{"id":"64867728-6992-4c61-bd15-7443993d7981","arxiv_id":"2501.06063","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"Bias voltage around 0.625 V reverses the sign of tunneling magnetoresistance in Fe3GaTe2/hBN/Fe3GaTe2 junctions at room temperature, reproducibly across devices.","lead":"This paper reports that the tunneling magnetoresistance of Fe3GaTe2/hBN/Fe3GaTe2 magnetic tunnel junctions flips sign at a bias voltage around 0.625 V, even at room temperature. The effect is consistent across several devices and temperatures, and the authors propose a spin-resolved density-of-states model to explain it.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No fatal flaw in the experimental TMR-inversion result; the load-bearing weakness is the theory's quantitative match, which rests on an unconstrained weighting function W(k) rather than a derived momentum-conservation model.","rationale":"I reviewed the experimental analysis in good faith: the spin-valve TMR shape, the temperature dependence of TMR and spin polarization, the consistency between spin polarization and anomalous Hall conductance, and the ruling out of Fowler–Nordheim tunneling below 3 V all support the central experimental observation of a reproducible bias-controlled TMR inversion. No credible evidence of a measurement artifact emerged. The reader's weakest assumption accurately identifies the theoretical model's ad hoc weighting function as the main soft spot, and I agree with that assessment. My read does not move the verdict: the paper remains CONDITIONAL, because the experimental result stands while the quantitative theoretical explanation is not yet independently validated. The proposed concrete test would settle whether the improved agreement is physics or fitting by removing the adjustable δ and using a momentum-conservation weight derived from an explicit twisted-interface calculation.","tokens_in":8110,"tokens_out":5608,"duration_ms":59741,"concrete_test":"Replace the fitted W(k) by a transmission weight obtained from a DFT/NEGF calculation of a twisted FGT/hBN/FGT supercell at the actual device twist angle(s), with no free δ, and recompute the first and second TMR inversion voltages from the same spin-resolved spectra. If the first inversion moves away from ~0.6 V or the second inversion remains absent, the quantitative agreement in Fig. 4c is attributable to the free parameter rather than to momentum-conservation physics.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The experimental claim that the first TMR polarity inversion occurs at 0.625±0.025 V across devices and temperatures appears internally consistent and is supported by the spin-valve data and the explicit exclusion of Fowler–Nordheim transport below 3 V. The load-bearing weakness is in the explanatory part of the paper. Equation (1) is written in terms of energy-resolved spin DOS and neglects in-plane momentum conservation; the revised calculation then introduces W(k)=δ/sqrt(δ²+k²) with δ=0.1×2π/Å to shift the predicted first inversion from 0.19 V to 0.52 V. No derivation fixes δ or connects it to an actual misalignment angle, and the manuscript does not state exactly how W(k) enters Eq. (1), making the green curve in Fig. 4c unreproducible from the text. The agreement is therefore a one-parameter fit rather than a validated prediction. The paper also concedes that the second inversion near 2.5 V is not captured by this model. This undermines the proposed mechanism (‘importance of in-plane momentum conservation’) but not the observed reproducibility of the inversion.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript reports room-temperature tunneling magnetoresistance (TMR) measurements in van der Waals magnetic tunnel junctions Fe3GaTe2/hBN/Fe3GaTe2. The central experimental claim is that the TMR sign inverts reproducibly at approximately 0.625±0.025 V across multiple devices and temperatures, and that a second inversion occurs near 2.5–3 V. The authors exclude Fowler–Nordheim tunneling and interfacial-state mechanisms and propose a theoretical explanation based on spin-resolved density of states at high bias, modified by a k-dependent weighting function W(k)=δ/sqrt(δ²+k²) intended to account for misalignment between the two electrodes. With δ chosen as 0.1×2π/Å, the predicted first inversion improves from 0.19 V to 0.52 V, approaching the measured value. The paper concludes that in-plane momentum conservation and electrode misalignment are important for quantitative TMR predictions.","tokens_in":8500,"tokens_out":2826,"duration_ms":29804,"significance":"If the experimental result holds, this is a valuable contribution: it demonstrates a room-temperature all-van-der-Waals magnetic tunnel junction with a reproducible voltage-controlled TMR inversion, a phenomenon with direct relevance for energy-efficient spintronic devices. The experimental part is careful and internally consistent: the inversion voltage is stable across devices and temperatures, the spin-valve data support the reported TMR values, and the authors explicitly rule out the main alternative mechanisms (Fowler–Nordheim transport below 3 V, interfacial chemical states, and magnon excitations as a sole cause). The theoretical part, however, is significantly weaker: the improved quantitative agreement is achieved by an ad hoc weighting function whose physical origin is not derived, and the model does not reproduce the second inversion. The paper should be credited for being explicit about these limitations, but as written the theoretical mechanism is not yet validated.","major_comments":[{"comment":"The weighting function W(k)=δ/sqrt(δ²+k²) is introduced in the text but it is not specified how it enters Eq. (1); the green curve in Fig. 4c is therefore not reproducible from the equations as written. Please write the modified current integrals explicitly, showing how W(k) multiplies the spin-resolved DOS terms and whether it is applied to both electrodes or only one.","section":"Theoretical model (p. 5, Eq. (1) and Fig. 4c)"},{"comment":"The parameter δ is a free parameter chosen as 0.1×2π/Å to shift the predicted first inversion voltage from 0.19 V to 0.52 V. No independent constraint, microscopic derivation, or connection to a specific crystallographic misalignment angle is provided. The improved agreement is therefore a one-parameter fit rather than a validated prediction of the momentum-conservation mechanism. Please either derive W(k) from a model of the misaligned interface, relate δ to an experimentally measurable quantity (e.g., twist angle), or present the result as a sensitivity study rather than an explanatory calculation.","section":"Theoretical model (p. 5, Fig. 4c)"},{"comment":"The model fails to reproduce the second TMR inversion near 2.5–3 V, and the manuscript attributes this discrepancy to hBN misalignment at high bias without any quantitative support. Since the title and abstract emphasize voltage-controlled inversions (plural) of TMR, the theoretical mechanism should at least provide a qualitative account of the second inversion, or the paper should explicitly narrow its claim to the first inversion only.","section":"Experimental bias dependence (Fig. 3d) vs theoretical model (Fig. 4c)"},{"comment":"The magnetization fit uses the empirical form M(T)=M(0)(1−(T/Tc)^α)^β with α=1.65, β=0.5, and Tc=310 K, but no error bars or fit residuals are shown for the extracted spin polarization or anomalous Hall conductance. Given that the critical exponent β=0.5 differs from previous reports (β=0.35–0.4), the fit quality and the sensitivity of β to the choice of Tc should be documented.","section":"Experimental data and fitting (Fig. 3b)"}],"minor_comments":[{"comment":"There is a typo in the sentence 'formatomicallyflat, defect-freeinterfaces' – it should read 'atomically flat, defect-free interfaces'.","section":"Introduction (p. 1)"},{"comment":"The statement 'the very large band gap of high quality hBN ensure the electronic transport would be tunneling for the whole investigated temperature range' should be rephrased for clarity and grammatical correctness.","section":"Introduction (p. 1)"},{"comment":"The 'anomaly at zero bias' is mentioned in the text but not clearly identified in the figure. Please specify what the anomaly is (e.g., a local maximum or minimum in the TMR-bias curve) and whether it is reproducible across devices.","section":"Fig. 3d and accompanying text (p. 4)"},{"comment":"The integration limits and the notation in Eq. (1) are not fully defined; in particular, the integration from μD to μS is presented with two lower limits in the displayed equation. Please clarify the convention for positive and negative bias and the relation μD−μS=eVb.","section":"Theoretical model (p. 5)"},{"comment":"The claim that 'none of the existing experimental works on 2D MTJs report intentional crystal alignment yet' is broad and would benefit from a specific reference or a more cautious wording, since the reader cannot easily verify the absence of reports in the cited literature.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The experimental core of this paper – the reproducible TMR inversion near 0.625 V – appears solid and is the main strength. The theoretical section, however, is currently more of a fitting exercise than a predictive model, because the weighting parameter δ is chosen to move the inversion voltage closer to the data without a derivation or independent constraint. This is a load-bearing weakness for the paper's explanatory claims, so a major revision is appropriate. In revision, the authors should either substantially strengthen the theory (e.g., derive W(k) from a microscopic model or twist-angle-dependent calculation, and address the second inversion) or explicitly reframe the theoretical part as a qualitative proposal. The experimental data are otherwise suitable for a good journal."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The one thing you should know: this is a credible experimental paper that reports a reproducible bias-controlled TMR inversion in an all-vdW MTJ, FGT/hBN/FGT, at room temperature. The inversion voltage is 0.625±0.025 V and holds across devices and temperatures. That part is solid. The theoretical explanation, however, is the soft spot: the improved match is achieved by introducing an ad hoc weighting function W(k)=δ/sqrt(δ²+k²) with δ chosen to move the predicted inversion from 0.19 V to 0.52 V. No independent constraint fixes δ, and the model doesn't reproduce the second inversion around 2.5 V. So treat the theory as illustrative, not validated.\n\nWhat's actually new: the material combination is new—FGT electrodes with hBN barrier—and they show the room-temperature polarity reversal is robust. They also carefully rule out Fowler-Nordheim tunneling below 3 V, interfacial states, and magnon effects. The agreement between the spin polarization extracted from TMR and the anomalous Hall conductance as a function of temperature is a nice check, even if the critical exponent discrepancy with bulk magnetization is left open.\n\nWhere it's soft: the theory. Equation (1) neglects in-plane momentum conservation, then they patch it with a k-dependent weight. The stress-test note is right: the text doesn't even state exactly how W(k) enters Eq. (1), so the green curve in Fig. 4c is not reproducible from the paper. Because δ is not derived from a misalignment angle or any independent input, the claim that the theory \"explains\" the inversion voltage is overstated. The paper acknowledges the second inversion isn't captured, which is honest, but it undercuts the proposed mechanism. These are fixable in revision if they present W(k) as a phenomenological model with clear limitations, or better, derive it from a scattering model.\n\nOverall: the experimental result is a useful data point for the 2D spintronics community. The paper deserves a serious referee; it shouldn't be desk rejected. But I'd want the theory section rewritten to be honest about the fit nature, and the methods expanded for both fabrication and DFT.\n\nWould I bring it to reading group? Probably yes, as a conversation starter about what counts as explanation in this field. I'd cite it for the experimental observation.\n\nRecommendation: send to peer review, but expect significant revision on the theory.","headline":"Solid experimental demonstration of reproducible bias-controlled TMR inversion in Fe3GaTe2/hBN/Fe3GaTe2 at room temperature, but the theory's quantitative match is a one-parameter fit, not a derived prediction.","tokens_in":8927,"tokens_out":2161,"would_cite":true,"duration_ms":19336,"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":"Bias voltage flips the magnetoresistance sign of a van der Waals magnetic tunnel junction at a reproducible 0.625 V.","keywords":["tunneling magnetoresistance","TMR inversion","Fe3GaTe2","hBN barrier","van der Waals heterostructure","magnetic tunnel junction","spin-resolved density of states","room-temperature spintronics"],"falsifier":"Fabricate two Fe3GaTe2/hBN/Fe3GaTe2 devices, one with the FGT flakes crystallographically aligned and one deliberately rotated (for example by 30 degrees), and measure the first TMR inversion voltage in each. The weighting model predicts the aligned device should approach the unweighted value near 0.19 V while the rotated device should invert near or above 0.5 V; observing no such difference would show the weighting function is not capturing the misalignment effect.","tokens_in":99,"feed_emoji":"🧲","tokens_out":10375,"duration_ms":149236,"temperature":0.7,"pith_summary":"This paper reports that the tunneling magnetoresistance (TMR) of Fe3GaTe2/hBN/Fe3GaTe2 magnetic tunnel junctions changes sign under applied bias at room temperature. The first reversal, from positive to negative TMR, occurs at 0.625±0.025 V and repeats across multiple devices and temperatures, while a second reversal back to positive occurs near 2.5 V. The authors argue that the mechanism is the energy-dependent spin-resolved density of states of the Fe3GaTe2 electrodes, not interface chemistry or barrier wave-function interference. They modify an existing tunneling-current model by weighting electronic states near the center of the momentum zone more heavily, approximating the effect of misaligned electrode lattices, and this moves the predicted first inversion voltage from about 0.19 V to 0.52 V. If the interpretation is right, bias-controlled TMR inversion is a reproducible, usable feature of two-dimensional magnetic tunnel junctions.","feed_headline":"Bias flips a 2D tunnel junction's magnetoresistance at 0.625 V","feed_subtitle":"In Fe3GaTe2/hBN/Fe3GaTe2, TMR flips from positive to negative at 0.625 V, reproducibly.","key_machinery":"The central object is the bias-dependent tunneling-current model built from spin-resolved densities of states. The parallel and antiparallel currents are integrals over the bias window of products of source and drain DOS, $\\rho^\\uparrow_D \\rho^\\uparrow_S + \\rho^\\downarrow_D \\rho^\\downarrow_S$ and $\\rho^\\uparrow_D \\rho^\\downarrow_S + \\rho^\\downarrow_D \\rho^\\uparrow_S$, with TMR defined as $I_P/I_{AP}-1$. To close the gap between the predicted 0.19 V inversion and the measured 0.625 V, the authors multiply the DOS at each in-plane momentum state by $W(k)=\\delta/\\sqrt{\\delta^2+k^2}$ with $\\delta=0.1\\times2\\pi/\\mathrm{\\AA}$, which suppresses states away from the $\\Gamma$ point, the zone center where in-plane momentum is conserved automatically. This weighting function is what carries the improved quantitative agreement, moving the predicted first inversion to about 0.52 V.","core_discovery":"The paper establishes that a van der Waals magnetic tunnel junction with Fe3GaTe2 electrodes and a four-layer hBN barrier shows two bias-driven inversions of TMR at room temperature: positive-to-negative at 0.625±0.025 V and negative-to-positive near 2.5±0.5 V. It claims the first threshold is reproducible across devices and temperatures, and that the inversion mechanism is the participation of higher-energy electrons whose majority and minority spin densities of states differ, so that spin injection and detection can be tuned by voltage. The paper rules out interface states, which would require chemical bonding across the FGT/hBN interface, and Fowler-Nordheim barrier effects, which do not set in below about 3 V. It then shows that a DOS-integral model, corrected by a momentum-dependent weighting factor meant to represent lattice misalignment of the two exfoliated electrodes, reproduces the first inversion voltage much more closely than the unweighted model.","pith_inferences":["A direct test the authors do not carry out: fabricate devices with a controlled crystallographic rotation between the two FGT electrodes. The model implies the first inversion voltage should move toward the unweighted value near 0.19 V when the flakes are aligned and remain near or above 0.5 V when they are rotated.","The paper does not emphasize this, but the near-temperature-independence of the 0.625 V threshold, despite strongly temperature-dependent spin polarization, suggests the band-structure features setting the inversion survive thermal broadening, which would make the effect practical at operating temperatures.","The same DOS-integral construction could serve as a screening tool for other van der Waals ferromagnet/insulator pairs, predicting which combinations invert TMR and at what bias, provided the weighting function can be derived from the misalignment angle instead of fitted."],"forward_implications":["A single Fe3GaTe2/hBN/Fe3GaTe2 junction can be switched between positive and negative TMR by choosing the bias voltage, adding a voltage-controlled handle on the resistance state.","The reproducible 0.625 V threshold becomes a quantitative benchmark against which refined theories of momentum conservation and lattice misalignment in van der Waals MTJs can be tested.","Because the inversion is tied to the spin-resolved DOS of Fe3GaTe2, other room-temperature two-dimensional ferromagnets with different band structures should show their own characteristic inversion voltages.","The model's failure to reproduce the second reversal marks the high-bias momentum dependence of the hBN barrier as the missing ingredient for a complete theory."],"supporting_citations":[{"why":"Supplies the spin-resolved DOS tunneling-current integrals and the baseline inversion-voltage calculation that this paper adapts.","marker":"[15]"},{"why":"Reports voltage-tunable spin injection and detection in Fe3GeTe2/hBN devices and documents the same theory-experiment gap in inversion voltage.","marker":"[14]"},{"why":"Establishes Fe3GeTe2/hBN/Fe3GeTe2 tunneling spin valves and the anomalous-Hall-conductance method used here to infer spin polarization.","marker":"[11]"},{"why":"Provides the Julliere formula whose always-positive TMR the observed sign inversion goes beyond.","marker":"[24]"},{"why":"Gives an earlier Fe3GaTe2 band-structure calculation that the present DFT and spin-resolved DOS analysis follows.","marker":"[41]"},{"why":"Documents the above-room-temperature ferromagnetism of the Fe3GaTe2 crystals used to build the devices.","marker":"[22]"},{"why":"Reports room-temperature tunable TMR in Fe3GaTe2-based heterojunctions with polarity reversal, motivating the quantitative mechanism study here.","marker":"[19]"}],"fun_headline_variants":["TMR flips sign at 0.625 V in Fe3GaTe2/hBN junction","Voltage tunes TMR polarity in 2D magnetic tunnel junction","Bias voltage flips TMR at 0.625 V in van der Waals stack","Two TMR inversions seen in Fe3GaTe2/hBN/Fe3GaTe2"],"cache_read_input_tokens":11008,"weakest_assumption_plain":"The quantitative agreement rests on the assumption that a single weighting factor $W(k)=\\delta/\\sqrt{\\delta^2+k^2}$ with $\\delta=0.1\\times2\\pi/\\mathrm{\\AA}$ correctly represents how lattice misalignment between the two Fe3GaTe2 electrodes reduces tunneling from states away from the zone center; if that factor does not capture the real momentum-conservation physics, the improved 0.52 V agreement is fitting rather than prediction.","fun_headline_variants_meta":{"raw":{"variants":["TMR flips sign at 0.625 V in Fe3GaTe2/hBN junction","Voltage tunes TMR polarity in 2D magnetic tunnel junction","Bias voltage flips TMR at 0.625 V in van der Waals stack","Two TMR inversions seen in Fe3GaTe2/hBN/Fe3GaTe2"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000712,"raw_usage":{"total_tokens":3185,"prompt_tokens":908,"completion_tokens":2277,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":524,"completion_tokens_details":{"reasoning_tokens":2182}},"tokens_in":524,"tokens_out":2277,"duration_ms":14617,"temperature":1.0,"reasoning_tokens":2182,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T21:04:42.584751+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Fabricate two Fe3GaTe2/hBN/Fe3GaTe2 devices, one with the FGT flakes crystallographically aligned and one deliberately rotated (for example by 30 degrees), and measure the first TMR inversion voltage in each. The weighting model predicts the aligned device should approach the unweighted value near 0.19 V while the rotated device should invert near or above 0.5 V; observing no such difference would show the weighting function is not capturing the misalignment effect.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the spin-resolved DOS tunneling-current integrals and the baseline inversion-voltage calculation that this paper adapts."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reports voltage-tunable spin injection and detection in Fe3GeTe2/hBN devices and documents the same theory-experiment gap in inversion voltage."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes Fe3GeTe2/hBN/Fe3GeTe2 tunneling spin valves and the anomalous-Hall-conductance method used here to infer spin polarization."},{"cited_title":"Tunneling between ferromagnetic films","cited_arxiv_id":null,"evidence_quote":"Provides the Julliere formula whose always-positive TMR the observed sign inversion goes beyond."},{"cited_title":"Tremendous tunneling magnetoresistance ef- fects based on van der waals room-temperature ferromag- net Fe3GaTe2 with highly spin-polarized fermi surfaces","cited_arxiv_id":null,"evidence_quote":"Gives an earlier Fe3GaTe2 band-structure calculation that the present DFT and spin-resolved DOS analysis follows."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Documents the above-room-temperature ferromagnetism of the Fe3GaTe2 crystals used to build the devices."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reports room-temperature tunable TMR in Fe3GaTe2-based heterojunctions with polarity reversal, motivating the quantitative mechanism study here."}],"review_version":1}