Chapter Four Phonons I.

Crystal vibrations
Lattice is not rigid. Atoms can move from equilibrium.
basis atom
r = ru1 ,u 2 ,u 3 + rj + u (t)
vibration
lattice point

When wave propagates in the solid,

there are one longitudinal and two transverse polarizations .

k
us-2 us-1 us us+1 us+2

us-1 us+1 us+2 us+3 us+4

k

a a
s-1 s s+1 s+2 s+3 s+4
s-3 s-2 s-1 s s+1 s+2
1D monatomic lattice

C M C M C M C M C
us-2 us-1 F us F us+1
L R

Fs = FR + FL = C(u s +1 − u s ) + C(u s −1 −u s )
d 2us
M 2 = C(u s +1 + u s −1 − 2u s )
dt

☻ Set of coupled, linear, second order differential equations.

☻ Hard to solve in general if M’s and C’s are different.
☻ Method : a trial solution (good guess)

u s (t) = A exp[i (kx s − ωt )] where xs=sa

 (
M(−ω 2 )A e iksa e − iω t = C A e ik(s +1)a + A eik(s −1)a − 2A e iksa e − iω t )
(
− Mω 2 = C e ika + e −ika − 2 )
= C 2(cos(ka) − 1)

ω2 =
2C
(1 − cos(ka ) ) = 2C  2 sin 2 ( ka ) 
M M 2 
4C ka
ω= sin Dispersion relation
M 2

1.0

π -1
sec ]
0.8

When k = ±
1/2

0.6
a
[(4C/M)

4C
ω= : maximum
0.4

M 0.2
ω

set the boundary of first BZ

0.0
-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5
-1
k [(2 π /a) m ]
For a small k (ka<<1) Long wavelength limit

4C ka  C  Ca Ca
ω≈ =  a  k = k= k
M 2  M  M/a λ
= vk continuum elastic wave limit
Dispersion : ω ≠ vk phase velocity vp
What is the wave velocity?
{ group velocity vg

ω
phase velocity v p =
k
dω the velocity of energy
group velocity v g ≡ = ∇ k ω (k)
dk propagation in the medium
 time increasing
(a) (d)
group velocity
dω
vg ≡ = ∇ k ω (k)
(b) (e) dk

(f) phase velocity

(c)
ω
vp =
k
time increasing

(d) (g)
 phase velocity group velocity
ω 4C/M sin( ka / 2 ) dω d sin( ka / 2 )
vp ≡ = vg ≡ = 4C/M
k k dk dk
a 4C sin( ka / 2 ) a 4C ka
= = cos( )
2 M ka / 2 2 M 2

sin(ka / 2)
v p = vs vg = vs cos(ka/2)
ka/2
π
When k= ± , vg=0 group velocity vanishes ! vp=(2/π)vs
a
(k lies on the boundary of first BZ)
r
Direct lattice r = na x̂
r 2π Bragg condition
Reciprocal lattice G = m x̂
r
a
1 π
BZ boundary G = ± x̂
2 a
 How about the motion of atoms?
u s (t) = A exp[i (ksa − ωt )] a plane wave

The relative displacement between two adjacent atoms

u s +1
For a fixed k, = e ika same
us
2π
For the other fixed k' = k + n = k+G
u s +1 a ika in2π
i(k + n2π /a)a
=e = e e = eika
us
The displacement can always be described by a wavevector
within the first BZ.
2π
At k=q, u s (t) = A exp[i (qsa − ω t )] same as k= +q
a
2π   2π 
At k= − q , u s (t) = A exp i ( − q)sa − ω t  same as k= − q
a  a 
= A exp[i (2π s − qsa − ω t )] = A exp[i (− qsa − ω t )]
 λ=4a/3 λ=4a/7

λ=4a
2π π
λ1 = 4a, k1 = =
4a 2a
4a 3π 2π π
λ 2 = , k 2 = = −
3 2a a 2a
4a 5π 2π π
λ3 = , k 3 = = +
5 2a a 2a
4a 7π  2π  π
λ 4 = , k 4 = = 2  −
7 2a  a  2a
 The displacement can always be described by a wavevector
within the first BZ.
Reduced zone scheme
k1 k2 k3 k4

1.0

0.8
sec )]
-1

0.6
1/2
ω [(4C/M

0.4

0.2

0.0
-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0
-1
k [(2π /a) m ]
Consider that k’s range over all reciprocal space,
the expression for ω(k) is periodic.

All the information is in the first Brillouin zone : the rest is repeated
with periodicity 2π/a – that is, the frequencies are the same for ω(k)
and ω(k+G) where G is any reciprocal lattice vector 2π
= n×
a

Periodicity in reciprocal space

r
r wavevector k is identical to the
r with
In fact, the motions of atoms
motion with wavevector k + G.
r
All independent vibrations are described by k' s inside BZ.
Group velocity of vibration wave

vg = vs cos(ka/2)
π
At the BZ boundary, k=± and λ = 2a
a
vg = 0 zero group velocity
z fundamentally different from elastic wave in a continuum
z any wave (vibration or others) is diffracted if k is on the zone boundary
z leads to standing wave with zero group velocity

1 .0
sec ]

0 .8
-1
1/2

0 .6
ω [(4C/M)

0 .4

0 .2

0 .0
-0 .5 -0 .4 -0 .3 -0 .2 -0 .1 0 .0 0 .1 0 .2 0 .3 0 .4 0 .5
-1
k [(2 π /a ) m ]
 vg=0 No propagation (standing wave)

Transverse (shear)
2a

2λ
Alternate atoms oscillate in opposite phases.
Longitudinal (sound)
Contribution of other nearest planes to ω
In metals the effective forces may be quite long range,
carried from ion to ion through conduction electron sea.

The generalization of the dispersion relation to p nearest planes,

2
ω = ∑ C p (1 − cos(pka))
2

M p >0
Solving for the interplanar force constant Cp

dkω cos(rka) = 2∑ Cp ∫ dk(1 − cos(pka))cos(rka)

π/a π/a
M∫ 2
− π/a − π/a
p >0

Cr
= − 2π
a
Ma π/a
Cp = − ∫
2
dkω cos(pka)
2π - π/a
More complicated lattices

1st and 2nd nearest neighboring couplings in 1D monatomic chain

Two dimensional monatomic rectangular lattice

How about two atoms per primitive basis in one dimension?

1D diatomic lattice
M2 M1 C M2 C M1 C M2 C M1 C M2 M1

us-1 vs-1 us vs us+1 vs+1

a
Fs = FR + FL
d 2us
M1 2 = C( v s − u s − u s + v s −1 ) = C( v s + v s −1 − 2u s )
dt
d 2 vs
M 2 2 = C(u s +1 − v s − v s + u s ) = C(u s +1 + u s − 2v s )
dt

A trial solution set

u s (t) = u exp[i (kx s − ωt )]

v s (t) = v exp[i (kx s − ωt )]
where xs=sa
 (
M1 (−ω 2 )u e iksa e − iω t = C(v e iksa + v e ik(s −1)a ) − 2Cue − iksa e − iω t )
− M ω u = Cv(1 + e ) − 2Cu
1
2 −ika

(2C − M ω ) u − C(1 + e ) v = 0
1
2 −ika

M (−ω )v e e
2
2
= (C(u e iksa −iω t
+ue ik(s +1)a iksa
)
) − 2Cve −iksa e −iω t
− M ω v = Cu (e + 1) − 2Cv
2
2 ika

− C(1 + e ) u + (2C − M ω ) v = 0
ika
2
2

(2C − M ω )(2C − M 1
2
2 ω 2
−) (
C 1 + e − ika
C )(
1 + e ika
=0 )
M 1M 2 ω 4 − 2C (M 1 + M 2 )ω 2 + C 2 (4 − (2 + 2 cos( ka ) )) = 0

ω −
4 2 C (M 1 + M 2 ) ω + 2C
2 2 2 sin 2
( ka/2 )
=0
M 1M 2 M 1M 2
 1 1  2 4C 2
ω − 2C 
4
+  ω + sin 2 ( ka/2 ) = 0
 M1 M 2  M 1M 2
Dispersion relation
2
 1 1   1 1  4 ka
ω 2 = C +  ± C  +  − sin 2 ( )
 M1 M 2   M 1 M 2  M 1M 2 2

 1 1   4 M 1M 2 2 ka 

ω = C
2
+  1 ± 1 − sin ( )

 M1 M 2   (M1 + M 2 )2
2 

 1 1 
2 C  +  + optical branch
M
 1 M 2 
2C
M2 Two branches
M1 > M2 correspond to
2C
± signs of
M1
dispersion relation

- acoustic branch

- π/a π/a
k
For a small k (ka<<1) long wavelength limit (λ>>a)
sin (ka/2) ~ ka/2
1/2 1/ 2
  1 1    4M1M 2  ka  
2

ω =  C +  1 ± 1 − 2   

  M 1 M 2    (M1 + M 2 )  2  
1/ 2

 M1 + M 2   
1/ 2

M1M 2 2
= C  1 ± 1 − (ka )  
 M1M 2    (M1 + M 2 )
2
 
1/ 2
 M1 + M 2    M1M 2 
≅ C  1 ± 1 − (ka ) + ... 
2

 M1M 2    2(M1 + M 2 ) 
2


C
acoustic branch ω− ≈ (ka ) ω ∝ k
2(M1 + M 2 )
as k 0
 1 1 
optical branch ω+ ≈ 2C +  ω is finite
 M1 M 2 
 At BZ boundary k=±π/a : sin (ka/2) ~ ±1

 1 1   4 M1M 2 

ω = C
2
+  1 ± 1 −
M
 1 M 2 
 (M 1 + M 2 )2 

 M1 + M 2   M − M2 
ω = C  1 ± 1 
 M1 + M 2 
 M 1M 2   
 1 1  1 1

= C  + 
 ± −
 M1 M 2  M1 M 2
M1>M2

acoustic  1 1   1 1  2C
ω − = C  +  −  −  =
branch  M1 M 2   M 2 M1  M1

optical  1 1   1 1  2C gap
ω + = C  +  +  −  =
branch  M1 M 2   M 2 M1  M2
How about M1=M2 ; back to monatomic chain

2C 2 ka 2C ka
ω= 1 ± 1 − sin ( ) = 1 ± cos( )
M 2 M 2

1.4 optical branch

1/2

1.2
G
(2C/M)

1.0
optical branch is
higher k values
0.8 folded back to
diatomic BZ
0.6
ω

0.4
acoustic branch
0.2

0.0
-1.00 -0.75 -0.50 -0.25 0.00 0.25 0.50 0.75 1.00
ka/2 (2π/a)
 2C
M2
M1 > M2 2C
M1

- π/a π/a k

-acoustic branch M2 M1 M2 M1 M2 M1 M2
k=0, ω=0
us-1 us-1 us vs us+1 vs+1
-optical branch
2C
k=0, ω= M1 + M 2
, maximum
M2 M2 M2 M2

M1 M1 M1
 M1
-acoustic branch M2
π 2C
k= , ω= M1
a

-optical branch
π 2C
k= , ω= M2
a

-acoustic branch (k, ω) -optical branch (k, ω)

 In three dimensions

p atoms per primitive cell 3p vibration branches

Acoustic (3) : LA (longitudinal)
TA1 (transverse)
TA2 (transverse)
Optic (3p-3) : LO (longitudinal)
TO (transverse)
Si-diamond structure
Points :data
Lines: model calculation

G. Dolling, in “Inelastic scattering of Neutrons in Solids and Liquids”, P.37, 1963.

So far, ω and k are continuous except that,
z k is restricted in the first BZ (discrete of lattice spacing)
z ω is forbidden in some gaps (splitting of acoustic and optical modes)
Additional quantization effects:
z k is quantized by finite crystal size.
z Energy of vibration mode is quantized by quantum effect.

discrete k values Finite crystal size

Eg. 1D monatomic chain w/. length l N+1 atoms, N bounds

a=l/N spacing

u s (k, t) = A exp[i (kx s − ωt )] where xs=sa

u1(k,t)= uN(k,t) exp[ik(Na)]=1

 2π   2π 
∴kNa=n2π n 
k= Na  = n   discrete !
   l 
 N=20, k=n(2π/20a)=(n/10)(π/a)
only 20 modes are allowed.

quantized energy of vibration mode

Each k has a corresponding ω,
what is the energy associated w/. this mode?

Some quantum systems:

photons: k=2π/λ photon energy E=hω

ω=ck mode energy Ek,ω= nk (hω)
Number of photon at (k,ω)
A particle in a box: λ = 2l
n
n
2π π
kn = = n 
λn l

En =
(hk n )
=n
2
h2π2
2

2m 2ml 2
Phonon: particle-like properties
number ni
Energy h ω i
r
wavevector k i
r r
crsytal momentum Pi = h k i Not a real momentum !

What is the real momentum? Physical momentum

eg. k=0, corresponding to translation of the whole crystal by u(t)
k=0
u(t) r
r d u (t) r r
Preal = Nm but Pcrystal = h k = 0
dt
eg. k 0, corresponding to relative motions of atoms around the equilibrium

k 0
r r r
Preal = 0 but Pcrystal = h k ≠ 0
r r
du (t)
Preal = M ∑ i
i dt
r du(t)
u s (t) = u(t)exp[iksa] Preal = M ∑ exp[iksa ]
dt s
du(t)  1 − exp[ikNa] 
=M  
dt  1 − exp[ika] 
2π n
Q discrete k = ± ∴ exp[ikNa] = exp[±i2π n] = 1
Na
r
Preal = 0 Only k=0 carries the real momentum.

Lattice vibrations
Specifying the vibrational states of the crystal by specifying
number of phonon in each state ki
(ki, ω) is determined by structure and binding
ni is determined by excitation (thermal, acoustic, etc..)
Simple harmonic oscillator:

1 2
U(x) = cx
2
 1 c
E n =  n + hω where ω =
 2 m
Zero point energy

PHONONS: quantized lattice vibrations

Lattice vibrations: r
wavevector k
()
r
frequency ω = f k Dispersion relation
 1
energy E n =  n +  hω SHO
 2
Phonons: can be thought as particles
interact with other electrons, phonons, etc…
Determination of phonon structure :
Neutron scattering – neutrons only scatter off atoms, not electrons
Primary tool for obtaining phonon dispersion relation
Elastic scattering
r r r' r r
r ∆k = G Bragg condition k = k + G
r'
k k r' r 2
h k '2
h 2k 2
k =k =
2M n 2M n

Inelastic scattering
phonon absorption
phonon emission r'
r r' r k
k k k
r r
q q

r' r r r 2 '2 r' r r r 2 '2

h k h 2k 2 h k h 2k 2
k + q = k + G and + hω q = k − q = k + G and − hωq =
2M n 2M n 2M n 2M n
Experimental measurements of dispersion curves

Dispersion curves ω as a function are measured by inelastic diffraction

If the atoms are vibrating then diffraction can occur with energy loss or
gain by scattering particles

In principle, can use any particle – neutrons from a reactor, X-rays from
a synchrotron, He atoms which scatter from surfaces, …

¾ Neutrons are most useful for vibrations

For λ ~ atomic size, energies ~ vibration energies

¾ X-ray only recently has it been possible to have enough resolution

meV resolution with KeV X-rays
 Experimental setup
detector
Triple axis : rotation of sample

sample

Selected
energy out
Selected
energy in
Single crystal
monochromator
Single crystal
monochromator

Neutrons or X-rays with

broad range of energies
 r r
Measure k, k' , and ∆E ω(1012Hz) [100]
Na
r
q and hωq for phonons T=90K

k(2π/a)
[110]

k(2π/a)

[111]

A triple axis neutron spectrometer at Brookhaven

k(2π/a)
Review
Fundamental concepts needed to understand the vibratory motion of atoms:
Normal mode of vibration : all atoms oscillate with the same frequency.
Only atomic vibrations with certain frequencies, determined by interatomic
forces, occur in any given solid. --- periodic arrangement of atoms
Normal mode displacements for these materials have an especially simple
form and are relatively easy to discuss.

If the displacements of atoms from their equilibrium sites are small,

the forces they exert on each other are proportional to their displacements,
as if the atoms were connected by ideal spring.
Atomic motions are simple harmonic.

substitute expressions for the forces into Newton’s 2nd law

generate a set of differential equations, one for each atom
seek solutions for which all atomic displacements have the same f
 v phase → ∞ u M
ω → constant →− 2
v group → 0 v M1 At BZ boundary
Center of mass is stationary
 1 1  ω = constant
2 C  +  + optical branch
M
 1 M 2  2C a
2C v phase =
M2 π
M2
M1 > M2 v group = 0
2C
M1 u=0

- acoustic branch ω = constant

2C a
- π/a 0 π/a k v phase =
M1 π
π v group = 0
Long wavelength limit : k <<
a
u v=0
C →1
ω = vs k vs = a
2(M1 + M 2 ) v