WorksPhysics and Astronomy
Steering Skyrmion Diffusion with Chiral Molecules
Abstract
Chiral molecules adsorbed on a magnetic film interact with its magnetization and can thereby alter the thermal motion of magnetic skyrmions. Here we develop a theoretical model in which an aligned layer of tilted molecules modifies the interfacial magnetic interactions according to its handedness. Within this model a skyrmion responds only to spatial variations of the coupling, so a uniform layer leaves its position unaffected, whereas boundaries and coverage gradients exert forces on it. Near a chiral region, these forces deflect the skyrmion transversely in a direction that reverses between enantiomers, which provides a direct signature of the chiral interaction. In extended periodic patterns, the same forces reshape the long-time diffusion, suppressing it in the direction in which the potential varies and enhancing it in the perpendicular direction. The net effect on the mean diffusion is set by the balance between the gyrotropic and dissipative responses of the skyrmion, as we show exactly for one-dimensional landscapes. Stochastic micromagnetic simulations and collective-coordinate calculations confirm these predictions and show how molecular orientation, spatial patterning, and skyrmion properties together determine the motion.
Introduction
Magnetic skyrmions are localized spin textures whose nonzero topological charge counts the winding of the magnetization across the film, and they behave as stable, particle-like quasiparticles in magnetic materials [1, 2, 3, 4, 5]. In thin layers, exchange, anisotropy and Dzyaloshinskii–Moriya interactions stabilize them at sizes from a few to hundreds of nanometers [6, 7, 8, 9], which makes them candidates for low-power storage, logic and spintronic devices [10, 11, 12]. Because the texture stays well defined while its center moves, the dynamics reduce to those of an effective particle [13, 14, 15, 16] that responds to currents, fields, interfaces and material inhomogeneities [17, 18]. At finite temperatures, the motion is Brownian and can be observed at room temperature in weakly pinned films [19, 20, 21, 22], leading to proposals for using thermal skyrmion motion as a resource for stochastic and probabilistic computing [23, 24, 25].
The response of a skyrmion to a force differs qualitatively from that of an overdamped particle. Its topological charge adds a gyrotropic term to the collective-coordinate dynamics, which rotates the velocity away from the force [26, 27, 28]. Under driving, this rotation causes the skyrmion Hall effect [29, 30], and in thermal motion it gives the long-time diffusion constant a characteristic dependence on the Gilbert damping [31, 32, 33]. Confinement can partially lift the resulting gyrotropic suppression, so defects and interactions can enhance diffusion [34, 35]. Controlling skyrmion diffusion therefore requires control over both the energy landscape and how its gradients are converted into motion.
Chiral molecular layers offer a way to engineer such structured magnetic environments. Chiral molecules exhibit spin-selective transport and polarization effects, collectively known as chiral-induced spin selectivity [36, 37, 38, 39], and adsorbed layers can modify the magnetization of a ferromagnet without an applied current [40]. Theoretically, molecular dipoles can modify magnetic anisotropy through interfacial spin-orbit coupling [41], while tilted chiral molecular potentials can produce handedness-dependent Hall responses [42] and spin-polarized Josephson currents [43]. More recently, experiments on CoFeB have shown that chiral polypeptides affect skyrmion stability and thermal dynamics [44], while nitrogen-vacancy magnetometry has revealed changes in skyrmion size, spacing, and shape with molecular handedness and field polarity [45]. These observations motivate identifying the interfacial interactions that distinguish molecular enantiomers and determining how they influence skyrmion motion.
Here we use the symmetry of interfacial DM interactions [46, 47, 48] to construct an enantiomer-odd response permitted by an aligned and tilted molecular layer. An axisymmetric skyrmion couples to the spatial curl of this response, so molecular boundaries and coverage variations generate a handed energy landscape, as illustrated in Figure 1(a). Gyrotropic motion converts the resulting force into a predominantly transverse deflection that reverses with molecular handedness. Extended patterns instead produce anisotropic diffusion, suppressing diffusion in the direction of the potential variation and enhancing it in the perpendicular direction. The balance between gyrotropic and dissipative motion determines how the mean diffusion changes. Handedness therefore controls the transient direction, while the long-time diffusion tensor is the same for the reflection-related patterns considered here. These observables test the mechanism without requiring a quantitative calibration of a particular molecular interface.
Figure 1. Chiral molecular interface and Néel skyrmion. (a) Skyrmion beneath an aligned layer of tilted chiral molecules of handedness , with in-plane tilt direction , and interfacial field . (b) Relaxed reference skyrmion in the film plane , with in color on the scale of (c), the in-plane magnetization as arrows and the radius dashed. (c) Azimuthal averages of and against the distance from the skyrmion center, with at and for scale.
Model
Magnetic energy
We describe the magnetic film by a unit vector field specifying the local magnetization direction and assume that it is uniform across the film thickness . In the absence of adsorbates, the energy density is
where is the exchange stiffness, is the effective perpendicular anisotropy, is the saturation magnetization and is a perpendicular magnetic field that we set to zero below.
The interfacial DM interaction is linear in magnetization gradients and selects a definite rotational sense. To describe the native interaction and the molecular contribution in a common form, we start from the energy of a nearest-neighbor bond along with DM vector , where is the spin at lattice site and is the lattice constant. For a texture that varies slowly on the lattice scale, expanding to first order in and summing the bonds gives the continuum density
where each vector has units of energy per area and absorbs the spin length and the lattice-to-continuum conversion. For a Rashba-type interfacial mechanism, an effective polar field gives [49, 50, 51], where is an effective spin-orbit coefficient. This form excludes a DM interaction of bulk type, which would require , because is always perpendicular to . We use the relation only to fix the geometry of the two channels below.
The original interface is represented by a field normal to the film, . The Rashba relation then gives and , where is the native DM constant. Substituting these vectors into Eq. (2) gives the Néel DM energy [46]
Its competition with the anisotropy is measured by . For , the regime studied here, the uniform state is the ground state and an isolated skyrmion is metastable. The resulting Néel skyrmion consists of a reversed core surrounded by a domain wall in which the magnetization rotates through the film plane, as the relaxed texture in Figure 1(b) shows. We characterize its size by the radius , at which changes sign, and by the exchange-anisotropy length that sets the scale of the wall. The anisotropy includes the local part of the demagnetizing field, and we omit nonlocal magnetostatic interactions.
An aligned layer of tilted chiral molecules defines a preferential in-plane direction , corresponding to the projection of their molecular axes onto the film plane, as illustrated in Figure 1(a). For a molecular axis tilted by an angle from the surface normal, we write . Together with the surface normal, defines a second in-plane vector , perpendicular to the tilt. Since this cross product is an axial vector, and the molecular handedness is a pseudoscalar, their product transforms as a polar vector. An enantiomer-odd contribution to the effective Rashba field can therefore be written as . At fixed molecular orientation, this contribution reverses with handedness and vanishes upon averaging over uniformly distributed in-plane tilt directions. The coefficient characterizes the interfacial spin-orbit response and depends on microscopic bonding, molecular tilt, and orientational order. We treat the resulting magnetic coupling phenomenologically, without assuming a particular microscopic mechanism. A tilted layer of either handedness can also produce an in-plane field along the tilt, which is the same for both enantiomers. Through the same Rashba relation it adds a coupling vector along . Section II D shows that the skyrmion responds to the
curl of the coupling, which involves only coverage gradients perpendicular to the coupling vector. This even contribution therefore acts on gradients along , whereas the odd one acts on gradients along , and § 1 C uses this difference to tell the two apart.
We retain this enantiomer-odd contribution and treat enantiomer-even modifications of the magnetic interface separately. Applying the Rashba relation to the in-plane field produces DM vectors normal to the film. Their coefficients along the two in-plane bond directions define the vector
We choose the sign assigned to so that , and write the relative coupling as , where is nonnegative. When the two channels share the same spin-orbit coefficient, this convention gives .
Substituting the corresponding bond vectors and into Eq. (2) gives
Unlike the original Néel interaction, this term couples only to spatial rotations of the in-plane magnetization. Adsorption may also modify the perpendicular magnetic anisotropy. We account for this through an enantiomer-even contribution , , where is a phenomenological parameter that can have either sign. This correction is quadratic in the molecular coupling and changes the effective anisotropy from to . We use as a reference, corresponding to at full coverage, and examine the dependence on below. A smaller, higher-gradient correction is included in the numerical implementation and discussed in Appendix A.
In the calculation, the molecules therefore enter through the single vector field
where is the local coverage normalized to unit peak. The handedness is carried by the sign of , so exchanging enantiomers at fixed tilt and coverage amounts to , while , , and are the same for both. We take , and hence , throughout. We consider three coverage profiles, a uniform layer, an isolated Gaussian patch representing a small group of molecules, and a periodic array of such patches representing an extended pattern. Comparing them separates the effect of the amount of coverage from that of its spatial structure. We use , where the anisotropy correction remains small compared with the odd channel.
The total energy density is . Under only changes, reversing its sign, because does not involve the molecules and is quadratic in . The native coupling also keeps its sign in this exchange, so a local change of by adsorption could alter the size and energy of the skyrmion but could not distinguish the enantiomers. The translational potential , the energy of the rigid texture as a function of the position of its center, therefore has the form up to a constant, where and do not depend on the handedness. Its odd part distinguishes the enantiomers, whereas the even part is common to both. We report the two parts separately.
Stochastic dynamics
The magnetization obeys the stochastic Landau–Lifshitz–Gilbert equation
where is the gyromagnetic ratio, is the vacuum permeability, and is the Gilbert damping. The field is the sum of the effective and thermal fields. The effective field is , where is the energy of the film. We compute it as the exact derivative of the discretized energy rather than by discretizing the continuum expression. This matters when and vary in space, because the derivative then contains gradients of the couplings themselves (see Appendix A). The thermal field is Gaussian and white in space and time, and the fluctuation-dissipation theorem fixes its amplitude [52, 53],
where and label Cartesian components of the thermal field, and label mesh cells of volume , is the Boltzmann constant and is the temperature.
We integrate Eq. (7) at unless stated otherwise, using the Stratonovich form of the Heun scheme with the same noise sample in the predictor and the corrector [54, 55]. The time step resolves the stiff exchange modes and is reduced further at low damping to control the stochastic integration error. We test thermal equilibrium against the exact Boltzmann distribution of a macrospin and against the magnetization of the extended film, with values and tolerances given in Appendix A.
The topological charge and the skyrmion position are both obtained from the density , evaluated with central differences. The charge is for the relaxed texture and stays near that value during the simulation. The position is a circular mean of the coordinate weighted by the signed density .
Reduction to the motion of a point
For transport we only need the position of the skyrmion, and the full magnetization dynamics reduces to it when the texture moves without changing shape. We set , where is the relaxed texture and is the position of its center. Projecting Eq. (7) onto the translational mode then gives an equation for alone [26],
where and are the gyrotropic and dissipative coefficients discussed below, is the force from the energy of the texture centered at , and the thermal force has covariance . In a clean film is flat and only the thermal force acts, whereas spatial variations of make position dependent and add the force . We use this rigid-texture approximation for long-time transport, knowing that it neglects internal modes, thermally induced deformations and interactions between skyrmions.
The second term on the left of Eq. (9) is ordinary friction. Its coefficient is a dimensionless measure of the magnetization gradients and generally grows with the circumference of the skyrmion wall. The first term acts perpendicular to the velocity, survives as and does no work, like the magnetic force on a moving charge. Its dimensionless coefficient is fixed by the topological charge, and the material enters both terms only through the common prefactor . For the reference skyrmion of Sec. III A, and , so at the ratio and the gyrotropic term dominates friction by nearly an order of magnitude.
Inverting Eq. (9) shows that a force along one direction produces a velocity rotated from it by nearly a right angle. Its components along the force and perpendicular to it are in the ratio
A skyrmion pushed by a chiral patch therefore moves almost perpendicular to the push. The adsorption geometry sets the axis of the force and the handedness sets its sign, whereas the angle of the deflection is set by the magnetic texture rather than by the molecular chemistry. The same inversion, together with the fluctuation-dissipation theorem, gives the free diffusion constant
which serves as the baseline for all simulations below [31, 32, 33]. For it reduces to , which grows with the damping. The thermal force vanishes as , while the gyrotropic response that converts it into motion stays finite, and the massless description has no cyclotron orbit. For a patterned film we report the long-time diffusion tensor , defined by for the displacement during a time , with all coefficients normalized by . Its mean fixes the displacement variance .
Spatial coupling and symmetry
We now determine the potential generated by the molecular coupling of Eq. (6). For a rigid skyrmion centered at , its enantiomer-odd part is . An axisymmetric texture can be written as and , where , with and the polar coordinates relative to and the helicity. The interaction of Eq. (5) then becomes , where . The coupling is independent of helicity and measures the circulation of around the skyrmion. For a uniform field this circulation vanishes, so a uniform molecular layer exerts no force on a rigid axisymmetric skyrmion.
The weight is concentrated in the skyrmion wall, as shown in Figure 1(c). For a coupling that varies slowly across the texture, expanding about and integrating over the azimuth gives
Thus, the odd potential is governed by the spatial curl of the molecular DM coupling, rather than its local magnitude. This holds exactly for a rigid axisymmetric texture, because by Stokes’ theorem the line integral of around a circle of radius centered at equals the flux of through the enclosed disk. The local expression above further assumes that the curl varies slowly across the skyrmion, and at shorter length scales the full convolution with the skyrmion texture must be retained.
For , with , the curl reduces to , where . A Gaussian molecular patch therefore creates an attractive and a repulsive lobe on opposite sides of the patch, perpendicular to the molecular tilt. Exchanging enantiomers reverses these lobes, whereas the even anisotropy contribution is unchanged. The result is consistent with reflection in the plane spanned by and , which reverses both the handedness and the skyrmion charge while preserving the energy (Appendix A). Reversing alone at zero field leaves the potential unchanged but reverses the gyrotropic response and hence the transverse deflection.
These forces do not produce sustained directed motion in equilibrium. The stationary distribution is proportional to , and gyrotropy generates only circulating probability currents, with no net flux through a periodic cell. Molecular handedness therefore controls the transient deflection following a change in the coupling, rather than a steady drift. The symmetric long-time diffusion tensor is even under at fixed potential. For the reflection-related patterns considered below, it is also identical for the two enantiomers.
Results
Free diffusion and the chiral selection rule
Unless stated otherwise, we use J/m, J/m, J/m, A/m and nm. These values give and are representative of CoFeB, but they define an idealized thin-film model rather than a calibration of the experimental stack. On the 1 nm mesh the relaxed isolated skyrmion of Figure 1(b) has radius 10.9 nm, charge and . § 1 C lowers to obtain a larger texture while keeping all other parameters fixed.
We first follow the reference skyrmion in a clean film, without molecules, using stochastic LLG simulations at several values of the damping, and compare the result with the Thiele description of Sec. II C. The mean squared displacement in Figure 2(a) is linear in lag, with a slope that grows with [31]. The massless collective-coordinate model misses only the short-time bend of the data, which comes from cyclotron motion, frequency-dependent inertia and gyrodamping and does not affect the asymptotic diffusion constant. Figure 2(b) collects the diffusion constants fitted to the long-time slopes and compares them, over a wider range of damping, with Eq. (11) and with the purely dissipative value obtained by setting . The fitted constants agree with the Thiele values within three to eighteen percent for , and each deviation is consistent with its statistical uncertainty. At the gyrocoupling reduces the Thiele value by the factor relative to the dissipative value. Equation (11) has its maximum at , and all our calculations lie below this turnover, where increasing the damping increases the diffusion (see Appendix A for the numerical checks).
Figure 2. Free skyrmion diffusion in the reference film. (a) Connected mean squared displacement of the skyrmion center against lag for several Gilbert damping values , from stochastic LLG simulations (solid) and the Thiele prediction (dashed). (b) Diffusion constant against from LLG (points), Eq. (11) (solid) and the dissipative limit (dotted). The dashed line marks the maximum of Eq. (11) at .
We test Eq. (12) by translating the relaxed texture across several coupling profiles and evaluating its energy. A uniform chiral field produces no position-dependent energy to numerical precision. A localized patch produces an attractive and a repulsive side, which exchange when is reversed, with a peak-to-peak energy that grows linearly in and reaches about at . Periodic and ramped profiles, which also have nonzero curl, act on the skyrmion as well. Spatial variation is therefore the source of the force, and uniform coverage is inert at linear order.
Motion in structured chiral layers
A localized patch converts molecular handedness into a transient deflection. We place the relaxed clean-film skyrmion at the center of a Gaussian patch of width 10 nm and switch on the coupling at , with . Figure 3(a) shows the resulting trajectories for both enantiomers and for the reference film. Because the odd potential crosses zero at the patch center with a finite slope, the initial force there is finite, and the two enantiomers move to opposite sides while the reference skyrmion only wanders thermally. Figure 3(b) shows the mean transverse displacement and its growth with the coupling. Figure 3(c) shows the mean transverse velocity over the first 1.5 ns. Its antisymmetric part is linear in and agrees with the initial-force Thiele prediction within one to six percent for all four coupling strengths. Repeating these calculations at zero temperature separates the response to the force from the thermal motion.
Figure 3. Skyrmion motion in chiral adsorbate potentials. (a)–(c) Transient response to a Gaussian patch of width switched on at , at , for (red), (blue) and the reference film (gray). (a) Trajectories of the skyrmion center, 16 per condition at , with displacements along and along . The cross marks the patch center, where all trajectories start. (b) Mean transverse displacement against time for , 0.0242, 0.0484 and 0.0969, darker for larger , with standard errors shaded. Dashed lines give the weak-coupling prediction of Eq. (13) for the two smallest couplings. (c) Mean transverse velocity over the first 1.5 ns against , with the initial-force Thiele prediction (line). (d) Long-time diffusion coefficients and of the periodic array at , from homogenization of the micromagnetic landscapes (lines) and from collective-coordinate trajectories of 6 s (points). The dotted line marks the free film.
In the simulations the coupling is switched on at , but the same displacement appears without any switching. In the Thiele equation the future motion depends only on the present position, so a skyrmion that starts at in a film already covered by molecules moves on average as in Figure 3(b). Tracking experiments can therefore measure this displacement by averaging trajectory segments that start at the same position. For weak coupling it follows from the free motion of the skyrmion,
where is the probability that a free skyrmion moves from to in a time , a Gaussian with diffusion coefficient , and is the mobility obtained by inverting Eq. (9). At short times this reduces to the initial-force Thiele prediction. For it follows the micromagnetic displacement over the full 6.2 ns within two standard errors, as the dashed lines in Figure 3(b) show, whereas for stronger coupling the skyrmion moves beyond the region where the force is nearly constant. Since the force of a localized patch integrates to zero, the displacement eventually saturates. For it reaches about 35 nm, three skyrmion radii, half of it within about 50 ns.
A check at zero temperature, with the skyrmion placed at six positions across a periodic profile, confirms the mobility itself. The measured velocities follow the prediction with slopes of 1.006 and 0.95 for the transverse and longitudinal components, with relative residuals of 0.06 and 0.24, the larger residual belonging to the weaker longitudinal component. In the gyrotropic limit the signed transverse response is along , where . The handedness thus sets the direction of the force, while the topological charge sets the sense in which it is converted into transverse motion. At the two enantiomers separate by about 13.3 nm after 6.2 ns, compared with 5.9 nm of root-mean-square thermal wandering in the reference film.
An isolated patch changes only the transient motion, whereas an extended pattern can modify the asymptotic diffusion. We therefore consider a periodic array of same-handed Gaussian patches with a common tilt, a period of 21.3 nm and a width of 0.235 times the period. We obtain its translational landscape directly from the micromagnetic energy of the rigid texture. At , the landscape varies by from peak to trough, compared with for the even part alone, and its one-dimensional projection carries 99.4 percent of the variance. The resulting skyrmion potential varies predominantly along and is nearly uniform along . We therefore distinguish diffusion along the potential variation, , from diffusion perpendicular to it, .
For a landscape that depends only on the component of along , inverting Eq. (9) gives and with the dissipative and gyrotropic mobilities and , so that , and with thermal velocities obeying . Motion along is hindered by the potential barriers, while gyrotropy generates an additional velocity along that follows the local potential gradient. This velocity vanishes on average in equilibrium, but its temporal correlations enhance the spreading. Periodic homogenization gives the exact result
where and is the Lifson–Jackson factor, with uniform averages over one period [56]. The normalized suppression along is independent of , although itself depends on . The enhancement along is the finite-barrier counterpart of confinement-enhanced gyrotropic diffusion. Exchanging and leaves unchanged, and the even adsorption term keeps the diagonal diffusivities equal for the reflection-related profiles studied here, so the tensor is the same for both enantiomers.
Landscapes are already known to enhance skyrmion diffusion, since defects and mutual interactions raise the long-time diffusion at low damping, where the gyrotropic term dominates [34, 35]. This work identifies what an axisymmetric skyrmion couples to, namely the curl of an enantiomer-odd interfacial channel rather than the coverage itself, and gives the resulting transport as a closed-form tensor. The response is therefore anisotropic rather than scalar. The mean diffusion rises or falls according to whether exceeds one, independently of the landscape, and the magnitude of the change is controlled by a pattern length scale set by the skyrmion form factor.
Figure 3(d) shows the two diffusion coefficients of the array against at . They are computed by periodic homogenization of the full two-dimensional landscape (Appendix C) and checked against stochastic collective-coordinate trajectories. At the homogenization gives and , in agreement with the stochastic estimates and . The one-dimensional result, and , differs by about 0.3 and 3 percent, a projection error that is separate from the statistical uncertainty of the trajectories. The two-dimensional values agree between the enantiomers and are stable under refinement of the cell solver. Diffusion perpendicular to the potential variation is almost eight times faster than along it, and the mean diffusion is 2.934 times the free value, a 193 percent enhancement. Varying the independent anisotropy coefficient over changes the directional coefficients by less than four percent and preserves this anisotropy. The effect is therefore governed by the odd interfacial coupling rather than by the particular even correction.
Enhanced and suppressed diffusion
From Eq. (14), . For any nonflat one-dimensional potential, the mean diffusion therefore increases when gyrotropic motion dominates and decreases when dissipation dominates, with the crossover exactly at . A two-dimensional landscape can shift this crossover, so its position must be calculated rather than assumed.
Figure 4(a) plots against , from this result over a range of and from the full micromagnetic landscapes of two textures, each followed as the damping varies. The first is the reference skyrmion. The second is a larger texture, obtained by lowering the anisotropy to , which has , and on the 0.5 nm mesh. Together they establish both regimes. In a pattern of period 48 nm, the larger texture gives at and 0.800 at . Changing only the damping thus turns a 22 percent enhancement into a 20 percent suppression, and the residual two-dimensional structure shifts the crossover from the projected to 0.258. These results describe transport within the rigid collective-coordinate model, not a finite-temperature micromagnetic simulation of the larger texture, and Appendix A reports the mesh and relaxation checks.
Figure 4. Control of the mean diffusion by damping and pattern period. (a) against the gyrotropic ratio , from Eq. (14) for (dashed) and (band), and from two-dimensional homogenization (solid) for the skyrmions of radius 10.9 nm and 44.7 nm in coverage periods of 21.3 nm and 48 nm, with , and . Circles mark for the smaller and for the larger skyrmion. (b) Thin-wall prediction of against the period of the coverage , for nm, nm, nm, mJ/m, and K. The vertical line marks the strongest coupling, near . Dotted horizontal lines mark the free film.
The size dependence comes from the wall. For an axisymmetric texture , which for a circular wall of width reduces to , and for this gives . The gyrotropic coefficient is fixed by the topological charge, whereas the dissipation grows with the length of the wall, so larger skyrmions approach the dissipative regime more easily. Figure 5 in Appendix A compares this law with the dissipation integral of textures relaxed over a range of , which approach it from above. The law underestimates by 11 percent for the reference skyrmion and by 0.2 percent at , so it is reliable for the large skyrmions where we use it. The radius alone does not decide the regime of a real film, however, because the wall width and the damping enter in the same way.
Figure 5. Test of the circular-wall law used in § 1 C. Dissipation integral against for nine textures relaxed for on the 1 nm mesh (filled circles) and for on the 0.5 nm mesh (open circle), compared with (line). Labels give the radii of the two transport textures. The inset shows against for each texture, in the shade of its point, together with (dashed).
The pattern must also vary on the length scale that the skyrmion samples. For sinusoidal coverage with wave number , period and first-harmonic amplitude , the odd potential has the amplitude , where is a Bessel function. For a narrow wall approaches , as Figure 5 shows, and the amplitude becomes . It peaks first at , that is at . Much shorter periods are averaged out by the skyrmion, and much longer ones weaken the coverage gradient. The diffusion response is therefore nonmonotonic in the period. Figure 4(b) shows against the period for a large skyrmion with and , obtained from Eq. (14)
with the finite wall width retained in the amplitude. For this skyrmion lies on the suppressed side, and the suppression is strongest near , where it reaches 26 percent at and 29 percent at . The strength of the effect is thus tied to a controllable length scale rather than to an independently chosen barrier height.
The experiments of Ref. [44] found that chiral molecular adsorption can either suppress or enhance skyrmion diffusion, depending on the magnetic-field polarity. Our model also permits both regimes, but through the balance of gyrotropic and dissipative motion rather than field reversal. Indeed, reversing the skyrmion charge at fixed potential leaves the symmetric diffusion tensor unchanged. The observed field-polarity dependence therefore cannot be explained by this mechanism alone.
Our results instead identify a distinct contribution of molecular chirality, a spatially varying odd interfacial coupling that produces a handed transverse deflection. We express its strength as the surface DMI coefficient . Our simulations use , comparable to the DMI of a heavy-metal interface, which makes the response clear above the thermal noise, and Figure 4(b) uses . For weaker molecular couplings the most sensitive measurements are those that grow linearly with . A straight edge of the molecular coverage running along is the simplest example. The exact form of Sec. II D gives its potential,
a ridge that extends about on each side of the edge with a height close to . The micromagnetic energy of the reference skyrmion reproduces this ridge within about one percent at its maximum. In a molecular stripe along , one edge therefore attracts the skyrmion and the other repels it, and the two swap between enantiomers. The skyrmion spends more time near the attracting edge by the factor , whereas changes of anisotropy are the same at both edges and do not contribute. Edges along respond only to the even in-plane contribution, which the same calculation confirms. A ridge of requires for , about two hundred times less than in our simulations. The mean diffusion instead changes quadratically, for a weak sinusoidal landscape of peak-to-trough variation , and therefore needs landscapes of order . The most favorable conditions are thus a film with weak pinning and soft perpendicular anisotropy, where is large and micrometer skyrmions can be followed optically, with molecules adsorbed directly on the magnetic layer in stripes aligned with their tilt. A periodic molecular pattern of either enantiomer should in addition produce different diffusion coefficients in the two in-plane directions. Such measurements could distinguish this mechanism from changes in anisotropy, pinning, or other adsorption-induced modifications of the magnetic film.
Conclusions
An aligned chiral adsorbate layer permits an odd interfacial spin coupling, and an axisymmetric skyrmion responds to its spatial curl. Because the coupling cancels within a uniform layer, boundaries and coverage variations are the source of translational forces. Near a localized patch these forces produce a handed transverse deflection, which we verified with stochastic micromagnetic simulations. In an extended pattern they suppress diffusion in the direction of the potential variation and enhance it in the perpendicular direction.
The balance between gyrotropic and dissipative mobilities decides whether this anisotropy raises or lowers the mean diffusion. The result is exact in one dimension, and independent calculations on the full two-dimensional landscapes confirm both regimes. The response persists when the even adsorption anisotropy is varied, and its magnitude depends on the pattern wavelength through averaging over the skyrmion wall, which offers a geometric way to strengthen or weaken the coupling. These results establish a theoretical mechanism for steering and redistributing thermal skyrmion motion. The handed effects grow linearly with the molecular coupling, whereas the change of diffusion grows quadratically, so straight edges of the coverage aligned with the molecular tilt offer the most sensitive test. Transient deflection and directional diffusion provide complementary observables, both for controlled patterns and for conditional comparison with molecular experiments.
Acknowledgments
We thank Artem Volosniev for useful discussions and comments, and Yossi Paltiel for discussions of how the chiral polypeptide layers are prepared and imaged. The simulations were performed on the scientific computing clusters of the Institute of Science and Technology Austria and the University of Hamburg. U.P. was funded by the European Union, in the framework of the Marie Skłodowska-Curie Actions - CISSE Doctoral Network, under Grant Agreement no. 101071886. Views and opinions expressed are, however, those of the author(s) only and do not necessarily reflect those of the European Union or the European Research Executive Agency (REA). Neither the European Union nor the granting authority can be held responsible for them.
Numerical methods and validation
The reference micromagnetic calculations use a square 1 nm mesh with periodic boundaries. The clean-film and localized-patch trajectories use a 96 nm box, and the small-skyrmion landscape uses a 64 nm box containing exactly three coverage periods. The larger texture is checked on 2 nm, 1 nm and 0.5 nm meshes, in a 384 nm box of eight periods for the two coarser meshes and a 288 nm box of six periods for the finest. Cropping the 1 nm texture between these boxes changes its radius and dissipative integral by less than the quoted precision. Periodic Gaussian image sums keep the coverage smooth across all boundaries.
At the box-size test gives , 1.04 and 1.03 for 64 nm, 96 nm and 128 nm boxes. The two larger boxes agree with the free Thiele prediction, and we use 96 nm for the baseline. These finite-temperature diffusivities carry ensemble sampling uncertainty, so we do not use the difference of two means to infer convergence at the percent level.
Every energy term carries the cell volume , where and are the mesh spacings. Exchange uses forward differences and the DM terms use central differences. The effective field is the exact derivative of this discrete energy, including the gradients of spatially varying coefficients, and finite-difference checks on random spin configurations agree with it to better than of the root-mean-square field.
The numerical implementation also contains the even higher-gradient correction , with the lattice constant nm and , evaluated with discrete periodic second derivatives in both the energy and the field. We treat it as part of the numerical model, not as a microscopically required compass interaction. Its contribution to the potential depth of the larger skyrmion is below on all three meshes, whereas varying tests the much larger local anisotropy contribution.
We minimize the larger texture in normalized Cartesian spin coordinates using the tangential projection of the analytic energy gradient, and we record both the optimizer status and the maximum torque . The final maximum torques are about 0.12, 0.23 and 1.49 A/m on the 2 nm, 1 nm and 0.5 nm meshes. The corresponding radii are 40.39 nm, 43.69 nm and 44.75 nm, and the dissipative integrals are 40.46, 44.55 and 45.80. Refining from 1 nm to 0.5 nm changes from 1.257 to 1.222 at and from 0.795 to 0.800 at . The crossover is therefore robust, while the numbers retain a discretization dependence of a few percent.
Figure 5 tests the circular-wall law of § 1 C. We minimize nine textures for in the same way on the 1 nm mesh, in boxes between 7.9 and 12.8 radii wide, with final maximum torques between 0.2 and 1.7 A/m. The texture reproduces the 1 nm radius and dissipative integral of the larger texture given above. Since is fixed by the material parameters, the law contains no free parameter. It underestimates by 14 percent at and by 0.2 percent at , and the wall profiles approach as grows.
We check the thermal-field normalization on independent uniaxial macrospins with . The simulated second moments, 0.3879 and 0.3477 at 300 K and 1000 K, differ from
the exact values 0.3898 and 0.3497 by less than 0.6 percent. On the 1 nm mesh the stochastic time steps are 20.75 fs for and 4.15 fs at , where the smaller step controls the low-damping integration error. At these steps the equilibrium magnetization of the extended film agrees with the spin-wave estimate within about 0.5 percent.
Further checks show that the energy is unchanged, to numerical precision, under the mirror operation combined with the axial spin transformation and enantiomer exchange. A nonuniform curl-free test field , with amplitude and wave number , produces a potential variation below for an axisymmetric texture in a sufficiently large box. FFT correlation of the discrete coupling fields with the texture densities reproduces explicit rigid translations to the same precision.
Position and displacement statistics
The position estimator weights the coordinate by the signed topological density through a circular mean that respects the periodic boundaries,
and similarly for , where is the coordinate of cell and is the box length along . Successive increments are unwrapped before displacements are accumulated. Normalizing by the signed charge removes the extensive positive thermal background that biases positive-weight estimators. In tests with component noise of amplitude 0.1, the position bias is about 0.024 nm, compared with 0.208 nm for a positive-weight estimator. The circular estimator also passes tests with the skyrmion placed on the periodic boundary. These tests characterize its numerical accuracy and do not prove exact unbiasedness at arbitrary temperature.
For the free-film trajectories we average displacement variances over time origins and independent trajectories, using the connected second moment of the displacement over a lag , because a mean drift would otherwise add a contribution quadratic in lag. Single-origin ensemble averages have the same expectation for stationary Brownian increments and are not biased downward, but they have a larger sampling uncertainty.
The stochastic collective-coordinate estimates use uncentered time-averaged second moments, with equilibrium initial conditions and zero mean drift expected by symmetry. We fit them through the origin over lags up to one fifth of the trajectory duration and use running diffusivities as a finite-lag diagnostic. Bootstrap resampling over whole trajectories retains the correlations between time origins. These estimates serve as independent checks of the asymptotic cell calculation of Appendix C rather than defining its long-time limit.
Figures 3(a)–3(c) instead report mean displacements with standard errors computed separately for each sign. The two signs share thermal-noise seeds, which reduces the variance of their difference but does not cancel fluctuations exactly in nonlinear dynamics. The separate errors are therefore conservative for the difference.
Periodic transport calculation
An independent homogenization calculation determines the long-time diffusion from the periodic landscape without fitting trajectories. We write and and choose the unit of time so that . The drift is then , and the stationary density is . The periodic correctors , one for each direction , obey
with zero spatial mean and give the effective symmetric tensor
where are the Cartesian unit vectors and is the average over one cell weighted by . We solve for the correctors with Fourier derivatives and a preconditioned iterative linear solver, check the residual, and refine the periodic grid independently of the micromagnetic mesh.
For a one-dimensional potential we define the normalized weight , with a uniform average over one period. The corrector derivatives are then and , and the identities yield Eq. (14), including its vanishing off-diagonal component. The normalization of is essential for this result.
The solver reproduces the exact tensor for a cosine potential to better than in both dissipative and gyrotropic test cases. The coefficients of the small-skyrmion landscape are unchanged on , and grids, those of the larger landscape change by less than on , and grids, and typical relative residuals of the linear system are below . These tolerances concern only the reduced transport calculation. They are independent of the micromagnetic mesh error and of the rigid-texture approximation.
References
- [1]T. H. R. Skyrme, A non-linear field theory, Proc. R. Soc. Lond. A 260, 127 (1961).
- [2]A. N. Bogdanov and D. A. Yablonskii, Thermodynamically stable “vortices” in magnetically ordered crystals, Sov. Phys. JETP 68, 101 (1989).
- [3]S. Mühlbauer, B. Binz, F. Jonietz, C. Pfleiderer, A. Rosch, A. Neubauer, R. Georgii, and P. Böni, Skyrmion lattice in a chiral magnet, Science 323, 915 (2009).DOI
- [4]X. Z. Yu, Y. Onose, N. Kanazawa, J. H. Park, J. H. Han, Y. Matsui, N. Nagaosa, and Y. Tokura, Real-space observation of a two-dimensional skyrmion crystal, Nature (London) 465, 901 (2010).DOI
- [5]S. Heinze, K. von Bergmann, M. Menzel, J. Brede, A. Kubetzka, R. Wiesendanger, G. Bihlmayer, and S. Blügel, Spontaneous atomic-scale magnetic skyrmion lattice in two dimensions, Nat. Phys. 7, 713 (2011).
- [6]J. Sampaio, V. Cros, S. Rohart, A. Thiaville, and A. Fert, Nucleation, stability and current-induced motion of isolated magnetic skyrmions in nanostructures, Nat. Nanotechnol. 8, 839 (2013).DOI
- [7]S. Woo, K. Litzius, B. Krüger, M.-Y. Im, L. Caretta, K. Richter, M. Mann, A. Krone, R. M. Reeve, M. Weigand, P. Agrawal, I. Lemesh, M.-A. Mawass, P. Fischer, M. Kläui, and G. S. D. Beach, Observation of room-temperature magnetic skyrmions and their current-driven dynamics in ultrathin metallic ferromagnets, Nat. Mater. 15, 501 (2016).
- [8]C. Moreau-Luchaire, C. Moutafis, N. Reyren, J. Sampaio, C. A. F. Vaz, N. Van Horne, K. Bouzehouane, K. Garcia, C. Deranlot, P. Warnicke, P. Wohlhüter, J.-M. George, M. Weigand, J. Raabe, V. Cros, and A. Fert, Additive interfacial chiral interaction in multilayers for stabilization of small individual skyrmions at room temperature, Nat. Nanotechnol. 11, 444 (2016).
- [9]O. Boulle, J. Vogel, H. Yang, S. Pizzini, D. de Souza Chaves, A. Locatelli, T. O. Menteş, A. Sala, L. Aballe, M. Foerster, M. Belmeguenai, Y. Roussigné, A. Stashkevich, S. M. Chérif, L. D. Buda-Prejbeanu, O. Klein, M. Chshiev, S. Auffret, I. M. Miron, and G. Gaudin, Room-temperature chiral magnetic skyrmions in ultrathin magnetic nanostructures, Nat. Nanotechnol. 11, 449 (2016).DOI
- [10]K. K. Mishra, A. H. Lone, S. Srinivasan, H. Fariborzi, and G. Seti, Magnetic skyrmionic spin: from fundamental physics to pioneering applications, Applied Physics Reviews 12, 011315 (2025).arxiv.org/abs/2308.11811
- [11]G. Finocchio, F. Büttner, T. Tomasello, M. Carpentieri, and M. Kläui, Magnetic skyrmions: from fundamental to applications, Journal of Physics D: Applied Physics 49, 423001 (2016).
- [12]T. Dohi, R. M. Reeve, and M. Kläui, Thin film skyrmionics, Annual Review of Condensed Matter Physics 13, 73–95 (2022).DOI
- [13]N. Nagaosa and Y. Tokura, Topological properties and dynamics of magnetic skyrmions, Nat. Nanotechnol. 8, 899 (2013).DOI
- [14]A. Fert, N. Reyren, and V. Cros, Magnetic skyrmions: advances in physics and potential applications, Nat. Rev. Mater. 2, 17031 (2017).DOI
- [15]C. Back, V. Cros, H. Ebert, K. Everschor-Sitte, A. Fert, M. Garst, T. Ma, S. Mankovsky, T. L. Monchesky, M. Mostovoy, N. Nagaosa, S. S. P. Parkin, C. Pfleiderer, N. Reyren, A. Rosch, Y. Taguchi, Y. Tokura, K. von Bergmann, and J. Zang, The 2020 skyrmionics roadmap, J. Phys. D: Appl. Phys. 53, 363001 (2020).
- [16]K. Everschor-Sitte, J. Masell, R. M. Reeve, and M. Kläui, Perspective: Magnetic skyrmions—overview of recent progress in an active research field, J. Appl. Phys. 124, 240901 (2018).
- [17]A. Fert, V. Cros, and J. Sampaio, Skyrmions on the track, Nat. Nanotechnol. 8, 152 (2013).DOI
- [18]J. Iwasaki, M. Mochizuki, and N. Nagaosa, Universal current-velocity relation of skyrmion motion in chiral magnets, Nat. Commun. 4, 1463 (2013).
- [19]J. Zázvorka, F. Jakobs, D. Heinze, N. Keil, S. Kromin, S. Jaiswal, K. Litzius, G. Jakob, P. Virnau, D. Pinna, K. Everschor-Sitte, L. Rózsa, A. Donges, U. Nowak, and M. Kläui, Thermal skyrmion diffusion used in a reshuffler device, Nat. Nanotechnol. 14, 658 (2019).DOI
- [20]L. Zhao, Z. Wang, X. Zhang, X. Liang, J. Xia, K. Wu, H.-A. Zhou, Y. Dong, G. Yu, K. L. Wang, X. Liu, Y. Zhou, and W. Jiang, Topology-dependent brownian gyromotion of a single skyrmion, Phys. Rev. Lett. 125, 027206 (2020).
- [21]T. Nozaki, Y. Jibiki, M. Goto, E. Tamura, T. Nozaki, H. Kubota, A. Fukushima, S. Yuasa, and Y. Suzuki, Brownian motion of skyrmion bubbles and its control by voltage applications, Appl. Phys. Lett. 114, 012402 (2019).DOI
- [22]T. Dohi, M. Weißenhofer, N. Kerber, F. Kammerbauer, Y. Ge, K. Raab, J. Zázvorka, M.-A. Syskaki, A. Shahee, M. Ruhwedel, T. Böttcher, P. Pirro, G. Jakob, U. Nowak, and M. Kläui, Enhanced thermally-activated skyrmion diffusion with tunable effective gyrotropic force, Nat. Commun. 14, 5424 (2023).
- [23]D. Prychynenko, M. Sitte, K. Litzius, B. Krüger, G. Bourianoff, M. Kläui, J. Sinova, and K. Everschor-Sitte, Magnetic skyrmion as a nonlinear resistive element: A potential building block for reservoir computing, Phys. Rev. Applied 9, 014034 (2018).DOI
- [24]D. Pinna, F. Abreu Araujo, J.-V. Kim, V. Cros, D. Querlioz, P. Bessiere, J. Droulez, and J. Grollier, Skyrmion gas manipulation for probabilistic computing, Phys. Rev.DOI
- [25]K. Raab, M. A. Brems, G. Beneke, T. Dohi, J. Rothörl, F. Kammerbauer, J. H. Mentink, and M. Kläui, Brownian reservoir computing realized using geometrically confined skyrmion dynamics, Nat. Commun. 13, 6982 (2022).
- [26]A. A. Thiele, Steady-state motion of magnetic domains, Phys. Rev. Lett. 30, 230 (1973).DOI
- [27]K. Everschor, M. Garst, B. Binz, F. Jonietz, S. Mühlbauer, C. Pfleiderer, and A. Rosch, Rotating skyrmion lattices by spin torques and field or temperature gradients, Phys. Rev. B 86, 054432 (2012).DOI
- [28]I. Makhfudz, B. Krüger, and O. Tchernyshov, Inertia and chiral edge modes of a skyrmion magnetic bubble, Phys. Rev. Lett. 109, 217201 (2012).
- [29]W. Jiang, X. Zhang, G. Yu, W. Zhang, X. Wang, M. Benjamin Jungfleisch, J. E. Pearson, X. Cheng, O. Heinonen, K. L. Wang, Y. Zhou, A. Hoffmann, and S. G. E. te Velthuis, Direct observation of the skyrmion Hall effect, Nat. Phys. 13, 162 (2017).DOI
- [30]K. Litzius, I. Lemesh, B. Krüger, P. Bassirian, L. Caretta, K. Richter, F. Büttner, K. Sato, O. A. Tretiakov, J. Förster, R. M. Reeve, M. Weigand, I. Bykova, H. Stoll, G. Schütz, G. S. D. Beach, and M. Kläui, Skyrmion Hall effect revealed by direct time-resolved X-ray microscopy, Nat. Phys. 13, 170 (2017).DOI
- [31]C. Schütte, J. Iwasaki, A. Rosch, and N. Nagaosa, Inertia, diffusion, and dynamics of a driven skyrmion, Phys. Rev. B 90, 174434 (2014).DOI
- [32]J. Miltat, S. Rohart, and A. Thiaville, Brownian motion of magnetic domain walls and skyrmions, and their diffusion constants, Phys. Rev. B 97, 214426 (2018).DOI
- [33]M. Weißenhofer and U. Nowak, Diffusion of skyrmions: the role of topology and anisotropy, New J. Phys. 22, 103059 (2020).
- [34]P. Rieger, M. Weißenhofer, and U. Nowak, Defect-enhanced diffusion of magnetic skyrmions, Phys. Rev. B 108, 144417 (2023).
- [35]D. Schick, M. Weißenhofer, L. Rózsa, J. Rothörl, P. Virnau, and U. Nowak, Two levels of topology in skyrmion lattice dynamics, Phys. Rev. Research 6, 013097 (2024).DOI
- [36]B. Göhler, V. Hamelbeck, T. Z. Markus, M. Kettner, G. F. Hanne, Z. Vager, R. Naaman, and H. Zacharias, Spin selectivity in electron transmission through self-assembled monolayers of double-stranded DNA, Science 331, 894 (2011).DOI
- [37]R. Naaman and D. H. Waldeck, Chiral-induced spin selectivity effect, J. Phys. Chem. Lett. 3, 2178 (2012).DOI
- [38]R. Naaman, Y. Paltiel, and D. H. Waldeck, Chiral molecules and the electron spin, Nat. Rev. Chem. 3, 250 (2019).DOI
- [39]B. P. Bloom, Y. Paltiel, R. Naaman, and D. H. Waldeck, Chiral induced spin selectivity, Chem. Rev. 124, 1950 (2024).DOI
- [40]O. Ben Dor, S. Yochelis, A. Radko, K. Vankayala, E. Capua, A. Capua, S.-H. Yang, L. T. Baczewski, S. S. P. Parkin, R. Naaman, and Y. Paltiel, Magnetization switching in ferromagnets by adsorbed chiral molecules without current or external magnetic field, Nat. Commun. 8, 14567 (2017).DOI
- [41]R. Alhyder, A. Cappellaro, M. Lemeshko, and A. Volosniev, Achiral dipoles on a ferromagnet can affect its magnetization direction, The Journal of Chemical Physics 159, 104103 (2023). [42] R. Alhyder, M. Lemeshko, and A. Cappellaro, Quantum transport in the presence of a chiral molecular potential, The Journal of Chemical Physics 162, 234106 (2025).
- [43]O. Kuliashov, A. Cappellaro, O. Millo, Y. Paltiel, M. Lemeshko, and R. Alhyder, Supercurrents in Josephson junctions with chiral molecular potentials, Physical Review B 114, 154510 (2026).DOI
- [44]Y. Kapon, F. Kammerbauer, T. Balland, S. Yochelis, M. Kläui, and Y. Paltiel, Effects of chiral polypeptides on skyrmion stability and dynamics, Nano Lett. 25, 306 (2025).DOI
- [45]B. Hondamuni, T. Balland, F. Kammerbauer, A. Moharana, Bindu, A. Singh, M. Ozeri, S. Yochelis, Y. Paltiel, O. Dhungel, Z. Kazi, K.-M. C. Fu, H. Watanabe, M. Kläui, A. Wickenbrock, N. Bar-Gill, A. Wittmann, and D. Budker, Magnetic microscopy of skyrmions in magnetic thin films with chiral overlayers, Phys. Rev. Appl. 26, 044075 (2026).
- [46]S. Rohart and A. Thiaville, Skyrmion confinement in ultrathin film nanostructures in the presence of Dzyaloshinskii–Moriya interaction, Phys. Rev. B 88, 184422 (2013).DOI
- [47]H. T. Nembach, J. M. Shaw, M. Weiler, E. Jué, and T. J. Silva, Linear relation between Heisenberg exchange and interfacial Dzyaloshinskii-Moriya interaction in metal films, Nat. Phys. 11, 825 (2015).
- [48]T. Srivastava, M. Schott, R. Juge, V. Křížáková, M. Belmeguenai, Y. Roussigné, A. Bernand-Mantel, L. Ranno, S. Pizzini, S.-M. Chérif, A. Stashkevich, S. Auffret, O. Boulle, G. Gaudin, M. Chshiev, C. Baraduc, and H. Béa, Large-voltage tuning of Dzyaloshinskii-Moriya interactions: A route toward dynamic control of skyrmion chirality, Nano Lett. 18, 4871 (2018).DOI
- [49]A. Crépieux and C. Lacroix, Dzyaloshinskii-Moriya interactions induced by symmetry breaking at a surface, J. Magn. Magn. Mater. 182, 341 (1998).
- [50]K.-W. Kim, H.-W. Lee, K.-J. Lee, and M. D. Stiles, Chirality from interfacial spin-orbit coupling effects in magnetic bilayers, Phys. Rev. Lett. 111, 216601 (2013).DOI
- [51]A. Kundu and S. Zhang, Dzyaloshinskii-Moriya interaction mediated by spin-polarized band with Rashba spin-orbit coupling, Phys. Rev. B 92, 094434 (2015).DOI
- [52]W. F. Brown, Thermal fluctuations of a single-domain particle, Phys. Rev. 130, 1677 (1963).
- [53]R. Kubo, The fluctuation-dissipation theorem, Rep. Prog. Phys. 29, 255 (1966).DOI
- [54]J. L. García-Palacios and F. J. Lázaro, Langevin dynamics study of the dynamical properties of small magnetic particles, Phys. Rev. B 58, 14937 (1998).
- [55]J. H. Mentink, M. V. Tretyakov, A. Fasolino, M. I. Katsnelson, and T. Rasing, Stable and fast semi-implicit integration of the stochastic Landau-Lifshitz equation, J. Phys.: Condens. Matter 22, 176001 (2010).DOI
- [56]S. Lifson and J. L. Jackson, Self-diffusion of ions in polyelectrolyte solutions, J. Chem. Phys. 36, 2410 (1962).




