{"id":"7b4203a2-38c9-473b-847d-ceeba0777a98","arxiv_id":"1908.06740","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":6,"one_line_summary":"In a BAs/CrI3 moiré bilayer, the magnetic proximity effect varies with interlayer atomic registry and produces spin-split minibands whose energy splitting is tunable by moiré periodicity and electric field.","lead":"A semiconductor sheet placed on a magnetic sheet experiences a magnetic effect that changes from spot to spot, creating a repeating pattern of electron states with separated spins. The pattern can be adjusted by twisting the sheets, stretching them, or applying an electric field, which may lead to new spin-based nanodevices.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Moiré miniband predictions rest on a rigid-lattice linear registry mapping; if in-plane reconstruction is significant, the central spin-splitting vs periodicity claim is quantitatively untested.","rationale":"The paper's central claim has two layers: (i) the magnetic proximity effect depends on interlayer atomic registry, which is supported by symmetry analysis and by DFT at three high-symmetry stackings with spin splittings 0, 2.6, and 12.7 meV, plus an interpolation check along one line; and (ii) in a moiré this registry dependence produces laterally modulated proximity fields and tunable miniband spin splittings, which relies entirely on the local approximation with a linear rigid mapping r(R) and no reconstruction. The second layer is the load-bearing step because all quantitative moiré predictions in Figs. 5 and 6 follow from mapping commensurate results onto the moiré via that assumption. The paper itself flags the reconstruction issue only qualitatively, and no code or data are provided to test it. This is not an internal inconsistency, but it is an untested extrapolation. The reader's weakest assumption identifies the same point, so I agree. Because the core registry-dependent physics is credible and has independent DFT support, the verdict should remain CONDITIONAL rather than ACCEPT or REJECT; the concrete relaxation test described above would settle whether the quantitative miniband predictions survive.","tokens_in":14314,"tokens_out":5272,"duration_ms":56845,"concrete_test":"Fit a registry-dependent interlayer potential (or a machine-learned potential) to the DFT stacking-energy surface of BAs/CrI3, then relax the full moiré at b=28a and b=58a including intralayer elasticity. Compute the displacement field u(R); if |u| or the soliton width is not negligible compared with the moiré period, rebuild Vτ(R) from the reconstructed local registries and recompute the miniband spin splitting as a function of b and of the Stark amplitude v0. The rigid-lattice prediction stands only if the resulting spin-splitting curve in Fig. 5(c) changes by less than about 20% and the electric-field threshold for spin separation in Fig. 6(b) is not shifted by a factor of two.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The decisive step is the local approximation in Section IV.A: the actual moiré is replaced by a rigid lattice with a linear mapping r(R), so the commensurate-registry functions Es(r) and Vτ(r) from Sections III.B-C become moiré potentials Vτ(R). All miniband results in Figs. 5-6 are consequences of this mapping. The paper explicitly concedes that in-plane relaxation 'would change quantitatively the profiles of spin splitting and band edge energy' and then asserts, without evidence, that 'the general picture ... is not changed.' That assertion is untested. If the layers reconstruct, r(R) is not the rigid value: the confinement centers, miniband widths, spin splitting versus moiré periodicity (Fig. 5c), and the electric-field competition in Fig. 6 are all renormalized. Since the central claim is a quantitative prediction of tunable miniband spin splitting, this untested mapping is load-bearing. The DFT evidence for registry dependence itself (A/B/H splittings and the single-line interpolation in Fig. 3b) is credible; the weak point is the extrapolation from commensurate bilayers to the full moiré.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript studies the magnetic proximity effect (MPE) in a van der Waals heterobilayer formed by a monolayer of the semiconductor BAs and a monolayer of the ferromagnet CrI3, with a focus on moiré superlattices. From DFT calculations of commensurate BAs/CrI3 bilayers and symmetry analysis, the authors show that the spin splitting induced in the BAs valence band depends strongly on the interlayer atomic registry: about 12.7 meV for H stacking, 2.6 meV for B stacking, and negligible for A stacking. They parameterize this registry dependence with the interpolation formula Eq. (1), validate it along one line in registry space, and similarly parameterize the registry-dependent band-edge energy as a moiré potential in Eq. (2). Under a rigid-lattice local approximation, these potentials are mapped to a moiré superlattice, giving spin-polarized minibands whose width and spin splitting depend on the moiré period, and an electric-field-tunable spin splitting that can lead to spatial separation of spin-up and spin-down states. The paper closes with proposed device concepts such as spin-polarized quantum dot arrays and one-dimensional spin waveguides.","tokens_in":14519,"tokens_out":5779,"duration_ms":66414,"significance":"If the results hold, the paper identifies a general and physically transparent mechanism: the magnetic proximity field in a vdW heterostructure is registry dependent, so a moiré pattern automatically produces a lateral modulation of that field, with observable consequences in miniband spin splitting. The central registry dependence is supported by independent evidence: C3 symmetry selection rules for interlayer hopping and direct DFT at the A, B, and H stackings. The interpolation Eq. (1) is checked against first-principles data along a line, and the moiré model is clearly stated. The electric-field-induced spin separation prediction is falsifiable and would be a distinctive experimental signature. The main uncertainties are the reliance on PBE-level DFT without error bars, the use of potentials fitted to the same high-symmetry configurations used to establish the effect, and the untested robustness of the miniband predictions to in-plane reconstruction.","major_comments":[{"comment":"The entire miniband calculation rests on the rigid-lattice local approximation, where the mapping between local registry r and position R is assumed linear. The text explicitly concedes that in-plane relaxation \"would change quantatively the profiles of spin splitting and band edge energy\" but then asserts without calculation that \"the general picture ... is not changed.\" This is load-bearing for the central quantitative claims: Fig. 5(c) presents the miniband spin splitting as a function of moiré periodicity, and Fig. 6(b) presents the electric-field competition, both of which would be renormalized if the registry mapping were not rigid. I request a quantitative assessment of the validity of the no-reconstruction assumption, for example an estimate of the adhesion energy versus the elastic energy cost for the relevant moiré periods, or a DFT relaxation of at least one large-period moiré supercell. In the absence of such evidence, the claims should be explicitly restricted to the rigid-lattice limit.","section":"IV.A"},{"comment":"The interpolation formula Eq. (1) is the first-principles-anchored expression for the registry dependence of the spin splitting, but it is validated only along a single line in the two-dimensional registry unit cell. Because the later moiré model uses the full two-dimensional behavior of the registry-dependent quantities, the authors should provide a more complete quantitative comparison, such as a color-map of the difference between Eq. (1) and DFT over the whole unit cell, or at least report a root-mean-square deviation along several high-symmetry paths. This would also address the concern that the two fitted amplitudes Es_H and Es_B already guarantee agreement at the three special points.","section":"Eq. (1) and Fig. 3(b)"},{"comment":"The quantitative values Es_H = 12.7 meV and Es_B = 2.6 meV are obtained from GGA-PBE DFT without a Hubbard U correction, even though monolayer CrI3 is a correlated magnetic insulator. The vdW-functional comparison in Appendix C checks the interlayer functional but does not test the dependence of the spin splitting on the exchange-correlation treatment of the Cr 3d states. A DFT+U or hybrid-functional calculation for at least the H and B stackings would give an indication of the error bar on the miniband spin splittings, which currently inherit the PBE values without any stated uncertainty.","section":"Appendix A and Appendix C"}],"minor_comments":[{"comment":"The heading \"Discussion and summery\" contains a typo; it should read \"Summary.\"","section":"Section V"},{"comment":"The paragraph describing the local approximation contains the typo \"quantatively\" instead of \"quantitatively.\"","section":"Section IV.A"},{"comment":"The statement that the translation vector r is well defined only in the 1x1 unit cell of BAs is somewhat confusing, since the commensurate supercell is a 2x2 supercell of BAs on a 1x1 cell of CrI3; please clarify the definition of the reference unit cell used for r.","section":"Section II"},{"comment":"The parameters {vτ, φτ} are given as approximately {83 meV, 152.3°} and {70 meV, 156°}, but no explicit uncertainty or fitting error is reported; please add the fitting residuals or a statement of the expected accuracy.","section":"Eq. (2)"},{"comment":"In the band-structure panels, the reciprocal-space labels K̃, Γ̃, and M̃ are used but not explicitly defined in the text; a brief definition of the mini-Brillouin-zone notation would improve readability.","section":"Fig. 5"}],"recommendation":"major_revision","confidential_remarks":"The paper is within scope for a condensed-matter physics journal. I do not see a damaging circularity in the main registry-dependence claim, which is independently supported by symmetry and direct DFT at the special stackings; the moiré miniband calculation is a model extrapolation built on that evidence. The main risk is overclaiming robustness to in-plane reconstruction. The requested quantitative check of the rigid-lattice assumption, or a clear caveat in the abstract, should determine whether the paper is publishable in its current form."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a good applied theory paper, not a revolution. The genuinely new piece is showing that the magnetic proximity effect in BAs/CrI3 depends strongly on stacking registry (0 to ~13 meV across A/B/H), and that a moiré pattern therefore produces a spatially modulated proximity field that shows up as spin-split minibands. That combination is new and useful.\n\nWhat it does well: the symmetry analysis for C3-symmetric stackings is clean and matches the DFT. The interpolation formula for Es(r) is checked against first-principles data along one line and works, which is real evidence. The paper is also unusually honest: it explicitly flags the rigid-lattice local approximation, the neglect of reconstruction, the role of SOC, and the vdW functional sensitivity. That honesty earns credit.\n\nWhere the soft spots are: the main one is exactly what the stress test says. The miniband spin splitting versus moiré periodicity (Fig. 5c) and the electric-field competition (Fig. 6) all go through the mapping r(R) that assumes no in-plane reconstruction. The paper concedes relaxation would change the quantitative profiles, then asserts the general picture survives without testing it. For a prediction with this much device flavor, that assertion is a gap. The interpolation also inherits some circularity: the moiré potential parameters are fitted at the same high-symmetry configurations whose spin splittings the long-period limit recovers, so Fig. 5c asymptotics are partly baked in. The electric-field part relies on a fitted interlayer-distance modulation and an assumed form; plausible, but softer. No data/code is provided for reproduction, though the model is simple enough that it's likely reproducible.\n\nBottom line: the central registry-dependence claim is well supported, and the moiré extension is a reasonable and clearly stated model, not a falsifiable first-principles supercell calculation. The paper deserves serious peer review. I'd send it out. The referee should ask for a sensitivity test of the rigid-lattice assumption (e.g., reconstructed r(R) from a continuum model) before the quantitative claims are taken at face value.","headline":"Solid registry-dependent magnetic proximity effect with a credible DFT anchor, but the moiré miniband predictions rest on a rigid-lattice approximation the paper doesn't quantitatively test.","tokens_in":15119,"tokens_out":1556,"would_cite":true,"duration_ms":14733,"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":"Atomic registry sets the magnetic proximity effect in moiré bilayers.","keywords":["magnetic proximity effect","moiré superlattice","van der Waals heterostructure","spin splitting","interlayer atomic registry","miniband","CrI3","BAs"],"falsifier":"Resolve the valence-band spin splitting of BAs in a BAs/CrI3 moiré as a function of position or moiré periodicity, using spin-resolved scanning tunneling microscopy or angle-resolved photoemission on samples with controlled twist and strain, and compare with the paper's prediction that the splitting goes from near zero in A-type regions to about 12.7 meV in H-type regions and grows with moiré periodicity. Observing sizeable splitting in A-type regions, or no growth of the miniband splitting with periodicity, would falsify the registry-modulation picture.","tokens_in":1832,"feed_emoji":"🧲","tokens_out":2609,"duration_ms":68685,"temperature":0.7,"pith_summary":"This paper tries to establish that the magnetic proximity effect at a van der Waals interface is not uniform: it depends sensitively on how the two atomic lattices are aligned, the interlayer atomic registry. In a moiré pattern the registry changes continuously from place to place, so the paper claims the magnetic proximity field is laterally modulated and appears as a spin splitting of moiré minibands. Because the moiré periodicity can be changed by relative twist or strain, and because the interlayer distance also varies across the moiré, the miniband spin splitting becomes mechanically and electrically tunable. A sympathetic reader would care because this turns moiré heterostructures into a platform for spin-polarized quantum dot arrays, spin filters, and spin memories with gate-level control.","feed_headline":"Atomic registry sets the magnetic proximity effect in moiré bilayers","feed_subtitle":"In BAs on CrI3, twist, strain, or a gate voltage tunes the spin splitting of moiré minibands.","key_machinery":"The machinery is the local-approximation map from a moiré location $\\mathbf{R}$ to a local interlayer translation vector $\\mathbf{r}(\\mathbf{R})$, treating each small region as a rigid, commensurate bilayer. With this mapping, the registry-dependent spin splitting $E_s(\\mathbf{r})$ and the spin-dependent band-edge energy $V_\\tau(\\mathbf{r})$, both expressed as sums over the first-shell reciprocal lattice vectors, become moiré potentials in the tight-binding Hamiltonian $H = H_0 + V_\\tau(\\mathbf{R})$. These two registry-dependent functions carry the entire argument: the miniband spin splitting, the exponential flattening of the miniband with moiré periodicity, and the electric-field response all follow from them.","core_discovery":"The central claim is that in a van der Waals bilayer made of a monolayer semiconductor (BAs) on a two-dimensional ferromagnet (CrI3), the magnetic proximity effect arises from spin-conserved interlayer hopping and depends strongly on the interlayer atomic registry. First-principles calculations show that the registry-dependent spin splitting at the BAs valence band edge ranges from essentially zero to about 12.7 meV between high-symmetry stackings, and the paper captures this with a compact registry-dependent formula $E_s(\\mathbf{r})$. In a long-period moiré, where the registry varies smoothly from place to place, the proximity field is therefore spatially modulated, producing a spin-dependent moiré potential and minibands whose spin splitting grows as the moiré periodicity increases and approaches the value of the local high-symmetry configuration. A perpendicular electric field, acting through the moiré modulation of interlayer distance, shifts the confinement centers and can even spatially separate the spin-up and spin-down carriers in the moiré.","pith_inferences":["Editorial inference: the same registry-dependent proximity mechanism should operate in other semiconductor/ferromagnet van der Waals pairs, though the quantitative splittings will depend on band alignment and interlayer hopping amplitudes, so the BAs/CrI3 numbers are a case study rather than a universal scale.","Editorial inference: if atomic reconstruction occurs in the moiré, the linear registry map used here would break down and the confinement centers and spin-splitting values would shift, so full-scale atomistic calculations including relaxation would test how robust the predicted miniband spin splitting is.","Editorial inference: the predicted electric-field-induced spin separation suggests a concrete experimental search using spin-resolved scanning tunneling microscopy or magneto-optical Kerr measurements, which should show spin-up and spin-down densities moving to opposite moiré regions as the gate voltage is swept."],"forward_implications":["In a long-period BAs/CrI3 moiré, the topmost valence miniband becomes flat and spin-split, with the splitting approaching about 12.7 meV as the band-edge state localizes around the H stacking region.","The miniband spin splitting can be tuned mechanically by relative twist and/or strain, because those control the moiré periodicity.","A perpendicular electric field monotonically reduces the miniband spin splitting by moving the confinement center from H to A registries, and at a specific field strength it spatially separates spin-up and spin-down carriers at different moiré locations.","The spin-polarized and well-localized miniband states form a hexagonal array of quantum dots that a uniaxial strain can turn into a one-dimensional array of spin waveguides.","Because the proximity-induced spin polarization follows the CrI3 magnetization, switching that magnetization with a small external field reverses the spin polarization of the miniband states."],"supporting_citations":[{"why":"Supplies the form of the registry-dependent spin-splitting formula $E_s(\\mathbf{r})$ that the paper adapts to model the moiré proximity effect.","marker":"[27]"},{"why":"Supplies the Fourier expansion of the band-edge energy over the first-shell reciprocal lattice vectors that becomes the moiré potential.","marker":"[28]"},{"why":"Establishes the ab initio local-approximation framework for moiré superlattice bands that the paper relies on for its tight-binding model.","marker":"[45]"},{"why":"Provides the measured interlayer-distance variation in a MoS2/WSe2 moiré that motivates the electric-field modulation of the proximity effect.","marker":"[20]"},{"why":"Applied the linear mapping from local registry to moiré position in van der Waals heterobilayers, the same mapping used here.","marker":"[26]"},{"why":"Earlier use of the local-approximation picture in moiré patterns of two-dimensional magnets, supporting the modeling strategy for BAs/CrI3.","marker":"[32]"},{"why":"Experimental demonstration of spin and valley control in monolayer TMDs stacked on ferromagnets, the empirical anchor for the magnetic proximity effect studied here.","marker":"[37]"},{"why":"Experimental observation of valley manipulation by optically tuning the magnetic proximity effect in WSe2/CrI3 heterostructures, situating the present work among proximity-effect measurements.","marker":"[38]"}],"fun_headline_variants":["Twist and strain tune spin splitting in magnetic moiré bilayers","Gate voltage separates spin-up and spin-down in moiré superlattices","Moiré modulation of proximity field controls spin-split minibands","Registry-dependent magnetic proximity yields tunable spin splitting","Electric field tunes spin splitting in moiré magnets"],"cache_read_input_tokens":17152,"weakest_assumption_plain":"The results stand or fall on the local approximation: the moiré is treated as two rigid, unreconstructed lattices where each location maps linearly to a local interlayer translation, so the registry-dependent physics of commensurate bilayers transfers directly to the moiré.","fun_headline_variants_meta":{"raw":{"variants":["Twist and strain tune spin splitting in magnetic moiré bilayers","Gate voltage separates spin-up and spin-down in moiré superlattices","Moiré modulation of proximity field controls spin-split minibands","Registry-dependent magnetic proximity yields tunable spin splitting","Electric field tunes spin splitting in moiré magnets"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001206,"raw_usage":{"total_tokens":4968,"prompt_tokens":945,"completion_tokens":4023,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":561,"completion_tokens_details":{"reasoning_tokens":3935}},"tokens_in":561,"tokens_out":4023,"duration_ms":26456,"temperature":1.0,"reasoning_tokens":3935,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:35:45.913119+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Resolve the valence-band spin splitting of BAs in a BAs/CrI3 moiré as a function of position or moiré periodicity, using spin-resolved scanning tunneling microscopy or angle-resolved photoemission on samples with controlled twist and strain, and compare with the paper's prediction that the splitting goes from near zero in A-type regions to about 12.7 meV in H-type regions and grows with moiré periodicity. Observing sizeable splitting in A-type regions, or no growth of the miniband splitting with periodicity, would falsify the registry-modulation picture.","supporting_citations":[{"cited_title":"Yu, G.-B","cited_arxiv_id":null,"evidence_quote":"Supplies the form of the registry-dependent spin-splitting formula $E_s(\\mathbf{r})$ that the paper adapts to model the moiré proximity effect."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the Fourier expansion of the band-edge energy over the first-shell reciprocal lattice vectors that becomes the moiré potential."},{"cited_title":"Zhang, C","cited_arxiv_id":null,"evidence_quote":"Provides the measured interlayer-distance variation in a MoS2/WSe2 moiré that motivates the electric-field modulation of the proximity effect."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Applied the linear mapping from local registry to moiré position in van der Waals heterobilayers, the same mapping used here."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Earlier use of the local-approximation picture in moiré patterns of two-dimensional magnets, supporting the modeling strategy for BAs/CrI3."},{"cited_title":"Zhong, K","cited_arxiv_id":null,"evidence_quote":"Experimental demonstration of spin and valley control in monolayer TMDs stacked on ferromagnets, the empirical anchor for the magnetic proximity effect studied here."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Experimental observation of valley manipulation by optically tuning the magnetic proximity effect in WSe2/CrI3 heterostructures, situating the present work among proximity-effect measurements."}],"review_version":1}