ARTICLES
PUBLISHED ONLINE: 11 MAY 2014 | DOI: 10.1038/NNANO.2014.94
Switching of perpendicular magnetization by
spin–orbit torques in the absence of external
magnetic fields
Guoqiang Yu1† *, Pramey Upadhyaya1†, Yabin Fan1, Juan G. Alzate1, Wanjun Jiang1, Kin L. Wong1,
So Takei2, Scott A. Bender2, Li-Te Chang1, Ying Jiang3, Murong Lang1, Jianshi Tang1, Yong Wang3,
Yaroslav Tserkovnyak2, Pedram Khalili Amiri1 and Kang L. Wang1 *
Magnetization switching by current-induced spin–orbit torques is of great interest due to its potential applications in
ultralow-power memory and logic devices. The switching of ferromagnets with perpendicular magnetization is of particular
technological relevance. However, in such materials, the presence of an in-plane external magnetic field is typically
required to assist spin–orbit torque-driven switching and this is an obstacle for practical applications. Here, we report the
switching of out-of-plane magnetized Ta/Co20Fe60B20/TaOx structures by spin–orbit torques driven by in-plane currents,
without the need for any external magnetic fields. This is achieved by introducing a lateral structural asymmetry into our
devices, which gives rise to a new field-like spin–orbit torque when in-plane current flows in these structures. The
direction of the current-induced effective field corresponding to this field-like spin–orbit torque is out-of-plane, facilitating
the switching of perpendicular magnets.
T
he breaking of structural symmetries in nanomagnetic systems
is of great interest for the development of ultralow-power
spintronic devices. Structural asymmetry in various magnetic
heterostructures has been engineered to reveal novel interactions
between electric currents and magnetization, resulting in spin–
orbit torques (SOTs) on the magnetization1–6 that are both fundamentally and technologically important. In recent experiments,
such SOTs have been used to realize current-induced magnetization
switching2–4,7 and domain-wall motion8–10. Typical heterostructures
exhibiting SOTs consist of a ferromagnet (F), on one side of which is
a heavy non-magnetic metal (NM) with strong spin–orbit coupling
and on the other an insulator (I) (shown schematically in Fig. 1a,
breaking mirror symmetry in the growth direction). In terms of applications, the use of SOTs in NM/F/I structures allows for a significantly lower write current compared to regular spin-transfer-torque
(STT) devices4, improving energy efficiency and scalability1–5,11
in SOT-based magnetic random access memory (SOT-MRAM)
when compared to STT-MRAM.
The ability to perform SOT-induced switching without using
external magnetic fields, in particular for perpendicularly magnetized ferromagnets (which are more scalable than in-plane
ones12), is important for applications. However, there are currently
no practical solutions to achieve this. In NM/F/I heterostructures
studied to date, the form of the resultant current-induced SOT
alone does not allow for deterministic switching of perpendicular
ferromagnets; the application of an additional external in-plane
magnetic field is also required2–4. (This is a very general feature of
SOT devices, which can be explained by symmetry-based arguments, as discussed in the following.) In such experiments, the
field allows each current direction to favour a particular orientation
for the out-of-plane magnetization component, resulting in
deterministic perpendicular switching. Although this in-plane
external field may in principle be integrated into the device by
engineering the material stack, it is undesirable from a practical
point of view. For memory applications, in particular, it reduces
the thermal stability of the perpendicular bit by lowering the
energy barrier between its stable states, reducing the retention time.
In this Article, we present a NM/F/I structure that exhibits zerofield current-induced switching of perpendicular magnetization.
Our device consists of a stack of Ta/Co20Fe60B20/TaOx layers, with
structural mirror asymmetry along the in-plane direction. The
lateral asymmetry, in effect, replaces the role of the external
in-plane magnetic field. We present experimental results on
current-induced SOT switching of perpendicular magnetization
without applied magnetic fields. We also present a symmetrybased analysis of SOT interactions and show that this type of
bias-field-free switching originates from lateral symmetry-breaking,
giving rise to a new field-like torque upon application of an in-plane
current. The results add to the understanding of SOTs and their
connection to symmetry-breaking and provide a potential
pathway towards the design of practical perpendicular SOT devices.
Symmetry-based analysis of current-induced SOTs
The current-induced SOT terms, which are physically allowed for a
particular device structure, can be determined based on their symmetry properties. Hence, symmetry arguments provide a tool for
designing magnetic material and device structures to realize particular switching characteristics. Figure 1 schematically illustrates how
lateral symmetry-breaking can give rise to current-induced perpendicular magnetization switching. The z axis points along the growth
direction, and the current is applied along the x axis. The figure
depicts the scenarios described in the following sections.
1
Department of Electrical Engineering, University of California, Los Angeles, California 90095, USA, 2 Department of Physics and Astronomy, University of
California, Los Angeles, California 90095, USA, 3 Center for Electron Microscopy and State Key Laboratory of Silicon Materials, Department of Materials
Science and Engineering, Zhejiang University, Hangzhou, 310027, China. † These authors contributed equally to this work. * e-mail: guoqiangyu@ucla.edu;
wang@seas.ucla.edu
NATURE NANOTECHNOLOGY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturenanotechnology
© 2014 Macmillan Publishers Limited. All rights reserved.
1
ARTICLES
NATURE NANOTECHNOLOGY
DOI: 10.1038/NNANO.2014.94
z
a
b
y
x
H´ap
´
HFL
y
´
HDL
y
´
HFL
y
´
HDL
y
HDL
y
m´
HDL
y
m´
HFL
y
HFL
y
Hap
J´
J´
m
m
J
J
c
d
HzFL´
J
HzFL
J
´
HFL
y
´
HDL
y
HzFL
HzFL
E
HDL
y
m´
E
HFL
y
J´
m
Over oxidation
Optimum oxidation
J
Under oxidation
Figure 1 | Schematics of mirror symmetry and current-induced effective fields corresponding to SOTs. a, Effective fields induced by current, and absence of
deterministic out-of-plane switching, in a perpendicular magnetic structure that is symmetric with respect to the x–z plane. The typical stack structure is nonmagnetic metal, ferromagnetic metal and oxide from bottom to top. The same direction of current allows for states with magnetization pointing up as well as
DL
FL
DL
down. The in-plane effective fields (HFL
y and Hy ) are permitted due to the structural inversion asymmetry along the z axis. Blue arrows indicate Hy and Hy and
′
DL′
their mirror reflections HFL
and
H
with
respect
to
the
x–z
plane.
It
is
important
to
note
the
difference
between
an
external
field
H
and
the
damping-like
field
y
y
ap
HDL
y ; the latter does not break the mirror symmetry with respect to the x–z plane as it depends on m and changes sign when magnetization is reversed, while the
former does break the aforementioned symmetry (see b). b, Symmetry with respect to the x–z plane is broken by applying an external magnetic field (Hap) along
the x direction. By fixing the direction of Hap (either positive or negative with respect to the x axis), a unique perpendicular magnetization state can be chosen for
FL′
each current direction. c, The new perpendicular effective field (HFL
z ) induced by the laterally asymmetric structure, and its mirror image (Hz ). The presence of
the new perpendicular effective field, induced by the lateral symmetry-breaking, uniquely determines the z component of the magnetization for a particular
direction of current, thereby allowing deterministic switching without external magnetic fields. d, Schematic representation of the ferromagnet/oxide interface,
illustrating its non-uniform oxygen content. The resulting non-uniform charge distribution may produce in-plane electric fields (E) along the interface, which in
turn can contribute to the observed HFL
z . Red spheres indicate oxygen atoms and perpendicular pink arrows correspond to the PMA in the magnetic layer.
Mirror symmetry-breaking along z axis only
Figure 1a depicts the case where mirror symmetry is broken only along
the z axis (that is, preserving mirror symmetries along the x and
y axes, similar to previous works2–4,7) in the absence of external
magnetic fields. In this case, the symmetry-breaking results in
current-induced SOTs, which, to quadratic order in m, consist of a
FL
field-like (FL) term TFL
c = gHy m × y and a damping-like (DL)
DL
=
g
H
m
×
m
×
y
(ref. 13). Here, m denotes a unit
term TDL
c
y
vector along the magnetization and g is the gyromagnetic ratio.
Equivalently, these torques can be expressed in terms of effective magDL
FL
DL
netic fields, namely HFL
y = Hy y and Hy = Hy m × y (depicted in
FL
DL
Fig. 1), where Hy and Hy represent the current-dependent proportionality constants for each term. Within simple microscopic
models14,15, such SOTs can be derived from the interfacial Rashba
2
interaction16 and/or the spin Hall effect in normal metals17–20. In
this case, the absence of deterministic perpendicular switching can
be understood by performing a mirror reflection with respect to the
x–z plane on the state with perpendicular magnetization component
Mz . 0, as illustrated in Fig. 1a. Under this transformation, the direction of the current density J is unaltered. However, magnetization
(being a pseudo-vector) reverses its components parallel to the x–z
plane, resulting in Mz , 0 in the mirror state. Consequently, if a particular current direction allows an equilibrium magnetization state
with Mz . 0, the same direction of current should also favour a
state with Mz , 0. Therefore, a given current direction does not
favour a unique perpendicular magnetization orientation, and no
deterministic switching is obtained. This argument suggests that, to
achieve current-induced switching of perpendicular magnetization,
NATURE NANOTECHNOLOGY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturenanotechnology
© 2014 Macmillan Publishers Limited. All rights reserved.
NATURE NANOTECHNOLOGY
ARTICLES
DOI: 10.1038/NNANO.2014.94
a
b
z
Ta
CoFeB
Ta
z
y
x
y
x
+I
Oxidation
θ = 90°
TaOx
CoFeB
Ta
I
−V
+V
Patterning
V
−I
tTa (nm)
c
1.6
1.7
d
1.8
1.9
2.0
1.8
15
1.5
12
RHall (normalized)
Hk (kOe)
1.5
1.2
0.9
0.6
tTa = 1.56 nm
1.65 nm
9
1.70 nm
6
1.82 nm
3
1.94 nm
1.99 nm
0
0.3
0
5
10
Position (mm)
15
20
−100
−50
0
Hz (Oe)
50
100
Figure 2 | Device geometry and magnetic perpendicular anisotropy. a, Procedure for growth and patterning of the devices. The sample structure is
Ta(5)/Co20Fe60B20/TaOx (wedge) (thickness in nanometres). The Ta layer on top of the CoFeB film was deposited with varying thickness across the wafer,
resulting in a wedge shape. The thickness of the top Ta layer before oxidation varies from 0.81 to 2.13 nm. After O2/Ar plasma oxidation and annealing, nonuniform PMA was realized. White arrows show the magnetization of Co20Fe60B20 , and dots in the TaOx schematically show the distribution of oxygen atoms.
b, Structure of one device in the array (scale bar, 10 mm) and the measurement configuration. Each individual device is designed to have a lateral asymmetry
due to the wedge in the TaOx. u is the angle between the Hall bar and the wedge direction (y axis). c, Effective perpendicular anisotropy field (Hk ) as a
function of position/thickness of the devices at room temperature. Owing to the non-uniform oxidation at the interface, which depends on the thickness of
the initially deposited Ta, a non-monotonic distribution of Hk is obtained. Note that the variation of Hk with respect to position (dHk/dy) can be positive or
negative, depending on the device location along the wedge. d, Perpendicular magnetization of Ta/CoFeB/TaOx measured by EHE at different locations along
the wedge. A good correlation is found between the out-of-plane coercivity and Hk , obtaining a maximum coercivity near the peak of the Hk distribution.
mirror symmetry with respect to the x–z plane also has to be broken.
This was achieved in previous works using an external magnetic field
Hap (refs 3,7) along the current direction, as depicted in Fig. 1b. The
mirror transformation in this case also reverses the external magnetic
field. Thus, by fixing the external field direction along the current, the
symmetry between magnetic states with opposite Mz components is
broken, allowing for a unique magnetic state. (For the case shown, a
positive/negative external field favours the state with positive/negative
Mz.) Formally, this scenario has also been explained by solving for the
equilibrium magnetization orientation in the presence of SOTs within
a single-domain model3.
asymmetry along the y axis consists of a varying thickness of the
insulating layer. The mirror transformation in this case reverses
the directions of both Mz and J (with respect to the thickness
wedge), hence associating each current direction with a unique Mz
orientation. Hence, breaking lateral structural inversion symmetry
can replace the role of the external bias field. This is also reflected
in the form of the allowed current-induced SOT terms. Using a
symmetry-based phenomenology21, the current-induced SOT terms
(up to quadratic order in m) due to mirror asymmetry along both
the y- and z-axes are given by
TSOT = gHyFL m × y + gHyDL m × m × y + gHzFL m × z
Mirror symmetry-breaking along both z and y axes
When the mirror symmetry along the y axis is also broken, a
particular current direction can uniquely determine the Mz
components. This is illustrated in Fig. 1c, where the structural
+ gHzDL m × m × z
(1)
Neglected in equation (1) are the symmetry-allowed dissipative
torques (inexpressible as DL terms) that are not essential to this
NATURE NANOTECHNOLOGY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturenanotechnology
© 2014 Macmillan Publishers Limited. All rights reserved.
3
NATURE NANOTECHNOLOGY
a
0
−80
−40
0
40
Hz (Oe)
3
0
80
−80
−40
0
40
Hz (Oe)
80
−80
−40
0
40
Hz (Oe)
80
−80
−40
0
40
Hz (Oe)
80
e
3
3
RHall (Ω)
RHall (Ω)
b
0
0
−3
−3
−80
−40
0
40
Hz (Oe)
80
c
f
RHall (Ω)
3
RHall (Ω)
Experiments were carried out on sputter-deposited Ta(5.0 nm)/
Co20Fe60B20(1.0 nm)/TaOx(wedge) films. Details of sample preparation are presented in the Methods. The oxide thickness and
oxygen content at the Co20Fe60B20/TaOx interface were changed
continuously across the wafer. The film was patterned into an
array of Hall bars (shown in Fig. 2a,b), with the transverse direction
(the y axis) along the TaOx wedge (Fig. 2a), breaking the mirror
symmetry with respect to the x–z plane. Thus, application of a
current along the Hall bars is expected to produce an out-of-plane
effective magnetic field, based on the symmetry arguments
already discussed.
The devices were characterized using extraordinary Hall effect
(EHE) measurements, as shown in Fig. 2b. The effective perpendicular anisotropy field (Hk) of the Co20Fe60B20 layer was determined using EHE measurements as a function of an in-plane
magnetic field. Figure 2c shows the measured Hk as a function of
position along the TaOx gradient. It shows a non-monotonic dependence of perpendicular magnetic anisotropy (PMA) on position,
indicating an increase on the thinner side of the wedge
(dHk/dy . 0) and a decrease on the thicker side (dHk/dy , 0).
The perpendicular magnetization, measured as a function of perpendicular magnetic field, is shown in Fig. 2d. For devices located
on the central region of the wedge with largest Hk , the loops are
square-shaped with a large coercivity. As expected, they become
less square-shaped and eventually turn into hard-axis-like loops
on both sides of the wedge where Hk is smaller. The observed
PMA is due to interfacial magnetic anisotropy between the
Co20Fe60B20 film and its adjacent TaOx and Ta layers, similar to
recent reports in other material systems22–24. The anisotropy associated with the TaOx interface is in turn affected by the appearance of
Fe–O and Co–O bonds at the interface23,24, exhibiting a non-monotonic dependence on oxygen content25,26. As a result, the change of
PMA across the wedge reflects the oxygen concentration gradient at
the Co20Fe60B20/TaOx interface across the wafer.
We next performed EHE measurements for different direct currents applied along the x axis. Figure 3a–c shows the results for one
device (device A, tTa ¼ 1.65 nm before oxidation) in the dHk/dy . 0
region. As expected, small currents have almost no influence on the
switching behaviour (Fig. 3a). At larger currents, the centres of the
hysteresis loops are shifted to the left for positive currents, indicating the presence of a perpendicular current-induced effective field
FL
FL
HFL
z = Hz z. The value of Hz can be extracted from the positive
+
negative
(HS− ) switching fields, that is,
(HS ) and
HzFL = − HS+ + HS− /2. For currents in the opposite direction, the
loops are shifted to the right. At I ¼+10 mA, the separation
between the two loops for this device (Fig. 3c) is
HzFL (10 mA) 2 HzFL (210 mA)≈ 22 Oe. It is interesting to compare
this to the measurements shown in Fig. 3d–f, which show the
EHE signal for a different device (device B, tTa ¼ 1.94 nm before
oxidation) in the dHk/dy , 0 region (that is, on the opposite side
of the anisotropy peak in Fig. 2c). In this case, the sign of the
current-induced field is opposite to device A for the same direction
of current. The separation between the two loops at I ¼+10 mA is
HzFL (10 mA)–HzFL (210 mA) ≈ 266 Oe for device B. Thus, by comparing the HzFL of these two devices, it is evident that dHzFL /dI . 0 in
the dHk/dy . 0 region, while dHzFL /dI , 0 in the dHk/dy , 0 region.
DOI: 10.1038/NNANO.2014.94
−3
−3
Experiment
4
d
3
RHall (Ω)
experiment (Supplementary Section 1). In the last two terms of
equation (1), HzFL and HzDL parameterize the strengths of the
current-induced effective fields arising from the additional
asymmetry, representing the new FL and DL SOT terms, respectively. The new FL term gives rise to a current-induced effective
field HzFL along the z axis (depicted in Fig. 1c) and can thus
facilitate current-induced deterministic switching of perpendicular
magnetization without external magnetic fields.
RHall (Ω)
ARTICLES
0
3
0
−3
−3
−80
−40
0
40
Hz (Oe)
80
Figure 3 | Effect of HFL
z induced by current at room temperature.
a–f, Perpendicular magnetization of Ta/CoFeB/TaOx , measured by EHE, while
currents of +1 mA (a,d), +6 mA (b,e) and +10 mA (c,f) are applied to the
devices. Red circles (black squares) indicate positive (negative) current
values. Results for a typical device (tTa ¼ 1.65 nm before oxidation, device A)
in the dHk/dy . 0 region are shown in a–c, and results for a typical device
(tTa ¼ 1.94 nm before oxidation, device B) in the dHk/dy , 0 region are
shown in d–f. The shift directions of the EHE loops with respect to current,
which reflect the directions of HFL
z induced by the current, are opposite for
these two devices with opposite signs of dHk/dy.
To quantify the HzFL induced by current, the values of HS+ and
HS− for the two devices are summarized in Fig. 4a,b for different
applied currents. The current-induced perpendicular field can be
obtained by
fitting
the
current
dependence
of
HzFL = − HS+ + HS− /2. Here, Fig. 4a corresponds to device A
(dHk/dy . 0) and Fig. 4b corresponds to device B (dHk/dy , 0).
For both cases, the resultant HzFL can be fitted well to a linear
curve, and can be expressed as HzFL ¼ bJ, where J is the applied
current density. The values of b, extracted similarly for all devices
measured along the wedge, are shown in Fig. 4c, with the largest
absolute value reaching 56 Oe per 1011 A m22. The plot also
shows dHk/dy as a function of position for comparison. It can be
seen that both the sign and magnitude of b, and hence HzFL , correspond well with dHk/dy. Thus, in addition to the sign and magnitude of the applied current, the current-induced HzFL also depends
on the sign and magnitude of dHk/dy. This, in turn, establishes
the correlation of HzFL with symmetry-breaking along the y axis,
confirming the central hypothesis of this work.
To further verify the role of symmetry-breaking, we conducted
additional experiments on samples with a uniform TaOx layer. As
expected, these samples showed a vanishingly small perpendicular
effective field. In addition, the dependence of b on the angle
NATURE NANOTECHNOLOGY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturenanotechnology
© 2014 Macmillan Publishers Limited. All rights reserved.
NATURE NANOTECHNOLOGY
a
40
ARTICLES
DOI: 10.1038/NNANO.2014.94
c
b 40
0.2
0
−20
20
β (Oe per 1011 Am−2)
Switching field (Oe)
20
0
−20
0
0
−20
−40
dH k /dy (Oe μm−1)
Switching field (Oe)
20
−0.2
−60
−40
−0.4
−40
−1
0
1
J (1011 Am−2)
−1
0
1
J (1011 Am−2)
0
5
10
15
Position (mm)
20
Figure 4 | Switching fields as a function of applied current densities, and b as a function of position of the Hall bar devices on the wafer. a,b, Measured
switching fields as a function of current for devices A (a) and B (b) (discussed in Fig. 3), with dHk/dy ¼ 0.09 Oe mm21 and 20.24 Oe mm21, respectively.
Blue lines are linear fits. The values of b, representing the perpendicular effective field (HFL
z ¼ bJ), are extracted from the slope, and are 12.7 Oe per
1011 A m22 and 235.4 Oe per 1011 A m22 for devices A and B, respectively. The red inverted triangles (black triangles) represent positive (negative) current
values. c, Red circles correspond to b and black squares correspond to dHk/dy, indicating a clear correlation between dHk/dy and HFL
z . This is in agreement
with the expected relationship between lateral symmetry-breaking and the strength of HFL
.
The
dashed
line
corresponds
to
zero
for
both sides of the
z
vertical coordinates.
between the wedge ( y axis) and the current flow direction was
studied (Supplementary Section 3). Both experiments further
confirm the central hypothesis of this work.
Discussion of physical mechanisms
The symmetry argument does not provide any details about the
microscopic mechanisms behind the current-induced effective
fields seen in the hysteresis loops. In principle, in addition to new
spin–orbit terms due to lateral symmetry-breaking, at least two
other mechanisms may also contribute to the observed results.
These are (1) current-induced magnetic (Oersted) fields and (2)
spin–orbit terms due to symmetry-breaking along the growth direction, in conjunction with the anisotropy gradient. However, these
were ruled out in our devices, as will be explained next.
The non-uniform oxidation of the TaOx layer may cause the
current density to be non-uniform along the width of our devices.
Hence, assuming there will be a larger current density on the less
oxidized side of the Hall bar, this asymmetry will produce a net perpendicular magnetic field within the Hall bar area. Considering that
the less oxidized part has a thicker metallic film and hence smaller
resistance, for a particular current direction, this Oersted field would
be expected to point in the same direction for all Hall bars, because
they all have an identical direction of the TaOx thickness gradient.
Therefore, based on the sample structure shown in Fig. 2a, the
Oersted field should result in a negative sign of b for all devices,
which is not the case in our experiments. This indicates that
Oersted fields are not the origin of the observed HzFL . In addition,
an estimate of the Oersted field induced by current in our structures
(Supplementary Section 4) indicates that its value is 16 times
smaller than the largest current-induced HzFL observed in
our experiments.
It should also be noted that, considering the size of the present
samples and their relatively low perpendicular anisotropy, magnetization reversal occurs via a micromagnetic process, such as nucleation and propagation of domain walls. The symmetry-based
argument presented earlier is also valid in the case of such micromagnetic reversal processes. For example, it has been shown recently
in domain-wall-mediated SOT-induced switching experiments that,
when lateral symmetry is broken by an applied in-plane magnetic
field fixing the magnetization in the wall, conventional SOTs can
result in deterministic switching27. In the present work no such
magnetic field is present; however, the lateral symmetry is broken
by the anisotropy gradient. Thus, in principle, it is conceivable
that the presence of lateral symmetry-breaking due to an anisotropy
gradient, in conjunction with conventional SOTs, might result in
the observed switching via a non-trivial nucleation process. We
ruled out this possibility by directly measuring the SOTs through
independent second-harmonic measurements5,13,28 in the presence
of large external magnetic fields. The applied field was larger than
the saturation field of our samples, ensuring single-domain behaviour during the experiment. For all devices measured, the first-harmonic signal could be fitted well using a single-domain model
(Supplementary Section 5). The results showed excellent agreement
between the current-induced HzFL extracted from second-harmonic
measurements (that is, where the samples were single domain) and
values extracted from hysteresis loop shifts (Supplementary Fig. 10).
Based on this agreement of the two independent measurements in
all tested samples (Supplementary Fig. 10), we believe that a new
spin–orbit field-like torque due to lateral symmetry-breaking is
the most likely origin of the present observations.
Microscopically, the new field-like torque appears to stem from the
lateral oxidation gradient at the Co20Fe60B20/TaOx interface, which can
induce Rashba-like spin–orbit coupling with the effective electric field
pointing along the wedging direction y. Namely,
a microscopic elec
tron Hamiltonian of the form H s · y × p (ref. 16) at the
Co20Fe60B20/TaOx interface could, in principle, account
for a fieldlike torque of the form TFL m × HFL
z m × y × J . Here,
s and p represent Pauli matrices and the electron’s momentum operator, respectively. As shown in Fig. 1d, an electric field along the
wedging direction could, in turn, originate from the redistribution of
charges near the interface depending on oxygen content, which is
also responsible for the non-monotonic Hk dependence on position26.
However, a more detailed understanding of these electric fields
and their possible contribution to HzFL is needed and requires
first-principles calculations.
We further also compared the magnitude of HzFL with the regular
effective field HyFL , which results from the inversion asymmetry
along the z axis1,5,9,29. The largest magnitude of HzFL was
4.7 Oe mA21 (¼ 56 Oe per 1011 A m22). The HyFL measured by
the second-harmonic method, on the other hand, was 170 Oe per
1011 A m22 for the same device (Supplementary Section 6). Thus,
HzFL is sizable compared to the regular (in-plane) field-like term.
Current-induced switching in the absence of external fields
Finally, we demonstrate that the perpendicular effective field HzFL ,
induced by an in-plane x-axis current, can be used to
NATURE NANOTECHNOLOGY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturenanotechnology
© 2014 Macmillan Publishers Limited. All rights reserved.
5
ARTICLES
a
NATURE NANOTECHNOLOGY
For device applications, the current-induced HzFL can be used in
three-terminal structures, where the perpendicular free layer is part
of a magnetic tunnel junction. It should also be noted that the
non-uniform oxidation method used to create lateral asymmetry
in this work is not the only approach that could be used for this
purpose. For integration into large arrays (for example, memory
chips), a more localized method of generating lateral asymmetry
may be more appropriate. We hope this work will motivate
research to develop such devices that make use of perpendicular
effective fields. In combination with conventional SOTs, these
could result in ultralow-power and scalable spintronic memory
and logic circuits.
4
RHall (Ω)
2
0
−2
−4
−15
−10
−5
0
5
10
15
I (mA)
b
DOI: 10.1038/NNANO.2014.94
4
Methods
RHall (Ω)
2
0
−2
−4
−15
−10
−5
0
5
10
15
I (mA)
Figure 5 | Switching of perpendicular magnetization by current in the
absence of external magnetic fields. a,b, Hall resistance as a function of
direct current for devices C (tTa ¼ 1.67 nm before oxidation) (a) and B
(tTa ¼ 1.94 nm before oxidation) (b) with dHk/dy ¼ 0.10 and
20.24 Oe mm21, respectively. No external magnetic field is applied. The
favoured magnetization direction corresponding to each current direction is
opposite for these two devices, due to the opposite orientations of the
current-induced HFL
z (that is, different signs of b ). All measurements were
carried out at room temperature.
deterministically switch perpendicular magnetization without the
assistance of external magnetic fields. This is shown in Fig. 5a,b
for two representative devices on different sides of the Hk peak in
Fig. 2c. We were able to reversibly switch the perpendicular
magnetization with currents of 6 mA (¼ 5.0 × 106 A cm22)
for device C (tTa ¼ 1.67 nm before oxidation; Fig. 5a) and 3 mA
(¼ 2.5 ×106 A cm22) for device B (Fig. 5b). For the first device
(dHk/dy . 0), positive currents favour a positive magnetization
(resulting in a negative Hall resistance RHall ), whereas for the
latter device (dHk/dy , 0) they favour a negative magnetization.
The favoured direction of magnetization for a particular current
direction depends on the sign of b and hence the location along
the wedge, as expected. Thus, currents of opposite polarities can
be used to switch the perpendicular magnetization in opposite
directions and the favoured direction of magnetization for each
current is determined by the sign of the lateral device asymmetry,
quantified by dHk/dy. We obtained similar results for several
other devices measured at different points along the wedge.
Owing to the correlation between dHk/dy and HzFL (Fig. 4c), we
expect that the current density required for switching can be
further reduced by increasing the value of dHk/dy, that is, by creating a larger structural asymmetry in the device. Moreover, the perpendicular HzFL can be used together with conventional SOTs to
switch the device. The conventional SOTs can in this case bring
the perpendicular magnetization into the sample plane, where it is
metastable30, and the new torques would determine the switching
direction. It should be noted that the new DL term in equation
(1) can, in principle, also promote switching of perpendicular
magnets by acting as a ‘negative’ damping. However, based on the
experimental EHE loops presented in this work, the observed
switching is attributed predominantly to the FL term.
6
The stack structure of Ta/CoFeB/TaOx was fabricated from
Ta(5 nm)/Co20Fe60B20(1 nm)/Ta(wedge) sputtered films. The metal layers were
deposited on a thermally oxidized wafer (on an area of 10 mm × 50 mm) by d.c.
magnetron sputtering at room temperature in an AJA International Physical Vapour
Deposition System. The deposition rates were 0.06 nm s21 for Ta and 0.03 nm s21
for Co20Fe60B20 at Ar pressures of 2 mtorr and 3 mtorr, respectively. The top Ta was
grown in a wedge shape, giving a continuous gradient of thickness along the length
of the sample. The thickness of the top Ta was varied from 0.81 to 2.13 nm across the
wafer, as shown in Fig. 2a. The TaOx layer was formed by exposing the sample to
radiofrequency O2/Ar plasma for 100 s. The top-wedged Ta layer was thus oxidized,
resulting in a change of oxidation level at the CoFeB/TaOx interface along the wedge
direction. The films were then annealed at 200 8C for 30 min to enhance their PMA.
The sample was subsequently patterned into an array of Hall bar devices (seven in
the width direction, with constant thickness of the top Ta layer, and 35 in the length
direction of the sample, varying its thickness) by standard photolithography and dry
etching techniques. The size of the Hall bars was fixed at 20 mm × 130 mm. The Hall
bar lengths were oriented along the width direction of the sample, resulting in a
varying top Ta thickness (hence oxidation) along the width of the Hall bars (that is,
y axis). The spacing between two adjacent Hall bar devices in the length direction
was 1.1 mm. A Keithley 6221 current source and a Keithley 2182A nanovoltmeter
were used in the extraordinary Hall voltage measurement. The external magnetic
field was generated by a Helmholtz coil driven by a Kepco power supply. All
measurements were carried out at room temperature. More than 24 devices similar
to device A and 16 devices similar to device B were tested in the same batch. Similar
behaviour was also observed in other batches of devices.
Received 28 October 2013; accepted 8 April 2014;
published online 11 May 2014
References
1. Miron, I. M. et al. Current-driven spin torque induced by the Rashba effect in a
ferromagnetic metal layer. Nature Mater. 9, 230–234 (2010).
2. Miron, I. M. et al. Perpendicular switching of a single ferromagnetic layer
induced by in-plane current injection. Nature 476, 189–193 (2011).
3. Liu, L. Q., Lee, O. J., Gudmundsen, T. J., Ralph, D. C. & Buhrman, R. A. Currentinduced switching of perpendicularly magnetized magnetic layers using spin
torque from the spin Hall effect. Phys. Rev. Lett. 109, 096602 (2012).
4. Liu, L. Q. et al. Spin-torque switching with the giant spin Hall effect of tantalum.
Science 336, 555–558 (2012).
5. Kim, J. et al. Layer thickness dependence of the current-induced effective
field vector in Ta vertical bar CoFeB vertical bar MgO. Nature Mater. 12,
240–245 (2013).
6. Fan, X. et al. Observation of the nonlocal spin–orbital effective field. Nature
Commun. 4, 1799 (2013).
7. Pai, C. F. et al. Spin transfer torque devices utilizing the giant spin Hall effect of
tungsten. Appl. Phys. Lett. 101, 122404 (2012).
8. Ryu, K. S., Thomas, L., Yang, S. H. & Parkin, S. Chiral spin torque at magnetic
domain walls. Nature Nanotech. 8, 527–533 (2013).
9. Emori, S., Bauer, U., Ahn, S. M., Martinez, E. & Beach, G. S. D. Current-driven
dynamics of chiral ferromagnetic domain walls. Nature Mater. 12,
611–616 (2013).
10. Haazen, P. P. J. et al. Domain wall depinning governed by the spin Hall effect.
Nature Mater. 12, 299–303 (2013).
11. Wang, K. L., Alzate, J. G. & Amiri, P. K. Low-power non-volatile spintronic
memory: STT-RAM and beyond. J. Phys. D 46, 074003 (2013).
12. Mangin, S. et al. Current-induced magnetization reversal in nanopillars with
perpendicular anisotropy. Nature Mater. 5, 210–215 (2006).
13. Garello, K. et al. Symmetry and magnitude of spin–orbit torques in
ferromagnetic heterostructures. Nature Nanotech. 8, 587–593 (2013).
14. Wang, X. & Manchon, A. Diffusive spin dynamics in ferromagnetic thin films
with a Rashba interaction. Phys. Rev. Lett. 108, 117201 (2012).
NATURE NANOTECHNOLOGY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturenanotechnology
© 2014 Macmillan Publishers Limited. All rights reserved.
NATURE NANOTECHNOLOGY
DOI: 10.1038/NNANO.2014.94
15. Manchon, A. Spin Hall effect versus Rashba torque: a diffusive approach.
Preprint at http://arxiv.org/abs/1204.4869 (2012).
16. Bychkov, Y. A. & Rashba, E. I. Properties of a 2d electron-gas with lifted spectral
degeneracy. JETP Lett. 39, 78–81 (1984).
17. Dyakonov, M. I. & Perel, V. I. Current-induced spin orientation of electrons in
semiconductors. Phys. Lett. A 35A, 459–460 (1971).
18. Hirsch, J. E. Spin Hall effect. Phys. Rev. Lett. 83, 1834–1837 (1999).
19. Zhang, S. F. Spin Hall effect in the presence of spin diffusion. Phys. Rev. Lett. 85,
393–396 (2000).
20. Sinova, J. et al. Universal intrinsic spin Hall effect. Phys. Rev. Lett. 92,
126603 (2004).
21. Hals, K. M. D. & Brataas, A. Phenomenology of current-induced spin–orbit
torques. Phys. Rev. B 88, 085423 (2013).
22. Schellekens, A. J., van den Brink, A., Franken, J. H., Swagten, H. J. M. &
Koopmans, B. Electric-field control of domain wall motion in perpendicularly
magnetized materials. Nature Commun. 3, 847 (2012).
23. Ikeda, S. et al. A perpendicular-anisotropy CoFeB–MgO magnetic tunnel
junction. Nature Mater. 9, 721–724 (2010).
24. Manchon, A. et al. Analysis of oxygen induced anisotropy crossover in
Pt/Co/MOx trilayers. J. Appl. Phys. 104, 043914 (2008).
25. Kita, K., Abraham, D. W., Gajek, M. J. & Worledge, D. C. Electric-field-control
of magnetic anisotropy of Co0.6Fe0.2B0.2/oxide stacks using reduced voltage.
J. Appl. Phys. 112, 033919 (2012).
26. Yang, H. X. et al. First-principles investigation of the very large perpendicular
magnetic anisotropy at Fe/MgO and Co/MgO interfaces. Phys. Rev. B
84, 054401 (2011).
27. Lee, O. J. et al. Central role of domain wall depinning for perpendicular
magnetization switching driven by spin torque from the spin Hall effect.
Phys. Rev. B 89, 024418 (2014).
28. Pi, U. H. et al. Tilting of the spin orientation induced by Rashba effect in
ferromagnetic metal layer. Appl. Phys. Lett. 97, 162507 (2010).
29. Khvalkovskiy, A. V. et al. Matching domain-wall configuration and spin–orbit
torques for efficient domain-wall motion. Phys. Rev. B 87, 020402 (2013).
ARTICLES
30. Bhowmik, D., You, L. & Salahuddin, S. Possible route to low current, high speed,
dynamic switching in a perpendicular anisotropy CoFeB–MgO junction
using spin Hall effect of Ta. 2012 IEEE International Electron Devices Meeting
(IEDM), paper no. 29.7.1–29.7.4 (2012).
Acknowledgements
This work was partially supported by the Defense Advanced Research Projects Agency
(DARPA) programme on Nonvolatile Logic (NVL) and in part by the National Science
Foundation Nanosystems Engineering Research Center for Translational Applications of
Nanoscale Multiferroic Systems (TANMS). This work was also supported in part by the
Function Accelerated nanoMaterial Engineering Center, one of six centres of
Semiconductor Technology Advanced Research network, a Semiconductor Research
Corporation programme sponsored by Microelectronics Advanced Research Corporation
and DARPA. P.U. and J.G.A. acknowledge partial support from a Qualcomm Innovation
Fellowship.
Author contributions
G.Q.Y. and P.U. jointly conceived the idea with contributions from Y.T., P.K.A. and K.L.W.
G.Q.Y. designed the experiments to test the idea, fabricated and measured the devices, with
contributions from Y.F., J.G.A., W.J., K.W., L.T.C., M.L., J.T., Y.J. and Y.W. P.U. performed
the theoretical analysis and modelling, with help from S.T., S.A.B. and Y.T. G.Q.Y., P.U.,
P.K.A. and K.L.W. wrote the paper. All authors discussed the results and commented on the
manuscript. The study was performed under the supervision of P.K.A. and K.L.W.
Additional information
Supplementary information is available in the online version of the paper. Reprints and
permissions information is available online at www.nature.com/reprints. Correspondence and
requests for materials should be addressed to G.Y. and K.L.W.
Competing financial interests
The authors declare no competing financial interests.
NATURE NANOTECHNOLOGY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturenanotechnology
© 2014 Macmillan Publishers Limited. All rights reserved.
7