{"id":"4119c6d2-b4b5-40c1-9ac3-840da0c0ee63","arxiv_id":"2507.04844","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"An orbital magnetic field induces an effective Zeeman term in moiré surface states, tilting the zeroth Landau level linearly with field strength.","lead":"This paper derives a new type of Zeeman splitting in the surface states of topological insulators when a moiré pattern is present: the orbital motion of electrons in a magnetic field couples different Dirac points and acts like an extra spin-dependent energy term. This makes the normally flat zeroth Landau level bend with magnetic field, which could be measured in tunneling experiments as a fingerprint of the moiré potential.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central tilt-of-zeroth-LL prediction is never numerically tested in the weak-field regime: the only simulation uses flux Φ0/3 per moiré cell, where l_B ≈ 0.69 L_m and the perturbative expansion behind Eq. (14) is uncontrolled.","rationale":"The reader's weakest assumption is that the magnetic field is treated as a weak perturbation but the only numerics use Φ0/3 per moiré cell; I agree that this is the load-bearing gap. The analytic downfolding in Eqs. (3)–(11) is internally coherent, and the Landau-level result (Eqs. 19–20) follows from the effective Dirac Hamiltonian once gz is known. No internal inconsistency or circularity is apparent, and no parameters are fitted to the claimed prediction, so the issue is not correctness of algebra but validation of the prediction in its stated regime. The manuscript itself flags the strong-field limitation in the Conclusion, which strengthens the concern rather than resolving it. Because the missing test is addressable—extract gz at small flux and compare to Eq. (14)—the appropriate verdict remains CONDITIONAL, not REJECT or ACCEPT. I therefore keep the reader's verdict unchanged. I do not raise concerns about the g∥ formula (Eq. 13) because the central claim concerns only gz; if the numerical test is supplied, it should also expose any such auxiliary issue.","tokens_in":8147,"tokens_out":11114,"duration_ms":138120,"concrete_test":"Run exact-diagonalization tight-binding simulations of the same BHZ + Dirac-comb model at small flux per moiré cell, e.g., Φ/Φ0 = 1/q with q = 5, 7, 9, using a magnetic supercell, and compute the lowest Landau band energy at Γ as a function of B. Fit E0 − gz^num B + cB^2 and check that the linear term dominates. Independently construct the zero-field Γ-point eigenstates and coupling matrices V_l^{αβ} from the tight-binding Hamiltonian, evaluate Eq. (14) to get gz^an, and compare with gz^num. If they agree within numerical accuracy and the B² correction is subdominant, the central claim is confirmed; if not, the perturbative derivation is not validated.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The claim that the zeroth Landau level tilts as E = E0 − gzBz (Eqs. 19–20) rests on the downfolded Zeeman coefficient gz in Eq. (14), obtained from V(E0−H′)^−1V† with minimal coupling and the commutator [πμ,πν] = i(e/ℏ)εμνλBλ. This is a small-field, small-coupling expansion: the magnetic length must be much larger than the moiré period and the inter-Dirac-point coupling treated as a perturbation. The numerical section instead uses a Dirac-comb moiré potential with flux Φ0/3 per moiré cell. Then B = h/(3eL_m^2), so l_B ≈ 0.69L_m, i.e., Landau and moiré length scales are comparable, and Hofstadter-type band mixing is expected. The authors' Conclusion admits this: 'Our theory treats the magnetic field as a perturbation to the moiré potential. When the magnetic field is large, the superlattice potential is instead a perturbation...' The tight-binding part never evaluates Eq. (14) from zero-field eigenstates, never extracts gz from a fit of E_LL0(B), and never compares a computed slope to the analytic formula. Therefore the paper's central quantitative prediction is unsupported in its regime of validity; the only numerical spectra shown are in the opposite regime.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript studies moiré surface states of topological insulators in a magnetic field, modeled as a two-dimensional Dirac fermion in a periodic scalar potential with several Dirac points at the Γ point. Its central claim is that minimal coupling, combined with the noncommutativity of the canonical momentum [πμ,πν]=i(e/ℏ)εμνλBλ, converts the second-order downfolding correction V(E0−H′)−1V† into an effective spin-dependent term gμνσμBν in the projected Hamiltonian (Eqs. (7)–(11)). Symmetry reduction to the point group Cnv gives g∥(σxBx+σyBy)+gzσzBz (Eq. (12)), with gz expressed in terms of the inter-Dirac-point coupling matrices V_l^{αβ} and energies E_l (Eq. (14)). The physical consequence is that the zeroth Landau level of each moiré Dirac point, otherwise field-independent, disperses linearly as E = E0 − gzBz (Eq. (20)). The analytic part is followed by a tight-binding simulation of a three-dimensional BHZ slab with a Dirac-comb moiré potential on its top surface (Eqs. (22)–(25)), and by a brief symmetry classification of the Dirac doublets.","tokens_in":8396,"tokens_out":62641,"duration_ms":580947,"significance":"Assessed at face value, the mechanism is an elegant and physically interesting route to an orbital-field-induced spin splitting: the field enters through the commutator of the gauge-invariant momentum, and virtual inter-Dirac-point processes promote this orbital coupling to a spin-Zeeman term. The prediction is sharp and, in principle, falsifiable: the slope of the zeroth Landau level versus B is fixed by microscopic band-structure quantities (the V_l and E_l), and the tilt should be absent in particle-hole-symmetric moiré potentials. The Landau-level solution (Eqs. (15)–(20)) is transparent and internally consistent, and the choice of a three-dimensional lattice simulation rather than a two-dimensional tight-binding model is well motivated by the topological obstruction to purely two-dimensional surface models. The paper's strength lies in this clean, generic formulation. Its present weaknesses are that the central g-factor formula appears, on direct re-derivation, to carry an overall factor-of-two error, and that the numerical simulation is executed in a field regime where the theory's perturbative premise fails, so the headline prediction is not actually tested.","major_comments":[{"comment":"The numerical simulation is performed at a magnetic flux of Φ0/3 per moiré unit cell (Fig. 1 caption), which gives a magnetic length l_B ≈ 0.69 L_m, comparable to the moiré period. In this regime the magnetic field is not a weak perturbation to the moiré potential, and the expansion behind Eq. (14), in which the field enters only through the commutator [πμ,πν] and the inter-Dirac-point coupling V is small relative to the energy separations E_l − E0, is uncontrolled; the paper itself identifies this as the opposite limit in the Conclusion ('When the magnetic field is large, the superlattice potential is instead a perturbation'). Because the simulation never evaluates Eq. (14) from zero-field eigenstates, never extracts gz from the slope of the zeroth Landau level as a function of B, and never compares a computed slope with the analytic formula, the central prediction (Eqs. (19)–(20)) is not validated in the regime where the derivation applies. I recommend adding a weak-field simulation (flux well below Φ0 per moiré cell) with an explicit linear fit of the zeroth-Landau-level energy and a comparison against Eq. (14); such a comparison must also separate the spin-independent O(π²) corrections generated by the same downfolding, whose expectation values are linear in B in the n = 0 Landau level.","section":"Tight-binding simulation (Eqs. (22)–(25); Fig. 1 caption)"},{"comment":"On direct evaluation of the downfolding in Eq. (7) with minimal coupling, the σzBz coefficient is (e/ℏ) Σ_l (E_l − E0)^(−1) Re[V_l^{xx}(V_l^{yy})* − V_l^{xy}(V_l^{yx})*], a factor of two smaller than Eq. (14). The check is simplest for a single high-energy level with V = v(σxπx + σyπy): the σz⊗Bz part of V V† is −(e/ℏ)v²σzBz (since σxσyπxπy + σyσxπyπx = iσz[πx,πy]), giving an induced coefficient v²(e/ℏ)/(E_l − E0), not 2v²(e/ℏ)/(E_l − E0). Equivalently, the contraction εαα′z εββ′z summed over the two orderings gives 2Re[V^{xx}V^{yy*} − V^{xy}V^{yx*}] but is multiplied by the factor 1/2 from the decomposition of πβπβ′ into symmetric and antisymmetric parts. Since Eq. (20) uses gz directly as the slope of the zeroth Landau level, this doubles the predicted tilt. I would ask the authors to re-derive Eq. (11) carefully and correct Eqs. (13) and (14) accordingly.","section":"Eqs. (11) and (14)"},{"comment":"The in-plane coefficient g∥ in Eq. (13) does not follow from the derivation of Eq. (11): in a strictly two-dimensional surface model the indices β and β′ in εββ′ν are in-plane only, so εββ′x = εββ′y = 0, and the mechanism of Eqs. (7)–(11) generates only the gzσzBz term, not g∥(σxBx + σyBy). Furthermore, the index structure of the displayed expression for g∥ (products such as V^{zx}V^{xy*}) is not compatible with the εαα′μ contraction in Eq. (11). Because the off-diagonal terms are later absorbed into a redefinition of π, this does not affect the final Landau levels (Eqs. (19)–(20)), but the statement that both g-factors arise from Dirac point coupling should either be corrected or be substantiated with an explicit three-dimensional mechanism.","section":"Eqs. (12)–(13)"}],"minor_comments":[{"comment":"Replace 'session' with 'section' in 'To close this session', 'in previous sessions', and 'as elaborated in the session of tight-binding simulation'.","section":"Throughout"},{"comment":"In the sentence 'Here z denotes the complex conjugate of complex number z', the symbol should be written as z̄ (complex conjugate); as printed it is indistinguishable from the variable z.","section":"After Eq. (11)"},{"comment":"Panels (d) and (e) are described in the text as plotting the Landau levels, but they appear to be schematic illustrations of E versus Bz; please state explicitly which panels of Fig. 1 are simulation results and which are drawings.","section":"Fig. 1"},{"comment":"The stated motivation for abandoning the k·p model, that the vector potential 'could become divergent at long distances', is not the real reason; minimal coupling in a fixed gauge is standard and well behaved. The actual motivation (a lattice regularization that permits Peierls phases and handles strong fields) should be stated.","section":"Tight-binding model simulation"},{"comment":"The units of U0 in the Dirac-comb potential are not specified; please give the strength in units of the hopping integral and state the effective on-site potential for the 2×2 moiré cell so that the simulation is reproducible.","section":"Eq. (23)"},{"comment":"The Introduction promises a discussion of experimental realizations and material candidates, but the Conclusion contains none; the only list of candidate materials appears in the tight-binding section. An order-of-magnitude estimate of gz from Eq. (14) with realistic parameters, and the corresponding field scale needed to observe the zeroth-Landau-level tilt, would strengthen the paper.","section":"Conclusion"},{"comment":"The double-group classification of Dirac doublets in the paragraph around Eq. (21) is not used elsewhere in the paper; using the angular-momentum labels to constrain the form of V_l^{αβ} would sharpen Eq. (14), or this paragraph could be shortened.","section":"Eq. (21) and following"}],"recommendation":"major_revision","confidential_remarks":"The factor-of-two discrepancy in Eqs. (11) and (14) is, in my reading, robust: I checked it both by operator identities in V(E0−H′)−1V† and by explicit evaluation in the n = 0 Landau level of a minimal two-level model. I would ask the editor to have the authors supply a corrected derivation and a weak-field numerical slope comparison, which would also settle the sign of gz. The manuscript has two independent load-bearing problems (the factor error and the absence of a validation in the theory's regime of validity), so major revision is warranted rather than minor revision; the mechanism itself is promising and within the journal's scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The useful core is the downfolding calculation in Eqs. (3)–(11): minimal coupling plus projection turns the noncommutativity of π into an effective Zeeman term, and the resulting g-factor formulas (Eqs. 11 and 14) are explicit in terms of inter-Dirac-point couplings. That is a standard technique, but the application to moiré surface states is new, and the prediction that the zeroth Landau level tilts as E0 − gzBz (Eq. 20) is concrete and testable. The symmetry analysis showing that particle-hole symmetry suppresses gz is also clean and useful.\n\nThe soft spot is the numerical section. The simulation uses flux Φ0/3 per moiré cell, so l_B ≈ 0.69L_m. That is the opposite regime from the weak-field expansion behind Eq. (14); the authors concede in the Conclusion that their theory treats B as a perturbation to the moiré potential and that for large B the superlattice potential is the perturbation. The spectra in Fig. 1 therefore cannot validate the predicted slope, and the paper never extracts gz from small-flux data or compares a fitted slope to Eq. (14). This is a genuine gap: the central quantitative prediction is unsupported in its regime of validity.\n\nThat said, the analytic part stands on its own. The g-factor is expressed directly in terms of band-structure elements, not fitted to the Landau level shift, so circularity is low. The reliance on Refs. [17,18] for the existence and structure of moiré surface states is shared prior work, not a problem. A small-flux tight-binding calculation with the same model would settle it, and I hope the authors do that and release the code.\n\nFor a referee: send it out. The mechanism and formulas are worth publishing, but the numerical verification needs to happen in the weak-field regime before the paper is complete. It is a moderate-impact theory paper, not a paradigm shift, and a serious referee should ask for the missing small-flux comparison.","headline":"Clean analytic derivation of an orbital-induced Zeeman effect for moiré surface states, but the numerical support is in the wrong regime and leaves the central slope prediction untested.","tokens_in":9023,"tokens_out":2785,"would_cite":true,"duration_ms":28596,"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":"The orbital effect of a magnetic field induces an effective Zeeman term in moiré surface states, tilting the zeroth Landau level linearly with Bz.","keywords":["induced Zeeman effect","moiré surface states","topological insulator","Landau levels","orbital effect","Dirac point coupling","minimal coupling","zeroth Landau level"],"falsifier":"Measure the zeroth Landau level energy of a particle-hole-asymmetric moiré surface state versus out-of-plane field at weak field: the paper predicts $E=E_0-g_zB_z$; observing a flat, field-independent $n=0$ level, or a slope incompatible with Eq. (14), would rule out the induced-Zeeman mechanism. Alternatively, a tight-binding simulation at flux much smaller than one flux quantum per moiré cell should reproduce the analytic $g_z$; failure there would also falsify the claim.","tokens_in":7864,"feed_emoji":"🧲","tokens_out":7299,"duration_ms":73406,"temperature":0.7,"pith_summary":"Surface electrons of a topological insulator can form moiré states when the top layer is rotationally misaligned, and these states host several Dirac points at the Brillouin-zone center. This paper shows that applying an out-of-plane magnetic field then produces more than the familiar orbital Landau quantization: through the non-commutativity of the minimal-coupling momenta, the coupling between coexisting Dirac points generates an effective Zeeman term in the low-energy Hamiltonian. The result is that the zeroth Landau level, which would otherwise stay pinned at the Dirac energy, tilts linearly with magnetic field, $E = E_0 - g_z B_z$. This matters because it makes the inter-Dirac-point coupling coefficients experimentally visible as a field-dependent shift of the $n=0$ level, with the slope $g_z$ computable from the band structure.","feed_headline":"Moiré surface Dirac points tilt the zeroth Landau level","feed_subtitle":"Orbital motion plus inter-Dirac-point coupling makes the n=0 Landau level shift linearly with field.","key_machinery":"The load-bearing object is the projection (downfolding) of the full moiré Hamiltonian onto a single Dirac doublet, $H_{\\rm eff}=H+V(E_0-H')^{-1}V^\\dagger$, together with the minimal-coupling replacement $k\\to\\pi=k-eA/\\hbar$ and the commutator $[\\pi_\\mu,\\pi_\\nu]=ie\\hbar^{-1}\\epsilon_{\\mu\\nu\\lambda}B_\\lambda$. The commutator is what converts the second-order-in-$\\pi$ coupling correction into a zeroth-order Zeeman term; the g-factor formula (Eqs. 11 and 14) is the quantitative content, expressing $g_z$ as a sum over inter-Dirac-point coupling coefficients $V_l^{\\alpha\\beta}$ divided by energy separations $E_l-E_0$.","core_discovery":"Starting from a massless Dirac fermion in a periodic moiré potential, the paper writes the $\\Gamma$-point Hamiltonian in the basis of Dirac doublets, $H_\\Gamma = \\begin{pmatrix} H & V \\\\ V^\\dagger & H' \\end{pmatrix}$, and projects onto the Dirac point at energy $E_0$. After minimal coupling $k\\to\\pi=k-eA/\\hbar$, the correction $V(E_0-H')^{-1}V^\\dagger$ is quadratic in $\\pi$; using $[\\pi_\\mu,\\pi_\\nu]=ie\\hbar^{-1}\\epsilon_{\\mu\\nu\\lambda}B_\\lambda$, this quadratic form collapses to a zeroth-order spin-dependent term. The effective Hamiltonian becomes $H_{\\rm eff}=E_0+\\hbar v_0(\\sigma\\times\\pi)\\cdot\\hat z + g_{\\mu\\nu}\\sigma_\\mu B_\\nu$, with $g_{\\mu\\nu}$ given by a sum over the coupling coefficients $V_l^{\\alpha\\beta}$ and energy denominators $E_l-E_0$ (Eqs. 11 and 14). Under point group $C_{nv}$ the surviving term is $g_z\\sigma_zB_z$, so the Landau levels are $E=E_0+\\operatorname{sgn}(n)\\sqrt{2ev_0^2|n|B_z\\hbar+g_z^2B_z^2}$ and, for the zeroth level, $E=E_0-g_zB_z$. The paper also shows that a nonzero $g_z$ requires particle-hole asymmetry of the spectrum around $E_0$, so the square-lattice potential with only first harmonics gives $g_z=0$ whereas a triangular lattice or higher harmonics gives a finite tilt.","pith_inferences":["Going beyond the paper, the same commutator-based projection should generate an effective Zeeman-like term at any time-reversal-invariant momentum where multiple Dirac points coexist, so twisted surfaces of other Dirac semimetals may show a similar $B_z$ tilt of their zeroth level.","The slope $g_z$ could be used as a spectroscopic ruler: measuring the $n=0$ level shift as a function of $B_z$ extracts the inter-Dirac-point coupling coefficients $V_l^{\\alpha\\beta}$ that are otherwise hard to isolate.","The paper's tight-binding check uses a flux of $1/3$ flux quantum per moiré cell, where the magnetic field is not a weak perturbation; a quantitative test of Eq. (14) would need smaller flux, where the analytic perturbative formula should hold."],"forward_implications":["The zeroth Landau level of moiré surface states moves with out-of-plane magnetic field as $E_0 - g_zB_z$, giving a direct experimental signature of the induced Zeeman term.","A triangular moiré lattice, which breaks particle-hole symmetry of the first-harmonic square potential, should show a nonzero $g_z$, while a symmetric square lattice should not.","Only the out-of-plane component $B_z$ changes the Landau-level energies; in-plane components can be absorbed into a redefinition of the canonical momentum.","When the magnetic field becomes large compared with the moiré potential, the Landau levels are expected to broaden into Chern bands with finite bandwidth, beyond the perturbative regime."],"supporting_citations":[{"why":"Proposes the moiré surface-state model with multiple Dirac points at $\\Gamma$ that this paper builds on.","marker":"[17]"},{"why":"Gives the Dirac doublet basis, renormalized velocity $v_0$, and Dirac point energies $E_l$ used in the projection.","marker":"[18]"},{"why":"Supplies the $C_{nv}$-invariant form of the Zeeman term used to reduce the general $g_{\\mu\\nu}$ to $g_z\\sigma_zB_z$ plus in-plane terms.","marker":"[6]"},{"why":"Introduces the Peierls phase used to couple the tight-binding model to the magnetic field.","marker":"[24]"},{"why":"Underlies the Peierls-substitution treatment of magnetic flux in the lattice simulation and the Hofstadter-band limit.","marker":"[25]"},{"why":"Justifies the use of a three-dimensional model because 2D tight-binding models cannot represent topological surface states.","marker":"[19]"}],"fun_headline_variants":["Moiré twist tilts zeroth Landau level with field","Orbital effect induces Zeeman term in moiré Dirac states","Triangular moiré lattices cause zeroth Landau tilt","Zeroth Landau level becomes field-tilted in moiré surfaces"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise, stated in the paper's conclusion, is that the magnetic field is a weak perturbation to the moiré potential; if the field is strong enough that Landau quantization dominates over the superlattice potential, the downfolding to a Dirac doublet and the formula for $g_z$ no longer apply.","fun_headline_variants_meta":{"raw":{"variants":["Moiré twist tilts zeroth Landau level with field","Orbital effect induces Zeeman term in moiré Dirac states","Triangular moiré lattices cause zeroth Landau tilt","Zeroth Landau level becomes field-tilted in moiré surfaces"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000537,"raw_usage":{"total_tokens":2603,"prompt_tokens":993,"completion_tokens":1610,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":609,"completion_tokens_details":{"reasoning_tokens":1531}},"tokens_in":609,"tokens_out":1610,"duration_ms":17162,"temperature":1.0,"reasoning_tokens":1531,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T19:39:20.185997+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the zeroth Landau level energy of a particle-hole-asymmetric moiré surface state versus out-of-plane field at weak field: the paper predicts $E=E_0-g_zB_z$; observing a flat, field-independent $n=0$ level, or a slope incompatible with Eq. (14), would rule out the induced-Zeeman mechanism. Alternatively, a tight-binding simulation at flux much smaller than one flux quantum per moiré cell should reproduce the analytic $g_z$; failure there would also falsify the claim.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the Dirac doublet basis, renormalized velocity $v_0$, and Dirac point energies $E_l$ used in the projection."},{"cited_title":"Bradlyn, L","cited_arxiv_id":null,"evidence_quote":"Introduces the Peierls phase used to couple the tight-binding model to the magnetic field."},{"cited_title":"Peierls, Z","cited_arxiv_id":null,"evidence_quote":"Underlies the Peierls-substitution treatment of magnetic flux in the lattice simulation and the Hofstadter-band limit."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Justifies the use of a three-dimensional model because 2D tight-binding models cannot represent topological surface states."}],"review_version":1}