Spin-torque diode effect in magnetic tunnel junctions.

Abstract

There is currently much interest in the development of 'spintronic' devices, in which harnessing the spins of electrons (rather than just their charges) is anticipated to provide new functionalities that go beyond those possible with conventional electronic devices. One widely studied example of an effect that has its roots in the electron's spin degree of freedom is the torque exerted by a spin-polarized electric current on the spin moment of a nanometre-scale magnet. This torque causes the magnetic moment to rotate at potentially useful frequencies. Here we report a very different phenomenon that is also based on the interplay between spin dynamics and spin-dependent transport, and which arises from unusual diode behaviour. We show that the application of a small radio-frequency alternating current to a nanometre-scale magnetic tunnel junction can generate a measurable direct-current (d.c.) voltage across the device when the frequency is resonant with the spin oscillations that arise from the spin-torque effect: at resonance (which can be tuned by an external magnetic field), the structure exhibits different resistance states depending on the direction of the current. This behaviour is markedly different from that of a conventional semiconductor diode, and could form the basis of a nanometre-scale radio-frequency detector in telecommunication circuits.

Full text

Page 1

© 2005 Nature Publishing Group
Spin-torque diode effect in magnetic tunnel
junctions
A. A. Tulapurkar1,2†, Y. Suzuki1,2,3, A. Fukushima1,2, H. Kubota1,2, H. Maehara4, K. Tsunekawa4,
D. D. Djayaprawira4, N. Watanabe4& S. Yuasa1,2
There is currently much interest in the development of ‘spin-
tronic’ devices, in which harnessing the spins of electrons (rather
than just their charges) is anticipated to provide new functional-
ities that go beyond those possible with conventional electronic
devices. One widely studied example of an effect that has its roots
in the electron’s spin degree of freedom is the torque exerted by a
spin-polarized electric current on the spin moment of a nano-
metre-scale magnet. This torque causes the magnetic moment to
rotate1–19at potentially useful frequencies. Here we report a very
different phenomenon that is also based on the interplay
between spin dynamics and spin-dependent transport, and
which arises from unusual diode behaviour. We show that the
application of a small radio-frequency alternating current to a
nanometre-scale magnetic tunnel junction20–22can generate a
measurable direct-current (d.c.) voltage across the device when
the frequency is resonant with the spin oscillations that arise
from the spin-torque effect: at resonance (which can be tuned
by an external magnetic field), the structure exhibits different
resistance states depending on the direction of the current. This
behaviour is markedly different from that of a conventional
semiconductor diode23, and could form the basis of a nano-
metre-scale radio-frequency detector in telecommunication
circuits.
Weperformedexperimentsonamagnetictunneljunction(MTJ)in
the structure Si (substrate)/PtMn (15nm)/CoFe (2.5nm)/Ru
(0.85nm)/CoFeB (3nm)/MgO (0.85nm)/CoFeB (3nm); see Fig. 1a.
This multi-layered film was further patterned into oval-shaped pillars
of dimension 200nm £ 100nm, using electron-beam lithography
and ion milling techniques. The bottom anti-ferromagnetically
coupled CoFe and CoFeB layers (the synthetic antiferromagnetic
layer) act as a pinned layer, while the top CoFeB layer acts as a free
layer, whose magnetization can be changed. The resistance of the
MTJ depends on the relative orientations of the pinned and free
layers. The present MTJ shows a giant tunnelling magnetoresistance
(TMR)duetothecrystallineMgO(001)tunnellingbarrier.Acurrent
passing throughtheMTJgetsspin-polarized by thepinnedlayer,and
exerts a torque on the free layer.
The experimental arrangement to measure the diode effect is
shown in Fig. 1a. A bias T is used to pass high-frequency current
(200MHz to 15GHz) through the MTJ and to measure the d.c.
voltage simultaneously. For all the experiments described here, the
external magnetic field was applied at an angle of 308 from the
pinned-layer magnetization axis within the film plane (see inset of
Fig.1b).InthisgeometrythesampleshowedagiantTMRof,100%,
as shown in Fig. 1b. We also measured microwave power from
the MTJ arising from the thermal fluctuations of the free-layer
magnetization24,25. The power was measured by a spectrum analyser,
by passing a d.c. current of 1mA using a bias T.
The radio frequency (r.f.) response of the MTJ was first tested
using a network analyser. The results obtained showed evidence of
magnetic resonance excited by r.f. current (results not shown).
LETTERS
Figure 1 | Experimental set-up and magnetoresistance. a, Schematic
diagramofthe experimental set-upandcross-sectionalview ofthemagnetic
tunneljunction(MTJ)device.Thethicknessesofvariouslayersofthedevice
in nanometres are given in brackets. The bottom CoFeB and CoFe layers,
coupled anti-ferromagnetically through the Ru layer, act as a pinned layer.
The top CoFeB layer acts as a free layer, the magnetization of which can be
changed. The pinned and free layers are separated by a tunnelling MgO
barrier. The experimental set-up measures the d.c. voltage produced across
the device on applying the r.f. current. b, The magnetoresistance of the
device, by applying magnetic field at 308 from the pinned-layer
magnetization. The arrows indicate the sweeping direction of the magnetic
field.
1Nanoelectronics Research Institute, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba 305-8568, Japan.2CREST, Japan Science and Technology
Agency (JST), 4-1-8 Honcho, Kawaguchi 332-0012, Japan.3Graduate School of Engineering Science, Osaka University, 1-3 Machikaneyama, Toyonaka, Osaka 560-8531, Japan.
4Electron Device Equipment Division, Canon ANELVA Corporation, 5-8-1 Yotsuya, Fuchu, Tokyo 183-8508, Japan. †Present address: Stanford Linear Accelerator Center, Stanford
University, California 94025, USA.
Vol 438|17 November 2005|doi:10.1038/nature04207
339

Page 2

© 2005 Nature Publishing Group
Current-induced resonance has recently been observed also in a
magneticdomainwallinthemegahertzfrequency range26.Thesignal
is, however, small, and its phase determination is prone to errors in
thecalibrationofthenetworkanalyser.Incontrast,wefoundthatthe
MTJ produces d.c. voltage because of its nonlinear behaviour and
thisdiode-effectmeasurementoffersphase-problem-freeresults.The
effect of alternating current(a.c.) on the precession of magnetization
inducedbylarged.c. currenthas recently19been studied. However,in
the present experiment, we excite the magnetization only with
alternating current, without applying a d.c. bias. The d.c. voltage
response measured by passing 0.55mA of r.f. current is plotted in
Fig. 2. The response shows a large resonance structure, whose
position depends on the magnetic field. Figure 3a shows the noise
power spectra having a large peak, along with a small side peak, the
positions of which also depend on the magnetic field.
The working principles of the spin-torque diode and the semi-
conductor p–n junction diode are compared in Fig. 4a. As shown in
Fig. 4a, when current flows from the n side to the p side, the space
charge region around the p–n junction is enlarged, and so the
resistance of the p–n junction is higher in this case. For the opposite
direction of current, the space charge region is shrunk, which gives
lower resistance. In the case of the spin-torque diode, the alternating
current passing through it exerts a torque on the free-layer spin
moment. When the frequency of the alternating current nears the
precession frequency of the free-layer spin-moment, the spin is tilted
towards the pinned-layer magnetization during the negative (or
positive) half of the alternating current. This configuration has low
resistance. During the next half of the alternating current, spin is
tilted in the opposite direction, which is a high-resistance state. The
difference in resistance during positive and negative currents pro-
duces d.c. voltage, in the case of both the diodes. In contrast to the
semiconductor diode, the spin-torque diode is resistant to the noise
because it produces d.c. voltage only in a narrow frequency range
around the resonance frequency, which can be tuned by applying a
magnetic field.
Furthermore, the spin-torque diode effect is phase-sensitive. A
small r.f. current (I ¼ Ia.c.sin(2pft) passing through MTJ exerts a
torque on the free-layer spin. As a result, the z component (parallel
to the pinned-layer magnetization) of the spin oscillates as
Sz¼ cosðvÞþAsinð2pftÞþBcosð2pftÞ, where A and B are respect-
ively the in-phase and 908-phase components of the motion of the
spin, Szdenotes the direction cosine of the spin and v is the angle
between the free and pinned layers. The resistance of the MTJ
depends on Szas follows: R ¼ R0þ 0.5*DR*(1 2 Sz), where R0is
the resistance of the sample when the free layer is parallel to the
pinned layer, and DR is the increase in the resistance when the free
layer is anti-parallel to the pinned layer. The d.c. voltage given by
the time-averaged value of I*R is Vd.c.¼ 2A*DR*Ia.c./4. Thus alter-
nating currentpassing through the MTJproduces d.c. voltage,which
is sensitive to the in-phase part of the oscillation of the magnetic
moment. Because this is an intra-sample detection, it gives a phase-
error-free spectrum, and a new method for performing ferromag-
netic resonance experiments, for which at present an oscillating
magnetic field is usually applied.
A torque exerted on the free-layer spin can in general be decom-
posedintotwodirectionsorthogonaltoit,thatis,s ˆfree£ (s ˆfree£ s ˆpin)
and (s ˆfree£ s ˆpin), where s ˆfreeand s ˆpinare unit vectors along the
magnetizations of the free and pinned layers respectively. The torque
from the spin-transfer effect1,2lies in the first direction. The torque
along the second direction is called the field-like term, and has been
predicted in different ways3,4. Owing to the symmetry differences
between these terms (when the free- and pinned-layer magnetiza-
tions are in-plane), the d.c. voltage produced by the spin-transfer
torque shows a peak, whereas the d.c. voltage produced by an
effective field shows a dispersion curve, as shown in Fig. 4b. If both
torques act on the spin, by superposition, we get d.c. voltage, as
shown in the bottom panel of Fig. 4b. For small and uniform
oscillation of the magnetic moment, the d.c. voltage is given
approximately by (see Supplementary Discussion):
Vd:c:<1
4DRsin2ðvÞI2
a:c:Re
ifg
ð f2
0ST2g
02f2Þ2ifag0Hd
0Hdg
0FT
??
ð1Þ
whereSTisthespintransferandFTistheeffectivefieldterm,perunit
current. g is the gyromagnetic ratio (g
Gilbert damping factor, Hdis the demagnetization field perpendicu-
lar to the free-layer plane, f0is the resonant frequency and f is the
frequencyofthe applied alternating current,Ia.c.. (We have neglected
the stray field effect for a uniform mode.)
The comparison of the d.c. voltage and thermal noise power
spectrum at 300Oe is shown in Fig. 3b. Each spectrum is composed
of a large resonance centred at about 7.5GHz and a small resonance
at lower frequency. Here we consider only the dominant resonance
(larger peak in the noise) as the uniform oscillation of the free layer
(see Supplementary Fig. 1). The d.c. voltage from this mode corre-
sponds to a combination of spin-transfer and effective-field torques
(comparewith thebottom trace inFig. 4b).Theresonancefrequency
obtained from fitting to a superposition-type spectrum coincides
with the main peak position in the thermal noise spectrum. The
frequency of the resonance position as a function of magnetic field is
plotted in Fig. 3c. For small v, and neglecting the marginal influence
of d.c. bias current, the resonance frequency is approximately given
byKittel’sequation27:f ¼ g
jHdipþHextjÞ?1=2where Hcis the coercivity and Hdipis the dipolar
field from the pinned layer. The fitting to this equation gives Hcþ
Hdip¼ 176Oe and Hd¼ 12.8kOe.
As mentioned above, because d.c.-measurement is phase-sensitive
detection, we can decompose the spectra into two sources of
torque—spin transfer and effective field—by using equation (1);
seeSupplementaryFig.2.Further,wealsomeasuredthed.c.response
by changing the r.f. current (see Fig. 3d), and found that the d.c.
voltage decreased linearly with the square of the current, as expected
from equation (1).
Thed.c.voltageoutputofanidealsemiconductordiodeisgivenby
Vd.c.< eVr.f.
the diode, kBis Boltzmann’s constant and T is the temperature. The
0¼ 2g/2p . 0), a is the
0½ðHcþjHdipþHextjÞðHcþHdþ
2/kBT,whereVr.f.isthehigh-frequencyvoltageappliedto
8
f (GHz)
Figure2|Direct-currentvoltagegeneratedbythedeviceinresponsetothe
alternating current. The d.c. voltage is plotted as a function of the
frequency of the a.c. current (0.55mA). The external magnetic fields are as
shown.Thed.c.voltageresultsfromtheresonantoscillationofthemagnetic
momentofthefreelayerbycurrent-inducedspin-transferandeffective-field
torques.
LETTERS
NATURE|Vol 438|17 November 2005
340

Page 3

© 2005 Nature Publishing Group
Figure 3 | Magnetic-field dependence of microwave power and a.c. current
dependence of d.c. voltage. a, Microwave power spectra of the device
measuredbyspectrumanalyser,usingad.c.biascurrentof1mA.Thepower
spectra with zero d.c. current are taken as background and subtracted from
the data. The external magnetic fields are as shown. b, Comparison of the
d.c.voltage spectrumand microwavepowerspectrumat300Oe. Thearrows
mark the positions of two peaks in the microwave power. The d.c. spectrum
also has two corresponding resonance frequencies. The shape of the d.c.
spectrum is a combination of peak and dispersion curves (also see Fig. 4b).
Thus, the resonance positions do not match with the maxima in the d.c.
voltage. c, The magnetic-field dependence of resonant frequency
corresponding to the larger peak in the microwave power spectra. The black
pointsrepresenttheexperimentaldatapoints,whiletheredcurveisthefitto
the data using Kittel’s equation. The inset shows the extrapolation of the
fitted curve to zero frequency. d, Direct-current voltage from a different
sample for the given a.c. currents. The inset shows the linear power
dependence of the d.c. voltage measured from peak to valley, as marked by
the arrow in the main panel.
Figure 4 | Principle of the spin-torque diode. a, Comparison of the
functioningofthesemiconductorp–ndiode(leftpanel)andthespin-torque
diode (right panel). In the semi-conductor diode, if positive voltage is
applied to the n side, the space charge region around the p–n junction is
enlarged, and the resistance is high. For the opposite polarity, the space
charge region is shrunk and the resistance is low. In the case of the spin-
torque diode, the free-layer magnetization (shown by thin black arrows)
oscillatesowingtothecurrent-inducedtorque.Theresistanceofthediodeis
less when the a.c. current is negative, because the free layer makes a smaller
angle with the pinned layer (shown by the thick blue arrow). When the
alternating current is positive, the resistance is larger, owing to the larger
angle.Thebottomtraceintherightpanelofashowstheschematicvariation
of the product of the current and the change in resistance. The dotted line
shows the average value of this product, which appears as d.c. voltage across
the spin-torque diode. b, Theoretical plot of the d.c. voltage spectrum
obtained from equation (1). The d.c. voltage shows a peak if the torque
induced by the a.c. current is due to the spin-transfer effect only. This is
shown by the black curve in the top panel. But if the torque is due to the
effective-fieldeffectonly,thed.c.voltage showsa dispersioncurve, asshown
by the red curve. If both the torques apply, the d.c. voltage shows a
superposition of peak and dispersion, as shown by the green curve in the
bottom panel.
NATURE|Vol 438|17 November 2005
LETTERS
341

Page 4

© 2005 Nature Publishing Group
spin-torque diode can perform r.f. detection better than the semi-
conductordiodeifthemagnetizationofthepinnedlayerlies458out-
of-plane. In this geometry, by increasing the ratio (Hd/Hc) of free-
layer magnetization, we can squeeze the elliptical trajectory of the
precessionofthemagneticmomenttoliein-plane.Thisenhancesthe
amplitude of in-plane oscillation, and produces large d.c. voltage.
The maximum d.c. voltage produced by spin-transfer in this case is
given by (see Supplementary Discussion):
Vd:c:< TMRV2
r:f:
Vc
ffiffiffiffiffiffi
Hd
Hc
r
where Vcis the critical voltage required to flip the magnetization1.
Thus, by increasing the Hd/Hcratio as well as by increasing
the magnetoresistance (as we have done here using a high-
quality crystalline MgO barrier), the spin-torque diode can be a
sensitive power detector. The sensitivity can also be enhanced by
increasingthespin-transferefficiency28,29,whichdecreasesthecritical
voltage.
Received 28 March; accepted 1 September 2005.
1. Slonczewski, J. C. Current-driven excitation of magnetic multilayers. J. Magn.
Magn. Mater. 159, L1– -L7 (1996).
Berger, L. Emission of spin waves by a magnetic multilayer traversed by a
current. Phys. Rev. B 54, 9353– -9358 (1996).
Zhang, S., Levy, P. M. & Fert, A. Mechanisms of spin-polarized current-driven
magnetization switching. Phys. Rev. Lett. 88, 236601 (2002).
Tatara, G. & Kohno, H. Theory of current-driven domain wall motion: spin
transfer versus momentum transfer. Phys. Rev. Lett. 92, 086601 (2004).
Kiselev, S. I. et al. Microwave oscillations of a nanomagnet driven by a spin-
polarized current. Nature 425, 380– -383 (2003).
Rippard, W. H., Pufall, M. R., Kaka, S., Russek, S. E. & Silva, T. J. Direct-current
induced dynamics in Co90Fe10/Ni80Fe20point contacts. Phys. Rev. Lett. 92,
027201 (2004).
Sun, J. Applied physics: spintronics gets a magnetic flute. Nature 425, 359– -361
(2003).
Myers, E. B., Ralph, D. C., Katine, J. A., Louie, R. N. & Buhrman, R. A. Current-
induced switching of domains in magnetic multilayer devices. Science 285,
867– -870 (1999).
Tsoi, M. et al. Generation and detection of phase-coherent current-driven
magnons in magnetic multilayers. Nature 406, 46– -48 (2000).
10. Tulapurkar, A. A. et al. Subnanosecond magnetization reversal in magnetic
nanopillars by spin angular momentum transfer. Appl. Phys. Lett. 85,
5358– -5360 (2004).
11.Yagami, K., Tulapurkar, A. A., Fukushima, A. & Suzuki, Y. Low-current spin-
transfer switching and its thermal durability in a low-saturation-magnetization
nanomagnet. Appl. Phys. Lett. 85, 5634– -5636 (2004).
12. Koch, R. H., Katine, J. A. & Sun, J. Z. Time-resolved reversal of spin-transfer
switching in a nanomagnet. Phys. Rev. Lett. 92, 088302 (2004).
13. Urazhdin, S., Birge, N. O., Pratt, W. P. Jr & Bass, J. Current-driven magnetic
2.
3.
4.
5.
6.
7.
8.
9.
excitations in permalloy-based multilayer nanopillars. Phys. Rev. Lett. 91,
146803 (2003).
14. O¨zyilmaz, B. et al. Current-induced magnetization reversal in high magnetic
fields in Co/Cu/Co nanopillars. Phys. Rev. Lett. 91, 067203 (2003).
15. Grollier, J. et al. Spin-polarized current induced switching in Co/Cu/Co pillars.
Appl. Phys. Lett. 78, 3663– -3665 (2001).
16. Katine, J. A., Albert, F. J., Buhrman, R. A., Myers, E. B. & Ralph, D. C. Current-
driven magnetization reversal and spin-wave excitations in Co/Cu/Co pillars.
Phys. Rev. Lett. 84, 4212– -4215 (2000).
17. Stiles, M. D. & Zangwill, A. Anatomy of spin-transfer torque. Phys. Rev. B 66,
014407 (2002).
18. Lee, K.-J., Deac, A, Redon, O., Nozie `res, J.-P. & Dieny, B. Excitations of
incoherent spin-waves due to spin-transfer torque. Nature Mater. 3, 877– -881
(2004).
19. Pufall, M. R., Rippard, W. H., Kaka, S., Silva, T. J. & Russek, S. E. Frequency
modulation of spin-transfer oscillators. Appl. Phys. Lett. 86, 082506 (2005).
20. Yuasa, S., Nagahama, T., Fukushima, A., Suzuki, Y. & Ando, K. Giant room-
temperature magnetoresistance in single-crystal Fe/MgO/Fe magnetic tunnel
junctions. Nature Mater. 3, 868– -871 (2004).
21. Parkin, S. S. P. et al. Giant tunnelling magnetoresistance at room temperature
with MgO (100) tunnel barriers. Nature Mater. 3, 862– -867 (2004).
22. Djayaprawira, D. D. et al. 230% Room temperature magnetoresistance in
CoFeB/MgO/CoFeB magnetic tunnel junctions. Appl. Phys. Lett. 86, 092502
(2005).
23. Sze, S. M. Physics of Semiconductor Devices Ch. 2 (John Wiley & Sons, New
York, 1981).
24. Smith, N. & Arnett, P. White-noise magnetization fluctuations in
magnetoresistive heads. Appl. Phys. Lett. 78, 1448– -1450 (2001).
25. Nazarov, A. V., Cho, H. S., Nowak, J., Stokes, S. & Tabat, N. Tunable
ferromagnetic resonance peak in tunneling magnetoresistive sensor structures.
Appl. Phys. Lett. 81, 4559– -4561 (2002).
26. Saitoh, E., Miyajima, H., Yamaoka, T. & Tatara, G. Current-induced resonance
and mass determination of a single magnetic domain wall. Nature 432,
203– -206 (2004).
27. Kittel, C. Introduction to Solid State Physics Ch. 16 (John Wiley & Sons,
Singapore, 1996).
28. Jiang, Y. et al. Substantial reduction of critical current for magnetization
switching in an exchange-biased spin valve. Nature Mater. 3, 361– -364 (2004).
29. Manschot, J., Brataas, A. & Bauer, G. E. W. Reducing the critical switching
current in nanoscale spin valves. Appl. Phys. Lett. 85, 3250– -3252 (2004).
Supplementary Information is linked to the online version of the paper at
www.nature.com/nature.
Acknowledgements A.A.T. thanks the Japan Society for the Promotion of
Science for the fellowship grant. S.Y. thanks the Japan Science and Technology
Agency (JST) for the PRESTO programme. A part of this work is supported by
the 21st Century COE programme by JSPS. We thank C. Chappert, T. Devolder,
W. Mizutani and M. Mizuguchi for their help.
Author Information Reprints and permissions information is available at
npg.nature.com/reprintsandpermissions. The authors declare no competing
financial interests. Correspondence and requests for materials should be
addressed to Y.S. (suzuki-y@mp.es.osaka-u.ac.jp).
LETTERS
NATURE|Vol 438|17 November 2005
342