# 19.

and using eqn 11.28 we have
C
p
per mole =

f
2
+ 1

R, (19.23)
from which we may derive
γ =
C
p
C
V
=
(
f
2
+ 1)R
f
2
R
= 1 +
2
f
. (19.24)
We can summarize our results for the heat capacity of gases, per atom/molecule,
as follows:
Gas Modes f E γ
Monatomic translational only 3
3
2
k
B
5
3
Diatomic translational and rotational 5
5
2
k
B
7
5
Diatomic translational, rotational and vibrational 7
7
2
k
B
9
7
19.2.4 The heat capacity of a solid
In a solid, the atoms are held rigidly in the lattice and there is no pos-
sibility of translational motion. However, the atoms can vibrate about
their mean positions. Consider a cubic solid in which each atom is con-
nected by springs (chemical bonds) to six neighbours (one above, one
below, one in front, one behind, one to the right, one to the left). Since
each spring joins two atoms, then if there are N atoms in the solid, there
are 3N springs (neglecting the surface of the solid, a reasonable approx-
imation if N is large). Each spring has two quadratic modes of energy
(one kinetic, one potential) and hence a mean thermal energy equal to
2 ×
1
2
k
B
T = k
B
T. Hence the mean energy of the solid is
E = 3Nk
B
T, (19.25)
and the heat capacity is ∂E/∂T = 3Nk
B
. Because R = N
A
k
B
, the
molar heat capacity of a solid is then expected to be 3N
A
k
B
= 3R.
x
z
y
Fig. 19.6 In a cubic solid, each atom is
connected by chemical bonds, modelled
as springs, to six nearest neighbours,
two along each of the three Cartesian
axes. Each spring is shared between
two atoms.
The equipartition theorem seems to be an extremely powerful tool for
evaluating thermal energies of systems. However, it does have some
limitations, and to discover what these are, it is worth thinking about
the assumptions we have made in deriving it.
206 Equipartition of energy
• We have assumed that the parameter for which we have taken the
energy to be quadratic can take any possible value. In the deriva-
tion, the variables x
i
could be integrated continuously from −∞
to ∞. However, quantum mechanics insists that certain quanti-
ties can only take particular ‘quantized’ values. For example, the
problem of a mass on a spring is shown by quantum mechanics
to have an energy spectrum which is quantized into levels given
by (n +
1
2
)ω. When the thermal energy k
B
T is of the same or-
der, or lower than, ω, the approximation made by ignoring the
quantized nature of this energy spectrum is going to be a very bad
one. However, when k
B
T ω, the quantized nature of the en-
ergy spectrum is going to be largely irrelevant, in much the same
way that you don’t notice that the diﬀerent shades of grey in a
newspaper photograph are actually made up of lots of little dots if
you don’t look closely. Thus we come to an important conclusion:
The equipartition theorem is generally valid only at high temper-
ature, so that the thermal energy is larger than the energy gap
between quantized energy levels. Results based on the equipar-
tition theorem should emerge as the high-temperature limit of
more detailed theories.
x
V x
x
Fig. 19.7 V (x) is a function which is
more complicated than a quadratic but
which has a minimum at x = x
0
.
• Everywhere we have assumed that modes are quadratic. Is that
always valid? To give a concrete example, imagine that an atom
moves with coordinate x in a potential well given by V (x), which is
a function which might be more complicated than a quadratic (see
for example Fig. 19.7). At absolute zero, the atom ﬁnds a potential
minimum at say x
0
(so that, for the usual reasons, ∂V/∂x = 0
and ∂
2
V/∂x
2
> 0 at x = x
0
). At temperature T > 0, the atom
can explore regions away from x
0
by borrowing energy of order
k
B
T from its environment. Near x
0
, the potential V (x) can be
expanded
3
as
3
Using a Taylor expansion; see Ap-
pendix B.
V (x) = V (x
0
) +

∂V
∂x

x
0
(x −x
0
) +
1
2

2
V
∂x
2

x
0
(x −x
0
)
2
+· · · ,
(19.26)
so that using

∂V
∂x

x
0
= 0, we ﬁnd that the potential energy is
V (x) = constant +
1
2

2
V
∂x
2

x
0
(x −x
0
)
2
+· · · , (19.27)
which is a quadratic again. This demonstrates that the bottom
of almost all potential wells tends to be approximately quadratic
(this is known as the harmonic approximation).
4 4
The argument that the bottom of al-
most all potential wells tends to be
(∂
2
V/∂x
2
)
x
0
turned out to be zero.
This would happen if, for example,
V (x) = α(x −x
0
)
4
.
If the temperature gets too high, the system will be able to access
positions far away from x
0
and the approximation of ignoring the
higher–order (cubic, quartic, etc.) terms (known as the anhar-
monic terms) in the Taylor expansion may become important.