Spin-polarized current-induced magnetization reversal in single nanowires
Derek Kelly,* Jean-Eric Wegrowe,
²
Trong-kha Truong, Xavier Hoffer, and Jean-Philippe Ansermet
‡
Ecole Polytechnique Federale de Lausanne, CH-1015 Ecublens, Switzerland
Received 13 September 2002; revised manuscript received 5 May 2003; published 17 October 2003
Using electrochemical deposition, 6 m long Ni nanowires, with typical diameters of the order of 80 nm,
are grown in ion-track etched membranes. Electric contacts are established during the growth, allowing resis-
tance measurements of a single magnetic wire. Whatever the angle of the applied magnetic field with the wire,
the full loops of magnetoresistance of a nickel nanowire can be described quantitatively on the basis of
anisotropic magnetoresistance of a uniform magnet, and exhibit a jump of the magnetization at the so-called
switching field. Hybrid wires made half with nickel and half with a Co/Cu multilayer were also produced. The
multilayer could be grown using either a single bath technique or a multiple bath setup, with the result of a
different magnetic anisotropy in the Co layers. When the multilayer is made of an optimal number of layers,
the two parts of the hybrid wire act as two resistances in series, having no magnetic interaction onto each other.
In contrast, the action of a current pulse on the nickel magnetization is to provoke a switch, when injected
before the unstable state of the hysteresis cycle has been reached. But the amount of applied field discrepancy
where the current still has an effect is given by a measured value H
max
, which appears to be substantially
dependent on the presence or not of a multilayer close enough to the nickel wire and on the orientation of the
magnetization in the multilayer. The role of the multilayer’s presence or state evidences the role of spin
polarization in the current-induced switches of nickel. This is confirmed by measurements of the amplitude of
H
max
in homogeneous nickel wires that exclude spurious effects such as the induced oersted-field, heating, or
a combination of the two to account for all the current-induced switches.
DOI: 10.1103/PhysRevB.68.134425 PACS numbers: 72.25.Ba, 72.25.Hg, 72.25.Pn
I. INTRODUCTION
Spin-transport in magnetic nanostructures is currently re-
ceiving increasing interest from the scientific community be-
cause of its importance in the design of new memory devices
capable of maintaining the persistent increase in memory
storage density and the accompanying need for information
processing speed. Historically, as early as the 1950’s, the
coupling of conduction electrons to spin waves
1
was invoked
in order to account for deviations from the simplest version
of the two-current model that describes transport in metallic
ferromagnets.
2
Ferromagnetic resonnance FMR experi-
ments provided information on spin-flip scattering rates due
to this coupling.
3
On the other hand, in the past two decades
a lot of progress has been done in instrumental techniques
providing physicists with capabilities of engineering struc-
tures on the nanometer scale. A size that is in principle small
enough to allow the magnetization to remain uniform, mak-
ing those samples magnetic single domain particles. Direct
studies on spin-dependent scattering emerged from the first
realizations of magnetic nanostructures and gave rise to the
discovery of spin injection.
4,5
Studies of the exchange field
coupling of a set of magnetic layers has brought to the dis-
covery of giant magnetoresistance, a new property of spin-
dependent electrical transport.
6–8
Tunnel,
9,10
ballistic
magnetoresistance,
11
and domain wall scattering
12
are also
concerned with the effect of a magnetic configuration on the
conduction electrons. Recently the coupling of conduction
electron spin to the exchange field was invoked to explain
the electrical resistance of domain walls: the spin of the elec-
trons follows almost but not exactly adiabatically the ex-
change field,
13
thus causing a slight spin-mixing
14
and con-
sequently an increase in resistance.
15
In the past decade or two, theoreticians pointed out the
possibility of the converse effect of the current on the mag-
netization. Namely they predicted that spin-polarized cur-
rents of the order of 10
7
to 10
8
A /cm
2
excite spin
waves
16–19
or propagate domain walls.
20
Recently it was also
suggested that currents could be switching magnetic domains
by longitudinal relaxation of the spin of the electrons.
21
In
spin valves injection into a magnet after spin polarization in
a spin polarizer it is thought that the injection of spins gen-
erates a torque,
22–27
an effective exchange interaction due to
longitudinal spin accumulation,
28
or both torque and effec-
tive field due to transverse spin accumulation.
29
From the
experimental point of view, first Berger et al.
30,31
have evi-
denced the action of a high current density on domain walls
in thin films. Recently Garcia et al. provoked domains shifts
causing magnetoresistance changes up to 300% in nickel
nanocontacts by injection of ballistic electrons.
32,33
Tsoi
et al. first inferred from I -V measurements that strong cur-
rents through point contacts into macroscopic Co/Cu multi-
layered thin films indeed excited spin waves,
34
later deter-
mined their high frequency nature, and suggested a
transverse polarization configuration of the waves.
35
Further,
magnon excitation in Fe/Cr/Fe trilayer films were directly
observed by Rezende et al.
36
in the dynamic response of a
multilayer, allowing to distinguish between the effects of
spin injection and Oersted field on the magnetization. Theeu-
wen et al.
37
measured a reduction of the GMR ratio of a
trilayer when traversed by an intense current in one sense but
not if reversed, as well as the appearance of distinct high
resistance GMR plateaus dependent on bias polarity and the
sense of field sweep, possibly accounted for in part by the
generation of incoherent magnons. Sun
38
attributed to mo-
mentum transfer onto ferromagnetic clusters the sudden re-
sistance change in manganite trilayers. Ralph et al. observed
PHYSICAL REVIEW B 68, 134425 2003
0163-1829/2003/6813/13442513/$20.00 ©2003 The American Physical Society 68 134425-1