{"id":"57c0f816-a78a-435a-a0e6-17174a7418f2","arxiv_id":"2412.15885","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":7,"one_line_summary":"In Mn2Au, microsecond current pulses switch the magnetic order by thermal strain, while nanosecond pulses switch it purely by spin-orbit torque and create a single aligned domain.","lead":"This paper shows that current pulses switch the magnetic order of the antiferromagnet Mn2Au by two different mechanisms depending on pulse length: slow thermal strain for long pulses, and a fast spin-orbit torque for nanosecond pulses. The nanosecond mechanism can align the entire magnetic order in one direction, which is what antiferromagnetic memory devices would need.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 2.5 ns 'purely NSOT' claim relies on a strain-axis discriminator computed only for 10 µs pulses; at 2.5 ns the thermal diffusion length is sub-micron, so the 45°-cross strain direction may differ and the geometry test may not exclude thermomagnetoelastic switching.","rationale":"The paper's long-pulse results are convincing: at 10 µs, the 0° and 45° crosses switch into mutually perpendicular orientations that match the COMSOL strain axes, cleanly identifying thermomagnetoelastic switching. The geometry-based idea is sound and the 100 ns resistance data in Fig. 4a provide a useful internal consistency check. However, the extension of this discriminator to 2.5 ns is the load-bearing step for the central claim. The COMSOL simulations are only shown for 10 µs, and the text merely asserts that strain decreases with pulse width without quantifying it at 2.5 ns. At 2.5 ns, heat has not diffused over the device length scale: using the Supplement's parameters, the thermal diffusion length is well below 1 µm in both MgO and Mn2Au, so the temperature distribution is set by the instantaneous Joule heating pattern rather than the quasi-steady-state pattern used to compute the strain axes. Consequently, the direction of the in-plane strain anisotropy in the 45° cross at 2.5 ns cannot be assumed identical to the 10 µs case. If the strain anisotropy at 2.5 ns were perpendicular to the current, the observed bright XMLD contrast would be explained by thermomagnetoelastic switching, and the geometric argument would no longer rule it out. The paper's own Fig. 4a shows that thermomagnetoelastic switching occurs at 100 ns at sufficiently high current density, so this mechanism is not inherently too slow to appear in the nanosecond regime. The missing 2.5 ns thermal-stress simulation is a concrete, testable gap rather than a demonstrated flaw; the verdict should therefore remain conditional, with the condition being a quantitative strain test at the relevant pulse width and geometry.","tokens_in":11515,"tokens_out":13239,"duration_ms":115495,"concrete_test":"Run the same COMSOL thermal-stress model used for Fig. 2d–f with the pulse width set to 2.5 ns and J = 3×10^12 A/m² for the 45° cross geometry, and extract the time-resolved in-plane strain anisotropy (e.g., epsilon_[110] − epsilon_[1-10]) at the cross center at t = 2.5 ns. Repeat with plausible upper and lower bounds of the unknown thermal parameters (e.g., k = 50–400 W/(m·K), c = 100–400 J/(kg·K)). If the sign of the center strain anisotropy at 2.5 ns is opposite to that at 10 µs (i.e., it would align the Néel vector perpendicular to current), or if its magnitude is not at least an order of magnitude below the 10 µs switching value, the observed bright XMLD contrast in Fig. 2j does not uniquely support NSOT, and the central claim is not established. If the sign stays parallel and magnitude is negligible, the discriminator survives.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim that 2.5 ns switching is purely NSOT rests on the observation that both 0° and 45° crosses show the Néel vector perpendicular to the current. This discriminator is valid only if the thermomagnetoelastic strain axis in the 45° cross remains parallel to the current at 2.5 ns, as COMSOL shows for 10 µs (Fig. 2d–f). No 2.5 ns simulation is presented. The thermal diffusion lengths at 2.5 ns are ≈0.2 µm in MgO and ≈0.7 µm in Mn2Au (using the Supplement's parameters), far smaller than the ~10 µm cross features, so the temperature field and strain anisotropy are not the quasi-steady-state ones used to establish the discriminator. If the 45°-cross strain anisotropy at 2.5 ns were perpendicular to the current, the observed bright XMLD contrast would be thermomagnetoelastic, not NSOT. The paper's own Fig. 4a shows thermomagnetoelastic switching can occur at 100 ns at higher current density, so this mechanism is not categorically absent in the nanosecond regime. The claim is therefore conditional on unverified strain behavior at 2.5 ns.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports an experimental study of current-induced Néel vector switching in epitaxial Mn2Au(001) devices, distinguishing two mechanisms: bulk Néel spin-orbit torque (NSOT) and thermomagnetoelastic strain. Using 0° and 45° cross geometries, the authors show that 10 µs current pulses produce Néel vector alignment whose direction follows the simulated strain axis (perpendicular to current in the 0° cross, parallel in the 45° cross), consistent with thermomagnetoelastic switching. For 2.5 ns pulses, both geometries show alignment perpendicular to the current, which the authors attribute to purely NSOT-driven switching; they further observe a predominantly single domain after the nanosecond pulse, which they interpret as directional (rather than axis) alignment. The paper also presents transverse resistance measurements showing a crossover at 100 ns where both mechanisms can operate at different current densities.","tokens_in":11864,"tokens_out":5313,"duration_ms":45105,"significance":"If the conclusions hold, the paper provides a valuable experimental protocol for distinguishing NSOT from thermomagnetoelastic switching in antiferromagnets, and strong evidence that bulk NSOT in Mn2Au can drive complete directional Néel vector alignment on nanosecond timescales. The geometry-based discriminator and the single-domain observation are conceptually clear and potentially reproducible. The use of XMLD-PEEM imaging, exchange-coupled Py readout, and resistance measurements gives a multi-modal characterization. However, the central nanosecond claim depends on assumptions about the strain direction at pulse widths for which no simulations are presented, and on the interpretation of the domain structure; these are load-bearing and not fully resolved.","major_comments":[{"comment":"The conclusion that 2.5 ns switching is purely NSOT rests on the assumption that the thermomagnetoelastic strain in the 45° cross is parallel to the current at that pulse width, as it is at 10 µs. The COMSOL simulations shown in Fig. 2d–f are only for 10 µs pulses, and at 2.5 ns the thermal diffusion length in MgO and Mn2Au (using the Supplement's parameters) is approximately 0.2 µm and 0.7 µm, respectively, much smaller than the cross dimensions; the strain field therefore need not have the same anisotropy as the quasi-steady-state case. If the 45° cross strain at 2.5 ns were perpendicular to the current, the observed bright XMLD contrast would be consistent with thermomagnetoelastic switching rather than NSOT. The authors should provide transient COMSOL simulations at 2.5 ns, or a quantitative argument for the strain direction at that pulse width, before claiming a purely NSOT mechanism.","section":"Results and Discussion, Fig. 2"},{"comment":"The abstract and summary state that for 'pulses shorter than 100 ns' switching is driven purely by NSOT, but Fig. 4a shows that for 100 ns pulses, NSOT switching at lower current densities is followed by thermomagnetoelastic switching at higher current densities. This indicates that the thermomagnetoelastic mechanism is not absent at 100 ns, and the phrase 'purely NSOT' should be qualified to the current-density range in which NSOT dominates. The manuscript should clarify the boundary of the NSOT-only regime and how the 100 ns crossover data relate to the abstract's claim.","section":"Fig. 4a and Summary"},{"comment":"The single-domain observation after 2.5 ns is used as evidence for NSOT because thermomagnetoelastic switching is argued to produce 180° domains. This argument assumes that a strong thermomagnetoelastic strain cannot select one of the two antiparallel directions along the strain axis, e.g., via inhomogeneous strain or a secondary torque. The authors should either provide a supporting calculation or explicitly state this symmetry assumption as a limitation, since the XMLD contrast alone cannot distinguish antiparallel directions except through the visibility of 180° domain walls.","section":"Supplemental Material II and Fig. S3"}],"minor_comments":[{"comment":"There are several typographical and grammatical errors, such as 'the compound investigate here' (Introduction) and 'could enabling novel applications' (Summary & Conclusion); these should be corrected.","section":"Throughout"},{"comment":"The text refers to 'Fig. 1g' of the main manuscript when describing the 10 µs switched domain configuration, but the corresponding image appears in Fig. 2g; please update the cross-reference.","section":"Supplemental Material I"},{"comment":"Reference [26] is a placeholder ('URL will be inserted by publisher') and must be resolved before publication.","section":"References"},{"comment":"The caption for panel d says the current flows 'from both right arms to both left arms', which is initially confusing; consider adding a schematic with current arrows in the figure or a more explicit description.","section":"Fig. 2 caption"}],"recommendation":"major_revision","confidential_remarks":"The stress-test concern regarding the 2.5 ns strain direction is well founded and should be addressed head-on. The paper's 10 µs discriminator is convincing, but the nanosecond 'purely NSOT' claim is not yet fully supported without simulations or an explicit estimate of the strain tensor at 2.5 ns. The 100 ns crossover data in Fig. 4a also sit uneasily with the abstract's wording. This is a strong and important manuscript, but these load-bearing points need to be resolved before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read the Mn2Au switching paper. The geometry trick is genuinely good: comparing 0° and 45° crosses lets them flip the predicted strain axis while keeping the NSOT axis fixed. The 10 µs data fall right where the strain model says, which is solid evidence that thermomagnetoelastic effects dominate at long pulses. That part I buy.\n\nThe stronger claim is that 2.5 ns pulses switch purely by bulk NSOT, with complete directional alignment into one domain. The evidence is not as clean. The COMSOL simulations that establish the strain-direction discriminator are done only for 10 µs pulses; the temperatures and strain at 2.5 ns are computed with parameters the authors admit are \"typical values for metals\" for Mn2Au, and no 2.5 ns strain simulation is shown. At 2.5 ns the thermal diffusion length is sub-micron, much smaller than the device features, so the strain anisotropy in the 45° cross could in principle differ from the quasi-steady-state direction. If it flipped, the observed perpendicular alignment would be thermomagnetoelastic, not NSOT. That is a real gap.\n\nThe paper does have a second line of argument: the 2.5 ns images show a single large domain rather than closed 180° loops, which is what you'd expect from NSOT, not from an easy-axis mechanism. That is suggestive, but it's visual interpretation of XMLD-PEEM with no quantitative statistics and no statement on how many devices were imaged. The NiFe bilayer experiment is a nice addition and does show one-to-one correspondence, but again it's a single demonstration.\n\nThe 100 ns data are actually the most convincing for the crossover: resistance traces show NSOT-like switching at lower current density and thermomagnetoelastic at higher current density, meaning both mechanisms are present in the nanosecond regime. That tells me the authors are not just seeing thermal effects everywhere. But it also means \"purely NSOT\" at 2.5 ns is a claim about the absence of a strain effect that they have not directly measured.\n\nThe citation pattern looks appropriate; the Zelezny prediction is tested against data, not fitted. No serious circularity.\n\nBottom line: this is a worthwhile paper for the antiferromagnetic spintronics community. The long-pulse mechanism separation is a solid contribution, and the NSOT claim is plausible and potentially important. It deserves a serious referee, but the referee should push for short-pulse strain simulations with a realistic parameter range, direct temperature/strain measurements if possible, and more statistics on the single-domain observation before the \"purely NSOT\" statement is accepted.","headline":"A clever geometry trick separates thermomagnetoelastic from NSOT switching at long pulses; the nanosecond pure-NSOT claim is plausible but rests on an untested strain-direction assumption and qualitative domain imaging.","tokens_in":12399,"tokens_out":5954,"would_cite":true,"duration_ms":50178,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":["75.70.Tj","75.50.Ee"],"model":"deepseek-v4-flash","headline":"Nanosecond pulses switch Mn2Au's Néel vector by pure spin-orbit torque.","keywords":["antiferromagnetic spintronics","Néel spin-orbit torque","Mn2Au","thermomagnetoelastic switching","antiferromagnetic domain switching","XMLD-PEEM imaging","ultrafast current pulses","current-induced switching"],"falsifier":"Measure the transient strain and temperature in the Mn2Au crosses during a 2.5 ns, ~3×$10^{12}$ A/m² current pulse (for example by time-resolved x-ray diffraction) and compute the magnetoelastic anisotropy using measured elastic constants of Mn2Au. If the induced strain is large enough to rotate the Néel vector into the observed perpendicular direction, the claim of pure NSOT switching is falsified.","tokens_in":11355,"feed_emoji":"⚡","tokens_out":7108,"duration_ms":58252,"temperature":0.7,"pith_summary":"This paper sets out to tell apart two mechanisms that can rotate the staggered magnetization (the Néel vector) of the antiferromagnet Mn2Au when a current pulse passes through it: a fast bulk Néel spin-orbit torque (NSOT) and a slower thermomagnetoelastic effect driven by current-induced heating and strain. The authors claim that, for pulses of 10 microseconds or longer, thermal strain dominates and aligns the Néel vector along an axis whose direction depends on device geometry, not on current direction. For pulses shorter than 100 nanoseconds, they claim the reorientation is purely NSOT, aligning the Néel vector along one specific direction perpendicular to the current and erasing 180-degree domain walls to leave a single large antiferromagnetic domain. The result matters because it provides experimental evidence that the bulk NSOT predicted for Mn2Au is strong enough for ultrafast antiferromagnetic switching, and it gives researchers a geometry-based way to avoid mistaking thermal effects for spin-orbit torques.","feed_headline":"Nanosecond pulses switch Mn2Au by pure spin-orbit torque","feed_subtitle":"A geometry-based cross test separates bulk Néel spin-orbit torque from slower thermal-strain switching.","key_machinery":"The load-bearing object is the staggered effective field produced by a current in a bulk antiferromagnet with inversion symmetry breaking: $\\mathbf{B}_A \\propto +\\hat{z}\\times \\mathbf{J}$ and $\\mathbf{B}_B \\propto -\\hat{z}\\times \\mathbf{J}$, where $\\hat z$ is the film normal and $\\mathbf J$ the in-plane current. These fields generate field-like torques that cant both sublattices the same way, and the resulting exchange torque rotates the Néel vector around the film normal; the torque is maximum when the Néel vector is parallel or antiparallel to $\\mathbf J$ and vanishes when it is perpendicular. The argument is carried experimentally by comparing two cross geometries: in a 0° cross, the thermal-strain easy axis lies perpendicular to the current, while in a 45° cross it lies parallel, so any mechanism that follows the current direction (NSOT) can be distinguished from one that follows the strain axis (thermomagnetoelastic). NSOT also removes 180° domain walls because it drives both antiparallel orientations toward the same perpendicular direction, whereas strain only fixes an axis and leaves both directions populated.","core_discovery":"The central discovery is that the two known current-induced switching mechanisms in Mn2Au can be cleanly separated by geometry and pulse width. For 10 µs and 1 ms pulses, the final Néel vector orientation follows the anisotropic thermal strain pattern computed for the device: perpendicular to current in a 0°-oriented cross and parallel to current in a 45°-oriented cross, meaning the thermomagnetoelastic effect is the driver. For 2.5 ns pulses, both geometries show the Néel vector aligned perpendicular to the current, which matches the NSOT torque direction and not the strain axis; moreover, the switched area is one connected domain with only small residual 180° walls. Electrical transport measurements place the crossover near 100 ns and show complete NSOT switching at current densities of 2–3 × $10^{12}$ A/m², more than five times the density needed for thermomagnetoelastic switching. The authors conclude that bulk NSOT, predicted a decade earlier, is sufficient to switch Mn2Au and to create a single-domain state.","pith_inferences":["If the thermal-strain assumption holds, the 0°/45° cross comparison could be adopted as a general diagnostic for separating intrinsic spin-orbit torques from thermomagnetoelastic effects in other collinear antiferromagnets.","The creation of a single connected AFM domain with a known Néel vector direction implies a memory bit that is not just axis-aligned but directionally encoded; exploiting it would require a readout that resolves antiparallel states, which XMLD contrast alone cannot do.","A direct test would be to measure Mn2Au's thermal conductivity, heat capacity, thermal expansion, Young's modulus, and Poisson ratio, then rerun the finite-element strain calculation; if the computed strain at 2.5 ns is large enough to reorient the Néel vector, the pure-NSOT conclusion would need revision.","The pulse-width dependence suggests a crossover scaling law: thermal switching should scale with deposited energy and heat diffusion time, while NSOT switching should scale with current density; mapping this crossover across pulse widths and device sizes could predict the fastest reliable NSOT switching point."],"forward_implications":["For pulse widths of 100 ns and below, Mn2Au devices can be switched by the bulk Néel spin-orbit torque alone, with no thermal-strain contribution at the switching current density.","NSOT switching yields a deterministic single antiferromagnetic domain: the Néel vector points along a specific in-plane direction perpendicular to the applied current, not merely along an easy axis.","Thermomagnetoelastic switching dominates for microsecond and millisecond pulses, so studies claiming SOT-driven switching in collinear antiferromagnets must rule it out using geometry or pulse-width dependence.","The 100 ns crossover implies an operating window where NSOT switching occurs at lower current densities than thermal switching; at 100 ns the same sample shows NSOT first and thermomagnetoelastic switching only at higher current densities.","Current densities around 2–3 × 10^12 A/m² are sufficient for complete NSOT switching in the nanosecond regime, establishing practical requirements for ultrafast antiferromagnetic memory writing."],"supporting_citations":[{"why":"Derives the staggered effective fields and the Néel spin-orbit torque direction in Mn2Au used to predict perpendicular alignment.","marker":"[9]"},{"why":"Shows that current-induced antiferromagnetic switching can be purely thermomagnetoelastic, the competing mechanism the paper must distinguish from NSOT.","marker":"[17]"},{"why":"Reports current-driven writing in Mn2Au devices, providing the prior experimental context this paper extends by identifying the switching mechanism.","marker":"[14]"},{"why":"Establishes the strong exchange coupling between Mn2Au's Néel vector and Ni80Fe20 magnetization used to visualize the switched single-domain state.","marker":"[23]"},{"why":"Provides the finite-element simulations of temperature and strain distributions whose geometry-dependent easy axis is compared with experiment.","marker":"[24]"},{"why":"Documents the initial four-domain Néel vector configuration of the Mn2Au films from which switching starts.","marker":"[25]"}],"fun_headline_variants":["Mn2Au switching: ultrafast NSOT wins at nanosecond pulses","Geometry separates spin-orbit torque from thermal strain switching","Nanosecond pulses enable pure Neel spin-orbit torque in Mn2Au","Cross test reveals bulk NSOT switching in antiferromagnet","Ultrafast NSOT switching of Mn2Au demonstrated"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The claim that nanosecond switching is purely NSOT rests on the assumption that current-induced thermal strain in the Mn2Au films is negligible at 2.5 ns pulse widths; the finite-element simulations supporting this use generic 'typical values for metals' because Mn2Au's thermal and elastic parameters are unpublished.","fun_headline_variants_meta":{"raw":{"variants":["Mn2Au switching: ultrafast NSOT wins at nanosecond pulses","Geometry separates spin-orbit torque from thermal strain switching","Nanosecond pulses enable pure Neel spin-orbit torque in Mn2Au","Cross test reveals bulk NSOT switching in antiferromagnet","Ultrafast NSOT switching of Mn2Au demonstrated"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000663,"raw_usage":{"total_tokens":2996,"prompt_tokens":877,"completion_tokens":2119,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":493,"completion_tokens_details":{"reasoning_tokens":2031}},"tokens_in":493,"tokens_out":2119,"duration_ms":12934,"temperature":1.0,"reasoning_tokens":2031,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T10:59:48.271046+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the transient strain and temperature in the Mn2Au crosses during a 2.5 ns, ~3×$10^{12}$ A/m² current pulse (for example by time-resolved x-ray diffraction) and compute the magnetoelastic anisotropy using measured elastic constants of Mn2Au. If the induced strain is large enough to rotate the Néel vector into the observed perpendicular direction, the claim of pure NSOT switching is falsified.","supporting_citations":[{"cited_title":"Relativistic N´ eel-Order Fields Induced by Elec- trical Current in Antiferromagnets,Phys","cited_arxiv_id":null,"evidence_quote":"Derives the staggered effective fields and the Néel spin-orbit torque direction in Mn2Au used to predict perpendicular alignment."},{"cited_title":"Direct Imaging of Current-Induced Antiferromag- netic Switching Revealing a Pure Thermomagnetoelastic Switching Mechanism in NiO.Nano Lett.21, 114 (2021)","cited_arxiv_id":null,"evidence_quote":"Shows that current-induced antiferromagnetic switching can be purely thermomagnetoelastic, the competing mechanism the paper must distinguish from NSOT."},{"cited_title":"R., Golias, 8 E., Sarpi, B., Veiga, L","cited_arxiv_id":null,"evidence_quote":"Reports current-driven writing in Mn2Au devices, providing the prior experimental context this paper extends by identifying the switching mechanism."},{"cited_title":"P., Backes, D., Veiga, L","cited_arxiv_id":null,"evidence_quote":"Establishes the strong exchange coupling between Mn2Au's Néel vector and Ni80Fe20 magnetization used to visualize the switched single-domain state."},{"cited_title":"J., Krizek, F., Bar- ton, L","cited_arxiv_id":null,"evidence_quote":"Documents the initial four-domain Néel vector configuration of the Mn2Au films from which switching starts."}],"review_version":1}