Chapter Twelve Ferromagnetism and Antiferromagnetism

In some materials, there are important interactions between magnetic

moments resulting in ordered magnetic phases in the absence of H.

M >> 0 M >0 M =0
Parallel alignment Anti-parallel Anti-parallel alignment
Ferromagnetism Ferrimagnetism Anti-Ferromagnetism
What are the interactions ?
(1) Dipole-dipole interaction Not Important
 
µ produces local field → align adjacent µ very weak
(2) “Exchange” interaction
QM interaction due to overlap of electron wavefunctions
  (Spin dependent)
U = −2J ∑ Si • S j
ij
 Ferromagnetism
M
T=300K
Ms

H
Hc

The typical magnetization curve

of the ferromagnet
A hysteresis loop of magnetization is present.
Ms : saturation magnetization
Mr = M(0) residual magnetization
Hc coercive field M(± Hc)=0
 Ms χ M 1
χ= ~ Curie-Weiss law
H T −θ
Ferromagnet
Paramagnet
T
Tc
Curie temperature
High T (T>Tc) : paramagnetic FCC

Low T (T<Tc) : ferromagnetic 1473K

1388K
1273K
Mo
1073K
1043K
Ms (emu/g)

627K

T(K)
273K
 Approaching Tc (Curie temperature)
High T (T>Tc) : paramagnetic χ ∝ (T − Tc )−γ
M s ∝ (Tc − T )
β
Low T (T<Tc) : ferromagnet
In mean field approximation : γ=1 and β=1/2.

Experimental data
γ β Tc(K)
Fe 1.33 0.34 1043 Ni
Co 1.21 1388
Ni 1.35 0.42 627
Gd 1.3 292
CrO2 1.63 387
CuBr3 1.21 0.37 33
EuS 0.33 17

Tc=354oC
How will Ms vary with temperature ?
What is Tc ?

Recall Langevin function

Ms/Mo
= L(a ) = coth (a ) −
M 1
Mo a
µH
where a = in the theory of
k BT paramagnetism

Weiss suggested that

a “molecular field HM” in addition to H
T/Tc
is acting on material. : H + HM= H + γM
µH M µγM M o M  k BT 
When H=0, a = = =  a
k BT k BT M o M o  µγM o 
M/Mo is a linear function of a with a slope proportional to T.
 M  k BT 
Three straight lines: =  a
T
T T M o  µγM o 
2
3
1 w/. T3>T2>T1
P
The intersection P gives spontaneous
M/Mo

fractional magnetization Ms/Mo

a=µH/kBT
The Curie temperature Tc :
k BTc 1
= Tc =
µγM o 3
Hence, M =  k B T a =  T a
M o  µγM o   3Tc 
achieved at that temperature.
Ms decreases with increasing T and
reaches 0 at T2.
When T is higher than T2, the
spontaneous magnetization vanishes
implying that T2 is Tc.
µγM o
3k B
intersects w/.
M
= L(a ) at P
Mo
determining Ms/Mo.
The law of corresponding states :
All ferromagnetic materials, which naturally have different values of
Mo and Tc have the same value of Ms/Mo for any particular value of T/Tc.

Very nearly, but not exactly, correct

M in the calculation is magnetic moment per unit volume, emu/cm3 .

n: # of atoms per unit volume, changes with temperature.
Exact statement : take M[emu/g] = M[emu/cm3] / ρ[g/cm3] instead of M.
All materials have the same values of Ms/Mo for the same value of T/Tc.

Μ  k BT   T 
=  a =  a
M o  µγρM o   3Tc 
Ms/Mo

The Weiss’s prediction of a law of

corresponding states is verified.
But the shape of the curve is wrong.
Weiss-Langevin : J=∞ T/Tc
As discussed in the previous chapter, Ch.11.
QM need to be taken into accounted for the source of M.
M 2J + 1  (2J + 1)a '  1  a' 
= B J (a' ) = coth   − coth  
Mo 2J  2J  2J  2J 
gJµ B H µ H H
where a ' = =
k BT k BT
M  k BT 
The straight line : =  a'
M o  µ HγρM o 
1  J +1
The slope of BJ(a’) at the origin is :  
3 J 
µ HγρM o  J + 1  g (J + 1)µ BγρM o
Curie temperature Tc =  =
3k B  J  3k B
Μ  J + 1  T  Μs  Ms / Mo 
=  a' = tanh  
M o  J  3Tc  J=1/2 Mo  T/Tc 
in fairly good agreement with experiment
 J=1/2 implies that the magnetic moment is due entirely to spin, g=2
and there is no orbital contribution.

Experimental data Ferromagnetism is due

g* g+ essentially to electron spin
with little or no contribution from
Fe 2.12 2.08 orbital motion of the electrons.
Co 2.22 2.18
At 0K, the spins on all the atoms are
Ni 2.2 2.09 parallel, in one direction. At a higher
Cu2MnAl 2.01 2.00 temperature, a certain fraction of the
total, determined by the Brillouin
Ni78Fe22 2.07 2.11 function, flip over into the other direction;
Ni79Mo5Fe16 2.12 2.10 the value of that fraction determines the
value of Ms.
Reduction of Ms (T≠0) is due
At 0K to excitation of spin waves.

At T≠0
 In the calculation shown above, we only consider the effect of the
molecular field and put the applied field H equals to zero.
Now we apply a magnetic field H, µ H (H + H M ) µ H (H + γρM )
a' = =
k BT k BT
Μ  k BT   k BT  µ H H  k BT  H
=  a'−  =  a'−
M o  µ HγρM o   µ HγρM o  k BT  µ HγρM o  γρM o

B(a’)
Curve is the Brillouin
J=1/2 function for J=1/2.
M/Mo

Lines 2 & 4 represent the

molecular field alone.
The dashed lines 2’ & 4’
represent the molecular
and a applied field.
H
γρM o a’
Above the Curie temperature T>Tc
For instance, T=1.2Tc, the effect of the applied field move the intersection
point of the field line and the magnetization curve from the origin to point B.

Μ  J +1
Near the origin = a'
M o  3J 
Μ µ H Μ o (J + 1) / 3k B J C'
Susceptibility χ = = =
H T − µ HγρΜ o (J + 1) / 3k B J T −θ
µ H Μ o (J + 1) µ HγρΜ o (J + 1)
C' = and θ=
3k B J 3k B J
θ is the temperature at which susceptibility becomes infinite.
Tc is the temperature at which the spontaneous magnetization becomes 0.
µ HγρM o  J + 1 
Tc =   identical
3k B  J 
 Below the Curie temperature T<Tc
For instance, T=0.5Tc, the effect of the applied field shift the intersection
point of the field line and the magnetization curve from point P to point P’.

Since the magnetization curve is nearly flat in this region, the increase
in relative magnetization from P to P’ is very slight.
“molecular field HM” : γρMs
spin
k BTc
H M = γρM s = Ms
µH M o
J=1/2
ferromagnetic paramagnetic
For Fe, µo=221.9emu/g

(1.38 ×10 )
erg/K (1043K )
M/Mo

-16
= (218emu/g )
( )
2.06 ×10-20 erg/Oe (221.9emu/g )

= 6.9 ×106 Oe
Co H M = 11.9 ×106 Oe
T(K)
Ni H M = 14.7 ×106 Oe
These fields are very much larger than any continuous field yet
produced in the laboratory.
However, the molecular field is in no sense a real field, but rather a
force tending to make adjacent atomic moments parallel to one another.

Ferromagnetic
N
Mo[emu/g] µH[µB] M o = μH
A
Fe 221.9 2.22 N : Avogadro’s number
Co 162.5 1.72 A : Atomic mass
Ni 57.5 0.60
µ B = 0.927 ×10 −20 erg/Oe

The huge difference between a ferromagnetic and paramagnetic is

due to the degree of alignment achieved and not to any large
difference in the size of moment per atom.
Exchange energy
Play an important part of the total energy of many molecules and
of the covalent bond in many solids.
Heisenberg showed that it also played a decisive role in ferromagnetism.

Ferromagnetism is due essentially to electron spin and reduction

of Ms (T≠0) is due to excitation of spin waves.
Quantized spin waves : Magnons
In general, consider N spins
 
N
U = −2J ∑ Si • S j
where J is called exchange integral and
spin angular momentum of atom i : Si
i, j=1
If J is positive, U is a minimum when the spins are parallel (φij=0)
and a maximum when they are anti-parallel (φij=π).

If J is negative, U is a maximum when the spins are parallel (φij=0)

and a minimum when they are anti-parallel (φij=π)
As we have already seen, ferromagnetism is due to the alignment of
spin moments on adjacent atoms.
J>0 for ferromagnetism

The ground state of a simple ferromagnet has all spins parallel

N 
U = −2J ∑ Si • Si +1
Treat Si as
classical vector U o = −2JNS2
i =1

What is the energy of the first excited state?

One particular spin is reversed.

OR
U = U o + 8JS 2
we can form an excitation of much lower energy if all spins share the reversal
The elementary excitations of a spin system
have a wave like form and called magnons.
Analog to lattice vibrations or phonons
 A spin wave on a line of spins

Side view

Top view

A classical derivation of the magnon dispersion relation

For the pth spin
 
( 
− 2JSp • Sp −1 + Sp +1 )
The magnetic moment at site p : µp = - gµBSp

µ

− 2J −
p 

( 
)   2J 
 • Sp −1 + Sp +1 = − µ p • − (  
Sp −1 + Sp +1 )
 gµ B   gµ B 
Effective magnetic field or exchange
Classical mechanics field on the pth spin Bp.

dSp  
 = µ p × Bp
dt
 
  2J       
(Sp-1 + Sp+1 ) = 2J (Sp × Sp-1 + Sp × Sp+1 )
dS p
 = − gµ BSp ×  −
dt  gµ B 
Assuming that Sp << S, Sp << S, and Sp = S
x y z
In Cartesian coordinate
dSpx
= (Sp Sp-1 + Sp Sp+1 − SpSp-1 − SpSp+1 ) 
2J y z y z z y z y = (
2JS y y
2S p − S p -1 − S p +1 )
y

dt 
dSpy
= (SzpSpx-1 + SzpSpx+1 − Spx Szp-1 − Spx Szp+1 ) = − ( 2Sp − Sp-1 − Spx+1 )
2J 2JS x x
dt  
dSzp
= (Sp Sp-1 + Spx Spy+1 − SpySpx-1 − SpySpx+1 ) = 0
2J x y
dt 
Spx = u exp[i(kpa − ωt )]
A trial solution set Spy = v exp[i(kpa − ωt )]
Szp = S
 − i ωu =
2JS
(2 − e − e )v = (1 − cos(ka ))v
−ika ika 4JS
 
− i ωv = −
2JS
(2 − e − e )u = − (1 − cos(ka ))u
−ika ika 4JS
 
4JS
For a set of solution
iω (1 − cos(ka ))
 =0
4JS
− (1 − cos(ka )) iω

2
 4JS
−ω + 
2
(1 − cos(ka )) = 0 Spx = u cos(kpa − ωt )
  
Spy = u sin (kpa − ωt )
ω=
4JS
(1 − cos(ka )) and u = −iv Szp = S

At long wavelength
ω ≅ 4JS
(ka )
2
( )
= 2JSa 2 k 2
ka<<1 2
Dispersion relation for magnons in a ferromagnet
in 1D w/. nearest-neighbor interactions

ω
2.0 = 1 − cos(ka )
4JS
1.5
k2 may be determined accurately
ω/4JS

by neutron scattering or by spin

1.0
wave resonance in thin films.
0.5
At long wave length limit,
0.0
0.00 0.25 0.50 0.75 1.00
( )
ω = 2JSa 2 k 2 = Dk 2
ka/π
Fe Co Ni

D 281 500 364 meVÅ2

T= 295K
by Neutron scattering
Magnon spectrum obtained by neutron scattering experiment

  '
kn kn

k

 2 k 2n  2 k '2n
= + ω k
2M n 2M n
  1
Quantization of spin waves ε k =  n k + ωk
 2
In thermal equilibrium the average value of the number of magnons
1
nk =
exp(ωk / k BT ) − 1
Planck distribution
3
The number of magnons from ω to ω+dω D(ω )dω = 
1 
 4πk dk
2

1  2π 
2
 1 
3 k ω 
D(ω ) =  π
2 dk 2 1
 4πk = 2 =
 2π  dω π
2 2 2 2
2JSa 2 2 JSa 4 JSa
2k

ω = (2JSa 2 ) k 2
3/ 2
1   
= 2  2 
ω
4π  2JSa  dω 2JSa 2
= 2k
dk 
∞ ∞
ω
3/ 2
Total number of dωD(ω ) n = dω 1   
magnons ∫
0
k
0
∫
2  2 
4π  2JSa  exp (ω / k BT ) − 1
 ω
= 0.0587(4π 2 )
∞ x
Let x = and ∫ dx x
k BT 0 e −1
∞
ω / k B T
3/ 2 ∞
d ω
3/ 2
1     k BT 
∫ dωD(ω ) n
0
k = 2  2 
4π  2JSa 

  
 ∫0 k BT exp (ω / k BT ) − 1
 k BT 
= 0.0587 
3/ 2
1 ( )
= 0.0587 4π 2
 2JS  a3
SC BCC FCC
The number N of atoms per unit volume:
1/a3 2/a3 4/a3
Q/a3

Fractional change of magnetization

∆M ∑ n k 0.0587  k BT 
3/ 2
= =  
M (0) NS SQ  2JS 
Bloch T3/2 law In agreement with
experimental data
 Tc=627K
Ni Mo=510 gauss

At T=60K ~ 0.1Tc
∆M
≈ 2 ×10 −3
Mo

∆M A=(7.5±0.2)x10-6 K-3/2 Ni
Bloch T3/2 law = AT 3/2
Mo A=(3.4±0.2)x10-6 K-3/2 Fe
This application of band theory to magnetic problems was made by
Stoner, Mott, and Slater in 1933. (the collective-electron theory)

Why µH(0K) are 2.22, 1.72, and 0.6 µB for Fe, Co, and Ni, respectively?

Electron distributions in free atoms

K Ca Sc Ti V Cr Mn Fe Co Ni Cu Zn
3d 0 0 1 2 3 5 5 6 7 8 10 10
4s 1 2 2 2 2 1 2 2 2 2 1 2
3d+4s 1 2 3 4 5 6 7 8 9 10 11 12

The occupation of energy levels is in accordance with

the Pauli exclusion principle.

When atoms are brought close together to form a solid, the position of
energy levels are profoundly modified.
overlapping of electron clouds
 Splitting
of levels

When d=do, the 3d levels are

spread into a band extending from
B to C, and the 4s levels are
spread into a much wider band
extending from A to D.
Why does the band spread wider in
4s than 3d?
A: 4s electrons are farther from the
nucleus.
Interatomic distance d

An important and difficult problem of the band theory is to calculate the

“shape” of energy bands, i.e., DOS : N(E) for the band.
 Density of states N(E)=dN/dE
The states are full of electrons, up to the Fermi energy at 0K.
The total number of spins in the subband with spins up is different
from that in the subband with spins down.

Net magnetization Molecular field or exchange interaction

The nonintegral values of the magnetic moments are easily explained

by the complex shapes of the bands.
 Ni

dN dN dN
dN
dE dE dE
dE
Co Fe
 4s and 3d bands in Cu

Fermi surface
2.2eV
7.1eV 3.46eV

3d 3d↑ 3d↓
4s 5 electrons 5 electrons
Filled-10electrons

4s and 3d bands in Ni
below Tc
above Tc

0.27 hole 0.54 hole

0.54
electron 0.54 5
4s 3d↑ 3d↓ electron 3d↑ 3d↓ electron
4.73 electron 4.46 electron
 n = number of 3d + 4s electrons per atom
x = number of 4s electrons per atom
n − x = number of 3d electrons per atom
The magnetic moment per atom for this series of transition metal is
predicted to increase as number of n-x of 3d electrons increase up
to n-x=5 and then decrease to zero as number reach 10.
At saturation (T=0K), the net moment is the difference between
spin-up states (n-x=5) and the spin-down states, n-x-5:
µ = [5 − (n − x − 5)]µ B = [10 − n + x ]µ B
For Ni, µ=0.6µB,
x=0.6
Hence,

µ = [10.6 − n ]µ B
The agreement is generally
good for Fe, Co, and Ni.
 Magnetic alloys : Alloys of the transition metals or rare earth metals

Chemical valence Z = N ↑ + N ↓ Valence electrons

Magnetization of the atom (
µ = N↑ − N↓ µB ) (
µ = 2N↑ − Z µB)
For the transition metals, N↑ is determined from the fact that the
spin-up d bands lie entirely above or below the Fermi level.

Magnetic valence Z m = 2 N ↑d − Z
Other contribution from the 4s band (actually 4sp hybrid band) N s↑ = 0.3µ B
The magnetic moment per atom ( )
µ = Z m + 2 N ↑s µ B = (Z m + 0.6 )µ B

Ex. Fe0.8Co0.2 alloy : Z m = 2 × 0.8 + 1× 0.2 = 1.8

µ = (1.8 + 0.6 )µ B = 2.4 µ B
The average magnetic valence concept can then be used to predict
the resulting magnetic moment when transition elements are alloyed.
 The Slater-Pauling curve :
The curve of magnetic moment per atom versus electron-to-atom ratio.

Assuming 0.3µB for the sp band spin-up electrons.

Magnetic valence Zm=2Nd+-Z
Band model contribution to magnetic moment per atom
Ericksson et al. considered the contributions of the 3d and 4sp hybrid bands
and obtained the magnetic moment resulting from the excess holes (Nh) in
the spin-up and spin-down 3d bands..

Parameter Fe Co Ni
Excess holes (Nh) 3.4 2.5 1.5

µspin3d 2.26 1.64 0.64

µspin4sp -0.07 -0.07 -0.02

µspintotal 2.19 1.57 0.62

µorbit3d 0.09 0.14 0.07

µ total 2.28 1.71 0.69

Experimental value 2.20 1.72 0.6
Criteria for the existence of ferromagnetism in a metal :

The electrons responsible must lie in partially filled bands in order

that there may be vacant energy levels available for electrons with
unpaired spins to move into.
Ruling out inner core electrons

The density of states in the band must be high, so that the increase
in energy caused by spin alignment will be small.
Ruling out valence electrons

The atoms must be the right distance apart so that exchange force
can cause the d-electron spins in one atom to align the spins in the
neighboring atom.

Only Fe, Co, and Ni of transition metals are ferromagnetic.

Many of rare earths are ferromagnetic below room temperature
due to spin imbalance in their 4f bands.
1839
1803
1885
1885
1945
1879
1896
1880
1878
1886
1879
1843
1828
1878
1907
 Bethe-Slater curve:
postulated variation of J
with the ratio ra/r3d.
ra :the radius of an atom
r3d : the radius of its 3d
shell of electrons

Exchange forces are responsible for magnetism, as well for

antiferromagnetism, and ferrimagnetism.
Antiferromagnetism
M =0
Spins are ordered in an antiparallel arrangement with zero net moment.
Below TN :Néel temperature
A small positive susceptibility present at all temperatures but varies in a
peculiar way with temperature.
a weak cusp at T=TN

1904-2000

1970
Anomalous paramagnetic
C
T > TN, χ∝
T +θ
like Curie temperature
w/. Tc = -TN
Paramagnetic
T < TN, anti-ferromagnetic order
Some antiferromagnetic materials

material TN(K) θ(K) χ(0)/χ(TN)

MnO 122 610 0.69
FeO 198 570 0.78
CoO 293 280
NiO 523 3000 0.67
Fe2O3 950 2000
Cr2O3 307 1070 0.76
FeS 613 857
FeCl2 24 -48 <0.2
FeF2 79 117 0.72
MnO2 84 116 0.93
α-Cr 310

Most are ionic compounds :oxides, sulphides, chlorides, and the like.
Manganese Oxide MnO: Ordered arrangements of spins of the Mn2+ ions.

A
B Antiferromagnetic arrangement
of A and B sublattices in 2D
A
B
 Considering only the nearest neighbor interaction AB
H mA = −γM B Molecular field acts on the ion A due to magnetization of B
H mB = −γM A Molecular field acts on the ion B due to magnetization of A
Μ C M A T = C ' ρ (H − γM B )
Above TN, χ= =
ρH T M BT = C ' ρ (H − γM A )
Curie law
(M A + M B )T = C ' ρ (2H − γ (M A + M B ))
MT = 2C ' ρH − C ' ργM
(T + C ' ργ )M = 2C ' ρH
Μ 2C '
Therefore, χ= =
ρH T + C ' ργ
When a field is applied above TN, each sublattice becomes magnetized
in the same direction as the field, but each sublattice set up a molecular
field in the opposite direction to the applied field, tending to reduce both
MA and MB.

Hence, the susceptibility χ is smaller than that of an ideal paramagnetic

in which the molecular field is zero.
Below TN, Each sublattice spontaneously magnetized, in zero applied
field, by the molecular field created by the other sublattice.

M = MA + MB = 0
At T=TN, M A TN = −C ' ργM B where H=0
MA
C ' ργ = − TN = TN
MB
The Néel temperature at which χ(T) is the maximum equals to θ.
Below TN, each sublattice is spontaneously magnetized to saturation
just as a ferromagnetic is.
MA  µH  Brillouin function
= B J,
 k T
M oA  B 

For the spontaneous magnetization in the absence of applied field,

H mA = −γM B = γM A = γρM A
MA  µ HγρM A 
= B J, 
M oA  k BT 
 MA MnF2
M oA

T
TN

MB
−
M oB

The net spontaneous magnetization is zero below TN.

However, an applied field can produce a small magnetization. χ
χ depends on the angle bet. the applied field and its spin axis.
A
B Spin H
H axis
A
B χ║
χ┴
Field at right angle to spin axis M
Assuming that the applied field turn each MB
MA
sublattice magnetization away from spin axis α H
by a small angle α.
This rotation immediately creates a HmA
magnetization M in the direction of H. HmB
 
The spins will rotate until that H mA + H mB
cancel the applied field H. Hm

2H mA sin α = H = 2γρM A sin α

and M = 2 M A sin α H = γρM
Therefore, M 1 C
χ⊥ = = = a constant
H γρ 2θ
Independent of temperature
Field parallel with spin axis
Assuming that the applied field H MB MA
increase the zero-field value of the A-
sublattice magnetization by ∆MA and ∆MB
decreases the corresponding value of H ∆MA
the B-sublattice by ∆MB.

A net magnetization in the direction of H

M = M A − M B = ∆M A + ∆M B

B(J,a’) nearly ∆M A = ∆a ' M oA B' (J, a o ' )

linear
∆M B ∆M A
µH
M oB M oA ∆a ' = (H a − γρ∆M B )
M/Mo

k BT
µH
= (H a − γρ∆M A )
k BT
ng µ H2
∆M A = (H a − γρ∆M A )B' (J, a o ' )
a’=µHH/kBT 2k BT
 M 2∆M A 2ng µ H2 B' (J, a o ' )
Therefore, χ11 = = =
Ha Ha 2k BT + ng µ H2 γρB' (J, a o ' )
temperature dependence
C
It becomes χ = 2θ at T=TN.
C
It reduces to χ ∝ at high temperatures.
T +θ
It approaches to zero as T approaches 0.
Theoretical calculation for J=1
For a powder specimen, there is
no preferred orientation of the
crystals.

χ p = χ11 cos 2 θ + χ ⊥ sin 2 θ

1 2
= χ11 + χ ⊥
3 3
Ferrimagnetism

Ferrimagnetic materials exhibit the phenomena of magnetic saturation

and hysteresis, like ferromagnetics.

The most important ferrimagnetic materials are certain double oxides of

iron and another metal, called ferrites.

The ferrites were developed into commercially useful materials. 1933

Si0.03Fe0.97 (Silicon ferrite) ….

Until 1948, ferrimagnetics were separated from ferromagnetics.

Néel provided the theoretical key to understand the ferrites.

They fall mainly into two groups with different crystal structures :

Cubic : MO•Fe2O3 where M is a divalent metal ion, Mn, Ni, Co, Fe,…
Ironferrite FeO •Fe2O3 is the oldest magnetic material to man.

Hexagonal : BaO •6Fe2O3 Barium ferrite

 Cubic ferrite
Hexagonal ferrite

metal ion Oxygen ion metal ion

Tetrahedral A site Octahedral B site Barium ferrite

Key: Lattice with bases of two magnetic atoms
Moments anti-align, but do not cancel
Eg. Magnetite Fe3O4=FeO • Fe2O3
parallel Fe2+ Fe3+ antiparallel

M ≠0

Why is it different from ferromagnetic ?

How to know this fact ?

Fe3+ ions have a spin of 5/2 and should contribute 5µB

Fe2+ ions have a spin of 2 and should contribute 4µB
expectation Effective moment (2×5+4) µB = 14 µB at T=0K

results It show only 4.1 µB only from Fe2+ ions

confirmed by Neutron diffraction
Magnetization : complex behaviors

FeO • Fe2O3

T > Tc, paramagnetic

MA + MB
χ= Let CA and CB be Curie constants for ions A and B.
ρH
M A T = CA ρ (H − γM B )
M BT = CB ρ (H − γM A )
For non-zero
(C A + C B )T − 2 ργC A C B
solutions MA and MB Tc = ργ C A C B χ=
At H=0 T 2 − Tc2
Below Tc MO • Fe2O3

M (emu/g) Cubic ferrite

T(oC)
M/Mo
 Below Tc Iron Garnets
M3+ is a trivalent metal ion.
M3 Fe5O12 3 Fe3+ ion on tetrahedral site d
2 Fe3+ ion on octahedral site a
3 M3+ ion on site c

Y3+ is a diamagnetic ion

At T=0K, resulting in 5µB
M decreases w/. increasing T and
reaches to zero at T=Tc.

Rare earth M3+ :are paramagnetic

and magnetized (c) opposite to the
resultant of the Fe3+ ions (a+d).
M drops rapidly w/. increasing T
The compensation temperature Due to weak c-a and c-d coupling.
at which the magnetization crosses zero. Then M passes through zero and
increase again due to Fe3+.
T/Tc T/Tc
Neutron magnetic scattering
An X-ray photon see the spatial distribution of electronic charge, whether
the charge density is magnetized or unmagnetized.
A neutron sees two aspects of a crystal:
the distribution of nuclei and the distribution of electronic magnetization.

Neutron A neutron can be inelastically scattered by the

k n’ magnetic structure,
kn With creation or annihilation of a magnon.
magnon spectra
 2 k 2n  2 k '2n
= + ωk
Magnon 2 Mn 2 Mn
Neutrons are uncharged, easily penetrate electron cloud,
and are scattered only by nucleus.
If the scattering atom or ion has a net magnetic moment,
that moment will interact with neutron beam, because
the neutron has a small magnetic moment, ~0.3µB.
 Neutrons M=1.67x10-27kg
) = 0.28
λ(A When E=80meV , λ=1Å
E(eV)

µ=5.4×10-4µB
 Body-centered tetragonal

BCT structure

MnF2 T=23K

T=300K

MnF2 TN=67K
 T=80K

TN=122K

MnO
T=293K
MnO

The first substance to be clearly recognized as antiferromagnetic in 1938.

However, the first direct evidence came from Neutron diffraction experiment
by Shull and Smart in 1949.
 Magnetic Domains

Ferromagnetic materials are not uniformly magnetized, but break up

into regions called domains. The magnetization for each domain will,
in general, have a different orientation.
Origin of domains – magnetic, exchange, and anisotropy energies

Magnetic field energy

Introducing
more domains

µo
E= ∫ dV decreasing the magnetic field energy
2
H
2
However,
it increases the wall energy.
 For a 180o magnetization reversal in one step,
an exchange energy per wall area has to be
overcome: 2 2
Wex = 2
JS
a
For a 180o magnetization reversal in N steps,
an exchange energy per wall area is reduced:

N 2 π 
2

Wex = 2 JS  
a: lattice constant
a N

Domain wall orientations

Bloch Wall Néel Wall

 Anisotropy Energy

The magnetic moments in a magnetic material tend to line up

preferentially along certain crystallographic direction.
The exchange energy depends on the orbital overlap of electronic
wavefunctions between electronic orbits on different sites.

With crystal anisotropy the rotation away from the easy axis
( )
costs extra energy U K = K1 α12α 22 + α 22α 32 + α 32α12 + K 2α12α 22α 32
where k1 and k2 are the magnetic anisotropy constants
 N spins in a Bloch wall with lattice constant a

U K = KNa
N 2 π 
2

Wex = JS  
N
2
a
The energy per unit area of the wall

N 2 π 
2

σ W = Wex + U K ≈ 2 JS   + KNa
a N
∂σ W π
2
 
Minimizing σW, = −JS 2   + Ka = 0
∂N  Na 

π JS
2 2 KJS2
N= and σ W = 2π
Ka 3 a

For Iron, σW ≈ 1 erg/cm2 (180o wall, N≈300)

 Magnetic hysteresis

Saturation
Remanent magnetization
magnetization

Coercive field
Saturation
magnetization Demagnetized
state
1/χ 1/χ

T
T
Diamagnetism Paramagnetism

Ferromagnetism

Ms χ M 1
χ= ~ Curie-Weiss law
H T −θ
Ferromagnet
Paramagnet
T
Tc
Curie temperature
A
B
A
B

Anti-ferromagnetism TN

Néel temperature

1/χ
A Ms
B
A
Ferrimagnet
B Paramagnet
T
Tc
Ferrimagnetism Curie temperature
Magnetic Devices

Permanent Magnets
Transformers
Magnetic Amplifiers
Data Storage
 Spin + Electronics
Spin

Magneto- Spin
Electronics Photonics a multidisciplinary field including
magnetism, semiconductor physics, optics,
mesoscopic physics, superconductivity
Quantum and new connections to other fields
Electronics

The central theme is

how to manipulate the spin degree of freedom
which interact with the solid-state environment.
Spin-dependent transport
Conventional transistors make use of current or voltage to
control the transmitted current.
Electronics

Nature 404,918 (2000)

Use the spin configurations to control current

Spintronics
 The most commonly built structures
for spin-dependent transport make use of
(1) Giant magnetoresistance effect (GMR)
(2) Tunneling magnetoresistance effect (TMR)

M.N. Baibich et al., Phys. Rev. Lett. 61, 2472

(1988).
Tunneling magnetoresistance effect (TMR)
Tunneling conductance across the barrier is proportional to the
product of density of states on both sides.

G ∝ N1,↑ N 2,↑ + N1,↓ N 2,↓

R parallel < Ranti− parallel

GaMnAs/AlAs/GaMnAs
tunneling junction
M. Tanaka and Y. Higo,
Phys. Rev. Lett. 87, 026602
(2001).
 Datta-Das Spin field-effect transistor
Datta and Das,
Appl. Phys. Lett. 56, 665 (1990).

gate

How to maintain spin coherence ?

 MRAM IBM MRAM images
Magnetic RAM chips use
magnetic rather than electrical
structures to store information, so
they do not need to be constantly
powered to retain data, like
current RAM technologies.
much faster and less expensive

1989 - IBM scientists made a string of key discoveries

about the "giant magnetoresistive" effect in thin-film structures.
2000 - IBM and Infineon established a joint MRAM development program.
2002 - NVE Announces Technology Exchange with Cypress Semiconductor.
2003 - A 128K bit MRAM chip was introduced, manufactured
win 0.18 technology.

2004
June - Infineon unveiled a 16M bit prototype based on 0.18
September - MRAM becomes a standard product in Freescale,
which has began sampling MR
 NEC, Toshiba claim

Magnetorestitive RAM could

replace flash memory and
DRAM by as early as 2010

One issue involves the

size of MRAM cells, which
tend to be bigger than
those of other memory
types.
New Technique could
allow them to develop
256M bit MRAMs by early
2006