{"id":"8cccca1a-4dd1-4bcb-ba39-485354cc7c68","arxiv_id":"2608.00488","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Far-infrared photons heat the electronic subsystem of half-filled magic-angle twisted bilayer graphene, melting the correlated insulator and producing a bolometric response with internal voltage responsivity near 1.7x10^7 V/W.","lead":"Absorbing a tiny amount of far-infrared light heats the electrons in magic-angle graphene and melts its insulating state, collapsing the resistance by a large factor. This turns the material into an extremely sensitive bolometer for long-wavelength radiation, with response that survives in magnetic fields of several tesla.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Hot-electron/lattice decoupling ratio is extrapolated from a 200 mK reference device to T_e≈10 K in the main device; the paper's own δ≈5 scaling of G_e-ph makes the 100 mK lattice-heating bound questionable.","rationale":"The reader's weakest assumption identifies the same linchpin: the thermal decoupling hierarchy measured on a different device and extrapolated to the main device. I agree that this is the most load-bearing concern, because the central claim is not simply that MATBG is photoresistive, but that weak THz radiation selectively overheats the low-heat-capacity electronic subsystem and melts the correlated insulator. If the lattice heats substantially, the response is still bolometric but the specific hot-electron mechanism, and the heat-capacity explanation for the giant responsivity, collapse. The paper deserves credit for honest limitations and for converging evidence: the steep dR/dT, negative photoresistance across 85–2140 μm, DC-heating equivalence over B–T space, polarization independence, antenna simulations, and independent Johnson-noise thermometry all support electronic heating in some regime. However, the Johnson-noise measurement only proves that T_e rises; it does not prove T_L stays cold. The quantitative bound in Eq. S34 depends on a single ratio measured at 200 mK with superconducting contacts, and the paper's own δ≈5 power law implies that G_e-ph is strongly temperature-dependent, so extrapolation to T_e≈10 K is not benign. This is an addressable experimental question, not a logical contradiction, so the reader's CONDITIONAL verdict remains appropriate. I would not change the verdict; the proposed in-situ heat-budget test would either validate or refute the central mechanism.","tokens_in":23408,"tokens_out":11023,"duration_ms":143160,"concrete_test":"On the main device geometry at T0=1.7 K and ν=2, measure the e-ph vs cross-plane conductance ratio in situ. Use the existing graphite back gate as a heater: pass current through the gate to raise the whole stack temperature, and calibrate ΔT_e via R_xx(T) thermometry. Separately, apply direct Joule heating through the TBG channel and measure the same ΔT_e. Record P_channel/P_heater for ΔT_e from 0.1 K to 10 K. If P_channel/P_heater remains ≲0.01 at ΔT_e≈10 K, the decoupling assumption holds; if it approaches or exceeds 0.1, T_L is not pinned to T0 and the hot-electron interpretation as stated is unsupported. A complementary direct check is to embed a second graphene layer as a phonon thermometer in the stack and read T_L during THz exposure.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central mechanism requires T_e to rise by ~10 K while T_L rises <100 mK, so that the resistance change is caused by selective electronic overheating. The only direct evidence for this hierarchy is Supplementary Note 11: on a reference device at T0=200 mK, θ≈1°, ν=2.6, the ratio of direct Joule power to graphite-heater power needed for the same ΔT_e is interpreted as G_e-ph/G_⊥≈10^-4 (Eq. S33), and this ratio is then used in Eq. S34 to bound ΔT_L in the main device (θ=1.01°, T0=1.7 K, ν=2). Two unquantified steps make this extrapolation load-bearing. First, the reference device uses superconducting Al contacts, which suppress Wiedemann–Franz cooling, whereas the main device does not; its heat budget can differ. Second, and more seriously, the paper's own heat-balance model P=Σ(T_e^δ−T_0^δ) with δ≈5 (Eq. S17) means the differential electron–phonon conductance G_e-ph=dP/dT_e scales as T_e^{δ−1}. Moving from 0.2 K to T_e≈10 K raises G_e-ph by roughly (10/0.2)^4≈6×10^6 if G_⊥ is nearly constant, and by ~1200 even relative to the main device's 1.7 K base. Unless G_⊥ grows by a comparable factor, the hierarchy G_e-ph≪G_⊥ fails at the operating point, so the lattice can no longer be treated as pinned to T0. The noise thermometry in Supplementary Note 10 measures T_e, not T_L, so it does not resolve this. If the hierarchy fails, the observed photoresponse could still be bolometric, but it would be phonon-mediated, invalidating the 'selectively overheat the electronic subsystem' claim and the heat-capacity argument for the large response.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports a giant photoresistive response in magic-angle twisted bilayer graphene (MATBG) tuned to half filling of the moiré band. Under continuous-wave far-infrared and millimetre-wave illumination (λ = 85–2140 μm), the resistance at ν = 2 drops dramatically, and the authors attribute this to selective overheating of the low-heat-capacity electronic subsystem, which suppresses the correlated insulating gap and drives an insulator-to-metal transition. Using dual-modulation transport and a DC Joule-heating comparison, they extract an absorbed-power voltage responsivity of R_V^abs = (1.7 ± 0.4) × 10^7 V/W and a thermal-fluctuation-limited NEP ~ 2 × 10^-15 W/√Hz. The response is broadband, polarization-independent, and robust to magnetic fields up to several tesla. The central mechanistic claim is that this is a hot-electron bolometric effect, distinct from single-particle photoconductivity and dual to superconducting hot-electron bolometers.","tokens_in":23851,"tokens_out":4775,"duration_ms":60068,"significance":"If the hot-electron mechanism is correct, the paper demonstrates a new class of bolometric detector based on a many-body correlated insulator, with internal responsivity far exceeding typical commercial bolometers and with magnetic-field compatibility. The work also provides a transport-based probe of the fragility of the correlated gap in MATBG. The paper has clear strengths: the photoresistance is measured with a dual-modulation technique; the DC-heating comparison in Fig. 3 and Supplementary Figures 7–8 provides a consistency check across filling, temperature, and magnetic field; the polarization dependence is explicitly tested; two additional MATBG devices are measured; and the data for the main figures are provided as Source Data. The main weakness is the extrapolation of the electron–phonon/substrate thermal hierarchy from a 200 mK reference device to the operating conditions of the main device, on which the selective electronic-overheating interpretation rests.","major_comments":[{"comment":"The hot-electron interpretation relies on the hierarchy G_e-ph ≪ G_⊥, quantified as ~10^-4 from a reference device at T0 = 200 mK. Using the paper's own heat-balance model P = Σ(T_e^δ − T_0^δ) with δ ≈ 5 (Eq. S17), the differential conductance G_e-ph = dP/dT_e scales as T_e^{δ−1}. Moving from 0.2 K to the claimed T_e ≈ 10 K raises G_e-ph by roughly (10/0.2)^4 ≈ 6×10^6, and even relative to the main device's 1.7 K base by ~1200. Unless G_⊥ grows by a comparable factor, the bound ΔT_L < 100 mK (Eq. S34) is not justified. The reference device also uses superconducting Al contacts; the insensitivity to suppressing superconductivity was demonstrated on that reference device, not on the main device, whose contacts are not specified as superconducting. This extrapolation is load-bearing for the claim of selective electronic overheating: without it, the photoresponse could still be bolometric bu","section":"Supplementary Note 11, Eqs. S17, S33–S34"},{"comment":"The incident-to-absorbed power conversion is determined by matching R_THz(ρ) with Δ(dV/dI)(P_abs) from DC Joule heating at ν = 2. This is a calibration convention rather than an independent measurement of absorbed power. The agreement across fillings, temperatures, and fields (Supplementary Fig. 8) establishes consistency of the two perturbations, but only under the assumption that both act through the same thermal variable (T_e). If a substantial component of the DC-heating response is lattice-mediated, the calibration would fold that same component into the THz channel. Consequently, the headline internal responsivity R_V^abs = (1.7 ± 0.4) × 10^7 V/W and NEP_TF ≈ 2 × 10^-15 W/√Hz are contingent on this assumption. The manuscript should state this explicitly and, ideally, provide an independent calorimetric or phonon-thermometry check.","section":"Supplementary Note 5, Fig. 3a–b"},{"comment":"The noise thermometry is performed on a separate TBG device at T0 = 3.9 K with a silicon lens, not on the main device. It measures T_e only and does not report T_L. The sublinear T_e(P) dependence and δ ≈ 5 fit are consistent with electron–phonon cooling but do not directly verify the T_e ≫ T_L hierarchy in the main device. The main-text statement that external illumination selectively overheats the electronic subsystem is therefore supported by inference from auxiliary devices and the DC-heating comparison, not by a direct measurement in the main device. Please clarify which quantities are measured in the main device and which are transferred from reference devices.","section":"Supplementary Note 10"},{"comment":"The practical external responsivity for ideal focusing is R_V^{NA=1} ≈ 75–370 V/W, orders of magnitude below the internal R_V^abs. The abstract and Discussion emphasize the internal responsivity and compare it with commercial bolometers, without making clear that the external figure of merit in the present device is set by diffraction-limited optics and poor absorption. While the distinction is stated in the Results, the abstract's 'millivolts per nW' phrasing refers to absorbed, not incident, power. This should be clarified to avoid overstating the device-level performance.","section":"Main text, 'Performance', and Supplementary Table 1"}],"minor_comments":[{"comment":"Typo: 'T ransport and photoresponse measurements' should be 'Transport and photoresponse measurements'.","section":"Methods, first paragraph"},{"comment":"The caption reads 'MA TBG' with an extra space; correct to 'MATBG'.","section":"Supplementary Figure 17 caption"},{"comment":"The caption notes the curve was acquired during a different cooldown and absolute values differ. Please state the scale or normalize to dark resistance so readers can compare with Fig. 1g without confusion.","section":"Figure 1h caption"},{"comment":"The expression for NEP_JN^F contains a ratio with V*/I* and dV/dI. The notation (dV/dI)* vs V*/I* is not defined explicitly; define these operating-point quantities.","section":"Supplementary Note 8, Eq. S14"},{"comment":"The statement says source data for Figs. 1–3 are provided, but not for the supplementary figures. If the main quantitative claims (e.g., DC-heating calibration, noise thermometry) are shown in supplementary figures, please include those source data as well.","section":"Data Availability"}],"recommendation":"major_revision","confidential_remarks":"This is a potentially important result from a strong group. The central measurement—giant, broadband, negative photoresistance at ν=2 with a quantitative match to DC Joule heating—is convincing as a bolometric response. The major revision should focus on the thermal-budget extrapolation: the paper needs either a direct measurement of T_L in the main device, a more careful scaling analysis of G_e-ph(T_e)/G_⊥(T_0), or a clearly stated caveat that the 'selective electronic overheating' interpretation is one of two possible thermal mechanisms. The internal responsivity claim is contingent on the absorbed-power calibration, which is itself based on the same thermal model. I would not reject on these grounds, but the load-bearing nature of the extrapolation warrants a major revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Let me give you the short version first. This is a real result: MATBG tuned to half filling shows a huge photoresistive response to weak THz/FIR radiation, with internal voltage responsivity around 1.7e7 V/W and an NEP near 1e-15 W/√Hz. The experiments are thorough, and the central observations—negative photoresistance at three wavelengths, quantitative agreement with DC Joule heating across fillings, fields, and temperatures, polarization independence, and independent noise thermometry—converge. The paper deserves a serious referee.\n\nThe genuinely new element is the mechanism claim: that long-wavelength photons selectively overheat the electronic subsystem, melting the correlated gap. That is interesting physics, and if true, it makes correlated flat-band insulators a bolometric platform with a magnetic-field advantage over superconducting detectors.\n\nNow the soft spots, in order of importance. The biggest is the thermal-budget argument. The claim that T_e rises by ~10 K while the lattice stays within ~100 mK of T0 rests on a single reference-device measurement at 200 mK, which gives G_e-ph/G_⊥ ~1e-4. But the paper itself uses δ≈5 electron-phonon scaling elsewhere. Moving from 0.2 K to 10 K raises the differential G_e-ph by roughly (10/0.2)^4 ≈ 6e6. Even at the main device's base temperature of 1.7 K, the ratio climbs to order unity. Unless the cross-plane conductance G_⊥ grows comparably, the hierarchy G_e-ph << G_⊥ fails at the operating point, and the lattice can no longer be assumed pinned to T0. The noise thermometry in Supp. Note 10 measures T_e, not T_L, so it doesn't close this gap. The consequence: the response may still be bolometric, but it could be phonon-mediated, and the 'selective overheating' interpretation is not established. This is load-bearing, and the authors should address it—ideally by direct T_L measurement or by showing the scaling does not apply.\n\nA secondary issue is calibration of absorbed power via matching THz to DC Joule heating; that's somewhat circular for the responsivity number, but acceptable given the match across B and T. The noise figures rely on fitted G_th and δ, and the response time is an RC estimate rather than a measurement. Those are minor.\n\nMy take: the device physics is solid, the mechanism needs more work. I'd send this to peer review; it's the kind of paper a good referee can improve. The authors are upfront that the flat spectrum doesn't pin down the absorption pathway and that the huge response needs a high-quality stack—those are honest caveats, and they strengthen rather than weaken the case for refereeing.","headline":"Impressive device result, but the hot-electron mechanism rests on an extrapolated thermal-decoupling ratio that likely fails at the operating point.","tokens_in":24520,"tokens_out":7561,"would_cite":true,"duration_ms":80810,"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":"At half filling of the moiré band, weak long-wavelength photons collapse the correlated gap in magic-angle twisted bilayer graphene, driving an insulator-to-metal transition and a bolometric response near 10^7 V/W.","keywords":["correlated insulator","magic-angle twisted bilayer graphene","hot-electron bolometer","terahertz photoresponse","far-infrared detection","moiré flat bands","intervalley coherence","bolometric detection"],"falsifier":"On the same ν=2 device at T0=1.7 K, measure T_e by Johnson-noise thermometry and T_L with an independent phonon thermometer under 0.14 THz irradiation strong enough to visibly reduce Rxx; observing ΔT_L of order 1 K when ΔT_e is ~10 K, or no ΔT_e at all, would overturn the hot-electron hierarchy and the electronic-melting interpretation.","tokens_in":23265,"feed_emoji":"🔥","tokens_out":7226,"duration_ms":81348,"temperature":0.7,"pith_summary":"This paper claims that a correlated insulator can be turned into a metal by weak, long-wavelength photons alone, without structural change or above-gap excitation. In magic-angle twisted bilayer graphene at half filling of the moiré band, the absorbed radiation heats the low-heat-capacity electron system by roughly ten kelvin while the lattice barely moves, collapsing the ~1 meV correlated gap and sharply lowering the resistance. The effect is broadband (85–2140 µm), equivalent to DC Joule heating at the same absorbed power, and yields an internal voltage responsivity of (1.7±0.4)×10^7 V/W with thermal-fluctuation noise near 2×10^-15 W/√Hz, robust to magnetic fields of several tesla. If correct, it establishes correlated flat bands as ultrasensitive detectors of faint far-infrared and terahertz radiation and as a direct probe of many-body order under weak thermal perturbation.","feed_headline":"Weak THz light melts a moiré insulator into a metal","feed_subtitle":"At half filling, a ~1 meV correlated gap turns faint long-wavelength photons into a megavolt-per-watt response.","key_machinery":"The load-bearing object is the many-body correlated gap at half filling of the flat moiré band—an intervalley-coherent insulating state with an energy scale of about 1 meV, the smallest energy in the system—and its extreme temperature sensitivity: it closes continuously as the electronic temperature approaches roughly 10 K. Radiation absorbed by the graphene is thermalized on femtosecond timescales into a hot Fermi–Dirac distribution at temperature T_e; because the electronic heat capacity is tiny and electron–phonon coupling is weak, T_e rises by ~10 K while T_L rises by less than ~100 mK, collapsing the gap and producing a large negative photoresistance. The argument is carried by a DC-hea","core_discovery":"The discovery is that at moiré filling ν=2, a weak beam of far-infrared or millimetre-wave photons collapses the correlated insulating state of MATBG through electronic heating rather than photocarrier generation. The resistance at ν=2 rises by over an order of magnitude on cooling from 10 K to 1.7 K, tracing a gap of about 1 meV that softens and vanishes near 10 K; illumination deposits power that raises the electronic temperature, restoring density of states at the Fermi level and driving the sample metallic. The paper shows that the photoresistance measured under 0.14 and 3.5 THz illumination matches the change in differential resistance produced by DC Joule heating at the same absorbed p","pith_inferences":["If the hot-electron hierarchy holds generally, any moiré system with a steep temperature-dependent resistance—Chern insulators, fractional states, or other correlated gaps—should show a similar bolometric collapse, making detector sensitivity a proxy for the fragility of the order.","The DC-heating equivalence implies the detector is self-calibrating in absorbed-power units, which could make cross-device sensitivity comparisons possible without absolute optical power calibration.","Because the transduction is thermal, engineering absorption (cavities, plasmonic metasurfaces) should raise external responsivity toward the internal 10^7 V/W value while preserving nanosecond-scale response, potentially enabling photon-counting in the far infrared.","Bolometric readout also offers a thermodynamic thermometer for the correlated phase: Rxx at ν=2 maps T_e, so the same device can measure electronic heat capacity and electron-phonon coupling in situ."],"forward_implications":["A correlated insulator can serve as a practical bolometer in the 85–2140 µm window, with internal voltage responsivity above 10^7 V/W and a thermal-fluctuation-limited NEP around 2×10^-15 W/√Hz.","Because the response is thermal, it is inherently broadband and polarization-insensitive in the present geometry; spectral flatness follows from heating, not resonant absorption.","The device remains sensitive in perpendicular magnetic fields up to several tesla, unlike superconducting hot-electron bolometers, and its intrinsic electron–phonon cooling time is picosecond, with the measured speed set by the RC time constant of the high-resistance geometry.","The same heating mechanism produces weaker but same-sign photoresistance at ν=-2, and the effect size tracks the quality of the correlated insulating state, so optimized stacks should show proportionally larger responses.","Time-resolved photoresistance measurements could interrogate the intrinsic suppression-and-recovery dynamics of the correlated order without high-energy interband excitation."],"supporting_citations":[{"why":"Supplies the theoretical picture of a strain-disordered, gapless intervalley-coherent phase, used to explain the non-Arrhenius closing of the gap with temperature.","marker":"8"},{"why":"Provides the dual-modulation photoresistance measurement technique and the bolometric comparison approach used for the THz response.","marker":"9"},{"why":"Establishes the prior superconducting MATBG photodetection result that this work contrasts and extends to the correlated-insulator side.","marker":"18"},{"why":"Original demonstration of the correlated insulator at half filling in magic-angle graphene, defining the state exploited here.","marker":"19"},{"why":"Supports the interaction-driven band-flattening picture and the ~1 meV correlated-gap scale with its temperature behavior.","marker":"33"},{"why":"Provides spectroscopic evidence for intervalley gaps and many-body resonances, grounding the identification of the IVC order.","marker":"35"},{"why":"Supplies low-temperature electronic heat-capacity measurements of graphene, underpinning the low-heat-capacity premise.","marker":"43"},{"why":"Provides thermal-property measurements of superconducting MATBG used to estimate electronic heat capacity and thermal conductance scales.","marker":"44"},{"why":"Demonstrates ultrafast Umklapp-assisted electron-phonon cooling in MATBG, setting the picosecond intrinsic response time and cooling channel.","marker":"45"},{"why":"Supplies the hot-electron bolometer framework and DC-heating calibration approach used to convert incident power to absorbed power.","marker":"48"}],"fun_headline_variants":["Tiny THz beam melts correlated insulator in twisted graphene","Moiré insulator flips to metal with faint photons","Hot electrons, not photons, flip twisted graphene from insulator to metal","Ultra-sensitive bolometer: THz heat melts moiré insulator to metal"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The main device's electrons are thermally decoupled from the lattice: deposited THz power raises electron temperature by about 10 K while the lattice rises less than ~100 mK, a hierarchy measured on a different reference device and extrapolated to the main sample.","fun_headline_variants_meta":{"raw":{"variants":["Tiny THz beam melts correlated insulator in twisted graphene","Moiré insulator flips to metal with faint photons","Hot electrons, not photons, flip twisted graphene from insulator to metal","Ultra-sensitive bolometer: THz heat melts moiré insulator to metal"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000839,"raw_usage":{"total_tokens":3490,"prompt_tokens":736,"completion_tokens":2754,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":480,"completion_tokens_details":{"reasoning_tokens":2692}},"tokens_in":480,"tokens_out":2754,"duration_ms":22011,"temperature":1.0,"reasoning_tokens":2692,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T00:51:21.486099+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"On the same ν=2 device at T0=1.7 K, measure T_e by Johnson-noise thermometry and T_L with an independent phonon thermometer under 0.14 THz irradiation strong enough to visibly reduce Rxx; observing ΔT_L of order 1 K when ΔT_e is ~10 K, or no ΔT_e at all, would overturn the hot-electron hierarchy and the electronic-melting interpretation.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the theoretical picture of a strain-disordered, gapless intervalley-coherent phase, used to explain the non-Arrhenius closing of the gap with temperature."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the dual-modulation photoresistance measurement technique and the bolometric comparison approach used for the THz response."}],"review_version":1}