709,"duration_ms":90114,"temperature":0.7,"pith_summary":"This paper is an engineering study of four prototype 13-micrometer-cutoff mercury cadmium telluride infrared detector arrays, built to see whether long-wave detectors can run warm enough for passive cooling in space. It claims that one of three pixel designs, called Design 2, suppresses the quantum-tunneling dark current that otherwise makes large reverse bias unusable. On that array, at 28 K and 350 mV applied reverse bias, the median dark current is 1.8 $e^-$/s, the median well depth is 81 ke-, and 90.3% of pixels meet the 75 ke- well-depth and 200 $e^-$/s dark-current operability threshold. If the claim holds, it clears the main obstacle to extending the cutoff to 15 micrometers and to replacing bulky cryogens with passive cooling on future infrared space missions.","feed_headline":"Pixel redesign tames dark current in 13-micron infrared detectors","feed_subtitle":"At 28 K and 350 mV, 90% of pixels hold 75,000 electrons with 1.8 e-/s median dark current.","key_machinery":"The load-bearing object is the Design 2 pixel: a mercury cadmium telluride photodiode structure, proprietary to the manufacturer, intended to reduce quantum tunneling dark current, hybridized to a 1024x1024 readout circuit. Its effect is visible as nearly linear signal-versus-time curves and a curvature histogram peaking near zero for the Design 2 array at 350 mV, in contrast to the strongly curved discharge of the other arrays. The supporting theory is the triangular-barrier band-to-band tunneling expression, in which the only fitted quantity $\\beta = E_g^{3/2}/\\mathcal{E}$ (band gap and junction electric field) sets the exponential suppression; the paper fits $\\beta$ pixel by pixel to high-bias current-bias data and shows that operable pixels in all arrays follow it above about 200 mV.","core_discovery":"The central discovery is that a proprietary experimental pixel structure (Design 2) reduces band-to-band tunneling in a 12.6-micrometer-cutoff mercury cadmium telluride photodiode array enough that the array can be operated at 350 mV reverse bias without the dark-current blowup seen in the other three arrays. For the Design 2 array measured at 28 K, the median dark current is 1.8 $e^-$/s and the median well depth is 81 ke-, with 90.3% of pixels above 75 ke- well depth and below 200 $e^-$/s dark current. The three other arrays, including a second experimental design, show median dark currents above 200 $e^-$/s at the same bias and operability below 1%; their discharge curves are strongly curved, which the paper attributes to band-to-band tunneling. Fits of the measured current-versus-bias curves to a triangular-barrier tunneling model indicate that band-to-band tunneling dominates above roughly 200 mV in all arrays, and that Design 2 effectively raises the parameter $\\beta$ controlling that tunneling.","pith_inferences":["The headline result is single-array evidence; the natural next test is whether a second Design 2 array from a different wafer reproduces 90% operability and 1.8 $e^-$/s median dark current.","Because the absolute electron counts depend on the calibration chain, an independent measurement of node capacitance or non-linearity could rescale the well-depth numbers; the qualitative conclusion that Design 2 suppresses tunneling is more robust than the absolute values.","If the residual inoperable pixels trace to the cross-hatch misfit-dislocation pattern, growth-process changes that reduce those dislocations should push operability above 90% without further pixel redesign.","The paper's identification of band-to-band tunneling could be checked by measuring dark current versus temperature at fixed high bias, since the tunneling model predicts a specific band-gap dependence distinct from thermal currents and multiplexer glow."],"forward_implications":["At 28 K and 350 mV, a Design 2 array is usable, so a 13-micrometer-cutoff space instrument could be passively cooled, saving the mass and lifetime cost of cryogens.","The same pixel strategy should transfer to 15-micrometer-cutoff arrays, the stated next step, because it pushes the onset of band-to-band tunneling to higher bias.","Standard and Design 1 arrays are limited to roughly 250 mV reverse bias and the smaller well depths that come with it; at 350 mV their operability collapses.","In the low-bias, low-temperature regime where thermal dark current dominates, median dark currents below 1 $e^-$/s mean the arrays are suitable for low-background astronomy with linear calibration.","For applications needing larger well depth, a constant-voltage CTIA readout or a higher operating temperature sidesteps the non-linear tunneling-current regime."],"supporting_citations":[{"why":"It establishes the prior 10-micrometer HgCdTe array baseline that the 13-micrometer devices extend.","marker":"[13]"},{"why":"It supplies the operability requirements (dark current below 200 e-/s and about 40 ke- well depth) used as the benchmark.","marker":"[14]"},{"why":"It provides the diffusion dark current model fitted to the temperature data.","marker":"[18]"},{"why":"It provides the generation-recombination dark current model fitted to the temperature data.","marker":"[19]"},{"why":"It supplies the triangular-barrier band-to-band tunneling expression used for the high-bias fits.","marker":"[20]"},{"why":"It supplies the trap-to-band tunneling model used for inoperable pixels.","marker":"[22]"},{"why":"It introduces the threshold-voltage parametrization of soft breakdown and activated traps used to fit inoperable pixel curves.","marker":"[8]"},{"why":"It gives the noise-squared versus signal method used to derive nodal capacitance for the ADU-to-electron conversion.","marker":"[31]"},{"why":"It defines the interpixel capacitance coupling parameter and the 1-8alpha correction applied to the measured capacitance.","marker":"[32]"},{"why":"It provides the debiasing non-linearity correction and the saturation/well-depth measurement procedure.","marker":"[33]"}],"fun_headline_variants":["13-micron HgCdTe array tames dark current at 1.8 e-/s","Pixel tweak cuts tunneling dark current in long-wave IR detectors","13 μm detector array hits 1.8 e-/s dark current with new pixel design","Reduced tunneling dark current at 13 μm paves way to 15 μm arrays","New pixel design lowers dark current to 1.8 e-/s in 13-μm HgCdTe"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The headline well-depth and dark-current numbers rest on the calibration chain that converts detector voltage to electrons — noise-squared capacitance, an interpixel-capacitance correction, and a fitted non-linearity slope — so a systematic error in that chain would rescale every quoted value proportionally.","fun_headline_variants_meta":{"raw":{"variants":["13-micron HgCdTe array tames dark current at 1.8 e-/s","Pixel tweak cuts tunneling dark current in long-wave IR detectors","13 μm detector array hits 1.8 e-/s dark current with new pixel design","Reduced tunneling dark current at 13 μm paves way to 15 μm arrays","New pixel design lowers dark current to 1.8 e-/s in 13-μm HgCdTe"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001297,"raw_usage":{"total_tokens":5330,"prompt_tokens":1018,"completion_tokens":4312,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":634,"completion_tokens_details":{"reasoning_tokens":4197}},"tokens_in":634,"tokens_out":4312,"duration_ms":28495,"temperature":1.0,"reasoning_tokens":4197,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T05:31:32.626236+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the well depth of the Design 2 array by integrating a calibrated photon flux of known rate to saturation and compare with the electron count from the noise-squared/IPC/non-linearity calibration; a disagreement beyond the quoted error would rescale the 81 ke- and 1.8 e-/s headline values. A second Design 2 array from a different wafer, tested at the same 28 K and 350 mV, would show whether the suppression reproduces.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It establishes the prior 10-micrometer HgCdTe array baseline that the 13-micrometer devices extend."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It supplies the operability requirements (dark current below 200 e-/s and about 40 ke- well depth) used as the benchmark."},{"cited_title":"Reine, A","cited_arxiv_id":null,"evidence_quote":"It provides the diffusion dark current model fitted to the temperature data."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It provides the generation-recombination dark current model fitted to the temperature data."},{"cited_title":"Kinch, ``Chapter 7 M etal- I nsulator- S emiconductor I nfrared D etectors,'' in Mercury Cadmium Telluride , R","cited_arxiv_id":null,"evidence_quote":"It supplies the triangular-barrier band-to-band tunneling expression used for the high-bias fits."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It supplies the trap-to-band tunneling model used for inoperable pixels."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It introduces the threshold-voltage parametrization of soft breakdown and activated traps used to fit inoperable pixel curves."},{"cited_title":"Mortara and A","cited_arxiv_id":null,"evidence_quote":"It gives the noise-squared versus signal method used to derive nodal capacitance for the ADU-to-electron conversion."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It defines the interpixel capacitance coupling parameter and the 1-8alpha correction applied to the measured capacitance."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It provides the debiasing non-linearity correction and the saturation/well-depth measurement procedure."}],"review_version":1}