Influence of top electrode on the current

advertisement
APPLIED PHYSICS LETTERS 87, 162502 共2005兲
Influence of top electrode on the current-induced magnetic switching
in magnetic nanopillars
T. Yanga兲 and J. Hamrle
Frontier Research System, RIKEN, 2-1 Hirosawa, Wako, Saitama 351-0198, Japan
and CREST, Japan Science and Technology Corporation, Japan
T. Kimura and Y. Otani
Institute for Solid State Physics, University of Tokyo, Kashiwa, Chiba 227-8581, Japan,
Frontier Research System, RIKEN, 2-1 Hirosawa, Wako, Saitama 351-0198, Japan
and CREST, Japan Science and Technology Corporation, Japan
共Received 18 March 2005; accepted 22 August 2005; published online 10 October 2005兲
Magnetic nanopillars with variable top electrodes were fabricated to clarify the roles of the spin
current and the spin accumulation in the current-induced magnetic switching. The critical switching
current is significantly increased when the size of the top electrode is comparable to that of the
nanopillar. This result implies that the dominant contribution in the current-induced magnetic
switching is not the spin accumulation, but the spin current. © 2005 American Institute of Physics.
关DOI: 10.1063/1.2093921兴
The magnetic nanopillar with a ferromagnet/nonmagnet/
ferromagnet layered structure is attractive for the application
to the magnetic random access memory, which has parallel
共P兲 and antiparallel 共AP兲 magnetic configurations, switchable by either an external field or a dc current.1 Since the first
dc current induced magnetic switching was experimentally
demonstrated using such nanopillars by Katine et al.,2 there
have been many experimental3–6 and theoretical7–15 studies
on this type of systems. Nevertheless, the roles of the spin
current and the spin accumulation have not been clarified
yet.
Usually, when a charge current je given by a sum of the
spin-up j↑ and the spin-down j↓ currents travels through alternating magnetic and nonmagnetic layers, the spin current
js 共=j↑ − j↓兲 is evolved as a consequence of the spindependent scattering. Accumulated spins at the interfaces accordingly build the split in the electrochemical potential
⌬␮ 共=␮↑ − ␮↓兲. As the simplest case, the spin-up and down
currents are correlated with the spin accumulation as in following one-dimensional diffusion equations16
j↑,↓ =
␴↑,↓ ⳵ ␮↑,↓
e ⳵x
共1兲
and
⌬ ␮ ⳵ 2⌬ ␮
=
,
␭2
⳵ x2
共2兲
where ␭, ␮, ␴, and e are, respectively, the spin-diffusion
length, the electrochemical potential, the electrical conductivity, and the electronic charge. Despite the earlier set of
diffusion equations, the roles of the spin accumulation and
the spin current in the current-induced magnetic switching
are still a point of controversy. Theories based on the spin
transfer1,8–15,17 suggest that the spin current transfers the
transverse component of the spin angular momentum to the
local magnetic moment at the interface upon flowing into the
adjacent magnetic layer, and thereby a torque is exerted on
a兲
Electronic mail: tyang@riken.jp
the local moment, resulting in spin-wave excitation or magnetic switching. Notably the theory based on the spin
accumulation7 claims that the nonequilibrium magnetization
of the accumulated spins applies an effective exchange field
on the local magnetization, described as the nonequilibrium
exchange interaction for the case of two magnetic layers
separated by a nonmagnetic one. The effective field thus
forces the local magnetization to take the orientation of the
accumulated spins. Remarkable is that even among the spintransfer theories, divergence appears in terms of the origin of
the spin current. Recently, it has been reported11–13 that the
spin accumulation theoretically gives rise to a transverse spin
current at the interface much larger than the spin current
carried by the polarized charge current. Therefore, we have
three possible contributions to the current-induced magnetic
switching, such as the spin current carried by the polarized
charge current 共hereafter the spin current兲, the transverse
spin current induced by the spin accumulation, and the effective field due to the spin accumulation. It is important to
clarify the roles of these three possible contributions for further theoretical and practical understandings. In this letter
our main purpose is to clarify the roles of the spin current
and the spin accumulation. If the spin current gives dominant
contribution, the polarization of the charge current should be
increased to reduce the critical switching current. If the spin
accumulation gives dominant contribution, no matter
whether it induces the effective field or the transverse spin
current, the spin accumulation should be enhanced.
The studied nanopillars comprise an electron-beam
evaporated layered structure of Cu 共60 nm兲 / Co 共40
nm兲 / Cu 共6 nm兲 / Co 共2.5 nm兲 / Au 共15 nm兲 / Cu 共50 nm兲. The
layers of Cu 共60 nm兲 and Cu 共50 nm兲 serve as the bottom and
top electrodes, respectively. The extended bottom Co 共40
nm兲 layer has a fixed magnetization during the currentinduced magnetic switching, while the top Co 共2.5 nm兲 layer
has freely switchable magnetization. For this study, the effect
of the spin current needs to be separated from the effect of
the spin accumulation. This can be realized by adjusting the
size of the top electrode.
0003-6951/2005/87共16兲/162502/3/$22.50
87, 162502-1
© 2005 American Institute of Physics
Downloaded 16 Oct 2005 to 134.160.214.75. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp
162502-2
Yang et al.
Appl. Phys. Lett. 87, 162502 共2005兲
FIG. 1. Schematic illustration of the grid-shaped top electrode and the current flowing paths 共the solid lines with big arrow兲.
Usually, the top electrode of a nanopillar is much wider
than the nanopillar itself to ensure good electric contact. According to the theoretical calculations,12,14 a wide top electrode facilitates the spin-flip scattering, and suppresses the
spin accumulation inside the nanopillar. By decreasing the
width of the top electrode down to that of the nanopillar, the
spin accumulation is increased while the spin current is
decreased.14 In this way, the effect of the spin current can be
separated from the effect of the spin accumulation by comparing the switching behaviors of nanopillars with the conventional wide top electrode 共⬃10 ␮m wide兲 and narrow top
electrode 共⬃100 nm wide兲, respectively.
However, it is technically difficult to align a top electrode of ⬃100 nm in width onto a nanopillar of ⬃100 nm.
To solve this problem, we designed a grid-shaped top electrode, as shown in Fig. 1. The slits of the grid are narrower
than the nanopillar whereas the width of the conductive
strips between the slits is comparable to the nanopillar. The
whole grid is much wider than the nanopillar so that the
contact between the top electrode and the nanopillar is assured. The length of the conductive strips is much larger than
the spin-diffusion length.
The detailed nanopillar fabrication process is described
in our previous work.18 The ion milling is carefully carried
out to avoid a step edge in the bottom Co layer producing
dipolar coupling to the top Co layer. Prior to the top electrode deposition, the Au surface of the nanopillar is cleaned
with ion milling and then quickly moved into the evaporation
chamber. The remaining Au capping layer is estimated to be
less than 15 nm after the final ion milling. This thickness is
well below the room temperature spin-diffusion length of
evaporated Au, roughly estimated to be 35 nm according to
the directly measured values.19–21 Although there is also possible spin-flipping at the Au/ Cu interface,20 the spin-memory
loss in the Au layer and at the interface should not be
significant.
Figures 2共a兲 and 2共b兲 show the scanning electron microscope 共SEM兲 images of a grid-shaped Cu electrode and a
conventional wide Cu electrode, respectively. Figure 2共c兲
shows the top-view image of the fabricated nanopillar with
the size of 130⫻ 70 nm2. Such a nanopillar partially covered
with the grid-shaped electrode is shown in Fig. 2共d兲. The
nanopillar cannot be seen when the nanopillar is fully covered with the single Cu strip. The slits in the grid are 100 nm
wide and the Cu strip is 150 nm wide.
The perpendicular transport properties of the pillars are
measured by means of lock-in technique at room temperature. To compensate the possible dipolar coupling field, the
differential resistance 共dV / dI兲 vs dc current 共Idc兲 loop is
measured with varying the applied field along the long axis
FIG. 2. SEM images of 共a兲 the grid-shaped Cu electrode, 共b兲 the wide Cu
electrode, 共c兲 a nanopillar 共top view兲, and 共d兲 a nanopillar half-covered with
the grid-shaped electrode, indicated by the dashed white open circle.
of the elliptic pillar. Since both the external field and the
dipolar coupling field cause shift and shrinkage of the dV / dI
vs Idc loop,22 the averaged critical switching current is determined from the widest loop in the measurement. We found
that only the fully covered nanopillars exhibited sharp
single-step switchings. So the results discussed hereafter are
for the fully covered nanopillars.
Figures 3共a兲 and 3共b兲 are the magnetoresistance 共MR兲
loops for two nanopillars with the wide and the grid-shaped
narrow top electrodes, respectively. The corresponding
dV / dI vs Idc loops are plotted in Fig. 3共c兲. For both nanopillars, sharp and full transitions between the AP and P states
are induced by either the applied field or the dc current. The
nanopillar with the wide top electrode has a reasonable resistance of about 1.5 ⍀ and a MR ratio of 4.8%. However,
the resistance of the nanopillar with the grid-shaped electrode is as high as 4.4 ⍀. This high value is caused by the
grid-shaped electrode, in which the electrical current flows
along the paths shown in Fig. 1. It can be understood that the
measured resistance includes the contribution of the top electrode as well as the pillar resistance. The two samples are
fabricated from the same chip. The only difference between
them is the top electrode.
Remarkable is a large difference in the critical switching
current seen as a change in width of the two dV / dI vs Idc
FIG. 3. MR loops for nanopillars with 共a兲 the wide top electrode, and 共b兲 the
grid-shaped top electrode. 共c兲 dV / dI vs Idc loops for nanopillars with the
wide top electrode 共dashed line兲, and the grid-shaped top electrode
共solid line兲.
Downloaded 16 Oct 2005 to 134.160.214.75. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp
162502-3
Appl. Phys. Lett. 87, 162502 共2005兲
Yang et al.
loops in Fig. 3共c兲. The relation 共兩I P→AP兩 + 兩IAP→P兩兲 / 2 yields
the averaged critical switching current densities of 5.1
⫻ 107 A / cm2 for the wide top electrode and 9.2
⫻ 107 A / cm2 for the grid-shaped narrow top electrode. The
latter is 1.8 times as large as the former. A similar result is
also obtained for nanopillars with a larger size of 160
⫻ 100 nm2. For six nanopillars with narrow top electrode
and four nanopillars with wide top electrode, the critical
switching current density is increased in average by a factor
of 1.7 when the wide top electrode is replaced with the narrow one.
One possible reason for the difference in the critical
switching current is the Oersted field due to the current flowing in the top electrode. According to the magnetization direction of the bottom Co layer, this Oersted field should
cause expansion or shrinkage of the dV / dI vs Idc loop. To
examine the influence of this Oersted field, the magnetization
of the bottom Co is reversed, and then the dV / dI vs Idc loop
is measured again. Almost the same result is obtained, meaning that the influence of the current in the top electrode is
negligible.
Theoretical calculations show that the size of the top
electrode affects the spin current14 and the spin
accumulation12,14 inside the nanopillar. Increase in the size of
the top electrode diminishes the spin accumulation at the
Cu/ Co共free layer兲 interface by a factor of about 4 共Ref. 14兲
or 2.12 In return, the spin current is increased at the same
position by a factor of about 2.14 This means that the wide
electrode provides a large volume for conduction-electronspin relaxation. Thus the spin accumulation is difficult to be
built-up inside the nanopillar. At the same time, the spin
current from the pillar is absorbed to the top electrode, leading to the enhanced spin current inside the nanopillar. This
can also be explained by employing the spin-flip resistance
RS = ␭ / 关␴A共1 − ␣2兲兴, where A is the cross-sectional area, and
␣ is the spin asymmetry. Increase in the area A decreases the
value of RS, and thus accelerates the spin-flip scattering.14
Taking into account the earlier discussion, present experimental results show that the critical switching current is
increased by promoting the spin accumulation 共decreasing
the spin current兲. In other words, the magnetic switching
becomes difficult to occur. This indicates that the spin current is the dominant factor in the current-induced magnetic
switching in the magnetic nanopillars, contrary to previous
theoretical conclusions.7,11–13 Those conclusions support that
the spin accumulation is dominant and, hence, should be
increased to reduce the critical switching current. We do not
deny the effects of the spin accumulation, but our experimental results show that the spin current contribution seems
dominant and more effective. Therefore, to reduce the critical switching current, the spin polarization of the charge current should be increased. The possible ways are to introduce
spin-flip scatterers in the electrode or to use capping layers
with short spin-diffusion length.5,10,14
In summary, with using the grid-shaped top electrode,
we succeeded in examining the contributions of the spin current and the spin accumulation. The results show that the
spin current carried by the polarized charge current makes
the dominant contribution to the current-induced magnetic
switching.
The authors are grateful to the Nanoscience Development and Support Team of RIKEN for their technical
support.
J. C. Slonczewski, J. Magn. Magn. Mater. 159, L1 共1996兲.
J. A. Katine, F. J. Albert, R. A. Buhrman, E. B. Myers, and D. C. Ralph,
Phys. Rev. Lett. 84, 3149 共2000兲.
3
J. Grollier, V. Cros, A. Hamzic, J. M. George, H. Jaffre’s, A. Fert,
G. Faini, J. B. Youssef, and H. Legall, Appl. Phys. Lett. 78, 3663 共2001兲.
4
J. Z. Sun, D. J. Monsma, D. W. Abraham, M. J. Rooks, and R. H. Koch,
Appl. Phys. Lett. 81, 2202 共2002兲.
5
S. Urazhdin, N. O. Birge, W. P. Pratt, Jr., and J. Bass, Appl. Phys. Lett. 84,
1516 共2004兲.
6
S. I. Kiselev, J. C. Sankey, I. N. Krivorotov, N. C. Emley,
R. J. Schoelkopf, R. A. Buhrman, and D. C. Ralph, Nature 共London兲 425,
380 共2003兲.
7
C. Heide, P. E. Zilberman, and R. J. Elliott, Phys. Rev. B 63, 064424
共2001兲.
8
M. D. Stiles and A. Zangwill, J. Appl. Phys. 91, 6912 共2002兲.
9
X. Waintal, E. B. Myers, P. W. Brouwer, and D. C. Ralph, Phys. Rev. B
62, 12317 共2000兲.
10
A. A. Kovalev, A. Brataas, and G. E. W. Bauer, Phys. Rev. B 66, 224424
共2002兲.
11
L. Berger, J. Appl. Phys. 89, 5521 共2001兲.
12
L. Berger, J. Magn. Magn. Mater. 278, 185 共2004兲.
13
A. Fert, V. Cros, J.-M. George, J. Grollier, H. Jaffrès, A. Hamzic,
A. Vaurès, G. Faini, J. B. Youssef, and H. Le Gall, J. Magn. Magn. Mater.
272-276, 1706 共2004兲.
14
J. Hamrle, T. Kimura, T. Yang, and Y. Otani, Phys. Rev. B 71, 094434
共2005兲.
15
J. Zhang, P. M. Levy, and S. Zhang, Phys. Rev. Lett. 93, 256602 共2004兲.
16
P. C. van Son, H. van Kempen, and P. Wyder, Phys. Rev. Lett. 58, 2271
共1987兲.
17
L. Berger, Phys. Rev. B 54, 9353 共1996兲.
18
T. Yang, T. Kimura, and Y. Otani, J. Appl. Phys. 97, 064304 共2005兲.
19
Y. Ji, A. Hoffmann, J. S. Jiang, and S. D. Bader, Appl. Phys. Lett. 85,
6218 共2004兲.
20
H. Kurt, Wen-C. Chiang, C. Ritz, K. Eid, R. Loloee, W. P. Pratt, Jr.,
and J. Bass, J. Appl. Phys. 93, 7918共2003兲.
21
A. B. Gougam, F. Pierre, H. Pothier, D. Esteve, and N. O. Birge, J. Low
Temp. Phys. 118, 447 共2000兲.
22
S. Urazhdin, H. Kurt, W. P. Pratt, Jr., and J. Bass, Appl. Phys. Lett. 83,
114 共2003兲.
1
2
Downloaded 16 Oct 2005 to 134.160.214.75. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp
Download