{"id":"61972fc6-e9cd-4384-ad8f-1cab04f8e2e5","arxiv_id":"2608.09614","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"A symmetric MgO/NiFe/MgO trilayer shows a self-induced spin-orbit torque attributed to a 1.8 nm magnetic dead layer at the bottom interface.","lead":"A nominally symmetric, heavy-metal-free MgO/NiFe/MgO sandwich still produces a spin-orbit torque, and the authors trace this to a thin magnetic dead layer at one interface. The result could make heavy-metal-free spin-orbit torque devices possible in standard magnetic stacks.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central claim is underspecified: the data cannot distinguish dead-layer-induced asymmetry from the intrinsic top/bottom (111) interface inequivalence that the paper's own DFT section concedes, so the 1.2 nm vs 1.8 nm fit does not uniquely validate the dead-layer mechanism.","rationale":"The paper has real strengths: a nominally symmetric MgO/NiFe/MgO trilayer, XPS evidence of bottom-interface oxidation, an independent SQUID-derived dead layer of about 1.8 nm, a non-monotonic thickness dependence of the damping-like torque, and DFT work showing NiFe can host a sizable intrinsic spin Hall conductivity. None of these, however, forces the magnetic dead layer to be the symmetry-breaking agent. The paper itself flags the unresolved top/bottom (111) surface-termination inequivalence and the resulting modification of interfacial SOC and orbital moments. Because the Kim-Lee mechanism predicts a net self-torque for any asymmetry in interfacial spin absorption, a structurally asymmetric interface could produce the same torque without a dead layer. The quantitative agreement in Fig. 3d therefore tests the assumed parameterization of Eq. 6, not the physical origin of the asymmetry. The reader's weakest assumption is exactly this identification problem, and the manuscript provides no dead-layer-free or dead-layer-inverted control. For these reasons the CONDITIONAL verdict is appropriate; my stress-test does not move the verdict, but it sharpens the condition: the authors should either rule out termination-driven asymmetry by structural characterization or demonstrate a control without the dead layer.","tokens_in":11102,"tokens_out":5817,"duration_ms":56251,"concrete_test":"Perform a likelihood-ratio test on the Fig. 3d thickness sweep: refit Eq. 6 with dead-layer thickness fixed to zero, allowing only the interfacial spin-mixing asymmetry (g_top - g_bot) and the other bulk parameters to vary, and compare Akaike/Bayesian information criteria to the published fit. If the zero-dead-layer fit is statistically indistinguishable (e.g., Delta AIC < 2), the extracted 1.2 nm is not a unique signature of a magnetic dead layer, and the central attribution is not established. If it is significantly worse, the dead-layer hypothesis gains genuine discriminating support.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing assumption is that the observed net damping-like torque can be uniquely attributed to the bottom magnetic dead layer. The paper's DFT section explicitly concedes that the (111) Ni3Fe/MgO top and bottom terminations may be intrinsically inequivalent because of ABC stacking and different interface hybridization, and that exact terminations were not characterized. Since the Kim-Lee mechanism requires only an asymmetry in interfacial spin absorption (g_top != g_bot), a purely electronic or structural interface asymmetry could generate a self-torque even with no magnetic dead layer. The quantitative fit in Fig. 3d does not resolve this: Eq. 6 contains at least C_self, g_top, g_bot, lambda_F, K, and t_eff as adjustable parameters, and the extracted ~1.2 nm dead layer is a derived parameter of that fit rather than an independent observable. With no control sample in which the dead layer is suppressed, moved to the top interface, or varied independently, the central causal attribution remains underdetermined. This limitation is not merely hypothetical; the paper itself calls out the possible top/bottom termination inequivalence and places its confirmation beyond scope.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports observation of a nonzero damping-like spin-orbit torque in a nominally symmetric MgO/NiFe/MgO trilayer, where conventional symmetry arguments forbid a net self-torque. Harmonic Hall measurements on a thickness wedge (4.5–14.4 nm) yield a DL effective field that peaks and then decreases with thickness. SQUID magnetometry and XPS depth profiling identify a ~1.8 nm magnetically dead layer at the bottom NiFe/MgO interface. The authors fit the thickness dependence with the Kim–Lee drift-diffusion expression, extracting a dead-layer thickness of ~1.2 nm, and support the mechanism with DFT calculations of the intrinsic spin Hall conductivity of Ni3Fe(111) and MgO/Ni3Fe interfaces. The central claim is that the naturally formed dead layer is the symmetry-breaking element that enables the forbidden self-torque.","tokens_in":11402,"tokens_out":5475,"duration_ms":54027,"significance":"If the causal attribution holds, the result is significant: it would provide a direct experimental validation of the Kim–Lee prediction that interfacial asymmetry in spin absorption alone can produce a net self-torque in a single ferromagnet, and it would reframe magnetic dead layers as functional spintronic elements. The experimental dataset is substantial and the paper does several things well: the harmonic Hall analysis includes a symmetry decomposition and explicit accounting for thermoelectric contributions; the dead-layer evidence from SQUID and XPS is independent of the transport model; and the thickness-series measurement is a suitable test of the Kim–Lee formalism. The main weakness is that the paper's own DFT discussion concedes an alternative symmetry-breaking source—intrinsic top/bottom (111) termination inequivalence—so the data do not uniquely identify the dead layer as the origin of the torque. The quantitative fit to Eq. 6 also lacks the parameter constraints needed to support the specific '1.2 nm vs 1.8 nm' agreement claimed in the abstract.","major_comments":[{"comment":"The manuscript explicitly concedes that 'the (111) Ni3Fe surface terminations at the top and bottom interfaces may not be equivalent' because of the ABC stacking sequence, with different bonding configurations and electronic hybridization, and that 'confirming the exact atomic terminations... lies beyond the scope of this work.' Since the Kim–Lee mechanism requires only an asymmetry in interfacial spin absorption (g_top ≠ g_bot), a purely structural/electronic interface asymmetry could produce the observed net damping-like torque without any magnetic dead layer. No control sample in which the dead layer is suppressed, moved to the top interface, or varied independently is reported. The wording in the abstract and the XPS/SQUID section—'we identify the symmetry-breaking origin' and 'we therefore attribute the observed self-torque to the interfacial asymmetry arising from the magnetic dead layer'—therefore overstates what the data establish. The authors should either provide such controls or reframe the central claim as demonstrating a self-torque consistent with dead-layer-induced asymmetry rather than uniquely identifying that mechanism.","section":"DFT Analysis, final paragraph before Conclusions"},{"comment":"The claimed quantitative agreement between the fitted dead-layer thickness (~1.2 nm) and the independently measured value (~1.8 nm) is not demonstrated in the main text. Equation (6) contains at least C_self, g_top, g_bot, λ_F, K, and the dead-layer correction entering through t_eff as adjustable parameters, yet no fixed values, constraints, starting points, parameter uncertainties, or uniqueness/identifiability analysis are reported. A correlated multi-parameter fit of a non-monotonic thickness dependence can easily absorb a 0.6 nm shift in the effective thickness. To make this point load-bearing, the authors must report the full parameter set with confidence intervals, a fit in which t_dead is fixed to the SQUID value of 1.8 nm with a comparison of goodness of fit, and a sensitivity analysis. As it stands, the 'agreement' between 1.2 nm and 1.8 nm is suggestive but does not constitute the quantitative validation claimed in the abstract.","section":"Eq. 6 and Fig. 3(d)"}],"minor_comments":[{"comment":"The phrase 'This mechanism provides direct evidence for substantial self-torque' overstates the inferential strength of a fit with unconstrained parameters; 'is consistent with' would be more appropriate.","section":"Introduction, paragraph after Eq. 6"},{"comment":"The linear fit of magnetic moment per area versus NiFe thickness used to extract the 1.8 nm dead layer should report the fit parameters, uncertainties, and the criterion used to identify the thickness intercept as the dead layer.","section":"Fig. 3(c)"},{"comment":"The DFT calculations are performed on an ordered Ni3Fe(111) surface, whereas the measured films are polycrystalline Ni80Fe20; a brief justification of the transferability of the computed spin Hall conductivity to the polycrystalline experimental system would strengthen the argument.","section":"DFT Analysis, Fig. 4"},{"comment":"The statement that the damping-like torque is 'comparable to HM-based systems' would be more convincing with a direct numerical benchmark against specific Pt/FM values cited in the references, rather than a qualitative comparison.","section":"Fig. 2(e)–(f) and text"},{"comment":"All symbols in Eq. (6), especially C_self, K, λ_F, and t_eff, should be defined in the main text rather than only in the supplementary information.","section":"Equation (6)"},{"comment":"Several references are missing journal identifiers (e.g., [5], [21], [35], [45], [50]); these should be completed before publication.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The stress-test concern is well placed: the paper's own DFT section provides an alternative symmetry-breaking route (top/bottom (111) termination inequivalence) that the experimental data cannot currently exclude. This does not invalidate the core observation of a non-zero self-torque in a symmetric trilayer, but it means the title and abstract overclaim the causal role of the dead layer. The fit-based extraction of the dead-layer thickness is also under-documented. I recommend major revision rather than rejection, because the observation and the characterization are valuable and the causal question could be addressed by reframing the claims or by adding a control experiment."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe paper reports a robust damping-like torque in a nominally symmetric MgO/NiFe/MgO trilayer with no heavy metal, and attributes it to a ~1.8 nm magnetic dead layer at the bottom interface. The harmonic Hall data look carefully analyzed, with a symmetry decomposition and field-dependence checks, and the SQUID and XPS evidence for an oxidized, magnetically dead bottom interface is independent and credible. The DFT spin Hall conductivity for Ni3Fe is a useful addition. If the observation holds, it is the first clean confirmation of the Kim-Lee interfacial-asymmetry mechanism in a simple trilayer, and that is significant.\n\nThe soft spot is the causal attribution. The paper's own DFT section concedes that the top and bottom (111) Ni3Fe/MgO terminations may be intrinsically inequivalent because of ABC stacking and interface hybridization. The Kim-Lee mechanism requires only an asymmetry in interfacial spin absorption, so a purely structural asymmetry could generate the self-torque even without a dead layer. The quantitative fit in Fig. 3d does not resolve this: Eq. 6 contains at least six adjustable parameters, and the extracted 1.2 nm dead-layer thickness is a derived parameter of that fit rather than an independent observable. The agreement with the SQUID value is suggestive, but it does not uniquely validate the dead-layer mechanism. There is no control sample with the dead layer suppressed, moved to the top interface, or varied independently. The paper itself flags the termination ambiguity and puts its resolution beyond scope—that is honest, but it means the central claim is underdetermined.\n\nThe thickness dependence of the torque and the DFT layer-dependence are consistent with the authors' story, and I would not be surprised if the dead layer is indeed the main symmetry breaker. But the current evidence cannot rule out an electronic or structural interface asymmetry as the source. A referee should ask for a more constrained model mapping, an estimate of the intrinsic top/bottom asymmetry from the DFT, and ideally a control sample.\n\nWho is this for? SOT spintronics experimentalists and theorists working on self-torque and interface engineering. The paper deserves serious peer review, and with the requested controls the mechanism could be pinned down. I would cite the observation, but not yet the dead-layer mechanism.","headline":"Robust self-torque in a symmetric trilayer is a real observation, but the dead-layer attribution is underdetermined by the data and the paper's own DFT section admits an alternative symmetry breaker.","tokens_in":11955,"tokens_out":3609,"would_cite":true,"duration_ms":29764,"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":"Forbidden spin-orbit torque observed in a symmetric ferromagnet trilayer.","keywords":["spin-orbit torque","self-induced torque","magnetic dead layer","NiFe","MgO/NiFe/MgO","harmonic Hall measurement","spin Hall conductivity","heavy-metal-free spintronics"],"falsifier":"A direct test is to make the two interfaces chemically and magnetically equivalent, for example by eliminating the bottom dead layer or by forming an identical dead layer at the top interface, and check whether the damping-like torque vanishes; a second test is high-resolution structural imaging of the (111) terminations to see whether they are equivalent.","tokens_in":10938,"feed_emoji":"🧲","tokens_out":6494,"duration_ms":58091,"temperature":0.7,"pith_summary":"This paper reports that a nominally symmetric MgO/NiFe/MgO trilayer, a structure where a net spin-orbit torque is forbidden by inversion symmetry, nonetheless shows a clear damping-like torque with magnitude comparable to heavy-metal/ferromagnet bilayers. The symmetry-breaking agent is identified as a naturally formed magnetic dead layer about 1.8 nm thick at the bottom NiFe/MgO interface, seen in magnetization and photoemission depth profiles. Fitting the thickness dependence of the torque to the drift-diffusion model of [1] yields a dead-layer thickness of about 1.2 nm, in quantitative agreement with the structural estimate. Density-functional calculations give NiFe an intrinsic spin Hall conductivity large enough to supply the observed spin currents. If correct, the result converts an interface defect usually treated as parasitic into a functional element that can generate heavy-metal-free spin-orbit torques in standard magnetic stacks.","feed_headline":"Forbidden spin-orbit torque observed in symmetric trilayer","feed_subtitle":"A parasitic dead layer at one interface supplies the missing asymmetry, and theory matches the measured thickness.","key_machinery":"The carrying mechanism is the drift-diffusion model of [1] applied to a single ferromagnetic layer, expressed in the fitting formula H_AD(t) = C_self / t_eff x Re[((g_top - g_bot) lambda_F tanh(t_eff / 2 lambda_F)) / ((g_top + g_bot) lambda_F coth(t_eff / lambda_F) + K)]. Here g_top and g_bot are the spin-absorption conductances of the two interfaces, and their difference is the source of the net torque: a spin current generated by the ferromagnet's intrinsic spin Hall effect precesses while diffusing, then reflects differently at the two interfaces so the top and bottom accumulations do not cancel. The magnetic dead layer is the physical object that makes g_top and g_bot unequal: a roughly 1.8 nm oxidized, magnetically inert region at the bottom interface detected by SQUID and XPS. In the experiment, this formula converts the measured thickness dependence into a dead-layer thickness and matches the structural value. The DFT calculation supplies the complementary ingredient, showing that the Ni3Fe(111) surface and the MgO/Ni3Fe(111) interface have an intrinsic spin Hall conductivity large enough to generate the required spin current.","core_discovery":"The central claim is that self-induced spin-orbit torque does not require a broken bulk or an engineered gradient: asymmetry in interfacial spin absorption alone can make the torque finite in a symmetric ferromagnet. In MgO(2)/NiFe(t)/MgO(3.5) trilayers with t from 4.5 to 14.4 nm, harmonic Hall measurements find a damping-like effective field that rises, peaks, and falls with thickness, the signature of a bulk spin current that precesses and diffuses before accumulating at the interfaces. X-ray photoemission shows Ni and Fe oxides only at the bottom NiFe/MgO interface, and SQUID magnetometry gives a magnetic dead layer of about 1.8 nm; fitting the torque data to the drift-diffusion equation of [1] with unequal top and bottom spin absorption yields about 1.2 nm. The paper therefore claims the first explicit confirmation that an interfacial dead layer creates the spin-absorption asymmetry required for a net self-torque, and that NiFe's intrinsic spin Hall conductivity, around 3 x $10^{2}$ (hbar/e) S/cm in the DFT (111)-surface calculation, an order below Pt, is sufficient to drive it. The field-like torque is non-negligible and roughly thickness-independent, consistent with the same model's prediction of transverse torques from spin precession even without interfacial asymmetry.","pith_inferences":["If the dead-layer mechanism holds, past control samples with nominally symmetric oxide/ferromagnet/oxide stacks may deserve re-examination, since hidden self-torques could have been subtracted or dismissed as artifacts.","A direct testable extension is to tune the bottom interface by controlled oxygen exposure or an ultrathin inserted oxide layer and map the torque against dead-layer thickness; the drift-diffusion formula makes the predicted curve quantitative.","Because NiFe's intrinsic spin Hall conductivity is an order below Pt yet still yields a measurable torque, the result suggests that modest spin Hall materials can be sufficient once interface absorption is asymmetric.","The finding implies that torque cancellation in symmetric ferromagnetic stacks is the exception rather than the rule whenever the two interfaces differ chemically or magnetically, even unintentionally."],"forward_implications":["Standard MgO/NiFe/MgO films, which contain no heavy metal, can produce damping-like torques at a magnitude comparable to Pt/ferromagnet bilayers.","The torque magnitude can be predicted from the stack's dead-layer thickness through the drift-diffusion formula, making structure-to-function mapping quantitative.","Interfacial oxidation, normally avoided in device growth, becomes a controllable parameter for engineering self-torques.","The measured field-like torque confirms that spin precession inside an ordinary ferromagnet yields transverse torque even without interfacial asymmetry.","Because the effect requires only a common ferromagnet and oxide barriers, existing oxide/ferromagnet/oxide stacks may already contain hidden self-torques that standard symmetry arguments would miss."],"supporting_citations":[{"why":"Supplies the drift-diffusion prediction that unequal interfacial spin absorption yields a net self-torque in a single ferromagnet; the paper quantitatively fits to this model.","marker":"[1]"},{"why":"Gives the symmetry decomposition of spin currents in a ferromagnet that the harmonic Hall analysis is built on.","marker":"[11]"},{"why":"States the cancellation condition for a symmetric stack and the routes to a non-zero net torque used to motivate the experiment.","marker":"[12]"},{"why":"Provides the microscopic spin-current form for ferromagnets used alongside the symmetry expansion.","marker":"[13]"},{"why":"Establishes the harmonic Hall measurement method the authors use to extract damping-like and field-like effective fields.","marker":"[46]"},{"why":"Supports the claim that spin polarizations collinear with the magnetization dephase and contribute no measurable torque.","marker":"[49]"},{"why":"Reports symmetric-interface cancellation behavior that the present stack is designed against and that the dead layer breaks.","marker":"[52]"},{"why":"Provides the reference platinum spin Hall conductivity used to put the NiFe value in context.","marker":"[61]"},{"why":"Supplies comparison materials whose spin Hall conductivity is similar to the computed NiFe value.","marker":"[62]"},{"why":"Documents that (111) surface terminations at top and bottom interfaces can be inequivalent, the caveat the paper raises about its own attribution.","marker":"[63]"}],"fun_headline_variants":["Forbidden torque traced to 1.8-nm dead layer at interface","Dead layer asymmetry unlocks self-induced spin-orbit torque","No heavy metal: dead layer breaks symmetry for spin torque","Interfacial dead layer supplies missing asymmetry for spin-orbit torque"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument depends on the magnetic dead layer at the bottom interface being the only meaningful difference between the two interfaces; if the top and bottom (111) Ni3Fe surfaces are intrinsically inequivalent, the torque could come from that structural asymmetry alone.","fun_headline_variants_meta":{"raw":{"variants":["Forbidden torque traced to 1.8-nm dead layer at interface","Dead layer asymmetry unlocks self-induced spin-orbit torque","No heavy metal: dead layer breaks symmetry for spin torque","Interfacial dead layer supplies missing asymmetry for spin-orbit torque"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000625,"raw_usage":{"total_tokens":2933,"prompt_tokens":1025,"completion_tokens":1908,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":641,"completion_tokens_details":{"reasoning_tokens":1836}},"tokens_in":641,"tokens_out":1908,"duration_ms":14036,"temperature":1.0,"reasoning_tokens":1836,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T14:03:22.318127+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct test is to make the two interfaces chemically and magnetically equivalent, for example by eliminating the bottom dead layer or by forming an identical dead layer at the top interface, and check whether the damping-like torque vanishes; a second test is high-resolution structural imaging of the (111) terminations to see whether they are equivalent.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the drift-diffusion prediction that unequal interfacial spin absorption yields a net self-torque in a single ferromagnet; the paper quantitatively fits to this model."},{"cited_title":"Davidson, V","cited_arxiv_id":null,"evidence_quote":"Gives the symmetry decomposition of spin currents in a ferromagnet that the harmonic Hall analysis is built on."},{"cited_title":"Kim, B.-G","cited_arxiv_id":null,"evidence_quote":"States the cancellation condition for a symmetric stack and the routes to a non-zero net torque used to motivate the experiment."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the microscopic spin-current form for ferromagnets used alongside the symmetry expansion."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the harmonic Hall measurement method the authors use to extract damping-like and field-like effective fields."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supports the claim that spin polarizations collinear with the magnetization dephase and contribute no measurable torque."},{"cited_title":"Choi et al., Nature Communications 16, 5859 (2025)","cited_arxiv_id":null,"evidence_quote":"Reports symmetric-interface cancellation behavior that the present stack is designed against and that the dead layer breaks."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the reference platinum spin Hall conductivity used to put the NiFe value in context."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies comparison materials whose spin Hall conductivity is similar to the computed NiFe value."},{"cited_title":"Jo et al., Nano Letters 24, 7100 (2024)","cited_arxiv_id":null,"evidence_quote":"Documents that (111) surface terminations at top and bottom interfaces can be inequivalent, the caveat the paper raises about its own attribution."}],"review_version":1}