betaFeedback

WorksPhysics and Astronomy

Inertial domain wall characterization in layered multisublattice antiferromagnets

Abstract

The motion of a Néel-like 180∘{180}^{\circ} domain wall induced by a time-dependent staggered spin-orbit field in the layered collinear antiferromagnet Mn2_2Au is explored. Through an effective version of the two sublattice nonlinear σσ-model which does not take into account the antiferromagnetic exchange interaction directed along the tetragonal c-axis, it is possible to replicate accurately the relativistic and inertial traces intrinsic to the magnetic texture dynamics obtained through atomistic spin dynamics simulations for quasistatic processes. In the case in which the steady-state magnetic soliton motion is extinguished due to the abrupt shutdown of the external stimulus, its stored relativistic exchange energy is transformed into a complex translational mobility, being the rigid domain wall profile approximation no longer suitable. Although it is not feasible to carry out a detailed follow-up of its temporal evolution in this case, it is possible to predict the inertial-based distance travelled by the domain wall in relation to its steady-state relativistic mass. This exhaustive dynamical characterization for different time-dependent regimes of the driving force is of potential interest in antiferromagnetic domain wall-based device applications.

Introduction

In those spintronics devices that rely on domain walls (DW) as information carriers, the objective is to have ultrafast magnetization dynamics with minimal response times for reasonable external stimuli in order to optimize its operability. Antiferromagnetic (AFM) magnetic solitons can move at velocities of the order of tens of km/s in the special relativity framework [1, 2], and superluminal-like regimes can be accessed for contracted magnetic textures whose extent is comparable to the atomic spacing [3, 4]. However, it is not only important how fast a magnetic texture can move, but also how long it takes for it to move stably at a certain speed. In particular, AFM show a low exchange-mediated static DW mass and a weak motion-based deformation tendency [5, 6], therefore it is usually accurate to describe their dynamics through a Newton-like second-order differential equation of motion [7, 8]. This being the case, in the presence of dissipation and external forces the precise magnetic soliton positioning is limited by inertial effects. In this context, it becomes essential to characterize in detail the DW evolution during acceleration and deceleration processes under different time-dependent stimuli. However, in the current literature there is a clear absence of analysis of inertial dynamic signatures of magnetic solitons in real AFM materials with invariant spin spaces where the complete set of interactions are taken into account, as well as certain controversy about the existence of claims about a hypothetical universal AFM DW-like massless behavior [9].

Among the most interesting systems, it is possible to highlight the case of Mn2Au\mathrm{Mn}_2\mathrm{Au} and CuMnAs\mathrm{CuMnAs}, two complex layered AFM that can be excited efficiently through current-induced spin-orbit (SO) fields [10, 11] and whose magnetic state can be characterized combining magnetoresistance effects with image characterization in real space [12, 13, 14, 15]. The experimental observation of DW in these type of materials [16, 17, 18], as well as proposals based on thermoelectric effects to characterize their positioning [19, 20] and the possibility of extrapolating the usual characterization techniques used in ferromagnets (FM), motivates their theoretical exploration for potential AFM DW-based racetrack memories, all-spintronics architectures, or memristive-like neuromorphic computing approaches [21, 22, 23].

In this work, we study the dynamics of a one-dimensional (1D) DW in one of the FM layers of the AFM Mn2Au\mathrm{Mn}_2\mathrm{Au} by means of staggered current-induced SO fields. For this purpose, the crystal and magnetic structure of Mn2Au\mathrm{Mn}_2\mathrm{Au} is introduced in § 2, as well as all the interactions present in the system. On the other hand, in Section III we discuss how it is possible to reduce the description of the system composed of four sublattices to a two sublattice nearest neighbours-based model through the inequivalence in symmetry of the magnetic and crystallographic unit cells. To analyze the magnetic soliton dynamics taking into account the real magnetic structure and interactions of Mn2Au\mathrm{Mn}_2\mathrm{Au}, we introduce an effective version of the nonlinear σ\sigma-model that does not take into account the AFM exchange interaction J2\mathcal{J}_2 along the tetragonal crystal axis, which has a null projection along the DW propagation direction, assumption that is supported by the atomistic spin dynamics simulations shown in Suppl. Notes I and II. Moreover, we demonstrate that it is possible, assuming that the magnetic texture behaves as a rigid entity during its motion, to reduce the Lorentz-invariant formalism to a Newton-like second-order differential equation of motion. To test our theoretical formalism, in Section IV we perform atomistic spin dynamics simulations that reveal the relativistic and inertial DW traces for SO field-based quasistatic processes. Notably, when the external stimulus is turned off abruptly interrupting the simulated steady-state motion, the magnetic soliton propagates further than expected via the rigid profile approximation. During the field-free deceleration regime, the relativistic exchange energy stored by the magnetic texture during its previous dynamic evolution is transformed into a complex translational mobility. In this line, we found a reproducible quasilinear correlation between the after-pulse distance travelled by the DW and its steady-state relativistic mass. Finally, the conclusions of our work are exposed in § 4.

Physical system

We consider the layered collinear AFM Mn2_2Au. This material is interesting because it is a good conductor, it has a strong magnetocrystalline anisotropy [24, 25, 26], and a Néel temperature well above room temperature [27]. To characterize the considered system, which is exposed in Figure 1 (a) [28], we write down the interactions that configure the energy, ww, for the conventional tetragonal unit cell of Mn2_2Au [4, 29], which is given by

Crystal and spin structure of Mn2Au tetragonal unit cell and related magnetic configurations

Figure 1. (a) Crystal and spin structure of the Mn2_2Au tetragonal unit cell along with the types of atoms and sublattices present in the system, where the magnetic Mn-based layers are numbered and their corresponding unit magnetization vectors, mi\mathbf{m}_i, and position vectors, ri\mathbf{r}_i, are indicated. Distribution of the exchange bonds of AFM origin, J1\mathcal{J}_1 and J2\mathcal{J}_2, and the one of FM nature, J3\mathcal{J}_3, as well as the in-plane xyxy basal lattice period a0a_0 and the out-of-plane height cc parameters. (b) Sketch of the distribution of the Néel-like DW magnetization, mi\mathbf{m}_i, and the SO field, HiSO\mathbf{H}^{\mathrm{SO}}_i, in each of the magnetic sublattices, denoted by A and B, together with the definition of the two types of Néel order parameters involving different layers, ℓα=(m3−m2)/2\boldsymbol{\ell}_{\alpha}=(\mathbf{m}_3-\mathbf{m}_2)/2 and ℓβ=(m1,3−m2,4)/2\boldsymbol{\ell}_{\beta}=(\mathbf{m}_{1,3}-\mathbf{m}_{2,4})/2. (c) View from the top of the unit cell along the zz-th spatial direction of the distribution of the first nearest neighbours of the Mn atom of layer 2 located at the position xix_i along the xx-th axis, characterized by m2(xi)\mathbf{m}_2(x_i). Those neighbours of its same sublattice, at a distance a0a_0, are denoted by m2(xi±1)\mathbf{m}_2(x_{i\pm1}), being mediated by the AFM exchange interaction J3\mathcal{J}_3, and those from layer 1, located at an in-plane spacing a0/2a_0/2, are represented by m1(xi±1/2)\mathbf{m}_1(x_{i\pm1/2}), and are connected through the AFM exchange contribution encoded by J1\mathcal{J}_1. (d) Description of the unit AFM vector, ℓ=(mA−mB)/2\boldsymbol{\ell}=(\mathbf{m}_{\mathrm{A}}-\mathbf{m}_{\mathrm{B}})/2, in terms of the polar out-of-plane ϵ\epsilon and in-plane azimuthal φ\varphi angles relative to the Cartesian coordinate system.

w=−∑⟨i,j⟩Jijmi⋅mj−K2⊥∑i(mi⋅z^)2−K2∥∑i(mi⋅y^)2−K4⊥2∑i(mi⋅z^)4−K4∥2∑i[(mi⋅u^1)4+(mi⋅u^2)4]−μ0μs∑imi⋅HiSO,(1)\begin{aligned} w ={}& -\sum_{\langle i,j\rangle}\mathcal{J}_{ij}\mathbf{m}_i\cdot\mathbf{m}_j-K_{2\perp}\sum_i(\mathbf{m}_i\cdot\hat{\mathbf{z}})^2 \\ &-K_{2\parallel}\sum_i(\mathbf{m}_i\cdot\hat{\mathbf{y}})^2-\frac{K_{4\perp}}{2}\sum_i(\mathbf{m}_i\cdot\hat{\mathbf{z}})^4 \\ &-\frac{K_{4\parallel}}{2}\sum_i\left[(\mathbf{m}_i\cdot\hat{\mathbf{u}}_1)^4+(\mathbf{m}_i\cdot\hat{\mathbf{u}}_2)^4\right] \\ &-\mu_0\mu_s\sum_i\mathbf{m}_i\cdot\mathbf{H}^{\mathrm{SO}}_i, \tag*{(1)} \end{aligned}

where the sum ⟨i,j⟩\langle i,j\rangle runs only over first nearest neighbours whose atomic positions are labeled by the indices i,ji,j, being represented the unit magnetic moment in the ii-th lattice position by mi\mathbf{m}_i. The symbols x^\hat{\mathbf{x}}, y^\hat{\mathbf{y}}, z^\hat{\mathbf{z}} refer to the unit vectors along the xx-, yy-, and zz-th spatial directions in the Cartesian coordinate system, while the unit vectors u^1,2\hat{\mathbf{u}}_{1,2} represent the in-plane xyxy-based directions u1=[110]\mathbf{u}_1=[110] and u2=[11ˉ0]\mathbf{u}_2=[1\bar{1}0]. Furthermore, as it can be seen in Figure 1 (a), the lattice constant along the xx- and yy-th directions in the basal planes is represented by a0=3.328a_0=3.328 Å while the size of the conventional unit cell along the zz-th direction is given by c=8.539c=8.539 Å [27]. Within the conventional unit cell there are two Mn atoms per each type of sublattice, A and B, giving rise to a total of four magnetic atoms. In fact, the magnetic moment of each of these Mn atoms, which coincides with the net contribution to each FM layer in the unit cell, would be given by μs=4μB\mu_s=4\mu_{\mathrm{B}} [25]. Additionally, as it can be seen in Figure 1 (a), there are three types of exchange contributions between magnetic moments mi\mathbf{m}_i and mj\mathbf{m}_j in the unit cell, which are represented

conventional unit cell there are two Mn atoms per each type of sublattice, A and B, giving rise to a total of four magnetic atoms. In fact, the magnetic moment of each of these Mn atoms, which coincides with the net contribution to each FM layer in the unit cell, would be given by μs=4μB\mu_s=4\mu_{\mathrm{B}} [25]. Additionally, as it can be seen in Figure 1 (a), there are three types of exchange contributions between magnetic moments mi\mathbf{m}_i and mj\mathbf{m}_j in the unit cell, which are represented by the exchange integrals Jij\mathcal{J}_{ij}. This set is composed by the J1\mathcal{J}_1, J2\mathcal{J}_2, and J3\mathcal{J}_3 parameters, where the first two are AFM, being J1=−(396 K) kB\mathcal{J}_1 = -(396\,\mathrm{K})\,k_{\mathrm{B}} and J2=−(532 K) kB\mathcal{J}_2 = -(532\,\mathrm{K})\,k_{\mathrm{B}}, and the third one is FM, being J3=(115 K) kB\mathcal{J}_3 = (115\,\mathrm{K})\,k_{\mathrm{B}} [25, 26, 27], where kBk_{\mathrm{B}} is the Boltzmann constant. On the other hand, it is possible to appreciate that there are two types of tetragonal anisotropies in the system, given by K4∥=1.8548×10−25 JK_{4\parallel}=1.8548\times10^{-25}\,\mathrm{J} and K4⊥=2K4∥K_{4\perp}=2K_{4\parallel}, and another two of uniaxial origin, denoted as K2⊥=−1.303×10−22 JK_{2\perp}=-1.303\times10^{-22}\,\mathrm{J} and K2∥=7K4∥K_{2\parallel}=7K_{4\parallel} [4, 20, 24]. The last term in Eq. (1) is the Zeeman-like contribution, where μ0\mu_0 denotes the vacuum permeability and HiSO\mathbf{H}^{\mathrm{SO}}_i expresses the staggered SO field on each ii-th lattice site which, because the locally broken inversion symmetry occurs along the zz-th spatial direction, is HA,BSO=±HySOy^\mathbf{H}^{\mathrm{SO}}_{\mathrm{A,B}}=\pm H^{\mathrm{SO}}_y\hat{\mathbf{y}} when the electric current density, j\mathbf{j}, is injected along j∥x^\mathbf{j}\parallel\hat{\mathbf{x}}, and HA,BSO=∓HxSOx^\mathbf{H}^{\mathrm{SO}}_{\mathrm{A,B}}=\mp H^{\mathrm{SO}}_x\hat{\mathbf{x}} when j∥y^\mathbf{j}\parallel\hat{\mathbf{y}} [10].

Theoretical Framework

In view of the different interactions present in the system, collected by Eq. (1), the magnetization is constrained in-plane for each Mn-based xyxy FM layer. Thus, the type of magnetic texture that can be stabilized in each of these planes is a 1D 180° Néel-like DW, as it is exposed in Figure 1 (b). When proposing how it is possible to approach the analytical characterization of the dynamics of a 1D DW in Mn2_2Au, it is worth noting that the conventional unit cell consists of four staggered magnetized layers along the zz-th spatial direction connected through two types of AFM exchange contributions, J1\mathcal{J}_1 and J2\mathcal{J}_2, which makes it difficult to define a unique Néel order parameter in the system. However, it is feasible to reduce its characterization to a two sublattice single staggered vector-based description due to the symmetric inequivalence of the magnetic and crystallographic unit cells. To carry out this discussion, let us use the numbering of the Mn planes in accordance with Figure 1 (a). This being the case, it is possible to differentiate two crystallographically identical Mn-based groups: one made up of planes 1 and 4, and the other by sheets 2 and 3. We note that, if an inversion transformation is carried out with respect to the unit cell center position, operation which would be characterized through their position vectors ri\mathbf{r}_i, it is possible to obtain that crystallographically the Mn atoms of plane 1 are transformed into those of the layer 4, and vice versa (this is, r1,4→−r4,1\mathbf{r}_{1,4}\rightarrow-\mathbf{r}_{4,1}). This can be extrapolated to the case of those Mn atoms that reside in the layers 2 and 3 (which would be represented by r2,3→−r3,2\mathbf{r}_{2,3}\rightarrow-\mathbf{r}_{3,2}). However, the crystallographic symmetry is not preserved if the AFM ordering of the magnetic moments in the Mn sites is taken into account because the magnetic moments that exist in the planes 1-4 and 2-3 are antiparallel with respect to each other within the exchange approximation. It is precisely this broken inversion symmetry that gives rise to the staggered SO field, HiSO\mathbf{H}^{\mathrm{SO}}_i, included in Eq. (1), in each type of magnetic sublattice, which allows to induce the AFM dynamics.

In this line, taking into account that Mn2_2Au is a magnetically-based centro-asymmetric AFM, it is possible to introduce four vectors in the system, one FM vector, ma\mathbf{m}_a, and three AFM vectors, li\mathbf{l}_i, as linear combinations of the four sublattice magnetization vectors, mi\mathbf{m}_i, giving this as a result: ma=(m1+m2+m3+m4)/4\mathbf{m}_a=(\mathbf{m}_1+\mathbf{m}_2+\mathbf{m}_3+\mathbf{m}_4)/4, la=(m1−m2−m3+m4)/4\mathbf{l}_a=(\mathbf{m}_1-\mathbf{m}_2-\mathbf{m}_3+\mathbf{m}_4)/4, lb=(−m1+m2−m3+m4)/4\mathbf{l}_b=(-\mathbf{m}_1+\mathbf{m}_2-\mathbf{m}_3+\mathbf{m}_4)/4, and lc=(m1+m2−m3−m4)/4\mathbf{l}_c=(\mathbf{m}_1+\mathbf{m}_2-\mathbf{m}_3-\mathbf{m}_4)/4 [30]. In fact, one of these AFM vectors, namely lb\mathbf{l}_b, can be chosen as the main one to define the system due to the specific magnetic symmetry of the Mn2_2Au unit cell. At this point, we have to remember that planes 1-3 and 2-4 are magnetically identical, the relative magnetization direction being parallel to each other. Due to this, we can introduce a two sublattice model made up of Mn-based layers 2 and 3 (of type B and A represented in Figure 1 (b), respectively) assuming that m1=m3\mathbf{m}_1=\mathbf{m}_3 and m2=m4\mathbf{m}_2=\mathbf{m}_4. This has a result that the AFM vectors are given now by la,c=0\mathbf{l}_{a,c}=0, lb=(m3−m2)/2\mathbf{l}_b=(\mathbf{m}_3-\mathbf{m}_2)/2, while the FM vector is represented by ma=(m3+m2)/2\mathbf{m}_a=(\mathbf{m}_3+\mathbf{m}_2)/2. This allows defining for Mn2_2Au the main AFM vector as lα=lb=(m3−m2)/2\mathbf{l}_\alpha=\mathbf{l}_b=(\mathbf{m}_3-\mathbf{m}_2)/2 and the magnetization vector as mα=ma=(m3+m2)/2\mathbf{m}_\alpha=\mathbf{m}_a=(\mathbf{m}_3+\mathbf{m}_2)/2 [30]. In this way, it is possible to describe the dynamics in the layered AFM Mn2_2Au through a two magnetic sublattice formalism taking into account only the two FM embedded layers 2 and 3, thus excluding from consideration the layers 1 and 4 for this purpose. We introduce also a more general definition of said variables in terms of the two types of magnetic layers of the system, A and B, as it is shown in Figure 1 (b), with which we have that l=(mA−mB)/2\mathbf{l}=(\mathbf{m}_{A}-\mathbf{m}_{B})/2 and m=(mA+mB)/2\mathbf{m}=(\mathbf{m}_{A}+\mathbf{m}_{B})/2, respectively, which is consistent with the lα\mathbf{l}_\alpha and mα\mathbf{m}_\alpha characterization.

To address the analytical description of the system, it is necessary to evaluate how many nearest neighbours exchange-based bonds have an impact on the inhomogeneous DW transition. For this, as it can be seen in Figure 1 (c), which shows a top view of the conventional unit cell, we will focus on the number of relevant first nearest neighbours for a Mn atom of layer 2 in an arbitrary position xix_i along the xx-th spatial direction, which is characterized by the unit magnetization vector m2(xi)\mathbf{m}_2(x_i). In this line, it is possible to observe that said atom has four intersublattice first neighbours on layer 1, at a distance a0/2a_0/2 along the xx-th axis, which are shown through m1(xi±1/2)\mathbf{m}_1(x_{i\pm1/2}), which are mediated by the interaction characterized by the AFM J1\mathcal{J}_1 parameter. Also, the aforementioned atomic position has two first intrasublattice neighbours along the xx-th spatial direction, at a distance a0a_0, characterized by m2(xi±1)\mathbf{m}_2(x_{i\pm1}), interacting through the FM exchange J3\mathcal{J}_3 contribution. It should be noted that both, the first intrasublattice neighbours in layer 2 along the yy-th axis and the only intersublattice first nearest neighbour, mediated by the exchange interaction J2\mathcal{J}_2, of layer 3 along the zz-th axis, are not taken into account because they do not impose any type of exchange penalty to determine the static or dynamic DW configuration in each of the sublattices of the system. In Suppl. Note I it can be found a simulations-based discussion about the role of the AFM exchange interaction J2J_2 in this regard. On the other hand, due to the relative order of magnitude of the uniaxial magnetic anisotropy constants compared to the tetragonal ones, being given by K2⊥/K4⊥=351K_{2\perp}/K_{4\perp}=351 and K2∥/K4∥=7K_{2\parallel}/K_{4\parallel}=7, the fourth-order anisotropy constants will be neglected in the main approximation from now on.

In this way, we readjust the energy given by Eq. (1) for the case of a two sublattice-based description in terms of the orthogonal set of unit vectors defined by l\boldsymbol{l} and m\boldsymbol{m}, which satisfy the conditions m2+l2=1\boldsymbol{m}^{2}+\boldsymbol{l}^{2}=1 and m⋅l=0\boldsymbol{m}\cdot\boldsymbol{l}=0. We take into account that even though these variables are constructed in terms of the magnetic sublattice types A and B, as discussed above, they refer from now on to the two FM embedded layers 2 and 3, in accordance with the definition of mα\boldsymbol{m}_{\alpha} and lα\boldsymbol{l}_{\alpha}, the latter being explicitly represented in Figure 1 (b). However, in our case we are interested in describing the dynamics of a single magnetic texture, which occurs entirely in a single sublattice of the system. Since the introduction of the AFM vectors m\boldsymbol{m} and l\boldsymbol{l} implicitly assumes that the motion of a single DW occurs across both magnetic sublattices, we must halve the resulting magnetic anisotropy and Zeeman-like energies. Therefore, the energy, ww, can be rewritten, taking into account the construction of the exchange part of the energy exposed in Suppl. Note II within an effective version of the nonlinear σ\sigma-model framework [31, 32, 33], due to the non-inclusion of the AFM exchange interaction along the c-axis of the system encoded by J2J_2, in the exchange limit [34, 35], as follows

w=12Am2+18a(∂xl)2+wa(l)−2γℏl⋅HSO(2)w=\frac{1}{2}A\boldsymbol{m}^{2}+\frac{1}{8}a\left(\partial_x\boldsymbol{l}\right)^{2}+w_a(\boldsymbol{l})-2\gamma\hbar\boldsymbol{l}\cdot\boldsymbol{H}^{\mathrm{SO}} \tag*{(2)}

where we introduced the homogeneous AFM exchange parameter, A=16∣J1∣A=16\lvert J_1\rvert, and the inhomogeneous FM-like exchange constant, given by a=8a02(J3+∣J1∣/2)a=8a_0^2(J_3+\lvert J_1\rvert/2). Here the term wa(l)w_a(\boldsymbol{l}) encapsulates the uniaxial anisotropic contributions of the system, wa(l)=∣K2⊥∣(l⋅z^)2−K2∥(l⋅y^)2w_a(\boldsymbol{l})=\lvert K_{2\perp}\rvert(\boldsymbol{l}\cdot\hat{\boldsymbol{z}})^2-K_{2\parallel}(\boldsymbol{l}\cdot\hat{\boldsymbol{y}})^2, and ∂x\partial_x expresses the variation of the order parameter along the xx-th spatial direction. This last statement is because it has been chosen that the current is injected along j∥x^\boldsymbol{j}\parallel\hat{\boldsymbol{x}}, so HSO=HySOy^\boldsymbol{H}^{\mathrm{SO}}=H_y^{\mathrm{SO}}\hat{\boldsymbol{y}}, inducing the DW motion along the xx-th spatial direction. Also, we have rewritten the Zeeman-like term of Eq. (1) taking into account that μ0μs=2γℏ\mu_0\mu_s=2\gamma\hbar, where γ\gamma represents the gyromagnetic ratio and ℏ\hbar is the reduced Planck constant.

Along the same line, it is possible to introduce how the Landau-Lifshitz-Gilbert (LLG) equations of the magnetic sublattice magnetization motions look like in the terms of l\boldsymbol{l} and m\boldsymbol{m} vectors within the exchange limit [35, 36], which are given by

l˙=γHmeff×l(3)\dot{\boldsymbol{l}}=\gamma\boldsymbol{H}^{\mathrm{eff}}_{m}\times\boldsymbol{l} \tag*{(3)}
m˙=(γHleff−αl˙)×l(4)\dot{\boldsymbol{m}}=\left(\gamma\boldsymbol{H}^{\mathrm{eff}}_{l}-\alpha\dot{\boldsymbol{l}}\right)\times\boldsymbol{l} \tag*{(4)}

where Hl,meff\boldsymbol{H}^{\mathrm{eff}}_{l,m} refers to the effective magnetic fields associated with the vector variables l,m\boldsymbol{l},\boldsymbol{m}. These effective fields can be expressed as Hl,meff=−12γℏδwδ(l,m)\boldsymbol{H}^{\mathrm{eff}}_{l,m}=-\frac{1}{2\gamma\hbar}\frac{\delta w}{\delta(\boldsymbol{l},\boldsymbol{m})}, where δ\delta represents the functional derivative, α\alpha the phenomenological Gilbert damping parameter, which accounts for the dissipation processes, and the dot over a variable points out its derivative with respect to time, tt. Through Eq. (3), it can be found that m=2ℏ(l˙×l)/A\boldsymbol{m}=2\hbar(\dot{\boldsymbol{l}}\times\boldsymbol{l})/A, expression which can be substituted in Eq. (4) to obtain a second-order differential equation only in terms of the unit staggered AFM vector l\boldsymbol{l}, which will be expressed by

l×[(∂x2l)−1vm2l¨+h−4a∂wa(l)∂l−ηl˙]=0(5)\boldsymbol{l}\times\left[\left(\partial_x^2\boldsymbol{l}\right)-\frac{1}{v_m^2}\ddot{\boldsymbol{l}}+\boldsymbol{h}-\frac{4}{a}\frac{\partial w_a(\boldsymbol{l})}{\partial\boldsymbol{l}}-\eta\dot{\boldsymbol{l}}\right]=0 \tag*{(5)}

where vmv_m represents the maximum magnon group velocity of the medium, which is given by vm=aA/(4ℏ)=2a02∣J1∣(J3+∣J1∣/2)/ℏ=43.39v_m=\sqrt{aA}/(4\hbar)=2a_0\sqrt{2\lvert J_1\rvert(J_3+\lvert J_1\rvert/2)}/\hbar=43.39 km/s, h\boldsymbol{h} encodes the reduced SO field as h=8γℏHySOy^/a\boldsymbol{h}=8\gamma\hbar H_y^{\mathrm{SO}}\hat{\boldsymbol{y}}/a, and η\eta denotes the dissipative parameter expressed as η=8αℏ/a\eta=8\alpha\hbar/a. Interestingly, it has been predicted that an AFM exchange interaction like J2J_2, perpendicular to the inhomogeneous DW transition, should govern the value of vmv_m related to the low-frequency acoustic branch, contrary to what is demonstrated in Suppl. Notes I, both within and outside of the standard nonlinear σ\sigma-model framework [8, 37, 38].

In accordance with what it is shown in Figure 1 (d), it is possible to parameterize through spherical coordinates the unit Néel order parameter taking into account which is the in-plane easy-axis direction, giving rise to l=(sin⁡φcos⁡ε,cos⁡φ,−sin⁡ε)\boldsymbol{l}=(\sin\varphi\cos\varepsilon,\cos\varphi,-\sin\varepsilon), where φ\varphi represents the azimuthal angle, which accounts for the rotation of the magnetization in the xyxy plane being measured from the yy-th axis, while ε\varepsilon expresses the polar angle, which describes the out-of-plane canting being characterized from the xyxy plane. Because we are working on the exchange limit, it is possible to assume that ε≃0\varepsilon\simeq0, whereby the reduced AFM vector can be expressed as l≃(sin⁡φ,cos⁡φ,0)\boldsymbol{l}\simeq(\sin\varphi,\cos\varphi,0). This being the case, it is possible to reduce Eq. (5) to a sine-Gordon wave-like equation [39, 40], with the following functional form

1vm2φ¨−(∂x2φ)+12Δ02sin⁡2φ+hsin⁡φ=−ηφ˙(6)\frac{1}{v_m^2}\ddot{\varphi}-\left(\partial_x^2\varphi\right)+\frac{1}{2\Delta_0^2}\sin2\varphi+h\sin\varphi=-\eta\dot{\varphi} \tag*{(6)}

where Δ0\Delta_0 stands for the DW width at rest, which is given by Δ0=a/(8K2∥)=a0(J3+∣J1∣/2)/K2∥=19.17\Delta_0=\sqrt{a/(8K_{2\parallel})}=a_0\sqrt{(J_3+\lvert J_1\rvert/2)/K_{2\parallel}}=19.17 nm, and where h=8γℏHySO/ah=8\gamma\hbar H_y^{\mathrm{SO}}/a denotes the reduced scalar SO field.

At this point, it is convenient to consider the magnetic texture dynamics within the framework of the well-known collective coordinates approach [41]. For this, it is usual to introduce what is known as Walker-like rigid profile through the angular variable that defines the spatio-temporal evolution of the magnetization, φ(x,t)=2arctan⁡exp⁡[(x−q(t))/Δ(t)]\varphi(x,t)=2\arctan\exp[(x-q(t))/\Delta(t)] [42], with qq being the DW center position and Δ\Delta being the dynamic DW width. Due to the Lorentz invariance shown by Eqs. (5) and (6), the magnetic soliton dynamics in AFM shows emergent special relativity signatures. In particular, the DW width decreases as the velocity of the magnetic texture, q˙\dot{q}, increases, which is given by the expression Δ(t)=Δ0β(t)\Delta(t)=\Delta_0\beta(t), where β(t)=1−(q˙(t)/vm)2\beta(t)=\sqrt{1-\left(\dot{q}(t)/v_{\mathrm{m}}\right)^2} represents the Lorentz factor. In order to avoid the excitation of internal modes of the magnetic texture, we focus on quasistatic processes, being its spatial extent variation governed entirely by the special relativity-based Lorentz factor, so we can neglect the time derivatives of the DW width Δ\Delta. In this way, we obtain that

1Δvm2q¨+ηΔq˙−h=0.(7)\frac{1}{\Delta v_{\mathrm{m}}^2}\ddot{q}+\frac{\eta}{\Delta}\dot{q}-h=0. \tag*{(7)}

As it can be seen in Eq. (7), we have a Newton-like second-order differential equation for the time evolution of the DW center position, qq, which explicitly shows the inertial nature of the magnetic texture [7, 8]. In the particular case in which a constant SO field is applied, it is possible to access a steady-state-like DW motion regime after the accommodation of the soliton to its new dynamic state. In this sense, we can reduce the previous equation to a compact expression that accounts for the steady-state DW velocity, which we denote from now on as v=q˙v=\dot{q}, which is given by

v=vm1+(vm/v0)2,(8)v=\frac{v_{\mathrm{m}}}{\sqrt{1+\left(v_{\mathrm{m}}/v_0\right)^2}}, \tag*{(8)}

where v0=hΔ0/ηv_0=h\Delta_0/\eta.

Relativistic and inertial domain wall dynamic signatures

To verify the predictions obtained above through our effective version of the nonlinear σ\sigma-model, we have performed atomistic spin dynamics simulations of the real crystallographic and magnetic conventional unit cell of Mn2_2Au, as it is shown in Figure 1 (a), taking into account all interactions of the system, as it is collected in Eq. (1). With this objective in mind, the system of the coupled LLG equations of motion of the local magnetic moments mi\mathbf{m}_i, being given by

1+α2γm˙i=−mi×Hieff−αmi×(mi×Hieff),(9)\frac{1+\alpha^2}{\gamma}\dot{\mathbf{m}}_i=-\mathbf{m}_i\times\mathbf{H}^{\mathrm{eff}}_i-\alpha\mathbf{m}_i\times\left(\mathbf{m}_i\times\mathbf{H}^{\mathrm{eff}}_i\right), \tag*{(9)}

is evaluated numerically site by site through a fifth-order Runge-Kutta method. Here, Hieff\mathbf{H}^{\mathrm{eff}}_i represents the effective field at each lattice position, which depends on the interactions exposed in Eq. (1), as Hieff=−1μ0μsδwδmi\mathbf{H}^{\mathrm{eff}}_i=-\frac{1}{\mu_0\mu_s}\frac{\delta w}{\delta\mathbf{m}_i}, and the damping parameter is α=0.001\alpha=0.001 [4, 20, 43]. In this case, the computational domain consists of 60000 cells along the xx-th propagation direction, one cell width with periodic boundary conditions along yy-th direction, and

one cell thick along the zz-th direction [44]. Due to their simple functional forms in terms of intrinsic parameters of the material, we can test the validity of our analytical formalism by comparing the values of the DW width at rest, Δ0\Delta_0, and the maximum magnon group velocity of the medium, vmv_{\mathrm{m}}, with the simulated ones. This being the case, we have found that the simulated values are Δ0=19.78\Delta_0=19.78 nm and vm=43.3v_{\mathrm{m}}=43.3 km/s. The theoretically-predicted DW width at rest Δ0\Delta_0 presents a good correspondence with the fitted rigid DW profile-based simulated value, differing only by a 3.1%, which possibly comes from the non-inclusion in the analytical model, for simplicity, of the in-plane tetragonal anisotropy contribution encoded by K4∥K_{4\parallel}. On the other hand, the maximum magnon group velocity vmv_{\mathrm{m}} obtained analytically and by simulations coincide in a 99.93%. With this satisfactory correspondence between the theory and simulations, which supports the non-inclusion in our formalism of the AFM exchange interaction given by J2J_2, we can explore the emergent special relativity signatures during steady-state DW dynamic processes. As it can be seen in Figure 2, the saturation of the magnetic texture velocity, vv, as it is predicted by Eq. (8), and the contraction of the DW width, Δ\Delta, in correspondence with the expression Δ=Δ0β\Delta= \Delta_0\beta, as the SO field, HySOH_y^{\mathrm{SO}}, increases are verified.

Comparison of the relativistic signatures for steady-state processes in Mn2Au

Figure 2. Comparison of the relativistic signatures for steady-state processes in Mn2_2Au obtained through atomistic spin dynamics simulations and theory. (a) Saturation of the velocity, vv, of the magnetic texture as the SO field, HySOH_y^{\mathrm{SO}}, increases, being based the analytical formalism in Eq. (8). (b) DW width, Δ\Delta, contraction as the speed vv of the magnetic soliton increases, the theoretical prediction coming from the combination of the relativistic expression Δ=Δ0β\Delta=\Delta_0\beta and Eq. (8).

To explore the inertial signatures on our system, we use the time-dependent SO field-based excitation regimes represented in Figure 3 (a). As it can be seen, there are three well differentiated regions. In the first one, which covers the interval t∈[0,tr)t \in[0,t_{\mathrm{r}}), being trt_{\mathrm{r}} the time taken to reach a constant value of the field of HySO=60H_y^{\mathrm{SO}} = 60 mT, which we denote as ramping time, the rest state of the magnetic texture is disturbed through a SO field that increases linearly with time. We denote this regime as region I, and each colored dashed line in Figure 3 corresponds to a certain trt_{\mathrm{r}} that defines the aforementioned domain. Consistently with Eq. (7), which implicitly shows the existence of a non-zero DW mass, the initial response of the magnetic texture to the external stimulus is fast,

Time-dependent staggered SO field-based excitation protocol and dynamic domain-wall properties

Figure 3. (a) Time-dependent staggered SO field-based excitation protocol, HySO(t)H_y^{\mathrm{SO}}(t), applied in each Mn-based FM layer of Mn2_2Au for different ramping times, trt_{\mathrm{r}}. Dynamic time evolution of the DW velocity vv (b) and width Δ\Delta (c) obtained through atomistic spin dynamics simulations for different ramping times trt_{\mathrm{r}}. Comparison of the dynamic time evolution of the velocity vv (d) and width Δ\Delta (e) of a Néel-like DW for atomistic spin dynamics simulations and the analytical expressions given by Eq. (7) and Δ=Δ0β\Delta= \Delta_0\beta for two ramping times, tr=30t_{\mathrm{r}} = 30 and 60 ps. Each vertical colored dashed line represents the end of the ramped process for the different ramping times trt_{\mathrm{r}}, while the dashed black line denotes the instant t=100t = 100 ps at which the driving SO field HySOH_y^{\mathrm{SO}} is abruptly turned off.

but not instantaneous, as it can be seen in Figure 3 (b, c). At the time when a constant field value of 60 mT is reached, that is, at t=trt = t_{\mathrm{r}}, the magnetic soliton tends to a steady-state regime (region II), which covers the interval t∈[tr,100 ps)t \in[t_{\mathrm{r}},100\ \mathrm{ps}). This upper limit is denoted by a black dashed line when appropriate in Figure 3. Thus, it is possible to observe in Figure 3 (b, c) that, in the region II, after a brief adaptation period to the new dynamic regime, which is a sample of the inertial nature of the process, the magnetic texture moves steadily at a speed of v=42.56v = 42.56 km/s. This is very close to the maximum magnon group velocity of the medium, denoting a 98% of it, while shrinking to a width of Δ=4.08\Delta= 4.08 nm, which represents a contraction of 80% with respect to the simulated DW width at rest. Finally, at a certain moment given by t=100t = 100 ps, the SO field is abruptly switched off. This makes it possible to observe that the magnetic texture is capable of initiating an after-pulse displacement in the absence of an external stimulus at the same time as its width expands until it stops completely, as it can be seen in Figure 3 (b, c). We denote this regime as region III, and it covers the interval t∈[100,140]t \in[100,140] ps.

The range of values considered for the ramping time trt_{\mathrm{r}} has been chosen to avoid the excitation of internal DW modes during the acceleration process, in such a way that the simulations were comparable to the scenario exposed in Section III through Eq. (7). As it can be seen in Figure 3 (d, Figure 3e), in these circumstances there is a great correspondence between the simulated and the theoretically-predicted velocity of the magnetic texture, vv, and its spatial extent, Δ\Delta, in regions I and II. However, if attention is paid to Figure 3 (b, Figure 3c) in the region III, it is possible to appreciate undulations once the SO field HySOH_y^{\mathrm{SO}} is turned off. These ripples observed in the simulated DW velocity and width in the region III cause longer decay times than those predicted through a simple Newton-like pseudoparticle behaviour. Therefore, we avoid the analytical evaluation of this region through Eq. (7) because the magnetic soliton is no longer fulfilling the imposed rigid profile constraint. However, the after-pulse displacement that the magnetic texture experiences in region III is related to the increase in the exchange-based relativistic DW mass obtained during the acceleration process in region I while its width shrinked. This is a purely inertial phenomenon, which is consistent with the massive pseudoparticle behavior captured by Eq. (7), in which the higher is the dynamic DW mass after turning off the external stimulus, the greater is the braking distance travelled by it. Interestingly, it is the fact that the magnetic soliton moves along a particular direction which sets how this stored relativistic exchange energy is transformed into a translational displacement, manifesting its massive particle-like behavior rather than being dissipated in a breather-like fashion or through the emission of spin waves without prolonging its mobility. For different values of the SO field during region II, we have found through simulations that there is a quasilinear relationship between the after-pulse distance travelled by the magnetic texture, xx, normalized to the its steady-state DW width, Δ\Delta, and its steady-state DW mass, mDWm_{\mathrm{DW}}, normalized to its rest state value, mDW0m_{\mathrm{DW}}^{0}, this is, mDW/mDW0=1/βm_{\mathrm{DW}}/m_{\mathrm{DW}}^{0}=1/\beta according to Eq. (7), which can be seen in Figure 4 and can be expressed as

Quasilinear correspondence between normalized braking distance and normalized domain wall mass

Figure 4. Quasilinear correspondence between the normalized relativistic DW mass, mDW/mDW0m_{\mathrm{DW}}/m_{\mathrm{DW}}^{0}, where mDWm_{\mathrm{DW}} and mDW0m_{\mathrm{DW}}^{0} represent the aforementioned mass in steady-state and at rest, respectively, and the normalized braking distance, xx, travelled by the magnetic texture once the SO field is turned off abruptly, x/Δx/\Delta, in terms of the steady-state DW width, Δ\Delta, characterized by a linear fitting with bb and cc represent the adjustment parameters, together with their corresponding uncertainties.

xΔ=bmDWmDW0+c,(10)\frac{x}{\Delta}=b\frac{m_{\mathrm{DW}}}{m_{\mathrm{DW}}^{0}}+c, \tag*{(10)}

where bb and cc are two fitting-dependent parameters, which are given, accompanied by their associated uncertainties, by b=13.81(27)b=13.81(27) and c=−13.92(89)c=-13.92(89). It is remarkable that the DW is capable of undertaking exchange-based after-pulse displacements of the order of 4 to 11 times greater than the DW width at rest, Δ0\Delta_0, for SO fields between 10-60 mT. This accurate prediction of the braking distance experienced by the magnetic soliton opens the door to implement low-energy consumption processes.

Conclusions

We addressed the theoretical characterization of the dynamic evolution of a 1D Néel-like DW in one of the FM sublattices of the layered collinear AFM Mn2_2Au driven by the current-induced SO fields. Despite the complexity of the system, we have exploited the symmetric inequivalence between the crystallographic and the magnetic unit cell to reduce its description to a two-sublattice problem. Since the AFM exchange interaction directed along the c-axis of the system, which is encoded through J2\mathcal{J}_2, has a null projection along the 1D inhomogeneous magnetic texture transition, it has no impact on the temperature-independent standard nonlinear σ\sigma-model. Because of this, we have worked on within an effective theory framework avoiding its inclusion, methodology which can be extrapolated to layered multisublattice AFM with different exchange-oriented contributions. In the rigid profile approximation, we have shown that it is possible to reduce the dynamic description to a Newton-like second-order differential equation of motion. Comparing our formalism with atomistic spin dynamics simulations, we have been able to replicate with a high degree of precision the relativistic and inertial signatures of the magnetic texture motion during quasistatic dynamic processes within the framework of our effective model. After the abrupt shutdown of the SO field in simulations, the rigid DW profile approach is no longer supported and our analytical formalism fails to describe the after-pulse inertial dynamic regime. Interestingly, during the deceleration process the relativistic exchange energy accumulated during the previous dynamic evolution of the magnetic texture is converted into translational mobility, rather than being released in a breather-like fashion or through the emission of spin waves with non-associated displacement. We have found a quasilinear relationship that allows us to predict, for the range of simulated SO fields, the value of the braking distance travelled by the DW through the knowledge of its relativistic mass before turning off the external stimulus. This detailed dynamic characterization of the 1D magnetic texture in the complex multisublattice antiferromagnet Mn2_2Au is of potential interest for AFM DW-based technology applications.

Acknowledgements

R.R.-E., K.Y.G., and R.M.O. thanks O. Chubykalo-Fesenko, S. Khmelevskyi, A. A. Sapozhnik, M. Jourdan, A. K. Zvezdin, and B. A. Ivanov for the fruitful discussions that have helped us to improve this manuscript. The work of R.M.O. and K.Y.G. was partially supported by the STSM Grants from the COST Action CA17123 “Ultrafast opto-magneto-electronics for nondissipative information technology”. K.Y.G. acknowledges support by IKERBASQUE (the Basque Foundation for Science) and the Spanish Ministry of Science and Innovation under grant PID2019-108075RB-C33/AEI/10.13039/501100011033.

Supplemental Material: “Inertial domain wall characterization in layered multisublattice antiferromagnets”

R. Rama-Eiroa; P. E. Roy; J. M. González; K. Y. Guslienko; J. Wunderlich; R. M. Otxoa

“Simulated exchange-based intrinsic domain wall properties”

To elucidate the role that each exchange interaction plays in the static and dynamic domain wall (DW) properties, we have performed atomistic spin dynamics simulations where we have varied, through a numerical scaling factor, given by κ\kappa, the magnitude of the exchange parameters of the system. By varying only one of these at each time, we can compare the analytical expressions shown in Section III of the main text with the results that we obtain from the simulations. First, and as it can be seen in Suppl. Fig. 1 (a), the simulated DW width at rest Δ0\Delta_0 does not depend on the scaling magnitude applied to the antiferromagnetic (AFM) J2\mathcal{J}_2 constant, since this interaction only guarantees that the contiguous sublattices are perfectly antiparallel, while J1\mathcal{J}_1 and J3\mathcal{J}_3 characterize the relative angle between in-plane spins and hence the spatial soliton extent. This is consistent with the expression that we have obtained in Section III of the main text, according to which the DW width at rest does not depend on the contribution governed by J2\mathcal{J}_2, that is,

Δ0=a0J3+∣J1∣/2K2∥\Delta_0 = a_0 \sqrt{\frac{\mathcal{J}_3 + |\mathcal{J}_1|/2}{K_{2\parallel}}}

As it can be appreciated, the correlation between the trend of the simulated and analytically-predicted DW width at rest Δ0\Delta_0 is clear. On the other hand, we can explore the spin wave (SW) dispersion relation of the system by varying only the magnitude of the J2\mathcal{J}_2 parameter, and see how this affects the functional form of the low-frequency acoustic branch from which it can be extracted the maximum magnon group velocity, vmv_{\mathrm{m}}, of the medium, which is characterized as

vm=(dfdk)max⁡v_{\mathrm{m}} = \left(\frac{\mathrm{d}f}{\mathrm{d}k}\right)_{\max}

where ff expresses the SW frequency and kk represents the associated wavenumber along the propagation direction. As it can be seen in Suppl. Fig. 1 (b), the low-frequency acoustic branch is independent of the magnitude associated with the exchange interaction governed by J2\mathcal{J}_2, while the high-frequency optical branch maintains its functional form for κJ2≠0\kappa\mathcal{J}_2 \ne0, but it shifts to higher frequencies as the value of the aforementioned combined variable increases. When κ=0\kappa= 0, the optical branch disappears, so it is strongly dependent on the magnitude of J2\mathcal{J}_2. However, the maximum slope of the differentiated SW dispersion relation is always found in the J2\mathcal{J}_2-independent acoustic branch, so the maximum magnon group velocity, vmv_{\mathrm{m}}, does not depend on the aforementioned interaction.

Now we can evaluate the dynamic behaviour of the magnetic texture under the action of a constant current-induced spin-orbit (SO) field, HySO=60H_y^{\mathrm{SO}} = 60 mT. Taking into account that, as we introduced in Eq. (8) in the main text, the steady-state speed, vv, has a functional form such that v=v(Δ0,vm)v = v(\Delta_0,v_{\mathrm{m}}), no dependency on J2\mathcal{J}_2 is expected. Accordingly, in Suppl. Fig. 1 (c) it is possible to see that there is a great correspondence between the simulated and analytically predicted values for the steady-state velocity vv for different numerical scalings in the exchange parameters J1\mathcal{J}_1, J2\mathcal{J}_2, and J3\mathcal{J}_3. While it does not depend on the intersublattice AFM exchange contribution J2\mathcal{J}_2, it is strongly dependent on the magnitude of the interlayer AFM exchange interaction J1\mathcal{J}_1, which is consistent with the analytic form proposed, and is affected by changes in the value of the intrasublattice ferromagnetic (FM) exchange term encoded by J3\mathcal{J}_3. On the other hand, because the theoretically-predicted steady-state DW width, Δ\Delta, is entirely conditioned by its rest state value, Δ0\Delta_0, the corresponding speed vv to the applied SO field HySOH_y^{\mathrm{SO}}, and the maximum magnon group velocity of the medium, vmv_{\mathrm{m}}, since Δ=Δ0β\Delta= \Delta_0\beta, being β\beta the Lorentz factor, again the AFM exchange interaction encoded through the J2\mathcal{J}_2 parameter should not play any role in its determination. Consequently, and as it can be seen in Suppl. Fig. 1 (d), the steady-state DW width follows the theoretically-predicted trend, only existing, as in the case of the DW width at rest Δ0\Delta_0 presented in Suppl. Fig. 1 (a), a small constant shifting between the simulated and the theoretical values. Thus, the role of J2\mathcal{J}_2 consists of acting as a hard-axis anisotropy, strongly constraining the magnetization in the Mn-based basal planes of the conventional unit cell, since this interaction is the dominant one in magnitude of the system, but it does not intervene in the static and dynamic DW configurational properties.

Comparison of simulated and theoretically predicted domain-wall widths, spin-wave dispersion relations, velocities, and domain-wall widths

Figure Supplementary Figure 1. (a) Comparison between the simulated and the theoretically predicted DW width at rest, Δ0\Delta_0, for different numerical scalings, κ\kappa, of, at each time, one of the exchange parameters of the system, J1\mathcal{J}_1, J2\mathcal{J}_2, and J3\mathcal{J}_3. (b) Simulated SW dispersion relations, including the high-frequency optical and low-frequency acoustic branches, characterized through the associated SW frequency ff and wavenumber kk, for different numerical scaling through the κ\kappa parameter of the AFM exchange interaction governed by J2\mathcal{J}_2. Comparison between the simulated and the analytically calculated steady-state velocity vv (c) and DW width Δ\Delta (d), for the case of a SO field of HySO=60H_y^{\mathrm{SO}}=60 mT, for different numerical scalings, controlled by the κ\kappa parameter, affecting, individually in each case, one of the exchange interactions of the system, characterized by J1\mathcal{J}_1, J2\mathcal{J}_2, and J3\mathcal{J}_3.

Exchange energy-based analytical description

To construct the exchange-based part of the energy, ww, specified in Eq. (2) of the main text, it is necessary to take into account the number of first nearest neighbours along the xx-th spatial direction and through which exchange interactions are mediated. For this, we will rely on Figure 1 (c) of the manuscript, and we will take as a reference an atom of the layer 2, according to the numbering of the sheets shown in Figure 1 (a) of the main text, in an arbitrary position xix_i, whose unit magnetization vector will be denoted as m2(xi)\mathbf{m}_2(x_i). Since, along the xx-th axis, the intra- and interlayer first nearest neighbours are at different distances, we will spatially denote those of layer 1 through xi±1/2x_{i\pm1/2}, characterizing the J1\mathcal{J}_1 interaction, and those of layer 2 through xi±1x_{i\pm1}, mediated by the J3\mathcal{J}_3 coefficient. This being the case, taking into account that each atom of layer 2 has four first nearest neighbours from layer 1 at a distance a0/2a_0/2 along the xx-th spatial direction, we can first characterize the J1\mathcal{J}_1 contribution to the exchange energy of the system through the introduction of the unit magnetization vectors m1(xi±1/2)\mathbf{m}_1(x_{i\pm1/2}) and their interaction with the magnetization vector of the reference atom m2(xi)\mathbf{m}_2(x_i), that is

wexc(J1)=−2J1m2(xi)⋅[m1(xi−1/2)+m1(xi+1/2)],(11)w_{\mathrm{exc}}(\mathcal{J}_1)=-2\mathcal{J}_1\mathbf{m}_2(x_i)\cdot\left[\mathbf{m}_1(x_{i-1/2})+\mathbf{m}_1(x_{i+1/2})\right], \tag*{(11)}

where it is important to note that the 2 factor that appears on the right-hand side corresponds to the fact that there are two first nearest neighbours on each side of the reference atom of the layer 2 located at xix_i.

Performing a one-dimensional Taylor series decomposition up to second order for the atoms belonging to layer 1 with respect to the reference position xix_i of the layer 2 atom, we will have

wexc(J1)=4∣J1∣m2(xi)⋅[m1(xi)+12(a02)2(∂x2m1(xi))].(12)w_{\mathrm{exc}}(\mathcal{J}_1)=4\lvert\mathcal{J}_1\rvert\mathbf{m}_2(x_i)\cdot\left[\mathbf{m}_1(x_i)+\frac{1}{2}\left(\frac{a_0}{2}\right)^2\left(\partial_x^2\mathbf{m}_1(x_i)\right)\right]. \tag*{(12)}

expression which can be rewritten in terms of the magnetization vector, m=(mA+mB)/2\mathbf{m} = (\mathbf{m}_A + \mathbf{m}_B)/2, and the Néel order parameter, l=(mA−mB)/2\mathbf{l} = (\mathbf{m}_A - \mathbf{m}_B)/2, where, in this case, due to the equivalence between the magnetic sublattices 1-3, which we will denote as type A, as it can be seen in Figure 1 (b) of the main text, and the correlation between the layers 2-4, being able to refer to them as type B, as explained in Section III of the manuscript, it is possible to redefine, in this case, the orthogonal set of unitary vectors m\mathbf{m} and l\mathbf{l}, as m=(m1+m2)/2\mathbf{m} = (\mathbf{m}_1 + \mathbf{m}_2)/2 and l=(m1−m2)/2\mathbf{l} = (\mathbf{m}_1 - \mathbf{m}_2)/2. Considering that m2+l2=1\mathbf{m}^2 + \mathbf{l}^2 = 1 and that we are working on the exchange limit regarding the exchange interaction J1\mathcal{J}_1 (m≪l\mathbf{m} \ll\mathbf{l}), it is possible to obtain that, after throwing out the spatial indices, the exchange interaction can be reduced to

wexc(J1)=8∣J1∣[m2−14(a02)2l⋅(∂x2l)].(13)w_{\mathrm{exc}}(\mathcal{J}_1) = 8\lvert\mathcal{J}_1\rvert\left[\mathbf{m}^2 - \frac{1}{4}\left(\frac{a_0}{2}\right)^2\mathbf{l}\cdot\left(\partial_x^2\mathbf{l}\right)\right]. \tag*{(13)}

Finally, taking into account that the Néel order parameter l\mathbf{l}, as well as the magnetization vector m\mathbf{m}, is a fixed length vector, it is possible to demonstrate that l⋅(∂x2l)=−(∂xl)2\mathbf{l}\cdot\left(\partial_x^2\mathbf{l}\right) = -\left(\partial_x\mathbf{l}\right)^2. Hence the exchange contribution encoded via J1\mathcal{J}_1 can be rewritten as

wexc(J1)=8∣J1∣[m2+a0216(∂xl)2].(14)w_{\mathrm{exc}}(\mathcal{J}_1) = 8\lvert\mathcal{J}_1\rvert\left[\mathbf{m}^2 + \frac{a_0^2}{16}\left(\partial_x\mathbf{l}\right)^2\right]. \tag*{(14)}

Similarly, the functional form of the intrasublattice FM exchange parameter J3\mathcal{J}_3 contribution to the system can be analyzed analogously to the case concerning layer 1. For this, the interaction of the unit magnetization vector of layer 2 in an arbitrary spatial position, m2(xi)\mathbf{m}_2(x_i), with its two first nearest neighbours along the xx-th axis located at a distance a0a_0 within the same sublattice must be taken into account, which will be given by

wexc(J3)=−J3m2(xi)⋅[m2(xi−1)+m2(xi+1)].(15)w_{\mathrm{exc}}(\mathcal{J}_3) = -\mathcal{J}_3\mathbf{m}_2(x_i)\cdot\left[\mathbf{m}_2(x_{i-1}) + \mathbf{m}_2(x_{i+1})\right]. \tag*{(15)}

Performing a Taylor series expansion up to second order along the xx-th spatial direction of the intrasublattice first nearest neighbours atoms at a distance a0a_0 from the reference atom of layer 2 located at the arbitrary position xix_i, it is possible to get that

wexc(J3)=−2J3m2(xi)⋅[m2(xi)+a022(∂x2m2(xi))].(16)w_{\mathrm{exc}}(\mathcal{J}_3) = -2\mathcal{J}_3\mathbf{m}_2(x_i)\cdot\left[\mathbf{m}_2(x_i) + \frac{a_0^2}{2}\left(\partial_x^2\mathbf{m}_2(x_i)\right)\right]. \tag*{(16)}

It is possible to rewrite this equation in terms of the AFM l\mathbf{l} and the magnetization m\mathbf{m} vectors given by the sublattice categories denoted by labels A and B, as we did earlier in Eq. (13), not being necessary to specify in this case which numerically-based layer of type A we are basing ourselves on to construct the vectors, since sublattices 1-3 are magnetically equivalent, as we pointed out previously. If we take into account, again, that we are working on the exchange limit, that the m\mathbf{m} and l\mathbf{l} variables form an orthogonal unit set (m⋅l=0\mathbf{m}\cdot\mathbf{l} = 0), and that both are fixed length vectors, we can get that

wexc(J3)=J3a02(∂xl)2.(17)w_{\mathrm{exc}}(\mathcal{J}_3) = \mathcal{J}_3a_0^2\left(\partial_x\mathbf{l}\right)^2. \tag*{(17)}

Finally, it is possible to group both contributions, given by Eqs. (14) and (S9), which will describe the static and dynamic DW configuration in one of the FM layers of the conventional tetragonal unit cell, giving rise to

wexc=12Am2+18a(∂xl)2,(18)w_{\mathrm{exc}} = \frac{1}{2}A\mathbf{m}^2 + \frac{1}{8}a\left(\partial_x\mathbf{l}\right)^2, \tag*{(18)}

where we have introduced the homogeneous AFM exchange parameter, A=16∣J1∣A = 16\lvert\mathcal{J}_1\rvert, and the inhomogeneous FM-like exchange constant, given by a=8a02(J3+∣J1∣/2)a = 8a_0^2(\mathcal{J}_3 + \lvert\mathcal{J}_1\rvert/2). It should be noted the absence of a potential contribution by the AFM exchange interaction encoded through the J2\mathcal{J}_2 parameter in the previous expression, in line with the analysis carried out in Suppl. Note I.

References

  1. [1]O. Gomonay, T. Jungwirth, and J. Sinova, Phys. Rev. Lett. 117, 017202 (2016).
  2. [2]T. Shiino, S.-H. Oh, P. M. Haney, S.-W. Lee, G. Go, B.-G. Park, and K.-J. Lee, Phys. Rev. Lett. 117, 087203 (2016).
  3. [3]H. Yang, H. Y. Yuan, M. Yan, H. W. Zhang, and P. Yan, Phys. Rev. B 100, 024407 (2019).
  4. [4]R. M. Otxoa, P. E. Roy, R. Rama-Eiroa, J. Godinho, K. Y. Glusko, and J. Wunderlich, Commun. Phys. 3, 190 (2020).
  5. [5]S. K. Kim, Y. Tserkovnyak, and O. Tchernyshov, Phys. Rev. B 90, 104406 (2014).
  6. [6]G. Tatara, C. A. Akosa, and R. M. Otxoa, Phys. Rev. Res. 2, 043226 (2020).
  7. [7]E. G. Tveten, A. Qaiumzadeh, O. A. Tretiakov, and A. Brataas, Phys. Rev. Lett. 110, 127208 (2013).
  8. [8]H. Y. Yuan, W. Wang, M.-H. Jung, and X. R. Wang, Phys. Rev. B 97, 211434 (2018).
  9. [9]S. Selzer, U. Atxitia, R. Ritzmann, D. Hinzke, and U. Nowak, Phys. Rev. Lett. 117, 107201 (2016).
  10. [10]J. Železný, H. Gao, K. Výborný, J. Zemen, J. Mašek, A. Manchon, J. Wunderlich, J. Sinova, and T. Jungwirth, Phys. Rev. Lett. 113, 157201 (2014).
  11. [11]P. Wadley, B. Howells, J. Železný, C. Andrews, V. Hills, R. P. Campion, V. Novák, K. Olejník, F. Maccherrozzi, S. S. Dhesi, et al., Science 351, 587 (2016).
  12. [12]K. Olejník, V. Schuler, X. Martí, V. Novák, Z. Kašpar, P. Wadley, R. P. Campion, K. W. Edmonds, B. L. Gallagher, J. Garcés, et al., Nat. Commun. 8, 15434 (2017).
  13. [13]M. Grzybowski, P. Wadley, K. Edmonds, R. Beardsley, V. Hills, R. Campion, B. Gallagher, J. S. Chauhan, V. Novak, T. Jungwirth, et al., Phys. Rev. Lett. 118, 057701 (2017).
  14. [14]X. Zhou, J. Zhang, F. Li, X. Chen, G. Shi, Y. Tan, Y. Gu, M. Saleem, H. Wu, F. Pan, et al., Phys. Rev. Appl. 9, 054028 (2018).
  15. [15]S. Y. Bodnar, Y. Skourski, O. Gomonay, I. Sinova, M. Kläui, and M. Jourdan, Phys. Rev. Appl. 14, 014004 (2020).
  16. [16]V. M. T. S. Barthem, C. V. Colin, R. Haettel, D. Dufeu, and D. Givord, J. Magn. Mat. 406, 289 (2016).
  17. [17]A. A. Sapozhnik, M. Filianina, S. Y. Bodnar, A. Lamirand, M.-A. Mawass, Y. Skourski, H.-J. Elmers, H. Zabel, M. Kläui, and M. Jourdan, Phys. Rev. B 97, 134429 (2018).
  18. [18]Z. Kašpar, M. Surynek, J. Zubáč, F. Krizek, V. Novák, R. P. Campion, M. S. Wernle, P. Gambardella, X. Marti, P. Němec, et al., Nat. Electr. 4, 30 (2021).
  19. [19]T. Janda, J. Godinho, T. Ostatnicky, E. Pfitzner, G. Ulrich, A. Hoehl, S. Reimers, Z. Šobáň, T. Metzger, H. Re-ichlova, et al., Phys. Rev. Mat. 4, 094413 (2020).
  20. [20]R. M. Otxoa, U. Atxitia, P. E. Roy, and O. Chubykalo-Fesenko, Commun. Phys. 3, 31 (2020).
  21. [21]S.-H. Yang, K.-S. Ryu, and S. Parkin, Nat. Nanotechnol. 10, 221 (2015).
  22. [22]S. Lequeux, J. Sampaio, V. Cros, K. Yakushiji, A. Fukushiman, R. Matsumoto, H. Kubota, S. Yuasa, and J. Grollier, Sci. Rep. 6, 1 (2016).
  23. [23]Z. Luo, A. Hrabec, T. P. Dao, G. Sala, S. Finizio, J. Feng, S. Mayr, J. Raabe, P. Gambardella, and L. J. Heyderman, Nature 579, 214 (2020).
  24. [24]A. B. Shick, S. Khmelevskyi, O. N. Myrasov, J. Wunderlich, and T. Jungwirth, Phys. Rev. B 81, 214209 (2010).
  25. [25]V. M. T. S. Barthem, C. V. Colin, H. Mayaffre, M.-H. Julien, and D. Givord, Nat. Commun. 4, 2892 (2013).
  26. [26]R. Masrour, E. K. Hli, M. Hamedoun, B. Benoussaf, A. Boutahar, and H. Lassri, J. Magn. Mat. 393, 600 (2015).
  27. [27]S. Khmelevskyi and P. Mohn, Appl. Phys. Lett. 93, 162503 (2008).
  28. [28]P. Wells and J. H. Smith, Acta Crystallogr. Sect. A 26, 379 (1970).
  29. [29]P. E. Roy, R. M. Otxoa, and J. Wunderlich, Phys. Rev. B 94, 014439 (2016).
  30. [30]E. A. Turov, A. V. Kolchanov, V. V. Men’shenin, I. F. Mirsaev, and V. V. Nikolaev, Symmetry and Physical Properties of Antiferromagnets (Cambridge International Science Publishing, Cambridge, 2010).
  31. [31]A. K. Zvezdin, JETP Lett. 29, 553 (1979).
  32. [32]V. G. Bar’yakhtar, B. A. Ivanov, and M. V. Chetkin, Sov. Phys. Uspekhi 28, 563 (1985).
  33. [33]E. M. Lifshitz and L. P. Pitaevskii, Statistical Physics, Course of Theoretical Physics, Vol. 9 (Pergamon Press, Oxford, 1980).
  34. [34]N. Papicolaou, Phys. Rev. B 51, 15062 (1995).
  35. [35]E. G. Tveten, T. Müller, J. Linder, and A. Brataas, Phys. Rev. B 93, 104408 (2016).
  36. [36]A. M. Kosovichev, B. A. Ivanov, and A. S. Kovalev, Phys. Rep. 194, 117 (1990).
  37. [37]A. K. Zvezdin and V. V. Kostyuchenko, J. Exp. Theor. Phys. 89, 734 (1999).
  38. [38]M. Arana, F. Estrada, D. S. Maior, J. B. S. Mendes, L. E. Fernandez-Outon, W. A. A. Macedo, V. M. T. S. Barthem, D. Givord, A. Azevedo, and S. M. Rezende, Appl. Phys. Lett. 111, 192409 (2017).
  39. [39]I. V. Bar’yakhtar and B. A. Ivanov, Solid State Commun. 34, 545 (1980).
  40. [40]A. F. Andreev and V. I. Marchenko, Sov. Phys. Uspekhi 23, 21 (1980).
  41. [41]R. Rajaraman, Solitons and Instantons (North-Holland, Amsterdam, 1982).
  42. [42]N. L. Schryer and L. R. Walker, J. Appl. Phys. 45, 5406 (1974).
  43. [43]O. Gomonay, T. Jungwirth, and J. Sinova, Phys. Rev.
  44. [44]It should be noted that the absence of periodicity along the z-th spatial direction in the simulated ultrathin film does not guarantee the condition m₁,₂ = m₃,₄ used in Section III, which allowed us to reduce the system description to a two-sublattice-based nonlinear σ-model.

Paper details

vlnc:W_rmqbnk3n
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 8

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