ARTICLE

Received 21 Apr 2015 | Accepted 26 Nov 2015 | Published 19 Jan 2016

DOI: 10.1038/ncomms10293

OPEN

Magnetic bilayer-skyrmions without skyrmion

Hall effect
Xichao Zhang1,2, Yan Zhou1,2 & Motohiko Ezawa3

Magnetic skyrmions might be used as information carriers in future advanced memories,

logic gates and computing devices. However, there exists an obstacle known as the skyrmion
Hall effect (SkHE), that is, the skyrmion trajectories bend away from the driving current
direction due to the Magnus force. Consequently, the skyrmions in constricted geometries
may be destroyed by touching the sample edges. Here we theoretically propose that the SkHE
can be suppressed in the antiferromagnetically exchange-coupled bilayer system, since the
Magnus forces in the top and bottom layers are exactly cancelled. We show that such a pair
of SkHE-free magnetic skyrmions can be nucleated and be driven by the current-induced
torque. Our proposal provides a promising means to move magnetic skyrmions in a perfectly
straight trajectory in ultra-dense devices with ultra-fast processing speed.

1 School

of Electronic Science and Engineering and Collaborative Innovation Center of Advanced Microstructures, Nanjing University, Nanjing 210093, China.
of Physics, University of Hong Kong, Pokfulam Road, Hong Kong, China. 3 Department of Applied Physics, University of Tokyo,
Hongo 7-3-1, Tokyo 113-8656, Japan. Correspondence and requests for materials should be addressed to Y.Z. (email: [email protected]) or to M.E.
(email: [email protected]).
2 Department

NATURE COMMUNICATIONS | 7:10293 | DOI: 10.1038/ncomms10293 | www.nature.com/naturecommunications

ARTICLE

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10293

he magnetic skyrmion, a topologically protected spin

texture with a quantized topological number18, has been a
prominent topic of condensed matter physics since the rst
experimental observations of skyrmion lattices in bulk noncentrosymmetric magnets9,10 and thin lms11. The creation and
transmission of an isolated magnetic skyrmion in thin lms is a
key for future skyrmionics, which utilizes skyrmions as
information carriers in advanced memories, logic gates and
computing devices1215. Recently, skyrmions were realized at
room temperature16,17. The strong spin orbit coupling at the
interface between the magnetic layer with perpendicular magnetic
anisotropy (PMA) and the underlying heavy metal layer provides
a sizeable DzyaloshinskiiMoriya interaction (DMI) to stabilize
skyrmions2,1821.
Skyrmions can be displaced by the spin-polarized current,
where the angular momentum of the spin-polarized current is
transferred from itinerant conduction electrons to the magnetic
moment of magnetic skyrmions2,3,22. However, a skyrmion
cannot move in a straight line along the driving current
direction, as it feels the Magnus force which shifts its trajectory
when moving1. This is referred to as the Skyrmion Hall effect
(SkHE). As a result, magnetic skyrmions may be destroyed at the
edges of nanotracks23,24. Especially, the distance that a skyrmion
can be moved drastically reduces as the nanotrack becomes
narrower. This will be a roadblock to realize skyrmionic
ultradense devices, where it is preferable to use a nanotrack
with width of tens of nanometres.
One approach of reducing the SkHE is to explore spin-wave (or
magnon)-driven skyrmion motion2531. It has been shown that a
skyrmion can be displaced by magnons induced by thermal
gradients in insulating chiral ferromagnets30,31. It has also been
shown that the skyrmion Hall angle may become zero for highenergy magnons29. However, as compared with the current-driven
skyrmion scheme, it is difcult to generate spin waves in a

nanometre-size nanotrack with spectral properties needed for

driving the motion of a skyrmion27,32. It is also difcult to realize
a skyrmionic nanocircuit based on thermal gradients. Therefore, we
focus on the current-driven skyrmion motion in this work. Here we
show that the SkHE is completely suppressed by considering two
perpendicularly magnetized ferromagnetic (FM) sublayers strongly
coupled via the antiferromagnetic (AFM) exchange interaction with
a heavy-metal layer beneath the bottom FM layer33,34 (Fig. 1ac).
When one skyrmion is created in the top FM layer, another
skyrmion is simultaneously created in the bottom FM layer when
the interlayer AFM exchange coupling is strong enough. We refer to
such a pair of AFM-coupled magnetic skyrmions as a magnetic
bilayer-skyrmion (Fig. 1d,e). We show a bilayer-skyrmion can travel
over arbitrarily long distances driven by spin currents without
touching nanotrack edges thanks to the complete suppression of the
SkHE (Fig. 2a,b). Our result provides a guideline for designing
realistic ultradense and ultrafast information processing, storage and
logic computing devices based on magnetic skyrmions.
Results
Hamiltonian. We investigate a bilayer system where the top and
bottom FM layers are coupled antiferromagnetically by the
exchange interaction, as illustrated in Fig. 1a. The Hamiltonian
for each layer reads


X
X
mti  mtj
D  mti mtj
Ht  Aintra
hi;ji

hi;ji

i
Xh
2
HDDI ;
K
1  mt;z

where t is the layer index (t T, B), mti represents

the local
 
magnetic moment orientation normalized as mti  1 at the site
i, and hi,ji runs over all the nearest neighbour sites in each layer.
The rst term represents the FM exchange interaction with FM

MTJ

Spin-polarized current

mp

x
y

Top FM layer

AFM-coupled
nanodisks

Insulating spacer
Bottom FM layer
+1

Top FM layer
Insulating spacer
Bottom FM layer

1
Lead

Lead

Nanotrack (CPP)

Heavy metal

mz

Charge current

Spin current

Top FM layer
Insulating spacer
Bottom FM layer

Electron flow
Electron flow

Lead

Lead

Nanotrack (CIP)

Heavy metal

Figure 1 | Schematics of the antiferromagnetically exchange-coupled bilayer systems and the bilayer-skyrmion. (a) The AFM-coupled bilayer nanodisk
with a diameter of 100 nm for bilayer-skyrmion creation. The spin-polarized current (polarization direction p  z) is injected into the top ferromagnetic
(FM) layer in the central circle with a diameter of 40 nm, which can be realized by means of a magnetic tunnel junction (MTJ) injector above the nanodisk.
(b) The AFM-coupled bilayer nanotrack (500  50  3 nm3) for the study of the motion of a bilayer-skyrmion driven by the current perpendicular to the
plane (CPP). The charge current ows through the heavy-metal substrate along the x-direction, which gives rise to a spin current (p y) perpendicularly
injected to the bottom FM layer because of the spin Hall effect. The skyrmion in the bottom FM layer is driven by the spin current, whereas the skyrmion in
the top FM layer moves accordingly due to the interlayer AFM exchange coupling. (c) The AFM-coupled bilayer nanotrack (500  50  3 nm3) for the
study of the motion of a bilayer-skyrmion driven by the in-plane current (CIP). The electrons ow towards the right in both the top and bottom FM layers,
that is, the corresponding charge currents ow along the  x direction. The skyrmions in both the top and bottom FM layers are driven by the current. In all
the models, the thickness of both the top FM layer, the bottom FM layer and the insulating spacer are equal to 1 nm. The initial state of the top FM layer is
almost spin-up (pointing along z) and that of the bottom FM layer is almost spin-down (pointing along  z). (d) Illustration of a pair of skyrmions, (that
is, the bilayer-skyrmion) in an AFM-coupled nanodisk. (e) Side view of the bilayer-skyrmion along the diameter of d. The colour scale represents the out-ofplane component of the magnetization, which is used throughout this paper.
2

NATURE COMMUNICATIONS | 7:10293 | DOI: 10.1038/ncomms10293 | www.nature.com/naturecommunications

ARTICLE

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10293

The sign of Ainter is negative for the interlayer AFM interaction.

We assume that the spins in the top FM layer are pointing
upward. Then the spins in the bottom FM layer are pointing
downward because of the interlayer AFM exchange coupling.
Bilayer skyrmions. The Hamiltonian equation (1) allows the
excitation of a spin texture carrying the topological number
(the skyrmion number) Qt in each FM layer. Spins swirl
continuously around the core and approach the ground-state
value asymptotically in the skyrmion spin texture. The skyrmion
numbers of the top and bottom FM layers are quantized,
and opposite since all the spins are inverted. We choose the
convention that QT  QB 1.
x
Bottom
Top

0.2 ns

73 nm

Bottom
Top

0.4 ns

Top

0 ns

146 nm

Bottom

Bottom
2

D = 3.5 mJ m

j=

331010

Top

156 nm

Bottom
Top

312 nm

Bottom

Top

0 ns

Bottom
Top

0.22 ns

Bottom

D = 3.5 mJ m2 j = 1,0001010 A m2 2 =  = 0.6

Am

c
800

600

800

CPP geometry
Top
Bottom

600

CIP geometry

v (m s1)

v (m s1)

Bottom

0.2 ns

Bottom

Top

0.22 ns

Top

0 ns

0.4 ns

Top

0 ns

CIP

Coupled
(Ainter=6 pJ m1)

CPP

Decoupled
(Ainter=0.06 pJ m1)

Coupled
(Ainter=6 pJ m1)

A skyrmion is characterized by its topological stability

in each FM layer. Even if the skyrmion spin texture is
deformed continuously, its skyrmion number cannot change
as it is quantized. It can be neither destroyed nor separated
into pieces, that is to say, a skyrmion is topologically
protected.
When the AFM interaction equation (2) is taken into account,
the skyrmions in the two layers are bound. Such a pair of
skyrmions may be called a bilayer-skyrmion. We estimate the
excitation energy based on the Hamiltonian Htotal HT HB
Hinter. By determining numerically the spin prole of each
skyrmion, we compare the energy of one bilayer-skyrmion state
with the energy of the homogeneous state in Fig. 3c as a function
of the DMI. The energy of the bilayer-skyrmion becomes lower
than that of the uniform ground state for D\4 mJ m  2 in a
typical sample, which is the threshold value. When the DMI
strength is much smaller than the threshold value, it is hard to
create a bilayer-skyrmion. On the other hand, when it is much
larger, a deformed bilayer-skyrmion is generated. As a result, it is
preferable to tune the DMI strength around the threshold value to
create an isolated bilayer-skyrmion.

Decoupled
(Ainter=0.06 pJ m1)

exchange stiffness Aintra. The second term represents the DMI

with the DMI vector D. The third term represents the PMA with
the anisotropic constant K. HDDI represents the dipoledipole
interaction. There exists an AFM exchange coupling between the
top and bottom FM layers,
X
Hinter  Ainter
mTi  mBi :
2

Top
Bottom

400

CPP geometry
Top
Bottom
CIP geometry
Top
Bottom

400
Decoupled
200

200

0
0.01

0.1

10

100

1,000

0.01

j (1010 A m2)

0.1

10

100

1,000

j (1010 A m2)

Figure 2 | Current-induced motion of skyrmions in the top and bottom FM layers of an AFM-coupled bilayer nanotrack. Top views of the motion of
skyrmions at selected interlayer exchange coupling constants and times driven by spin currents with (a) the current-perpendicular-to-plane (CPP) injection
geometry and (b) the current-in-plane (CIP) injection geometry. The size of the AFM-coupled bilayer nanotrack is 500  50  3 nm3, where
D 3.5 mJ m  2. The skyrmions are initially created by the MTJ skyrmion injector placed on the top FM layer at x 100 nm. For the CPP case, the spin
current in the bottom FM layer is applied along z but polarized along y. The skyrmion in the bottom FM layer moves along the nanotrack driven by the
spin current, whereas the skyrmion in the top FM layer moves accordingly due to the interlayer AFM exchange coupling. For the CIP case, the skyrmions in
both the top and bottom FM layers are driven by in-plane spin-polarized currents. The skyrmions velocities in the top (open symbols) and bottom (solid
symbols) FM layers as functions of the current density j for (c) the large interlayer AFM exchange stiffness Ainter  6 pJ m  1 and (d) the small interlayer
AFM exchange stiffness Ainter  0.06 pJ m  1. For the CIP case, the square, triangle and star symbols indicate different values of the non-adiabatic torque
constant b of 0.15, 0.3 and 0.6, respectively. The cross symbol denotes the decoupling as well as the destruction of skyrmions in the top and bottom FM
layers because of a large driving current density and a small interlayer AFM exchange coupling, where the velocities are calculated just before the
destruction of skyrmions.
NATURE COMMUNICATIONS | 7:10293 | DOI: 10.1038/ncomms10293 | www.nature.com/naturecommunications

ARTICLE
b

Current on

0.07 ns

2.0 ns
Ainter=0 pJ m1, Ainter=2 pJ m1,
j=3x1012 A m2 j=7x1012 A m2

Interlayer exchange (pJ m1)

0.0
0.5
1.0

0.5

0.5
0.0

Pulse on

0.5
1.0

0.0

0.5

D=4 mJ m2

6
4
2
0
0.3

1.0
t (ns)

1.5

2.0

Top
Bottom

1,870

1.0

Top

Aint

1,860
Pulse off

Ainter=2 pJ m1,
j=7x1012 A m2

0.0

Current off

0.5 ns

Pulse on

10

13

17

Current density (x 1012 A m2)

Etotal (x1019 J)

0.03 ns

Top
Bottom

0.5

Pulse off
1
Ainter=0 pJ m ,
j=3x1012 A m2

1.0
t (ns)

1,880
1,890
Bilayer AFM state

1,900
1,910

1.5

Bilayer skyrmion

Bottom
8

D (mJ m2)

2.0

Interlayer exchange (pJ m1)

mz

0 ns

c
1.0

Pulse off

+1

Pulse off

Top Bottom

Skyrmion number

Top Bottom

Skyrmion number

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10293

D=4 mJ m2

6
4
2
0
Aint

0.3

10

13

17

Current density (x 1012 A m2)

Figure 3 | Creation of skyrmions in the AFM-coupled bilayer nanodisk and time evolution of the skyrmion number. (a) Injection of skyrmions in the
AFM-coupled bilayer nanodisk (D 4 mJ m  2) with/without the interlayer AFM exchange coupling. A 500-ps-long spin-polarized current is injected into
the top FM layer followed by a 1,500-ps-long relaxation. The spin-polarized current injection region in the top FM layer and the corresponding region in the
bottom FM layer are denoted by solid and dash yellow circles, respectively. The interlayer AFM exchange coupling constant Ainter is set as 0 or  2 pJ m  1,
whereas the corresponding interface AFM exchange coupling constant s equals to 0 or  2 mJ m  2. (b) The evolution of the skyrmion number of the top
and bottom FM layers in the nucleation process of skyrmions corresponding to a. (c) Total micromagnetic energy Etotal (including the intralayer exchange,
interlayer exchange, dipoledipole, anisotropy and DMI energy) for a bilayer-skyrmion and the AFM-coupled ground state as a function of the DMI
constant D. Relaxed state of (d) the top FM layer and (e) the bottom FM layer after the injection of a 500-ps-long spin-polarized current for various current
density j and interlayer AFM exchange coupling constant Ainter.

Skyrmion Hall effect. To understand the origin of the SkHE, we

analyse the dynamics of a skyrmion based on the Thiele equation.
The Thiele equation describes the centre-of-mass motion of a
skyrmion without affecting its form in each FM layer3,22,3538.
Suppressing the layer index, the Thiele equation for the
skyrmion driven by the in-plane current (CIP) reads,




3
G v s  v d D bvs  av d rVr 0;
.
(d)
(s)
in each FM layer. Here, v R is the drift velocity; v is the
velocity induced by spin-polarized current; G (0,0,G) is the
gyromagnetic coupling vector, where G 4pQ with the skyrmion
number Q; D represents the dissipative force; V(r) is the
conning potential induced by the sample edges; aC0.1 is
the Gilbert damping constant and b is the strength of the
non-adiabatic torque. The rst term on the left hand side of
equation (3) corresponds to the Magnus force. We consider a
skyrmion moving along the x axis far away from the edge.
s
s
Thus, we set v x 6 0; vy 0 and V 0, to nd


b G2 a  b
us ;
v d

x
a
a G2 a2 D2 x


4
ab
d
s
u :
v y DG 2
G a2 D2 x
Consequently, a skyrmion undergoes a transverse motion
d
= 0 because it carries a non-zero skyrmion number (Ga0).
v y
4

Similarly, the Thiele equation for the skyrmion driven by the

current perpendicular to the plane (CPP) reads36,
Gv d  Dav d jspin 0;

where jspin is the spin current induced by the charge current in

the heavy-metal layer. The rst term on the left hand side of
equation (5) corresponds to the Magnus force. We nd
aD
G
j
vyd 2
j :
6
v d
x 2
2 2 spin
G a2 D2 spin
G
a D 
d
= 0 for Ga0,
A skyrmion undergoes a transverse motion v y
similar to the CIP case.
It is concluded that a moving skyrmion feels the Magnus force
and bends towards the direction perpendicular to the driving
current. The direction of the Magnus force is opposite when the
sign of the skyrmion number Q is opposite. For instance, if a
skyrmion moves along the x direction it will be bent towards
the y (  y) direction in the top (bottom) FM layer.
A skyrmion is protected topologically only when it is far away
from the edge. When the skyrmion touches an edge, the
boundary condition is violated and the topological stability is
lost. The skyrmion number Q is no longer quantized and may
change continuously. We present a typical example of the
annihilation process of the skyrmion because of the SkHE in
Supplementary Figs 1 and 2 and Supplementary Movies 1 and 2.
It is found by micromagnetic simulations that a skyrmion can

NATURE COMMUNICATIONS | 7:10293 | DOI: 10.1038/ncomms10293 | www.nature.com/naturecommunications

ARTICLE

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10293

only travel 2 nm along the nanotrack at a driving current density

of 1  1010 A m  2 when the nanotrack width is reduced to 24 nm
for typical Co|Pt material parameters.
Motion of a bilayer-skyrmion in an AFM-coupled nanotrack.
The skyrmions in the two FM layers of the bilayer system move
simultaneously when they are strongly bound by the interlayer
AFM exchange coupling. As a result, the skyrmion in the top FM
layer follows the motion of the skyrmion in the bottom FM layer
even when the current is injected only into the bottom FM layer.
There is no SkHE for a bilayer-skyrmion with sufcient interlayer
AFM coupling strength. This is conrmed by micromagnetic
simulations in Fig. 2a,b. When the interlayer AFM exchange
coupling is strong enough, two skyrmions are tightly bound and
the Magnus forces acting on the skyrmions between the top and
bottom FM layers are exactly cancelled. Accordingly, the bilayerskyrmion moves in a straight line along the nanotrack for
sufciently strong AFM exchange coupling (see Supplementary
Movies 3 and 4 for details). However, for zero or sufciently small
interlayer AFM exchange coupling, a skyrmion in the top FM
layer moves left-handed and the skyrmion in the bottom FM
layer moves right-handed (see Supplementary Movies 58 for
details). Once the skyrmions in the top and bottom FM layers are
decoupled, the SkHE becomes active, leading to the destruction of
skyrmions in the top FM layer and/or bottom FM layer by
touching the nanotrack edges as shown in the decoupled case of
Fig. 2a,b.
The relation between the magnitude of the injected current and
the bilayer-skyrmion velocity is shown in Fig. 2c,d. There are
some prominent features as shown in Fig. 2c,d. First, the velocity
is proportional to the injected current density, following the jv
relation of a conventional single skyrmion. The CPP scheme has
been proven to be much more efcient to drive the skyrmions
than the CIP method, consistent with the reported results of
isolated skyrmions2,3. Second, for weak coupling case in Fig. 2d
(here Ainter  0.06 pJ m  1), the bilayer-skyrmion still moves as
a whole for small and medium current densities and only
decoupled and destroyed because of the SkHE for large current
density. This is because the SkHE increases as the current
increases, which acts as the repulsive force between two
skyrmions. Third and most important, the AFM-coupled
bilayer-skyrmion can move at a high speed of a few hundreds
to a few thousands metres per second without a transverse shift
with almost the same low-current density threshold feature as a
single skyrmion in typical Co|Pt material. This can be seen from
the low-current regime of Fig. 2c,d (cf. ref. 3) and the comparison
with the single skyrmion as given in Supplementary Tables 1
and 2.
These results demonstrate that the bilayer-skyrmion may
possess multiple advantages such as enhanced jv slope especially
at higher driving current density, small depinning threshold
current density (see Supplementary Tables 1 and 2 for details)
and maximum speed of up to B1,000 m s  1. It is noted that for
slightly different material parameters, similar conclusions remain
as for the Co|Pt system.
Creation of a bilayer-skyrmion by vertical spin current. We
proceed to discuss how to create a bilayer-skyrmion. We propose
two different ways. In the rst method, we inject the spinpolarized current (polarized along  z) perpendicularly into the
central circle region of the top FM layer, as illustrated in Fig. 1a.
We demonstrate how the spin textures develop and evolve in
Fig. 3a. The spins start to ip in both FM layers following the spin
current injection only in the top FM layer. When there is no
interlayer AFM exchange coupling, a skyrmion is formed only in

the top FM layer (see Supplementary Movie 9 for details). By

contrast, a skyrmion is formed accordingly in the bottom FM
layer upon the current injection in the presence of the interlayer
AFM exchange coupling (see Supplementary Movie 10 for
details).
Figure 3b shows the evolution of the skyrmion number
corresponding to Fig. 3a. In the process of creating a bilayerskyrmion, the skyrmion number of both top and bottom FM
layers oscillate at the initial stage for to0.2 ns, and rapidly
increases to 1. The skyrmion numbers remain stable even
when the current is switched off, demonstrating that the bilayerskyrmion is topologically protected. During this process, the spins
in the top and bottom FM layers are always anti-parallel.
A comment is in order. The saturated skyrmion number is not
exactly Q 1 but Q 0.89, because there is a background
skyrmion-number contribution of  0.09 from the tilting edge
spins. Accordingly, the calibrated skyrmion number is Q 0.98,
which is almost unity within numerical accuracy.
We present a nucleation phase diagram of a bilayer-skyrmion
as functions of the current density and the interlayer AFM
exchange coupling in Fig. 3d,e. When the magnitude of the
injected current is strong enough, the bilayer-skyrmion is created.
This is due to the fact that spin ip costs a certain energy. On the
other hand, if the interlayer AFM exchange coupling is too
strong, the bilayer-skyrmion is suppressed because of the fact that
the nucleation eld and the coercivity of the top and bottom FM
layers increase with the interlayer AFM exchange coupling,
leading to a larger threshold current density for nucleation.
Creation of a bilayer-skyrmion from a bilayer-domain wall
pair. A skyrmion can also be created from a domain wall (DW)
pair by using a junction geometry13,14. In this scenario, we rst
create an AFM-coupled DW pair within an AFM-coupled bilayer
nanotrack, where a local CPP injection with  z direction is
applied on the top FM layer. We show how the spins start to ip
in the top FM layer and subsequently in the bottom FM layer
driven by the AFM exchange coupling force in Supplementary
Fig. 3a. Then, the bilayer DW pair is shifted by applying the CPP
injection, as shown in the process from t 50 ps to t 120 ps in
Supplementary Fig. 3b. Here we consider the vertical injection of
a spin current towards z and polarized along y in the bottom
FM layer. The CPP injection moves the bilayer-DW along the
nanotrack. When the bilayer-DW arrives at the junction interface
(t 170 ps), both the end spins of the DW are pinned at the
junction, whereas the central part of the DW continues to move
because of spin-transfer torques (STTs) in the wide part of the
nanotrack. Therefore, the structure is deformed into a curved
shape and a bilayer-skyrmion texture forms at t 190 ps (see
Supplementary Fig. 3 and Supplementary Movie 11 for details).
Consideration for experiments. There are already some experimental ways to host and manipulate DWs in systems similar to
the proposed device in this work. A prototypical example based
on the three-dimensional implementation of the racetrack
memory architecture is the Co|Ni multilayers with strong PMA
separated by an AFM layer (for example, Ru, Cr and so on).
Such devices can be fabricated by atomic layer deposition and
patterned with Electron Beam Lithography and ion milling18,33,39.
Strong PMA is derived from the spin-orbit coupling at the interface
of the heavy metal (for example, Pt) and FM material (Co in this
case). Various material properties such as the total MS and the
interfacial coupling of the synthetic antiferromagnetic layers can be
controlled within the reach of leading-edge material fabrication
and deposition techniques18,33. It is very likely that interfacial DMI
might be present and controlled by atomically engineering the

NATURE COMMUNICATIONS | 7:10293 | DOI: 10.1038/ncomms10293 | www.nature.com/naturecommunications

ARTICLE

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10293

interface and material composition. Therefore, it is promising to

engineer a number of materials and design properties such as
the interfacial DMI (in the range of 0.1B10 mJ m  2) and the
interlayer AFM exchange coupling (  0.1B  10 pJ m  1) to host
bilayer-skyrmions16,18,33,40.
Discussion
The relation between the velocity and the injected current for a
bilayer-skyrmion is the same as that for a monolayer-skyrmion,
as shown in Fig. 2c,d. The critical current density required to
drive the bilayer-skyrmion into motion in the CPP case equals
B0.01  1010 A m  2, at which the speed of the bilayer-skyrmion
is B0.15 m s  1. While in the CIP case, the critical current
required to drive the skyrmion is approximately ten times larger
than that of the CPP case. The critical current density is in good
agreement with the value of a single isolated skyrmion in the
monolayer nanotrack shown in the previous analysis3, indicating
that the interlayer AFM exchange coupling between the top and
bottom skyrmions does not increase the depinning current of the
skyrmions (see Supplementary Tables 1 and 2 for details).
It is also worth noting that the interlayer AFM exchange
coupling does not produce a mass of the bilayer-skyrmion. When
the driving current is suddenly turned off, the bilayer-skyrmion
stops high-speed motion immediately (see Supplementary Movie
12 for details).
We also investigate the case where the DMI constant D is
different between the top and bottom FM layers. It is found that
the results of the current-driven motion of a bilayer-skyrmion do
not change much as the DMI only changes the size of the
skyrmion (see Supplementary Movies 13 and 14 for details). The
massless property and its robustness make the bilayer-skyrmion
an ideal candidate for practical applications.
The maximum velocity of a monolayer-skyrmion is typically
much less than 102 m s  1, limited by the edge conning force
BD2/J (see ref. 41 for details). Remarkably, the maximum
velocity of a bilayer-skyrmion is much larger than a monolayerskyrmion. The bilayer-skyrmion can move along the central line
of the nanotrack at a high speed of a few hundred metres
per second.
In summary, we have presented a solution of inhibiting the
Hall effect of skyrmions without affecting the topological
protection, by exploring a device comprised of antiferromagnetically exchange-coupled bilayer nanodisks and nanotracks.
Compared with the mostly investigated skyrmion in the
monolayer system, the bilayer-skyrmion exhibits distinct characteristics with regard to the current-transport behaviour and
robustness. First, it can move strictly along the direction of the
spin-polarized current ow, which is necessary for ultradense
memory applications. This is in high contrast with the case of the
monolayer-skyrmion, where the skyrmion is destroyed by the
edges of the nanotracks. Second, it will be robust against magnetic
eld perturbations, which might be generated externally or
internally within the device circuitry, as the net magnetic moment
of the bilayer-skyrmion is zero. Our proposal of transporting
skyrmion for arbitrarily long distances at much enhanced velocity
may be useful for versatile applications such as ultradense
memory and ultrafast information processing. Similar ideas can
be extended to a multilayer or a superlattice where the skyrmions
are strongly coupled to realize a better manipulation of skyrmions
in conned geometries.
Methods
LandauLifshitzGilbert (LLG) equation for CPP. We may apply the
spin-polarized CPP injection from the magnetic tunnel junction or the spin
Hall effect in heavy-metal layer3,15,24,36,42. For the CPP case, the spin current is
perpendicular to the lm plane, whereas the charge current can be either in the lm
6

plane such as the spin Hall effect scenario or perpendicular to the lm plane such
as the perpendicular magnetoresistive random access memory (MRAM) where
a perpendicularly magnetized polarizer is incorporated. The dynamics of the
magnetization mi at the lattice site i is governed by the LLG equation. In the case of
the CPP injection, we numerically solve the LLG equation extended into the
following form,
dmi
dmi
 jgjmi Heff
i ami 
dt
dt
jgjumi pmi  jgju0 mi p;

with the layer index suppressed. Here, Heff

i  @Htotal =@mi is the effective
magnetic eld induced by the Hamiltonian Htotal HT HB Hinter; g is the
gyromagnetic ratio; a is the Gilbert damping coefcient originating from the spin
relaxation; u is the STT coefcient; u0 is the eld-like out-of-planeSTT
 coefcient
  j jpj
with m0
and p represents the electron polarization direction. We have u m e2dM
S
0

the vacuum magnetic permittivity, d the lm thickness of FM layer, MS the

saturation magnetization and j the current density. We take  z direction of the
spin polarization for creating the skyrmion, whereas y direction of the spin
polarization for moving the skyrmion. The STT is induced either by injection from
a magnetic tunnel junction polarizer or by the spin Hall effect3,36,42.
LLG equation for CIP. For the CIP case, the current is applied along the nanotrack
axial direction3,36,42. We numerically solve the LLG equation with the addition of
the STT term in the following form,
dmi
dmi
 jgjmi Heff
i ami 
dt
dt
jpja3

jr  rmi
2eMS
jpja3 b

mi jr  rmi ;
2eMS

where b is the strength of the non-adiabatic torque and a is the lattice constant.
Micromagnetic modelling. The three-dimensional micromagnetic simulations are
performed using the well-established Object Oriented MicroMagnetic Framework
developed at the National Institute of Standards and Technology43. Additional
extension module is employed to include the DMI in the simulation2,3,44. The
time-dependent magnetization dynamics is governed by the LLG equation
including both antidamping-like and eld-like STTs43,4548. The average energy
density E is a function of vectorial magnetization M, which includes the intralayer
exchange, the interlayer exchange, the anisotropy, the applied eld (Zeeman),
the demagnetization (dipoledipole) and the DMI energy terms. For
micromagnetic simulations, the standard intrinsic magnetic parameters are
adopted from refs 2,3: Gilbert damping coefcient a 0.3, gyromagnetic ratio
g  2.211  105 m A  1 s  1, saturation magnetization MS 580 kA m  1,
intralayer exchange stiffness Aintra 15 pJ m  1, DMI constant D 0B6 mJ m  2
and PMA K 0.8 MJ m  3 unless otherwise specied. The interlayer exchange
coefcient Ainter is set from 0 to  10 pJ m  1, whereas the corresponding
interface exchange coefcient s equals from 0 to  10 mJ m  2 (s Ainter/1 nm
in our case). The negative value of the constant s denotes that the interface is
antiferromagnetically exchange-coupled. The eld-like out-of-plane STT coefcient
u0 is set to zero, and the Oersted eld is neglected for simplicity. The polarization
rate (|p|) of the spin current used in all simulations is xed at 0.4. All the samples
are discretized into tetragonal volume elements with the size of 2  2  1 nm3 in
the simulation, which is sufciently smaller than the typical exchange length
(B4.3 nm) as well as the skyrmion size, ensuring a balance between numerical
accuracy and computational efciency. For all simulation of current-driven
skyrmions reported throughout this paper, the skyrmions are rst created at the
designed spot of the nanotrack (x 100 nm) by a local spin CPP of the top FM
layer. Then the system is relaxed to an energy minimum state without applying any
current. Next, we start the timer and the spin current is injected into the nanotrack
in the CIP or CPP scheme as shown in Fig. 1. For the conguration of CIP, the
electrons ow towards the right in both the top and bottom FM layers, that is, the
charge currents ow towards the left, whereas for the conguration of CPP, the
electrons ow upward along z only in the bottom FM layer.

References
1. Nagaosa, N. & Tokura, Y. Topological properties and dynamics of magnetic
skyrmions. Nat. Nanotech 8, 899911 (2013).
2. Fert, A., Cros, V. & Sampaio, J. Skyrmions on the track. Nat. Nanotech. 8,
152156 (2013).
3. Sampaio, J., Cros, V., Rohart, S., Thiaville, A. & Fert, A. Nucleation, stability
and current-induced motion of isolated magnetic skyrmions in nanostructures.
Nat. Nanotech. 8, 839844 (2013).
4. Iwasaki, J., Mochizuki, M. & Nagaosa, N. Current-induced skyrmion dynamics
in constricted geometries. Nat. Nanotech. 8, 742747 (2013).

NATURE COMMUNICATIONS | 7:10293 | DOI: 10.1038/ncomms10293 | www.nature.com/naturecommunications

ARTICLE

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms10293

5. Buttner, F. et al. Dynamics and inertia of skyrmionic spin structures. Nat. Phys.
11, 225228 (2015).
6. Romming, N., Kubetzka, A., Hanneken, C., von Bergmann, K. &
Wiesendanger, R. Field-dependent size and shape of single magnetic skyrmion.
Phys. Rev. Lett. 114, 177203 (2015).
7. Jiang, W. et al. Blowing magnetic skyrmion bubbles. Science 349, 283286
(2015).
8. Koshibae, W. et al. Memory functions of magnetic skyrmions. Jpn J. Appl. Phys.
54, 053001 (2015).
9. Muhlbauer, S. et al. Skyrmion lattice in a chiral magnet. Science 323, 915919
(2009).
10. Yu, X. Z. et al. Real-space observation of a two-dimensional skyrmion crystal.
Nature 465, 901904 (2010).
11. Heinze, S. et al. Spontaneous atomic-scale magnetic skyrmion lattice in two
dimensions. Nat. Phys. 7, 713718 (2011).
12. Tchoe, Y. & Han, J. H. Skyrmion generation by current. Phys. Rev. B 85, 174416
(2012).
13. Zhou, Y. & Ezawa, M. A reversible conversion between a skyrmion and a
domain-wall pair in a junction geometry. Nat. Commn. 5, 4652 (2014).
14. Zhang, X., Ezawa, M. & Zhou, Y. Magnetic skyrmion logic gates: conversion,
duplication and merging of skyrmions. Sci. Rep. 5, 9400 (2015).
15. Zhang, X. et al. Magnetic skyrmion transistor: skyrmion motion in a
voltage-gated nanotrack. Sci. Rep. 5, 11369 (2015).
16. Moreau-Luchaire, C. et al. Skyrmions at room temperature: From magnetic
thin lms to magnetic multilayers. Preprint at http://arxiv.org/abs/1502.07853
(2015).
17. Woo, S. et al. Observation of room temperature magnetic skyrmions and their
current-driven dynamics in ultrathin Co lms. Preprint at http://arxiv.org/abs/
1502.07376 (2015).
18. Parkin, S. & Yang, S. Memory on the racetrack. Nat. Nanotech. 10, 195198
(2015).
19. Bogdanov, A. N. & Yablonskii, D. A. Thermodynamically stable vortices in
magnetically ordered crystals. The mixed state of magnets. Sov. Phys. JETP 68,
101103 (1989).
20. Bogdanov, A. & Hubert, A. Thermodynamically stable magnetic vortex states in
magnetic crystals. J. Magn. Magn. Mater. 138, 255269 (1994).
21. Roessler, U. K., Bogdanov, A. N. & Peiderer, C. Spontaneous skyrmion ground
states in magnetic metals. Nature 442, 797801 (2006).
22. Iwasaki, J., Mochizuki, M. & Nagaosa, N. Universal current-velocity relation of
skyrmion motion in chiral magnets. Nat. Commun. 4, 1463 (2012).
23. Purnama, I., Gan, W. L., Wong, D. W. & Lew, W. S. Guided current-induced
skyrmion motion in 1D potential well. Sci. Rep. 5, 10620 (2015).
24. Zhang, X. et al. Skyrmion-skyrmion and skyrmion-edge repulsions in
skyrmion-based racetrack memory. Sci. Rep. 5, 7643 (2015).
25. Iwasaki, J., Beekman, A. J. & Nagaosa, N. Theory of magnon-skyrmion
scattering in chiral magnets. Phys. Rev. B 89, 064412 (2014).
26. Schutte, C. & Garst, M. Magnon-skyrmion scattering in chiral magnets. Phys.
Rev. B 90, 094423 (2014).
27. Zhang, X. et al. All-magnetic control of skyrmions in nanowires by a spin wave.
Nanotechnology 26, 225701 (2015).
28. Schwarze, T. et al. Universal helimagnon and skyrmion excitations in metallic,
semiconducting and insulating chiral magnets. Nat. Mat. 14, 478483 (2015).
29. Schroeter, S. & Garst, M. Scattering of high-energy magnons off a magnetic
skyrmion. Low Temp. Phys. 41, 817825 (2015).
30. Kong, L. & Zang, J. Dynamics of an insulating skyrmion under a temperature
gradient. Phys. Rev. Lett. 111, 067203 (2013).
31. Lin, S.-Z., Batista, C. D., Reichhardt, C. & Saxena, A. AC current generation in
chiral magnetic insulators and skyrmion motion induced by the spin Seebeck
effect. Phys. Rev. Lett. 112, 187203 (2014).
32. Chumak, A. V., Vasyuchka, V. I., Serga, A. A. & Hillebrands, B. Magnon
spintronics. Nat. Phys. 11, 453461 (2015).
33. Yang, S. H., Ryu, K. S. & Parkin, S. Domain-wall velocities of up to 750 m s  1
driven by exchange-coupling torque in synthetic antiferromagnets. Nat.
Nanotech. 10, 221226 (2015).
34. Gilbert, D. A. et al. A new reversal mode in exchange coupled
antiferromagnetic/ferromagnetic disks: distorted viscous vortex. Nanoscale 7,
98789885 (2015).
35. Thiele, A. A. Steady-state motion of magnetic domains. Phys. Rev. Lett. 30,
230233 (1973).

36. Tomasello, R. et al. A strategy for the design of skyrmion racetrack memories.
Sci. Rep. 4, 6784 (2014).
37. Everschor, K. et al. Rotating skyrmion lattices by spin torques and eld or
temperature gradients. Phys. Rev. B 86, 054432 (2012).
38. Schulz, T. et al. Emergent electrodynamics of skyrmions in a chiral magnet.
Nat. Phys. 8, 301304 (2012).
39. Purnama, I., Kerk, I. S., Lim, G. J. & Lew, W. S. Coupled Neel domain wall
motion in sandwiched perpendicular magnetic anisotropy nanowires. Sci. Rep.
5, 8754 (2015).
40. Yang, H., Thiaville, A., Rohart, S., Fert, A. & Chshiev, M. Anatomy of
Dzyaloshinskii-Moriya interaction at Co/Pt interfaces. Preprint at
http://arxiv.org/abs/1501.05511 (2015).
41. Iwasaki, J., Koshibae, W. & Nagaosa, N. Colossal spin transfer torque effect on
skyrmion along the edge. Nano Lett. 14, 44324437 (2014).
42. Khvalkovskiy, A. V. et al. Matching domain-wall conguration and
spin-orbit torques for efcient domain-wall motion. Phys. Rev. B 87, 020402(R)
(2013).
43. Donahue, M. J. & Porter, D. G. OOMMF users guide, version 1.0. Interagency
Report NISTIR 6376 (1999).
44. Rohart, S. & Thiaville, A. Skyrmion connement in ultrathin lm
nanostructures in the presence of Dzyaloshinskii-Moriya interaction. Phys. Rev.
B 88, 184422 (2013).
45. Gilbert, T. L. A Lagrangian formulation of the gyromagnetic equation of the
magnetization eld. Phys. Rev. 100, 1243 (1955).
46. Landau, L. & Lifshitz, E. On the theory of the dispersion of magnetic
permeability in ferromagnetic bodies. Physik. Z. Sowjetunion 8, 153169
(1935).
47. Thiaville, A., Rohart, S., Jue, E., Cros, V. & Fert, A. Dynamics of
Dzyaloshinskii domain walls in ultrathin magnetic lms. Europhys. Lett. 100,
57002 (2012).
48. Thiaville, A., Nakatani, Y., Miltat, J. & Suzuki, Y. Micromagnetic understanding
of current-driven domain wall motion in patterned nanowires. Europhys. Lett.
69, 990 (2005).

Acknowledgements
Y.Z. acknowledges the support by National Natural Science Foundation of China
(Project No. 1157040329), the Seed Funding Program for Basic Research and Seed
Funding Program for Applied Research from the HKU, ITF Tier 3 funding (ITS/171/13,
ITS/203/14), the RGC-GRF under Grant HKU 17210014 and University Grants
Committee of Hong Kong (Contract No. AoE/P-04/08). M.E. acknowledges the support
by the Grants-in-Aid for KAKENHI No. 25400317 and No. 15H05854. M.E. is very
much grateful to N. Nagaosa for many helpful discussions on the subject. X.Z. thanks
J. Xia for useful discussions. Y.Z. thanks Professor F.C. Zhang for many discussions on
skyrmions.

Author contributions
M.E. conceived the idea and designed the project. Y.Z. coordinated and supervised the
work. X.Z. performed the numerical simulations and M.E. carried out the theoretical
analysis. All authors discussed the results and prepared the manuscript.

Additional information
Supplementary Information accompanies this paper at http://www.nature.com/
naturecommunications
Competing nancial interests: The authors declare no competing nancial interests.
Reprints and permission information is available online at http://npg.nature.com/
reprintsandpermissions/
How to cite this article: Zhang, X. et al. Magnetic bilayer-skyrmions without skyrmion
Hall effect. Nat. Commun. 7:10293 doi: 10.1038/ncomms10293 (2016).
This work is licensed under a Creative Commons Attribution 4.0
International License. The images or other third party material in this
article are included in the articles Creative Commons license, unless indicated otherwise
in the credit line; if the material is not included under the Creative Commons license,
users will need to obtain permission from the license holder to reproduce the material.
To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

NATURE COMMUNICATIONS | 7:10293 | DOI: 10.1038/ncomms10293 | www.nature.com/naturecommunications