Upload
others
View
8
Download
0
Embed Size (px)
Citation preview
1
Energy bands for electrons in crystals
2
Outline
Recall the solution for the free electron gas in a box of size
LLL
• States classified by k with E(k) = ħ2k2/2m
• Periodic boundary conditions lead to k= integer (2 / L), etc.
• Pauli Exclusion Principle, Fermi Statistics
• Electrons in crystals
• First step - NEARLY free electrons in a crystal
• Simple picture of degenerate perturbation lead to energy gaps
• Band structures in solids
3
Schrödinger Equation
• How can we solve the Schrödinger Eq.
where V(r) has the periodicity of the crystal?
2
2( ) ( ) ( )
2V r r E r
m
− + =
• We will consider simple cases as an introduction
♦ Nearly Free Electrons
4
Consider one dimensional example
• If the electrons can move freely on a line from 0 to L (with no
potential),
Of which we have seen before:
• Schrödinger Eq. in 1-d with V = 0
• Solution with f (x) = 0 at x = 0, L
0 L
2 2
2( ) ( )
2
dx E x
m dx − =
2 2
*
0
exp( ) 2( ) , , 0,1,2, 3, ...
( ) ( ) 1, ( )2
L
ik xx k m m
LL
kx x dx E k
m
= = =
= =
Boundary conditions
5
Electrons on a line
• For electrons on a line, the energy is just the kinetic energy
• Values of k fixed by the line, k = m (2/L), m = 0,1,...
2 2( ) / 2E k k m=
E
kkF0-kF
Filled
states
Empty
states
• The lowest energy state is for electrons to fill the lowest statesup to the Fermi energy and Fermi momentum kF with twoelectrons in each state.• This is a metal– the electrons can conduct electricity asdiscussed before.
6
Why some materials insulators or semiconductors?
• To answer this question we need consider the effect of
crystal on electrons.
• In a crystal, the nuclei are arranged in a periodic crystalline
array.
• This leads to the modification of the energies of the
electrons and the different behaviors in different crystals.
7
Electrons in Crystals
• Simplest extension of the Electron Gas model
Very weak potentials
with crystal periodicity
• Nearly Free electron Gas –
– Very small potential variation with the periodicity of the crystal
• We will first consider electrons in one dimension.
8
Electrons in a periodic potential
2
02
pH
m=
for free electrons
2 2
02
kE
m=
a weak periodic potential
0( )H H V r= + V( )=V( )r r R+
31' exp[ ( ') ] (r)k V k i k k r V d r
V= −
This is zero unless the difference is a reciprocal lattice vector.
9
Electrons in a periodic potential
0 0 0( ) +E k E k V k E V= = +
1st order perturbation theory:
V0 is assumed to be zero for simplicity.
only a constant energy-
shift
2nd order perturbation theory:
2
0
' 0 0
'( ) +
( ) (k')k k G
k V kE k E
E k E= +
=−
10
Electrons in a periodic potential
Degenerate perturbation theory
A divergence in perturbation theory
11
Degenerate Perturbation Theory
0( )k H k E k=
0( )k G H k G E k G+ + = +
*
Gk H k G V+ =
Gk G H k V+ =
k k G = + +
*
0
0
( )
( )
G
G
E k VE
V E k G
= +
2
0 0( ( ) )( ( ) ) 0
GE k E E k G E V− + − − =
12
Degenerate Perturbation Theory
( ) ( )E k E k G= +
22
0( ( ) ) 0
GE k E V− − =
At the zone boundary
0( ) ( )
GE k E k V
=
There is a gap opening up at
the zone boundary.
=
( )1
2k k G
= +
13
In one dimension
0
2( ) cos( )
xV x V
a
=
( )1
2k k G
= +
exp( )ikxk
L=
exp( )ikxk G
L
−+ =
( ) ( )
( ) ( )
/ 2cos /
/ 2 sin /
i x a i x a
i x a i x a
e e a x a a
e e a i x a a
−
+
−
−
= + =
= − =
14
In one dimension
15
Nearby the zone boundary
2
2 2
0
2
2 2
0
2( + ) [( ) ]
2
2( + ) [( ) ]
2
n n nE
a m a a
n n nE
a m a a
+
−
= + +
− = − +
*
0
0
G
G
E VE
V E
+
−
=
2
0 0( )( ) 0
GE E E E V
+ −− − − =
2 222 2 2 22
{ [( ) ] )} ( )2 2
G
n nE V
m a m a
+ − = +
Not on the zone boundary
2 222 2 22
[( ) ] ( )2 2
G
n nE V
m a m a
= + +
δ is small
2 2 2 2
2 2 1( ) [1 ( ) ]
2 2G
G
n nE V
m a m m a V
= +
>1 for weak potentialThe dispersion is quadratic in δ.
2 2 2
2
*( )
2 2G
nE V
m a m
= +*
22 1
1 ( )G
mm
n
m a V
=
17
Nearly free electrons on a line
Energy gap: energies
at which there are no
waves that can travel
through crystal
18
Nearly free electrons on a line
plotted in repeated zone scheme
19
Why materials insulating (or semiconducting)?
• If there are just the right number of electrons to fill the lower
band and leave the upper band empty, the Fermi level lies in
the gap.
• Electrons are not free to move in the gap!
• Only if one adds an energy as large as the gap energy can an
electron be raised to the upper band where it can move.
20
Summary: Electrons in Crystals
• Real Crystal –
Potential variation in accord
with the periodicity of the crystal
Attractive (negative) potential
around each nucleus
• Periodic potential leads to:
– Electron bands – E(k) different from free
electron bands
– Band Gaps
21
Outline
• Electrons in crystals
Electrons in a periodic potential
Bloch theorem
• Quantitative calculations for nearly free electrons
Energy bands and standing waves at the Brillouin
zone boundary
Energy gaps
• Tight Binding Model
• Metals vs. insulators – simple arguments
22
Bloch theorem (1928)
( ) (R) ( )R
T r C r =
( ) ( ) ( )R
T H r H r R r R = + +
( ) ( )V r V r R= + ( ) ( )H r H r R= +
[ , ] 0R
T H =There are simultaneous eigen-functions of H
and all translation operator.
( ) ( )R
T r r R = +
Translation operator
= ( ) ( )= ( )R
H r r R HT r +
23
( ) ( ) ( )R
T r r R r = + =
a ik=
ln (R)=a RC
( ) ( ) exp[ ] ( )R
r R T r ik R r + = =
1 2 1 2R R R RT T T
+=
1 2 1 2C(R ) (R ) (R R )C C= +
1 2 1 2ln (R )+ln (R ) ln (R R )C C C= +
lnC is linear in R.
exp[ ] ( )R
PT i R r=
Bloch theorem (1928)
24
• Bloch theorem
( ) ( ) exp( )k k
r u r ik r =
• Key points:
-- Each state is labeled by a wave vector k
-- k can be restricted to the first Brillouin zone:
where is just another
periodic function
( ) ( ) exp[ ( ) ]
'( ) exp( )
k G k G
k
r u r i k G r
u r i k r
+ + = +
=
'( ) ( ) exp( )k k G
u r u r i G r+
=
uk(r) is a periodic function
Bloch theorem (1928)
25
Bloch theorem (1928)
• Thus a wave function in a crystal can always be written as
where uk(r) is a periodic function and k is restricted to the first
Brillouin zone
• k becomes continuous in the limit of a large system.
( ) ( ) exp( )k k
r u r ik r =
26
Bloch wave packet
Real part of wave function Condensed Matter Physics by Marder (2000)
27
Potential energy as a periodic function
• We have seen in Chapter 2 that any periodic function can be
expressed as Fourier series
where G is a reciprocal lattice vector.
( ) exp( )G G
f r f iG r=
( ) exp( )G G
U r U iG r= *
G GU U
−=
G GU U
−=
• Thus
since U(r) is real,
• If the crystal is symmetric, i.e., U(-r) = U(r), then
( ) ( )f r T f r+ =
• To check: A periodic function satisfies
where 1 2 3 1 1 2 2 3 3
( , , ) : integeri
T n n n n a n a n a n= + +
28
Schrödinger equation
• In a perfect crystal
2
2[ exp( )] ( ) ( )
2 G GU iG r r r
m− + =
2, 0,1,2, ...k m m
L
= =
• Note that we do NOT assume is periodic! It is a wave!
• What is k?
Similar to the problem of electrons in a box, we assume (r) is
periodic in a large box of volume L3. This leads to
( ) ( ) exp( )k
r C k ik r = • Now expand
29
Schrödinger equation - continued
• The Schrödinger equation becomes
22
( )[ exp( ) exp[ ( ) ]
where ( /( )exp( ), 2 )
kk G
k k
GC k i k r U i k G r
k i mC k r k
+ +
= =
( ) ( ) ( ) 0k GG
C k U C k G − + − =
• To satisfy the equal sign for any given r, the coefficient in the
above equation must be equal to zero, we then obtain the central
equation (eq. 7-27):
[( ) ( ) ( )]exp[ ] 0kk G G
C k U C k G ik r − + − =
• This can be written as, by re-labeling the sums,
30
Central equation
• How do we interpret the equation?
( ) ( ) ( ) 0k GG
C k U C k G − + − =
( ) ( ) exp[ ( ) ]k G
r C k G i k G r = − −
Yet to be determined
• If UG ≠ 0 then each k is mixed with (k – G) where G is a
reciprocal lattice vector. The wave function is
2 2( ) ( ) exp( ), ( / 2 )
k kr C k ik r m k = = =
• If UG = 0 (no potential – free electrons) then each k is
independent and each wave function is
• Consider the case where the potential UG is very weak. Then we
can find a good approximate solution. This is the “nearly free
electron” approximation.
31
Solving central equation
( ) ( ) ( ) 0k GG
C k U C k G − + − =The central equation
• For k near the Brillouin zone boundary, i.e., k → /a, the wave
exp(ik·r) is mixed strongly with exp[i(k-G)·r], where G is the one
(and only one) vector that leads to
│k│~ │( k – G )│
• Let U = UG =U-G for that G. Retain only equations in the central
equation that contain C(k) and C(k-G)
32
Solving central equation
• Leads to two coupled equations:
( ) ( ) ( ) 0
( ) ( ) ( ) 0
( )0
( )or
k
k G
k
k G
C k U C k G
C k G U C k
U
U
−
−
− + − =
− − + =
−=
−
The central equation ( ) ( ) ( ) 0k GG
C k U C k G − + − =
2 2 1/21 1( ) [ ( ) ]
2 4
and ( ) [( ) / ] ( )
k k G k k G
k
U
C k G U C k
− −= + − +
− = −
• Solution:
33
Example in 1D
• At the zone boundary: k = /a, k-G = -/a and
2 2( / 2 )( / )
k k Gm a
−= =
kU
= • Solution for k = /a:
Each root describes an energy band
2 2 1/2(1 / 2)( ) [ (1 / 4)( ) ]
k k G k k GU
− −= + − +
the energies are given by
( ) [( ) / ] ( ) ( )k
C k G U C k C k − = − = • Eigenvectors:
34
Example in 1D
( ) ( )exp( )
( )[exp( ) exp( )]
( )[exp( / ) exp( / )]
kx C k ik x
C k ikx ikx iGx
C k i x a i x a
=
= −
= −
• The wave function:
• The same solution as the a weak periodic potential
• Even though the potential is extremely strong, the electrons
still behave almost as if they do not feel the atoms at all.
• They still almost form plane-wave eigenstates with the
only modification being the periodic Bloch function.
35
• The energy is always (K) = (ħ2/2m)K2 for free electron bands
• We define the reduced k by
k = K – G
• Introduce the effects of potential, i.e., band gaps for periodic
potential
Nearly-free electron approximation in general
crystals
36
Brief summary
• We have solved the “central equation” in the “nearly-free
electron approximation”
• The results applied to all types of crystals but we have
assumed the potential is “weak” which is not always true
• Which conclusions will always apply in all crystals?
-- The Bloch theorem
-- Standing waves, energy bands and energy gaps at the BZ
boundary
37
Metals vs. Insulators
• How can we use the results so far to determine which crystals will
be metals? Which are insulators?
38
Metals vs. Insulators
• How can we know when the Fermi energy will be in the band? in the gap?
• We often need to sum or integrate over k to find total quantities,
for example the total number of filled states in the bands.
39
Sums and integrals over k points
• We can use the idea of periodic boundary conditions on box of
size L L L
Exactly the same as for phonons, electrons in a box,...
• Volume per k point = (2π/L)3
• Total number of k points in BZ:
• Rules:
3 3 3
k-point cell3
2
22
BZV L L
N Na a
L
= = = =
k-point cell
32
k
N N
dkL
=
→
40
Metals vs. Insulators
• Electrons obey the exclusion principle.
2 electrons per primitive cell of the crystal fill a band. Any
additional electrons must go into the next band, and so forth.
• An odd number of electrons per primitive cell always leads to a
partially filled band –a metal.
• An even number may lead to an insulator.
41
Metals vs. Insulators
• An even number of electrons per cell may lead to an insulator if
the Fermi energy is in a gap everywhere in the BZ.
42
Fermi sea in 2D square lattice
Weak periodic potential Strong periodic potential
The area of the Fermi sea remains fixed as the strength of the
periodic potential is changed.
43
Equi-energy contours in 2D square lattice
From tight-binding
model
44
Fermi surface
Weak periodic
potential
Strong periodic
potential
45
Metals vs. Insulators
No periodic potential Strong periodic potential
Metal Insulator
46
Intermediately strong periodic potential
47
Intermediately strong periodic potential
48
Metals vs. Insulators
• A band holds two electrons per cell of the crystal
• A crystal with an even number of electrons per cell may be an
insulator!
Electrons “frozen”, i.e., electrons can’t move around the
energy gap for any excitations of electrons.
• Therefore a crystal with an odd number of electrons per cell
must be a metal!
49
Metals vs. Insulators
Examples
• Na: 1 valence electron/atom => 1 valence electron/cell
• NaCl: 1 + 7 = 8 valence electrons/cell
• Xe: 8 valence electrons/cell
• Mg: 2 electrons/cell, bands overlap, not very good metal
• Si: 4 valence electrons/atom => 8 electrons/cell, pure
crystal an insulator at absolute zero.
• CoO: 7 + 8 = 15 valence electrons/cell, but an insulator…
50
Mott-Hubbard insulator
strong correlation, insulating,
atomic limitU t
non-interacting, metallic,
band-limitt U
U W (band width)
metal-to-insulator transition
, ,
j i i i
i j i
H t c c U n n
Hubbard model
51
Metals vs. Insulators
Indirect band gap direct band gap
52
Direct/Indirect transitions
53
Optical Properties of Insulators
GaAs with 1.44 eV band
gap looks black since it
absorbs all frequencies of
visible light.
CdS with 2.6 eV band gap
looks reddish since it
absorbs violet and blue
light.
54
Optical Properties of Metals
Since metals are very conductive, photons excite the electrons
which then re-emit light.
Noble metals looks shiny because there is no insulating oxidized
layer.
Silver looks brighter than copper and gold because the band
width of silver is greater than that of gold or copper.
Impurity could have a strong influence on their optical properties.
55
Summary
• “central equation”: general for ALL
• We solved the problem in the “nearly-free electron
approximation” where we assume the potential is “weak”
• Some results applied to ALL types of crystals:
-- The Bloch theorem
-- Standing waves and gaps at the BZ boundary
-- Energy bands
• We can predict that most materials must be metals, and
other materials can be insulators – simply by counting
electrons!