Magnetoresistance 24.10.

2002 1

Giant Magnetoresistance

Why magnetic layers separated by non-magnetic metals ought to exhibit a

negative magnetoresistance.

In a ferromagnet, it is reasonable to assume that spins which are polarized parallel or

antiparallel to the magnetization have different resistivities. This is often referred to as
the Mott picture.1 It is based on an approximate picture of the electronic structure of a
transition-metal ferromagnet. The sp bands are assumed to carry the current. The
conduction electrons scatter into the d bands which are split by exchange. One of the d
band is full, the other has unoccupied states at the Fermi level. (Figure below from2).
Hence, in working out the scattering amplitudes, the density of the final states may,
grossly speaking, favor or forbid scattering.
Magnetoresistance 24.10.2002 2

Therefore we write the resistivities of spins parallel or antiparallel to the magnetization as

:

ρ↑ = ρ(1 − β) ρ↓ = ρ(1 + β)

Consider a system of two ferromagnetic layers, once in a parallel configuration, once in

an antiparallel alignment, and the electrons do not flip their spins going from one layer to
the other.

In the parallel configuration, one spin In the antiparallel configuration, each spin
polarization experiences ρ ↑ through both channel experiences ρ ↑ and ρ ↓ in series.
layers, the other spin polarization ρ ↓ in The effective resistance is
both layers. The effective resistance is :

2 ρ ↑ ρ↑ ρ↑ + ρ↑
ρ↓↓ = ρ ↑↑ = 2
= ρ (1 − β ) ρ↑↓ = =ρ
ρ↑ + ρ↑ 2
Magnetoresistance 24.10.2002 3

The effective resistance is greater in the antiparallel configuration than in the parallel
configuration, no matter what the sign of β is. This model is directly inferred from the
two-current model, where the current of both spin orientations are assumed separate. This
is a reasonable model because a scattering event where the electron flips its spin while
keeping its velocity is very rare.

Practically, in order to have a non-negligible resistance value to observe, the

measurements are performed usually in a geometry where the current flows parallel to the
layers (CIP-GMR). If the layer thickness is larger than the electronic mean free path, the
simplest picture shown above fails and the CIP-GMR is severely diminished. Therefore,
structuring on a nanometer scale is necessary. Below are shown the historical data of the
first observation of giant magnetoresistance in Fe/Cr multilayers.

R/R (H = 0)

(Fe 30 Å/Cr 18 Å ) 30

0.8 H S

0.7 (Fe 30 Å/Cr 12 Å ) 35

0.6 HS
(Fe 30 Å/Cr 9 Å

0.5 HS applied field

– 40 – 30 – 20 – 10 0 10 20 30 40
Magnetic field (kG)

Figure 1 : Magnetoresistance of a Fe/Cr superlattice. This effect is now obtained at room

temperature and fields of about hundred Gauss. (from Baibich 1988 )
Magnetoresistance 24.10.2002 4

The two-current model

Bibliography
Mott introduced the idea of “spin-up” and “spin-down” currents in ferromagnets (N.F. Mott, Proc. Roy.
Soc. 153, 699(1936) and Adv. Phys. 13, 325(1964) ). It was refined in the late 1960’s by :
- “Evidence for Two Current Conduction Iron”, I.A. Campbell, A. Fert, A.R. Pomeroy, Phil. Mag
997 (1967)
- “Tow-current conduction in Nickel” A. Fert, I.A. Campbell, Phys. Rev. Lett. 21(16) 1190 (1968)
- I.A. Campbell, A. Fert, in Ferromagnetic Materials, Wohlfarth Editor, vol III, chapter 9
- A. Fert, I.A. Campbell, J.Phys. F. Meal Phys. Vol 6(5) 849 (1976)

Experimental challenge

The two-current model was introduced to elucidate the departure from Mathiessen’s law
of the resistivity of ferromagnets doped with impurities.

Mathiessen’s law (Annalen Phys. 7, 761 and 892 (1864) )

The basic idea of Mathiessen’s law is expressed in Kittel’s “Introduction to Solid State
Physics”. The book of Ashcroft and Mermin gives a detailed discussion of the limitation
of this law in terms of semiclassic transport theory.

The basic idea is that the scattering cross section or the relaxation rates in Drude’s model
of resistivity corresponding to various scattering center, add :
1 1 1
= +
τ τ1 τ 2
so that the contributions to the resistivity add :
ρ = ρ1 + ρ 2
Kittel shows the example of the temperature dependence of two samples of potassium.
Magnetoresistance 24.10.2002 5

The low temperature limit is assumed to be due to varying concentrations of impurities,

while the temperature dependent component is thought of as that of the pure metal.

Note : use of Mathiessen rule may be used to estimate the resistivity of a sample when
the temperature dependence of the pure metal is known

Now, when a ferromagnet is doped with impurities, deviations from Mathiessen’s rule
were observed. For example Campbell, Fert and Pomeroy report a deviation :

∆ρ (T ) = ρ (T ) − ρ 0 − ρ P (T )
where ρ 0 is the low temperature limit of the resistivity and ρ P (T ) is the temperature
dependence of the pure metal.

The two-current model

We assume at low temperature that the current is carried in parallel by electrons with
spin-up and spin-down electrons. The introduction of impurities generates for each spin
direction a residual resistivity ρ↑ (0) and ρ↓ (0) . The overall resistivity is then :
Magnetoresistance 24.10.2002 6

ρ↑ (0) ρ↓ (0)
ρ (0) =
ρ↑ (0) + ρ↓ (0)

Now, having assumed this, one has to ask ourselves how correct this is. It turns out that,
as the temperature rises, processes can mix momentum from one spin channel to the
other. There are two processes known to produce this. One is the elastic scattering of
electrons by spin waves. The other is a mutual spin-flip process between two interacting
electrons. The description of these mechanisms is beyond the scope of this section.

We are going to describe these processes by a detailed balance equation describing the
momentum of each spin channel. The rate of exchange of momentum between the spin
channels is assumed given as τ ↑↓ . That is, with the usual notation:
d ( N− − N+ ) (N −
− N+ )
=−
dt τ↑↓
If we denote the average velocity in each channel by v and the electric field is E, then
we write :

e 1 1
−
m
E = v↑ +
τ
(v − v )
2τ ↑↓ ↑ ↓
e 1 1
−
m
E = v↓ +
τ
(v − v )
2τ ↑↓ ↓ ↑

Then for each spin channel we define

i↑,↓ = nev↑,↓
m
ρ ↑ ,↓ =
ne τ ↑,↓
2

while the so-called spin-mixing effect gives a term :

m
ρ↑↓ =
ne 2τ ↑↓
Finally the resititivity is calculated as :

E
ρ (T ) =
i↑ + i↓

The result is (PRL 1968)

ρ↑ ρ↓ + 4 ρ↑↓ ( ρ↑ + ρ↓ )
ρ=
ρ↑ + ρ↓ + 4 ρ↑↓
Magnetoresistance 24.10.2002 7

Each spin resistivity can be written :

ρ↑,↓ (T ) = ρ↑,↓ (0) + ρi↑,↓ (T )

where ρi↑,↓ (T ) are meant as the intrinsic resistivity of each spin channel of the pure
metal.

Estimate of the spin-dependent scattering parameters

Fert and Campbell continued their analysis by expressing the low temperature expension
of the resistivities. At low enough temperatures we have :

ρ↑ (0), ρ↓ (0) >> ρ i↑ (T ), ρi↓ (T )

To first order one gets :

(α − µ )
2 2
 α −1
ρ = ρ (0) + ρi (T ) + ρi (T ) + ρ ↑↓  
(1 + α ) α +1
2

where
ρi ↑ (T)ρi ↓ (T)
ρi (T) =
ρi ↑ (T) + ρi ↓ (T)

Hence, the deviation from Mathiessen’s rule is :

(α − µ )
2 2
 α −1
∆ρ = ρi (T ) + ρ↑↓  
(1 + α ) α +1
2

In these formulas, the so-called spin asymmetries are :

ρ↓ (0) ρ i ↓ (T )
α= µ=
ρ↑ (0) ρ i ↑ (T )

Typical data are shown below. The squares are resistivity values of the pure sample of
Ni, the circles are the temperature-dependent part of the resistivity of Ni doped with 1400
ppm of Cr, the circles, Ni doped with 700 ppm of Cr.
Magnetoresistance 24.10.2002 8

Theses authors then analysed their data so as to recognize in the temperature-dependent

part of the resistivity of doped samples as including a correction proportional to ρi (T )
and the rest as attributed to ρ↑↓ (T ) .

The soundness of the model was tested by analysing ternary alloys. Their results were
summarized by Fert and Campbell in J. Phys. F.

The two-current model in the limit where no spin-mixing takes place was used by most authors to analyse
giant magnetoresistance data. Prof. M. Gijs included the temperature dependent term successfully in his
temperature CPP-GMR measurements.
Magnetoresistance 24.10.2002 9

Engineering GMR materials

Typically, superlattices or multilayers of a magnetic metal alternating with a non-
magnetic metal have been used.

One way of getting the antiparallel alignment is to have the spacer layer just thin enough
to produce a magnetic coupling mediated by the spacer (the so-called exchange coupling)
which can be made antiferromagnetic. (Fig. 2a)

In the so-called spin-valve structure (Fig. 2b), one layer has its magnetization pinned,
while the other is freer to be reversed.

Granular materials with ferromagnetic ultrafine grains in a non-magnetic matrix (Fig. 2c)
have been shown to have GMR also. 3,4

A remarkable result was obtained with a combination of Co clusters and permalloy layers
spaced with silver (Fig. 2d). A record magnetoresistive sensitivity to the applied magnetic
field of 6.5% per Oersted was achieved !5
Magnetoresistance 24.10.2002 10

Engineering MR Materials

Various GMR nanostructures (left) and their magnetoresistance

behaviour (right – note that all the hotizontal scales are different).
(a) antiferromagnetically coupled multilayer;
(b) spin – valve structure;
(c) granular alloy;
(d) hybrid nanostructure including clusters and layers. See text for
details.
a M

magnetoresistance
NM
antiferromagnetically
M
coupled multilayers
NM
M
magnetic field
b NiFe
Cu
spin valve structure NiFe

FeMn

granular material

d
Ag Co clusters
NiFe
hybrid cluster/layer
Ag Co clusters
NiFe

Barthélémy et al., Physics World Nov. 1994

Figure 2 : Nanostructures for GMR. From top to bottom : - multilayers make use of an
antiferromagnetic coupling among the layers, - spin valve bi-layer structures have one
layer pinned magnetically; - granular materials; - hybrids. (after Physics World, Nov.
1994 p. 34)

Perpendicular Spin Transport

While the two-current model used above predicts a "giant magnetoresistance", some of
the actual issues required extensive research :

- How far can a spin travel before it flips ?

Magnetoresistance 24.10.2002 11

- What is the relative importance of bulk and interface scattering in producing

GMR ?

- What is spin dependent : the scattering potential itself, or the occupation of

the states ?

- Do the reflections of the electronic wave functions at interfaces play a role ?

Experiments in which the current was perpendicular to the interfaces have been
particularly useful for shedding light on these issues.6 Perpendicular transport has also
practical advantages over transport parallel to the interfaces : layers can be thicker when
the current is perpendicular rather than parallel to the interfaces. This is relevant for
practical applications, which must consider the ease of manufacturing and the reliability
of the materials. The reason is as follows. When the current is parallel to the layers, the
electrons experience two layers provided their separation is of the order of the electron
mean free path. When the current is perpendicular to the layers, the spins undergo a
diffusion process which extends over distances much longer that the electron mean free
path since the spin flip events are rare. Indeed, electron spin resonance experiments of
the 60's and the 70's determined that electrons can travel a long way without flipping their
spin.

A tough challenge
A small square cut out of thin film of a magnetic multilayer, of size in the sub-millimeter
range, has a resistance in the nano-ohm range ! Its measurement has been carried out by
few. 7,8 Others have relied on advance lithographic techniques in order to reduce the
surface area so as to bring the resistance to more accessible values. 9 This is quite an
extensive process.10 However, when the columns are very thin, the current density is no
longer even throughout the column. Furthermore, the field sensitivity is reduced
considerably.

Instead of using advanced lithography in order to obtain a thin column, some groups have
produced multilayers in the form of wires by electrodeposition of multilayers in pores of
nanoporous membranes. 11,12,13 (Figure 3a, below) Others have used grooved substrates
and a straight deposition (Figure 3b) so that they measured GMR with current-at-an-
angle(CAP) 14 or they deposited the layers sideways, so as to have a structure in which
the current was mostly perpendicular to the layers. 15 (Figure 3c)
Magnetoresistance 24.10.2002 12

40nm
Polycarbonate

Cu layer

Co layer

Cu film

direction of
multilayer deposition
(100)
4µm
0.17µm 0.20µm

)
(1
11

11
(1
71°

)
InP
4.3µm
C IP
CAP
I

Si InP

Figure 3 : Special ways of measuring spin transport perpendicular to the planes : A)

multilayered nanowires in nanoporous membranes, B) straight deposition and C) oblique
deposition, on grooved substrates.

Several groups have studied CPP-GMR. They determined in particular the so-called spin
asymmetry for majority and minority spins. Experimental results are summarized in the
following table. The other parameters given in the table will be explained later.
Magnetoresistance 24.10.2002 13

Table 1
Bulk ( β ), interface ( γ ) spin assymetry parameters and spin diffusion lengths of Co/Cu
and Co/Py multilayers
β γ lsf (Cu) lsf (Co) lsf (Py)ref
(nm) (nm) (nm)
________________________________________________________________________
__

16
0.45 0.7 40 >10

17
0.4 0.8 140 44

18
0.5 0.8 varies with doping

19
0.5 0.7

20
0.9 0.9 140 3

21
0.7 0.7 - 5.5

There is a close correlation between the spin asymmetry β and the magnetic moment per
atom as displayed on a Stoner-Pauling plot. This correlation has not been accounted for
theoretically, as far as we know.
Magnetoresistance 24.10.2002 14

The contributions of the parameters β and γ can be seen in the following figure, in
which the GMR is estimated if bulk or bulk and interface effects are present. This figure
suggests also the importance of making nanostructures with sizes less than the spin
diffusion length. Only when the typical sizes of the device is less than the spin diffusion
length can their be any spin-dependent effect. We see therefore the importance of
understanding the spin-scattering phenomena in order to plan to avoid or master them to
ones purpose. (see the chapter « spin flip scattering »)

0.30

0.25

0.20
²R/R

0.15

0.10

0.05

0.00
20 40 60 80 100
Layer thickness [nm]
Magnetoresistance 24.10.2002 15

∆R
Figure : Magnetoresistance of Co/Cu multilayers of equal Co and Cu thicknesses.
R max
Predictions for : (.......) bulk spin-dependent scattering only ( β = 0. 4 ); (----) same with
interface scattering ( γ = 0. 7 ); (____) same with also spin diffusion effects ( lsf =40 nm). (data
Doudin 1996)

Colossal Magnetoresistance (CMR)

The reader may have a first introduction to this topic by reading a recent article in
Physics World.22 A complete survey of the practical understanding of these systems and
the state of the art in their synthesis can be found in the critical review of Coey, Viret and
von Molnar.23 The topic is linked to GMR for historical reasons. The perovskites that
display the so-called colossal magnetoresistance were studied in the 1950's. 24 In the 90's,
the search for any kind of "giant" magnetoresistance was fiercely under way. In 1994, a
group showed that it was possible to "tune" the structure of manganite perovskites so that
the apparent metal to insulator transition that is affected by magnetic field would occur
around room temperature. 25 In terms of magnetoresistance ratio as used for AMR or
GMR, this transition would represent 10'000% or more. Hence, the authors coined it
"colossal".

In fact, at this point of the understanding of CMR, the connection with spin-dependent
transport in nanostructure relates to the notion of spin injection and tunnel experiments.
These materials appear to be half metallic ferromagnet, meaning that only one spin band
crosses the Fermi level, hence are expected to have very strong tunnel magnetoresistance.

At a more fundamental level, the connection between spin transport and CMR arises in
the mechanism called the double-exchange. The following quote from a Science paper
may suffice in giving a flavor of the current thinking as to the mechanisms of CMR.26

“Although the theoretical understanding of the CMR phenomenon is still incomplete,

double exchange (DE) mechanism (1), electron-phonon coupling (2), and orbital ordering
effects (3) are commonly adopted as the main ingredients. The DE mechanism links the
electronic to the magnetic transition and describes the hopping of electrons in eg orbitals.
The DE mechanism links the electronic to the magnetic transition and describes the
hopping of electons in eg orbitals between neighboring Mn3+ and Mn 4+ sites with strong
on-site Hund’s coupling by an O 2− ion. This charge transport is enhanced in the
ferromagnetic state when the local Mn d-shell spins are parallel. In turn, the hopping
electrons promote ferromagnetic order because they tend to preserve their spin direction.
The random spin disorder in the crossover regime around TC can be removed at least
Magnetoresistance 24.10.2002 16

partially by applying a magnetic field. This shifts Tc to higher temperatures and causes
CMR. However, the DE mechanism alone is insufficient to correctly describe the high-
temperature transport properties and to quantify the large resistance drop. At such
temperatures, the insulator-like behavior results from electron localization caused by
electron-lattice coupling, originating from local Jahn-Teller distortions at the Mn3+ site.
These distortions also lift the degeneracy of the eg orbital order (and disorder) into play.

Other materials, simpler in structure, present the apparent similar complexity of magnetic
and transport behavior. Mauger for example treats the case of Europium Chalcogenides.
He describes the interplay of magnetism and conduction with a model in which the
ferromagnetism is mediated by an indirect exchange via the conduction electrons. The
conduction electrons themselves have their spin bands split by the coupling to the
magnetic lattice.
References:
A.Mauger, Phys. Stat. Sol. (b) 84, 761(1977)
A. Mauger et al., Le Journal de Physique, 39(1978) 1125
Mauger and C. Godart, PHYSICS REPORTS 141, 51 (1986)

Spin tunneling

Slonczewski carried a quantum mechanical calculation of spin between two

ferromagnets. ( 27 ) He used a free-electron model for the conduction electrons, a
rectangular barrier potential, and an internal exchange energy in the magnetic layers of
the form :

rr
− hσ

The directions of the molecular fields in the layers differ by an angle θ and their
magnitudes are the same. The junction conductance is found to vanish when the
magnetization of the layers are anti-parallel if only one spin band is present at the Fermi
level. This is the case of the so-called half metallic ferromagnets to be discussed below.
Magnetoresistance 24.10.2002 17

In the imperfect case where both spin bands cross the Fermi level, the junction
conductance has the form :

(
G = Gfbf 1+ Pfb 2 cosθ )

with P fb =
(k↑ − k ↓ )(κ 2 − k↑ k ↓ )
(k↑ + k ↓ )(κ 2 + k↑ k ↓ )
where k↑ and k↑ are the spin up and spin down momentum values in the metals, and

i κ is the imaginary momentum in the barrier. Gfbf is proportional to e −2κd where d is

the barrier thickness. The first factor of Pfb is a fractional spin polarization. The second

one is conceptually novel relative to Jullière's model. It expresses a dependence of the

effective relative polarization on the height of the barrier. It can vanish and even change
sign!

This simple model provides a basis for deriving two other effects. First, an effective
junction exchange coupling is found, of the form :

E(coupling) = − J cos θ

where the coupling constant varies in sign and magnitude with the barrier height. Second,
a dynamical term is derived which represents damping for one sign of the applied
voltage, and excitation of oscillation of the magnetization for the other sign of the applied
voltage.

S.T. Chui considered the difference in spin up and spin down chemical potentials, ∆µ , at

the ferromagnet/insulator interface. It depends on whether the magnetization vectors of

Magnetoresistance 24.10.2002 18

the adjacent layers are parallel or antiparallel. He points out that these chemical potential
differences result in a large bias dependence of the magnetoresistance. ( 28 )

Veyayev et al. have indicated the possibility of strongly enhanced tunneling

magnetoresistance obtained by adding to the tunnel structure a thin paramagnetic layer.
Selective multiple reflections of either spin up or spin down electrons in this layer result
in very large MR effects at specific values of the thickness of the paramagnetic layer, in a
similar fashion to what happens in optical filters. ( 29)

Butler et al. cautioned against using the simple potential barrier models as discussed
above. A calculation of the electronic structure of spin dependent structures such as
Fe/Ge/Fe or Fe/GaAs/Fe indicate that the local density of states at the Fermi level at the
interfacial metal layer differs significantly from its value in the bulk of the metal.
Furthermore the band structure of the majority and minority spins in the semiconductor
barrier are quite different, although the barrier remains non-magnetic. ( 30) These
considerations confrm the difficulties that Alvarado et al. had when they used magnetic
tips to study spin dependent tunneling. They saw in the early 90’s that the hydridization
of the wave function of the tip with the d states of the substrates could lead to great
variation in the spin-dependent current. Spin-dependent tunneling is still at the
developmental stage in year 2000.

Unlike giant magnetoresistance, tunnel magnetoresistance is still today (2000) a subject

of controversy. In addition to the subtle questions of electronic structure at the interface,
there are pratical issues concerning the quality of the tunnel barrier. There could be pin-
holes in which the current passes. They could be so small that the electrons would
traverse them ballistically. Then the magnetoresistance would be determined by ballistic
electrons jumping from one magnetic domain to another (see Garcia).
Magnetoresistance 24.10.2002 19

Spin disorder scattering

In the 1960’s, the transport properties of rare earth metals and of magnetic semiconductors were under
close scrutiny. In both cases, a peak in resistivity was observed at Tc. The change of resistivity with
magnetic field was quite large. It was interpreted by many as arising from spin-disorder scattering. The
case of metals was treated in a powerfully clear paper by De Gennes and Friedel. 31 The review article of C.
Haas on magnetic semiconductors contains numerous and valuable information …32

The general idea is that the spin of the conduction electrons scatter with localized magnetic spins, owing to
the exchange coupling between the two. (see my notes on this)

We follow de Gennes and Friedel. We assume that the conduction electrons can be treated as free electrons
with an effective mass m. We assume the presence of a lattice of magnetic ions of spins S, separated by a
distance d, with Z nearest neighbors, c is their concentration and the coupling between two localized
moments is given by
r r
2 JSR ⋅ SR '
The coupling of the conduction electrons to the localized moments is assumed to be of the form :
r r
( r
Η = ∑ G δ R − rp SR ⋅ sp
R ,p
)( r
)
In both cases, the transfer of energy from the conduction electron spins to the magnetization is neglected.
Whether such a transfer is possible at high current density in nanostructure is a subject of current studies,
therefore, beyond the scope of this introduction. In semiconductors, only small k values can be considered,
whereas large k must be included, as do Friedel and de Gennes. However, it turns out that the theoretical
treatment of Haas for small k does not change much to the overall picture (see figure below) and we follow
the work of the former, since we are concerned here with the general physical picture.

Figure :

resistivity at zero field of

Eu(0.95)Gd(0.05)Se (line a)
compared with the theoretical
prediction according to de Gennes
and Friedel (curve b) and C.Haas
(curve c). (ref : review of C. Haas)

Friedel and de Gennes apply the Born formula for scattering to get :
Magnetoresistance 24.10.2002 20

2 2
r r r r r
∑ exp(iκ ⋅ R) ( s )β
dσ  m  r
= 
dΩ  2πh 2 
∑p
α ,β
α α
R
p ⋅ SR − sp ⋅ SR

Here α and β are meant as the spin state of the overall system (electron plus localized moments).
r
pα is the probability to find the system in the state α in a thermal equilibrium. h κ is the momentum
change of the electron in the collision. The sum over the final states β can be operated readily. There
remains :

2 2
r r r r r r r r
dσ  m 
= 
dΩ  2πh 2 
∑p
α ,β
α α
R ,R '
r
∑ exp(iκ ⋅ (R − R ')G 2
(s ⋅S
p R
r
− sp ⋅ SR )( s ⋅ S
p R'
r
− sp ⋅ SR ' )α

Thus, there is a thermal average of the expectation value of a correlation function that involves the electron
spin operator and the spin of local moments at R and R’. The average of the operators on the electron spin
1
subspace yields s p (s p + 1) that factors a correlation function for the local moments :
2
r r r r
SR ⋅ SR ' − SR SR '
T T T
The reader familiar with the fluctuation-dissipation theorem can expect readily a link between this
correlation function and the susceptibility of the magnetic system. Thus, where a peak of susceptibility is
found, one expects likewise a peak in resistivity, as the scattering cross-section increases there.

If the moments are uncorrelated, then the scattering at one site is independent of the others, in particular, it
is isotropic and the cross-section, hence the relaxation time, takes on the form :
r
1 S2
∝1−
τ S(S + 1)
This shows clearly the connection between the resistivity and the magnetic order parameter.

Magnon Drag and Magnetic Polarons

C. Haas points out an important limitation of the approach : the electron occupation probability of the
system at equilibrium in a given array of magnetic moment is used for these calculations. The effect of the
current on the equilibrium magnetization and how this in turn changes the electron populations is not taken
into account. This effect was expect by Haas and others quoted in his review. The effect is referred to as
magnon drag. The accumulated literature on this topic is rather scarce, even today. The review of Haas
gives an account of the general idea. Magnon drag was described in the framework of Boltzmann’s
equation by Baylin. 33 Evidence of a strong interaction of dc current with spin waves was shown in bulk
magnetic semiconductors. 34 Experimental evidence of magnon drag appears in a few works on
thermopower. 35 The effect of magnon drag on electron mass and mobility was worked out. 36 A two-band
model was applied to the calculation of the thermopower. 37 A phenomenological description of Magnon
drag and its impact on the thermoelectric power was given by Berger et al. 38

De Gennes and Friedel give an estimate for inelastic scattering, that is, transfer of energy from the spin of
the incoming electrons and the magnetization. There calculation shows that in fact, one should consider
inelastic scattering at low temperature. They refer to work on neutron scattering for an estimate of this
effect. 39
Magnetoresistance 24.10.2002 21

Clearly, the case of weakly doped EuSe requires a

different mechanism. The magnetoresistance in
this case is what is nowadays referred to as
“colossal” magnetoresistance. 40

A possibly more drastic departure from the

equilibrium electron populations giving rise
possibly to colossal magnetoresistance, may arise
from a strong correlation of conduction electron
spins and the lattice of moments, leading to the
formation of so-called magnetic polarons. (see
notes on this) Research is still going on in this
area. 41

Figure :
resistivity versus temperature at various fields, of
Eu(0.99)Gd(0.01)Se

Magneto- impedance effects

Above, we have followed the classical literature that was concerned with the description
of the so-called Walker modes. These were observed in ferromagnetic insulators such as
ferrites.

Starting in the early 90s, there has been an intense interest in very large variations of the
impedance of a magnetic wire. These effects arise in soft magnetic wires and ribbons at
frequencies when the skin depth becomes small and a small change is field is sufficient to
change the domain configuration and thus the differential susceptibility. Changes of
impedance of the order of several 100% with as little as 10 Oe could be obtained.

For possible application in magnetoresistive memories, see e.g.

- Asymmetrical magneto-impedance effect in twisted amorphous wires for sensitive
magnetic sensors, T. Kitho, K. Mohri, IEEE Trans. Magnetics 31(6) 3137 (1995
- Pulse response bistable magneto-impedance effect in amorphous wires, M. Noda,
L.V. Panina, K. Mohri, IEEE Trans. Magn. 31(6) 3167(1995)

For typical results see :

- symposium on magneto-impedance, I. K. Schuller chairman, J.Appl. Phys.
76(19)6198(1994)
Magnetoresistance 24.10.2002 22

- in particular “Giant magneto-impedance and magneto-inductive effects in

amorphous alloys, same reference

For the basic theory, see:

- Theory of giant magneto-impedance effect in amorphous wires with different
types of magnetic anisotropy, N.A. Usov, A.S. Antonov, A.N. Lagar’kov, JMMM
185(1998)159-173
- (going to FMR frequencies) Impedance of a ferromagnetic sandwich strip, A.
Sukstanskii, V. Korenivski, A. Gromov, J.Appl. Phys. 89(1) 775(2001)
Magnetoresistance 24.10.2002 23

1
N.F. Mott, The Electrical Conductivity of Transition metals, Proc. Roy. Soc. A153(1936)699
2
"Magnetismus von Festkörpern und Grezflächen", 8-19 März 1993, Vorlesungmanuskripte, tel. 02461 61-
60-48, fax 02461 61-24-10
3
J.X. Xiao, J.S. Jiang, C.L. Chien, Phys. Rev. Lettt. 68(5), 3749(1992)
4
A.E. Berkowitz, J.R. Mitchell, M.J. Carey, A.P. Young, S. Zhang, F.E. Spada, F.T. Parker, A. HJutten,
G. Thomas, Phys. Rev. lett. 68(25)3745(1992)
5
Holody-P Steren-LB Morel-R Fert-A Loloee-R Schroeder-PA
PHYSICAL REVIEW B-CONDENSED MATTER 1994, Vol 50, Iss 17, pp 12999-13002
6
M.A.M. Gijs, G.E.W. Bauer, "Perpendicular Giant Magnetoresistance of Magnetic Multilayers, to be
published in "Advances in Physics"
7
W.P. Pratt Jr., S.F. Lee, J.M. Slaughter, R. Loloee, P.A. Scroeder, J. Bass, Phys. Rev. Lett. 66,
3060(1991)
8
P. Dauguet, P. Gandit, J. Chaussy, J. Appl. Phys. 79,5823(1996)
9
M.A.M. Gijs, J.B. Giesbers, S.K.J. Lenczowski, Phys. Rev. Lett. 70,3343 (1993)
10
W. Vavra, S.F. Cheng, A. Fink, J.J. Krebs, G.A. Prinz, Appl. Phys. Lett. 66, 2579(1995)
11
A. Blondel, J.P. Meier, B. Doudin, J-Ph Ansermet, Appl. Phys. Lett. 65,3019(1994)
12
L. Piraux, J.M. George, J.F. Despres, C. Leroy, E. Ferain, R. Legras, K. Ounadjela, A. Fert, Appl. Phys.
Lett. 65, 2488(1994)
13
K. Liu, K. Nagodawithana, P.C. Searson, C.L. Chien, Phys. Rev. B51, 7381(1885)
14
T. Ono, T. Shinjo, J. Phys. Soc. Jpn. 64, 363(1995)
15
M.A.M. Gijs, M.T. Johnson, A. Reinders, P.E. Huisman, R.J.M. van de Veerdonk, S.K.J. Lenczowski,
R.M.J. van Gansewinkel, Appl. Phys. Lett. 66,1839(1995)
16
B. Doudin, A. Blondel, J-Ph Ansermet, J. Appl. Phys. 79 (1996) 6090
17
L. Piraux. S. Dubois, C. Marchal, J.M. Beuken, L. Filipozzi, J.F. Despres, K. Ounadjela, A. Fert, JMMM
156, 317-320(1996)
18
Yang, Q., Holody P., Lee SF. , Henry LL., Loloee R., Schroeder PA., Pratt Jr. WP., Bass J., Phys. Rev.
Lett,. 72, 3274(1994)
19
M. Howson et al. cited in J.Bass, W.P. Pratt, P.A. Schroeder, "Current-Perpendicular (CPP)
Giant Magnetoreistance -- Larger and Simpler than Current-In-Plane (CIP) Giant
Magnetoreistance", Comments on condesned Matter Physics, in press
20
Piraux, Banuyls Journées Louis Néel du Magnétisme
21
W.P. Pratt, S.D. Steenwyk, S.Y. Hsu, W.C. Chiang, A.C. Schaefer, R. Loloee, J. Bass, to be
published
22
Colossal magnetoresistance, Josep Fontbuberta, Physics World, February 1999, page 33
23
J.M. D. Coey, M. Viret, S. von Molnar, Mixed-valence manganites, Adv. Phys. 1999, vol. 48(2) 167-293
24
J. Volger, Physica XX, 49-66 (1954)
25
Thousand fold Change in Resistivity in magnetoresistive La-Ca-Mn-O Films, S. Jin, T.H. Tiefel, M.
McCormack, R.A. Fastnacht, R. Ramesh, L.H. Chen, Science, 264(1994) 413
26
Spatially Inhomogeneous Metal-Insulator Transition in Doped Manganites, M. Faeth, et al. Science 285,
1540(1999)
27
J.C. Slonczewski, Conducance and echange coupling of two ferromagnets separated by a tunneling
barrier, Phys. B39(10) 1989
28
Chui-ST, Bias Dependence in Spin-Polarized Tunneling, PHYSICAL REVIEW B-CONDENSED
MATTER 1997, Vol 55, Iss 9, pp 5600-5603
29
A. Veyayev, N. Ryzhanova, C. Lacroix, L. Giacomoni, B. Dieny, "Resonance, in tunneling through
magnetic valve tunnel junctions", Europhys. Lett. 39(2) 219-224(1997)
30
W.H. Butler, X.G. Zhang, Xindong Wang, J. van Ek, J.M. MacLaren, "Electronic Structure of
FM/semiconductor/FM spin tunneling structures", J. Appl. Phys. 81(8( 5518(1997)
31
P.G. de Gennes, J. Friedel, J. Phys. Chem. Solides, 1958, pp. 71-77, vol. 4, « Anomalies de résistivité
dans certains métaux magnétiques »
32
C Haas, « Magnetic Semiconductors », Critical Reviews in Solid State Sciences, vol. 1(1), pp. 47-98
33
M. Bailyn, Phys. Rev 126, 2040 (1962)
Magnetoresistance 24.10.2002 24

34
B. Vuarl, E.E. Thomas, « helicon-spin wave interaction in the magnetic semiconductor Ag/x Cd/1-x Cr/2
Se/4, APL 12(1) 14 (1968)
35
F.J. Blatt, D.J. Flood, V. Rowe, P.A. Schroeder, J.E. Cox, Magnon-drag Thermopower in Iron, PRL
18(11), 395 (1967)
G. Haacke, L.C. Beegle, Anomalous Thermoelectric power of FeCr2S4 near the Curie
Temperature, PRL 17(8), 427(1966)
36
Alba Theumann, Effect of Magnon Drag on Electron Mass and Mobility, PRB 1(11) 4400(1970)
37
W.M. MacInnes, K. Schroeder, PRB4(11) 4091(1971)
38
G.N. Grannemann, L. Berger, Magnon-drag Peltier effect in a Ni-Cu alloy, PRB13(5) 2072(1976)
39
Van Vleck J.H., Phys. Rev. 55, 924 (1939); De Gennes P.G., C. R. Adad. Sci. Paris 244, 755 (1957)
40
S. von Molnar, S. Methfessel, JAP 38(3) 959 (1967)
41
see eg. S. Pathak, S. Satpathy, PRB 63, 214413 (2001); D.J. Garcia et al. PRB 65, 134444(2002) ; M.
Umehara, PRB 61(18) 12209 (2000-II)