Current-induced motion of nanoscale magnetic torons
over the wide range of the Hall angle

Abstract

Current-driven dynamics of spin textures plays a pivotal role in potential applications for electronic devices. While two-dimensional magnetic skyrmions with topologically nontrivial spin textures have garnered significant interest, their practical use is hindered by the skyrmion Hall effect — a transverse motion to the current direction that occurs as a counteraction to the topological Hall effect of electrons by an emergent magnetic field arising from the Berry phase effect. Here, we explore current-driven dynamics of three-dimensional topological spin textures known as magnetic torons, composed of layered skyrmions with two singularities called Bloch points at their ends. Through extensive numerical simulations, we show that the torons also exhibit a Hall motion, but surprisingly over a wide range spanning from the zero Hall effect, a purely longitudinal motion, to the perfect Hall effect, a purely transverse motion accompanied by no longitudinal motion. Such flexible and controllable behaviors stem from anisotropic potential barriers on the discrete lattice, which can be particularly relevant for nanoscale torons recently discovered. Our results not only provide an experimental method to probe topology of three-dimensional magnetic textures but also pave the way for future developments in topological spintronics beyond the realm of skyrmions.

Figure 1: Schematic illustrations of the topological spin textures and their drift motions driven by an electron flow. a Real-space spin configuration of a skyrmion and the skyrmion Hall effect. The small arrows represent the spins and their color displays the $z$ component (see the inset). Skyrmions show a transverse drift motion as well as a longitudinal one with the velocity $\mathbf{v}$ as a reaction to the topological Hall effect in the electron flow $\tilde{\mathbf{j}}_{e}$. b Real-space spin configurations of a magnetic toron, given by a pair of hedgehog (magenta sphere) and antihedgehog (cyan sphere) connected by the skyrmion string (gray line) (left). Corresponding real-space distribution of the stream of the EMF (right). The hedgehog and antihedgehog correspond to a source and sink of the EMF, respectively. c Toron Hall effect. A magnetic toron shows a transverse drift motion similar to the skyrmion Hall effect. The blue wavy surface represents the potential for the torons from a discrete lattice structure. d Perfect toron Hall effect. Torons exhibit only a transverse motion without longitudinal one because of strong one-dimensional anisotropy in the lattice potential. e Zero toron Hall effect. Torons exhibit only a longitudinal motion accompanied by no transverse motion. The left and right panels correspond to the situations realized in Figs. 3d,e and 3a–c, respectively.

Research in the field of topological spin textures has opened a new era for electronics and spintronics by leveraging the synergy between electronic and magnetic properties through topology. Of particular interest are two-dimensional magnetic skyrmions with swirling spin textures characterized by a topological invariant called the skyrmion number (Fig. 1a). The skyrmion was originally hypothesized in the field of particle physics [1, 2], and later theoretically predicted to be realized in chiral magnets [3, 4, 5, 6] and actually observed in experiments in a form of a triangular array [7, 8]. Owing to their inherent topological stability, skyrmions have drawn attention as entirely new information carriers in next-generation spintronic devices. They also offer a significant advantage in terms of energy consumption, as they can be driven at much lower electric current densities [9]. However, the current-driven skyrmions exhibit a transverse motion relative to the current direction, known as the skyrmion Hall effect [9, 10, 11, 12]. This is caused by the Magnus force as a counteraction to the topological Hall effect of electrons, as schematically shown in Fig. 1a, arising from the emergent magnetic field (EMF) by the Berry phase effect, $\mathbf{b}(\mathbf{r})$, given by $b_{i}(\mathbf{r})=\frac{1}{2}\epsilon_{ijk}\mathbf{S}(\mathbf{r})\cdot(\partial_{j}\mathbf{S}(\mathbf{r})\times\partial_{k}\mathbf{S}(\mathbf{r}))$, where $\epsilon_{ijk}$ is the Levi-Civita symbol and $\mathbf{S}(\mathbf{r})$ represents the spin at spatial position $\mathbf{r}$ [13, 14]. This lateral motion presents challenges in device applications of magnetic skyrmions, although various strategies have been proposed, such as utilizing antiferromagnetic skyrmions that do not exhibit the skyrmion Hall effect [15, 16].

In three-dimensional space, skyrmions manifest as string-like structures, stacked in the out-of-plane direction. This skyrmion string may terminate in some cases, and at these endpoints, unique three-dimensional topological spin textures emerge. They are known as the Bloch points and were observed by using 3D tomographic imaging techniques [17, 18, 19, 20]. These Bloch points are also characterized by a topological invariant $Q_{\mathrm{m}}$ called the monopole charge, which is defined by a surface integral of the EMF $\mathbf{b}(\mathbf{r})$ on a sphere surrounding the Bloch point, and classified into two types depending on the sign of the monopole charge: magnetic hedgehogs and antihedgehogs with positive and negative monopole charges, respectively [14, 21]. They typically appear in pairs at both ends of a skyrmion string and such a pair is called the magnetic toron (Fig. 1b) [22]. Unlike skyrmions, the hedgehogs and antihedgehogs have singularities with vanishing spin length at their cores, which can be regarded as sources and sinks in terms of the EMF, thus acting as emergent magnetic monopoles and antimonopoles, respectively.

Recent experiments have demonstrated that such magnetic torons are stabilized by forming periodic arrangements called the hedgehog lattices (HLs) in chiral magnets MnSi_1-xGe_x [23, 24, 25, 26] and a nonchiral magnet SrFeO₃ [27]. Notably, these HLs have remarkably short magnetic pitches, with hedgehogs and antihedgehogs spanning only a few nanometers in size, suggesting a giant EMF around them caused by a strong twist of the spin textures. Due to the EMF as well as the topological robustness, Bloch points are expected to exhibit peculiar electromagnetic responses like skyrmions, but their current-driven dynamics remains largely unexplored, except for an isolated Bloch point or toron in the long-wavelength limit [28, 29]. Given that smaller texture sizes are preferable for EMF-based devices, it is imperative to investigate the current-driven dynamics of these short-pitch HLs to advance future spintronic devices beyond the capabilities of skyrmions.

Here, we examine the responses of nanoscale magnetic torons to an applied electric current by extensive numerical simulations. We show that, similar to skyrmions, the torons exhibit a Hall motion, which we call the toron Hall effect (Fig. 1c). Strikingly, we discover that the toron Hall effect can be controlled by the electric current and the magnetic field in a wide range including two extremes: a purely transverse motion without any longitudinal one (perfect toron Hall effect; Fig. 1d) and an exclusively longitudinal motion with no transverse one (zero toron Hall effect; Fig. 1e). These unique behaviors stem from the modulation of potential barriers on the discrete lattice, which is particularly relevant for the nanoscale torons realized in experiments. Furthermore, we reveal that the responses to the current act as efficient electrical probes for topological characteristics hidden behind the spin textures that are challenging to observe experimentally.

Toron Hall effect

Figure 2: Hedgehog lattice and topological transition in the magnetic field. a-c Real-space spin configurations and distributions of magnetic torons in the 3$Q$-HL with $N_{\mathrm{t}}=4$ at $H=0$ (a), the 3$Q$-HL with $N_{\mathrm{t}}=2$ at $H=0.3$ (b), and the topologically-trivial $2Q$ state with $N_{\mathrm{t}}=0$ at $H=1.0$ (c). Here, $N_{\mathrm{t}}$ denotes the total number of the torons in the MUC represented by the black cubes. The graphical notations are common to those in Fig. 1. The lower bar displays the phase diagram while increasing the magnetic field $H$. d-g Real-space distributions of the magnetic torons within the MUC for the ground state at $H=0$ (d), $H=0.3$ (e), $H=0.45$ (f), and $H=0.6$ (g). Here, $d_{\mathrm{t}}$ denotes the length of the skyrmion string in the $y$ direction. As $H$ increases, the hedgehogs and antihedgehogs approach each other along the skyrmion strings (d-f), and half of them disappear with pair annihilation at the transition at $H=H_{\rm topo}\simeq 0.465$ (f,g).

Figure 3: Toron Hall effect. a,b The velocities of spin textures for the longitudinal (a) and transverse motions (b) under the electron flow in the $x$ direction ($\tilde{\mathbf{j}}_{e}\parallel\hat{\bf x}$) on the plane of the magnetic field $H$ and the current density $\tilde{j}_{e}$. The white lines represent the threshold current densities, $\tilde{j}_{e,\mathrm{c}\parallel}$ (a) and $\tilde{j}_{e,\mathrm{c}\perp}$ (b). The dashed and solid gray vertical lines denote the values of $H$ where $d_{\mathrm{t}}$ and $N_{\mathrm{t}}$ show discontinuous changes, respectively; see Fig. 2. c The Hall angle of the drift motion, $\theta_{\mathrm{Hall}}$ [Eq. (1)]. In the black and white regions, the torons exhibit the zero and perfect Hall effects, respectively, while they are immobile in the gray region. d,e The velocities of spin textures for the longitudinal (d) and transverse (e) motions under the electron flow in the $y$ direction ($\tilde{\mathbf{j}}_{e}\parallel\hat{\bf y}$). The latter is always zero in this parameter region, meaning that the zero toron Hall effect always occurs for $\tilde{j}_{e}>\tilde{j}_{e,\mathrm{c}\parallel}$ (above the white curve in d). $v_{\parallel}$ cannot be defined in the gray area. f $\tilde{j}_{e}$ dependence of $\theta_{\mathrm{Hall}}$ at three representative values of $H$ for $\tilde{\mathbf{j}}_{e}\parallel\hat{\mathbf{x}}$ (top) and $\tilde{\mathbf{j}}_{e}\parallel\hat{\mathbf{y}}$ (bottom). For $\tilde{\mathbf{j}}_{e}\parallel\hat{\mathbf{x}}$, the perfect toron Hall effect with $\theta_{\mathrm{Hall}}=\pi/2$ occurs for small $\tilde{\mathbf{j}}_{e}$, and it can be switched to the zero toron Hall effect with $\theta_{\mathrm{Hall}}=0$ by increasing $\tilde{j}_{e}$ at $H=0.3$. In contrast, for $\tilde{\mathbf{j}}_{e}\parallel\hat{\mathbf{y}}$, only the zero toron Hall effect with $\theta_{\mathrm{Hall}}=0$ is observed.

Figure 4: Potential barrier for magnetic torons. a,b Energy change per site, $\Delta E$, for an in-plane spatial translation of the 3$Q$-HL by $\bm{\Delta}\mathbf{r}=(\Delta x,\Delta y,0)$ measured from that for the ground state at $H=0$ (a) and $H=0.5$ (b). The potential is equivalent in the $x$ and $y$ directions at $H=0$, while it becomes almost one-dimensional at $H=0.5$. The gray dashed lines denote the intersections of the $xy$ plane and the unit cubes composed of the lattice sites. $\Delta E$ along the red and green dashed lines as well as the definitions of the potential barriers $V_{x}$ and $V_{y}$ are presented in the top and right panels, respectively. c $H$ dependence of the potential barriers in the direction of $\mu$, $V_{\mu}$. The behaviors of $V_{x}$ and $V_{y}$ qualitatively explain those of the threshold current densities of $v_{\parallel}$ for $\tilde{\mathbf{j}}_{e}\parallel\hat{\bf x}$ and $\hat{\bf y}$ in Figs. 3a and 3d, respectively.

To elucidate the current-induced dynamics of magnetic torons, we consider a spin model for a metallic chiral magnet on a simple cubic lattice, and perform a real-time simulation in an electric current based on the Landau-Lifshitz-Gilbert equation. The model includes effective interactions of Ruderman-Kittel-Kasuya-Yosida and Dzyaloshinskii-Moriya (DM) types, which has been shown to approximately reproduce the HLs discovered in MnSi_1-xGe_x [30, 31, 32] (see Methods). In the absence of the electric current, the model stabilizes a HL represented by the superposition of three spin density waves, called the $3Q$-HL, at zero magnetic field (Fig. 2a) (see Supplementary Note 1). The magnetic unit cell (MUC) contains eight Bloch points (hedgehogs and antihedgehogs), which are connected in four pairs by the skyrmion strings (Fig. 2d); namely, there are four torons denoted as $N_{\mathrm{t}}=4$. In the magnetic field applied in the $z$ direction, $H$, the hedgehogs and antihedgehogs move toward each other along the skyrmion strings (Figs. 2e,f), and the half of them disappear with pair annihilation at $H=H_{\mathrm{topo}}\simeq 0.465$, leading to the topological transition to the $3Q$-HL with $N_{\mathrm{t}}=2$ (Figs. 2b,g). By further increasing $H$, the system turns into a $2Q$ state at $H=H_{\mathrm{c}}\simeq 0.935$, which is topologically trivial with no Bloch points ($N_{\rm t}=0)$ (Fig. 2c).

Let us first discuss drift motions caused by an electron flow $\tilde{\mathbf{j}}_{e}$ in the $x$ direction perpendicular to the skyrmion strings (see Methods). We find that the magnetic torons exhibit a transverse motion to the electron flow due to the Magnus force as well as a longitudinal one over the wide range of the HL phases under the magnetic field. We call this the toron Hall effect. Remarkably, however, the velocities of the longitudinal and transverse motions, $v_{\parallel}$ and $v_{\perp}$, respectively, show largely different and complicated dependences on $H$ and $\tilde{j}_{e}=|\tilde{\mathbf{j}}_{e}|$. Figures 3a and 3b summarize $v_{\parallel}$ and $v_{\perp}$, respectively, on the $H$-$\tilde{j}_{e}$ plane. The longitudinal motion occurs above a certain threshold current density $\tilde{j}_{e,\mathrm{c}\parallel}$ represented by the white curve; $\tilde{j}_{e,\mathrm{c}\parallel}$ overall increases with $H$, except significant reduction near the topological transition at $H_{\mathrm{topo}}\simeq 0.465$. Above $\tilde{j}_{e,\mathrm{c}\parallel}$, $v_{\parallel}$ monotonically increases with $\tilde{j}_{e}$. Meanwhile, the transverse motion also occurs above a threshold $\tilde{j}_{e,\mathrm{c}\perp}$, but it shows a sharp dip at $H\simeq 0.225$ and becomes vanishingly small for $H\gtrsim H_{\mathrm{topo}}$. The dip corresponds to a shortening of the skyrmion strings (Fig. 2); we will discuss this later. In addition, $v_{\perp}$ exhibits nonmonotonic behavior while increasing $\tilde{j}_{e}$ across $\tilde{j}_{e,\mathrm{c}\parallel}$; most significantly, around $H=0.3$, it becomes nonzero above $\tilde{j}_{e,\mathrm{c}\perp}$, but vanishes for $\tilde{j}_{e}>\tilde{j}_{e,\mathrm{c}\parallel}\sim 0.12$. We also note that $v_{\perp}$ vanishes in the weak field region where $H\lesssim 0.2$ in this range of $\tilde{j}_{e}$ due to the weak Magnus force stemming from small EMF (see Supplementary Note 1).

These distinct behaviors of $v_{\parallel}$ and $v_{\perp}$ indicate that the system exhibits the toron Hall effect in a wide range including two extreme limits. One is a purely transverse motion without any longitudinal one ($v_{\parallel}=0$ and $v_{\perp}\neq 0$), appearing in the region where $\tilde{j}_{e,\mathrm{c}\perp}<\tilde{j}_{e}<\tilde{j}_{e,\mathrm{c}\parallel}$. We call this the perfect toron Hall effect. The other is opposite: a purely longitudinal motion accompanied with no transverse one ($v_{\parallel}\neq 0$ and $v_{\perp}=0$), occurring in the region where $v_{\perp}$ vanishes despite $\tilde{j}_{e}>\tilde{j}_{e,\mathrm{c}\parallel}$. We call this the zero toron Hall effect. These two extreme behaviors can be explicitly visualized by plotting the Hall angle defined by

$$\theta_{\mathrm{Hall}}=\arctan\left(\frac{v_{\perp}}{v_{\parallel}}\right),$$ (1)

which represents the net direction of the drift motion with respect to the electron flow (Fig. 3c). The perfect and zero toron Hall effects correspond to $\theta_{\mathrm{Hall}}=\pi/2$ and $\theta_{\mathrm{Hall}}=0$, respectively, shown by the white and black regions. The former occurs in the wide range of $H\gtrsim 0.2$, especially for small $\tilde{j}_{e}$ in the $3Q$-HL with $N_{\mathrm{t}}=2$ after the topological transition, while the latter appears in the $3Q$-HL with $N_{\mathrm{t}}=4$ for $\tilde{j}_{e}>\tilde{j}_{e,\mathrm{c}\parallel}$.

Notably, one can achieve the zero toron Hall effect over a wider range of $H$ and $\tilde{j}_{e}$ simply by changing the current direction. We demonstrate this by taking the electron flow in the $y$ direction along the skyrmion strings ($\tilde{\mathbf{j}}_{e}\parallel\hat{\bf y}$). In this case also, the longitudinal motion occurs above nonzero threshold $\tilde{j}_{e,\mathrm{c}\parallel}$ (Fig. 3d); $\tilde{j}_{e,\mathrm{c}\parallel}$ shows a different $H$ dependence from the $\tilde{\mathbf{j}}_{e}\parallel\hat{\bf x}$ case with characteristic dip structures, which will be discussed later. We note that $v_{\parallel}$ cannot be defined in the $2Q$ phase since the spin modulation in the $y$ direction is absent (see Supplementary Note 1). In stark contrast, the transverse motion is not observed at all in this parameter region (Fig. 3e). Thus, with this current setting, the zero toron Hall effect appears in the entire region for $\tilde{j}_{e}>\tilde{j}_{e,\mathrm{c}\parallel}$.

Our results show that the toron Hall effect is much more flexible than the skyrmion Hall effect. It can be controlled by the amplitudes and directions of the magnetic field and the electric current. Most strikingly, it includes two extremes, the zero and perfect toron Hall effects, which have never been reported in the skyrmion Hall effect. By choosing appropriate parameters, one can even achieve switching between the two extremes, as demonstrated in Fig. 3f: By increasing $\tilde{j}_{e}$ for $\tilde{\mathbf{j}}_{e}\parallel\hat{\bf x}$ at $H=0.3$, the perfect toron Hall effect is observed for $0.091\lesssim\tilde{j}_{e}\lesssim 0.115$ (Supplementary Video 1), followed by the zero toron Hall effect for $0.115\lesssim\tilde{j}_{e}\lesssim 0.181$ (Supplementary Video 2), and finally, the toron Hall effect with $\theta_{\mathrm{Hall}}\sim\frac{\pi}{5}$ appears for $\tilde{j}_{e}\gtrsim 0.181$ (Supplementary Video 3).

We note that the toron Hall effect vanishes when we turn off the DM-type interaction in our model for the $3Q$-HL (see Supplementary Note 5). Nonetheless, this does not mean that the toron Hall effect never occur in nonchiral magnets. In general, the net EMF can be nonzero even in nonchiral cases in the presence of spin-orbit coupling, leading to the topological Hall effect as observed in SrFeO₃ [33], allowing the toron Hall effect as its counteraction.

Anisotropic potential barrier

The peculiar toron Hall effect originates from the energy potential for torons on the discrete lattice. Drift motions of nanoscale torons are strongly affected by the lattice discretization due to their short magnetic pitch and singular structures of constituting Bloch points. The magnetic field influences the spin textures, resulting in energy potential modulation in an anisotropic manner. This causes the different and complicated $H$ dependences of $\tilde{j}_{e,\mathrm{c}\parallel}$ and $\tilde{j}_{e,\mathrm{c}\perp}$, as demonstrated below.

When $H=0$, the energy potential for torons on the $xy$ plane is fourfold rotational symmetric reflecting the lattice symmetry (Fig. 4a). Here, we plot the energy increase per site $\Delta E$ by shifting the spin texture with $\bm{\Delta}\mathbf{r}=(\Delta x,\Delta y,0)$ from the ground state. The potential oscillates with the lattice period, and the heights of the potential barrier in the $x$ and $y$ directions are equivalent, $V_{x}=V_{y}$. In this case, an electric current required for torons to overcome the potential barrier is equal for $\tilde{\mathbf{j}}_{e}\parallel\hat{\bf x}$ and $\tilde{\mathbf{j}}_{e}\parallel\hat{\bf y}$, resulting in $\tilde{j}_{e,\mathrm{c}\parallel}\simeq 0.09$ in both cases (Figs. 3a,d). By introducing $H$, $V_{x}$ increases but $V_{y}$ decreases; namely, the potential develops a one-dimensional anisotropy through the modulation of the spin texture, giving rise to different threshold current densities for the different current directions. Figure 4b represents an example at $H=0.5$, where $V_{y}$ almost vanishes. Such a highly anisotropic potential causes the unique drift motions including the zero and perfect toron Hall effects found above. On one hand, for $\tilde{\mathbf{j}}_{e}\parallel\hat{\bf x}$, the longitudinal drift motion is hindered by large $V_{x}$, while the transverse motion is easily driven with vanishingly small $V_{y}$, resulting in $\tilde{j}_{e,\mathrm{c}\perp}<\tilde{j}_{e,\mathrm{c}\parallel}$. This leads to the perfect toron Hall effect, as illustrated in Fig. 1d. On the other hand, for $\tilde{\mathbf{j}}_{e}\parallel\hat{\bf y}$, the longitudinal drift motion is caused by a small current density, while the transverse motion is largely suppressed, resulting in $\tilde{j}_{e,\mathrm{c}\parallel}<\tilde{j}_{e,\mathrm{c}\perp}$. This brings about the zero toron Hall effect, as illustrated in the left panel of Fig. 1e. The zero toron Hall effect for $\tilde{\mathbf{j}}_{e}\parallel\hat{\bf x}$ in Fig. 3c is also caused by the anisotropic potential, but in more complicated competition between the longitudinal driving force from the electric current and the induced transverse Magnus force (the right panel of Fig. 1e).

Remarkably, the energy potential becomes almost one-dimensional ($V_{y}\simeq 0$) in the entire $3Q$-HL phase with $N_{\mathrm{t}}=2$ after the topological transition (Fig. 4c). The origin of the vanishingly small $V_{y}$ lies in the suppression of the higher harmonics of the spin structure factor in the $y$ direction; the suppression beyond the Nyquist wave number makes the spin texture less susceptible to the lattice discretization (see Supplementary Note 2). It is worth noting that the $H$ dependences of the potential barriers qualitatively explain those of the critical current densities: $V_{x}$ behaves as $\tilde{j}_{e,\mathrm{c}\parallel}$ for $\tilde{\mathbf{j}}_{e}\parallel\hat{\bf x}$ (Fig. 3a), and $V_{y}$ as $\tilde{j}_{e,\mathrm{c}\perp}$ for $\tilde{\mathbf{j}}_{e}\parallel\hat{\bf x}$ (Fig. 3b) and $\tilde{j}_{e,\mathrm{c}\parallel}$ for $\tilde{\mathbf{j}}_{e}\parallel\hat{\bf y}$ (Fig. 3d) (for $V_{z}$, see Supplementary Note 3). This supports our argument based on the potential barrier on the discrete lattice. We note that $V_{x}$ does not follow the significant reduction of $\tilde{j}_{e,\mathrm{c}\parallel}$ for $\tilde{\mathbf{j}}_{e}\parallel\hat{\bf x}$ near the topological transition in Fig. 3a, but this is accounted for by the Magnus force in the direction longitudinal to the current arising from the transverse drift motion of torons.

Probing hidden topology

Our results indicate that the $H$ dependences of the threshold current densities, $\tilde{j}_{e,\mathrm{c}\parallel}$ and $\tilde{j}_{e,\mathrm{c}\perp}$, can be a sensitive probe of the topological transition with toron annihilation. This can be complementary to the elastic constant measurements [21]. Furthermore, they are useful for probing more detailed changes in the topological objects, namely the length changes in the skyrmion strings. For instance, for $\tilde{\mathbf{j}}_{e}\parallel\hat{\bf x}$, $\tilde{j}_{e,\mathrm{c}\perp}$ exhibits a sharp dip at $H\simeq 0.225$ (Fig. 3b), where the lengths of the skyrmion strings change from $d_{\mathrm{t}}=7$ to $6$ in two out of four torons and from $5$ to $4$ in the rest (Fig. 2). We also observe a small anomaly in $\tilde{j}_{e,\mathrm{c}\parallel}$ (Fig. 3a). A similar sharp dip is also found in $\tilde{j}_{e,\mathrm{c}\parallel}$ for $\tilde{\mathbf{j}}_{e}\parallel\hat{\bf y}$ (Fig. 3d), but in this case, a smaller dip is additionally seen at $H\simeq 0.765$ in the $3Q$-HL with $N_{\mathrm{t}}=2$, where the length of the skyrmion string reduces from $6$ to $5$ in the remaining two torons. These behaviors are understood as follows. The Bloch points locate at the interstitial positions on the lattice so that their cores with vanishing spin length avoid the lattice sites, and hence the energy potential is strongly modulated when the Bloch points traverse between the unit cubes of the lattice with the shortening of the skyrmion strings. Thus, the measurement of the threshold current densities can serve as a probe of not only topological transitions but also detailed changes of topological objects in the lattice spacing scale that are usually hidden in the macroscopic properties [32, 34]. We note that this holds even for nonchiral cases without the DM-type interaction (see Supplementary Note 5).

In experiments, the threshold current densities under the magnetic field can be measured straightforwardly. In the present simulation, $\tilde{j}_{e}\sim 0.1$ corresponds to the electric current density of $\sim 10^{12}$ A/m², when assuming the energy unit as $1$ meV, the lattice constant as $0.5$ nm, and the spin polarization of the current as $0.2$ [11] (see Methods). This is comparable to a typical value required for magnetic domain walls to drive [35]. Thus, the threshold current measurements would provide microscopic information of the topology, complementary to the real-space measurements such as the Lorentz transmission electron microscopy [25] and 3D tomographic imaging techniques [19, 20]. Such measurements for nanoscale HLs discovered in MnSi_1-xGe_x and SrFeO₃ are eagerly awaited for the verification of our theoretical prediction.

Concluding remarks

Our comprehensive study of the current-induced dynamics of nanoscale magnetic torons has revealed the extensive controllability of their drift motions spanning from the zero to perfect toron Hall effect. We clarified that such prominent current-induced dynamics arises from the potential barrier on the discrete lattice which can be highly anisotropic through the modulation of the spin textures by the magnetic field. We also found that the field dependence of the threshold current serves as a concise and sensitive probe of the microscopic changes of topological objects.

While the measurement of the threshold currents is experimentally straightforward, their actual values as well as the $H$ dependences depend on the crystal structure and its symmetry of the real materials. Our results for the simple cubic lattice need modifications to apply to MnGe with the $B20$ structure, while they will be rather straightforwardly extended to the cubic perovskite SrFeO₃. Nevertheless, we believe that the flexible and controllable toron Hall effect in the wide range of the Hall angle is universally observed for HLs with nanoscale torons because of the short magnetic pitch comparable to the lattice constant and the singularity at the cores of the Bloch points. Extensions of our studies to crystallographic groups other than cubic are intriguing for further explorations. For instance, hexagonal systems may allow zero and perfect toron Hall effects even at zero field because of the symmetry of the energy potential.

Magnetic torons are promising information carriers due to their topological robustness. Our finding of flexible and controllable drift motions will increase their advantage over skyrmions. In the case of skyrmions, it was shown that the current-driven motions are almost free from the lattice discretization, and the Hall motion inevitably occurs, which has hindered the applications of skyrmions to spintronic devices [9, 10, 11]. The extremely high controllability of the toron Hall effect is expected to resolve these issues and lead to future device applications. We note, however, that such a skyrmion motion is partly due to the real-space scale of spin textures much larger than the lattice spacing. Recently, nanoscale skyrmions have been discovered beyond the conventional DM mechanism [36, 37, 38, 39]. Such short-pitch skyrmions would be more susceptible to the lattice discretization compared to conventional long-pitch ones, while they are anticipated to be less sensitive than magnetic torons due to the absence of singularities. Detailed comparisons, including other topological spin textures, will be left to future studies.

Methods

Model for metallic chiral magnets

We adopt an effective spin model for metallic chiral magnets with itinerant electrons coupled to localized spins. The electron-spin coupling brings about effective long-range interactions between the localized spins known as the Ruderman-Kittel-Kasuya-Yosida interaction [40, 41, 42]. By extending the argument, the previous studies systematically derived the generalized and multiple-spin interactions in the presence of the spin-orbit coupling, and showed that such extensions stabilize HLs as well as skyrmion lattices with short magnetic pitches [43, 44, 45, 30, 31, 46, 32, 34]. In this study, we consider the symmetry-adapted two-spin interactions for a cubic chiral magnet by following the previous studies [31, 32]. The Hamiltonian reads

$$\mathcal{H}=\sum_{\begin{subarray}{c}i,j\\ g_{3}({\bf r}_{ij})\leq r_{\rm c}\end{subarray}}{\bf S}_{i}^{{\mathsf{T}}}(t){\mathsf{J}}_{ij}{\bf S}_{j}(t)-\sum_{i}HS_{i,z}(t),$$ (2)

where

$$g_{3}({\bf r})=|x|+|y|+|z|,$$ (3)

with

$$\mathsf{B}_{{\bf Q}_{1}}=\left(\begin{array}[]{ccc}-J(1-\Delta)&0&0\\ 0&-J(1+2\Delta)&\mathrm{i}D\\ 0&-\mathrm{i}D&-J(1-\Delta)\end{array}\right),$$ (7)

$$\mathsf{B}_{{\bf Q}_{2}}=\left(\begin{array}[]{ccc}-J(1-\Delta)&0&-\mathrm{i}D\\ 0&-J(1-\Delta)&0\\ \mathrm{i}D&0&-J(1+2\Delta)\end{array}\right),$$ (11)

$$\mathsf{B}_{{\bf Q}_{3}}=\left(\begin{array}[]{ccc}-J(1+2\Delta)&\mathrm{i}D&0\\ -\mathrm{i}D&-J(1-\Delta)&0\\ 0&0&-J(1-\Delta)\end{array}\right),$$ (15)

$$\bar{L}_{\gamma}({\bf q})=\frac{L_{\gamma}(q_{x})}{L_{\gamma}(0)}\frac{L_{\gamma}(q_{y})}{L_{\gamma}(0)}\frac{L_{\gamma}(q_{z})}{L_{\gamma}(0)},$$ (16)

$$L_{\gamma}(q)=\frac{\sinh\gamma}{\cosh\gamma-\cos q}.$$ (17)

Here, $\mathbf{S}_{i}(t)$ denotes the classical spin at the site $i$ located at $\mathbf{r}_{i}$ and time $t$ ($|\mathbf{S}_{i}(t)|=1$); $J$, $\Delta$, and $D$ represent the coefficients for the isotropic exchange, anisotropic exchange, and DM interactions, respectively. The interactions exhibit spatial oscillations arising from the itinerant nature of electrons; we set the wave vectors $\mathbf{Q}_{1}=(Q,0,0)$, $\mathbf{Q}_{2}=(0,Q,0)$, and $\mathbf{Q}_{3}=(0,0,Q)$, which are related with each other by threefold rotational symmetry about the [111] axis. Furthermore, we introduce the spatial decay of the interaction parametrized by $\gamma$ as a function of $\mathbf{r}_{ij}=\mathbf{r}_{j}-\mathbf{r}_{i}$ and reduce the numerical cost by limiting the summation in real space for $g_{3}(\mathbf{r}_{ij})\leq r_{\mathrm{c}}$ by following the previous studies [31, 47]. The second term in Eq. (2) represents the Zeeman coupling to the magnetic field $\mathbf{H}=(0,0,H)$. We take $J$ as the energy unit, set the lattice constant unity, and choose $\Delta=0.3$, $D=0.3$, $Q=\pi/6$, $\gamma=0.6Q$, and $r_{\mathrm{c}}=16$. We perform the calculations for the $N=12^{3}$-site system under the periodic boundary condition, which corresponds to the single MUC. We confirm that the results are qualitatively intact for $N=60^{3}$.

Landau-Lifshitz-Gilbert equation

To elucidate the current-induced dynamics of the HL, we study the real-time dynamics of spins by solving the Landau-Lifshitz-Gilbert (LLG) equation given by

$$\frac{d\mathbf{S}_{i}(t)}{dt}=-\mathbf{S}_{i}(t)\times\mathbf{h}_{i}^{\mathrm{eff}}(t)-\alpha\mathbf{S}_{i}(t)\times\frac{d\mathbf{S}_{i}(t)}{dt}+\bm{\tau}_{i}(t),$$ (18)

where $\alpha$ is the Gilbert dam**, ${\bf h}^{\rm eff}_{i}(t)$ is the mean magnetic field at time $t$, defined as

$${\bf h}^{\rm eff}_{i}(t)=\frac{\partial\mathcal{H}(t)}{\partial{\bf S}_{i}(t)}.$$ (19)

Note that the length constraint of $|{\bf S}_{i}(t)|=1$ is deferred only in the calculation of this derivative. In Eq. (18), $\bm{\tau}_{i}(t)$ is the torque arising from an electric current, which is given by [48, 11]

$$\bm{\tau}_{i}(t)=-\left[\left(\tilde{\mathbf{j}}_{e}\cdot\hat{\bm{\delta}}\right)\mathbf{S}_{i}(t)+\beta\left(\mathbf{S}_{i}(t)\times\left(\tilde{\mathbf{j}}_{e}\cdot\hat{\bm{\delta}}\right)\mathbf{S}_{i}(t)\right)\right].$$ (20)

Here, the effect of the electric current is incorporated as the spin transfer torque; $\tilde{\mathbf{j}}_{e}$ represents the normalized velocity of conduction electrons,

$$\tilde{\mathbf{j}}_{e}=-\frac{pa^{2}}{2e}\mathbf{j},$$ (21)

where $p$ is the spin polarization, $a$ is the lattice constant (we set $a=1$), $e>0$ is the elementary charge, and $\mathbf{j}$ is the electric current. In Eq. (20), $\beta$ is the nonadiabatic coefficient, and $\hat{\bm{\delta}}=(\hat{\delta}_{x},\hat{\delta}_{y},\hat{\delta}_{z})$ is introduced to approximate the spatial derivative on the discrete lattice as

$$\hat{\delta}_{\mu}{\bf S}_{{\bf r}_{i}}(t)=\frac{1}{2}\left({\bf S}_{{\bf r}_{i}+\hat{\bm{\mu}}}-{\bf S}_{{\bf r}_{i}-\hat{\bm{\mu}}}\right).$$ (22)

In this study, we numerically solve Eq. (18) by using the fourth-order Runge-Kutta method with the time step $\Delta t=0.1$. We take $\alpha=0.04$ and $\beta=0.02$, which are typical values for ferromagnets [48, 49].

Calculating velocity

We calculate the velocities of drift motions of the spin textures by using the time series of the Fourier components of spins. By assuming that spin structures in continuum space exhibit the rigid drift motion, a real-space and real-time spin configuration can be given by $\mathbf{S}(\mathbf{r},t+\Delta t)=\mathbf{S}(\mathbf{r}-\mathbf{v}\Delta t,t)$, where $\mathbf{S}(\mathbf{r},t)$ is the spin at the position $\mathbf{r}$ and time $t$, and $\mathbf{v}$ is the velocity of the spin texture. The Fourier transform of spins is then given by

$$\mathbf{S}(\mathbf{q},t+\Delta t)=\int\mathbf{S}(\mathbf{r},t+\Delta t)e^{-\mathrm{i}\mathbf{q}\cdot\mathbf{r}}d^{3}\mathbf{r}=\mathbf{S}(\mathbf{q},t)e^{-\mathrm{i}\mathbf{q}\cdot\mathbf{v}\Delta t},$$ (23)

where $\mathbf{q}$ is the wave vector. Hence, the velocity at $t$ can be calculated by the phase difference between $\mathbf{S}(\mathbf{q},t)$ and $\mathbf{S}(\mathbf{q},t+\Delta t)$. In the actual calculations, we use the above relation for the Fourier components for the ordering wave vectors $\mathbf{Q}_{1}$, $\mathbf{Q}_{2}$, and $\mathbf{Q}_{3}$ calculated from $\mathbf{S}_{i}(t)$ obtained by the LLG simulation as

$$v_{\mu}=-\frac{1}{Q(t_{f}-t_{i})}\int_{t_{i}}^{t_{f}}~{}\frac{d}{dt}\arg\left(S_{{\bf Q}_{\eta},\mu^{\prime}}(t)\right)dt,$$ (24)

where $\eta=1,2,3$ for $\mu=x,y,z$, respectively, and $\mu^{\prime}$ denotes the largest component of ${\bf S}_{{\bf Q}_{\eta}}$ determined by the exchange anisotropy in the model in Eq. (2). For instance, at ${\bf q}={\bf Q}_{1}$, the exchange anisotropy is strong in the $S_{y}$ direction, leading to the largest amplitude in $S_{{\bf Q}_{1},y}$. The integral in Eq. (24) is taken from $t_{i}=500$ to $t_{f}=1000$.

Calculating potential barrier

We compute the potential barrier for magnetic torons by the energy change with shifting their positions from the ground state. For this calculation, we first approximately express the ground-state spin configuration by a superposition of three elliptical spirals using

$${\bf S}_{i}^{\mathrm{ellip}3Q}\propto\left(\begin{array}[]{c}\psi_{2}\varepsilon_{2}\sin\mathcal{Q}_{2}+\psi_{3}\cos\mathcal{Q}_{3}\\ \psi_{1}\cos\mathcal{Q}_{1}+\psi_{3}\varepsilon_{3}\sin\mathcal{Q}_{3}\\ \psi_{1}\varepsilon_{1}\sin\mathcal{Q}_{1}+\psi_{2}\cos\mathcal{Q}_{2}+m\end{array}\right),$$ (28)

where $\mathcal{Q}_{\eta}={\bf Q}_{\eta}\cdot{\bf r}_{i}+\varphi_{\eta}$, $\psi_{\eta}$ and $\varepsilon_{\eta}$ respectively represent the long axis and the ellipticity of the constituting spirals, and $\varphi_{\eta}$ is the phase. We estimate these parameters by minimizing the cost function defined by

$$U(\{\psi_{\eta},\varepsilon_{\eta},\varphi_{\eta}\},m)=$$

$$\frac{1}{N}\sum_{i}\left(1-{\bf S}^{\rm LLG}_{i}\cdot{\bf S}^{{\rm ellip}3Q}_{i}(\{\psi_{\eta},\varepsilon_{\eta},\varphi_{\eta}\},m)\right),$$ (29)

where ${\bf S}_{i}^{{\rm LLG}}$ is the ground-state spin configuration obtained by numerically solving the LLG equation and ${\bf S}_{i}^{{\rm ellip}3Q}$ is that generated from Eq. (28). The optimal state $\{{\bf S}_{i}^{{\rm ellip}3Q}\}^{*}$ shows the cost function less than $6\times 10^{-4}$ for all $H$, and hence $\{{\bf S}_{i}^{\rm LLG}\}$ is well reproduced by Eq. (28) (see Supplementary Note 4).

Then, once the parameters in Eq. (28) is obtained by minimizing Eq. (29), we introduce a spatial translation by $\Delta{\bf r}$ through a phase shift from $\varphi_{\eta}$ to $\varphi_{\eta}-\Delta\varphi_{\eta}$ [50]. Specifically, for the present $3Q$-HL case, a spatial translation by $\Delta{\bf r}=(\Delta x,\Delta y,\Delta z)$ is represented by a phase shift as

$$\Delta{\bf r}=\frac{1}{Q}\left(\Delta\varphi_{1},\Delta\varphi_{2},\Delta\varphi_{3}\right).$$ (30)

Thus, by shifting the phase $\varphi_{\eta}$ in the optimal spin state $\{{\bf S}_{i}^{{\rm ellip}3Q}\}^{*}$, we compute the energy change per site for the model in Eq. (2) measured from that for the optimal spin configuration $\{{\bf S}_{i}^{{\rm ellip}3Q}\}^{*}$, denoted by $\Delta E$.

In the ground state, the Bloch points are located at the interstitial positions of the lattice sites, and the energy increases when they traverse across the unit cubes composed of the lattice sites. This implies that the $3Q$-HL undergoes the potential energy approximately given by

$$\Delta E(\Delta{\bf r})=\sum_{\mu=x,y,z}\frac{V_{\mu}}{2}(1-\cos(2\pi\Delta\mu)).$$ (31)

By assuming this form, we obtain the potential barrier $V_{\mu}$ as $V_{\mu}=\Delta E(\hat{\bm{\mu}}/2)$.