Chapter Nine

Fermi surfaces and Metals

To get band structure of real crystals, turns on weak periodic potential.

Band gaps open up at BZ edges.
To calculate electronic properties, put in electrons (Fermions).
fill them up to Fermi energy F.
At T=0, the Fermi surface separates the unfilled orbits from the filled orbits.
The electrical properties of the metal are determined by the shape of the
Fermi surface, because the current is due to change in the occupancy of
states near the Fermi surface.
Aluminum (v=3)

Copper (v=1)

Zone 1

Zone 1

Zone 2

Zone 3

1D chain

(2h2/2ma2)

Extended-zone scheme

25

16
9
4
1
-3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0

k(2/a)

/a

-/a

k
5

(2h2/2ma2)

Reduced-zone scheme
25

Free electrons (k) :

16

modulated by lattice
periodicity
9
4
1
-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0

k(/a)

All in the first BZ.

Two dimensional square lattice

2.0

E( h /2ma )

1.5

2 2

1.0

0.5
0.0
-1.0
-0.5

k(
x

0.5

0.0

/a
)

0.0

0.5

-0.5
1.0

-1.0

( / a

ky

1.0

Construction of Brillouin zones : bisect all G

/a

-/a

1D

k
5

2D

ky
3

translate region into 1st zone by G

to form reduced zones

3
2a

3d

1st zone

2b

2d

kx

3a
2c
3

3d

2c
2d

2b

2a
2nd zone

3a
3rd zone

Construction of free electron Fermi surface

ky
1st zone
3

Fully occupied

2
3

kF

1
2

kx

2nd zone

3rd zone

3d

2
3

3
3a

The shaded regions are filled with

electrons and are lower in energy
than the unshaded regions.
hole-like

electron-like

One electron per primitive cell v=1

Free electron

With the crystal potential,

the energy inside the first Brillouin zone
is lower close to the zone boundary.

+ weak periodic Potential

constant
So the Fermi surface is extended
toward the zone boundary as it get
closer.

[010]
Fermi surface
[100]
Fermi surface is distorted
from a sphere
near the zone boundary.
BCC Li
A cusp is caused by interaction

Two electrons per primitive cell v=2

1st band Fermi surface

Free electron

[110]

[100]

2nd band Fermi surface

(1) kF << kBZ No dispersion

Constant contours are spheres.

kF

BZ boundary

(2) kF <~ kBZ Fermi surface bulges toward BZ boundary

k > kF
kF

e
rf e

ro
t
c
le

K bulges toward
BZ boundary
due to dispersion.

kF/a

(3) k = kBZ Bragg scatterings open energy gap

1
v(k) = k (k) = 0 at zone boundary.
h

Standing waves

(4) k > kBZ electron states in second or higher bands

corresponding to higher order Brillouin zones of k space.

Nearly free electrons

The interaction of the electron with periodic potential of the crystal
causes energy gaps at the zone boundary.
Fermi surfaces will intersect zone boundaries perpendicularly.

The crystal potential will round out sharp corners in the Fermi surface.

Fermi surface: surface in k-space separates filled and unfilled states

Only metals have Fermi surfaces.
Important because electronic properties depend on electron states near F
within kBT
Heat capacity
Electric conductivity
Thermal conductivity

2
D( F )k 2BT
Ce =
3

1 2 2
= e v F D( F )
3

=LT

Volume of Fermi surface only depends on conduction electron density.

Shape of Fermi surface depends on strength of periodic potential and
size of kF relative to kBZ

Density of states depends on (k) , actually

State number between and +d in band

r
r
k
vg =
h

In two dimensions, D(k ) = L uniform in k-space

2
r
r
D( )d = D(k)dk x dk y = D(k) 2 kdk
Area of k-space

+d

= dkk
dk
d
d
d
= dk
hv g
= dk

Therefore,

1
L
D( ) = 2 dk
path integral along constant contour
k
2

 L
In three dimensions, D(k ) =
2

uniform in k-space

1
L
D( ) = 2 dS
k
2

area integral over constant surface

Crystal is not cubic,
[010]

Crystal is cubic,

[100]

Q
R
D()

P
P

D()

multiple peaks

How to determine the Fermi surface?

Magnetic field response direct probe of the Fermi surface

r
r
r r
dk
F = qv B = h
dt

and

r r
F v = 0

Magnetic field drives electrons in k-space along constant contours.

electron orbit
r
B

r
r
F dk

electron orbit
dt

r
v k

Free electron
hole orbit
r
B

Free hole

r
v k
r
r
F dk
dt
Nearly free electron
r
B

hole orbit

r
v k
r
r
F dk

dt

Period of orbit
Lorentz force
Period

r
r
r r
dk
h
F=h
= qv B dt = r r dk
dt
qv B
1
h2
T = dt =
dk
qB k
constant
1

and

=
k = k =
k
k
k
1

h2
h2
h2
k

dk =
dk
S( , k z )
PeriodT =

qB

qB
k qB
where S is the k-space area enclosed by the orbit in its plane
Free electron

hk F
v=
= vF
m

h
2 m
(2 k F ) =
T=
eBv F
eB
2 eB cyclotron
=
c =
T
m frequency

De Haas-van Effect : oscillation of the magnetic moment of a metal

as a function of magnetic field
(1930)
-M/H (106)
Bi (1930)&(1932)

H is along [111] direction noble metal

Bi

R()
T=1.3K

A111(belly)/A111(neck)=51 Ag

Magnetoresistance of Ga

Calculation of energy bands

z The tight-binding method
z The Wigner-Seitz method
z The pseudopotential method extension of the OPW method
Orthogonalized plane-wave

Push isolated atoms together to form crystal

An isolated atom

very far away

two isolated atoms

Two atoms move closer to each other.

Two energy levels

A-B

A+ B

1.2
1.0
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
-0.6
-0.8
-1.0
-1.2
-1.4

A+ B
A-B

-6

-4

-2

Solid with N atoms has N allowed energy states.

When more atoms are brought together,

the degeneracies are further split to form
bands ranging from fully bonding to fully
antibonding.
Different orbitals can lead to band overlap.

There are two idealized situations for which wave functions can be
expressed in a simple manner and
an energy band calculation can be carried out with relative case.
Energies are far above the maxima of potential energy.
Nearly Free Electrons
Energies are deep within the potential wells at nuclei.
Tightly Bound Electrons

Influence of the periodic potential depends both on

the magnitude of this potential and on the opportunities for atoms
to interact which varies with the interatomic spacing.

Tight binding method (Linear Combination of Atomic Orbitals)

For an interatomic spacing which permits some overlaps between atoms
(but not very much), the bands can be stimulated.

It is quite good for the inner electrons of atoms, but not for the
conduction electrons.
the d bands of the transition metals
the valence band of diamond-like materials
inert gas crystals

Tight binding method

Free atoms s

introduced by Bloch in 1928

Overlapping s

1952

1905~1983
r
Let kr ( r ) is the ground state of an electron moving in the potential U(r)
of an isolated atom.

r h2 2
r r
r
r
r
r
H atom k ( r ) =
+ U( r ) k ( r ) = E k k ( r )
2m

Let kr ( r ) is for the electron moving in the whole

r crystal that contains
N isolated atoms.

Atoms are at lattice sites

rj (j=1,.,N)

r
r r
r
r
A trial wavefunction ( r ) = C k k ( r rj )
r
k

N linear combinations

[ (

r r
r r
1
=
exp ik rj k ( r rj )
N j

satisfying Bloch condition

[ (

r r
r r r
r r
1
kr ( r + T) =
exp ik rj kr ( r + T rj )
N j
r-(r -T)
r r
r r r
r rj r
1
exp ik T exp ik rj T kr ( r + T rj )
=
j
N
r r
r
= exp ik T kr ( r )

) [ ( (

))

Schrdinger equation

r
r
r
r
r
r
H crystal ( r ) = [H atom + U( r )] k ( r ) = k k ( r )
r
k

r
Where U( r ) contains all corrections to the atomic potential required to
produce the full periodic potential of the crystal.

First order energy

) (

r r
r r
1
kr H crystal kr = exp ik rj exp ik rm m H crystal j
N j m
r r
r r
(
where m = ( r rm ) and j = r rj )

r r
U( r rn )

r
U lattice ( r )

r r
r r
U( r rn ) = U lattice ( r rm )
m n

r r
r
r r
m H crystal j = dV ( r rm )[H atom + U( r )] ( r rj )

Rewrite the first order energy

H crystal
r
k

r
k

r r r
r r r
r r
r
1
= exp ik (rj rm ) dV (r (rm rj )) H atom + U( r + rj ) ( r )
N j m
r r
r
r
r
r
= exp ik m dV ( r m )[H atom + U( r )] ( r )
m

r r
r
r on the same atom
r
dV ( r m )H atom ( r ) = dV ( r )H atom ( r )
r
r r
r r
r
r r
r
r
r
dV

(
r

U(
r
)

(
r
)
=
dV

(
r
)

U(
r
)

(
r
)
+
dV

(
r

U(
r ) ( r )

Overlap, up to nearest neighbors

r
r
dV ( r )H crystal ( r ) -

m=0

Therefore,

k = k H crystal k

r r
= exp ik
n.n.

m0, n.n.

For a simple cubic,

the nearest neighbor atoms = (a, 0,0), (0, a, 0), (0, 0, a)
r r
k = exp ik
n.n.

[
= [2cos(k a ) + 2cos(k a ) + 2cos(k a )]

= exp(ik x a ) + exp(- ik x a ) + exp(ik y a ) + exp(- ik y a ) + exp(ik z a ) + exp(- ik z a )

x

Along L, [111]

6 k + 6
-
12

An energy band width is 12 .

The weaker the overlap is, the narrower the energy band is.
Constant energy surfaces in the BZ of a SC lattice

Reduced zone scheme

Periodic zone scheme

k = 2 [cos(k x a ) + cos(k y a ) + cos(k z a )]

By series expansion

x2 x4
cos x = 1 + ...
2! 4!

For ka<<1,

k 2 a 2 k 2y a 2 k 2 a 2
+ 1 z
k = 2 1 x + 1
2
2
2

k 2a 2

= 2 3
2

= 6 + k 2 a 2
The constant energy surfaces in the neighborhood of k=0 are spherical.
Effective mass

2
m = h

d 2 k
dk 2

2
= h

d 2 6 + k 2 a 2
dk 2

h2
= 2 2
2 a

The weaker the overlap is, the narrower the energy band is and
the higher the effective mass is.

For a face-centered cubic,

the nearest neighbor atoms = .5a(1, 1,0), .5a(0, 1, 1), .5a(1, 0, 1)

r r
k = exp ik
n.n.

ik x a + ik y a
ik x a + ik y a
ik x a ik y a
ik x a ik y a

+
exp
exp
exp
exp
+
+
+

2
2
2
2

ik y a + ik z a
ik y a + ik z a
ik y a ik z a
ik y a ik z a
+ exp
+ exp
+ exp
+
= exp
2
2
2
2

exp ik z a + ik x a + exp ik z a + ik x a + exp ik z a ik x a + exp ik z a ik x a

2
2
2
2

ik y a
ik y a
k y a
ik z a
ik z a
k x a

+ exp
exp
2cos
+ 2cos 2 exp 2 + exp 2 +
2
2
2

=
ik x a
ik x a

k z a
+
2
cos
exp
exp

2
2
2

k x a k ya
k ya k za
k z a k x a
+ cos
cos
= 4 cos
cos
+ cos
cos

2 2
2 2
2 2

The energies
of an s-band in a FCC crystals

r- space

k- space

Along [ kx=2/a, (01), ky=kz= 0 ]

= 4 [1 + 2 cos( )]

Along L [ kx= ky=kz= 2/a, (00.5) ]

= 12 cos 2 ( )

Along K [ kx= ky= 2/a, (03/4), kz= 0 ]

= 4 [cos 2 ( ) + 2 cos( )]

quite successful for the weakest type of interaction bet. neighboring atoms

d states

Energy gap of the "metallic" single-walled carbon nanotubes

Mod. Phys. Lett. B18, 769 (2004)

Based upon the Slater-Koster tight-binding calculations, we investigated

electronic properties of the "metallic" single-walled carbon nanotubes
(SWNTs) in detail. Our results show that tube curvature may produce an
energy gap at the Fermi level for zigzag and chiral "metallic" SWNTs, and
this effect decreases with the increasing of either the radius or the chiral
angle. Our calculated results are in good agreement with experiments

Calculations and applications of the complex band structure for

carbon nanotube field-effect transistors
PRB70, 045322 (2004)
Using a tight binding transfer matrix method, we calculated the complex
band structure for armchair and zigzag carbon nanotubes (CNTs). The
imaginary part of the complex band structure connecting the conduction and
valence band forms a loop, which can profoundly affect the characteristics of
nanoscale electronic devices made with CNTs. We then study the quantum
transport in carbon nanotube field-effect transistors (CNTFETs) with the
complex band structure effects. A complete picture of the complex band
structure effect on the performance of semiconductor zigzag CNTFETs is
drawn.

Wigner-Seitz method
The cellular method by Wigner and Seitz in 1933

1963

polyhedron structure in real space

quite successful for the simple alkali metals

1902~1995

1935 Kimball extended to nonmetallic materials such as diamond, Si, Ge,

h2 2

+
U(r)

k (r) = k k (r)
2m

and Bloch function k (r) = e

r r
ik r

uk (r)

1
2
(
)
i
k
U(r)

+
h
+
uk (r) = k uk (r)
2m

start with the easiest-found solution at k=0, uo(r) within a single primitive cell
Wigner-Seitz
then, construct the approximation solution

k (r) = e

r r
ik r

uo (r)

boundary condition: , are continuous

The first approximation of the cellular method is the replacement of

periodic potential U(r) within the WS primitive cell by a potential V(r)
with spherical symmetry about the origin.

Radial functions for

k=0, metal Na
3S orbital of free Na atom
3S conduction band in metal Na

as

eigenenergy -5.15eV (atom)

r (Bohr units)
In real metal Na,

k = e

r r
ik r

r
h 2k 2
u o ( r ) and k = o +
2m

As we know, F=3.1eV for Na.

The average KE per e- is 0.6 F=1.86eV

1.15eV less

-8.20eV (k=0)

metal
Metal is more 0eV
Fermi level
stable than
free atom.
-6.3eV
average energy
-8.2eV

k=0 state

Two major difficulties with the cellular method :

The computational difficulties involved in numerically satisfying
the boundary condition over the surface of the WS primitive cell.
The cellular method potential has a discontinuous derivative midway
between lattice points but the actual potential is quite flat there.

Later, a modification
muffin-tin potential

Pseudopotential methods

The theory of pseudopotential began as an extension of OPW method.

The orthogonalized plane-wave (OPW) k
valence electrons + core electrons

k (r) = eik r + b c kc (r)

by C. Herring (1940)

satisfying Bloch condition w/. wavevector k

Outside the core, the potential energy that acts on conduction electron is
relatively weak. Potential due to the singly charged positive ion cores is
reduced markedly by the electrostatic screening of other conduction
electrons.

v (r) = v (r) ( dr c (r)kv (r) ) c (r)

k

exact valence wave function

H kv = kv kv

and

H k
v

v
c
v
c
c
v
c

dr k (r )k (r ) H k = k k c dr k (r )k (r ) k

H kc = kc kc

and

h2 2
pseudo v
(H + VR ) k = k k = + V k
2m

VR kv = kv kc
c

)( dr

effective Schrdinger eq.

adding VR to U : partial cancellation

c
k

(r) k (r) kc

cancellation theorem

The pseudopotential for a problem

is neither unique nor exact.
On Empty Core Model (ECM)
Unscreened potential :
V(r)=
Na

0,

for r<Re

-e2/r, for r>Re

Later, screening effect :

Thomas-Fermi dielectric function

Empirical Pseudopotential Method (EPM)

band structure

Coefficients V(G) are deduced from theoretical fits to

measurements of the optical reflectance and absorption of crystal.
Chapter 15.
Charge density map can be plotted from the wavefunctions
generated by the EPM, in excellent agreement with X-ray
diffraction determination,
giving an understanding of the bonding and have great
predictive value for proposed new structures and compounds.

Numerical calculation of band structures using the first-principal

non-local pseudo-potential.
A. Zunger and M.L. Cohen, Phys. Rev. B20, 4082 (1979).

Si

Numerical calculation of Density of states

Some physical quantities obtained from

numerical calculation and experimental
measurement , respectively.
Lattice Cohesive
Bulk
constant energy modulus
()
(eV)
(Mbar)

Si

5.45
5.43

4.84
4.63

0.98
0.99

Ge

5.66
5.65

4.26
3.85

0.73
0.77

Diamond

3.60
3.57

8.10
7.35

4.33
4.43

Free atoms

APPROACHES

Atomic s

Metallic crystals
Free electrons

Tight-Binding model

Nearly Free electron model

Overlapping s

Free electrons + periodic potential

Essence of
crystal binding

Many body Treatments

Pseudopotentials
Band structures
Full treatment

Essence of
energy gaps /
transport

Experimental methods in Fermi Surface studies

H is along [111] direction noble metal
de Haas-van effect :
Oscillation of the magnetic moment
of a metal as a function of
magnetic field (H).
(1930)

A111(belly)/A111(neck)=51 Ag

Quantization of orbits in a magnetic field

Review:
a charge q of mass m in a magnetic field B
2
r
r
1 r q r
Hamiltonian
H=
p A , where B = A
2m
c
Total momentum

r qr
r
r
v
p total = p kinetic + p field = hk + A
c

Bohr-Sommerfeld relation orbits are quantized in a magnetic field

r
r
1

=
+
)2h
p
d
r
(
n
total
2
r r q r r
= hk d r + A d r
c
the first term

r r q r r
r
qr r r
hk d r = c r B d r = c B r d r
qr
q

= B 2(area)n = 2 B
c
c

r
r
dk q r r q d r r
h
= v B =
B
dt c
c dt
r qr r
hk = r B
c
the second term

r v
r
r
A

d
r
=

d
a

r r
= B da = B

r
r
q
1

=
+
p
d
r
n

2h
B
total
c
2

1 2hc
1

B = n +
= n + 4.14 10 7 Tesla m 2
2 q
2

magnetic flux through the orbit in real space

How about in k- space ?
2

hc
Sn
A n =
qB

hc
r =
k
qB
2

hc
1 2hc

Sn = n +
B = BA n = B
2 q

qB

Therefore,
the area of an orbit in k space Sn is quantized in magnetic field B
2

1 2hc qB 1
1 2q

=
+
Sn = n +
n
B

2 q hc B
2 hc

Different orbits can have the same area by changing magnetic field B.
For instance,
S
1 2q
The nth orbit in magnetic field Bn
= n +
Bn
2 hc
The (n+1)th orbit in magnetic field Bn+1
1
1 2q

=
S

B n +1 Bn hc

Equal increments of 1/B reproduce orbits w/ the same area.

Bi, T=1.6K
Steele and
Babiskin,
PR98, (1955)

-M/H (106)

Bi (1930)&(1932)

oscillation of the magnetic moment of a metal as a function of magnetic field

Low temperature and high magnetic field

De Haas-van Effect
Onsager: the change in 1/B through a single period (1/B)
(1952) was determined by 1 = 2q 1

B hc S
where S is any extremal cross-sectional area of the Fermi
surface in a plane normal to magnetic field.

1 2q

B The normal line of the orbital area is along the

Sn = n +
direction of magnetic field B.
2 hc

Quantized of the closed orbits in a magnetic field B. (L.D. Landau)

Free electrons model
an electron in a cubical box of side L in magnetic field B z
h2 2
1
n (k z ) =
k z + n hc
2
2m

where c =

eB
mc

cyclotron frequency
n=1,2,positive integer

only need to consider kx and ky

The number of levels with energy for a given n and kz
2 m
2 m
2 eB the area between
(k 2 ) = 2kk = 2 =
c =
successive orbits
h
h
hc
L
D(k) =

the number of level per unit area

in k-space
2

eL2
L 2 eB
B
=

2 hc 2 hc

depending on B

in the absence of B

in the presence of B

The number of orbital levels on a circle is a constant,

independent of n.

Does Fermi level change with magnetic field B?

States w/. k kF are occupied at T=0, N is conserved.
Fermi sphere volume in k-space occupied
by electrons in the ground states

splitting into many Landau levels

F
F

partly s+1
empty s
filled
s
F
s-1

hc
0.5hc

g()
B=0

B1

increasing
B2
slightly

At the critical fields Bs,

no partly filled level
2
eL
and N = s
Bc
2 hc

How to put electrons in the energy levels?

The number of levels with energy for a given n
= 0.5B (assumption)

In a 2D system with N=50

eL2
L 2 eB
B
=

2 hc 2 hc

w/o. consideration of spin

When all levels are fully occupied from n=1 to s, total energies of e-

eL2
1
s 2 eL2
B(n )hc =
Bhc

2
2 2hc
n =1 2hc
s

When s+1 level is partly occupied by decreasing B slightly,

Energy for

Energy for

e- in

e- in

s+1 level

1
eL2
N s
B hc (s + )
2
2hc

2
2
s
eL
the lower levels
Bhc
2 2hc

Oscillation of total
electronic energy

U(B)

Magnetic moment at T=0K

U
oscillates.
B
1 2e

w/. period =
B hcS
where S is the extremal
area of the Fermi surface
normal to the direction of B

Information of the Fermi surface

Oscillation of the magnetic moment
De Haas van Alphen effect

shape and size

What are the extremal areas ?

When magnetic field is along z-axis,
the area of a Fermi surface cross section at height kz is S(kz), and
the extremal area Se are the values of S(kz) at the kz where
dS/dkz=0, stationary wrt. small changes in kz.
Along k1-axis,
three extremal orbits :
(1),(2) area peaks and (3) area dip
Along k2-axis,
only one extremal orbit : (4) area peak

Fermi surface
FCC lattice BCC reciprocal lattice
Cu, Ag, Au, monovalent metal w/. FCC structure

k F = 3 n
2

1/ 3

1/3

2 4
= 3 3
a

4.90
=
a

2
The distance between hexagonal faces is a

3=

10.88
a

The Fermi surface neck out to meet these faces.

2
12.57
2=
a
a

The distance between square faces is

The Fermi surface does not neck out to meet these faces.

Experimental data on Au by Shoenberg

Period of 1/B for the magnetic moment
1/B111=2.0510-9 gauss-1

S=4.66 1016 cm-2 (belly)

1/B111=6.10-8 gauss-1

S=1.6 1015 cm-2 (neck)

1/B100=1.9510-9 gauss-1

S=4.90 1016 cm-2

Sbelly/Sneck=29

a: a closed particle orbit

b: a closed hole orbit
c: an open orbit

2DEG
10 nm, GaAs Cap
8 nm, spacer AlGaAs
15 nm, - doping layer, Si

60 nm, spacer AlGaAs

0.2eV
0.3

E1

1500 nm, GaAs

2DEG
Ec

buffer layer

Rxx ()

EF
energy
2.0

0.14

0.1

T=0.4K

0.12

1.5

0.10

0.0
1

Rxy (h/e )

0.08

1.0

0.06

Rxx (k)

0.2

0.4

0.6

0.8

0.0

Rxx=45/ (LT), 5k/ (RT)

-0.5

11

-2

n=1.4x10 (cm )
6

=0.99x10 ( cm /Vsec)

-1.0
-6

3
-1

0.02
0.00
0.0

2
H

0.04

0.5

T=0.4K

0.2

-4

-2

H (Tesla)

4
-1

(Tesla )

0.45m

Source

Drain

0.3m 0.6m

Quantized Conductance Transport

GaAs/AlGaAs heterostructures (Dr.Umansky provided)
n=1.41011/cm2
Mean free path l=13.6 m
=2.2106cm2/Vs (0.3K)
Split gates confined QPC

2e 2
12900
; N =
G = N
R
h

dgap=0.3m and lchannel=0.5 m

12

20.0k

11

18.0k

T=0.3K

16.0k

10
9
8

12.0k

2
G (2e /h)

14.0k

R ( )

10.0k
8.0k

1D

6.0k

2D

4.0k
2.0k
0.0
-1.1

6
5
4
3
2
1

-1.0

-0.9

-0.8

-0.7

-0.6

-0.5

VSG (V)

-0.4

-0.3

-0.2

-0.1

0.0

0
-1.1

-1.0

-0.9

-0.8

-0.7

VSG (V)

-0.6

-0.5

-0.4