betaFeedback

WorksPhysics and Astronomy

Micromagnetic description of twisted spin spirals in the B20 chiral magnet FeGe from first principles

Abstract

Using the model of classical Heisenberg exchange and Dzyaloshinskii-Moriya (DM) interaction we show that the ground state of B20 FeGe chiral magnet is a superposition of twisted helical spin-density waves formed by different sublattices of the crystal. Such twisted spin-density waves propagate in the same direction but with different phases and different directions of the rotation axes. We derive an advanced micromagnetic expression describing the exchange and DM interaction for such magnetic structures. By employing first-principles calculations based on density functional theory and using our micromagnetic model we show that the magnitude of the spin-spiral twist in B20 FeGe is of the same order as global spiraling. While the energy difference between the ground state of twisted spirals and the FM state is in good agreement with the experimental results, for the spin spirals without a twist it is smaller by factor three. In addition, we verify our results by employing spin dynamics simulations. This calls for new experiments exploring the ground state properties of B20 chiral magnets.

Introduction

Non-centrosymmetric B20 chiral magnets have been reported to have the interesting property of breaking inversion symmetry, leading to various types of magnetic structures, ranging from spin spirals over skyrmions [1, 2, 3, 4] and bobbers [5, 6] to the three-dimensional (3D) lattice of 3D magnetic textures [7], and thus holding great potential for innovative spintronic applications [8]. Due to the advent of different magnetic phases that can be tuned by temperature, magnetic fields, and other material parameters [9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19], the B20 systems not only host a rich phase diagram but their complex magnetic order and nontrivial topology in momentum space imprint also on unique transport and optical phenomena [2, 20, 21, 22, 23, 24, 25].

Irrespective of the crucial relevance for explaining chiral magnetism in B20 compounds, however, a complete understanding of the underlying magnetic structures and magnetic interactions stabilizing them has been remarkably elusive. While, for instance, the magnetic order at the ground state of many B20 compounds is often interpreted as a homogeneous helix [26, 27] caused by relativistic Dzyaloshinskii-Moriya (DM) interactions [28], such a model ignoring higher-order magnetic interactions [29] and temperature effects [30] fails to explain the 3D texture of a few nanometers observed in B20 MnGe. Furthermore, a now well-known and well-accepted assumption made by Bak and Jensen 40 years ago [27] that DM interaction stabilizes a homogeneous helix in B20 materials is incomplete. In particular, as shown by Chizhikov and Dmitrienko [31, 32, 33], the DM vector that is perpendicular to the component responsible for helical spiraling gives rise to intersublattice canting in B20 compounds. Such an effect can be viewed as the superposition of several helical spin-density waves in each sublattice propagating in the same direction but having different phases and different directions of the rotation axes (see Figure 1).

Superposition of twisted spin-density waves and rotation-axis geometry

Figure 1. (a) Superposition of twisted spin-density waves formed by different sublattices (AA, BB, CC, . . .) of the crystal. Such spirals propagate in the same directions but with different (b) directions of the rotation axes erot(β,α)\mathbf{e}_{\mathrm{rot}}(\beta,\alpha) and cone angles θ\theta and phases ϕ\phi. The magnetic moment S\mathbf{S} in (c) is defined by cone angle θ\theta with respect to erot=R(β,α)ez\mathbf{e}_{\mathrm{rot}}=\mathcal{R}(\beta,\alpha)\mathbf{e}_{z} and by phase ϕ\phi and q⋅R\mathbf{q}\cdot\mathbf{R} with respect to ex′=R(β,α)ex\mathbf{e}_{x}^{\prime}=\mathcal{R}(\beta,\alpha)\mathbf{e}_{x}, where R(β,α)\mathcal{R}(\beta,\alpha) is the rotation matrix mapping the ez\mathbf{e}_{z} axis to the rotation axis erot\mathbf{e}_{\mathrm{rot}}. A spin spiral with twist Ω={β,α,θ,ϕ}\Omega=\{\beta,\alpha,\theta,\phi\} can be characterized by orthogonal vectors em(Ω)\mathbf{e}_{m}(\Omega), m=1,2,3m=1,2,3 [Eq. (1)].

In this work, we derive an advanced micromagnetic energy equation describing the exchange and DM interactions for the spin spirals with twists formed by the magnetic moments of different sublattices in a crystal. While the micromagnetic energies of the exchange (EexE^{\mathrm{ex}}) and DM interaction (EdmE^{\mathrm{dm}}) for the trivial (without twist) spin spirals have well-known dependencies on their wave vector q\mathbf{q},

Eex=Eex(q0,q⊗2,…),Edm=Edm(q1,q⊗3,…),E^{\mathrm{ex}} = E^{\mathrm{ex}}(\mathbf{q}^{0},\mathbf{q}^{\otimes2},\ldots), \qquad E^{\mathrm{dm}} = E^{\mathrm{dm}}(\mathbf{q}^{1},\mathbf{q}^{\otimes3},\ldots),

where q⊗p\mathbf{q}^{\otimes p} is the pp-fold tensor product of q\mathbf{q} vector with itself, we show that the coupling between different sublattices gives rise to the twist of spin spirals with new contributions in q\mathbf{q} to the energies of magnetic interactions:

Eex=Eex(q0,q1,q⊗2,q⊗3,…),Edm=Edm(q0,q1,q⊗2,q⊗3,…).E^{\mathrm{ex}} = E^{\mathrm{ex}}(\mathbf{q}^{0},\mathbf{q}^{1},\mathbf{q}^{\otimes2},\mathbf{q}^{\otimes3},\ldots), \qquad E^{\mathrm{dm}} = E^{\mathrm{dm}}(\mathbf{q}^{0},\mathbf{q}^{1},\mathbf{q}^{\otimes2},\mathbf{q}^{\otimes3},\ldots).

Interestingly, the last expression implies also that even when ∣q∣=q=0|\mathbf{q}|=q=0, the ground state magnetic order might not be ferromagnetic (FM). To substantiate the importance of such an effect in real materials, we employ density functional theory calculations for B20 FeGe since the energies without and with spin-orbit coupling in this compound are well described by the exchange and DM interactions [34], in contrast to MnGe, for which higher-order interactions are expected to be pivotal [35]. We compute the atomistic interaction parameters via multiple-scattering theory as implemented in the Korringa-Kohn-Rostoker (KKR) Green’s function method [36, 37, 38], and from them we determine the corresponding intersublattice micromagnetic parameters entering the advanced micromagnetic energy equation. By minimization of the micromagnetic energy, we determine the strength of global spiraling, as well as local twist between helices in different sublattices. We show that the magnitude of the spin spiral twist in B20 FeGe is of the same order as global spiraling and it reduces, by three times, the energy difference between the helical ground state and the FM state, resulting in good agreement with the experimentally measured saturation magnetic field. Finally, we verify the results of our micromagnetic model by employing spin-dynamics simulations, that in addition to the twist of the spin spiral in B20 structures, indicate their small nonhomogeneity. However, since the energy gain due to the nonhomogeneity of the spin waves is much smaller than that due to their twist, for the sake of simplicity we ignore this effect in our micromagnetic model.

Spin-spiral twist: Micromagnetic model

We consider a one-dimensional spin-wave formed by conical homogeneous spin spirals in each sublattice AA of a crystal. For each spin spiral we assume the same wave vector q\mathbf{q} but a different rotation axis erotA(βA,αA)\mathbf{e}_{\mathrm{rot}}^{A}(\beta^{A},\alpha^{A}), cone angle θA\theta^{A}, and phase ϕA\phi^{A} (see Figure 1 and Appendix A). We define the orientation of any classical spin SiA\mathbf{S}_{i}^{A} of atom ii in sublattice AA as

SiA=e1Acos⁡(q⋅RiA)+e2Asin⁡(q⋅RiA)+e3A,(1)\mathbf{S}_{i}^{A}=\mathbf{e}_{1}^{A}\cos\left(\mathbf{q}\cdot\mathbf{R}_{i}^{A}\right)+\mathbf{e}_{2}^{A}\sin\left(\mathbf{q}\cdot\mathbf{R}_{i}^{A}\right)+\mathbf{e}_{3}^{A}, \tag*{(1)}

where orthogonal vectors enA=en(βA,αA,θA,ϕA)\mathbf{e}_{n}^{A}=\mathbf{e}_{n}(\beta^{A},\alpha^{A},\theta^{A},\phi^{A}) (n=1,2,3n=1,2,3) characterize the spin-spiral twist in sublattice AA with a condition that ∣SiA∣=1\lvert\mathbf{S}_{i}^{A}\rvert=1 (see Figure 1(c) and Appendix B). The upper index AA enumerates also the basis atom (i=1i=1) with its atomic position rA=R1A\mathbf{r}^{A}=\mathbf{R}_{1}^{A} in the chemical unit cell. RiA=rA+τi\mathbf{R}_{i}^{A}=\mathbf{r}^{A}+\boldsymbol{\tau}_{i} defines the position of atom ii in sublattice AA, where τi\boldsymbol{\tau}_{i} is a translation vector.

We examine the effect of spin-spiral twist on the exchange and DM interactions following the effective spin-lattice Hamiltonian:

EAB=−1NA∑i=1NA∑j=1NBSiAJijABSjB,(2)E^{AB}=-\frac{1}{N_{A}}\sum_{i=1}^{N_{A}}\sum_{j=1}^{N_{B}}\mathbf{S}_{i}^{A}\mathcal{J}_{ij}^{AB}\mathbf{S}_{j}^{B}, \tag*{(2)}

where the sum over ii is restricted to the number of magnetic atoms NAN_{A} in sublattice AA of the magnetic unit cell and jj runs over atoms in sublattice BB of whole crystal structure. Tensor JijAB\mathcal{J}_{ij}^{AB} represents the strength of the exchange JijABJ_{ij}^{AB} and DM interaction DijABD_{ij}^{AB} between two sites ii and jj of sublattices AA and BB, respectively:

JijAB=(JDz−Dy−DzJDxDy−DxJ)ijAB.(3)\mathcal{J}_{ij}^{AB}= \begin{pmatrix} J & D_{z} & -D_{y}\\ -D_{z} & J & D_{x}\\ D_{y} & -D_{x} & J \end{pmatrix}_{ij}^{AB}. \tag*{(3)}

If the magnetic structure in each sublattice has characteristic length scales much larger than the underlying crystal lattice, then the sum over NAN_{A} (over the magnetic unit cell) in Eq. (2) can be approximated by the integral [39], and the energy density of the magnetic interactions (between atoms in the chemical unit cell rA\mathbf{r}^{A} and all other atoms RjB\mathbf{R}_{j}^{B}) has a form

EAB(q)≈−12π∑j=1NB∫02πdω SA(q⋅rA+ω)J1jABSB(q⋅RjB+ω).(4)E^{AB}(\mathbf{q})\approx-\frac{1}{2\pi}\sum_{j=1}^{N_{B}}\int_{0}^{2\pi}\mathrm{d}\omega\,\mathbf{S}^{A}\left(\mathbf{q}\cdot\mathbf{r}^{A}+\omega\right)\mathcal{J}_{1j}^{AB}\mathbf{S}^{B}\left(\mathbf{q}\cdot\mathbf{R}_{j}^{B}+\omega\right). \tag*{(4)}

Considering the magnetic structure given by Eq. (1), the total energy after taking an integral in Eq. (4) has a form

EAB(q)≈−∑j=1NBJ1jAB:[C+AB+12C−ABcos⁡(q⋅R1jAB)+12C−ABsin⁡(q⋅R1jAB)].(5)E^{AB}(\mathbf{q})\approx-\sum_{j=1}^{N_{B}}\mathcal{J}_{1j}^{AB}:\left[\mathcal{C}_{+}^{AB}+\frac{1}{2}\mathcal{C}_{-}^{AB}\cos\left(\mathbf{q}\cdot\mathbf{R}_{1j}^{AB}\right)+\frac{1}{2}\mathcal{C}_{-}^{AB}\sin\left(\mathbf{q}\cdot\mathbf{R}_{1j}^{AB}\right)\right]. \tag*{(5)}

where R1jAB=RjB−rA\mathbf{R}_{1j}^{AB}=\mathbf{R}_{j}^{B}-\mathbf{r}^{A}, the notation “:” stands for the inner product of tensors of the same rank, and tensors C+AB=e1A⊗e1B+e2A⊗e2B\mathcal{C}_{+}^{AB}=\mathbf{e}_{1}^{A}\otimes\mathbf{e}_{1}^{B}+\mathbf{e}_{2}^{A}\otimes\mathbf{e}_{2}^{B}, C−AB=e1A⊗e1B−e2A⊗e2B\mathcal{C}_{-}^{AB}=\mathbf{e}_{1}^{A}\otimes\mathbf{e}_{1}^{B}-\mathbf{e}_{2}^{A}\otimes\mathbf{e}_{2}^{B}, and C⊥AB=e1A⊗e2B−e2A⊗e1B\mathcal{C}_{\perp}^{AB}=\mathbf{e}_{1}^{A}\otimes\mathbf{e}_{2}^{B}-\mathbf{e}_{2}^{A}\otimes\mathbf{e}_{1}^{B} characterize a twist between spin spirals in sublattices AA and BB. By expanding cosine and sine functions in Taylor series with respect to q⋅R\mathbf{q}\cdot\mathbf{R} (see Appendix C) we arrive at the following micromagnetic expression for the total energy of spin spirals with twist:

EAB(q)≈M(0)AB:CAB+12∑p=0∞M(p)AB:(C±AB⊗Q(p)),(6)E^{AB}(\mathbf{q})\approx\mathcal{M}^{(0)AB}:\mathcal{C}^{AB}+\frac{1}{2}\sum_{p=0}^{\infty}\mathcal{M}^{(p)AB}:\left(\mathcal{C}_{\pm}^{AB}\otimes\mathcal{Q}^{(p)}\right), \tag*{(6)}

where the sum over pp defines the order of Taylor expansion, C2pAB=C+AB\mathcal{C}_{2p}^{AB}=\mathcal{C}_{+}^{AB}, C2p+1AB=C−AB\mathcal{C}_{2p+1}^{AB}=\mathcal{C}_{-}^{AB}, and Q(p)=q⊗qp\mathcal{Q}^{(p)}=\mathbf{q}\otimes\mathbf{q}^{p} is the pp-fold tensor product of q\mathbf{q} vector with itself. The tensor M(p)AB\mathcal{M}^{(p)AB} of rank (p+1)(p+1) characterizes the exchange and DM interactions between all sites of sublattices AA and BB:

M(p)AB=(A(p)ABDz(p)AB−Dy(p)AB−Dz(p)ABA(p)ABDx(p)ABDy(p)AB−Dx(p)ABA(p)AB).(7)\mathcal{M}^{(p)AB} = \begin{pmatrix} \mathcal{A}^{(p)AB} & \mathcal{D}_{z}^{(p)AB} & -\mathcal{D}_{y}^{(p)AB} \\ -\mathcal{D}_{z}^{(p)AB} & \mathcal{A}^{(p)AB} & \mathcal{D}_{x}^{(p)AB} \\ \mathcal{D}_{y}^{(p)AB} & -\mathcal{D}_{x}^{(p)AB} & \mathcal{A}^{(p)AB} \end{pmatrix}. \tag*{(7)}

The components of M(p)AB\mathcal{M}^{(p)AB} are tensors of rank (p)(p) and they are related to JijAB\mathcal{J}_{ij}^{AB} and R1jAB\mathcal{R}_{1j}^{AB}:

Mkk′(2p)AB=(−1)p+1(2p)!∑j=1NBJ1jABR1j(2p)AB,(8)\mathcal{M}_{kk'}^{(2p)AB} = \frac{(-1)^{p+1}}{(2p)!} \sum_{j=1}^{N_B} \mathcal{J}_{1j}^{AB} \mathcal{R}_{1j}^{(2p)AB}, \tag*{(8)}
Mkk′(2p+1)AB=(−1)p+1(2p+1)!∑j=1NBJ1jABR1j(2p+1)AB.(9)\mathcal{M}_{kk'}^{(2p+1)AB} = \frac{(-1)^{p+1}}{(2p+1)!} \sum_{j=1}^{N_B} \mathcal{J}_{1j}^{AB} \mathcal{R}_{1j}^{(2p+1)AB}. \tag*{(9)}

where R1j(p)AB=(R1jAB)⊗p\mathcal{R}_{1j}^{(p)AB} = \left(\mathbf{R}_{1j}^{AB}\right)^{\otimes p} is the pp-fold tensor product of R1jAB\mathbf{R}_{1j}^{AB} with itself (see Appendix C). The micromagnetic tensors A(p)AB\mathcal{A}^{(p)AB} and D(p)AB=(Dx(p)AB,Dy(p)AB,Dz(p)AB)\mathcal{D}^{(p)AB} = \left(\mathcal{D}_{x}^{(p)AB},\mathcal{D}_{y}^{(p)AB},\mathcal{D}_{z}^{(p)AB}\right) in Eq. (7) describe the micromagnetic exchange and DM interactions, respectively, between all ions of sublattices AA and BB. Also, those tensors for different pairs of sublattices are related by symmetries of the underlying crystal structure and for B20 FeGe we list them in Appendix D.

The micromagnetic energy given by Eq. (6) is a central equation of this work and it can be decomposed into the exchange and DM interactions (see Appendix E). From this equation it is clear that the twist of the spin spirals gives rise to the energies of the exchange and DM interactions, which are both functions of Q(p)=⊗pq\mathbf{Q}^{(p)} = \otimes^p \mathbf{q}, where p=0,1,2,⋯p = 0, 1, 2, \cdots. This is in sharp contrast to trivial spin spirals, for which the exchange and DM interactions are functions of only Q(2p)\mathbf{Q}^{(2p)} and Q(2p+1)\mathbf{Q}^{(2p+1)}, respectively.

The main challenge is to find the twist tensors C±AB\mathcal{C}_{\pm}^{AB} and C±AB\mathcal{C}_{\pm}^{AB} minimizing the total energy. Using this model, we examine the twist of the spin-density waves in B20 FeGe, noting that it becomes difficult to carry out with full rigor, because of the 16 independent parameters (ϕA,βA,αA,θA)(\phi^{A},\beta^{A},\alpha^{A},\theta^{A}) characterizing the spin spiral’s twist in four sublattices (A=1,2,3,4)(A = 1, 2, 3, 4) of the crystal. On the other hand, the proposed micromagnetic model has many attractive qualitative features since it gives solutions in terms of a few micromagnetic multisublattice parameters taking into account the symmetries of the crystal structure, whereas the atomistic model becomes too difficult to carry out for a large supercell in which the magnetic structure is commensurate and due to the long-range character of the magnetic interactions in the itinerant magnets. For our micromagnetic model, we assume homogeneous spin density waves in each sublattice of the crystal; as we demonstrate by means of spin-dynamics simulations (see Appendix A), the gain in total energy due to the inhomogeneity of the spin-density waves is much smaller than that due to the twist.

Computational details

We compute the microscopic parameters of Heisenberg exchange JijJ_{ij} and DM Dij\mathbf{D}_{ij} interaction for B20 FeGe by employ-

ing first-principles calculations via multiple-scattering theory as implemented in the KKR Green’s function method [36, 37, 38]. In this framework, JijJ_{ij} and Dij\mathbf{D}_{ij} are obtained from the collinear state by applying infinitesimal rotations of the magnetic moments [40, 41]. All electronic structure calculations are performed for the experimental lattice parameter and atomic positions of FeGe [42, 43] in the local density approximation [44], using 72×72×7272 \times72 \times72 k-points in the full Brillouin zone with Fermi broadening of 100K. The micromagnetic parameters are computed using Eq. (8), for which the summation is truncated above a maximal interaction radius of Rmax⁡=12aR_{\max} = 12a, where aa is the lattice parameter.

First-principles results and discussions

The atomistic parameters of the exchange interaction JijJ_{ij} and the absolute values of the DM interaction vectors ∣Dij∣|\mathbf{D}_{ij}| as a function of the distance ∣Rij∣|\mathbf{R}_{ij}| between interacting pairs in B20 FeGe are shown in Figure 2 (a) and (Figure 2b). While both atomistic interaction parameters decay fast with ∣Rij∣|\mathbf{R}_{ij}|, their corresponding micromagnetic parameters, A(2)=12∑ijJijRij2A^{(2)} = \frac{1}{2}\sum_{ij} J_{ij}R_{ij}^{2} and D(1)=−∑ijRij⋅DijD^{(1)} = -\sum_{ij}\mathbf{R}_{ij}\cdot\mathbf{D}_{ij}, respectively (for more details see § 1.3) have diminishing oscillatory behavior with respect to the summation cutoff ∣rij∣<Rmax⁡|\mathbf{r}_{ij}| < R_{\max}, see Figure 2 (c) and (Figure 2d). The calculated value of the magnetic moment per Fe atom is 1.11 μB1.11\,\mu_{\mathrm{B}} [34], which is slightly larger than experimental value of ∼1.0 μB\sim1.0\,\mu_{\mathrm{B}} [18, 26, 42].

Figure 2. (a) Exchange interaction parameters JijJ_{ij} and (b) absolute values of DM interaction vectors ∣Dij∣|\mathbf{D}_{ij}| between Fe atoms as a function of the interatomic distance (between Fe atoms) ∣Rij∣|\mathbf{R}_{ij}| (in units of the lattice parameter aa). (c) Micromagnetic spin stiffness A(2)A^{(2)} and (d) micromagnetic spiralization D(1)D^{(1)} as a function of Rmax⁡R_{\max} (in units of the lattice parameter aa), up to which contributions from the atomistic parameters (Jij(J_{ij} and Dij\mathbf{D}_{ij} corresponding to all ∣RijAB∣<Rmax⁡)|\mathbf{R}_{ij}^{AB}| < R_{\max}) are included. Note that parameters JijJ_{ij} and ∣Dij∣|\mathbf{D}_{ij}| in (a) and (b) are multiplied by ∣Si∣∣Sj∣|\mathbf{S}_{i}||\mathbf{S}_{j}|. The inset in (b) presents cos⁡θij\cos\theta_{ij} for the first few shells, where θij\theta_{ij} is the angle between Rij\mathbf{R}_{ij} and Dij\mathbf{D}_{ij}.

Spin wave formed by identical spirals

We first consider the simplest case in which the spin spirals in all sublattices are flat (θA=π/2\theta^{A} = \pi/2), have the same phase ϕ\phi and the same direction of their rotation axes (erotA∥q\mathbf{e}_{\mathrm{rot}}^{A} \parallel\mathbf{q}). In this case ∣e3A∣=0|\mathbf{e}_{3}^{A}| = 0, emA∥erot\mathbf{e}_{m}^{A} \parallel\mathbf{e}_{\mathrm{rot}}, and Eq. (6) after the summation over all pairs of sublattices in the B20 structure gives

EI(q)=∑p=0∞A‾‾(2p):Q‾‾(2p)+D‾‾(2p+1):[e^rot⊗Q‾‾(2p+1)]=A(0)+D(1)q+A(2)q2+OI(q)(10)\begin{aligned} E_{\mathrm{I}}(\mathbf{q}) = \sum_{p=0}^{\infty} \underline{\underline{A}}^{(2p)}:\underline{\underline{Q}}^{(2p)}+\underline{\underline{D}}^{(2p+1)}:[\hat{\mathbf{e}}_{\mathrm{rot}}\otimes\underline{\underline{Q}}^{(2p+1)}] \\ &= A^{(0)}+D^{(1)}q+A^{(2)}q^{2}+O_{\mathrm{I}}(\mathbf{q}) \tag*{(10)} \end{aligned}

where q=∣q∣q=|\mathbf{q}| and OI(q)O_{\mathrm{I}}(\mathbf{q}) depicts the anisotropic higher-order contributions to the exchange and DM interaction, see Eq. (13). Tensors A‾‾(2p)=∑ABA‾‾(2p)AB\underline{\underline{A}}^{(2p)}=\sum_{AB}\underline{\underline{A}}^{(2p)AB} and D‾‾(2p+1)=∑ABD‾‾(2p+1)AB\underline{\underline{D}}^{(2p+1)}=\sum_{AB}\underline{\underline{D}}^{(2p+1)AB} characterize the micromagnetic exchange and DM interactions of order pp, respectively, between magnetic ions in whole crystal. Note that since A‾‾(2p+1)AB=−A‾‾(2p+1)BA\underline{\underline{A}}^{(2p+1)AB}=-\underline{\underline{A}}^{(2p+1)BA} and D‾‾(2p)AB=−D‾‾(2p)BA\underline{\underline{D}}^{(2p)AB}=-\underline{\underline{D}}^{(2p)BA}, these odd contributions to the total energy after the summation over all pairs of sublattices for such trivial spin spirals cancel out. The energy contributions from the even terms A‾‾(2p)AB=A‾‾(2p)BA\underline{\underline{A}}^{(2p)AB}=\underline{\underline{A}}^{(2p)BA} and D‾‾(2p+1)AB=D‾‾(2p+1)BA\underline{\underline{D}}^{(2p+1)AB}=\underline{\underline{D}}^{(2p+1)BA} for such magnetic order remain. For p≤2p\leq2 the obtained micromagnetic interaction tensors reduce to the identity matrices (see Appendix D),

A‾‾(2)=12I2∑AB∑i≠jJijAB∣RijAB∣2=A(2)I2,(11)\underline{\underline{A}}^{(2)}=\frac{1}{2}\mathcal{I}_{2}\sum_{AB}\sum_{i\neq j}\mathcal{J}_{ij}^{AB}|\mathbf{R}_{ij}^{AB}|^{2}=A^{(2)}\mathcal{I}_{2}, \tag*{(11)}
D‾‾(1)=−I2∑AB∑i≠jDijAB⋅RijAB=D(1)I2.(12)\underline{\underline{D}}^{(1)}=-\mathcal{I}_{2}\sum_{AB}\sum_{i\neq j}\mathbf{D}_{ij}^{AB}\cdot\mathbf{R}_{ij}^{AB}=D^{(1)}\mathcal{I}_{2}. \tag*{(12)}

As follows from Eq. (12), each bond RijAB\mathbf{R}_{ij}^{AB} has the largest contribution to the spiralization parameter D(1)D^{(1)} if DijAB∥RijAB\mathbf{D}_{ij}^{AB}\parallel\mathbf{R}_{ij}^{AB}, and it is zero if DijAB⊥RijAB\mathbf{D}_{ij}^{AB}\perp\mathbf{R}_{ij}^{AB}. Note that while ∣Dij∣|\mathbf{D}_{ij}| are largest for the first few nearest neighbors, see Figure 2(b), their contribution to the total energy of the homogeneous flat spin spiral ∼Rij⋅Dij∼cos⁡θij\sim\mathbf{R}_{ij}\cdot\mathbf{D}_{ij}\sim\cos\theta_{ij}, where θij\theta_{ij} is an angle between Rij\mathbf{R}_{ij} and Dij\mathbf{D}_{ij}, is negligibly small, see the inset in Figure 2(b) and Figure 2(d). In contrast, as demonstrated by Chizhikov and Dmitrienko [31, 32, 33], such normal components of the DM vectors, Dij⊥Rij\mathbf{D}_{ij}\perp\mathbf{R}_{ij}, might give rise to the twist between spin spirals formed by different sublattices, which we discuss in the following sections.

While the lower-order contributions to the total energy, Eq. (10), of the exchange (p=0,2p=0,2) and DM (p=1p=1) interactions do not depend on the direction of the wave vector q\mathbf{q} [45], the higher-order contributions are anisotropic

OI(q)=D1(3)(qx2,qy2,qz2)⋅q+D2(3)(qz2,qx2,qy2)⋅qO_{\mathrm{I}}(\mathbf{q})=D_{1}^{(3)}(q_{x}^{2},q_{y}^{2},q_{z}^{2})\cdot\mathbf{q}+D_{2}^{(3)}(q_{z}^{2},q_{x}^{2},q_{y}^{2})\cdot\mathbf{q}
OI(q)=+D3(3)(qy2,qz2,qx2)⋅q+A1(4)(qx4+qy4+qz4)\phantom{O_{\mathrm{I}}(\mathbf{q})=}+D_{3}^{(3)}(q_{y}^{2},q_{z}^{2},q_{x}^{2})\cdot\mathbf{q}+A_{1}^{(4)}(q_{x}^{4}+q_{y}^{4}+q_{z}^{4})
OI(q)=+A2(4)(qx2qy2+qx2qz2+qy2qz2)+⋯ .(13)\phantom{O_{\mathrm{I}}(\mathbf{q})=}+A_{2}^{(4)}(q_{x}^{2}q_{y}^{2}+q_{x}^{2}q_{z}^{2}+q_{y}^{2}q_{z}^{2})+\cdots. \tag*{(13)}

The corresponding micromagnetic parameters for these higher-order terms, Dn(3)D_{n}^{(3)} and An(4)A_{n}^{(4)}, are given in Appendix D. Since the contribution to the total energy of these terms for small values of ∣q∣|\mathbf{q}| is small, ∣OI(qmin⁡)∣<0.1 (μeV)≪∣EI(qmin⁡)∣≈9 (μeV)|O_{\mathrm{I}}(\mathbf{q}_{\min})|<0.1\ (\mu\mathrm{eV})\ll|E_{\mathrm{I}}(\mathbf{q}_{\min})|\approx9\ (\mu\mathrm{eV}), we neglect them. Also, as follows from Figure 3(a), energies obtained for q001\mathbf{q}_{001} and q111\mathbf{q}_{111} using Eq. (10) without OI(q)O_{\mathrm{I}}(\mathbf{q}) contributions (black solid lines) are in good agreement with the atomistic model (black symbols) given by Eq. (2).

Figure 3. (a) The total energy (per unit cell) of (b) the exchange and DM interactions, computed using the atomistic (circles) and micromagnetic (solid lines) models for the superposition of helical spin-density waves in B20 FeGe without twist (in black) and with twist (in blue and red) as a function of q=∣q∣q = |\mathbf{q}|. The twist parameters of the spirals in each sublattice AA minimizing the total energy {ϕA}\{\phi^{A}\} (blue curve) and {ϕA,βA,αA}\{\phi^{A}, \beta^{A}, \alpha^{A}\} (red curves) for (c) q001\mathbf{q}_{001} and (d) q111\mathbf{q}_{111}. (e) Illustration showing the relative orientations of the rotation axes erotA\mathbf{e}_{\mathrm{rot}}^{A} of the spin spirals in each sublattice AA for q001\mathbf{q}_{001} and q111\mathbf{q}_{111}. Vertical arrows in (a) show the energy difference between the helical ground state and the FM state, corresponding to saturation magnetic fields of 0.04 and 0.12 T for spin spirals without and with twists, respectively. The solid green line stands for the experimental pitch ∣qexp⁡∣=9 μm−1|\mathbf{q}^{\exp}| = 9\ \mu\mathrm{m}^{-1}. δE\delta E is the total energy difference of the spin spiral with twist at qmin⁡\mathbf{q}^{\min} (the theoretical pitch) and qexp⁡\mathbf{q}^{\exp}.

Spin spirals with a phase difference

Now we consider a case in which flat spirals (θA=π/2\theta^A=\pi/2) in each sublattice AA are allowed to have different phases ϕA\phi^A while the directions of the rotation axes erotA=erot(β,α)=q/∣q∣\mathbf{e}_{\mathrm{rot}}^A=\mathbf{e}_{\mathrm{rot}}(\beta,\alpha)=\mathbf{q}/|\mathbf{q}| remain the same. For such a choice of the magnetic structure the total energy of the magnetic interactions between atoms in sublattices AA and BB, Eq. (6), has the form

EIIAB(q,ϕAB)=cos⁡ϕAB∑p(A(2p)AB:Q(2p)+D(2p+1)AB:Q(2p+1)⊗erot)+sin⁡ϕAB∑p(−A(2p+1)AB:Q(2p+1)+D(2p)AB:Q(2p)⊗erot)(14)E_{\mathrm{II}}^{AB}(\mathbf{q},\phi^{AB})=\cos\phi^{AB}\sum_p\left(\mathcal{A}^{(2p)AB}:\mathcal{Q}^{(2p)}+\mathcal{D}^{(2p+1)AB}:\mathcal{Q}^{(2p+1)}\otimes\mathbf{e}_{\mathrm{rot}}\right) +\sin\phi^{AB}\sum_p\left(-\mathcal{A}^{(2p+1)AB}:\mathcal{Q}^{(2p+1)}+\mathcal{D}^{(2p)AB}:\mathcal{Q}^{(2p)}\otimes\mathbf{e}_{\mathrm{rot}}\right) \tag*{(14)}

where ϕAB=ϕB−ϕA\phi^{AB}=\phi^B-\phi^A is the phase difference between spin spirals in sublattices AA and BB. Note that if ϕAB=0\phi^{AB}=0, then EIIAB(q)=EIAB(q)E_{\mathrm{II}}^{AB}(\mathbf{q})=E_{\mathrm{I}}^{AB}(\mathbf{q}), see Eq. (10). As follows from Eq. (14), the phase difference ϕmin⁡AB(q)\phi_{\min}^{AB}(\mathbf{q}) corresponding to the lowest total energy of the magnetic interactions between two sublattices

ϕmin⁡AB(q)=arctan⁡[D(0)AB⋅erot−A(1)AB⋅q+⋯A(0)AB+erotT⋅D(1)AB⋅q+⋯]\phi_{\min}^{AB}(\mathbf{q})=\arctan\left[\frac{\mathcal{D}^{(0)AB}\cdot\mathbf{e}_{\mathrm{rot}}-\mathcal{A}^{(1)AB}\cdot\mathbf{q}+\cdots}{\mathcal{A}^{(0)AB}+\mathbf{e}_{\mathrm{rot}}^{T}\cdot\mathcal{D}^{(1)AB}\cdot\mathbf{q}+\cdots}\right]

is nonzero if D(2p)AB\mathcal{D}^{(2p)AB} or A(2p+1)AB\mathcal{A}^{(2p+1)AB} are nonzero tensors. To explain the emergence of the phase difference let us assume first only two spin spirals formed by sublattices AA and BB. When each atom ii in sublattices AA and BB possesses an inversion symmetry with respect to their relative positions, RiiAB=−Rii−1AB\mathbf{R}_{ii}^{AB}=-\mathbf{R}_{ii-1}^{AB}, and magnetic interactions, JijAB=(Jii−1AB)T\mathcal{J}_{ij}^{AB}=(\mathcal{J}_{ii-1}^{AB})^{T}, as shown in Figure 4(a), then the summation in Eq. (8) gives D(2p)AB=A(2p+1)AB=0I2p+1\mathcal{D}^{(2p)AB}=\mathcal{A}^{(2p+1)AB}=0\mathbb{I}_{2p+1}; therefore, ϕmin⁡AB(q)=0\phi_{\min}^{AB}(\mathbf{q})=0. However, when atoms in one of the sublattices are displaced from their symmetric positions by δ\boldsymbol{\delta}, see Figure 4(b), then ϕmin⁡AB(q)=δ⋅q\phi_{\min}^{AB}(\mathbf{q})=\boldsymbol{\delta}\cdot\mathbf{q} if JiiAB=(Jii−1AB)T\mathcal{J}_{ii}^{AB}=(\mathcal{J}_{ii-1}^{AB})^{T} and the magnetic moments remain the same directions as before the displacement (SiB=SiB′\mathbf{S}_i^B=\mathbf{S}_i^{B'}). If JijAB≠(Jii−1AB)T\mathcal{J}_{ij}^{AB}\ne(\mathcal{J}_{ii-1}^{AB})^{T}, then SiB≠SiB′\mathbf{S}_i^B\ne\mathbf{S}_i^{B'}, and ϕmin⁡AB(q)≠δ⋅q\phi_{\min}^{AB}(\mathbf{q})\ne\boldsymbol{\delta}\cdot\mathbf{q} in general.

Illustrations of spin-density waves and displacements of Fe and Ge ions

Figure 4. (a) and (b) Illustrations of the spin-density wave formed by spin spirals in sublattices AA and BB. Spin spirals in (a) are identical, and those in (b) have phase difference ϕAB=δ⋅q\phi^{AB}=\boldsymbol{\delta}\cdot\mathbf{q} due to the displacement of atoms in sublattice B→B′B\rightarrow B' by δ\boldsymbol{\delta} such that SiB=SiB′\mathbf{S}_i^B=\mathbf{S}_i^{B'}. (c) Displacements of Fe and Ge ions from high-symmetry sites of the NaCl structure (smaller spheres) to lower-symmetry sites of the B20 structure (larger spheres) along cube diagonals nA\mathbf{n}^A, as shown in (d).

Similarly, the emergence of the relative phases ϕAB\phi^{AB} between spin spirals in different sublattices of the B20 structure is expected because atoms in this structure are displaced with respect to the high-symmetry sites [given by space group Fm3mFm3m, see Figure 4(c)]. Such displacement of atoms in each sublattice AA can be defined as δA=nAuM\boldsymbol{\delta}^A=\mathbf{n}^A u^M, where the parameter uMu^M stands for the Wyckoff positions of magnetic ions in the B20 structure (space group P213P2_13) and nA\mathbf{n}^A is the symmetry direction at each site in sublattice AA (see Appendix D). Finding phase ϕA\phi^A for each spiral AA in a B20 chiral magnet for a given direction of q\mathbf{q} requires taking into account the magnetic interactions between all pairs of sublattices. Since only a phase difference ϕAB\phi^{AB} enters Eq. (14), we are allowed to choose ϕ1=0\phi^1=0. Assuming erot∥q\mathbf{e}_{\mathrm{rot}}\parallel\mathbf{q} and minimizing (numerically) the total energy, Eq. (14), for q001\mathbf{q}_{001} and q111\mathbf{q}_{111} with respect to the phases ϕA\phi^A, we obtain

ϕ1=ϕ2=0,ϕ3=ϕ4=φ001(q001),\phi^1=\phi^2=0,\quad\phi^3=\phi^4=\varphi_{001}(\mathbf{q}_{001}),
ϕ1=0,ϕ2=ϕ3=ϕ4=φ111(q111)(15)\phi^1=0,\quad\phi^2=\phi^3=\phi^4=\varphi_{111}(\mathbf{q}_{111}) \tag*{(15)}

Therefore, due to the symmetries of the B20 structure the only unknown parameter φ(q)\varphi(\mathbf{q}) remains for the two cases of q\mathbf{q}. The above equations indicate also that the phase difference ϕAB\phi^{AB} between spirals in sublattices AA and BB can be defined in terms of the relative atomic displacements from the high-symmetry sites projected on erot∥q\mathbf{e}_{\mathrm{rot}}\parallel\mathbf{q}:

ϕmin⁡AB(q)=erot⋅(nB−nA)φ(q)\phi_{\min}^{AB}(\mathbf{q})=\mathbf{e}_{\mathrm{rot}}\cdot(\mathbf{n}^B-\mathbf{n}^A)\varphi(\mathbf{q})

Since, due to the symmetries, the only unknown parameter φ(q)\varphi(\mathbf{q}) remains, the total energy in Eq. (14) can be easily minimized, and φmin⁡(q)\varphi_{\min}(\mathbf{q}) that includes the interactions between magnetic atoms in all sublattices of the crystal can be computed. The energies of the exchange and DM interactions as a function of ∣q∣|\mathbf{q}| and φmin⁡(q)\varphi_{\min}(\mathbf{q}) for q001\mathbf{q}_{001} or q111\mathbf{q}_{111} are shown as blue circles (atomistic model) and blue lines (micromagnetic model) in Figure 3(a) and Figure 3(b). The corresponding φmin⁡(q)\varphi_{\min}(\mathbf{q}) for q001\mathbf{q}_{001} and q111\mathbf{q}_{111} are shown in Figure 3(c) and Figure 3(d), respectively. As follows from Figure 3 the energies corresponding to φmin⁡(q)\varphi_{\min}(\mathbf{q}) for the two directions of q\mathbf{q} are almost the same (a small difference is due to the higher-order contributions; see § 1.3). The obtained energies are lower than in the case of φ(q)=0\varphi(\mathbf{q})=0, EIIAB(qmin⁡,φmin⁡)<EIAB(qmin⁡)E_{\mathrm{II}}^{AB}(\mathbf{q}_{\min},\varphi_{\min})<E_{\mathrm{I}}^{AB}(\mathbf{q}_{\min}), as the energy gain of the DM interaction dominates the energy loss of the exchange contribution, see Figure 3(b). Also, the obtained phase difference φmin⁡(qmin⁡)≈−0.8∘\varphi_{\min}(\mathbf{q}_{\min})\approx-0.8^\circ, is of the same order as the angle between magnetic moments of nearest atoms formed due to the global spiraling of ≈1.28∘\approx1.28^\circ [46].

Spirals with different cones

Now we consider a case in which in addition to the phase difference φAB\varphi^{AB} between spin spirals each of them is allowed to have a cone angle θA\theta^A between each SiA\mathbf{S}_i^A and erotA\mathbf{e}_{\mathrm{rot}}^A that might differ from π/2\pi/2, see Figure 5 (a). The total energy of the magnetic interactions for such magnetic structure has the form

Illustration of the superposition of spin spirals A and B with different cone angles and different orientations of the rotation axes

Figure 5. Illustration of the superposition of spin spirals AA and BB with: (a) different cone angles and (b) different orientations of the rotation axes. The rainbow colors for the vectors represent different orientations of the vector chirality χiAB=SiA×SiB\boldsymbol{\chi}_i^{AB} = \mathbf{S}_i^A \times\mathbf{S}_i^B component normal to q\mathbf{q}.

EIIIAB(q,φAB,θA,θB)=cos⁡θAcos⁡θBEFMAB+sin⁡θAsin⁡θBEIIAB(q,φAB)E_{\mathrm{III}}^{AB}\left(\mathbf{q},\varphi^{AB},\theta^A,\theta^B\right) = \cos\theta^A\cos\theta^B E_{\mathrm{FM}}^{AB} + \sin\theta^A\sin\theta^B E_{\mathrm{II}}^{AB}\left(\mathbf{q},\varphi^{AB}\right)

where EFMAB=−∑ijJijAB=A(0)ABE_{\mathrm{FM}}^{AB}=-\sum_{ij}J_{ij}^{AB}=\mathcal{A}^{(0)AB} and EIIAB(q,φAB)E_{\mathrm{II}}^{AB}\left(\mathbf{q},\varphi^{AB}\right) are the total energies, respectively, of the FM state and spin-spiral state with phase difference φAB\varphi^{AB}, see Eq. (14). In such a case the chirality vectors χijAB=SiA×SjB\boldsymbol{\chi}_{ij}^{AB}=\mathbf{S}_i^A\times\mathbf{S}_j^B of the spin-spiral state have components that are normal and parallel to the direction of the wave vector q\mathbf{q}, χijAB=χij∥qAB+χij⊥qAB\boldsymbol{\chi}_{ij}^{AB}=\boldsymbol{\chi}_{ij\parallel\mathbf{q}}^{AB}+\boldsymbol{\chi}_{ij\perp\mathbf{q}}^{AB}, see Figure 5 (a). Case θA=θB=0∘\theta^A=\theta^B=0^\circ (θA=θB=90∘\theta^A=\theta^B=90^\circ) corresponds to the FM (flat spin spiral) state. Any deviations of the cone angles θA\theta^A and θB\theta^B from π/2\pi/2 reduces the energy contribution of EIIAB(q,φAB)E_{\mathrm{II}}^{AB}\left(\mathbf{q},\varphi^{AB}\right) and the total energy is in between the total energy of the FM state and the total energy of the flat spin spirals. This is in particular because conical spin spirals reduce the contribution to the total energy of the DM vectors parallel to q\mathbf{q}, as Dij⋅χij∥qAB=sin⁡θAsin⁡θBDij⋅χijAB\mathbf{D}_{ij}\cdot\boldsymbol{\chi}_{ij\parallel\mathbf{q}}^{AB}=\sin\theta^A\sin\theta^B\mathbf{D}_{ij}\cdot\boldsymbol{\chi}_{ij}^{AB}. Also, the components of DM vectors normal to q\mathbf{q} on average over the magnetic unit cell give zero contribution to the total energy, as vector chirality components χij⊥qAB\boldsymbol{\chi}_{ij\perp\mathbf{q}}^{AB} rotate by 2π2\pi over the whole magnetic unit cell, see Figure 5 (a). From this, we conclude that the competition between the exchange and DM interactions favors the emergence of flat spin spirals rather than those with θA≠π/2\theta^A\ne\pi/2.

Spirals with different rotation axes

Finally, we consider a case in which flat spirals in each sublattice in addition to the phase difference φAB\varphi^{AB} are allowed to have different orientations of their rotation axes erotA\mathbf{e}_{\mathrm{rot}}^A, characterized by angles βA\beta^A and αA\alpha^A as illustrated in Figure 1. Like in the case discussed above, different orientations of the rotation axes in two sublattices give rise to χij∥qAB\boldsymbol{\chi}_{ij\parallel\mathbf{q}}^{AB} and χij⊥qAB\boldsymbol{\chi}_{ij\perp\mathbf{q}}^{AB} vector chirality components, see Figure 5 (b). While the component χij∥qAB\boldsymbol{\chi}_{ij\parallel\mathbf{q}}^{AB} caused by the DM interaction vectors parallel to q\mathbf{q} gives rise to the global spiraling, the components χij⊥qAB\boldsymbol{\chi}_{ij\perp\mathbf{q}}^{AB} might arise due to the DM interaction vectors DijAB\mathbf{D}_{ij}^{AB} normal to q\mathbf{q}. Although the contribution of DM interaction from χij⊥qAB\boldsymbol{\chi}_{ij\perp\mathbf{q}}^{AB} varies over the magnetic unit cell, it does not vanish as χij⊥qAB\boldsymbol{\chi}_{ij\perp\mathbf{q}}^{AB} rotates around q\mathbf{q} by only π\pi.

The directions of the spin-spirals rotation axes formed by different sublattices of the B20 structure can be obtained from the minimization of the total energy, Eq. (6), for a given wave vector q\mathbf{q}. In this case the total energy of the magnetic interactions is a function of 13 parameters (q,ϕA,βA,αA)(\mathbf{q},\phi^A,\beta^A,\alpha^A) for A=1,2,3,4A=1,2,3,4. As we will demonstrate, the number of unknown parameters can be reduced due to symmetries, but let us first briefly discuss the case ∣q∣=0|\mathbf{q}|=0, for which the total energy given by Eq. (6) has the form

EIV(0)=12∑AB(e1A⋅e1B+e2A⋅e2B)A(0)AB+12∑AB(e1A×e1B+e2A×e2B)⋅D(0)AB(16)E_{\mathrm{IV}}^{(0)}=\frac{1}{2}\sum_{AB}\left(\mathbf{e}_1^A\cdot\mathbf{e}_1^B+\mathbf{e}_2^A\cdot\mathbf{e}_2^B\right)\mathcal{A}^{(0)AB} +\frac{1}{2}\sum_{AB}\left(\mathbf{e}_1^A\times\mathbf{e}_1^B+\mathbf{e}_2^A\times\mathbf{e}_2^B\right)\cdot\mathbf{D}^{(0)AB} \tag*{(16)}

From this equation it follows that the contribution from the DM interaction takes maximal values when erotA=emA⊥emB=erotB\mathbf{e}_{\mathrm{rot}}^A=\mathbf{e}_m^A\perp\mathbf{e}_m^B=\mathbf{e}_{\mathrm{rot}}^B. Note that in the case with erotA=emA⊥emB=erotB\mathbf{e}_{\mathrm{rot}}^A=\mathbf{e}_m^A\perp\mathbf{e}_m^B=\mathbf{e}_{\mathrm{rot}}^B we have (emA×emB)=erotAsin⁡φAB(\mathbf{e}_m^A\times\mathbf{e}_m^B)=\mathbf{e}_{\mathrm{rot}}^A\sin\varphi^{AB}, and only D(0)AB∥erotA\mathbf{D}^{(0)AB}\parallel\mathbf{e}_{\mathrm{rot}}^A contributes to the total energy for ∣q∣=0|\mathbf{q}|=0 (see § 1.3). However, since D(0)AB∦D(0)AB\mathbf{D}^{(0)AB}\not\parallel\mathbf{D}^{(0)AB} in B20 FeGe (see Table III (PDF p. 11) in Appendix D), the DM interaction, in addition to the phase φAB\varphi^{AB}, might lead to erotB∥erotC\mathbf{e}_{\mathrm{rot}}^B\parallel\mathbf{e}_{\mathrm{rot}}^C.

M‾‾(p)AB\underline{\underline{\mathcal{M}}}^{(p)AB}AA-BB(x,y,z)(x,y,z)AA-BB(x,y,z)(x,y,z)AA-BB(x,y,z)(x,y,z)AA-AA(x,y,z)(x,y,z)
D‾‾(0)AB\underline{\underline{\mathcal{D}}}^{(0)AB} (meV)1-2(−0.43,0.76,0)(-0.43,0.76,0)1-3(0,−0.43,0.76)(0,-0.43,0.76)1-4(0.76,0,−0.43)(0.76,0,-0.43)1-1(0,0,0)(0,0,0)
2-1(0.43,−0.76,0)(0.43,-0.76,0)2-4(0,0.43,0.76)(0,0.43,0.76)2-3(−0.76,0,−0.43)(-0.76,0,-0.43)2-2(0,0,0)(0,0,0)
3-4(−0.43,−0.76,0)(-0.43,-0.76,0)3-1(0,0.43,−0.76)(0,0.43,-0.76)3-2(0.76,0,0.43)(0.76,0,0.43)3-3(0,0,0)(0,0,0)
4-3(0.43,0.76,0)(0.43,0.76,0)4-2(0,−0.43,−0.76)(0,-0.43,-0.76)4-1(−0.76,0,0.43)(-0.76,0,0.43)4-4(0,0,0)(0,0,0)
A‾‾(1)AB\underline{\underline{\mathcal{A}}}^{(1)AB} (meVÅ)1-2(−22.39,17.99,0)(-22.39,17.99,0)1-3(0,−22.39,17.99)(0,-22.39,17.99)1-4(17.99,0,−22.39)(17.99,0,-22.39)1-1(0,0,0)(0,0,0)
2-1(22.39,−17.99,0)(22.39,-17.99,0)2-4(0,22.39,17.99)(0,22.39,17.99)2-3(−17.99,0,−22.39)(-17.99,0,-22.39)2-2(0,0,0)(0,0,0)
3-4(−22.39,−17.99,0)(-22.39,-17.99,0)3-1(0,22.39,−17.99)(0,22.39,-17.99)3-2(17.99,0,22.39)(17.99,0,22.39)3-3(0,0,0)(0,0,0)
4-3(22.39,17.99,0)(22.39,17.99,0)4-2(0,−22.39,−17.99)(0,-22.39,-17.99)4-1(−17.99,0,22.39)(-17.99,0,22.39)4-4(0,0,0)(0,0,0)

Table III (PDF p. 11). Micromagnetic tensors D‾‾(0)AB\underline{\underline{\mathcal{D}}}^{(0)AB} and A‾‾(1)AB\underline{\underline{\mathcal{A}}}^{(1)AB} between different pairs of sublattices AA and BB of B20 FeGe.

The energies (neglecting small higher-order anisotropic contributions) of the magnetic interactions in B20 FeGe minimized numerically with respect to the twist angles (ϕA,βA,αA)(\phi^A,\beta^A,\alpha^A) as a function of q001\mathbf{q}_{001} and q111\mathbf{q}_{111} are shown as red lines in Figure 3(a) and Figure 3(b). While the minimized total energy does not depend on the direction of q\mathbf{q}, the corresponding twist’s angles of the spirals in each sublattice AA depend on it, such that

∑A4nA⋅erotA=0,(∑A4erotA)∥q\sum_A^4 \mathbf{n}^A\cdot\mathbf{e}_{\mathrm{rot}}^A=0,\qquad\left(\sum_A^4\mathbf{e}_{\mathrm{rot}}^A\right)\parallel\mathbf{q}
φmin⁡AB(q)=(nB⋅erotB−nA⋅erotA)φ(q)\varphi_{\min}^{AB}(\mathbf{q})=\left(\mathbf{n}^B\cdot\mathbf{e}_{\mathrm{rot}}^B-\mathbf{n}^A\cdot\mathbf{e}_{\mathrm{rot}}^A\right)\varphi(\mathbf{q})

As the result, only three parameters (ϕ,β,α)(\phi,\beta,\alpha) out of 12 {ϕA,βA,αA}\{\phi^A,\beta^A,\alpha^A\} remain independent. As an example, in the case of q001\mathbf{q}_{001} we obtain

ϕ1=ϕ2=0,ϕ3=ϕ4=ϕ(q),\phi^1=\phi^2=0,\qquad\phi^3=\phi^4=\phi(\mathbf{q}),
β1=β2=β3=β4=β(q),\beta^1=\beta^2=\beta^3=\beta^4=\beta(\mathbf{q}),
α1=α(q),α2=α(q)+π,\alpha^1=\alpha(\mathbf{q}),\qquad\alpha^2=\alpha(\mathbf{q})+\pi,
α4=−α(q),α3=−α(q)−π(17)\alpha^4=-\alpha(\mathbf{q}),\qquad\alpha^3=-\alpha(\mathbf{q})-\pi \tag*{(17)}

A similar result can be shown for q111\mathbf{q}_{111}. In this case erot1∥q∥n1\mathbf{e}_{\mathrm{rot}}^1\parallel\mathbf{q}\parallel\mathbf{n}^1, erot2⋅q=erot3⋅q=erot4⋅q\mathbf{e}_{\mathrm{rot}}^2\cdot\mathbf{q}=\mathbf{e}_{\mathrm{rot}}^3\cdot\mathbf{q}=\mathbf{e}_{\mathrm{rot}}^4\cdot\mathbf{q} and erot2⋅n2=erot3⋅n3=erot4⋅n4\mathbf{e}_{\mathrm{rot}}^2\cdot\mathbf{n}^2=\mathbf{e}_{\mathrm{rot}}^3\cdot\mathbf{n}^3=\mathbf{e}_{\mathrm{rot}}^4\cdot\mathbf{n}^4. The orientations of the rotation axes for the spin spirals in each sublattice of B20 FeGe for q001q_{001} and q111q_{111} are shown in Figure 3(e). The energy difference of about 0.028 meV between the helical ground state with twist and the FM state is three times lower than in the case without twist, corresponding to a saturation magnetic field of B=0.12B = 0.12 T [47], which is in good agreement with the experimental value obtained for FeGe [18, 26, 48].

Conclusions

In this work, we studied the magnetic interactions between sublattices of non-centrosymmetric crystal using the micromagnetic model. In particular, we showed that due to the exchange and DM interactions the ground state spin spirals formed by sublattices of B20 FeGe have different phases and different orientations of their rotation axis (see Eq. (1)). We developed a micromagnetic model describing the exchange and DM interactions for such twisted spin-density waves (Eq. (6)), which gives a solution in terms of a few micromagnetic multi-sublattice interaction parameters taking into account symmetries of the crystal structure, whereas the atomistic model becomes too difficult to carry out due to the long range of the magnetic interactions. To uncover the importance of the spin-spiral twist for the energetics of the materials with multi-sublattices without inversion symmetry we compute the micromagnetic exchange and DM interaction parameters for B20 FeGe, as other higher-order interactions in this compound are expected to be small [34, 35]. From the minimization of the micromagnetic total energy, we obtain the ground state parameters characterizing the twist of the spin-density waves in each sublattice of the crystal. In particular, we show that the magnitude of this effect in B20 FeGe is of the same order as global spiraling and, it lowers the total energy (compare to the FM state) three times compared to the case when it is ignored.

While the twist of the spin spirals in B20 FeGe significantly reduces the total energy of the exchange and DM interactions, the obtained period of the magnetic modulations λmin⁡DFT=2π/qmin⁡DFT\lambda_{\min}^{\mathrm{DFT}} = 2\pi/q_{\min}^{\mathrm{DFT}} remains two times larger than the experimental observation λmin⁡exp=2π/qmin⁡exp\lambda_{\min}^{\mathrm{exp}} = 2\pi/q_{\min}^{\mathrm{exp}}, see Figure 3(a). A possible reason for this discrepancy might stem from the failure of DFT to capture the effect of electronic correlations which can dramatically change the overall behavior of the system, as was recently demonstrated using dynamical mean-field theory [49]. While the exchange interaction parameters JijJ_{ij} obtained by DFT explain the experimental Curie temperature of FeGe very well [34], the electronic correlations might have a stronger impact on the strength of the DM interaction. As shown in Figure 7 in Appendix A, not only does the period of the spin spiral become shorter when the strength of the DM interaction is increased, but also the corresponding twist angles ϕmin⁡\phi_{\min} and βmin⁡\beta_{\min} become larger, lowering the total energy more prominently compared with the case when the twist is ignored. On the other hand, since the difference in total energy for qmin⁡DFTq_{\min}^{\mathrm{DFT}} and qmin⁡expq_{\min}^{\mathrm{exp}} due to the twist is small, E(qmin⁡DFT,Ω)−E(qmin⁡exp,Ω)=8μE(q_{\min}^{\mathrm{DFT}},\Omega)-E(q_{\min}^{\mathrm{exp}},\Omega)=8\mueV (see Figure 3(a)), we speculate that it can be overtaken by higher-order magnetic interactions [35] or by quantum fluctuations [50] shifting qmin⁡DFTq_{\min}^{\mathrm{DFT}} towards qmin⁡expq_{\min}^{\mathrm{exp}}.

Energy, period, and twist angles as functions of the DMI scaling factor

Figure 7. (a) and (b) The total energy of the exchange and DM interaction as a function of spin spiral period λ\lambda, computed for q∥(001)q \parallel(001) using scaling factors ηDMI=1\eta_{\mathrm{DMI}} = 1 and ηDMI=2\eta_{\mathrm{DMI}} = 2, respectively, for the length of the DM interaction vectors Dij\mathbf{D}_{ij} of FeGe. The energies of the SD simulations (large gray circles) and atomistic model (small circles) are obtained using the effective J~ijAB\widetilde{J}_{ij}^{AB} and J~ijAB\widetilde{\mathbf{J}}_{ij}^{AB} interaction parameters, respectively. The micromagnetic energies are minimized for spin spirals without twist (blue lines) and with a twist (red lines). The diamonds indicate λmin⁡\lambda_{\min} corresponding to the minimum of the total energy Emin⁡E_{\min}. (c) Period λmin⁡\lambda_{\min}, (d) energy minimum Emin⁡E_{\min}, and (e) corresponding twist angles (φmin⁡,βmin⁡,αmin⁡)(\varphi_{\min},\beta_{\min},\alpha_{\min}) of the spin spirals with twist as a function of the scaling factor ηDMI\eta_{\mathrm{DMI}} for the length of Dij\mathbf{D}_{ij}.

Acknowledgments

We thank F. Rybakov and N. Kiselev for fruitful discussions. We acknowledge financial support from the DARPA TEE program through a MIPR grant (#HR0011831554) from DOI and from Deutsche Forschungsgemeinschaft (DFG) through SPP 2137 “Skyrmionics” (Project BL 444/16) and the Collaborative Research Centers SFB 1238 (Project C01), as well as computing resources at the supercomputers JURECA at Jülich Supercomputing Centre and JARA-HPC from RWTH Aachen University (Projects jara0224 and jara3dmagic).

A Spin-dynamic simulations

Here, by employing atomistic spin-dynamics simulations [51], we test whether the exchange and DM interactions in B20 chiral magnets give rise to the twist of the spin-density waves, as described by Eq. (1). To prove the concept and to reduce the computational burden we consider first the magnetic interactions up to the fourth nearest neighbors, ∣RijAB∣≤Rmax⁡=a|R_{ij}^{AB}| \le R_{\max}=a, where aa is the lattice parameter. We use interaction parameters obtained for B20 FeGe and to reduce the size of the magnetic unit cell we multiply the DM interaction vectors by a factor of 10, see Figure 6(a). We perform atomistic spin-dynamics simulations at zero temperature using cubic supercells of size Nλ×Nλ×NλN_{\lambda} \times N_{\lambda} \times N_{\lambda}, where Nλ∈{30,…,60}N_{\lambda} \in\{30,\ldots,60\}. Our simulations confirm that the spin spirals form a twist at the ground state in all chosen supercells. Also, no modulations within the same sublattice in directions normal to the direction of q001q_{001} were observed; thus, a one-dimensional magnetic supercell 1×1×Nλ1 \times1 \times N_{\lambda} is sufficient.

Figure 6. (a) Exchange interaction parameters JijJ_{ij} and absolute values of DM interaction vectors ∣Dij∣|D_{ij}| between Fe atoms as a function of the interatomic distance ∣Rij∣|R_{ij}| (in units of the lattice parameter aa). (b) The total energy of the exchange and DM interaction, computed using the micromagnetic model (solid lines), atomistic model (circles), and spin-dynamics (SD) simulations (diamonds) for the spin spiral propagating along the (001) direction as a function of λ=2π/∣q∣\lambda= 2\pi/|q|. The black curve stands for the spin spirals without twist. Blue and red curves represent twisted spin spirals characterized by (φ)(\varphi) and (φ,αA,β)(\varphi,\alpha^{A},\beta), respectively, see Eqs. (15) and (17). (c) Twist parameters φ\varphi, β\beta, and αA\alpha^{A} minimizing the micromagnetic total energy as a function of λ\lambda. (d) Difference between the twist parameters obtained from the spin-dynamics simulations ΩiA={φiA,βiA,αiA}=ΩA(ziA)\Omega_i^{A} = \{\varphi_i^{A},\beta_i^{A},\alpha_i^{A}\} = \Omega^{A}(z_i^{A}) and twist parameters obtained from the micromagnetic model ΩA={φA,βA,αA}=const\Omega^{A} = \{\varphi^{A},\beta^{A},\alpha^{A}\} = \mathrm{const} as a function of the atomic position along the zz-direction in a 1×1×361 \times1 \times36 magnetic unit cell.

Figure 6(b) shows the total energies obtained from the minimization of the micromagnetic energy (solid lines), from the atomistic model (circles), and from the spin-dynamics simulations (diamonds) as a function of the period of the magnetic modulation λ=2π/∣q∣\lambda= 2\pi/|\mathbf{q}| for the spin spirals propagating in the zz direction. Since the period of the magnetic modulations is rather small, we include micromagnetic parameters M(p)AB\mathcal{M}^{(p)AB} up to order p=7p=7, which results in perfect agreement with an atomistic model. As shown in Figure 6(b) the total energy minimum of the spin spiral is reduced by a factor of 14 due to the phase difference φ\varphi (blue curve) and by a factor of 32 due to the deviation of the rotation axes erotA(βA,αA)\mathbf{e}_{\mathrm{rot}}^{A}(\beta^{A},\alpha^{A}) from the direction of q\mathbf{q} (red curve). The period of the magnetic modulations λmin⁡\lambda_{\min} becomes shorter due to such a twist as well. The corresponding twist angles φ(λ)\varphi(\lambda), β(λ)\beta(\lambda), and α(λ)\alpha(\lambda) minimizing the micromagnetic total energy as a function of λ\lambda are shown in Figure 6(c).

Energies obtained by spin-dynamics simulations are only slightly lower than the energies predicted by our micromagnetic model, see Figure 6(b). This small difference is because in the micromagnetic model we assume homogeneous spirals for each sublattice, e.g., twist angles ΩA={φA,βA,αA}=const\Omega^{A} = \{\varphi^{A},\beta^{A},\alpha^{A}\} = \mathrm{const} for any site RiA\mathbf{R}_{i}^{A}. However, spin-dynamics simulations show that ΩiA=ΩA+ΔΩiA(RiA)\Omega_{i}^{A} = \Omega^{A} + \Delta\Omega_{i}^{A}(\mathbf{R}_{i}^{A}), where ΩiA=f(RiA)\Omega_{i}^{A} = f(\mathbf{R}_{i}^{A}) is a function of the atomic position RiA\mathbf{R}_{i}^{A} [52], see Figure 6(d). As can been seen from Figure 6, the energy gain due to such nonhomogeneity of the spin spirals ΔΩiA\Delta\Omega_{i}^{A} is much smaller than the energy gain due to a twist ΩA\Omega^{A}, as well as ΔΩiA≪ΩA\Delta\Omega_{i}^{A} \ll\Omega^{A}.

On the other hand, the computational burden in the spin-dynamic simulations of the ground state properties of FeGe due to long range of the magnetic interactions can be significantly reduced by replacing the larger number of the realistic interaction parameters JijABJ_{ij}^{AB} by a smaller number of the atomistic effective interaction parameters J~ijAB\widetilde{J}_{ij}^{AB} with the condition that they both provide the same micromagnetic interaction tensors M(p)AB\mathcal{M}^{(p)AB} (see Appendix D). Such atomistic effective interaction parameters J~ijAB\widetilde{J}_{ij}^{AB} can be obtained by solving

the following system of equations with different orders pp:

(−1)p+1(2p)!∑s=1Nsh∑j=1NBsJ~1j(kk′)ABsR1j(2p)ABs=Mkk′(2p)AB,\frac{(-1)^{p+1}}{(2p)!}\sum_{s=1}^{N_{\mathrm{sh}}}\sum_{j=1}^{N_{B^{s}}}\widetilde{J}_{1j(kk')}^{AB^{s}}\mathcal{R}_{1j}^{(2p)AB^{s}} = \mathcal{M}_{kk'}^{(2p)AB},
(−1)p+1(2p+1)!∑s=1Nsh∑j=1NBsJ~1j(kk′)ABsR1j(2p+1)ABs=Mkk′(2p+1)AB.\frac{(-1)^{p+1}}{(2p+1)!}\sum_{s=1}^{N_{\mathrm{sh}}}\sum_{j=1}^{N_{B^{s}}}\widetilde{J}_{1j(kk')}^{AB^{s}}\mathcal{R}_{1j}^{(2p+1)AB^{s}} = \mathcal{M}_{kk'}^{(2p+1)AB}.

where the sum over ss is restricted to the number of the interacting (effective) shells NshN_{\mathrm{sh}} with radii Rs=∣RijAB∣≤Rmax⁡R^{s} = |\mathbf{R}_{ij}^{AB}| \leq R_{\max} (shell index ss represent all bonds of the same length connected by symmetries) and jj runs over sites of sublattice BsB^{s} within the same RsR^{s}.

The effective interaction parameters J~ijAB\widetilde{J}_{ij}^{AB} and D~ijAB\widetilde{\mathbf{D}}_{ij}^{AB} for FeGe obtained as a fit to the micromagnetic tensors J(p)AB\mathcal{J}^{(p)AB} (p≤2p \leq2) and D(p)AB\mathcal{D}^{(p)AB} (p≤1p \leq1) are given in Table I. As shown in Figure 7(a), the energies corresponding to these effective parameters obtained by SD simulations are in excellent agreement with the prediction of our micromagnetic model, as well as with the energies obtained by the atomistic model applied to the predicted magnetic structure with realistic interaction parameters. However, while the prediction of our micromagnetic model is in excellent agreement with the SD simulations, showing a significant reduction of the total energy of the spin spirals due to the twist, the obtained period of the magnetic modulations λmin⁡DFT≈140\lambda_{\min}^{\mathrm{DFT}}\approx140 (nm) remains two times larger than the experimental observation.

ssAA-BBRsR_sRijAB\mathbf{R}_{ij}^{AB}J~ijAB\widetilde{J}_{ij}^{AB}D~ijAB\widetilde{\mathbf{D}}_{ij}^{AB}
11-20.613(0.23, -0.27, -0.5)3.311(-0.013, -0.295, 1.226)
21-20.914(0.23, 0.73, -0.5)1.474(0.183, 0.088, -0.481)
31-20.957(-0.77, -0.27, -0.5)-0.006(0.045, -0.175, -0.332)
41-21.173(-0.77, 0.73, -0.5)-0.263–
51-21.355(1.23, -0.27, -0.5)1.997–
61-21.384(0.23, -1.27, -0.5)-0.101–
71-21.515(1.23, 0.73, -0.5)-1.421–
81-21.541(0.23, -0.27, -1.5)1.694–
11-11.000(-1, 0, 0)4.733(-0.127, 0.039, 0.108)
21-11.414(-1, -1, 0)-1.904–
31-11.414(-1, 0, 1)-0.820–

Table 1. The effective interaction parameters J~ijAB\widetilde{J}_{ij}^{AB} and D~ijAB\widetilde{\mathbf{D}}_{ij}^{AB} (where ss stands for the radius RsR_s of the shell with all bonds ∣RijAB∣=Rs\lvert\mathbf{R}_{ij}^{AB}\rvert=R_s connected by symmetries) between atoms of sublattices A=1A=1 and B=1,2B=1,2, obtained as a fit to the micromagnetic parameters J(p)AB\mathcal{J}^{(p)AB} (p≤2p\leq2) and D(p)AB\mathcal{D}^{(p)AB} (p≤1p\leq1). The effective interaction parameters between atoms of all other sublattices can be obtained by using symmetries (see Appendix D).

A possible reason for this discrepancy might stem from the failure of DFT to capture the effect of electronic correlations, which can have a more drastic effect on DM interaction than on the exchange, which provides good agreement with the experimental Curie temperature [34]. Therefore, we also study (using the micromagnetic model and SD simulations) the dependences of the spin-spiral total energies, corresponding period, and twist angles on strength of the DM interaction vectors ηDij\eta\mathbf{D}_{ij}, where η=0.9,…2.1\eta=0.9,\ldots2.1. As shown in Figure 7, not only does the period λmin⁡\lambda_{\min} of the spin spiral become shorter when the strength of the DM interaction is increased, reaching the experimental λ=70\lambda=70 (nm) at η≈2\eta\approx2, but also the corresponding twist angles ϕmin⁡\phi_{\min} and βmin⁡\beta_{\min} become larger, lowering the total energy Emin⁡E_{\min} more prominently than in the case when the twist of the spin spirals is ignored.

Magnetic structure

We define the magnetic moment SiAS_i^A at site ii of sublattice AA as

SiA=R(βA,αA)(sin⁡θAcos⁡[q⋅RiA+ϕA]sin⁡θAsin⁡[q⋅RiA+ϕA]cos⁡θA)=e1Acos⁡(q⋅RiA)+e2Asin⁡(q⋅RiA)+e3A,\mathbf{S}_i^A=\mathcal{R}(\beta^A,\alpha^A)\begin{pmatrix}\sin\theta^A\cos[\mathbf{q}\cdot\mathbf{R}_i^A+\phi^A]\\ \sin\theta^A\sin[\mathbf{q}\cdot\mathbf{R}_i^A+\phi^A]\\ \cos\theta^A\end{pmatrix} =\mathbf{e}_1^A\cos(\mathbf{q}\cdot\mathbf{R}_i^A)+\mathbf{e}_2^A\sin(\mathbf{q}\cdot\mathbf{R}_i^A)+\mathbf{e}_3^A ,

where the parameter ΩA={βA,αA,θA,ϕA}\Omega^A=\{\beta^A,\alpha^A,\theta^A,\phi^A\} represents a set of spin-spiral twist angles and R(βA,αA)\mathcal{R}(\beta^A,\alpha^A) is a rotation matrix mapping the ez\mathbf{e}_z axis to the rotation axis erotA\mathbf{e}_{\mathrm{rot}}^A. Vectors emA=emA(ΩA)\mathbf{e}_m^A=\mathbf{e}_m^A(\Omega^A) characterizing the twist of the spin spiral in sublattice AA are defined as

e1A=sin⁡θAR(βA,αA)(excos⁡ϕA+eysin⁡ϕA),\mathbf{e}_1^A=\sin\theta^A\mathcal{R}(\beta^A,\alpha^A)(\mathbf{e}_x\cos\phi^A+\mathbf{e}_y\sin\phi^A),
e2A=sin⁡θAR(βA,αA)(eycos⁡ϕA−exsin⁡ϕA),\mathbf{e}_2^A=\sin\theta^A\mathcal{R}(\beta^A,\alpha^A)(\mathbf{e}_y\cos\phi^A-\mathbf{e}_x\sin\phi^A),
e3A=cos⁡θAR(βA,αA)ez=cos⁡θAerotA.\mathbf{e}_3^A=\cos\theta^A\mathcal{R}(\beta^A,\alpha^A)\mathbf{e}_z=\cos\theta^A\mathbf{e}_{\mathrm{rot}}^A .

Trigonometric expansion

Since (q⋅R)p=Q(p):R(p)(\mathbf{q}\cdot\mathbf{R})^p=\mathcal{Q}^{(p)}:\mathcal{R}^{(p)}, where Q(p)=q⊗p\mathcal{Q}^{(p)}=\mathbf{q}^{\otimes p} and R(p)=R⊗p\mathcal{R}^{(p)}=\mathbf{R}^{\otimes p} are the pp-fold tensor products of q\mathbf{q} and R\mathbf{R} vectors with themselves, respectively, we expand the cosine and sine functions of q⋅R\mathbf{q}\cdot\mathbf{R} in the form

−cos⁡q⋅R=∑p=0∞(−1)p+1(2p)!Q(2p):R(2p),-\cos\mathbf{q}\cdot\mathbf{R}=\sum_{p=0}^{\infty}\frac{(-1)^{p+1}}{(2p)!}\mathcal{Q}^{(2p)}:\mathcal{R}^{(2p)},
−sin⁡q⋅R=∑p=0∞(−1)p+1(2p+1)!Q(2p+1):R(2p+1),-\sin\mathbf{q}\cdot\mathbf{R}=\sum_{p=0}^{\infty}\frac{(-1)^{p+1}}{(2p+1)!}\mathcal{Q}^{(2p+1)}:\mathcal{R}^{(2p+1)},

where the sum over pp defines the order of Tailor expansion and and notation “::” denotes the inner product between two tensors. Using this form for the expansion of the trigonometric functions the atomistic interaction tensors J1jAB\mathcal{J}_{1j}^{AB} entering the total energy expression (Eq. (5) in the main text) and corresponding distance between interacting sites R1jAB\mathbf{R}_{1j}^{AB} can be combined in the micromagnetic interaction tensors M(p)AB\mathcal{M}^{(p)AB}.

−∑j=1∞J1jAB:C+ABcos⁡q⋅R1jAB=∑p=0∞∑j=1∞(−1)p+1(2p)!(J1jAB:C+AB)(Q(2p):R1j(2p)AB)=∑p=0∞(−1)p+1(2p)!∑j=1∞(J1jAB⊗R1j(2p)AB):(C+AB⊗Q(2p))=∑p=0∞M(2p)AB:(C+AB⊗Q(2p)AB),\begin{aligned} -\sum_{j=1}^{\infty}\mathcal{J}_{1j}^{AB}:\mathcal{C}_{+}^{AB}\cos\mathbf{q}\cdot\mathbf{R}_{1j}^{AB} =\sum_{p=0}^{\infty}\sum_{j=1}^{\infty}\frac{(-1)^{p+1}}{(2p)!}\left(\mathcal{J}_{1j}^{AB}:\mathcal{C}_{+}^{AB}\right)\left(\mathcal{Q}^{(2p)}:\mathcal{R}_{1j}^{(2p)AB}\right) \\ &=\sum_{p=0}^{\infty}\frac{(-1)^{p+1}}{(2p)!}\sum_{j=1}^{\infty}\left(\mathcal{J}_{1j}^{AB}\otimes\mathcal{R}_{1j}^{(2p)AB}\right):\left(\mathcal{C}_{+}^{AB}\otimes\mathcal{Q}^{(2p)}\right) \\ &=\sum_{p=0}^{\infty}\mathcal{M}^{(2p)AB}:\left(\mathcal{C}_{+}^{AB}\otimes\mathcal{Q}^{(2p)AB}\right), \end{aligned}
−∑j=1∞J1jAB:C−ABsin⁡q⋅R1jAB=∑p=0∞∑j=1∞(−1)p+1(2p+1)!(J1jAB:C−AB)(Q(2p+1):R1j(2p+1)AB)=∑p=0∞(−1)p+1(2p+1)!∑j=1∞(J1jAB⊗R1j(2p+1)AB):(C−AB⊗Q(2p+1))=∑p=0∞M(2p+1)AB:(C−AB⊗Q(2p+1)AB),\begin{aligned} -\sum_{j=1}^{\infty}\mathcal{J}_{1j}^{AB}:\mathcal{C}_{-}^{AB}\sin\mathbf{q}\cdot\mathbf{R}_{1j}^{AB} =\sum_{p=0}^{\infty}\sum_{j=1}^{\infty}\frac{(-1)^{p+1}}{(2p+1)!}\left(\mathcal{J}_{1j}^{AB}:\mathcal{C}_{-}^{AB}\right)\left(\mathcal{Q}^{(2p+1)}:\mathcal{R}_{1j}^{(2p+1)AB}\right) \\ &=\sum_{p=0}^{\infty}\frac{(-1)^{p+1}}{(2p+1)!}\sum_{j=1}^{\infty}\left(\mathcal{J}_{1j}^{AB}\otimes\mathcal{R}_{1j}^{(2p+1)AB}\right):\left(\mathcal{C}_{-}^{AB}\otimes\mathcal{Q}^{(2p+1)}\right) \\ &=\sum_{p=0}^{\infty}\mathcal{M}^{(2p+1)AB}:\left(\mathcal{C}_{-}^{AB}\otimes\mathcal{Q}^{(2p+1)AB}\right), \end{aligned}

and

−∑j=1∞J1jAB:CAB=∑j=1∞(J1jAB:CAB)(Q(0):R1j(0)AB)=−10!∑j=1∞(J1jAB⊗R1j(0)AB):(CAB⊗Q(0))=M(0)AB:CAB,\begin{aligned} -\sum_{j=1}^{\infty}\mathcal{J}_{1j}^{AB}:\mathcal{C}^{AB} =\sum_{j=1}^{\infty}\left(\mathcal{J}_{1j}^{AB}:\mathcal{C}^{AB}\right)\left(\mathcal{Q}^{(0)}:\mathcal{R}_{1j}^{(0)AB}\right) \\ &=-\frac{1}{0!}\sum_{j=1}^{\infty}\left(\mathcal{J}_{1j}^{AB}\otimes\mathcal{R}_{1j}^{(0)AB}\right):\left(\mathcal{C}^{AB}\otimes\mathcal{Q}^{(0)}\right) \\ &=\mathcal{M}^{(0)AB}:\mathcal{C}^{AB}, \end{aligned}

where tensors C+AB=e1A⊗e1B+e2A⊗e2B\mathcal{C}_{+}^{AB}=\mathbf{e}_1^A\otimes\mathbf{e}_1^B+\mathbf{e}_2^A\otimes\mathbf{e}_2^B, C−AB=e1A⊗e2B−e2A⊗e1B\mathcal{C}_{-}^{AB}=\mathbf{e}_1^A\otimes\mathbf{e}_2^B-\mathbf{e}_2^A\otimes\mathbf{e}_1^B, and CAB=e3A⊗e3B\mathcal{C}^{AB}=\mathbf{e}_3^A\otimes\mathbf{e}_3^B characterize the twist between spin spirals in sublattices AA and BB.

D Symmetries and micromagnetic interaction tensors

Magnetic and nonmagnetic atoms in B20 compounds are located at the 4a4a Wyckoff positions, r1(u)=(u,u,u)\mathbf{r}^{1}(u)=(u,u,u), r2(u)=(0.5−u,1−u,0.5+u)\mathbf{r}^{2}(u)=(0.5-u,1-u,0.5+u), r4(u)=(0.5+u,0.5−u,1−u)\mathbf{r}^{4}(u)=(0.5+u,0.5-u,1-u), and r3(u)=(1−u,0.5+u,0.5−u)\mathbf{r}^{3}(u)=(1-u,0.5+u,0.5-u), where the parameters uFe=0.135u^{\mathrm{Fe}}=0.135 and uGe=−0.158u^{\mathrm{Ge}}=-0.158 stand, respectively, for Fe and Ge atoms in the B20 FeGe compound. Each sublattice AA in B20 compounds has only one threefold rotation axis (see Figure 8), which can be defined as

Nearest Fe and Ge atoms in the B20 structure

Figure 8. Nearest Fe (light red spheres) and Ge (blue spheres) atoms at each Fe site (darker red spheres) of the B20 structure. Red and blue lines indicate bonds connected by threefold rotation symmetry with respect to the symmetry direction nA\mathbf{n}^{A}. All red (blue) bonds RijAB\mathbf{R}_{ij}^{AB} (RijAB′\mathbf{R}_{ij}^{AB^{\prime}}) are connected to Rij12\mathbf{R}_{ij}^{12} by TωABA\mathcal{T}_{\omega_{AB}}^{A} (−TωABB-\mathcal{T}_{\omega_{AB}}^{B}) symmetry. Dashed arrows indicate bonds at site A=Fe1A=\mathrm{Fe1} that appear at other sites B≠AB\ne A.

nA=∂rA(u)∂u∈{(111),(1ˉ11),(11ˉ1),(111ˉ)}.\mathbf{n}^{A}=\frac{\partial\mathbf{r}^{A}(u)}{\partial u}\in\{(111),(\bar{1}11),(1\bar{1}1),(11\bar{1})\}.

All symmetries at any site of any sublattice AA in the B20 structure can be generalized by the following symmetry operation

TAB=TωABA=(nA⊗nA)∘RωAB[111],(18)\mathcal{T}^{AB}=\mathcal{T}_{\omega_{AB}}^{A}=\left(\mathbf{n}^{A}\otimes\mathbf{n}^{A}\right)\circ\mathcal{R}_{\omega_{AB}}^{[111]}, \tag*{(18)}

where RωAB[111]\mathcal{R}_{\omega_{AB}}^{[111]} is a threefold rotation matrix around the n1=[1,1,1]\mathbf{n}^{1}=[1,1,1] axis on angle ωAB=2π3k\omega^{AB}=\frac{2\pi}{3}k (k=0,1,2k=0,1,2), “∘\circ” is the elementwise (Hadamard) product, and operation (nA⊗nA)\left(\mathbf{n}^{A}\otimes\mathbf{n}^{A}\right) maps sublattice 1 to any other sublattice AA, such that any bond RijAB\mathbf{R}_{ij}^{AB} between sites i′i' and j′j' of any sublattices AA and BB, respectively, can be represented as RijAB=TAB⋅Rij12\mathbf{R}_{ij}^{AB}=\mathcal{T}^{AB}\cdot\mathbf{R}_{ij}^{12} or as RijAA=TAA⋅Rij11\mathbf{R}_{ij}^{AA}=\mathcal{T}^{AA}\cdot\mathbf{R}_{ij}^{11} if A=BA=B. Note that the successive three fold rotations around nA\mathbf{n}^{A} gives RijAB→RikAC→RilAD\mathbf{R}_{ij}^{AB}\rightarrow\mathbf{R}_{ik}^{AC}\rightarrow\mathbf{R}_{il}^{AD}, where the order of sublattices BB, CC, DD is such that nA⋅(RijAB×RikAC)≥0\mathbf{n}^{A}\cdot(\mathbf{R}_{ij}^{AB}\times\mathbf{R}_{ik}^{AC})\geq0 and nA⋅(RikAC×RilAD)≥0\mathbf{n}^{A}\cdot(\mathbf{R}_{ik}^{AC}\times\mathbf{R}_{il}^{AD})\geq0. Also, it is easy to show that the symmetry operation −TωABB-\mathcal{T}_{\omega_{AB}}^{B} at sublattice AA gives RijAB=−TωABBRij12=−RijBA\mathbf{R}_{ij}^{AB}=-\mathcal{T}_{\omega_{AB}}^{B}\mathbf{R}_{ij}^{12}=-\mathbf{R}_{ij}^{BA}. It represents double counting and therefore can be omitted; otherwise, a factor of 1/21/2 has to be used in the energy calculations. For instance, any site of sublattice AA has six nearest neighbors in total; however, only three of them are related by threefold rotation symmetry nA\mathbf{n}^{A}, and the other three appear at sites of

other sublattices BB, CC, and DD (see Figure 8). The symmetry operations TωABA\mathcal{T}_{\omega_{AB}}^{A} and −TωABB-\mathcal{T}_{\omega_{AB}}^{B} for the bonds are illustrated in Table II (PDF p. 11), and those for the micromagnetic vectors D(0)AB\mathcal{D}^{(0)AB} and A(1)AB\mathcal{A}^{(1)AB} are given in Table III (PDF p. 11).

AA-BBT0∘A\mathcal{T}^{A}_{0^\circ}RAB\mathbf{R}^{AB}AA-BBT120∘A\mathcal{T}^{A}_{120^\circ}RAB\mathbf{R}^{AB}AA-BBT240∘A\mathcal{T}^{A}_{240^\circ}RAB\mathbf{R}^{AB}AA-A′A'T0∘A\mathcal{T}^{A}_{0^\circ}RAA′\mathbf{R}^{AA'}T120∘A\mathcal{T}^{A}_{120^\circ}RAA′\mathbf{R}^{AA'}T240∘A\mathcal{T}^{A}_{240^\circ}RAA′\mathbf{R}^{AA'}
1-2T0∘1\mathcal{T}^{1}_{0^\circ} (x,y,z)(x,y,z)1-3T120∘1\mathcal{T}^{1}_{120^\circ} (x,y,z)(x,y,z)1-4T240∘1\mathcal{T}^{1}_{240^\circ} (y,z,x)(y,z,x)1-1T0∘1\mathcal{T}^{1}_{0^\circ} (x′,y′,z′)(x',y',z')T120∘1\mathcal{T}^{1}_{120^\circ} (z′,x′,y′)(z',x',y')T240∘1\mathcal{T}^{1}_{240^\circ} (y′,z′,x′)(y',z',x')
−T0∘2-\mathcal{T}^{2}_{0^\circ} (x,y,zˉ)(x,y,\bar z)−T120∘3-\mathcal{T}^{3}_{120^\circ} (zˉ,x,y)(\bar z,x,y)−T240∘4-\mathcal{T}^{4}_{240^\circ} (y,zˉ,x)(y,\bar z,x)−T0∘1-\mathcal{T}^{1}_{0^\circ} (xˉ′,yˉ′,zˉ′)(\bar x',\bar y',\bar z')−T120∘2-\mathcal{T}^{2}_{120^\circ} (zˉ′,xˉ′,yˉ′)(\bar z',\bar x',\bar y')−T240∘3-\mathcal{T}^{3}_{240^\circ} (yˉ′,zˉ′,xˉ′)(\bar y',\bar z',\bar x')
2-1T0∘2\mathcal{T}^{2}_{0^\circ} (xˉ,yˉ,zˉ)(\bar x,\bar y,\bar z)2-4T120∘2\mathcal{T}^{2}_{120^\circ} (zˉ,x,y)(\bar z,x,y)2-3T240∘2\mathcal{T}^{2}_{240^\circ} (yˉ,zˉ,x)(\bar y,\bar z,x)2-2T0∘2\mathcal{T}^{2}_{0^\circ} (xˉ′,yˉ′,zˉ′)(\bar x',\bar y',\bar z')T120∘2\mathcal{T}^{2}_{120^\circ} (zˉ′,xˉ′,yˉ′)(\bar z',\bar x',\bar y')T240∘2\mathcal{T}^{2}_{240^\circ} (yˉ′,zˉ′,xˉ′)(\bar y',\bar z',\bar x')
−T0∘1-\mathcal{T}^{1}_{0^\circ} (xˉ,yˉ,zˉ)(\bar x,\bar y,\bar z)−T120∘4-\mathcal{T}^{4}_{120^\circ} (z,x,y)(z,x,y)−T240∘3-\mathcal{T}^{3}_{240^\circ} (yˉ,zˉ,x)(\bar y,\bar z,x)−T0∘2-\mathcal{T}^{2}_{0^\circ} (xˉ′,yˉ′,zˉ′)(\bar x',\bar y',\bar z')−T120∘2-\mathcal{T}^{2}_{120^\circ} (zˉ′,xˉ′,yˉ′)(\bar z',\bar x',\bar y')−T240∘2-\mathcal{T}^{2}_{240^\circ} (yˉ′,zˉ′,xˉ′)(\bar y',\bar z',\bar x')
3-4T0∘3\mathcal{T}^{3}_{0^\circ} (x,yˉ,zˉ)(x,\bar y,\bar z)3-1T120∘3\mathcal{T}^{3}_{120^\circ} (x,zˉ,yˉ)(x,\bar z,\bar y)3-2T240∘3\mathcal{T}^{3}_{240^\circ} (y,z,xˉ)(y,z,\bar x)3-3T0∘3\mathcal{T}^{3}_{0^\circ} (xˉ′,yˉ′,zˉ′)(\bar x',\bar y',\bar z')T120∘3\mathcal{T}^{3}_{120^\circ} (zˉ′,xˉ′,yˉ′)(\bar z',\bar x',\bar y')T240∘3\mathcal{T}^{3}_{240^\circ} (yˉ′,zˉ′,xˉ′)(\bar y',\bar z',\bar x')
−T0∘4-\mathcal{T}^{4}_{0^\circ} (x,yˉ,zˉ)(x,\bar y,\bar z)−T120∘1-\mathcal{T}^{1}_{120^\circ} (zˉ,x,yˉ)(\bar z,x,\bar y)−T240∘4-\mathcal{T}^{4}_{240^\circ} (y,z,xˉ)(y,z,\bar x)−T0∘3-\mathcal{T}^{3}_{0^\circ} (xˉ′,yˉ′,zˉ′)(\bar x',\bar y',\bar z')−T120∘3-\mathcal{T}^{3}_{120^\circ} (zˉ′,xˉ′,yˉ′)(\bar z',\bar x',\bar y')−T240∘4-\mathcal{T}^{4}_{240^\circ} (yˉ′,zˉ′,xˉ′)(\bar y',\bar z',\bar x')
4-3T0∘4\mathcal{T}^{4}_{0^\circ} (x,yˉ,zˉ)(x,\bar y,\bar z)4-2T120∘4\mathcal{T}^{4}_{120^\circ} (zˉ,xˉ,yˉ)(\bar z,\bar x,\bar y)4-1T240∘4\mathcal{T}^{4}_{240^\circ} (yˉ,zˉ,xˉ)(\bar y,\bar z,\bar x)4-4T0∘4\mathcal{T}^{4}_{0^\circ} (xˉ′,yˉ′,zˉ′)(\bar x',\bar y',\bar z')T120∘4\mathcal{T}^{4}_{120^\circ} (zˉ′,xˉ′,yˉ′)(\bar z',\bar x',\bar y')T240∘4\mathcal{T}^{4}_{240^\circ} (yˉ′,zˉ′,xˉ′)(\bar y',\bar z',\bar x')
−T0∘3-\mathcal{T}^{3}_{0^\circ} (xˉ,yˉ,zˉ)(\bar x,\bar y,\bar z)−T120∘1-\mathcal{T}^{1}_{120^\circ} (z,x,yˉ)(z,x,\bar y)−T240∘4-\mathcal{T}^{4}_{240^\circ} (yˉ,zˉ,xˉ)(\bar y,\bar z,\bar x)−T0∘4-\mathcal{T}^{4}_{0^\circ} (xˉ′,yˉ′,zˉ′)(\bar x',\bar y',\bar z')−T120∘4-\mathcal{T}^{4}_{120^\circ} (zˉ′,xˉ′,yˉ′)(\bar z',\bar x',\bar y')−T240∘4-\mathcal{T}^{4}_{240^\circ} (yˉ′,zˉ′,xˉ′)(\bar y',\bar z',\bar x')

Table II (PDF p. 11). Symmetry relations between bonds Rij′AB=TαABA⋅Rij′12\mathbf{R}^{AB}_{ij'}=\mathcal{T}^{A}_{\alpha_{AB}}\cdot\mathbf{R}^{12}_{ij'} and Rij′AA′=TαABA⋅Rij′11′\mathbf{R}^{AA'}_{ij'}=\mathcal{T}^{A}_{\alpha_{AB}}\cdot\mathbf{R}^{11'}_{ij'} in the B20 structure.

Since symmetry-related bonds Ri′j′AB\mathbf{R}_{i'j'}^{AB} and symmetry-related DM vectors Di′j′AB\mathbf{D}_{i'j'}^{AB} can be defined via Rij12\mathbf{R}_{ij}^{12} or Rij11\mathbf{R}_{ij}^{11} and Dij12\mathbf{D}_{ij}^{12} or Dij11\mathbf{D}_{ij}^{11}, respectively, all micromagnetic tensors M(p)AB\mathcal{M}_{(p)}^{AB} can be also defined in terms of M(p)12\mathcal{M}^{(p)12} or M(p)11\mathcal{M}^{(p)11}:

A(0)AB=A(0)12,A(1)AB=TABA(1)12,A(2)AB=TABA(2)12(TAB)T,A(3)AB=TAB(TAB(A(3)12)T)(TAB)T.(19)\begin{aligned} \mathcal{A}^{(0)AB}&=\mathcal{A}^{(0)12},\qquad\mathcal{A}^{(1)AB}=\mathcal{T}^{AB}\mathcal{A}^{(1)12},\\ \mathcal{A}^{(2)AB}&=\mathcal{T}^{AB}\mathcal{A}^{(2)12}(\mathcal{T}^{AB})^{T},\\ \mathcal{A}^{(3)AB}&=\mathcal{T}^{AB}\left(\mathcal{T}^{AB}\left(\mathcal{A}^{(3)12}\right)^{T}\right)(\mathcal{T}^{AB})^{T}. \tag*{(19)} \end{aligned}

The same operations are valid for tensors D(p)\mathcal{D}^{(p)}. Note if A=BA=B the only symmetry operation T0A\mathcal{T}_{0}^{A} is required to transform M(0)11\mathcal{M}^{(0)11} to M(0)AA\mathcal{M}^{(0)AA}. Below we list tensors M(p)12\mathcal{M}^{(p)12} and M(p)11\mathcal{M}^{(p)11} for B20 FeGe as well as the sum of M(p)AB\mathcal{M}^{(p)AB} over all pairs of sublattices:

ABABA(0)AB\mathcal{A}^{(0)AB}A(1)AB\mathcal{A}^{(1)AB}D(0)AB\mathcal{D}^{(0)AB}
11−12.06-12.06(0,0,0)(0,0,0)(0,0,0)(0,0,0)
12−13.37-13.37(−22.39,17.99,0)(-22.39,17.99,0)(−0.43,0.76,0)(-0.43,0.76,0)
∑AB\sum_{AB}−208.68-208.68(0,0,0)(0,0,0)(0,0,0)(0,0,0)

Table 2.

ABABA(2)AB\mathcal{A}^{(2)AB}D(1)AB\mathcal{D}^{(1)AB}
11(−15.77−23.96−23.96−23.96−15.77−23.96−23.96−23.96−15.77)\begin{pmatrix}-15.77&-23.96&-23.96\\-23.96&-15.77&-23.96\\-23.96&-23.96&-15.77\end{pmatrix}(−1.201.010.370.37−1.201.011.010.37−1.20)\begin{pmatrix}-1.20&1.01&0.37\\0.37&-1.20&1.01\\1.01&0.37&-1.20\end{pmatrix}
12(23.15−40.360.00−40.365.190.000.000.00111.78)\begin{pmatrix}23.15&-40.36&0.00\\-40.36&5.19&0.00\\0.00&0.00&111.78\end{pmatrix}(−0.04−1.17−0.00−0.82−1.800.000.000.001.94)\begin{pmatrix}-0.04&-1.17&-0.00\\-0.82&-1.80&0.00\\0.00&0.00&1.94\end{pmatrix}
∑AB\sum_{AB}497.42I1497.42\mathcal{I}_{1}−4.38I1-4.38\mathcal{I}_{1}

Table 3.

A(3)12\mathcal{A}^{(3)12}D(2)12\mathcal{D}^{(2)12}
(−305.74−14.100.00−14.1081.110.000.000.0077.91−14.1081.110.0081.11345.570.000.000.00−140.590.000.0077.910.000.00−140.5977.91−140.590.00)\begin{pmatrix}-305.74&-14.10&0.00\\-14.10&81.11&0.00\\0.00&0.00&77.91\\-14.10&81.11&0.00\\81.11&345.57&0.00\\0.00&0.00&-140.59\\0.00&0.00&77.91\\0.00&0.00&-140.59\\77.91&-140.59&0.00\end{pmatrix}(17.270.070.000.07−4.890.000.000.00−0.935.42−2.560.00−2.56−18.610.000.000.000.170.000.00−4.720.000.006.54−4.726.540.00)\begin{pmatrix}17.27&0.07&0.00\\0.07&-4.89&0.00\\0.00&0.00&-0.93\\5.42&-2.56&0.00\\-2.56&-18.61&0.00\\0.00&0.00&0.17\\0.00&0.00&-4.72\\0.00&0.00&6.54\\-4.72&6.54&0.00\end{pmatrix}
A(3)11=∑ABA(3)AB=0I3\mathcal{A}^{(3)11}=\sum_{AB}\mathcal{A}^{(3)AB}=0\mathcal{I}_{3}D(2)11=∑ABD(2)AB=0I3\mathcal{D}^{(2)11}=\sum_{AB}\mathcal{D}^{(2)AB}=0\mathcal{I}_{3}

Table 4.

Here, Ip\mathcal{I}_{p} is identity tensor of rank pp. As we can see, the sum of the above tensors M(p)AB\mathcal{M}^{(p)AB} over all pairs of sublattices of B20 structure gives the scalar matrices due to the symmetry of the B20 structure characterized by the point group TT. Similarly, summing up the higher-order tensors D(3)AB\mathcal{D}^{(3)AB} and A(4)AB\mathcal{A}^{(4)AB} over all pairs of sublattices gives

D‾‾(3)=∑ABD‾‾(3)AB=((D1(3)000D2(3)000D3(3))(0D2(3)0D2(3)00000)(00D3(3)000D3(3)00)(0D3(3)0D3(3)00000)(0D1(3)0D1(3)0000D2(3))(00D2(3)0D3(3)000D1(3))),(20)\underline{\underline{\mathcal{D}}}^{(3)}=\sum_{AB}\underline{\underline{\mathcal{D}}}^{(3)AB}= \left( \begin{pmatrix} D^{(3)}_1&0&0\\ 0&D^{(3)}_2&0\\ 0&0&D^{(3)}_3 \end{pmatrix} \begin{pmatrix} 0&D^{(3)}_2&0\\ D^{(3)}_2&0&0\\ 0&0&0 \end{pmatrix} \begin{pmatrix} 0&0&D^{(3)}_3\\ 0&0&0\\ D^{(3)}_3&0&0 \end{pmatrix} \begin{pmatrix} 0&D^{(3)}_3&0\\ D^{(3)}_3&0&0\\ 0&0&0 \end{pmatrix} \begin{pmatrix} 0&D^{(3)}_1&0\\ D^{(3)}_1&0&0\\ 0&0&D^{(3)}_2 \end{pmatrix} \begin{pmatrix} 0&0&D^{(3)}_2\\ 0&D^{(3)}_3&0\\ 0&0&D^{(3)}_1 \end{pmatrix} \right), \tag*{(20)}
A‾‾(4)=∑ABA‾‾(4)AB=((A1(4)000A2(4)000A2(4))(0A2(4)0A2(4)00000)(00A2(4)000A2(4)00)(0A2(4)0A2(4)00000)(0A1(4)0A1(4)0000A2(4))(00A2(4)0A2(4)000A1(4))).\underline{\underline{\mathcal{A}}}^{(4)}=\sum_{AB}\underline{\underline{\mathcal{A}}}^{(4)AB}= \left( \begin{pmatrix} A^{(4)}_1&0&0\\ 0&A^{(4)}_2&0\\ 0&0&A^{(4)}_2 \end{pmatrix} \begin{pmatrix} 0&A^{(4)}_2&0\\ A^{(4)}_2&0&0\\ 0&0&0 \end{pmatrix} \begin{pmatrix} 0&0&A^{(4)}_2\\ 0&0&0\\ A^{(4)}_2&0&0 \end{pmatrix} \begin{pmatrix} 0&A^{(4)}_2&0\\ A^{(4)}_2&0&0\\ 0&0&0 \end{pmatrix} \begin{pmatrix} 0&A^{(4)}_1&0\\ A^{(4)}_1&0&0\\ 0&0&A^{(4)}_2 \end{pmatrix} \begin{pmatrix} 0&0&A^{(4)}_2\\ 0&A^{(4)}_2&0\\ 0&0&A^{(4)}_1 \end{pmatrix} \right).

where D1(3)=−140.86D^{(3)}_1=-140.86 meVÅ3^3, D2(3)=27.16D^{(3)}_2=27.16 meVÅ3^3, D3(3)=27.92D^{(3)}_3=27.92 meVÅ3^3, A1(4)=−1195.94A^{(4)}_1=-1195.94 meVÅ4^4, and A2(4)=−126.49A^{(4)}_2=-126.49 meVÅ4^4.

Micromagnetic energies

The total energy of the spin spirals with twist, Eq. (6), can be unfolded into the exchange and DM contributions:

EexAB(q)=A(0)AB⋅(e3A⋅e3B)+12∑m,n=12(emA⋅enB)[δmnA(0)AB+εmnA(1)AB⋅q+δmnA(2)AB:q⊗2+⋯ ]E_{\mathrm{ex}}^{AB}(\mathbf{q})=\mathcal{A}^{(0)AB}\cdot\left(\mathbf{e}_{3}^{A}\cdot\mathbf{e}_{3}^{B}\right)+\frac{1}{2}\sum_{m,n=1}^{2}\left(\mathbf{e}_{m}^{A}\cdot\mathbf{e}_{n}^{B}\right)\left[\delta_{mn}\mathcal{A}^{(0)AB}+\varepsilon_{mn}\mathcal{A}^{(1)AB}\cdot\mathbf{q}+\delta_{mn}\mathcal{A}^{(2)AB}:\mathbf{q}^{\otimes2}+\cdots\right]
EdmAB(q)=D(0)AB⋅(e3A×e3B)+12∑k∑m,n=12(emA×enB)k[δmnDk(0)AB+εmnDk(1)AB⋅q+δmnDk(2)AB:q⊗2+⋯ ]E_{\mathrm{dm}}^{AB}(\mathbf{q})=\mathcal{D}^{(0)AB}\cdot\left(\mathbf{e}_{3}^{A}\times\mathbf{e}_{3}^{B}\right)+\frac{1}{2}\sum_{k}\sum_{m,n=1}^{2}\left(\mathbf{e}_{m}^{A}\times\mathbf{e}_{n}^{B}\right)_{k}\left[\delta_{mn}\mathcal{D}_{k}^{(0)AB}+\varepsilon_{mn}\mathcal{D}_{k}^{(1)AB}\cdot\mathbf{q}+\delta_{mn}\mathcal{D}_{k}^{(2)AB}:\mathbf{q}^{\otimes2}+\cdots\right]

where indexes k=x,y,zk=x,y,z represent the components of the micromagnetic DM tensor Dk(p)AB\mathcal{D}_{k}^{(p)AB} and symbols εmn\varepsilon_{mn} and δmn\delta_{mn} are the antisymmetric two-dimensional Levi-Civita tensor and Kronecker delta function, respectively, for which each index m,nm,n takes values 1, 2.

References

References

  1. [1]S. Mühlbauer, B. Binz, F. Jonietz, C. Pfleiderer, A. Rosch, A. Neubauer, R. Georgii, and P. Böni, Skyrmion lattice in a chiral magnet, Science 323, 915 (2009), https://science.sciencemag.org/content/323/5916/915.full.pdf.DOI
  2. [2]A. Neubauer, C. Pfleiderer, B. Binz, A. Rosch, R. P. G. Niklowitz, and P. Böni, Topological hall effect in the a phase of mnsi, Phys. Rev. Lett. 102, 186602 (2009).pubmed.ncbi.nlm.nih.gov/19518895
  3. [3]X. Z. Yu, Y. Onose, N. Kanazawa, J. H. Park, J. H. Han, Y. Matsui, N. Nagaosa, and Y. Tokura, Real-space observation of a two-dimensional skyrmion crystal, Nature 465, 901 (2010).DOI
  4. [4]X. Z. Yu, N. Kanazawa, Y. Onose, K. Kimoto, W. Z. Zhang, S. Ishiwata, Y. Matsui, and Y. Tokura, Near room-temperature formation of a skyrmion crystal in thin-films of the helimagnet FeGe, Nature materials 10, 106 (2011).DOI
  5. [5]F. Zheng, F. N. Rybakov, A. B. Borisov, D. Song, S. Wang, Z. A. Li, H. Du, N. S. Kiselev, J. Caron, A. Kovács, M. Tian, Y. Zhang, S. Blügel, and R. E. Dunin-Borkowski, Experimental observation of chiral magnetic bobbers in B20-type FeGe, Nature Nanotechnology 1, (2018), arXiv:1706.04654.
  6. [6]F. N. Rybakov, A. B. Borisov, S. Blügel, and N. S. Kiselev, New type of stable particlelike states in chiral magnets, Phys. Rev. Lett. 115, 117201 (2015).DOI
  7. [7]T. Tanigaki, K. Shibata, N. Kanazawa, X. Yu, Y. Onose, H. S. Park, D. Shindo, and Y. Tokura, Real-Space Observation of Short-Period Cubic Lattice of Skyrmions in MnGe, Nano Letters 15, 5438 (2015), arXiv:1503.03945.
  8. [8]A. Fert, V. Cros, and A. Sampaio, Skyrmions on the track, Nature Nanotechnology 8, 152 (2013).DOI
  9. [9]H. Wilhelm, M. Baenitz, M. Schmidt, U. K. Rößler, A. A. Leonov, and A. N. Bogdanov, Precursor phenomena at the magnetic ordering of the cubic helimagnet fege, Phys. Rev. Lett. 107, 127203 (2011).DOI
  10. [10]F. Pedrazzini, H. Wilhelm, D. Jaccard, T. Jarlborg, M. Schmidt, M. Hanfland, L. Akselrud, H. Q. Yuan, U. Schwarz, Y. Grin, and F. Steglich, Metallic state in cubic fege beyond its quantum phase transition, Phys. Rev. Lett. 98, 047204 (2007).pubmed.ncbi.nlm.nih.gov/17358807
  11. [11]K. Shibata, J. Iwasaki, N. Kanazawa, S. Aizawa, T. Tanigaki, M. Shirai, T. Nakajima, M. Kubota, M. Kawasaki, H. S. Park, D. Shindo, N. Nagaosa, and Y. Tokura, Large anisotropic deformation of skyrmions in strained crystal, Nature Nanotechnology 10, 589 (2015).DOI
  12. [12]M. Lee, W. Kang, Y. Onose, Y. Tokura, and N. P. Ong, Unusual hall effect anomaly in mnsi under pressure, Phys. Rev. Lett. 102, 186601 (2009).pubmed.ncbi.nlm.nih.gov/19518894
  13. [13]R. Ritz, M. Halder, C. Franz, A. Bauer, M. Wagner, R. Bamler, A. Rosch, and C. Pfleiderer, Giant generic topological hall resistivity of mnsi under pressure, Phys. Rev. B 87, 134424 (2013).DOI
  14. [14]M. Deutsch, O. L. Makarova, T. C. Hansen, M. T. Fernandez-Diaz, V. A. Sidorov, A. V. Tsvyashchenko, L. N. Fomicheva, F. Porcher, S. Petit, K. Koep­ernik, U. K. Rößler, and I. Mirebeau, Two-step pressure-induced collapse of magnetic order in the mnge chiral magnet, Phys. Rev. B 89, 180407 (2014).DOI
  15. [15]U. K. Rößler, Ab initio study on magnetism and pressure-induced transitions in cubic MnGe, Journal of Physics: Conference Series 391, 012104 (2012).
  16. [16]K. Shibata, X. Z. Yu, T. Hara, D. Morikawa, N. Kanazawa, K. Kimoto, S. Ishiwata, Y. Matsui, and Y. Tokura, Towards control of the size and helicity of skyrmions in helimagnetic alloys by spin–orbit coupling, Nature Nanotechnology 8, 723 (2013).DOI
  17. [17]S. V. Grigoriev, N. M. Potapova, S.-A. Siegfried, V. A. Dyadkin, E. V. Moskvin, V. Dmitriev, D. Menzel, C. D. Dewhurst, D. Chernyshov, R. A. Sadykov, L. N. Fomicheva, and A. V. Tsvyashchenko, Chiral properties of structure and magnetism in Mn1−xFexGe compounds: When the left and the right are fighting, who wins?, Phys. Rev. Lett. 110, 207201 (2013).
  18. [18]C. S. Spencer, J. Gayles, N. A. Porter, S. Sugimoto, Z. Aslam, C. J. Kinane, T. R. Charlton, F. Freimuth, S. Chadov, S. Langridge, J. Sinova, C. Felser, S. Blügel, Y. Mokrousov, and C. H. Marrows, Helical magnetic structure and the anomalous and topological hall effects in epitaxial b20 Fe1−yCoyGe films, Phys. Rev. B 97, 214406 (2018).DOI
  19. [19]S. X. Huang and C. L. Chien, Extended skyrmion phase in epitaxial FeGe(111) thin films, Phys. Rev. Lett. 108, 267201 (2012).
  20. [20]F. Jonietz, S. Mulbauer, C. Pfleiderer, A. Neubauer, W. Munzer, A. Bauer, T. Adams, R. Georgii, P. Boni, R. A. Duine, K. Everschor, M. Garst, and A. Rosch, Spin Transfer Torques in MnSi, Science 1648 (2011).
  21. [21]N. Kanazawa, Y. Onose, T. Arima, D. Okuyama, K. Ohoyama, S. Wakimoto, K. Kakurai, S. Ishiwata, and Y. Tokura, Large topological hall effect in a short-period helimagnet mnge, Phys. Rev. Lett. 106, 156603 (2011).pubmed.ncbi.nlm.nih.gov/21568591
  22. [22]G. Chang, S. Y. Xu, B. J. Wieder, D. S. Sanchez, S. M. Huang, I. Belopolski, T. R. Chang, Z. Zhang, A. Bansil, H. Lin, and M. Z. Hasan, Unconventional Chiral Fermions and Large Topological Fermi Arcs in RhSi, Physical Review Letters 119, 1 (2017).
  23. [23]D. Takane, Z. Wang, S. Souma, K. Nakayama, T. Nakamura, H. Oinuma, Y. Nakata, H. Iwasawa, C. Cacho, T. Kim, K. Horiba, H. Kumigashira, T. Takahashi, Y. Ando, and T. Sato, Observation of Chiral Fermions with a Large Topological Charge and Associated Fermi-Arc Surface States in CoSi, Physical Review Letters 122, 76402 (2019), 1809.01312.
  24. [24]M. Yao, K. Manna, Q. Yang, A. Fedorov, V. Voroshinin, B. Valentin Schwarze, J. Hornung, S. Chattopadhyay, Z. Sun, S. N. Guin, J. Wosnitza, H. Borrmann, C. Shekhar, N. Kumar, J. Fink, Y. Sun, and C. Felser, Observation of giant spin-split Fermi-arc with maximal Chern number in the chiral topological semimetal PtGa, Nature Communications 11, 1 (2020).DOI
  25. [25]D. S. Sanchez, I. Belopolski, T. A. Cochran, X. Xu, J. X. Yin, G. Chang, W. Xie, K. Manna, V. Süß, C. Y. Huang, N. Alidoust, D. Multer, S. S. Zhang, N. Shumiya, X. Wang, G. Q. Wang, T. R. Chang, C. Felser, S. Y. Xu, S. Jia, H. Lin, and M. Z. Hasan, Topological chiral crystals with helicoid-arc quantum states, Nature 567, 500 (2019).DOI
  26. [26]L. Ludgren, O. Beckman, V. Attia, S. P. Bhattacherjee, and M. Richardson, Helical Spin Arrangement in Cubic FeGe, Physica Scripta 1, 69 (1970).DOI
  27. [27]P. Bak and M. H. Jensen, Theory of helical magnetic structures and phase transitions in MnSi and FeGe, Journal of Physics C: Solid State Physics 13, L881 (1980).DOI
  28. [28]I. E. Dzyaloshinskii, Sov. Phys. JETP 19, 960 (1964).
  29. [29]S. Okumura, S. Hayami, Y. Kato, and Y. Motome, Magnetic hedgehog lattices in noncentrosymmetric metals, Phys. Rev. B 101, 144416 (2020).DOI
  30. [30]E. Mendiwe-Tapia, M. dos Santos Dias, S. Grytsiuk, J. B. Staunton, S. Blügel, and S. Lounis, Short period magnetization texture of b20-mnge explained by thermally fluctuating local moments, Phys. Rev. B 103, 024410 (2021).
  31. [31]V. E. Dmitrienko and V. A. Chizhikov, Weak antiferromagnetic ordering induced by dzyaloshinskii-moriya interaction and pure magnetic reflections in mnsi-type crystals, Phys. Rev. Lett. 108, 187203 (2012).pubmed.ncbi.nlm.nih.gov/22681111
  32. [32]V. A. Chizhikov and V. E. Dmitrienko, Frustrated magnetic helices in mnsi-type crystals, Phys. Rev. B 85, 014421 (2012).DOI
  33. [33]V. A. Chizhikov and V. E. Dmitrienko, Multishell contribution to the dzyaloshinskii-moriya spiraling in mnsi-type crystals, Phys. Rev. B 88, 214402 (2013).DOI
  34. [34]S. Grytsiuk, M. Hoffmann, J.-P. Hanke, P. Mavropoulos, Y. Mokrousov, G. Bihlmayer, and S. Blügel, Ab initio analysis of magnetic properties of the prototype B20 chiral magnet FeGe, Phys. Rev. B 100, 214406 (2019).DOI
  35. [35]S. Grytsiuk, J. P. Hanke, M. Hoffmann, J. Bouaziz, O. Gomonay, G. Bihlmayer, S. Lounis, Y. Mokrousov, and S. Blügel, Topological-chiral magnetic interactions driven by emergent orbital magnetism, Nature Communications 11, 1 (2020), 1904.02369.DOI
  36. [36]https://jukr-fz-juelich.de.
  37. [37]D. S. G. Bauer, Ph.D. thesis, RWTH Aachen (2013).
  38. [38]N. Papanikolaou, R. Zeller, and P. H. Dederichs, Conceptual improvements of the KKR method, Journal of Physics: Condensed Matter 14, 2799 (2002).DOI
  39. [39]Since q · R_i^A = q · (r^A + τ_i) = q · r^A + ω_i and q · R_j^B = q · (R_j^B + τ_i) = q · R_j^B + ω_i, where R_j'^B = R_j^B + τ_i, τ_i is translation vector (along the magnetic unit cell) and ω_i is an angle between two spins separated by τ_i, the summation over the magnetic unit cell can be approximated by an integral ≈ 1/(2π) ∫_0^{2π} dω S(q · r^A + ω) ∑_{ij} S(q · R_j^B + ω) assuming that ω_i ≪ 2π.
  40. [40]A. Liechtenstein, M. Katsnelson, V. Antropov, and V. Gubanov, Local spin density functional approach to the theory of exchange interactions in ferromagnetic metals and alloys, Journal of Magnetism and Magnetic Materials 67, 65 (1987).DOI
  41. [41]H. Ebert and S. Mankovsky, Anisotropic exchange coupling in diluted magnetic semiconductors: Ab initio spin-density functional theory, Phys. Rev. B 79, 045209 (2009).DOI
  42. [42]R. Wäppling and L. Häggström, Mössbauer study of cubic FeGe, Physics Letters A 28, 173 (1968).DOI
  43. [43]B. Lebech, J. Bernhard, and T. Freltoft, Magnetic structures of cubic FeGe studied by small-angle neutron scattering, Journal of Physics: Condensed Matter 1, 605 (1989).DOI
  44. [44]J. P. Perdew and Y. Wang, Pair-distribution function and its coupling-constant average for the spin-polarized electron gas, Phys. Rev. B 46, 12947 (1992).pubmed.ncbi.nlm.nih.gov/10003333
  45. [45]Note, A^(2) : Q^(2) = A^(2)_L : Q^(2) = A^(2)q^2 and for erot || q D^(1) : Q^(1) ⊗ E_rot = D^(1) T_2 : Q^(1) ⊗ E_rot = D^(1)q, where I_p is identity matrices of rank p.
  46. [46]The angle between magnetic moments of nearest atoms formed due to the global spiraling can be defined as ≈ 2πa/(4λ) = 1.28°, where a is lattice parameter, λ ≈ 70a is an experimental period of the spin-spiral modulation and factor 4 counts 4 magnetic ions in the unit cell.
  47. [47]The saturation magnetic field B || q corresponds to the Zeeman energy E_Z(q_min) = −m · B = E(q_min) − E_FM, where |m| = 4μ_B = 4 × 57.8838 (μeV/T) is the total magnetic moment in the unit cell. Since the saturation magnetic filed leads to m || B we get |B| = −E_Z(q_min)/|m|.
  48. [48]A. Bauer and C. Pfleiderer, General aspects of skyrmion lattices in chiral magnets, in Topological Structures in Ferroic Materials: Domain Walls, Vortices and Skyrmions, edited by J. Seidel (Springer International Publishing, Cham, 2016) pp. 1–28.DOI
  49. [49]V. Borisov, Y. O. Kvashnin, N. Nattalis, D. Thonig, P. Thunstrom, M. Pereiro, A. Bergman, E. Sjöqvist, A. Delin, L. Nordström, and O. Eriksson, Theory of heisenberg and anisotropic exchange interactions in magnetic materials with correlated electronic structure and significant spin-orbit coupling, (2020), arXiv:2011.08209 [cond-mat.str-el].DOI
  50. [50]A. A. Povzner, A. G. Volkov, T. A. Nogovitsyna, and S. A. Bessonov, Spin Fluctuations and Concentration Magnetic Transitions in Chiral Helical Ferromagnets Fe1−xCoxSi, Physics of the Solid State 62, 92 (2020).DOI
  51. [51]https://github.com/spirit-code/spirit, .
  52. [52]The twist angles of the spirals in each sublattice A are defined locally between the nearest magnetic sites R_i^A and R_{i−1}^A of the same sublattice: φ_i^A = arccos(S_i^A · S_{i−1}^A) − q · R_{i−1,i−1}^AA, and α_i^A and β_i^A are polar coordinates of the rotation axis e_rot^A(α_i^A, β_i^A) = S_i^A × S_{i−1}^A.

Paper details

vlnc:W_885knchw
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