You are on page 1of 14

REVIEWS OF MODERN PHYSICS, VOLUME 80, OCTOBER–DECEMBER 2008

Nobel Lecture: Origin, development, and future of spintronics*


Albert Fert
Unité Mixte de Physique CNRS/Thales, 91767, Palaiseau, France
and Université Paris-Sud, 91405, Orsay, France
共Published 17 December 2008兲

DOI: 10.1103/RevModPhys.80.1517

OVERVIEW understood from the typical band structure of a ferro-


magnetic metal shown in Fig. 1共a兲. The splitting between
Electrons have a charge and a spin, but until recently, the energies of the “majority spin” and “minority spin”
charges and spins have been considered separately. In directions 共spin up and spin down in the usual notation兲
conventional electronics, the charges are manipulated by makes that the electrons at the Fermi level, which carry
electric fields but the spins are ignored. Other classical the electrical current, are in different states for opposite
technologies, magnetic recording, for example, are using spin directions and exhibit different conduction proper-
the spin but only through its macroscopic manifestation, ties. This spin-dependent conduction was proposed by
the magnetization of a ferromagnet. This picture started Mott 共1936兲 to explain some features of the resistivity of
to change in 1988 when the discovery 共Baibich et al., ferromagnetic metals at the Curie temperature. How-
1988; Binash et al., 1989兲 of the giant magnetoresistance ever, in 1966, when I started my Ph.D. thesis, the subject
共GMR兲 of the magnetic multilayers opened the way to was still almost completely unexplored. My supervisor,
an efficient control of the motion of the electrons by Ian Campbell, proposed that I investigate it with experi-
acting on their spin through the orientation of a magne- ments on Ni- and Fe-based alloys and I had the privilege
tization. This rapidly triggered the development of a
to be at the beginning of the study of this topic. I could
new field of research and technology, today called spin-
confirm that the mobility of the electrons was spin de-
tronics and, like the GMR, exploiting the influence of
pendent and, in particular, I showed that the resistivities
the spin on the mobility of the electrons in ferromag-
of the two channels can be very different in metals
netic materials. Actually, the influence of the spin on the
doped with impurities presenting a strongly spin-
mobility of the electrons in ferromagnetic metals, first
dependent scattering cross section 共Fert and Campbell,
suggested by Mott 共1936兲, had been experimentally dem-
1968, 1971, 1976兲. In Fig. 1共b兲, I show the example of the
onstrated and theoretically described in my Ph.D. thesis
spin up 共majority spin兲 and spin down 共minority spin兲
almost 20 years before the discovery of 1988. The GMR
was the first step on the road of the exploitation of this resistivities of nickel doped with 1% of different types of
influence to control an electrical current. Its application impurities. It can be seen that the ratio ␣ of the spin
to the read heads of hard disks greatly contributed to down resistivity to the spin up one can be as large as 20
the fast rise in the density of stored information and led for Co impurities or, as well, smaller than 1 for Cr or V
to the extension of the hard disk technology to consum- impurities, consistent with the theoretical models devel-
er’s electronics. Then, the development of spintronics re- oped by Jacques Friedel for the electronic structures of
vealed many other phenomena related to the control these impurities. The two-current conduction was rap-
and manipulation of spin currents. Today this field of idly confirmed by other groups and, for example, ex-
research is expanding considerably, with very promising tended to Co-based alloys by Loegel and Gautier 共1971兲
new axes like the phenomena of spin transfer, spintron- in Strasbourg.
ics with semiconductors, molecular spintronics, or In my thesis, I also worked out the so-called two-
single-electron spintronics. current model 共Fert and Campbell, 1968, 1971, 1976兲 for
the conduction in ferromagnetic metals. This model is
based on a picture of spin up and spin down currents
FROM SPIN-DEPENDENT CONDUCTION IN coupled by spin mixing, i.e. by momentum exchange.
FERROMAGNETS TO GIANT MAGNETORESISTANCE Spin mixing comes from momentum exchange between
the two channels by spin-flip scattering, mainly from
GMR and spintronics take their roots from previous
electron-magnon scattering which increases with tem-
research on the influence of the spin on the electrical
perature and equalizes partly the spin up and spin down
conduction in ferromagnetic metals 共Mott, 1936; Fert
currents at room temperature 共the degree of equaliza-
and Campbell, 1968, 1971, 1976; Loegel and Gautier,
1971兲. The spin dependence of the conduction can be tion depends on the ratio between the “spin mixing re-
sistivity” and the resistivity兲. The two-current model is
the basis of spintronics today, but, surprisingly, the inter-
*The 2007 Nobel Prize for Physics was shared by Albert Fert pretation of the spintronics phenomena is generally
and Peter Grünberg. This paper is the text of the address given based on a simplified version of the model neglecting
in conjunction of the award. spin mixing and assuming that the conduction by two

0034-6861/2007/80共4兲/1517共14兲 1517 ©The Nobel Foundation, 2007


1518 Albert Fert: Nobel Lecture: Origin, development and …

(a)
α ≈ 0.3
(b)

α ≈ 20
(c)

FIG. 1. 共Color兲 Basics of spintronics. 共a兲 Schematic band structure of a ferromagnetic metal showing the energy band spin splitting.
共b兲 Resistivities of the spin up and spin down conduction channels for nickel doped with 1% of several types of impurities
共measurements at 4.2 K兲 共Fert and Campbell, 1968, 1971, 1976兲. The ratio ␣ between the resistivities ␳0↓ and ␳0↑ of the spin ↓ and
spin ↑ channels can be as large as 20 共Co impurities兲 or, as well, smaller than 1 共Cr or V impurities兲. 共c兲 Schematic for spin-
dependent conduction through independent spin ↓ and spin ↑ channels in the limit of negligible spin mixing 关␳↑↓ = 0 in the
formalism of Fert and Campbell 共1968, 1971, 1976兲兴.

independent channels is parallel, as illustrated by Fig. port equations tell us that the relative orientation of lay-
1共c兲. It should be certainly useful to revisit the interpre- ers A and B can be felt by the electrons only if their
tation of many recent experiments by taking into ac- distance is smaller than the electron mean-free path,
count the spin mixing contributions. Note that spin mix-
ing, i.e., momentum exchange between the two channels (a) ρ↓A ρ↓B
by electron-magnon collisions 共electron-electron scatter- αA = > 1 αB = <1
ρA↑ ρB↑
ing inside the electron system兲, should not be confused
with the spin-lattice mechanism relaxing the spin accu- spin
mulation 共the relaxation of the spin accumulation to the
lattice comes from spin-orbit without direct contribution
from electron-magnon scattering兲. spin
As a matter of fact, some experiments of my thesis ρAB >> ρA+ ρB Type # 1
with metals doped with two types of impurities 共Fert and
Campbell, 1968, 1971, 1976兲 were already anticipating
the GMR. This is illustrated by Fig. 2. Suppose, for ex-
(b) αA > 1 αB > 1
ample, that nickel is doped with impurities of Co which
scatter strongly the electrons of the spin down channel spin
and with impurities of rhodium which scatter strongly
the spin up electrons. In the ternary alloy Ni共Co+ Rh兲,
spin
that I call type No. 1, the electrons of both channels are
strongly scattered either by Co or by Rh, so that the ρAB ≈ ρA+ ρB Type # 2
resistivity is strongly enhanced. In contrast, there is no
such enhancement in alloys of type No. 2 doped with
impurities 共Co and Au, for example兲 scattering strongly FIG. 2. 共Color兲 Experiments on ternary alloys based on the
the electrons in the same channel and leaving the second same concept as that of the GMR 共Fert and Campbell, 1968,
1971, 1976兲. 共a兲 Schematic for the spin-dependent conduction
channel open. The idea of GMR is the replacement of
in alloys doped with impurities of opposite scattering spin
the impurities A and B of the ternary alloy by magnetic
asymmetries 共␣A = ␳A↓ / ␳A↑ ⬎ 1, ␣B = ␳B↓ / ␳B↑ ⬍ 1, ␳AB Ⰷ ␳A + ␳B兲
layers A and B in a multilayer, the antiparallel magnetic and experimental results for Ni共Co1−xRhx兲 alloys of total con-
configuration of the layers A and B corresponding to the centration 1%. 共b兲 Same for alloys doped with impurities of
situation of an alloy of type No. 1, while the configura- similar scattering spin asymmetries 共␣A = ␳A↓ / ␳A↑ ⬎ 1, ␣B
tion with a parallel configuration corresponds to type = ␳B↓ / ␳B↑ ⬎ 1, ␳AB ⬇ ␳A + ␳B兲 and experimental results for
No. 2. This brings the possibility of switching between Ni共Au1−xCox兲 alloys of total concentration 1%. In GMR the
high and low resistivity states by simply changing the impurities A and B are replaced by magnetic layers, the situ-
relative orientation of the magnetizations of layers A ation of 共a兲 关共b兲兴 corresponding to the antiparallel 关parallel兴
and B from antiparallel to parallel. However, the trans- magnetic configurations of adjacent magnetic layers.

Rev. Mod. Phys., Vol. 80, No. 4, October–December 2008


Albert Fert: Nobel Lecture: Origin, development and … 1519

that is, practically, if they are spaced by only a few nm. Agnés Barthélémy and Frédéric Petroff, as well as in the
Unfortunately, in the 1970s, it was not technically pos- small but rapidly increasing community working in the
sible to make multilayers with layers as thin as a few nm. field, we had the exalting impression of exploring a wide
I put some of my ideas in the fridge and, in my team at virgin country with so many amazing surprises in store.
the Laboratoire de Physique des Solides of the Univer- On the experimental side, two important results were
sité Paris-Sud, from the beginning of the 1970s to 1985 I published in 1990. Parkin et al. 共1990兲 demonstrated the
worked on other topics like the extraordinary Hall ef- existence of GMR in multilayers made by the simpler
fect, the spin Hall effect, the magnetism of spin glasses, and faster technique of sputtering 共Fe/ Cr, Co/ Ru, and
and amorphous materials. Co/ Cr兲, and found the oscillatory behavior of the GMR
In the mid 1980s, with the development of techniques due to the oscillations of the interlayer exchange as a
like molecular beam epitaxy 共MBE兲, it became possible function of the thickness of the nonmagnetic layers.
to fabricate multilayers composed of very thin individual Also in 1990 Shinjo and Yamamoto 共1990兲, as well as
layers and I could consider trying to extend my experi- Dupas et al. 共1990兲, demonstrated that GMR effects can
ments on ternary alloys to multilayers. In addition, in be found in multilayers without antiferromagnetic inter-
1986, I saw the beautiful Brillouin scattering experi- layer coupling but composed of magnetic layers of dif-
ments of Grünberg et al. 共1986兲 revealing the existence ferent coercivities. Another important result, in 1991,
of antiferromagnetic interlayer exchange couplings in was the observation of large and oscillatory GMR ef-
Fe/ Cr multilayers. Fe/ Cr appeared as a magnetic multi- fects in Co/ Cu, which became the archetypical GMR
layered system in which it was possible to switch the system 关Fig. 4共a兲兴. The first observations 共Mosca et al.,
relative orientation of the magnetization in adjacent 1991兲 were obtained in my group by my Ph.D. student
magnetic layers from antiparallel to parallel by applying Dante Mosca with multilayers prepared by sputtering at
a magnetic field. In collaboration with the group of Michigan State University and at about the same time in
Alain Friederich at the Thomson-CSF company, I the group of Stuart Parkin at IBM 共Parkin et al., 1991兲.
started the fabrication and investigation of Fe/ Cr multi- Also in 1991, Dieny et al. 共1991兲 reported the first obser-
layers. The MBE expert at Thomson-CSF was Patrick vation of GMR in spin valves, i.e., trilayered structures
Etienne, and my three Ph.D. students, Frédéric Nguyen in which the magnetization of one of the two magnetic
Van Dau first and then Agnès Barthélémy and Frédéric layers is pinned by coupling with an antiferromagnetic
Petroff, were also involved in the project. This led us in layer while the magnetization of the second one is free.
1988 to the discovery 共Baibich et al., 1988兲 of very large The magnetization of the free layer can be reversed by
magnetoresistance effects that we called GMR 关Fig. very small magnetic fields, so that the concept is now
3共a兲兴. Effects of the same type in Fe/ Cr/ Fe trilayers used in most applications.
were obtained practically at the same time by Grünberg Other developments of the research on magnetic mul-
at Jülich 共Binash et al., 1989兲 关Fig. 3共b兲兴. The interpreta- tilayers and GMR at the beginning of the 1970s are de-
tion of the GMR is similar to that described above for scribed in the Nobel lecture of my co-laureate Peter
the ternary alloys and is illustrated by Fig. 3共c兲. The first Grünberg, with, in particular, a presentation of the vari-
classical model of the GMR was published by Camley ous devices based on the GMR of spin-valve structures
and Barnas 共1989兲 and I collaborated with Levy and 共Parkin, 2002; Chappert et al., 2007兲. In the read heads
Zhang for the first quantum model 共Levy et al., 1990兲 in 共Fig. 5兲 of the hard disk drives 共HDDs兲, the GMR sen-
1991. sors based on spin valves have replaced the anisotropic
I am often asked if I was expecting such large MR magnetoresistance 共AMR兲 sensors in 1997. The GMR,
effects. My answer is yes and no: on the one hand, a very by providing a sensitive and scalable read technique,
large magnetoresistance could be expected from an ex- has led to an increase of the areal recording density by
trapolation of my preceding results on ternary alloys, on more than two orders of magnitude 共from ⬇1 to
the other hand one could fear that the unavoidable ⬇ 600 Gbit/ in.2 in 2007兲. This increase opened the way
structural defects of the multilayers, interface roughness, both to unprecedented drive capacities 共up to 1 ter-
for example, might introduce spin-independent scatter- abyte兲 for video recording or backup and to smaller
ings canceling the spin-dependent scattering inside the HDD sizes 共down to 0.85-in. disk diameter兲 for mobile
magnetic layers. The good luck was finally that the scat- appliances like ultralight laptops or portable multimedia
tering by the roughness of the interfaces is also spin de- players. GMR sensors are also used in many other types
pendent and adds its contribution to the “bulk” one 共the of applications, mainly in the automotive industry and
bulk and interface contributions can be separately de- biomedical technology 共Freitas et al., 2003兲.
rived from CPP-GMR experiments兲.
CPP-GMR AND SPIN ACCUMULATION PHYSICS
THE GOLDEN AGE OF GMR
During the first years of the research on GMR, the
Rapidly, our papers reporting the discovery of GMR experiments were performed only with currents flowing
attracted attention for their fundamental interest as well along the layer planes, in the geometry one calls CIP
as for the many possibilities of applications, and the re- 共current in plane兲. It is only in 1993 that experiments of
search on magnetic multilayers and GMR became a very CPP-GMR began to be performed, that is experiments
hot topic. In my team, reinforced by the recruitment of of GMR with the current perpendicular to the layer

Rev. Mod. Phys., Vol. 80, No. 4, October–December 2008


1520 Albert Fert: Nobel Lecture: Origin, development and …

≈ 80%

(c)
(b)

FIG. 3. 共Color兲 First observations of giant magnetoresistance. 共a兲 Fe/ Cr共001兲 multilayers 共Baibich et al., 1988兲 关with the current
definition of the magnetoresistance ratio MR= 100共RAP − RP兲 / Rp, MR= 85% for the 共Fe 3 nm/ Cr 0.9 nm兲 multilayer兴. 共b兲
Fe/ Cr/ Fe trilayers 共Binash et al., 1989兲. 共c兲 Schematic of the mechanism of the GMR. In the parallel magnetic configuration
共bottom兲, the electrons of one of the spin directions can go easily through all the magnetic layers and the short circuit through this
channel leads to a small resistance. In the antiparallel configuration 共top兲, the electrons of each channel are slowed down every
second magnetic layer and the resistance is high. From Chappert et al., 2007.

planes. This was done first by sandwiching a magnetic definitely higher than in CIP 共the CPP-GMR will prob-
multilayer between superconducting electrodes by Bass, ably be used in a future generation of read heads for
Pratt, and Shroeder at Michigan State University 共Pratt hard disks兲, but also subsists in multilayers with rela-
et al., 1991; Bass and Pratt, 1999兲, and, a couple of years tively thick layers, up to the micron range 共Piraux et al.,
after, in a collaboration of my group with Luc Piraux at 1994; Fert and Piraux, 1999兲, as it can be seen in Figs.
the University of Louvain, by electrodepositing the 4共c兲 and 4共d兲. In a theoretical paper with Thierry Valet
multilayer into the pores of a polycarbonate membrane 共Valet and Fert, 1993兲, I showed that, owing to spin ac-
共Piraux et al., 1994; Fert and Piraux, 1999兲 关Figs. cumulation effects occurring in the CPP geometry, the
4共b兲–4共d兲兴. In the CPP geometry, the GMR is not only length scale of the spin transport becomes the long spin

Rev. Mod. Phys., Vol. 80, No. 4, October–December 2008


Albert Fert: Nobel Lecture: Origin, development and … 1521

CPP-GMR (b)

6 nm

(a)

60 nm
10

8
CIP-GMR
MR ratio (%) 6

4 400 nm

2
(c) (d)
0
0 100 200 300 400 500
Co thickness (nm)
FIG. 4. 共Color兲 共a兲 Variation of the GMR ratio of Co/ Cu multilayers in the conventional current in plane 共CIP兲 geometry as a
function of the thickness of the Cu layers 共Mosca et al., 1991兲. The scaling length of the variation is the mean-free path 共short兲. 共b兲
Structure of multilayered nanowires used for CPP-GMR measurements. 共c兲 CPP-GMR curves at 77 K for 共Permalloy
12 nm/copper 4 nm兲 multilayered nanowires 共solid lines兲 and 共cobalt 10 nm/copper 5 nm兲 multilayered nanowires 共dotted lines兲
共Piraux et al., 1994; Fert and Piraux, 1999兲. 共d兲 Variation of the CPP-GMR ratio of Co/ Cu multilayered nanowires as a function of
the thickness of the Co layers 共Piraux et al., 1994; Fert and Piraux, 1999兲. The scaling length of the variation is the spin diffusion
length 共long兲. The inset shows the curves for Permally/copper at 4.2 K.

diffusion length in place of the short mean-free path for the figure兲, while, far from the interface on the other
the CIP geometry. Actually, the CPP-GMR has revealed side, it is equally distributed in the two channels. With
the spin accumulation effects which govern the propaga- the current direction and the spin polarization of the
tion of a spin-polarized current through a succession of figure, there is accumulation of spin up electrons 共and
magnetic and nonmagnetic materials and plays an im- depletion of spin down for charge neutrality兲 around the
portant role in all the current developments of spintron- interface, or, in other words, a splitting between the
ics. The diffusion current induced by the accumulation Fermi energies 共chemical potentials兲 of the spin up and
of spins at the magnetic-nonmagnetic interface is the spin down electrons. This accumulation diffuses from
mechanism driving a spin-polarized current at a long dis- the interface in both directions to the distance of the
tance from the interface, well beyond the ballistic range SDL. Spin flips are also generated by this out of equilib-
共i.e., well beyond the mean-free path兲 up to the distance rium distribution and a steady splitting is reached when
of the spin diffusion length 共SDL兲. In carbon molecules, the number of spin flips is just what is needed to adjust
for example, the spin diffusion length exceeds the mi- the incoming and outgoing fluxes of spin up and spin
cron range and, as we will see in the section on molecu- down electrons. To sum up, there is a broad zone of spin
lar spintronics, strongly spin-polarized currents can be accumulation which extends on both sides to the dis-
transported throughout long carbon nanotubes. tance of the SDL and in which the current is progres-
The physics of the spin accumulation occurring when sively depolarized by the spin flips generated by the spin
an electron flux crosses the interface between a ferro- accumulation.
magnetic and a nonmagnetic material is explained in Figure 6 is drawn for the case of spin injection, i.e., for
Fig. 6. Far from the interface on the magnetic side, the electrons going from the magnetic to the nonmagnetic
current is larger in one of the spin channels 共spin up in conductor. For electrons going in the opposite direction

Rev. Mod. Phys., Vol. 80, No. 4, October–December 2008


1522 Albert Fert: Nobel Lecture: Origin, development and …

FIG. 5. 共Color兲 GMR head for


hard disk. From Chappert et al.,
2007.

FIG. 6. 共Color兲 Schematic rep-


resentation of the spin accumu-
lation at an interface between
a ferromagnetic metal and a
nonmagnetic layer. 共a兲 Spin up
and spin down currents far
from an interface between fer-
romagnetic and nonmagnetic
conductors 共outside the spin-
accumulation zone兲. 共b兲 Split-
ting of the chemical potentials
EF↑ and EF↓ at the interface.
The arrows symbolize the spin
flips induced by the spin-split
out of equilibrium distribution.
These spin flips control the pro-
gressive depolarization of the
electron current between the
left and the right. With an op-
posite direction of the current,
there is an inversion of the spin
accumulation and opposite spin
flips, which polarizes the cur-
rent when it goes through the
spin-accumulation zone. 共c兲
Variation of the current spin
polarization when there is an
approximate balance between
the spin flips on both sides
共metal/metal兲 and when the
spin flips on the left side are
predominant 共metal/semicon-
ductor without spin-dependent
interface resistance, for ex-
ample兲. From Chappert et al.,
2007.

Rev. Mod. Phys., Vol. 80, No. 4, October–December 2008


Albert Fert: Nobel Lecture: Origin, development and … 1523

(a) (b)

FIG. 7. 共Color兲 共a兲 Principle of the magnetic random access memory 共MRAM兲 in the basic “cross point” architecture. The binary
information “0” and “1” is recorded on the two opposite orientations of the magnetization of the free layer of magnetic tunnel
junctions 共MTJs兲, which are connected to the crossing points of two perpendicular arrays of parallel conducting lines. For writing,
current pulses are sent through one line of each array, and only at the crossing point of these lines the resulting magnetic field is
high enough to orient the magnetization of the free layer. For reading, one measures the resistance between the two lines
connecting the addressed cell. Schematic from Chappert et al., 2007. 共b兲 High magnetoresistance, TMR= 共Rmax − Rmin兲 / Rmin, mea-
sured by Lee et al. 共2007兲 for the magnetic stack: 共Co25Fe75兲80B20共4 nm兲 / MgO共2.1 nm兲 / 共Co25Fe75兲80B20共4.3 nm兲 annealed at 475 ° C
after growth, measured at room temperature 共closed circles兲 and low temperature 共open circles兲.

共spin extraction兲, the situation is similar except that a interface, increase the proportion of spin on the semi-
spin accumulation in the opposite direction progres- conductor side, and shift the depolarization from the
sively polarizes the current in the nonmagnetic conduc- metallic to the semiconductor side 共the same conclusions
tor. In both the injection and extraction cases, the spin appear also in a paper by Rashba兲 共Rashba, 2000; Fert
polarization subsists or starts in the nonmagnetic con- and Jaffrès, 2001兲. Spin injection through a tunnel bar-
ductor at a long distance from the interface. This physics rier has now been achieved successfully in several ex-
can be described by new types of transport equations periments but the tunnel resistances are generally too
共Valet and Fert, 1993兲 in which the electrical potential is large for an efficient transformation of the spin informa-
replaced by a spin- and position-dependent electro- tion into an electrical signal 共Fert et al., 2007兲.
chemical potential. These equations can be applied not
only to the simple case of a single interface but to a
multi-interface system with overlap of the spin accumu- MAGNETIC TUNNEL JUNCTIONS AND TUNNELING
lations at successive interfaces. They can also be ex- MAGNETORESISTANCE (TMR)
tended to take into account band bending and high cur-
rent density effects 共Yu and Flatté, 2002; Fert et al., An important stage in the development of spintronics
2007兲. has been the research on the tunneling magnetoresis-
The physics of spin accumulation play an important tance 共TMR兲 of the magnetic tunnel junctions 共MTJ兲.
role in many fields of spintronics, for example, in one of The MTJs are tunnel junctions with ferromagnetic elec-
the most active fields of research today, spintronics with trodes and their resistance is different for the parallel
semiconductors. In the case of spin injection from a and antiparallel magnetic configurations of their elec-
magnetic metal into a nonmagnetic semiconductor 共or trodes. Some early observations of TMR effects, small
spin extraction for the opposite current direction兲, the and at low temperature, had been already reported by
much larger density of states in the metal makes that Jullière 共1975兲, but they were not easily reproducible
similar spin accumulation splittings on the two sides of and actually could not be really reproduced for 20 years.
the interface, as shown in Fig. 6, lead to a much larger It is only in 1995 that large 共⬇20% 兲 and reproducible
spin accumulation density and to a much larger number effects were obtained by Moodera and Miyasaki’s
of spin flips on the metallic side. The depolarization is groups on MTJ with a tunnel barrier of amorphous alu-
therefore faster on the metallic side and the current is mina 共Miyazaki and Tezuka, 1995; Moodera et al., 1995兲.
almost completely depolarized when it enters the semi- From a technological point of view, the interest of the
conductor, as shown in Fig. 6共c兲. This problem was first MTJ with respect to the metallic spin valves comes from
raised by Schmidt et al. 共2000兲. I came back to the theory the vertical direction of the current and from the result-
with my co-worker Henri Jaffrès to show that the prob- ing possibility of a reduction of the lateral size to a sub-
lem can be solved by introducing a spin-dependent in- micronic scale by lithographic techniques. The MTJs are
terface resistance, typically a tunnel junction, to intro- at the basis of a new concept of magnetic memory called
duce a discontinuity of the spin accumulation at the MRAM 共magnetic random access memory兲 and are

Rev. Mod. Phys., Vol. 80, No. 4, October–December 2008


1524 Albert Fert: Nobel Lecture: Origin, development and …

schematically represented in Fig. 7共a兲. The MRAMs are


expected to combine the short access time of the
semiconductor-based RAMs and the nonvolatile charac-
ter of the magnetic memories. In the first MRAMs, put
onto the market in 2006, the memory cells are MTJs P
with an alumina barrier. The magnetic fields generated
by “word” and “bit” lines are used to switch their mag-
netic configuration, see Fig. 7共a兲. The next generation of
MRAM, called ST-RAM, based on MgO tunnel junc-
tions and switching by spin transfer, is expected to have
a much stronger impact on the technology of computers.
The research on the TMR has been very active since
1995 and the most important step was the recent transi-
tion from MTJ with an amorphous tunnel barrier 共alu-
mina兲 to single crystal MTJ and especially MTJ with a
MgO barrier. In the CNRS/Thales laboratory we
founded in 1995, the research on TMR was one of our
main projects and, in collaboration with a Spanish
group, we obtained one of the very first results 共Bowen
AP
et al., 2001兲 on MTJ with epitaxial MgO. However, our
TMR was only slightly larger than that found with alu-
mina barriers and similar electrodes. The important
breakthrough came in 2004 at Tsukuba 共Yuasa et al.,
2004兲 and IBM 共Parkin et al., 2004兲 where it was found
that very large TMR ratios, up to 200% at room tem-
perature, could be obtained from MgO MTJ of very
high structural quality. TMR ratios of about 600% have
been now reached 共Lee et al., 2007兲 关Fig. 7共b兲兴. In such
MTJ, the single crystal barrier filters the symmetry of
the wave functions of the tunneling electrons 共Mathon
and Umerski, 1999; Mavropoulos et al., 2000; Zhang and FIG. 8. 共Color兲 Physics of TMR illustrated by the decay of
Butler, 2004兲, so that the TMR depends on the spin po- evanescent electronic waves of different symmetries in a
larization of the electrodes for the selected symmetry. MgO共001兲 layer between cobalt electrodes calculated by
The high spin polarization obtained by selecting the Zhang and Butler 共2004兲. The ⌬1 symmetry of the slowly de-
symmetry of the tunneling waves with a single crystal caying tunneling channel is well represented at the Fermi level
barrier is a very good illustration of what is under the of the spin conduction band of cobalt for the majority spin
word “spin polarization” in a spintronic experiment. In direction and not for the minority spin one, so that a good
the example of Fig. 8, taken from Zhang and Butler connection by tunneling between the electrodes exists only for
共2004兲, one sees the density of states of evanescent the parallel magnetic configuration when a ⌬1 channel can be
connected to both electrodes 共above兲. In the antiparallel con-
waves functions of different symmetries, ⌬1, ⌬5, etc., in a
figuration 共below兲, both the spin up and spin down ⌬1 channels
MgO共001兲 barrier between Co electrodes. The key point are poorly connected on one of the sides. This explains the
is that, at least for interfaces of high quality, an evanes- very high TMR of this type of junction.
cent wave function of a given symmetry is connected to
the Bloch functions of the same symmetry at the Fermi
level of the electrodes. For Co electrodes, the ⌬1 sym- of a given magnetic conductor in a MTJ depends on the
metry is well represented at the Fermi level in the ma- symmetry selected by the barrier and, depending on the
jority spin direction subband and not in the minority barrier, can be positive or negative, large or small. In the
one. Consequently, a good connection of the slowly de- same way the spin polarization of metallic conduction
caying channel ⌬1 with both electrodes can be obtained depends strongly on the spin dependence of the scatter-
only in their parallel magnetic configuration, which ex- ing by impurities, as illustrated by Fig. 1共b兲.
plains the very high TMR. Other types of barrier can There are other promising directions to obtain large
select other symmetries than the symmetry ⌬1 selected TMR and experiments in several of them are now led by
by MgO共001兲. For example, a SrTiO3 barrier predomi- Agnès Barthélémy 共much more than by myself兲 in the
nantly selects evanescent wave functions of ⌬5 symme- CNRS/Thales laboratory. First, we tested ferromagnetic
try, which are connected to minority spin states of cobalt materials which were predicted to be half metallic, i.e.,
共Velev et al., 2005; Bowen et al., 2006兲. This explains the metallic for one of the spin directions and insulating for
negative effective spin polarization of cobalt we had ob- the other one, in other words 100% spin polarized. Very
served in SrTiO3-based MTJ 共De Teresa et al., 1999兲. high spin polarization 共95%兲 and record TMR 共1800%兲
This finally shows that there is no intrinsic spin polariza- have been obtained by our Ph.D. student Martin Bowen
tion of a magnetic conductor. The effective polarization with La2/3Sr1/3MnO3 electrodes 共Bowen et al., 2003兲 but

Rev. Mod. Phys., Vol. 80, No. 4, October–December 2008


Albert Fert: Nobel Lecture: Origin, development and … 1525

Au
LBMO
10mV 3K 1 LSMO
180
1
160
After +2V
R( )

3 2
140
2
120 After -2V
3
4
100

-4 -2 0 2 4 4
H (kOe)
FIG. 9. 共Color兲 Four state resistance of a tunnel junction com-
posed of a biferroic tunnel barrier 共La0.1Bi0.9MnO3兲 between a
ferromagnetic electrode of La2/3Sr1/3MnO3 and a nonmagnetic
electrode of gold. The states 1–4 correspond to the magnetic
共white arrows兲 and electric 共black arrows兲 polarizations repre-
sented on the right side of the figure. From Gajek et al., 2007. FIG. 10. 共Color兲 Illustration of the spin-transfer concept intro-
duced by Slonczewski 共1996兲. A spin-polarized current is pre-
pared by a first magnetic layer F with an obliquely oriented
the Curie temperature of this manganite 共around 350 K兲
spin polarization with respect to the magnetization axis of a
is too low for applications. It now turns out from recent
second layer F2. When this current goes through F2, the ex-
results in Japan 共Ishikawa et al., 2006兲 that ferromagnets change interaction aligns its spin polarization along the mag-
of the family of the Heusler alloys also present very netization axis. As the exchange interaction is spin conserving,
large TMR ratios with still 90% at room temperature the transverse spin polarization lost by the current has been
共Ishikawa et al., 2006兲. Another interesting concept that transferred to the total spin of F2, which can also be described
we are exploring is spin filtering by tunneling through a by a spin-transfer torque acting on F2. This can lead to a mag-
ferromagnetic insulator layer 共Leclair et al., 2002; Ramos netic switching of the F2 layer or, depending on the experimen-
et al., 2007兲. This can be described as the tunneling of tal conditions, to magnetic oscillations in the microwave fre-
electrons through a barrier of spin-dependent height if quency range. From Chappert et al., 2007.
the bottom of the conduction band is spin split, which
gives rise to a spin dependence of the transmission prob- a torque acting on its magnetic moment. This torque can
ability 共spin filtering兲. Very high spin filtering coefficients induce an irreversible switching of this magnetic mo-
have been found at low temperature with Eus barriers ment or, in a second regime, generally in the presence of
共Leclair et al., 2002兲 at MIT and at Eindhoven. Promis- an applied field, it generates precessions of the moment
ing results with insulating ferromagnets of much higher in the microwave frequency range.
Curie temperature have been recently obtained, see, for The first evidence that spin transfer can work was in-
example, Ramos et al. 共2007兲. Some of the magnetic bar- dicated by experiments of spin injection through point
riers we have recently tested in MTJ are also ferroelec- contacts by Tsoi et al. 共1998兲 but a clear understanding
tric, so that the MTJs present the interesting property of came later from measurements 共Katine et al., 2000; Grol-
four states of resistance corresponding to the P and AP lier et al., 2001兲 performed on pillar-shaped metallic
magnetic configurations and to the two orientations of trilayers 关Fig. 11共a兲兴. In Figs. 11共b兲 and 11共c兲, I present
the ferroelectric polarization 共Gajek et al., 2007兲, as examples of our experimental results in the low field
shown in Fig. 9. regime of irreversible switching, for a metallic pillar and
for a tunnel junction with electrodes of the ferromag-
MAGNETIC SWITCHING AND MICROWAVE netic semiconductor Ga1−xMnxAs. For metallic pillars or
GENERATION BY SPIN TRANSFER tunnel junctions with electrodes made of a dilute ferro-
magnetic transition metal like Co or Fe, the current den-
The study of the spin-transfer phenomena is one of sity needed for switching is around 106 – 107 A / cm2,
the most promising new directions in spintronics today which is still slightly too high for applications, and an
and also an important research topic in our CNRS/ important challenge is the reduction of this current den-
Thales laboratory. In spin-transfer experiments, one ma- sity. The switching time has been measured in other
nipulates the magnetic moment of a ferromagnetic body groups and can be as short as 100 ps, which is very at-
without applying any magnetic field but only by transfer tractive for the switching of MRAM. For the tunnel
of spin angular momentum from a spin-polarized cur- junction of Fig. 11共c兲, the switching current is only about
rent. The concept, which has been introduced by Slonc- 105 A / cm2 and smaller than that of the metallic pillar by
zewski 共1996兲 and appears also in papers by Berger two orders of magnitude. This is because a smaller num-
共1996兲, is illustrated in Fig. 10. As described in the cap- ber of individual spins is required to switch the smaller
tion of the figure, the transfer of a transverse spin cur- total spin momentum of a dilute magnetic material.
rent to the “free” magnetic layer F2 can be described by In the presence of a large enough magnetic field, the

Rev. Mod. Phys., Vol. 80, No. 4, October–December 2008


1526 Albert Fert: Nobel Lecture: Origin, development and …

(a) 14,6
(b)

dV/dI( )
Free Co layer
Cu spacer 14,5
Fixed Co layer

14,4
-2 0
I (mA)

Power (pW/GHz)
550000 (c) (d)
( )

3
Resistance

500000
2
450000
1
400000
0
5
-1.0x10 4
-5.0x10 0.0 4
5.0x10 5
1.0x10 3,5 4,0
Current density (A.cm-2 ) Frequency (GHz)
FIG. 11. 共Color兲 Experiments of magnetic switching and microwave generation induced by spin transfer from an electrical dc
current in trilayered magnetic pillars. 共a兲 Schematic of a trilayered magnetic pillar. 共b兲 Switching by spin transfer between the
parallel and antiparallel magnetic configurations of a Co/ Cu/ Co metallic pillar 共Grollier et al., 2001; Grollier, 2003兲. The switching
between parallel and antiparallel orientations of the magnetizations of the two magnetic layers of the trilayer is detected by
irreversible jumps of the resistance at a critical value of the current. The critical current density is of the order of 107 A / cm2. 共c兲
Switching by spin transfer of a pillar-shaped tunnel junction composed of electrodes of the dilute ferromagnetic semiconductor
GaMnAs separated by a tunnel barrier of InGaAs 共Elsen et al., 2006兲. The critical current is about 100 times smaller than in the
Py/Cu/Py pillar. Similar results have been obtained by Hayakawa et al. 共2005兲. 共d兲 Typical microwave power spectrum of a
Co/ Cu/ Py pillar 共Py= Permalloy兲 共Boulle, 2006; Boulle et al., 2007兲.

regime of irreversible switching of the magnetization of portant applications. Switching by spin transfer will be
the “free” magnetic layer in a trilayer is replaced by used in the next generation of MRAM 共ST-RAM兲 and
a regime of steady precessions of this free layer mag- will bring great advantages in terms of precise address-
netization sustained by the spin-transfer torque 共Rippart ing and low energy consumption. The generation of os-
et al., 2004兲. As the angle between the magnetizations of cillations in the microwave frequency range will lead to
the two magnetic layers varies periodically during the the design of spin-transfer oscillators 共STOs兲. One of the
precession, the resistance of the trilayer oscillates as a main interests of the STOs is their agility, that is, the
function of time, which generates voltage oscillations in possibility of changing rapidly their frequency by tuning
the microwave frequency range. In other conditions, the a dc current. They can also have a high quality factor.
spin-transfer torque can also be used to generate an os- Their disadvantage is the very small microwave power
cillatory motion of a magnetic vortex. of an individual STO, metallic pillar, or tunnel junction.
The spin-transfer phenomena raise a series of various The solution is certainly the synchronization of a large
theoretical problems. The determination of the spin- number of STOs. The possibility of synchronization has
transfer torque is related to the solution of spin- been already demonstrated for two nanocontacts induc-
transport equations 共Kovalev et al., 2002; Slonczewski, ing spin-transfer excitations in the same magnetic layer
2002; Barnas et al., 2005; Stiles and Miltat, 2006兲, while 共Kaka et al., 2005; Mancoff et al., 2005兲. In our labora-
the description of the switching or precession of the tory we are exploring theoretically and experimentally a
magnetization raises problems of nonlinear dynamics concept of self-synchronization of a collection of electri-
共Stiles and Miltat, 2006兲. All these problems interact cally connected STOs by the rf current components they
and, for example, some of our recent results show that it induce 共Grollier et al., 2006兲.
is possible to obtain very different dynamics 共with, for
applications, the interest of oscillations without applied
field兲 by introducing strongly different spin relaxation SPINTRONICS WITH SEMICONDUCTORS AND
times in the two magnetic layers of a trilayer to distort MOLECULAR SPINTRONICS
the angular dependence of the torque 共Boulle, 2006;
Boulle et al., 2007兲. Spintronics with semiconductors 共Jonker and Flatté,
The spin-transfer phenomena will have certainly im- 2006; Awschalom and Flatté, 2007兲 is very attractive as it

Rev. Mod. Phys., Vol. 80, No. 4, October–December 2008


Albert Fert: Nobel Lecture: Origin, development and … 1527

can combine the potential of semiconductors 共control of currents in structures composed of only nonmagnetic
current by gate, coupling with optics, etc.兲 with the po- semiconductors. Actually the SHE can be also found in
tential of the magnetic materials 共control of current by nonmagnetic metals 共Vila et al., 2007; Seki et al., 2008兲
spin manipulation, nonvolatility, etc.兲. It should be pos- and the research is also very active in this field. I men-
sible, for example, to gather storage, detection, logic, tion that, already in the 1970s, I had found very large
and communication capabilities on a single chip that SHE induced by resonant scattering on spin-orbit-split
could replace several components. New concepts of levels of nonmagnetic impurities in copper 共Fert et al.,
components have also been proposed, for example, the 1981兲.
concept of spin field-effect transistors 共spin FETs兲 based Several groups have tried to probe the potential of
on spin transport in semiconductor lateral channels be- spintronics with semiconductors by validating experi-
tween spin-polarized sources and drains with control of mentally the concept of spin FET 共Datta and Das, 1990兲
the spin transmission by a field-effect gate 共Datta and described above. Both ferromagnetic metals and ferro-
Das, 1990兲. Some nonmagnetic semiconductors have a magnetic semiconductors have been used for the source
definite advantage on metals in terms of spin-coherence and the drain, but the results have been relatively poor.
time and propagation of spin polarization on long dis- In a recent review article, Jonker and Flatté 共2006兲 note
tances 共Jonker and Flatté, 2006; Awschalom and Flatté, that a contrast larger than about 1% 共i.e., 关RAP
2007兲. However, as it will be discussed below, the long − RP兴 / RP ⬎ 1%兲 has never been observed between the
standing problem of the spin FET is still far from being resistances of the parallel and antiparallel magnetic ori-
solved. entations of the source and the drain, at least for lateral
Spintronics with semiconductors is currently devel- structures. We have recently proposed 共Fert et al., 2007兲
oped along several roads. this can be understood in the models 共Fert and Jaffrès,
共i兲 The first road is by working on hybrid structures
2001兲 I had developed with Henri Jaffrès to describe the
associating ferromagnetic metals with nonmagnetic
spin transport between spin-polarized sources and
semiconductors. As this has been mentioned in the sec-
drains. In both the diffusive and ballistic regimes, a
tion on spin accumulation, Schmidt et al. 共2000兲 have
strong contrast between the conductances of the two
raised the problem of “conductivity mismatch” to inject
configurations can be obtained only if the resistances of
a spin-polarized current from a magnetic metal into a
the interfaces between the semiconductor and the
semiconductor. Solutions have been proposed by the
source or drain are not only spin dependent but also
theory 共Rashba, 2000; Fert and Jaffrès, 2001兲 and one
chosen in a relatively narrow window. The resistances
knows today that the injection 共extraction兲 of a spin-
polarized current into 共from兲 a semiconductor can be must be larger than a first threshold value for spin injec-
achieved with a spin-dependent interface resistance, tion 共extraction兲 from 共into兲 metallic source 共drain兲, and
typically a tunnel junction. Spin injection and extraction smaller than a second threshold value to keep the carrier
through a tunnel contact has been now demonstrated in dwell time shorter than the spin lifetime. For vertical
spin LEDs and magneto-optical experiments 共Stephens structures with a short distance between source and
et al., 2004; Jonker and Flatté, 2006; Awschalom and drain, the above conditions can be satisfied more easily
Flatté, 2007兲. and relatively large magnetoresistance can be observed,
共ii兲 Another road for spintronics with semiconductors as illustrated by the results I present in Fig. 12. However,
is based on the fabrication of ferromagnetic semiconduc- the results displayed in Fig. 12共c兲 show that the magne-
tors. The ferromagnetic semiconductor Ga1−xMnxAs 共x toresistance drops rapidly when the interface resistance
⬇ a few percent兲 has been discovered 共Ohno et al., 1996兲 exceeds some threshold value. This can be explained by
by the group of Ohno in Sendai in 1996 and, since this the increase of the dwell time above the spin lifetime.
time, has revealed very interesting properties, namely, Alternatively, the magnetoresistance also drops to zero
the possibility of controlling the ferromagnetic proper- when an increase of temperature shortens the spin life-
ties with a gate voltage, and also large TMR and TAMR time and increases the ratio of the dwell time to the spin
共tunneling anisotropic magnetoresistance兲 effects. How- lifetime. For most experiments on lateral structures, it
ever, its Curie temperature has reached only 170 K, well turns out that a part of the difficulties comes from too
below room temperature, which rules out most practical large interface resistances giving rise to too short dwell
applications. Several room-temperature ferromagnetic times. Min et al. 共2006兲 have arrived at similar conclu-
semiconductors have been announced but the situation sions for the particular case of silicon-based structures
is not clear on this front yet. and propose interesting solutions to lower the interface
共iii兲 The research is now very active on a third road resistances by tuning the work function of the source
exploiting spin-polarized currents induced by spin-orbit and the drain.
effects, namely, the spin Hall 共Zhang, 2000; Kato et al., A recently emerging direction is spintronics with mol-
2004; Koenig et al., 2007兲, Rashba, or Dresselhaus ef- ecules 共see Fig. 13兲. Very large GMR- or TMR-like ef-
fects. In the spin Hall effect 共SHE兲, for example, spin- fects are predicted by the theory, especially for carbon-
orbit interactions deflect the currents of the spin up and based molecules in which a very long spin lifetime is
spin down channels in opposite transverse directions, expected from the very small spin-orbit coupling. Prom-
thus inducing a transverse spin current, even in a non- ising experimental results have been published during
magnetic conductor. This could be used to create spin the last years on spin transport in carbon nanotubes

Rev. Mod. Phys., Vol. 80, No. 4, October–December 2008


1528 Albert Fert: Nobel Lecture: Origin, development and …

40
(a) (b)
30
Ga1-xMnxAs (30 nm)
FIG. 12. 共Color兲 Spintronics

TMR (%)
20 with semiconductors illustrated
GaAs (10Å) by experimental results 共Mat-
AlAs (15-19Å) 10 tana et al., 2003; Fert et al.,
2007兲 on the structure repre-
0 sented on the right side and
composed of a GaAs layer
GaAs (5-10nm) -300 -200 -100 0 100 200 300 separated from the GaMnAs
Magnetic Field (Oe) source and drain by tunnel bar-
50 riers of AlAs. 共a兲 MR curve at
(c) 4.2 K showing a resistance dif-
TMR ( % ) 40
b GaAs QW=6nm ference of 40% between the
AlAs (15-19Å)
b

T=4K parallel and antiparallel mag-


30
GaAs (10Å) netic configurations of the
Ga1-xMnxAs (300 20 source and the drain. 共b兲 MR
nm) 1.45nm 1.95nm
ratio as a function of the resis-
10 1.7nm tance of the tunnel barriers.
0
1E-3 0.01 0.1
ARAP ( .cm2 )
K)

共Cottet et al., 2006; Hueso et al., 2007兲. In particular, my CONCLUSION


recent work 共Hueso et al., 2007兲 in collaboration with a
group in Cambridge on carbon nanotubes between a fer- In less than 20 years, we have seen spintronics in-
romagnetic source and drain made of the metallic man- creasing considerably the capacity of our hard disks, ex-
ganite L1/3Sr1/3MnO3 has shown that the relative differ- tending the hard disk technology to mobile appliances
ence between the resistances of the parallel and like cameras or portable multimedia players, entering
antiparallel configurations can exceed 60%–70%, well the automotive industry and biomedical technology and,
above what can be obtained with semiconductor chan- with TMR and spin transfer, getting ready to enter the
nels. This can be explained not only by the long spin RAM of our computers or the microwave emitters of
lifetimes of the electrons in carbon nanotubes but also our cell phones. The research of today on the spin-
by their short dwell time related to their high Fermi transfer phenomena, on multiferroic materials, on spin-
velocity 共a definite advantage on semiconductors兲. The tronics with semiconductors, and molecular spintronics,
research is currently very active in this field and, in par- opens fascinating new fields and is also very promising
ticular, graphene-based devices are promising. of multiple applications. Another perspective, out of the

(b)
FIG. 13. 共Color兲 Spintronics
MR=72% with molecules as illustrated.
La2/3Sr1/3MnO3 La2/3Sr1/3MnO3 Left: Artistic view of spin trans-
(LSMO) (LSMO) port through a carbon nanotube
between magnetic electrodes
共courtesy of T. Kontos兲. Right:
Magnetoresistance experimen-
tal results 共Hueso et al., 2007兲 at
4.2 K on carbon nanotubes be-
tween electrodes made of the
ferromagnetic metallic oxide
MR=60% (c) La2/3Sr1/3MnO3. A contrast of
72% and 60% is obtained be-
CNT tween the resistances for the
parallel 共high field兲 and antipar-
allel 共peaks兲 magnetic configu-
(a) rations of the source and drain.
P P
The voltage difference at con-
stant current can reach 60 mV.

Rev. Mod. Phys., Vol. 80, No. 4, October–December 2008


Albert Fert: Nobel Lecture: Origin, development and … 1529

scope of this lecture, should be the exploitation of the Grollier, J., V. Cros, and A. Fert, 2006, Phys. Rev. B 73,
truly quantum-mechanical nature of spin and the long 060409共R兲.
spin coherence time in confined geometry for quantum Grollier, J., V. Cros, A. Hanzic, J. M. George, H. Jaffres, A.
computing in an even more revolutionary application. Fert, G. Faini, J. Ben Youssef, and H. Le Gall, 2001, Appl.
Spintronics should take an important place in the sci- Phys. Lett. 78, 3663.
ence and technology of our century. Grünberg, P., R. Schreiber, Y. Pang, M. B. Brodsky, and H.
Sowers, 1986, Phys. Rev. Lett. 57, 2442.
Hayakawa, J., et al., 2005, Jpn. J. Appl. Phys., Part 2 44, L1267.
REFERENCES
Hueso, L. E., J. M. Pruneda, V. Ferrari, G. Brunell, J. P.
Valdes-Herrera, B. D. Simmons, P. B. Littlewood, E. Arta-
Awschalom, D. D., and M. E. Flatté, 2007, Nat. Phys. 3, 153.
cho, A. Fert, and N. D. Mathur, 2007, Nature 共London兲 445,
Baibich, M. N., J. M. Broto, A. Fert, F. Nguyen Van Dau, F.
410.
Petroff, P. Etienne, G. Creuzet, A. Friederich, and J. Cha-
Ishikawa, T., et al., 2006, Appl. Phys. Lett. 89, 192505.
zelas, 1988, Phys. Rev. Lett. 61, 2472.
Jonker, B. T., and M. E. Flatté, 2006, in Nanomagnetism, ed-
Barnas, J., A. Fert, M. Gmitra, I. Weymann, and V. K. Dugaev,
ited by D. L. Mills and J. A. C. Bland 共Elsevier, New York兲, p.
2005, Phys. Rev. B 72, 024426.
227.
Bass, J., and W. P. Pratt, 1999, J. Magn. Magn. Mater. 200, 274.
Jullière, M., 1975, Phys. Lett. 54A, 225.
Berger, L., 1996, Phys. Rev. B 54, 9353.
Kaka, S., et al., 2005, Nature 共London兲 437, 389.
Binash, G., P. Grünberg, F. Saurenbach, and W. Zinn, 1989,
Phys. Rev. B 39, 4828. Katine, J. A., et al., 2000, Phys. Rev. Lett. 84, 3149.
Boulle, O., 2006, Ph.D. thesis 共Université Paris-Sud兲. Kato, Y., R. C. Myers, A. C. Gossard, and D. D. Awschalom,
Boulle, O., V. Cros, J. Grollier, L. G. Pereira, C. Deranlot, F. 2004, Science 306, 1910.
Petroff, G. Faini, J. Barnas, and A. Fert, 2007, Nat. Phys. 3, Koenig, M., S. Wiedmann, C. Bruene, A. Roth, H. Buhmann,
492. L. W. Molenkamp, X.-L. Qi, and S.-C. Zhang, 2007, Science
Bowen, M., M. Bibes, A. Barthélémy, J.-P. Contour, A. Anane, 318, 766.
Y. Lemaitre, and A. Fert, 2003, Appl. Phys. Lett. 82, 233. Kovalev, A. A., A. Brattas, and G. E. W. Bauer, 2002, Phys.
Bowen, M., V. Cros, F. Petroff, A. Fert, C. Martínez Boubeta, Rev. B 66, 224424.
J. L. Costa-Krämer, J. V. Anguita, A. Cebollada, F. Briones, Leclair, P., J. K. Ha, J. M. Swagten, J. T. Kohlhepp, C. H. Van
J. M. de Teresa, L. Morellón, M. R. Ibarra, F. Güell, F. Peiró, de Vin, and W. J. M. de Jonge, 2002, Appl. Phys. Lett. 80, 625.
and A. Cornet, 2001, Appl. Phys. Lett. 79, 1655. Lee, Y. M., J. Hayakawa, S. Ikeda, F. Matsukura, and H. Ohno,
Bowen, M., et al., 2006, Phys. Rev. B 73, 140408. 2007, Appl. Phys. Lett. 90, 212507.
Camley, R. E., and J. Barnas, 1989, Phys. Rev. Lett. 63, 664. Levy, P. M., S. Zhang, and A. Fert, 1990, Phys. Rev. Lett. 65,
Chappert, C., A. Fert, and F. Nguyen Van Dau, 2007, Nature 1643.
Mater. 6, 813. Loegel, B., and F. Gautier, 1971, J. Phys. Chem. Solids 32,
Cottet, A., T. Kontos, S. Sahoo, H. T. Man, W. Belzig, C. 2723.
Bruder, and C. Schönenberger, 2006, Semicond. Sci. Technol. Mancoff, F. B., et al., 2005, Nature 共London兲 437, 393.
21, 578. Mathon, J., and A. Umerski, 1999, Phys. Rev. B 60, 1117.
Datta, S., and B. Das, 1990, Appl. Phys. Lett. 56, 665. Mattana, R., J.-M. George, H. Jaffrès, F. Nguyen Van Dau, A.
De Teresa, J. M., A. Barthélémy, A. Fert, J. P. Contour, F. Fert, B. Lépine, A. Guivarch, and G. Jézéquel, 2003, Phys.
Montaigne, and A. Vaures, 1999, Science 286, 507. Rev. Lett. 90, 166601.
Dieny, B., V. S. Speriosu, S. S. P. Parkin, B. A. Gurney, D. R. Mavropoulos, Ph., N. Papanikolaou, and P. H. Dederichs, 2000,
Wilhoit, and D. Mauri, 1991, Phys. Rev. B 43, 1297. Phys. Rev. Lett. 85, 1088.
Dupas, C., P. Beauvillain, C. Chappert, J. P. Renard, F. Trigui, Min, B.-C., K. Motohashi, C. Lodder, and R. Jansen, 2006,
P. Veillet, E. Velu, and D. Renard, 1990, J. Appl. Phys. 67, Nature Mater. 5, 817.
5680. Miyazaki, T., and N. Tezuka, 1995, J. Magn. Magn. Mater. 139,
Elsen, M., O. Boulle, J.-M. George, H. Jaffrès, R. Mattana, V. 231.
Cros, A. Fert, A. Lemaître, R. Giraud, and G. Faini, 2006, Moodera, J. S., L. R. Kinder, T. M. Wong, and R. Meservey,
Phys. Rev. B 73, 035303. 1995, Phys. Rev. Lett. 74, 3273.
Fert, A., and I. A. Campbell, 1968, Phys. Rev. Lett. 21, 1190. Mosca, D. H., F. Petroff, A. Fert, P. A. Schroeder, W. P. Pratt,
Fert, A., and I. A. Campbell, 1971, J. Phys. 共France兲 32, C1-46. and R. Loloee, 1991, J. Magn. Magn. Mater. 94, L1.
Fert, A., and I. A. Campbell, 1976, J. Phys. F: Met. Phys. 6, Mott, F., 1936, Proc. R. Soc. London, Ser. A 153, 699.
849. Ohno, H., et al., 1996, Appl. Phys. Lett. 69, 363.
Fert, A., A. Frièderich, and A. Hamzic, 1981, J. Magn. Magn. Parkin, S. S. P., 2002, in Spin Dependent Transport in Magnetic
Mater. 24, 231. Naostructures, edited by S. Maekawa and T. Shinjo 共Taylor
Fert, A., J.-M. George, H. Jaffrès, and R. Mattana, 2007, IEEE and Francis, London兲, p. 237.
Trans. Electron Devices 54, 921. Parkin, S. S. P., R. Bhadra, and K. P. Roche, 1991, Phys. Rev.
Fert, A., and H. Jaffrès, 2001, Phys. Rev. B 64, 184420. Lett. 66, 2152.
Fert, A., and L. Piraux, 1999, J. Magn. Magn. Mater. 200, 338. Parkin, S. S. P., N. More, and K. P. Roche, 1990, Phys. Rev.
Freitas, P. P., H. Ferreira, D. Graham, L. Clarke, M. Amaral, V. Lett. 64, 2304.
Martins, L. Fonseca, and J. S. Cabral, 2003, Europhys. News Parkin, S. S. P., et al., 2004, Nature Mater. 3, 862.
34 共6兲, 225. Piraux, L., J. M. George, J. F. Despres, C. Leroy, E. Ferain, R.
Gajek, M., M. Bibes, S. Fusil, K. Bouzehouane, J. Fontcuberta, Legras, K. Ounadjela, and A. Fert, 1994, Appl. Phys. Lett. 65,
A. Barthélémy, and A. Fert, 2007, Nature Mater. 6, 296. 2484.
Grollier, J., 2003, Ph.D. thesis 共Université Paris-Sud兲. Pratt, W. P., et al., 1991, Phys. Rev. Lett. 66, 3060.

Rev. Mod. Phys., Vol. 80, No. 4, October–December 2008


1530 Albert Fert: Nobel Lecture: Origin, development and …

Ramos, A. V., et al., 2007, Appl. Phys. Lett. 91, 122107. Tsoi, M., A. G. M. Jansen, J. Bass, W. C. Chiang, M. Seck, V.
Rashba, E. I., 2000, Phys. Rev. B 62, R16267. Tsoi, and P. Wyder, 1998, Phys. Rev. Lett. 80, 4281.
Rippart, W. H., et al., 2004, Phys. Rev. Lett. 92, 027201. Valet, T., and A. Fert, 1993, Phys. Rev. B 48, 7099.
Schmidt, G., et al., 2000, Phys. Rev. B 62, R4790. Velev, J. P., et al., 2005, Phys. Rev. Lett. 95, 216601.
Seki, T., et al., 2008, Nature Mater. 7, 125. Vila, L., T. Kjimura, and Y. Otani, 2007, Phys. Rev. Lett. 99,
Shinjo, T., and H. Yamamoto, 1990, J. Phys. Soc. Jpn. 59, 3061.
226604.
Slonczewski, J. C., 1996, J. Magn. Magn. Mater. 159, L1.
Yu, Z. G., and M. E. Flatté, 2002, Phys. Rev. B 66, 201202.
Slonczewski, J. C., 2002, J. Magn. Magn. Mater. 247, 324.
Stephens, J., et al., 2004, Phys. Rev. Lett. 93, 097602. Yuasa, S., et al., 2004, Nature Mater. 3, 868.
Stiles, M. D., and J. Miltat, 2006, in Spin Dynamics in Confined Zhang, S., 2000, Phys. Rev. Lett. 85, 393.
Magnetic Structures, III, edited by B. Hillebrands and A. Zhang, X. G., and W. H. Butler, 2004, Phys. Rev. B 70, 172407.
Thiaville 共Springer, Berlin兲.

Rev. Mod. Phys., Vol. 80, No. 4, October–December 2008