WorksPhysics and Astronomy
Orbital Angular Momentum Currents and Textures in a Multiorbital Double Helix
Abstract
The microscopic origin of chirality induced spin selectivity (CISS) in biomolecular structures has traditionally been discussed in terms of spin orbit coupling and spin angular momentum. However, orbital angular momentum (OAM) is increasingly recognized as a more direct electronic response to molecular chirality. A helical hopping network built only from s orbitals can support an itinerant orbital moment and an orbital Edelstein response, but such a basis has no local atomic OAM and therefore omits the complementary contribution arising from coherence among active p orbitals. Here we construct a six-band multiorbital double helix model parameterized by characteristic B form DNA geometry and hopping scales. The model incorporates major minor groove asymmetry, an axial base pair offset representing stagger, and compositional inequivalence between the two strands, providing structural control parameters with which the orbital response can be tuned. Inter strand hybridization generates a full three component atomic OAM texture in the corotating local helical frame, including a radial component absent in the corresponding single helix model. We separately evaluate the itinerant OAM using the Bloch wave packet formulation of orbital magnetization and identify an additional inter-strand circulation channel unique to the double-helix connectivity. The resulting atomic and itinerant responses have distinct microscopic origins, magnitudes, and temperature dependences, establishing a multiorbital framework for geometry controlled angular momentum generation and transport in chiral molecular conductors.
Introduction
Chirality-induced spin selectivity (CISS) has revealed sizeable spin-dependent responses in DNA, helical proteins, molecular layers, and other chiral structures built largely from light, nonmagnetic elements [1, 2, 3, 4]. Microscopic descriptions commonly combine structural chirality with spin–orbit coupling (SOC) [5, 6, 7], together with contacts, dephasing, or other nonequilibrium ingredients needed to evade reciprocity constraints in ideal coherent two-terminal transport [8, 9]. The comparatively weak atomic SOC of organic molecules and the difficulty of accounting for the magnitude and diversity of CISS signals [10] motivate a complementary question: how strongly can chirality act first on the orbital degree of freedom?
Orbital angular momentum (OAM) can respond directly to structural chirality without requiring atomic SOC. Yoda, Yokoyama, and Murakami showed that chiral electronic structures can develop current-induced orbital magnetization, or an orbital Edelstein response [11, 12]. More generally, orbital textures associated with hybridization, crystal symmetry, and the geometrical structure of Bloch states have emerged as central ingredients of orbitronics, with both intra-atomic and inter-site contributions playing distinct roles [13, 14]. Such orbital responses may subsequently be converted into spin polarization through SOC or through interfacial orbital-to-spin conversion, providing a possible intermediate stage between structural chirality and the experimentally detected spin response [15, 16, 17].
Two microscopic OAM mechanisms should be distinguished in a helical conductor. The itinerant contribution is the self-rotation of a propagating Bloch wave packet through the real-space hopping network. Göbel, Schimpf, and Mertig showed that this mechanism alone produces an orbital moment and orbital Edelstein response in a helical model containing only orbitals [18]; related first-principles work finds substantial orbital textures in DNA-like chiral structures [19]. A complementary atomic contribution appears when the local basis itself carries OAM. In a -orbital helix, Slater–Koster hybridization among differently oriented local orbitals generates momentum-dependent inter-orbital coherences and hence an on-site OAM texture, as also emphasized in multiorbital helical-chain transport [20] and in recent analyses of chirality-encoded molecular wave functions [21].
In our previous single-helix model [22], a minimal local basis generated atomic OAM without atomic SOC. The azimuthal and longitudinal textures were finite, whereas the radial component vanished because of the phase structure of the single-strand Bloch Hamiltonian. Time-reversal symmetry forced the equilibrium OAM accumulation to vanish but allowed a conventional equilibrium OAM current; under a longitudinal field, the same odd-in- texture produced a linear orbital Edelstein polarization while the first-order field-induced OAM current vanished by parity.
Here we extend that framework to an asymmetric B-DNA-inspired double helix. The second strand introduces inter-strand hybridization, strand inequivalence, major–minor groove asymmetry, axial stagger, and new closed inter-site circulation paths. We construct a six-band -orbital tight-binding Hamiltonian and show that inter-strand hybridization releases the single-helix phase constraint, producing a full three-component atomic OAM texture with a finite radial component. We then determine the equilibrium conventional OAM current and the driven response through second order in a longitudinal electric field: a linear atomic orbital Edelstein polarization is accompanied by a quadratic field-induced OAM current. Separately, we evaluate the itinerant OAM with the Bloch-wave-packet formulation of orbital magnetization, calculate its longitudinal texture and linear and quadratic responses, and identify a double-helix plaquette with a purely radial vector area as an additional transverse itinerant channel. The model therefore places atomic and itinerant OAM on the same microscopic footing while keeping their distinct physical origins explicit.
Tight-Binding Hamiltonian Framework
Slater–Koster p-Orbital Framework
We represent the electronic structure of the double helix by a minimal orthogonal basis of three local -like orbitals (, , ) associated with each coarse-grained nucleotide site. The kinetic charge transfer between any two sites in the lattice is governed by the generalized Slater–Koster framework. For an arbitrary bond vector connecting two sites, possessing a unit direction , the transfer integral between a local -orbital oriented along and another along is dictated by the fundamental overlap integrals and :
This formulation yields the effective transfer matrices used below. The geometrical parameters are chosen from characteristic B-form DNA structural scales, while the on-site energies and hopping amplitudes constitute a representative tight-binding parameterization consistent with the energy scales used in model descriptions of DNA charge transport [5, 23]. Table I in the appendix separates the structural input from the model’s effective electronic parameters. We derive their geometrical dependence in the subsequent sections.
Double-Helix Geometry and Vertical Shear
The structural scaffolding of B-form DNA is parameterized as two discrete, intertwined helices identified by the strand index and the sequential nucleotide index . In a global Cartesian frame , the spatial coordinates of the lattice sites are defined as:
where defines the helix radius, denotes the axial rise per nucleotide, and represents the intra-strand azimuthal twist angle. The phase angle governs the relative azimuthal separation of the second strand, inducing the asymmetric formation of the major and minor grooves characteristic of B-DNA. The parameter strictly defines the longitudinal displacement between the two complementary bases at a single rung. While simplified models often assume , realistic DNA geometries in solution exhibit a finite vertical shear, crystallographically termed base-pair stagger. This vertical shear deviates from strict planarity, possessing an average characteristic value of Å [24].
To accurately evaluate the transfer integrals, a localized, co-rotating cylindrical coordinate frame must be established at every discrete site to orient the minimal orthogonal orbital basis :
where and . This local rotational mapping ensures that the resulting Slater–Koster hybridizations encode the molecular geometry. We denote the intra-strand bond by and the inter-strand rung by . The model is represented in Figure 1. Figure 1 summarizes the geometrical parameters and the co-rotating local -orbital basis used throughout the tight-binding construction. The intra-strand nearest-neighbor distance is . The unit bond vector connecting adjacent intra-strand sites yields specific projections onto the local basis, generating the direction cosines , , and .

Figure 1. Schematic representation of the double helix geometry and local orbital basis. The red and blue backbones represent the two helical strands wrapping around the central -axis, with radius and axial rise . The structural asymmetry is defined by the minor groove angle and the major groove angle , as well as an inter-strand vertical shear . Position vectors and locate site on each strand, with and representing the intra-strand and inter-strand bond vectors, respectively. At each lattice site, the local electronic states are modeled using , , and orbitals.
Inter-Strand Coupling and Structural Asymmetry
The two complementary strands are bridged by the inter-strand coupling matrix . The rung vector has a transverse magnitude and, due to the vertical shear, a longitudinal component . Its length is . The direction cosine characterizing the out-of-plane tilt is . We determine the projections of the normalized vector onto the local bases of strand 1 and strand 2. Substituting these geometric overlaps into the inter-strand Slater–Koster parameters yields the fully coupled coupling block:
The purely diagonal elements are analytically formulated as:
where , while and are the inter-strand - and -type Slater–Koster integrals. The in-plane structural asymmetry dictated by the major and minor groove offset defines the transverse cross-couplings:
The presence of the vertical shear activates out-of-plane cross-couplings between the orbitals and the transverse and orbitals across the gap:
Consequently, the matrix representing hopping from strand 1 to strand 2 is not symmetric in its local-orbital indices, . Hermiticity of the full Hamiltonian is nevertheless preserved because the reverse hopping block is .
Energy Spectrum and Analytical Eigenfunctions
Intra-Strand Kinetics and Compositional Asymmetry
We construct the intra-strand Hamiltonian governing electron propagation along a single helical backbone by considering a tight-binding framework based on local -orbitals . Following the Slater–Koster parameterization developed for helical architectures in [22], the intra-strand hopping parameters account for spatial chirality through a coordinate transformation that links neighboring sites across a step height and turning angle . Because B-DNA is composed of two distinct nucleotide sequences forming an asymmetric double helix, the on-site atomic energies of the two strands are inherently different. We define the strand-dependent on-site energies as for orbital on strand . Applying the Bloch ansatz , where is the number of unit cells and denotes the local orbital spinor, we obtain two distinct Bloch matrices, and , for the isolated strands:
Spatial chirality manifests through the purely imaginary, odd-parity momentum hybridization channels and , which are identical for both strands due to their matching local rotational frames, while the real coupling maintains even parity under momentum inversion. The complete analytical definitions of these matrix elements and the underlying Slater–Koster parameters are detailed in Appendix A.
Bonding–Antibonding Representation of the 6-Band Hamiltonian
The total system requires the full Bloch matrix formed by the two structurally and compositionally asymmetric strands:
where is the inter-strand hopping matrix. To expose the bonding–antibonding structure, we apply the unitary transformation , where is the identity matrix. This gives
where , , , and . (19) expresses the Hamiltonian in the bonding–antibonding basis, but it does not, in general, block-diagonalize the system. The diagonal blocks, and , would define independent bonding (+) and antibonding (−) sectors only if the two off-diagonal blocks vanished. Exact decoupling, therefore, requires simultaneously , . The first condition corresponds to two equivalent strands, , whereas the second requires a Hermitian inter-strand coupling matrix, . In the present double-helix model, strand inequivalence gives , while the asymmetric cross-orbital inter-strand hoppings generated by the double-helix geometry produce . Consequently, the bonding and antibonding sectors remain coupled, and the labels + and − are not exact quantum numbers of the full Hamiltonian. They may still characterize the dominant bonding or antibonding content of individual eigenstates when the corresponding mixing is weak.
The strength of the bonding–antibonding mixing cannot be inferred solely from the fact that and are nonzero. The magnitude of the off-diagonal matrix elements relative to the energy separation between the corresponding states controls it. The orbital and bonding character of the exact six-band eigenstates must therefore be determined directly from their eigenvector weights.
Characteristic Equation and Adjugate Spinors
Because the system cannot be decoupled, we must evaluate the full sixth-order characteristic secular equation:
Expanding this determinant yields a sixth-degree polynomial in energy . The exact analytical energy bands correspond to the six real roots of this polynomial, which are numerically plotted in Figure 2.
Figure 2. Electronic band structure of the fully coupled six-band B-DNA double-helix model. Panel (a) isolates the lowest energy states: (blue), (red), corresponding primarily to the longitudinal orbitals. Panel (b) details the remaining higher-energy manifolds: (green), (orange), (purple), (cyan), dominated by hybridized transverse modes. The structural and compositional asymmetry lifts the degeneracies of the corresponding symmetric model.
The derived energy spectrum, illustrated in Figure 2, demonstrates the impact of the coupled structural and compositional asymmetries. Because and , bonding and antibonding character is mixed rather than defining two exact, independent sectors. The resulting bands show avoided crossings and lifted orbital degeneracies. The lower-energy manifold in panel (a) is energetically separated due to the deep on-site energy
of the longitudinal mode, while the transverse modes in panel (b) hybridize heavily due to the chiral inter-strand overlap. For each exact energy eigenvalue , the corresponding six-component analytical Bloch spinor is formally defined using the algebraic adjugate matrix formulation. The unnormalized spinor components correspond to any nonzero column of the classical adjoint matrix . Denoting as the minor of the characteristic matrix, the eigenvector components are analytically expressed as:
where is a fixed row index chosen such that the resulting vector is nontrivial. The normalized eigenfunctions are subsequently obtained by dividing by the vector norm, yielding the local orbital amplitudes for the first strand and for the second strand.
Atomic Orbital Angular Momentum Textures
The first contribution considered is the local, or atomic, orbital angular momentum associated with the on-site -orbital manifold. This quantity is defined by the usual angular-momentum operators acting within the local basis on each strand and is distinct from the itinerant contribution associated with inter-site wavepacket circulation, which will be treated separately in Sec. VI.
Local Atomic OAM Operators and Textures
We compute the exact momentum-dependent local atomic OAM textures for band , defined generally as . The local OAM operators operate independently on the localized orbitals of each strand. For the full double-helix basis, the expectation values require evaluating the inter-orbital coherences across both structural backbones. The three spatial orbital textures are explicitly computed as:
The textures were obtained numerically and are depicted in Figure 3. Figure 3 shows that inter-strand coupling generates additional orbital structure beyond the isolated single helix. In particular, panel (a) displays a finite radial texture . Although smaller than the azimuthal and longitudinal components, it is a sensitive signature of the incompatible phases introduced by asymmetric inter-strand hybridization, as discussed below.
Figure 3. Momentum-dependent orbital angular momentum textures for the six-band double-helix model. The color coding follows the band indexing () established in Fig. 2. Panels (a), (b), and (c) display the radial, azimuthal, and longitudinal components, respectively. Time-reversal symmetry imposes the antisymmetric orbital texture . Crucially, panel (a) demonstrates the emergence of a finite radial orbital texture , which is absent in the corresponding isolated single-helix model.
The band-specific behavior in panels (b) and (c) reflects the hybridization strength of the specific transverse modes, showing enhanced structure near avoided crossings. The six band-resolved textures reflect the lifted degeneracies and hybridization of the asymmetric double helix.
Symmetry Origin of the Radial Atomic OAM
To clarify the origin of the nonzero radial component in the double-helix geometry, it is useful to consider first a symmetric reference configuration defined by and . In this limit, the inter-strand cross-couplings , , and vanish. For the basis ordering , the diagonal unitary transformation then transforms the full Bloch Hamiltonian into a real matrix. Its eigenvectors may consequently be chosen real in this representation. Since the radial orbital angular momentum is determined by the relative phase between the and amplitudes,
the radial orbital texture vanishes in this symmetric reference limit.
Departing from this symmetric reference configuration introduces incompatible phase constraints through the inter-strand hybridizations. The major–minor groove asymmetry generates the transverse cross-coupling , while a finite vertical shear activates the additional out-of-plane couplings and . These terms are, in general, incompatible with the orbital rephasing that renders the intra-strand chiral hybridizations real. Consequently, the full Hamiltonian cannot, in general, be made simultaneously real by a diagonal rephasing of the orbital basis. The eigenvectors may therefore acquire nontrivial relative phases between their orbital components. In particular, the coherence can acquire a finite imaginary part, producing the nonzero radial orbital texture observed in Figure 3. Both the groove asymmetry and the vertical shear can therefore contribute to the phase structure responsible for the radial atomic texture; disentangling their individual quantitative contributions is beyond the scope of the present work.
Atomic OAM Transport and Response
Throughout this section, denotes the local atomic OAM defined in Sec. IV. Equilibrium currents and field-induced responses of this contribution are treated separately from the itinerant OAM introduced later in Sec. VI.
Equilibrium Atomic Orbital Current
A conventional OAM current is not, in general, forbidden in equilibrium by time-reversal symmetry because both the band velocity and a time-reversal-odd OAM texture change sign under . To rigorously examine these properties within the asymmetric B-DNA double helix, we project the microscopic current-density operator onto the individual fully hybridized Bloch bands. The orbital angular momentum current operator associated with a spatial component is defined through the canonical anti-commutator relation [25]:
where is the one-dimensional velocity operator. Within the semiclassical adiabatic wave-packet approximation, in which a Bloch state moves with the band group velocity while carrying its local on-site orbital moment, we approximate the band-resolved conventional current as:
The reduction of the projected anticommutator current to this band-velocity–times–orbital-texture form, together with the assumptions entering the adiabatic approximation, was derived explicitly for the corresponding multiorbital single-helix problem in Ref. [22]. We therefore use the same approximation here and focus on the new consequences of inter-strand hybridization and double-helix geometry. By aggregating the individual band current densities weighted by the equilibrium Fermi-Dirac distribution function , the total macroscopic equilibrium orbital angular momentum current is expressed as:
Under strict time-reversal symmetry, the energy dispersion relation must obey , which naturally forces the band group velocity to act as a strictly odd function of momentum, . As demonstrated via the time-reversal symmetry properties of the real-space hopping parameters, the orbital angular momentum textures are similarly constrained to be perfectly odd functions under momentum inversion, . Crucially, the product of these two odd quantities is an even function of momentum throughout the Brillouin zone. Because the equilibrium distribution is also even, symmetry does not require the integral over the Brillouin zone to cancel. Consequently, time-reversal symmetry does not force the conventional equilibrium orbital-current integral to vanish: a finite persistent-like equilibrium OAM current is symmetry-allowed in the double helix. Because orbital angular momentum is not generally conserved in the presence of orbital hybridization, the conventional equilibrium current does not, by itself, imply an accumulation of OAM at the molecular boundaries; a complete continuity equation must also include the corresponding orbital-torque term.
Linear Atomic Orbital Edelstein Response
For the atomic OAM textures introduced in Sec. IV, the unbiased double-helix Hamiltonian preserves time-reversal symmetry (). This symmetry mandates that the energy dispersion is strictly even, , while the intrinsic orbital angular momentum (OAM) textures are antisymmetric, . Consequently, integrating these textures over the symmetric Brillouin zone confirms that the net OAM identically vanishes in thermal equilibrium.
To drive a nonzero macroscopic response, the field produces a nonequilibrium odd-in- occupation correction; the static Hamiltonian need not itself lose TR symmetry. We apply a uniform longitudinal electric field and evaluate the nonequilibrium electron dynamics using the semiclassical Boltzmann Transport Equation (BTE) within the relaxation time approximation. For each band, the steady-state nonequilibrium distribution obeys
where is the elementary charge, is the phenomenological momentum-relaxation time, and is the equilibrium Fermi-Dirac distribution at chemical potential and temperature , with the Boltzmann constant. We use BZ for the first Brillouin zone and expand . To first order,
where is the band group velocity. Because the equilibrium energy is an even function of momentum, the band velocity is strictly odd. Therefore, the derivative is odd, and hence . We can now evaluate the macroscopic nonequilibrium responses at the first order. The linear orbital Edelstein response is defined as the expectation value of the OAM operator over the perturbed distribution:
In this expression, the integrand is the product of the odd OAM texture and the odd distribution . The resulting integrand is therefore an even function of , surviving the integration over the symmetric Brillouin zone and yielding a robust, nonzero orbital Edelstein polarization in the linear regime. Defining , the linear orbital Edelstein susceptibility is
Figure 4 shows the temperature dependence of the three components of the linear atomic orbital Edelstein susceptibility. For all components, the absolute magnitude of the response decreases monotonically with increasing temperature, as expected from the progressive thermal broadening of the Fermi surface. The different decay scales reflect the distinct orbital sectors sampled at the corresponding chemical potentials. The latter are held fixed throughout each temperature sweep. The values used for the radial, azimuthal, and longitudinal responses are, respectively, eV, eV, eV, eV. Two azimuthal curves are shown because sizeable responses occur in two separate energy regions, one near each of the two quoted chemical potentials.
Figure 4. Temperature dependence of the linear atomic Edelstein susceptibility at the chemical potentials that maximize the response. The different components reflect distinct orbital sectors and their thermal sampling near avoided crossings. (a) Radial response. (b) Azimuthal response near eV (solid line) and near eV (dashed line). The azimuthal component is appreciable in two distinct energy windows. (c) Longitudinal response.
In contrast, we evaluate the longitudinal nonequilibrium orbital current density generated in this same linear regime:
Here, the integrand consists of the product of three odd functions: the velocity , the OAM texture , and the distribution correction . The product of three odd functions remains strictly odd. When integrated over the symmetric limits of the Brillouin zone, this term exactly cancels out. Therefore, parity dictates that while the applied electric field successfully generates a finite Edelstein OAM polarization (a static accumulation of angular momentum), the continuous nonequilibrium orbital current completely vanishes at the first order ().
Second-Order Atomic OAM Response
Because the first-order field-induced correction to the orbital current vanishes by parity, the leading nonequilibrium correction appears at second order in the electric field. To fully characterize the transport behavior of the
helical channel, we extend the semiclassical BTE formalism to the second order in the electric field (). Recursively inserting into the transport equation allows us to capture the second-order nonequilibrium correction driven by the higher-order gradients of the distribution function:
Parity constraints strictly dictate which second-order macroscopic observables are permissible. Because is an odd function, its momentum derivative must be strictly even. Consequently, the second-order distribution is even in momentum. Applying this to the second-order OAM polarization, we find:
The integrand is an overall odd function of , dictating that the orbital Edelstein response is forbidden at second order by parity (). In contrast, we evaluate the nonequilibrium orbital current at the second order. The orbital current operator scales as the product of velocity and angular momentum, yielding the integral:
Because the product of velocity and OAM texture is strictly even, the total integrand is an even function of momentum and survives the Brillouin zone integration. To evaluate this rigorously and avoid severe numerical artifacts from directly differentiating the sharp Fermi-Dirac distribution, we integrate by parts twice over the periodic Brillouin zone. Since , , and are periodic at the zone boundaries, the boundary terms vanish. Defining the band current kernel , we obtain the numerically stable form
This form shows that the second-order orbital current depends directly on the momentum-space curvature of the orbital-velocity product within the occupied states. The field-induced orbital current to second order is , so the conductivity is given by:
Figure 5 shows the corresponding quadratic conductivities. Similar to Figure 4, each conductivity has a chemical potential fixed so it maximizes the conductivity; for the radial, azimuthal, and longitudinal conductivities, the chemical potentials are, respectively (in eV), , , , . All components decrease monotonically in magnitude as the temperature increases, reflecting the progressive thermal averaging of the momentum-resolved orbital-current structure.
Figure 5. Temperature dependence of the second-order orbital current conductivity for (a) radial, (b) azimuthal (solid: eV, dashed: eV), and (c) longitudinal components. As in Fig. 4, the azimuthal conductivity is appreciable in two chemical-potential windows.
Itinerant Orbital Angular Momentum
Having completed the analysis of the local atomic orbital angular momentum, we now turn to the itinerant contribution. Both quantities originate from the same multiorbital tight-binding Hamiltonian, but they describe physically distinct forms of electronic circulation. The atomic contribution is associated with the intra-site orbital operator acting within the local -orbital basis and is therefore determined by coherences between the amplitudes on a given site. The itinerant contribution instead originates from the motion of the electronic wave packet between distinct lattice sites and depends explicitly on the spatial geometry of the hopping network.
This distinction is particularly relevant for a double helix. In the local atomic sector, the second strand modifies the orbital hybridization and allows a radial OAM component that is absent in the corresponding isolated single-helix model. In the itinerant sector, the second strand changes the connectivity of the lattice itself: nearest-neighbor propagation along the two strands together with inter-strand hopping generates elementary closed circulation paths that are absent in a nearest-neighbor single helix. As shown below, these inter-strand loops possess a radial vector area in the co-rotating frame and therefore provide a new geometrical channel for a local radial itinerant OAM.
Modern-theory definition of itinerant orbital angular momentum
The position operator is not well defined for extended Bloch states in a periodic system, and the itinerant circulation cannot therefore be evaluated directly from the naive operator . The modern theory of orbital magnetization [26, 27] provides an origin-independent description in terms of the intrinsic orbital magnetic moment carried by a Bloch wave packet. For a cell-periodic Bloch eigenstate with energy , the intrinsic orbital magnetic moment is
where denotes the elementary charge. The corresponding itinerant orbital angular momentum is related to the orbital magnetic moment through
where is the electron mass and for pure orbital motion. In this section, denotes the full six-band Bloch Hamiltonian of Eq. (18). Using perturbation theory,
the three Cartesian components of the itinerant OAM may be written as
where denotes a cyclic permutation of . Thus,
The longitudinal component therefore represents only one component of the general itinerant orbital moment.
The full modern-theory expression is vectorial, although the screw symmetry of an infinite helix implies that a spatially uniform transverse component cannot survive as a bulk laboratory-frame vector. This does not exclude a transverse itinerant texture that rotates together with the molecular frame.
Bulk itinerant OAM and local circulating textures
The distinction between a bulk transverse moment and a local transverse texture is essential in a helical system. Consider a local itinerant moment written in the co-rotating frame as
Under one screw operation, and , where denotes a rotation about the global axis; the same relation holds for . Therefore and may remain finite and independent of in the co-rotating frame even though their Cartesian components rotate along the structure and vanish upon averaging over complete turns, . The longitudinal component, by contrast, is referred to the fixed global helix axis and may survive the spatial average, .
The modern-theory expression in Eq. (41) determines the bulk itinerant orbital moment. To understand the spatial origin of possible local transverse components, it is useful to complement this reciprocal-space formulation by examining the circulating part of the bond-current network generated by the same tight-binding Hamiltonian. This real-space construction is not presented as an additional contribution to the modern-theory OAM. Rather, it provides a local geometrical interpretation of the inter-site circulation underlying the itinerant response.
Local transverse circulation enabled by double-helix connectivity
The second strand introduces elementary closed inter-site paths that do not exist in the nearest-neighbor single helix. For an oriented cycle carrying a circulating component , the associated itinerant OAM is
where is the oriented vector area. Equation (44) is a geometrical interpretation of the same inter-site circulation described by the modern-theory expression, not an additional contribution to it.
Appendix B shows that the elementary double-helix plaquette has a vector area strictly radial, i.e., collinear with the local co-rotating radial direction. Hence a band state carrying circulation around this path has a local radial itinerant OAM. Time-reversed partners carry opposite circulation and cancel in equilibrium, whereas the odd-in- occupation shift induced by a longitudinal field makes a first-order radial response symmetry-allowed. We use this construction only to establish the transverse channel geometrically; its magnitude is not evaluated here. The explicit numerical calculation below is restricted to the longitudinal component, whose dominance in the present nearest-neighbor parameterization is discussed in Appendix B 5.
Itinerant OAM of the Double Helix
We now apply the general modern-theory expression derived above to the full six-band tight-binding Hamiltonian of the asymmetric double helix. Although the itinerant orbital angular momentum is in general a vector quantity, in the present work we evaluate explicitly its longitudinal component . As justified by the connectivity and hopping hierarchy discussed in Appendix B 5, this component is expected to provide the dominant itinerant contribution for the nearest-neighbor double-helix model considered here.
The Bloch Hamiltonian of the molecular system is parameterized by the longitudinal crystal momentum . Nevertheless, the longitudinal orbital moment is determined by the transverse position-weighted velocity matrix elements entering Eq. (41). These matrix elements are fixed by the full three-dimensional real-space embedding of the hopping network.
To generate them without introducing additional physical Bloch quantum numbers, we temporarily augment the longitudinal phase factor by auxiliary transverse phase variables and . These variables act only as generators of the Cartesian bond-displacement factors and ; they do not represent independent crystal momenta of the quasi-one-dimensional molecule. Thus, for a hopping contribution between two basis states and , connected by the spatial displacement , we define the auxiliary extension
where the sum includes all hopping processes connecting the corresponding basis states. The auxiliary extension reduces to the physical quasi-one-dimensional Hamiltonian on the line .
The transverse derivatives entering the orbital-moment formula are then generating derivatives evaluated on this physical line,
Equations (46) and (47) emphasize that these transverse generating derivatives are not independent phenomenological parameters. They are fixed by the same Slater–Koster hopping amplitudes used to construct , weighted by the actual Cartesian displacements of the corresponding bonds. In this sense, and are compact momentum-space generators of the position-weighted transverse hopping matrix elements of the embedded quasi-one-dimensional lattice. The three-dimensional geometry of the helix therefore enters the itinerant orbital moment through the physical spatial factors and , without assigning the molecule fictitious transverse translational symmetry.
The exact eigenvalue problem of the full Hamiltonian is
We define the transverse interband matrix elements
and
The longitudinal itinerant orbital angular momentum of band is then obtained directly from the modern-theory expression as
with for pure orbital motion. Equation (51) is the expression used for the numerical evaluation of the itinerant OAM texture shown below.
This form is particularly convenient computationally. At each momentum , the full Hamiltonian is diagonalized to obtain and , the matrices and are constructed from Eqs. (46) and (47), and the interband sum in Eq. (51) is evaluated for each of the six bands. No assignment of an exact bonding or antibonding quantum number is required.
The geometrical content of Eq. (51) becomes apparent by considering two hopping contributions entering the product of transverse derivatives. Their displacement factors occur in the antisymmetric combination .
The longitudinal itinerant OAM therefore probes oriented areas projected onto the transverse plane. In a helix, successive intra-strand hopping vectors rotate around the molecular axis, so propagation along the molecular backbone naturally carries a circulating component about the global direction. The modern-theory expression in Eq. (51) is the reciprocal-space representation of this helical itinerant circulation.
Because the Hamiltonian contains no magnetic field and preserves time-reversal symmetry, , the itinerant OAM texture is odd under momentum reversal,
This relation provides a direct numerical check of the calculation. In particular, the itinerant texture must vanish at the time-reversal-invariant momentum , and the equilibrium Brillouin-zone average of vanishes in the absence of an occupation imbalance.
Accordingly, we use Eq. (51) to calculate the six band-resolved longitudinal itinerant textures . We leave the transverse itinerant components as symmetry-allowed secondary channels, as their detailed quantitative evaluation lies beyond the scope of the present work.
Figure 6 shows the six band-resolved longitudinal itinerant textures. A prominent feature is the localized resonant enhancement of occurring near exclusively for the hybridized transverse branches (), while the remaining bands are negligible. This sharp resonance peak arises directly from the interband energy denominator in Eq. (51). Despite the presence of these sharp interband resonances, the complete itinerant texture rigorously preserves global time-reversal symmetry (), adhering to the odd-parity relation from Eq. (52).
Figure 6. Longitudinal itinerant OAM texture obtained from Eq. (51). The colors label the same six bands as Figure 2; orbital and bonding assignments refer only to dominant eigenstate character. The texture is odd under as required by time-reversal symmetry.
Itinerant Orbital Linear Edelstein Response
The nonequilibrium longitudinal itinerant accumulation is obtained from the same first-order occupation shift as the atomic response,
Since both and are odd in momentum, the integrand is even and the response is symmetry-allowed.
Figure 7 depicts the thermal evolution of the longitudinal itinerant orbital Edelstein susceptibility, . The magnitude of the itinerant susceptibility decreases monotonically with temperature, approaching zero as thermal broadening progressively averages the sharply momentum-dependent itinerant texture. In evaluating this response, the chemical potential is fixed at eV (near the energy where the bands exhibit an avoided crossing) rather than at the values of that maximize the local atomic Edelstein responses in Sec. V B. While the local atomic orbital response is driven by intra-site – coherences concentrated primarily near the band centers, the itinerant response is governed by the avoided crossing of the energy bands near 1.14 eV.
Figure 7. Temperature dependence of the longitudinal itinerant orbital Edelstein response susceptibility, (in units of ), evaluated at the optimal chemical potential eV.
Second-order itinerant orbital-current response
The longitudinal itinerant OAM texture can be placed on the same nonequilibrium footing as the atomic OAM discussed in Sec. V. Within the same semiclassical intra-band approximation, define the band-resolved itinerant current kernel
Both factors are odd under , so is even and its first-order field-induced correction vanishes. At second order, is even. The leading field-induced itinerant OAM current is therefore
The integrand is now even in momentum and the response is symmetry-allowed. Using the periodicity of the Brillouin zone and integrating by parts twice gives
Defining the quadratic itinerant orbital-current conductivity through
we obtain
which is the itinerant counterpart of the second-order atomic orbital-current conductivity derived in Sec. V. The same parity argument also shows that the second-order itinerant OAM accumulation vanishes,
because the itinerant texture is odd whereas is even. Thus the longitudinal itinerant sector exhibits the same response hierarchy as the atomic one: a finite linear Edelstein polarization but no linear field-induced orbital current, followed at second order by a finite orbital-current response but no quadratic OAM accumulation.
Atomic versus Itinerant Orbital Angular Momentum
The longitudinal channel provides the common sector in which the two orbital contributions can be compared without identifying them. The atomic quantity measures on-site circulation within the local -orbital basis, whereas measures inter-site wave-packet circulation through Eq. (51). This distinction also clarifies the relation to -orbital helical models: such models retain the itinerant channel but, by construction, cannot describe the atomic OAM generated by coherence among local orbitals. In the present multiorbital double helix, both mechanisms coexist in the same Hamiltonian. Their momentum dependence and Edelstein susceptibilities therefore quantify physically distinct contributions to the total orbital response.
Figure 8 compares the longitudinal atomic and itinerant responses at K. The linear susceptibilities follow different chemical-potential dependences: the atomic contribution tracks local inter-orbital coherence, whereas the itinerant response is enhanced near interband avoided crossings. For the representative parameters, the atomic contribution dominates the linear response throughout most of the plotted range. The
Figure 8. Longitudinal linear Edelstein susceptibilities (, left axis, in units of ) and second-order orbital-current conductivities (, right axis, in units of ) as functions of chemical potential at K. Solid dark- and light-green curves denote the atomic and itinerant linear responses, respectively; dashed dark-blue and cyan curves denote the corresponding second-order orbital-current conductivities. The two orders exhibit distinct chemical-potential dependences: the linear response is predominantly atomic, whereas the itinerant quadratic conductivity develops pronounced resonant structure and sign changes, with its strongest enhancement in the – eV window.
quadratic response displays a markedly different hierarchy: develops pronounced resonant enhancements and sign changes in selected chemical-potential windows, particularly around – eV, where its magnitude substantially exceeds the atomic contribution. Thus, the relative importance of atomic and itinerant OAM depends not only on chemical potential but also on the response order.
The combined analysis also suggests a channel hierarchy. In the atomic sector, the calculated longitudinal texture is the largest, followed by the azimuthal component, whereas the radial component is much smaller but is a sensitive symmetry diagnostic of inter-strand phase frustration. In the itinerant sector, longitudinal winding is expected to dominate because it relies on the strong intra-strand network, whereas radial itinerant circulation requires weaker inter-strand transfer. We retain the transverse itinerant channels as symmetry-allowed possibilities but do not evaluate them quantitatively in this work.
Discussion and Conclusions
We have developed a six-band tight-binding description of atomic and itinerant orbital angular momentum in an asymmetric double helix with characteristic B-form DNA geometry. Strand inequivalence and asymmetric inter-strand hopping prevent exact separation into bonding and antibonding sectors and, more importantly for the orbital response, generate relative phases among the local -orbital amplitudes. These phases activate a radial atomic texture that is absent in the corresponding single-helix model, while the azimuthal and longitudinal atomic components remain finite.
The driven atomic response follows a simple parity hierarchy. Time-reversal symmetry makes the OAM texture odd in momentum, so a longitudinal electric field produces a finite first-order orbital Edelstein polarization but no first-order field-induced OAM current. At second order the parity is reversed: the OAM accumulation vanishes whereas a quadratic orbital-current response is allowed. The itinerant sector exhibits the same hierarchy in the longitudinal channel. Figure 8 compares the atomic and itinerant linear susceptibilities and quadratic current conductivities on common chemical-potential axes, showing that the relative importance of the two microscopic mechanisms depends on the response order and on the portion of the multiorbital spectrum being sampled.
The double helix also changes the geometry of itinerant circulation. Longitudinal is generated by the ordinary helical winding along each strand and is evaluated explicitly with the modern Bloch-wave-packet expression. The second strand additionally creates elementary inter-strand plaquettes whose oriented vector area is purely radial in the co-rotating frame. This establishes a symmetry-allowed local radial itinerant channel distinct from the radial atomic texture: the former is an inter-site circulation effect, whereas the latter arises from on-site -orbital coherence. Because the plaquette channel necessarily involves weaker inter-strand transfer, we expect it to be subleading to longitudinal itinerant winding except where interband resonances enhance it; its full quantitative evaluation is left for future work.
The present model deliberately isolates orbital physics. It uses nearest-neighbor hopping, a phenomenological relaxation time, and no explicit spin–orbit coupling, contacts, or dephasing mechanism. These ingredients will be required to connect the calculated OAM directly to spin-selective two-terminal observables. Within those limitations, the main result is robust: a double helix is not merely two copies of a single helical channel. Inter-strand coupling creates new atomic phase structure and new itinerant circulation topology, providing independent microscopic routes to geometry-controlled orbital polarization and nonlinear orbital transport in chiral molecular conductors.
Acknowledgments
We acknowledge financial support from the Universidad San Francisco de Quito through the Poligrant POLI 41981. EV acknowledges support from the Doctoral program in Advanced Materials USFQ. We also acknowledge the use of AI-assisted tools for language refinement and discussion of theoretical directions during the preparation of this manuscript.
Microscopic Intra-Strand Parameters and Slater–Koster Expressions
This appendix provides the explicit structural and algebraic elements governing the intra-strand tight-binding kinetics summarized in Sec. II. It also summarizes in Table 1 all the B-DNA parameters used in the response and conductivity computations.
| Parameter | Value |
| Geometric Dimensions | |
| Radius () | 10.0 Å |
| Axial rise () | 3.4 Å |
| Azimuthal twist () | 36° |
| Groove asymmetry () | 150° |
| Vertical shear () | 0.15 Å |
| On-Site Energies: Strand 1 | (eV) |
| Radial () | 1.00 |
| Azimuthal () | 1.15 |
| Longitudinal () | −1.00 |
| On-Site Energies: Strand 2 ( eV) | (eV) |
| Radial () | 1.15 |
| Azimuthal () | 1.30 |
| Longitudinal () | −0.85 |
| Intra-Strand Hopping | (meV) |
| Radial-radial () | −20.0 |
| Azimuthal-azimuthal () | 10.0 |
| Longitudinal-longitudinal () | 50.0 |
| Chiral radial-azimuthal () | 3.5 |
| Chiral radial-longitudinal () | −14.3 |
| Chiral azimuthal-longitudinal () | 43.9 |
| Inter-Strand Hopping | (meV) |
| Symmetric radial () | −18.0 |
| Symmetric azimuthal () | −8.0 |
| Symmetric longitudinal () | 10.0 |
| Parity-breaking radial-azimuthal () | −7.5 |
| Parity-breaking radial-longitudinal () | −0.075 |
| Parity-breaking azimuthal-longitudinal () | 0.020 |
Table 1. Geometric dimensions and representative tight-binding parameters used for the B-DNA-inspired double-helix model. The structural values follow characteristic B-form DNA scales, while the electronic parameters provide an effective multiorbital description consistent with model charge-transport energy scales [5, 23].
Modeling the B-DNA single helical backbone with radius , axial rise , and an angular twist between successive nucleotides, the intra-strand nearest-neighbor spatial distance is given by:
The unit bond vector connecting adjacent sites along strand is projected onto the local cylindrical coordinate basis . These projections define the direction cosines , , and :
By applying the Slater–Koster parameterization for -orbital overlaps and defining the parameter , the microscopic intra-strand hopping elements are analytically evaluated as:
These matrix elements dictate the dispersion profiles and generate the momentum-dependent, imaginary odd-parity hybridization channels and real even-parity coupling used to formulate the Bloch matrix blocks in Eq. (17). The complete form of the couplings and the on-site terms is given by
Geometrical Origin of Transverse Itinerant OAM
This appendix gives the geometrical and symmetry analysis of the radial itinerant channel introduced in the main text.
Single-helix winding and the new circulation topology of the double helix
The distinction between the single and double helix is particularly transparent at the nearest-neighbor level. An isolated single helix possesses a directed helical propagation path and therefore supports itinerant circulation around the global molecular axis. This winding motion gives the longitudinal component already identified in the single-helix problem.
However, a nearest-neighbor single strand contains no elementary closed inter-site plaquette. Its connectivity is simply , and consequently, there is no local nearest-neighbor circulation cycle analogous to the one introduced below. The double helix changes this topology even without extending the hopping range. Nearest-neighbor hopping along each strand, combined with inter-strand hopping, generates the elementary closed path
This plaquette represents a circulation channel that is absent in the corresponding nearest-neighbor single-helix model.
Vector area of the elementary double-helix plaquette
Using the coordinates defined for the two strands,
and
the vector area of the plaquette in Eq. (76) is
Introducing the azimuthal direction associated with the center of the plaquette,
and the corresponding co-rotating radial unit vector , the vector area reduces to , where
For this elementary inter-strand plaquette, , . Thus, the new elementary circulation path introduced by the second strand possesses a purely radial vector area in its natural co-rotating frame. A circulating electronic current around this plaquette can therefore generate a local radial itinerant OAM, . Because rotates by from one unit cell to the next, this local radial itinerant component forms a helical texture and need not produce a nonzero uniform transverse moment in the laboratory frame. Figure 9 displays this elementary directed cycle and its radial oriented area.

Figure 9. Elementary inter-strand plaquette of the double helix associated with the radial itinerant orbital-angular-momentum channel. The directed circulation follows the path . The corresponding oriented vector area is collinear with the local co-rotating radial direction , demonstrating that this elementary plaquette generates a purely radial itinerant OAM contribution (only its direction indicated). The arrows indicate the oriented local circulation associated with this OAM channel; this circulating component is compatible with and superimposed on the direction of the longitudinal bias-driven transport.
The existence of the radial loop geometry does not require the major–minor groove asymmetry or the vertical shear. Indeed, in the symmetric geometrical limit and ,
which remains finite. The second strand, therefore, creates a radial itinerant circulation channel, even in the symmetric double helix. Structural and compositional asymmetries enter at the electronic level by modifying the band-resolved circulation amplitudes; time-reversal symmetry still enforces equilibrium cancellation, while a driven state may select a preferred circulation.
Equilibrium cancellation and nonequilibrium selection of loop circulation
A finite plaquette area does not by itself generate an orbital moment. In the absence of magnetic fields or other explicit time-reversal-symmetry breaking, the equilibrium system contains time-reversed electronic partners with opposite circulating currents. Denoting the band-resolved plaquette circulation by , time-reversal symmetry requires
Because the equilibrium Fermi–Dirac distribution is even, , the equilibrium average of the plaquette circulation vanishes,
The radial plaquette geometry therefore defines an available itinerant OAM channel rather than an equilibrium radial orbital polarization.
A longitudinal nonequilibrium drive changes this conclusion. Within the relaxation-time Boltzmann description introduced previously, the distribution becomes
where . The applied field therefore preferentially populates one propagation direction over its time-reversed partner. In an asymmetric chiral double helix, the propagation direction can be correlated with a particular circulation sense around the inter-strand plaquette. If the band-resolved plaquette current obeys (83), the nonequilibrium correction
contains an even integrand and is therefore symmetry-allowed to remain finite. Combining the driven plaquette circulation with (44) and the radial vector area gives a local field-induced radial itinerant OAM. This mechanism must be distinguished from spontaneous equilibrium circulation. The applied electric field does not act as an electromotive force around the microscopic plaquette. Instead, it generates an asymmetric occupation of propagating states. Chirality and structural or compositional asymmetry can correlate the direction of propagation with the local circulation carried by those states, so that the cancellation between opposite loop-current partners is incomplete away from equilibrium.
The physical sequence is therefore: double-helix loop geometry radial circulation channel, followed by nonequilibrium occupation imbalance + structural/compositional asymmetry preferred loop circulation .
Consequently, the second strand plays different roles in the atomic and itinerant sectors. For the atomic OAM, inter-strand hybridization modifies the relative phases of the local -orbital amplitudes and activates the radial coherence entering . For the itinerant OAM, the second strand creates a new closed inter-site circulation path with a radial vector area. The two radial textures therefore arise from the same double-helix Hamiltonian but from physically distinct microscopic mechanisms.
Connection with the longitudinal itinerant OAM
The radial plaquette circulation described above is complementary to the longitudinal itinerant OAM associated with propagation around the global helix axis. We evaluate the latter explicitly in Sec. VI D from the Bloch-wave-packet orbital-moment expression, and it already exists for an isolated single helix.
At the nearest-neighbor level, the distinction between the two geometries can therefore be summarized as single-helix winding , whereas double-helix inter-strand plaquette under nonequilibrium circulation selection.
Longitudinal versus transverse itinerant channels
The longitudinal itinerant component arises from propagation along the strong intra-strand hopping network: each nearest-neighbor step carries a transverse displacement as the backbone winds around the global axis. By contrast, the elementary radial channel necessarily uses the weaker inter-strand transfer amplitudes to close the plaquette. For the parameterization of Table 1, this connectivity provides a natural power-counting argument for retaining as the dominant itinerant component in the explicit calculation. It is not an exact symmetry statement: interband denominators can enhance a transverse channel near avoided crossings, and longer-range hopping may generate additional circulation paths. The present model does not identify a comparably direct nearest-neighbor geometry for an azimuthal itinerant component.
References
- [1]R. Naaman, Y. Paltiel, and D. H. Waldeck, J. Phys. Chem. Lett. 11, 3660 (2020).
- [2]B. P. Bloom, Y. Paltiel, R. Naaman, and D. H. Waldeck, Chem. Rev. 124, 1950 (2024).
- [3]C. D. Aiello, J. M. Abendroth, M. Abbas, A. Afanas’ev, S. Agarwal, A. S. Banerjee, D. N. Beratan, J. N. Belling, B. Berche, A. Botana, J. R. Caram, G. L. Celardo, G. Cuniberti, A. Garcia-Etxarri, A. Dianat, I. Diez-Perez, Y. Guo, R. Gutierrez, C. Herrmann, J. Hihath, S. Kale, P. Kurian, Y.-C. Lai, T. Liu, A. Lopez, E. Medina, V. Mujica, R. Naaman, M. Noormandipour, J. L. Palma, Y. Paltiel, W. Petuskey, J. C. Ribeiro-Silva, J. J. Sáenz, E. J. G. Santos, M. Solyanik-Gorgone, V. J. Sorger, D. M. Stemer, J. M. Ugalde, A. Valdes-Curiel, S. Varela, D. H. Waldeck, M. R. Wasielewski, P. S. Weiss, H. Zacharias, and Q. H. Wang, ACS Nano 16, 4989 (2022).
- [4]H. J. Eckvahl, G. Copley, R. M. Young, M. D. Krzysianiak, and M. R. Wasielewski, J. Am. Chem. Soc. 146, 24125 (2024).
- [5]S. Varela, V. Mujica, and E. Medina, Phys. Rev. B 93, 155436 (2016).
- [6]R. Gutierrez and G. Cuniberti, Acta Phys. Pol. A 127, 185 (2015).
- [7]A.-M. Guo and Q.-F. Sun, Proc. Natl. Acad. Sci. U.S.A. 111, 11658 (2014).
- [8]X. Yang, C. H. van der Wal, and B. J. van Wees, Nano Lett. 20, 6148 (2020).
- [9]X. Yang, Connecting chirality and spin in electronic devices, Ph.D. thesis, University of Groningen, Groningen, Netherlands (2020).DOI
- [10]S. Varela, R. Gutierrez, G. Cuniberti, E. Medina, and V. Mujica, J. Chem. Phys. 161, 114111 (2024).
- [11]T. Yoda, T. Yokoyama, and S. Murakami, Sci. Rep. 5, 12024 (2015).
- [12]T. Yoda, T. Yokoyama, and S. Murakami, Nano Lett. 18, 916 (2018).
- [13]D. Go, D. Jo, C. Kim, and H.-W. Lee, Phys. Rev. Lett. 121, 086602 (2018).
- [14]T. P. Cysne, L. M. Canonico, M. Costa, R. B. Muniz, and T. G. Rappoport, npj Spintronics 3, 39 (2025).
- [15]S. Varela, M. Peralta, V. Mujica, B. Berche, and E. Medina, SciPost Phys. Core 6, 044 (2023).
- [16]M. Ozeri, J. Xu, G. Bauer, L. A. B. Olde Olthof, G. Kimbell, A. Wittmann, S. Yochelis, J. W. A. Robinson, V. Paltiel, and O. Millo, J. Phys. Chem. Lett. 14, 4941 (2023).
- [17]S. Mishra, A. K. Mondal, E. Z. B. Smolinsky, R. Naaman, M. Maeda, T. Nishimura, T. Taniguchi, T. Yoshida, K. Takayama, and E. Yashima, Angew. Chem. Int. Ed. 59, 14671 (2020).
- [18]B. Göbel, L. Schimpf, and I. Mertig, Phys. Rev. Res. 7, 033180 (2025).
- [19]Y. Liu, J. Xiao, J. Koo, and B. Yan, Nat. Mater. 20, 638 (2021).
- [20]Y. Yatabe and H. Akera, J. Chem. Phys. 159, 074701 (2023).
- [21]T. Georgiou, J. L. Palma, V. Mujica, R. N. Schwartz, and L.-S. Bouchard, J. Chem. Phys. 163, 204104 (2025).
- [22]D. Cordova, B. Berche, and E. Medina, J. Chem. Phys. 165, 054309 (2026).
- [23]L. G. D. Hawke, G. Kalosakas, and C. Simserides, Eur. Phys. J. E 32, 291 (2010).
- [24]X. Shui, L. McFail-Isom, G. G. Hu, and L. D. Williams, Biochemistry 37, 8341 (1998).
- [25]J. Shi, P. Zhang, D. Xiao, and Q. Niu, Phys. Rev. Lett. 96, 076604 (2006).
- [26]D. Xiao, J. Shi, and Q. Niu, Phys. Rev. Lett. 95, 137204 (2005).
- [27]O. Busch, I. Mertig, and B. Göbel, Phys. Rev. Res. 5, 043052 (2023).






