{"id":"02903bcf-ad21-490d-b825-90fe34e13bb3","arxiv_id":"2506.13448","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"DFT and Monte Carlo calculations predict that MnBi2Te4/MnBr3 heterostructures raise the MnBi2Te4 Curie temperature to 72 K when the Hubbard U in MnBr3 is 3 eV, with coupling strength controlled by U-driven structural distortions.","lead":"This computational paper predicts that stacking a monolayer of the magnetic topological insulator MnBi2Te4 on a monolayer of MnBr3 can raise the magnetic ordering temperature of the MnBi2Te4 layer from about 13 K to 72 K, a four- to fivefold increase. The result depends on the assumed strength of electronic correlations (Hubbard U) in MnBr3, which the authors propose as a knob for controlling interlayer magnetic coupling.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"MC spin model is not fully specified: without single-ion anisotropy terms, 2D isotropic Heisenberg cannot yield finite TC; all quoted TC values rest on hidden Hamiltonian terms.","rationale":"The paper's argument has a clear causal chain: U2 controls MnBr3 distortions, distortions control interlayer Jz, and Jz controls whether the MC heat capacity shows one or two transitions and how high TC is. Most of the supporting DFT analysis (charge transfer, LDOS, Table I) is internally consistent, and the ML-MBT validation to about 13 K is an encouraging independent check. The weakest link is the step where finite-temperature order is obtained: in two dimensions, the MC simulation must include a term that the printed Hamiltonian omits. Since the loaded claim is a factor-of-four-to-five TC enhancement, this is the single most load-bearing point. The paper may well have the anisotropy in the unstated PASP output; the fix is documentation and a robustness test, not a fundamental reworking. I therefore keep the reader's CONDITIONAL verdict rather than moving to reject or accept. I agree with the reader's weakest-assumption identification, though I would center it on the undisclosed anisotropic term rather than on the choice of U2, which is a legitimate design parameter in a DFT+U study.","tokens_in":12200,"tokens_out":4329,"duration_ms":44436,"concrete_test":"Ask the authors to output the complete PASP Hamiltonian for U2 = 3.0 eV, including all single-ion anisotropy coefficients, and the fit RMSE on the 100 held-out configurations. Then rerun the MC heat-capacity calculation with (i) all anisotropy coefficients set to zero and (ii) the easy-axis coefficient scaled by 0.5x and 2.0x. If case (i) yields no magnetic transition or case (ii) shifts either peak by more than about 20%, the TC values depend sensitively on the undisclosed term and the central claim should be treated as conditional. If the peaks are stable under these perturbations, the concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claim is the Monte Carlo-derived Curie temperatures (72 K, 56 K, 158 K), but the effective spin Hamiltonian actually used in the Monte Carlo step is not fully disclosed. Equation (1) in Section III.B contains only bilinear isotropic Heisenberg terms, yet a strictly isotropic 2D Heisenberg magnet has no finite-temperature order by the Mermin-Wagner theorem. The paper's own ML-MBT validation gives TC about 13 K, so the PASP Hamiltonian must contain some additional anisotropy or symmetry-breaking term; that term is never written down, parameterized, or validated. If the hidden term is absent or incorrectly fitted, the finite-TC values, and hence the claimed four-to-fivefold enhancement and the single-versus-split transition behavior, are unsupported. The absence of reported training/test error for the ML fit compounds this: there is no check that the spin model reproduces the DFT energies for the heterostructure, especially for the strong interlayer couplings that control the U2 = 3.0 eV result.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript proposes a van der Waals heterostructure consisting of monolayer MnBi2Te4 (ML-MBT) and monolayer MnBr3, and claims, on the basis of DFT+U calculations, machine-learned spin-Hamiltonian fitting (PASP), and Monte Carlo simulations, that the Curie temperature of the ML-MBT layer is enhanced by a factor of four to five, from about 13 K to 72 K, at a Hubbard U2 = 3.0 eV, with a single unified magnetic transition. At U2 = 5.0 eV the authors find that the layers magnetically decouple, producing two transitions at 56 K and 158 K. The proposed control mechanism is that the strength of U2 modifies structural distortions in MnBr3, which in turn set the interlayer ferromagnetic exchange interactions.","tokens_in":12439,"tokens_out":3794,"duration_ms":43209,"significance":"If the central prediction holds, the paper offers an interesting materials-design route: using electronic correlations in one layer to control interlayer magnetic coupling and the magnetic transition temperature of a topological-insulator monolayer. The computational pipeline is forward and non-circular: the Hubbard parameters are inputs, the Heisenberg exchanges are fitted to DFT energies, and the ML-MBT benchmark gives a TC of about 13 K, consistent with previous work. The claimed sensitivity of the result to U2 is honestly presented, and the authors explicitly note the difficulty of determining U2 from first principles. The main weaknesses are the incomplete disclosure of the spin Hamiltonian actually used in the Monte Carlo step and the lack of a quantitative justification for the specific U2 = 3.0 eV choice, both of which are load-bearing for the numerical TC values.","major_comments":[{"comment":"The spin Hamiltonian written in the paper contains only bilinear isotropic Heisenberg exchange terms. As the authors themselves note in the Introduction, the Mermin-Wagner theorem forbids finite-temperature order in a two-dimensional isotropic Heisenberg model, so the finite Curie temperatures reported in Fig. 3 (13 K, 72 K, 56 K, 158 K) must arise from additional anisotropy or symmetry-breaking terms in the PASP Hamiltonian that are never written down, parameterized, or validated. Please provide the complete Hamiltonian actually used in the Monte Carlo simulations, including any single-ion anisotropy, Dzyaloshinskii-Moriya, or other spin invariants, together with their fitted values, and show how the quoted TCs depend on those terms. Without this, the numerical transition temperatures are not reproducible from the information given in the paper.","section":"§III.B, Eq. (1)–(4)"},{"comment":"The paper states that 500 random spin configurations were generated, with 400 used for training and 100 for testing, but no test error or fitting accuracy is reported anywhere. This is especially important because the central conclusion rests on the interlayer exchange couplings, which are small in magnitude compared with the intralayer couplings and are claimed to change substantially between U2 = 3.0 and 5.0 eV. Please report the root-mean-square error or a parity plot for the training and test sets, and specify the accuracy of the fitted interlayer couplings J_z. This is needed to establish that the machine-learned Hamiltonian faithfully reproduces the DFT energies, including for the configurations that determine the interlayer interactions.","section":"§II, PASP fitting; §III.B, Fig. 2"},{"comment":"The central quantitative claim—the four- to fivefold enhancement of TC to 72 K and the single-transition behavior—is obtained only for U2 = 3.0 eV, while U2 = 5.0 eV gives a qualitatively different result (two transitions at 56 K and 158 K). The paper offers no independent determination of U2, such as constrained random-phase approximation or a fit to an experimental observable. Since the choice U2 = 3.0 eV is the linchpin of the main result, please either provide a first-principles or experimental justification for this value, or explicitly reframe the abstract and introduction so that the 72 K enhancement is presented as a conditional prediction of a parameter scan rather than as the paper's headline result. The summary does acknowledge the U2-dependence, but the framing in the Abstract and Introduction overstates the robustness of the 72 K value.","section":"§III.B, Table I and Fig. 3; §IV"},{"comment":"The mechanism claim that structural distortions at U2 = 3.0 eV cause the enhanced interlayer coupling rests on the LDOS difference quantity Δdos, whose definition depends on the assignment of 'central' and 'nearest' Mn atoms in regions A and B. The paper does not state the numerical criteria used to define the two triangular regions or how many nearest neighbors are averaged. Please specify these definitions precisely, because the correlation between Δdos and J_z is central to the proposed seesaw mechanism and is otherwise not quantitatively assessable.","section":"§III.C, Fig. 5"}],"minor_comments":[{"comment":"The convergence criterion 'total energy is less than 1×10⁻7 eV' should read '1×10⁻7 eV/atom' or equivalent, and the superscript formatting is corrupted in the manuscript text.","section":"§II, Methods"},{"comment":"The text contains a line break inside 'Néel temperature' ('Né el') and several other OCR-type artifacts (e.g., 'U1 = 5.0 eV is fixed' appears with missing characters in Fig. 2 caption area). Please proofread the final text.","section":"Abstract and Introduction"},{"comment":"The black dashed curve is described as the 'numerical sum' of the isolated monolayer heat-capacity curves, but adding heat capacities does not yield the heat capacity of a non-interacting two-layer system unless the energy scales are comparable and the layer degrees of freedom are independent. As a schematic reference this is fine, but the wording should be softened to avoid implying a quantitatively rigorous decoupled limit.","section":"§III.B, Fig. 3"},{"comment":"The notation J_a^μ h_μ is not self-contained: the basis functions h_μ are deferred to the Supplemental Material, but the main text should at least state the order of the expansion and the number of independent parameters per shell, so that the reader can see how the 9+9+4+8+4+14 terms in Eqs. (2)–(4) are counted.","section":"§III.B, Eq. (2)–(4)"}],"recommendation":"major_revision","confidential_remarks":"The core physics idea is timely and the forward pipeline is a strength, but the two load-bearing gaps—the undisclosed anisotropy terms in the Monte Carlo Hamiltonian and the unjustified U2 = 3.0 eV choice—need to be addressed before publication. I do not see this as a reject situation: both are fixable within the manuscript's scope. I would be willing to review a revised version."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper's real novelty is the MBT/MnBr3 combination and the claim that the Hubbard U on the MnBr3 layer controls interlayer coupling through subtle structural distortions. The seesaw behavior between intralayer and interlayer exchange is a nice piece of physics, and the ML-MBT validation (Tc about 13 K) is a good benchmark. The electron-counting argument for FM interlayer coupling is consistent with prior literature. So there is genuine content here.\n\nThe soft spots are real but addressable. The biggest one: Equation (1) writes only isotropic Heisenberg terms, yet a strictly isotropic 2D Heisenberg model has no finite-Tc order. PASP must be fitting single-ion anisotropy or other symmetry-lowering terms somewhere, because the same pipeline gives 13 K for ML-MBT, but those terms are never shown, parameterized, or validated. The training/test behavior of the ML fit is also unreported. That means a reader cannot reproduce or fully trust the specific Tc numbers (72, 56, 158 K) from the text. This is the kind of omission a referee should catch and the authors can fix.\n\nSecond, the central prediction hangs on U2 = 3.0 eV. The authors are honest that 5.0 eV gives a different qualitative answer and even suggest using the transition as a probe of U, but that leaves the 72 K claim conditional. They need a stronger argument for why 3.0 eV is physical, or at least a systematic scan over reasonable U values with a discussion of which regime is likely.\n\nOverall, this is a solid computational design study, not a finished result. The mechanism is plausible and the system is timely. I would send it to peer review and ask for the full Hamiltonian, ML test error, and a more robust U2 justification. The counting of isolated monolayers in the MC benchmark is also slightly crude but acceptable.","headline":"New heterostructure, new mechanism, and a sane computational pipeline, but the headline Tc values are less solid than they look because the spin Hamiltonian used in Monte Carlo is never fully disclosed, and the key number is conditional on a single Hubbard U setting.","tokens_in":12999,"tokens_out":3532,"would_cite":true,"duration_ms":37248,"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":"This paper claims that stacking a monolayer of MnBi2Te4 with one of MnBr3 raises the MnBi2Te4 layer's Curie temperature from about 13 K to 72 K, and that the Hubbard U on the MnBr3 layer decides whether the two layers order together or…","keywords":["MnBi2Te4","MnBr3","van der Waals heterostructure","Curie temperature enhancement","Hubbard U","interlayer magnetic coupling","magnetic phase separation","Monte Carlo simulation"],"falsifier":"Recompute the transition temperatures with the full spin Hamiltonian written out explicitly, including single-ion anisotropy and any other symmetry-breaking terms, and report the error of the machine-learning fit on the held-out 100 spin configurations; if the 72 K transition does not survive an anisotropy-corrected simulation, the enhancement claim is not settled. In parallel, a measured magnetization-versus-temperature curve of a grown stack that shows a single transition near 72 K would support the picture, while two separate transitions near 56 K and 158 K would match the decoupled $U_2 = 5.0$ eV case.","tokens_in":11999,"feed_emoji":"🧲","tokens_out":11850,"duration_ms":105122,"temperature":0.7,"pith_summary":"This paper argues that a van der Waals stack of one monolayer of MnBi2Te4 and one monolayer of MnBr3 can raise the Curie temperature of the MnBi2Te4 layer from about 13 K to 72 K, a four- to fivefold increase. The control knob is the Hubbard U on the Mn-3d orbitals of MnBr3: at a moderate value (3.0 eV) the MnBr3 layer develops slight structural distortions that break its intralayer symmetry, strengthen the interlayer ferromagnetic exchange, and make the whole stack order in a single magnetic transition. At a stronger value (5.0 eV) the distortions disappear, the interlayer coupling weakens, and the two layers order separately at 56 K and 158 K. A sympathetic reader would care because the result points to a practical route for engineering higher-temperature two-dimensional magnets and suggests that magnetic transition behavior could be used to read off the strength of electronic correlations in a material.","feed_headline":"Stacking lifts MnBi2Te4's Curie temperature from 13 K to 72 K","feed_subtitle":"Moderate correlations couple layers into one magnet; stronger correlations split them into 56 K and 158 K transitions.","key_machinery":"The argument is carried by a Heisenberg spin Hamiltonian fitted to density-functional energies through a machine-learning procedure and then simulated with Metropolis Monte Carlo on a $9 \\times 9 \\times 1$ supercell. Within that Hamiltonian the load-bearing quantity is the interlayer exchange $J_z^{B1}$, the dominant coupling between the Mn atoms of MnBr3 and the Mn atoms of ML-MBT; its sign and magnitude decide whether the stack shows one transition or two. The mechanism behind its variation is a seesaw between intralayer and interlayer exchange: as $U_2$ grows, the intralayer nearest-neighbor exchange $J_1^A$ strengthens while $J_z^{B1}$ weakens, and the structural asymmetry measured by Mn-Br bond-length and bond-angle differences and by the local-density-of-states splitting $\\Delta_{\\mathrm{dos}} = \\mathrm{LDOS}(\\text{central Mn}) - \\mathrm{LDOS}(\\text{nearest Mn})$ shrinks, vanishing near 3.8 eV. The Hamiltonian is built from Heisenberg exchange terms within each layer and across the interface, and the Monte Carlo simulation converts those fitted exchange constants into the reported transition temperatures.","core_discovery":"The central discovery is that interfacing a monolayer of MnBi2Te4 (ML-MBT) with a monolayer of MnBr3 gives a ferromagnetically coupled stack in which ML-MBT orders at 72 K when the Hubbard $U_2$ on MnBr3 is 3.0 eV, compared with roughly 13 K for the isolated monolayer. The authors attribute the enhancement to $U_2$-dependent structural distortions in the MnBr3 layer: at 3.0 eV the two triangular regions of the MnBr3 supercell become inequivalent, producing a measurable asymmetry in Mn-Br bond lengths and bond angles, a splitting in the local density of states, and a strong interlayer exchange $J_z^{B1} = -6.91$ meV. Raising $U_2$ to 5.0 eV restores a nearly symmetric structure, drops that exchange to $-2.09$ meV, and drives the intralayer exchange $J_1^A$ from $-9.51$ to $-14.21$ meV; the result is two separate magnetic transitions at 56 K and 158 K, which the paper reads as the two layers decoupling. The same electron transfer from ML-MBT to MnBr3 is spin-polarized and weakens as $U_2$ grows, tying the charge redistribution to the magnetic coupling.","pith_inferences":["The paper's own numbers imply a sharp crossover near $U_2 \\approx 3.8$ eV where the LDOS splitting vanishes; a systematic scan of chemically substituted analogues with different effective $U$ could look for the predicted jump from one magnetic transition to two.","Because the mechanism works through structural distortion, biaxial strain on the MnBr3 layer could act as a substitute for tuning $U_2$: straining the layer toward the 3.0-eV geometry might reproduce the single-transition state even in a sample with stronger correlations.","The 72 K transition could also be read as proximity-induced ordering of the MBT layer by the high-$T_C$ MnBr3 rather than an intrinsic enhancement; a layer-resolved magnetization or element-specific measurement would separate the two readings.","If the anisotropy terms are as small as the paper's isotropic model implies, the transition temperatures should depend strongly on the simulation cell size; a finite-size scaling check would put a number on that sensitivity."],"forward_implications":["Heterostructure assembly becomes a tunable lever: the same two materials can be made to order as one magnet at 72 K or as two magnets at 56 K and 158 K depending on the effective correlation strength in the MnBr3 layer.","A double-peaked heat capacity, rather than a single peak, becomes a fingerprint of interlayer decoupling and could be used in experiments to infer which $U$ regime a sample sits in.","The spin-polarized charge transfer from ML-MBT to MnBr3 means the interface simultaneously modifies the magnetic order and the electronic structure of the topological layer, so the stack is a candidate platform for proximity-tuned topological or spintronic devices.","The electron-counting rule that favors interlayer ferromagnetism at type-II/type-I interfaces suggests the design principle may transfer to other pairs of d-electron van der Waals monolayers, not only MnBi2Te4 and MnBr3."],"supporting_citations":[{"why":"Electron-counting rule used to predict that the type-II/type-I interface between ML-MBT and MnBr3 favors interlayer ferromagnetic coupling.","marker":"[44]"},{"why":"Supplies the machine-learning fitting and Monte Carlo simulation procedure that converts DFT energies into exchange constants and Curie temperatures.","marker":"[41]"},{"why":"Reported MnBr3 monolayer as a high-Curie-temperature (~200 K) Dirac half-metal, the component whose correlations the paper tunes.","marker":"[35]"},{"why":"Earlier ML-MBT exchange constants and ~13 K Curie temperature used to validate the fitting method before adding the MnBr3 layer.","marker":"[47]"},{"why":"Documents the near-isotropic magnetic anisotropy and low ordering temperature of ML-MBT that the heterostructure is designed to overcome.","marker":"[24]"},{"why":"Prediction of manganese trihalides as two-dimensional Dirac half-metals, supporting the choice of MnBr3 as a topological magnetic partner.","marker":"[42]"},{"why":"Supplies the GGA+U scheme that defines the Hubbard $U_2$ parameter whose variation drives the reported transition changes.","marker":"[39]"}],"fun_headline_variants":["Stacking MnBr3 lifts MnBi2Te4's Curie point to 72 K","MnBi2Te4/MnBr3: fivefold Tc boost via Hubbard U","One magnet at U=3 eV, two at U=5 eV in MnBi2Te4/MnBr3","Hubbard U tunes MnBi2Te4/MnBr3 from one magnet to two","Stacking with MnBr3 raises MnBi2Te4's Tc fivefold"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is that the Monte Carlo spin Hamiltonian the paper fits is complete enough to describe a two-dimensional magnet, including the magnetic anisotropy that allows order at finite temperature; the paper does not display those terms, and the quoted transition temperatures of 13, 72, 56, and 158 K would shift if the missing terms are wrong.","fun_headline_variants_meta":{"raw":{"variants":["Stacking MnBr3 lifts MnBi2Te4's Curie point to 72 K","MnBi2Te4/MnBr3: fivefold Tc boost via Hubbard U","One magnet at U=3 eV, two at U=5 eV in MnBi2Te4/MnBr3","Hubbard U tunes MnBi2Te4/MnBr3 from one magnet to two","Stacking with MnBr3 raises MnBi2Te4's Tc fivefold"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001059,"raw_usage":{"total_tokens":4531,"prompt_tokens":1119,"completion_tokens":3412,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":735,"completion_tokens_details":{"reasoning_tokens":3291}},"tokens_in":735,"tokens_out":3412,"duration_ms":24782,"temperature":1.0,"reasoning_tokens":3291,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T20:01:59.234309+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the transition temperatures with the full spin Hamiltonian written out explicitly, including single-ion anisotropy and any other symmetry-breaking terms, and report the error of the machine-learning fit on the held-out 100 spin configurations; if the 72 K transition does not survive an anisotropy-corrected simulation, the enhancement claim is not settled. In parallel, a measured magnetization-versus-temperature curve of a grown stack that shows a single transition near 72 K would support the picture, while two separate transitions near 56 K and 158 K would match the decoupled $U_2 = 5.0$ eV case.","supporting_citations":[{"cited_title":"Xiao, and B","cited_arxiv_id":null,"evidence_quote":"Electron-counting rule used to predict that the type-II/type-I interface between ML-MBT and MnBr3 favors interlayer ferromagnetic coupling."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the machine-learning fitting and Monte Carlo simulation procedure that converts DFT energies into exchange constants and Curie temperatures."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Reported MnBr3 monolayer as a high-Curie-temperature (~200 K) Dirac half-metal, the component whose correlations the paper tunes."},{"cited_title":"Zhao , L","cited_arxiv_id":null,"evidence_quote":"Earlier ML-MBT exchange constants and ~13 K Curie temperature used to validate the fitting method before adding the MnBr3 layer."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Prediction of manganese trihalides as two-dimensional Dirac half-metals, supporting the choice of MnBr3 as a topological magnetic partner."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the GGA+U scheme that defines the Hubbard $U_2$ parameter whose variation drives the reported transition changes."}],"review_version":2}