Chapter 5: Phonons II
Beyond the Debye Model
Chris Wiebe
Phys 4P70
The Debye Model
Last time, we looked at the
low-T and high-T limits of the
Debye Model
The Debye Model gave:
High T: C = 3R
Low T C ~ (T/D)3
The Debye Model was a much
better approximation than the
Einstein Model (one
frequency)
Another look at the Einstein Model
We can get the same result for the Einstein Model if we take our
density of states, and indicate that we can only have one frequency
We can do this by : D () = N( O)
U =
h , p
exp(h / kT ) 1
U = d ( D ( ))
p
N oscillators at 0
exp(h / kT ) 1
U = 3 d ( N ( 0 ))
exp(h / kT ) 1
h 0
U = 3N
exp(h0 / kT ) 1
This is the same result that we got before
Density of States
The whole key to calculating
the heat capacity is calculating
the density of states, D()
For the Debye Model in 3D,
this is easy:
VK 2 dK
D( ) =
2 2 d
V ( / v) 2
D( ) =
2 2
V ( / v) 2
D( ) =
2 2
d
( / v)
d
1 V 2
=
v 2 2 v 3
Cut-off
frequency
D
Density of States
In real systems, we
have to integrate over
the entire dispersion
curve to get the
density of states
If these contain terms
where the slope goes
to zero, then there is
a discontinuity in the
density of states
Remember:
2
VK dK
D( )d =
2 2 d
These are called Van
Hove singularities
(see Problem 1)
Anharmonic effects
1.
2.
3.
4.
In the theory we have been covering so far for phonons, we have
assumed that the potential energy is kx2 (so there is a Hookes
law restoring force F= -kx between atoms). Another way of saying
this is that this is harmonic.
However, real systems have anharmonic effects (the potential
may not be exactly kx2, or may contain higher order terms)
In terms of physical phenomena, what this assumes is that:
Two lattice waves cannot interact
A wave cannot change as a function of time
There is no thermal expansion of the lattice
The force constants between the atoms do not change as a
function of temperature or pressure
In reality, all of these effects are real and they can modify our
dispersion curves (and how we look at phonons)
Thermal Expansion
Anharmonic effects can be significant for physical properties
Example: thermal expansion
As you heat up a solid, the lattice expands (due to the average
displacement of the atoms increasing with higher thermal energy kT)
Lets assume the simplest anharmonic potential for the local potential
energy of each atom: U(x) = cx2-gx3
(so, we are
modifying
this picture
a little)
Assume that each atom lies in a potential that looks like kx2
but modified slightly by an anharmonic term gx3
(also assume a classical model all energies can be accessed)
Thermal Expansion
We can calculate the average displacement by using the Boltzmann
distribution function, as we did before
< x >=
dx exp(U ( x) / kT ) x
dx exp(U ( x) / kT )
Substituting for our anharmonic potential, we have:
2
3
dx
exp(
U
(
x
)
/
kT
)
x
dx
exp(
(
cx
gx
) / kT )x
= dx exp((cx 2 ) / kT ) exp( gx 3 / kT )x dx exp((cx 2 ) / kT )x[1 + gx 3 / kT ]
= dx exp((cx 2 ) / kT )[ x + gx 4 / kT ]
(for small displacements)
Thermal Expansion
So what we have is:
2
4
dx
exp(
(
cx
)
/
kT
)
[
x
+
gx
/ kT ]
= dx exp((cx 2 ) / kT )x + dx exp((cx 2 ) / kT )( gx 4 / kT )
Odd function of x
Even function of x
So, we can forget about one of these terms
2
2
4
dx
exp(
(
cx
)
/
kT
)
x
+
dx
exp(
(
cx
)
/
kT
)
(
gx
/ kT )
= 0 + (3 1/ 2 / 4)( g / c 5 / 2 )(kT ) 3 / 2
(tabulated integral)
Thermal Expansion
Now, solving for the denominator:
2
3
dx
exp(
U
(
x
)
/
kT
)
=
dx
exp(
(
cx
gx
) / kT )
= dx exp((cx 2 ) / kT ) exp( gx 3 / kT ) dx exp((cx 2 ) / kT )[1 + gx 3 / kT ]
= dx exp((cx 2 ) / kT ) + dx exp((cx 2 ) / kT )( gx 3 / kT )
Even function of x
So we have:
Odd function of x (so it is = 0)
2
1/ 2
dx
exp(
(
cx
)
/
kT
)
=
(
kT
/
c
)
Thermal Expansion
Putting it all together, we have:
< x >=
dx exp(U ( x) / kT ) x
dx exp(U ( x) / kT )
(3 1/ 2 / 4)( g / c 5 / 2 )(kT ) 3 / 2
=
(kT / c)1/ 2
3g
< x >= 2 kT
4c
Thermal expansion is related to anharmonic effects
(if g = 0, then there is no thermal expansion!)
The lattice constant increases as a function of x
Thermal Expansion
Note: this is just a general argument for why lattices expand as a
function of temperature (which is due to an anharmonic effect!)
In general, not all lattices have to exhibit this behaviour
Example: water contracts as it freezes (actually, it is less dense than
the liquid phase!)
Example: some non-cubic materials expand in some directions, contract
in others (complicated interatomic forces)
Overall though, the total volume tends to increase as the temperature
increases.
Ice is less dense
than water, so it
floats!
(negative thermal
expansion)
Nb3Sn has
a transition
to a
tetragonal phase
where the a axis
contracts,
c expands
as T increases
Thermal Conductivity
What is thermal conductivity?
What is heat?
Heat is the spontaneous flow of energy from an object at a higher
temperature to an object at a lower temperature
Lets say that you have a solid with a temperature gradient across it
dT/dx (ie. One end is at a higher temperature than the other)
The amount of energy transmitted across per unit area per unit time, jV,
is then:
dT
jV = K
dx
Flux of thermal
energy
(per unit area
per unit time)
T2
T1
Heat flow (T1 > T2)
in x-direction
Thermal Conductivity
Coefficient, K
(units: power per unit length per unit temp)
Thermal Conductivity
What conducts heat in a solid?
Heat flow is a random process in the example above, energy flows
in, and doesnt proceed directly to the other end (it diffuses randomly
throughout the solid)
We will see that for solids heat flow is largely determined by phonons
(lattice vibrations carry the energy throughout the lattice), but free
electrons can carry heat as well (this is important for metals)
From the kinetic theory of gases, we know that
Thermal conductivity
1
K = Cvl
3
Average particle velocity
Mean free path of a particle
between collisions
Heat capacity/unit volume
We are going to show that this is true for phonons as well
Thermal Conductivity
What is the mean free path?
In a gas, this is the average length that a gas particle
travels between collisions with other gas molecules
What does this mean for a solid?
This is the phonon mean free path the average
distance that a lattice vibration travels before it collides
with another phonon
If we assume that these phonons can move like particles
throughout the lattice (much like a gas), and can collide
with other phonons, then we can use the same theory as
for the thermal conductivity of gases