betaFeedback

WorksPhysics and Astronomy

Spin swapping mediated by spin-orbital textures in centrosymmetric multiorbital materials

Abstract

Spin swapping converts a primary spin current into a secondary one by interchanging its flow and spin-polarization directions, attributed to spin-orbit scattering or spin-band splitting. Here we uncover a fundamentally distinct route to spin swapping arising from the orbital degree of freedom in the spin pumping to centrosymmetric multiorbital materials without spin-band splitting. When the dynamical proximity ferromagnet injects a longitudinal spin current flowing normal to the interface, spin-orbit-entangled orbital textures redirect the injected spins into longitudinal and transverse orbital currents. Their subsequent conversion back to the spin sector generates transverse spin currents with propagation and polarization directions swapped relative to the primary spin current. The swapping efficiency achieves 50%50\% for Ge and Si. Governed by a fourth-rank tensor constrained by crystalline symmetry and without requiring spin-orbit scattering or inversion symmetry breaking, our finding reveals spin-orbital textures as hidden intermediate channels of spin swapping with unexplored material choices and functionalities.

Introduction

Spintronics requires spin currents with diverse spin polarizations. As a useful functional transport phenomenon for this purpose, spin swapping converts a primary spin current into a secondary one by interchanging the directions of spin flow and spin polarization [1]. Two scenarios are possible: for a primary spin current JS,αJ^{\mathrm{S},\alpha} flowing along jj with spin polarization along α≠j\alpha\ne j, spin swapping generates a secondary component JS,jαJ^{\alpha}_{\mathrm{S},j}, while with α=j\alpha=j the primary spin current JS,ααJ^{\alpha}_{\mathrm{S},\alpha} generates JS,ββJ^{\beta}_{\mathrm{S},\beta} (β≠α\beta\ne\alpha) [1]. Originally proposed as an extrinsic effect arising from spin-orbital impurity scattering [1, 2], it has since been predicted in disordered conductors [3], superconductors [4, 5], magnetic bilayers [6], and ferromagnets [7]. Beyond this extrinsic mechanism, spin swapping can also arise intrinsically from band spin splitting by spin-orbit coupling (SOC) in bulk and at interfaces [8, 9, 10]. Observed in Rashba-interface heterostructures and ferromagnet/nonmagnet bilayers [11, 12, 13, 14, 15, 16, 17], spin swapping has been harnessed to generate nontrivial (out-of-plane) spin polarizations for spin-orbital torque and anisotropic information transport. Recently, spin swapping in ferromagnetic metals was predicted to be mediated by orbital textures [18], but in this mechanism the band exchange spin splitting is essential, which is therefore absent in broader nonmagnetic materials.

Spin pumping provides an efficient and general route to driving the primary spin current [19, 20, 21]: coherent magnetization precession driven by ferromagnetic resonance (FMR) in a proximity ferromagnet injects into the adjacent conductor a spin current that flows predominantly normal to the magnetic interface, with its spin polarization aligned with the saturation magnetization Figure 1(a). Within this conventional picture, spin pumping alone does not generate spin-swapping transport. Such descriptions, however, formulate angular-momentum transfer primarily in the spin sector and do not resolve the spin-orbital textures carried by multiorbital Bloch states in conductors; they thus leave open broader, yet unexplored, spin and orbital dynamics [22, 23, 24, 25, 26] in materials with intrinsic spin-orbital textures, even when inversion and time-reversal symmetries are preserved. Here we unify these two spin phenomena, i.e., spin pumping and spin swapping, within a single mechanism.

Spin and orbital swapping in a multiorbital material pumped by a precessing magnetization

Figure 1. Spin and orbital swapping in a multiorbital material pumped by a precessing magnetization M(t)\mathbf{M}(t). (a) plots a pumped conventional spin current JSyzJ^{z}_{S_y} into conductors, polarized along the saturation magnetization direction y^\hat{\mathbf{y}} and flowing normal to the interface. (b): The intrinsic orbital texture induces the spin texture by SOC, giving rise to a spin-swapping current JSzyJ^{y}_{S_z}, polarized along z^\hat{\mathbf{z}} and flowing along y^\hat{\mathbf{y}}. Meanwhile, a longitudinal and transverse orbital currents JLyzJ^{z}_{L_y} and JLzyJ^{y}_{L_z} are generated.

In this work, we predict spin and orbital swapping in centrosymmetric multiorbital materials in the pumping configuration and trace them to the locked spin and orbital textures. Orbital textures are ubiquitous in multiorbital materials owing to crystal-field effects and momentum-dependent orbital hybridization [27, 28, 29, 30, 31]. Through the atomic spin-orbit interaction, these orbital textures give rise to momentum-dependent spin textures nearly aligned with their orbital counterparts in centrosymmetric crystals. This enables spin pumping to generate additional swapped spin and orbital currents, as illustrated in Figure 1(b). For a saturation magnetization parallel to the interface (Ms∥y^\mathbf{M}_{\mathrm{s}} \parallel\hat{\mathbf{y}}), the conventional spin current JSyzJ^{z}_{S_y} is accompanied by a swapped current JSzyJ^{y}_{S_z}, whereas for a saturation magnetization normal to the interface (Ms∥z^\mathbf{M}_{\mathrm{s}} \parallel\hat{\mathbf{z}}), the conventional spin current JSzzJ^{z}_{S_z} additionally generates JSxxJ^{x}_{S_x} and JSyyJ^{y}_{S_y}. Beyond swapped spin currents, this effect also generates swapped orbital currents by a mechanism distinct from orbital pumping [22– [26]. We interpret these phenomena as the interconversion of spin and orbital currents mediated by spin-orbital textures and governed by fourth-rank polar tensors from Neumann’s principle. Such swapping responses are universally allowed by crystalline symmetry: performing the symmetry analysis for group-IV semiconductors (Ge and Si) and beyond (e.g., GaAs, GaN, CuInSe2_2, and SnSe) establishes the general symmetry-allowed nature of spin swapping.

General formalism

Here we illustrate the general principle of the spin and orbital swapping in the pumping configuration. To this end, we consider a ferromagnet|multiorbital-material heterostructure as in Figure 1 to show how the FMR in the ferromagnet pumps swapped spin and orbital currents into the adjacent multiorbital materials. We consider a multiband Hamiltonian in centrosymmetric materials described by the matrix Hf(k)\mathcal{H}_{f}(\mathbf{k}) under the orbit-spin product basis ∣O∈{O1,O2,O3,⋯ }⟩⊗∣s∈{↑,↓}⟩\lvert O \in\{O_{1},O_{2},O_{3},\cdots\}\rangle\otimes\lvert s \in\{\uparrow,\downarrow\}\rangle. The eigenstates φn,k\varphi_{n,\mathbf{k}} with the dispersion εn(k)\varepsilon_{n}(\mathbf{k}) for band nn obey Hf(k)φn,k=εn(k)φn,k\mathcal{H}_{f}(\mathbf{k})\varphi_{n,\mathbf{k}}=\varepsilon_{n}(\mathbf{k})\varphi_{n,\mathbf{k}}, viz., the unitary transformation Uf(k)=(φ1,k,φ2,k,⋯ )U_{f}(\mathbf{k})=(\varphi_{1,\mathbf{k}},\varphi_{2,\mathbf{k}},\cdots) diagonalizes the Hamiltonian matrix according to Uf†(k)Hf(k)Uf(k)=diag⁡(ε1(k),ε2(k),⋯ )U_{f}^{\dagger}(\mathbf{k})\mathcal{H}_{f}(\mathbf{k})U_{f}(\mathbf{k})=\operatorname{diag}(\varepsilon_{1}(\mathbf{k}),\varepsilon_{2}(\mathbf{k}),\cdots). We shall take the Luttinger Hamiltonian [32, 33] in the valence band as a prototypical example, so εnh(k)=−εn(−k)\varepsilon^{h}_{n}(\mathbf{k})=-\varepsilon_{n}(-\mathbf{k}) denotes the dispersion of holes in band nn with the wave vector k\mathbf{k}, which are all doubly degenerate due to the Kramers degeneracy. Our theory can be routinely extended to the conduction band by particle-hole transmutation. With h^n(k)\hat{h}_{n}(\mathbf{k}) denoting the annihilation operator of holes in band nn, the free Hamiltonian of holes read

H^f=∑n∑kεnh(k)h^n†(k)h^n(k).(1)\hat{H}_{f}=\sum_{n}\sum_{\mathbf{k}}\varepsilon^{h}_{n}(\mathbf{k})\hat{h}^{\dagger}_{n}(\mathbf{k})\hat{h}_{n}(\mathbf{k}). \tag*{(1)}

To find the spin and orbital textures under the eigenstates in the underlying multiorbital materials, we now define the matrix elements of the spin and orbital operators. The matrices for the spin operator under the orbit-spin product basis ∣On⟩⊗∣sa⟩\lvert O_{n}\rangle\otimes\lvert s_{a}\rangle are diagonal in the orbital indices and independent of wave vectors, which, in terms of the Pauli matrices σ\boldsymbol{\sigma}, are expressed by

Σ∣On⊗sa⟩∣Om⊗sb⟩=δnmσsasb.(2)\boldsymbol{\Sigma}_{\lvert O_{n}\otimes s_{a}\rangle\lvert O_{m}\otimes s_{b}\rangle}=\delta_{nm}\boldsymbol{\sigma}_{s_{a}s_{b}}. \tag*{(2)}

In the spin operator S^=(1/V)∑n,m∑kSk∣nm⟩hh^n,k†h^m,k\hat{\mathbf{S}}=(1/V)\sum_{n,m}\sum_{\mathbf{k}}\mathbf{S}^{h}_{\mathbf{k}\lvert nm\rangle}\hat{h}^{\dagger}_{n,\mathbf{k}}\hat{h}_{m,\mathbf{k}} with crystal volume VV,

Skh=−(Uf†(−k)ΣUf(−k))T/2(3)\mathbf{S}^{h}_{\mathbf{k}}=-\left(U_{f}^{\dagger}(-\mathbf{k})\boldsymbol{\Sigma}U_{f}(-\mathbf{k})\right)^{T}/2 \tag*{(3)}

denotes the spin matrix of holes in the eigenstate basis. The orbital matrix with elements Lmn\mathcal{L}_{mn} in the orbit-spin product basis ∣On⟩⊗∣sa⟩\lvert O_{n}\rangle\otimes\lvert s_{a}\rangle around the high-symmetry point k0\mathbf{k}_{0} is derived in terms of the k⋅p\mathbf{k}\cdot\mathbf{p} theory in the Supplementary Information Sec. I [34]. We explicitly calculate the orbital matrices of pp-orbital electrons in the Luttinger model and show that the SOC renormalizes Lmn\mathcal{L}_{mn} away from the atomic orbital angular momentum (OAM) used in previous calculations [35, 36] (see expressions and derivations in Supplementary Information Sec. II [34]). The spatially-averaged OAM operator L^=(1/V)∑n,m∑kLk∣nm⟩hh^n,k†h^m,k\hat{\mathbf{L}}=(1/V)\sum_{n,m}\sum_{\mathbf{k}}\mathbf{L}^{h}_{\mathbf{k}\lvert nm\rangle}\hat{h}^{\dagger}_{n,\mathbf{k}}\hat{h}_{m,\mathbf{k}}, in which

Lkh=−(Uf†(−k)LUf(−k))T(4)\mathbf{L}^{h}_{\mathbf{k}}=-\left(U_{f}^{\dagger}(-\mathbf{k})\mathcal{L}U_{f}(-\mathbf{k})\right)^{T} \tag*{(4)}

represents the OAM matrix of holes in the eigenstate basis. The spin matrix Skh\mathbf{S}^{h}_{\mathbf{k}} and OAM matrix Lkh\mathbf{L}^{h}_{\mathbf{k}} describe the spin and orbital structures, in which the diagonal elements denote the spin and orbital angular momentum carried by holes of associated bands, while the off-diagonal terms are correlations between different bands; their dependence on k\mathbf{k} defines the spin and orbital textures.

Combined with the velocity, these textures can further lead to spin/orbital-current textures. The spin-current operator is J^S,αi=(1/V)∑n,m∑kJS,αh,i(k)∣nm⟩h^n,k†h^m,k\hat{J}^{i}_{S,\alpha}=(1/V)\sum_{n,m}\sum_{\mathbf{k}}\mathcal{J}^{h,i}_{S,\alpha}(\mathbf{k})_{\lvert nm\rangle}\hat{h}^{\dagger}_{n,\mathbf{k}}\hat{h}_{m,\mathbf{k}}, in which the spin-current textures read

JS,αh,i(k)=−14[Uf†(−k){Σα,Vi(−k)}Uf(−k)]T.(5)\mathcal{J}^{h,i}_{S,\alpha}(\mathbf{k})=-\frac{1}{4}\left[U_{f}^{\dagger}(-\mathbf{k})\left\{\boldsymbol{\Sigma}_{\alpha},\mathcal{V}_{i}(-\mathbf{k})\right\}U_{f}(-\mathbf{k})\right]^{T}. \tag*{(5)}

Here, V(k)=(1/ℏ)∂Hf(k)/∂k\mathcal{V}(\mathbf{k})=(1/\hbar)\partial\mathcal{H}_{f}(\mathbf{k})/\partial\mathbf{k} denotes the velocity matrix. On the other hand, the spatially-averaged orbital-current operator read J^L,αi=(1/V)∑n,m∑kJL,αh,i(k)∣nm⟩h^n,k†h^m,k\hat{J}^{i}_{L,\alpha}=(1/V)\sum_{n,m}\sum_{\mathbf{k}}\mathcal{J}^{h,i}_{L,\alpha}(\mathbf{k})_{\lvert nm\rangle}\hat{h}^{\dagger}_{n,\mathbf{k}}\hat{h}_{m,\mathbf{k}}, in which

JL,αh,i(k)=−12[Uf†(−k){Lα,Vi(−k)}Uf(−k)]T.(6)\mathcal{J}^{h,i}_{L,\alpha}(\mathbf{k})=-\frac{1}{2}\left[U_{f}^{\dagger}(-\mathbf{k})\left\{\mathcal{L}_{\alpha},\mathcal{V}_{i}(-\mathbf{k})\right\}U_{f}(-\mathbf{k})\right]^{T}. \tag*{(6)}

defines the orbital-current textures. These textures are crucial for the generation of spin and orbital swapping.

At the interface located at the xx-yy plane with z=0z=0 between the ferromagnet and the multiorbital materials, the spin degree of freedom of holes interacts with the dynamical magnetization via the interfacial exchange interaction. We focus on the driving force by the FMR in the microwave-driven ferromagnet. That is, the dynamical magnetization M(t)=∑ζ=±M(ζ)e−iζωt\mathbf{M}(t)=\sum_{\zeta=\pm}\mathbf{M}^{(\zeta)}e^{-i\zeta\omega t} precessing with frequency ω\omega is uniform in the ferromagnet region (z<0z<0), which is real by M(+)=M(−)∗\mathbf{M}^{(+)}=\mathbf{M}^{(-)*}. In this scenario, the system exhibits translational invariance in the x^\hat{\mathbf{x}} and y^\hat{\mathbf{y}} directions, so the parallel component of the wave vector κ=kxx^+kyy^\boldsymbol{\kappa}=k_x\hat{\mathbf{x}}+k_y\hat{\mathbf{y}} is conserved. The coupling between the holes and magnetization due to the interfacial exchange interaction JexJ_{\mathrm{ex}} then reads

H^int=Jex∫Vdrs^(r)⋅M(t)δ(z=0)=∑ζ=±∑m,n∑κ,kz,kz′Gkz′kz(ζ)(κ)∣nm⟩e−iζωth^n,kz′†(κ)h^m,kz(κ).(7)\hat{H}_{\mathrm{int}}=J_{\mathrm{ex}}\int_Vd\mathbf{r}\hat{\mathbf{s}}(\mathbf{r})\cdot\mathbf{M}(t)\delta(z=0) =\sum_{\zeta=\pm}\sum_{m,n}\sum_{\mathbf{\kappa},k_z,k'_z}G_{k'_z k_z}^{(\zeta)}(\mathbf{\kappa})\lvert n m\rangle e^{-i\zeta\omega t}\hat{h}_{n,k'_z}^{\dagger}(\mathbf{\kappa})\hat{h}_{m,k_z}(\mathbf{\kappa}). \tag*{(7)}

The dynamical magnetization then scatters the holes between wave-vector components kzk_z and kz′k'_z and across different valence bands mm and nn, governed by the amplitude

Gkz′kz(ζ)(κ)=−ℏJex2Lz(Uf†(−κ,−kz)ΣUf(−κ,−kz′))T⋅M(ζ),G_{k'_z k_z}^{(\zeta)}(\mathbf{\kappa})=-\frac{\hbar J_{\mathrm{ex}}}{2L_z}\left(U_f^{\dagger}(-\mathbf{\kappa},-k_z)\boldsymbol{\Sigma}U_f(-\mathbf{\kappa},-k'_z)\right)^{T}\cdot\mathbf{M}^{(\zeta)},

in which LzL_z is the thickness of the multiorbital materials along interface normal z^\hat{\mathbf{z}}-direction. When kz=kz′k_z=k'_z, the magnetization couples directly to the spin textures, but generally it induces more dynamics involving kz≠kz′k_z\ne k'_z.

In terms of the typical scattering process illustrated in Figure 2, it is then expected that the hole correlation ⟨h^n,kz′†(κ)h^m,kz(κ)⟩\langle\hat{h}_{n,k'_z}^{\dagger}(\mathbf{\kappa})\hat{h}_{m,k_z}(\mathbf{\kappa})\rangle between the wave vectors {kz,kz′}\{k_z,k'_z\} and the bands {m,n}\{m,n\} can be induced in the scattering process: the hole with wave vector kzk_z in the band mm first absorbs (ζ=−\zeta=-) or emits (ζ=+\zeta=+) a Kittel magnon of frequency ω\omega and transfers to the intermediate state with wave vector qzq_z in the band m′m' with energy εm′h(κ,qz)=εmh(κ,kz)−ζℏω\varepsilon_{m'}^{h}(\mathbf{\kappa},q_z)=\varepsilon_m^{h}(\mathbf{\kappa},k_z)-\zeta\hbar\omega, which then emits or absorbs the Kittel magnon and becomes the final state with wave vector kz′k'_z in band nn, governed by the energy conservation εm′h(κ,qz)=εnh(κ,kz′)−ζℏω\varepsilon_{m'}^{h}(\mathbf{\kappa},q_z)=\varepsilon_n^{h}(\mathbf{\kappa},k'_z)-\zeta\hbar\omega. For the pumped waves propagating to the positive half space z>0z>0, we need to concern the wave vectors {kz,kz′}\{k_z,k'_z\} with positive group velocities, viz., kz=qz,mm′(ζ)k_z=q_{z,mm'}^{(\zeta)} and kz′=qz,nm′(ζ)k'_z=q_{z,nm'}^{(\zeta)} are roots of εmh(κ,qz,mm′(ζ))=εnh(κ,qz,nm′(ζ))=εm′h(κ,qz)+ζℏω\varepsilon_m^{h}(\mathbf{\kappa},q_{z,mm'}^{(\zeta)})=\varepsilon_n^{h}(\mathbf{\kappa},q_{z,nm'}^{(\zeta)})=\varepsilon_{m'}^{h}(\mathbf{\kappa},q_z)+\zeta\hbar\omega with positive group velocities ∂kzεm,kzh(κ)∣kz=qz,mm′(ζ)>0\left.\partial_{k_z}\varepsilon_{m,k_z}^{h}(\mathbf{\kappa})\right|_{k_z=q_{z,mm'}^{(\zeta)}}>0 and ∂kzεn,kzh(κ)∣kz=qz,nm′(ζ)>0\left.\partial_{k_z}\varepsilon_{n,k_z}^{h}(\mathbf{\kappa})\right|_{k_z=q_{z,nm'}^{(\zeta)}}>0. The quantum coherence between the hole initial state ∣m,qz,mm′(ζ)⟩\lvert m,q_{z,mm'}^{(\zeta)}\rangle and final state ∣n,qz,nm′(ζ)⟩\lvert n,q_{z,nm'}^{(\zeta)}\rangle induced by magnon-assisted scattering channels gives rise to the pumped spin and orbital currents. Particularly, for spatially uniform currents, only contributions with qz,mm′(ζ)=qz,nm′(ζ)q_{z,mm'}^{(\zeta)}=q_{z,nm'}^{(\zeta)} survive, which, together with the energy conservation, requires the bands mm and nn in the established coherence to be degenerate.

Figure 2. Illustration of two typical hole-magnon scattering processes induced by the interfacial ss-dd exchange interaction. The coherent magnons induce the hole coherence between different wave vectors and bands.

We substantiate the above expectation by setting up the quantum kinetic equation and solving the driven density matrix of holes (see details in the Supplementary Information Sec. III [34]), which accounts microscopically for the spin and orbital dynamics of holes when driven by the dynamical exchange magnetic field JexM(t)J_{\mathrm{ex}}\mathbf{M}(t). Accordingly, we obtain the DC spin “SS” and orbital “LL” currents at z>0z>0 pumped by the FMR of the proximity ferromagnet

JS/L,αi=ℏ2Jex24(2π)3∑ζ=±∑{m,n}∑m′∫dk dqz(φm′,−qz∗(−κ)Σ⋅M(ζ)φm,−qz(−κ))(φn,−qz∗(−κ)Σ⋅M(ζ)∗φm′,−qz(−κ))×JS/L,αh,i(q~z,κ)∣nmf(εm′,qzh(κ))−f(εm′,qzh(κ)+ζℏω)(∂q~zεm,q~zh(κ))2.(8)J_{S/L,\alpha}^{i}=\frac{\hbar^{2}J_{\mathrm{ex}}^{2}}{4(2\pi)^{3}}\sum_{\zeta=\pm}\sum_{\{m,n\}}\sum_{m'}\int dk\,dq_z\left(\varphi_{m',-q_z}^{*}(-\mathbf{\kappa})\boldsymbol{\Sigma}\cdot\mathbf{M}^{(\zeta)}\varphi_{m,-q_z}(-\mathbf{\kappa})\right)\left(\varphi_{n,-q_z}^{*}(-\mathbf{\kappa})\boldsymbol{\Sigma}\cdot\mathbf{M}^{(\zeta)*}\varphi_{m',-q_z}(-\mathbf{\kappa})\right) \times\mathcal{J}_{S/L,\alpha}^{h,i}(\tilde{q}_z,\mathbf{\kappa})\bigg|_{nm}\frac{f\left(\varepsilon_{m',q_z}^{h}(\mathbf{\kappa})\right)-f\left(\varepsilon_{m',q_z}^{h}(\mathbf{\kappa})+\zeta\hbar\omega\right)}{\left(\partial_{\tilde{q}_z}\varepsilon_{m,\tilde{q}_z}^{h}(\mathbf{\kappa})\right)^2}. \tag*{(8)}

Here, the summation over the intermediate hole states {qz,m′}\{q_z,m'\} in Figure 2 is implied; the braces in {m,n}\{m,n\} indicate they are degenerate-band pairs, the wave vector q~z=qz,mm′(ζ)=qz,nm′(ζ)\tilde{q}_z=q_{z,mm'}^{(\zeta)}=q_{z,nm'}^{(\zeta)} indicates the energy conservation in the scattering process and positive group velocities in the transport, and f(εm,kh)=1/(exp⁡((εm,kh−μ)/(kBT))+1)f(\varepsilon_{m,\mathbf{k}}^{h})=1/(\exp((\varepsilon_{m,\mathbf{k}}^{h}-\mu)/(k_{\mathrm{B}}T))+1) is the equilibrium Fermi-Dirac distribution of holes at temperature TT and chemical potential μ\mu.

In multiorbital systems without symmetry-protected band degeneracy, such as Rashba systems with broken inversion P\mathcal{P} symmetry [8, 9, 10, 37] and ferromagnetic metals with broken time-reversal T\mathcal{T} symmetry [18], the electron bands are nondegenerate at generic momenta, forcing m=nm=n in Eq. (8). By contrast, in centrosymmetric systems preserving time-reversal symmetry, the combined PT\mathcal{P}\mathcal{T} symmetry enforces Kramers degeneracy, giving rise to twofold-degenerate subspaces at every wave vector. We can then choose a proper band basis within the degenerate subspace to render the spin current texture matrix JS,αh,i(k)∣nm=δmn(Sk,αhVi(−k))∣nn\mathcal{J}_{S,\alpha}^{h,i}(\mathbf{k})|_{nm} = \delta_{mn}(S_{\mathbf{k},\alpha}^{h}\mathcal{V}_{i}(-\mathbf{k}))|_{nn} diagonal, which also requires m=nm=n. Thus, the pumping process can be understood as magnon-induced non-equilibrium redistribution δnn(k)\delta n_{n}(\mathbf{k}) of holes in band nn across different wave vectors, giving rise to tensorial spin current due to the underlying spin textures. The same analysis applies to the orbital current, with spin texture replaced by the corresponding orbital texture.

The general formula (8) is reduced to the longitudinal spin current when considering a free electron model with dispersion εkh=ℏ2k2/(2m∗)\varepsilon_{\mathbf{k}}^{h} = \hbar^{2}k^{2}/(2m^{*}) and effective mass m∗m^{*} [20, 38], in which the band indices reduce to spin indices {m,n,m′}∈{↑,↓}\{m,n,m'\} \in\{\uparrow,\downarrow\}. The summation over the spin indices {m,n,m′}\{m,n,m'\} then simplifies the pumped spin current to JS,iDC,α∝∫dκdqzF(κ,qz)(M(+)×M(+)∗)αVi(−q~z,−κ)J_{S,i}^{\mathrm{DC},\alpha} \propto\int d\boldsymbol{\kappa}dq_{z}F(\boldsymbol{\kappa},q_{z})(\mathbf{M}^{(+)}\times\mathbf{M}^{(+)*})_{\alpha}\mathcal{V}_{i}(-\widetilde{q}_{z},-\boldsymbol{\kappa}), where F(κ,qz)F(\boldsymbol{\kappa},q_{z}) is an even function of wave vectors. The spin polarization of the pumped current is aligned with the saturation magnetization, along the direction of M(+)×M(+)∗\mathbf{M}^{(+)}\times\mathbf{M}^{(+)*}. Further, since Vx,y(−q~z,−κ)\mathcal{V}_{x,y}(-\widetilde{q}_{z},-\boldsymbol{\kappa}) are odd under κ→−κ\boldsymbol{\kappa}\to-\boldsymbol{\kappa}, only the longitudinal spin current flowing normal to the interface survives upon integration over the in-plane momenta [20, 38]. This process can be understood in terms of spin-dependent scattering, whereby holes are transferred from states with spin antiparallel to the saturation magnetization to states with spin parallel to it. By contrast, in multiorbital materials with spin-orbital textures, the summation over band indices becomes considerably more intricate and generally cannot be simplified in the same manner. As a consequence, the spin polarization of the pumped current is no longer necessarily aligned with the saturation magnetization, and spin current flowing in the other directions may emerge. Application to group-IV semiconductors.—Group-IV semiconductors provide a prototypical model system because of orbital textures originating from multiple pp-orbitals in the valence band [32, 33, 39, 40, 41, 42, 43].

orbital transport in these materials has been of particular interest because an intrinsic orbital Hall effect was originally predicted for pp-doped Si [35] and has recently been investigated experimentally [44]. When the atomic SOC is explicitly retained, the angular momentum injected by FMR is not confined to the spin sector, but interplays with the orbital degree of freedom. We focus on the valence bands with three pp-orbitals, incorporating the SOC between the pp-orbitals and spins. The Luttinger Hamiltonian [32, 33] around the Γ\Gamma point under the basis of ∣Px,↑⟩|P_x,\uparrow\rangle, ∣Px,↓⟩|P_x,\downarrow\rangle, ∣Py,↑⟩|P_y,\uparrow\rangle, ∣Py,↓⟩|P_y,\downarrow\rangle, ∣Pz,↑⟩|P_z,\uparrow\rangle, and ∣Pz,↓⟩|P_z,\downarrow\rangle reads

H^f(r)=Ak^2−(A−B)∑ik^i2T^i2−C2∑i≠jk^ik^j{T^i,T^j}+λ3(T⋅Σ−E6×6)(9)\hat{H}_{f}(\mathbf{r}) = A\hat{k}^{2} - (A-B)\sum_{i}\hat{k}_{i}^{2}\hat{T}_{i}^{2} - \frac{C}{2}\sum_{i\ne j}\hat{k}_{i}\hat{k}_{j}\{\hat{T}_{i},\hat{T}_{j}\} + \frac{\lambda}{3}(\mathbf{T}\cdot\mathbf{\Sigma}-E_{6\times6}) \tag*{(9)}

in which the coefficients {A,B,C}\{A,B,C\} [32, 33], the atomic orbital T\mathbf{T}, and the spin-Σ\mathbf{\Sigma} matrices are detailed in Method. E6×6E_{6\times6} is the identity matrix, and λ\lambda denotes the energy splitting at the Γ\Gamma point between the spin-split-off band and heavy-hole/light-hole bands.

In Ge, the strong SOC with λ=290\lambda=290 meV [40] separates the spin-split-off band from the heavy-hole and light-hole bands, while the heavy-hole and light-hole dispersions split away from the Γ\Gamma point due to the crystal field effect. All these bands remain twofold degenerate at generic momenta, enforced by the combined PTPT symmetry. With a light doping, the spin-split-off band is negligibly populated and thus contributes little to the pumped current. Furthermore, because the heavy-hole group velocity is much smaller than that of the light holes, the heavy-hole states dominate the pumping response, as the current contribution in Eq. (8) is inversely proportional to the square of the group velocity [∂qzεm,q~zh(κ)]2[\partial_{q_z}\varepsilon_{m,\tilde{q}_z}^{h}(\boldsymbol{\kappa})]^2. While our calculation includes all the contributions, we here trace the spin- and orbital-swapping currents induced by the spin and orbital textures of heavy-hole states.

By the strong SOC, the spin and orbital textures of heavy-hole bands {Ha,Hb}\{H_a,H_b\} are locked to the wave-vector direction. Figure 3(a) and (b) illustrate the spin and orbital textures in one heavy-hole band HaH_a, showing that the spin texture is nearly radially inward, whereas the OAM texture is nearly radially outward. Both textures are reversed for the other heavy-hole band HbH_b. Magnon-assisted scattering preferentially increases the hole populations with spins aligned with the saturation magnetization, leading to a momentum-dependent redistribution and imbalance of holes.

Figure 3. Spin and orbital textures vs. scattering-induced population changes of holes in the degenerate heavy-hole manifold. (a) and (b) plot the spin texture Skh\mathbf{S}^{h}_{\mathbf{k}} and orbital texture Lk\mathbf{L}_{\mathbf{k}} of holes in one of the heavy-hole band HaH_a. The spin texture is nearly radially inward, whereas the OAM texture is nearly radially outward. (c) and (d) illustrate the changes in hole population δnHa\delta n_{H_a} and δnHb\delta n_{H_b} in the two hole bands when the magnon spin is normal to the interface. (e) and (f) show the other situation when the magnon spin is parallel to the interface. Orange and green regions indicate an increase (δn>0\delta n > 0) and a decrease (δn<0\delta n < 0) of hole population.

When the saturation magnetization is normal to the interface as in Figure 3(c) and (d), the hole population in the band HaH_a increases for kz<0k_z<0 and decreases for kz>0k_z>0, i.e., δnHa(kz<0)>0\delta n_{H_a}(k_z<0)>0 and δnHa(kz>0)<0\delta n_{H_a}(k_z>0)<0. The opposite behavior occurs in the band HbH_b, where the hole occupation decreases for kz<0k_z<0 and increases for kz>0k_z>0, i.e., δnHb(kz<0)<0\delta n_{H_b}(k_z<0)<0 and δnHb(kz>0)>0\delta n_{H_b}(k_z>0)>0. The positive group velocities of the nearly parabolic heavy-hole band select kz>0k_z > 0 for the pumped spin/orbital current at z>0z > 0. Therefore, by the opposite spin textures, the depletion of holes δnHa(kz>0)<0\delta n_{H_a}(k_z > 0) < 0 in band HaH_a and the enhanced population δnHb(kz>0)>0\delta n_{H_b}(k_z > 0) > 0 in band HbH_b contribute constructively to a negative conventional spin current JSzzJ^z_{S_z}, accompanied by the swapped spin-current components JSxxJ^x_{S_x} and JSyyJ^y_{S_y}.

When the saturation magnetization ∥y^\parallel\hat{\mathbf{y}} is parallel to the interface, as in Figure 3(e) and (f), the hole population in the band HaH_a decreases for ky>0k_y > 0 and increases for ky<0k_y < 0, i.e., δnHa(ky>0)<0\delta n_{H_a}(k_y > 0) < 0 and δnHa(ky<0)>0\delta n_{H_a}(k_y < 0) > 0, whereas that in band HbH_b increases for ky>0k_y > 0 and decreases for ky<0k_y < 0, i.e., δnHb(ky>0)>0\delta n_{H_b}(k_y > 0) > 0 and δnHb(ky<0)<0\delta n_{H_b}(k_y < 0) < 0. We again examine the intraband spin-current textures in the kz>0k_z > 0 hemisphere.

textures in the kz>0k_z > 0 hemisphere. The depletion of holes in the ky>0k_y > 0 region and the enhanced population in the ky<0k_y < 0 region in a single band HaH_a contribute to both the conventional and swapped spin currents [Figure 3(e)], which are the same for the other band HbH_b [Figure 3(f)], leading to the conventional spin current JSyzJ^z_{S_y} and the swapped spin current JSzyJ^y_{S_z}. The same mechanism extends to orbital-current generation by replacing the spin-current texture with the corresponding orbital-current texture. These results demonstrate that the interplay between the spin and orbital textures and the magnon-induced hole scattering (Figure 2) determines the allowed spin and orbital current components.

We numerically calculate the spin and orbital currents injected into the bulk Ge by the FMR of yttrium iron garnet (YIG) [45, 46, 47]. The Luttinger-Hamiltonian parameters for Ge are A=−1.85×10−37A = -1.85 \times10^{-37} J · m2^2, B=−2.96×10−38B = -2.96 \times10^{-38} J · m2^2, and C=−2.08×10−37C = -2.08 \times10^{-37} J · m2^2 [33, 40]. The calculated parameters for the orbital matrix are given in Method. The applied magnetic field μ0H0=0.05\mu_0 H_0 = 0.05 T corresponds to a FMR frequency ω=2π×9.1\omega= 2\pi\times9.1 GHz. The driven transverse magnetization is 0.2Ms0.2M_s, with μ0Ms=0.177\mu_0 M_s = 0.177 T for YIG. The exchange coupling Jex=5.4×10−3J_{\rm ex} = 5.4 \times10^{-3} A−1^{-1} · s−1^{-1} corresponds to an exchange spin splitting ∼1\sim1 meV.

To trace the origin of the orbital currents and swapped spin current, Figure 4 plots all calculated non-zero spin-current and orbital-current components as a function of SOC strength λ\lambda (noting λ=290\lambda= 290 meV for Ge [40]) at room temperature T=300T = 300 K and chemical potential μ=50\mu= 50 meV. When the saturation magnetization is normal to the interface of YIG|Ge, increasing the SOC strength gradually induces the swapped spin currents JSxxJ^x_{S_x} and JSyyJ^y_{S_y}, as well as the orbital-swapping currents JLxxJ^x_{L_x}, JLyyJ^y_{L_y}, and JLzzJ^z_{L_z}, in addition to the primary longitudinal spin current JSzzJ^z_{S_z} [Figure 4(a) and (b)]. The spin swapping in this configuration achieves a high efficiency ≳50%\gtrsim50\%. Notably, JSxxJ^x_{S_x} and JSyyJ^y_{S_y} have equal magnitudes, as do JLxxJ^x_{L_x} and JLyyJ^y_{L_y}. These relations are dictated by the cubic crystal symmetry described by the OhO_h point group (shown below). When the saturation magnetization (along y^\hat{\mathbf{y}}) is parallel to the interface, increasing the SOC strength induces the swapped spin current JSzyJ^y_{S_z} and the orbital currents JLzyJ^y_{L_z} and JLyzJ^z_{L_y}, in addition to the primary spin current JSyzJ^z_{S_y} [Figure 4(c) and (d)]. The spin swapping efficiency is ≳20%\gtrsim20\% in this case. The pumped orbital current is one order of magnitude larger than the spin current due to the large orbital OAM carried by holes in Ge (see Table II in Method). The pumped orbital current in Si is smaller (see details in the Supplementary Information [34]).

Pumped spin and orbital currents in Ge for different magnetization configurations

Figure 4. Pumped spin and orbital currents in Ge by treating the SOC strength λ\lambda as a parameter for different magnetization configurations. (a) and (b) show the first scenario of spin and orbital swapping when the saturation magnetization is normal to the interface. (c) and (d) show the second scenario when the saturation magnetization is parallel to the interface.

ElementE0E_0 (eV)λ\lambda (eV)PP (Å·eV)m0D1/ℏ2m_0D_1/\hbar^2m0D2/ℏ2m_0D_2/\hbar^2
Si4.070.049.07−1.33−1.31
Ge0.890.2910.01−7.39−5.57
Sn−0.410.779.5214.41−16.67
C13.040.0013.78−0.95−0.95

Table II. Kane-model parameters and orbital parameters for group-IV elements [40].

Orbital-texture-mediated spin swapping was recently predicted in centrosymmetric ferromagnets [18], where exchange spin splitting is essential and broken time-reversal symmetry renders the bands nondegenerate at generic momenta. Here, by contrast, PTPT symmetry preserves Kramers-degenerate pairs with opposite spin and orbital textures, so that spin pumping into non-magnetic centrosymmetric multiorbital materials generates swapped spin and orbital currents through an explicit spin →\to orbital →\to spin intermediate channel, governed solely by crystalline symmetry rather than by magnetic order or spin-band splitting.

Symmetry analysis and wide material choices

These phenomena can also be interpreted as the interconversion between spin and orbital currents. Due to the SOC, the pumped longitudinal spin current is converted into a longitudinal orbital current flowing perpendicular to the interface, as well as an orbital Hall current propagating parallel to the interface. The conversion from the orbital current in turn induces the transverse components of the spin current. The conversion from the spin current JS\mathbf{J}_{S} to the orbital current JL\mathbf{J}_{L} is expressed as JL,αi=NαβijJS,βjJ_{L,\alpha}^{i}=N_{\alpha\beta}^{ij}J_{S,\beta}^{j}, governed by spin-orbital conversion matrix NαβijN_{\alpha\beta}^{ij}. Because the spin current JS,βjJ_{S,\beta}^{j} and orbital current JL,αiJ_{L,\alpha}^{i} are second-order axial tensors, in which the subscript indicates the angular-momentum polarization and the superscript represents the spatial indices, the coefficients [Nαβij][N_{\alpha\beta}^{ij}] form a fourth-order polar tensor with 3×3×3×3=813\times3\times3\times3=81 elements. It is constrained by the crystal point-group symmetry

Nαβij=∑m,n,η,λ={x,y,z}RimRjnRαηRβλNηλmn,(10)N_{\alpha\beta}^{ij}=\sum_{m,n,\eta,\lambda=\{x,y,z\}}R_{im}R_{jn}R_{\alpha\eta}R_{\beta\lambda}N_{\eta\lambda}^{mn}, \tag*{(10)}

where RR is the transformation matrix of the symmetry operation.

The group-IV semiconductors such as Ge, Si, and diamond exhibit the 3ˉm\bar{3}m (OhO_h) point-group symmetry. The symmetry analysis implies that the nonzero orbital-spin conversion tensor [Nαβij][N_{\alpha\beta}^{ij}] has four independent coefficients {N1,N2,N3,N4}\{N_1,N_2,N_3,N_4\} as shown in Table I. When the saturation magnetization is perpendicular to the interface as in Figure 4(a), the primary pumped spin current JS,zzJ_{S,z}^{z} is converted into orbital currents JL,xx=JL,yy=N2JS,zzJ_{L,x}^{x}=J_{L,y}^{y}=N_2J_{S,z}^{z} and JL,zz=N1JS,zzJ_{L,z}^{z}=N_1J_{S,z}^{z}. In contrast, when the saturation magnetization along y^\hat{y} is parallel to the interface as in Figure 4(c), the primary spin current JS,yzJ_{S,y}^{z} can generate a longitudinal orbital current JL,yz=N3JS,yzJ_{L,y}^{z}=N_3J_{S,y}^{z} and a swapped orbital current JL,zy=N4JS,yzJ_{L,z}^{y}=N_4J_{S,y}^{z}. The same symmetry analysis applies to the conversion from the orbital current to the swapped spin current. Our numerical results for Ge in Figure 4 are consistent with these symmetry-allowed current components and their relations. Furthermore, extending the symmetry analysis beyond group-IV semiconductors to materials with different crystal symmetries, e.g., GaAs, GaN, CuInSe2_2, and SnSe, reveals that analogous swapping channels are also symmetry-allowed (see the Supplementary Information Sec. IV [34]). These findings suggest that spin and orbital swapping in the pumping configuration can occur in a broad range of multiorbital materials.

iα\jβi\alpha\backslash j\betaxxxxyxyxzxzxxyxyyyyyzyzyxzxzyzyzzzzz
xxxxN1N_1000N2N_2000N2N_2
yxyx0N3N_30N4N_400000
zxzx00N3N_3000N4N_400
xyxy0N4N_40N3N_300000
yyyyN2N_2000N1N_1000N2N_2
zyzy00000N3N_30N4N_40
xzxz00N4N_4000N3N_300
yzyz00000N4N_40N3N_30
zzzzN2N_2000N2N_2000N1N_1

Table 1. Spin-orbital interconversion tensor [Nαβij][N_{\alpha\beta}^{ij}].

Conclusion and discussion

In summary, we uncover an orbital-mediated mechanism of spin and orbital swapping driven by spin pumping to centrosymmetric multiorbital materials. Spin-orbit-entangled textures redirect the injected angular momentum into swapped spin and orbital currents, without requiring spin-band splitting or extrinsic spin-orbit scattering. Microscopic calculations for a ferromagnet|group-IV semiconductor heterostructure demonstrate the high efficiency of this mechanism, 20% to 50%, which depends on spin-orbit coupling and magnetization orientation. Fabrication of heavy-metal electrodes on the lateral faces of the crystal favors future detection of the predicted effects.

A fourth-rank spin-orbital conversion tensor establishes the crystalline-symmetry constraints on these responses beyond the specific Hamiltonian model. These results establish spin swapping in spin-degenerate band structures and connect it to the orbital structure of Bloch states. By identifying orbital textures as a source of angular-momentum redirection, our work opens a new way for the realization of spin and orbital swapping with broadened material choices and provides a framework for controlling coupled spin and orbital transport through band structure and crystal symmetry.

Methods and materials

Luttinger Hamiltonian.—The material calculations in this work are based on the Luttinger k⋅p\mathbf{k}\cdot\mathbf{p} Hamiltonian. For the valence bands with three pp-orbitals in group-IV semiconductors, the Luttinger Hamiltonian [32, 33] around the Γ\Gamma point under the orbit-spin product basis of ∣Px,↑⟩\lvert P_x,\uparrow\rangle, ∣Px,↓⟩\lvert P_x,\downarrow\rangle, ∣Py,↑⟩\lvert P_y,\uparrow\rangle, ∣Py,↓⟩\lvert P_y,\downarrow\rangle, ∣Pz,↑⟩\lvert P_z,\uparrow\rangle, and ∣Pz,↓⟩\lvert P_z,\downarrow\rangle reads

H^f(r)=Ak^2−(A−B)∑ik^i2I^i2−C2∑i≠jk^ik^j{I^i,I^j}+λ3(I⋅Σ−E6×6),(11)\hat{H}_{f}(\mathbf{r})=A\hat{k}^{2}-(A-B)\sum_i\hat{k}_i^2\hat{I}_i^2-\frac{C}{2}\sum_{i\ne j}\hat{k}_i\hat{k}_j\{\hat{I}_i,\hat{I}_j\}+\frac{\lambda}{3}\left(\mathcal{I}\cdot\mathbf{\Sigma}-E_{6\times6}\right), \tag*{(11)}

in which the coefficients [32, 33]

A=ℏ22m0+ℏ2m02∑s≠pπpsxπsxxεp−εs,A=\frac{\hbar^2}{2m_0}+\frac{\hbar^2}{m_0^2}\sum_{s\ne p}\frac{\pi^x_{ps}\pi^x_{sx}}{\varepsilon_p-\varepsilon_s},
B=ℏ22m0+ℏ2m02∑s≠pπpsyπsxyεp−εs,B=\frac{\hbar^2}{2m_0}+\frac{\hbar^2}{m_0^2}\sum_{s\ne p}\frac{\pi^y_{ps}\pi^y_{sx}}{\varepsilon_p-\varepsilon_s},
C=ℏ2m02∑s≠pπpsxπsyy+πpsyπsyxεp−εs.(12)C=\frac{\hbar^2}{m_0^2}\sum_{s\ne p}\frac{\pi^x_{ps}\pi^y_{sy}+\pi^y_{ps}\pi^x_{sy}}{\varepsilon_p-\varepsilon_s}. \tag*{(12)}

Here, εs\varepsilon_s is the energy extrema of the other bands at the Γ\Gamma point;

πmn=∑j={x,y,z}πmnje^j=⟨m∣(p+ℏ4m0c2(σ×∇V0))∣n⟩\begin{aligned} \pi_{mn}=\sum_{j=\{x,y,z\}}\pi^j_{mn}\hat{e}_j \\ &=\langle m\mid\left(\mathbf{p}+\frac{\hbar}{4m_0c^2}\left(\boldsymbol{\sigma}\times\nabla V_0\right)\right)\mid n\rangle \end{aligned}

represents the matrix elements of the momentum operator between the states ∣m⟩,∣n⟩∈∣Px,y,z,s={↑,↓}⟩\lvert m\rangle,\lvert n\rangle\in\lvert P_{x,y,z},s=\{\uparrow,\downarrow\}\rangle, in which m0m_0 is the mass of free electrons, V0V_0 is the periodic crystal potential, and σ^\hat{\sigma} denotes the Pauli spin operators. The last term in Eq. (11) represents the SOC [48, 49] with coupling strength λ\lambda between the atomic orbital I\mathcal{I} matrices and the spin-Σ\mathbf{\Sigma} matrices, which contain the components

Ix=(0000000000000000−i000000−i00i000000i00),Iy=(0000i000000i000000000000−i000000−i0000),Iz=(00−i000000−i00i000000i0000000000000000),\mathcal{I}_x= \begin{pmatrix} 0&0&0&0&0&0\\ 0&0&0&0&0&0\\ 0&0&0&0&-i&0\\ 0&0&0&0&0&-i\\ 0&0&i&0&0&0\\ 0&0&0&i&0&0 \end{pmatrix}, \quad \mathcal{I}_y= \begin{pmatrix} 0&0&0&0&i&0\\ 0&0&0&0&0&i\\ 0&0&0&0&0&0\\ 0&0&0&0&0&0\\ -i&0&0&0&0&0\\ 0&-i&0&0&0&0 \end{pmatrix}, \quad \mathcal{I}_z= \begin{pmatrix} 0&0&-i&0&0&0\\ 0&0&0&-i&0&0\\ i&0&0&0&0&0\\ 0&i&0&0&0&0\\ 0&0&0&0&0&0\\ 0&0&0&0&0&0 \end{pmatrix},
Σx=(010000100000000100001000000001000010),Σy=(0−i0000i00000000−i0000i00000000−i0000i0),Σz=(1000000−10000001000000−10000001000000−1).(13)\mathbf{\Sigma}_x= \begin{pmatrix} 0&1&0&0&0&0\\ 1&0&0&0&0&0\\ 0&0&0&1&0&0\\ 0&0&1&0&0&0\\ 0&0&0&0&0&1\\ 0&0&0&0&1&0 \end{pmatrix}, \quad \mathbf{\Sigma}_y= \begin{pmatrix} 0&-i&0&0&0&0\\ i&0&0&0&0&0\\ 0&0&0&-i&0&0\\ 0&0&i&0&0&0\\ 0&0&0&0&0&-i\\ 0&0&0&0&i&0 \end{pmatrix}, \quad \mathbf{\Sigma}_z= \begin{pmatrix} 1&0&0&0&0&0\\ 0&-1&0&0&0&0\\ 0&0&1&0&0&0\\ 0&0&0&-1&0&0\\ 0&0&0&0&1&0\\ 0&0&0&0&0&-1 \end{pmatrix}. \tag*{(13)}

Renormalized orbital matrix by SOC.—As addressed in the main text, the SOC with the bands out of the subspace of interest efficiently renormalizes the orbital matrix of holes, which in the orbital-spin product basis ∣Px,y,z⟩⊗∣s={↑,↓}⟩\lvert P_{x,y,z}\rangle\otimes\lvert s=\{\uparrow,\downarrow\}\rangle is found to be (see Supplementary Information Sec. II [34])

Lx=m0ℏ2(000i3(D1−D2)−13(D1−D2)000−i3(D1−D2)0013(D1−D2)−i3(D1−D2)i3(D1−D2)23(D1−D2)0−2i3(2D1+D2)0−13(D1−D2)02i3(D1+D2)00−2i3(2D1+D2)0002i3(2D1+D2)23(D1−D2)0013(D1−D2)023i(2D1+D2)23(D1−D2)0),\mathcal{L}_x=\frac{m_0}{\hbar^2}\begin{pmatrix} 0&0&0&\frac{i}{3}(D_1-D_2)&-\frac{1}{3}(D_1-D_2)&0\\ 0&0&-\frac{i}{3}(D_1-D_2)&0&0&\frac{1}{3}(D_1-D_2)\\ -\frac{i}{3}(D_1-D_2)&\frac{i}{3}(D_1-D_2)&\frac{2}{3}(D_1-D_2)&0&-\frac{2i}{3}(2D_1+D_2)&0\\ -\frac{1}{3}(D_1-D_2)&0&\frac{2i}{3}(D_1+D_2)&0&0&-\frac{2i}{3}(2D_1+D_2)\\ 0&0&0&\frac{2i}{3}(2D_1+D_2)&\frac{2}{3}(D_1-D_2)&0\\ 0&\frac{1}{3}(D_1-D_2)&0&\frac{2}{3}i(2D_1+D_2)&\frac{2}{3}(D_1-D_2)&0 \end{pmatrix},
Ly=m0ℏ2(0−2i3(D1−D2)0−13(D1−D2)2i3(2D1+D2)02i3(D1−D2)0−13(D1−D2)002i3(2D1+D2)013(−D1+D2)00−13(D1−D2)0−13(D1−D2)000013(D1−D2)−2i3(2D1+D2)0−13(D1−D2)00−2i3(D1−D2)0−2i3(2D1+D2)013(D1−D2)2i3(D1−D2)0),\mathcal{L}_y=\frac{m_0}{\hbar^2}\begin{pmatrix} 0&-\frac{2i}{3}(D_1-D_2)&0&-\frac{1}{3}(D_1-D_2)&\frac{2i}{3}(2D_1+D_2)&0\\ \frac{2i}{3}(D_1-D_2)&0&-\frac{1}{3}(D_1-D_2)&0&0&\frac{2i}{3}(2D_1+D_2)\\ 0&\frac{1}{3}(-D_1+D_2)&0&0&-\frac{1}{3}(D_1-D_2)&0\\ -\frac{1}{3}(D_1-D_2)&0&0&0&0&\frac{1}{3}(D_1-D_2)\\ -\frac{2i}{3}(2D_1+D_2)&0&-\frac{1}{3}(D_1-D_2)&0&0&-\frac{2i}{3}(D_1-D_2)\\ 0&-\frac{2i}{3}(2D_1+D_2)&0&\frac{1}{3}(D_1-D_2)&\frac{2i}{3}(D_1-D_2)&0 \end{pmatrix},
Lz=m0ℏ2(23(D1−D2)0−2i3(2D1+D2)00−13(D1−D2)0−23(D1−D2)0−23i(2D1+D2)−13(D1−D2)02i3(2D1+D2)023(D1−D2)00i3(D1−D2)02i3(2D1+D2)0−23(D1−D2)−i3(D1−D2)00−13(D1−D2)0i3(D1−D2)00−13(D1−D2)0−i3(D1−D2)000).(14)\mathcal{L}_z=\frac{m_0}{\hbar^2}\begin{pmatrix} \frac{2}{3}(D_1-D_2)&0&-\frac{2i}{3}(2D_1+D_2)&0&0&-\frac{1}{3}(D_1-D_2)\\ 0&-\frac{2}{3}(D_1-D_2)&0&-\frac{2}{3}i(2D_1+D_2)&-\frac{1}{3}(D_1-D_2)&0\\ \frac{2i}{3}(2D_1+D_2)&0&\frac{2}{3}(D_1-D_2)&0&0&\frac{i}{3}(D_1-D_2)\\ 0&\frac{2i}{3}(2D_1+D_2)&0&-\frac{2}{3}(D_1-D_2)&-\frac{i}{3}(D_1-D_2)&0\\ 0&-\frac{1}{3}(D_1-D_2)&0&\frac{i}{3}(D_1-D_2)&0&0\\ -\frac{1}{3}(D_1-D_2)&0&-\frac{i}{3}(D_1-D_2)&0&0&0 \end{pmatrix}. \tag*{(14)}

Particularly, disregarding the SOC, i.e. λ=0\lambda=0, the coefficients D1=D2D_1=D_2, leading to L=(m0D/ℏ2)I\mathcal{L}=(m_0D/\hbar^2)\mathcal{I} with D=D1+D2D=D_1+D_2, where I\mathcal{I} denotes the atomic orbital angular momentum in Eq. (11). This approximation was used to obtain orbital dynamics in semiconductors [26, 33, 35]. For the group-IV semiconductors with the diamond structure, such as Si, Ge, Sn, and diamond, we use parameters in the multiband Kane Hamiltonian [39, 41, 43, 50] to calculate the orbital matrix, as well as its associated orbital density and orbital-current density operators of the valence bands, as listed in Table II [40]. The constants m0D1/ℏ2=−7.39m_0D_1/\hbar^2=-7.39 and m0D2/ℏ2=−5.57m_0D_2/\hbar^2=-5.57 differ for Ge, but they are almost the same for Si and diamond. Intriguingly, m0D1/ℏ2=14.41m_0D_1/\hbar^2=14.41 and m0D2/ℏ2=−16.67m_0D_2/\hbar^2=-16.67 can have different signs for Sn. The difference between D1D_1 and D2D_2 arises from the atomic SOC, which modifies the orbital angular momentum of pp-orbital electrons and causes it to deviate from that of the atomic one. The summation m0(D1+D2)/ℏ2=−12.96m_0(D_1+D_2)/\hbar^2=-12.96 for Ge is approximately equal to the result m0D/ℏ2=−11.2m_0D/\hbar^2=-11.2, given in Refs [32, 33].

Funding

This work was supported by the National Natural Science Foundation of China under Grants No. 12374109 and 12474100, as well as the National Key Research and Development Program of China under Grant No. 2023YFA1406600. H.W. acknowledges the support of Swiss National Science Foundation (Grant No. 10004526).

SUPPLEMENTARY DATA

Supplementary data are available at NSR online.

Conflict of interest statement

None declared.

[email protected]

References

  1. [4]C. Espedal, P. Lange, S. Sadjina, A. G. Mal’shukov, and A. Brataas, Spin hall effect and spin swapping in diffusive superconductors, Physical Review B 95, 054509 (2017).DOI
  2. [5]L. J. Kamra and J. Linder, Inverse spin-hall effect and spin swapping in spin-split superconductors, Physical Review Letters 132, 226002 (2024).DOI
  3. [6]H. B. M. Saidaoui and A. Manchon, Spin-swapping transport and torques in ultrathin magnetic bilayers, Physical Review Letters 117, 036601 (2016).DOI
  4. [7]C. Ortiz Pauyac, M. Chshiev, A. Manchon, and S. A. Nikolaev, Nonlocal spin Hall and spin swapping torques in diffusive ferromagnets, Physical Review Letters 120, 176802 (2018).DOI
  5. [8]S. Sadjina, A. Brataas, and A. G. Mal’shukov, Intrinsic spin swapping, Physical Review B 85, 115306 (2012).DOI
  6. [9]J. Borge and I. V. Tokatly, Ballistic spin transport in the presence of interfaces with strong spin-orbit coupling, Physical Review B 96, 115445 (2017).DOI
  7. [10]S. Li, K. Shen, and K. Xia, Interfacial spin hall effect and spin swapping in fe-au bilayers from first principles, Physical Review B 99, 134427 (2019).DOI
  8. [11]H. Kim, S. Karube, J. Borge, J. Kim, K. Kondou, and Y. Otani, Evidence for spin swapping from modulation of transverse resistance in magnetic heterostructures with rashba interface, Applied Physics Letters 116, 122403 (2020).DOI
  9. [12]Q. Fu, L. Liang, W. Wang, L. Yang, K. Zhou, Z. Li, C. Yan, L. Li, H. Li, and R. Liu, Observation of nontrivial spin-orbit torque in single-layer ferromagnetic metals, Physical Review B 105, 224417 (2022).DOI
  10. [13]J. Ryu, R. Thompson, J. Y. Park, S.-J. Kim, G. Choi, J. Kang, H. B. Jeong, M. Kohda, J. M. Yuk, J. Nitta, K.-J. Lee, and B.-G. Park, Efficient spin-orbit torque in magnetic trilayers using all three polarizations of a spin current, Nature Electronics 5, 217 (2022).DOI
  11. [14]B. K. Hazra, B. Pal, J.-C. Jeon, R. R. Neumann, B. Göbel, B. Grover, H. Deniz, A. Styeryovedov, H. Meyerheim, I. Mertig, S.-H. Yang, and S. S. P. Parkin, Generation of out-of-plane polarized spin current by spin swapping, Nature Communications 14, 4549 (2023).
  12. [15]F. Wang, G. Shi, K.-W. Kim, H.-J. Park, J. G. Jang, H. R. Tan, M. Lin, Y. Liu, T. Kim, D. Yang, S. Zhao, K. Lee, S. Yang, A. Soumyanarayanan, K.-J. Lee, and H. Yang, Field-free switching of perpendicular magnetization by two-dimensional PtTe2/WTe2 van der Waals heterostructures with high spin Hall conductivity, Nature Materials 23, 768 (2024).DOI
  13. [16]K. Kim, W. Y. Shim, and S.-B. Choe, Huge self-spin swapping effect with asymmetric spin-sink structures, NPG Asia Materials 17, 16 (2025).DOI
  14. [17]M. Wu, K. Kondou, T. Chen, S. Nakatsuji, and Y. Otani, Identifying the origin of out-of-plane spin polarization in the noncollinear antiferromagnet mn3ge, Physical Review B 114, L020403 (2026).arxiv.org/abs/2603.24029
  15. [18]H.-J. Park, H.-W. Ko, G. Go, J. H. Oh, K.-W. Kim, and K.-J. Lee, Spin swapping effect of band structure origin in centrosymmetric ferromagnets, Physical Review Letters 129, 037202 (2022).
  16. [19]Y. Tserkovnyak, A. Brataas, and G. E. W. Bauer, Spin pumping and magnetization dynamics in metallic multilayers, Phys. Rev. B 66, 224403 (2002).DOI
  17. [20]Y. Tserkovnyak, A. Brataas, G. E. W. Bauer, and B. I. Halperin, Nonlocal magnetization dynamics in ferromagnetic nanostructures, Reviews of Modern Physics 77, 1375 (2005).
  18. [21]A. Brataas, Y. Tserkovnyak, G. E. W. Bauer, and P. J. Kelly, Spin pumping and spin transfer, in Spin Current, edited by S. Maekawa, S. O. Valenzuela, E. Saitoh, and T. Kimura (Oxford University Press, Oxford, 2012) pp. 87–135.
  19. [22]H. Hayashi, D. Go, S. Haku, Y. Mokrousov, and K. Ando, Observation of orbital pumping, Nat. Electron. 7, 646 (2024).arxiv.org/abs/2304.05266
  20. [23]D. Go, K. Ando, A. Pezo, S. Blügel, A. Manchon, and Y. Mokrousov, Orbital pumping by magnetization dynamics in ferromagnets, Physical Review B 111, L140409 (2025).arxiv.org/abs/2309.14817
  21. [24]N. N. Keller, A. Bose, N. Soya, E. Hauth, F. Kammerbauer, R. Gupta, H. Hayashi, H. Kashiki, G. Jakob, S. Krishnia, K. Ando, and M. Kläui, Identification of orbital pumping from spin pumping and rectification effects, Nano Lett. 25, 13462 (2025).arxiv.org/abs/2502.08269
  22. [25]H. Wang, M.-G. Kang, D. Petrosyan, S. Ding, R. Schlitz, L. J. Riddiford, W. Legrand, and P. Gambardella, Orbital pumping in ferrimagnetic insulators, Phys. Rev. Lett. 134, 126701 (2025).DOI
  23. [26]C. Cai, H. Wang, and T. Yu, Evanescent orbital pumping by magnetization dynamics free of spin-orbit coupling, Sci. China Phys. Mech. Astron. 69, 237511 (2026).arxiv.org/abs/2504.05827
  24. [27]D. Go, D. Jo, H.-W. Lee, M. Kläui, and Y. Mokrousov, Orbital currents in solids, EPL 135, 37001 (2021).
  25. [28]D. Jo, D. Go, G.-M. Choi, and H.-W. Lee, Spintronics meets orbitronics: Emergence of orbital angular momentum in solids, npj Spintronics 2, 19 (2024).DOI
  26. [29]R. Burgos Atencia, A. Agarwal, and D. Culcer, Orbital angular momentum of bloch electrons: equilibrium formulation, magneto-electric phenomena, and the orbital hall effect, Adv. Phys. X 9, 2371972 (2024).
  27. [30]K. Ando, Orbitronics: Harnessing orbital currents in solid-state devices, J. Phys. Soc. Jpn. 94, 092001 (2025).DOI
  28. [31]T. P. Cysne, L. M. Canonico, M. Costa, R. B. Muniz, and T. G. Rappoport, Orbitronics in two-dimensional materials, npj Spintronics 3, 39 (2025).DOI
  29. [32]J. M. Luttinger, Quantum theory of cyclotron resonance in semiconductors: General theory, Phys. Rev. 102, 1030 (1956).
  30. [33]J. M. Luttinger and W. Kohn, Motion of electrons and holes in perturbed periodic fields, Phys. Rev. 97, 869 (1955).DOI
  31. [34]In the Supplementary Information, we derive the renormalized orbital operators by spin-orbit interaction and the OAM matrix for group-IV semiconductors, set up the scattering theory for spin-orbit pumping, and address the wide material choice for spin swapping.
  32. [35]B. A. Bernevig, T. L. Hughes, and S.-C. Zhang, Orbitronics: The intrinsic orbital current in p-doped silicon, Phys. Rev. Lett. 95, 066601 (2005).
  33. [36]J. Sohn, J. M. Lee, and H.-W. Lee, Dyakonov-perel-like orbital and spin relaxations in centrosymmetric systems, Physical Review Letters 132, 246301 (2024).DOI
  34. [37]K. Chen and S. Zhang, Spin pumping in the presence of spin-orbit coupling, Phys. Rev. Lett. 114, 126602 (2015).pubmed.ncbi.nlm.nih.gov/25860763
  35. [38]Y. Tserkovnyak, A. Brataas, and G. E. W. Bauer, Enhanced gilbert damping in thin ferromagnetic films, Physical Review Letters 88, 117601 (2002).pubmed.ncbi.nlm.nih.gov/11909427
  36. [39]E. O. Kane, Band structure of indium antimonide, J. Phys. Chem. Solids 1, 249 (1957).DOI
  37. [40]P. Lawaetz, Valence-band parameters in cubic semiconductors, Phys. Rev. B 4, 3460 (1971).DOI
  38. [41]R. Winkler, Spin-Orbit Coupling Effects in Two-Dimensional Electron and Hole Systems (Springer, Berlin, 2003).
  39. [42]D. Culcer, J. Sinova, N. A. Sinitsyn, T. Jungwirth, A. H. MacDonald, and Q. Niu, Semiclassical spin transport in spin-orbit-coupled bands, Phys. Rev. Lett. 93, 046602 (2004).pubmed.ncbi.nlm.nih.gov/15323781
  40. [43]E. O. Kane, The k · p method, Semicond. Semimetals 1, 75 (1966).
  41. [44]R. Matsumoto, R. Ohshima, Y. Ando, D. Go, Y. Mokrousov, and M. Shiraishi, Observation of giant orbital hall effect in si, arXiv:2501.14237 10.48550/arXiv.2501.14237 (2025), arXiv:2501.14237.arxiv.org/abs/2501.14237
  42. [45]Y. Kajiwara, K. Harii, S. Takahashi, J. Ohe, K. Uchida, M. Mizuguchi, H. Umezawa, H. Kawai, K. Ando, K. Takanashi, S. Maekawa, and E. Saitoh, Transmission of electrical signals by spin-wave interconversion in a magnetic insulator, Nature 464, 262 (2010).DOI
  43. [46]J.-C. Rojas-Sánchez, M. Cubukcu, A. Jain, C. Vergnaud, C. Portemont, C. Ducruet, A. Barski, A. Marty, L. Vila, J.-P. Attané, E. Augendre, G. Desfonds, S. Gambarelli, H. Jaffrès, J.-M. George, and M. Jamet, Spin pumping and inverse spin hall effect in germanium, Physical Review B 88, 064403 (2013).
  44. [47]E. Santos, J. E. Abrão, J. L. Costa, J. G. S. Santos, G. Rodrigues-Junior, J. B. S. Mendes, and A. Azevedo, Negative orbital hall effect in germanium, Physical Review Applied 22, 064071 (2024).DOI
  45. [48]G. Dresselhaus, A. F. Kip, and C. Kittel, Cyclotron resonance of electrons and holes in silicon and germanium crystals, Phys. Rev. 98, 368 (1955).DOI
  46. [49]G. Dresselhaus, Spin-orbit coupling effects in zinc blende structures, Phys. Rev. 100, 580 (1955).DOI
  47. [50]T. B. Bahder, Eight-band k · p model of strained zinc-blende crystals, Phys. Rev. B 41, 11992 (1990).DOI

Paper details

vlnc:W_df7aa544
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