{"id":"c7aa521e-2264-4dd5-949f-d74d7527da12","arxiv_id":"1908.04845","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"TCAD models reproduce the high-power response saturation of TeraFET detectors and attribute it to gate leakage in AlGaAs/InGaAs and AlGaN/GaN HFETs and to avalanche and velocity saturation in Si MOSFETs.","lead":"This paper uses TCAD simulations to model terahertz detectors called TeraFETs and to reproduce their response saturation at high THz power. It attributes the saturation to gate leakage in III-V HFETs and to avalanche and velocity saturation in silicon MOSFETs.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Validation of the saturation mechanism rests on normalized, frequency-mismatched comparisons; without absolute agreement the mechanism attribution is not established.","rationale":"The paper's internal toggling of transport models, avalanche, velocity saturation, and barrier tunneling is a reasonable way to explore what the TCAD model can produce, and the simulations appear internally consistent in showing that certain mechanisms produce saturation. The load-bearing weakness is external validation: the comparison to experiment is normalized in the vertical direction and performed at a different frequency. This means the apparent agreement does not test the absolute response magnitude or the exact high-power onset, which are essential if the paper claims to explain the experimentally observed saturation. The reader's weakest assumption identified the same issue, and the proposed check directly tests whether the simulated saturation occurs at the same absolute signal levels as the measured one. Since the paper is already CONDITIONAL on this kind of validation, the appropriate stress-test outcome is to keep the verdict unchanged while sharpening the required test.","tokens_in":7494,"tokens_out":3767,"duration_ms":37052,"concrete_test":"Obtain the raw experimental data from But et al. (SPIE 2013) at 0.6 and 1.07 THz. Using the known or separately simulated frequency response of the detector, convert the measured absolute response to 0.3 THz and compare directly with the simulated drain response in Figs. 2, 4, and 6 without y-normalization. If the absolute simulated value at Va = 1 V differs from the frequency-scaled measurement by more than a factor of 3, or if a TCAD run at 0.6 THz shows that the saturation onset in Va shifts relative to the 0.3 THz run, then the mechanism attribution is not experimentally established.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The core claim (Section III) is that TCAD identifies gate leakage for the two HFETs and velocity saturation/avalanche for Si MOSFETs as causes of the experimentally observed response saturation. The only experimental support is Figs. 2, 4, and 6. Each caption states that 'the analytical and measured data are normalized to the range of the simulated results,' so the absolute detector response is never compared to experiment. The measured data are at 0.6 and 1.07 THz, while all simulations are at 0.3 THz. The TeraFET response and its saturation onset can depend on frequency through the plasma-wave impedance, gate-channel coupling, and rectification efficiency; normalizing away the vertical scale removes any test of whether the simulated magnitude or the Va at which saturation begins matches experiment. In addition, the Si MOSFET model is described as an 'exemplary' SOI device with default material parameters, not calibrated to the specific measured detector. The mechanism attribution is therefore built on a qualitative shape match of normalized curves, not on a validated quantitative reproduction of the measured high-power response.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript presents Synopsys Sentaurus TCAD models for three TeraFET detector types: AlGaAs/InGaAs HFETs, AlGaN/GaN HFETs, and SOI MOSFETs. The authors simulate the rectified drain response versus THz signal amplitude at 0.3 THz while toggling physical mechanisms (hydrodynamic vs. drift-diffusion transport, avalanche, velocity saturation, gate barrier tunneling) on and off. The simulated responses are compared with measured data from the literature at 0.6 and 1.07 THz and with an analytical theory. The central claim is that the saturation of the detector response at high intensities (above about 1 V) is caused by gate leakage in the two HFET material systems and by velocity saturation and avalanche effects in Si MOSFETs.","tokens_in":7725,"tokens_out":4934,"duration_ms":51824,"significance":"The systematic mechanism-toggling methodology is a genuine strength: it offers a transparent way to attribute a nonlinear response effect to specific physical mechanisms, and the spatially resolved current-density, electric-field, and impact-ionization profiles give mechanistic insight that goes beyond a simple circuit-level fit. If the validation were secure, the material-specific conclusions would be practically useful for high-power THz detection and for developing compact TeraFET models. However, the experimental validation as presented is weakened by vertical normalization of the measured and analytical curves to the simulated range, by the mismatch between simulated and measured frequencies, and by the use of an uncalibrated 'exemplary' Si MOSFET model. The mechanistic conclusions are plausible but are not yet established by the evidence shown.","major_comments":[{"comment":"The captions state that 'the analytical and measured data are normalized to the range of the simulated results.' This introduces a free vertical scaling factor for every comparison curve. With this normalization, no absolute detector response is ever compared with experiment, so the abstract's claim that the models are 'validated over a wide dynamic range' and the Section III claim that the models 'explain the experimentally observed response saturation' are not supported by the data as presented. The authors should plot absolute measured and simulated responses on a common scale, or provide an independently derived calibration factor and demonstrate that it is not tuned per curve.","section":"Figs. 2, 4, 6 captions"},{"comment":"All simulations are performed at 0.3 THz, while the measured data cited from [14] were taken at 0.6 and 1.07 THz. The TeraFET response magnitude, and in particular the onset and sharpness of saturation, can depend on frequency through the plasma-wave impedance and the gate-channel coupling. Because the vertical scale is normalized, the comparison cannot reveal whether the simulated frequency dependence is correct. Please simulate at the measured frequencies or justify quantitatively that the saturation onset is frequency-independent over this range.","section":"Figs. 2, 4, 6; Sections II-A, II-B, II-C"},{"comment":"The AlGaAs/InGaAs HFET model is stated to have been 'validated' by comparing simulated I-V characteristics and the gate-bias dependence of the THz response with measured data in [18], which is listed as an unpublished conference paper. The calibration parameters and validation plots are therefore not available for inspection in this manuscript. The present paper's validation chain depends on that reference, so the essential calibration results should be included in the paper or in a supplementary document.","section":"Section II-A and reference [18]"},{"comment":"The Si MOSFET model is explicitly described as an 'exemplary' SOI device using default material parameter files, not calibrated to the specific measured detector from [14]. The conclusion that velocity saturation and avalanche are responsible for the Si MOSFET response saturation is therefore not tied to the measured device. A sensitivity analysis over device parameters, or calibration to the measured device's I-V and response data, is needed to support the attribution for this material system.","section":"Section II-C"}],"minor_comments":[{"comment":"The phrase 'AlGaAs/InGaAs and AlGaN/GaN and silicon TeraFETs' should read 'AlGaAs/InGaAs, AlGaN/GaN, and silicon TeraFETs' for clarity.","section":"Abstract"},{"comment":"The legend text 'slope ~ Va s' and 'slope ~ V2 a' appears to contain formatting artifacts; these should read 'slope ~ V_a' and 'slope ~ V_a^2', respectively.","section":"Figs. 2, 4, 6 legends"},{"comment":"The sentence 'uses the same dimensions with the AlGaAs/InGaAs HFET TCAD model' should be 'uses the same dimensions as the AlGaAs/InGaAs HFET TCAD model.'","section":"Section II-B"},{"comment":"The x-axis label 'position (m)' with tick values around 3.0E-6 is technically meters, but the authors should confirm whether the intended unit is micrometers and make the axis label unambiguous.","section":"Figs. 3 and 5"},{"comment":"Reference [18] is listed as 'to be published'; if it has appeared by the time of resubmission, the citation should be updated with full publication details.","section":"Reference [18]"}],"recommendation":"major_revision","confidential_remarks":"The manuscript's validation claim depends on normalized comparisons and an unpublished calibration reference. In my view the mechanistic modeling is publishable if the authors can supply absolute measured-versus-simulated response curves, address the frequency mismatch, and either calibrate the Si MOSFET model or clearly reframe the Si result as a qualitative demonstration on an exemplary device. The editor may also wish to consider whether an unpublished conference paper should be the sole basis for model calibration."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know about arXiv:1908.04845. First, it is a useful simulation study: the authors build TCAD models for three TeraFET flavors and toggle physical mechanisms (hydrodynamic vs drift-diffusion transport, barrier tunneling, velocity saturation, impact ionization) to see which one produces the observed high-power response saturation. The result—gate leakage for the two HFETs, velocity saturation plus avalanche for the SOI MOSFET—is a plausible material-system-dependent attribution that is genuinely new as a simulation result. Second, the validation does not yet support the strength of the conclusion. Every experimental comparison in Figs. 2, 4, and 6 is vertically normalized to the simulated range, and the measurements were taken at 0.6 and 1.07 THz while the simulations run at 0.3 THz. That removes any test of absolute magnitude or the voltage at which saturation begins, and it lets a frequency-dependent response be matched by construction. The Si MOSFET model is an 'exemplary' device with default Sentaurus parameters, not the measured detector. The TCAD calibration is also delegated to an unpublished conference paper [18].\n\nWhat the paper does well: the simulations are internally consistent. The current-density profiles below the gate show larger gate leakage with hydrodynamic transport in both HFETs, and the electric-field and displacement-current profiles with and without the avalanche model give a concrete picture of how impact ionization could produce saturation in the MOSFET. That kind of model-toggling is a reasonable way to identify candidate mechanisms, and the authors are transparent about what is switched on and off.\n\nThe soft spots are real but proportionate. The normalization concern is the main one: it pre-conditions the agreement, so the paper should be read as demonstrating that the mechanisms can reproduce the saturation shape, not that the TCAD models quantitatively reproduce the measured high-power response. The frequency mismatch is related; since TeraFET response and saturation onset can depend on frequency through the plasma-wave coupling, simulating at the measured frequencies would strengthen the claim. The conclusion overstates by saying the models 'explain the experimentally observed response saturation' when the evidence is a shape match of normalized curves. A sentence of hedging would not fix it—the figures need to show absolute comparison.\n\nDespite these weaknesses, this is a serious paper for the TeraFET community. It deserves peer review, and a competent referee can ask for the missing comparisons. I would not cite it in my own work until the validation is tightened, but I would bring it to a reading group.","headline":"Plausible TCAD attribution of TeraFET saturation mechanisms, but the experimental validation relies on normalized, frequency-mismatched comparisons and needs reinforcement before the mechanism claim is settled.","tokens_in":8201,"tokens_out":2087,"would_cite":false,"duration_ms":20475,"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":"TCAD simulations trace TeraFET high-power saturation to material-specific mechanisms: gate leakage in III-V HFETs, velocity saturation and avalanche in silicon MOSFETs.","keywords":["TeraFET","terahertz detection","TCAD modeling","response saturation","gate leakage","hydrodynamic transport","avalanche effect","AlGaN/GaN HFET"],"falsifier":"Measure the drain response of the same AlGaAs/InGaAs or AlGaN/GaN detector at 0.3 THz over a wide power range without renormalizing the ordinate and compare absolute volts of response to the TCAD prediction; if the measured curve saturates at a different level than simulated, or if the simulated curve does not flatten when gate leakage is suppressed, the attributed mechanism is wrong. For the Si MOSFET, directly measuring the impact-ionization or substrate current during high-intensity illumination would test the avalanche claim.","tokens_in":7331,"feed_emoji":"⚡","tokens_out":3938,"duration_ms":38399,"temperature":0.7,"pith_summary":"This paper uses TCAD device simulations of three TeraFET material systems to establish why the detector response stops growing at high incident THz power, roughly above a 1 V signal magnitude. By switching physical mechanisms on and off in the simulations, it attributes the saturation to gate leakage current in AlGaAs/InGaAs and AlGaN/GaN HFETs, and to velocity saturation together with avalanche-generated carriers in Si MOSFETs. If correct, the result means that high-power roll-off is not a single universal effect but a set of material-dependent effects, and that TCAD models can cover the full dynamic range where the analytical Taylor-expansion response formula breaks down.","feed_headline":"Why TeraFET detectors saturate at high THz power","feed_subtitle":"III-V devices roll off through gate leakage; silicon devices roll off through velocity saturation and avalanche.","key_machinery":"The key machinery is the mechanism-toggle TCAD experiment: the same detector structure is simulated with hydrodynamic versus drift-diffusion transport, and with velocity saturation, avalanche, and gate barrier tunneling separately included or excluded. Hydrodynamic transport is the load-bearing ingredient for the HFET result, because it accounts for energy transport across heterointerfaces and thereby produces the gate leakage current that the drift-diffusion model misses. For the silicon MOSFET, the avalanche model plays that role: switching it off removes the enhanced displacement current near the drain and largely removes the saturation. Toggling these mechanisms is what connects each predicted saturation curve to a specific physical cause.","core_discovery":"The paper's central claim is that full-device TCAD models reproduce the experimentally observed saturation of the TeraFET drain response to intense THz radiation, and that the saturation mechanism differs by material system. In the III-V HFETs, hydrodynamic transport predicts substantially larger transverse electron current below the Schottky gate than drift-diffusion transport; this gate leakage rectifies the input THz signal and flattens the response, while the drift-diffusion model fails to saturate. In the SOI MOSFET, where gate leakage is blocked by the oxide, the paper identifies the avalanche (impact ionization) model together with velocity saturation as the cause: high THz fields generate carriers that enhance the displacement current near the drain, acting like a leakage path. The analytical theory, in contrast, is limited to an intermediate power range because it uses only the next terms of a Taylor expansion.","pith_inferences":["If gate leakage is the saturation knob in III-V TeraFETs, then a detector with a higher Schottky barrier or a gate insulator should push the saturation onset to higher THz powers; this design consequence is implied by the paper's mechanism but is not tested there.","The avalanche interpretation for the Si MOSFET predicts that the saturation threshold should depend on channel length and doping through the electric-field profile, so a series of devices with varied geometry could separate avalanche from velocity-saturation contributions experimentally.","A hybrid modeling strategy, analytical at low and intermediate power and TCAD for the high-power regime, could yield compact models valid over the whole dynamic range; the paper points toward such models but does not build them."],"forward_implications":["The analytical Taylor-expansion response formula is valid only over an intermediate power range and cannot reproduce high-intensity saturation, so TCAD is required for full dynamic-range TeraFET modeling.","Hydrodynamic transport, not drift-diffusion, is necessary to capture THz response saturation in III-V HFETs, because it includes energy transport across heterointerfaces and hence the gate leakage that causes the roll-off.","In Si MOSFETs, gate leakage is negligible, and the observed saturation is associated with avalanche-generated carriers and velocity saturation, with the gate oxide field staying below the breakdown limit.","The mechanism-by-mechanism TCAD approach attributes saturation to different device physics in different material systems, rather than to a single universal high-power effect."],"supporting_citations":[{"why":"Supplies the measured TeraFET response data showing saturation at high THz intensities that the TCAD models are compared against.","marker":"[14]"},{"why":"One of the prior proposed explanations of the saturation effect that the paper's mechanism attribution addresses.","marker":"[15]"},{"why":"Reports saturation of AlGaN/GaN HEMT photoresponse to intense THz radiation, providing experimental context for the GaN model.","marker":"[16]"},{"why":"Another report on AlGaN/GaN photoresponse to high-intensity THz radiation, part of the experimental background.","marker":"[17]"},{"why":"The previously validated TCAD model for TeraFET detectors on which the present simulation setup builds.","marker":"[18]"},{"why":"Supplies the hydrodynamic transport, avalanche, velocity-saturation, and barrier-tunneling models that are toggled on and off in the simulations.","marker":"[19]"},{"why":"Describes the 130 nm AlGaAs/InGaAs pHEMT process used to set the HFET model's structural parameters.","marker":"[20]"},{"why":"Provides the analytical high-intensity response theory whose limited range motivates the TCAD approach.","marker":"[21]"}],"fun_headline_variants":["TCAD reveals why TeraFETs saturate: gate leakage vs avalanche","TeraFET saturation: different physics for III-V and silicon","Model pinpoints TeraFET roll-off: leakage, velocity, avalanche","Why TeraFET response flattens: material-specific physics","TCAD models match TeraFET saturation across dynamic range"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The validation is established by normalizing the measured and analytical curves to the range of the simulated results, and the measurements were taken at 0.6 and 1.07 THz while the simulations run at 0.3 THz, so if that normalization hides large absolute discrepancies, the mechanism attribution built on the validation is unsupported.","fun_headline_variants_meta":{"raw":{"variants":["TCAD reveals why TeraFETs saturate: gate leakage vs avalanche","TeraFET saturation: different physics for III-V and silicon","Model pinpoints TeraFET roll-off: leakage, velocity, avalanche","Why TeraFET response flattens: material-specific physics","TCAD models match TeraFET saturation across dynamic range"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000594,"raw_usage":{"total_tokens":2733,"prompt_tokens":849,"completion_tokens":1884,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":465,"completion_tokens_details":{"reasoning_tokens":1794}},"tokens_in":465,"tokens_out":1884,"duration_ms":11786,"temperature":1.0,"reasoning_tokens":1794,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T13:30:19.271731+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the drain response of the same AlGaAs/InGaAs or AlGaN/GaN detector at 0.3 THz over a wide power range without renormalizing the ordinate and compare absolute volts of response to the TCAD prediction; if the measured curve saturates at a different level than simulated, or if the simulated curve does not flatten when gate leakage is suppressed, the attributed mechanism is wrong. For the Si MOSFET, directly measuring the impact-ionization or substrate current during high-intensity illumination would test the avalanche claim.","supporting_citations":[{"cited_title":"The dynamic range of THz broadband FET detectors,","cited_arxiv_id":null,"evidence_quote":"Supplies the measured TeraFET response data showing saturation at high THz intensities that the TCAD models are compared against."},{"cited_title":"Nonlinear photoresponse of field effect transistors terahertz detectors at high irradiation intensities,","cited_arxiv_id":null,"evidence_quote":"One of the prior proposed explanations of the saturation effect that the paper's mechanism attribution addresses."},{"cited_title":"Saturation of photoresponse to intense THz radiation in AlGaN/GaN HEMT detector,","cited_arxiv_id":null,"evidence_quote":"Reports saturation of AlGaN/GaN HEMT photoresponse to intense THz radiation, providing experimental context for the GaN model."},{"cited_title":"AlGaN/GaN HEMTs photoresponse to high intensity THz radiation,","cited_arxiv_id":null,"evidence_quote":"Another report on AlGaN/GaN photoresponse to high-intensity THz radiation, part of the experimental background."},{"cited_title":"An efficient TCAD model for TeraFET detectors,","cited_arxiv_id":null,"evidence_quote":"The previously validated TCAD model for TeraFET detectors on which the present simulation setup builds."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the hydrodynamic transport, avalanche, velocity-saturation, and barrier-tunneling models that are toggled on and off in the simulations."},{"cited_title":"A High- Performance 0.13-m AlGaAs/InGaAs pHEMT Process Using Sidewall Spacer Technology,","cited_arxiv_id":null,"evidence_quote":"Describes the 130 nm AlGaAs/InGaAs pHEMT process used to set the HFET model's structural parameters."},{"cited_title":"Plasmonic terahertz detector response at high intensities,","cited_arxiv_id":null,"evidence_quote":"Provides the analytical high-intensity response theory whose limited range motivates the TCAD approach."}],"review_version":1}