betaFeedback

WorksPhysics and Astronomy

Electron States Emerging at Magnetic Domain Wall of magnetic semiconductors with strong Rashba effect

Abstract

In the present article, we explore the electron properties of magnetic semiconductors with strong Rashba spin-orbit coupling taking into account the presence of domain walls at the sample surface. We consider antiphase domain walls separating domains with both in-plane and out-of-plane magnetization as well as noncollinear domain walls. First, we propose the model and unveil general physical picture of phenomenon supported by analytical arguments. Further, we perform a comprehensive tight-binding numerical calculations to provide a profound understanding of our findings. A domain wall separating domains with any polarization directions is demonstrated to host a bound state. What is more interesting is that we predict that either of these domain walls also induces one-dimensional resonant state. The surface energy spectrum and spin polarization of the states are highly sensitive to the magnetization orientation in the adjacent domains. The spectral broadening and the spatial localization of the resonant state depend significantly on a relation between the Rashba splitting and the exchange one. Our estimation shows that chiral conducting channels associated with the long-lived resonant states can emerge along the magnetic domain walls and can be accessed experimentally at the surface of BiTeI doped with transition metal atoms.

INTRODUCTION

In designing next generation of the spintronic technologies, two novel approaches to non-volatile memory and information processing can be highlighted. In one of these, electrically manipulated magnetic domain wall (DW), rather than domains, serves as an active element [1, 2]. Another approach is based on generation of spin polarized dissipationless current [3]. Recent theoretical studies [3, 4, 5] and experimental findings [5, 6] have predicted and demonstrated that these approaches can be successfully combined with each other in both magnetically doped topological insulator (Bi,Bi)2_2Te3_3 thin films and intrinsic antiferromagnetic TI MnBi2_2Te4_4 flakes, where DWs host robust in-gap bound electron states. These topologically protected states may impact on the realization of peculiar quantum phenomena such as quantum anomalous Hall effect and axion insulator phase [3]. The linear dispersion relation, spin-momentum locking and quantized conductivity are special properties of these states. However, their characteristic features such as group velocity, spin polarization direction and spatial localization are determined by both the electrostatic conditions at the TI surface or the interfaces between TI and trivial insulator and the relative orientation of the spontaneous magnetization in the domains separated by DW [7, 8, 9, 10, 11, 12]. Thus, in semiconductor materials with topologically nontrivial band structure, network of the unique channels for chiral electron waves can be created and manipulated via the surface/interface potential and the spatially-varying magnetization texture [7, 8, 9, 10, 11, 12]. In turn, an accumulation of carriers at the bound state makes possible to move a magnetic DW together with the conducting channel using an external electric field.

A natural question is whether there are other solids with electron properties similar to those of magnetic TIs. Among these candidates, materials containing high-Z (heavy) elements where strong spin-orbit coupling (SOC) and spatial inversion symmetry breaking are of prominent interest. According to the Rashba predictions [13, 14, 15], owing to the structural inversion asymmetry, electrons propagating along the surface or interface naturally experience an embedded potential gradient perpendicular to the boundary. Hereby, even in nonmagnetic system, the spin-degeneracy of the two-dimensional (2D) band structure is lifted out, without violating the time reversal symmetry. This symmetry can be broken by doping with magnetic impurities or by designing magnetic heterostructures. It bears intriguing consequences for electrical transport properties of materials wherein the Rashba splitting of the surface states intertwines with an exchange splitting. To date there are two prominent examples of real diluted ferromagnetic (FM) semiconductor with the highest observed Rashba SOC: Mn-doped GeTe [16, 17] and V-doped BiTeI [18, 19] [20]. The works [16] have experimentally witnessed the entanglement and manipulation of the magnetic order and SOC splitting in multiferroic Rashba semiconductor Mn-doped GeTe and the presented direct evidence for the existence of a strong magnetoelectric coupling between the electric polarization and the magnetization in this material. Such a coupling can result in unconventional spin textures depending on electric biasing under switching, which readily indicates that the (Ge,Mn)Te samples are far from the ideal monodomain phase. Transport measurement on (Ge,Mn)Te film has demonstrated that the Berry curvature dominantly contributes to the anomalous Hall conductivity [21]. As for V-doped BiTeI, the angle-resolved photoemission spectroscopy (ARPES) studies [18, 19] [20] have demonstrated the giant Rashba-type splitting and the huge exchange gap of about 90 meV for Bi0.985V0.015TeI\mathrm{Bi}_{0.985}\mathrm{V}_{0.015}\mathrm{TeI} at the Γ\Gamma-point. Thereat such a sizable gap does not correspond to the observed very weak total out-of-plane magnetization [19]. That can be explained by the spontaneous formation of the magnetic domains with opposite spin polarization below the Curie temperature.

The surface of the as-grown diluted magnetic semiconductors with strong Rashba SOC can be highly inhomogeneous, with multi-domain structure [17, 19]. It consists of fragments distinguished by magnitude and orientation of electric and magnetic polarization as well as spatial scale and shape. These materials can display electronic properties, which dramatically differ from those of the ideal uniform counterpart. The main reason for this is thought to be boundaries between the fragments, i.e., DWs separating ferroelectric or FM domains. It raises the question of whether DW can support a spatial confinement of electron density and how it affects the electronic spin-polarized transport. However, an influence of DWs on electron states and the corresponding contribution to the transport and other properties of the magnetic Rashba systems remain still poorly understood. In the present article we provide theoretical study of electron states in the presence of a nonuniform exchange field caused by a single magnetic DW at the surface of the FM material with the Rashba SOC. Such a special case has not been solved so far, seriously hampering the analysis and interpretation of experimental data. Here, by combining the analytic formalism and detailed numerical calculations, we thoroughly investigate the spin-resolved spectral and spatial characteristics of the Rashba semiconductors with magnetic order. It is found that the DW texture creates two different types of the quasi-one-dimensional (1D) spin-polarized electron states in terms of their properties. First, a purely bound state confined to DW is formed. It has a quasi-parabolic dependence of energy vs momentum near the 2D surface band continuum. Second, a peculiar resonant state with a quasi-linear spectrum and a spin-momentum locking appears as a prominent redistribution of an electron density around DW. We establish the universality of the resonant state that emerges in the class of materials under consideration irrespective of the vector texture of magnetic DW. Along with that, the velocity and spin polarization of such a state non-trivially depends on both mutual magnetization orientation in adjacent domains and orientation relative to the surface. The DW-induced resonant state and its features resemble a canonical situation in magnetic TI, where a bound edge state emerges at DW for the topological reason [7, 10, 11, 22]. The dissimilarity proves to be crucial at relatively large exchange splitting, when the resonant state energy is strongly broadened due to coupling to the 2D quasiparticle continuum. In the context of the transport properties, one should expect that the magnetic DWs play no lesser important role in the Rashba-type semiconductors than in TIs [23]. Among conceptual consequences, it implies a certain extension of the principle of bulk-boundary correspondence. Our investigation proposes a new physical insight on quantum transport in magnetic spin-orbit materials.

The rest of the paper is organized as follows. In Section II we introduce the model combining effects of Rashba SOC and a DW-type magnetization distortion on the electronic properties of a magnetic semiconductor. By the example of a single antiphase DW we give a general understanding of the origin of the states at DW and analytically estimate their spectral features. In Section III, we perform comprehensive tight-binding calculations to provide a profound insight into momentum- and spin-resolved spectral density of both 2D and 1D states. We investigate the bound and resonant electron states appearing at sharp interfaces between domains with opposite out-of-plane (Subsection IIIA) and in-plane (Subsection IIIB) magnetizations. Similarly, we address the situation with a noncollinear DW structure (Subsection IIIC). Section IV contains detailed discussion on the possibility of experimental observation of the resonant states in real magnetic semiconductors with strong Rashba effect in particular BiTeI. We summarize the main results in Section V.

Model and general relations: Analytical treatment.

We study a system of 2D Rashba electrons appearing at the surface of a three-dimensional semiconductor due to inversion symmetry violation. These electrons experience an exchange field which may be caused by either localized moments in the semiconductor host or a magnetic material layer interfaced with the semiconductor via proximity effect. Our analysis is based on the minimal model of electrons moving along the surface (x,y)(x,y) and affected by both SOC and exchange field [14, 15, 24], which is described by the 2D low energy Hamiltonian:

H(k)=βk2σ0−α([k×σ]⋅ez)+J(M⋅σ),(1)H(\mathbf{k}) = \beta k^2\sigma_0 - \alpha([\mathbf{k} \times\boldsymbol{\sigma}] \cdot\mathbf{e}_z) + J(\mathbf{M} \cdot\boldsymbol{\sigma}), \tag*{(1)}

where k=(kx,ky)\mathbf{k} = (k_x,k_y) is the momentum, σ=(σx,σy,σz)\boldsymbol{\sigma} = (\sigma_x,\sigma_y,\sigma_z) is the Pauli matrix vector, σ0\sigma_0 is a unit 2×22 \times2 matrix. The first term, expanded to quadratic order in kk around the Γˉ\bar{\Gamma}-point (the 2D Brillouin zone center), represents an electron kinetic energy, where the coefficient β=1/2m∗\beta= 1/2m^* is inversely proportional to the effective mass m∗m^*. We will set ℏ=1\hbar= 1 everywhere. In the vector product term, α\alpha is the Rashba SOC parameter, ez\mathbf{e}_z is the direction of the SOC field along which inversion symmetry is broken.

The electron spins are coupled to magnetization M=(Mx,My,Mz)\mathbf{M} = (M_x,M_y,M_z) via an exchange integral JJ, which is assumed to be isotropic and, for definiteness, positive. The parameter α\alpha is uniform in the (x,y)(x,y) plane, while the magnetization M=M(x,y)\mathbf{M} = \mathbf{M}(x,y) is in general position-dependent. In a locally homogeneous surface region, where the magnetization M\mathbf{M} does not vary, the energy spectrum of the Hamiltonian (1) is given by E±(k)=βk2±JMz2+(αkx−JMy)2+(αky+JMx)2E_\pm(\mathbf{k}) = \beta k^2 \pm\sqrt{JM_z^2 + (\alpha k_x - JM_y)^2 + (\alpha k_y + JM_x)^2}. Figure 1 illustrates the dispersion E±(k)E_\pm(\mathbf{k}) for a nonmagnetic case (Fig. 1.a) as well for the out-of-plane (Fig. 1.b) and in-plane (Fig. 1.c) the magnetization orientations. Note that, in experimental situations, the 2D band structure E±(k)E_\pm(\mathbf{k}) of the semiconductor surface can be controlled by electric gating as well as by external magnetic field applied to the sample.

To learn the role of a magnetic texture in the surface state modification, let us for simplicity restrict ourselves to a discussion on Rashba fermions coupled to an exchange field of a single magnetic DW approximated with the piece-wise constant profile M(x)=M(r)h(x)+M(l)h(−x)\mathbf{M}(x) = \mathbf{M}^{(r)}h(x) + \mathbf{M}^{(l)}h(-x) where h(x)h(x) denotes the Heaviside function. In the following, we use (r)(r) and (l)(l) to label the model parameters for the right and left semi-infinite domains, respectively. The DW line runs in the ey\mathbf{e}_y direction for concreteness. The magnetization of both domains is assumed to remain in (x,z)(x,z) plane, i.e. My(r)=My(l)=0M_y^{(r)} = M_y^{(l)} = 0. Such magnetization M(x)\mathbf{M}(x) breaks translation symmetry along the ex\mathbf{e}_x direction and can cause the appearance of peculiar 1D states. The solutions of the corresponding problem, H(x,ky)Θ(x,ky)=ϵ(ky)Θ(x,ky)H(x,k_y)\Theta(x,k_y) = \epsilon(k_y)\Theta(x,k_y), can be constructed by suitably matching at the interface x=0x = 0 the spinor components of the envelope wave-function Θ(x,ky)\Theta(x,k_y) inherent to homogeneous domain regions around DW. The eigen energies and functions of the problem depend parametrically on the momentum kyk_y that is preserved. For each value of ϵ\epsilon there are four characteristic wave-vectors which are given by

[q1,2(r)(ky,ϵ)]2=ky2−12β2[2βϵ+α2+α4+4α2βϵ+4β2((Δz(r))2+(Δx(r))2+2αkyΔx(r))],[q1,2(l)(ky,ϵ)]2=ky2−12β2[2βϵ+α2+α4+4α2βϵ+4β2((Δz(l))2+(Δx(l))2−2αkyΔx(l))],(2)\begin{aligned} [q_{1,2}^{(r)}(k_y,\epsilon)]^2 &= k_y^2 - \frac{1}{2\beta^2}\left[2\beta\epsilon+ \alpha^2 + \sqrt{\alpha^4 + 4\alpha^2\beta\epsilon+ 4\beta^2\left((\Delta_z^{(r)})^2 + (\Delta_x^{(r)})^2 + 2\alpha k_y\Delta_x^{(r)}\right)}\right],\\ [q_{1,2}^{(l)}(k_y,\epsilon)]^2 &= k_y^2 - \frac{1}{2\beta^2}\left[2\beta\epsilon+ \alpha^2 + \sqrt{\alpha^4 + 4\alpha^2\beta\epsilon+ 4\beta^2\left((\Delta_z^{(l)})^2 + (\Delta_x^{(l)})^2 - 2\alpha k_y\Delta_x^{(l)}\right)}\right], \tag*{(2)} \end{aligned}

where Δx,z(r,l)=JMx,z(r,l)\Delta_{x,z}^{(r,l)} = JM_{x,z}^{(r,l)}. For the sake of definiteness, we assume Δx,y(r,l)>0\Delta_{x,y}^{(r,l)} > 0. The linear combination of the corresponding exponential functions, ∼exp⁡[±q1,2(r,l)(ky,ϵ)x]\sim\exp[\pm q_{1,2}^{(r,l)}(k_y,\epsilon)x], which satisfies the boundary conditions, provides the sought solution.

To have reference points, we begin with the study of the case of an easy-axis anisotropy, when the out-of-plane magnetization M=ezMz\mathbf{M} = \mathbf{e}_z M_z (Mx=My=0M_x = M_y = 0) is preferred in the ground state. The system can be characterized by two energy scales, namely, the Rashba SOC splitting, ESO=kSO2/2m∗=α2/4βE_{\mathrm{SO}} = k_{\mathrm{SO}}^2/2m^* = \alpha^2/4\beta, being the energy difference between the crossing point and the band minimum at M=0\mathbf{M} = 0, kSO=m∗α=α/2βk_{\mathrm{SO}} = m^*\alpha= \alpha/2\beta is the minimum position in the momentum space. The magnetization M=ezMz\mathbf{M} = \mathbf{e}_zM_z lifts the spin degeneracy at the Mammers point and opens the exchange gap $2\Delta_z = 2JM_z$ at k=0k = 0 between the upper and lower spin-split subbands, $E_\pm(k) = \beta k^2 \pm Δz2+α2ky2\sqrt{\Delta_z^2+\alpha^2 k_y^2}, in the spectrum of the Hamiltonian (1) (see Figure 1.b). The lower subband exhibits one minimum or two minima depending on whether the system is in the exchange-dominated regime, Δz>2Eso\Delta_z > 2E_{\mathrm{so}}, or in the SOC-dominated regime, Δz<2Eso\Delta_z < 2E_{\mathrm{so}}. When Δz>2Eso\Delta_z > 2E_{\mathrm{so}} and ϵ<−Δz\epsilon< -\Delta_z, [q1(r,l)(ky,ϵ)]2[q_1^{(r,l)}(k_y,\epsilon)]^2 and [q2(r,l)(ky,ϵ)]2[q_2^{(r,l)}(k_y,\epsilon)]^2 are positive. In turn, when 2Eso>Δz2E_{\mathrm{so}} > \Delta_z and ϵ<−Eso(1+(Δz/2Eso)2)\epsilon< -E_{\mathrm{so}}(1+(\Delta_z/2E_{\mathrm{so}})^2), [q1(r,l)(ky,ϵ)]2[q_1^{(r,l)}(k_y,\epsilon)]^2 and [q2(r,l)(ky,ϵ)]2[q_2^{(r,l)}(k_y,\epsilon)]^2 are complex conjugated each other. In both cases, the bound electron state (indexed with ’B’) with the energy ϵ=ϵB(⊥)(ky)\epsilon=\epsilon_B^{(\perp)}(k_y) below the conduction band bottom occurs at the out-of-plane DW [’(⊥)(\perp)’] on the localization scale ∼∣Re⁡q1,2(r,l)(ky,ϵ)∣−1\sim|\operatorname{Re}q_{1,2}^{(r,l)}(k_y,\epsilon)|^{-1}. In particular, in the case of the antiphase DW with profile M(x,y)=ezMzsgn⁡(x)M(x,y)=e_zM_z\operatorname{sgn}(x), when q1(r)(ky,ϵ)=q1(l)(ky,ϵ)=q1,2(ky,ϵ)q_1^{(r)}(k_y,\epsilon)=q_1^{(l)}(k_y,\epsilon)=q_{1,2}(k_y,\epsilon), there appears the state with the dispersion relation ϵB(⊥)(ky)\epsilon_B^{(\perp)}(k_y), which is asymmetric with respect to ky=0k_y=0 [25]. At ky=0k_y=0 the energy ϵB(⊥)(0)\epsilon_B^{(\perp)}(0) satisfies the equation

Band structure of the semiconductor surface with spin-orbit Rashba splitting in three different scenarios

Figure 1. Band structure of the semiconductor surface with spin-orbit Rashba splitting in three different scenarios: no magnetic order (a), subjected to the uniform exchange field with out-of-plane direction (b) and in-plane direction (c).

βϵ{[q1(0,ϵ)]2+[q2(0,ϵ)]2}+Δz2=0.\beta\epsilon\left\{[q_1(0,\epsilon)]^2+[q_2(0,\epsilon)]^2\right\}+\Delta_z^2=0.

Interestingly, when Δz/2Eso≪1\Delta_z/2E_{\mathrm{so}}\ll1, the dispersion curve ϵB(⊥)(ky)\epsilon_B^{(\perp)}(k_y) approaches the energy level −Eso-E_{\mathrm{so}} occupying the interval ∣ky∣<kso|k_y|<k_{\mathrm{so}}. In this limit, one obtains ϵB(⊥)(0)=−(Eso+Δz2/2Eso)\epsilon_B^{(\perp)}(0)=-(E_{\mathrm{so}}+\Delta_z^2/2E_{\mathrm{so}}) and q1,2(0,ϵB(⊥)(0))=k(⊥)±iksoq_{1,2}(0,\epsilon_B^{(\perp)}(0))=k^{(\perp)}\pm ik_{\mathrm{so}} (where k(⊥)=Δz/αk^{(\perp)}=\Delta_z/\alpha), in other words, the envelope function decays slowly away from DW and simultaneously quickly oscillates. Thus, the magnetic DW acts as a short-range linear defect for Rashba’s fermions, which causes the bound state ϵB(⊥)(ky)\epsilon_B^{(\perp)}(k_y) below the 2D band continuum.

Apart from the bound electron state, we are interested in another specific type of 1D electron state, which can arise at the out-of-plane magnetic DW. The energy ϵ\epsilon of this state lies inside the local exchange gap. For definiteness, we will talk about the antiphase DW, M(x,y)=ezMzsgn⁡(x)M(x,y)=e_zM_z\operatorname{sgn}(x). Note that, in the energy range E+(ky)>ϵ>E−(ky)E_+(k_y)>\epsilon>E_-(k_y), the value q1(ky,ϵ)q_1(k_y,\epsilon) is real, while the value q2(ky,ϵ)q_2(k_y,\epsilon) is imaginary. Therefore, on the one hand, the spinor wave with the wave-vector q2(ky,ϵ)q_2(k_y,\epsilon) propagates along the surface and, having reached the magnetic DW, is partly transmitted through the interface or partly reflected from it. On the other hand, on each side of the DW, there is an evanescent wave ∼exp⁡[±q1(ky,ϵ)x]\sim\exp[\pm q_1(k_y,\epsilon)x]. The full envelope function is expressed as a superposition of the above propagating and evanescent waves. It describes an electron state that is placed within the local exchange gap and degenerate with the continuum of the lower subband E−(ky)E_-(k_y). So one can consider this state as a resonant one (’R’). Within the scattering theory, the characteristics of the resonant state are contained in the reflection and transmission coefficients, which can be analytically estimated for the sharp DW. Under conditions of a relatively weak exchange energy, Δz≪4Eso\Delta_z\ll4E_{\mathrm{so}}, cumbersome calculations give us the expression for the reflection coefficient as a function of energy and momentum

R(⊥)(ϵ,ky)=Δz4Esoϵ+Re⁡[ϵR(⊥)(ky)]ϵ−ϵR(⊥)(ky).(3)R^{(\perp)}(\epsilon,k_y)=\frac{\Delta_z}{4E_{\mathrm{so}}}\frac{\epsilon+\operatorname{Re}[\epsilon_R^{(\perp)}(k_y)]}{\epsilon-\epsilon_R^{(\perp)}(k_y)}. \tag*{(3)}

where Re⁡[…]\operatorname{Re}[\ldots] denotes the real part. Writing down Eq. (3) under the stipulation Δz/4Eso≪1\Delta_z/4E_{\mathrm{so}}\ll1 and ∣ky∣/kso≪1|k_y|/k_{\mathrm{so}}\ll1 we keep main non-vanishing terms in the dispersion relation

ϵR(⊥)(ky)=αky+ωR(⊥)−iΓR(⊥).(4)\epsilon_R^{(\perp)}(k_y)=\alpha k_y+\omega_R^{(\perp)}-i\Gamma_R^{(\perp)}. \tag*{(4)}

In this limit, the resonant state inside the local exchange gap, ∣ϵR(⊥)(ky)∣<Δz|\epsilon_R^{(\perp)}(k_y)|<\Delta_z, is specified by linear momentum-energy dependence with the group velocity α\alpha, small energy shift upwards ωR(⊥)=Δz2/4Eso\omega_R^{(\perp)}=\Delta_z^2/4E_{\mathrm{so}} and scanty broadening width ΓR(⊥)=Δz2/8Eso\Gamma_R^{(\perp)}=\Delta_z^2/8E_{\mathrm{so}}. Such a smallness of ΓR(⊥)\Gamma_R^{(\perp)} justifies the expansion (4) for the resonance. A notable feature is that the resonance narrows significantly faster than the exchange gap closes. Near the resonance, the squared reflection probability takes the Lorentzian form, ∣R(⊥)(ϵ,0)∣2≈(ΓR(⊥))2(ϵ−ωR(⊥))2+(ΓR(⊥))2|R^{(\perp)}(\epsilon,0)|^2\approx\frac{(\Gamma_R^{(\perp)})^2}{(\epsilon-\omega_R^{(\perp)})^2+(\Gamma_R^{(\perp)})^2}, with the distance from the resonance, it goes to the value ∣R(⊥)(ϵ,0)∣2≈(Δz/4Eso)2|R^{(\perp)}(\epsilon,0)|^2\approx(\Delta_z/4E_{\mathrm{so}})^2, and it passes through zero (an antire­sonance) at ϵ=−ωR(⊥)\epsilon=-\omega_R^{(\perp)}. The spatial distribution of the envelope function of the state (4) exhibits the significant localized component around DW, ∼a1exp⁡(±κ(⊥)∣x∣)\sim a_1\exp(\pm\kappa^{(\perp)}|x|), and the propagating wave of a minor magnitude, ∼a2exp⁡(±2iksox)\sim a_2\exp(\pm2ik_{\mathrm{so}}x), i.e. ∣a2/a1∣∼Δz/4Eso|a_2/a_1|\sim\Delta_z/4E_{\mathrm{so}}. Thus, the in-gap resonant state acquires a quasi-bound character in the direction perpendicular to DW and propagates along DW with the slight attenuation. Moreover, the state is highly spin-polarized along the exe_x axis. Similarly to the above analysis one can address the case of an easy-plane anisotropy. The in-plane magnetization does not open an energy gap in the Rashba spectrum, however, it displaces the crossing point position in momentum space from the Brillouin zone center, as seen in Figure 1.c. Therefore, for the sharp antiphase DW with the domain polarization directed normal to DW, M(x,y)=exMxsgn⁡(x)M(x,y)=e_xM_x\operatorname{sgn}(x) due to the mirror symmetry with respect to the (y,z)(y,z) plane, the projected 2D bands are given by the four gapless branches, E±(r)(kx=0,ky)=βky2±(αky+Δx)E_\pm^{(r)}(k_x=0,k_y)=\beta k_y^2\pm(\alpha k_y+\Delta_x) and E±(l)(kx=0,ky)=βky2±(αky−Δx)E_\pm^{(l)}(k_x=0,k_y)=\beta k_y^2\pm(\alpha k_y-\Delta_x) (where Δx=JMx\Delta_x=JM_x) which are derived from each domain in pairs. Note that these dependencies attain the minimum values at ∣k∣=kso|k|=k_{\mathrm{so}}, namely, E±(r)(kx=0,ky=±kso)=−Eso∓ΔxE_\pm^{(r)}(k_x=0,k_y=\pm k_{\mathrm{so}})=-E_{\mathrm{so}}\mp\Delta_x and E±(l)(kx=0,ky=∓kso)=−Eso∓ΔxE_\pm^{(l)}(k_x=0,k_y=\mp k_{\mathrm{so}})=-E_{\mathrm{so}}\mp\Delta_x. The characteristic feature of the spectrum projection is an energy-momentum window restricted by the four branches E±(r)(kx=0,ky)E_\pm^{(r)}(k_x=0,k_y) and E±(l)(kx=0,ky)E_\pm^{(l)}(k_x=0,k_y), which in the limit Δx/4Eso≪1\Delta_x/4E_{\mathrm{so}}\ll1 acquires the shape of an almost perfect rhombus. Solving the problem of a low-energy scattering of the Rashba electrons on the in-plane DW [’(∥)(\parallel)’], we find that, within the rhombus-like window, there is a 1D resonant state. Under the condition Δx/4Eso≪1\Delta_x/4E_{\mathrm{so}} \ll1, the dispersion relation for this state reads

ϵR(∥)(ky)=λ[ky2−(k(∥))2+ωR(∥)−iΓR(∥)],(5)\epsilon_R^{(\parallel)}(k_y)=\lambda\left[k_y^2-\left(k^{(\parallel)}\right)^2+\omega_R^{(\parallel)}-i\Gamma_R^{(\parallel)}\right], \tag*{(5)}

where ωR(∥)=Δx2/4Eso\omega_R^{(\parallel)}=\Delta_x^2/4E_{\mathrm{so}}, ΓR(∥)=Δx3/8Eso2\Gamma_R^{(\parallel)}=\Delta_x^3/8E_{\mathrm{so}}^2, k(∥)=Δx/αk^{(\parallel)}=\Delta_x/\alpha and λ≈4β(Δx/Eso)2\lambda\approx4\beta(\Delta_x/E_{\mathrm{so}})^2. Remarkably, the energy band ϵR(∥)(ky)\epsilon_R^{(\parallel)}(k_y) exists in a small portion of the 1D Brillouin zone, ∣ky∣<k(∥)\lvert k_y\rvert<k^{(\parallel)}, and has an effective mass mR∗m_R^* which greatly exceeds that of the 2D surface band, mR∗≈(Eso/Δx)2m∗m_R^*\approx(E_{\mathrm{so}}/\Delta_x)^2m^*. The width of such a flattened band is extremely small, ∼Δx4/Eso3\sim\Delta_x^4/E_{\mathrm{so}}^3, i.e., it is narrower than the resonance broadening ΓR(∥)\Gamma_R^{(\parallel)}. Thus, the heavy fermion emerging at the DW has the finite lifetime, in addition to the weak dispersion. Therefore, the state (5) will form a sharp peak in DOS near zero energy. The resonant state is highly spin-polarized along the ez\boldsymbol{e}_z direction, in other words, it is almost chiral. And the polarization sign depends on whether the in-plane DW has a head-to-head texture or a tail-to-tail one.

The in-plane DW also hosts the bound state. One can show that the antiphase DW, M(x,y)=exMxsgn⁡(x)\boldsymbol{M}(x,y)=\boldsymbol{e}_xM_x\operatorname{sgn}(x), creates the state exponentially localized at the interface. It has a parabolic-like spectrum with a negative effective mass, ϵB(∥)(ky)\epsilon_B^{(\parallel)}(k_y), which is symmetric with respect to ky=0k_y=0 and fixed within the interval −Eso−Δx<ϵB(∥)(ky)<2Δx−Eso−Δx2/4Eso-E_{\mathrm{so}}-\Delta_x<\epsilon_B^{(\parallel)}(k_y)<2\Delta_x-E_{\mathrm{so}}-\Delta_x^2/4E_{\mathrm{so}}.

The essential physics of the origin of the resonant states captured by our low-energy effective model is the following. Based on the standard Hamiltonian for magnetic semiconductor with strong Rashba-like SOC, we modify the exchange part in Eq. (1) by introducing a spatially varying magnetization, which gives a natural way to segment the surface into regions with distinct topological indicator values. Formally, when β→0\beta\to0 or m∗→∞m^*\to\infty, the Hamiltonian (1) is of the same form that describes pure Dirac fermions for the 2D TI, where the constant α\alpha represents their velocity [26]. In such a case, the 1D bound state with a linear spectrum spanning the exchange gap is known to be induced by the interface between two magnetic domains with opposite out-of-plane magnetization directions [23]. In turn, chiral symmetry of the linearized Hamiltonian 1 allows for the existence of a state with a flat band at zero energy at the in-plane DW [27]. Recently [10, 11], we have shown that a single antiphase DW on the surface of the magnetic TI with an easy-plane anisotropy hosts the chiral dispersionless state with a zero energy. Magnetic DWs with more complex textures also can create the bound states in TI [10, 11, 22]. In any case, such states (placed inside a global energy gap) are protected by the change in topological invariant across the magnetic DW. The similar effect can occur at the magnetic DW in the Rashba semiconductor. Indeed, under the condition Δx,z/4Eso≪1\Delta_{x,z}/4E_{\mathrm{so}}\ll1, the dominant contribution to the formation of the 1D resonant states at DW comes from the 2D Rashba states with small momenta near k=0\boldsymbol{k}=0, ∣ky∣≪kso\lvert k_y\rvert\ll k_{\mathrm{so}}, and our results reproduce those of the TI theory. However, the role of higher order terms in the momentum grows with an increase in the exchange field. Taking in account the quadratic term, k2k^2, in Eq. (1), the bound state defined in linearized approximation is transformed into the resonant (quasi-bound) state with the spectra (4) and (5), which acquires the small energy shift and broadening. In turn, the envelope function of the resonant state becomes rather strongly localized near DW within pre-asymptotic region. Under the weak out-of-plane exchange field, the existence of the in-gap linear-in-momentum state (4) is protected owing to almost quantized change of a topological invariant at the interface, C(r)−C(l)≈1−Δz/4EsoC^{(\mathrm{r})}-C^{(\mathrm{l})}\approx1-\Delta_z/4E_{\mathrm{so}}. The state (5) with flattened spectrum can be attributed to the approximate chiral symmetry of the model with the in-plane magnetic anisotropy. The effect of resonance broadening of the 1D in-gap state ϵR(⊥,∥)(ky)\epsilon_R^{(\perp,\parallel)}(k_y) is due to its coexistence with a continuous branch E−(k)E_-(\boldsymbol{k}) of the 2D Rashba states. As the exchange splitting increases, the energy shift ϵR(⊥,∥)\epsilon_R^{(\perp,\parallel)} increases and the resonance becomes broader, which implies that the state completely loses its meaning in the limit Δx,z/4Eso≫1\Delta_{x,z}/4E_{\mathrm{so}}\gg1.

Tight-binding simulations. Results.

To gain our insight into the electron states hosted at the magnetic DWs in Rashba semiconductor surface, we resort to numerical simulations based on a tight-binding approximation. This method allows us to reproduce in detail the features of these states for an arbitrary spatial profile of the DW texture in the whole region of the model parameters, which cannot be performed analytically and which is highly time-consuming within ab initio technique.

According to tight-binding formalism we perform a lattice regularization of the continuum Hamiltonian (1) via the substitution kx,y→1asin⁡(kx,ya)k_{x,y}\rightarrow\frac{1}{a}\sin(k_{x,y}a) and kx,y2→2a2[1−cos⁡(kx,ya)]k_{x,y}^2\rightarrow\frac{2}{a^2}[1-\cos(k_{x,y}a)], where aa is a square lattice constant and use recursive technique for the Green functions [28, 29]. Having obtained the retarded Green function G(x,x′,ky,E+iη)G(x,x',k_y,E+i\eta) for a certain magnetization distribution, we can calculate the corresponding one-particle spectral function A(ky,E)=−1πlim⁡η→0+Im⁡Tr⁡∫dxG(x,x,ky,E+iη)A(k_y,E)=-\frac{1}{\pi}\lim_{\eta\to0+}\operatorname{Im}\operatorname{Tr}\int dxG(x,x,k_y,E+i\eta) and spin polarized spectral function Si(ky,E)=−1πlim⁡η→0+Im⁡Tr⁡∫dxσiG(x,x,ky,E+iη)S_i(k_y,E)=-\frac{1}{\pi}\lim_{\eta\to0+}\operatorname{Im}\operatorname{Tr}\int dx\sigma_iG(x,x,k_y,E+i\eta). The total DOS and the spin density components in dependence on the state energy EE can be calculated from the expressions ρ(E)=2πa∫−π/aπ/adkyA(ky,E)\rho(E)=\frac{2\pi}{a}\int_{-\pi/a}^{\pi/a}dk_yA(k_y,E) and Si(E)=2πa∫−π/aπ/adkySi(ky,E)S_i(E)=\frac{2\pi}{a}\int_{-\pi/a}^{\pi/a}dk_yS_i(k_y,E), respectively. We assume periodic boundary conditions in the ey\boldsymbol{e}_y direction, along which the DW is running.

Out-of-plane DW structure

The results obtained from analysis of the spectral functions A(ky,E)A(k_y,E) and S(ky,E)S(k_y,E) in the case of the out-of-plane magnetic DW are demonstrated in Figure 2. As has been discussed above, in the magnetic Rashba semiconductor with perpendicular-to-surface uniaxial anisotropy the surface electron spectrum exhibits an opening of the exchange gap 2Δz2\Delta_z, which is clearly reflected in Figure 2a,b as well as in Fig. S1 of Supplemental Material for different values of Δz/4ESO\Delta_z/4E_{\rm SO}. Figure 2c,d exemplifies spin textures of opposite chirality for the projection of the inner and outer branches of the 2D spectrum, E+(ky)E_+(k_y) and E−(ky)E_-(k_y), respectively. In addition, in Figure 2e,f one can see the van Hove-like singularity in the vicinity of the conduction band bottom, ρ2D(⊥)(E)=a22πβ(1+Δz24ESO2+Δz2ESO)−1/2\rho_{2D}^{(\perp)}(E)=\frac{a^2}{2\pi\beta}\left(1+\frac{\Delta_z^2}{4E_{\rm SO}^2}+\frac{\Delta_z^2}{E_{\rm SO}}\right)^{-1/2}, which is typical for DOS of the Rashba system. As can be observed in Figure 2, regardless of the relation between the exchange splitting and the Rashba one, a separation of the surface into two domains with opposite magnetization directions, M(x,y)=ezMzsgn⁡(x)\mathbf{M}(x,y)=\mathbf{e}_zM_z\operatorname{sgn}(x), leads to an appearance of the 1D bound state at DW. The energy branch ϵB(∥)(ky)\epsilon_B^{(\parallel)}(k_y) defining this state, is split off from the band continuum E−(ky)E_-(k_y), it is well-defined and clearly distinguished in the spectral picture of Figure 2. Its dispersion is not symmetric with respect to ky=0k_y = 0, i.e., the largest energy separation is placed at a finite value of momentum kyk_y. With an increase in the ratio Δz/4Eso\Delta_z/4E_{\mathrm{so}}, the minimum of the dependence ϵB(⊥)(ky)\epsilon_B^{(\perp)}(k_y) becomes deeper and moves closer to the middle of the Brillouin zone (see Fig. S1). The spin texture of the bound state, shown in Figure 2.c,d under the relatively small exchange field, has a complicated composition in the momentum space: the positive (negative) in-plane spin polarization, SxB(⊥)(ky,ϵ)>0S_{xB}^{(\perp)}(k_y,\epsilon) > 0 (SxB(⊥)(ky,ϵ)<0S_{xB}^{(\perp)}(k_y,\epsilon) < 0), occurs mainly near ky≈−kso2−(k(⊥))2k_y \approx-\sqrt{k_{\mathrm{so}}^2-(k^{(\perp)})^2} (ky≈kso2−(k(⊥))2k_y \approx\sqrt{k_{\mathrm{so}}^2-(k^{(\perp)})^2}). This fact is manifested as a presence of two peaks with opposite sign in the spin density SxB(⊥)(ϵ)S_{xB}^{(\perp)}(\epsilon), which contribute to DOS of the bound state, ρB(⊥)(ϵ)\rho_B^{(\perp)}(\epsilon), below the 2D band edge [see Figure 2.e,f].

Spectral characteristics of the electron states at the surface of the magnetic Rashba semiconductor divided into two out-of-plane domains

Figure 2. Spectral characteristics of the electron states at the surface of the magnetic Rashba semiconductor divided into two out-of-plane domains. (a) and (d) Spectral function intensity. The projection of the 2D domain states is shown in azure-green color and the 1D states stemmed from DW are in cherry color. (b) and (e) Spin polarized spectral function intensity. The positive/negative spin polarization along the exe_x axis is represented in blue/red color. (c) and (f) Energy dependence of 2D DOS (blue curve), 1D DOS (green curve) and the Sx(E)S_x(E) component of spin density (red curve). The bound state ϵB(⊥)(ky)\epsilon_B^{(\perp)}(k_y) and resonant state ϵR(⊥)\epsilon_R^{(\perp)} are indexed in the figure. The upper panels is for Δz/4ESO=0.1\Delta_z/4E_{\rm SO}=0.1, the lower panels is for Δz/4ESO=0.4\Delta_z/4E_{\rm SO}=0.4.

The most remarkable feature in the weak exchange field regime, Δz/4Eso≪1\Delta_z/4E_{\mathrm{so}} \ll1, plotted in Figure 2.a,c, is the pronounced spectral branch connecting the 2D inner and outer bands through the gap. It shows the almost linear energy-momentum dependence within the local exchange gap and hardly visible broadening. Such a spectral feature is naturally associated with the in-gap resonant state induced by the out-of-plane magnetic DW, which is analytically described by Eq. (4) for the dispersion ϵR(⊥)(ky)\epsilon_R^{(\perp)}(k_y). However, with the increasing exchange gap size, this branch broadens substantially until its signature disappears at Δz≳Eso\Delta_z \gtrsim E_{\mathrm{so}} [see Figure 2.b,d and Fig. S1; note that spin resolution allows us to better see the resonance even at Δz/4Eso=0.4\Delta_z/4E_{\mathrm{so}} = 0.4]. Another important feature: the unidirectional resonant state ϵR(⊥)(ky)\epsilon_R^{(\perp)}(k_y) possesses a nearly perfect spin polarization SxR(⊥)(ky,ϵ)>0S_{xR}^{(\perp)}(k_y,\epsilon) > 0 [Figure 2.c-f]. In other words, it is a right-handed chiral state. Apparently, the out-of-plane DW with a negative magnetization gradient hosts a left-handed chiral state, because a change M(x,y)=ezMzsgn⁡(x)→M(x,y)=−ezMzsgn⁡(x)\mathbf{M}(x,y) = \mathbf{e}_z M_z \operatorname{sgn}(x) \to\mathbf{M}(x,y) = -\mathbf{e}_z M_z \operatorname{sgn}(x) leads to inversion of sign of both the electron velocity dϵR(⊥)(ky)/dky\mathrm{d}\epsilon_R^{(\perp)}(k_y)/\mathrm{d}k_y and spin polarization SxR(⊥)(ky,ϵ)S_{xR}^{(\perp)}(k_y,\epsilon).

The behavior of the resonant state under variation in the parameters of the Hamiltonian (1) may be related to its topological properties. In the case of the antiphase out-of-plane DW, it is natural to consider the difference between the Chern numbers of the right and left domains, C(r)C^{(r)} and C(l)C^{(l)}, respectively. The dependence C(r)−C(l)C^{(r)}-C^{(l)} as a function of the chemical potential position μ\mu is shown in Figure 3. One can see that this dependence changes markedly with variation in the ratio Δz/4Eso\Delta_z/4E_{\mathrm{so}}. When the chemical potential is fixed inside the exchange gap, ∣μ∣<Δz|\mu| < \Delta_z, one obtains:

Topological probe of the magnetic Rashba semiconductor surface

Figure 3. Topological probe of the magnetic Rashba semiconductor surface. (a) and (b) The difference between the Chern numbers of the right and left domains C(r)−C(l)C^{(r)}-C^{(l)} as a function of the chemical potential position μ\mu for two values of the ratio Δz/4ESO\Delta_z/4E_{\rm SO}, 0.1 and 0.4, respectively.

C(r)−C(l)=1−ΔzΔz2+Ω2(μ)(6)C^{(r)}-C^{(l)}=1-\frac{\Delta_z}{\sqrt{\Delta_z^2+\Omega^2(\mu)}} \tag*{(6)}
Ω2(μ)=8Eso2[1+μ2Eso+(1+μ2Eso)2+Δz2−μ24Eso2](7)\Omega^2(\mu)=8E_{\mathrm{so}}^2\left[1+\frac{\mu}{2E_{\mathrm{so}}}+\sqrt{\left(1+\frac{\mu}{2E_{\mathrm{so}}}\right)^2+\frac{\Delta_z^2-\mu^2}{4E_{\mathrm{so}}^2}}\right] \tag*{(7)}

The adjusted domains have contrasting topological properties in the reciprocal space which, in the weak exchange field regime, Δz/4Eso≪1\Delta_z/4E_{\mathrm{so}} \ll1, results in what the Chern number undergoes the change closer to an unit quantity. This change associated with the out-of-plane magnetic DW gives rise to the appearance of the in-local-gap resonant state. However, at large exchange field, namely, Δz/4Eso≳1\Delta_z/4E_{\mathrm{so}} \gtrsim1, the difference C(r)−C(l)C^{(r)}-C^{(l)} (Eq. (6)) becomes poorly quantized, as a consequence, the resonance is being eroded because its width significantly increases.

In-plane DW structure

The spectral function of the Rashba fermions coupled to the in-plane magnetic DW of a tail-to-tail type, M(x,y)=exMxsgn⁡(x)\mathbf{M}(x,y)=\mathbf{e}_xM_x\operatorname{sgn}(x), contributed from 2D and 1D surface electron states is presented in Figure 4. Figure S2, in addition to Figure 4, allows us to imagine the evolution of the system band structure with increasing the exchange field size Δx=JMx\Delta_x=JM_x. The 2D spectrum consists of four energy branches pairwise stemmed from the right and left domains. The corresponding projections are described by energy spectra E±(r)(kx=0,ky)=βky2±(αky+Δx)E_{\pm}^{(r)}(k_x=0,k_y)=\beta k_y^2\pm(\alpha k_y+\Delta_x) and E±(l)(kx=0,ky)=βky2±(αky−Δx)E_{\pm}^{(l)}(k_x=0,k_y)=\beta k_y^2\pm(\alpha k_y-\Delta_x) [Figure 4.a,b]. The images clearly show the bound state with the dispersion ϵB(∣∣)(ky)\epsilon_B^{(||)}(k_y) between the bands E−(∣∣)E_-^{(||)} and E−(∣∣)E_-^{(||)} in the vicinity of their bottom. The energy band ϵB(∣∣)(ky)\epsilon_B^{(||)}(k_y) is symmetric with respect to ky=0k_y=0 and has a negative effective mass. When the ratio Δx/4Eso\Delta_x/4E_{\mathrm{so}} increases, the maximum value ϵB(∣∣)(0)\epsilon_B^{(||)}(0) decreases. The spin texture of the bound state S(ky,ϵ)S(k_y,\epsilon) is depicted in Figure 4.c,d. Here, the in-plane component SxB(∣∣)(ky,ϵ)S_{xB}^{(||)}(k_y,\epsilon) is antisymmetric, hence SxB(∣∣)(ϵ)=0S_{xB}^{(||)}(\epsilon)=0. At the same time, the out-of-plane component SzB(∣∣)(ky,ϵ)S_{zB}^{(||)}(k_y,\epsilon) is purely negative and its value increases with rising Δx/4Eso\Delta_x/4E_{\mathrm{so}}. As seen in Figure 4.e,f, DOS associated with the bound state, ρB(∣∣)(ϵ)\rho_B^{(||)}(\epsilon), displays a pronounced peak at ϵ≈ϵB(∣∣)(0)\epsilon\approx\epsilon_B^{(||)}(0).

Spectral characteristics of electron states on the surface of a magnetic Rashba semiconductor divided into two in-plane domains

Figure 4. Spectral characteristics of the electron states on the surface of the magnetic Rashba semiconductor divided into two in-plane domains. (a) and (d) Spectral function intensity. The projection of the 2D domain states is shown in azure-green color and the 1D states stemmed from DW are in cherry color. (b) and (e) Spin polarized spectral function intensity. The positive/negative spin polarization along the exe_x axis is represented in blue/red color. (c) and (f) Energy dependence of 2D DOS (blue curve), 1D DOS (green curve) and Sz(E)S_z(E) component of spin density (red curve). The bound state ϵB(II)(ky)\epsilon^{(\mathrm{II})}_{B}(k_y) and resonant states ϵR(II)(ky)\epsilon^{(\mathrm{II})}_{R}(k_y) and ϵR(III)(ky)\epsilon^{(\mathrm{III})}_{R}(k_y) are indexed in the figure. The upper panels is for Δx/4ESO=0.1\Delta_x/4E_{\mathrm{SO}}=0.1, the lower panels is for Δx/4ESO=0.4\Delta_x/4E_{\mathrm{SO}}=0.4.

As mentioned earlier, in the case of the surface divided into two in-plane domains, the 2D spectrum exhibits the window in the shape of an improper rhombus around the origin (ky=0,E=0)(k_y=0,E=0) restricted by the curves E=E±(r)(kx=0,ky)E=E_{\pm}^{(r)}(k_x=0,k_y) and E=E±(l)(kx=0,ky)E=E_{\pm}^{(l)}(k_x=0,k_y) [see Figure 4.a,b]. The resonant state is found to occur inside the window. The corresponding weakly dispersing band ϵR(∣∣)(ky)\epsilon_R^{(||)}(k_y) bridges between the nodes with momenta (kx=0,ky=k(∣∣))(k_x=0,k_y=k^{(||)}) and (kx=0,ky=−k(∣∣))(k_x=0,k_y=-k^{(||)}). At the small ratio Δx/4Eso≪1\Delta_x/4E_{\mathrm{so}}\ll1, in Figure 4 one can observe the emergence of the almost flat band ϵR(∣∣)(ky)\epsilon_R^{(||)}(k_y) with a large positive effective mass and small broadening, which goes over the momentum range in the 1D Brillouin zone, ∣ky∣<k(∣∣)≪π/a|k_y|<k^{(||)}\ll\pi/a. Such features of the resonant state spectrum are consistent with the analytical estimations in Eq. (5). With increasing size Δx\Delta_x, the resonant band is up-shifting, its width, ϵR(∣∣)(k(∣∣))−ϵR(∣∣)(0)\epsilon_R^{(||)}(k^{(||)})-\epsilon_R^{(||)}(0), and existence interval, 2k(II)2k^{(\mathrm{II})}, increase as well as the broadening ΓR(II)\Gamma^{(\mathrm{II})}_{R}. When the exchange splitting becomes comparable to or bigger than the SOC splitting ESOE_{\mathrm{SO}}, the resonant state image blurs and then disappears [see Figure 4b,Figure 4d and Fig. S2]. Strikingly, Figure 4c,Figure 4d shows that the resonant state acquires predominantly spin polarization along the eze_z axis axis rather than in the plane, especially at small Δx/4ESO\Delta_x/4E_{\mathrm{SO}}. The polarization sign depends on whether the in-plane DW has a head-to-head configuration with SzR(II)(ky,ϵ)>0S^{(\mathrm{II})}_{zR}(k_y,\epsilon)>0 or a tail-to-tail one with SzR(II)(ky,ϵ)<0S^{(\mathrm{II})}_{zR}(k_y,\epsilon)<0. The numerical simulation catches one more resonant state with energy dispersion ϵR(III)(ky)\epsilon^{(\mathrm{III})}_{R}(k_y) located near the edges of the 2D branches E±(r)(kx=0,ky)E^{(\mathrm{r})}_{\pm}(k_x=0,k_y) and E±(l)(kx=0,ky)E^{(\mathrm{l})}_{\pm}(k_x=0,k_y) at ky≈±ksok_y\approx\pm k_{\mathrm{so}}, which can be seen at Δx/4ESO=0.1\Delta_x/4E_{\mathrm{SO}}=0.1. Albeit, at Δx/4ESO=0.4\Delta_x/4E_{\mathrm{SO}}=0.4 the state ϵR(III)(ky)\epsilon^{(\mathrm{III})}_{R}(k_y) merges with the state ϵR(II)(ky)\epsilon^{(\mathrm{II})}_{R}(k_y).

The above trends are corroborated by the behavior of the total DOS and the spin density. In the weak exchange field regime [Figure 4e], we observe three sharp peaks related to the 1D bound states and the two resonant ones induced by DW. With increasing the size Δx\Delta_x [Figure 4f], the bound state peak is well maintained, while the resonant states become smeared.

Figure 5 illustrates the spatial profile of the resonant state ϵR(L)(ky)\epsilon_R^{(L)}(k_y). Its probability density is composed of three components: one exponentially decays with distance from DW, ∼exp⁡(±k(∥)∣x∣)\sim\exp(\pm k^{(\parallel)}|x|), the other rapidly oscillates, ∼cos⁡(4ksox+2ϕ)\sim\cos(4k_{\mathrm{so}}x+2\phi), the third is the interference of the two formers, exp⁡(−k(∥)∣x∣)cos⁡(2ksox+ϕ)\exp(-k^{(\parallel)}|x|)\cos(2k_{\mathrm{so}}x+\phi), where k(∥)=Δx/α<ksok^{(\parallel)}=\Delta_x/\alpha<k_{\mathrm{so}}, ϕ\phi is a phase shift. At small Δx/4Eso\Delta_x/4E_{\mathrm{so}}, k(∥)k^{(\parallel)}, the probability density is strongly concentrated around DW on the scale ∣x∣≲[2k(∥)]−1|x|\lesssim[2k^{(\parallel)}]^{-1}, where the first component prevails. The increase of Δx/4Eso\Delta_x/4E_{\mathrm{so}} causes a drop in the amplitude and localization length of the first component [see Figure 5]. Thus, if the exchange splitting reaches comparably large values, the resonant state disappears in both momentum and real space.

Probability density for the resonant state

Figure 5. Probability density for the resonant state ϵR(II)(ky)\epsilon^{(\mathrm{II})}_{R}(k_y), which corresponds with the red violet points in Figure 4b,Figure 4e. The ratio Δx/4ESO\Delta_x/4E_{\mathrm{SO}} has been taken as 0.1 in green and 0.4 in blue.

Noncollinear DW structure

We now discuss the surface supporting two domains with different magnetization direction confined in the (x,z)(x,z) plane, which are separated by sharp DW at x=0x=0. It is convenient to express the magnetization vectors in spherical coordinates as M(r,l)=M0(sin⁡ϑ(r,l),0,cos⁡ϑ(r,l))\mathbf{M}^{(r,l)}=M_0(\sin\vartheta^{(r,l)},0,\cos\vartheta^{(r,l)}). Then the sought states generated by the noncollinear DW are specified by a pair of the angles ϑ(r)\vartheta^{(r)} and ϑ(l)\vartheta^{(l)} between the vectors M(r)\mathbf{M}^{(r)} and M(l)\mathbf{M}^{(l)} and the ez\mathbf{e}_z axis, respectively. In the weak exchange field regime, Δ0/4Eso=0.2\Delta_0/4E_{\mathrm{so}}=0.2 (Δ0=JM0\Delta_0=JM_0), Figure 6 depicts the redistribution of spectral weight which demonstrates how the prominent features of the energy structure of the 2D and 1D surface states are modified under reconfiguration of the DW texture. Evidently, the noncollinear DW [ϑ(r)≠ϑ(l)\vartheta^{(r)}\neq\vartheta^{(l)}] hosts always the bound state, the dispersion of which, ϵB(L)(ky)\epsilon_B^{(L)}(k_y), closely follows modifications in the dispersion of the 2D band near its bottom. In the case when the out-of-plane components of the domain magnetizations M(r)\mathbf{M}^{(r)} and M(l)\mathbf{M}^{(l)} have opposite signs, i.e., cos⁡ϑ(r)⋅cos⁡ϑ(l)<0\cos\vartheta^{(r)}\cdot\cos\vartheta^{(l)}<0, as is plotted in Figure 6a,Figure 6b, the in-gap topologically protected resonant state with the quasi-linear spectrum ϵR(L)(ky)\epsilon_R^{(L)}(k_y) is realized. The term ‘topologically protected resonant state’ is used here in the sense that has been described in Subsection III A. Further, passing through the phase with zero exchange splitting in the left domain, ϑ(l)=−π/2\vartheta^{(l)}=-\pi/2 [Figure 6c], the 1D state ϵR(L)(ky)\epsilon_R^{(L)}(k_y) is transformed into the trivial state, cos⁡ϑ(r)cos⁡ϑ(l)>0\cos\vartheta^{(r)}\cos\vartheta^{(l)}>0, which is illustrated in Figure 6d,Figure 6e. Note, the spectral branch of the trivial state does not span the gap but clearly originates from the DW. After the exchange gap in the left domain reverses its sign again, ϑ(l)=π/2\vartheta^{(l)}=\pi/2 [Figure 6g], the state ϵR(L)(ky)\epsilon_R^{(L)}(k_y) gets back to the topological character [Figure 6h]. It should be noted that the weakly damped fermion excitation at ky=0k_y=0 propagates along the noncollinear DW at the velocity α(∠)=αsin⁡(ϑ(r)+ϑ(l)2)\alpha^{(\angle)}=\alpha\sin\left(\frac{\vartheta^{(r)}+\vartheta^{(l)}}{2}\right). Depending on the angles, the velocity may changes not only in the value but also in the direction in the range of ∣α(∠)∣≤α|\alpha^{(\angle)}|\le\alpha.

Spectral function intensity of electron states at the surface of a magnetic Rashba semiconductor divided into two domains with noncollinear magnetizations

Figure 6. Spectral function intensity of the electron states at the surface of the magnetic Rashba semiconductor divided into two domains with noncollinear magnetizations M(r)\mathbf{M}^{(r)} and M(l)\mathbf{M}^{(l)} confined in the (x,z)(x,z) plane. It is possible to follow the evolution of the energy spectrum when the vector M(l)\mathbf{M}^{(l)} rotates while the direction of M(r)\mathbf{M}^{(r)} is fixed (ϑ(r)=π/4\vartheta^{(r)}=\pi/4). The projection of the 2D domain states is made in azure-green color and the 1D states stem from DW are in cherry color. The bound state ϵB(L)(ky)\epsilon_B^{(L)}(k_y) and the resonant states ϵR(L)(ky)\epsilon_R^{(L)}(k_y) are indexed in the figure. In the upper part of the panels, the red arrows indicate the magnetization directions in the right (blue ribbon) and left (green ribbon) domains. The ratio Δ0/4Eso\Delta_0/4E_{\mathrm{so}} used in these plots is 0.2.

Figure 6, together with Fig. S3, gives a hint at the possibility to control the electron properties by applying external magnetic field to the system under consideration. For example, in the case of the Rashba magnetic semiconductor with an easy plane anisotropy, the magnetization in the domains is tuned from the in-plane to out-of-plane direction under increasing external field, h=ezh\mathbf{h}=e_z h, perpendicular to the surface. At the same time, the almost flat band ϵR∥(ky)\epsilon_R^{\parallel}(k_y) and its accompanying sharp peak in DOS are shifted by the value Δ0cos⁡ϑ(r,l)\Delta_0\cos\vartheta^{(r,l)} with respect to the initial position near zero energy (at h=0\mathbf{h}=0) towards higher or lower energy depending on the direction of the field h\mathbf{h}. The magnetic configuration and the corresponding band structure at a value ∣h∣|\mathbf{h}|, exceeding the anisotropy field but not exceeding the saturation field, are given in Fig. 6.d. Similar reasoning can be used if external field aligned along the exe_x direction is applied to the material with the easy axis anisotropy. In this case, the typical magnetic configuration and related electron structure are exemplified in Fig. 6.h.

Discussion

The surface electron properties of the polar semiconductor BiTeI possessing the large Rashba splitting ESO≈100E_{\mathrm{SO}}\approx100 meV and doped with magnetic atoms were thoroughly studied in the works [18, 19] [20]. The doping of BiTeI with vanadium had been found to leads to the surface out-of-plane magnetic ordering and opening of the anomalously large exchange gap which, according to the ARPES measurements [18, 19] [20], reaches 2Δz=1252\Delta_z=125 meV at the Kramers point at the V-concentration of 2% and temperature 20 K. Interplay of the extraordinary large Rashba splitting and the sizable exchange gap would be ideally suited to satisfy the requirements for the realization and control of spin-polarized conductivity. However, in the Bi1−x_{1-x}Vx_xTeI material the situation has its own specifics. Initially, when no magnetic field is applied, the observed total out-of-plane spin polarization of the samples is turned out to be very weak which does not match such a large gap. The authors of Ref. 19 rationalized this seeming contradiction by assumption of the spontaneous formation of the surface ferromagnetic domains of almost equal area which are oppositely oriented below the Curie temperature Tc=130T_c=130 K. As was shown above, the out-of-plane DW induces the partially polarized bound state beyond the band continuum. What is even more interesting is that the chiral resonant state with Dirac-like dispersion exists inside the exchange gap. The quasi 1D unidirectional channel localized along DW can manifest itself, provided that the Rashba splitting energy exceeds the exchange one. This channel carries a weakly decaying fermion excitation, surviving for relatively long time τ(⊥)∼1/ΓR(⊥)\tau^{(\perp)}\sim1/\Gamma_R^{(\perp)}.

During this time, in a ballistic transport regime with the Fermi energy level being within the gap, ∣μ∣<Δz|\mu|<\Delta_z, the excitation passes the distance lR⊥=ατR(⊥)l_R^{\perp}=\alpha\tau_R^{(\perp)}. Authors of Ref. 19, studied the dependence of the surface exchange gap on temperature TT and concentration xx of magnetic impurities via the ARPES measurements in V- and Mn-doped BiTeI samples. For example, at x≈2%x\approx2\% and T=100T=100 K, they estimated gap as 2Δz≈802\Delta_z\approx80 meV. This means that Δz/4ESO≈0.1\Delta_z/4E_{\mathrm{SO}}\approx0.1, hence, according to the dispersion relation of Eq. (4), the broadening is about two orders of magnitude smaller than the gap, ΓR(⊥)≈2Δz⋅10−2\Gamma_R^{(\perp)}\approx2\Delta_z\cdot10^{-2}. Taking into account that BiTeI has Rashba parameter αR=3.8\alpha_R=3.8 eV·Å [30], the characteristic length lR(⊥)l_R^{(\perp)} can be roughly evaluated as large as 50 μ\mum. The latter even surpasses the inelastic scattering length observed in the TI thin film Cr-doped (Bi,Sb)2_2Te3_3 in the QAH regime at low temperatures [31]. Of course, the detailed analysis should include other dissipation mechanisms connected with the bulk bands and mid-gap impurity states as well as random magnetic multi-domain structure, which may reduce the length lR(⊥)l_R^{(\perp)}.

For semiconductors with the strong Rashba effect, experimental information about the surface properties modification triggered by DW still lacks so far, which would motivate experimentalists to perform relevant investigations. The developed theory may suggest adequate experimental approaches to test the signatures of the peculiar surface states in these materials. The combination of scanning tunneling microscopy and spectroscopy (STM/STS) and magnetic force microscopy (MFM) provides powerful tool for this purpose. In the virgin phase a sample possesses a multi-domain structure [17, 19]. MFM can select a single DW on the sample surface and identify its vector and spatial configuration, while STM/STS can map the electron density surrounding the DW through differential conductance measurements. We consider the surface of BiTeI doped with transition metal atoms as the object of such a study. When the Fermi level lies lower (higher) than or near to the 2D conduction band bottom (valence band top), STM/STS provides the signature of the bound state ϵB(⊥,∥)(ky)\epsilon_B^{(\perp,\parallel)}(k_y), which is shifted towards the bigger binding energy with increasing the exchange gap size 2Δz(T)2\Delta_z(T), i.e., with decreasing temperature TT. By tuning the Fermi level to the exchange gap, the tunneling conductivity can be locally enhanced in a narrow stripe stretching along DW if, as argued above, the evanescent component of the envelope function of the resonant state ϵR(⊥)(ky)\epsilon_R^{(\perp)}(k_y) dominates over the extended one. Moving away from the DW at the distance longer than about ∼[k(⊥)]−1\sim[k^{(\perp)}]^{-1}, the STM/STS signal significantly decays. This situation could be observed in the Bi1−x_{1-x}(V,Mn)x_xTeI samples under the optimal doping x≈2x\approx2–3% over the temperature region where the magnetic order exists. Indeed, in accordance with the dependence 2Δz(T)2\Delta_z(T) deduced in Ref. 20, the condition of the relatively weak exchange energy is satisfied at 0<T<Tc0 < T < T_c. Thereat, it is hard to detect a trace of the resonant state in close proximity to T≈TcT \approx T_c where the Fermi-Dirac distribution broadening surpasses the gap. Thus, there are reasons to regard the surface of diluted magnetic semiconductor on the base of BiTeI as a possible platform for realizing the quasi-robust chiral conducting channels at rather high temperatures, above ∼100\sim100 K.

In the system with out-of-plane anisotropy, both the 1D resonant states hosted at DWs and the lower subband of the 2D Rashba state simultaneously contribute to the surface electric transport when ∣μ∣<Δz|\mu| < \Delta_z. In this energy range, the presence of a magnetization has weak effect on the 2D subband E−(k)E_-(\mathbf{k}). The existence and characteristics of the resonant state are sensitive to the surface magnetic moments arrangement. In turn, the latter can be driven by an external magnetic field. Figure 6 and S3 demonstrates how the resonant state spectrum depends on the orientation of the domain magnetizations with respect to each other and to the surface normal. Furthermore, in a field-sweep experiments, during a magnetization reversal process under the field h=ezh\mathbf{h} = e_z h the concentration of DWs increases, reaching a maximum at the coercive field hch_c. Correspondingly, the concentration of the in-gap resonant states crossing the Fermi level increases. Since the magnetic domain scale is assumed to be much larger than the resonant state localization length ∼[k(⊥)]−1\sim[k^{(\perp)}]^{-1}, the sample surface can be considered as a network of the 1D percolating channels. Therefore, an enhancement of a longitudinal conductivity σxx(h)\sigma_{xx}(h) can be detected in the vicinity of the field HcH_c, provided the resonant state contribution dominates over the 2D subband one. Similar behavior of σxx(h)\sigma_{xx}(h) has been disclosed and substantiated in TI Bi2_2(Te,Se)3_3 doped with Mn [23]. In contrast, typically ferromagnets show a decrement of the longitudinal conductivity near the coercive field associated with the increased electron scattering at DWs [32]. Thereby, we predict that the Rashba magnetic semiconductor with an easy-axis anisotropy can manifest an peculiarity in magnetoresistance.

As for the resonant state with the flattened highly spin-polarized band ϵR(⊥)(ky)\epsilon_R^{(\perp)}(k_y) caused by the in-plane DW, it gives rise to a sharp DOS peak in the vicinity of E=0E = 0, which can be fairly easily identified by STM/STS. Albeit, whether such a singularity can enhance the inter-electron interactions to provoke instabilities at the DW (such as charge density wave or superconductivity) is an open question. To verify it one needs to fabricate the Rashba material possessing an easy plane magnetic anisotropy.

Conclusion

In summary, we present a theoretical investigation of electron states induced by an inhomogeneous magnetization at the surface of a semiconductor with the strong Rashba splitting. We predict that a single magnetic DW hosts not only the bound state but also, what is especially noteworthy, the 1D chiral resonant state. The properties of the resonant state depend profoundly on the characteristic parameters of the model and the magnetization orientation in adjacent domains. With decreasing ratio of the exchange energy to the Rashba splitting energy, the resonance broadening gets narrower and the state becomes more localized at DW. The dispersion relation ranges from a linear spectrum that tends to cross the local exchange gap to almost flat-band spanning the Kramers degeneration nodes when the domain magnetization varies from out-of-plane to in-plane. We substantiate the physical reason why the 1D chiral resonant state is guaranteed to appear at the antiphase DW. Our finding expands the realm of systems in which the convergence of magnetic order and SOC leads to the creation of peculiar electron states. The estimations demonstrate that the weakly damped resonant states can be materialized at the pristine cleaved surface of BiTeI doped with transition metal atoms and their signatures can be sought out in the magnetotransport and STM/STS studies. The unique surface electron states, whose possibility we predict in diluted magnetic semiconductors with strong Rashba splitting, being manipulated by both electrostatic and magnetic means open new opportunities for use in spintronic devices.

Acknowledgements

I.P.R acknowledges financial support from the Ministry of Education and Science of the Russian Federation within State Task No. FSWM-2020-0033 (in the part of tight-binding calculations). E.V.C acknowledges Saint-Petersburg State University for a research project 95442847.

References

  1. [1]S. S. P. Parkin, M. Hayashi, and L. Thomas, Science 320, 190 (2008).
  2. [2]S. Parkin and S.-H. Yang, Nature Nanotechnology 10, 195 (2015).
  3. [3]C.-Z. Chang, C.-X. Liu, and A. H. MacDonald, Rev. Mod. Phys. 95, 011002 (2023).
  4. [4]N. Varnava, J. H. Wilson, J. H. Pixley, and D. Vanderbilt, Nature Communications 12, 3998 (2021).
  5. [5]Y.-F. Zhao, R. Zhang, J. Cai, D. Zhuo, L.-J. Zhou, Z.-J. Yan, M. H. W. Chan, X. Xu, and C.-Z. Chang, Nature Communications 14, 770 (2023).
  6. [6]K. Yasuda, M. Mogi, R. Yoshimi, A. Tsukazaki, K. S. Takahashi, M. Kawasaki, F. Kagawa, and Y. Tokura, Science 358, 1311 (2017).
  7. [7]V. N. Men’shov and E. V. Chulkov, JETP Letters 117, 147 (2023).
  8. [8]V. N. Men’shov, I. A. Shvets, and E. V. Chulkov, JETP Letters 110, 771 (2019).
  9. [9]V. N. Men’shov, I. A. Shvets, and E. V. Chulkov, Phys. Rev. B 106, 205301 (2022).
  10. [10]I. P. Rusinov, V. N. Men’shov, and E. V. Chulkov, Phys. Rev. B 104, 035411 (2021).
  11. [11]V. N. Men’shov, I. P. Rusinov, and E. V. Chulkov, JETP Letters 114, 699 (2021).
  12. [12]V. N. Men’shov, I. A. Shvets, and E. V. Chulkov, Phys. Rev. B 99, 115301 (2019).
  13. [13]Y. A. Bychkov and E. I. Rashba, Journal of Experimental and Theoretical Physics Letters 39, 66 (1984).
  14. [14]A. Manchon, H. C. Koo, J. Nitta, S. M. Frolov, and R. A. Duine, Nature Materials 14, 871 (2015).
  15. [15]G. Bihlmayer, P. Noël, D. V. Vyalikh, E. V. Chulkov, and A. Manchon, Nature Reviews Physics 4, 642 (2022).
  16. [16]J. Krempaský, S. Muff, F. Bisti, M. Fanciulli, H. Volfová, A. P. Weber, N. Pilet, P. Warnicke, H. Ebert, J. Braun, F. Bertran, V. V. Volobuev, J. Minár, G. Springholz, J. H. Dil, and V. N. Strocov, Nature Communications 7, 13071 (2016).
  17. [17]J. Krempaský, S. Muff, J. Minár, N. Pilet, M. Fanciulli, A. P. Weber, E. B. Guedes, M. Caputo, E. Müller, V. V. Volobuev, M. Gmitra, C. A. F. Vaz, V. Scagnoli, G. Springholz, and J. H. Dil, Phys. Rev. X 8, 021067 (2018).
  18. [18]I. I. Klimovskikh, A. M. Shikin, M. M. Otrokov, A. Ernst, I. P. Rusinov, O. E. Tereshchenko, V. A. Golyashov, J. Sánchez-Barriga, A. Y. Varykhalov, O. Rader, K. A. Kokh, and E. V. Chulkov, Scientific Reports 7, 3353 (2017).
  19. [19]A. M. Shikin, A. A. Rybkina, I. I. Klimovskikh, O. E. Tereshchenko, A. S. Bogomyakov, K. A. Kokh, A. Kimura, P. N. Skirdkov, K. A. Zvezdin, and A. K. Zvezdin, 2D Materials 4, 025055 (2017).
  20. [21]R. Yoshimi, K. Yasuda, A. Tsukazaki, K. S. Takahashi, M. Kawasaki, and Y. Tokura, Science Advances 4, ea at9989 (2018).
  21. [22]E. K. Petrov, V. N. Men’shov, I. P. Rusinov, M. Hoffmann, A. Ernst, M. M. Otrokov, V. K. Dugaev, T. V. Menshchikova, and E. V. Chulkov, Phys. Rev. B 103, 235142 (2021).
  22. [23]J. G. Checkelsky, J. Ye, Y. Onose, Y. Iwasa, and Y. Tokura, Nature Physics 8, 729 (2012).
  23. [24]G. Bihlmayer, O. Rader, and R. Winkler, New Journal of Physics 17, 050202 (2015).
  24. [25]F. Ronetti, K. Plekhanov, D. Loss, and J. Klinovaja, Phys. Rev. Res. 2, 022052 (2020).
  25. [26]M. Z. Hasan and C. L. Kane, Reviews of Modern Physics 82, 3045 (2010).
  26. [27]C.-K. Chiu, J. C. Y. Teo, A. P. Schnyder, and S. Ryu, Rev. Mod. Phys. 88, 035005 (2016).
  27. [28]M. P. L. Sancho, J. M. L. Sancho, J. M. L. Sancho, and J. Rubio, Journal of Physics F: Metal Physics 15, 851 (1985).
  28. [29]J. Henk and W. Schattke, Computer Physics Communications 77, 69 (1993).
  29. [30]K. Ishizaka, M. S. Bahramy, H. Murakawa, M. Sakano, T. Shimojima, T. Sonobe, K. Koizumi, S. Shin, H. Miyahara, A. Kimura, K. Miyamoto, T. Okuda, H. Namatame, M. Taniguchi, R. Arita, N. Nagaosa, K. Kobayashi, Y. Murakami, R. Kumai, Y. Kaneko, Y. Onose, and Y. Tokura, Nature Materials 10, 521 (2011).
  30. [31]P. Deng, C. Eckberg, P. Zhang, G. Qiu, E. Emmanouilidou, G. Yin, S. K. Chong, L. Tai, N. Ni, and K. L. Wang, Nature Communications 13, 4246 (2022).
  31. [32]P. M. Levy and S. Zhang, Phys. Rev. Lett. 79, 5110 (1997).

Paper details

vlnc:W_ftbcmr7b
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents