Вы находитесь на странице: 1из 10

4.

6 Discussion and Problems 21

4.6
Discussion and Problems

Discussion

1. What does the phonon dispersion, that is, the vibrational frequency as a
function of wave number ω (k), look like for an infinite chain of atoms
with one atom per unit cell?

2. What does it look like for two atoms per unit cell? Why does one speak
of optical and acoustic branches?

3. The phonon dispersion for a one-dimensional chain with two atoms per
unit cell is given by (4.10) and (4.16), respectively. What about the am-
plitude of the vibrations?

4. Why is the movement of the atoms in the chain the same when multiples
of 2π/a are added to, or subtracted from, the wave number?

5. Explain the meaning and use of periodic boundary conditions to de-


scribe the properties of finite solids.

6. What predictions does the so-called Dulong-Petit law make about the
heat capacity of a solid and its temperature dependence (and why)?
How does it compare to the experiment?

7. At room temperature, do metals have a noticeably higher heat capacity


than insulators because of the mobile electrons?

8. Explain the Einstein model for the heat capacity of a lattice. How do its
predictions compare to the experiment?

9. Explain the Debye model for the heat capacity of a lattice. In which
respect does it work better than the Einstein model and why?

10. What is the definition of the Debye temperature (or frequency) and what
are the typical values?

11. Which one has a higher Debye temperature, lead or diamond, and why?

12. The thermal conductivity of an insulator has a maximum at about 10%


of the Debye temperature. Why does it decrease for lower temperatures?
Why does it decrease for higher temperatures?

13. Explain in a microscopic model why a solid undergoes thermal expan-


sion.
22 4 Thermal Properties of the Lattice

14. Thermal expansion is a so-called anharmonic process. Why is it called


like this?

Problems

1. One-dimensional chain with one atom per unit cell: We have determined
the phonon dispersion relation for an infinite chain of atoms with lattice
spacing a and one atom per unit cell (mass M). The result is (4.10). For
longitudinal vibrations, sketch how the atoms would move for k close to
0 (k  π/a) and for k at the Brillouin zone boundary (k = π/a).

The solution is given in Fig. 4.15. There are different ways of arriving at
this. Formally, one can consider (4.8): u is the amplitude of the vibration
of the atoms. It is a complex number and can basically have any desired
value. The phase difference between the atom in unit cell n and the unit
cell n + 1 is ka. If k is very small, then this phase difference is also very
small and the atoms move all in phase. If k = π/a the phase difference
is π and the atoms move exactly out of phase.
The other way of getting this result is by noting that the wave length λ
is given by 2π/k. If k is very small, this wavelength is very large and
the atoms in a small region of space will move in phase. For k = π/a,
λ = 2a and the vibration of every other atom must be in phase.

a
k << π/a
(a) λ >> a

(b) k = π/a
λ = 2a

Fig. 4.15 Movement of the atoms in a one-dimensional chain with one


atom per unit cell. (a) for k = 0 and (b) for k = π/a.

2. One-dimensional chain with two atoms per unit cell: For two atoms per unit
cell of length b, we get two branches in the dispersion, the acoustic and
the optical branch. The solutions are given by (4.16). (a) Plot these so-
lutions inside the first Brillouin zone for M2 = 0.2M1 , M2 = 0.9M1 and
M2 = M1 . (b) Consider the case where M1 6= M2 . For longitudinal
vibrations, sketch how the atoms would move for k close to 0 and for k
at the Brillouin zone boundary (k = π/b). (c) Which movement corre-
4.6 Discussion and Problems 23

sponds to which solution of (4.16)? (d) Explain what happens in the case
of M2 = M1 discussed in (a).

(a) The dispersion for different mass ratios is plotted in Fig. 4.16. What is
remarkable about the dispersion, in contrast to a one-dimensional chain
with only one atom per unit cell, is the fact that there is a forbidden
energy interval, that is, a gap in the permitted energy values. This gap
is only missing for M2 = M1 because this effectively is the case of only
one atom per unit cell (see (d)).

M 2 /M 1 =0.2 M 2 /M 1 =0.9 M 2 /M 1 =1
2.5
frequency ω (arb. units)

2.5
4

2.0 2.0

3
1.5 1.5

2
1.0 1.0

1 0.5
0.5

0 0.0 0.0
0 π/b 2π/b 3π/b 4π/b 0 π/b 2π/b 3π/b 4π/b 0 π/b 2π/b 3π/b 4π/b

Fig. 4.16 Phonon dispersion for a one-dimensional chain with two


atoms per unit cell and different mass ratios. The thick curve for the
case M2 = M1 shows the dispersion for a chain with only one atom
per unit cell and lattice constant b/2.

(b) The equations of motion for the two atoms are given by (4.13). If we
consider only one kind of atom, that is, we pick either the un (t) or vn (t),
we get exactly the same result as for the one atomic chain: for k ≈ 0,
un (t) and un+1 (t) vibrate almost in phase and for k = π/b they vibrate
exactly out of phase (see Fig. 4.17). For the second atom in the chain
there are in principle two possibilities: either they vibrate in phase or
out of phase with the first one. This needs more detailed analysis, done
in the following.
(c) For k = 0 the assignment is simple: The vibration where all the atoms
move in phase must be the acoustic mode and the vibrations where the
two atoms in the unit cell move out of phase must be the optical mode.
These vibrations are shown in Fig. 4.17a and (b).
24 4 Thermal Properties of the Lattice

b
k << π/b
(a) λ >> b

k << π/b
(b) λ >> b

(c) k = π/b
λ = 2b

k = π/b
(d)
λ = 2b

Fig. 4.17 Movement of the atoms in a one-dimensional chain with two


atom per unit cell. (a) and (b) for k = 0, showing the acoustic and
optical vibration, respectively. (c) and (d) show the vibrations at the
Brillouin zone boundary k = π/b.

It is useful to do this a little more formally. By insertion of k into (4.16)


we find that for k = 0 we get the two solutions

ω2 = 0
γ
ω2 = 2 , (4.54)
MR
where MR is the reduced mass of the two atoms
M1 M2
MR = . (4.55)
M1 + M2
This means that the vibration of the two atoms is effectively reduced to
the vibration of one atom with the reduced mass and a spring constant
which is a factor of two stronger.
If we are interested in the phase and amplitude difference of the two
atoms in the unit cell, we can use (4.14). The amplitudes of the individ-
ual atom’s motions are u and v, respectively. These are complex num-
bers. We are free to choose one of them in amplitude and phase but the
other one is completely determined. Using (4.14) we get

v 2γ − ω 2 M1 γ(1 + eikb )
= −
= (4.56)
u γ (1 + e ikb ) 2γ − ω 2 M2
4.6 Discussion and Problems 25

For k = 0 we have the two cases (4.54). Inserting k and ω into (4.56)
gives 1 and − M1 /M2 for the acoustic and optical mode, respectively.
For the acoustic mode this means that the two atoms in the unit cell
move in phase and with the same amplitude. For the optical mode they
move exactly out of phase (because of the minus sign) and the relative
amplitude is given by the mass ratio.
For the Brillouin zone boundary we have k = π/b (or k = −π/b). In
this case the two possible vibrational frequencies are


ω2 =
M1

ω2 = . (4.57)
M2

At first glance, this appears surprising: Each of the vibrational frequen-


cies depends only on the mass of one atom, not on the mass of the other.
To see why this is so, we again calculate the relative phase phase dif-
ference of vibrations between two unit cells and within the unit cell.
The atoms in unit cell n and n + 1 must vibrate exactly out of phase
for k = π/b (see problem 4.1). The relative amplitude in the unit cell can
be calculated again using (4.56) which gives a ratio if 0 and ∞ for the first
and second solution in (4.57), respectively. This means that for the solu-

tion with ω 2 = M , the M2 atoms do not vibrate at all, which explains
1
why the actual mass M2 is irrelevant for the vibrational frequency. For
the second solution of (4.57), the opposite is true. The atoms M2 vibrate
while the atoms M1 stand still. These vibrations are shown in Fig. 4.17c
and (d).
(d) When M2 = M1 , we are not really dealing with a one-dimensional
chain with two atoms per unit cell (with length b) anymore but rather
with a chain of only one atom per unit cell (with length b/2). Consistent
with this, the forbidden energy region, which are a characteristic feature
of two different atoms per unit cell, vanishes, as seen in Fig. 4.16. The
two solutions at π/b are now degenerate and, indeed, an inspection of
Fig. 4.17 shows that the atomic motions of these solutions are exactly
equal when M2 = M1 .
Since the "real" lattice constant is now b/2 and not b, the reciprocal lat-
tice constant is not 2π/b but 4π/b. The expected vibration for a chain
with one atom per unit cell and a lattice constant b/2 is shown as a thick
line in Fig. 4.16. But what about the additional thin branch? This is just
an artifact of choosing too big a unit cell. It does not add any physically
new vibrations. Take for example the "optical" mode at k = 0. Compar-
ing Fig. 4.17b and Fig. 4.17b shows that this "optical" mode at k = 0 is
26 4 Thermal Properties of the Lattice

exactly the same as the mode at the zone boundary for a chain with only
one atom per unit cell. The only reason it is there is that we have arti-
ficially introduced a too large unit cell. Compare this result to problem
2.6d where we have unsuccessfully tried to use an artificially increased
unit cell for observing x-ray diffraction with a longer wavelength.

3. Optical phonons: Optical phonons are called “optical” because they can in
some cases couple to electromagnetic radiation. This can, for example,
happen in a one-dimensional chain with two ions of opposite charge
per unit cell. A vibration of these two ions “against” each other would
correspond to a changing dipole and could be excited by the changing
electric field of an electromagnetic wave. Using energy and momentum
conservation, show that such an excitation can only work for the optical
phonon branch.

The momentum for photons is

h hν ω
h̄k = = = . (4.58)
λ c c
Hence we can write down a dispersion relation for photons which is

ω = ck (4.59)

When a phonon is excited by the photon, all the energy and k of the
photon must be transferred to the phonon, thus

ωphonon = ωphoton (4.60)

and

kphonon = kphoton . (4.61)

This has a simple graphical interpretation: Both relations can only be


fulfilled at the same time when the dispersion curves for the photons
and for the phonons cross.
The dispersion curves for photons and acoustic phonons never cross be-
cause the slope of the former is c, the speed of light, and of the latter ν,
the speed of sound in the material, which is of course much smaller.
The dispersion curves for photons and optical phonons do always cross.
The k value for the crossing can be estimated by using (4.59) for a typical
phonon vibrational frequency. Let ω ≈ 1014 Hz. Then it follows that

ω 1014 Hz
k= = = 0.33 × 106 m−1 . (4.62)
c 3 × 108 ms−1
4.6 Discussion and Problems 27

This is very small compared to the typical size of a Brillouin zone which
is π/10−9 m−1 . Note that the absorption from "optical" phonons does,
in fact, not usually lie in the visible region of the energy spectrum but in
the infra red. Strong absorption at the frequency of the optical phonons
is observed for many ionic crystals. We come back to this in chapter 9.

4. Periodic boundary conditions: Periodic boundary conditions lead to a re-


striction of the possible k values. Consider a linear chain of copper
atoms. The length of the chain should be 1 cm, the lattice spacing should
be 0.36 nm and the force constant 50 Nm−1 . Calculate the smallest pos-
sible finite wave number k. What would be the vibrational angular fre-
quency for this wave number? What would be the corresponding energy
in electron volts?

The smallest possible wave number is

2π 2π
k= = −2 = 628 m−1 . (4.63)
L 10 m
This is very small compared to the typical size of a Brillouin zone which
is π/10−9 m−1 . For such small values of k we can surely assume the
dispersion to be acoustic and calculate the vibrational frequency using
(4.11)
s
50 Nm−1
r
γ
ω= ak = 3.6 × 10−10 m × 628 m−1
M 63 × 1.66 × 10−27 kg
= 4.94 × 106 Hz = 3.25 × 10−9 eV. (4.64)

Consistent with the very small k value, this is a very small energy.

5. The Debye model: The phonon dispersion for a one-dimensional chain


of atoms is given by (4.10) and shown in Figure 4.1. What would the
dispersion look like in the Debye model?

In the Debye model it is assumed that the linear (acoustic) dispersion


relation is valid for all frequencies, up to the highest possible frequency
which is given by the size of the Brillouin zone. Don’t be mislead by
the fact that (4.43) contains the number of atoms N and the volume V of
the solid. This is the same thing because the first Brillouin zone contains
exactly N k-points. The resulting dispersion is shown in Fig. 4.18.

6. Atomic force constants and Debye temperature: (a) Estimate the value of
the force constant γ and the angular frequency ω for the vibrations of
the atoms in copper. Use that Young’s modulus Y = 130 GPa and that
the cubic lattice constant is 0.36 nm. (b) Estimate the angular frequency
28 4 Thermal Properties of the Lattice

frequency ω (arb. units)

0
-2π 0 2π
a a
first Brilloin zone

k = 2π/λ

Fig. 4.18 Thick line: phonon dispersion for a one-dimensional chain of


atoms in the Debye model. Thin line: actual dispersion.

for the vibrations that correspond to the Debye temperature of 343 K


and compare to the result in (a). (c) Having obtained γ, estimate the
amplitude of the vibrations as a fraction of the lattice spacing at room
temperature.

(a) Using (4.27) we get that

γ = Ya = 130 × 109 Pa 3.6 × 10−10 m = 46.8 Nm


r s
γ 46.8 Nm
ω= = = 2.11 × 1013 Hz. (4.65)
M 63.5 × 1.66 × 10−27 kg

(b) The conversion is

k B ΘD 1.38 × 10−23 JK−1 300 K


ωD = = = 4.49 × 1013 Hz, (4.66)
h̄ 1.05 × 10−34 Js
which is quite similar to the result in (a).
(c) The amplitude of the vibrations is taken from (4.5)
s s
2k B T 2 × 1.38 × 10−23 JK−1 300 K
xmax = = = 1.33 × 10−11 m,
γ 46.8 Nm
(4.67)

which is a few percent of a lattice spacing.

7. (*) Heat capacity of layered solids: Imagine a layered solid with a strong
intralayer interaction and a very weak interlayer interaction, such as
4.6 Discussion and Problems 29

graphite. For graphite, the observed heat capacity between 15 and 50 K


is not proportional to T 3 but to T 2 . Show that you indeed get a T 2 tem-
perature dependence if you regard graphite as a purely two-dimensional
solid, neglecting any interaction between the sp2 bonded sheets.

(a) If the bonding between the layers is very weak, this is likely to lead to
a soft spring constant between them. This is indeed the case for graphite
and it is for example reflected in a strongly anisotropic value of Young’s
modulus (see Fig. 3.4). But this also means that inter-layer vibrations are
still excited at very low temperatures while the intra-layer vibrations are
already frozen out.
(b) In order to explain the heat capacity in the range between 15 K and
50 K, we have to consider a purely two-dimensional situation. The calcu-
lation proceeds in the same way as that for the three-dimensional solid
and it is therefore only briefly sketched here. The task is to calculate the
mean energy (4.37). The main difference to the three-dimensional case
lies in the density of states. The total number of states for a circle with
radius n or |k| is given by
2
L|k|

A
N = πn2 = π = ω2 , (4.68)
2π 4πν2
where A is the area of the layer. The density of states is then
dN ωA
g(ω )dω = dω = dω. (4.69)
dω 2πν2
The Debye frequency can be obtained from the condition (4.42) to be

2 N 2
ωD = 12π ν . (4.70)
A
With (4.69) and (4.70), (4.37) becomes
3Ah̄ ω2
Z ωD
h Ei = dω, (4.71)
2πν2 0 eh̄ω/k B T − 1
or, using the same substitution as in the three-dimensional case
2 Z xD x2

T
h Ei = 18Nk B T dx. (4.72)
ΘD 0 ex −1
For low temperatures, the integration is again carried out to infinity and
the heat capacity becomes proportional to T 2 .
(c) At lower temperatures (below 5 K) the intra-layer vibrations are
frozen out, too, and one recovers the T 3 behavior of a three-dimensional
solid.
30 4 Thermal Properties of the Lattice

8. Thermal expansion: Explain why the coefficient of thermal expansion for


a solid vanishes at T = 0.

The physical reason for the vanishing coefficient of thermal expansion is


the same as for the vanishing heat capacity at 0 K: At zero temperature
no phonons are excited, that is, all the quantum mechanical oscillators
are in their ground state. At a slightly higher temperature, the thermal
energy will still be insufficient to excite the oscillators to a higher state
and they remain in their ground state. Therefore there is no thermal
expansion, the oscillators are not even excited in the harmonic part of
the potential.
Alternatively, one can argue more formally using one of the Maxwell
relations from thermodynamics
   
∂S ∂V
=− (4.73)
∂p T ∂T p

The right side of the equation is the desired thermal expansion at con-
stant pressure. According to the third law of thermodynamics, the en-
tropy at T = 0 is constant (or zero). It does therefore not depend on the
pressure and hence the left side of the expression vanishes.

9. Thermal expansion: In a Bragg reflection experiment using copper, a sharp


peak is observed at an angle of 25.23◦ at 300 K. At 500 K, the same peak
is observed at an angle of 25.14◦ . Use this information to calculate the
coefficient of linear expansion for copper.

We observe the same Bragg reflection with the same order (n) and wave-
length at both temperatures. From Bragg’s law (2.4) we get

2dˆ sin θ̂ = 2d sin θ (4.74)


d sin θ̂
= ,
dˆ sin θ
where the quantities with a hat refer to 500 K and those without a hat to
300 K. Using (4.50) we also have that
d
= 1 + α∆T (4.75)

so that
sin 25.23◦
   
sin θ
α = − 1 /∆T = − 1 /200 K (4.76)
sin θ sin 25.14◦
= 1.67 × 10−5 .

Вам также может понравиться